Formation, Reactivity and Decomposition of Aryl Phospha‐Enolates

Abstract Two lithium phospha‐enolates [RP=C(Si i Pr3)OLi]2 were prepared by reaction of triisopropyl silyl phosphaethynolate, i Pr3SiPCO, with aryl lithium reagents LiR (R=Mes: 1,3,5‐trimethyl phenyl; or Mes*: 1,3,5,‐tri‐tertbutyl phenyl). Monomer/dimer aggregation of the enolates can be modulated by addition of 12‐crown‐4. Substitution of lithium for a heavier alkali metal was achieved through initial formation of a silyl enol ether, followed by reaction with KO t Bu to form the corresponding potassium phospha‐enolate [MesP=C(Si i Pr3)OK]2. On addition of water, the enolates are protonated to afford RP=C(Si i Pr3)(OH). For the sterically less demanding system (R=Mes), this phospha‐enol rapidly tautomerises to the corresponding acyl phosphine MesP(H)C(Si i Pr3)(O), which on heating extrudes CO. In contrast, bulkier phospha‐enol (R=Mes*) is stable to rearrangement at room temperature and thermally decomposes to RH and i Pr3SiPCO.


Introduction
Lithium enolates are integral synthetic intermediates with wide ranging applications. [1] A key characteristic that drives the diverse reactivity of the enolate functionality is 1,3-delocalisation of negative charge between the β-carbanion and the adjacent carbonyl. Modification of this well-established functional group by replacing this CÀ H group with a diagonallyrelated phosphorus atom, to create a phospha-enolate, has the potential to disrupt this intramolecular interaction. This modification can significantly alter resonance stabilization and hence, reactivity.
Initial steps to alkali-metal phospha-enolates included the addition of lithium phosphide to phenyl acyl chloride to form a 2-phospha-1,3-dionate A, in a complex sequence of steps relying on the tautomerisation of a bis-acyl phosphine (Figure 1). [2] Dimeric in the solid state, the three central heterocycles are not co-planar, indicating no delocalisation throughout the molecule. Dehalogenation of a mono-acyl phosphine produced a sodium phospha-enolate B, which was shown to react as a phosphide despite the fact that the negative charge is distributed equally over the oxygen and phosphorus atoms. [3] A similar synthetic approach resulted in the formation of phospha-enolate C. [4] The reactivity of this tungsten-supported example varies depending on the nature of the added substrate; MeI gives rise to phosphorus-centered alkylation, where Me 3 SiCl results in silylation of the oxygen atom.
The phosphaethynolate anion (PCO À ) [5] has been shown to act an effective chemical precursor to a number of phosphoruscontaining compounds including phosphinecarboxamides (H 2 PC(O)NR(H)), [6] and more recently, metal cyaphido-complexes ([M]À C � P). [7,8] It can also be employed as a precursor to phospha-enolates. The addition of an organo-lithium to a borylsubstituted phosphaethynolate produced a structurally authenticated lithium phospha-enolate D, [9] a reaction that includes a migration of the supporting boryl group from the oxygen to the carbon atom.
Here, we present an extension of this reactivity to the reaction of a silyl phosphaethynolate with aryl lithium reagents to form silyl-substituted lithium phospha-enolates. Furthermore, we explore the fundamental reactivity of these phosphaenolates in salt metathesis and hydrolysis reactions, as well as their thermal stability and decomposition pathways.

Results and Discussion
Addition of an aryl-lithium reagent LiR (R = Mes or Mes*) to an in situ generated solution of the silyl phosphaethynolate i Pr 3 SiOCP in toluene leads the formation of lithium phosphaenolates 1 a and 1 b (Scheme 1). [10] The reduction of the C � P bond order is immediately evident by 31 P NMR spectroscopy, where the starting phosphaethynolate resonance ( 31 P{ 1 H}: À 360 ppm) is replaced by a single new downfield signal ( 31 P{ 1 H} 1 a: 132.0 ppm; 1 b: 159.5 ppm), indicating quantitative product formation. The aryl group adds to the phosphorus center, as evidenced by the presence of CÀ P coupling in the 13 C{ 1 H} NMR resonance attributed to the ipso-carbon atom (1 a: 13 C{ 1 H}: 137.55 ppm; d, 1 J C-P = 58.9 Hz). The silyl group was found to migrate from oxygen to carbon. Micro-solvation by coordinating solvents is typical in the isolation of lithium enolates, [11] however whilst dioxane is present during the synthesis of 1 a and 1 b, 1 H NMR spectroscopy confirms none is present in the isolated products; both are base-free enolates.
