Unexpected Isomerization of Hexa‐tert‐butyl‐octaphosphane

Abstract Octaphosphane {cyclo‐(P4 tBu3)}2 (1) undergoes an unexpected isomerization reaction to the constitutional isomer 2,2′,2′′,2′′′,3,3′‐hexa‐tert‐butyl‐bicyclo[3.3.0]octaphosphane (2) in the presence of Lewis acidic metal salts. The mechanism of this reaction is discussed and elucidated with DFT calculations. The results underline the fascinating similarity between phosphorus‐rich and isolobal carbon compounds. The new bicyclic octaphosphane 2 shows a dynamic behavior in solution and reacts with [AuCl(tht)] (tht=tetrahydrothiophene) to give a mono‐ ([AuCl(2‐κP 3)], 3) and a dinuclear complex ([(AuCl)2(2‐κP 3,κP 3′)], 4). With cis‐[PdCl2(cod)] (cod=1,5‐cyclooctadiene), the chelate complex ([PdCl2(2‐κ2 P 2,P 2′)], 5) with a different coordination mode of the ligand was obtained.


Experimental Section General Remarks
All experiments were carried out under nitrogen using standard Schlenk techniques. Toluene, dichloromethane, diethyl ether, pentanes as well as hexanes (isomeric mixtures) and acetonitrile were dried and degassed with an MB SPS-800 Solvent Purification System (MBRAUN) and kept over molecular sieves (4 Å). THF was distilled from potassium and benzophenone and kept over molecular sieves (4 Å). The compounds {cyclo-(P4 t Bu3)}2 (1) [1] and [AuCl(tht)] [2] were prepared as described in the literature. All other chemicals were purchased and used without further purification.
Elemental Analysis. A VARIO EL (HERAEUS) microanalyzer was used to determine the elemental composition of the compounds.
Luminescence. Luminescence was tested at 254 and 366 nm using a common UV lamp.

Reaction Study of the Isomerization of Octaphosphanes 1 and 2
The reactions were carried out in sealed glass NMR tubes. The respective octaphosphane and ZnCl2 were added and evacuated, then 0.6 mL C6D6 were added. The suspension was kept at 70 °C in a thermostat without stirring. For the reaction with 0.1 equivalents of ZnCl2, the right amount of ZnCl2 was added by preparing a stock solution of ZnCl2 in water, adding this to the NMR tube and evaporating the solvent. The composition of the reaction mixture was determined based on the relative intensity of all t Bu protons from 1 H{ 31 P} NMR spectroscopy. Heating was stopped for approximately 30 minutes for each measurement. The amounts of the starting materials in the different reactions were:

Synthesis of [(AuCl)2(2)]·2THF (4·2THF)
A solution of [AuCl(tht)] (58 mg, 180.9 mmol, 2.0 eq) in 8 mL THF was added dropwise to a solution of octaphosphane 2 (53 mg, 89.9 mmol, 1.0 eq) in 7 mL THF resulting in a deep yellow solution. Subsequently, the solvent was evaporated under reduced pressure, the residue dissolved in 3 mL THF and carefully layered with 12 mL of hexanes. After three days at 3 °C, large colorless crystals had formed which were isolated by decantation, washed with 2 mL of hexanes and dried in vacuum. The compound showed no luminescent behavior. The crystals can be dissolved in THF and the solvent removed under reduced pressure yielding the solvent free complex as a colorless powder.

Reaction of Octaphosphane 2 with [(AuCl)2(2)] (3)
10 mL of THF were added to a mixture of octaphosphane 2 (12 mg, 20 mmol, 1.0 eq) and complex of 4 (22 mg, 20 mmol, 1.0 eq). After 5 minutes, the suspension became clear and turned pale yellow. The solvent was removed under reduced pressure, the resulting residue dried in vacuum and investigated with NMR spectroscopy. The substance was analytically identical to 3.

Synthesis of [PdCl2(2)] (5)
1,4-Dioxane (5 mL) was added to a mixture of [PdCl2(cod)] (68.9 mg, 241.3 mmol, 3.3 eq) and octaphosphane 2 (43.8 mg, 74.1 mmol, 1.0 eq). Heating to reflux for 2 h resulted in a deep red solution. After cooling to rt, the solvent was removed under reduced pressure, the residue washed twice with 10 mL hexanes each and extracted 3 times with 5 mL toluene each. Evaporation of the solvent of the combined phases lead to a brown powder, which was dried in vacuum for 15 minutes.   [3] Every signal labeling in this publication also extends to the chemically equivalent yet magnetically nonequivalent nuclei i.e. PA for PA, PA' as well as PA'' and so forth. The 31 P NMR signals are labeled alphabetically starting from the nucleus which is most deshielded. All 31 P NMR spectra were measured with a 90° pulse and reduced acquisition (0.6 s) as well as d1 time (1.0 s). At least 1000 scans were necessary to obtain acceptable spectra. The samples were usually saturated solutions of the respective compound. Suitable spectra for simulation required about 100,000 scans. Simulation of the NMR spectra was performed using the DAISY module implemented in the program TopSpin version 3.6.1 (BRUKER, BioSpin GmbH, Rheinstetten). For this, the 1 JPP coupling constants were set negative. [4] NMR        Figure S11. 1

