Main group catalysis for H₂ purification based on liquid organic hydrogen carriers

Abstract

Molecular hydrogen (H₂) is one of the most important energy carriers. In the midterm future, a huge amount of H₂ will be produced from a variety of hydrocarbon sources through conversion and removal of contaminants such as CO and CO₂. However, bypassing these purification processes is desirable, given their energy consumption and environmental impact, which ultimately increases the cost of H₂. Here, we demonstrate a strategy to separate H₂ from a gaseous mixture of H₂/CO/CO₂/CH₄ that can include an excess of CO and CO₂ relative to H₂ and simultaneously store it in N-heterocyclic compounds that act as liquid organic hydrogen carriers (LOHCs), which can be applied to produce H₂ by subsequent dehydrogenation. Our results demonstrate that LOHCs can potentially be used for H₂ purification from CO- and CO₂-rich crude H₂ in addition to their well-established use in H₂ storage.

A H₂ purification system has been achieved based on LOHCs, which goes beyond the well-established use in H₂ storage.

INTRODUCTION

Molecular hydrogen (H₂) is an essential reductant that has been widely used in, e.g., petroleum refineries, the industrial production of ammonia and methanol, and the chemical industry. Moreover, H₂ is one of the most promising energy carriers of the future, given its high stability and thus transportability, its high gravimetric energy density, and the low environmental impact of its combustion product compared to those of hydrocarbon-based energy sources (1–4). These features make H₂ an attractive candidate for the construction of a greener and sustainable economy, which is commonly referred to as the “hydrogen economy” (3). Thus, it can be expected that a huge amount of H₂, on a magnitude of more than 10¹² standard cubic feet per year, will be produced from a wide range of hydrocarbon and renewable resources (1, 2). In this context, H₂ production combined with CO₂ capture and storage from hydrocarbon resources such as petroleum, coal, natural gas, and biomass represents a pragmatic choice for the midterm future due to the limited supply of renewable energy (2), while the electrolysis of water using electricity obtained from renewable resources seems to be an alternative option in the long-term future (5). The predominant contemporary route to H₂ production includes the intensive purification of crude H₂, which is a gaseous mixture of H₂, CO, CO₂, and other components that is produced by gasification, reforming, and/or water-gas shift (WGS) (process I in Fig. 1A). Purification processes such as pressure swing adsorption (PSA), membrane separation, and cryogenic separation critically determine the purity of the H₂, which is sometimes required to exceed 99.99% for fuel cells, and influence the total energy consumption of the H₂ production process, making it cost inefficient. Notable advances have been made to improve the efficiency, H₂ recovery rate, and reproducibility of H₂ purification processes (4). Nevertheless, an approach that could fundamentally solve all these challenging issues remains to be found (1). Thus, although H₂ can currently be stored after or during the process I in Fig. 1A (6), we envisaged a solution where H₂ could be stored in its carrier directly from crude H₂, which often includes more CO than H₂, without the requirement for any of the aforementioned shift and purification processes (process II in Fig. 1A) (7). Moreover, the recovery of H₂ after our proposed path ultimately leads to the production of highly pure H₂.

Fig. 1. Research background and concept of this study.

(A) Simplified schemes of representative contemporary routes of H₂ purification followed by H₂ storage (process I) and a conceptually novel route involving the simultaneous separation and storage of H₂ from crude H₂ (process II); WGS, water-gas shift; PSA, pressure swing adsorption. (B) Schematic illustration of the general concept behind LOHCs and the critical differences between well-established methods and this work. (C) Potential reactions among H₂, CO, CO₂, LA (Lewis acid/acidic part), and/or LB (Lewis base/basic part). Dashed arrows represent backward reactions that do not always occur under the same conditions as the corresponding forward reaction.

