Download PDF
Synlett 2023; 34(18): 2187-2192
DOI: 10.1055/a-2110-5359
cluster
Modern Boron Chemistry: 60 Years of the Matteson Reaction

Remote Back Strain: A Strategy for Modulating the Reactivity of Triarylboranes

Mahiro Sakuraba
a   Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
,
Taichi Morishita
a   Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
,
Taiki Hashimoto
a   Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
,
Sensuke Ogoshi∗
a   Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
,
Yoichi Hoshimoto  ∗
a   Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
b   Center for Future Innovation (CFi), Division of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan
› Author Affiliations

This project was supported by the Environment Research and Technology Development Fund (JPMEERF20211R01 to Y.H.) of the Environmental Restoration and Conservation Agency of the Ministry of the Environment of Japan, and Grants-in-Aid for Transformative Research Area (A) Digitalization-driven Transformative Organic Synthesis (JSPS KAKENHI Grant 22H05363 to Y.H.). Y.H. acknowledges financial supports from the Yazaki Memorial Foundation for Science and Technology, the Kansai Research Foundation for Technology Promotion, and the Takeharakenzai Alpsclean Co., Ltd. M.S. gratefully acknowledges a JST SPRING grant (JPMJSP2138).
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Abstract

A strategy for modulating the Lewis acidity of triarylboranes is proposed based on the concept of remote back strain. Steric repulsion and noncovalent interactions, both generated between the aryl meta-substituents of triarylboranes, are found to be critical for determining the strength of the remote back strain. Applying this concept, we synthesized B[2,6-F2-3,5-(TMS)2-C6H]3 and the liquid B[2,6-F2-3,5-(allyl)2-C6H]3 and we demonstrated their superior catalytic activity for the hydrogenation of quinoline relative to B(C6F5)3 or B(2,6-F2C6H3)3. Moreover, we established the first example of the catalytic hydrogenation of quinoline by using B[2,6-F2-3,5-(allyl)2-C6H]3 in the presence of a gaseous 1:1:1 molar mixture of H2, CO, and CO2.


#

Key words

triarylboranes - Lewis acids - frustrated Lewis pairs - hydrogenation - boron catalysis - hydrogen separation

Triarylboranes are representative Lewis acids that are widely applied as catalysts, activators, sensors, and bioimaging agents.[1] Especially in the field of main-group catalysis, recent progress on frustrated Lewis pairs (FLPs)[2] has brought significant structural diversification to halogenated triarylboranes beyond the archetypical B(C6F5)3. Accordingly, several strategies have been proposed for controlling the reactivity of triarylboranes, specifically through the control of their Lewis acidity.[3] [4] These strategies focus on regulating the accessibility (a kinetic aspect) and energy (a thermodynamic aspect) of the empty p orbitals at the boron center. In the latter case, strategies that substitute meta-F and/or para-F atoms in B(C6F5)3 with more- or less-electron-withdrawing substituents have been applied to prepare more- or less-electrophilic triarylborane derivatives through regulation of their electron affinity (i.e., their intrinsic Lewis acidity).[1a] [b] , [3] [5] Strategies that regulate the steric repulsion between Lewis base (LB) counterparts (front strain) through modulation of the size of the ortho-substituents on the aryl groups in the triarylborane have also been widely explored (Figure [1]; left).[6] Decreasing the front strain through the introduction of smaller ortho-substituents should enhance the global and effective Lewis acidity,[3] [5] whereas increasing this strain should result in the opposite outcome.[1] [2] [4] [7] For example, Ashely, O’Hare, and their co-workers demonstrated that heteroleptic species with the formula B(C6Cl5) n (C6F5)3–n (n = 1–3), equipped with C6Cl5 groups that are larger and more electron-withdrawing than the C6F5 groups, tend to exhibit increased intrinsic Lewis acidity and decreased effective Lewis acidity with increasing n.[7b] In addition, Soós and co-workers have developed a size-exclusion design to prepare heteroleptic triarylboranes, BAr2Ar′ with sterically demanding Ar′ groups, such as 2,6-Cl2C6H3 and 2,4,6-Me3C6H2 (Mes), for the catalytic hydrogenation of unsaturated molecules such as imines and quinolines.[7c] [e] These authors have also reported that the substitution of the meta-H atoms in the Ar′ groups with Cl atoms causes a negligible change in the Lewis acidity and catalytic activity.[7f] On the other hand, the Lewis acidity, and hence the catalytic reactivity, of triarylboranes has rarely been modulated by regulating the intramolecular repulsion between the Ar groups of tetrahedral LB–borane adducts (back strain) (Figure [1]; right).

