Synthesis of acyl(chloro)phosphines enabled by phosphinidene transfer

Acyl(chloro)phosphines RC(O)P(Cl)(t-Bu) have been prepared by formal insertion of tert-butyl phosphinidene (t-Bu–P) from t-BuPA (A = C14H10 or anthracene) into the C–Cl bond of acyl chlorides.

Acyl(chloro)phosphines RC(O)P(Cl)(t-Bu) have been prepared by formal insertion of tert-butyl phosphinidene (t-Bu-P) from t-BuPA (A ¼ C 14 H 10 or anthracene) into the C-Cl bond of acyl chlorides.
We show that the under-explored acyl(chloro)phosphine functional group provides an efficient method to prepare bis(acyl)phosphines, which are important precursors to compounds used industrially as radical polymerization initiators. Experimental and computational investigations into the mechanism of formation of acyl(chloro)phosphines by our synthetic method reveal a pathway in which chloride attacks a phosphonium intermediate and leads to the reductive loss of anthracene from the phosphorus center in a P(V) to P(III) process. The synthetic applicability of the acyl(chloro)phosphine functional group has been demonstrated by reduction to an acylphosphide anion, which can in turn be treated with an acyl chloride to furnish dissymmetric bis(acyl)phosphines.
Treatment of t-BuPA with benzoyl chloride (PhC(O)Cl) in dichloromethane at 23 C (Scheme 1) led to an instant colour change from colourless to bright yellow, and the quantitative formation of 1 as a yellow powder, isolated in 82% yield. The species exhibited a singlet resonance at d 114 ppm in the 31  Collectively, these data support the formulation of 1 as (tertbutylchlorophosphanyl)(phenyl)methanone, PCl(t-Bu)(C(O)Ph).
As 1 could not be isolated as single crystals suitable for structural analysis, a transition metal complex of this phosphine was prepared. Treatment of 1 with half an equivalent of [Ru(p-cymene)Cl 2 ] 2 in tetrahydrofuran led to the formation of coordination complex 2. Layering the reaction mixture with pentane at 23 C gave 2 as a red-orange crystalline material in 77% yield. Multinuclear NMR spectroscopy studies, in addition to a single crystal X-ray diffraction study conrmed the formulation of 2 as Ru(p-cymene)Cl 2 (PCl(t-Bu)(C(O)Ph)) ( Fig. 1). Racemic 2 crystallizes in the P2 1 /c space group, and the solid state structure showed C-O, P-C, and P-Ru bond lengths of 1.221(4), 1.875(4) and 2.3178(8)Å respectively, which are in line with the expected values. 38 Ru-P binding of acylphosphines to ruthenium has been observed in the coordination chemistry of related P(III) ligands. 39,40 Previous studies on thermal phosphinidene transfer have established that thermal loss of a phosphinidene does not occur when the R-group of an RPA compound is not p-donating. 35 Therefore, the mechanism of acylphosphine formation is more likely to proceed associatively via nucleophilic attack by phosphorus at the electrophilic carbonyl group of benzoyl chloride. We envisioned that replacement of the chloride for a less nucleophilic anion may lead to the isolation of the putative cationic acylphosphonium intermediate. To this end, the reaction of equimolar solutions of t-BuPA and benzoyl triate ( (Scheme 1) as a white powder in 75% yield. A singlet resonance is observed at d 127 ppm in the 31  Compound 3[OTf] decomposes slowly in solution at ambient temperature, yielding an anthracenyl(acyl)hydridophosphonium triate salt as the major product (see ESI †). Similar C-H activation of arenes by phosphorus-containing electrophiles has been described for metal-stabilized phosphenium triates. [42][43][44] Despite its instability in solution, 3[OTf] can be stored in the solid state at À35 C for weeks without decomposition and puried by recrystallization from saturated dichloromethane/ pentane solutions, also at À35 C. Treatment of 3[OTf] with a chloride source (tetra-n-butylammonium chloride) in chloroform at 23 C quantitatively generated 1, [TBA][OTf] and anthracene (Scheme 1). This reactivity is consistent with nucleophilic attack of the phosphonium center in 3[OTf] by chloride, inducing reductive loss of anthracene.
The mechanism leading to the formation of 1 was probed via competition experiments involving the addition of two different para-substituted benzoyl chloride substrates in a 1 : 1 ratio to t-BuPA. The product ratio was determined by quantitative 31 P{ 1 H} NMR spectroscopy and a series of such experiments is summarised in the Hammett plot (Fig. 2). A small negative value for r of À2.36 is consistent with build-up of positive charge in the transition state. 45 This is particularly evident for electron withdrawing groups, such as CF 3 , where the reaction is accelerated compared to electron rich para-substituted benzoyl chlorides. These electronic effects are further evidence of phosphine attack at the electrophilic carbon atom of the acylphosphine functional group in the rate limiting step of the reaction.
