Transition‐Metal‐Mediated Functionalization of White Phosphorus

Abstract

Recently there has been great interest in the reactivity of transition‐metal (TM) centers towards white phosphorus (P₄). This has ultimately been motivated by a desire to find TM‐mediated alternatives to the current industrial routes used to transform P₄ into myriad useful P‐containing products, which are typically indirect, wasteful, and highly hazardous. Such a TM‐mediated process can be divided into two steps: activation of P₄ to generate a polyphosphorus complex TM‐P_n, and subsequent functionalization of this complex to release the desired phosphorus‐containing product. The former step has by now become well established, allowing the isolation of many different TM‐P_n products. In contrast, productive functionalization of these complexes has proven extremely challenging and has been achieved only in a relative handful of cases. In this review we provide a comprehensive summary of successful TM‐P_n functionalization reactions, where TM‐P_n must be accessible by reaction of a TM precursor with P₄. We hope that this will provide a useful resource for continuing efforts that are working towards this highly challenging goal of modern synthetic chemistry.

The activation of P₄ by transition‐metal complexes has become well established, due to a desire to find new and improved routes for its transformation into useful P‐containing compounds. However, the next necessary step—functionalization of the resulting metal‐bound P_n moieties—has proven extremely challenging and has only occasionally been achieved. Herein, a comprehensive review of these important transformations is provided.

1. Introduction

The element phosphorus is essential for life, serving as a building block for DNA and of the cellular energy carrier ATP in all living organisms. ^[1] In addition, synthetic phosphorus compounds have a huge impact on daily life, ^[1] being found in (among other things) detergents, fertilizers, insecticides, food products and flame retardants. Organophosphorus derivatives in particular play a crucial role in the chemical and pharmaceutical industries. The industrial precursor for these synthetic compounds is white phosphorus (P₄), the most reactive allotrope of the element, which is produced from phosphate rock on a megaton scale annually,[ ² , ³ ] via reaction of the mineral apatite with quartz sand and coke in an electric arc furnace (Scheme 1a). ^[1] While most of the P₄ produced worldwide is re‐oxidized to provide high‐purity phosphate materials, a significant fraction (ca. 18 %) is used to prepare the myriad valuable organophosphorus compounds that are required by modern society. Unfortunately, the synthesis of these target organophosphorus derivatives is a multistep process, almost always involving the initial chlorination of P₄ to PCl₃, followed by subsequent functionalization with Grignard or organolithium reagents (Scheme 1b).[ ² , ³ ] Triphenylphosphane (PPh₃), for example, is one of the most synthetically important organophosphorus compounds, and is prepared industrially by high‐temperature reaction of chlorobenzene with PCl₃ in the presence of molten sodium. ^[4] As well as requiring extremely toxic (Cl₂), corrosive (PCl₃) and pyrophoric (Na) reagents, this route also generates huge amounts of inorganic salt waste (NaCl), which is accumulated as a by‐product. Thus, sustainability and safety concerns each provide a powerful impetus for the urgent overall improvement of such processes. ^[5]

Scheme 1.

Production of white phosphorus (P₄) from the calcium phosphate part of apatite minerals (a) and the synthesis of organophosphorus compounds via PCl₃ (b) or PH₃ (c) (R=organic residue; X=Cl, Br, I; THPC=tetrakishydroxymethylphosphonium chloride).

In some specific cases alternative methods can be used to transform P₄; however, these often suffer from similar problems. For example, alkyl phosphanes and phosphonium salts can be prepared through hydrophosphination of alkenes and ketones (Scheme 1 c), but this requires the intermediacy of extremely toxic PH₃ gas. ^[5] As such, the development of alternative routes that avoid the use of chlorine gas and circumvent the highly toxic intermediates PCl₃ or PH₃ is currently a topic of great interest. ^[6] In pursuit of this challenging goal, much emphasis has been placed on understanding the fundamental reactivity of P₄ towards reactive metal centers. It is hoped that studying these reactions may ultimately pave the way to effective catalytic methods for converting P₄ directly to organophosphorus derivatives.

1.1. Activation of white phosphorus

The controlled and consecutive cleavage of P−P bonds within the P₄ tetrahedron (often referred to as P₄ activation) plays a crucial role in the formation of reactive P_n (n=1–4) units, potentially suitable for further functionalization. As outlined in several reviews, a large number of reactive main group element or transition‐metal compounds has been applied to the activation and degradation of the P₄ molecule.[ ³ , ⁷ , ⁸ ] Scheme 2 illustrates conceivable degradation pathways starting with the initial formation of “butterfly‐P₄” species. The stepwise cleavage of further P−P bonds results in cyclic, branched and linear P_n fragments stabilized by transition metals or main group compounds.

Scheme 2.

A selection of possible reaction pathways for the activation of white phosphorus (P₄). Only the P_n backbones of the fragments are shown; charges and substituents are omitted for clarity.

The transition‐metal‐mediated activation of P₄ in particular has attracted considerable attention over the last several decades, and has given rise to a plethora of fascinating complexes bearing highly versatile P_n units.[ ³ , ⁷ ] A complete description of all the P_n ligands known in the literature would exceed the scope of this review. Nevertheless, an overview of common structural P_n motifs is illustrated in Figure 1. Monophosphido ligands, P₂ dumbbells or cyclo‐P₃ rings can be derived from fragmented P₄ molecules (n≤3). Tetraphosphido ligands (n=4) are typically observed either as intact, metal‐bound P₄ tetrahedra, or as partially degraded “butterfly” species, P₄ rings and chains. Sometimes, the aggregation of multiple phosphorus atoms is also observed (n≥5), which may result in aromatic cyclo‐P₅ and cyclo‐P₆ ligands, or even extended polyphosphorus cages. In addition to the basic structures summarized in Scheme 2, bridging motifs, metal‐phosphorus multiple bonding, and structures with varying hapticity may also be observed.

Figure 1.

Structural motifs of selected transition‐metal complexes bearing P_n ligands; [M]=transition‐metal complex fragment.

2. Transition‐Metal‐Mediated Functionalization of White Phosphorus

If P₄ activation represents the first step of its transformation into potential organophosphorus products, then the second step is functionalization of the resulting P_n moieties through interaction with suitable reagents. However, whereas the activation of P₄ has now been extensively investigated, subsequent functionalization remains far less explored. In the last few years, interest in the transfer and incorporation of P₄‐derived phosphorus atoms into organic or main group substrates has grown substantially, yet the controlled and selective functionalization of activated phosphorus units is still challenging. In this review, we aim to provide a comprehensive overview of these transition‐metal‐mediated functionalizations of white phosphorus. Throughout the review the term “functionalization” will be used to refer to reactions that involve formation of new chemical bonds between phosphorus and a non‐metal or metaloid element (c.f. P₄ “activation”, where the P atoms only form new bonds to the transition metal). Reactions in which P₄ is functionalized by a main group reagent in the absence of a transition metal will not be discussed here, but have been reviewed previously.[ ⁸ , ⁹ ]

The main part of the review is divided into the following sections: section 2.1 discusses the activation and functionalization of P₄ in one step. The following sections discuss the functionalization of transition‐metal coordinated P_n units. The functionalization of P₁, P₂, und P₃ moieties is described in sections 2.2. and 2.3. A large number of publications have focused on complexes with P₄ units, and these results are described in section 2.4. Section 2.5. describes the functionalization of larger P_n units with five or more P atoms. Finally, the functionalization of the P₄ molecules by transition‐metal‐generated p‐block element radicals is described as an alternative for P₄ functionalization in section 2.6.

2.1. One‐Step activation and functionalization

The first transition‐metal‐mediated P₄ functionalization reaction was reported in 1974 by Green and co‐workers. ^[10] They described the reaction of [Cp₂MoH₂] (1) with an excess of P₄ in hot toluene, affording the deep red diphosphene complex [Cp₂Mo(η²‐P₂H₂)] (2), which was crystallographically characterized by Canillo et al. three years later (Scheme 3a). ^[11] This discovery represented a landmark in phosphorus chemistry, since not only P₄ activation, but also functionalization was observed. Specifically, four P−P bonds of the P₄ tetrahedron are cleaved and, simultaneously, new P−H bonds are formed by transfer of the hydride ligands from the metal center to phosphorus.

Scheme 3.

Early transition metallocene‐mediated activation of P₄ with concomitant functionalization.

