An osmium(II) methane complex: Elucidation of the methane coordination mode

Abstract

The activation of inert C─H bonds by transition metals is of considerable industrial and academic interest, but important gaps remain in our understanding of this reaction. We report the first experimental determination of the structure of the simplest hydrocarbon, methane, when bound as a ligand to a homogenous transition metal species. We find that methane binds to the metal center in this system through a single M···H-C bridge; changes in the ¹J_CH coupling constants indicate clearly that the structure of the methane ligand is significantly perturbed relative to the free molecule. These results are relevant to the development of better C─H functionalization catalysts.

A detailed experimental determination shows how the simplest hydrocarbon, methane, binds to a homogenous transition metal species.

INTRODUCTION

The efficient and scalable conversion of hydrocarbon feedstocks (such as petroleum and natural gas) to higher-value commodity or specialty chemicals is an important goal of modern catalysis, and in part for this reason, the activation of C─H bonds by molecular transition metal complexes has been the subject of significant and growing interest since the 1980s (1–13). Particularly impressive has been the use of directing groups to control the functionalization of specific C─H bonds (14–16). Despite these advances, the development of mild and selective C─H activation methods that do not rely on directing groups remains an important challenge (2, 5, 17–21).

One mechanism by which alkane C─H bonds can be activated by a transition metal complex involves, as a first step, coordination of the alkane to the metal center. Such compounds, which are known as σ-alkane complexes, were initially studied by matrix isolation methods (22, 23) and have subsequently been studied by a variety of other techniques (24–31). Despite these advances, there is little detailed experimental information about exactly how the alkane binds to the metal, especially in solution. One significant challenge in studying σ-alkane complexes in solution has been their low stability; the metal-alkane interaction is very weak, usually on the order of ~10 kcal mol⁻¹. Consequently, solution information on the smallest and most weakly coordinating alkanes, particularly methane, is scarce, and in no case has the coordination mode of the methane ligand been determined experimentally (32–34).

In 1998, we reported that the low-temperature protonation of the osmium(II) methyl complex (C₅Me₅)Os(dmpm)(CH₃), where dmpm is the chelating diphosphine Me₂PCH₂PMe₂, affords the methyl/hydride complex [(C₅Me₅)Os(dmpm)(CH₃)H]⁺ (35). This compound was the first example of an alkyl/hydride complex that, on the nuclear magnetic resonance (NMR) time scale, is in dynamic equilibrium with its alkane tautomer: Interconversion of the methyl/hydride complex with [(C₅Me₅)Os(dmpm)(CH₄)]⁺ occurs at a rate of 170 s⁻¹ at −100°C. Density functional theory (DFT) calculations indicated that this σ-methane tautomer is about 5 kcal mol⁻¹ higher in energy than the methyl/hydride complex and that variation of the ancillary ligands can reverse this energy difference so that the σ-methane tautomer is the predominant species in solution (36–38).

Guided by those DFT calculations, we report here the synthesis of a new osmium(II) σ-methane complex and its characterization by low-temperature NMR spectroscopy. Isotopic labeling studies have enabled us to obtain detailed experimental information about how the methane ligand binds to the metal center and to what extent this interaction changes its structure.

RESULTS AND DISCUSSION

In our earlier work, we found that [(C₅Me₅)Os(dmpm)(CH₃)H]⁺ is in equilibrium with the σ-methane tautomer [(C₅Me₅) Os(dmpm)(CH₄)]⁺ in solution but that the equilibrium concentration of the latter is too low to detect directly (35). This result suggested that modifying the ancillary ligands to decrease the electron richness of the osmium center should lower the energy of the σ-methane tautomer by disfavoring oxidative addition of the C─H bond. We used benchmarked DFT calculations to evaluate the relative energies of [(C₅R₅)Os(diphosphine)(CH₃)H]⁺ and [(C₅R₅)Os(diphosphine)(CH₄)]⁺ complexes bearing a variety of substituted cyclopentadienyl and diphosphine ligands (38). The calculations, which confirmed our initial expectations, predicted that replacement of the dmpm ligand with the more electron-withdrawing phosphine (CF₃)₂PCH₂P(CF₃)₂ (dfmpm) should render the σ-methane complex more stable than the methyl/hydride tautomer by 1.7 kcal mol⁻¹, while keeping the barrier for methane dissociation large enough (12.1 kcal mol⁻¹) to observe the complex at low temperatures. The calculations suggested that the stabilizing effect of dfmpm on the methane ligand is a result of a higher ratio of CH σ → Os d donation to Os d → CH σ* backdonation (38).

