Radical-initiated P,P-metathesis reactions of diphosphanes: evidence from experimental and computational studies

Abstract

By combining the diphosphanes Ar₂P–PAr₂, where Ar = C₆H₅, 4-C₆H₄Me, 4-C₆H₄OMe, 3,5-C₆H₃(CF₃)₂, it has been shown that P,P-metathesis generally occurs rapidly under ambient conditions. DFT calculations have shown that the stability of unsymmetrical diphosphanes Z₂P–PZ′₂ is a function of the difference between the Z and Z′ substituents in terms of size and electronegativity. Of the mechanisms that were calculated for the P,P-metathesis, the most likely was considered to be one involving Ar₂P˙ radicals. The observations that photolysis increases the rate of the P,P-metatheses and TEMPO inhibits it, are consistent with a radical chain process. The P,P-metathesis reactions that involve (o-Tol)₂P–P(o-Tol)₂ are anomalously slow and, in the absence of photolysis, were only observed to take place in CHCl₃ and CH₂Cl₂. The role of the chlorinated solvent is ascribed to the formation of Ar₂PCl which catalyses the P,P-metathesis. The slow kinetics observed with (o-Tol)₂P–P(o-Tol)₂ is tentatively attributed to the o-CH₃ groups quenching the (o-Tol)₂P˙ radicals or inhibiting the metathesis reaction sterically.

---

Introduction

Dynamic and reversible chemical reactions have garnered increasing attention in medicine and biology,¹ supramolecular chemistry² and materials science,^3–11 and they are key to the rapidly growing field of dynamic covalent chemistry (DCC).^1,12,13 In DCC, disulfides play a leading role, by virtue of the weakness of the S–S bond and their ready availability via chemical synthesis and their presence in natural products.^12,14–16 In the last decade, several other main group element–element bonds (O–O,^17,18 N–N¹⁹ and Se–Se^9,20,21) have been investigated for their potential in dynamic chemistry with a view to their application in the field of polymeric materials. A notable omission from this group is diphosphanes (P₂R₄) which contain a P–P bond whose strength (ca. 52 kcal mol⁻¹)²² is intermediate between S–S (ca. 62 kcal mol⁻¹)²³ and Se–Se (ca. 40 kcal mol⁻¹)²² and therefore diphosphanes should be candidates for DCC.

Diphosphanes have received considerable attention as targets themselves and also as undesired byproducts in the synthesis of tertiary phosphines.^24–31 Several methods to prepare diphosphanes are available, including: Wurtz-type reductive coupling of chlorophosphines (with Li, Na, K, Mg and Hg);^32–34 salt metathesis between chlorophosphine and LiPR₂;^26,27 dehydrocoupling of HPR₂;^35–39 chlorosilane elimination;⁴⁰ P–N/P–P bond metathesis.⁴¹ The chemistry of diphosphanes can be divided into two categories: reactions where the P–P bond remains intact and reactions where the P–P bonds are cleaved.⁴² Reactions involving P–P cleavage are pertinent to the chemistry described in this article and include additions of X₂ to the R₂P–PR₂ to give monophos species R₂PX (e.g. X = H, Cl) and additions of R₂P–PR₂ to C=C or C≡C bonds (diphosphination) to give diphos species (Scheme 1). It has also been shown that diphosphanes containing P–N bonds can add to C=S or C=O bonds.^43,44 Mechanisms have been proposed for some of the P–P cleavage reactions including radical chain processes.^43,45

Scheme 1 Some reactions involving P–P cleavage of tetraphenyldiphosphane.⁴²

Diphosphane metathesis has not been systematically investigated, but an early ³¹P NMR study by Harris et al. showed that rapid metathesis occurred between Me₂P–PMe₂ and (F₃C)₂P–P(CF₃)₂ in CH₂Cl₂ to form Me₂P–P(CF₃)₂.^46,47 Gilheany et al.⁴⁸ have shown that the calculated high energy barrier to P,P-metathesis via a concerted mechanism is not compatible with the observed rapid kinetics for the Me₂P–P(CF₃)₂ system and suggested that pathways involving impurities in the diphosphanes may facilitate the P,P-metathesis process. Grubba et al.²⁶ have succeeded in the preparation of a variety of pure, unsymmetrical diphosphanes containing P–C(alkyl) and P–N bonds. By employing low temperatures, they avoided unwanted P,P-metatheses that would lead to symmetrical diphosphanes contaminating the products. Moreover, they reported that R₂PCl or R₂PLi can catalyse diphosphane metathesis. We have reported that diphosphane metathesis was a side reaction in the attempted synthesis of some unsymmetrical diphosphanes.²⁷

In this study, we have investigated the P,P-metathesis reactions of tetra-aryldiphosphanes (Scheme 2) using ³¹P{¹H} NMR spectroscopy and computational methods, with a view to exploring the dynamic nature of these P–P bonds. This has provided some new insights into the diphosphane metathesis reaction.

Scheme 2

Results and discussion

Thermodynamics of diphosphane metathesis

Experimental equilibrium studies. The tetra-aryldiphosphanes A₂–E₂ (Chart 1) used in this study, were chosen in order to sample a range of steric and electronic effects. All have been previously reported and were prepared by modified literature procedures for the reductive coupling of the corresponding chlorophosphines by Mg in THF (see ESI† for details).^26,34 In each of the diphosphane samples, residual chlorophosphine was not detected by ³¹P NMR spectroscopy and was therefore estimated to be less than 0.05% (see ESI†).

