Facile Synthesis of a Stable Side‐on Phosphinyne Complex by Redox Driven Intramolecular Cyclisation

Abstract Alkyne complexes with vicinal substitution by a Lewis acid and a Lewis base at the coordinated alkyne are prospective frustrated Lewis pairs exhibiting a particular mutual distance and, hence, a specific activation potential. In this contribution, investigations on the generation of a WII alkyne complex bearing a phosphine as Lewis base and a carbenium group as Lewis acid are presented. Independently on potential substrates added, an intramolecular cyclisation product was always isolated. A subsequent deprotonation step led to an unprecedented side‐on λ5‐phosphinyne complex, which is interpreted as highly zwitterionic according to visible absorption spectroscopy supported by TD‐DFT. Low‐temperature 31P NMR and EPR spectroscopic measurements combined with time‐dependent IR‐spectroscopic monitoring provided insights in the mechanism of the cyclisation reaction. Decomposition of the multicomponent IR spectra by multivariate curve resolution and a kinetic hard‐modelling approach allowed the derivation of kinetic parameters. Assignment of the individual IR spectra to potential intermediates was provided by DFT calculations.

The X-band EPR spectrum was recorded on a Bruker EMX CW-micro spectrometer equipped with an ER 4131VT Digital Temperature Control System and an ER 4119HS-WI high-sensitivity optical resonator. The calibration of the g value was performed using DPPH (2,2diphenyl-1-picrylhydrazyl) (g = 2.0036±0.00004). A sample of the reaction mixture of 3c with Jutzi acid in CH 2 Cl 2 was filled into a J. Young EPR tube under inert conditions and measured 90 minutes after the start of the reaction at 300 K. Analysis of the hyper fine coupling was performed using the simulation program EPRsim32. [S1] Figure S1. X-band EPR spectrum of reaction mixture 3c with [H(Et 2 O) 2 ][(B(C 6 F 5 ) 4 ] at 300 K after 90 min; black = experiment, red = simulation with g iso = 2.003; hyperfine coupling to 127 I(S = 5/2, 100%) with A iso = 25 10 -4 cm -1 . Hyperfine coupling to 183 W was not accounted for because of restricted resolution.

Cyclic voltammetry and spectro-electrochemical measurements
Cyclic voltammetry was performed using a Princeton Applied Research VersaSTAT 3. A three electrode arrangement with a glassy carbon working electrode, a platinum wire counter electrode and an Ag/AgCl in CH 3 CN reference electrode and 0.15 M n Bu 4 NPF 6 as supporting electrolyte was employed. The ferrocene/ferrocenium (Fc/Fc + ) redox couple was used as internal standard.  Figure S2. Cyclic voltammetry of complex 3c (black) and 3d (red) measured in CH 2 Cl 2 (referenced against ferrocene/ferrocenium). 3c shows a reversible oxidation at E 1/2 = +0.011 V. 3d shows a reversible oxidation at E 1/2 = +0.38 V. Both are putative W II/III redox process. [S2] Spectro-electrochemical data were also acquired with a Princeton Applied Research VersaSTAT 3 potentiostat. The optically transparent thin-layer electrochemical (OTTLE) cell was home-built. It comprised a Pt working and counter electrode and a thin silver wire as a pseudoreference electrode sandwiched between two CaF 2 windows of a conventional liquid IR cell. The working electrode was positioned in the centre of the spectrometer beam. The build-up followed the design of Hartl and co-workers and modifications of Winter et al. [S3] Figure S3. IR spectro-electrochemical measurement of 3c. Changes of the spectra during oxidation (left) and reduction (right) shows reversible W(II)/W(III) oxidations steps. The CO vibration increases from 1923 cm -1 to 2085 cm -1 S4