Mass spectrometry measurements
Mass spectrometry measurements were carried out as ESI-MS with a BRUKER DALTONICS FT-ICR-MS spectrometer (Type APEX II, 7 Tesla).
The identity of 2 is supported by mass spectrometry. Two ions attributable to the monoxide of 2 are detected as adducts with H + or Na + , respectively. They occur together with the higher oxides as a series of six peaks [M(P8 t Bu6Ox)] + (M = H, Na; x = 1-6) separated by m/z = 16. No peaks for [M(P8 t Bu6Ox)] + ions with x = 7 or 8 are detectable. This is in accordance with the observations from mass spectrometry of octaphosphane 1 [1] as well as the literature, which indicate that these octaphosphanes can be oxidized to the hexoxide but no further. [5] Obviously, the tendency of the P atoms that carry no tBu group to be oxidized is extremely low. Another signal series that can be assigned to [Na(P8 t Bu6)2Ox] + with x = 2-9 is found. No peak for the unoxidized octaphosphane is detectable. This could be due to lower ionization efficiency for the parent octaphosphane 2 or because it is swiftly oxidized under the conditions of the MS measurement.

X-Ray Diffraction Single Crystal Measurements
Single crystal X-ray diffraction data were collected with a GEMINI CCD diffractometer (RIGAKU). The radiation source was a molybdenum anode (Mo-Kα, λ = 0.71073 Å). The absorption corrections were carried out semiempirically with the SCALE3 ABSPACK module [6] . All structures were solved by dual space methods with Sir-92 [7] and SHELXT-2017. Structure refinement was done with SHELXL-2015/2018 [8] by using full-matrix least-square routines against F 2 . All hydrogen atoms were calculated on idealized positions. The pictures were generated with the program ORTEP3 [9] . In all pictures, hydrogen atoms and solvent molecules were omitted for clarity and thermal ellipsoids are shown with 50% probability. CCDC 1956729 (2) For the crystal structure of 3: The complex consists of one octaphosphane molecule coordinated to an AuCl fragment close to a THF molecule. The THF molecule is partially replaced by a second AuCl fragment, which thus corresponds to the bimetallic complex 4. The THF molecule and the Cl atom of the second fragment could only be refined isotropically. The occupancy was refined to 4.442%.

•
For the crystal structure of 4: There are contacts between the CH3 (H11) group in the complex and the CH2 (H25A) group of the solvent molecule. This is most likely no attractive intermolecular interaction but results from close packing of the molecules. The dihedral angle λ in all molecular structures was extracted from the crystal structures as the mean value of the two dihedral angles between the two bridgehead phosphorus atoms and two phosphorus atoms which are connected to them and part of the same ring. Figure S17. Illustration of the determination of the dihedral angle λ. The grey circles mark one set of atoms which were used for the determination.

Powder X-ray Diffraction Measurements
Powder X-ray diffraction data were collected on a STADI-P diffractometer (STOE) with a silicon solid-state detector Methyn-1K (DECTRIS) at room temperature. The radiation source was a copper anode (Cu-Kα, λ = 1.540598 Å) combined with a germanium single crystal monochromator. Samples were measured in sealed glass capillaries (inner diameter 0.5 mm, HILGENBERG) with Debye-Scherrer geometry. Processing of the raw data was carried out with the diffractometer software WinXPow [10] (STOE). The theoretical powder pattern from the single crystal structures was calculated using MERCURY [11] .

Benchmark Study
The method for optimizing the molecular structures of the gold(I) complexes was benchmarked based on the molecular structure of the trinuclear complex of octaphosphane 1 obtained from single crystal XRD measurements and was also used for the gold complexes presented in this work. [1] The BP86 functional [16,17] in combination with the def2-SVP [18,19] basis set was used due to the best agreement with the experimental data.

Librational Profile of Octaphosphane 2
These calculations were carried out using the BP86 [16,17] functional in combination with the def2-TZVP [18,19] basis set. A single point calculation of the optimized molecular geometries using the def2-TZVP basis set in combination with the PBE0 [20,21] functional was performed to obtain the final energies. The transition state was located by a relaxed surface scan of the dihedral angle between the bridgehead P atoms and two of the directly connected P atoms, leaving the second one unconstrained. The transition state structures were verified as such by showing only one negative eigenvalue of the Hesse matrix, which corresponds to the described chemical motion.
Scheme S1. Libration of octaphosphane 2. The transition state corresponds to the average structure in solution. The remaining structures (left and right) correspond to the enantiomeric structures found in the solid state. Figure S24. Librational profile of octaphosphane 2 with the relative energy (left) and the second, non-fixed dihedral angle (right) as a function of the fixed angle λ.

Mechanism of the Isomerization Reaction
These calculations were carried out using the B3LYP [22,23] functional in combination with the def2-TZVP [18,19] basis set.

Comparison of Isomers of E8R6
The energies of different isomers were calculated with the combinations of functional and basis sets shown in Table S2. The relative energies of the different structures are displayed in Table S3.