Chart 1

When two of the diphosphanes A₂–E₂ were mixed in CDCl₃ at ambient temperature, a reaction ensued to give an equilibrium mixture of homodiphosphanes X₂ and Y₂, and the heterodiphosphane XY (Scheme 2). The data for the diphosphane metathesis reactions are given in Table 1.

Table 1 Data for the P,P-metathesis equilibria shown in Scheme 2

Entry	X2	Y2	K	Time^a/min	^31P NMR data for XY
a Time by which equilibrium had been established according to ^31P NMR spectroscopy. b In THF. c In CDCl3. d In C6D6. e Estimated value (see details in ESI†).
1^b	A2	B2	4	30	−14.9	−16.4	J PP = 162 Hz
2^c	A2	C2	5	20	−15.9	−18.4	J PP = 159 Hz
3^d	A2	D2	30^e	90	−15.9	−8.5	J PP = 177 Hz
4^d	B2	C2	4	15	−16.8	−18.8	J PP = 156 Hz
5^b	B2	D2	15	20	−11.7	−15.7	J PP = 165 Hz
6^b	C2	D2	3	20	−10.5	−15.4	J PP = 173 Hz
7^c	A2	E2	4	700	−16.6	−35.3	J PP = 152 Hz
8^c	B2	E2	4	1000	−19.1	−36.2	J PP = 149 Hz
9^c	C2	E2	4	2800	−19.0	−36.9	J PP = 148 Hz
10^c	D2	E2	30	4000	−17.9	−29.5	J PP = 156 Hz

---

The ³¹P{¹H} NMR data for the XY species (Table 1) show that the chemical shifts for the two signals are close to the respective X₂ and Y₂ with ¹J_XY ≃160 Hz. For example, Fig. 1 is the ³¹P{¹H} NMR spectrum obtained after mixing diphosphanes B₂ and C₂ (entry 4, Table 1) and shows the singlets for the reactants along with an AB pattern for the product BC; the spectra for the other diphosphane combinations are given in the ESI.†

Fig. 1 The ³¹P{¹H} NMR spectrum of the P,P-metathesis reaction between (p-Tol)₂P–P(p-Tol)₂ (B₂) and (p-Anisyl)₂P–P(p-Anisyl)₂ (C₂).

The pure heterodiphosphane AE was prepared in order to establish the P,P-metathesis equilibrium with A₂ and E₂ (Scheme 2) from the opposite direction. Grubba et al.²⁶ have prepared a range of heterodiphosphanes (Z₂P–PZ′₂) by the addition of Z₂PLi to Z′₂PCl at low temperatures. However, our attempts to extend this route to the preparation of AE were unsuccessful, even with the reaction mixture maintained at −78 °C throughout; instead, an equilibrium mixture of A₂, E₂ and AE was obtained. This contrasting behaviour may be due to the lack of thermodynamic stability of heterodiphosphane AE compared with the Grubba heterodiphosphanes²⁶ (see below). The desired AE was prepared in >96% purity (after recrystallisation from methanol) via the protection/deprotection method (Scheme 3) that we previously developed for the synthesis of heterodiphosphanes.^27,49 Solutions of pure AE in CDCl₃ do indeed equilibrate to a mixture of A₂, E₂ and AE (see below for further details).

Scheme 3

The diphosphane equilibrium constants for the P,P-metatheses (Scheme 2) fall in the range 3–30 (see Table 1), which corresponds to ΔG values between −0.5 and −2 kcal mol⁻¹, demonstrating that the diphosphanes in this study are very close in energy and insensitive to the steric and electronic effects of the aryl substituents. The equilibria that do not involve E₂ were established rapidly (see Table 1) despite the dilution (12.5 mM) which is a useful property for the potential application of diphosphanes in dynamic chemistry.¹ The anomalously slow P,P-metatheses involving E₂ are discussed later.

Computational equilibrium studies. To understand better the factors that control the stability of heterodiphosphanes with respect to the corresponding homodiphosphanes, DFT calculations were carried out to determine the thermodynamics of the hypothetical equilibria shown in Scheme 4 in which the range of P-substituents Z and Z′ is greatly extended from the substituted aryls in Scheme 2. The values for ΔG are collected in Table 2, and for some of the combinations (Table 2, entries 1–3 and 7–10), the calculated |ΔG| is less than 3 kcal mol⁻¹ and therefore, these particular P,P-metathesis equilibria are predicted to be finely balanced. The calculated ΔG of 1.2 kcal mol⁻¹ (Table 2, entry 22) for A₂/E₂ is remarkably close to the experimental value of −0.8 kcal mol⁻¹ obtained from the experimentally determined value of K = 4 in CDCl₃ solution (Table 1, entry 7).

