The chemistry of multiple bonded p-block complexes has been the subject of intense research since the seminal work of West and Yoshifuji, reporting the first heavy main group multiple bonds (Si=Si and P=P, respectively) in 1981.¹ The weaker π-bond of the heavier p-block elements renders such species highly reactive. Harnessing this reactivity can lead to “transition metal-like” reactivity, such as the activation/functionalization of relatively inert substrates and catalysis.² This reactivity is of particular interest when observed for highly abundant, low-toxicity elements such as silicon and aluminum, as they present potential alternatives to generally more scarce and toxic transition metal systems which are used heavily in industry. As a result, both the search for new multiple bonded main group complexes and subsequent investigations into their reactivity presents a fruitful avenue to new abundant element-mediated synthetic approaches.

The industrial production of useful monophosphorus (P1) products typically relies on the use of white phosphorus (P₄) as a chemical feedstock, which is then oxidised by elemental chlorine to give PCl₃ and PCl₅ (alongside many other by-products) and subsequent functionalization.³ To mitigate the use of hazardous chlorine gas and improve the atom economy of such chemistry, the direct functionalization of P₄ to phosphorus-based products is highly desirable.⁴ As such, the stoichiometric reactivity of both main group⁵ and transition metal-based organometallic complexes⁶ towards P₄ has been well-explored. While P−P bond cleavage and the formation of new element-phosphorus compounds are now well known for many elements of the periodic table, the subsequent release of P1 products is far rarer.⁴

The controlled activation of P₄ by main group compounds is most commonly observed for those in a low oxidation state, as they possess an amphiphilic nature (an empty p orbital and a lone pair of electrons). For example, both neutral and anionic aluminum(I) reagents have been shown to react with P₄ yielding various formal reduction products such as the P₄ insertion isomers VII⁷ and IX⁸ (Figure 1) and the P₄ fragmentation compounds VIII⁹ and X.¹⁰ Meanwhile, the reactivity of unsaturated main group complexes towards white phosphorus is limited to four examples: that of two disilenes (I and IV, Figure 1),¹¹ a diboraallene (III)¹², and a phosphasilenyl-tetrylene reported recently by our group.¹³

Since our 2017 report of the first dialumene (Al=Al),¹⁴ we and others have been exploring the reactivity of this class of compound towards various substrates such as alkenes, alkynes, CO₂, and H₂.¹⁵ In this contribution, we disclose our efforts to extend the dialumene reactivity to white phosphorus. We demonstrate that both aryl- and silyl-substituted dialumenes selectively afford novel P₄ activation products which are amenable to further functionalization on exposure to electrophiles, affording useful phosphines.

Aryl/silyl stabilized dialumenes 1 ([Tripp(IiPr)Al]₂) and 2 ([SitBu₂Me(IiPr)Al]₂) were synthesized in good yields following the reported procedures (Tripp=2,4,6-iPr₃-C₆H₂; IiPr={MeCN(iPr)}₂C).^{14, 15} Treatment of 1 with one equivalent of white phosphorus in toluene at room temperature gave a brown solid after workup (Figure 2). Over time it was apparent via NMR spectroscopic monitoring that this initial product was converting into a new compound, and it was possible to grow crystals of the product. SC-XRD data¹⁶ confirms this compound to be [Tripp(IiPr)Al]₂P₄ (3-cis, Figure 2), indicating that one molecule of P₄ has inserted into the Al=Al double bond of 1. It features a butterfly-like geometry with adjacent IiPr and Tripp ligands on the Al centers positioned in a cis manner. To the best of our knowledge, this is a new mode of P₄ activation with the aluminum centers. This reactivity is analogous to the recently reported insertion of P₄ into a Si=Si double bond (V and VI, Figure 1).^11c

