Lewis‐Acidic Properties and Catalytic Behavior of Hydroboration‐ and Diolysis‐Derived NHC‐Stabilized Borenium Cations

ABSTRACT

In this work, we report the synthesis and characterization of a series of thirteen N‐heterocyclic carbene‐stabilized borenium cations featuring varied steric and electronic properties of the supporting NHC ligands (IMe₂, IMes, SIMes). The series comprises borenium species bearing distinct substituents, including hydrides, carbon‐based groups (sp³‐ and sp²‐hybridized), and diol‐derived moieties forming five‐ to seven‐membered rings. The Lewis acidity was quantified in terms of Gutmann's Acceptor Number (AN), revealing that most of the prepared boreniums exhibit higher acidity than the benchmark Lewis acid B(C₆F₅)₃. These borenium species were subsequently evaluated as catalysts in two model transformations—homocoupling of diazo derivatives and hydrosilylation of acetophenones. The markedly different catalytic behavior observed across the series highlights the decisive influence of the structural and electronic factors on the global Lewis acidity of the boron center and, consequently, on its catalytic performance.

A series of thirteen (N‐heterocyclic carbene) NHC‐stabilized borenium cations is synthesized from NHC→BH₃ complexes via hydroboration and condensation with diols. The novel compounds are fully characterized. Eight crystal structures are presented. Kinetic and NMR data show how the structural features translate to catalytic activity in hydrosilylation and diazo‐homocoupling reactions. Particularly, the presence of a dioxaboracycle.

1. Introduction

Tricoordinate boron compounds have been for long recognized as archetypal Lewis acids. In a specific case given, the resultant Lewis acidity‐derived reactivity relies on factors that can be separately described, even if they cannot be separately assessed experimentally (Figure 1). For example, it has been evidenced that the cyclic boronic esters (of diols) display higher Lewis acidity strength than the corresponding boronic acids (the pK _a lowers by about 2–4 units from acid to ester) [1, 2, 3]. This can be explained by the restricted rotation around the B‐O bonds in cyclic esters, preventing the oxygen lone pairs from optimal alignment with the empty boron p‐orbital. A different, complementary explanation associates Lewis acidity strength with the energy of geometry reorganization from flat planar to tetrahedral upon adduct formation [4]. Formation of a five‐membered cyclic ester results in narrowing of the O‐B‐O angle with respect to the idealized 120° in non‐cyclic boronic acids (and non‐cyclic esters). In turn, this results in lowering of the reorganization energy, making it easier to adopt tetrahedral geometry (and form an adduct) with any given Lewis base. This concept formed a basis for the THC (tetrahedral boron character) scale, which relies exclusively on structural data [5]. The intrinsic factors controlling the Lewis acidity of the boron center in rigid 1,3,2‐dioxaborolane rings are reflected in dioxaborol‐2‐ylium moieties in some of the structures presented in this work.

FIGURE 1.

Factors controlling the resultant Lewis acidity of the boron center.

One of the most common experimental Lewis acidity assessment methods is the measurement of the acceptor number (AN)—utilizing Et₃PO (TEPO) as a ³¹P NMR probe [6, 7, 8]. The presence of oxygen atoms directly bound to the boron center can affect the AN, as exemplified by the series B(OC₆F₅)₃: 88.4 > BC₆F₅(O‐C₆F₅)₂: 87.5 > B(C₆F₅)₂OC₆F₅: 86.4 > B(C₆F₅)₃: 78.9. This is a consequence of the raising hardness and therefore, stronger interaction with the Et₃PO—a hard Lewis base in terms of the Persons’ HSAB theory [9]. The AN scale, like all other experimental scales, is dependent on the Lewis base‐partner. Thus, it does not correlate linearly with complementary scales, like computational FIA (fluoride ion affinity), for which a much smaller and charge‐denser F^– anion is used to assess the adduct formation energy [10]. The FIA scale also has its shortcomings, particularly in the context of cationic Lewis acids [11].

