Metalate-Mediated Functionalization of P4 by Trapping Anionic [Cp*Fe(CO)2(η1-P4)]- with Lewis Acids.

The development of selective functionalization strategies of white phosphorus (P4) is important to avoid the current chlorinated intermediates. The use of transition metals (TMs) could lead to catalytic procedures, but these are severely hampered by the high reactivity and unpredictable nature of the tetrahedron. Herein, we report selective first steps by reacting P4 with a metal anion [Cp*Fe(CO)2]- (Cp*=C5(CH3)5), which, in the presence of bulky Lewis acids (LA; B(C6F5)3 or BPh3), leads to unique TM-substituted LA-stabilized bicyclo[1.1.0]tetraphosphabutanide anions [Cp*Fe(CO)2(η1-P4⋅LA)]-. Their P-nucleophilic site can be subsequently protonated to afford the transient LA-free neutral butterflies exo,endo- and exo,exo-Cp*Fe- (CO)2(η1-P4H), allowing controllable stepwise metalate-mediated functionalization of P4.

Organophosphorus compounds (OPCs) play a crucial role in synthetic chemistry,1 but are typically produced through large‐scale halogenation of white phosphorus (P4 → PCl3) and subsequent salt elimination, generating equimolar halogenide waste. However, this is avoidable by direct functionalization of the P4 tetrahedron. This desirable avenue has been scrutinized for both main‐group compounds2 and transition‐metal (TM) complexes, ultimately in search of catalytic conversions.3 The challenge is to control the unpredictable reactivity of P4. To date, steps invoking reducing neutral metal complexes have been reported by, for example, the groups of Scheer (Fe),4 Driess (Fe,5 Co,6 and Ni),7 and Cummins (Nb)8 and cations by Peruzzini and co‐workers (Fe,9 Ru,10 Rh, and Ir).11 However, in spite of its inherent electrophilic character, reactions of P4 with TM anions have hardly been considered.

In 2002, Ellis and co‐workers provided the first insights by reacting P4 with a naphthalene‐stabilized titanate that afforded the all‐inorganic metallocene [(η5‐P5)2Ti]2− (A; Scheme 1).12 Later, Wolf and co‐workers described the formation of iron polyphosphides B and C through P4‐aggregation induced by anionic [Cp*Fe(η4‐C10H8)]− (Cp*=C5(CH3)5),13 and recently the preparation of dinuclear cobalt tetraphosphido complex D by reaction of P4 with [Co(BIAN)(cod)]− (BIAN=1,2‐bis(2,6‐diisopropylphenylimino)‐acenaphthene, cod=1,5‐cyclooctadiene).14 The group of Ruiz reported the fragmentation of P4 into P2 (E), facilitated by the triply bonded molybdenate [Mo2Cp2(μ‐PCy2)(μ‐CO)2]− (Cp=C5H5).15, 16

Scheme 1

Anionic metal‐mediated P4 functionalization products (counter cations omitted). Cp=C5H5, Cp*=C5(CH3)5, BIAN=1,2‐bis(2,6‐diisopropylphenylimino)acenaphthene.

The marked unpredictability when using metal anions is reminiscent to the often uncontrolled reactions of P4 with carbanions.17 For this, we developed a selective functionalization strategy by trapping the initial P4 adduct (using sterically encumbered ArylLi) with Lewis acids to give stabilized bicyclo[1.1.0]tetraphosphabutane anions ([ArylP4⋅LA]−; LA=B(C6F5)3 or BPh3)) that can be substituted or fragmented to novel OPCs containing P4, P3, and P1 units.18 Herein, we show this approach to also allow the [Cp*Fe(CO)2]− anion to functionalize P4 in a controlled manner, providing the first examples of LA‐stabilized TM(η1‐P4)− butterfly anions (1 a and 1 b; Scheme 1), and report on their P‐nucleophilicity by protonation experiments.

To selectively generate anionic TM(η1‐P4)− from P4, we reasoned that a stable, bulky metalate with well‐defined nucleophilic character would be required, for which the readily available Li[Cp*Fe(CO)2] was considered a good candidate.19 To capture the incipient phosphide, we opted for B(C6F5)3 as strong LA. Indeed, addition of a solution of P4 in toluene to a cooled (0 °C) mixture of Li[Cp*Fe(CO)2] and B(C6F5)3 instantaneously gave the novel LA‐stabilized bicyclo[1.1.0]tetraphosphabutanide 1 a (δ 31P{1H}: −65.0 (P1), −107.1 (P4), −340.7 (P2/P3) ppm), which could be isolated in 55 % yield as a dark yellow powder (Scheme 2).

Scheme 2

Synthesis of Lewis‐acid‐stabilized [Cp*Fe(CO)2(η1‐P4)]− butterfly anions. Solvent: toluene for 1 a and THF for 1 b.