More than 20 years later, Stephan and co‐workers provided a further example of such a fragmentation/hydrogenation process (Scheme 3 b). Reaction of the tantalocene trihydride complex [Cp₂TaH₃] (3) with P₄ results in the hydridodiphosphene complex [Cp₂Ta(H)(η²‐P₂H₂)] (4). ^[12] Interestingly, Chirik et al. found that an analogous reaction of the related zirconium dihydride complex [Cp*₂ZrH₂] with P₄ proceeds differently, and does not give a diphosphene complex. Instead, only P₄ activation to the [1.1.0]‐tetraphosphabicyclobutane‐1,4‐diyl (“butterfly‐P₄”) complex [Cp*₂Zr(η²‐P₄)] occurs, along with reductive elimination of H₂. ^[13] They further described the treatment of the sterically more encumbered dinuclear complex [Cp*Cp′ZrH₂]₂ (5, Cp′=η⁵‐C₅H₄ tBu) with P₄ (Scheme 3 c). The product molecule [{Cp*Cp′Zr}₂(μ₂,η²:η²‐P₄H₂)] (6) features a bridging P₄ chain best described as a P₄H₂ ⁴⁻ tetraanion. Lappert and co‐workers used the zirconium diphosphido complex [Cp₂Zr(P(SiMe₃)₂)₂] (7) for a related insertion of a rearranged P₄ scaffold into both Zr−P bonds to yield the hexaphosphane‐3,5‐diide complex 8 (Scheme 3 d). ^[14]

A related approach for the one‐step activation and functionalization of P₄ using late transition metals was initially reported by Peruzzini et al. in 1998 (Scheme 4). ^[15] Rhodium(III) and iridium(III) hydride complexes [(triphos)MH₃] (9 a: M=Rh, 9 b: M=Ir; triphos=1,1,1‐tris(diphenylphosphanylmethyl)ethane) enable the direct hydrogenation of P₄ to PH₃, if conducted in a closed system. The stoichiometric by‐products are the highly stable cyclo‐P₃ compounds 10. When carrying out the reaction of 9 a with P₄ at lower temperature, or in an open system, the evolution of dihydrogen gas and an isolable intermediate species [(triphos)Rh(η¹:η²‐HP₄)] (11 a) were observed. Further mechanistic studies performed with the kinetically more stable dihydridoethyl iridium complex [(triphos)IrH₂(Et)] (9 c) revealed the initial formation of a butterfly compound [(triphos)Ir(H)(η²‐P₄)], which slowly isomerizes to [(triphos)Ir(η¹:η²‐HP₄)] (11 b).

Scheme 4.

Functionalization of P₄ mediated by rhodium and iridium triphos complexes. Reaction conditions: (a) for 9 a: THF, 70 °C; for 9 b THF, 120 °C. (b) for 9 a: open system, THF, 70 °C, ‐H₂ or closed system, THF, 40 °C, ‐H₂; for 9 c: open system, THF, reflux, ‐C₂H₆. (c) for 11 a, 11 b: THF, 70 °C; for 11 c, 11 d, 11 e: THF, 60 °C, 20 atm H₂. (d) for 11 a, 11 e: +MeOTf; for 11 c: +HBF₄⋅OMe₂.

Only a year later, Peruzzini et al. successfully extended this concept to P−C bond formation by reporting on analogous hydrocarbon‐substituted tetraphosphido rhodium complexes [(triphos)Rh(η¹:η²‐RP₄)] (11 c: R=Me; 11 d: R=Et, 11 e:=Ph) derived from the corresponding ethylene complexes [(triphos)Rh(R)(η²‐C₂H₄)] (12 a) and P₄ (Scheme 4). ^[16] During the reaction the labile ethylene ligand is released while the alkyl and aryl moieties previously bound to the metal center in 12 a selectively migrate to the activated P₄ scaffold. Notably, the reaction of P₄ with the corresponding hydrido‐ethylene derivative [(triphos)Rh(H)(η²‐C₂H₄)] (12 b) does not afford the expected product 11 a. Instead, the ethylene ligand inserts into the Rh−H bond and successively gives the ethyltetraphosphido species 11 d. Moreover, the pressurization of 11 c, 11 d and 11 e with H₂ at 60 °C induces the formation of 10 a along with the phosphanes RPH₂ in moderate yields. The reactivity of complexes 11 was further explored through reactions with electrophiles. ^[17] The reaction of 11 a and 11 e with MeOTf or MeI gave the doubly functionalized and highly temperature sensitive cations [(triphos)Rh(η¹:η²‐MeRP₄)]⁺ (13 a: R=H, 13 b: R=Ph). The fact that 13 a is also obtained by treating 11 c with HBF₄⋅OMe₂ supports the idea that in this system electrophilic attack generally takes place at the already‐functionalized phosphorus atom.

More recently, a joint study by Power, Wolf and co‐workers demonstrated that the low‐coordinate cobalt‐tin cluster 14 serves as a potent agent for one‐step P₄ activation and functionalization (Scheme 5 a). ^[18] During the reaction, the P₄ tetrahedron is selectively inserted into the rhombohedral Co₂Sn₂ cluster core of 14, during which one of the bulky, tin‐bound terphenyl substituents undergoes a migration to phosphorus, thereby forming a new P−C bond. The product 15 bears a terphenyl‐substituted P₄ chain and represented the first example of a molecular cluster compound containing phosphorus, cobalt and tin.

Scheme 5.

Arylation of P₄ by a low‐coordinate cobalt‐tin cluster (a) and ruthenium‐mediated halogenation (b) and subsequent alkylation of P₄ (c).

An even more recent collaboration by the groups of Caporali and Grützmacher dealt with the chlorination of P₄ by the 16 valence electron species [Cp*RuCl(PCy₃)] (16, Scheme 5b). ^[19] Promoted by two equivalents of 16, migration of two chloride ligands from ruthenium to an activated P₄ unit yields the dinuclear complex [Cp*Ru(PCy₃)(μ₂,η²:η⁴‐P₄Cl₂)RuCp*] (17), containing a planar and unsymmetrically bridging 1,4‐dichlorotetraphosphabutadiene ligand. A selective exchange of the chloro substituents with alkyl groups was achieved by salt metathesis with nBuLi (Scheme 5 c). The product was the tetranuclear compound [(Cp*Ru)₄(μ₃,η²:η²:η⁴‐P₄ nBu₂)₂] (18), which features two coplanar [P₄ nBu₂] moieties.

2.2. Functionalization of P₁ and P₂ ligands

In comparison to the one‐step reactions mentioned above, more versatile transformations can be made feasible by separating P₄ activation from the subsequent functionalization step. The following section deals with functionalizations of P₁ and P₂ ligands derived from P₄, which date back to the pioneering work of Scherer et al. in 1991. Oxidation of the Ni₂WP₂ complex 19 with (Me₃SiO)₂ affords 20 a, the first complex of PO, the heavier congener of the ubiquitous nitric oxide (NO, Scheme 6a). ^[20] Oxidation of 19 with S₈ similarly affords the isoelectronic PS species 20 b, which undergoes partial loss of CO and rearrangement from μ₃:η¹ to μ₂:η² coordination of the PS ligands when heated to 100 °C (21, Scheme 6 b). ^[21] Mays and co‐workers found that a similar μ₃‐PO compound (23 a) is formed if the trinuclear species 22 is exposed to atmospheric oxygen (Scheme 6 c). ^[22] The corresponding oxidation with elemental sulfur is fully reversible and leads to 23 b in the presence of an excess of S₈ in CS₂ (Scheme 6 d). ^[23] The reverse reaction to produce 22 takes place in common organic solvents in the absence of an excess S₈. Scherer et al. also reported on the synthesis of E=P−P=E ligands in 25 by oxidation of the P₂ dumbbells in the Fe₄P₄ cluster 24 with elemental sulfur or grey selenium (Scheme 6 e). ^[24]

Scheme 6.

Oxidation of P₄‐derived P₁ and P₂ ligands in the coordination sphere of transition metals.

A series of remarkable functionalizations was performed by Cummins and co‐workers using early transition metals triply bound to a terminal phosphido (P³⁻) ligand. The molybdenum phosphide 26 was reacted with monomeric acetone peroxide, elemental sulfur and mesityl azide (MesN₃), giving complexes with terminally‐bound phosphorus monoxide (27 a), phosphorus monosulfide (27 b) and iminophosphenium (27 c) ligands (Scheme 7 a).[ ²⁵ , ²⁶ ] In addition, the phosphaalkyne AdC≡P (Ad=1‐adamantyl) adds to the Mo≡P triple bond in 26 to yield the cyclo‐CP₂ complex 28 (Scheme 7 b). ^[27]

Scheme 7.

P₁ functionalization mediated by the molybdenum phosphido complex 26.

