B–B vs. B–H Bond Activation in a (μ‐Hydrido)diborane(4) Anion upon Cycloaddition with CO₂, Isocyanates, or Carbodiimides

Abstract

The intriguing (μ‐hydrido)diboranes(4) with their prominent pristine representative [B₂H₅]⁻ have mainly been studied theoretically. We now describe the behavior of the planarized tetraaryl (μ‐hydrido)diborane(4) anion [1H]⁻ in cycloaddition reactions with the homologous series of heterocumulenes CO₂, iPrNCO, and iPrNCNiPr. We show that a C=O bond of CO₂ selectively activates the B−B bond of [1H]⁻, while the μ‐H ligand is left untouched ([2H]⁻). The carbodiimide iPrNCNiPr, in contrast, neglects the B−B bond and rather adds the B‐bonded H⁻ ion to its central C atom to generate a formamidinate bridge across the B₂ pair ([3]⁻). As a hybrid, the isocyanate iPrNCO combines the reactivity patterns of both its congeners and gives two products: one of them ([4H]⁻) is related to [2H]⁻, the other ([5]⁻) is an analog of [3]⁻. We finally propose a mechanistic scenario that rationalizes the individual reaction outcomes and combines them to a coherent picture of B–B vs. B–H bond activation.

Unprecedented cycloaddition reactions of two isomeric (hydrido)diborane(4) anions with a homologous heterocumulene series proceed via B–B or B–H bond activation. Our diborane starting materials exhibit sterically exposed BHB cores, allowing us to contribute experimental evidence to the body of theoretical work on this class of [B₂H₅]⁻ anion analogs.

Introduction

The activation of C=E double bonds in (cyclo)addition reactions is the key step of countless organic syntheses (E=CR₂, NR, O). In the cases of corresponding C−E′ single bonds, however, more forcing reaction conditions and/or the addition of transition‐metal catalysts are typically required (E′=CR₃, NR₂, OR). Practically useful examples of single‐bond activations have been reported predominantly for C−X (X=Cl, Br, I) or arene C−H bonds. ^[1]

Due to the smaller nuclear attraction potential of the boron atom compared to the carbon atom, the bonding orbitals of B−H and B−B single bonds are radially more extended and energetically higher than those of their C−H and C−C analogs.[ ² , ³ ] This leads to increased reactivity, so that, for example hydroboration reactions can already occur spontaneously at room temperature.[ ⁴ , ⁵ ] As an important additional feature, boron atoms can easily expand their coordination numbers from c.n.=3 to 4. Consequently, diboranes containing electron‐precise B−B bonds are known with B(sp²)–B(sp²), B(sp²)–B(sp³), and B(sp³)–B(sp³) cores. Such structural diversity allows for a fascinating variety of chemical conversions, as adduct formation does not per se have either a stabilizing or destabilizing influence on a diborane scaffold: The vacant B(p_z) orbitals of diboranes(4) render their B−B bonds kinetically vulnerable. Coordination of a Lewis base can lead to kinetic protection, but, on the other hand, induces rehybridization of the boron center, thereby lengthening the B−B bond and again making it susceptible to cleavage. The ambiguity of Lewis‐base coordination to diboranes(4) becomes immediately apparent when bis(pinacolato)diboron (B₂pin₂) is compared with parent B₂H₄: (i) While free B₂pin₂ is stable under ambient conditions, addition of KOtBu leads to heterolytic cleavage of the B−B bond in presence of suitable substrates, which is widely exploited for borylation and diboration reactions. ^[6] (ii) Free B₂H₄ can only be investigated by low‐temperature matrix spectroscopy (3 K; solid neon) and partially compensates for its electron deficiency by forming a doubly H‐bridged C _2v structure.[ ⁷ , ⁸ , ⁹ ] Various ligands proved to be suitable for the preparation of diadducts (L)H₂B–BH₂(L) that are sufficiently stable for X‐ray crystallography and further derivatizations. ^[10] Obviously, the chemistry of B–B two‐electron‐two‐center (2e2c) bonds is far from being well explored or fully understood and still holds many surprises. ^[11] Major drawbacks for further studies include a lack of efficient synthesis protocols and the dilemma that sterically demanding or π‐donating substituents, which are often mandatory to obtain isolable compounds, inevitably impair also desired diborane‐substrate interactions.

