Boron-catalysed hydrogenolysis of unactivated C(aryl)–C(alkyl) bonds

The hydrogenolysis of C–C bonds is one of the most important processes in the petroleum industry. These transformations typically rely on heterogeneous catalysts and take place at high temperatures and high pressures with limited selectivity. Employing homogeneous transition metal catalysts, while allowing the hydrogenolysis of C–C bonds to proceed under much milder conditions, is only suitable for substrates containing strained C–C bonds or directing groups. Here we report that a borenium complex can catalyse the selective hydrogenolysis of unstrained C(aryl)–C(alkyl) bonds of alkylarenes in the absence of directing groups at ambient temperature, affording the corresponding alkanes and arenes. Mechanistic studies suggest a reaction pathway that involves a synergistic activation of dihydrogen by the borenium complex and alkylarenes, followed by retro-Friedel–Crafts reaction to cleave the C(aryl)–C(alkyl) bonds. The synthetic utility of this protocol is demonstrated by the conversion of post-consumer polystyrene into valuable benzene and phenylalkanes with mass recovery rates above 90%. Hydrogenolysis of unactivated C(aryl)–C(alkyl) bonds is a challenging task even in the presence of metal catalysts. Now, an approach using a boron catalyst is described that facilitates the hydrogenolysis of alkylarenes under mild conditions, and its utility is demonstrated by degrading polystyrene waste into benzene and phenylalkanes.

The hydrogenolysis of C-C bonds is one of the most important processes in the petroleum industry. These transformations typically rely on heterogeneous catalysts and take place at high temperatures and high pressures with limited selectivity. Employing homogeneous transition metal catalysts, while allowing the hydrogenolysis of C-C bonds to proceed under much milder conditions, is only suitable for substrates containing strained C-C bonds or directing groups. Here we report that a borenium complex can catalyse the selective hydrogenolysis of unstrained C(aryl)-C(alkyl) bonds of alkylarenes in the absence of directing groups at ambient temperature, affording the corresponding alkanes and arenes. Mechanistic studies suggest a reaction pathway that involves a synergistic activation of dihydrogen by the borenium complex and alkylarenes, followed by retro-Friedel-Crafts reaction to cleave the C(aryl)-C(alkyl) bonds. The synthetic utility of this protocol is demonstrated by the conversion of post-consumer polystyrene into valuable benzene and phenylalkanes with mass recovery rates above 90%.
The catalytic hydrogenolysis of C-C bonds, a process involving rupture of C-C bonds via reaction with dihydrogen, plays a central role in the petroleum industry, upgrading low-value feedstocks to high-value fuels or commodity chemicals [1][2][3] . Recently, the hydrogenolysis of C-C bonds has also attracted attention as a viable way for the chemical recycling of polyolefin plastic waste 4,5 . Due to the large bond dissociation energy and non-polar nature of C-C bonds, industrial hydrogenolysis of C-C bonds typically takes place at temperatures above 300 °C with highpressure hydrogen (40-200 bar). The substantial energy requirements of this process provide a strong incentive to develop catalytic systems that can operate at ambient conditions. Several reactions resulting in the homogeneous hydrogenolysis of C-C bonds have been reported using transition metal catalysts under milder conditions 6 . The key step in these transformations is the cleavage of C-C bonds via oxidative addition, which takes place prior to the activation of dihydrogen. However, to ensure that the oxidative addition of sterically shielded C-C bonds can outcompete more accessible C-H bonds, specifically designed substrates, such as those incorporated with strained C-C bonds [7][8][9] or preinstalled directing groups [10][11][12][13][14] , are required, thereby limiting the scope of this approach.
In recent decades, systems based on main-group elements have attracted increasing interest as a result of their potential to mimic or even transcend the reactivities of their transition-metal counterparts in catalysis [15][16][17] . To circumvent the inherent limitation associated with transition-metal catalysts in C-C bond hydrogenolysis, we set out to explore the possibility of catalysts based on main-group elements for the hydrogenolysis of C(aryl)-C(alkyl) bonds of alkylarenes, a transformation also known as hydrodealkylation and widely applied in the production of arenes in the petroleum industry ( Fig. 1a) 2,3,18 . The presence of unsaturated arene moieties in this transformation, however, represents an extra challenge because of competing arene hydrogenation reactions 19 . The first example of homogeneous hydrogenolysis of unactivated C(aryl)-C(alkyl) bonds was reported by Milstein and co-workers 10 , in which methylarenes with two pendant phosphine arms can be converted to the corresponding arenes and methane in the presence of rhodium catalysts under dihydrogen at 150 °C (Fig. 1b). The two phosphine moieties, which cannot be readily removed, are indispensable for the observed reactivity, as formation of stable metallacycle intermediates provides the driving force for the oxidative addition of the C(aryl)-C(alkyl) bond 20 .