Lithium enolates 1 a and 1 b readily crystallize from standing hexane solutions; Figure 2 depicts the resulting solid-state structures, and key bond metrics are collated in Table 1. Like the direct boryl analogue D, both are dimeric enolates, featuring a Li 2 O 2 4-membered ring in the center of a fused planar tricyclic system. The internal LiÀ OÀ Li angles of this planar motif are close to 90°. Short Li1···C2 contacts in both 1 a (2.412(2) Å) and 1 b (2.422(2) Å) indicate interaction of the cation with the aryl group, providing intramolecular stabilization in the absence of an external coordinated base. The P=C bond lengths of 1 a and 1 b are 1.712(1) Å and 1.714(1) Å, which are slightly elongated for phosphaalkene bonds indicative of delocalisation of electrons from the P=C into the CÀ O bond. [12] The magnitude of this interaction can be calculated using Natural Bond Order (NBO) analysis. [13] Such analysis for 1 b gives an interaction between the oxygen-based lone pair and the PÀ C π* orbital of 272.3 kJ mol À 1 (see Supporting Information for full details), a value consistent with moderate delocalization across the PÀ CÀ O unit.
Analogous products could not be isolated from reactions of i Pr 3 SiOCP with LiTipp or Li Tipp Ter (Tipp = 2,4,6-tri-isopropyl phenyl; Ter = 2,6-bis(2.6-diisopropylphenyl)phenyl). The phosphaethynolate i Pr 3 SiOCP is a transient species, and isomerizes to the corresponding phosphaketene i Pr 3 SiPCO over time in solution. [14] Addition of LiMes to a toluene solution containing only the thermodynamically favoured phosphorus-bonded isomer gave a complex, intractable mixture of products, implying that the formation of enolates 1 a/1 b occur via the oxygen-bonded isomer. Similarly, using the triphenylsilyl analogue of the phosphaethynolate, which is known to exist as a mixture of Ph 3 SiOCP and Ph 3 SiPCO, gave an intractable mixture of products. We therefore hypothesize that formation of 1 a/1 b involves initial addition of LiR across the C � P bond of i Pr 3 SiOCP to form an unobserved intermediate. This is immediately followed by a 1,2 migration of the -Si i Pr 3 group from oxygen to carbon, driven by the oxophilicity of the lithium counter-cation. This reactivity is analogous to what we have previously observed for a boryl-substituted phosphaethynolate on reaction with aryl-lithium reagents. [9] Addition of 12-crown-4 to 1 a or 1 b effectively sequesters the lithium cation, breaking up the dimeric structure to form monomeric salts 2 a and 2 b, respectively. An upfield shift of the 31 P{ 1 H} NMR resonance provides evidence of ether coordination ( 31 P{ 1 H} 1 a: 132.0 ppm; 2 a: 108.    À 0.15 ppm, 2 J H-Si = 6.7 Hz) support the silylation. Expectedly due its low molecular weight and additional silyl group, compound 3 a was isolated a pale-yellow oil with a melting point below À 30°C. The bulkier Mes* derivative, 3 b, can be crystallized from hexane as colorless prisms, and the resulting solid-state structure is shown in Figure 3. The P1=C1 bond length of 3 b, at 1.687(1) Å, is the shortest in our series here, and more the in the range expected of a true phosphaalkene. [12] This greater degree of multiple bond character is also evident in the CÀ P coupling constants which increase relative to 2 a/2 b (3 a: 1 J C-P = 93.2 Hz; 3 b: 1 J C-P = 95.1 Hz). It follows that an NBO analysis indicates less delocalization over the PÀ C-O unit in 3 b, with the magnitude of the interaction between the oxygen-based lone pair and PÀ C π* orbital being less than half that of 2 b (2 b: 354.7 kJ mol À 1 ; 3 b: 122.2 kJ mol À 1 ). The new C1À O1À Si2 fragment is significantly deviated from the expected bent geometry, with an angle of 160.10(9)°.
