Chalcogen‐Transfer Rearrangement: Exploring Inter‐ versus Intramolecular P−P Bond Activation

Abstract tert‐Butyl‐substituted diphospha[2]ferrocenophane has been used as a stereochemically confined diphosphane to explore the addition of O, S, Se and Te. Although the diphosphanylchalcogane has been obtained for tellurium, all other chalcogens give diphosphane monochalcogenides. The latter transform via chalcogen‐transfer rearrangement to the corresponding diphosphanylchalcoganes upon heating. The kinetics of this rearrangement has been followed with NMR spectroscopy supported by DFT calculations. Intermediates during rearrangement point to a disproportionation/synproportionation mechanism for the S and Se derivatives. Cyclic voltammetry together with DFT studies indicate ferrocene‐centred oxidation for most of the compounds presented.


Introduction
Chalcogen transfer is af undamental reactioni nm olecular chemistry and chalcogenophosphoranes (phosphine chalcogenides) are well established reagents for this purpose. [1] Although the preparation of phosphorus sulfides, as an example, is known since centuries, mechanistic detailso nh ow chalcogen atoms are transferred are still quite limited with some light shed by as tudy addressing the oxidation of white phosphorus with sulfur in the melt and at low temperature under thermal and photochemical conditions. [2] The scenarioi s complicated by initial chalcogena ddition to either ap hospho-rus lone pair,o ri nsertion into aP -P bond, potentially followed by subsequent interconversion via chalcogen migration proceedingi ntra-or intermolecularly (Scheme 1). Diphosphanes and diphosphane monochalcogenides are suitable model compoundsfor the study of such processes featuring aphosphorus lone pair as well as aP -P bond. For the case of sulfur,d iphosphane monosulfides have been obtained via desulfurization of diphosphane disulfides, [3] oxidation of diphosphanes with elemental chalcogen, [4,5] or synproportionationo fd iphosphanes and diphosphanedisulfides. [4] For the heavierc halcogen selenium, the corresponding diphosphane monoselenide was reported using the oxidationr oute, [6] and via metathesis of metal selenide and halophosphanes. [7,8] In ac omprehensive study, [8] the steric influence on the relative ratio of the isomeric diphosphane monochalcogenides A and diphosphanylchalcoganes B has been exploredf or Ch = S, Se andT e, including the thermal conversion A!B whichh ad been studied independently for Ch = Ob efore. [9] The aspect of ring-straino nt his transformation has been demonstrated elegantly for an o-carborane bridgedd iphosphane with Ch = Se. [10] While diphosphanylchalcogane isomer B is observed exclusively for Ch = Te , [8] the opposite is true for oxygen,w here diphosphane monochalcogenide A is dominant, except for stereo-electronically unique organofluoro substituents. [11] Based on previousr esults, [12] we considered tert-butyl substituted diphospha [2]ferrocenophane 1 as suitable diphosphane to study chalcogen-transferr eactions as it combines two stereogenic P III centres with lone pairs and an on-polar > P-P < bond. This compound is available as as ingle diastereomer with trans orientation of the tert-butyl groups adjacent to the P-stereogenic centres, [13] whichs hould allow facile monitoring of epimerizationv ia 31 PNMR spectroscopy.H ere we reporto ur investigation concerning addition vs.i nsertion of chalcogen atoms to the diphosphane unit in 1 and its stereochemical course.