Conceptually, our group considers (μ‐hydrido)diborane(4) anions as an ideal class of compounds to study the reactivity of B−B single bonds. The bridging H⁻ ion is the least sterically demanding ligand available to saturate both B(p_z) orbitals simultaneously. Concurrently, H⁻ can easily shift to a terminal position, thereby exposing one tricoordinated boron center to nucleophilic attack. For the parent system [B₂H₅]⁻, some knowledge has already been gained from experiments and quantum‐chemical calculations: the anion was generated in an argon matrix at 8 K and its B−B bond was identified as a strong non‐classical electron‐donor site.[ ¹² , ¹³ ] In an extreme view, [B₂H₅]⁻ can be described as a protonated diborene, H⁺/[H₂B=BH₂]²⁻, which further underscores the pronounced donor potential of the B−B bond.[ ¹⁴ , ¹⁵ ] Despite their high promise for the discovery of fundamentally important reactivity patterns, isolable derivatives of [B₂H₅]⁻ are virtually unknown, the only exception being Tamao's molecule, which bears two bulky 1,1,3,3,5,5,7,7‐octaethyl‐s‐hydrindacen‐4‐yl (Eind) substituents.[ ¹⁶ , ¹⁷ ]

We recently disclosed the synthesis of the tetraaryl (μ‐hydrido)diborane(4) anion [1H]⁻, which possesses a sterically exposed BHB three‐membered ring. Its alkali‐metal salts M[1H] (M⁺=Li⁺–K⁺) are available in near quantitative yields through deprotonation of the tetraaryl diborane(6) 1H₂[ ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ ] using sterically demanding Brønsted bases (LiC(SiMe₃)₃, NaN(SiMe₃)₂, KN(SiMe₃)₂; Scheme 1). ^[15]

Scheme 1.

Synthesis of organic [B₂H₅]⁻ derivatives M[1H] through deprotonation of the B₂H₆ derivative 1H₂; carbon atoms marked with asterisks bear tBu substituents. (i) THF, room temperature.

Herein, we describe the highly selective reaction of M[1H] with the abundant C₁ building block CO₂ and with its analogs iPrNCO and iPrNCNiPr. This substrate scope allows us to systematically assess how the steric demand of the heterocumulene together with subtle changes in its electronic structure influence the reaction pattern of the BHB core.

Results and Discussion

According to quantum‐chemical calculations, [1H]⁻ with its μ‐H‐supported B−B bond is the thermodynamically most favorable among several conceivable isomers. ^[5] Only 5.1 and 6.6 kcal mol⁻¹ higher in energy lie the isomeric minimum structures [1 aH]⁻ and [1 bH]⁻, in which two formal Wagner–Meerwein rearrangements have converted the dibenzo[g,p]chrysene‐type scaffold of [1H]⁻ into the 9,9′‐difluorene‐type frameworks of [1 aH]⁻ and [1 bH]⁻ (Figure 1 a). While in [1 aH]⁻ the hydride substituent is located at a terminal position, it remains in a bridging mode in [1 bH]⁻. We already previously found evidence that [1 aH]⁻ is indeed thermally accessible at room temperature and does significantly contribute to the reactivity of the system.[ ⁵ , ²⁶ ] The shift of the H⁻ ligand between bridging and terminal positions decisively influences the nodal structures of the respective frontier orbitals such that [1H]⁻/[1 bH]⁻ and [1 aH]⁻ can interact with an incoming heterocumulene in different ways. Four conceivable modes of interaction are exemplarily shown for CO₂ in Figure 1. The HOMO of [1H]⁻/[1 bH]⁻ features lobes of π symmetry at its B−B bond ^[5] and can thus behave as a σ donor toward a CO₂ molecule approaching the B−B bond along a trajectory perpendicular to the bond vector (B–B bond activation). It becomes obvious that an overall constructive interaction with the LUMO of CO₂ should be possible, since the lobe at the central C atom (bonding interaction) is considerably larger than the lobes at the peripheral O atoms (antibonding interactions; Figure 1 b). Alternatively, the bonding B–H orbital of [1 aH]⁻ could act as the nucleophile toward the C atom of CO₂ (B–H bond activation; Figure 1 c). ^[27] In any case, one of the O atoms is an ideally positioned Lewis base to form an adduct with the (resulting) tricoordinated B atom of [1H]⁻/[1 aH]⁻/[1 bH]⁻. We will show that the B–B bond activation scenario is encountered with CO₂, the B–H bond activation scenario with iPrNCNiPr, and both at the same time with iPrNCO.