The chemical space of neutral Lewis acidic organoboron compounds is complemented by the relatively stable cationic species of high Lewis acidity strength. The enhanced electrophilic properties make borenium species promising candidates as catalysts [12, 13, 14]. The behavior and nomenclature of tri‐ and di‐coordinate boron cations were first reviewed by Nöth et al. [15, 16, 17]. In principle, when starting from a neutral tricoordinate boron, displacement of a substituent (anion) by a neutral donating ligand, such as pyridine or lutidine, generates a borenium cation. For this to happen, the neutral donor must be sterically bulky, and the displaced substituent must be a good leaving group (anion), so that the nucleophilic dissociation is favored over Lewis acid‐base adduct formation [17]. In other words, the neutral Lewis base is separating the cationic boron from direct interaction with the weakly coordinating anion (WCA), and a salt is formed. In the case of the NHC‐coordinated boranes, the hydride anion takes the role of a leaving group, being substituted by a weakly coordinating anion, resulting in a tricoordinate NHC→borenium cation in a salt with a WCA [18]. The reaction is driven by the thermodynamic stability of the byproduct (for example, H₂ gas or triphenylmethane). ¹¹B NMR studies reveal that, when triflic acid is used (HOSO₂CF₃), the boron atom is often firmly bonded and cannot be distinguished as a cation [19]. In the case of triflimic acid (HNTf₂; HN(SO₂CF₃)₂), the direct interaction is observed (for example, by diffractometry), but the compound acts as an ionic pair. In the case of Ph₃C⁺ B(C₆F₅)₄ ^–, the two boron centers (cationic and anionic) have no possibility of contact [19].

The hydride abstraction strategy can also be used to generate single B‐H species, like the [IMes→B(H)C₆F₅]⁺ [B(C₆F₅)₄]^– salt [20]. Chiral NHC‐stabilized borenium variants have been obtained as camphor‐ and oxazole‐ring analogues with 9‐borabicyclo[3.3.1]nonane (9‐BBN) and also from (+)‐diisopinocampheylborane, among others [21]. Cyclized, planarized boreniums can also be obtained by consecutive hydride abstraction/dehydrogenative borylation cycles from N‐benzyl substituted NHC→BH₃ adducts [22]. NHC‐stabilized borenium cations can be also accessed by other methods, e.g., from diarylboron halides [23, 24] and by halogen abstraction [25] or substitution [26].

We have been interested in the synthetic applicability of simple NHC→BH₂ ⁺ WCA^– species in the context of their reactivity in hydroboration. We envisioned that those intermediates could be used as starting points to obtain diverse borenium compounds in which the Lewis acidity and catalytic properties originate from the structure of the hydroboration and diolysis partners, in addition to the effects originating from a specific NHC ligand (and possibly the WCA counterion). Herein, we present a series of tricoordinate NHC‐stabilized borenium cations derived from IMes→BH₃, SIMes→BH₃, and IMe₂→BH₃ borane complexes, as salts with WCAs (IMes = 1,3‐dimesitylimidazol‐2‐ylidene, SIMes = 1,3‐dimesitylimidazolidin‐2‐ylidene, IMe₂ = 1,3‐dimethylimidazol‐2‐ylidene). The strong σ‐electron donor character of NHCs assures the stability of such complexes. AN of the prepared compounds is assessed in DCM‐d₂, and the catalytic activity is screened in two model reactions. Obtained compounds are structurally diverse, including bulky groups, unsaturated bonds, and dioxaboracycles (cationic analogues of cyclic boronic esters). They also differ in the donor strength of the NHC moiety.

2. Results and Discussion

2.1. Synthesis

The title borenium compounds were obtained via hydroboration and diolysis (hydride‐to‐alkoxy substitution with a diol) in reactions between an NHC→BH₂ ⁺ NTf₂ ^– salt and unsaturated alkanes or diols, respectively (Scheme 1, method 1). Tetracoordinate NHC→BH₃ species were obtained by typical methods; IMes→BH₃ from isolated, free persistent carbene reacted with BH₃·SMe₂ [27], SIMes→BH₃ from carbene generated in situ and IMe₂→BH₃ from doubly N‐alkylated imidazolium treated with NaBH₄ [28]. These compounds are stable to moisture in air, unable to hydroborate, and engage in the reaction only after abstraction of one hydride anion by a WCA‐conjugated Brønsted acid, such as triflimic acid HNTf₂ [29, 30]. Obtained NHC→BH₂ ⁺ NTf₂ ^– salts are white powders and can be stored for months inside an Ar‐filled glovebox with minimal to no decomposition. Compounds 1b’ and 1g’ were synthesized by using Ph₃C⁺ B(C₆F₅)₄ ^– in place of HNTf₂, with the generation of Ph₃CH byproduct, which can be removed from the mixture by washing with hexane [31].

SCHEME 1.