Results and Discussion
As as tarting point of our investigation an improved synthesis of diphospha [2]ferrocenophane 1 has been developed furnishing analyticallyp ure 1 by simple recrystallization in good yields (70 %). Our approache mploys oxidative P-P bond formation from the corresponding bisphosphanide using various oxidants (SiBr 4 ,C Cl 4 ,C HCl 3 ,C HBr 3 ,P bCl 2 )o ut of which PbCl 2 gave best results (Scheme 2). The advantage over ap ublished route via reductiveP -P bond formation accompanied by formationo f by-products,i st he ease of purification requiringn oc olumn chromatography as in the previousmethod. [13] To explore whether one chalcogen atom (Ch = O, S, Se, Te ) can be transferred selectively to diphosphane 1 by stoichiometric control, we employed reagentsc ontaining the respective chalcogen-chalcogen bonds. While H 2 O 2 or tert-butyl hydroperoxide, S 8 and grey seleniumf urnish cleanly monooxidized diphosphane monochalcogenides 2a-c at room temperature, no reaction is observed with elemental tellurium under these conditions (Scheme 3). Nevertheless, at elevated temperature partial oxidation of 1 by tellurium can be observed (33 % yield besides unreacted 1). Unlike for the other chalcogens, oxidation of 1 with tellurium leads exclusively to the ring expanded telluradiphospha [3]ferrocenophane 3d.
Upon monooxidation, the 31 PNMR chemicals hift of 20.6 ppm in 1, [13] changes to lower field for the tetracoordinate phosphorus atom in 2,w hile its trivalent counterparti ss hifted to higher field (Table 1). Chemical inequivalence of the 31 P nucleii n2 entailss ignal splitting, for which the value of 1 J PP increases with the size of the chalcogen (Table 1). For 2c the 77 Se chemical shift at À311.7 ppm showsc oupling to the directly bonded tetracoordinate andt he trivalentp hosphorus atoms ( 1 J SeP = 725 Hz, 2 J SeP = 30 Hz). Identity and purity of the above mentioned monooxidized diphosphanes was furtherc orroborated by 1 H, 13 CNMR, mass spectrometry and elementala nalysis.S ingle crystals suitable for X-ray diffraction have been obtained for 2a,b (Figure 1). Remarkably the P-P bond lengths, 2.240(3) in 2a and 2.246 (2) in 2b are almost identicalt ot he one in non-oxidized 1.
[13a] Compared with the latter,t he tetracoordinate phosphorus Scheme1.Illustrationofa ddition,insertion and migration of chalcogen atoms for the formation and interconversion of isomeric diphosphanemonochalcogenides A and diphosphanylchalcoganes B.
In light of the thermally induced rearrangement of diphosphanyl monochalcogenides 2b,c!3b,c,o utlined above, we wondered, whether twofold oxidized compounds 4 would  behavel ikewise (Scheme 5). Indeed, thiodiphosphorane 4b rearranges in as imilar and stereospecific fashion to 5b.T he transformation seemst ob em uch more facile sincei to ccurs at lower temperature (170 8C) and is completed after af ew hours. The rearrangemento ft he selenodiphosphorane 4c proceeds accordingly yielding 5c.S urprisingly,t he isomerization of 4c takes place at higher temperature (150 8C) compared with 2c. In contrast to the monooxidized phosphorane 2a which shows no rearrangement to the correspondingd iphosphanylchalcogane, ar earrangement of twofold oxidized 4a to 5a could be observed at high temperature (190 8C) over ap eriod of several days.

