A Highly Reactive Geminal P/B Frustrated Lewis Pair: Expanding the Scope to C−X (X=Cl, Br) Bond Activation

Abstract

The geminal frustrated Lewis pair tBu₂PCH₂B(Fxyl)₂ (1; Fxyl=3,5‐(CF₃)₂C₆H₃) is accessible in 65 % yield from tBu₂PCH₂Li and (Fxyl)₂BF. According to NMR spectroscopy and X‐ray crystallography, 1 is monomeric both in solution and in the solid state. The intramolecular P⋅⋅⋅B distance of 2.900(5) Å and the full planarity of the borane site exclude any significant P/B interaction. Compound 1 readily activates a broad variety of substrates including H₂, EtMe₂SiH, CO₂/CS₂, Ph₂CO, and H₃CCN. Terminal alkynes react with heterolysis of the C−H bond. Haloboranes give cyclic adducts with strong P−BX₃ and weak R₃B−X bonds. Unprecedented transformations leading to zwitterionic XP/BCX₃ adducts occur on treatment of 1 with CCl₄ or CBr₄ in Et₂O. In less polar solvents (C₆H₆, n‐pentane), XP/BCX₃ adduct formation is accompanied by the generation of significant amounts of XP/BX adducts. FLP 1 catalyzes the hydrogenation of PhCH=NtBu and the hydrosilylation of Ph₂CO with EtMe₂SiH.

Introduction

Sterically demanding main group Lewis acids and bases that are unable to neutralize each other through adduct formation (frustrated Lewis pairs, FLPs) can still act synergistically on a third molecule and thereby exhibit reactivity commonly associated with transition metal complexes (e.g., H₂ activation).1, 2, 3, 4, 5, 6, 7 To date, combinations of suitable organophosphines and organoboranes have been by far the most popular FLPs. Adjustment of their chemical behavior is possible through variation of the substituent patterns and/or the bridging unit between the reactive centers. A frequently employed substituent on boron is the C₆F₅ ring; the phosphine fragments often carry tert‐butyl or mesityl groups. Multiple bimolecular (i.e., unbridged) FLPs do exist and are synthetically more conveniently accessible than their monomolecular (i.e., bridged) congeners.1, 2, 3, 4, 5, 6, 7 However, the preorganization of Lewis acidic and basic sites that is achievable through the introduction of a linker can significantly aid in the fine‐tuning of FLP reactivity, and thus makes the additional synthetic effort worthwhile. For example, Erker and co‐workers studied a series of compounds R₂P(CH₂)_nB(C₆F₅)₂ (n=2–4) and found the ethylene‐ and butylene‐bridged species to be active FLPs (e.g., for H₂ cleavage), whereas the propylene derivative showed no indication of typical FLP activity.8, 9, 10, 11, 12, 13, 14

Methylene‐bridged P/B pairs differ fundamentally from the abovementioned C₂‐, C₃‐, and C₄‐linked compounds, because a one‐atom spacer leads to less conformational flexibility of the molecular scaffold and thus to a well‐defined P⋅⋅⋅B distance. Moreover, the degree of intramolecular P/B interaction should be small, because formation of a P−B σ bond would result in a strained three‐membered ring and, in contrast to phosphinoboranes (C₀ species),15, 16 P=B π donation is not possible. Thus, in a geminal P/B FLP, the two reactive sites should be perfectly preoriented for small‐molecule activation.

Our initial attempts at the synthesis of a first geminal P/B FLP relied on the nucleophilic substitution of EtOB(C₆F₅)₂ with tBu₂PCH₂Li.17 However, the successful formation of the methylene bridge was accompanied by a cyclization reaction, during which the phosphorus atom displaced an ortho‐fluorine atom of one of the C₆F₅ groups. The obtained zwitterionic five‐membered heterocycle A is no longer an FLP (Scheme 1).17, 18, 19 Shortly thereafter, Erker et al. used the hydroboration of (C₆F₅)₂PCH=CHMe and (C₆F₅)₂PC≡CMe with HB(C₆F₅)₂ to make (C₆F₅)₂PCH(Et)B(C₆F₅)₂ and (C₆F₅)₂PC(=C(H)Me)B(C₆F₅)₂, respectively.20, 21 These geminal FLPs did not undergo the undesired cyclization reaction, likely because the nucleophilicities of the phosphorus atoms are tamed by their electron‐withdrawing C₆F₅ substituents. In an alternative approach, Slootweg, Lammertsma, and co‐workers avoided cyclization by employing ClBPh₂ instead of EtOB(C₆F₅)₂, thereby synthesizing tBu₂PCH₂BPh₂.22, 23

Scheme 1.

