Agostic interaction and intramolecular proton transfer from the protonation of dihydrogen ortho metalated ruthenium complexes

Abstract

Protonation of the ortho-metalated ruthenium complexes RuH(H₂)(X)(PⁱPr₃)₂ [X = 2-phenylpyridine (ph-py) (1), benzoquinoline (bq) (2)] and RuH(CO)(ph-py)(PⁱPr₃)₂ (3) with [H(OEt₂)₂]⁺[BAr′₄]⁻ (BAr′₄ = [(3,5-(CF₃)₂C₆H₃)₄B]) under H₂ atmosphere yields the corresponding cationic hydrido dihydrogen ruthenium complexes [RuH(H₂)(H-X)(PⁱPr₃)₂][BAr′₄] [X = phenylpyridine (ph-py) (1-H); benzoquinoline (bq) (2-H)] and the carbonyl complex [RuH(CO)(H-ph-py)(PⁱPr₃)₂][BAr′₄] (3-H). The complexes accommodate an agostic CH interaction characterized by NMR and in the case of 1-H by x-ray diffraction. Fluxional processes involve the hydride and dihydrogen ligands in 1-H and 2-H and the rotation of the phenyl ring displaying the agostic interaction in 1-H and 3-H. NMR studies (lineshape analysis of the temperature-dependent NMR spectra) and density functional theory calculations are used to understand these processes. Under vacuum, one equivalent of dihydrogen can be removed from 1-H and 2-H leading to the formation of the corresponding cationic ortho-metalated complexes [Ru(H₂)(THF)(X)(PⁱPr₃)₂]⁺ [X = ph-py (1-THF), bq (2-THF)]. The reaction is fully reversible. Density functional theory calculations and NMR data give information about the reversible mechanism of CH activation in these ortho-metalated ruthenium complexes. Our study highlights the subtle interplay between key ligands such as hydrides, σ-dihydrogen, and agostic bonds, in CH activation processes.

Catalytic transformation of alkanes and arenes via activation of an inert CH bond is of considerable interest and remains a challenge to chemists (1–13). Since the 1980s, many information have been gathered on the stoichiometric transformations of a CH bond at a transition metal center (14–26). Different mechanisms are operative depending on the metal, the ligand set and the nature of the media in which the reaction is performed. They differ, inter alia, by the way the RH moiety interacts with the transition metal before activation. Oxidative addition proceeds from a σ complex where the CH bond interacts as a Lewis base with the metal, whereas σ-bond metathesis does not require precoordination of the substrate. An alternative mechanism based on the properties of σ complexes is now under consideration for late transition metals. Such a σ-complex assisted metathesis mechanism (σ-CAM) allows substrate functionalization by σ-ligand substitution. It involves the interconversion of at least two σ complexes and no change in oxidation state, thanks to the intermediacy of secondary interactions (27).

The inert character of the CH bond is reflected by its poor properties as a ligand, even though alkane complexes have been observed (28, 29). It is thus necessary to promote the interaction of one particular CH bond to observe selective activation. A major breakthrough was proposed by Murai in 1993 with the selective insertion of an olefin into the aromatic CH bond ortho to an activating ketone group (Eq. 1) (30). In this system, the key intermediate is an ortho-metalated complex resulting from ortho CH bond activation, thanks to chelating assistance with the donor group. Coordination of the olefin, olefin insertion into Ru–H and CC coupling are the subsequent steps needed to close the catalytic cycle as proposed by Kakiuchi and Murai (8).

We have shown that, when using the bis-dihydrogen complex, Ru(H)₂(H₂)₂(PCy₃)₂, as catalyst precursor, the CC coupling of ethylene with acetophenone or benzophenone was catalytic at room temperature (31, 32). The activity was recently improved by replacing the tricyclohexylphosphines by two tricyclopentylphosphines (33). Moreover, we were able to isolate key intermediates of general formula RuH(H₂)(o-C₆H₅R)(PR′₃)₂, (R = COCH₃, COC₆H₅) that proved to be ortho-metalated species (31). These compounds with a ketone chelating group present a very limited solubility in most of the solvents. It was thus easier to perform an in-depth study on analogous complexes, with the R substituent replaced by an aromatic N-heterocycle. In such a case, chelation is assisted by nitrogen coordination to the metal center. Indeed, we recently reported the properties of a series of ortho-metalated ruthenium hydrido complexes Ru(H)(H₂)(X)(PiPr₃)₂ [X = 2-phenylpyridine (ph-py), benzoquinoline (bq), phenylpyrazole (ph-pz)] resulting from CH activation of the corresponding functionalized arene (34). These compounds display remarkable exchange couplings between the hydride and the dihydrogen ligand. Despite the nonactivity of these species toward the Murai reaction, we believed that a study focused on protonation could bring some general information, especially on the interplay between key ligands such as σ-dihydrogen or agostic bonds, as well as on hydrogen transfer processes.