General synthetic scenarios used to obtain intermediates 1a , 2a , 3a , 1bH, and 1hH, and the title compounds 1b , 1b’ , 1c , 1d , 1e , 1f , 1 g , 1g' , 1 h , 2b , 3b , 3c , 3d , 3e , and 3g .

The boron center in the IMes→BH₂ ⁺ directly interacts with the NTf₂ ^– anionic species. The single ¹¹B NMR wide signal δ _B = –17.5 ppm (FWHM ≈ 440 Hz) is evidence of a single tetrahedral boron species present. The same is true for the IMe₂→BH₂ ⁺ NTf₂ ^– derivative. However, the reaction between the saturated SIMes→BH₃ and 1 eq. HNTf₂ is not selective, presumably producing a H‐bridged dimer and some unidentified byproducts in addition to the SIMes→BH₂ ⁺ NTf₂ ^–. For this reason, we were unable to isolate and purify compound 2a and excluded it from the assessment of catalytic performance. Nonetheless, upon triethylphosphine oxide (TEPO) addition to the sample, a single ³¹P{¹H} NMR signal at δ _P = 84.6 ppm indicates the formation of [SIMes→BH₂·TEPO]⁺ adduct. The in situ generated SIMes→BH₂ ⁺ NTf₂ ^– mixtures reacted with acetylenes and diols in our protocols resulted further in complicated mixtures that indeed contain some small amounts of the desired products. However, we were unable to isolate and characterize any of those, with the one exception of compound 2b .

In the case of derivatives 1b , 1b’ and 1 h , an alternative pathway was also explored (Scheme 1, method 2) [32]. Combining the free carbene with CatBH or BBN results in the neutral NHC→BHR₂ adduct in the first step (compounds 1bH and 1hH). Abstracting the H^– by HNTf₂ or Ph₃C⁺ B(C₆F₅)₄ ^– in the second step yields the final product. This method seems to produce fewer impurities, but is limited by the availability of the starting carbene and mono‐hydroborane derivative.

2.2. X‐ray Diffractometry

For compounds 1b , 1b’ , 1d , 1f , 2b , 3b , 3e, and 1hH, single crystals of sufficient quality for X‐ray diffraction were obtained from DCM/hexane mixtures inside the glovebox. Crystal data for all structures, data collection, and details of refinement are given in Table S2 in the Supporting Information.

Comparison of parent structures 1b’ , 1b , 2b, and 3b (Figure 2) allows to assess the influence of NHC and anion characters on the molecular conformation in the crystal phase. The shortest B–C_NHC bond of 1.547(5) Å for compound 3b shows the extent of steric repulsion from the mesityl‐ substituents to the 1,3,2‐benzodioxaborol‐2‐yl. Particularly, when compared with 1b bond length (1.561(2) Å), for in both cases the dihedral angle between the NHC and the 1,3,2‐benzodioxaborol‐2‐yl moieties is quite low (13.2(2)° and 5.9° for 1b and 3b , respectively). On the other hand, the stronger donor properties of SIMes versus IMes result in diminished imidazol‐dioxaborol conjugation and longer B–C_NHC bond distance (1.583(3) Å) in the case of 2b , what in turn allows the 1,3,2‐benzodioxaborol‐2‐yl to be rotated almost perpendicularly to the NHC ring (89.8(3)°), ultimately resulting in a denser crystal packing. In the case of 1b’ , the large perfluorinated anion dominates as the source of interactions governing the crystal packing, which results in a dihedral angle around B–C_NHC bond of 64.3(4)° (NHC→B bond length of 1.572(4) Å). Regardless of the rotation around the B–C_NHC bond, the mesityl‐ moieties retain a perpendicular conformation to the 5‐membered diazacycle in all cases. This is also true for the structures of 1f (Figure 3) and 1d (Figure 4). In the case of 1f , the 7‐membered dioxaborepine cycle adopts a chair‐like conformation. The O‐B‐O angle is therefore wider 130.6(1)°) and corresponds to a lower AN for this derivative in comparison with the catechol‐derived compounds. This is consistent with the previously mentioned geometry reorganization energy theory [4] and results for parent neutral boronic esters [33]. The structure of 1d reveals B–C_NHC bond length of 1.580(4) Å, which is most probably caused by a significant repulsion from the styryl groups against mesityls.

FIGURE 2.