Formation of the zwitterionic heterocycle A from EtOB(C₆F₅)₂ and tBu₂PCH₂Li.

Even though the above P−C−B Lewis pairs proved to be capable of activating a variety of small molecules, we still remained interested in the development of geminal FLPs featuring strongly Lewis acidic and strongly Lewis basic centers. Bearing in mind that the Gutmann acceptor number of B[3,5‐(CF₃)₂C₆H₃]₃ (B(Fxyl)₃) is comparable to that of B(C₆F₅)₃,24 we first developed facile routes to the borane building blocks XB(Fxyl)₂ (X=H, MeO, F, Cl, Br)25 and now report the synthesis of tBu₂PCH₂B(Fxyl)₂ (1; Scheme 2). We further show that 1 is highly reactive toward a broad selection of substrates commonly employed in FLP chemistry. Moreover, unprecedented transformations were observed on treatment of 1 with CX₄ (X=Cl, Br). Depending on the solvent employed, we isolated either the adduct tBu₂P(X)CH₂B(CX₃)(Fxyl)₂ or its formal dihalocarbene‐elimination product tBu₂P(X)CH₂B(X)(Fxyl)₂.

Scheme 2.

Reactions performed with the aim to synthesize the geminal P/B FLP 1. i) n‐Heptane, 16 h, room temperature; ii) C₆D₆, 16 h, room temperature; iii) C₆H₆, 16 h, room temperature; iv) n‐heptane/C₆H₆, 3 h, room temperature.

Results and Discussion

Synthesis of the geminal FLP tBu₂PCH₂B(Fxyl)₂ (1)

Using the protocol published by Slootweg, Lammertsma et al.22 as a guideline, we first tried to prepare tBu₂PCH₂B(Fxyl)₂ (1) by treatment of tBu₂PCH₂Li26, 27 with (Fxyl)₂BCl.25 Unfortunately, the reaction gave a complex mixture of inseparable products; the same result was obtained with (Fxyl)₂BBr as starting material. We therefore switched from tBu₂PCH₂Li to the less nucleophilic tBu₂PCH₂Sn(nBu)₃ (Scheme 2). Even though the reaction with (Fxyl)₂BBr was again not selective, we were able to isolate a few single crystals of 2, the cyclic adduct between our target compound 1 and one equivalent of the borane reactant. We next tested the complementary approach, that is, the combination of tBu₂PCH₂Li with the less electrophilic borane (Fxyl)₂BOMe.25 This reaction furnished 1 as the main product, albeit in the form of its LiOMe adduct 3 (Scheme 2). Addition of Me₃SiCl to a C₆D₆ solution of 3 led to decomposition rather than to the liberation of free 1. (Fxyl)₂BF25 is a similarly mild electrophile to (Fxyl)₂BOMe, but LiF has an exceptionally high lattice energy. Thus, the synthesis of the desired FLP 1 was finally achieved from tBu₂PCH₂Li and (Fxyl)₂BF in 65 % yield (Scheme 2).

