Synthesis and Thermally and Light Driven Cleavage of an N-Heterocyclic Diphosphine with Inorganic Backbone

. A diphosphine with an unsupported PP bond connecting two carbon-free “inorganic” 1,3,2,4,5-diazaphosphadisilolidine rings was prepared by reductive coupling of a P-chloro-substituted monocyclic precursor molecule. VT-EPR studies revealed that the diphosphine exists in solution, like other compounds of this kind, in dynamic equilibrium with the corresponding phosphinyl radicals. Determination of the radical concentration from the EPR spectra permitted to calculate ther-mochemical parameters for the homolytic PP bond fission. The results


Introduction
Phosphinyl radicals are divalent phosphorus compounds of general composition R 2 P • that are distinguished by an open shell electron structure with a formal 7 VE electron configuration.Known examples span the whole range from transient species, which occur merely as short-lived reaction intermediates (e.g.H 2 P • ), [1] over persistent radicals that possess extended life times in condensed phase and are easily detected spectroscopically, [2] to thermally stable compounds that can be isolated in crystalline form. [3]The creation of long-lived phosphinyl radicals can be accomplished by different routes.Reduction of chlorophosphine precursors enabled the generation of the first persistent phosphinyl radicals 1a,b (Scheme 1) by the group of Lappert in 1976, [4] and was also employed by the groups of Cummins, [5] Bertrand, [6] and Ishida and Iwamoto [7] in their syntheses of the isolable crystalline radicals 2-4.Of these species, 2 and 3 benefit from electronic stabilization by two transition metal imido or guanidinato moieties, respectively, while 4 is considered an electronically unperturbed radical that owes its existence exclusively to kinetic stabilization 1 disclose that both the enthalpy and entropy of dissociation are higher than in topologically related bi(diazaphospholidines).The impact of the entropy term allows explaining that, regardless of the presence of an energetically rather stable PP bond, the onset of dissociation is observable even at ambient temperature.Irradiation experiments showed that radical formation cannot only be induced thermally, but also by photolysis.
An alternative to radical generation via redox processes is the homolytic fission of the PP bond of diphosphines {5} 2 or {6} 2 , respectively.This process is usually initiated by thermal activation, and fission products 5, 6 and their precursors exist usually in a temperature dependent equilibrium (Scheme 2). [9]he prevalence of the dimers at low temperature and in the solid state indicates radicals 1a, [10] 5, 6 to be, unlike 2-4, thermodynamically unstable towards dimerization.9e,11] Interpretation of these data with the help of computational studies suggested that the stabilization of phosphinyl radicals is predominantly steric in origin, while electronic effects seemed to have only minor impact.In accord with an earlier postulate derived from theoretical models, [10]  data corroborated that radical formation is facilitated when the dimers are loaded with strain energy that can be discharged upon bond fission.A considerable fraction of this strain energy is attributed to unfavorable distortions of bond and torsional angles that are enforced by a high degree of steric congestion in the overcrowded dimer and can be healed by structural relaxation of the sterically more flexible free radical. [10,11]A closer inspection revealed, however, that the situation is more complicated as bulky substituents may not only favor radical formation, but can also contribute to the stabilization of the dimers by dispersion forces. [11,12]Due to the action of many variables, a concise and detailed understanding of the factors governing the energetic and entropic contributions to radical stabilization is yet out of reach, and further studies are needed.Cyclic radicals still seem a viable entry point as their higher conformational rigidity is expected to facilitate insight.Following this rationale, we report here on a study of the PP bond fission of bis-heterocyclic diphosphine {7} 2 (Scheme 2) featuring a carbon-free, purely "inorganic" cyclic backbone.The molecular structure relates topologically to those of bi(diazaphospholidines) 6b-6d, but formal replacement of the ethanediyl moiety by a larger and less flexible Si 2 Me 4 unit is expected to provide for angular distortions and conformational preferences that differ from those of 6b-6d.Moreover, we established for the first time that irradiation with UV-light can induce an isothermal shift of the radical/diphosphine equilibrium to generate a photo-stationary state characterized by an increased radical concentration.