Scheme 4

Table 2 Calculated ΔG values for the diphosphane metathesis equilibria Scheme 4

Entry	Z	Z′	ΔG/kcal mol^−1
1	Me	H	−0.5
2	Me	NH2	−2.9
3	Me	OH	−2.9
4	Me	OMe	−3.7
5	Me	F	−6.6
6	Me	CF3	−10.5
7	Ph	H	0.8
8	Ph	Me	−0.7
9	Ph	NH2	1.3
10	Ph	NMe2	−0.8
11	Ph	OMe	−4.7
12	Ph	F	−6.1
13	Ph	CF3	−4.4
14	^tBu	H	−13.4
15	^tBu	Me	−3.3
16	^tBu	Ph	−4.0
17	^tBu	NH2	−6.8
18	^tBu	NMe2	−6.8
19	^tBu	OMe	−11.2
20	^tBu	F	−15.4
21	^tBu	CF3	−3.4
22	o-Tol	Ph	1.2

---

Two gross trends emerge from the data in Table 2: (1) a bias in favour of the XY species is calculated when there is a large electronegativity difference between the X and Y fragments; thus Me₂P–P(CF₃)₂ is predicted (and experimentally observed⁵⁰) to be highly favoured (Table 2, entry 6); (2) a strong bias in favour of the XY species is calculated when there is a large difference in steric bulk between the X and Y fragments; thus ^tBu₂P–PH₂ is predicted to be highly favoured (Table 2, entry 13). The largest calculated ΔG in favour of the XY species is for ^tBu₂P–PF₂ (Table 2, entry 20), at an apparent apotheosis of both trends. The observed trends in the calculated relative stabilities of the XY species in Scheme 4 can be rationalised as follows: (a) electronegative Z substituents (such as F) will lead to repulsive electrostatic interactions between the δ+ charges generated on the adjacent P atoms in diphosphane X₂; (b) bulky Z′ substituents (such as ^tBu) will lead to steric congestion in diphosphane Y₂; (c) in XY, both of the P–P bond-destabilising effects identified in (a) and (b) will be reduced and, in some cases, a destabilising, repulsive, electrostatic (δ+/δ+) interaction is replaced by a stabilising, attractive, electrostatic (δ+/δ−) interaction.

The calculated ΔG values for some of the Grubba heterodiphosphanes²⁶ (Table 2, entries 16 and 18) suggest that these should have high thermodynamic stability with respect to the corresponding homodiphosphanes. By contrast, the heterodiphosphine AE, as reported above, is calculated to be of similar energy to the homodiphosphanes A₂ and E₂ (Table 2, entry 22).

Computational mechanistic studies of diphosphane metathesis

The mechanism of the diphosphane metathesis shown in Scheme 5 has been previously investigated computationally by Gilheany et al.⁴⁸ The lowest energy route that they could find involved the ionic intermediate shown in Scheme 5 but the calculated barrier of 36.5 kcal mol⁻¹ was too high to explain the rapidity of the P,P-metathesis that is observed experimentally.

Scheme 5 Gilheany mechanism for a P,P-heterometathesis.

We have investigated the mechanism of diphosphane metathesis computationally using the degenerate exchange reactions of Me₂P–PMe₂ (Scheme 6). Tetramethyldiphosphane was chosen in order to limit the number of rotamers and thus the conformational noise in the calculations. It was hoped to distinguish between the three plausible pathways (i–iii) shown in Scheme 7: (i) a [2 + 2] concerted process involving a 4-membered intermediate or transition state; (ii) nucleophilic attack by one diphosphane on another to give an ionic intermediate (similar to the one suggested in Scheme 5); (iii) a [2 + 1] radical chain process.

Scheme 6

Scheme 7

The overall reaction profile calculated for the thermal rearrangement of Me₂P–PMe₂ (shown in Fig. 2(a) proceeds via intermediate int1 depicted in its lowest energy conformation in Fig. 2(b). The structure of int1 can be viewed in terms of the two resonance forms int1A and int1B shown in Fig. 2(c) (further details are given in the ESI†). The ionic form int1A is analogous to the intermediate suggested for the P,P-heterometathesis shown in Scheme 5 and the covalent int1B contains an unusual central P(V) moiety. The profile shown in Fig. 2 corresponds to pathway (ii) of Scheme 7, and notably, attempts to locate a synchronous, σ-bond metathesis pathway (corresponding to pathway (i) in Scheme 7) were unsuccessful; no imaginary normal modes corresponding to the 2 bond-breaking/forming vibrations are present along the relaxed, symmetric P₄ ring expansion surface. The energy of int1 of 44.1 kcal mol⁻¹ for the PMe₂ species is lower for the analogous PPh₂ intermediate (ΔG = 38.1 kcal mol⁻¹, see ESI†) but this is calculated to be not lowered by solvation with CH₂Cl₂. Thus, the calculated very high barrier is incompatible with the rapidity of the Ar₂P–PAr₂ metathesis reactions that is observed experimentally.

Fig. 2 (a) Reaction free energy profile for the P,P-homometathesis of Me₂P–PMe₂. Calculations performed at the PBE0-D3BJ/def2-TZVPP/PBE0-D3BJ/def2-SVP level of theory. (b) Calculated structures of ts1 and int1. Note: ts1 and ts2 are the same; key distances given in Å. (c) Resonance forms of int1.

The radical chain pathway (iii) in Scheme 7 requires homolysis of a P–P bond. Thermolysis is not viable under ambient conditions because the calculated P–P bond enthalpy is of the order of 52 kcal mol⁻¹. However, DFT calculations for Ph₂P–PPh₂ (A₂) suggest that excitation from S₀ → S₁ or S₀ → S₂ would result in rapid homolysis of the P–P bond, yielding phosphanyl radicals (Ar₂P˙) via intersystem crossing from S₁ or S₂ to the first excited triplet state (T₁) of A₂; the T₁ state is dissociative, and would lead to spontaneous radical formation. From the Ar₂P˙, the P,P-metathesis barrier generating XY species is only ∼5 kcal mol⁻¹, such that once radicals are generated an equilibrium would be expected to be established rapidly (see ESI† for details).

The DFT results suggested that Ar₂P˙ radicals (generated photolytically) were viable intermediates in the P,P-metathesis reactions and so this hypothesis was pursued experimentally.

