Synthesis, reactivity and computational analysis of halophosphines supported by dianionic guanidinate ligands

All material supplied via JYX is protected by copyright and other intellectual property rights, and duplication or sale of all or part of any of the repository collections is not permitted, except that material may be duplicated by you for your research use or educational purposes in electronic or print form. You must obtain permission for any other use. Electronic or print copies may not be offered, whether for sale or otherwise to anyone who is not an authorised user. Synthesis, Reactivity, and Computational Analysis of Halophosphines Supported by Dianionic Guanidinate Ligands Brazeau, Allison; Hänninen, Mikko M.; Tuononen, Heikki; Jones, Nathan; Ragogna, Paul


Introduction
A major theme in the field of synthetic main group chemistry is to push the limits of the bonding environments surrounding a central p-block element and impose a stress that will then allow for access to unprecedented yet controlled reactivity. There are many variables associated with forcing such a condition within a molecule that include modifications to the electronics, sterics, charge and ring strain.
Over the past decade there has been a shift in research efforts to design main group molecules that are capable of reactivity previously reserved for the transition metals. Highlights include the ability of "frustrated Lewis pairs," low-valent group 13 species, and heavy group 14 analogues of N-heterocyclic carbenes, alkenes and alkynes to activate small molecules, including H2, NH3, PH3, and P4. Power et al. were the first to demonstrate the bifurcation of dihydrogen by a main group metal, with the addition of H2 across the GeGe bond in unsaturated digermynes at ambient temperature and pressure to give a combination of digermenes and germanes. 1 Several years later he also showed the addition of dihydrogen to distannynes to give tin(II) hydrides. [2][3][4][5][6][7][8] A notable surge in the reported reactivity of "frustrated Lewis pairs" (FLP) was kickstarted by Stephan et al., where they reported metalfree reversible hydrogen activation by the unquenched reactivity of a phosphine-borane ((C6H2Me3)2P(C6F4)B(C6F5)2), which features a Lewis acidic boron and Lewis basic phosphorus. 9 Since then, many different FLP systems have been developed 10 and reactivity with a variety of substrates has been observed, including B-H bonds, 11 CO2, 12,13 unsaturated bonds 10 and even catalytic, metal-free hydrogenation. [14][15][16][17][18] The silylene LSi (L = CH{(C=CH2)(CMe)(2,6-i Pr2C6H3N)2}) will oxidatively add ammonia to give four-coordinate LSi(H)NH2 19 and is the only reported main group molecule that has been successfully used in the activation of PH3, yielding LSi(H)PH2. 20 Phosphenium cations [R2P] + are a class of phosphorus-containing molecules that have a rich history of reactivity, including reactions with Lewis bases, 21-23 inorganic and organic unsaturated bonds, 24 and as ligands on transition metals. [25][26][27][28][29][30][31][32] Phosphenium cations are also known to catenate, which is demonstrated by the many examples from Burford et al. in which a [R2P] + fragment is inserted into cyclic and acyclic P-P bonds, to expand the ring or chain by an additional P atom to give catenapolyphosphorus cations. 33 Weigand has recently shown the ability of phosphenium cations to activate small molecules by novel examples of controlled reactivity with P4 using a N-heterocyclic phosphenium cation (NHP) supported within a strained four-membered ring system to form cationic P5 clusters. 34,35 Substitution reactions on acyclic cations of the type [LPCl2] + and [L2PCl] 2+ to replace the Cl with CNor N3-groups have also been accomplished by this same group. 36 The phosphenium cation in particular presents a unique environment where the molecule is Lewis amphoteric given that the cationic phosphorus atom can be an electron donor or electron acceptor, which opens the floodgates for unique structures, bonding arrangements and reactivity. The NHP is isovalent with the N-heterocyclic carbene (NHC) but has inverse electronic properties, in that they are poor -donors but have excellent -accepting capabilities (Chart 1a).