Although dfmpm has been synthesized previously (39), the reported method is not easily scalable to quantities greater than ~1 mmol. Accordingly, we developed an alternative preparation (see the Supplementary Materials) that provides multigram quantities of dfmpm from the commercially available diphosphine Cl₂PCH₂PCl₂; a key improvement of our protocol is the use of the Ruppert-Prakash reagent (CF₃SiMe₃/CsF) as a CF₃ source rather than (CF₃)₂PI (40).

The osmium methyl complex (C₅Me₅)Os(dfmpm)(CH₃) was prepared (see the Supplementary Materials) in 54% overall yield in two steps from (C₅Me₅)₂Os₂Br₄ (Fig. 1) (41, 42); the second step, alkylation of (C₅Me₅)Os(dfmpm)Br, required the use of dimethylzinc because methyllithium and methylmagnesium reagents deprotonate the CH₂ group in the backbone of the dfmpm ligand, and trimethylaluminum activates the C─F bonds of the phosphine. The Os-CH₃ resonance in the ¹H NMR spectrum is a triplet (J_HP = 7.5 Hz) of septets (J_HF = 2 Hz); the septet splitting shows that there is a remarkably strong five-bond coupling between the methyl protons and 6 of the 12 fluorine atoms of the dfmpm ligand. The ¹J_CH coupling constant for the Os-CH₃ group is 131 Hz. The monomeric nature of the complex is confirmed by its single-crystal x-ray structure of (C₅Me₅)Os(dfmpm)(CH₃), which features an Os─C bond length of 2.177 (3) Å and a diphosphine bite angle of 70.14 (2)° (Fig. 2).

Fig. 1. Synthesis of (C₅Me₅)Os(dfmpm)(CH₃) and the isolated yields for each step.

We have previously reported a preparation for the (C₅Me₅)₂Os₂Br₄ starting material (41).

Fig. 2. An ORTEP view of the x-ray crystal structure of (C₅Me₅)Os(dfmpm)(CH₃).

All atoms are shown as 30% probability ellipsoids except for hydrogen atoms, which are shown as arbitrarily sized spheres.

Protonation of (C₅Me₅)Os(dfmpm)(CH₃) with either trifluoromethanesulfonic acid or bis(trifluoromethanesulfonyl)amine at −110°C in CDCl₂F (43) gives a product 1 that exhibits a singlet at the unusual chemical shift of δ −1.94 in the ¹H NMR spectrum. This value lies between the ¹H NMR chemical shift for the parent methyl resonance (δ 0.36) and the ¹H NMR chemical shifts of related osmium hydrides (δ −10 to −20) (42). Protonation of the ¹³C-labeled isotopolog (C₅Me₅)Os(dfmpm)(¹³CH₃) causes the ¹H NMR resonance at δ −1.94 to split into a doublet with J_CH = 126 Hz; the ¹³C NMR spectrum of the same species features a binomial pentet at δ −45.11 with the same coupling constant (Fig. 3). The size of the average C─H coupling constant, which is similar to the values of 124 and 127 Hz seen for Rh(PONOP)(CH₄)⁺ and (C₅H₅)Os(CO)₂(CH₄), respectively (32–34), rules out the possibility that the protonation product 1 is a classical osmium cis-methyl/hydride complex in which the hydride and methyl hydrogen atoms are exchanging rapidly on the NMR time scale. If this were the case, the three ¹J_CH coupling constants within the Os-CH₃ group would be approximately equal to the 131 Hz value seen in (C₅Me₅)Os(dfmpm)(¹³CH₃), whereas the ²J_CH coupling constant between the methyl carbon and the Os─H group would be close to zero: Exchange averaging would produce a J_CH coupling constant of at most ~100 Hz, which is inconsistent with the observed J_CH of 1. We therefore conclude that 1 is the methane coordination complex [(C₅Me₅)Os(dfmpm)(¹³CH₄)]⁺.

Fig. 3. ¹H and ¹³C NMR resonances for the coordinated methane ligand in [(C₅Me₅)Os(dfmpm)¹³CH₄]⁺ at −110°C in CDCl₂F.

A small quartet at δ −45.4 in the ¹³C NMR spectrum is due to residual (unprotonated) (C₅Me₅)Os(dfmpm)(¹³CH₃).