The selectivity of the Murai's reaction relies on the existence of an agostic precursor, which corresponds to the very first step of the activation process of the arene substrate. We have already communicated about the characterization of such a complex, namely [RuH(H₂)(H-ph-py)(PⁱPr₃)₂]⁺[BAr′₄]⁻ (1-H), showing two coordinated σ-bonds: an agostic CH bond and an H₂ ligand (35). 1-H was prepared by protonation of the ortho-metalated complex Ru(H)(H₂)(ph-py)(PⁱPr₃)₂. We now report the results of a combined experimental and theoretical study on a series of agostic complexes aimed at describing their structure, the influence of the various ligands, as well as the CH activation processes occurring within these systems. Part of this work has been communicated (35).

Results

For further details, see supporting information (SI) Text, Figs. 7–12, and Tables 2–6.

Synthesis and Properties of Agostic Complexes.

Protonation of RuH(H₂)(X)(PⁱPr₃)₂ [X = (ph-py) (1), (bq) (2)] and RuH(CO)(phpy)(PⁱPr₃)₂ (3) by [H(OEt₂)₂]⁺[BAr′₄]⁻ (BAr′₄ = [(3,5-(CF₃)₂C₆H₃)₄B]) at −20°C in THF, under an atmosphere of 1 bar H₂ in the case of 1 and 2, yields the corresponding agostic cationic species [RuH(H₂)(H-ph-py)(PⁱPr₃)₂]⁺[BAr′₄]⁻ (1-H), [RuH(H₂)(H-bq)(PⁱPr₃)₂]⁺[BAr′₄]⁻ (2-H), and [RuH(CO)(H-ph-py)(PⁱPr₃)₂]⁺[BAr′₄]⁻ (3-H) (see Chart 1). Throughout the paper, we will specifically analyze the data concerning 1-H, because it is the most interesting example (for detailed data, see SI Text). As previously communicated, the x-ray structure of 1-H features two coordinated σ-bonds, agostic CH and HH, that are mutually cis (35). The ¹H NMR-spectra display in the aromatic region at room temperature broad peaks at 6.18 and 7.95 ppm for complex 1-H (see Fig. 1). Decoalescence and sharpening of the peaks are observed when lowering the temperature. The NMR spectra below 213 K show new peaks at 4.14 and 8.22 ppm, and 7.87 and 8.30 ppm, corresponding respectively to the protons H9 and H5 on one side and H6 and H8 on the other side of the agostic phenyl ligand. The peak at 4.14 ppm was identified as H9 using an HH-COSY spectrum (see SI Text). The coupling between H8 and H9 is clearly observed by the cross peak at 4.14 ppm/8.30 ppm. The rate of the exchange between H5 and H9 was determined by lineshape analysis and reflects the dynamics of the phenyl ring rotation. The dependence on temperature can be expressed by the Arrhenius type equation: k = A exp(−E_a/RT) where k is the corresponding exchange rate, A is the frequency factor, E_a is the activation energy of the observed process, R is the ideal gas constant (8.314 J/(mol·K)), and T is the temperature (34). We found that k = 10^(11.3±0.3)exp((−35.6 ± 1.8) kJ·mol⁻¹/RT) s⁻¹ (173 K ≤ T ≤ 303 K; k = 18,000 s⁻¹ at 263 K). The value of the activation energy of the phenyl rotation process is the addition of the bond dissociation energy of the agostic interaction plus the intrinsic barrier of the phenyl ring rotation itself. The present value of E_a (35.6 kJ mol⁻¹) in complex 1-H is low, in agreement with the weak coordination of the CH bond. The chemical shift found for the frozen agostic proton H9 (4.14 ppm) is similar but shifted upfield compared with that observed by Crabtree and coworkers (36) in an agostic diphenylpyridine ligand (5.83 ppm). The agostic carbon was identified at 111.2 ppm for 1-H, with a coupling constant J_CH of 128 Hz (THF-d₈, 183 K) as determined by a 1D-gs-HMQC experiment. The value of 128 Hz is relatively high for a CH coupling constant of an agostic interaction that suggests a very weak activation of the CH bond (see below) (37).

Chart 1.

Structures of the ortho-metalated complexes 1–3, 1-THF, and 2-THF and the corresponding agostic complexes 1-H, 2-H, and 3-H.

Fig. 1.

Aromatic region of the experimental (top line) and simulated (bottom line) ¹H-NMR spectra (500.13 MHz) of [RuH(H₂)(H-ph-py)(PⁱPr₃)₂]⁺[BAr′₄]⁻ (1-H) under 800 mbar H₂ in THF-d₈. Only the exchange between H5 and H9 was simulated.