A stoichiometric reaction between 3 a and KO t Bu at 60°C for 18 h gave clean conversion to corresponding potassium enolate 4, which exhibits a 31 P{ 1 H} NMR resonance 104.7 ppm (Scheme 2). Retention of the mesityl group is evident in the 1 H NMR spectrum of 4, with singlet signals at 6.71, 2.45 and 2.13 ppm, in a ratio of 2 : 6 : 3. The expected doublet and septet resonances corresponding to the i Pr groups coalesce into one singlet resonance, which could not be deconvoluted by 1 H-1 H COSY experiments. No reactivity was not observed between silylated 3 a and NaO t Bu or LiO t Bu.
Reaction of bulkier silyated enolate 3 b with KO t Bu did not lead to the isolation of a product analogous with 4. Several equivalents of KO t Bu were required to consume 3 b, and from these mixtures a colourless crystalline solid was isolated. This product is insoluble in most common laboratory solvents and thus its full characterisation has thus far eluded us.
Addition of degassed water to base-free enolates 1 a and 1 b results in protonation of the enolate with elimination of LiOH, forming 5 a ( 31 P{ 1 H} 151.3 ppm) and 5 b ( 31 P{ 1 H} 169.1 ppm), respectively (Scheme 3). In the case of mesitylsubstituted 5 a, whilst initial protonation forms a phospha-enol, this tautomerizes through a 1,3 H-shift to the corresponding acyl phosphine isomer 5 a' within a few hours in solution. This transformation can be monitored by in situ 1 H and 31 P NMR spectroscopy ( Figures S32 and S33), with the appearance of a proton-coupled 31 P NMR signal at À 20.0 ppm (d, 1 J P-H = 217 Hz). A similar rearrangement was seen for the analogous protonated form of D, although the rearrangement in that case was found to proceed slowly in benzene but rapidly in pyridine. [9] When attempting to isolate 5 a, this rapid rearrangement was also observed in neat samples, making it very challenging to obtain pure samples of 5 a. Density functional theory (DFT) calculations indicate that acyl phosphine 5 a' is 10.6 kJ mol À 1 lower in energy than phospha-enol 5 a, providing an energetic driving force for the rapid tautomerization. On further standing at room temperature, mixtures of 5 a and 5 a' formed trace amounts of a third product, resonating at À 175.9 ppm (d, 1 J P-H = 207 Hz) in the 31 P NMR spectrum, and heating this mixture results in full conversion to this third compound (see below). No 1,3tautomerization was seen in the case of Mes*, and 5 b was isolated as a colorless solid. Whilst not observed, the corresponding 5 b' tautomer is calculated to be 7.3 kJ mol À 1 lower in energy than 5 b, and we attribute the stability to rearrangement to the larger Mes* preventing the tautomerization through steric interactions.
Heating a toluene or hexane solution of mesityl-substituted enolate 1 a, or its silylated analogue 3 a, for several days results in the formation of a small amount of a new phosphine product, 6, observed at À 175.9 ppm (d, 1 J P-H = 214 Hz) in the 31 P { 1 H} NMR spectrum. Higher conversions to 6 were achieved using lithium enolate 1 a, where the maximum conversion which could be achieved was reproducibly around 33 % in hexane, plateauing after 7 days. A new mesityl environment is evident in the 1 H NMR spectrum of the product mixtures, with singlet resonances at 6.76 (1 a: 6.77; 3 a: 6.82), 2.46 (1 a: 2.40; 3 a: 2.55) and 2.10 (1 a: 2.08; 3 a: 2.18), corresponding to the aryl and ortho/para methyl groups, respectively. Heating base adduct 2 a or indeed bulkier analogues 1 b, 2 b or 3 b did not produce any observable analogous reaction. The same product 6 can be generated cleanly by heating the tautomeric mixture of 5 a and 5 a', allowing for further NMR studies. Initially we considered that 6 could be MesPH 2 from further protonation of 5 a', however comparison of chemical shifts allowed us to discount this decomposition pathway (MesPH 2 : 31 P{ 1 H} À 156 ppm, 1 J P-H 206 Hz). [16] Long range 1 H-13 C HMBC interactions between PÀ H and the methine CÀ H, as well as the ortho-CH 3 of the mesityl group indicated all the functional groups remain on a single molecule, and so we considered the generation of MesP(H)(Si i Pr 3 ) through the extrusion of CO from 5 a' (Scheme 3). Whilst no evidence for carbon monoxide formation was found by in situ 13 C{ 1 H} NMR experiments, crystals of 6 grown at low temperature suitable for X-ray diffraction revealed that the identity of 6 is indeed this secondary phosphine ( Figure 4).