Zoom Image
Figure 1 A schematic representation of the front strain and back strain generated between triarylborane (BAr3) and Lewis bases (LBs)

We recently investigated the hydrogenation of 2-methylquinoline by using a gaseous mixture of H2, CO, and CO2, and have demonstrated that the catalyst turnover number (TON) increased significantly when the meta-substituent R in B(2,6-Cl2C6H3)(2,6-F2-3,5-R2-C6H)2 was changed from F (TON 1000) to Cl (TON 1400) and, finally, to Br (TON 1520).[8] Given the nearly identical energy levels of the p orbitals on each boron center, we speculated that the size and shape of the meta-substituents (R in Figure [2]) might have a substantial impact on the Lewis acidity and reactivity of triarylboranes including BAr2Ar′ through control of the remote back strain.

Zoom Image
Figure 2 Triarylboranes B1 –B3 used in this work. The energy levels of the LUMOs (eV), calculated at the RI-DSD-PBEP86-D3BJ/ma-Def2-QZVPP//PBEh-3c/Def2-SVP level (gas phase), are also shown. a LUMO+1.

Here, we report the synthesis of novel homoleptic boranes B(2,6-F2-3,5-R2-C6H)3 (B1 : R = TMS; B2 : R = allyl) to evaluate the impact of the remote back strain on the global and effective Lewis acidity by comparison with B3 (R = H) (Figure [2]). We discuss the degree of remote back strain based on the changes in the relative Gibbs energies (ΔG°) for the formation of an adduct with selected LBs such as Et3P=O, H2O, CO, THF, and NMe3. In addition, we calculated deformation energies (E DEF)[3a] [c] [9] to evaluate the energetic penalty paid for the change in conformation at the boron center from trigonal planar to tetrahedral upon formation of the adduct. We also report the application of these boranes in the catalytic hydrogenation of quinoline by using H2 or H2/CO/CO2.

To explore the relationship between the remote back strain and Lewis acidity, model molecules B1 and B2 , which contain meta-TMS and allyl groups, respectively, were designed and synthesized. A preliminary computational study revealed that the intrinsic Lewis acidities of B1 and B2 can be expected to be comparable (Figure [2]). In fact, there is an insignificant difference with respect to the energy level of the molecular orbitals including p orbitals of B1 (LUMO = –0.20 eV) and B2 (LUMO + 1 = –0.22 eV), whereas the differences compared with B3 (LUMO = –0.30 eV) need to be carefully considered. Triarylboranes bearing electron-withdrawing substituents at the meta-positions would be inappropriate for this study because of the significant differences in the energy levels of the p orbitals on their boron atoms, as exemplified by B(2,3,5,6-F4C6H)3 (LUMO = –1.16 eV; calculated by using identical conditions to those shown in Figure [2]). In addition, a simulation using (2,6-F2-3,5-R2-C6H)BH2 (R = TMS or allyl) as model compounds to better understand the percentage of buried volume (%V bur)[10] revealed that the meta-TMS groups occupy a larger area (Δ%V bur +2.6–4.7%) around the B-aryl groups than do the meta-allyl groups [Supporting Information (SI), Figure S7]. We also confirmed that the influence of the buttressing effect,[11] caused by the meta-TMS and allyl groups, is probably negligible under the current conditions (SI; Figure S8).

Zoom Image
Scheme 1 (A) Synthesis of B1 and B2 ; yields of the isolated products are shown. For the detailed reaction conditions, see SI. (B) Molecular structure of B1 with thermal ellipsoids at 30% probability; H atoms and solvated C6H14 are omitted for clarity. (C) Molecular structure of CH3CN–B2 with thermal ellipsoids at 30% probability; H atoms and disordered atoms are omitted for clarity. Only one of two crystallographically independent molecules found in the asymmetric unit is shown.
Zoom Image
Scheme 2 (A) Reaction between B1 –B3 and Et3P=O (0.3 equiv) in CH2Cl2. (B) Molecular structure of Et3P=O–B1 with thermal ellipsoids at 30% probability; H atoms and disordered atoms are omitted for clarity. Selected bond length (Å) and angles (°): B–O 1.540(4), P–O 1.494(3), average B–CAr 1.639, B-O-P 149.7(2).