Molecular structure view of the 1,3,2‐benzodioxaborol‐2‐yl moiety‐bearing compounds 1b , 1b’ , 2b , and 3b . Thermal ellipsoids are shown at the 30% probability level. Anions were omitted for clarity. For compound 2b , only one of the two independent molecules is shown. Selected bond lengths (Å) and angles (°): 1 b' {C2‐B1 1.572(4), C2‐N3‐C7‐C15 74.8(4), N3‐C2‐B1‐O10 64.3(4)}, 1b {C2‐B1 1.561(2), C2‐N6‐C8‐C23 89.6(2), N6‐C2‐B1‐O10 13.2(2)}, 2b {C2A‐B1A 1.583(3), C2A‐N3A‐C7A‐C11A 90.7(3), N3A‐C2A‐B1A‐O9A 89.8(3)}, 3b {C2‐B1 1.547(5), N3‐C2‐B1‐O9 5.9}.

FIGURE 3.

Molecular structure view of the dioxaborepine moiety‐bearing compound 1f . Thermal ellipsoids are shown at the 30% probability level. Anion is omitted for clarity. Selected bond lengths (Å) and angles (°): C2‐B1 1.595(2), O10‐B1‐O9 130.6(1), N3‐C2‐B1‐O10 133.5(2).

FIGURE 4.

Molecular structure view of compound 1d . Thermal ellipsoids are shown at the 30% probability level. Anion and one of the two independent molecules are omitted for clarity. Selected bond lengths (Å) and angles (°): C2A‐B1A 1.580(4), C16A‐B1A‐C16A 124.2, C2A‐N3A‐C7A‐C12A 107.1.

We have also obtained a single crystal of intermediate 1hH (Figure 5), the hydride adduct of 1 h . The B–C_NHC bond length is significantly longer than in the borenium structures and equals 1.638(2) Å. The boron atom is about 0.474 Å out‐of‐plane (defined by the three carbon atoms C2A, C25A, and C29A), which is expected to align with after the hydride abstraction in 1 h

FIGURE 5.

Molecular structure view of compound 1 h H. Thermal ellipsoids are shown at the 30% probability level. Anion and one of the two independent molecules are omitted for clarity. Selected bond lengths (Å) and angles (°): C2A‐B1A 1.638(2), C2A‐N3A‐C7A‐C13A 75.4(2), the B1A boron atom is about 0.474 Å out‐of‐plane defined by atoms C2A, C25A, and C29A.

2.3. Determination of the Acceptor Number and FIA, HIA Indices

Most of the obtained title borenium compounds (depicted in Scheme 2) are rather strong Lewis acids, and their AN has been assessed in DCM (AN _DCM = 20.4) at room temperature. Measurement in a solvent of lower AN, like benzene (AN _benzene = 8.2) or THF (AN _THF = 8.0), was not possible due to solubility issues. The AN of the obtained compounds is presented in Table 1.

SCHEME 2.

The scope of studied compounds. All boreniums are salts with the NTf₂ ^– anion unless otherwise indicated. Preparative method and corresponding isolated yield are specified.

TABLE 1.

Characterization of the obtained borenium cations, their Acceptor Number (AN) in DCM, Fluoride Ion Affinity and Hydride Ion Affinity (at 298.15 K) in gas phase and in DCM (at the PW6B95(D3)/def2‐QZVPP level of theory—See Supporting Information).

	NHC	R2	WCA	11B NMR / ppm	TEPODCM / ppm	AN	FIA / kJ·mol−1	FIA DCM b / kJ·mol−1	HIA / kJ·mol−1	HIA DCM b / kJ·mol−1
1a	IMes	H	−NTf2	−17.5	84.5	96.5	714	288	751	325
1b		Cat.	−NTf2	24.9	85.1	97.7	672	251	646	227
1b'			−B(C6F5)4	24.7	84.7	96.9
1c		(‐CH = CH‐tBu)	−NTf2	54.2	81.3	89.4	611	192	628	208
1d		(‐CH = CH‐Ph)	−NTf2	53.7	83.2	93.5	629	214	647	231
1e		(‐O)2BINAP	−NTf2	25.1	83.1	93.3	656	244	634	227
1f		(‐OCH2)2Ph	−NTf2	21.2	75.2 a	75.8 a	614	194	590	170
1g		(‐O)2biphenyl	−NTf2	22.5	82.1	91.0	652	239	633	220
1g'			−B(C6F5)4	22.1	81.7	90.2
1h		BBN	−NTf2	83.0	53.7 a	28.2 a	641	216	660	235
2a	SIMes	H	−NTf2	−18.0	84.6	96.7	740	319	777	355
2b		Cat.	−NTf2	25.3	85.0	97.6	676	261	653	237
3a	IMe2	H	−NTf2	−18.0	85.0	97.6	753	285	788	321
3b		Cat.	−NTf2	25.2	84.5	96.4	681	236	651	209
3c		(‐CH = CH‐tBu)	−NTf2	55.3	81.7	90.2	648	210	668	226
3d		(‐CH = CH‐Ph)	−NTf2	53.2	83.2	93.5	646	216	668	236
3e		(‐O)2BINAP	−NTf2	24.9	83.0	93.1	702	266	677	242
3g		(‐O)2biphenyl	−NTf2	23.2	82.7	92.4	679	243	656	221
4a		B(C6F5)3	n/a	59.7	76.6	78.9	442	231	477	264