To this end, we focused on the use of liquid organic hydrogen carriers (LOHCs), which have been widely used for H₂ storage and transportation (8–12). H₂ storage systems with LOHCs are based on a reaction sequence in which a H₂ lean state (H₀-LOHC) is hydrogenated to produce a H₂-rich state (H_x-LOHC), followed by a subsequent dehydrogenation of H_x-LOHC after storage/transport that regenerates H₂ and H₀-LOHC (Fig. 1B). The use of LOHCs has been extensively researched, as the technical, environmental, and economic advantages of H₂ storage using LOHCs are widely accepted (8). Furthermore, the pool of potential candidates for H₀-LOHCs has recently been expanded from the well-studied aromatic hydrocarbons to include heteroaromatics (11–13), cyclic dipeptides (14), amides (15, 16), cyclic ureas (17), and oligoesters (10), some of which provide hydrogenated products (H_x-LOHC) with H₂ storage capacity [H₂ weight % (wt %)] values that exceed the practical guidelines proposed by the European Union (5.0 wt %) and the U.S. government (5.5 wt %) (8). However, hitherto reported H₂ storage systems using LOHCs have predominantly been based on the use of transition metal catalysts. This presents a critical issue for H₂ separation in the presence of CO and CO₂, both of which can severely inhibit transition metal–catalyzed hydrogenation reactions (6, 9). We have successfully showed a strategy for H₂ separation from multicomponent gas mixtures such as H₂/CO/CO₂ via the main group–catalyzed hydrogenation of organic molecules in 2017 (7). Note here that Breakman-Danheux et al. (18) in 1996 and Jorschick et al. (19) in 2019 have independently reported transition metal–based heterogeneous compounds that have been applied to the hydrogenation of hydrocarbon-based LOHCs using contaminated H₂ including CO, CO₂, and/or gaseous hydrocarbons, which led to a marked suppression of the catalytic activity by CO despite the great excess of H₂ present.

Against this background, we have focused on main group catalysis (20, 21) including the use of frustrated Lewis pairs (FLPs) that are composed of Lewis bases (LBs) and triaryl boranes as the Lewis acids (LAs) (22–24). Triaryl boranes of the type Bⁿ such as B(C₆F₅)₃ (B¹) have been reported to catalyze the hydrogenation of N-heteroaromatic compounds such as 2-methylquinoline (Qin) under diluted conditions (25–27). FLPs are well known to mediate the heterolytic cleavage of the H─H bond to generate [LB─H][H─LA] species (Fig. 1C, I) (23). The subsequent proton/hydride transfer from [LB─H][H─LA] to N-heteroaromatic compounds facilitates the storage of H₂ (26). CO₂ fixation by FLPs has also been widely studied and found to proceed in either a reversible or irreversible manner (e.g., Fig. 1C, II) (28, 29). The hydrogenation of CO₂ has been reported in the presence of FLPs that are composed of B¹ and nitrogen-based LBs (Fig. 1C, III) (29–32). CO can reversibly bind to the boron center (Fig. 1C, IV), which would kinetically affect the H₂ cleavage step (29, 33). Moreover, these gases contain a certain amount of H₂O, which often triggers the decomposition of triaryl boranes to yield, e.g., [LB─H][HO─LA], although sophisticated strategies to minimize the influence of H₂O have been reported (34, 35). The reactions shown in Fig. 1C (II to VI) can seriously affect the progress of the targeted catalytic hydrogenation in the presence of CO, CO₂, and H₂O when the undesired paths are irreversible (or nearly irreversible). Therefore, a suitable triaryl borane that exhibits sufficient reactivity toward H₂ in cooperation with LBs, yet simultaneously avoids the aforementioned irreversible deactivation paths, would be highly desirable. Note that Voicu et al. (36) successfully applied an FLP comprising B¹ and P^tBu₃ to the microfluidic separation of ethylene and ethane.

Here, we demonstrate the direct storage of H₂ in N-heterocyclic compounds such as H₄-Qin and 2,6-dimethylpiperidine (H₆-Lut) under various mixed gas conditions including H₂, CO, CO₂, and CH₄ via a shelf-stable Bⁿ-catalyzed hydrogenation of Qin and 2,6-lutidine (Lut), respectively. Furthermore, the same Bⁿ also catalyzes the dehydrogenation from H₄-Qin to produce H₂ with concomitant generation of Qin. The molar compositions of H₂/CO/CO₂ used in this work (1/1/1, 1/5/1, and 1/1/5) were based on the molar composition of the typical crude H₂ produced by hydrocarbon resources (H₂/CO/CO₂ = 1/1/0.2 to 1/2/0.5) and the typical PSA off-gas (H₂/CO/CO₂ = 1/0.1/2), albeit these compositions vary depending on the feedstock (4). In addition, these gases include up to 0.9 mmol of H₂O (table S2), which should be considered under the applied conditions.