Whilst 6 has been reported as a reaction intermediate, [17] to the best of our knowledge it has not yet been structurally authenticated or isolated. The unexpected formation of this compound through carbon monoxide extrusion from an acylphosphine to our knowledge only has one literature precedent: on warming from À 90°C to room temperature, t BuP(H)C(O)(Cl) extrudes CO to form t BuP(H)(Cl). [18] Bulkier hydrolysis product 5 b does not undergo the same 1,3 rearrangement to a phosphine tautomer as 5 a at room temperature. Under the same thermal conditions for the formation of 6, 5 b instead decomposes to Mes*H and i Pr 3 SiPCO, presumably through a cyclic 5-membered transition state (CÀ PÀ CÀ O-H). In the absence of base, i Pr 3 SiPCO is stable and does not undergo any dimerization or cyclisation. [14] In the DFT optimized structure of 5 b ( Figure S43) the interatomic distance between the ipsocarbon atom and hydroxyl proton is reasonably short at 2.232 Å, and with a CÀ P=C angle of 99.4°, the À OH is geometrically poised for deprotonation. This stark difference in thermal decomposition pathway gives further support for the formation of 6 being solely from tautomer 5 a', and not 5 a.
Given the relatively high, reproducible conversion rates, and that reactivity on heating is limited only to two analogous compounds, it seems unlikely that the formation of 6 from 1 a or 3 a is from protonation via trace amounts of water present in the dried and degassed solvents. Further, this reactivity could be observed across different batches of reagents. We hypothesize that 1 a and 3 a deprotonate the glassware surface, and that this is the origin of 6 when heating these extremely reactive compounds.

Conclusion
We have shown that the addition of LiMes or LiMes* across the C � P bond in i Pr 3 SiOCP triggers migration of the silyl group from the oxygen to the carbon atom, resulting in the isolation of lithium phospha-enolates 1 a and 1 b. The dimeric structures of these enolates can be broken up into their monomeric components by addition of 12-crown-4 (2 a and 2 b). By utilizing a salt metathesis pathway, silylated silyl enol ethers (3 a and 3 b) can be prepared, from which the corresponding potassiated metal exchange product 4 was synthesized. Hydrolysis of the enolate by addition of degassed water produces phospha-enols 5 a and 5 b, and through an intra-molecular rearrangement the corresponding acyl-phosphine forms over the course of hours for smaller Mes-substituted example 5 a. Heating 5 a/5 a' cleanly extrudes CO from the acyl-phosphine, forming silyl phosphine.
NMR spectra were acquired on a Bruker Avance Neo 600 MHz NMR spectrometer with a broadband helium cryoprobe ( 13 C 151 MHz) or a Bruker AVIII 400 MHz NMR spectrometer ( 1 H 400 MHz, 1 H-29 Si HMBC 80 MHz, 7 Li 156 MHz, 31 P 162 MHz) at 295 K unless otherwise stated. 1 H and 13 C{ 1 H} NMR spectra were referenced to residual protic solvent resonance ( 1 H NMR C 6 D 6 : δ = 7.16 ppm; 13 C NMR C 6 D 6 : δ = 188.06 ppm). 7 Li NMR were externally referenced to LiCl in D 2 O (1 M). 29 Si NMR were externally referenced to Me 3 SiH. 31 P NMR were externally referenced to an 85 % solution of H 3 PO 4 using the spectrometer reference values. For 1 H-29 Si HMBC experiments a SiÀ H coupling constant of 6 Hz and a relaxation delay of 1.5 s was used to generate the 2D spectra. Elemental analyses were performed by London Metropolitan University. Samples (~10 mg) were submitted in vacuum sealed ampoules (solids) or double packaged vials sealed with electrical tape (oils).