^a For compounds 1f and 1h , the AN was calculated from the extrapolated ³¹P NMR chemical shift according to a titration procedure (see Supporting Information).

^b Solvent effects have been approximated using the PCM model.

The great majority of the title compounds indeed display a very high acceptor number, with the flat catechol‐derived boreniums 1b , 1b’ , 2b , and 3b at the top. Larger dioxaborepine ring‐containing examples 1e , 1 g , 1g’ , 3e , and 3 g have the AN a few units lower due to a nonflat 7‐member ring. However, the non‐cyclic boreniums 1d and 3d are on the same level, with the t‐Bu derivatives 1c and 3c a few units lower due to steric effect. As expected, the 1a , 2a , and 3a dihydrido‐boreniums display a very high AN, having small ‐H substituents. Compound 1f represents a mid‐point in our series with a lower AN of 75.8 caused by the chair‐like conformation of the dioxaborepine ring. Finally, the measurement for the most sterically crowded 1 h reveals AN comparable with the solvent. Moreover, for 1f and 1 h , we have observed a dependence of δ _P on the TEPO:LA ratio, for those cases the AN value was assessed in a titration experiment. For further details, see Supporting Information.

The ¹¹B NMR chemical shift for the title compounds corresponds with the typical values for tricoordinated boreniums, and it does not correlate with the AN results. However, in the case of dihydridoboreniums 1a , 2a, and 3a , the δ _B  =  –18 ppm, which is consistent with a tetrahedral coordination sphere (B‐anion contact in solution). For the oxygen‐containing boracycles, the ¹¹B NMR is measured around δ _B = 22–25 ppm, which is indicative of flat planar geometry, where boron is connected to oxygen heteroatoms. The presence of oxygen atoms negates any electron‐withdrawing effect from the aromatic moieties. In contrast, boreniums 1c , 1d , 3c , and 3d display ¹¹B around δ _B = 55 ppm, indicating an influence of the unsaturated substituents directly connected to the boron atom. ¹¹B NMR δ _B = 83 ppm for compound 1 h is typical for alkyl‐substituted boron species [34].

The FIA and HIA values were calculated at the DFT level of theory in the gas phase and DCM according to the methodology specified by Greb et al. (Table 1) [35, 36]. FIA for the benchmark compound 4a (B(C₆F₅)₃) was calculated to be 442 kJ·mol⁻¹, which is an acceptable result considering the computational effort (in comparison to the value of 448.1 kJ·mol⁻¹ obtained at the DLPNO‐CCSD(T)/cc‐pVQZ level) [35]. Similarly, HIA for 4a (our work: 477 kJ·mol⁻¹, 481 kJ·mol⁻¹ at DLPNO‐CCSD(T)/aug‐cc‐pVQZ) [36]. Both indices are highly dependent on the volume of substituents present, emphasizing the cooperative steric effect of the mesityl‐ and R² substituents. The origin of this effect might be connected more to the intramolecular repulsion between moieties (rotational obstruction) necessary for the trigonal‐to‐tetrahedral geometry conversion (upon adduct formation) and less dependent on the repulsion toward the anion partner.

