Impact of Ligands and Metals on the Formation of Metallacyclic Intermediates and a Nontraditional Mechanism for Group VI Alkyne Metathesis Catalysts

The intermediacy of metallacyclobutadienes as part of a [2 + 2]/retro-[2 + 2] cycloaddition-based mechanism is a well-established paradigm in alkyne metathesis with alternative species viewed as off-cycle decomposition products that interfere with efficient product formation. Recent work has shown that the exclusive intermediate isolated from a siloxide podand-supported molybdenum-based catalyst was not the expected metallacyclobutadiene but instead a dynamic metallatetrahedrane. Despite their paucity in the chemical literature, theoretical work has shown these species to be thermodynamically more stable as well as having modest barriers for cycloaddition. Consequentially, we report the synthesis of a library of group VI alkylidynes as well as the roles metal identity, ligand flexibility, secondary coordination sphere, and substrate identity all have on isolable intermediates. Furthermore, we report the disparities in catalyst competency as a function of ligand sterics and metal choice. Dispersion-corrected DFT calculations are used to shed light on the mechanism and role of ligand and metal on the intermediacy of metallacyclobutadiene and metallatetrahedrane as well as their implications to alkyne metathesis.


{[C6H3(C6H4SiEt2O)3]W≡N}2 (Nitride1)
Cat5 (40 mg/0.043 mmol) was dissolved in 0.6 ml of C6D6 to give a yellow suspension. Benzonitrile (4.7 μl/0.046 mmol) was added at room temperature resulting in the solution darkening to purple-red. After 3 h at room temperature, the solution had become yellowbrown in color and the formation of mesityl-phenylacetylene could be seen forming along with the consumption of Cat5•PhCN. After 6 h the solution was bright yellow and all resonances associated with Cat5•PhCN had disappeared. Slow evaporation of the solvent led to the deposition of pale-yellow/colorless crystals of Nitride1 suitable for X-ray diffraction studies. Yield = 17 mg (49%). S10 Supporting NMR spectra          Figure S21. Long-range coupling 1 H-13 C gHMBC NMR (500 MHz, C6D6, 25 °C) spectrum of MTd2 generated in-situ by the addition of excess 5-decyne (#) to a solution of Cat3. 4-tolyl-1-hexyne impurity is denoted by *. Figure S22. 13 C NMR (C6D6) of the reaction of Cat5 with 6 eq. of 1-methoxy-4-(phenylethynl)benzene. The absence of resonances between 220-280 suggests that an all aryl MCBD is not the resting state.

Rate of Alkyne Metathesis using Mo(VI) and W(VI)-based SiP Catalysts
Under an inert atmosphere, a 0.2 mM/0.002 mM solution of substrate/catalyst was prepared in 0.6 mL C6D6. The percentage of substrate conversion was monitored by integration of the OCH3 resonance periodically at set time points. The MestraNova fitting function was used to assist in the integration. S40 Figure S41. Dynamic scrambling of 1-methoxy-4-(phenylethynl)benzene (0.1 mM in C6D6) catalyzed by 2 mol% of Cat1 and Cat4 at rt monitored by 1 H NMR. Figure S42. Stacked 1 H NMR plots for kinetics experiments using Cat5 (Left, Black), Cat3 (Center, Red) and Cat 4 (Right, Blue). Integrations were obtained by "Generalized Lorentzian" fitting in mnova NMR software. S41 Figure S43. Substrate Scope using Cat5.

Crystallographic Information
Crystallographic data are summarized in Tables S1-S4. Suitable crystals for X-ray analysis of SiP Et , Pre2, Cat3-5, MTd2, MCBD2-6, Cat5•PhCN and Nitride1 were placed MiTeGen pins, coated in oil. The X-ray intensity data collection was carried out on a Bruker APEXII DUO CCD area detector using graphite-monochromated Mo-Kα radiation (λ = 0.71073 Å) or Cu-Kα radiation (λ = 1.54184 Å) at 90.0(5) K. Frames were integrated using SAINT, 6 producing a listing of non-averaged F 2 and σ(F 2 ) values. The intensity data were corrected for Lorentz and polarization effects and for absorption using SADABS. 7 The initial structure was determined by intrinsic phasing using SHELXT. 8 The further structure determination was performed by difference Fourier methods and refined by full-matrix least squares using SHELXL 9 or olex2.refine for Nitride1. All reflections were used during refinements. Non-hydrogen atoms were refined anisotropically and hydrogen atoms were visible in difference maps, but placed in idealized positions and treated with riding models. Disordered phenyl groups successfully modeled. Disordered solvent was removed using the SQUEEZE procedure for Cat3, MCBD4 and MTd2.    (11) 9.7250 (6) 13.5425 (4) 10.7732 (8)