Figure 1.

a) Schematic representations and relative Gibbs free energies [kcal mol⁻¹] of the three isomers [1H]⁻ (dibenzo[g,p]chrysene‐type) and [1 aH]⁻, [1 bH]⁻ (9,9′‐difluorene‐type). b) Possible frontier‐orbital interactions between [1H]⁻ or [1 bH]⁻ and the model heterocumulene CO₂. c) Possible frontier‐orbital interactions between [1 aH]⁻ and the model heterocumulene CO₂.

The three substrates were treated with all three available alkali‐metal salts M[1H] (M⁺=Li⁺–K⁺) in order to examine possible counter‐cation effects, which have shown to be relevant for the reactivities of several comparable anionic boron compounds.[ ²² , ²⁸ , ²⁹ , ³⁰ ]

Moreover, we employed initial reaction temperatures of both −78 °C and room temperature throughout to avoid relating products of kinetic reaction control to those of thermodynamic reaction control. In the present cases, the counter cations had little influence and the reaction temperature did not play any role. Therefore, we will mainly describe the room‐temperature reactions of K[1H] as this salt provided the most comprehensive set of information (differing behavior of Li[1H]/Na[1H] is mentioned where appropriate).

When a solution of K[1H] in [D₈]THF was placed under a blanket of CO₂ (1 atm), the initial intense yellow color faded immediately. NMR spectroscopy showed a quantitative conversion to one single product (K[2H]; Scheme 2 a), which could not be reverted to the starting materials through heating. ^[31] The ¹H and ¹³C{¹H} NMR spectra of K[2H] are each characterized by two sets of signals of equal intensity, proving that the (on average) C _2v symmetry of the starting material K[1H] is broken in K[2H]. An extremely broad, unresolved proton signal, sharpening up upon ¹¹B decoupling, indicates that a B⋅⋅⋅B‐bridging H⁻ ligand persists in K[2H]. However, the resonance now appears at δ(¹H)=4.76 ppm, almost 6 ppm downfield‐shifted from the corresponding resonance of K[1H] (δ(¹H)=−0.99 ppm), ^[22] which points toward a significantly altered electronic situation within the B(μ‐H)B core. A low‐field resonance at δ(¹³C)=195.5 ppm is consistent with the presence of a B–O–C(O)–B fragment.[ ²⁹ , ³² ] In line with that, two ¹¹B NMR resonances are found in the region of tetracoordinated boron nuclei (δ(¹¹B)=−1.6, −5.1 ppm). ^[33]

Scheme 2.

a) Cycloaddition reactions between K[1H] and CO₂ or iPrNCNiPr, furnishing the B–B activation product K[2H] or the B–H activation product K[3], respectively; carbon atoms marked with asterisks bear tBu substituents. b) Model compounds Na₂[A] and B used for comparison; NHC=1,3‐bis(2,6‐diisopropylphenyl)imidazol‐2‐ylidene. (i) [D₈]THF, room temperature.

X‐ray crystallography on the centrosymmetric dimer {[K(pmdta)][2H]}₂⋅THF finally confirmed the molecular structure of the cycloaddition product (Figure 2; PMDTA: N,N,N′,N′′,N′′‐pentamethyldiethylenetriamine; acronyms of coordinated solvent molecules are written in lowercase letters). The μ‐H ligand was located in the difference Fourier map. Also, the proposed B–O–C(O)–B moiety was found; its endocyclic O–C single bond has a length of 1.354(2) Å, while the exocyclic C=O double bond is significantly shorter (1.224(2) Å). The B−C and B−O bond lengths amount to 1.601(3) and 1.524(3) Å, respectively. We note that the B−O bond of K[2H], which is inert toward excess CO₂ over hours, is shorter by 0.065 Å than the B−O bond of a comparable cycloadduct Na₂[A] (Scheme 2 b), which readily inserts a second molecule of CO₂. ^[29] Boron has a covalent radius of 0.84 Å. ^[34] While the B⋅⋅⋅B distance of the starting material [K(thf)₂][1H] (1.651(6) Å) ^[22] is within the binding regime, that of [2H]⁻ falls clearly out of this range (2.141(3) Å). It is also instructive to compare [2H]⁻ with B (Scheme 2 b), a formal [2+2] cycloadduct of CO₂ with the B=B double bond of an NHC‐supported diborene (NHC=1,3‐bis(2,6‐diisopropylphenyl)imidazol‐2‐ylidene). B is described as still featuring a B–B single bond with a concomitant B⋅⋅⋅B distance of 1.80(2) Å. ^[35] To sum up, NMR and crystallographic data are in full agreement with the formal [2+2] cycloaddition of one C=O double bond of CO₂ across the B–B single bond of [1H]⁻, whereupon this latter bond is broken and a central five‐membered BOCBH heterocycle is constructed.