RESULTS

The interconversion between Qin and H₄-Qin was used as a model LOHC system to separate H₂ under these mixed gas conditions (Fig. 2). As expected, well-established transition metal complexes based on Rh (TM¹), Ru (TM²), or Ir (TM³) (37) did not catalyze the hydrogenation of Qin (1.5 M in toluene) using a gaseous mixture of H₂/CO/CO₂ (4 atm each; runs 1 to 3), whereas H₄-Qin was efficiently yielded when H₂ (99.95% purity, 4 atm) was used in the cases of TM² and TM³ (38). A higher yield of H₄-Qin (12%) was observed when 1 mole percent (mol %) B¹ was subjected to these mixed gas conditions (run 4A), although the deleterious influence of the contaminants (CO, CO₂, and/or H₂O) was again observed compared to the result obtained using pure H₂ (run 4B). Encouraged by this result, we explored the triaryl boranes B² to B⁶, which have been developed by Stephan et al. (39), Ashley et al. (40), and Soós et al. (26, 41, 42) (runs 5 to 9). The decreasing Lewis acidity exhibited by the boron centers when the C₆F₅ group in B¹ were replaced with three para–H-C₆F₄ groups (B²) or with two para–H-C₆F₄ groups and a 2,6-Cl₂-C₆H₃ group (B⁵) was found to be critical, and H₄-Qin was afforded in 63 and 81% yield when B² and B⁵ were used, respectively (runs 5 and 8). We thus carried out additional structural modifications via the substitution of the meta-F atoms with respect to the boron atom in B⁵ with Cl atoms (B⁷), H and Cl atoms (B⁸), Br atoms (B⁹), and (CF₃)₂C₆H₃ groups (B¹⁰) (runs 10 to 13). B⁹ showed the best result, affording H₄-Qin in 84% yield even in the presence of CO and CO₂ (run 12). Moreover, a significant enhancement in the hydrogenation of Qin was observed when the reactions were conducted using B⁵, B⁷, B⁹, and B¹⁰ in the absence of solvent (fig. S17); B⁹ exhibited a catalyst turnover number (TON) of 1520 at 100°C in the presence of H₂/CO/CO₂ (30 atm each), which is far higher than the TONs obtained using B⁵ (1000), B⁷ (1400), or B¹⁰ (1340) (runs 8, 10, 12, and 13; see also fig. S18). Note that the TON eventually reached to 2960 when the B⁹-catalyzed hydrogenation of Qin was carried out under the solvent-free conditions including H₂ (85 atm). The differences observed among B⁵, B⁷, B⁹, and B¹⁰ can be rationalized in terms of the electronic and steric properties of the meta-substituents, i.e., their electron-withdrawing ability, which influences the electron-accepting ability of the boron center, and their steric size, which should cause intramolecular steric repulsion among the introduced aryl groups (table S4 and fig. S44) (43). In this context, the larger size of the Br atoms in B⁹ compared to the F (B⁵) and Cl (B⁷) atoms can be expected to play a key role in maintaining high activity under the applied mixed gas conditions by destabilizing the four coordinated boron species that would be formed during the reactions involving CO, CO₂, and/or H₂O.

Fig. 2. Optimization of the reaction conditions.

General conditions for the catalytic hydrogenation of Qin: A mixture of Qin (2.5 mmol, 1.5 M in toluene) and Bⁿ (1 mol %) was treated with H₂/CO/CO₂ (4 atm each; conditions A) or H₂ (4 atm; conditions B) at 100°C. Yields of H₄-Qin were determined by GC analysis. The molecular structures of B⁹ and B¹⁰ were determined by single-crystal x-ray diffraction analysis and are shown with thermal ellipsoids at 30% probability (H atoms are omitted for clarity). BINAP, 2,2′-bis(diphenylphosphino)-1,1′-binaphthyl. *denotes 10 mol % catalyst. †denotes catalyst turnover number (TON) after a period of 48 hours under solvent-free conditions using 0.1 mol % Bⁿ and H₂/CO/CO₂ (30 atm each). ‡denotes catalyst TON after a period of 48 hours under solvent-free conditions using 0.05 mol % B⁹ and H₂ (85 atm).

Both B⁷ and B⁹ exhibited high stability toward air and moisture. B⁹ can be stored under ambient conditions (22°C, ca. 30% humidity) for at least 1 year without any apparent decomposition, while very minor levels (ca. 1%) of decomposition were observed for B⁷ after 1 year of storage (figs. S14 and S15).