The hydride region in the NMR spectrum shows at room temperature one broad signal at −13.7 ppm for the hydride and the dihydrogen ligands in fast mutual exchange. Further evidence of the presence of a dihydrogen ligand was supported by T₁ measurements (T_1min of 16 ms at 243 K, THF-d₈, 300 MHz) and a J_HD value of 23 Hz obtained by deuteration of 1-H. This J_HD value leads to a calculated HH distance of 0.93 Å, whereas the T_1min measurement leads to a value of 0.89 Å by using a fast rotation model (38). Decoalescence is observed at low temperature, leading to a triplet at −14.12 ppm (J_PH = 18 Hz) for the hydride resonance and to a broad signal at −13.02 ppm for the dihydrogen ligand. The Arrhenius parameters for this process have been determined by lineshape analysis (see SI Text).

The activation energy of the classical hydride dihydrogen exchange in complexes 1-H and 2-H (47 kJ mol⁻¹) is similar to those found for complexes 1 and 2 (40.0 and 42.3 kJ mol⁻¹, respectively) (34). The ligands and the charge have little effect on the observed classical exchange processes. The spectra of complexes 1-H and 2-H do not present temperature-dependent quantum-mechanical exchange couplings as were displayed by complexes 1 and 2. The nonclassical hydride dihydrogen exchange process might be too fast to be observed by liquid state NMR spectroscopy in 1-H and 2-H. Due to a reduced electron density on the positively charged complexes, the dihydrogen exchange barrier might be very low and therefore the quantum mechanical exchange fast (>10⁴ Hz) to be observed by liquid-state NMR spectroscopy (39).

We have previously described that the dihydrogen ligand in 1 can easily be substituted by N₂, O₂, or C₂H₄ (34). In the context of the Murai reaction, it was interesting to test the protonation of the ethylene complex RuH(C₂H₄)(ph-py)(PⁱPr₃)₂ (4). Because 4 loses immediately ethylene in solution, the protonation with [H(OEt₂)₂]⁺[BAr′₄]⁻ was performed under an atmosphere of 800 mbar ethylene in THF-d₈. When the reaction was carried out below room temperature, a new complex, [Ru(H₂)(THF)(ph-py)(PⁱPr₃)₂]⁺ (1-THF), was formed exclusively (see below and Chart 1). At higher temperatures, a mixture of products was observed and formation of ethane was detected.

Proton Transfer Reactions.

1-H and 2-H are not stable in solution. They transform into new orthometalated complexes [Ru(H₂)(THF)(X)(PⁱPr₃)₂]⁺ [X = ph-py (1-THF), bq (2-THF)] if H₂ is removed from the THF solution by pumping or upon heating to 65°C in a THF-d₈ solution inside an NMR tube (see Chart 1 and Scheme 1). THF adducts could not be isolated but their properties were studied by NMR spectroscopy and density functional theory (DFT) calculations (see below). All of the data support the preferred geometry shown in Scheme 1. The NMR data obtained for 1-THF are in agreement with the presence of a single dihydrogen molecule and the absence of any hydride or agostic proton in the coordination sphere of the metal. A broad signal was observed at −7.53 ppm which displays a short T_1min value of 11 ms at 263 K and 300 MHz. Upon deuteration, the hydride signal transforms into a 1:1:1 triplet (J_HD = 27 Hz) in agreement with the presence of an unstretched dihydrogen ligand. Moreover, the presence of a metalated carbon is inferred from the resonance at 169 ppm in the ¹³C{¹H} NMR spectrum. The reaction can be reversed when a solution of 1-THF is placed under H₂. Therefore, this reaction represents an easy and reversible CH activation process at room temperature. It is worth noting that a similar process was reported by Crabtree and coworkers (36) for an iridium complex with a diarylpyridine ligand and by Milstein and coworkers (40–42) for a rhodium complex with an aryl phosphane ligand. It is interesting that, upon placing 1-H in a deuterium atmosphere, exclusive deuteration of the ortho positions of the phenyl ring is observed, demonstrating the presence of an equilibrium between 1-H and a metalated intermediate.

Scheme 1.

Protonation in THF of RuH(H₂)(ph-py)(PⁱPr₃)₂ (1).

Computational Studies

Structure of the Agostic Complexes.