Mechanism
To understand the above mentioned chalcogen-transfer rearrangement, we set out to explore the energetics of the reaction and its intermediates for the transformation 2!3,d escribed above. Investigatingt he mechanismo ft he insertion with DFT methods, established that the process is slightly exothermica nd in general 3 is more stable than 2.T he energetic preference increases in the order a < b < c < d.( The calculated values vary between 11.3-(À30.5) kJ mol À1 ,d epending on the appliedDFT functionals Table S4-S6 in the Supporting Information). Firstly,w ec onsidered am onomolecular reactionm echanism for the transformation 2b!3b as depicted in Scheme S1 in the Supporting Information. (Similar processes werecalculated for 2c!3c and 2d!3d transformations,t abulated data in Ta ble S4-S6 in the Supporting Information). The rate limiting step is the chalcogen insertion and the reaction barrier of this   step (190.8 kJ mol À1 )i sc omparable with the bonding energy of the P-P bond (~200 kJ mol À1 ). Similar mechanisma nd slightly lower barriers were obtained for the 4!5 rearrangements (more details in the Supporting Information Scheme S4). It should be highlighted that these barriers should be thermally accessible at the reaction temperature (over 100 8C), but do not rule out other mechanistic pathways. Besides ah ypothetic intramolecular rearrangement also ab imolecular mechanism needs to be considered. In the literature ac oncerted rearrangementh as been proposed for the rearrangement of acyclic diphosphane monochalcogenidet ob isphosphanylchalcogane. [8] Unfortunately,a ll our attempts to localize the corresponding transition states for the concerted bimolecular rearrangements failed in our hands. Based on the optimized molecular structures, stericr epulsion between the tert-butyl groups and the ferrocenem oieties precludesaconcerted bimolecular mechanism showingn oe nergetic preference over the monomolecular rearrangementoutlined in Scheme S1. In support of as tepwise (non-concerted) rearrangement, trackingt he transformation 2b!3b via 31 P-NMR spectroscopy revealed the transient occurrence of intermediate 5b along with desulfurized 1 in equal amounts. The identityo fi ntermediately formed 5b has been confirmed by comparison with independently synthesized 5b (v.i.). Similarly,f or the selenium transfer rearrangement 2c!3c,t racking the rearrangement by NMR spectroscopy confirmed 5c along with deselenised 1 as intermediates. Consistent with the proposed reaction pathway, comproportionation of isolated 5c and 1 afforded 3c together with 2c,w hichw ere formed in equal amounts. Therefore, we conclude adisproportionation/synproportionation type mechanism for the chalcogen-transfer rearrangementu nder discussion (Scheme 6). The bimolecularn ature of the rate determining step of this non-concerted chalcogen-transfer rearrangement was established by NMR studies, running the reaction 2c!3c at differentc oncentrationsw hich confirmed as econdorder kinetics for selenium transfer (Figure6).
The suggested non-concerted bimolecular mechanism was investigated computationally as well. We have considered different pathways (more details in the Supporting Information) and generally it could be established that, the chalcogen transfer steps exhibit moderate barriers (DG # = 57.7-69.0 kJ mol À1 ) for Ch = Se, which is comparable with the experimentally obtained value and the bimolecular nature of the rate limiting step. On the other hand, it should be noted that in all investigated cases the chalcogen insertion into the P-P bond remain over 160 kJ mol À1 (Scheme S4 in Supporting Information), which is not consistent with our experimental observations.
To resolve this ambiguity,f urther calculation and experiments were carried out. TD-DFT calculations indicate electron transfer from the lone pairs of the chalcogens towards the P-P antibonding orbital in case of 2b,c and 4b,c,w hichr esults in elongation of the P-P bond. Therefore, possible photoactivation cannot be excluded. Despite the promising DFT results, running the reaction 2c ! 3c at different temperatures under exclusion of light (brown glass NMR tubes; Figure S11, the Supporting Information) revealn os ignificant influence by light and the determined value of the activation energy (E A = 76 (AE 3) kJ mol À1 )i sc omparable with the prior one. Also running the reaction 2c!3c at room temperature, but under illumination with UV-light (Hg-lamp, 15 W) shows no [3]ferrocenophane formation after severalhours of illumination.
