Inversion of Configuration at the Metal in Diastereomeric Imido Alkylidene Monoaryloxide Monopyrrolide (MAP) Complexes of Molybdenum

Abstract

The two diastereomers of Mo(NAr)(CHCMe₂Ph)(2,5-dimethylpyrrolide)(1) ((S_MR₁)-2 and (R_MR₁)-2, respectively, where 1 is an enantiomerically pure (R) phenoxide and Ar = 2,6-diisopropylphenyl), form adducts with PMe₃. One of these ((R_MR₁)-2(PMe₃)) has been isolated. An X-ray structure reveals that PMe₃ has added trans to the pyrrolide; it is a model for where an olefin would attack the metal. Trimethylphosphine will catalyze slow interconversion of (S_MR₁)-2 and (R_MR₁)-2 via formation of weak PMe₃ adducts. Reactions between (S_MR₁)-2 or (R_MR₁)-2 and ethylene yield Mo(NAr)(CH₂)(Me₂Pyr)(1) species in which the configuration at Mo is inverted by ethylene at a rate of the order of the NMR time scale at 22 °C via formation of metalacyclobutane intermediates with imido and aryloxide ligands in axial positions. A reactant olefin is proposed to approach Mo and the product olefin leave Mo trans to the pyrrolide.

Recently we discovered¹ that 14e Mo(NR)(CHR′)(OR″)(η¹-pyrrolide) (MonoAlkoxidePyrrolide or MAP) complexes can be prepared through addition of one equivalent of alcohol to Mo(NR)(CHR′)(pyrrolide)₂ complexes.² Of particular interest in terms of enantioselective metatheses are diastereomers that are formed when the alkoxide or aryloxide is enantiomerically pure. A dramatic example is Mo(NAr)(CHCMe₂Ph)(Me₂Pyr)(1) (2, where Ar = 2,6-diisopropylphenyl, Me₂Pyr = 2,5-dimethylpyrrolide, and 1 is derived from (R)-1H where R = Si(t-Bu)Me₂) that serves as an initiator for an asymmetric ring-closing of an intermediate in the enantioselective synthesis of the Aspidosperma alkaloid, quebrachamine,³ a reaction that yielded no product when several chiral Mo(NR)(CHR′)(diolate) catalysts⁴ were employed. Initiator 2 can be prepared in situ and is effective (95% ee) at relatively low loadings as a 7:1 mixture of diastereomers. Two issues that arise concern the relative reactivity of diastereomers and their interconversion through inversion of configuration at the metal. Therefore we turned to an exploration of reactions between the diastereomers and 2e donors such as PMe₃, which are models⁵ for the initial unobserved olefin adduct,⁶ and the conditions under which inversion of configuration at Mo might take place.

A 7:1 mixture of the two diastereomers of Mo(NAr)(CHCMe₂Ph)(Me₂Pyr)(1) ((S)-2 and (R)-2, respectively^3,7) is generated upon addition of (R)-1H to Mo(NAr)(CHCMe₂Ph)(Me₂Pyr)₂.¹ Pure (S)-2³ and pure (R)-2⁸ have both been isolated and structurally characterized. Each is unchanged in C₆D₆ or THF-d₈ after a week at 22 °C or 40 °C. The alkylidene is in the syn orientation in the solid state and in solution judging from the J_CH value (118 Hz in (S)-2 and 122 Hz in (R)-2).

Trimethylphosphine (15 equiv) was added to pure (S)-2 ([Mo] = 0.1 M, pentane). The solution was stored for several hours at 22 °C and crystals of a phosphine adduct were isolated in good yield (75%). An X-ray structural study revealed that the product is (R)-2(PMe₃), not (S)-2(PMe₃). The overall structure is closest to a square pyramid with the syn-alkylidene in the apical position (Figure 1) and PMe₃ trans to the pyrrolide. The N2-Mo-P1 angle is 165.00(10)° and the N1-Mo-O1 angle is 158.12(14)°, while the angles between C45 and the four other atoms bound to Mo are 100.31(17)° (to N1), 106.29(15)° (to N2), 100.75(15)° (to O1), and 85.61(12)° (to P1). The Mo-P1 distance (2.5703(11) Å) is relatively long and the Mo-P bond likely to be relatively weak.

Figure 1.