Screening of reaction conditions
We began our study by choosing (1,1-dimethylpropyl)benzene (2) as the model substrate. Combining 2 with 10 mol% of 1a under 1 bar of dihydrogen in bromobenzene resulted in the formation of benzene and isopentane with 97% and 91% yields, respectively, as determined by gas chromatography-mass spectrometry (GC-MS) after 24 h at 25 °C (Supplementary Table 3). Increasing the pressure of dihydrogen to 20 bar allows the reaction to reach full conversion with a much lower catalyst loading (1 mol%). No arene hydrogenation products were formed under these reaction conditions as indicated by the 13 C{ 1 H} NMR spectrum of the crude reaction mixture ( Supplementary Fig. 4). Other haloarenes, such as C 6 H 5 Cl, C 6 H 5 F or o-C 6 H 4 F 2 , are also suitable solvents for the hydrodealkylation reaction. In the absence of dihydrogen, both 2 and 1a remained unchanged in C 6 D 5 Br after 24 h at 25 °C. Replacing 1a with less electrophilic [IMe 4 B(H)Cb][B(C 6 F 5 ) 4 ] (IMe 4 = 1,3, 4,5-tetramethylimidazol-2-ylidene, Cb = o-carboran-1-yl, 1b) 34 resulted in lower yields (23% and 5% for benzene and isopentane, respectively) even at 50 °C (Supplementary Table 3). Neutral organoborane B(C 6 F 5 ) 3 and strong Brønsted acid CF 3 SO 3 H are completely ineffective as catalysts. As inductively coupled plasma mass spectrometry analysis revealed 1a containing trace amounts of transition metals (Supplementary Table 1), a number of transition-metal salts were examined as catalysts but none of them showed any activity (Supplementary Table 3).

Evaluation of substrate scope
With our optimized conditions in hand, we explored the scope of this transformation, starting with the alkyl moieties of the alkylarenes (Fig. 2). Tertiary-alkyl-substituted benzenes, such as t-butylbenzene (3) and 1-phenyl-adamantane (4), were efficiently converted to benzene and the corresponding alkanes. Secondary-alkyl-substituted benzenes (5)(6)(7)(8)(9)(10) are also suitable substrates. Interestingly, for 1-(4-chlorophenyl)-1-phenylethane (8), the formation of benzene is favoured over electron-poor chlorobenzene, in agreement with the proposed role of arenes as π bases in the dihydrogen activation process. For 1-phenyltetralin (9), selective hydrogenolysis of the exo-ring C(aryl)-C(alkyl) bond was observed. Cumene (11) suffers from During the review process, another example of hydrogenolysis of C(aryl)-C(alkyl) bonds was published, which employed rhodium catalysts using a similar directing-group-assisted strategy 14 . Here we present a borenium-catalysed hydrogenolysis of unstrained C(aryl)-C(alkyl) bonds of alkylarenes in the absence of directing groups at ambient temperature. Experimental and theoretical studies suggest a reaction pathway that involves a synergistic activation of dihydrogen by the borenium complex and alkylarenes, followed by retro-Friedel-Crafts reaction to cleave the C(aryl)-C(alkyl) bonds. This catalytic system can be utilized for the upcycling of post-consumer polystyrene into valuable benzene and phenylalkanes at 60 °C under 5 bar of dihydrogen with mass recovery rates above 90%.

Design plan
Our hydrodealkylation approach was inspired by frustrated Lewis pair chemistry pioneered by Stephan and Erker [21][22][23][24][25] . One of the most important discoveries in this field is that dihydrogen can be heterolytically cleaved when simultaneously interacting with an electron donor and an electron acceptor [26][27][28][29] . We speculated whether an alkylarene, as a π-electron donor, would be capable of activating dihydrogen in the presence of a Lewis acid, which would lead to the formation of a hydride species and a Wheland complex (I, Fig. 1c). A retro-Friedel-Crafts reaction of the Wheland complex I would yield the arene and a carbenium intermediate 30,31 . Subsequent hydride abstraction from the generated hydride species by the carbenium intermediate would afford the alkane and the Lewis acid catalyst, thus completing the catalytic cycle. Given that alkylarenes are much weaker electron donors than Lewis bases typically applied in frustrated Lewis pair chemistry (such as amines and phosphines), we reasoned that an exceptionally strong Lewis acid with high hydride affinity would be needed for smooth dihydrogen activation under mild conditions 28,29 . Furthermore, steric protection around the Lewis acid centre would be necessary to hinder the hydride transfer from the Lewis acid centre to the Wheland complex I, a side reaction that would lead to the hydrogenation of aromatic rings 32 . Recently, we reported the synthesis of an N-heterocyclic carbene (1,3-bis(2,3, 4,5,6-pentafluorobenzyl)imidazol-2-ylidene, IBn F )-stabilized o-carboranyl-substituted borenium complex 1a, which is Lewis acidic enough to activate methane under relatively mild conditions 33 .   Article https://doi.org/10.1038/s41929-022-00888-y competing transalkylation, which affords alkylated bromobenzene as a by-product due to the reaction between 2-propyl cation intermediate and solvent bromobenzene. Similarly, diphenylmethane (12) also undergoes transalkylation to yield benzene and alkylated bromobenzene. When the phenyl groups of diphenylmethane were replaced with bulkier mesityl groups, the reaction proceeded smoothly to yield mesitylene and 1,2,3,5-tetramethylbenzene in o-C 6 H 4 F 2 . Toluene and ethylbenzene failed to undergo hydrogenolysis even with 20 mol% of 1a at 50 °C under 20 bar of dihydrogen ( Supplementary Fig. 35).