B1 was prepared by following the procedure shown in Scheme [1]A.[12] A stepwise introduction of two TMS groups was achieved with 1,5-Br2-2,4-F2-C6H2 (1) and TMSOTf by a silver-catalyzed transmetalation.[13] Iodination of the product afforded 2TMS , which was converted into B1 by treatment with i-PrMgCl in THF, followed by the addition of BF3∙OEt2. Eventually, B1 (95–98% purity) was obtained in the form of colorless crystals in 35% yield from a hot hexane solution, and analytically pure B1 was isolated in 4% yield after repeated recrystallization. The molecular structure of B1 was unambiguously confirmed by multinuclear NMR and single-crystal X-ray diffraction (SC-XRD) analyses (Scheme [1]B).[14] We also synthesized B2 in 23% yield from 2Allyl , which was prepared by a catalyst-free one-pot reaction between allyl bromide and a Grignard reagent generated from 1.[15] Interestingly, B2 is a liquid under ambient conditions and was therefore characterized by multinuclear NMR analysis. The introduction of the allyl groups onto the ortho-fluorinated triarylborane motif was further confirmed by SC-XRD analysis of the CH3CN–B2 adduct (Scheme [1]C).[14] The CH3CN–B2 adduct can be stored under ambient conditions (22 ℃, ~30% humidity) for at least seven days without apparent decomposition. The coordinated CH3CN moiety can be removed in vacuo after dissolving the adduct in toluene.

We then evaluated the effective Lewis acidity of B1 –B3 by using the Gutmann–Beckett method,[4] i.e., by comparing the deviation in chemical shifts (ΔδP) upon the formation of Et3P=O–B n (n = 1–3) relative to free Et3P=O (δP = 51.4 in CD2Cl2). A larger ΔδP value is expected for triarylboranes that exhibit higher Lewis acidity toward Et3P=O. As shown in Scheme [2]A, the value of ΔδP follows the order B2 (ΔδP = +20.3) < B1 (ΔδP = +21.2) ≈ B3 (ΔδP = +21.5), showing that the LUMO(+1) levels alone cannot be used to rationalize the relative Lewis acidities of B1 –B3 . The formation of Et3P=O–B n was also confirmed by using 11B NMR spectroscopy, and resonances that indicate the generation of four-coordinate boron species were observed at δB = –0.4 (B1 ), –0.3 (B2 ), and –1.4 (B3 ). Cooling of a 1:1 mixture of B1 and Et3P=O to –30 ℃ afforded colorless crystals suitable for SC-XRD analysis.[14] The molecular structure of Et3P=O–B1 is shown in Scheme [2]B. A 1:1 mixture of Et3P=O and B2 remained a liquid.

We then reproduced the geometrical parameters obtained for Et3P=O–B1 from the SC-XRD analysis by using DFT calculations at the PBEh-3c/Def2-SVP level (gas phase).[3a] The final single-point energy was calculated at the RI-DSD-PBEP86-D3BJ/ma-Def2-QZVPP, CPCMC(CH2Cl2) level and combined with the thermal corrections from the PBEh-3c calculations. Similarly, structural optimizations and single-point energy calculations were conducted for the corresponding adducts containing B2 and B3 based on the optimized geometry of Et3P=O–B1 . In addition, theoretical calculations were conducted on the reaction between B1 –B3 and a variety of LBs, including H2O, CO, THF, and NMe3. From the obtained geometrical parameters, we calculated τδ(B) to evaluate the degree of geometrical deviation from the ideal tetrahedral geometry at the boron center.[16] [17] The parameter τδ(B) is calculated from the relationship {360 – (α + β)/141 × β/α}, where α and β are the largest and second-largest C–B–C angles, respectively. When the boron center adopts an ideal tetrahedral geometry, the maximum value of τδ(B) is 1.0 and smaller values suggest a more distorted tetrahedral geometry. The values of τδ(B), ΔG° (ΔE°), and E DEF with respect to [B n + LB] (+0.0 kcal mol–1) are summarized in Figure [3].

Zoom Image
Figure 3 Relative Gibbs energies (kcal mol–1), ΔG°/ΔE° with respect to [B n + LB], and the deformation energies, E DEF, calculated for the formation of LB–B n at the RI-DSD-PBEP86-D3BJ/ma-Def2-QZVPP//PBEh-3c/Def2-SVP level (gas phase). The final single-point energies were calculated by using the CPCMC(CH2Cl2) implementation for the corresponding results in (A).