Qualitative experimental kinetic studies of diphosphane metathesis

The aryl diphosphane metathesis reactions with A₂–D₂ occurred readily (typically <1 h) under ambient conditions in common organic solvents (C₆D₆, THF, PhMe, PhCl, CDCl₃ and CD₂Cl₂). By contrast, when the diphosphane E₂, containing o-tolyl substituents, was one of the reaction partners, P,P-metathesis was initially only observed to take place in CDCl₃ (typically over 6–12 h) and CD₂Cl₂ (typically over more than 24 h); the anomalous behaviour observed for the P,P-metatheses involving E₂ will be discussed after consideration of the P,P-metatheses involving A₂–D₂.

When cold (−78 °C) THF solutions of B₂ and C₂ were mixed and the P,P-metathesis reaction monitored by low-temperature ³¹P NMR spectroscopy, it was found that only traces of BC were detected at −80 °C after 1 h. By raising the temperature in increments of 20 °C, it was established that the P,P-metathesis at −20 °C progressed at a convenient rate to monitor the approach to P,P-metathesis equilibrium over a period of 2 h by ³¹P NMR spectroscopy (see Fig. 3). Although a detailed kinetic study was not attempted, the data do not appear to fit a reaction order of 1, 2 or 0.5 (see ESI†).

Fig. 3 (a) Plot of heterodiphosphane BC formation as a function of time over 2 h; (b) ³¹P{¹H} NMR spectra of the reaction between B₂ and C₂ at −20 °C in THF, shown in 10 min increments over the course of 1 h (with inverse-gated decoupling). See Fig. 1 for the peak assignments.

Evidence for radicals in the general P,P-metathesis reactions

In light of the computational results above suggesting that radicals may be involved in the mechanism of P,P-metathesis, experiments were carried out to determine if there was evidence of their presence. Two pieces of experimental evidence support the involvement of radicals in the P,P-metathesis process.

(1) When A₂ and C₂ were mixed in the presence of the radical scavengers 2,2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) or tri(t-butyl)phenol (TTBP), the progress of the P,P-metathesis reactions was greatly inhibited. Thus, a mixture of A₂ and C₂ proceeded smoothly to the metathesis equilibrium over a period of 2 h in CDCl₃, at −20 °C (with 35% conversion at 1 h); by contrast, when the same A₂/C₂ mixture was prepared in the presence of 5 equiv. of TEMPO, there was no formation of AC detected after 1 h. Moreover, when the reaction mixture containing TEMPO was allowed to warm to ambient temperature, <10% of AC was detected even after 16 h. Instead, in the presence of TEMPO, two singlet ³¹P resonances at +30.5 and +31.5 ppm were observed (ca. 10% of the total signal intensity), consistent with the formation of Ar₂P(O)-TEMPO (Ar = Ph, p-anisyl) adducts (Scheme 8) which would be expected to be formed by the interception of the phosphanyl radicals (Ar₂P˙) by TEMPO.⁵¹

Scheme 8

(2) When A₂ and B₂ were mixed in a standard NMR tube in THF at ambient temperatures, equilibration was essentially complete within ca. 15 min. However, when the same A₂/B₂ mixture in THF was prepared in an amberised NMR tube, to minimise the transmittance of UV and visible radiation, the equilibration to the same degree took ca. 150 min. This is consistent with diphosphane photodissociation being a significant component of the mechanism of P,P-metathesis.

The UV–vis spectra of A₂, C₂ and E₂ are shown in Fig. 4. Each spectrum has a major absorbance between 225 and 250 nm, and a shoulder at ca. 270 nm. TD-DFT calculations for A₂ were carried out using a range of functionals (CAM-B3LYP, PBE0 and M06-2X) and good agreement with experiment was found with CAM-B3LYP. These calculations suggest that the major peak in each spectrum corresponds to coincident aromatic π → π* and S₀ → S₂ (see Fig. 5(b)) transitions and the shoulder peak corresponds to S₀ → S₁ (see Fig. 5(a)). The transition electron density plots (Fig. 5) indicate that both transitions result in a significant delocalisation of electron density from the P lone pairs. The first triplet state, accessible from either S₁ or S₂via intersystem crossing, is dissociative, leading to spontaneous formation of phosphinyl radicals. There are small differences in the three spectra shown in Fig. 4. As expected, the p-OMe groups in C₂ produce a bathochromic shift in the π → π* band and there is a long ‘tail’ that extends towards the visible region, which explains why C₂ is pale yellow.

Fig. 4 UV–vis spectra of A₂, C₂ and E₂ in CH₂Cl₂ (20 μM).

Fig. 5 Transition electron density difference plots with the CAM-B3LYP/def2-TZVPP//PBE0-D3BJ/def2-SVP level of theory. (a) S₀ → S₁ and (b) S₀ → S₂. Red and blue correspond to areas of depleted and excess electron density, respectively. Isosurface plotted at a contour of 0.003 au.

The evidence in support of radicals being present led to the radical chain mechanism proposal in Scheme 9 where homolysis of the Ar₂P–PAr₂ is followed by attack of the Ar₂P˙ on the second diphosphane Ar′₂P–PAr′₂. The termination steps are then the radical couplings with P–P bond formation.

Scheme 9

The photodissociation requires UV radiation and it appears that there is sufficient background UVA-UVB radiation (280–400 nm) for the initiation step to proceed.