We envisaged designing a NHP within a highly strained four-membered ring so that the phosphorus atom could be probed for new and unobserved reactivity. While several four-membered rings incorporating NHP are known (Chart 1b) they are all of the type P-N-X-N where X = P, 34,37 Si or Al, 35,38 all electropositive elements, however the example where X = C has surprisingly been absent from the literature. We sought to extend this series and fill the void by using the nitrogen-based, chelating, guanidinate ligand (Chart 2a,b), which despite its wide spread use for supporting metallic elements spanning the periodic table, little is known about the role in stabilizing group 15 centres in the +3 oxidation state. Jones et al. have synthesized several (monoanionic)guanidinate pnictogen(III) dihalides (pnictogen = Pn = P, As, Sb) as precursors to amido-dipnictenes (Chart 3c). 39 Related (monoanionic)amidinate ligands (Chart 2c) have been used in stabilizing heavier group 15 atoms (As, Sb, Bi) [40][41][42][43] as shown in Chart 3a,b. We have previously reported a (monoanionic)guanidinate ligand capable of supporting a dicationic arsenic centre with additional stabilization from a Lewis base (Chart 3c). 44 For the synthesis of an NHP it would be ideal to have the "free" cation, which therefore calls for the use of a (dianionic)guanidinate (Chart 2b).
There is only one previous report of a (dianionic)guanidinate being used to support a group 15 atom involving antimony 45 (Chart 3d), and one example of a phosphorus centre 39 being supported by a guanidinate ligand. In this context, we report the first synthesis and comprehensive characterization of a series of chlorophosphines supported by homo-and heteroleptic (dianionic)guanidinates. The reactivity of the model compound tris(2,4,6-trimethylphenyl)guanidinato chlorophosphine (2,4,6-trimethylphenyl = Mes) has been studied extensively. We have explored the tendency of these compounds to retain the halogen and therefore the reluctancy to form the "free" NHP. A base-stabilized NHP however was easily accessible and computational analysis has provided insight on this rather counterintuitive result and gave valuable data on the nature of the electronics about the phosphorus atom, and why these particular systems are resistant to halide abstraction. Novel chemically instigated carbodiimide elimination was observed in conjuction with metal coordination, resulting in the first structurally characterized metal complex with a cationic iminophosphine ligand. A study between chlorophosphine and a gentle oneelectron reductant was explored to give the clean synthesis of the reductively coupled product featuring a dimeric structure with -N,N bridging guanidinates and a P-P bond. Upon investigating the thermal stability of the tris(Mes)guanidinato chlorophosphine we discovered the thermally induced ejection of carbodiimide and subsequent insertion of chloro(imino)phosphine into the P-N bond of the initial diaminochlorophosphine resulting in a new ring expansion product with a -N,N bridging guanidinate and a -bridging N-Mes. This ring expansion chemistry is extremely rare within P(III)-N chemistry.
These observations collectively give a deeper perception on the nature and reactivity of diaminochlorophosphines constrained in a four-membered ring.

Experimental Section
General Procedures. All manipulations were performed in an inert atmosphere in a nitrogen filled MBraun Labmaster dp glovebox or using standard Schlenk techniques unless stated otherwise.

9.
A THF (3mL) solution of Cp2Co (0.43 g, 2.27 mmol) was added to a colorless THF (3 mL) solution of 1PCl (1.08 g, 2.27 mmol). The reaction mixture turned a dark brown color and after stirring for 48 h at rt copious amounts of green precipitate were visible. The green precipitate was separated by centrifugation and the dark solution was decanted and dried in vacuo to give an off-white solid. The product was washed with CH3CN (2 x 6 mL) to remove remaining Cp2Co and the suspension was centrifuged. The solid white product was dissolved in CH2Cl2 (3 mL) and precipitated with the addition of CH3CN, the solution was decanted and discarded, and this process was repeated once more. After the second precipitation the vial was placed in the freezer (-30 C) for 45 min, the slightly colored solution was discarded and the white precipitate was dried in vacuo. X-ray quality crystals were grown at -30 C from the liquid diffusion of CH3CN into a concentrated CH2Cl2 solution of the bulk material over a oneweek period. Yield: 61%; dp: 310-320 C. 1

11.
