Sterically constrained tricyclic phosphine: redox behaviour, reductive and oxidative cleavage of P–C bonds, generation of a dilithium phosphaindole as a promising synthon in phosphine chemistry†‡

The redox behaviour of sterically constrained tricyclic phosphine 3a was investigated by spectroelectrochemistry. The data suggested a highly negative reduction potential with the reversible formation of a dianionic species. Accordingly, 3a reacted with two equivalents of Li/naphthalene by reductive cleavage of a P–C bond of one of the PC4 heterocycles. The resulting dilithium compound 5 represents a phosphaindole derivative with annulated aromatic C6 and PC4 rings. It is an interesting starting material for the synthesis of new heterocyclic molecules, as was shown by treatment with Me2SiCl2 and PhPCl2. The structures of the products (6 and 7) formally reflect ring expansion by insertion of silylen or phosphinidene fragments into a P–C bond of 3a. Treatment of 3a with H2O2 did not result in the usually observed transfer of a single O atom to phosphorus, but oxidative cleavage of a strained PC4 ring afforded a bicyclic phosphinic acid, R2PO2H.

Electrochemical and spectroelectrochemical data

General electrochemical procedures
All electrochemical (EC) and in situ UV-vis spectroelectrochemical (in situ SEC) measurements were performed in a Glovebox Pure Lab HE GP-1 SR (Innovative Technology, USA) within an atmosphere of purified nitrogen (<0.1ppm O 2 ; <0.1ppm H 2 O). The glovebox was equipped with military grade BNC feedthroughs in a homemade gas tight flange for low noise electrical connection of electrochemical and spectroelectrochemical cells inside. UV-vis-NIR spectroscopy inside the glovebox was performed via another homemade, gas tight flange containing four light-tight chambers with optical grade fused silica windows and four 74-UV (Ocean Optics Inc., USA) collimating lenses with SMA 905 connectors for optical fibers. EC and SEC cells were connected to a PGSTAT302 (Metrohm Autolab, Utrecht, The Netherlands) E= ±10 V, U= ±35 V with an auxiliary voltage monitor (BK Precision 2831E, Yorba Linda, CA USA). NOVA Software (Metrohm Autolab) Version 1.11.2 was used to control the potentiostat, magnetic stirring and to trigger the spectrometer. UV-vis spectra recorded by OceanView software (Version 1.4.1, Ocean Optics) were triggered by a DG1032Z arbitrary waveform function generator (Rigol, Beijing, China) controlled by the Nova Software (1.10.4, Fa Metrohm Autolab) using a DAC164 analog/digital converter. Data analysis of the spectral and electrochemical data was carried out using OriginPro 2019 (OriginLab Cooperation, Northampton, MA, USA).

Solvent preparation / supporting electrolyte preparation
THF was dried by refluxing with potassium and benzophenone as an indicator until a midnight blue color was established, CH 3 CN and PhF were dried by heating with CaH 2 to reflux conditions for at least 6 h and distillation under Argon onto vacuum activated molecular sieves (CH 3 CN: 3 Å ; PhF, THF: 4 Å, 1·10 -3 mbar, 350 °C, 12 h). Additionally, all solvents were stored over vacuum activated molecular sieves for at least 14 days. [ n Bu 4 N][OTf] was pre-dried at least 5 times by dissolving in CH 2 Cl 2 and evaporating the solvent to high vacuum at 80 °C. Final drying and removal of HOTf traces was achieved by dissolving in dry benzene and refluxing this solution in a soxleth apparatus for 5 days with vacuum activated molecular sieves in the extraction thimble renewed every day. Prior to each measurement the solvent is passed through a Pasteur pipette with an activated (1·10 -3 mbar, 350 °C, 24 h) aluminum oxide bed (D = 5 mm; L = 70 mm) in the glove box before the supporting electrolyte is added. In PhF as a solvent a substrate-based non-reversible reduction I at E P (red) = -2.62 V (vs. E 1/2 (Cp 2 Fe(Cp 2 Fe + )) and an non-reversible oxidation II at E P (ox) = 1.12 V were observed as broad peaks in the CV (Figure SI.1.). Square wave voltammetry revealed half-wave potentials according to the CV at E 1/2 (red) = -2.75 V for the reduction and E 1/2 (red) = 0.72 V for the oxidation reaction. Follow-up reactions are observed after the oxidation reaction III in the second cycle as non-reversible re-reduction peaks III´ and III´´ at E P (re-red) = -1.07 V and E P (re-red) = -1.55 V. Therefore, PhF is not usable as solvent for further spectroelectrochemical investigations due to a low diffusion coefficient of 3a in the reduction reaction I and the follow-up reactions III´´ and III´´´ after the substrate oxidation III. In CH 3 CN as a solvent phosphine 3a showed a reasonable sharp non-reversible reduction I peak E P (red) = -2.36 V (vs. E 1/2 (Cp 2 Fe(Cp 2 Fe + ); Figure SI.2. dotted curve) attributed to the two-electron reduction mechanism of 3a to [3a'] 2-. The non-reversible oxidation III was observed at E P (ox) = 1.28 V (dashed line). Additionally, extended CV measurements of the reduction reaction I show the known two-electron re-oxidation II (solid line) of dianion [3a'] 2to 3a' and back to 3a (vide infra) at E P (re-ox) = -0.7 V. SWV measurements allow an assignment of the half-wave potentials E 1/2 (SWV) = -2.38 V to the substrate-based reduction I and E 1/2 (SWV) = 1.15 V to substrate-based oxidation II.