Cummins and co‐workers also subsequently demonstrated the impressive synthetic potential of the anionic niobium complex 29, which is isoelectronic with 26 (Scheme 8). The Nb≡P triple bond in 29 participates in further solvent‐dependent P₄ activation. Trapping of 0.5 equivalents of P₄ in weakly coordinating solvents affords [(cyclo‐P₃)Nb(N[Np]Ar)₃]⁻ (Np=CH₂ tBu, Ar=3,5‐Me₂C₆H₃), which is structurally related and isolobal to 28. ^[28] In THF, however, addition of the entire P₄ tetrahedron occurs and concomitant migration of one amide ligand onto phosphorus gives the amino functionalized cyclo‐P₅ anion 30 (Scheme 8 a). Moreover, the anionic nature of 29 opened up avenues to salt metathesis reactions with electrophiles. Treatment of 29 with divalent group 14 element salts at low temperatures provides the dinuclear compounds 31 containing bridging μ₂,η³:η³‐cyclo‐EP₂ (E=Ge, Sn, Pb) triangles (Scheme 8 b). ^[29] The η²‐phosphanyl phosphinidene complexes 32 were accessible by reacting 29 under similar conditions with chlorophosphanes (Scheme 8 c). ^[30] Silylation and stannylation (compounds 33) at the nucleophilic phosphorus atom were achieved by treating 29 with Me₃ECl (E=Si, Sn, Scheme 8 d). Cummins and co‐workers further reported that 29 enables the remarkable transformation of acyl chlorides into the corresponding phosphaalkynes (Scheme 8 e). ^[31] In fact, the authors described a complete synthetic cycle involving an initially formed niobacyclic intermediate 34, [2+2] fragmentation to give the phosphaalkynes P≡C‐R (R=tBu, Ad) and the niobium(V) oxo product 35, and finally the recycling of 35 by step‐wise deoxygenation, P₄ activation and reduction. ^[32] The reaction of 29 with the chloroiminophosphane ClP=NMes* (Mes*=2,4,6‐tBu₃C₆H₂) gives 36, which bears the diphosphorus analogue of an organic azide ligand (P=P=N‐Mes*) coordinating through the P=P unit in an η²‐fashion (Scheme 8 f). ^[33] Remarkably, compound 36 serves as a precursor for the thermal release of formal [P≡P] units, which can be quantitatively trapped by using 1,3‐cyclohexadiene to form the Diels‐Alder adduct 37 via a double diene addition. The by‐product of this process is the niobium imide complex [(Mes*N)Nb([Np]NAr)₃] (38).

Scheme 8.

Phosphorus functionalization mediated by the niobium phosphide anion 29. (i) Recovery of starting material 29 from 35 proceeds by: 1.+Tf₂O in Et₂O at −35 °C; 2.+[Mg(thf)₃(C₁₄H₁₀)]/‐Mg(OTf)₂, ‐C₁₄H₁₀ in THF at −100 °C; 3.+0.25 equiv P₄ in THF at r.t.; 4.+Na‐amalgam/‐Hg in THF at r.t.

Moreover, Cummins and co‐workers also presented additional functionalizations of the terminal PO ligand in the above‐mentioned molybdenum complex 27 a, mediated by addition of more oxophilic metal species. The nucleophilic attack of one methyl group in [Cp₂ZrMe₂] at phosphorus gives the Mo^IV complex 39 in up to 75 % yield (Scheme 9a). ^[26] An uncommon phospha‐Wittig reaction takes place upon treatment of 27 a with the silyl phosphinidene complex 33 a (Scheme 9 b). ^[34] This O=P/Nb=PSiiPr₃ metathesis generates the oxo niobium compound 35 along with the silyl substituted diphosphenido molybdenum complex 40. In solution, 40 decomposes within days to the phosphido molybdenum complex 26 and the unstable diphosphene 41. Elevated temperatures accelerate this decomposition reaction. The reactive intermediate 41 readily oligomerizes to a mixture of the phosphinidene trimer 42 and tetramer 43, or can be trapped with dienes to form the [2+4] cycloaddition products 44.

Scheme 9.

Transformations of the phosphorus monoxide ligand P=O promoted by combinations of two metal complexes (Ar=3,5‐Me₂C₆H₃; R=C(CD₃)₂Me; Np=CH₂ tBu).

In 2000, Scheer and co‐workers described the functionalization of the chromium complex 45 with group 15 halides (Scheme 10 a). ^[35] While reactions with PCl₅ and PCl₃ both lead to the cyclo‐P₃ complex 46 and the dinuclear chromium chloride 47, the reactions with ECl₃ (E=As, Sb) are very unselective. A complex mixture of products is obtained, including 46, 47, the cyclo‐EP₂ complex 48 and various triple‐decker compounds 49. Interestingly, Ruiz and co‐workers found that the closely‐related heavier group 6 complex anions 50 can readily be functionalized with electrophiles affording the methyl‐ or stannyldiphosphenyl bridged species 51 (Scheme 10 b). ^[36]

Scheme 10.

Functionalization of P₂ units mediated by dinuclear group 6 complexes.

2.3. Functionalization of P₃ ligands

The first functionalization of a P₃ ligand was reported by Peruzzini and Stoppioni in 1986. Highly electrophilic alkylating agents MeOTf and [Me₃O]BF₄ were used for the methylation of the cyclo‐P₃ moiety in the group 9 triphos complexes 52 (Scheme 11a). ^[37] The products 53 a contain methyltriphosphirene ligands coordinating in an η³‐mode. It is noteworthy that these alkylations represented the first examples of successful transition‐metal‐mediated functionalization of any polyphosphorus ligand with carbon‐based electrophiles. Four years later, Huttner and co‐workers performed the analogous reaction with [Et₃O][BF₄], which gave the corresponding ethylated complex 53 b. ^[38] The protonation of 52 with HOTf gives a different outcome (Scheme 11 b). ^[39] Spectroscopic and crystallographic investigations indicated that H⁺ interacts weakly with the heteroatomic CoP₃ cluster core in 54 and is most likely located between both phosphorus and cobalt.

Scheme 11.

Reactivity of neutral cyclo‐P₃ complexes with electrophiles (top) and nucleophiles (bottom); [Ni]=[Ni(C₆H₂ tBu₃)] (triphos=1,1,1‐tris(diphenylphosphanylmethyl)ethane).

The reactivity of the cyclo‐P₃ ligand toward main group nucleophiles was explored by Scheer and co‐workers 30 years later using a related nickel complex (Scheme 11 c). ^[40] According to variable temperature ³¹P NMR studies, the reaction of the nickel cyclo‐P₃ sandwich compound 55 with LiPPh₂ initially forms the intermediate triphosphirene species 56 a, which then rapidly incorporates a second equivalent of 55 and concomitantly rearranges to the heptaphosphane compound 57 a. Since crystallization and purification of 57 a was unsuccessful due to its high sensitivity, protonation with HBF₄ was investigated to afford the more stable neutral species 58. The two nickel centers in 58 are bridged by a remarkable bicyclic P₆ ligand with an exocyclic PPh₂ substituent. By contrast, the reaction of 55 with LiNMe₂ gives the η²‐triphosphirene complex 56 b as isolable main product, and 57 b was detected only in minor quantities by ³¹P NMR spectroscopy.

Further η²‐triphosphirene complexes were obtained by Piro and Cummins by reacting the di‐ and trinuclear cyclo‐P₃ complex anions 59 a and 59 b with electrophiles (Scheme 12). ^[27] Treatment of dinuclear 59 a with Ph₃SnCl affords the P‐stannylated compound 60 a, and the reaction of trinuclear 59 b with 1‐adamantanecarbonyl chloride gives the analogous, yet thermally unstable, P‐acyclated species 60 b. When 59 b is reacted with ClP=NMes*, one [W(CO)₅] fragment is lost and a shift of the second [W(CO)₅] moiety to the iminophosphane P is observed. The product 61 features a Mes*NP[W(CO)₅]⁺ unit that circumambulates around the unsaturated triphosphorus cycle in solution at ambient temperature.

Scheme 12.

Functionalization of cyclo‐P₃ units with electrophiles mediated by oligonuclear complex anions (Ar=3,5‐Me₂C₆H₃; Np=CH₂ tBu; Mes*=2,4,6‐tBu₃C₆H₂).

Cummins and co‐workers further demonstrated the exceptional utility of anionic niobium complexes for phosphorus transfer by using the niobate 62 (Scheme 13), which bears phenolato instead of the more established anilido ligands (cf. 29, 59). The reaction of 62 with Ph₃SnCl yields the stannyltriphosphirene complex 63, where the Ph₃Sn⁺ moiety rapidly migrates around the cyclo‐P₃ ring in solution even at −90 °C (Scheme 13 a). ^[41] Subsequent liberation of the triphosphirene molecule from the metal center was achieved by converting 63 with the oxidant pyridine‐N‐oxide in the presence of the trapping agent 1,3‐cyclohexadiene. This procedure gave the uncommon Diels‐Alder adduct 64 along with the niobium oxo dimer [ONb(ODipp)₃]₂. Remarkably, 64 serves as a P₃ ³⁻ synthon and readily transfers its cyclo‐P₃ unit onto [ClRh(PPh₃)₃]. This reaction involves chloride abstraction from rhodium to eliminate Ph₃SnCl, and the release of 1,3‐cyclohexadiene from the diphosphene by [4+2] retrocycloaddition which ultimately affords the cyclo‐P₃ rhodium complex 65.