Gas phase FIA values for the R² = Cat. variants favor the 3b (FIA = 681 kJ·mol⁻¹) by few kJ·mol⁻¹ over 1b (FIA = 672 kJ·mol⁻¹) and 2b (FIA = 676 kJ·mol⁻¹), whereas 3c (FIA = 648 kJ·mol⁻¹) and 3d (FIA = 646 kJ·mol⁻¹) are about 17–19 and about 36–37 kJ·mol⁻¹ over 1d (FIA = 629 kJ·mol⁻¹) and 1c (FIA = 611 kJ·mol⁻¹) respectively. Quite a high value for 3e (702 kJ·mol⁻¹) is surprising considering the skewed 7‐membered dioxaboracycle. The equally unexpected high value of 641 kJ·mol⁻¹ for 1 h results from the small size of fluoride in comparison with TEPO. FIA values for the remaining variants present as expected and are in qualitative agreement with the measured AN. The difference between gas phase FIA and HIA reflects the properties of the anions. In all cases where boron is included in a dioxaboracycle, the HIA < FIA by about 20–30 kJ·mol⁻¹. For the remaining borenium cations, this trend is reversed (HIA > FIA by about 17–23 kJ·mol⁻¹), and the largest differences for the IMe₂→BH₂ ⁺ cations are around 37 kJ·mol⁻¹.

2.4. Catalytic Activity Assessment

2.4.1. Homocoupling of Bis(4‐methylphenyl)diazomethane

Diazo‐compounds are very useful intermediates due to their rather high reactivity. Some of them undergo very slow homocoupling to the corresponding ethenes even at room temperature under light exposure. During our previous investigations, we noticed that this reaction can also be catalyzed with organoboron Lewis acids, including boreniums [37]. Therefore, it can be used as a model reaction to compare the catalytic potency of the currently investigated compounds. Similar to previously, the process is not selective toward the ethene product in all cases, and it can also yield a bis‐substituted hydrazine, as revealed by the NMR studies (Scheme 3).

SCHEME 3.

Homocoupling of bis(4‐methylphenyl)diazomethane. Model reaction catalyzed by the title compounds yields a mixture of azine and ethene products in a variable ratio (see Table 2).

A few of the title compounds do not catalyze the aforementioned reaction to a reasonable extent. Compounds 1a , 1d , 1 g , 1g’ , 1 h , 3a , and 3c require 10–20 mol. % or higher loadings to produce a visible amount of the product over 24 h. Therefore, those compounds were excluded from the kinetic experiments. We were, however, able to ascertain the NMR product distribution in all cases, regardless of the conversion. Dihydridoboreniums 1a and 3a (similarly to 2a ) are very reactive and can engage in some side reactions, nullifying their catalytic properties in this process. Poor performance of 1 h can be explained by a very low AN = 28.2 and the greatest steric crowding around the boron center. Other excluded compounds do not share any common characteristics. The majority of the remaining compounds are excellent catalysts, and the reaction progress was monitored by UV‐Vis, which is a more suitable and faster technique than NMR.

For the kinetic UV‐Vis measurements, bis(4‐methylphenyl)diazomethane starting material was selected, with λ _max. = 535 nm, for the products do not absorb at this wavelength region. Table 2 presents the results of the kinetic measurements (tracking the consumption of the starting material). For two comparably weaker catalysts in the series, 1c and 1f , the catalyst loading was doubled (to 1 mol. %) for practical reasons. However, higher loadings would render the results incomparable. Table S1 in the Supporting Information presents additional details of the data fitting.

TABLE 2.

Diazo‐homocoupling kinetic measurements. ^a

	Cat. / mol.%	k obs. a / 10– 5 s– 1	prod. distrib. / % b
			ethen	azine
1a	—	‐ **	100	0
1b	0.5	35.4	100	0
1b’	0.5	42.4	70	30
1c	1.0	3.54	100	0
1d	—	‐ **	0	100
1e	0.5	11.2	100	0
1f	1.0	4.48	100	0
1g	—	‐ **	90	10
1g’	—	‐ **	100	0
1h	—	‐ **	100	0
2b	0.5	25.9	100	0
3a	—	‐ **	100	0
3b	0.5	39.4	0	100
3c	—	‐ **	100	0
3d	0.5	6.24	99	trace
3e	0.5	21.6	0	100
3g	0.5	35.2	5	95
4a c	0.5	0.821	100	0

^*Conditions: [diazo] = 10 mm, [cat.] = 0.05∼0.1 mm, T = 20 °C, t = 0–90 min.

^**Kinetics too slow to measure at comparable conditions.

^a First‐order fitting,

^b NMR assessment after 24 h,

^c 4 a = tris(pentafluorophenyl)‐borane.