The two degenerate propagating steps, where Ar₂P˙ and Ar′₂P˙ react with their respective dimers Ar₂P–PAr₂ and Ar′₂P–PAr′₂, are omitted from Scheme 9. The reverse P,P-metathesis process would also be initiated by photodissociation of Ar₂P–PAr′₂. The P,P-metathesis reactions generally proceed with a high degree of chemoselectivity but traces of Ar₂PH were often evident after several days of reaction which would arise from termination via abstraction of an H radical, presumably from the solvent.

The radical mechanism for diphosphane metathesis in Scheme 9 is reminiscent of the radical mechanism for disulfide metathesis (Scheme 10) arrived at from calculations by Ruipérez et al.⁵² and from experimental observations by Asua et al.⁵³ DFT and MD calculations were applied to two mechanisms of diaryldisulfide metathesis: a [2 + 2] concerted metathesis and a [2 + 1] radical chain reaction initiated by sulfanyl radicals (ArS˙).⁴⁷ The transition state in the concerted process was calculated to be too high in energy to be viable, whereas the radical process was viable provided the ArS˙ radicals could be accessed. This reasoning mirrors ours for the mechanisms proposed for diphosphane metatheses shown in Scheme 7.

Scheme 10

It has been reported that the diaryldisulfide metathesis between diphenyldisulfide and di(4-aminophenyl)disulfide proceeds spontaneously to equilibrium in ca. 12 h; this was accelerated by UV photolysis, and TEMPO completely inhibited the equilibration.⁵³ These observations made on the disulfide metathesis reactions (and analogous diselenide studies)⁵⁴ are very similar to our observations on the diphosphane metatheses.

Anomalous kinetics with (o-Tol)₂P–P(o-Tol)₂

The four P,P-metathesis reactions between tetra(o-tolyl)diphosphane (E₂) and A₂–D₂ did not appear to proceed at all in THF, PhMe, PhCl, PhF or C₆D₆ under ambient conditions after 72 h but do proceed in the chloroalkanes CH₂Cl₂ and CHCl₃, albeit slowly (see Table 1, entries 7–10). To probe this anomalous behaviour further, we have investigated the A₂/E₂ system (Fig. 6) in more detail including approaching the equilibrium starting from an A₂/E₂ mixture and starting from isolated AE.

Fig. 6 ³¹P{¹H} NMR spectra for the mixture of A₂ and E₂: (a) in toluene after 48 h, showing no detectable P,P-metathesis; (b) in CDCl₃ after 6 h, showing equilibrium with AE has been established.

The progress of the A₂/E₂ metathesis in CDCl₃ was followed by ³¹P NMR spectroscopy (see ESI† for details) with a spectrum recorded every 10 min for 12 h (see Fig. 7). The approach to equilibrium was measured from integration of the ³¹P NMR signals to produce the plot shown in Fig. 7. Analysis of the curve revealed that it was not consistent with the integrated equations for an order of reaction of 0, ½, 1 or 2; this indicates that the kinetics are complex, which would be expected if more than one mechanistic pathway was operating (see below).

Fig. 7 (a) Plot of the data obtained from integration of the ³¹P{¹H} NMR signals for the AE formed in the reaction between diphosphanes A₂ and E₂; (b) ³¹P{¹H} NMR spectra in CDCl₃ obtained by sampling over 12 h. See Fig. 6 for the peak assignments.

Evidence for radicals in the anomalous (o-Tol)₂P–P(o-Tol)₂ metathesis reactions

There are several pieces of evidence supporting a role for radicals in the P,P-metathesis equilibrium shown in Fig. 6:

(1) The lack of any reaction between A₂ and E₂ in THF after 72 h was replaced by 30% AE formation when the solution was irradiated with a UV lamp for 2 h. Similarly, no equilibration was observed in a THF solution of pure AE over a period of 72 h but proceeded smoothly over 2 h upon irradiation of the same solution with a UV lamp. This suggests that the photodissociation mechanism shown in Scheme 6 for the general P,P-metathesis (Table 1, entries 1–6) is viable for the A₂/E₂ metathesis but requires a more intense source of UV radiation than background UV for the initiation step.

(2) The slow A₂/E₂ metathesis equilibration in CDCl₃ which takes place over a period of 12 h was extinguished in the presence of 4 equiv. of the radical scavenger TEMPO, as no AE was detected after 16 h.

(3) Pulse-sonication (20 kHz) at 0 °C of a CHCl₃ or CH₂Cl₂ solution of A₂ and E₂ reduced the time to establish equilibrium from 12 h to <1 h. However, sonolysis had no accelerating effects on the reaction of A₂ with E₂ in THF or toluene. Fritze and von Delius reported the metathesis of disulfides (Scheme 10) in chloroform was accelerated by sonolysis and suggested that this proceeded via RS˙ radicals formed from the reaction of disulfides with Cl˙ generated by sonolytic chloroform degradation.¹⁶

(4) The onset of equilibration of solutions of AE in CDCl₃ was preceded by an induction period of up to 6 h; moreover, the time required to establish the P,P-metathesis equilibrium from mixtures of A₂/E₂ in CDCl₃ varied from 6 to 12 h. These observations of rate-inconsistency are typical of radical processes.⁵⁵

In addition to the evidence for a radical-promoted process, it is shown below that there is evidence of chlorophosphine involvement in the A₂/E₂ metathesis.