Cyclic voltammetry data of 3a in PhF, CH 3 CN and THF
Further investigation of the substrate based reduction I and the substrate based oxidation reaction II were conducted by a comparison of the full potential window CVs (dashed line) versus the partial CV scan for the oxidation follow-up processes ( Figure   [OTf] at a D = 3 mm platinum disc electrode to assign processes of the full potential window cycle to follow-up oxidation processes II and I´ of the substrate reduction process I. Unfortunately, the oxidation III of phosphine 3a leads to three follow-up reduction reactions III´ / III´´ / III´´´ E P (red) = -0.79 / -1.44 / -1.81 V indicating unwanted reactivity of the oxidation product in acetonitrile (Figure SI.3.). In acetonitrile an additional re-oxidation reaction I´ E P (ox) = 0.55 V ( Figure SI.4.) after the already assigned re-oxidation reaction II is also found, rendering acetonitrile as not suitable solvent to handle the two-electron reduction (I) product [3a'] 2due to further reactions. Based on this detailed peak assignment acetonitrile is not suitable for the reduction of phosphine 3a although the potential range of CH 3 CN (in blank measurements) lead to another prediction.

S6
CV measurements in THF do not allow for the observation of the oxidation reaction due to the limited oxidative potential window, but the reductive stability avoid side reactions after multiple reduction I cycles (Figure SI.5.). This enables an exhaustive re-oxidation II after the reduction product [3a'] 2is formed ( Figure SI.6.). Figure SI.5. shows a CV measurement cycling exclusive around the nonreversible reduction peak I. No changes starting from the second cycle are observed is in accordance with the mechanism hypothesis of the two-electron reduction followed by a very fast chemical followup reaction forming [3a'] 2without the formation of electro-active side products: 3a + 2 e -→ [3a'] 2-(EEC mechanism, see manuscript). Additionally, the back reaction from dianion [3a'] 2to 3a via an EEC mechanism with a slower chemical step. Investigation by composited CV cycles (Figure SI.6.). producing the two-electron reduction I product [3a'] 2in a first cycle followed multiple cycles to show the exhaustive properties of the re-oxidation II due to limited amounts of [3a'] 2in the diffusion layer from the first cycle.

Additional in situ UV-vis spectroelectrochemical data and mechanistic
investigation of the reduction of 3a   In addition to the in situ UV-vis MPCA experiment ( Figure 5, manuscript) a background experiment was designed in order to differentiate between diffusion and re-oxidation of dianion [3a'] 2to 3a. Therefore, a chronoamperametric pulse to trigger the reduction I reaction 3a → [3a'] 2was applied, followed by a chronopotentiometric pulse controlling the current to zero. During the chronopotentiometric pulse no reduction or re-oxidation happened. Only diffusion effect from the thinlayer area to the bulk compartment of the double-compartment cuvette-cell can take place. The losses by diffusion effects during the time with zero current conditions show only slight changes on the concentration of [3a'] 2in the cuvette cell. According to this finding, the reduction I and re-oxidation II in the in situ UV-vis MPCA measurement Figure 4 (manuscript) is suitable to proof the stability, spectroscopic and chemical reversibility of the reactions between 3a and [3a'] 2-.
In addition to the half-life time measurement of the intermediate 3a´ for the EEC reoxidation-mechanism ([3a'] 2-→ 3a) the EEC reduction-mechanism (3a → [3a'] 2-) was further investigated. The EEC hypothesis of the reduction is given by a CV experiment with an elevated scan rate of v = 10 V/s ( Figure  SI10). The reduction peak I has significantly broadened but remains up to 10 V/s non-reversible. Separation of the EE and C part of the mechanism was not possible due to the fast chemical follow-up reaction.

Preparative electrochemical reduction of phosphine 3a
Preparative electrochemical synthesis was used to gain UV-vis data of the reduction product dianion [3a´] 2on a preparative scale. Based on the assumed low diffusion coefficient deduced from the CV data potentiostatic conditions were chosen. Therefore, a solution of 0.28 mmol phosphine 3a in 35 mL THF with 0. Electrolyses was performed potentiostatic with E = -3.00 V at room temperature insight a nitrogen filled glove box. Figure SI11 shows the current over time and after integration the charge consumption. After 60 min of electrolysis time an in situ CV was measured by switching from the preparative to the analytical Pt disc electrode. The potentiostatic synthesis was stopped after 20 h to avoid contamination from the counter electrode compartment. Due to the low current a conversion of only 34% based on a two-electron process with 100% current yield was achieved. In situ CV before the electrolysis ( Figure SI12) show a very broad peak I for the reduction process, due to the high concentration of phosphine 3a in the preparative electrolysis cell. The reoxidation peak II also showing a reasonable broadening and low intensity. After 23% conversion and a drastic color change from yellow to deep red (insert figure SI13) the CV show a significant increase of the reoxidation peak II due to the formed bulk concentration of [3a´] 2from the preparative reduction. Further low intensity peaks (*) indicate side reactions between decompositions products from the counter electrode compartment caused by the prolonged electrolysis time.   Figure S24: 1 H-NMR spectrum of 8 C 6 D 6 Figure S25: 13 C{ 1 H}-NMR spectrum of 8 in C 6 D 6 (211.5 ppm = Cr(CO) 6 ) S18 Figure S26: 31 P{ 1 H}-NMR spectrum of 8 in C 6 D 6 2.5 NMR-Spectra of Compound 9 Figure S27: 1 H-NMR spectrum of 9 in C 6 D 6 S19 Figure S28: 13 C{ 1 H}-NMR spectrum of 9 in C 6 D 6 Figure S29: 31 P{ 1 H}-NMR spectrum of 9 in C 6 D 6 S20 2.6 NMR-Spectra of Compound 10 Figure