The presence of a PCH₂B backbone in compound 1 is confirmed by a doublet at 2.08 ppm (2 H; ² J(H,P)=3.1 Hz) in the ¹H NMR spectrum with ¹H–¹³C HMBC cross‐peaks to the signals of the C(CH₃)₃ groups at P and the B‐aryl ipso‐carbon atoms. Moreover, the CH₂ ¹³C resonance is significantly broadened due to the interaction of the C atom with the quadrupolar ¹¹B nucleus. The triorganoborane28 and ‐phosphine29 moieties give rise to resonances at δ(¹¹B)=63 ppm and δ(³¹P)=25.9 ppm, in accord with an FLP nature of the compound. Correspondingly, the crystal lattice of 1 contains monomeric molecules with intramolecular P⋅⋅⋅B distances of 2.900(5) Å (Figure 1). For comparison, the calculated molecular structures of tBu₂PCH₂B(C₆F₅)₂ in its ring‐opened and ring‐closed forms show P⋅⋅⋅B distances of 2.89 and 2.04 Å, respectively.22 The measured P1‐C1‐B1 angle of 1 is 114.9(3)°, and the sum of angles about the B center is 359.8°. Any significant σ interaction between P and B should lead to compression of the P1‐C1‐B1 angle from the ideal value of 107.5° and to pyramidalization of the B atom, which is not observed in the present case.

Figure 1.

Molecular structure of 1 in the solid state; displacement ellipsoids are drawn at 50 % probability. The disordered CF₃ groups are displayed in only one of two positions. H atoms are omitted for clarity. Selected bond lengths [Å], atom⋅⋅⋅atom distances [Å], and bond angles [°]: P1−C1 1.867(4), B1−C1 1.569(6); P1⋅⋅⋅B1 (intramolecular) 2.900(5), P1⋅⋅⋅B1 (intermolecular) 7.918(5); P1‐C1‐B1 114.9(3), C1‐B1‐C11 121.5(4), C1‐B1‐C21 119.5(4), C11‐B1‐C21 118.8(4).

As a first test of the reactivity of 1, we attempted the targeted syntheses of 2 and 3. Single crystals of the bromoborane adduct 2 (85 %) grew after equimolar solutions of 1 (in n‐heptane) and (Fxyl)₂BBr (in C₆H₆) had been slowly combined at room temperature. The air‐sensitive compound proved to be insoluble in common inert NMR solvents (for the NMR data of a corresponding BBr₃ adduct of 1, see compound 12 b below). However, the constitution of 2 was unequivocally confirmed by X‐ray crystallography (see the Supporting Information for more details).

The addition of solid MeOLi to a solution of 1 in C₆D₆ furnished small amounts of 3 (NMR spectroscopic monitoring). The low conversion is probably due to solubility issues. The ¹¹B NMR spectrum of 3 is characterized by a resonance at 1.5 ppm, typical of tetracoordinate boron species.28 In C₆D₆, the ³¹P{¹H} NMR signal of 3 appears as a 1:1:1:1 quartet with a chemical shift of 22.3 ppm. The quartet collapses to a singlet on addition of THF or H₃CCN to the sample. We therefore attribute the resonance fine structure in neat C₆D₆ to ³¹P–⁷Li coupling (¹ J=88 Hz) and thus to contact ion pairs, which are separated in the presence of coordinating solvent molecules. A cyclic contact ion pair in which the Li⁺ ion is chelated by the P atom and the BOMe moiety is also observed in the solid‐state structure of 3 (see the Supporting Information for more details).

Reactions of 1 with selected substrates

For a thorough assessment of its chemical behavior, compound 1 was treated with 14 different reagents (Scheme 3).

Scheme 3.

Reactions of 1 with selected substrates. i) Reversible at room temperature. ii) Dynamic association/dissociation equilibrium in solution. iii) Et₂O, room temperature. iv) C₆H₆ or n‐pentane, room temperature.

The standard FLP substrate, H₂, reacted in the usual manner2, 3 with activation of the H−H bond (<1 atm, room temperature). Product 4 is characterized by a ³¹P NMR resonance at 60.1 ppm (¹ J(P,H)=444 Hz) and an ¹¹B NMR signal at −10.8 ppm (¹ J(B,H)=88 Hz). The ¹H NMR spectrum shows a doublet of triplets for the PH proton (4.08 ppm), due to coupling with the ³¹P nucleus and the CH₂ bridge protons. The BH proton gives rise to the expected 1:1:1:1 quartet at 2.99 ppm. H₂ addition is not reversible up to a temperature of 120 °C. Nevertheless, the imine PhCH=NtBu can be hydrogenated quantitatively in the presence of catalytic amounts of 4 already at 80 °C (p(H₂)<1 atm, 20 mol % catalyst loading; see ref. 10 for related P/B FLP‐mediated hydrogenation reactions).