In a pressure tube 1PCl (0.54 g, 1.1 mmol) was dissolved in CDCl3 (3 mL) and heated in an oil bath at 90 C overnight. The solvent was removed in vacuo, producing a wax-like colorless product.
This was washed with pentane (6 mL), which produced a white powder. The suspension was centrifuged and the decanted solution was transferred to a vial and kept in the freezer (-35 C) for 1 h.
The centrifuged white solid was dissolved in CH2Cl2, transferred to a vial and dried in vacuo to give a sticky white solid. Resuspension in pentane and drying in vacuo produced a fine white powder. The cooled solution had a white precipitate, which was isolated by decanting the pentane and drying in vacuo to give a white powder, which was combined with the previous isolated product. (ranked intensity)) 452 (3), 563(9), 600(11), 722(15), 853 (5), 950 (7), 981 (10), 1056(4), 1140 (14), 1183 (8), 1222(1), 1475 (6), 1608 (12), 1649 (2), 2919 (13). FT-Raman (cm -1 (ranked intensity)) 86 (6), 204 (7), 396 (14), 421 (8) In an attempt to isomerize to one product, an aliquot of the reaction mixture was transferred to an NMR tube and heated at 90 C overnight. The 31 P{ 1 H} NMR spectrum revealed that two peaks were still present in approximately the same ratio, but that the resonances had now shifted downfield approximately 30 ppm to P = 210.9 and 211.1. There were no observable N-H peaks in the 1 H NMR spectra of 1PCl-5PCl and the IR spectra of the solids lacked N-H vibrations in the range of  = 3100-3500 cm -1 . 44,71 Given these data, the compounds were assigned as the chlorophosphines 1PCl, 2PCl, 3PCl, 4PCl, and 5PCl isolated in low to good yields (Table 1). X-ray quality crystals were grown from samples of the bulk powders for 1PCl-4PCl and subsequent X-ray diffraction experiments confirmed the synthesis of the strained, four-membered, cyclic diaminochlorophosphines.  68 Analogous results were observed using the sources of metal triflates with 1PCl over 7 days at 100 C. Only compound 1PCl was employed as a model system in subsequent reactivity studies given the simplicity of its solution 1 H NMR spectrum. In the case of 5PCl the addition of Me3SiOTf results in several products as evidenced by the many signals in the 31 P{ 1 H} NMR spectrum.
Given the difficulty associated with removing chloride from the chlorophosphines, we looked to the bromo-derivatives, as heavier halides generally undergo more facile metathesis. 72 The bromophosphine 1PBr was synthesized in a similar manner to 1PCl (Scheme 2, Table 1) and subjected to identical metathesis conditions. Again no reaction or multiple products were observed, none of which corresponded to a downfield chemical shift expected of an NHP in the 31 P{ 1 H} NMR spectra.
Given the puzzling failure of the salt metathesis routes, the Lewis acids AlCl3 and GaCl3 were employed as halide abstracting reagents for 1PCl. These reactions were also monitored by 31 P{ 1 H} NMR spectroscopy and revealed the formation of single products with upfield chemical shifts (P  10, Table 1). The volatiles were removed in vacuo and the bulk powders were redissolved in CDCl3 to obtain the 1 H NMR spectra, which in both cases showed sharp peaks indicating that there was no longer a slow exchange process on the NMR time scale, and furthermore that all symmetry in the molecule was lost. Given the upfield chemical shift in the 31 P{ 1 H} NMR spectra and the lack of symmetry noted in the 1 H NMR spectra, it was hypothesized that halide abstraction was not effected, rather a Lewis acid (AlCl3, GaCl3)/Lewis base (N or P) adduct was formed (Scheme 3). Single crystals suitable for X-ray diffraction studies were obtained for the product containing AlCl3 and the solid-state structure confirmed an NAl adduct of AlCl3 from the exocyclic nitrogen of the guanidinate ligand (6Al).