1,2‐Bis(di‐4‐tolylmethylene)hydrazine product is exclusively observed for catalysts 1d , 3b, and 3e , and a mixture of products for 1b’ , 1g , 3d (trace azine), and 3g . One can see from Table 2 that none of the catalysts lacking aromatic rings in the structure of the ‐R² substituents yield the azine product ( 1a , 1c , 1h , 3a, and 3c ). This suggests that some kind of π‐π or p‐π interaction is responsible for the selectivity control. All of the boreniums included in the kinetic experiments are better catalysts than B(C₆F₅)₃ ( 4a ), with 1c , 1e , 1f , and 3d having comparable k _obs. For the discussed model reaction, catalysts 1b , 1b’ , 1e , 2b , 3b , 3e, and 3 g display the fastest kinetics, having the observed kinetic constant an order of magnitude higher than the remaining compounds. The pattern here is clear, indicating the importance of flat, 5‐ and 7‐member dioxaboracycles. The contrasting results for 1 g (and 1g’ ) against 3 g can be explained by the significant repulsion from the mesityl‐ substituents on the labile 7‐membered cycle, which is causing the cycle to adopt a less flat conformation and more of a skewed chair‐like conformation. The BINOL moiety is less prone to this effect, as the k _obs. for 1e and 3e are much more comparable. Overall, the catechol‐derived boreniums are all very good catalysts for this model reaction, with 3b yielding azine product exclusively. In summary, the catalyst choice enables control over the reaction outcome and product distribution.

2.4.2. Hydrosilylation of Acetophenone

Reductions of aldehydes, ketones, imines, and enamines are standard model reactions used to assess the catalytic properties of Lewis acids (like BCF) [38, 39]. The readily available acetophenone was selected to react with Et₃SiH to form a silyl ether as the product, which can be easily hydrolyzed to 1‐phenylethanol and isolated via chromatography (Scheme 4, Table 3). The majority of the title compounds do not catalyze this process, even at 10 mol. % loading (and only traces of the product are found in some cases after 24 h at 55°C). In many of these cases, there is trace evidence of deoxygenation of acetophenone to ethylbenzene to a negligible degree. The hydrosilylation product can be obtained in high yield only for catalysts 1d , 3c , 3d , and clearly observed in the case of 1c (up to 29%), and a small amount (6%) for 1 g (after 24 h, at elevated temp.). Compounds 1c , 1d , 3c , 3d represent pairs of parent structures derived from phenylacetylene and t‐butyl acetylene in 2:1 hydroboration with IMes→BH₂ ⁺ NTf₂ ^– and IMe₂→BH₂ ⁺ NTf₂ ^–. Those are the only non‐cyclic and non‐oxygen‐containing structures in our series. The mechanism of this reaction was previously confirmed to proceed in the first step via hydride abstraction from Et₃SiH to the Lewis acid catalyst, and hydride donation from it in the last step [40, 41].

SCHEME 4.

Hydrosilylation of acetophenone catalyzed by the title compounds in a model reaction, optionally followed by hydrolysis to benzyl alcohol.

TABLE 3.

NMR yield for the silyl ether product of the hydrosilylation of acetophenone at r.t ^a .

Cat.	cat. 1 mol. %	cat. 10 mol. %
1a	0	0 (0) a
1b	0	0 (trace) a
1b'	0	0 (trace) a
1c	17	34 (29) b
1d	89 (85) b	98 (92) b (100) a
1e	0	trace (trace) a
1f	0	0 (trace) a
1g	0	0 (6) a
1g'	0	0 (0) a
1h	0	0 (0) a
2a	0	0 (0) a
2b	0	0 (0 a a
3a	trace	trace (trace) a
3b	0	0 (trace) a
3c	90 (84) b	99 (94) b
3d	99 (80) b	100 (85) b
3e	0	trace (trace) a
3g	0	0 (0) a
4a b	80 (100) a	84 (88) b
3 d c	—	100 (88) b
3 d d	—	100 (78) b
3d e	—	100 (69) b
3d f	—	100 (73) b
3d g	—	0

^*General reaction conditions: acetophenone (1 eq.), Et₃SiH (2 eq.), C₆D₆ for NMR reactions (0.5 mL) or toluene (for 1.5 mmol scale), 1 h.

^** 4a = tris(pentafluorophenyl)borane.

^a NMR yield of the reaction conducted over 24 h at 55 °C with 10 mol. % cat. loading.

^b Isolated yield of the benzyl alcohol (scale 1.5 mmol).

^c Reaction with 2,2‐dimethyl‐1‐phenylpropan‐1‐one.

^d Reaction with cyclohexyl(phenyl)methanone.

^e Reaction with 4‐chloroacetophenone.

^f Reaction with 4‐bromoacetophenone.

^g Reaction with di‐4,4´‐tolylmethanone.

