A robust tetravalent phosphaza-adamantane scaffold for molecular and macromolecular chemistry

Tetraarylmethanes and adamantanes are very rare examples of rigid, four-way, anionic connectors that play a scaffolding role in multiple areas of molecular and materials chemistry. We report the synthesis of a tetravalent phosphaza-adamantane cage, (PNSiMe3)4(NMe)6 (2), that shows unusually high ambient, thermal, and redox stability due to its unique geometry. It nevertheless participates in fourfold functionalization reactions on its periphery. The combination of a robust core but a reactive corona makes 2 a convenient inorganic scaffold upon which tetrahedral molecular and macromolecular chemistry can be reliably constructed. This potential is exemplified by the unprecedented synthesis of a tetracationic tetraphosphinimine (3) and the first porous all-P/N polyphosphazene network (5).

The unique geometry of tetrahedral, tetravalent molecules makes them valuable scaffolds in synthetic chemistry. Their ability to connect four functional groups in a rigid and well-separated tetrahedral arrangement has allowed development of new optoelectronic materials, 1 thermally-stable energetic compounds, 2 catalysts with enhanced robustness or multi-catalytic sites, [3][4][5][6] bioactive polypeptide frameworks, 7 and pharmaceuticals. 8 In crystalline reticular chemistry, tetrahedral cages are privileged secondary bonding units as their high symmetry facilitates packing, [9][10][11][12][13][14][15] and in amorphous reticular chemistry, tetrahedral connectors have been used to construct hyper-crosslinked polymers or porous organic polymers. [16][17]  The two most-studied families of tetrahedral, tetravalent linkers are the tetraarylmethanes (A) and adamantanes (B) -both featuring a carbon skeleton. Our interest in the reactivity of geometrically constrained p-block amides [18][19][20][21][22] led us to the family of phosphorus-nitrogen cages (C) reported by Holmes nearly 6 decades ago. [23][24][25][26] Their quantitative, one-step, multi-gram synthesis from commodity reagents (PCl3, RNH2) is appealing from a practical perspective, and their high molecular symmetry makes them inherently suited for evolving a four-directional functional platform. Quadruple oxidation of some derivatives of C with azides, sulfur, and oxygen has also been reported, [27][28][29][30][31][32] but no subsequent reactivity was possible since the resulting compounds do not feature sufficiently labile bonds. Salts of the binary polyanion P4N6 10have also been reported, but their high temperature solid-state synthesis (>600 o C elemental melt) and insolubility have precluded further use in synthetic chemistry. [33][34][35] We envisioned that conversion of C to a masked form of tetra-anion D would allow solution-phase tetravalent chemistry with a new inorganic synthon. Specifically, if the P III atoms in compound 1 could be oxidized to P V silylphosphinimines, the exo-cage N-Si bonds may be polar enough to engage in subsequent covalent metathesis with element halides. Here we validate this hypothesis and report the synthesis, structure, and reactivity of 2 as a new, remarkably robust, electron-rich, tetrahedral scaffold for molecular and macromolecular chemistry. Scheme 1. Synthesis of 1 and its sequential oxidation by Me3SiN3 to give 2.
We expected the four-fold oxidation of 1 with four equivalents of Me3SiN3 to be facile given that the analogous reaction of (Me2N)3P is complete at 55 o C in a few hours. 36 To our surprise, while single and double oxidation of 1 occurred smoothly (Scheme 1), giving 1' and 1'', complete oxidation was not achieved even upon refluxing with Me3SiN3 in toluene overnight. The reaction progress can be easily monitored through signal multiplicities observed in the 31 P NMR spectra (Figure 2), and doing so over two half lives revealed the bimolecular rate constants of successive oxidations to be >100, 14.3, 2.3, and 0.15 M -1 h -1 ( Figure S10, SI). The dramatic deceleration as a function of extent of oxidation is likely an electronic rather than steric effect, as it is also observed when the reaction is performed with a less hindered benzyl azide. 30 Even using a 10-fold excess of Me3SiN3, the reaction proved to be lethargic, requiring 12 weeks at 100 o C to achieve quantitative conversion to the tetraphosphinimine 2. Following removal of excess Me3SiN3 and recrystallization, 2 was reproducibly isolated in 60-80% crystalline yield and comprehensively characterized.
The title compound crystallizes in the tetragonal space group I41/a, and its molecular structure  Figure S11, SI). In contrast, the density functional theory calculated geometry of an isolated molecule of 2 shows a P=N bond length of 1.546 Å and a P=N-Si angles of 132 o , in line with expectations for an imine. We conclude that the experimentally observed bond angle distortion arises from intermolecular forces in the lattice. A view of the sub van der Waals interactions between molecules reveals no interactions involving any of the skeletal P or N atoms, but rather numerous contacts between peripheral Me groups on adjacent units ( Figure S12, SI). Given that SiR3 groups and methyl groups adjacent to heteroatoms are known to be very polarizable, 37 we interpret the lattice energy in 2 as being primarily a result of dispersion forces. Such dispersion-held lattices are ubiquitous in hydrocarbon chemistry, 38 but their dominance is unexpected in heteroatom-dense species like 2, featuring a high number of lone pairs (10), double bonds (4), and polar σ bonds (12 P-N, 4 Si-N bonds). In this context, the title compound is electronically quite distinct from the well-known tetrahedral scaffolds B, which lack polarizable skeletal lone pairs, but it exhibits similar intermolecular forces and physical properties due to symmetry. For example, compound 2 is soluble in all tested hydrocarbon, ethereal, halocarbon, nitrile, and aromatic solvents and sublimes at ca. 150 o C at atmospheric pressure.  We hypothesize that the stability of 2 is at least partly an emergent property of its cage geometry.