The P−P bond lengths (av. 2.239 Å) vary in the range from 2.197(2) Å to 2.298(2) Å, which fall in the range of typical P−P single bond (cf. 2.1891(7)-2.2937(7) Å^11c in the P₄ insertion product of Si=Si VI, or 2.209(5) Å^5b in P₄), and are slightly longer than the observed in a formal [P₄]⁴⁻ obtained from the reduction by Al(I), IX (av. 2.29 Å).⁸ But they are longer than the bonds in VII (ca. 2.13 Å) due to the presence of a delocalized π-bond within the [P₄]⁴⁻ unit.⁷ The Al−P bond lengths (av. 2.365 Å) of 3-cis are in the range of those in the P₄Al₆ framework of VIII (2.31–2.42 Å),⁹ but shorter than those of IX (av. 2.37 Å)^11c and VII (av. 2.54 Å).⁷ Additionally, they are longer than that in the P₄ fragmentation product from the alumanyl anion, X (2.246(1) Å) bearing highly polarized Al^δ+-P^δ− bonds.¹⁰

The ³¹P{¹H} NMR spectrum of 3-cis showed three signals at −55.1 ppm (d, P3), −138.3 ppm (d, P1 and P2) and −159.4 ppm (td, P4), which are shifted downfield compared to that observed for VI (−44.0 ppm (dd), 106.6 ppm (dt), −217.6 ppm (dt))^11c and X (−295.4 ppm).¹⁰ They appeared at higher chemical shifts than that of IX (78.6 ppm)⁸ and VII (374.2 ppm (br), 54.3 ppm (br)).⁷ Compound 3-cis is highly thermostable without any change in C₆D₆ when heated to 80 °C for 72 hours.

We next wanted to elucidate the identity of the initial product formed from the reaction of 1 and P₄ (Figure 2). Based on the geometry of the starting material 1, in which the Tripp and NHC moieties are arranged in a trans configuration about the Al=Al bond, we propose that this geometry is obtained upon initial reaction with P₄ giving 3-trans. The ³¹P{¹H} NMR data showed four signals at −52.7 ppm (d, P3), −128.8 ppm (dd, P1), −138.7 ppm (dd, P2), and −147.8 (ddd, P4), which shifted slightly downfield than the observed in 3-cis. P1 and P2 featuring quartet peaks at different shifts indicating they are in asymmetric chemical environments. It was not possible to isolate X-ray quality crystals of 3-trans, so to provide evidence of its formation, we calculated the ³¹P{¹H} NMR chemical shifts and coupling constants and compared them to the results of the experimental ³¹P{¹H} NMR spectrum. The simulated spectrum of the proposed 3-trans reproduces the experimentally observed pattern (Figure S61 in the supporting information). Quantum chemical calculations on the proposed mechanism for the reaction of compound 1 with P₄ giving 3-trans and 3-cis are presented in Figure S58 in the supporting information. We propose that the formation of 3-cis takes place via the initial formation of 3-trans that was observed by multinuclear NMR spectroscopy, which then isomerized to the energetically more favoured isomer 3-cis. The overall process is exergonic by 84.5 kcal mol⁻¹ relative to 1+P₄. Relaxed surface scans show that the NHC dissociation from Al1 in 3-trans and 3-cis, is barrierless, and results in the essentially similar intermediates 3′ and 3′′ (Figure 3), which only differ from each other by the relative orientation of the aryl substituent at Al1. Thus, we propose that the 3-trans to 3-cis isomerization proceeds via the NHC dissociation from 3-trans to form 3′, which is only endergonic by 14.6 kcal mol⁻¹. 3′ isomerizes to 3′′ at 13.9 kcal mol⁻¹, followed by re-association of the NHC at Al1.

3-cis is calculated to be energetically more favourable by 4.3 kcal mol⁻¹, therefore although the low NHC dissociation energies are expected to allow 3-trans and 3-cis to exist in thermodynamic equilibrium under ambient conditions, only 3-cis can be observed when the equilibrium is reached. In summary, these results indicate that the reaction of 1 and P₄ first goes via a trans conformation 3-trans which slowly isomerises to give the thermodynamic product 3-cis at −30 °C after 7 days or at room temperature after 72 hours (Figure 2).