Figure 2.

Solid‐state structures of the anionic parts of {[K(pmdta)][2H]}₂⋅THF, [K(thf)₂][3], [K(dme)₃][K][4H]₂, and [Li(thf)₃][5]; most C‐bonded H atoms and all tBu substituents are omitted for clarity; B: green, C: black, H: gray, N: blue, O: red.

The addition of iPrNCNiPr (1 equiv) to K[1H] in [D₈]THF resulted in the selective formation of K[3] (Scheme 2 a). The ¹H NMR spectrum shows only one set of tBu(C₆H₃) and iPr signals with an overall integral ratio suggesting an equimolar reaction between the two starting materials. A resonance assignable to a B‐bonded H atom is not detectable, even upon ¹¹B decoupling. Instead, we observe a singlet resonance at δ(¹H)=8.07 ppm (1 H), showing a cross‐peak to a signal at δ(¹³C)=158.8 ppm in the ¹H‐¹³C HSQC experiment. The associated C−H moiety is obviously not part of an aryl ring or an alkyl group, but can rather be assigned to a formic acid derivative.[ ³⁶ , ³⁷ ] Apart from the higher symmetry of K[3] compared to the CO₂ cycloadduct K[2H], a notable upfield shift of the resonance assignable to H‐1 becomes evident (K[3]: 7.77 ppm vs. K[2H]: 8.15/8.04 ppm; cf. Scheme 2 a). In our experience, the absence of aryl signals with chemical shift values of δ(¹H) >8 ppm in the spectrum of K[3] is a strong indication that this compound has a 9,9′‐difluorene‐type scaffold. Dibenzo[g,p]chrysene‐type scaffolds, such as that of K[2H], consistently feature more deshielded H‐1 protons due to the magnetic anisotropy effect prevalent in the cove regions ^[38] of polycyclic aromatic hydrocarbons.[ ³⁹ , ⁴⁰ ] The ¹¹B NMR spectrum of K[3] shows one signal at δ(¹¹B)=0.7 ppm; both B centers are thus chemically equivalent and tetracoordinated. ^[33]

X‐ray diffraction on single crystals of [K(thf)₂][3] ultimately proved that the reaction of K[1H] with iPrNCNiPr is indeed accompanied by two Wagner–Meerwein‐type rearrangements (cf. Figure 2). The two B centers are spanned by a largely symmetrical formamidinate bridge (C–N=1.319(5), 1.308(5) Å) and positioned at a distance of 1.821(6) Å. Thus, the five‐membered BNCNB core of K[3] still contains a covalent B–B single bond. ^[34] The formamidinate‐H atom originates from the previously B⋅⋅⋅B‐bridging H⁻ ligand of K[1H], as confirmed by a deuteration experiment using K[1D] (NMR spectroscopic control). It is thus plausible to assume that the encounter complex between the two starting materials resembles the scenario outlined in Figure 1 c for the model heterocumulene CO₂: We propose a charge flow from the bonding B–H orbital of the rearranged [1 aH]⁻ to the LUMO of iPrNCNiPr, accompanied by the formation of the first B−N bond. The second B–N adduct is formed at the newly generated tricoordinated B center after intramolecular rotation within the formamidinate ligand (Scheme 2 a). The resulting B(sp³)–B(sp³) diborane salts M[3] (M⁺=Li⁺–K⁺) featuring a central five‐membered BNCNB heterocycle are accessible in almost quantitative yields, ^[41] thus adding new members to a promising, ^[42] but experimentally little explored diborane class. ^[43]