Above −110°C, [(C₅Me₅)Os(dfmpm)(CH₄)]⁺ releases methane (figs. S1 and S2) by a process that is first order in 1. An Eyring analysis (fig. S3) indicated that ΔH^‡ = 16 ± 2 kcal mol⁻¹ and ΔS^‡ = 18 ± 12 cal mol⁻¹ K⁻¹ and that ΔG^‡ = 12.8 ± 0.1 kcal mol⁻¹ at −100°C (44, 45). The positive entropy of activation is consistent with a dissociative mechanism. The experimental value for ΔG^‡ of 1 is in good agreement with the calculated binding energies of the model complex [(C₅H₅)Os(dfmpm)(CH₄)]⁺ (12.1 kcal mol⁻¹) and other σ-methane complexes (32–34). Having determined the identity and stability of the protonated complex 1, we then turned to isotopic labeling studies to determine its structure.

There are four principal coordination modes by which methane could bind to a single metal center (Fig. 4): through one (κ¹), two (κ²), or three (κ³) bridging hydrogen atoms or through both the hydrogen and carbon atom of a single C─H bond (η²) (24, 46). Because the bridging and terminal H nuclei exchange rapidly even at −130°C, it is not possible to determine directly which of these four structures is present, but the structure can be determined indirectly by an isotopic perturbation of equilibrium (IPE) (47, 48) study of the ¹³CH₄, ¹³CH₃D, ¹³CH₂D_2, and ¹³CHD₃ isotopologs. Nearly 50 years ago, Calvert and Shapley (49) elegantly demonstrated the utility of IPE for determining the static structure of a triosmium complex containing an agostic methyl group, and more recently, Ball and coworkers (50, 51) have used IPE to study the structures of the alkane complexes (C₅H₄Prⁱ)Re(CO)₂(pentane) and (C₅H₅)Re(CO)₂(cyclopentane). The IPE method relies on the fact that, if hydrogen and deuterium nuclei equilibrate between two sites with different force constants, the H and D nuclei will preferentially occupy the site(s) with the weaker and the stronger force constants, respectively, because this arrangement minimizes the zero point contribution to the total energy. In a methane ligand, the C─H bond(s) that bridge to the metal should have lower C─H stretching frequencies and weaker force constants; consequently, the bridging site(s) should preferentially be occupied by H.

Fig. 4. The four possible coordination modes of a σ-methane ligand.

The nonstandard descriptors proposed in 1996 (24) are listed in (A), but for reasons given in (46), here, we will use the descriptors given in (B).

At a given temperature, the exchange-averaged ¹H NMR chemical shift of each isotopolog of [(C₅Me₅)Os(dfmpm)(CH₄)]⁺ can be expressed as a function of three parameters (see the Supplementary Materials): the terminal and bridging ¹H NMR chemical shifts for a static structure [δ_H(t) and δ_H(b)] and the zero point energy change that arises upon swapping an H/D pair between the two sites (ΔE). Similar equations apply for the exchange-averaged J_CH coupling constant of each isotopolog, which depends on the parameters J_CH(t) and J_CH(b) and the same energy term ΔE. Because each methane binding mode has a different ratio of bridging and terminal hydrogen atoms (except κ¹ versus η², which we discuss later), each binding mode also has a unique set of equations for the exchange-averaged chemical shifts and J_CH constants of the ¹³CH₄, ¹³CH₃D, ¹³CH₂D₂, and ¹³CHD₃ isotopologs. Because these systems of equations (52) each contain three variables, measuring the chemical shifts and coupling constants of at least three isotopologs allows us to determine the binding mode, chemical shift, and coupling constants at the slow exchange limit.

The IPE equations predict that the exchange-averaged chemical shifts of the ¹³CH₄, ¹³CH₃D, ¹³CH₂D₂, and ¹³CHD₃ methane adducts will show the following trend: For the κ¹ (and η²) binding modes, which contain more terminal sites than bridging sites, each successive D for H substitution will cause a larger upfield change in δ than the previous substitution. For the κ² binding mode, which contains an equal number of terminal and bridging sites, each successive D for H substitution will cause approximately the same upfield change. Last, for the κ³ binding mode, which contains fewer terminal sites than bridging sites, each successive D for H substitution will cause a smaller upfield change in δ than the previous substitution. The exchange-averaged C─H coupling constant will follow the same pattern: Each successive D for H substitution will produce a change in the averaged coupling constant that is unique to the ratio of terminal and bridging hydrogen atoms.