Crystallographic Details
Single crystals suitable for X-ray diffraction analysis were selected in Fomblin YR-1800 perfluoropolyether oil (Alfa Aesar) at ambient temperature and mounted on a glass fibre. During the measurement the samples were cooled to 173(2) K. Diffraction data were collected on a Bruker-Nonius Apex X8, Bruker D8 Quest-diffractometer and a Bruker Kappa Apex II diffractometer using graphite monochromated Mo-Kradiation (λ = 0.71073 Å).
Structure solutions were found by direct methods (SHELXS-97) [S4] and were refined by fullmatrix least-squares procedures on F2 (SHELXL-97). [S5] All non-hydrogen atoms were refined anisotropically, and hydrogen atoms were included at calculated positions with fixed thermal parameters.   Figure S7. Molecular crystal structure of the complex 4c-OTf (left) and 5 (right) (50% thermal ellipsoids). Hydrogen atoms and the anion CF 3 SO 3 of 4c + have been omitted for clarity. Selected bond lengths (Å) and angles (°): 4c + : P1-C1 1.753(4), P1-C28 1.808(4), C23-C28 1.386 (7), C16-C23 1.517 (7), C2-C16 1.501(6), C1-C2 1.341(6), C2-C16-C17 124.8(4), C2-C16-C23 107.0(3), C23-C16-C17 103.4(4). 5: C1-P1 1.743(10), P1-C28 1.772(10), C28-C23 1.395(15), C23-C16 1.426(15), C16-C2 1.401(13), C2-C1 1.383(13), C2-C16-C17 135.4(10), C2-C16-C23 116.3(9), C23-C16-C17 108.1(9).   For the chemometric analysis of this multicomponent spectral data a kinetic hard-model approach was applied. [S6] A large number of possible reaction mechanisms were analysed (Scheme S2). The reaction orders are optimized within the interval [1,2] if no specifications are given by the assumptions about the reaction system. This can cause broken orders. The results of all reaction mechanisms in Scheme S2 were compared with respect to the reconstruction error of the spectral data and the error of the kinetic model fit. Among all tested reaction mechanisms the one in Scheme S1 leads to optimal results. A B C D E S11 Scheme S2. Analysed reaction mechanism for the reaction of 3c with Jutzi acid. The model with best fit is marked. The integration of a species X, which due to its low concentration does not appear as an S12 independent band in the IR spectrum, does not lead to a reduction of errors in the kinetic hard-model approach. The series of spectra was split into two separate date sets. One covers the range for CO vibration around 1950 cm -1 and the other the range for the less sensitive CN vibration above above 2100 cm -1 . Both data sets were analysed with the peak group analysis. [S7] Pure component spectra and concentration profile have been extracted, see Figure S8. Every CN band can be assigned to a CO band, so that the formation of a W III -species at around 2083 cm -1 can be excluded.. It is important to remark that without further assumptions only qualitative but no quantitative concentration-time curves can be obtained by the peak group analysis. Nevertheless, an approximate trend can be extracted. This is shown in particular by the profile for 1933 cm -1 (red) and 1992 cm -1 (blue) which must be zero from a chemical point of view. The unusually broad band at 1976 cm -1 and the corresponding bands at 2106 cm -1 /2138 cm -1 , which formed immediately after the addition of the acid, seem conspicuous. The reason for this is the cyanide ligand, which besides the phosphine can also bind the proton, which leads to two isomers. [S8] The analysis clearly confirms that there was no complete oxidation of the tungsten centre. Instead the positive charge is stabilized by mesomerism between the ligand and the metal. This enables a significantly durable S13 intermediate with bands at 1992 cm -1 and 2187 cm -1 . The product has a CO vibration at 1944 cm -1 and forms slowly over a period of several days.

S7
It is possible to apply another multivariate curve resolution method, namely a kinetic hardmodelling approach from Section 4.1 for the area around 1950 cm -1 . The model in Scheme S3 is used. In comparison to the reaction of 3c with Jutzi acid it does not contain a dead-end but an equilibrium between the long-lived intermediate C CN  A green solution of 3b (10 mg, 9.4 µmol) in CD 2 Cl 2 (0.5 mL) in a J. Young NMR tube was treated with [H(Et 2 O) 2 ][B(C 6 F 5 ) 4 ] (7.8 mg, 9.4 µmol) solute in CD 2 Cl 2 (0.2 mL) at -80°C. The upper part of the solution colour changed slowly from green to red. The sample was recorded with 128 Scans at different temperature starting at -80 °C using Bruker Avance 500 MHz ( Figure S9). The temperature was increased by 20 K after each measurement. At -40 °C a Proton coupled 31 P NMR-spectrum was measured and is shown in Figure S10.

5.2.
Reaction mixture after protonation of 3c with [H(Et 2 O) 2 ][(B(C 6 F 5 ) 4 ] monitored by 31 P NMR spectroscopy A green solution of 3c (10 mg, 9.4 µmol) in CD 2 Cl 2 (0.5 mL) in a J. Young NMR tube was treated with [H(Et 2 O) 2 ][B(C 6 F 5 ) 4 ] (7.8 mg, 9.4 µmol) solute in CD 2 Cl 2 (0.2 mL) at -80°C. The upper part of the solution colour changed slowly from green to red. The sample was recorded with 128 Scans at different temperature starting at -80 °C using Bruker Avance 500 MHz ( Figure S11). The temperature was increased by 20 K after each measurement. At -80 °C a Proton coupled 31 P NMR-spectrum was measured and is shown in Figure S12. S15 Figure S11. Reaction sequence of 3c with [H(Et 2 O) 2 ][(B(C 6 F 5 ) 4 ] in 31 P NMR spectroscopy. Reaction mixture was heated stepwise from -80 °C to room temperature. Note the temperature shift of species B from -2.5 ppm at -80 °C to -1.0 ppm at -20 °C.        The calculations were carried out using either the G09RevE.01 [S9] or the ORCA 4.11 [S10] program package applying DFT. The molecular geometries of complex cation 4c + and neutral 5 as well as of the potential intermediates IM1 + (open-shell triplet and closed shell singlet), syn-IM2 + and anti-IM2 + (which were not isolated) were optimized without truncation and symmetry constraints in the gas phase using either the PBE0 functional [S11] or the long-range corrected CAM-b3lyp [S12]  Frequency calculations were performed with a smaller def2-SVP basis set in order to identify S32 all stationary points as minima. The final enthalpies (ΔH) were calculated using the total electronic energy from the higher level calculation and the thermal correction to enthalpy from the frequency calculation. TD-DFT calculations for complex 5 and extraction of the difference densities were performed with Orca ( Figure S44).

S33
Natural resonance theory analysis (NRT): The implementation of NRT to transition metal complexes frequently poses problems, which arise from the orthogonalization procedure for the d-orbitals. The identification of a starting reference resonance structure for complex 5 turned out to be impossible. Hence, a simplified, truncated model complex bearing NH 3 /NH 2 -ligands was used for the analysis. NBO6, the standard functional b3lyp and a SDD basis set were used. Due to the strong delocalization full density matrix (FDM) and a stabilization threshold of 30 kcal/mol were applied. Nine leading resonance structures for the optimized structure ( Figure S45) are depicted in Figure  S46.