Unlike H₂, EtMe₂SiH adds to 1 in a reversible manner at room temperature in C₆D₁₂ solution (the sterically more demanding Et₃SiH does not react at all). According to NMR spectroscopy, the association/dissociation equilibrium shifts toward quantitative formation of the Si−H activation product 5 only if excess EtMe₂SiH is supplied (approximately 10 equiv). The NMR spectra of 5 are consistent with the presence of a hydridoborate ion (δ(¹¹B)=−13.2 ppm; ¹ J(B,H)=82 Hz) and a silylphosphonium ion (δ(²⁹Si)=10.6 ppm; ¹ J(Si,P)=40 Hz). Further proof of the proposed molecular structure was gained by X‐ray crystallography (see the Supporting Information for more details). In contrast to its behavior in solution, crystalline 5 does not tend to lose silane at room temperature, even under dynamic vacuum. Under hydrolytic conditions, the silane adduct 5 cleanly transforms into the H₂ adduct 4.

The reaction between 1 and CO₂, another standard FLP substrate, takes a similar course to the reaction between tBu₂PCH₂BPh₂ and CO₂.22 An almost‐planar, five‐membered, air‐ and moisture‐stable heterocycle with an exocyclic C=O double bond is formed (6). The corresponding ¹³C NMR signal appears at 168.3 ppm, in good agreement with the shift reported for the literature‐known system mentioned above (167.8 ppm). An analogous structure to 6 is obtained from 1 and CS₂ (7). Compound 7 has a red‐purple color, characteristic of phosphine–CS₂ adducts.30, 31, 32 CS₂ activation by P/B Lewis pairs is far less common than CO₂ activation, and the only known examples are the addition of CS₂ to tBu₂PN≡Btmp (Htmp=2,2,6,6‐tetramethylpiperidine)33 and Et₂PC(Ph)=C(nBu)B(nBu₂).34

Whereas aldehydes have already been reported to react with P/B FLPs,12, 35, 36 the Ph₂CO adduct 8 is a rare example of an activated ketone. In a related case, Ph₂CO undergoes a [2+2] cycloaddition with the phosphinoborane tBu₂PBFlu (HBFlu=9‐borafluorene). The primary product then undergoes heterolytic cleavage of the P−B bond to furnish tBu₂PCPh₂OBFlu.16 The room‐temperature ¹H NMR spectrum of 8 shows poorly resolved phenyl resonances. Steric repulsion between the Ph and tBu substituents likely restricts intramolecular motion and/or causes an association/dissociation equilibrium between FLP 1, the ketone, and 8. To clarify this point, we also recorded NMR spectra of 8 at elevated temperatures. The ³¹P NMR signal (84.4 ppm) became severely broadened at 50 °C and completely vanished at 80 °C; similarly, the ¹¹B NMR resonance of 8 (4.9 ppm) was no longer detectable in the high‐temperature spectrum. Both signals reappeared when the sample was cooled back to room temperature. Moreover, the colorless solution of 8 adopts the yellow color of free 1 on heating, but becomes colorless again on cooling. Adduct formation of the FLP with Ph₂CO is thus a reversible dynamic process. Accordingly, compound 8 is hydrolyzed much more readily than compound 6. As a major hydrolysis product, we identified tBu₂P(H)CH₂B[OB(Fxyl)₂](Fxyl)₂ by X‐ray crystallography and NMR spectroscopy (see the Supporting Information for more details). This species is formally derived from (Fxyl)₂BOH by O−H addition to 1.

Geminal FLP 1 efficiently catalyzes the hydrosilylation of Ph₂CO with EtMe₂SiH (12 mol % catalyst loading, room temperature, 30 min, C₆D₆).37 Note that 1 not only interacts with Ph₂CO, but also with EtMe₂SiH (cf. 5), the other reagent of the hydrosilylation sequence.