Evidence for chlorophosphine intermediacy in the anomalous (o-Tol)₂P–P(o-Tol)₂ metathesis reactions

Under the conditions where a mixture of A₂ and E₂ in CDCl₃ would reach equilibrium with AE within 12 h, the presence of H₂O (100 equiv.) suppressed the reaction to the extent that no AE was detected even after 20 h (indicating that <1% AE had been formed). Instead, several minor signals (ca. 20% in total) were present and those at δ+ 23.5, +21.4 were tentatively assigned to Ar₂P(O)H (Ar = Ph or o-Tol). These observations are consistent with the water reacting rapidly with the Ar₂PCl compounds that are formed. It is notable that the addition of water to the P,P-metathesis reaction mixture of A₂ with C₂ in CDCl₃ had no discernible effect on the rapid formation of AC which indicates that Ar₂PCl is not involved in the P,P-metathesis for this typical system.

When a solution of A₂ alone in CHCl₃ was irradiated with UV, some conversion to Ph₂PCl was observed directly. The stoichiometric reaction of A₂ with (o-Tol)₂PCl in CHCl₃ immediately produced a mixture of A₂, AE, and Ar₂PCl (Ar = Ph, o-Tol) (Scheme 11) along with several unidentified P-containing species. After 4 h, E₂ was detected along with small amounts of secondary phosphines Ar₂PH (Ar = Ph, o-Tol). These observations confirm that Ar₂PCl compounds are plausible promoters of P,P-metathesis.

Scheme 11 P,P-Metathesis promoted by Ar₂PCl.

To explore whether Ar₂PCl could catalyse P,P-metathesis, a sub-stoichiometric amount (5 mol%) of Ph₂PCl was added to a chloroform solution of pure AE. The equilibration (Scheme 11) was then monitored by ³¹P{¹H} spectroscopy and compared with the same process carried out in the absence of Ph₂PCl. In the absence of Ph₂PCl, equilibrium was established in ca. 70 h, whereas in the presence of 5 mol% of Ph₂PCl, equilibrium was established in ca. 7 h (see ESI†).

The key observations on the P,P-metathesis reaction between E₂ and A₂ are that: (1) the solvent (CHCl₃ or CH₂Cl₂) is critical for the reaction to proceed under ambient conditions; (2) the reaction has the characteristics of being radical-initiated; (c) Ar₂PCl catalyses the P,P-metathesis.

It is known that CHCl₃ is a source of Cl˙ radicals (especially under photolysis^56–60 or sonolysis^61–64) and therefore the process shown in Scheme 12 is proposed in which the role of Cl˙ is twofold: (a) to generate Ar₂PCl which promotes the P,P-metathesis process (Scheme 11); (b) to form Ar₂P˙ and initiate a radical chain P,P-metathesis process similar to Scheme 9. DFT studies (see ESI†) show that initiation by attack of Cl˙ on Ar₂P–PAr₂ generating Ar₂P˙ (Scheme 12) is feasible: ΔG = −35.3 kcal mol⁻¹ and ΔG^‡ = 3.4 kcal mol⁻¹ (from the Ph₂P–PPh₂⋯Cl˙ association complex).

Scheme 12

The slow rate of the P,P-metathesis reaction between the o-tolyl diphosphane E₂ and A₂ contrasts with the rapid rate of the P,P-metathesis reaction between the isomeric p-tolyl diphosphane B₂ and A₂. The explanation for the lower rates therefore lies in the proximity of the o-CH₃ group to the P-reaction centre. We have proposed that photolytic Ph₂P˙ radical formation initiates the P,P-metathesis reactions of A₂ (see above) and, since the UV spectrum of E₂ resembles that of A₂ (see Fig. 4), both Ph₂P˙ and (o-Tol)₂P˙ radicals should be similarly accessible. We propose two tentative explanations for the role of the o-CH₃ groups in E₂. (1) The bulky o-tolyl substituents may be shielding E₂ from radical attack by Ph₂P˙ radicals or hindering radical attack by (o-Tol)₂P˙ on A₂. Some support for a steric effect of an ortho-alkyl group on the rate of radical-initiated P,P-metathesis comes from calculations of ΔG^‡ for Ar₂P˙ attack on Ar₂P–PAr₂ where Ar = Ph, o-C₆H₄Me or o-C₆H₄^tBu which gave values of 5.5, 6.7 and 13.3 kcal mol⁻¹ respectively (see ESI† for details). These show that, although the calculated ΔG^‡ values for Ph and o-C₆H₄Me are similar, for o-C₆H₄^tBu, the calculated value is significantly greater. (2) Alternatively, the o-CH₃ groups may provide a means of quenching the radical chain by H˙ intramolecular abstraction from the proximal CH₃ group to produce a benzylic radical by a process that is illustrated in Scheme 13. Calculations show that this would be highly unfavourable energetically (by 14.4 kcal mol⁻¹) for the ground state (o-Tol)₂P˙ radical but the initially formed P-centred diphosphane radical would have sufficient energy for H transfer to take place (see ESI† for details). The energetic C-centred radical could then abstract H˙ and the diphosphane reform thereby quenching the P,P-metathesis.

Scheme 13

Both of these explanations are speculative and further work would be required to establish their veracity.

Conclusions

DFT calculations have shown that the thermodynamic stability of unsymmetrical diphosphanes (Z₂P–PZ′₂), with respect to their symmetrical analogues (Z₂P–PZ₂ and Z′₂P–PZ′₂), should increase with increasing difference in the electronegativity and/or steric bulk of the PZ₂ and PZ′₂ constituents. In these terms, the observed stability of Me₂P–P(CF₃)₂ and ^tBu₂P–PPh₂ can be understood.