Although suitable single crystals for X-ray diffraction studies could not be grown for the GaCl3 adduct (6Ga) an identical structure was assigned based on identical 1 H NMR spectra and a similar chemical shift in the 31 P{ 1 H} NMR spectrum. It should be noted that while halide abstraction with these four-membered diaminochlorophosphines appear to be non-trivial, the analogous reaction of those with unsaturated fivemembered rings are generally facile. 28,[73][74][75][76][77] One major difference between these two species is the additional stabilization gained from the delocalization of a 6π-electron system in the case of the fivemembered ring, contrary to the 4π-electrons present in the four-membered ring. While aromaticity is not a necessity for the isolation of NHPs 25,34,35,37,38,67 and has been found to be a weak factor in their stabilization, 75 the lack thereof in our four-membered diaminochlorophosphines may be a contributing factor to the difficulty in halide abstraction.
e) 1PBr f) 6Al     Table 3 lists the atomic partial charges for the studied chlorophosphines as obtained from the natural population analysis (NPA) of their Kohn-Sham electron densities. 59 It is necessary to stress that the absolute values of the calculated charges have no physical meaning and it is only their relative magnitudes which yield useful information about structure-induced changes in the electron distribution.
The atomic partial charges of phosphorus and carbon in 5PCl and 12PCl-15PCl display only small variations, which is in contrast to those calculated for nitrogen and chlorine. These both show a more distinct dependence on the electron withdrawing/donating nature of N-substituents, most notably the all-Ph derivative 12PCl has the least charge concentrated on the electronegative elements within the molecular framework, thereby yielding the least polar P-Cl bond. In contrast, the N-alkyl substituted variants 5PCl, 14PCl and 15PCl display a less uniform charge distribution and consequently a more ionic P + -Cl  interaction. These results pinpoint N-alkyl derivatives of the target chlorophosphines as the most favorable candidates for halide abstraction. The increased reactivity of the cyclohexyl derivative 5PCl towards Me3SiOTf was also observed experimentally (vide supra).
A comparison of charge distributions calculated for 5PCl and 12PCl-15PCl with those of compounds for which halide abstraction is reported to take place reveals, as expected, that the P-Cl interaction is significantly more ionic in the latter systems.
The currently characterized chlorophosphines 16PCl-19PCl display equal and roughly 0.10 units greater charge separation within the P-Cl bond as compared to the identically N-substituted species with a P-N-C-N backbone (see Table 3). Consequently, chlorophosphines with a P-N-C-N backbone appear, by their nature, to be resilient to salt metathesis. To analyze the origin of this phenomenon in detail, we turned our attention to the frontier orbital structure of the corresponding NHPs.  Figure 5 shows the lowest unoccupied molecular (Kohn-Sham) orbitals (LUMOs) and orbital energies of NHPs with different backbones but identical N-substituents. This orbital is of particular interest in the study of the nature of bonding in equivalent chlorophosphines as it interacts directly with an electron pair from the halide anion to form the P-Cl bond. Although the overall morphology of the orbitals in Figure 5 is rather similar, a -type molecular orbital (MO) with a major contribution from the pz atomic orbital (AO) of phosphorus and a smaller admixture of AOs from the two flanking nitrogen nuclei, their orbital energies are vastly different with the LUMO of our target cations residing by far the lowest. Frontier MO theory based arguments predict that NHPs based on the P-N-C-N backbone should be the best electron acceptors in the series and form the most covalent P-X bonds. The observed differences in the charge distributions of the investigated halophosphines can therefore be correlated with the differing orbital characteristics of the corresponding NHPs. This holds not only for systems shown in Figure 5 but also for other chlorophosphines examined in the current work (see ESI). In addition, we note that the morphology of the LUMO is consistent with the possibility to affect the details of the P-Cl interaction by changing the exocyclic substituents from -(alkyl) to -type (aryl). Given that our target cations appear to be much better electron acceptors than other known NHPs, the details of the P-X interaction in the corresponding halophosphines should be inferable not only from atomic partial charges but also from calculated bond strengths. We examined the nature of the P-Cl bond in compounds 5PCl and 12PCl-15PCl with the help of energy decomposition analysis (EDA) which partitions the bonding interaction between a cationic NHP and a chloride anion into physically meaningful components (see Table 4). 60 It should be noted here that the instantaneous interaction energy (Eint) given by the EDA is not (the negative of) bond dissociation energy, which takes into account the energy gained from relaxation of the fragment geometries upon bond breaking.