Mechanistic investigations
To understand the mechanism of this hydrodealkylation reaction, we conducted kinetic experiments to determine the order of the reaction in each reactant by monitoring the hydrogenolysis of 6 in C 6 D 5 Br by 1  . This is consistent with our proposed reaction pathway entailing synergistic activation of dihydrogen with 1a and 6. The involvement of dihydrogen in the rate law also indicates that the activation of dihydrogen takes place either prior to or during the turnover-determining step. Replacing dihydrogen with dideuterium resulted in the formation of Article https://doi.org/10.1038/s41929-022-00888-y bond in the turnover-determining step. To gauge the impact of electron density of the aryl ring on the rate of the hydrogenolysis reaction, we examined the initial rates of a series of para-substituted (1,2,3,4-tetra hydronaphthalen-1-yl)benzene, and created a Hammett plot by plotting log(k X /k H ) against substituent parameters σ p (Fig. 3b). A negative Hammett parameter ρ of −2.46 was observed, indicative of increased reactivity for electron-rich arenes, consistent with the proposed role of the arenes as π bases in the activation of dihydrogen.
To provide further insight into the mechanism, we investigated the hydrogenolysis of 6 with density functional theory (M06-2X) calculations (Fig. 3c) 36 . Catalyst 1a (the [B(C 6 F 5 ) 4 ] − anion was omitted during the calculation) first forms a high-energy dihydrogen adduct 1a-H 2 with H 2 , which in turn reacts with 6 to afford Wheland complex 6-H + and neutral borane 1a-H via transition state TS1. At TS1, the H-H distance is elongated to 1.22 Å with one hydrogen atom (H1) interacting with the borenium centre (B-H1 = 1.25 Å) and the other hydrogen atom (H2) interacting with the ipso-carbon atom of one of the phenyl rings of 6 (C-H2 = 1.32 Å). A similar transition state was also proposed by Grimme and Stephan for a B(C 6 F 5 ) 3 -mediated hydrogenation of anilines 32 . Natural bond orbital analysis of TS1 shows that 0.79e charge transfers from the σ-bonding orbital of H1-H2 to the empty p orbital of B1 with the σ* antibonding orbital of H1-H2 receiving 0.34e charge from the π orbitals of one of the phenyl rings of 6 ( Supplementary Fig. 137). Natural population analysis revealed that the H1 atom is neutral with 0.00e, and the H2 atom bears a positive charge of +0.44e, indicating the heterolytic nature of the H-H bond activation. In agreement with experimental observation, the formation of 1-phenylethylium and benzene from the Wheland complex 6-H + is the turnover-determining step which needs to overcome an overall free energy barrier of 28.1 kcal mol −1 , comparable to the experimental value of 25.5 ± 2.2 kcal mol −1 (298 K). The subsequent barrier for the generation of ethylbenzene and catalyst 1a is very small (6.1 kcal mol −1 ). The overall reaction is exergonic with ΔG = −10.5 kcal mol −1 at 298 K. Alternative reaction pathways in which the protonation occurs on the para, meta or ortho carbon of the phenyl ring in the dihydrogen activation step were also explored ( Supplementary Fig. 138). While the free energy barriers for the dihydrogen activation occurring on meta or ortho carbon (28.0 and 29.1 kcal mol −1 , respectively) are higher compared to TS1, the one for para carbon (25.1 kcal mol −1 ) is 2.0 kcal mol −1 lower compared to TS1. However, the following intramolecular proton migrations on the phenyl ring to form intermediate 6-H + need to overcome higher overall free energy barriers in comparison with TS1 (27.9, 29.4 and 27.8 kcal mol −1 for proton migration from para to meta, meta to ortho, and ortho to ipso positions, respectively), rendering these pathways less kinetically favourable.