The unsymmetrical diphosphanes (Ar₂P–PAr′₂) have been shown to be in finely balanced equilibria with their symmetrical analogues Ar₂P–PAr₂ and Ar′₂P–PAr′₂ which reflects their insensitivity to the aryl substituents employed (Me, MeO, CF₃). Significantly, the P,P-metathesis equilibria with the tetra-aryldiphosphanes were generally established rapidly (<10 min) at ambient temperatures, in a range of solvents, even in dilute solution; furthermore, the rate was increased by photolysis and inhibited by radical scavengers. The experimental and computational evidence points to the involvement of Ar₂P˙ radicals generated photochemically.

Exceptional behaviour has been observed in the P,P-metathesis reactions involving (o-Tol)₂P–P(o-Tol)₂. These were only observed under UV-radiation or when the reactions were carried out in CHCl₃ and CH₂Cl₂. The evidence supporting the involvement of Ar₂PCl in the metatheses with (o-Tol)₂P–P(o-Tol)₂ in chlorinated solvents includes the effect of the presence of H₂O: no P,P-metathesis is observed and the formation of hydrolysis products Ar₂P(=O)H is detected. By contrast, the P,P-metatheses involving other Ar₂P–PAr₂ proceed uninhibited by the presence of water and no Ar₂P(=O)H species are detected.

The facility with which diphosphanes undergo P,P-metathesis is similar to the S,S-metathesis with disulfides and Se,Se-metathesis with diselenides. This augurs well for the application of diphosphanes in dynamic covalent chemistry.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was funded by the Engineering and Physical Sciences Research Council through the EPSRC Centre for Doctoral Training in Advanced Composites for Innovation and Science (grant number EP/L0160208/1). We thank the Department of Chemistry at Oxford and the Centre for Computational Chemistry in Bristol for access to computing facilities.