The B⁹-catalyzed hydrogenation of Qin (1 mol % catalyst, without solvent) also proceeded to furnish H₄-Qin in >99 and 94% yield using CO-rich (H₂/CO/CO₂ = 4/20/4 atm; a model of syngas) and CO₂-rich (H₂/CO/CO₂ = 4/4/20 atm; a model of industrial off-gas) mixtures, respectively, although a longer reaction time was required in both cases (Fig. 3A). These results imply that CO and CO₂ kinetically affect the catalytic activity of B⁹ toward hydrogenation, with obvious suppression when an excess of the contaminant CO₂ with respect to H₂ is present. The coexistence of CH₄ did not hamper the progress of the reaction.

Fig. 3. Direct H₂ purification/storage from contaminated H₂ gas based on the present catalytic process.

(A) Exploration of the gas composition of the H₂ source. Yields of H₄-Qin were determined using GC analysis. (B) Catalytic dehydrogenation of H₄-Qin (3.8 mmol) to Qin in the absence of solvent. Yields of Qin were determined by GC analysis. *denotes 2 hours. †denotes yield of recovered H₂ based on the collected volume (V = 164 ml, 7.32 mmol). (C) H₂ purification based on the B⁹-catalyzed hydrogenation of Qin (1.0 mmol) under mixed gas conditions and subsequent catalytic dehydrogenation. Yield of recovered H₂ based on the collected volume (V = 39.0 ml, 1.74 mmol); H₂ purity (%) = (molar amount of H₂)/(sum of the molar amounts of H₂, CO, and CO₂) × 100. (D) B⁷-catalyzed hydrogenation of Lut using a variety of H₂ sources that were dried over 4-Å MS before use. ‡indicates that B⁹ was used.

We also explored the optimal conditions for the catalytic dehydrogenation of H₄-Qin to afford Qin (Fig. 3B) (44, 45). Without solvent, 2 mol % B⁵, B⁷, and B⁹ successfully catalyzed the production of Qin in 84 to 88% yields at 200°C for 6 hours, whereas B¹ and B¹⁰ exhibited inferior results (18 and 68%, respectively) under otherwise identical reaction conditions. The recovery of H₂ [193% yield; CO and CO₂ were not detected by gas chromatography (GC) analysis; fig. S20] was confirmed during the B⁹-catalyzed dehydrogenation of H₄-Qin (2 hours) to Qin in 90% yield. Under the applied reaction conditions, the hydrogenation of Qin and the dehydrogenation of H₄-Qin could be catalyzed simultaneously by B⁹. Further optimization of the reaction equipment so that the recovered H₂ can be efficiently removed could thus be expected to increase the dehydrogenation efficiency.

To develop a strategy for the purification of the contaminated H₂, we designed a reaction system based on a B⁹-catalyzed hydrogenation/dehydrogenation sequence starting from H₂/CO/CO₂ [1/1/1 molar ratio; H₂ purity = (molar amount of H₂)/(sum of the molar amounts of H₂, CO, and CO₂) × 100 = 33.3%] as a feed gas (Fig. 3C). In the presence of 2 mol % B⁹, H₂ was directly stored in H₄-Qin (0.95 mmol, 95%) from H₂/CO/CO₂ (5.4 equivalents each) via the hydrogenation of Qin (1.0 mmol). After a simple evacuation, dehydrogenation of the obtained H₄-Qin was carried out to generate H₂ (1.74 mmol, 174%) with a concomitant regeneration of Qin (0.91 mmol, 91%). Thus, a significant increase in the H₂ purity from 33.3 to 99.9% was demonstrated by the efficient removal of CO (not detected by GC analysis) and CO₂ (detected in ca. 0.1%; fig. S21) via a single cycle of the B⁹-catalyzed hydrogenation/dehydrogenation sequence. The complete removal of CO in a single cycle would be especially noteworthy, as the removal of CO remains challenging in the well-developed multistep, multibed PSA and membrane technologies (4).

We further explored the catalytic activity of B⁷ and B⁹ toward the hydrogenation of Lut to afford H₆-Lut under the mixed gas conditions (Fig. 3D). This further investigation revealed that the H₂ storage capacity could be increased from 2.7 (H₄-Qin) to 5.3 wt % (H₆-Lut). Note that Lut has been a challenging substrate in previously reported organoborane-catalyzed hydrogenations using H₂ even under diluted conditions (46, 47). In the presence of B⁷ (10 mol %) and the absence of solvent, H₆-Lut was formed in 81% yield using H₂ [20 atm; dried over 4-Å molecular sieves (MSs) before use], while a decrease in yield was observed for B⁹ (run 1). Without the dehydration of H₂, the hydrogenation of Lut also proceeded to afford H₆-Lut in 72% under identical conditions (fig. S22). Moreover, B⁷ exhibited promising results for the simultaneous separation and storage of H₂ in H₆-Lut from CO- and CO₂-contaminated H₂ gas (runs 2 to 4), albeit an excess of H₂ with respect to the contaminants was present.