The τδ(B) values show that the geometries around each boron center in the LB–B n adducts, when the same LB is used, are nearly identical, regardless of differences in the meta-substituents. Moreover, the formation of Et3P=O–B1 (ΔG° = –8.0 kcal mol–1) is energetically more favorable than the formation of both Et3P=O–B3 (ΔG° = –6.8 kcal mol–1) and Et3P=O–B2 (ΔG° = –6.0 kcal mol–1) (Figure [3]A). Considering the relative energies of their LUMO(+1) levels (Figure [2]) and that TMS is sterically more demanding than the H and allyl substituents (Figure S7), we attribute the stabilization of Et3P=O–B1 to the noncovalent interactions (NCIs) arising from the TMS groups.[18] In fact, we found that several NCIs exist between the P-Et groups and meta-R groups in Et3P=O–B n (n = 1, R = TMS; n = 2, R = allyl), although a corresponding NCI could not be confirmed in Et3P=O–B3 (SI; Figures S14–S16). Furthermore, in the case of Et3P=O–B1 , NCIs also exist between the individual meta-TMS groups. The multiple NCIs that exist in Et3P=O–B1 would render its formation more exothermic than that of B3 . The NCIs between the individual meta-TMS groups probably play a critical role in enhancing the global and effective Lewis acidity of B1 with respect to B2 . When LB in the LB–B n adducts is H2O, CO, THF, or NMe3 (Figures [3]B–E), the ΔG° values increase in the order LB–B3 (m-H) < LB–B1 (m-TMS) < LB–B 2 (m-allyl). These results show good consistency with those obtained by using the Gutmann–Beckett method (Scheme [2]A). NCIs that exist between such small LBs and the meta-substituents are not expected to play a significant role in stabilizing the LB–B n adducts. Instead, the NCIs that exist between the individual meta-TMS groups should play a key role in making the formation of LB–B1 a more-exothermic process than the case of LB–B2 . These results are also consistent with the calculated E DEF values. For the LB–B n adducts explored in this work, E DEF increases in the order B3 (m-H) < B1 (m-TMS) < B 2 (m-allyl). Thus, E DEF can be used as an indicator to compare the relative strength of remote back strain when the front strain is comparable.

Subsequently, we examined the catalytic activity of B1 –B3 in the hydrogenation of quinoline (Qin) to give 1,2,3,4-tetrahydroquinoline (H4-Qin) (Scheme [3]). The catalytic hydrogenation of 2-substituted and/or 8-substituted derivatives of Qin has been developed by using H2 and triarylboranes, such as B(C6F5)3. The reaction proceeds through the generation of an FLP species in situ from boranes and the quinoline derivative (or a produced cyclic amine).[19] However, the hydrogenation of Qin itself remains challenging as it tends to form stable classical Lewis adducts with triaryboranes[20] and, consequently, the generation of an FLP species becomes unfavorable after the removal of the 2- and/or 8-substituents. In fact, it has been reported that B(C6F5)3 is an ineffective catalyst for the hydrogenation of Qin (TON = 0).[7d] In this context, modulation of the front strain based on the size-exclusion design has been demonstrated as an effective solution, and the heteroleptic catalyst B(2,3,5,6-F4 C6H)2(Mes) exhibits a TON of ~20 for the formation of H4-Qin.[7d]

Zoom Image
Scheme 3 (A) Catalytic hydrogenation of quinoline (Qin) in the presence of 5 mol% of B n ; yields of 1,2,3,4-tetrahydroquinoline (H4-Qin) were determined by using gas chromatography. † Catalyst TON after 24 h, using 2 mol% of B2 and 20 atm H2. (B) B2 -catalyzed hydrogenation of Qin using H2/CO/CO2 (20 atm each) as a model for crude H2.

The B3 -catalyzed hydrogenation of Qin initially furnished H4-Qin in 20% yield after eight hours at 100 ℃ (Scheme [3]A).[21] This result shows that lowering the intrinsic Lewis acidity of the catalyst is a key to promoting the reaction, as B3 has the virtually same front strain as B(C6F5)3. The hydrogenation of Qin was significantly accelerated by using catalysts with meta-TMS or meta-allyl groups, which furnished H4-Qin in yields of 42% and >99%, respectively. The TON reached 87 after 24 hours when 2 mol% of B2 was employed, clearly demonstrating the benefits of modulating the remote back strain. Finally, we applied B2 to the catalytic hydrogenation of Qin by using a 1:1:1 molar gaseous mixture of H2, CO, and CO2, a model of the crude H2 that is produced from a variety of carbon-based resources such as natural gas or biomass (Scheme [3]B).[8] [22] H4-Qin was obtained in >99% yield, demonstrating that Qin can be potentially used as an organic material that simultaneously separates and stores H2 directly from crude H2.