Computational details General Remarks
All optimizations of intermediates and transition states were calculated using restricted B3LYP-D3/def2SVP-LANL2DZ(M) level of the theory 10 in implicit solvent (benzene) using CPCM as solvation model 11 as implemented in Gaussian09. Frequency calculations, at the same level of theory, were used to obtain thermal corrections (at 298K) and to characterize optimized structures as transition states (only a single imaginary frequency) or intermediate (if no imaginary frequencies were found). Intrinsic reaction coordinate (IRCs) calculations were undertaken to ensure transition states connected illustrated ground states. Single point energy calculations using B3LYP-D3/def2TZVP-SDD(M) with solvent corrections calculated in implicit solvent (benzene) using CPCM were also performed on all structures. 12 For comparison, single point energy calculations with restricted PBEPBE/def2TZVP-SDD(M) were calculated with solvent corrections calculated in implicit solvent (benzene) using CPCM as a solvation model. All reported charges are from Mulliken population analysis. 13 All 3-D structures were generated using CYLview. 14 Noncovalent interaction (NCI) analysis, also known as reduce density gradient (RDG) method, was performed on Multiwfn to study the possible effect of noncovalent interaction in the metallatetrahedrane intermediates. 15 Extension distance of 0 Bohr, medium quality grid (totally about 512000 points) were set by default. Further visualization of the color-filled RDG isosurface was realized by VMD, where RDG isosurface and color range were set as 0.5, and -0.035 to 0.2, respectively. 16 The energy decomposition analysis calculations were performed using the secondgeneration absolutely localized molecular orbitals 17 (ALMO-EDA) method implemented in Q-Chem 5.0. 18 The HF/6-311G(d,p) method was used as employed by Liu. 19 This method decomposes the through-space interaction energies between the ligand and substrate into the energetic components including the Pauli repulsion energy (∆EPauli), the electrostatic energy (∆Eelstat), the polarization energy (∆Epol) and the charge transfer energy (∆Ect.) Distortion energies (∆Edist), or the energy required to distort the geometry of the starting intermediate to the transition S53 state geometry as described by the distortion-interaction model, 20

Choice of Computational Method
The computational method of B3LYP-D3/def2TZVP-SDD(M)-CPCM(benzene)//B3LYP-D3/def2SVP-LANL2DZ(M)-CPCM(benzene) was chosen as it provides an excellent balance between accuracy and computational cost, all while reproducing structural features from x-ray crystal structures. Pseudopotentials such as LANL2DZ and SDD are very commonly used on the metal center to reduce computational cost while maintaining the accuracy of the calculation. 21 Furthermore, pseudopotentials such as used LANL2DZ and SDD have been shown to be effective in describing Mo and W compounds. 22  Table S5. List of Bond lengths and angles for MCBD1-comp, optimized using B3LYP-D3/def2SVP-LANL2DZ(Mo)-CPCM(benzene).

Substrate-Dependence on Reaction Rates
For comparison of the experimental results of the apparent substrate-dependence on reaction rates, we have computed the lowest energy pathway to product formation for an aryl-propyne substrate with the less sterically hindered SiMe ligand for both tungsten and molybdenum. These pathways are given in Figures S58 and S59 below. While the barrier to [2+2]-cycloaddition is lower in energy for tungsten compared to molybdenum (9.2 vs. 16.6 kcal/mol), it is clear from our computations that the tungsten system would be slower due to the thermodynamically stabilized MCBD intermediate [W]-ent-C' (-0.3 kcal/mol). From here, the MCBD will undergo retro-[2+2] to yield product, with a relative energy barrier of 12.6 kcal/mol for tungsten and only 7.0 kcal/mol for molybdenum. These computations suggest that the reaction rates are dependent on the substrate for tungsten.     Table S6. List of Bond lengths and angles for MCBD6-comp, optimized using B3LYP-D3/def2SVP-LANL2DZ(M)-CPCM(benzene) compared to those of the x-ray crystal structure of MCBD6. Similar to that observed in MCBD6, the tolyl group is pointing away from the basal arene to minimize steric crowding.    Figure S61. Energetics of MCBD and MTd formation via [2+2] cycloaddition for tungsten with SiP Ph system computed at the B3LYP-D3/def2TZVP-SDD(W)-CPCM(benzene)// B3LYP-D3/def2SVP-LANL2DZ(W)-CPCM(benzene)level of theory (outside parenthesis) and the PBEPBE/def2TZVP-SDD(W)-CPCM(benzene)// B3LYP-D3/def2SVP-LANL2DZ(W)-CPCM(benzene)level of theory (inside parenthesis).