Hypotheticallyt he oxidantp roperties of chalcogens, may furnish oxidationo ft he ferroceneu nit which subsequently may undergo electron transfer with the neighbouring diphosphane unit, thus giving rise to radical speciese ven in the absence of light. Therefore, we explored the redox properties of central compounds of this investigationb yc yclic voltammetry (CV). In the proposedd isproportionation/synproportionation mechanism (Scheme6)d iphosphane 1 is oxidized by 5,t herefore we explored the electrochemical oxidation behaviour of 1 as as tartingp oint. Diphosphane 1 shows two distinct oxidation events with peak potentials at 0.29(1) Va nd 0.97(1) V( vs. Fc + /Fc;b lue line in Figure S1). The first current response shows reversible redox behaviour,i ft he voltage sweep is reverted before the second redox process. The second current response is reversible as well. However,w ith decreasing sweeping rate upon going through both redox processes in a single potentials weep,t he nature of both processes and the first redox process in particular becomes irreversible in the reductions weep ( Figure S1). As iron or phosphorus centred oxidation is known to be feasible in related compounds, [12b, 15d] DFT calculations were performed where the calculated spin density (at B3LYP/6-311G**//wB97XD/6-311 + G**) of 1 + + is localized at the ferrocene unit. Similarly,t he CVs of chalcogeno and dichalcogenop hosphoranes 2a-c and 4a-c have been recorded( Figures S2-3 and Figure S5,T he Supporting Information), indicating reversible iron centred oxidation as the primary redox event for most compounds (2a:0 .43(2) V; 4a: 0.54(1) V; 2b:0 .48(1) V; 4b:0 .61(1) V; all vs. Fc + /Fc). The assignment to iron centred oxidation is in accordance with the calculated spindensity of the corresponding radicalc ations. An exception are selenophosphoranes 2c and 4c showing irreversible electro-chemical responses for the oxidation (Figure S2 and S3) The reason of this observation can be explained with the increased energy level of the lone pairs of the selenium (compared to So rO ), which have more significant contribution to the HOMO, than the ferrocene unit ( Figure S7). The impact of the rearrangement 2!3 on the electronic situation of the ferroceneu nit may also be assessed by CV for the sulfur speciesw here the ferrocene-based oxidationp eak potential changes from 0.48(2) V( vs. Fc + /Fc)f or 2b to 0.32(1) V( vs.F c + /Fc) for 3b.I na greement with the reversible process the Fe centre has significant contribution to the spindensity distribution of the corresponding radical 3b + .S eleniuma nalogue 3c reveals af errocene-centred oxidation at ap eak potential of 0.33(1) V( vs.F c + /Fc) as well, which is reversible if the potential sweep is switched to reversed scan direction right after first oxidation by avoiding the oxidationo ft he P-Se-P scaffold. The tellurium analogue 3d shows aq uasi-reversible ferrocene based redox process at ap eak potential of 0.37(2) V( vs. Fc + /Fc), if the scan directioni ss witchedr ight after the oxidation ( Figure S4). In summary,t he electrochemical investigation outlined above give no indication for redox based radicalf ormation during oxidation andrearrangementof1 or 2 with chalcogens. As af inal check to prove or disprove the involvement of radicals peciesi nt he thermalc halcogen-transferr earrangement 2!3 we performed the reaction in an EPR spectrometer at high temperature in as ealed EPR tube which did not give any indication for radical formation, unlike in other cases. [12b]

Conclusion
We reported the addition versus insertion of chalcogens to a stereochemically confined diphosphane 1 taking advantage of the ferrocenophane scaffold. The resulting diphosphanemonochalcogenides 2 undergo ac halcogen-transfer rearrangement to the corresponding diphosphanylchalcoganes 3 for Sa nd Se. The analogous behaviour is observed for twofold oxidized chalcogenodiphosphoranes 4 which rearrange stereospecifically to asymmetrically substituted 5.T he mechanism of the chalcogen-transferrearrangementwas found to be bimolecular and non-concerted based on NMR and DFT studies. Detection and identification of by-products during rearrangement suggest ad isproportionation/synproportionationm echanism for 2b!3b,w hile no evidence for radicali ntermediates or photoactivation could be found. The general relevance of our results emerges from the prevalenceo fP III ÀP III units in all phosphorus allotropes with the competition of addition to P III atoms versus insertioni nto P-P-bonds being of utmosti mportance for P 4 activation or phosphorene functionalization being as urging topic in sustainable and materials chemistry.I nf uture work we intend to generalize the stereospecific addition/insertion of other amphiphilic species into PÀPb onds of moderately strained oligophosphanes aiming at new routest oc hiral organophosphanes.