Crystals of 1 a, suitable for single‐crystal X‐ray diffraction, were grown from Et2O after slow (1 min) addition of 12‐crown‐4. The molecular structure revealed the unprecedented metalphosphido‐borane (Figure 1; [Li(12‐crown‐4)2]+ counter cation omitted) with a bicyclic P4 core (P1−P2−P3−P4 98.21(6)ο) showing a slightly shorter transannular P2−P3 bond (2.1676(13) Å) compared to the peripheral P1−P2/P1−P3 (2.2324(13)/2.2100(12) Å) and P4−P2/P4−P3 (2.2091(13)/2.2252(13) Å) bonds, as is common for P4 butterfly‐type derivatives.18, 20, 21 Compound 1 a features a non‐symmetric substitution pattern with the tetraphosphide unit being flanked by the Cp*Fe(CO)2 moiety (Fe1−P1 2.3192(11) Å) and the B(C6F5)3 (P4−B1 2.080(4) Å) Lewis acid. The “Lewis” bond is marginally longer than that found in the related organyl‐substituted anion Li[Mes*P4⋅B(C6F5)3] (P‐B 2.064(2) Å),18a whereas the coordination bond connecting the iron complex is shorter than that observed in the neutral symmetric [{Cp“′(CO)2Fe}(μ,η1:1‐P4)] (Fe‐P 2.3552(19) Å).20a

Figure 1

Molecular structure of 1 a − in the crystal (displacement ellipsoids are set at 30 % probability; H atoms, [Li(12‐crown‐4)2]+ counter cation and non‐coordinated Et2O molecules are omitted for clarity). Selected bond lengths [Å] and angles [ο]: P1−P2/P3 2.2324(13)/2.2100(12), P4−P2/P3 2.2091(13)/ 2.2252(13), P2−P3 2.1676(13), Fe1−P1 2.3192(11), P4−B1 2.080(4), C1−O1 1.151(5); P1−P2−P3−P4 98.21(6).

Having established the formation of the [Cp*Fe‐ (CO)2(η1‐P4)]− butterfly anion, we wondered whether the weaker Lewis acid BPh3 would, likewise, enable its isolation.22 Indeed, addition of Li[Cp*Fe(CO)2] to a THF solution of P4 and BPh3 at 0 °C afforded 1 b (δ 31P{1H}: −46.6 (P4), −84.3 (P1), −337.0 (P2/P3) ppm), which was isolated as a brown powder in 86 % yield (Scheme 2). Notably, the Li[Cp*Fe(CO)2]/BPh3 combination in THF results in a higher yield than Li[Cp*Fe(CO)2]/B(C6F5)3 in toluene (86 % and 55 %, respectively), likely owing to better solubility of the anions in THF. Its use as a solvent for the synthesis of 1 a, however, is precluded, owing to formation of the reactive THF⋅B(C6F5)3 adduct.23

Phosphides 1 a and 1 b are the first examples of isolable TM‐generated P4 butterfly anions. ωB97X‐D/6–311+G(2d,p)//6‐31G(d) calculations (Scheme 3) revealed the nucleophilic addition of [Cp*Fe(CO)2]− to P4 to cause exothermic cleavage (ΔE=−15.6 kcal mol−1) of one P−P bond. The resulting [Cp*Fe‐ (CO)2(η1‐P4)]− “butterfly” anion is stabilized by a significant −59.5 kcal mol−1 as the B(C6F5)3 adduct (1 a) and by −35.1 kcal mol−1 (ΔE) with the weaker bonding BPh3 (1 b). The stronger bond with B(C6F5)3, as compared to BPh3, is reflected in the observed 19.3 ppm downfield shift of the 31P{1H} NMR resonance for the P1 atom (1 a vs. 1 b), owing to the larger electron‐withdrawing effect of the fluorinated triarylborane. This difference in inductive effect also leads to a weaker Fe1−P1 bond in the B(C6F5)3 adduct. ETS–NOCV analysis revealed a smaller total bonding energy to that in the BPh3 adduct (ΔΔE=13.7 kcal mol−1)24 and showed a lower contribution for σ donation in the orbital interaction terms (1 a: −101.6; 1 b: −107.0 kcal mol−1; π backdonation 1 a: −10.6; 1 b: −10.9 kcal mol−1).25 The highest occupied molecular orbital (HOMO) reflects the lone pair on the boron‐coordinated wing‐tip P atom and is expectedly lower in energy for 1 a (−0.17 eV) compared to 1 b (−0.15 eV), suggesting P‐nucleophilic character for both.

Scheme 3

Relative ωB97X‐D/6–311+G(2d,p)//6‐31G(d) energies (in kcal mol−1) for the formation of anionic 1 a and 1 b and computed HOMOs (contour value=0.05). [Fe]=Cp*Fe(CO)2.