To gain insight into the reaction mechanism for the present hydrogenation of N-heteroaromatic compounds in the presence of CO, CO₂, and H₂O, preliminary mechanistic studies were conducted using Qin (Fig. 4). First, we monitored the progress of the conversion of Qin to H₄-Qin using B¹, B⁷, or B⁹ under each condition using solely H₂ or H₂/CO/CO₂ (Fig. 4A). The production of H₄-Qin exhibited a zeroth-order dependence on the concentration of Qin with rate constants (k_obs) as follows: 3.08(29) × 10⁻⁴ mol m⁻³ s⁻¹ (H₂) and 3.56(60) × 10⁻⁶ m⁻³ s⁻¹ (H₂/CO/CO₂) (B¹); 1.71(6) × 10⁻⁴ m⁻³ s⁻¹ (H₂) and 1.66(5) × 10⁻⁴ m⁻³ s⁻¹ (H₂/CO/CO₂) (B⁷); and 2.21(13) × 10⁻⁴ m⁻³ s⁻¹ (H₂) and 1.93(10) × 10⁻⁴ m⁻³ s⁻¹ (H₂/CO/CO₂) (B⁹). Moreover, these results suggest that H₄-Qin itself does not affect the rate of hydrogenation, as neither an increase nor decrease in the rate was observed increasing conversion to H₄-Qin. Thus, the influence of CO and/or CO₂ is almost negligible for the B⁷- and B⁹-catalyzed hydrogenation processes, at least under conditions that do not involve excess amounts of CO/CO₂ with respect to H₂ (vide supra). In stark contrast, the B¹-catalyzed process was significantly inhibited in the presence of CO and/or CO₂. Control experiments using H₂/CO (10 atm each) and H₂/CO₂ (10 atm each) clarified that both CO and CO₂ affect the catalytic activity of B¹ and that contamination with CO₂ is especially deleterious (Fig. 4B). We also confirmed the kinetic orders in catalyst B⁷ [1.2(1)] and B⁹ [1.4(1)] under the H₂/CO/CO₂ atmosphere, demonstrating that these triaryl boranes do catalyze the formation of H₄-Qin (Fig. 4C).

Fig. 4. Mechanistic experiments.

(A) Kinetic profiles of the concentration of H₄-Qin (mole meter⁻³) with respect to reaction time (10³ s) obtained from the hydrogenation of Qin (1.5 M in toluene) in the presence of Bn (n = 1, 7, and 9) and different H₂ sources (pure H₂ or H₂/CO/CO₂; 10 atm each). (B) Influence of the gas composition on the B¹-catalyzed hydrogenation of Qin. Each experiment was pressurized with H₂ (10 atm) and/or CO_x (10 atm; x = 1 and/or 2). Average yields of independent three runs are shown with SEs. (C) Profile of lnk_obs with respect to ln[Bn] (n = 7 and 9). (D) Detailed analysis of the Bn-catalyzed hydrogenation of Qin (n = 1 and 9) in the presence of H₂/CO₂ (2.5 atm each). Product yields were calculated using ¹⁹F NMR analysis with C₆H₅CF₃ as the internal standard. In the case of B¹, several unidentified resonances were observed (for details, see fig. S38). (E) Calculated free energy profiles for the formation of OC─Bn (n = 1 and 9) [kilocalorie mole⁻¹; ωB97X-D/6-311+G(d,p)//ωB97X-D/6-31G(d,p) level]. The gas phase–optimized structure of OC─B⁹ and selected structural parameters for OC─Bn (n = 1 and 9) are also shown. (F) Hydrogenation of Qin using [Qin─H][HO─Bn] (n = 1 and 9). Yields of H₄-Qin were determined via GC analysis.