Hydrogenolysis of polystyrene
Besides small organic molecules, we also examined the possibility of applying this hydrogenolysis system to recycle aromatic chemicals from polystyrene, the most widely used aromatic plastic 37,38 . Stirring a solution of laboratory-grade polystyrene (PS1, 106 mg; M w = 3.10 × 10 5 g mol −1 ; polydispersity index (PDI) = 2.01) and 51 μmol of 1a (5 mol% per styrene unit) in 10 ml of o-C 6 H 4 F 2 under 5 bar of dihydrogen at 60 °C for 16 h resulted in the formation of 52 mg of benzene (49 wt%) as determined by GC analysis (Fig. 4). This corresponds to 65% of yield based on the phenyl moieties, comparable to the state-of-art heterogeneous polystyrene hydrogenolysis catalyst Ru/Nb 2 O 5 (65% yield of benzene at 300 °C) 39 . Importantly, benzene is the only low-boiling-point (<200 °C) product that can be detected by GC analysis (Supplementary Figure 58). This selectivity is in stark contrast to the complex monoarene products obtained from pyrolysis or hydrocracking of polystyrene [39][40][41] . In addition to benzene, long-chain phenylalkanes were obtained as 47 mg of colourless oil (44 wt%) after the removal of the catalyst by flash column chromatography. Gel-permeation chromatography analyses showed that the M w of this oil is 674 g mol −1 with a PDI of 1.64, indicating that the main chain of polystyrene is also cleaved during the hydrogenolysis process. We tentatively assumed that the main-chain C-C bonds could be cleaved via β scission of the carbenium intermediates 42 , releasing terminal alkenes which could subsequently be hydrogenated in the presence of 1a (for the proposed reaction pathway, see Supplementary Fig. 80) 43 . The 1 H NMR analyses confirmed the absence of olefins in the phenylalkane product, as no signals were observed in the region between 3.5 and 6.5 ppm. The integration ratio of the aromatic and aliphatic proton signals is 1:3.95, which led us to estimate that ∼81% of phenyl moieties of polystyrene were replaced with hydrogen atoms (for deduction details, see Supplementary Methods: Procedure for borenium-catalysed hydrogenolysis of polystyrene), correlating relatively well with the yield of benzene determined by GC analysis. The low percentage of CH 3 resonances (0.8-1.0 ppm) among the total integration of the aliphatic protons (∼6%) indicates that the phenylalkane product is mainly composed of phenyl-substituted linear alkanes, which are valuable precursors for the production of biodegradable surfactants 3,44 . Taken together, ∼93% of the weight of polystyrene was successfully converted to valuable aromatic chemicals. To test if this catalytic system could be utilized to upcycle waste polystyrene, we carried out the hydrogenolysis reaction with post-consumer polystyrene from a single-use plastic cup (PS2, M w = 1.76 × 10 5 g mol −1 ; PDI = 2.12) and expanded polystyrene foam (PS3, M w = 2.86 × 10 5 g mol −1 ; PDI = 2.21). Comparable yields of benzene (49 and 47 wt% for PS2 and PS3, respectively) and phenylalkanes (42 wt%, M w = 600 g mol −1 , PDI = 1.45 for PS2; 44 wt%, M w = 671 g mol −1 , PDI = 1.54 for PS3) from these post-consumer polystyrene plastics were obtained, revealing remarkable tolerance of the catalytic system to common processing impurities.

Conclusions
In the presence of a boron Lewis acid catalyst, we have achieved the hydrogenolysis of the C(aryl)-C(alkyl) bonds of unactivated alkylarenes Article https://doi.org/10.1038/s41929-022-00888-y without directing groups. The potential of this catalytic system was demonstrated by its application to the conversion of post-consumer polystyrene into valuable benzene and long-chain phenylalkanes under near-ambient temperature with selectivity-a process that had previously remained elusive for homogeneous transition-metal systems.
Although the turnover numbers of the catalyst still need improvement, the approach reported herein enables a top-down strategy for the synthesis of organic molecules from complex precursors, which remains a challenge for homogeneous transition-metal systems.

General procedure for borenium-catalysed hydrodealkylation of alkylarenes
Inside a nitrogen-filled glovebox, 1a and alkylarene were dissolved in C 6 H 5 Br (0.5 ml) and transferred into the well of an autoclave. After the head of the autoclave was assembled, the autoclave was taken out of the glovebox and connected to a dihydrogen gas cylinder. The connection line was flushed with dihydrogen three times. Then the autoclave was pressurized to 20 bar of dihydrogen and allowed to stand at room temperature for 24 h. The autoclave was then cooled to −40 °C in a cold ethanol bath for 15 min, followed by releasing the dihydrogen pressure.
After removing the head of autoclave, octane and diphenylmethane were added into the well of an autoclave as internal standards. The resulting reaction mixture was then subjected to GC-MS analysis. The identities of products were confirmed by comparison of the mass spectra and retention times of the products with those of authentic compounds.

Data availability
All data supporting the findings of this study are available within the Article and its Supplementary Information or from the corresponding authors upon reasonable request.