To gain further insight into the structure and the dynamic properties of the agostic complexes, model systems [Ru(H)(H-ph-py)(L)(PMe₃)₂]⁺ (L = H₂, 1q-H; L = CO, 3q-H; L = C₂H₄, 4q-H; L = vacant site, 5q-H) were optimized at the B3PW91 level. In all these complexes, the agostic CH bond is trans to the hydride ligand as was the case in the structure of 1-H.³⁵ The optimized geometry for 1q-H is in excellent agreement with the experimental data. The agostic Ru···C bond distance [2.528(3) Å, 1-H; 2.513 Å, 1q-H] and the dihedral angle between the two rings [27.5(4)°, 1-H; 24.1°, 1q-H] are particularly well reproduced. The agostic CH bond is only slightly elongated in comparison to the other aromatic CH bonds within the phenyl ring (1.123 Å, 1q-H), as a result of the weak interaction with the metal center. The DFT calculations confirm the presence of an H₂ ligand trans to N. The HH bond distance is longer than in 1-H [0.82(4) Å, 1-H; 0.946 Å, 1q-H] but compares well with the value deriving from NMR data (0.93 Å, J_HD; 0.89 Å, T_1min).

In the case of 3-H, and in the absence of any x-ray data, it was difficult to ascertain the geometry of the ligands around the ruthenium center. As formation of an agostic interaction and subsequent activation of the CH bond are critically influenced by the respective nature of the ligand cis and trans to the agostic bond, DFT calculations have been performed on [Ru(H)(H-ph-py)(CO)(PMe₃)₂]⁺ with the hydride trans (3q-H) and cis (3qcis-H) to the agostic bond. Even though the corresponding experimental systems were not characterized, the calculations were also carried out on the two isomers of [Ru(H)(H-ph-py)(C₂H₄)(PMe₃)₂]⁺ (4q-H and 4qcis-H) and [Ru(H)(H-ph-py)(PMe₃)₂]⁺ (5q-H and 5qcis-H). A structure corresponding to 1qcis-H could not be found as a local minimum, and the optimization procedure yielded 1q-H for different starting geometries featuring an H₂ ligand trans to the agostic bond. The DFT calculations indicate that the isomers with the hydride trans to the agostic bond are more stable than the other isomers (Table 1). The energy difference of 25.1 kJ mol⁻¹ between 3q-H and 3qcis-H is large enough to set 3q-H as a model for 3-H. This finding is in agreement with the preferred structure proposed from the NMR studies. Selected geometrical parameters pertinent to the agostic interaction are given in Table 1. For the carbonyl and ethylene species, the relative energies of the two isomers are similar.

Table 1.

Selected bond distances (Å) for the agostic interaction, calculated chemical shift δ_H (ppm) of the agostic proton, calculated coupling constant J_CH (Hz), relative energy ΔE (kJ mol⁻¹), and activation energy ΔE^#(Ph) (kJ mol⁻¹) for the phenyl-ring rotation for the calculated agostic species [Ru(H)(H-ph-py)(L)(PMe₃)₂]⁺

	1q-H, L = H2	3q-H, L = CO	3qcis-H, L = CO	4q-H, L = C2H4	4qcis-H, L = C2H4	5q-H, no L	5qcis-H, no L
CH	1.123	1.119	1.112	1.122	1.134	1.125	1.277
Ru·C	2.513	2.553	2.568	2.548	2.445	2.506	2.120
Ru·H	1.918	1.954	1.980	1.913	1.839	1.908	1.620
δH	4.6	4.2	3.2	4.9	2.7	6.7	−21.9
JCH	120.6	122.6	120.0	121.5	112.3	121.0	78.7
ΔE	—	0.0	25.1	0.0	23.7	0.0	35.9
ΔE#(Ph)	35.0	30.4	35.9	29.5	51.8	35.8	83.3

In this context, NMR measurements are particularly suited to investigate agostic interactions. Low temperature experiments allow access to useful information on the dynamic of the system (chemical shifts), and the magnitude of the agostic interaction can be monitored by J_CH values. NMR calculations of the chemical shift of the agostic proton and of the J_CH coupling constant were carried out with the hybrid PBE1PBE functional and IGLOO-II basis sets (see SI Text). In the case of 1-H, low temperature NMR data yielded an agostic interaction characterized by a chemical shift of 4.14 ppm and a J_CH coupling constant of 128 Hz. The calculated values of 4.6 ppm and 120.6 Hz for 1q-H compare well with the experimental values. In the case of 3-H, the low temperature NMR data yielded a proton chemical shift of 4.31 ppm and a J_CH coupling constant of 130 Hz for the agostic bond. The calculated values of 4.2 ppm and 122.6 Hz for 3q-H are also in better agreement with experimental NMR data than those of 3qcis-H (3.2 ppm and 120 Hz).

Phenyl Ring Rotation.