Scheme 13.

Phosphorus transfer reactions promoted by the anionic niobium cyclo‐P₃ complex 62; [Nb]=[Nb(ODipp)₃] (1,3‐CHD=1,3‐cyclohexadiene).

In a different approach, 62 was reacted with the iodo molybdenum(IV) species [IMo(N[tBu]Ar)₃] (Ar=3,5‐Me₂C₆H₃, Scheme 13 b), which acts as a P⁻ abstractor to form [PMo(N[tBu]Ar)₃] (c.f. 26, Scheme 7). ^[42] In this manner, the dinuclear cyclo‐P₄ cluster 67 was quantitatively obtained, presumably via an irreversible dimerization of an intermediate P₂ species 66. In the presence of the trapping agent 1,3‐cyclohexadiene, an equilibrium with the Diels–Alder product 68 was detected by ³¹P NMR spectroscopy. Complex 68 could not be isolated as a pure compound due to this equilibrium with 66, which irreversibly dimerizes to 67. However, upon addition of the oxidizing agent pyridine‐N‐oxide, liberation of the diphosphene ligand occurs, affording the above‐mentioned double cycloaddition product 37 (c.f. Scheme 8). Cummins and co‐workers also reported on the facile synthesis of the fascinating binary interpnictogen molecules EP₃ (69, E=As, Sb) via salt metathesis reactions of 62 with ECl₃ (Scheme 13 c). ^[43] The niobium dichloride by‐product [(DippO)₃NbCl₂(thf)] can easily be recycled to the cyclo‐P₃ precursor 62 by reduction in the presence of P₄.

2.4. Functionalization of P₄ ligands

As the formation of P₄ ligands is the most common result of transition‐metal‐mediated P₄ activation,[ ³ , ⁷ ] it is not surprising that phosphorus functionalization has mostly been explored for these complexes. These reactions are particularly versatile, depending on the precise nature of the P₄ ligand. Thus, the following section is divided into four parts correlating with the successive degradation of the P₄ tetrahedron: tetrahedral P₄ ligands, [1.1.0]bicyclotetraphosphane‐1,4‐diyl compounds (“butterfly‐P₄” ligands), cyclo‐P₄ units, and catena‐P₄ species.

2.5. Tetrahedral P₄ ligands

Stoppioni, Peruzzini and co‐workers reported on the hydrolytic disproportionation of intact P₄ tetrahedra in the coordination sphere of ruthenium. Such reactivity is remarkable given that free P₄ is well known to be indefinitely stable in water at room temperature. ^[7] The authors found that [CpRu(PPh₃)₂(η¹‐P₄)]⁺ (70 a) almost quantitatively forms the phosphane complex 71 a upon reaction with 100 equiv H₂O (Scheme 14 a). ^[44] By‐products are oxophosphorus species such as phosphorus acid (H₃PO₃) and phosphoric acid (H₃PO₄). Substitution of the triphenylphosphane ligands for bidentate 1,2‐(bisdiphenylphosphanly)ethane (dppe), or the sodium salt of meta‐sulfonated triphenylphosphane (TPPMS=Ph₂P(m‐C₆H₄SO₃Na) (compounds 70 b and 70 c, respectively), resulted in formation of minor quantities of hydroxyphosphane complexes such as 72 b,c and 73 b,c as side‐products. ^[45] The composition of the final mixtures strongly depends on the solvent, the temperature and the amount of H₂O used. When 73 b is dissolved in DCM, it reacts with its own PF₆ ⁻ counter anion and gives the fluorodihydroxyphosphane complex [CpRu(dppe){PF(OH)₂}] PF₂O₂ (74) by F/OH substitution. Lapinte, Peruzzini and co‐workers also examined the reaction of the slightly bulkier complex [Cp*Ru(dppe)(η¹‐P₄)]⁺ (70 d) with an excess of iodine in CHCl₃ at room temperature. ^[46] Three equivalents of PI₃ are released while one molecule of PI₃ remains bound to the metal center to form [Cp*Ru(dppe)(PI₃)]⁺ (75).

Scheme 14.

Hydrolysis and halogenations of P₄ in the coordination sphere of mononuclear (a) and dinuclear (b) ruthenium complexes; [Ru]=[CpRu(PPh₃)₂]; [Ru′]=[CpRu(dppe)] (dppe=1,2‐bis(diphenylphosphanyl)ethane); [Ru′′]=[CpRu(TPPMS)₂] (TPPMS=Ph₂P(m‐C₆H₄SO₃Na)); [Ru′′′]=[Cp*Ru(dppe)].

The dicationic diruthenium complex 76 displays a very complex hydrolysis behavior. When treated with 100 equiv H₂O, in a similar manner to 70 a, a diphosphane complex 77 is obtained along with 71 a, 72 a, 73 a and H₃PO₃ as by‐products (Scheme 14 b). ^[47] With a much higher excess of water (500 equiv), the reaction becomes selective and gives rise to H₃PO₃ and the remarkable 1‐hydroxytriphosphane complex 78 in 93 % isolated yield. ^[48] Reducing the amount of water to only 20 equiv slows down the reaction rate significantly and affords two different compounds as major products: ^[49] the 1,1,4‐tris(hydroxy)tetraphosphane complex 79 and a dinuclear species 80, which contains a bridging diphosphane and a P(OH)₃ ligand. The reaction mixture further contains small amounts of several other species, namely 71 a, 72 a, 73 a, 77 and H₃PO₃. A different reactivity is observed when 76 is first oxidized with iodine in the presence of traces of water. ^[50] The initial product is the monocationic diruthenium complex 81, which is stabilizing a cyclic (P₄H₂I)⁻ anion. In THF, the iodide anion dissociates from the tetraphosphorus ligand, resulting in the ruthenium‐substituted [1.1.0]bicyclotetraphosphane 82 that further hydrolyzes to the triphosphane complex 83 and phosphorous acid.

Krossing comprehensively studied the iodination of the homoleptic silver complex [Ag(η²‐P₄)₂]⁺ (84), which features two intact P₄ tetrahedra bound in an η²‐fashion and is paired with the very weakly coordinating aluminate anion [Al(OR^F)₄]⁻ (R^F=C(CF₃)₃, Scheme 15 a). ^[51] According to low temperature in situ ³¹P NMR experiments, 84 reacts with 3.5 equivalents of iodine even at −78 °C to give the cationic pentaphosphorus cage P₅I₂ ⁺ (85) along with AgI, PI₃ and P₄ as by‐products. Quantum chemical calculations indicated that the formation of 85 may proceed via two different pathways: the insertion of an in situ generated PI₂ ⁺ cation into one P−P bond of white phosphorus, or the intermediate formation of a naked P₅ ⁺ cation, which then reacts with I₂. When the reaction mixture is warmed above −40 °C, 85 rapidly decomposes to the subvalent binary cation P₃I₆ ⁺ (86) and several unidentified by‐products. The Raman and ³¹P NMR spectroscopic data support the intermediate formation of P₂I₄ from the remaining PI₃ and P₄, which reacts with P₅I₂ ⁺ (85) ultimately affording P₃I₆ ⁺ (86) and P₄.

Scheme 15.

Halogenation of P₄ by silver complexes (a,b) and arylation of P₄ with organolithium compounds in the coordination sphere of N‐heterocyclic carbene (NHC) gold cations.

In the same work, a different approach, namely the in situ reaction of [Ag(CH₂Cl₂)][Al(OR^F)₄] (87[Al(OR^F)₄]), P₄ and various halogenating agents was also presented (Scheme 15 b). The lighter congener of 85, P₅Br₂ ⁺ (88), was synthesized almost quantitatively by stirring 87[Al(OR^F)₄] with equimolar amounts of P₄ and PBr₃ at −78 °C for 8 h. An NMR scale reaction of 87[Al(OR^F)₄] with white phosphorus and bromine in CS₂ at room temperature gave several colorless crystals after redissolving the crude mixture in CDCl₃ for an NMR experiment. XRD analysis revealed that these crystals consisted of a dichlorodiorganophosphonium cation [Cl₂P(CDCl₂)₂]⁺ (89), as its [(R^FO)₃Al‐F‐Al(OR^F)₃]⁻ salt. It must be noted that, in this case, the [Al(OR^F)₄]⁻ anion present in the starting material decomposed to the fluoride bridged [(R^FO)₃Al‐F‐Al(OR^F)₃]⁻ anion. 89 is suggested to form via double insertion of a very electrophilic intermediate P⁺ unit into the C‐Cl bond of CDCl₃. Finally, the reaction of 87[Al(OR^F)₄], P₄, and I₂ in a molar ratio of 1:2:3.5 resulted in a mixture containing P₂I₄ and 86[(R^FO)₃Al‐F‐Al(OR^F)₃]. Both were identified by XRD analysis performed on single crystals, which were obtained from concentrated CS₂ extracts of the crude product.