This study proposes a strategy of modulating the remote back strain of triarylboranes to finely tune their Lewis acidity by varying the size and shape of their aryl meta-substituents. A comparison of the effective and global Lewis acidities of B(2,6-F2-3,5-R2-C6H)3 (B1 , R = TMS; B2 , R = allyl; B3 , R = H) revealed that consideration of both the steric repulsion and noncovalent interactions generated between R groups is essential for successfully modulating the remote back strain. In this context, the deformation energy serves as an informative indicator for comparing the strength of the remote back strain between such triarylboranes. The triarylboranes studied here were also used as catalysts for the hydrogenation of quinoline, and a large improvement in the catalytic activity was observed when the catalyst was changed from B3 (m-H) to B1 (m-TMS) or B2 (m-allyl). Subsequently, the B2 -catalyzed hydrogenation of quinoline using a 1:1:1 molar gaseous mixture H2, CO, and CO2 was achieved for the first time. These results can be expected to pave the way for further progress in advanced main-group catalytic processes and molecular-based H2 purification technologies.


#

Conflict of Interest

Y.H. has a pending patent application (JP2023-95773) related to the results discussed in this study.

Supporting Information

    Supporting information for this article is available online at https://doi-org.accesdistant.sorbonne-universite.fr/10.1055/a-2110-5359.
  • Supporting Information

Corresponding Authors

Sensuke Ogoshi
Department of Applied Chemistry, Faculty of Engineering, Osaka University
Suita, Osaka 565-0871
Japan   

Yoichi Hoshimoto
Department of Applied Chemistry, Faculty of Engineering, Osaka University
Suita, Osaka 565-0871
Japan   

Publication History

Received: 30 May 2023

Accepted after revision: 14 June 2023

Accepted Manuscript online:
14 June 2023

Article published online:
14 September 2023

© 2023. Thieme. All rights reserved

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany


   Permissions and Reprints All articles of this category
Zoom Image
Figure 1 A schematic representation of the front strain and back strain generated between triarylborane (BAr3) and Lewis bases (LBs)
Zoom Image
Figure 2 Triarylboranes B1 –B3 used in this work. The energy levels of the LUMOs (eV), calculated at the RI-DSD-PBEP86-D3BJ/ma-Def2-QZVPP//PBEh-3c/Def2-SVP level (gas phase), are also shown. a LUMO+1.
Zoom Image
Scheme 1 (A) Synthesis of B1 and B2 ; yields of the isolated products are shown. For the detailed reaction conditions, see SI. (B) Molecular structure of B1 with thermal ellipsoids at 30% probability; H atoms and solvated C6H14 are omitted for clarity. (C) Molecular structure of CH3CN–B2 with thermal ellipsoids at 30% probability; H atoms and disordered atoms are omitted for clarity. Only one of two crystallographically independent molecules found in the asymmetric unit is shown.
Zoom Image
Scheme 2 (A) Reaction between B1 –B3 and Et3P=O (0.3 equiv) in CH2Cl2. (B) Molecular structure of Et3P=O–B1 with thermal ellipsoids at 30% probability; H atoms and disordered atoms are omitted for clarity. Selected bond length (Å) and angles (°): B–O 1.540(4), P–O 1.494(3), average B–CAr 1.639, B-O-P 149.7(2).
Zoom Image
Figure 3 Relative Gibbs energies (kcal mol–1), ΔG°/ΔE° with respect to [B n + LB], and the deformation energies, E DEF, calculated for the formation of LB–B n at the RI-DSD-PBEP86-D3BJ/ma-Def2-QZVPP//PBEh-3c/Def2-SVP level (gas phase). The final single-point energies were calculated by using the CPCMC(CH2Cl2) implementation for the corresponding results in (A).
Zoom Image
Scheme 3 (A) Catalytic hydrogenation of quinoline (Qin) in the presence of 5 mol% of B n ; yields of 1,2,3,4-tetrahydroquinoline (H4-Qin) were determined by using gas chromatography. † Catalyst TON after 24 h, using 2 mol% of B2 and 20 atm H2. (B) B2 -catalyzed hydrogenation of Qin using H2/CO/CO2 (20 atm each) as a model for crude H2.