As shown above, we have evidenced by NMR an exchange process of the H5, H9 protons due to phenyl ring rotation. We have located, on the potential energy surface, a transition state (TS) for the phenyl ring rotation, 1q-H^TSROT (Fig. 2), at 35.0 kJ mol⁻¹ above 1q-H (Table 1). In 1q-H^TSROT, the phenyl ring is perpendicular to the pyridine ring (dihedral angle 92.7°), and the CH bond, which was agostic in 1q-H (1.123 Å), now presents a normal value (1.086 Å). The agostic interaction has thus disappeared in the TS. This finding is confirmed by the shortening of the Ru-H distance trans to the agostic (1.571 Å, 1q-H; 1.547 Å, 1q-H^TSROT). The Ru-N bond also shortens in 1q-H^TSROT (2.123 Å, 1q-H; 2.101 Å, 1q-H^TSROT) as a result of the relief of strain in the agostic interaction. The HH distance in the H₂ ligand remains unchanged (0.94 Å). The calculated value for the activation barrier associated to phenyl ring rotation (35.0 kJ mol⁻¹) is in excellent agreement with the experimental value obtained for ortho-H exchange in 1-H (35.6 kJ mol⁻¹). The calculations thus support phenyl ring rotation as the origin of the observed H-site exchange process.

Fig. 2.

B3PW91 optimized TS geometries. (Left) 1q-H^TSROT for phenyl-ring rotation within the agostic complex 1q-H. (Right) 1q-H^TSTR, for H-transfer from the agostic complex 1q-H.

The calculated activation barrier of 30.4 kJ mol⁻¹ for 3q-H^TSROT, is not in such a good agreement with the experimental value (44.5 kJ mol⁻¹, for 3-H). To gain further insight in the properties of these systems, calculations were also carried out on complexes not experimentally observed. The transition state for phenyl ring rotation of all of the agostic complexes were optimized (Table 1). They exhibit similar geometrical features with perpendicular phenyl and pyridine rings. When a hydride ligand is trans to the agostic bond, the activation barrier for phenyl ring rotation is rather insensitive to the actual nature of the ligand cis to the agostic bond (see 1q-H, 3q-H, 4q-H, and 5q-H in Table 1). When the hydride ligand is cis to the agostic bond, the activation barrier experiences larger variation depending on the nature of the trans ligand. Absence of the latter drastically increases the activation barrier (83.3 kJ mol⁻¹, 5qcis-H), and the reduction of the activation barrier correlates with the π-accepting properties of the ligand (51.8 kJ mol⁻¹ for L = C₂H₄; 35.9 kJ mol⁻¹ for L = CO); the more π-acid the ligand trans to the agostic, the easier the rotation. It is interesting to note that, in 4qcis-H, the C₂H₄ ligand lies in the equatorial plane defined by Ru, pyridine, and hydride. Therefore, the agostic bond is not competing for the same t_2g orbital in the back-donation interactions. The situation is different with CO in 3qcis-H as the latter possesses two π-accepting MOs, and there is thus competition with the CH agostic bond for the same t_2g orbital. This competition weakens the agostic interaction as illustrated by both the geometrical parameters and the values of the activation barrier for phenyl ring rotation.

Proton Transfer Studies.

The complex 1-H, featuring two coordinated σ-bonds in cis position, is not stable in solution and transforms into an orthometalated complex 1-THF stabilized by solvent (THF) coordination (Scheme 1). We first examined the hydrogen transfer from 1q-H (the model for 1-H), to 1q-H₂ (the model for 1-THF) but replacing THF by H₂ (see Scheme 2). The transition state, 1q-H^TSTR, for the hydrogen transfer from the agostic carbon to Ru has been located on the potential energy surface and lies 75.4 kJ mol⁻¹ above 1q-H (Fig. 2). The transition state connects to the bis–dihydrogen complex 1q-H₂ at 8.2 kJ mol⁻¹ above 1q-H. The Ru-C bond is almost formed in the TS (2.152 Å) and one hydrogen atom of the H₂ ligand that was cis to the agostic bond in 1q-H, has been transferred to the hydride ligand trans to the forming Ru-C bond to accommodate the change in nature of the ligand from weak (agostic) to strong (aryl). This new HH interaction in 1q-H^TSTR is typical of an H₂ ligand (0.862 Å). As a consequence, there is now a basic site (the hydride) cis to the transferring H and the H···H distance of 1.711 Å is typical of a dihydrogen bond but the transferring H is also close to Ru (1.56 Å). The TS in 1q-H^TSTR could thus be described as a Ru^IV dihydride dihydrogen complex. Whatever the nature of the TS, the final product is a Ru^II bis–dihydrogen complex and no Ru^IV intermediate could be located on the PES.

Scheme 2.

Proton transfer from the agostic species 1q-H to the orthometalated species 1q-H₂ and 1q-Me₂O (energies in kJ mol⁻¹).