FLP 1 not only traps compounds containing a C=O bond, but also adds to the C≡N bond of H₃CCN to give the five‐membered cyclic compound 9. The only comparable example of a P/B‐mediated H₃CCN activation was described by Nöth and co‐workers, who again used the species tBu₂PN≡Btmp. At room temperature, they observed kinetically controlled formation of the imine fragment PC(CH₃)=NB. On thermal treatment, the imine tautomerized to the thermodynamically preferred enamine PC(=CH₂)N(H)B.33 In the case of 9, we found a proton resonance at 1.88 ppm (d, ³ J(H,P)=4.9 Hz) with an integral of 3 H, assignable to a CH₃ group. The corresponding ¹³C NMR signal was observed at 26.5 ppm (d, ² J(C,P)=47 Hz). The molecular structure of 9 in the solid state shows an endocyclic C−N distance of 1.258(10) Å and an exocyclic C−C distance of 1.505(10) Å, which are typical values of C=N bonds38 and C(sp²)−C(sp³) single bonds,39 respectively. We therefore conclude that 9 is the imine rather than the enamine tautomer. In contrast to the adduct of Nöth et al., 9 is thermally stable up to 120 °C.

Reactions of P/B FLPs with terminal alkynes are governed by the basicity of the phosphine: FLPs containing less basic phosphines tend to add to the C≡C bond, whereas the use of strongly basic phosphines (e.g., tBu₃P) results in deprotonation of the alkyne to give phosphonium alkynylborate salts.40 Accordingly, 1 cleaves the terminal C−H bonds of Me₃SiCCH and PhCCH with generation of 10 a and 10 b, respectively. Phosphine protonation is evidenced by doublets of multiplets at about 53 ppm in the ³¹P NMR spectra with ¹ J(P,H) coupling constants of 450 Hz. The corresponding ¹H resonances appear at about 5 ppm as doublets of triplets (¹ J(H,P)=450 Hz, ³ J(H,H)=4.5 Hz). ¹¹B NMR signals are observed at −14.5 ppm. As a further characteristic, the BC≡C signals are broadened beyond detection in the ¹³C{¹H} NMR spectrum. A ¹H–¹³C HMBC experiment, however, revealed chemical shifts of 131.9 ppm (10 a) and 109.8 ppm (10 b). The proposed molecular structures of 10 a and 10 b were further corroborated by X‐ray crystallography (see the Supporting Information).

Stephan and co‐workers trapped N₂O with a bimolecular P/B FLP to obtain tBu₃PN=NOB(C₆F₅)₃.41 Although kinetically stable, the compound loses N₂ with formation of the phosphine oxide adduct tBu₃P=OB(C₆F₅)₃ on photolysis or heating to 135 °C. In contrast, the intramolecular phosphine oxide adduct 11 was already generated when an n‐pentane solution of 1 was stored under N₂O at 4 °C in the dark. The ¹¹B NMR resonance of 11 appears at 7.5 ppm and thus in the typical shift range of tetracoordinate boron nuclei.28 Compared to the ³¹P{¹H} NMR resonance of 1 (25.9 ppm), the signal of 11 is shifted to lower field (113.1 ppm). In the solid state, 11 has a P=O bond length of 1.576(2) Å and a B−O bond length of 1.612(3) Å. Both these bonds are significantly longer than those of the related intramolecular adduct tBu₂P(μ‐O)(μ‐C₆H₄)B(C₆F₅)₂ featuring a five‐membered heterocycle (P=O 1.546(2), B−O 1.550(2) Å).42