Syntheses
As a precursor to diphosphine {7} 2 , we first synthesized previously known chlorophosphine 8 [13] following a reported two-step procedure involving double amination of 1,2dichloro-1,1,2,2-tetramethyl disilane with tert-butylamine [14] and subsequent ring closure with phosphorus trichloride.For-mation of the diphosphine was then accomplished by reductive coupling of 8 with magnesium in THF, using sonication in an ultrasound bath to activate the metal (Scheme 3).Work-up afforded a moderate yield of crystalline product, which was identified by spectroscopic and analytical data.Both 8 (the crystal structure of which was previously unknown) and {7} 2 were further characterized by single-crystal X-ray diffraction studies.

Crystallographic Studies
The availability of crystallographic data for 8 (Figure 1) and 2-thiocyanato-1,3-di-tert-butyl-4,4,5,5-tetramethyl-1,3,2,4,5diazaphosphadisilolidine (9) [15] (Scheme 4) permitted comparing the structural features of the heterocyclic rings in the sterically strained dimer {7} 2 (Figure 2) and the less strained monocyclic "molecular halves".Inspection of selected metric parameters (Table 1) reveals that the Si-Si and Si-N distances in all compounds are essentially indistinguishable within experimental error.The P-N distances in the dimer are slightly elongated by some 0.06 Å, but similar effects occur also in monocyclic diazaphospholenes and diazaphospholidines and are attributable to the modulation of hyperconjugation between the nitrogen lone-pairs and the various exocyclic P-X bonds rather than steric effects. [16]We conclude therefore, that any steric strain associated with the formation of the dimeric struc-  a) Data from reference [15] .b) X = Cl (8), N (9), P ({7} 2 ).c) Sum of all bond angles at the central atom.1.
ture in {7} 2 leaves the bond lengths in the heterocycles untouched.
A similar invariance is also obvious for the endocyclic bond angles in the five-membered rings, whereas the pronounced asymmetry of the exocyclic N-P-P angles in {7} 2 (Table 1) gives a first indication for a sterically induced structural distortion in the dimer.Further angular variations, which tend to flatten the pyramidal coordination at the phosphorus and induce a slight pyramidalization of the coordination sphere around the nitrogen atoms, are too small to claim significance on their own account.However, in combination with additional warps in torsional angles, these deformations effect a clearly visible change from an envelope conformation of the fivemembered rings with ecliptic alignment of the Me 2 Si-units in the monomers 8, 9 to a twist conformation with slightly staggered SiMe 2 moieties in {7} 2 .