In terms of electronic structure and steric demand, iPrNCO occupies an intermediate position between the two previously addressed substrate molecules. When it reacts with K[1H], two isomeric products are formed in competing reactions (Scheme 3 a): K[4H] is related to the CO₂ adduct K[2H] (cf. Scheme 4), whereas K[5] is an analog of K[3]. According to ¹H NMR spectroscopy on the reaction mixture, K[4H] and K[5] are produced in a stoichiometric ratio of 1.5:1. Apart from a necessary symmetry break due to the switch from an NCN to an NCO bridge, the NMR data of K[5] do not differ significantly from those of K[3]. Importantly, the resonances of H‐1/1′ (7.74/7.73 ppm) again appear below 8 ppm, in agreement with a 9,9′‐difluorene‐type molecule. The most revealing NMR features of the second product K[4H] are its two signals in the ¹¹B NMR spectrum: The first is broad and has a chemical shift value of δ(¹¹B)=13.7 ppm, the second appears as a doublet at δ(¹¹B)=−18.5 ppm with a coupling constant ¹ J(B,H)=84 Hz that is characteristic of a terminal H substituent. ^[33] The ¹H and ¹³C{¹H} NMR spectra of K[4H] indicate the presence of two borafluorene units. Their chemical shift values are more pronouncedly different than in the case of K[5]; an assignment was achieved by means of a cross‐peak between the ¹¹B doublet‐resonance and the H‐1′ signal in the ¹H‐¹¹B HMBC experiment.

Scheme 3.

a) Cycloaddition reaction between K[1H] and iPrNCO, furnishing both the B–B activation product K[4H] and the B–H activation product K[5]; carbon atoms marked with asterisks bear tBu substituents. b) C–E are rare examples of structurally characterized boron compounds containing a chelate ligand related to that in K[4H]; NHC=N,N‐dimesitylimidazol‐2‐ylidene. (i) [D₈]THF, room temperature.

Scheme 4.

The computed, energetically favorable primary product [4 bH]⁻ of the cycloaddition reaction between K[1H] and iPrNCO and its rearrangement cascade to form K[4H]; carbon atoms marked with asterisks bear tBu substituents.

We succeeded in the isolation of K[4H] through recrystallization of the blend of crude reaction products from a 1:1 mixture of THF:DME ([K(dme)₃][K][4H]₂; DME: 1,2‐dimethoxyethane). Single crystals of [5]⁻ could be grown in the form of its Li⁺ salt [Li(thf)₃][5] (Figure 2). The molecular structure of [5]⁻ is formally derived from the structure of [3]⁻ by replacing one NiPr moiety with an O atom to form a five‐membered BNCOB heterocycle. The B–B (1.801(7) Å) and B–N bond lengths (1.605(6) Å) of [5]⁻ are the same as those of [3]⁻ within the experimental error margins; the length of the B–O bond is 1.652(6) Å. In the isomeric reaction product [4H]⁻, the iPrNCO core is linked to one borafluorenyl substituent via a B−C bond and acts toward the other borafluorene as a κ²‐N,O chelating ligand (B–N=1.586(6) Å, B–O=1.570(6) Å). The B−O bond is considerably shorter (by 0.082 Å) than that of [5]⁻, suggesting a stronger interaction, which may be one reason for the dominance of [4H]⁻ over [5]⁻ in the product mixture (note that the corresponding reaction with Na[1H] gives exclusively Na[4H]). On the other hand, [4H]⁻ should suffer from considerable angular strain due to its four‐membered ring structure, which may be one reason why related compounds, such as C–E, are very rare (Scheme 3 b). ^[44] All in all, the preferential formation and stability of [4H]⁻ remains astounding, given that Breher et al. explicitly mentioned that their compounds C and D are readily thermally decomposed. ^[44]