The serendipitous finding of the (Fxyl)₂BBr adduct 2 drew our attention to the possibility of trapping BCl₃ and BBr₃, too. Previously Uhl and co‐workers prepared cyclic adducts between BX₃ (X=F, Cl, Br, I) and the P/Al FLP Mes₂PC[=C(H)Ph]AltBu₂.23a Interestingly, the products with X=F and Cl proved to be thermally stable and could be stored at room temperature, whereas the adducts with X=Br and I decomposed above 0 °C.23a In the case of FLP 1, both the BCl₃ adduct 12 a and the BBr₃ adduct 12 b are isolable under ambient conditions. We did not observe any signs of substituent scrambling between the two B atoms of 12 a or 12 b. BX₃ binding results in downfield shifts of the ³¹P NMR resonances from 25.9 ppm in free 1 to 39.4 and 38.6 ppm in 12 a and 12 b, respectively (broadened 1:1:1:1 quartets). In turn, the FLP ¹¹B NMR signals experience an upfield shift from 63 ppm (1) to 35 ppm (12 a) or 34 ppm (12 b), attributable to a certain degree of intramolecular X−B coordination. Likely due to magnetic anisotropy effects,28 the ¹¹B NMR chemical shifts of the trihalogenated boron atoms differ by as much as 17.4 ppm between 12 a (7.2 ppm, ¹ J(B,P)=150 Hz) and 12 b (−10.2 ppm, ¹ J(B,P)=140 Hz). Adducts 12 a and 12 b both crystallize from n‐alkanes in the monoclinic space group P2₁/c (see Figure 2 for a plot of the molecular structure of 12 b). The P−BX₃ bond lengths of 12 a and 12 b are identical (2.002(2) Å versus 2.000(6) Å). In each molecule, the B1−X distance is remarkably longer than the B2−X distance (12 a: B1−Cl1 2.361(3), B2−Cl1 1.925(2) Å; 12 b: B1−Br1 2.408(7), B2−Br1 2.093(6) Å). By the same token, the B1 atoms are much less pyramidalized than the corresponding trihalogenated B2 atoms [sums of angles around boron: 12 a: 352° (B1), 328° (B2); 12 b: 350° (B1), 326° (B2)]. We therefore conclude that 12 a and 12 b are essentially phosphine adducts of BCl₃ and BBr₃ with additional weak interactions between the FLP B centers and the bridging halogen atoms.

Figure 2.

Molecular structure of 12 b in the solid state; displacement ellipsoids are drawn at 50 % probability. The disordered CF₃ groups are displayed in only one of two positions. H atoms are omitted for clarity. Selected bond lengths [Å] and angles [°]: P1−B2 2.000(6), B1−Br1 2.408(7), B2−Br1 2.093(6), B2−Br2 1.980(6), B2−Br3 1.990(6); C1‐B1‐C11 118.8(4), C1‐B1‐C21 114.5(5), C11‐B1‐C21 116.7(5), Br1‐B2‐Br2 109.3(3), Br1‐B2‐Br3 106.3(3), Br2‐B2‐Br3 110.4(3).

Combinations of Lewis acids and bases (usually AlCl₃ with amines) are known to facilitate the electrophilic borylation of arenes by boron halides. These reactions can be performed with a broad variety of aromatic compounds and most often involve borenium salts, such as [Cl₂B(amine)]⁺[AlCl₄]⁻, as the actual borylating agents.43, 44, 45, 46, 47, 48, 49, 50 On thermal treatment, the BX₃ adducts 12 a and 12 b could conceivably undergo B−X heterolysis with formation of borenium species tBu₂P(BX₂ ⁺)‐ CH₂(X⁻)B(Fxyl)₂. We therefore examined the reactivity of 12 b toward electron‐rich o‐xylene in C₆D₆. According to NMR spectroscopy, no conversion occurred at 60 °C (4 h) or 100 °C (1 h). Maintaining a temperature of 100 °C for 16 h led to quantitative decomposition of the FLP scaffold, while o‐xylene remained inert. We attribute this result to one of the following factors: 1) Phosphine‐supported borenium cations51, 52 may be less active borylating agents than their amine‐supported congeners. 2) Due to the high fluorophilicity of borenium electrophiles, the presence of CF₃ groups in the FLP could effect unwanted side reactions. Indeed, the thermolized sample gave rise to a prominent broad ¹¹B NMR signal at 24 ppm, which lies in a similar range to the ¹¹B resonances of FBBr₂ (30 ppm) and F₂BBr (20 ppm).28 3) As discussed above, the interaction between the (Fxyl)₂B moiety and the BBr₃ bromine atom in 12 b may be too weak to induce B−Br bond heterolysis.