To obtain the isotopologs needed for the IPE study, we treated a 1:1 mixture of (C₅Me₅)Os(dfmpm)(¹³CH₃) and (C₅Me₅)Os(dfmpm)(¹³CHD₂) with a 1:2 mixture of protio- and deutero-acid (53) in CDCl₂F at −110°C. At this temperature, four doublets are observed in the upfield region of the ¹H NMR spectrum (Fig. 5), one for each methane isotopolog: δ −1.94 (Os–¹³CH₄), −2.33 (Os–¹³CH₃D), −2.86 (Os–¹³CH₂D₂), and −3.63 (Os–¹³CHD₃). The chemical shift differences seen for each successive D for H substitution are thus 0.39, 0.53, and 0.77 parts per million (ppm). The C─H coupling constants of the four doublets are 126.5, 123.7, 120.0, and 114.6 Hz, respectively, so the differences are 2.8, 3.7, and 5.4 Hz. The same trends in δ and ¹J_CH are seen at −120°C and −130°C, but the differences seen for each substitution are larger, as expected from the Boltzmann statistics. The increasingly large changes in the chemical shifts and coupling constants seen for each successive D for H substitution is definitive evidence that only one hydrogen atom of the coordinated methane ligand bridges to the osmium center in the ground state structure and thereby proves that the binding mode of methane must be κ¹ (or η²).

Fig. 5. ¹H NMR resonances for the coordinated methane ligands in a mixture of the isotopologs [(C₅Me₅)Os(dfmpm)(¹³CH_4-xD_x)]⁺ at −110°C in CDCl₂F.

Solving the system of equations for the κ¹/η² configuration also enables us to determine the following NMR parameters of the terminal and bridging sites: δ_H(t) = 0.39 (5), δ_H(b) = −8.92 (17), and ΔE = 0.264 (5) kcal mol⁻¹. The chemical shift of the terminal hydrogen atoms δ_H(t) is close to the chemical shift of free alkanes, and the chemical shift of the bridging hydrogen atom δ_H(b) is similar to the chemical shifts of related osmium(II) hydrides of the form (C₅Me₅)Os(diphosphine)H (42). The zero point energy difference ΔE between having H or D in the bridging position is about twice the 0.13(1) kcal mol⁻¹ value reported by Calvert and Shapley (49) in their agostic methyl complex but is very similar to the 0.23(3) kcal mol⁻¹ value reported by Ball and coworkers (51) for (C₅H₄Prⁱ)Re(CO)₂(pentane).

When the IPE equations are solved for the C─H coupling constants, we find that J_CH(t) = 141 ± 3 Hz and J_CH(b) = 83 ± 11 Hz; the difference in ¹J_CH between the bridging and terminal C─H bonds is large (almost 60 Hz) and shows how strongly the methane molecule is desymmetrized by its interaction with the osmium center (54). Comparing J_CH(t) and J_CH(b) to the value seen for free methane (125 Hz) (32) shows that the one bridging C─H bond is significantly weakened, whereas the three terminal C─H bonds are significantly strengthened (or have considerably more s-character) (55).

Having determined that one hydrogen atom is bound to the metal center, we turned to the question whether the binding mode is better described as κ¹-H or η²-C,H; i.e., is the carbon atom also bonded to the metal (56)? The IPE data for complex 2 is consistent with both of these coordination modes, but DFT calculations on the model complex [(C₅H₅)Os(dfmpm)(CH₄)]⁺ suggest that there is little or no direct overlap between osmium and carbon, as evidenced by the large Os─H─C angle of 115° and by the long Os-C_methane distance (2.571 Å, c.f. the sum of the covalent radii ~2.20 Å) (38, 57).

Last, we wish to address why the ¹³C NMR shift of the σ-methane ligand in 2 is shielded by ~41 ppm relative to free methane. For (C₅H₄ⁱPr)Re(CO)₂(n-pentane), a similar strong shielding of the ¹³C NMR resonance for the bound σ-pentane ligand was proposed to indicate the presence of an η²-C,H interaction (50). We wish to propose an alternative possibility: that the observed shielding of the ¹³C NMR shift of the σ-methane ligand in 2 is a direct consequence of the weakness of the metal-alkane binding. As a result, the d⁶ osmium center is effectively five-coordinate rather than six-coordinate, which lowers the energy of one empty, σ*-character d-orbital; this lowering generates low-lying paramagnetic excited states, which mix into the ground state and perturb the observed NMR chemical shifts. This phenomenon, which is known as a diamagnetic anisotropy effect, is responsible for the strongly shielded ¹H NMR chemical shifts of certain five-coordinate d⁶ iridium hydrides (58) and probably also for the shielding of weakly bound ligands in other d⁶ pseudo-octahedral compounds, such as (C₅H₄Prⁱ)Re(CO)(PF₃)Xe (59). Relativistic effects may also contribute to the shielding terms for third-row transition metal complexes (58, 60–62).