The bis–dihydrogen complex 1q-H₂ presents two potential labile ligands and, upon substitution of one H₂ ligand by Me₂O (as a model for THF), two different complexes were optimized. The isomer with Me₂O cis to the phenyl ring, 1qcis-Me₂O, lies at 45.0 kJ mol⁻¹ above 1q-H, whereas the other isomer with Me₂O trans to the phenyl ring, 1q-Me₂O, lies at 25.0 kJ mol⁻¹ above 1q-H (Scheme 2). Substitution of the H₂ ligand trans to the strongest σ-donor ligand (aryl) is clearly favored. The complex 1q-Me₂O is 16.8 kJ mol⁻¹ less stable than the bis–dihydrogen complex 1q-H₂, yet the experimental observations point to the formation of an orthometallated complex with only one coordinated H₂ molecule. It should be noted that 1-THF was obtained when H₂ was removed from the THF solution by pumping or upon heating at 65°C. Heating is needed to overcome the activation barrier for proton transfer. In solution with a good coordinating solvent such as THF, we may assume that the more labile H₂, trans to the aryl, dissociates and is replaced by a solvent molecule. It is interesting to note that the two complexes, 1q-H and 1q-H₂, are very close in energy, in agreement with the reversibility of the reactions observed experimentally. The labeling experiments also support the proposed mechanism. Reaction of 1-H with D₂ leads to exclusive deuteration of the ortho positions of the phenyl ring in agreement with an equilibrium between 1-H and a bis–dihydrogen metalated intermediate analogous to the model 1q-H₂. A succession of H(D)-transfer reactions, coupled with phenyl ring rotation will eventually incorporate deuterium exclusively at the ortho-positions.

Discussion

The Agostic Interaction.

Formation of an agostic interaction of the CH bond ortho to the activating group is a key step in the Murai's reaction. The complexes 1-H, 2-H, and 3-H present such an interaction, and NMR studies highlighted various aspects of the interaction. The strength of any agostic interaction is generally inferred from the reduction of the J_CH coupling constant and the high field shift of the proton resonance. In the present case, both indicators are very close. The calculated trend for the J_CH coupling constant is faithfully reproduced but the proton chemical shift for 3q-H (4.2 ppm) is calculated lower than 1q-H (4.6 ppm) contrary to experiment. However, proton chemical shifts are difficult to compute accurately, particularly when interaction with a metal is present.

It is always difficult to estimate the intrinsic strength of an agostic interaction because definition of a strictly agostic-free configuration is not obvious. In our complexes, the energy of the H-site exchange process of the ortho hydrogen atoms was evaluated from NMR studies at variable temperature. The exchange was shown to proceed along a TS featuring a phenyl ring perpendicular to the pyridine ring (Fig. 2). In such a configuration the CH ortho bonds are clearly not agostic. The activation barrier could thus serve to estimate the strength of the agostic interaction in 1-H and 3-H. However, the phenyl-pyridine ligand possesses an intrinsic rotation barrier which was estimated to be 19.6 kJ mol⁻¹ from DFT(B3PW91) calculations. This allows the estimation of the strength of the agostic interaction in 1-H and 3-H, namely 16 kJ mol⁻¹ and 24.9 kJ mol⁻¹, respectively. As expected, these agostic interactions are weak in nature, and the calculated values compare well with the experimental data of 41.8 ± 25 kJ mol⁻¹ obtained for intramolecular W···HC agostic interactions in W(CO)₃(PCy₃)₂ by Hoff and coworkers (43). The above results point to a slightly stronger interaction in 3-H than in 1-H, contrary to what has been inferred from the proton chemical shift and J_CH coupling constant values; this illustrates the difficulty to obtain quantitative information on such weak interactions.

To gain further insight, calculations on systems not observed experimentally were carried out. In these complexes, an agostic interaction is always present but the nature of the ligands cis and trans to the agostic bond is varied. The value of the activation barrier for phenyl-ring rotation can be used as an indicator of the strength of the agostic interaction. From the results in Table 1, for a given ligand trans to the agostic (i.e., hydride), changing the nature of the cis ligand does not drastically alter the general features of the agostic interaction (see 1q-H, 3q-H, 4q-H, and 5q-H). The CH bond distances have very similar values (≈1.12 Å), and the J_CH coupling constants have values lying within 2 Hz. Also, the activation barriers for phenyl-ring rotation are very close with the lowest value (29.5 kJ mol⁻¹, 4q-H), only 6.3 kJ mol⁻¹ smaller than the highest one (35.8 kJ mol⁻¹, 5q-H).