Lammertsma and co‐workers recently reported that the N‐heterocyclic carbene (NHC) gold complex 90 readily reacts with aryl lithium compounds at low temperatures (Scheme 15 c). ^[52] The controlled P−C bond formation and concomitant cleavage of one P−P bond gives rise to the proposed intermediate butterfly species 91. Immediate addition of a second [(NHC)Au] fragment, derived through formal loss of P₄ from a second equivalent of 90, affords the cationic complex 92 in high yield.

2.6. [1.1.0]Bicyclotetraphosphan‐1,4‐diyl ligands (“butterfly‐P₄” ligands)

The first functionalizations of [1.1.0]bicyclotetraphosphan‐1,4‐diyl (“butterfly‐P₄”) ligands were reported by Scherer et al. Thermolysis of the diiron complexes 93 in the presence of diphenylacetylene affords the triphospholyl species 94 a in moderate yields (Scheme 16 a). ^[53] In the case of the related compound containing a sterically more demanding pentaisopropylcyclopentadienyl ligand (93 b), the remarkable P₁₁ cage compound 95 is also formed in small quantities. Scheer and co‐workers later extended this concept to other alkynes. Phenylacetylene gives a mixture of the monophospholyl (96) and 1,2,3‐triphospholyl species (94 b). ^[54] The reaction of 93 a with the phosphaalkyne tBuC≡P produces several compounds (Scheme 16 b). ^[55] While the tetraphospholyl (97) and the 1,2,4‐ triphospholyl (98) complexes are the main products, minor amounts of pentaphosphaferrocene 99 and the dinuclear triphosphaallyl complex 100 can also be isolated. It is proposed that key steps in these reactions are [3+1] fragmentation of the butterfly framework and subsequent addition of one or two equivalents of alkyne.

Scheme 16.

Addition reactions of unsaturated organic molecules to P₄ butterfly complexes; [Ni]=[CpNi(IMes)] (IMes=1,3‐bis(2,4,6‐trimethylphenyl)imidazolin‐2‐ylidene); [Fe]=[Cp′′′Fe].

Wolf and co‐workers found that the N=C and C=S bonds of the heterocumulene phenyl isothiocyanate (PhNCS) reversibly insert into a P−P bond of the P₄ butterfly scaffold of the dinuclear nickel complex 101 (Scheme 16 c). ^[56] The products are the two isomeric bicyclo[3.1.0]heterohexane species 102 and 103, which can be isolated as pure compounds, although they slowly equilibrate with the starting materials 101 and PhNCS in solution.

Protonation of the iron butterfly compound 93 a was also investigated by Scheer and co‐workers. According to ³¹P NMR and computational studies, the acidic proton selectively attacks the more nucleophilic, metal‐bound (“wing tip”) P atom to give the cation 104 (Scheme 17 a). ^[57] A similar observation was made by Lammertsma and co‐workers. ^[58] The reaction of the anionic Lewis acid‐stabilized P₄‐butterfly compound 105 with [Me₃NH][BPh₄] initially forms the intermediate wing tip protonated species 106 (Scheme 17 b). Immediate loss of the amine‐borane adduct Me₃N⋅BAr₃ (Ar=Ph, C₆F₅) leads to the formation of the neutral bicyclo[1.1.0]tetraphosphabutane isomers exo,endo‐107 and exo,exo‐107. The two isomers were calculated to lie close in energy and readily undergo Lewis acid‐catalyzed isomerization. Moreover, they decompose within one day due to a lack of kinetic stabilization.

Scheme 17.

Iron‐mediated protonation of P₄‐butterfly ligands; HX=[(Et₂O)H][BF₄], [(Et₂O)₂H][Al(OC(CF₃)₃]; [Fe]=[Cp′′′Fe(CO)₂]; [Fe′]=[Cp*Fe(CO)₂].

As also reported by Scheer and co‐workers, the reaction of [Cp′′₂Zr(κ² P‐P₄)] (108, Cp′′=1,3‐tBu₂C₅H₃) with the monochlorosilylene 109 in toluene at room temperature gives compounds 110 and 111, which are remarkable phosphorus/silicon analogues of benzene and cyclobutadiene, respectively (Scheme 18 a). ^[59] Computational studies indicated that 110 possesses considerable aromatic character, whereas 111 is weakly antiaromatic. The aromaticity in both compounds is substantially influenced by the additional donating nitrogen lone pairs of the bidentate PhC(NtBu)₂ substituents.

Scheme 18.

Synthesis of phosphorus/silicon analogues of benzene (a) and phosphorus‐rich cage compounds (b) by Zr‐mediated P₄ functionalization (L=[PhC(NtBu)₂], [Zr]=[Cp′′₂Zr], Cp′′=1,3‐tBu₂C₅H₃).

Scheer and co‐workers further reacted 108 with tBuC≡P in boiling toluene to access phosphorus‐rich cage compounds (Scheme 18 b). ^[60] According to ³¹P NMR spectroscopic analysis the major product [Cp′′₂Zr(κ² P‐P₃CtBu)] (112) is formed along with minor amounts of the carbon/phosphorus cages 113 and 114, which both incorporate a cuneane‐like P₅C₃ or P₆C₂ subunit. The authors suggested that 112 is formed by elimination of a P₂ unit from 108 and subsequent reaction with tBuC≡P. The released P₂ species may be trapped by multiple phosphaalkyne molecules to give metal‐free cages such as 113. 114 derives from the formal addition of two equivalents of tBuC≡P to the starting material 108. The products were successfully separated by fractional crystallization and column chromatography.

2.7. cyclo‐P₄ ligands

Functionalization at cyclo‐P₄ ligands is less common than at tetrahedral and butterfly‐P₄ ligands, and was first demonstrated by Scheer and co‐workers (Scheme 19). ^[61] When a tenfold excess of PCl₅ is reacted with the pentanuclear cyclo‐P₄ complex 115, the main products in the reaction mixture are [WCl₄(MeCN)₂], PCl₃, [W(CO)₅(PCl₃)], and the dinuclear tetrachlorodiphosphane complex 116. However, these species are only found in minor quantities when the reaction is performed with a smaller amount of PCl₅ (4 equiv). The main product in this case is 117, in which a [WCl(CO)₂(MeCN)₂] fragment is coordinated by a chlorinated cyclo‐P₄ ligand through its triphosphaallyl subunit. The isolation of 117 as a pure compound was not successful, because it readily reacts with traces of moisture to give the corresponding hydrolysis product 118.

Scheme 19.

Tungsten‐ and iron‐promoted transformations of cyclo‐P₄ ligands; [W]=[W(CO)₅].

Mézailles and co‐workers investigated the reactivity of the end‐deck cyclo‐P₄ iron complex 119 toward electrophiles (Scheme 19 b). ^[62] The P‐borylated Lewis adduct 120 is formed in an equilibrium reaction upon treatment of 119 with B(C₆F₅)₃ in DCM.

Recently, Scheer and co‐workers reported on the N‐heterocyclic carbene‐induced ring contraction of the end‐deck cyclo‐P₄ cobalt sandwich complex 121 (Scheme 20 a). ^[63] The treatment of 121 with two equivalents of 1,3,4,5‐tetramethylimidazol‐2‐ylidene (NHC) leads to selective abstraction of one phosphorus cation from the four‐membered tetraphosphorus ring. The resulting ionic product consists of a [(NHC)₂P]⁺ cation and a [Cp′′′Co(η³‐cyclo‐P₃)]⁻ anion (122, Cp′′′=1,2,4‐tBu₃C₅H₂). This methodology is also applicable to some triple‐decker sandwich complexes bearing cyclo‐P₆ middle decks (see Scheme 26, section 2.5).

Scheme 20.

Functionalization of the cyclo‐P₄ ligands in the cobalt sandwich complex 121; [Co]=[(1,2,4‐tBu₃C₅H₂)Co].

Scheme 26.

Functionalization of the cyclo‐P₆ middle deck in early transition‐metal triple‐decker complexes; [Mo]=[Cp*Mo], [V]=[Cp*V].