POV-ray drawing of (R)-2(PMe₃). Thermal ellipsoids are displayed at 50% probability level. Hydrogen atoms are omitted.

When (R)-2(PMe₃) is dissolved in C₆D₆ or toluene-d₈, largely (R)-2 and free PMe₃ are observed at room temperature. At a concentration of initial (R)-2(PMe₃) equal to 10 mM at −30 °C largely (R)-2(PMe₃) is observed, while at 40 °C only free PMe₃ and (R)-2 are observed. Over a period of 24 h at 22 °C (R)-2 is converted into an equilibrium mixture of (R)-2 and (S)-2 (K_eq = [(S)-2]/[(R)-2] = 2.0 at 40 °C). The approach to equilibrium at 40 °C depends on PMe₃ concentration to the first order (as shown in runs between 10 and 50 mM total PMe₃ concentration) and follows classic behavior⁹ with k_RS + k_SR = 9.0×10⁻³ M⁻¹s⁻¹. In THF-d₈ at 40 °C k_RS + k_SR = 14×10⁻³ M⁻¹s⁻¹ (K_eq = 2.0) and in 1:1 acetonitrile-d₃/C₆D₆ at 40 °C k_RS + k_SR = 24×10⁻³ M⁻¹s⁻¹ (K_eq = 0.8). (In the absence of PMe₃ in the last experiment ~5% inversion can be observed after several days at 60°C; therefore, inversion by acetonitrile is extremely slow compared to inversion by PMe₃.) All results are consistent with PMe₃-catalyzed interconversion of (R)-2 and (S)-2.

Inversion at Mo is also catalyzed by PPhMe₂ and (neat) pyridine-d₅. For 1 M PPhMe₂ in C₆D₆ at 40 °C k_RS + k_SR = 1.5×10⁻⁵ M⁻¹s⁻¹, 600 times slower than the PMe₃-catalyzed reaction. Inversion by pyridine-d₅ and (very slowly) acetonitrile (see above), rule out any required reaction between the 2e donor ligand and some ligand bound to the metal (e.g., between the phosphine and the alkylidene to yield an intermediate and unobservable ylide complex). The lack of any dramatic solvent effect argues against ionization to yield a four-coordinate cation with pyrrolide or aryloxide as the anion.

Mixtures of largely (S)-2, (S)-2(PMe₃), and PMe₃ (2 equiv) in toluene-d₈ between −30 °C and 22 °C were examined by ¹H NMR and formation of (R)-2 and (R)-2(PMe₃) was monitored. After ~6 h at 22 °C four phosphine adducts were observed, (S)-2(PMe₃) (δH_α = 15.51 ppm at −30 °C, J_CH = 121 Hz, J_HP = 5 Hz), (S)-2′(PMe₃) (~20% of total (S) adduct; δH_α = 13.94 ppm, J_HP = 6 Hz), (R)-2′(PMe₃) (~5% of total (R) adduct; δH_α = 14.86 ppm, J_HP = 8 Hz), and (R)-2(PMe₃) (δH_α = 13.83 ppm, J_CH = 122 Hz, J_HP = 7 Hz), along with (S)-2 and (R)-2; the amount of each depends upon time, temperature, concentration, and the amount of PMe₃ present. In all four adducts coupling of the alkylidene H_α proton to ³¹P is lost as the temperature is raised, equilibria shift toward phosphine free species, and (S)-2 and (R)-2 interconvert.

In order to further identify the adducts, ¹⁵N- or ¹³C-labeled analogs were prepared. In analogs that contain ¹⁵N-labeled dimethylpyrrolide couplings of ¹⁵N to ³¹P are 24.1 Hz in (S)-2(PMe₃), 26.5 Hz in (S)-2′(PMe₃), 31.6 Hz in (R)-2′(PMe₃), and 26.5 Hz in (R)-2(PMe₃). Since pyrrolide is trans to PMe₃ in (R)-2(PMe₃), we propose that it is trans to the pyrrolide in all four adducts. In analogous ¹³C_α-labeled neopentylidene complexes (R)-2′(PMe₃) (<5% of total adducts) was identified as an anti species (J_CH = 143 Hz) while (S)-2′(PMe₃) is syn (J_CH = 120 Hz); the difference between (S)-2′(PMe₃) and (S)-2(PMe₃) is ascribed to different conformations of the aryloxide in the crowded environment. NOESY studies confirm (R) and (S) assignments.