When the ligand in trans is varied, while keeping the same ligand in cis (i.e., hydride), the agostic interaction experiences larger variation. In the extreme case of 5qcis-H, where no trans ligand is present, strong CH activation is observed. The CH bond is strongly elongated and the Ru-C bond is formed (Table 1). The proton chemical shift (−21.9 ppm) is also in agreement with a strong Ru···H interaction. The complex could be considered as a Ru^IV dihydride. The relative energy between the two isomers (35.9 kJ mol⁻¹) is in agreement with the description of 5qcis-H as a Ru^IV complex lying at higher energy than the Ru^II isomer 5q-H. With a strong σ-donor and π-acceptor such as CO (3qcis-H) the situation is not very different from that for the pure σ-donor H (3q-H). Changing the trans ligand for a weaker σ-donor, such as C₂H₄ (4qcis-H), strengthens the agostic interaction. In particular, the J_CH coupling constant is significantly reduced (112.5 Hz) and the activation barrier for phenyl-ring rotation is substantially larger (51.8 kJ mol⁻¹). All of the above results point to a critical influence of the σ-donor properties of the ligand trans to the agostic bond to tune this interaction.

The agostic interaction is generally described as the result of two synergistic electron transfers: σ-donation from the CH bond to the metal and π back-donation from the metal in the σ*(CH) orbital. Within the natural bonding orbital (NBO) scheme, these two contributions have been evaluated with the second-order donor–acceptor perturbation procedure of NBO (44). These energetic stabilizations reflect the magnitude, not in an absolute way, of the electron transfer. As a result of these hyperconjugative interactions, the natural localized molecular orbital (NLMO) of the σ(CH) agostic bond does contain some Ru character and the extent of the hyperconjugative interaction is reflected by the weight of the parent NBO in the NLMO. Stronger delocalization is associated to lower NBO weight. The results clearly show that donation from the σ(CH) bond toward the σ*(Ru-L_t) antibond (between Ru and the ligand L_t trans to the agostic bond) is the major contribution to the agostic interaction. For the systems with the hydride trans to the agostic, the characteristics of the agostic interaction are very similar. The actual nature of the cis ligand only slightly modifies the electronic impact of the trans ligand on the strength of the agostic. For 3qcis-H, switching the role of H and CO does not lead to a significant alteration of the agostic interaction as the σ-donor properties of H and CO are very similar. It clearly illustrates that the π-accepting properties of the trans ligand have no strong influence on the agostic bond. The major difference observed in 4qcis-H is the large increase in the σ-donation contribution from σ(CH). This allows significant hyperconjugative delocalization as illustrated by the Ru-content of the NLMO (4.9%) and by the lower weight of the parent NBO in the NLMO (90.9%). There is thus clearly a stronger agostic interaction in 4qcis-H originating from the lower σ-donor character of the ethylene ligand trans to the agostic bond.

Proton Transfer Reactions.

In the Murai's system, after coordination of the ortho CH bond, CH activation is necessary to obtain an ortho-metalated complex as an intermediate before the CC coupling step. To design more efficient catalysts, it is thus desirable to better understand the factors influencing this CH bond breaking step. To test the influence of the presence of a cis hydride on the hydrogen transfer, the two TS, 5qcis-H^TSTR and 5q-H^TSTR, originating from the agostic complexes with one vacant site remaining in the coordination sphere of the metal, 5qcis-H and 5q-H, respectively, have been optimized. The most striking differences between the two systems lie in the values of the activation barriers: 0.5 kJ mol⁻¹ for 5qcis-H^TSTR and 122.8 kJ mol⁻¹ for 5q-H^TSTR. From the geometrical parameters, both TS could be described as Ru^IV complexes (agostic CH bond broken and formation of strong Ru-H and Ru-C bonds), but the difference in activation barrier is difficult to explain. One may argue that the absence of a ligand trans to the forming RuC bond in 5qcis-H^TSTR, whereas a strong σ-donor is present in 5q-H^TSTR, is the reason for the easier CH activation in the former. As a matter of fact, the activation barrier for 1q-H^TSTR is 75.4 kJ mol⁻¹ with an H₂ ligand trans to Ru–C.

To further test the influence of the nature of the ligand trans to the agostic bond in the proton transfer, the TS with Me₂O trans to the agostic, 6qcis–Me₂O^TSTR, has been optimized. The activation barrier is only 6.4 kJ mol⁻¹, thus showing the critical influence of the nature of the trans ligand. The agostic complex 6qcis–Me₂O, before proton transfer, exhibits a very strong agostic interaction as inferred from the CH bond distance (1.200 Å) and the calculated J_CH coupling constant (91.3 Hz). These results indicate that there is a good correlation between the strength of the agostic interaction and the activation of that bond. A strong agostic interaction correlates with an easy CH activation process.