The reactivity of 121 towards carbon based nucleophiles was also studied, by the same group. ^[64] Treatment of 121 with tBuLi and LiCH₂SiMe₃ in THF at −80 °C gave the axially substituted tetraphosphido complexes [Cp′′′Co(η³‐P₄R)]⁻ (123, R=tBu, CH₂SiMe₃), respectively as initial kinetic products (Scheme 20 b). The anions 123 are metastable, however they can be sufficiently stabilized by trapping the Li⁺ counter cations with 12‐crown‐4 or [2.2.2]‐cryptand in the cold reaction mixture. By this manner, the authors were able to isolate the [Li(12‐crown‐4)₂]‐salt of 123 a as a mixture with its equatorial isomer 123 a′, and pure 123 b as its [Li([2.2.2]‐cryptand)]‐salt at room temperature. Note that a corresponding equatorial isomer of 123 b was not detected. In the absence of complexing crown ethers, 123 a decomposes upon warming to room temperature to give a mixture of products including the abovementioned cyclo‐P₃ sandwich anion 122, and the dinuclear complex [(Cp′′′Co)₂(μ,η³:η³‐P₈CH₂SiMe₃)]⁻ (124 a), which features an alkyl substituted bicyclo[3.3.0]octaphosphane ligand. Interestingly, a different product mixture is obtained from the decomposition of 123 b. Besides the analogous P₈ tBu complex 124 b and 122, in this case, [Cp′′′Co(η³‐P₅ tBu₂)]⁻ (125) is also formed, which bears a 1,2‐diorgano substituted cyclo‐P₅ ligand.

Scheer and co‐workers also investigated ring expansions of 121 upon reaction with the pnictinidene complexes [Cp*E{W(CO)₅}₂] (E=P, As, Scheme 20 c). ^[65] The insertion of the phosphinidene into the P₄ ring is followed by a shift of one [W(CO)₅] unit and affords the two isomeric η⁴‐cyclo‐P₅ species 126 a and 126 b, which differ only in the orientation of the tungsten pentacarbonyl and the Cp* substituents. Interestingly, when 121 is reacted with the analogous arsinidene, all substituents previously bound to arsenic migrate to phosphorus resulting in several η⁴‐cyclo‐P₄As isomers (127 a‐c), which differ in the location of the arsenic atom within the five‐membered ring.

Very recently, Wolf and co‐workers described a highly selective P₄ functionalization and subsequent fragmentation in the coordination sphere of a cobalt α‐diimine complex (Scheme 21). ^[66] The reaction of the cyclo‐P₄ cobaltate anion [LCo(η⁴‐P₄)]⁻ (128, L=bis(2,6‐diisopropylphenyl)phenanthrene‐9,10‐diimine) with R₂PCl (R=Cy, tBu, Ph, Mes, N(iPr)₂) quantitatively gives the neutral cyclo‐P₅R₂ complexes 129 in up to 77 % isolated yield. Depending on the substituent R, different reaction outcomes are observed upon treatment of 129 with cyanide salts ([K(18‐crown‐6)]CN, [Et₄N]CN, [nBu₄N]CN). When R=Cy, tBu, Ph or N(iPr)₂ reaction with two equivalents of CN⁻ induces a remarkable [3+2] fragmentation, resulting in formation of the anionic cyclotriphosphido cobalt complex 130 and 1‐cyanodiphosphan‐1‐ide anions 131. By contrast, if 129 bears bulky mesityl substituents, the reaction reaches full conversion with only one equivalent of CN⁻. The product in this case is 132, which features a rearranged P₅Mes₂ ligand. The authors suggested that similar cyclotetraphosphido complexes may be key intermediates in the fragmentations that ultimately lead to 130 and 131, but that the bulky mesityl substituent in 129 hinders such onward reactivity.

Scheme 21.

Functionalization of a cyclo‐P₄ ligand with diorganochlorophosphanes R₂PCl (R=Cy, tBu, Ph, Mes, N(iPr)₂) mediated by a low valent α‐diimine cobalt complex, and subsequent rearrangement and fragmentation reactions (L=bis(2,6‐diisopropylphenyl)phenanthrene‐9,10‐diimine).

2.8. P₄ chains

Wolf and co‐workers demonstrated the first functionalization of P₄ chains in the coordination sphere of 3d metalate anions. Reaction of the diiron compound 133 with one equivalent of Et₃N⋅HCl or [H(Et₂O)₂][BAr^F ₄] (Ar^F=3,5‐(CF₃)₂C₆H₃) affords the protonated ferrate 134 in moderate yields (Scheme 22). ^[67] Crystallographic, spectroscopic and computational investigations indicated that the proton is highly mobile, and simultaneously bound to both iron and phosphorus. Treatment of 133 with an excess of Me₃SiCl results in liberation of the phosphorus scaffold from the iron center to give a mixture of PH(SiMe₃)₂, P(SiMe₃)₃ and the nortricyclane compound P₇(SiMe₃)₃ (135) in a ratio of 1:1:10.

Scheme 22.

Functionalization of the bridging P₄ chain in the ferrate 133 with electrophiles (Ar=4‐ethylphenyl; Ar^F=3,5‐(CF₃)₂C₆H₃).

The heterodinuclear complex 136, which features a catena‐P₄ unit, readily undergoes P−P condensation reactions with chlorophosphanes (Scheme 23). ^[68] The reaction with tBuPCl₂ affords a cyclo‐P₅ cobalt complex 137 with a concomitant chloride shift from P to Ga. By contrast, a different outcome is observed upon reaction of 136 with the dialkylmonochlorophosphanes R₂PCl (R=iPr, Cy, tBu). In this case, the N‐heterocyclic gallylene [Ga(nacnac)] (nacnac=CH[CMeN(2,6‐iPr₂C₆H₃)]₂) is released, affording the mononuclear cyclo‐P₅R₂ cobalt complexes 138. Variable temperature NMR studies on the reaction with iPr₂PCl revealed the formation of two intermediate species, likely being constitutional isomers, of which the more abundant could be crystallographically identified as the neutral catena‐P₅ iPr₂ complex 139.

Scheme 23.

Functionalization of the catena‐P₄ unit in the cobaltate complex 136 with chlorophosphanes; [Co]=[CoBIAN] (BIAN=bis(mesityl)iminoacenaphthene); [Ga]=[Ga(nacnac)] (nacnac=CH[CMeN(2,6‐iPr₂C₆H₃)]₂).

2.9. Functionalization of P_n ligands (n≥5)

To date the functionalization of white phosphorus‐derived P_n ligands with n≥5 has been only scarcely explored. Scheer and co‐workers treated the pentaphosphaferrocene [Cp*Fe(η⁵‐P₅)] (140 a) with a set of main group nucleophiles and thus obtained P‐functionalized η⁴‐P₅ ferrate complexes. ^[69] Distinct reactivity was observed depending on the nucleophile. While LiCH₂SiMe₃, LiNMe₂, and LiPH₂ gave rise to mononuclear complexes 141 (Scheme 24 a, minor product for LiPH₂), the formation of dinuclear complexes 142 (Scheme 24 b) occurred upon reaction with NaNH₂ and LiPH₂ (major product). The iodination of 140 a and its heavier congener [Cp*Ru(η⁵‐P₅)] (140 b) was also investigated by the same group (Scheme 24 c). ^[70] Layering dichloromethane solutions of 140 with three equivalents of iodine dissolved in MeCN selectively gives the monocationic complexes 143 as insoluble crystalline solids. Compounds 143 feature a unique tripodal cyclo‐P₃(PI₂)₃ ligand in an all‐cis conformation.

Scheme 24.

Functionalization of a cyclo‐P₅ ligand by nucleophiles (a,b), iodination (c) and thf ring‐opening (d); [Fe]=[Cp*Fe], [Ru]=[Cp*Ru], [Sm]=[L₂Sm], L=N,N′‐bis(2,6‐diisopropylphenyl)formamidinate).

Roesky and co‐workers reported on the Sm^II induced reduction and concomitant functionalization of the cyclo‐P₅ ligand in 140 a (Scheme 24 d). ^[71] The trinuclear complex 144 was obtained in 59 % yield by reacting 140 a with two equivalents of the bulky samarium complex [L₂Sm(thf)₂] (L=N,N′‐bis(2,6‐diisopropylphenyl)formamidinate) in refluxing heptane for seven days. During the reaction the P₅ unit is reduced and ring opening of a coordinated thf molecule occurs. The formation of a new P−C bond ultimately affords an oxidobutyl substituted pentaphosphido ligand bridging the [Cp*Fe], [L₂Sm] and [L₂Sm(thf)] fragments in an η⁴:η³:η¹‐fashion.

Furthermore, the reactivity of 140 a was also investigated toward silylenes in a very recent collaboration between the groups of Scheer and Roesky. ^[72] Treatment of 140 a with one equivalent of the sterically encumbered silylene [LSi{N(SiMe₃)₂}] (145, L=[PhC(NtBu)₂]) affords the Si‐substituted η⁴‐P₅ compound 146 a in 79 % isolated yield (Scheme 25 a). A different reaction outcome is observed when the less bulky silylene [LSi(Cl)] (109, three equivalents) is reacted with 140 a. A simultaneous extrusion and insertion process results in the selective formation of the phosphasilene species 148 and [Cp*Fe(η⁴‐P₄SiL)] (147), which was formally described as a P₄ ⁴⁻ ligand bridging between [Cp*Fe]⁺ and [LSi]³⁺ cations. According to a ³¹P{¹H} VT‐NMR experiment, in this case, 146 b, a cyclo‐P₅ species analogous to 146 a, is formed only as an intermediate at −70 °C. The reaction of 140 a with an equimolar amount of the formal Si^I compound [LSi‐SiL] gives [Cp*Fe{η⁴‐P₅(SiL)₂}] (149) via double ring expansion of the cyclo‐P₅ ring to afford a remarkable seven‐membered cyclo‐Si₂P₅ ring (Scheme 25 b). Considerable quantities of 147 and [L₂Si₂P₂] (111, see Scheme 18, section 2.4.2) were detected as by‐products in the ³¹P{¹H} NMR spectrum of the crude reaction mixture.