Radical Generation
The formation of radicals through thermally induced PP bond homolysis was monitored by recording EPR spectra of a solution of {7} 2 in anhydrous and degassed toluene.As in case of bi(diazaphospholidines) {6b,c} 2 and acyclic tetraaminodiphosphines {5c,d} 2 , [9d,11] an EPR signal (g = 2.0038) was already visible at ambient temperature (293 K) and increased steadily in intensity upon warming the solution up to a maximum temperature of 373 K.The signal displays a characteristic multiplet pattern arising from hyperfine coupling of the electron spin with the nuclear spins of phosphorus ( 31 P, I = ½), nitrogen ( 14 N, I = 1) and silicon ( 29 Si, I = ½, nat.abundance 4.6 %) that is typical for diaminophosphinyl radicals and was analyzed by simulation of the spectrum recorded at 363 K (Figure 3).The magnitude of the hyperfine couplings to the phosphorus nucleus [a( 31 P) = 75.4G] is larger and that to the nitrogen nuclei [a( 14 N) = 2.2 G] smaller than in diazaphospholidinyl radicals 6b,c [a( 31 P) = 60.9-63.8G, a( 14 N) = 3.7-4.3G], [11] but the phosphorus hyperfine coupling matches the values reported for acyclic diaminophosphinyls [e.g.a( 31 P) = 76.2G (5a), 75.5 G (5c) [11] ].The observation of a signal multi-ARTICLE plicity revealing hyperfine coupling of the electron spin in 7 with two equivalent 14 N nuclei indicates that the coordination planes of the phosphorus and both nitrogen atoms are, like in diazaphospholidinyls, close to coplanar.The observation of a significantly smaller 29 Si hyperfine coupling of 3.2 G than in 5a [a( 29 Si) = 12.2 G [11] ] is attributable to a different orientational preference of the N-Si bond, which lies in the NPN plane in 7 (see below) and is nearly perpendicular in 5a. [11]ecause the changes in EPR signal intensity were fully reversible upon cooling, we conclude that radicals 7 and diphosphine {7} 2 undergo, as in other reported cases, [9,11] reversible dynamic equilibration.Following a previously described procedure, [9e] we evaluated thermochemical parameters for the dissociation {7} 2 i 2 7 from the temperature dependent variation of the equilibrium constant K = c(7) 2 /c({7} 2 ) (see Figure 4).
The resulting value of ΔG Diss 295 of 90( 5) kJ•mol -1 , which translates into an equilibrium exponent pK Diss 295 = 15.9(10)reveals that the equilibrium composition matches at ambient temperature roughly that observed for 6c, but is further shifted to the side of the diphosphine than in 6b (Table 2).It should be noted that similar Gibbs enthalpies of dissociation were also determined for acyclic tetraaminodiphosphines {5a} 2 and {5d} 2 , whereas for the bi(diazaphospholidine) {6d} 2 bearing identical N-substituents as {7} 2 no thermally induced radical formation is reported.The substantially higher dissociation enthalpy ΔH Diss of 135.2(26) kJ•mol -1 for {7} 2 implies that PP bond homolysis is in this case energetically less favorable than in all reference compounds in Table 2, and it is only due to the compensation of the unfavorable energy term by  an unusually large positive entropy contribution [ΔS Diss = 154(8) J•K -1 •mol -1 ] that radical formation at ambient temperature is at all observable.
Having previously reported that bi(diazaphospholidines) and bi(diazaphospholenes) are also accessible by photolysis driven dehydrocoupling of secondary phosphine precursors, [17] we wanted to explore the feasibility of this route for the preparation of {7} 2 .The secondary phosphine (Me 2 Si(tBu)N) 2 PH (10)  was readily prepared by LiBEt 3 H reduction of 8 and identified by its NMR spectroscopic data. [18]Irradiation of a hexane solution of this species with a medium pressure mercury lamp produced indeed the expected target diphosphine {7} 2 , but the product was unstable under the reaction conditions and decomposed eventually into yet unidentified products, rendering this approach synthetically not useful.Attempting to monitor the photolysis reaction by EPR spectroscopy, we noted, however, that spectra of reaction mixtures containing {7} 2 recorded under continuous UV irradiation displayed the EPR signal of 7, which vanished when the irradiation source was turned off.Hypothesizing that radicals 7 may arise from a photochemical cleavage of {7} 2 , we recorded ambient temperature EPR spectra of a hexane solution of the pure dimer both under continuous UV irradiation (at a wavelength of 365 nm hitting the low energy tail of the first electronic transition of {7} 2 cen-ARTICLE tered at 304 nm) and in the dark.The outcome of this study revealed that the intensity of the EPR signal increases indeed significantly when the spectrum is recorded under continuous irradiation (Figure 5a).Time dependent monitoring of the EPR signal discloses that the raise in intensity comes abruptly when the irradiation is switched on, and then slowly adjusts to a new equilibrium level.Switching the light off immediately reestablishes the original signal strength (Figure 5b, c).Similar results were also obtained with a sample of {5c} 2 (see Supporting Information).We attribute the incidental raise in signal intensity to a shift in the thermal equilibrium composition caused by formation of an excess concentration of radicals via photochemically induced PP bond homolysis.The excess radical concentration is maintained as long as the light source is on, but returns back to the thermal equilibrium value when it is switched off.The immediate restitution of the original equilibrium concentration implies that the surplus radicals recombine at a rate faster than the time resolution of the instrument.The non-equilibrium rad-ical concentration level reached during the irradiation period must then be considered a photo-stationary state in which the thermal balance between radical formation and annihilation is disturbed by the generation of excess radicals through the photolysis process.The slow decay of the radical concentration during the early stage of the irradiation period is less well understood.We hypothesize that this phenomenon is possibly connected with the dissipation of the recombination energy produced by the excess radicals, which may accelerate radical recombination due to local heating until a new, stable temperature gradient in the sample is reached.It should be noted that the generation of phosphinyl radicals by UV-induced reduction of chlorophosphine precursors with electron rich olefins is known, [4,9b,19] but a light driven shift of PP-bond homolysis equilibria is to the best of our knowledge unprecedented.