Quantum chemical calculations were employed to study the formation of 1. These were performed using ORCA 4.0.0.2 46 at the B3LYP/Def2-TZVP level of theory and the results are summarised in Fig. 3. According to our calculations, initial nucleophilic attack of t-BuPA on benzoyl chloride proceeds with an activation barrier of +17.9 kcal mol À1 TS1, in which the phosphorus atom occupies a tetrahedral environment, leads to an intermediate comprising a phosphonium and chloride ion-pair,  I1. The distance between the phosphorus and chlorine atoms is 2.55Å which is signicantly longer than the sum of the singlebond covalent radii, 2.10Å. 38 However, the phosphorus center features a trigonal bipyramidal structure instead of the ideal tetragonal coordination sphere of a 4-coordinate phosphonium species, suggestive of a strong interaction between the phosphonium centre and chloride. A second transition state in the reaction pathway TS2 was located, which corresponds to formation of a P-Cl bond and concomitant loss of anthracene to give 1. TS2 had a relative Gibbs free energy of +15.6 kcal mol À1 , which is lower than the Gibbs free energy for TS1. The loss of anthracene in TS2 is highly concerted and appears to proceed by a cheletropic extrusion pathway. This type of formal reductive elimination from l-5 phosphoranes to give a phosphorus(III) product has previously been described in the literature. [47][48][49][50] Finally, we investigated the possibility of a phosphenium chloride salt intermediate (I2, see ESI †), which would arise from loss of anthracene from cation 3 and is the formal reverse of the McCormack reaction. [49][50][51] We found the relative Gibbs free energy of intermediate I2 to be 65.0 kcal mol À1 which is signicantly higher than either TS1 of TS2, ostensibly ruling it out as an intermediate in the reaction mechanism (see ESI †).
The higher Gibbs free energy of the rst transition state compared to the second is consistent with experiment, where a phosphonium intermediate is not observed when benzoyl chloride is used as the acylating reagent. Thus it is indicated that, when switching from chloride to triate, there is a change in the rate determining step from TS1 to TS2, which are in proximity (17.9 and 15.6 kcal mol À1 , respectively) as determined by DFT calculations.
Efforts to further derivatise 1 were undertaken by reduction of the P-Cl bond. Treatment of compound 1 with 2 equivalents of freshly-prepared sodium naphthalenide 52 led to the formation of a new species with concomitant formation of naphthalene and sodium chloride (Scheme 2). Addition of 15-crown-5 to the reaction mixture led to the isolation of [Na(15-crown-5)]4 as a crystalline orange powder in 49% yield. The connectivity of anion 4 as [OC(Ph)Pt-Bu] À was conrmed by a single crystal Xray diffraction experiment (Fig. 4). In the solid state, the P-C and C-O bond lengths are 1.756(2) and 1.271(2)Å, respectively.
The short C-P bond and long C-O bond are consistent with delocalization of electron density from a phosphide lone pair into the p*(C]O) orbital.
To quantify the degree of electron delocalization over the acylphosphide functional group, the electronic structure was evaluated using natural bond orbital (NBO) methods. 53 Natural resonance theory (NRT) analysis, implemented within the NBO package, showed that the primary resonance structure (4a, Fig. 5) had a contribution of 39%. A second resonance structure, containing a carbon-phosphorus double bond and carbonoxygen single bond (4b, Fig. 5), closely followed with a contribution of 33%, while the remaining resonance structures each had less than 2% weight. The C-O and C-P natural bond orders were remarkably similar at 1.51 and 1.49, respectively. The high degree of delocalization of the phosphide lone pair into the carbonyl unit seems to be a result of the anionic charge and of having two lone pairs at P, one of which is higher in energy and has more p-character and is thus more prone to delocalization, since for neutral acylphosphine-containing molecules, minimal delocalization has been observed. 54,55 The combination of [Na(15-crown-5)]4 and AdC(O)Cl (Ad ¼ 1adamantyl) in tetrahydrofuran generated 5, which was isolated as a crystalline yellow solid in 55% yield (Scheme 2). In the 31 P { 1 H} NMR spectrum of 5, a singlet resonance is observed at d 37 ppm. Two diagnostic doublet resonances are observed in the downeld region of the 13 C{ 1 H} NMR spectrum at d 225 (J PC ¼ 50 Hz) and 214 ppm (J PC ¼ 46 Hz), which are assigned to the two acyl carbon atoms. The IR spectrum of 5 shows two C]O absorption peaks at 1665 and 1629 cm À1 . These spectroscopic data inform our formulation of 5 as OC(Ph)P(C(O)Ad)t-Bu. Single crystals of 5 were grown from a concentrated pentane solution and analyzed by an X-ray diffraction study, which featured 5 as a single (R)-enantiomer (see ESI †) in the chiral space group P2 1 (Fig. 6).
The formation of 5 from [Na(15-crown-5)]4 provides an approach for preparing dissymmetric bis(acyl)phosphines. 56 Interestingly, a previous report of a (W(CO) 5 ) P-protected acylphosphide anion describes C-O instead of C-P bond formation upon treatment with an acyl chloride, to give an ester-containing phosphaalkene. 57 In summary, we have described a novel reaction of t-BuPA with acyl chlorides to give acyl(chloro)phosphine-containing products, which were not accessible previously. 36,37 Experimental mechanistic studies agree well with computational insights, and clearly demonstrate the presence of an intermediate acylphosphonium species. The acyl(chloro)phosphine functional group can be reduced to the corresponding acylphosphide anion, which reacts further with an acyl chloride to provide access to a dissymmetric bis(acyl)phosphine. Other synthetic pathways to related acylphosphine derivatives are the subject of on-going investigations.

Conflicts of interest
There are no conicts to declare.