To probe the utility of anions 1 a and 1 b as nucleophilic reagents, protonation experiments were performed by using the mild acid [Me3NH][BPh4] in THF.26 Addition of the acid to the most reactive phosphide 1 b (1:1 stoichiometry) showed, in the 31P{1H} NMR spectrum, the instantaneous formation of two new bicyclo[1.1.0]tetraphosphabutanes, which were identified as the neutral protonated LA‐free exo,endo and exo,exo isomers of Cp*Fe(CO)2(η1‐P4H) (2, 1:1.2 ratio; Scheme 4). Simulation of the 31P NMR resonances27 confirms the expected AMX2 spin systems for exo,endo‐2 (δPA 70.6, δPM −41.8, δPX −335.5 ppm; 1 J PA,PX=−194.5, 1 J PM,PX=−198.4, 2 J PA,PM=27.6 Hz) and exo,exo −2 (δPA −19.8, δPM −226.7, δPX −355.2 ppm; 1 J PA,PX=−166.4, 1 J PM,PX=−149.9, 2 J PA,PM=243.3 Hz). The isomers could be distinguished through the difference in the 2 J PA,PM (ΔJ=215.7 Hz) and 1 J P,H (145.9 Hz endo‐PH; 109.3 Hz exo‐PH) coupling constants. The 1H NMR spectrum showed a resonance for only the endo‐PH isomer (−1.14 ppm; 1 J H,P=152.9 Hz). The 11B{1H} NMR spectrum revealed two signals at 27.3 and −8.4 ppm, attributed to the amine–borane adduct Me3N⋅BPh3 in equilibrium with its constituents and Li[BPh4]. The phosphanes decompose within 24 h, owing to a lack of steric protection, which is common for neutral bicyclic tetraphosphanes bearing “small” substituents.18a,18c

Scheme 4

Top: Protonation of 1 a and 1 b. [Fe]=Cp*Fe(CO)2, LA=B(C6F5)3 (1 a) or BPh3 (1 b). Bottom: 31P{1H} NMR spectrum (162.0 MHz, [D8]THF, 297 K) recorded directly after mixing 1 b and [Me3NH][BPh4]. Insets show expanded experimental and simulated27 (inverted) regions. The resonance signals marked with an asterisk (*) correspond to unidentified side products.

Protonation of the less reactive 1 a, likewise, gave a mixture of exo,endo‐2 and exo,exo‐2 (1:1.2 ratio) with Me3N⋅B(C6F5)3 (δ 11B{1H}=−3.0 ppm) as the sole byproduct. We resorted to DFT calculations to obtain more insight into the product formation from the different precursors 1 a and 1 b (Scheme 5).28 Protonation of the anions by Me3NH+ to give 1 aH or 1 bH was calculated to be quite exothermic (ΔE=−76.8 and −89.0 kcal mol−1, respectively), as expected.29 Subsequent cleavage of the exo‐cyclic P−B bonds by the liberated NMe3 is driven by the formation of the amine–borane adduct and gives exo,endo‐2. This reaction is more exothermic for BPh3 (b, ΔE=−7.3 kcal mol−1) than B(C6F5)3 (a, ΔE=−0.2 kcal mol−1). The exo,exo‐2 isomer was computed to be almost equally stable (ΔE=−0.2 kcal mol−1) and is likely formed experimentally through Lewis or Brønsted acid enhanced isomerization30 in light of the high trigonal and turnstile inversion barriers of 53.7 and 58.3 kcal mol−1, respectively.31

Scheme 5

Relative ωB97X‐D/6–311+G(2d,p)//6‐31G(d) energies (in kcal mol−1) for the formation of 2. [Fe]=Cp*Fe(CO)2.

The selective protonation at the wing‐tip P atoms of the complexed P4 anions 1 a and 1 b confirms their P‐nucleophilic character and provides a simple route to hitherto scarce non‐symmetrical neutral TM‐complexed P4 derivatives. The reactivity is analogous to the organyl‐substituted congeners and should, therefore, be extendable to alkylations and possibly [3+1] fragmentations, on which we reported recently.18c The present report lies the foundation for the isolation of new TM‐mediated P4‐functionalized products.

In conclusion, reacting anionic Li[Cp*Fe(CO)2] with P4 in the presence of either the B(C6F5)3 or BPh3 Lewis acid provides facile access to unique metal‐substituted bicyclo[1.1.0]tetraphosphabutanide anions. Their P‐nucleophilic site can be protonated, affording the novel transient LA‐free tetraphosphanes exo,endo‐ and exo,exo‐Cp*Fe(CO)2(η1‐P4H). The controlled and selective formation of these intriguing new anionic and neutral derivatives enables the selective functionalization of white phosphorus by anionic metalates to be explored.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.