However, when comparing bonding trends across multiple similar systems, the snapshot-type picture given by the EDA procedure is sufficient.  Table 4 lists the calculated EDA interaction energies along with their division into individual contributions from Pauli repulsion (EPauli) and electrostatic (Eelstat) and orbital interactions (Eorb). The trend in Eint values confirms that the P-Cl bond is markedly stronger in all chlorophosphines with a P-N-C-N backbone as compared to systems previously reported to undergo salt metathesis. The only exception are the fully alkyl substituted 5PCl and 15PCl whose interaction energy equals that of 19PCl.
This is in accord with experimental observations, which demonstrated the increased reactivity of 5PCl over other halophosphines studied in the present work. Changes in the percentage of Eorb from the total attractive interactions (Eorb + Eelstat) show that as the P-Cl interaction weakens it also becomes more electrostatic in nature. These results are fully in accord with the picture gleaned from the calculated charge distributions and LUMO energies. A correlation analysis on the calculated P-Cl charge difference (Table 3) and the percentage of covalent (orbital-type) character in the P-Cl bond (Table 4) yields a linear relationship with a correlation coefficient R 2 = 0.93 (Figure 6a). An equally good linear regression (Figure 6b) is obtained if the LUMO energies of NHPs (see ESI) are plotted against the EDA interaction energies (Table 4). Investigation of the unoccupied orbitals of the studied chlorophosphines offers a rationale for the role of bpy in capturing the targeted NHP as a base-stabilized cation. As shown in Figure 7 for 12PCl, both of its two lowest unoccupied orbitals have suitable morphologies to accept electron density from a coordinating Lewis base. A geometry optimization for 12PCl with bpy was ran and led to the formation of an adduct with a slightly (0.05 Å) longer P-Cl bond as compared to 12PCl. This is consistent with the antibonding nature of the orbitals in Figure 7, which in turn implies a weakened P-Cl interaction in the adduct. Consequently, an EDA calculation conducted for the bpy complex of 12PCl yields a P-Cl interaction energy of 482 kJ mol 1 which is significantly less than that obtained for free 12PCl and similar to that found for chlorophosphines with either P-N-Si-N or P-N-C=C-N backbones (Table 4).

Figure 7.
Two of the lowest unoccupied molecular orbitals of 12PCl.
The collective results from computational investigations enable us to conclude that the exceptionally good electron acceptor properties of the target NHP cations render the corresponding halophosphines reluctant to halide abstraction, unless there is additional stabilization from a Lewis base.
The P-X bonds are not only stronger but also more covalent than in analogous systems known to undergo salt metathesis. Consequently, triflate-based reagents show no or only limited reactivity with 1PCl-5PCl, most likely simply due to increased energy required to break the P-Cl bond.

Conclusion
In summary, 1PCl-4PCl and 1PBr represent the first examples of N-heterocyclic chlorophosphines supported by (dianionic)guanidinate ligands. The experimentally observed reluctancy to form the corresponding N-heterocyclic phosphenium cations was rationalized by computational studies. Analysis of the charge distributions, orbital structures and relative P-Cl bond energies of the computationally studied compounds give corroborating evidence of the strong P-X bond and strong Lewis acidity of the cationic species. Reactivity studies of 1PCl demonstrate the high degree of strain induced by the four-membered ring, which leads to chemically and thermally induced carbodiimide elimination, as well as a novel ring expansion by insertion of chloro(imino)phosphine into a P-N bond of 1PCl. These results demonstrate the electronic nature and alluring reactivity of diaminochlorophosphines restricted in a four-membered ring.