All data are consistent with inversion at Mo in fluxional five-coordinate adducts. In terms of TBP intermediates (as shown in equation 1; SP species analogous to the observed structure of (R)-2(PMe₃) in the solid state are alternatives) L enters trans to the pyrrolide (Pyr) to give (R)(L_Pyr). A series of Berry pseudorotations or (equivalent) turnstile rearrangements¹⁰ permutes the alkylidene and imido positions to give (S)(L_Pyr), from which L leaves trans to Pyr to give (S)-2. Entry and exit of L only trans to Pyr in the two diastereomers is consistent with the structure of (R)-2(PMe₃) and is the most unifying proposal in our opinion. Catalyzed inversion at M by PMe₃ is the most efficient, and a PMe₃ adduct is the only one that can be isolated.

(1)

We then turned to an exploration of reactions between (S)- or (R)-Mo(NAr)(CHCMe₂Ph)(Me₂Pyr)(1) and ethylene, which is often generated in ring-closing reactions of terminal or 1,1-disubstituted olefins. These reactions give rise to observable (S)- or (R)-Mo(NAr)(CH₂)(Me₂Pyr)(1) (67% at 10 °C, δH_α = 12.35, 12.13; 33% at 10 °C, δH_α = 12.94, 12.24) and Mo(NAr)(CH₂CH₂CH₂)(Me₂Pyr)(1) (at −70 °C δH_α = 6.16, 5.69, 5.24, 5.03; δH_β = 0.74, −0.16). The two interconvert readily on the NMR time scale (at 22 °C) through gain and loss of ethylene, respectively. Although Mo(NAr)(CH₂CH₂CH₂)(Me₂Pyr)(1) loses ethylene readily, and therefore is not likely to be isolable, its NMR parameters are analogous to those of W(NAr)(CH₂CH₂CH₂)(Me₂Pyr)(1), which has been isolated and shown in an X-ray structure to be an essentially undistorted TBP with apical imido and aryloxide ligands and dimethylpyrrolide in the equatorial plane.¹¹ (Reactions involving ethylene and Mo and W MAP species will be reported in detail separately.) It is clear that inversion at the metal can be fast on the NMR time scale near room temperature.

We propose that an olefin attacks the metal in MAP species trans to the pyrrolide ligand to form an intermediate metalacyclobutane that contains the pyrrolide and two carbons of the resulting metallacycle in equatorial positions (e.g., equation 2). The product olefin (ethylene in this example) then leaves trans to the pyrrolide to generate the new alkylidene (M=CHR in this case) with the opposite configuration at M. In effect, the reactant olefin enters “trans” to the pyrrolide and the product olefin leaves “trans” to the pyrrolide, all via an intermediate TBP

(2)

with axial imido and aryloxide ligands, inverting the configuration at M with each metathesis step. This proposal is consistent with recent calculations⁶ if it is assumed that pyrrolide is a “donor” relative to the aryloxide (the “acceptor,” relatively), and with the preferred approach of PMe₃ described above.

It seems likely that inversion at M by an olefin will be faster than a reaction that involves rearrangement of a five-coordinate olefin adduct, in part because a metalacyclobutane ring forms rapidly. (There is only one reported observation of an olefin adduct of a high oxidation state alkylidene.¹²) That being said, it also should be noted that rearrangement of the five-coordinate metalacyclobutane itself might compete with loss of olefin in circumstances where the metalacyclobutane lifetime is relatively long, i.e., in tungstacyclobutane species. The speed of metathesis/inversion at M is expected to vary widely and exceptionally finely as steric interactions generated in reactions between a given diastereomer and a given olefin become more significant and unavoidable.

To the best of our knowledge it has not been possible to probe “inversion” at a catalytically competent tetrahedral transition metal center to which four ligands are covalently bound, especially if one of those ligands is exchanged during catalysis/inversion. Inversion at M in imido alkylidene complexes is a central issue in the context of enantioselective metathesis reactions that involve MAP species.^3,8 Lastly, controlling inversion at the metal is key to the long sought goal of employing “stereogenic-at-metal” complexes for possibly more efficient enantioselective reactions.¹³

Supplementary Material

1_si_001. Supporting Information Available.

Experimental details for the synthesis of all compounds and the X-ray structural study. Supporting Information is available free of charge via the Internet at http://pubs.acs.org.