Quantum‐chemical calculations were performed for all isomeric structure types [2H]⁻–[5]⁻. ^[45] Each structure type was computed with each of the three substrate fragments CO₂, iPrNCNiPr, and iPrNCO (see the Supporting Information for graphical representations and Gibbs free energies of all computed structures). In the cases of CO₂ and iPrNCNiPr, the energetically most favorable heterocyclic cores correspond to those found in the reaction products K[2H] and K[3], respectively. For iPrNCO, [4H]⁻ represents the global minimum (ΔG=0); [5]⁻ is higher in energy by only ΔG= +4.5 kcal mol⁻¹. Energetically in between lies the isomer [4 bH]⁻ (ΔG=+2.7 kcal mol⁻¹; Scheme 4), which has not been observed, likely because it rearranges to [4H]⁻ without significant activation barrier: We assume that iPrNCO is not necessarily attacked by the B−H bond of [1 aH]⁻, but can alternatively activate the B−B bond of the energetically comparable B(μ‐H)B isomer [1 bH]⁻ (Figure 1 a) from the side opposite to the bridging H⁻ ligand. The BNCBH core in the resulting primary product [4 bH]⁻ is a structural analog of the BOCBH core in the CO₂ cycloadduct [2H]⁻ (Scheme 2 a). A mere shift of the bridging H⁻ ion to a terminal position unlocks the rigid ring system and enables a flip about the C−N bond, which brings the (now tricoordinated) B atom into close vicinity to the electron lone‐pair of the carbonyl function. Subsequent B–O adduct formation establishes the BNCO ring of [4H]⁻. ^[46] All in all, we note a pleasantly good agreement between experiment and theory. One seeming inconsistency relates to the CO₂ cycloadduct [2H]⁻. While NMR spectroscopy and X‐ray crystallography reveal a dibenzo[g,p]chrysene periphery for [2H]⁻, both in solution and in the solid state, the corresponding computed anion structure is energetically less favorable by ΔG=+16.5 kcal mol⁻¹ relative to a 9,9′‐difluorene‐type structure (cf. [4 bH]⁻ with O instead of NiPr in Scheme 4). However, even this apparent discrepancy can be explained by the following scenario, which draws a coherent picture of all available experimental and theoretical results: ^[47] A mapping of the molecular surfaces of [1H]⁻ and [1 bH]⁻ showed that the B−B bond of the thermodynamically less favorable, less abundant isomer [1 bH]⁻ is more easily accessible than for [1H]⁻. ^[5] This is mainly due to different spatial orientations of the tBu groups relative to the B−B bonds. We assume that the small CO₂ molecule can nevertheless react similarly well with both isomers. The product distribution is thus dictated by the relative abundancies of [1H]⁻ vs. [1 bH]⁻ and dominated by the experimentally observed K[2H], which, once formed, does not rearrange to the computed global minimum structure. ^[31] The iPrNCNiPr molecule, on the other hand, is so bulky that it entirely neglects the B−B bond and prefers to react with the protruding B−H bond of [1 aH]⁻ to afford K[3]. The intermediate‐sized iPrNCO has the option to use a similar reaction trajectory as iPrNCNiPr (B–H bond activation yielding K[5]). Alternatively, it can perform a CO₂‐like B–B bond activation, with the distinct restriction to the sterically less demanding μ‐H component [1 bH]⁻ in the equilibrium of isomeric starting materials (yielding K[4H]). This selectivity for [1 bH]⁻ over [1H]⁻ is further enhanced by the fact that the formation of the thermodynamic primary product [4 bH]⁻ requires a reaction at the sterically more encumbered C=N bond of iPrNCO (cf. Scheme 4).

Conclusion

Interest in μ‐(hydrido)diboranes(4) with their electron‐precise B−B bonds and bridging H⁻ ligands goes back many decades, but to date most of the knowledge has come from quantum‐chemical calculations on the simplest representative, the elusive anion [B₂H₅]⁻. It has been suggested that the B−B bond acts as a non‐classical electron‐donor site, but also that a mere shift of the μ‐H atom to a terminal position fundamentally alters the frontier orbitals of the system and generates a H⁻ donor with a nearby Lewis acidic B center. Experimental insight into the unique properties of the central BHB unit has been hampered by the lack of synthetically accessible (μ‐hydrido)diborane(4) derivatives, in which the BHB core is not perturbed by electron‐donating or sterically protecting substituents. Our representative K[1H] with planarized dibenzo[g,p]chrysene‐type scaffold comes close to this ideal and reacts straightforwardly with various substrates, such as the heterocumulenes CO₂, iPrNCO, and iPrNCNiPr in either B–B or B–H bond activation reactions. The actual activation scenario is governed by two factors, (i) the steric demand of the substrate molecule and (ii) the ready isomerization of [1H]⁻ with bridging H⁻ ligand to its isomer [1 aH]⁻ with terminal B−H bond: The small CO₂ molecule can still undergo a formal [2+2] cycloaddition with the sterically loaded B−B bond of [1H]⁻, thereby forming a BOCBH heterocycle (K[2H]). The bulky substrate iPrNCNiPr most easily interacts with the B−H bond of [1 aH]⁻ to give the B(sp³)–B(sp³) diborane salt K[3] with a BNCNB ring. Both B–B and B–H bond activations are observed for the hybrid substrate iPrNCO, leading to the two isomeric and almost isoenergetic products K[4H] and K[5]. Taken together, it is experimentally demonstrated here for the first time that H⁻ ligand‐supported B–B single bonds show cycloaddition reactivities that would normally be associated with B=B double bonds, thus opening new perspectives for organoborane synthesis.

Conflict of interest

The authors declare no conflict of interest.

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

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

Supplementary

T. Trageser, D. Bebej, M. Bolte, H.-W. Lerner, M. Wagner, Angew. Chem. Int. Ed. 2021, 60, 13500.