Here, we have prepared a new methane coordination complex [(C₅Me₅)Os(dfmpm)(CH₄)]⁺ and provided the first experimental evidence for how methane binds to a transition metal complex: In this system, the binding is through one hydrogen atom. The methane ligand is considerably distorted by its interaction with the metal, as indicated by a difference of almost 60 Hz between the ¹J_CH coupling constants of the bridging and terminal C─H bonds. The differences in ¹J_CH between the bridging and terminal C─H bonds in the frozen structure are too large to be due solely to dispersion forces (bond polarization or van der Waals interactions), so there must be considerable covalent d-σ mixing between osmium and the bridging C─H bond. We propose that the shielding of the ¹H and ¹³C NMR resonances for the methane ligand may be a result of diamagnetic anisotropy from the pseudo–five-coordinate metal fragment (possibly augmented by relativistic effects), rather than an indication of direct osmium-carbon overlap. These results provide insights into the solution-phase structure and stability of σ-alkane species, which are important intermediates in the oxidative addition of saturated alkanes to transition metal centers.

MATERIALS AND METHODS

Unless stated otherwise, all manipulations were performed under argon or vacuum using standard Schlenk line or glove box techniques. Glassware was oven- or flame-dried and allowed to cool under vacuum or argon. Solvents were distilled under nitrogen from sodium/benzophenone (pentane and diethyl ether), magnesium (methanol), or sodium (toluene) and sparged with argon for 30 to 60 s immediately before use. Benzene-d₆ was purchased from Sigma-Aldrich or Cambridge Isotope Laboratories in 1-ml ampoules and used without purification. The following starting materials were obtained from commercial sources and used as received unless stated otherwise: dimethylzinc (1.2 M solution in toluene; Strem), iodomethane-¹³C (Sigma-Aldrich), iodomethane-¹³C,d₂ (Sigma-Aldrich), lithium powder (Sigma-Aldrich), bis(trifluoromethanesulfonyl)amine (dried over MgSO₄ and sublimed; Matrix Scientific), bis(trifluoromethanesulfonyl)amine lithium salt (Oakwood Chemicals), sulfuric acid–d₂ (Alfa Aesar), aluminum powder (-40 +325 mesh; Alfa Aesar), iodine (Alfa Aesar), dibromomethane (Sigma-Aldrich), hydrochloric acid (2 M in diethyl ether; Sigma-Aldrich), phosphorus(III) chloride (Sigma-Aldrich), phosphorus(V) oxychloride (Sigma-Aldrich), CF₃SiMe₃ (Oakwood Chemicals), cesium fluoride (Oakwood Chemicals), and trifluoromethanesulfonic acid (SynQuest Laboratories). The compounds (C₅Me₅)₂Os₂Br₄ (42) and CDCl₂F (43) were prepared according to published procedures. The CDCl₂F was distilled and stored at −20°C in a sealed glass vessel over Linde 3 Å or 4 Å molecular sieves.

NMR spectra were acquired on Varian spectrometers (Unity 400, Unity Inova 400, Unity 500, VXR 500, and Unity Inova 600) or Bruker (500 and 600 MHz) spectrometers at room temperature, unless specified otherwise. Low-temperature NMR spectra were acquired on a Varian Unity Inova 600 spectrometer. Positive chemical shifts indicate shifts to higher frequency relative to a chemical shift standard. ¹H and ¹³C spectra were referenced to SiMe₄ by setting appropriate shifts to residual solvent signals for benzene-d₅ (¹H δ 7.16 and ¹³C δ 128.06) or dichlorofluoromethane (¹H δ 7.47 and ¹³C δ 104.2). ¹⁹F and ³¹P spectra were referenced externally to 15% CFCl₃ in CDCl₃ and 85% H₃PO₄ in H₂O, respectively. For spectra collected below room temperature, the NMR probe temperature was calibrated with neat methanol (63). NMR spectra were processed with the MestReNova NMR software package. Fourier transform infrared spectra were acquired on a Thermo Nicolet IR200 spectrometer as Nujol mulls between KBr plates and were processed using the OMNIC software package with automatic baseline corrections. Melting points were acquired on a Thomas-Hoover Uni-Melt apparatus in sealed capillaries under argon. Elemental analyses were performed by the University of Illinois Microanalysis Laboratory and are reported as the average of two replicates. X-ray crystallographic data were collected by the University of Illinois George L. Clark X-Ray Facility and 3M Materials Laboratory.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S35

References