In order to gain insight into the reaction pathway, representative catalysts 1b , 1 g , 1d , 3d , and 3e were combined with excess amounts of acetophenone and Et₃SiH separately and analyzed by NMR (See Figures S6–S11 in the Supporting Information). In the case of 3d , in reaction with Et₃SiH, there is clear evidence of adduct formation, as indicated by ¹¹B NMR signals observed in the region corresponding to tetrahedral boron. For 3e , there is evidence of catalyst decomposition, most probably resulting from the hydrosilylation of the catalyst B‐O bonds. Interestingly, cation 3e is also engaging in a C═O→B adduct formation in the presence of acetophenone, which is not observed for 3d . Catalysts 1b and 1g are not showing any interaction with either reagent in NMR, demonstrating the extent of the steric hindrance from the two mesityl‐ moieties, having an impact on interaction even with Et₃SiH. This outcome is somewhat unexpected, considering the observed strong interaction with TEPO for the majority of IMes‐stabilized boreniums. Therefore, catalyst 1d was also investigated under excess of both reagents, manifesting only slow conversion to hydride adduct after 30 min. (Figure S8, bottom), which confirms the steric influence and is in agreement with the observed hydrosilylation efficiency (slightly lower than 3c and 3d after 1 h). The somewhat lower efficiency for 1c can be also attributed to the combined steric hindrance from mesityl‐ and t‐Bu moieties, which is reflected in a lower AN assessed for this compound (AN = 89.4) and lower HIA, with respect to 1d , 3c , and 3d . For those four cases, the hydrosilylation was performed on a larger scale (1.5 mmol), and the product was hydrolyzed to alcohol and isolated by flash chromatography. In the case of 1 h (catalyst derived from 9‐BBN), no product was obtained even after 24 h at 55°C. This result stands in contrast to the very good performance displayed by the lutidine‐stabilized [9‐BBN‐LB]⁺[NTf₂]^– borenium, investigated by Denmark et al. in the same process [40]. The very low AN measured for 1 h (AN = 28.2) further confirms the combined steric effect. The NHC→BH₂ ⁺ species 1a , 2a , and 3a are not active as catalysts for this reaction. Most probably because they are undergoing reduction in the presence of Et₃SiH. In the case of 3a , traces of the product were observed, most probably resulting from the marginal pathway catalyzed by the in situ‐formed product of acetophenone hydroboration.

We have submitted other starting materials to this protocol with 3d as the best catalyst. Phenyl‐tert‐butyl ketone and phenyl‐cyclohexyl ketone were selected to represent higher steric hindrance. 4‐Chloroacetophenone and 4‐bromoacetophenone—to probe EWG presence and di‐4,4´‐tolylmethanone as a benzophenone representative. In all the mentioned cases (except benzophenone), the reaction was quite rapid and quantitative (see Table 3, bottom entries). The reaction with di‐4,4´‐tolylmethanone gave traces of an unidentified product, suggesting 3d enters a stoichiometric reaction with the starting material.

3. Conclusion

We have synthesized and characterized a series of 13 diverse borenium cationic compounds in pure form and assessed their catalytic behavior in two benchmark model reactions. The great majority of the title compounds are characterized by a very high AN (Acceptor Number) around 90–98 units, indicating a Lewis acidity significantly stronger than that of the neutral benchmark Lewis acid B(C₆F₅)₃. Both calculated FIA and HIA indices confirm this trend. However, the catalytic performance is diverse and depends strongly on the structural properties and thus the variable factors affecting the global Lewis acidity of the boron center and the overall stability of the borenium cations. All of the investigated catalysts are active in diazo‐homocoupling reaction, though with different efficiency and leading to a different product distribution. Notably, the choice of the catalyst allows for the control over the reaction outcome, leading, instead of tetraarylethylene, to the formation of the corresponding bis(diarylmethylene)hydrazine. Simplifying, catalysts with cationic boron included in a dioxacycle are performing much better in this reaction. In contrast, the situation is somewhat reversed in the case of hydrosilylation. Only the non‐cyclic, non‐oxygen‐containing variants are active in this process. This outcome shows that the catalytic reactivity of boreniums can be controlled by the structural factors to a great extent. This information can be particularly useful while preparing the catalyst directly in the reaction media, considering that all the title compounds were synthesized from the same in situ‐generated dihydridoborenium species.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information [42, 43, 44, 45, 46, 47, 48, 49]. CCDC Deposition Numbers [50].