References

1. S. Ulrich, Acc. Chem. Res., 2019, 52, 510–519.
2. X. Yan, F. Wang, B. Zheng and F. Huang, Chem. Soc. Rev., 2012, 41, 6042–6065.
3. S. Nevejans, N. Ballard, M. Fernández, B. Reck, S. J. García and J. M. Asua, Polymer, 2019, 179, 121670.
4. H. Geng, Y. Wang, Q. Yu, S. Gu, Y. Zhou, W. Xu, X. Zhang and D. Ye, ACS Sustainable Chem. Eng., 2018, 6, 15463–15470.
5. M. Chen, L. Zhou, Y. Wu, X. Zhao and Y. Zhang, ACS Macro Lett., 2019, 8, 255–260.
6. J. Xia, T. Li, C. Lu and H. Xu, Macromolecules, 2018, 51, 7435–7455.
7. J. Liu, Y. Liu, Y. Wang, J. Zhu, J. Yu and Z. Hu, Mater. Today Commun., 2017, 13, 282–289.
8. C. Luo, X. Shi, Z. Lei, C. Zhu, W. Zhang and K. Yu, Polymer, 2018, 153, 43–51.
9. M. Irigoyen, A. Fernández, A. Ruiz, F. Ruipérez and J. M. Matxain, J. Org. Chem., 2019, 84, 4200–4210.
10. J.-C. Lai, L. Li, D.-P. Wang, M.-H. Zhang, S.-R. Mo, X. Wang, K.-Y. Zeng, C.-H. Li, Q. Jiang, X.-Z. You and J.-L. Zuo, Nat. Commun., 2018, 9, 2725.
11. I. Azcune and I. Odriozola, Eur. Polym. J., 2016, 84, 147–160.
12. S. P. Black, J. K. M. Sanders and A. R. Stefankiewicz, Chem. Soc. Rev., 2014, 43, 1861–1872.
13. P. Frei, R. Hevey and B. Ernst, Chem. – Eur. J., 2019, 25, 60–73.
14. R. Martin, A. Rekondo, A. R. De Luzuriaga, P. Casuso, D. Dupin, G. Cabañero, H. J. Grande and I. Odriozola, Smart Mater. Struct., 2016, 25, 084017.
15. Y. Amamoto, H. Otsuka, A. Takahara and K. Matyjaszewski, Adv. Mater., 2012, 24, 3975–3980.
16. U. F. Fritze and M. von Delius, Chem. Commun., 2016, 52, 6363–6366.
17. Y. Chen and R. P. Sijbesma, Macromolecules, 2014, 47, 3797–3805.
18. M. V. Encina, E. Lissi, M. Sarasúa, L. Gargallo and D. Radic, J. Polym. Sci., Polym. Lett. Ed., 1980, 18, 757–760.
19. K. L. Berkowski, S. L. Potisek, C. R. Hickenboth and J. S. Moore, Macromolecules, 2005, 38, 8975–8978.
20. H. Xu, W. Cao and X. Zhang, Acc. Chem. Res., 2013, 46, 1647–1658.
21. C. Liu, J. Xia, S. Ji, Z. Fan and H. Xu, Chem. Commun., 2019, 55, 2813–2816.
22. N. K. Kildahl, J. Chem. Educ., 1995, 72, 423–424.
23. I. V. Koval, Russ. Chem. Rev., 1994, 63, 735–750.
24. S. Burck, D. Gudat and M. Nieger, Angew. Chem., Int. Ed., 2004, 43, 4801–4804.
25. S. Burck, K. Götz, M. Kaupp, M. Nieger, J. Weber, J. S. auf der Günne and D. Gudat, J. Am. Chem. Soc., 2009, 131, 10763–10774.
26. N. Szynkiewicz, Ł. Ponikiewski and R. Grubba, Dalton Trans., 2018, 47, 16885–16894.
27. D. L. Dodds, M. F. Haddow, A. G. Orpen, P. G. Pringle and G. Woodward, Organometallics, 2006, 25, 5937–5945.
28. G. Bettermann, H. Buhl, R. Schmutzler, D. Schomburg and U. Wermuth, Phosphorus Sulfur, 1983, 18, 77–80.
29. N. A. Giffin, A. D. Hendsbee, T. L. Roemmele, M. D. Lumsden, C. C. Pye and J. D. Masuda, Inorg. Chem., 2012, 51, 11837–11850.
30. D. Gudat, Acc. Chem. Res., 2010, 43, 1307–1316.
31. M. Blum, O. Puntigam, S. Plebst, F. Ehret, J. Bender, M. Nieger and D. Gudat, Dalton Trans., 2016, 45, 1987–1997.
32. A. H. Cowley, Chem. Rev., 1965, 65, 617–634.
33. H. Nöth and H.-J. Vetter, Chem. Ber., 1963, 96, 1479–1484.
34. K. Issleib and W. Seidel, Chem. Ber., 1959, 92, 2681–2694.
35. A. K. King, A. Buchard, M. F. Mahon and R. L. Webster, Chem. – Eur. J., 2015, 21, 15960–15963.
36. S. Molitor, J. Becker and V. H. Gessner, J. Am. Chem. Soc., 2014, 136, 15517–15520.
37. M. C. Fermin and D. W. Stephan, J. Am. Chem. Soc., 1995, 117, 12645–12646.
38. V. P. W. Böhm and M. Brookhart, Angew. Chem., Int. Ed., 2001, 40, 4694–4696.
39. R. Waterman, Organometallics, 2007, 26, 2492–2494.
40. G. Fritz, in Advances in Inorganic Chemistry, ed. H. J. Emeleus and A. G. Sharpe, Academic Press, 1987, vol. 31, pp. 171–214.
41. K. O. Feldmann and J. J. Weigand, J. Am. Chem. Soc., 2012, 134, 15443–15456.
42. A. D. Gorman, J. A. Cross, R. A. Doyle, T. R. Leonard, P. G. Pringle and H. A. Sparkes, Eur. J. Inorg. Chem., 2019, 2019, 1633–1639.
43. N. A. Giffin, A. D. Hendsbee and J. D. Masuda, Dalton Trans., 2016, 45, 12636–12638.
44. N. Szynkiewicz, L. Ponikiewski and R. Grubba, Chem. Commun., 2019, 55, 2928–2931.
45. S. Kawaguchi and A. Ogawa, Asian J. Org. Chem., 2019, 8, 1164–1173.
46. R. K. Harris, E. M. Norval and M. Fild, Dalton Trans., 1979, 825–831.
47. A. A. M. Ali and R. K. Harris, Dalton Trans., 1988, 2775–2780.
48. A. Molloy, G. Sánchez-Sanz and D. Gilheany, Inorganics, 2016, 4, 36.
49. D. L. Dodds, J. Floure, M. Garland, M. F. Haddow, T. R. Leonard, C. L. McMullin, A. G. Orpen and P. G. Pringle, Dalton Trans., 2011, 40, 7137–7146.
50. L. R. Avens, L. V. Cribbs and J. L. Mills, Inorg. Chem., 1989, 28, 205–211.
51. T. Heurich, Z.-W. Qu, S. Nožinović, G. Schnakenburg, H. Matsuoka, S. Grimme, O. Schiemann and R. Streubel, Chem. – Eur. J., 2016, 22, 10102–10110.
52. J. M. Matxain, J. M. Asua and F. Ruipérez, Phys. Chem. Chem. Phys., 2016, 18, 1758–1770.
53. S. Nevejans, N. Ballard, J. I. Miranda, B. Reck and J. M. Asua, Phys. Chem. Chem. Phys., 2016, 18, 27577–27583.
54. S. Ji, W. Cao, Y. Yu and H. Xu, Angew. Chem., Int. Ed., 2014, 53, 6781–6785.
55. X. Xu and P. D. Pacey, Phys. Chem. Chem. Phys., 2001, 3, 2836–2844.
56. B. M. Harvey and P. E. Hoggard, Photochem. Photobiol., 2014, 90, 1234–1242.
57. K. Tsujikawa, H. Segawa, K. Kuwayama, T. Yamamuro, T. Kanamori, Y. T. Iwata and T. Ohmori, Forensic Toxicol., 2020, 38, 475–480.
58. K. E. Maudens, S. M. R. Wille and W. E. Lambert, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci., 2007, 848, 384–390.
59. C. Crabb and L. McDonald, US Pat, 3641169A, 1968.
60. S. Kawai, Yakugaku Zasshi, 1966, 86, 1125–1132.
61. H. M. Hung and M. R. Hoffmann, J. Phys. Chem. A, 1999, 103, 2734–2739.
62. V. Sáez, M. D. Esclapez, P. Bonete, D. J. Walton, A. Rehorek, O. Louisnard and J. González-García, Ultrason. Sonochem., 2011, 18, 104–113.
63. B. Park, E. Cho, H. Park and J. Khim, Jpn. J. Appl. Phys., 2011, 50, 07HE10.
64. M. Lim, Y. Son, J. Yang and J. Khim, Jpn. J. Appl. Phys., 2008, 47, 4123.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1dt01013a

---