Computational Studies
DFT calculations on radical 7 and diphosphine {7} 2 were carried out at the ωB97x-D/cc-pVDZ level of theory that had already been employed in earlier studies on radicals 5, 6. [11] The subsequent discussion is based on molecular geometries of 7 and {7} 2 that were obtained by energy optimization using the solid-state structures of 8 and {7} 2 as starting points, and established as local minima on the energy hypersurface by frequency calculations.Computed and experimental molecular structures of {7} 2 are in good agreement, and the computations succeed in particular in the critical prediction of the P-P distance, yielding a value of 2.297 Å that is very close to the experimental result of 2.3034(4) Å.The computed molecular structure of radical 7 is characterized by an essentially planar conformation of the heterocycle and an eclipsed alignment of the methyl substituents in 4,5-position.
Computation of the magnetic properties of radical 7 yields values of 85.7 G and 3.5 G for the hyperfine couplings a( 31 P) and a( 14 N).Even if these figures lie somewhat above the experimental data, they are a good match to the values computed for 5, 6, [11] where a very similar overestimation of the experimental data was achieved, and must be considered quite reasonable results in view of the fact that no conformational averaging was performed.The localization of 77 % and 20 % of the computed spin density on the p orbitals of the phosphorus and the nitrogen atoms, respectively, reproduces the results obtained on 5b,c and reveals that the modification in the heterocyclic backbone leaves the electronic structure of the radical unaffected.
While the computational model is known to fail in reproducing quantitatively the energetics of the dissociation, [11] comparison of calculated dissociation enthalpies for {7} 2 (163.2kJ•mol -1 ) and bi(diazaphospholidines) {6b} 2 (158.4 kJ•mol -1 ) [11] and {6c} 2 (157.4 kJ•mol -1 ) [11] confirms that PP bond fission in the former is expected to be slightly more endoergic, even if the observed difference is underestimated.Moreover, as had been noted before, the variation in entropy contributions is not reproduced very well, and a reliable analysis of the impact of steric factors on the dissociation energetic values is thus currently unfeasible.

Figure 1 .
Figure 1.Representation of the molecular structure of 8 in the crystal.Hydrogen atoms are omitted for clarity.Thermal ellipsoids are drawn at the 50 % probability level.Selected distances and angles are listed in Table1.

Figure 2 .
Figure 2. Representation of the molecular structure of {7} 2 in the crystal.Hydrogen atoms are omitted for clarity.Thermal ellipsoids are drawn at the 50 % probability level.Selected distances and angles are listed in Table1.

4 Figure 4 .
Figure 4. Van tЈHoff plot of R ln(K Diss (T)) vs. 1/T.Data points are represented as diamonds, and the straight line is the result of a linear regression analysis.ΔH Diss = 135.2(26)kJ•mol -1 and ΔS Diss = 154(8) J•(K•mol) -1 were calculated from the slope and intersection of the regression line, respectively.

Figure 5 .
Figure 5. (a) Expansion of the low field multiplet component in EPR spectra of a hexane solution of {7} 2 recorded at 298 K without (blue trace) and with (black trace) continuous UV-irradiation at 386 nm.(b) Time dependence of the signal strength at a fixed magnetic field of 3330.2G. UV-irradiation at 386 nm was switched on 10 sec and switched off 40 sec after the measurement had been started.(c) Monitoring the signal strength as in (b) during three consecutive cycles with 10 sec irradiation periods.Vertical axes represent arbitrary units.
in Table 1.Scheme 4. Molecular structure of 9.ARTICLE