Complexes 1-H and 2-H exhibit reactivity patterns in agreement with reversible CH bond breaking and bond making processes. The calculations have located the TS for H-transfer in 1q-H (Fig. 2) and the main features of this TS is the reorganization of the hydrides in the coordination sphere of the metal. Concomitant to the CH breaking process, there is a transfer of one hydrogen of the H₂ ligand that was cis to the agostic in 1q-H to the hydride trans to the agostic. In 1q-H^TSTR, there is now an optimal organization of the weak and strong ligands to accommodate both the formation of Ru-C (trans to H₂) and the transfer of H as a proton to the basic nearby hydride (trans to N). The geometrical features of this TS could be interpreted in two different ways. The short Ru-H distances for the transferring H and the cis hydride (1.56 and 1.60 Å, respectively), together with the already formed Ru-C bond (2.152 Å) could allow the description of 1q-H^TSTR as a Ru^IV dihydrogen–dihydride complex. The product of the transfer being a bis–dihydrogen Ru^II complex, this process would be akin to the oxidative addition hydrogen transfer mechanism as described by Goddard and coworkers (45) in the case of hydroarylation catalysts. Another alternative, as the H···H distance between the transferring H and the cis hydride is 1.711 Å, is the formation of a very stretched H₂ ligand in the TS, or at least of a dihydrogen bonded stabilized H-transfer. This would correspond to a Ru^II bis–dihydrogen TS and the transfer could be described as a prototype of a σ-CAM mechanism, as described by Perutz and Sabo-Etienne (27).

In summary, we have examined the protonation of a series of ortho-metalated Ru complexes by combining NMR and DFT studies. These complexes can serve as models for one of the key species in the Murai reaction. Our findings on the resulting agostic complexes indicate a critical influence of the σ-donor properties of the ligand trans to the agostic bond to tune this interaction and give rise to subsequent CH activation. We hope that further studies will enable us to develop new catalysts with reasonably strong agostic bonds that promote easier CH activation processes.

Materials and Methods

All reactions were carried out under argon by using Schlenk glassware and vacuum line or glove box techniques. Complexes RuH(H₂)(X)(PⁱPr₃)₂ [X = 2-phenylpyridine (ph-py) (1), benzoquinoline (bq) (2)], and RuH(CO)(ph-py)(PⁱPr₃)₂ (3) were prepared according to the procedures described in ref. 34. [H(OEt₂)₂]⁺[BAr′₄]⁻ (BAr′₄ = [(3,5-(CF₃)₂C₆H₃)₄B]) was prepared according to ref. 46. See SI Text for 2-H, 3H, and 2-THF. All of the calculations have been performed with the Gaussian03 package, at the B3PW91 level; see SI Text for more details.

[RuH(H₂)(H-ph-py)(PⁱPr₃)₂]⁺[Bar′4]⁻ (1-H).

A Fisher-Porter bottle was loaded with RuH(H₂)(ph-py)(PⁱPr₃)₂ (1) (0.46 g, 0.49 mmol) and [H(OEt₂)₂]⁺[BAr′₄]⁻ (0.79 g, 0.79 mmol) and cooled to −78°C. After ≈10 min, 15 ml of THF was added, yielding a pale orange solution whereby the solids remained essentially insoluble at this temperature. The bottle was then immersed in liquid nitrogen and the heterogeneous mixture thoroughly degassed during three freeze and thaw cycles. The bottle was then pressurized with H₂ (3 bar) at −78°C. Warming to room temperature over a period of ≈45 min, led to the formation of an orange solution; no gas evolution was observed during the reaction even at room temperature. Subsequently, the solution was pressurized with H₂ (3 bar). Orange crystals of 1-H, suitable for x-ray diffraction analysis, were obtained at room temperature under H₂ (0.54 g, 70%). The compound is best stored under an atmosphere of H₂ at −20°C; it rapidly decomposes in the solid state in vacuo. See SI Text for NMR data.

[Ru(H₂)(ph-py)(THF)(PⁱPr₃)₂]⁺[Bar′₄]⁻ (1-THF).

Complex 1 (20 mg) was mixed with 35 mg [H(OEt₂)₂]⁺[BAr′₄]⁻ in a NMR-tube inside a glove-box. THF (1.5 ml) was condensed into the tube and slowly warmed up to room temperature. Significant gas evolution was observed (degassing from time to time ensured that no overpressure was in the tube). Complete protonation was achieved by using an ultrasonic bath until total dissolution and no more gas evolution (≈5 min). THF was evaporated, and the solid was washed three times with 1 ml of pentane. The resulting solid was dried for 1 h at 10⁻⁶ torr. After this time, 1 ml of THF-d₈ was condensed on the solid, and the sample was flame sealed. See SI Text for NMR data.

Abbreviations

DFT

density functional theory

TS

transition state.