Homo- and heterodehydrocoupling of phosphines mediated by alkali metal catalysts

Catalytic chemistry that involves the activation and transformation of main group substrates is relatively undeveloped and current examples are generally mediated by expensive transition metal species. Herein, we describe the use of inexpensive and readily available tBuOK as a catalyst for P–P and P–E (E = O, S, or N) bond formation. Catalytic quantities of tBuOK in the presence of imine, azobenzene hydrogen acceptors, or a stoichiometric amount of tBuOK with hydrazobenzene, allow efficient homodehydrocoupling of phosphines under mild conditions (e.g. 25 °C and < 5 min). Further studies demonstrate that the hydrogen acceptors play an intimate mechanistic role. We also show that our tBuOK catalysed methodology is general for the heterodehydrocoupling of phosphines with alcohols, thiols and amines to generate a range of potentially useful products containing P–O, P–S, or P–N bonds.

ether (MTBE) and stirring for further 5 minutes. The upper organic phase was decanted from the formed gel. To the remaining gel 20 mL of CH2Cl2 was added and the mixture agitated well for additional 5 minutes. The resultant mixture was then filtered through a frit equipped with celite.
After washing the celite with CH2Cl2 (2 x 30mL) the organic phases were combined, dried over MgSO4 and filter via a cannula to a second 250 mL flask and the solvent was removed in vacuo to obtain the corresponding phosphine oxide. Another 250 mL flask equipped with gas inlet and addition funnel was charged with a solution of the above phosphine oxide (1 eq.) in 30 mL THF. This solution was added over a period of 15 minutes to a 1 M solution of DIBAL-H in hexane (3 eq.) and stirred for 30 min at ambient temperature. To the flask 50 mL freshly degassed MTBE was added via the addition funnel slowly. After cooling the solution to 0 °C, 30 mL degassed 2N aq. NaOH was added via the addition funnel followed by 10 mL sat. aq. NaCl. The solution was stirred for additional 5 minutes and warmed to room temperature. Stirring was subsequently stopped and the layers allowed to separate. The organic layer was then transferred via cannula to a second 250 mL flask charged with MgSO4. After stirring for 10 minutes the mixture was filtered via a cannula to a third 250 mL flask and the solvent was removed in vacuo to obtain the phosphines.
X-ray Crystallography X-ray diffraction quality crystals of 2j and 3e·(THF) were obtained by vapour diffusion of hexanes into a THF solution of the compound. X-ray diffraction experiments for compounds 2j and 3e·(THF) were carried out at 100 K on a Bruker APEX II diffractometer using Mo Kα radiation (λ = 0.71073 A). The data collections were performed using a CCD area detector from a single crystal mounted on a glass fibre. Intensities were integrated 7 and absorption corrections based on equivalent reflections using SADABS 8 were applied. The structures were all solved using direct methods and structures were refined against all F2 using ShelXL2013 9 or Olex2 10 . All of the nonhydrogen atoms were refined anisotropically. Hydrogen atoms were calculated geometrically and refined using a riding model. Compound 2j is a racemic twin and the absolute structure was not determined due to low Friedel pair coverage. The X-ray structure of [CyP]4 (2j) was previously reported by Burford and coworkers at 153 K 11 .    were determined by 31 P{ 1 H} NMR spectroscopy using a capillary of PCl3 as a calibration standard.

Supplementary
c Full conversion due to the hydrophosphination products. S10  Reaction was performed with 0.1 mmol Ph2PH, 0.01 mmol tBuOK, 0.1 mmol HA-5 and 0.5 mL THF in a J. Young NMR tube and heated at 50 °C in the NMR spectrometer (temperature was set to 15 °C lower than the boiling point of THF) and the instrument was programmed to automatically collect data over 16 h at 30 min intervals. Supplementary Figure 6 shows that during the reaction, there were no other P-containing species involved as reaction intermediates apart from the adduct 3a, and minor amounts of compound 5a by-product are also observed.    Attempted radical trapping experiments were performed by adding 5 equiv. of the radical trap reagent 1,4-cyclohexadiene to a 2:1 stoichiometric mixture of 1a and HA-2 in THF. Upon the addition of 10 mol% of either tBuOK or K[PhNNPh] to the preceding mixture the full conversion to 2a and hydrazobenzene and the complete lack of the H-atom abstraction product (benzene) was observed by 13 C{ 1 H} NMR (Fig. S27). The lack of formation of benzene even in the presence of added K[PhNNPh] radical anion in these experiments is suggestive that the secondary phosphine 1a is a far more efficient radical trap than 1,4-cyclohexadiene and far more likely to engage in Hatom abstraction. We can see from the above Figure, with more DTBP (radical source) the HA-2 mediated homodehydrocoupling reaction of 1a produced more HA-2 hydrophosphination product and less dehydrocoupling product 2a together with the fact that hydrophosphination product cannot easily be converted into 2a support a radical mechanism for HA-2 as hydrogen acceptor.

Computational methods
All calculations were carried out by using the Gaussian 16 program. 14 All structures were optimised at the M062X 15 level of density functional theory (DFT) with the TZVP 16 basis set. All optimised structures were characterised either as energy minima without imaginary frequencies or transition states with only one imaginary frequency by frequency calculations; and the imaginary mode connects the initial and the final states. The thermal corrections to Gibbs free energy at 298 K from the frequency analysis are added to the total electronic energy, and we therefore used the corrected Gibbs free energy (DG) at 298 K for our discussions and comparisons. In our previous study, 17 we have done benchmark calculations comparing different functionals on the basis of the experimental observation and found that the M062X method gave the best agreement between theory and experiment in phosphine related chemistry; and this is also the reason for our choice to use M062X in our current study. In our computations we always used the whole molecules without any simplification or constraints.

Coupling reaction free energy
It is found that the direct homocoupling of Ph2PH to Ph2P-PPh2 without hydrogen acceptor [2 equiv. Ph2PH = Ph2P-PPh2 + H2] is endergonic by 2.64 kcal·mol -1 , the thermodynamic equilibrium under stoichiometric conditions is less than 2% product; and therefore, hydrogen acceptor or other oxidants are needed to shift this reaction towards product formation. In the presence of an HA it is found that all of the dehydrocoupling reactions of 2 equiv. of 1a to 2a becomes exergonic and therefore thermodynamically downhill (Supplementary Figure 30). Apart from benzophenone (HA-1) with lower exergonic reaction energy (-3.48 kcal·mol -1 ), other hydrogen acceptors have higher exergonic reaction energies (-10 to -13 kcal·mol -1 ).