FLP 1 was unable to heterolytically cleave the B−X bond of BX₃ and form a phosphine‐coordinated borenium/haloborate ion pair. Yet, 1 readily splits the C−Br bond of PhCH₂Br to afford the benzylphosphonium bromoborate zwitterion 13. The ¹¹B NMR signal of compound 13 (−0.9 ppm) appears at considerably higher field relative to the corresponding resonance of 12 b (34 ppm). Accordingly, the B−Br bond length of 13 (2.16(2) Å) is shorter by 0.25 Å than the B1⋅⋅⋅Br1 distance in 12 b.

Compared to the latter conversion, which took the expected course, the outcome of the reaction between 1 and CBr₄ is less predictable. Given the considerable stability of the [CBr₃]⁻ ion,53, 54 abstraction of Br⁺ from CBr₄ by the phosphine site (cf. the Corey–Fuchs reaction55) and immediate trapping of [CBr₃]⁻ by the boron center offers a conceivable alternative to the tribromomethylation of the phosphorus atom. Therefore, we finally investigated the behavior of 1 toward CBr₄ and also included CCl₄ in our study (cf. the Appel reaction56). Addition of CX₄ (X=Cl, Br) to 1 in Et₂O indeed provided the C−X‐activated species 14 a and 14 b, featuring halophosphonium ions in combination with trihalomethanide‐coordinated boron atoms. Single crystals were grown at 4 °C (14 a) or room temperature (14 b). Both compounds are remarkably stable at room temperature in the solid state and in ethereal solutions; even in undried THF, they are not hydrolyzed. Moreover, they do not undergo rearrangement reactions, such as the Matteson homologation.57 NMR spectra were recorded in [D₈]THF. The ³¹P chemical shifts of 14 a (129.0 ppm) and 14 b (122.3 ppm) are similar, although the molecules contain different halogen substituents. The ¹¹B NMR resonances appear in the typical region of tetracoordinate boron nuclei, that is, −4.8 ppm (14 a) and −4.1 ppm (14 b). The CX₃ carbon atoms attached to boron are not detectable in the 1D ¹³C{¹H} NMR spectrum, likely due to quadrupolar broadening. Their chemical shifts were therefore determined from cross‐peaks with the CH₂ proton signals in ¹H–¹³C HMBC NMR spectra. We found values of 113.7 (14 a) and 76.2 ppm (14 b), which are intermediate between those of LiCX₃ [146 ppm (Cl); 101 ppm (Br)] on the one hand and HCX₃ [80 ppm (Cl); 14 ppm (Br)] on the other.54 These NMR features nicely reflect the fact that the covalent character of the B−C bonds lies between those of Li−C and H−C bonds.

Compounds 14 a and 14 b are isostructural in the solid state. Thus, only the molecular structure of 14 b is discussed here (Figure 3; see the Supporting Information for more details of that of 14 a). Contrary to all other open‐chain adducts of 1, 14 b adopts a B1−C1 s‐trans conformation (P1‐C1‐B1‐C10 178.3(3)°). The P1−Br1 bond length is 2.174(1) Å, and the B1−C10 (1.688(6) Å) and B1−C1 bonds (1.692(6) Å) have essentially the same lengths. The CBr₃ fragment is fully pyramidalized with Br−C10−Br bond angles ranging between 104.8(2) and 105.6(2)°.

Figure 3.

Molecular structure of 14 b in the solid state; displacement ellipsoids are drawn at 50 % probability. H atoms are omitted for clarity, the Fxyl and tBu groups are represented by the C atoms attached to the reactive centers. Selected bond lengths [Å], bond angles [°], and torsion angle [°]: P1−Br1 2.174(1), B1−C1 1.692(6), B1−C10 1.688(6); P1‐C1‐B1 130.0(3), C1‐B1‐C10 101.8(3); P1‐C1‐B1‐C10 178.3(3).

The addition of CX₄ to 1 in Et₂O gives 14 a or 14 b as the sole products. Yet, less polar solvents, such as C₆H₆ and n‐pentane, effect a different result: alongside each CX₄ adduct, a second species is generated in an approximately equimolar quantity. These compounds were identified as the formal X₂ adducts 15 a (X=Cl) and 15 b (X=Br) by NMR spectroscopy and X‐ray crystallography (we note in this context that attempts to synthesize 15 b directly from 1 and Br₂ failed). Compounds 15 a and 15 b are likely formed because dihalocarbene extrusion from [CX₃]⁻ successfully competes with boron coordination of the anion under these conditions.