Scheme 25.

Functionalization of the cyclo‐P₅ ligand in pentamethylpentaphosphaferrocene 140 a by heavy carbene analogues. [Fe]=[Cp*Fe], L=[PhC(NtBu)₂, nacnac=CH[CMeN(2,6‐iPr₂C₆H₃)]₂), dme=dimethoxyethane, NHC=1,3,4,5‐tetramethylimidazol‐2‐ylidene.

Roesky and co‐workers also studied the reaction of 140 a toward related low valent aluminium compounds. ^[73] Treatment of 140 a with an equimolar amount of [Al(nacnac)] (nacnac=CH[CMeN(2,6‐iPr₂C₆H₃)]₂) at room temperature afforded the triple‐decker type complex [(nacnacAl)(μ₂,η³:η⁴‐P₅)FeCp*] (150 a), which features a bent cyclo‐P₅ middle‐deck (Scheme 25 c). By contrast, reacting 140 a with [(Cp*Al)₄] led to the Al‐Fe cluster compound 151, containing a bridging monophosphido ligand (η³‐P) and a tetraphosphorus chain. A related [4+1] fragmentation of the cyclo‐P₅ ring in 140 a was also observed in the abovementioned reactions with silylenes (c.f. 149). The authors suggest that the reaction proceeds via the intermediate formation of 150 b and subsequent regioselective insertion of [Cp*Al] fragments into two adjacent P−P bonds. In fact, the donor stabilized compound 150 c was detected by NMR spectroscopy in the presence of dimethoxyethane (dme) and 150 d could even be successfully isolated and characterized by trapping with 1,3,4,5‐tetramethylimidazol‐2‐ylidene (NHC).

As mentioned in section 2.4.3 (Scheme 20), Scheer and co‐workers have demonstrated that the N‐heterocyclic carbene 1,3,4,5‐tetramethylimidazol‐2‐ylidene (NHC) is a valuable reagent for the contraction of a cyclo‐P₄ ring. ^[63] In the same work, this concept was also applied to the early transition‐metal triple‐decker sandwich complexes [(Cp*M)₂(μ,η⁶:η⁶‐P₆)] (152, M=V, Mo, Scheme 26). Thus, the reaction of the dimolybdenum species 152 a with two equivalents of NHC selectively extracts a phosphorus cation from the cyclo‐P₆ middle deck, affording the bis(NHC)‐supported P^I cation [(NHC)₂P]⁺ and the molybdate anion 153. According to the crystallographic data, the anticipated cyclo‐P₅ ring in 153 is best described as separated P₃ and P₂ units. The reaction of the divanadium complex 152 b with NHC is less selective. The two ionic species [(NHC)₂P][(Cp*V)₂(μ,η⁶:η⁶‐P₆)] (154) and [(NHC)₂P][(Cp*V)₂(μ,η⁵:η⁵‐P₅)] (155) were isolated as a co‐crystalline mixture from a THF extract. While 154 is probably formed by one electron reduction of the starting material 152 b, 155 derives from phosphorus cation abstraction from 152 b. The third identified product was the neutral complex 156, which features two bridging ligands: a triphosphaallylic P₃ chain and an NHC‐P phosphinidenide unit.

Separately, Cummins and co‐workers found that the diniobium octaphosphide complex 157 possesses a reactive phosphinidene moiety, which readily hydrolyzes to give the mononuclear phosphanyl‐substituted heptaphosphide complex 158 and the oxo niobium species 159 a (Scheme 27). ^[74] The former compound could also be synthesized in a more selective manner by protonation of 157 with two equivalents of 2,6‐dimethylpyridinium iodide. In this case, the corresponding niobium diiodide complex is the stoichiometric by‐product. Remarkably, 157 also undergoes Nb=P/O=C metathesis with ketones. ^[75] While the resulting alkyl substituted phosphaalkene complexes 160 are stable for up to several days at ambient temperature, the corresponding aryl derivatives immediately undergo an electrocyclic rearrangement ultimately affording the saturated organophosphorus cluster compounds 161. The carbophosphorus cluster can be liberated from niobium by treatment with two equivalents of pyridine‐N‐oxide in the presence of an excess of 1,3‐cyclohexadiene, leading to the Diels–Alder product 162. ^[76] Moreover, [4+2] cycloaddition with the niobium‐bound diphosphene moiety in 161 also takes place when 2,3‐dimethylbutadiene or spiro[2.4]hepta‐4,6‐diene are used as dienes, and the niobium oxo compounds 159 a and 159 b can be recycled by step‐wise deoxygenation, reduction and P₄ activation. Furthermore, Cummins and co‐workers described the remarkable reactivity of 162 towards ten equivalents of iodine, which cleanly affords four molecules of PI₃ along with the bis(diiodophosphanyl)‐substituted hydrocarbons 163 and 164.

Scheme 27.

Niobium‐mediated functionalization of a P₈ framework; [Nb]=[Nb(OR′)₃] (OR′=(adamantane‐2‐ylidene)(mesityl)methanolate); Ar=Ph, 4‐Cl‐C₆H₄, 4‐Me‐C₆H₄, 4‐OMe‐C₆H₄, 4‐NMe₂‐C₆H₄, 4‐CF₃‐C₆H₄). (i) Recovery of starting material 157 (0.5 equiv) proceeds by: 1. +O(OCCF₃)₂, +2 equiv Me₃SiI/‐Me₃SiO(OCCF₃) in Et₂O; 2. +SmI₂/‐SmI₃ in THF; 3. +P₄ in toluene.

2.10. Radical functionalization of P₄

While the reactions described thus far have all proceeded through direct coordination of P_n fragments, it has also proven possible for transition metals to mediate the functionalization of P₄ in an outer sphere manner, not necessarily involving the formation of intermediate (poly)phosphorus complexes. In particular, it has been found that transition metals can be used to induce the formation of free carbon‐ (or other main group element‐) centered radicals, which induce successive, homolytic P−P bond cleavage reactions, resulting in stepwise degradation of the P₄ molecule. A related, metal‐free concept was originally demonstrated by Barton and co‐workers, using alkyl radicals generated by the decomposition of pyridine thione oxycarbonyl esters (Barton's PTOC esters). ^[77] Cummins and co‐workers used the three‐coordinate Ti^III complex [Ti(N[tBu]Ar)₃] (165, Ar=3,5‐C₆H₃Me₂) for stoichiometric halogen radical abstraction from main group element halides RX (RX=PhBr, CyBr, Me₃SiI, Ph₃SnCl) to give the Ti^IV species 166 (Scheme 28 a). ^[78] The concomitantly‐formed R^. radicals successively break down the P₄ tetrahedron, ultimately affording the respective phosphanes PR₃ in certain cases. While quantitative conversion is observed for the heavier group 14 element halides Me₃SiI and Ph₃SnCl, considerable amounts of the diphosphanes P₂R₄ are found as by‐products in the analogous reactions with CyBr and PhBr. However, with an excess (5 equiv) of 165 and RX, these reactions also become quantitative. When more sterically demanding aryl groups such as Mes and Ar* (2,6‐Mes₂C₆H₃) are used, the stepwise P₄ degradation does not proceed to completion, but instead results in the triphosphirane P₃Mes₃ or the bicyclo[1.1.0]tetraphosphabutane (butterfly) species exo,endo‐Ar*₂P₄, respectively.

Scheme 28.

Radical functionalization of P₄ mediated by early (a) and late (b) transition metals. (i) only for Cp^BIG; (ii) MCp^R=NaCp^BIG, NaCp′′′, LiCp*, NaCp^iPr4; (iii) only for MCp^R=NaCp′′′, LiCp* (Cp^BIG=C₅(4‐nBu‐C₆H₄)₅; Cp′′′=C₅H₂ tBu₃; Cp*=C₅Me₅; Cp^iPr4=C₅HiPr₄).