We were also interested to see whether the silyl substituted dialumene ([SitBu₂Me(IiPr)Al]₂ 2) would react differently towards P₄. Treatment of one equivalent of white phosphorus with 2 in benzene at room temperature after 2 hours yielded a pale orange powder after work up (Figure 4). Recrystallisation of the orange powder and subsequent SC-XRD analysis shows the solid state structure to be [SitBu₂Me(IiPr)Al]₂P₄ 4, which possesses an analogous Al₂P₄ cage motif to 3-cis, but the NHC and silyl ligands are in a trans configuration (Figure 4). The distances between the Al centers and NHC (Al1−C1 2.066(3) Å, Al2−C12 2.103(3) Å) as well as P−P bond lengths (av. 2.237 Å) in Figure 4 were comparable with those of 3-cis (A1−C16 2.100(6) Å, Al2−C42 2.095(6) Å, P−P av. 2.239 Å). The angles of P−Al−P in 4 (av. 93.33°) are also similar to 3-cis (av. 93.44°). The ³¹P{¹H} NMR spectrum shows a multiplet for P4 at −118.5 ppm which is significantly downfield shifted compared with the signal of P4 in 3-trans (−147.8 ppm) and 3-cis (−159.4 ppm).

Contrary to 3-trans, it appears that 4 with trans configuration does not convert into the cis isomer on standing in solution nor on heating. The variable temperature ³¹P{¹H} NMR experiments indicated a minor shift of the signals on heating, including a coalescing of the signal for P2/P4 but on cooling it reverts to the original shifts (Figure S1 and S36 in the supporting information). Quantum chemical calculations show that the formation of 4 from the reaction of 2 and P₄ is exergonic by 73.8 kcal mol⁻¹. The trans isomer of 4 is more favoured by 4.1 kcal mol⁻¹ compared to the cis isomer. When 4 was heated to 80 °C for 2 hours, an intractable mixture of products was observed in the ¹H and ³¹P{¹H} NMR spectra (Figure S33 in the supporting information).

Considering the highly reduced and unsaturated nature of the P₄²⁻ unit in 3-trans/cis and 4, we were curious as to whether further functionalization was possible. Thus, reactions with the electrophiles Me₃SiCl, Me₃SiI, HCl (4 M in dioxane) and CH₃COCl were conducted (Figure 5-6, and Table S1-3 in the supporting information). Treating 3-trans with an excess of Me₃SiX (X=Cl, I) in C₆D₆ at 80 °C gave aluminum dihalides Tripp(IiPr)AlX₂ 5 and P(SiMe₃)₃, which was confirmed by in situ ¹H and ³¹P{¹H} NMR spectroscopy. PH₃ and P(OCCH₃)₃ were formed from the reactions of 3-trans with 4 M HCl in dioxane and CH₃COCl, respectively, at ambient temperature. When 3-cis was used, P(SiMe₃)₃ was formed with high P conversion rate under milder conditions (Table S2 in the supporting information), as well as PH₃ or P(OCCH₃)₃ with a comparable P conversion rate.

The reaction of 4 with Me₃SiX (X=Cl or I) gives comparable conversion rates to 3-trans at lower temperatures (Figure 6). The results were tracked by in situ ¹H and ³¹P{¹H} NMR spectroscopy, showing that P(SiMe₃)₃ and [SitBu₂Me(NHC)AlX₂ (6) were formed. Compound 4 reacted with HCl (4 M in dioxane), and CH₃COCl at room temperature forming PH₃, and P(OCCH₃)₃, respectively. These results demonstrate that the P₄ activation products by dialumenes can release P1 compounds on reaction with electrophiles under mild conditions. Additionally, the by-product LAlX₂ is the precursor to the dialumenes^{14, 15} allowing for the potential regeneration of the starting material.

In conclusion, we have utilised dialumenes ([L(IiPr)Al]₂, L=Tripp (1), or SitBu₂Me (2)) in the activation and subsequent functionalisation of white phosphorus (P₄). Their reactivity towards white phosphorus resulted in the reduction of P₄ to P₄²⁻ across the Al=Al double bond. Depending on the dialumene employed, we were successful in isolating the P-rich alanes in cis or trans configurations. This is the first report to elucidate the mechanism of the reaction of P₄ with a π-bond resulting in cis configuration from trans, supported with DFT calculations. In addition, subsequent reactivity studies with electrophiles demonstrated that both [Tripp(IiPr)Al]₂P₄ 3-trans/cis and [SitBu₂Me(IiPr)Al]₂P₄ 4 acted as a P³⁻ source, giving access to various useful phosphines. The mechanism of the formation of AlP₄Al core is currently under investigation. This is another step forward in the field of main group aided P₄ functionalisation and further reactivity studies of 3-trans/cis and 4 are currently ongoing.