First, we note that the lowest occupied molecular orbital (LUMO) of 2 is a combination of four P-N σ* antibonding orbitals, and is confined to a region inside the cage, where it is inaccessible for covalent intermolecular interactions (Figure 4, left). As phosphinimine degradation in protic media involves initial coordination of the solvent to the phosphorus atom, giving a five-coordinate intermediate, followed by generation of an acidic proton that catalyzes solvolysis, [42][43] we propose that the cage-like nature of 2 affords a measure of geometric protection against such reactions. Consistently, degradation of 2 was indeed observed when solutions were spiked with added proton sources (macroscopic amounts of water or acetic acid). Second, we propose that it is an intrinsic geometric feature of small cages that all connected vertices undergo some distortion to accommodate a perturbation at any vertex. The molecular rigidity engendered by this cumulative distortion energy cost may enhance kinetic protection against reaction coordinates involving changes to geometry or coordination number at cage vertices.  The reaction of 2 with four equivalents of Ph3PBr2 showed sequential metathesis (see Figure S17, SI, for spectra of intermediates) to yield the fully substituted product [(P(NPPh3))4(NMe)6][Br]4 (3, Scheme 3). The 31 P NMR spectrum of the cation shows two resonances of equal integration, corresponding to the expected A4X4 spin system and the 1 H NMR spectrum confirmed the loss of all silyl resonances. Single crystal X-ray diffraction unambiguously confirms the molecular structure of 3 in the solid state, but positional disorder arising from four monoatomic anions packing with a 174 atom-large cation significantly mars data quality and precludes discussion of metric parameters ( Figure S18, SI).
The cation in 3 belongs to the well-known family of bis(phosphino)iminium cations (PPN + cations), 44 but is the first all P/N example featuring a +4 molecular charge. 45 Despite this high charge, 3 exhibits no sensitivity towards ambient atmosphere in the solid or solution phases, and, interestingly, also features a LUMO comprised of P-N σ* antibonding orbitals localized primarily inside the cage ( Figure S19, SI).
Formation of 3 demonstrates that despite the stability of the core, quadruple functionalization on the periphery of 2 is possible to build molecular constructs extending from its central cage motif. We therefore envisioned the use of 2 as a platform for the synthesis of networked materials, as is known for scaffolds A and B. Thus, 2 was combined with two equivalents of p-n BuPhPCl2, which was selected as the electrophile due to the solubilizing nature of the linear butyl chain. Monitoring the THF solution by 31 P NMR spectroscopy first showed AM3X and A2M2X2 spin systems expected for the monoand di-substituted cages (4' and 4'', Figure S20, SI). With continued heating, the sharp resonances for these intermediates were replaced by a very broad upfield set, suggesting formation of macromolecular THF vacuum species. Indeed, the reaction mixture formed a translucent gel over time (compound 5, Figure 5), which was isolated by decanting the mother liquor leaching several times with DCM, and drying under vacuum.
Solid state 31 P NMR spectroscopy showed only resonances in the -25 to 55 ppm range, suggesting loss of all P-Cl environments (c.f. P-Cl resonances seen in the 140-160 ppm range for 4' and 4'', Figure S20, SI). Solid state 1 H NMR spectroscopy corroborated this view by showing loss of the Me3Si groups, collectively suggesting that exhaustive substitution was achieved to give 5 ( Figure S10, SI).
The porosity of the gel was confirmed by fully reversible absorption of ca. 100% of its mass in THF at room temperature ( Figure 5). The solvent swollen gel is supple, whereas the evacuated solventfree material is brittle and insoluble in all solvents tested. No evidence of degradation was noted when 5 was exposed to ambient atmosphere for ca. 6 months. While polyphosphazenes and cyclophosphazanes have previously been crosslinked by hydrocarbons, thereby unlocking a vast array of hybrid organic-inorganic functional network materials, 46-51 5 is, to the best of our knowledge, the only demonstrably porous material whose skeleton is constructed exclusively from inorganic phosphorusnitrogen bonds. In this context, conversion of 2 to 5 represents a topological generalization of hitherto linear (1D) and cyclic (2D) polyphosphazenes into the cage (3D) dimension.
In summary, we have achieved the complete oxidation of 1 to access the new tetrahedral, tetravalent inorganic scaffold 2 that shows high thermal, air, and redox stability due to its unique cage geometry. Despite the robustness of its core, 2 remains amenable to coronal decoration as shown by metathesis reactions yielding four-fold extended molecular constructs such as 3 and new classes of inorganic network materials such as 5. The latter is unprecedented as the first porous framework made exclusively from P=N/P-N bonds. These results provide proof-of-principle that a rich molecular or macromolecular covalent chemistry may be built upon this phosphaza-adamantane scaffold, which moreover benefits from the practical convenience of a 31 P NMR spectroscopic handle to accelerate analysis in either the solution or solid phases. Investigations into more expeditious syntheses of 2, its use as a precursor to new inorganic materials, and applications as a secondary bonding unit in crystalline reticular chemistry are underway.