Scheer reported on the synthesis of the organic exo,exo‐substituted P₄ butterfly compounds 167 by the one‐pot reactions of P₄ with metal‐generated cyclopentadienyl radicals (Scheme 28 b). ^[79] The required radicals were formed via three different pathways: (i) The treatment of NaCp^BIG (Cp^BIG=C₅(4‐nBu‐C₆H₄)₅) with CuBr led to precipitation of metallic copper along with dark blue {Cp^BIG}^. radicals, which were detected by EPR spectroscopy and selectively give 167 a upon addition of P₄. (ii) The reaction of the alkali cyclopentadienide salts MCp^R (=NaCp^BIG, NaCp′′′, LiCp*, NaCp^iPr4) with FeBr₃ afforded the intermediate Fe^III complexes 168, which readily transfer Cp^R. radicals onto the P₄ tetrahedron, giving 167 upon loss of FeBr₂. (iii) The corresponding Fe^II complexes 169 synthesized from MCp^R (=NaCp′′′, LiCp*) and FeBr₂ resulted in the P₄ butterfly species 167 b when reacted with P₄. The reaction mechanism in this case is suggested to involve disproportionation of 169 into the Fe^III complexes 168 b, which undergo the abovementioned reaction with P₄, and Fe^I intermediates 170 that form the dinuclear species 171 bearing a bridging catena‐P₄ ligand.

Furthermore, Yakhvarov and Budnikova have reported on electrochemical methods for radical functionalization of P₄.[ ⁸⁰ , ⁸¹ ] Their work has recently been reviewed in detail.[ ⁸² , ⁸³ ] The reaction principle is based on the electrocatalytic C−P bond formation mediated by bipyridine (bpy) nickel complexes. Figure 2 exemplifies a suggested schematic catalytic cycle for the nickel‐promoted transformation of P₄ into organophosphorus compounds. The proposed mechanism involves the cathodic electrogeneration of active Ni⁰ complexes from the corresponding Ni^II species, followed by the oxidative addition of aryl halides ArX. ^[81] The resulting organonickel aryl complex [NiX(Ar)(bpy)] is inert towards P₄. However, after electrochemical one‐electron reduction, the radical species [Ni(Ar)(bpy)]^. immediately incorporates the P₄ molecule. ^[83] Subsequent aqueous work‐up ultimately affords tertiary phosphanes and phosphane oxides. Note that the metal ions generated from the electrochemically soluble (sacrificial) anode are required to stabilize anionic phosphido intermediates and thus prevent undesired phosphorus polymerization processes. Depending on the anode material, different organophosphorus products are formed. ^[84] While a zinc anode mainly leads to the formation of tertiary phosphanes, an aluminium anode instead results in phosphane oxide formation. By contrast, use of a magnesium anode gives cyclic polyphosphorus compounds, such as (PhP)₅.

Figure 2.

Proposed mechanism for nickel‐electrocatalyzed arylation of white phosphorus in DMF or MeCN carried out in an undivided cell (bpy=2,2′‐bipyridine).

Very recently, Wolf and co‐workers described the direct and photocatalytic synthesis of triarylphosphanes and tetraarylphosphonium salts from P₄, aryl iodides, and triethylamine as a terminal electron donor (Scheme 29). ^[85] Based on the phosphorus atoms (0.25 equiv P₄), the reaction uses an excess of aryl iodide (11 equiv) and Et₃N (14.4 equiv) and is catalyzed by the commercially available photocatalyst [Ir(dtbbpy)(ppy)₂][PF₆] (172[PF₆], 2.2 mol %, dtbbpy=4,4′‐di‐tert‐butyl‐2,2′‐bipyridine, ppy=2‐(2‐pyridyl)phenyl) under blue LED light (455 nm). Depending on the steric and electronic nature of the aryl iodide substrate the formation of either the triarylphosphanes or tetraarylphosphonium salts is favored (Scheme 29). The resulting organophosphorus compounds can be isolated in up to 71 % yield. Both electron‐withdrawing groups and additional steric bulk at the ortho‐position support the formation of phosphanes over phosphonium salts. Further increase of the steric demand of the substrate gives even less substituted phosphanes: the secondary phosphane Mes₂PH is formed from mesityl iodide, and use of 2,6‐dimesitylphenyl iodide (Ar*I) gives rise to the primary phosphane Ar*PH₂. These observations are in line with mechanistic studies performed on the reaction using iodobenzene, which support a stepwise mechanism that sequentially produces primary phosphanes (PhPH₂), secondary phosphanes (Ph₂PH), tertiary phosphanes (Ph₃P) and finally the phosphonium salts (Ph₄P⁺). According to emission quenching experiments and redox potential measurements it is likely that the excited state of the photocatalyst 172 ⁺ is reductively quenched by Et₃N, generating the neutral complex 172, which in turn reduces PhI to the corresponding phenyl radical Ph^⋅. White phosphorus is then rapidly consumed by these radicals. As well as aryl‐substituted products, the same methodology could also be used to prepare the tin‐substituted phosphane P(SnPh₃)₃, starting from Ph₃SnCl.

Scheme 29.

Photocatalytic functionalization of P₄ to triarylphosphanes and tetraarylphosphonium salts.

3. Summary and Outlook

Building on the pioneering work of Green, Stoppioni and Peruzzini in the 1970s and 1980s, the transition‐metal‐mediated functionalization of white phosphorus has developed greatly over the past several decades. Scientists from around the globe have contributed to this branch of phosphorus chemistry, and demonstrated the synthetic potential of P₄ functionalization for the formation of diverse and unprecedented phosphorus compounds. A considerable number of transition‐metal complexes bearing P_n ligands derived from P₄ activation have been used for subsequent P₄ functionalization (sections 2.2–2.5). By contrast, only a few hydrido or alkyl complexes have shown the potential to both activate and functionalize P₄ in a single reaction (section 2.1), and even fewer complexes are currently capable of promoting outer sphere, radical functionalization (section 2.6). To date, neutral complexes have been employed more often for P₄ functionalization than ionic ones. However, out of the charged systems, anionic complexes have generally proven to be better platforms than cations. This may be attributed to the fact that by far the most common reactants for P₄ functionalizations are electrophiles. Attack at nucleophilic phosphorus sites is often accompanied by metathetical halide abstraction, which provides the driving force for these reactions. Hydrolyses, oxidations, cycloadditions, and reactions with nucleophiles have been reported much less frequently. Many of these functionalizations have given rise to remarkable new mono‐ or oligophosphorus complexes. However, the liberation of these P‐rich species from the complexing metal centers is challenging and thus far has seldom been achieved. Nevertheless, some fascinating compounds, such as EP₃ (E=As, Sb) prepared by Cummins and the P‐Si analogue of benzene reported by Scheer have been synthesized via these approaches. It is worth noting that, while P₄ functionalization at both early and late transition metals has been established to similar extents, release from the metal has mostly been observed at early transition‐metal systems. This is probably due to the higher oxophilicity of these metals, which can be exploited in ligand liberation reactions with oxidizing agents (e.g. pyridine‐N‐oxide).

Despite the growing number of successful phosphorus functionalization reactions, the ultimate goal, namely the general circumvention of chlorine gas and PCl₃ in the industrial formation of useful organophosphorus species, is still far from being reached. Nevertheless, we hope that careful evaluation of the above‐mentioned literature may help chemists to begin to predict the outcome of their prospective functionalization reactions, and hence further accelerate the progress being made in this fundamental area of modern phosphorus chemistry.

Conflict of interest

The authors declare no conflict of interest.

Biographical Information

Christian M. Hoidn was born in Zwiesel, Germany, in 1991 and studied chemistry at the University of Regensburg, Germany, where he received his bachelor's degree in 2013 and his M.Sc. diploma in 2015. Recently, he completed his doctoral studies under the supervision of Prof. Dr. Robert Wolf, which focused on the synthesis and characterization of low‐valent 3d metal complexes and their application for the activation and functionalization of white phosphorus. He was awarded with a Ph.D. scholarship of the Foundation of German Business (Stiftung der Deutschen Wirtschaft, sdw).

Biographical Information

Daniel Scott earned his PhD from Imperial College London under the supervision of Dr. Andrew Ashley and Prof. Matthew Fuchter. He subsequently completed an EPSRC doctoral prize fellowship at the same institution, and is currently an Alexander von Humboldt fellow working at the University of Regensburg within the research group of Prof. Dr. Robert Wolf. His research interests revolve around the activation and functionalization of small molecules, mediated by both main group systems and low‐valent transition‐metal complexes.

Biographical Information

Robert Wolf started his research career at the University of Cambridge, UK, under the guidance of Dominic S. Wright. He was awarded a PhD from Leipzig University for work in phosphorus chemistry supervised by Evamarie Hey‐Hawkins. After postdoctoral research with Philip P. Power (UC Davis, USA) and Koop Lammertsma (VU Amsterdam, The Netherlands), he started his independent career at the University of Münster (mentor: Werner Uhl). He became Professor of Inorganic Chemistry at the University of Regensburg in 2011. In 2017, he received an ERC Consolidator Grant for the development of new methods for the functionalization of white phosphorus.

C. M. Hoidn, D. J. Scott, R. Wolf, Chem. Eur. J. 2021, 27, 1886.