The differences in the 1D NMR spectra of 15 a/15 b compared to 14 a/14 b are surprisingly small and therefore not very diagnostic. More information regarding the chemical constitution of 15 a and 15 b can be gained from the 2D NMR spectra: the ¹H–¹³C HMBC cross‐peaks observed between the CH₂ proton signals and the CX₃ carbon resonances in the cases of 14 a and 14 b are absent in the spectra of 15 a and 15 b. Definite proof for the postulated structures of 15 a and 15 b stems from X‐ray crystallography, which clearly identified the two compounds as formal Cl₂ and Br₂ adducts. As in the cases of 14 a and 14 b, the molecular structures of 15 a and 15 b are rather similar, and we therefore restrict ourselves to the discussion of that of 15 b (Figure 4; see the Supporting Information for more details of that of 15 a). As expected, the P1−Br2 bond length of 15 b (2.167(2) Å) is virtually the same as that of 14 b (2.174(1) Å). In turn, the B1−Br1 bond length (2.135(8) Å) agrees with that of 13 (2.16(2) Å). Br1 and Br2 approach each other rather closely, such that the Br1⋅⋅⋅Br2 distance (3.581(1) Å) becomes shorter than the sum of the van der Waals radii of two Br atoms (3.8 Å).38

Figure 4.

Molecular structure of 15 b in the solid state; displacement ellipsoids are drawn at 50 % probability. H atoms are omitted for clarity, the Fxyl and tBu groups are represented by the C atoms attached to the reactive centers. Selected bond lengths [Å], atom⋅⋅⋅atom distance [Å], and bond angle [°]: P1−Br2 2.167(2), B1−Br1 2.135(8); Br1⋅⋅⋅Br2 3.581(1); P1‐C1‐B1 127.7(6).

Finally, we note that 15 b was also obtained (albeit in low yields) from the reaction between 1 and HCBr₃ in n‐pentane, whereas 1 did not activate H₂CBr₂, HCCl₃, or H₂CCl₂ (in n‐pentane or in the respective neat halomethane).

Conclusion

The length of the bridging unit in a monomolecular FLP greatly influences the chemical behavior. The bridge governs the conformational flexibility of the FLP scaffold, the ring size of transition states during small‐molecule activation, and the charge separation and dipole moment in the activation products. Thus, geminal FLPs should be particularly reactive, but only a few examples have been reported until now. Especially the combination of highly Lewis acidic boranes and highly basic phosphines in methylene‐bridged P/B FLPs is synthetically challenging: commonly used C₆F₅ boranes readily undergo o‐F substitution by the phosphine to form zwitterionic five‐membered rings containing tetracoordinate B and P atoms.

Recently, the (Fxyl)₂B (Fxyl=3,5‐(CF₃)₂C₆H₃) building block became available as an alternative to the (C₆F₅)₂B moiety. This granted us access to the geminal FLP tBu₂PCH₂B(Fxyl)₂ (1), which features a strong Lewis base and a strong Lewis acid. Compound 1 does not show any indications of P⋅⋅⋅B interaction in solution or in the solid state and can therefore be regarded as a genuine FLP. We have shown that 1 readily reacts with all standard FLP substrates, including H₂, EtMe₂SiH, CO₂/CS₂, Ph₂CO, and H₃CCN. Most importantly, 1 activates certain alkyl halides, such as CCl₄, CBr₄, and HCBr₃, through heterolysis of the C−X bonds. In this way, unprecedented X₃C borates were isolated and structurally characterized. We are currently investigating the suitability of such X₃C borates for the introduction of X₃C substituents into organic molecules through Suzuki‐type C−C coupling reactions.

Supporting information

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

K. Samigullin, I. Georg, M. Bolte, H.-W. Lerner, M. Wagner, Chem. Eur. J. 2016, 22, 3478.