Next, the influence of CO₂, CO, and H₂O was investigated using B¹ and B⁹. The hydrogenation of Qin was carried out in the presence of each borane (50 mol %) using H₂/CO₂ (2.5 atm each) at 100°C in toluene-d₈ and analyzed using multinuclear nuclear magnetic resonance (NMR) spectroscopy. In the case of B¹, the resultant mixture included C₆F₅H (20%), an equilibrium mixture of [H₄-Qin─COOB¹] and [H₄-Qin─H][H₃-Qin─COOB¹] (12%) (28, 29), and several unidentified compounds (Fig. 4D and fig. S38). Precipitation of the nitrogen-boron adduct [H₄-Qin─B¹] was also confirmed. We separately confirmed that C₆F₅H was not formed in the absence of CO₂ (fig. S41). These results clarify that CO₂ triggers the irreversible decomposition of B¹ to yield C₆F₅H via protodeboronation from both [H₄-Qin─COOB¹] and [H₄-Qin─H][H₃-Qin─COOB¹] under heating conditions. In stark contrast, when B⁹ was used, the generation of H₄-Qin and recovery of B⁹ in >99% yields were observed under otherwise identical conditions. These results are consistent with the fact that the B⁹-catalyzed hydrogenation of Qin is not irreversibly inhibited by CO₂.

Density functional theory calculations were carried out at the ωB97X-D/6-311+G(d,p)//ωB97X-D/6-31G(d,p)//gas phase level of theory to shed light on the observed kinetic suppression of the hydrogenation of Qin by CO (48). The relative Gibbs energies (kilocalorie mole⁻¹) for OC─Bn (Lewis pairs comprising CO and Bn and TS─Bn (saddle point species) with respect to [CO + Bn] (n = 1 and 9) are shown in Fig. 4E. The coordination of CO to the boron atoms in both B¹ and B⁹ is an endothermic process (29, 33), and coordination to the latter is far less favorable from a kinetic and thermodynamic perspective. The standard Gibbs free energies for the formation of OC─Bn are +3.2 (n = 1) and +8.1 (n = 9) kcal mol⁻¹, and the activation energies to overcome TS─Bn are +10.2 (n = 1) and +14.3 (n = 9) kcal mol⁻¹. These results rationalize the experimental results, i.e., the observations that contamination with excess CO kinetically affects both the B¹- and B⁹-catalyzed hydrogenation of Qin under the applied conditions, with this suppression being significant in the former case. The differences in the stability of OC─Bn should be related to the degree of geometric deviation from the ideal tetrahedral geometry around their boron centers, which can be evaluated on the basis of the value of τ_δ(B) [τ_δ = {360 – (α + β)/141 × β/α}, where α and β are the largest and second largest C─B─C angles obtained from the gas phase–optimized structures of OC─Bn] (49). More efficient orbital overlap between the lone pair on the carbon atom in CO and the p orbital on the boron atom in Bn should result in higher stabilization of OC─Bn adducts, adopting a more ideal tetrahedral geometry [τ_δ(B) = 0.9 to 1.0] and a linear arrangement of the B─C─O atoms (∡B─C─O ≈ 180°). In the present study, the lower τ_δ(B) of 0.88 for OC─B⁹ indicates that its boron atom adopts a more distorted tetrahedral geometry compared to that of OC─B¹ [τ_δ(B) = 0.92], and the B─C─O atoms in OC─B⁹ are confirmed to exhibit a bent alignment (169.9° versus 179.6° in OC─B¹). These results thus demonstrate the effective destabilization of OC─B⁹ due to the increased steric repulsion between CO and the 2,6-Cl2-C6H3 group introduced on B⁹, which eventually results in the reduced impact of CO on the B⁹ hydrogenation of Qin.

We further evaluated the influence of H₂O on the hydrogenation (Fig. 4F). GC analysis confirmed that no conversion of Qin occurred in the presence of H₂ (4 atm) at 100°C when 10 mol % [Qin─H][HO─B¹] was used. Although H₄-Qin was furnished in 13% yield when 10 mol % [Qin─H][HO─B⁹] was used, the low yield again confirmed the deleterious influence of H₂O. On the basis of these results and the stability of B⁹ toward moisture at ambient conditions (vide supra), [Qin─H][HO─B⁹] was not generated under the applied conditions shown in Figs. 2 and 3, although H₂O might be present as a contaminant.

DISCUSSION

The present results demonstrate a proof of concept for a H₂ purification technology based on LOHCs that goes beyond their well-established use in H₂ storage. This technology can be expected to change the industrial value of crude H₂ containing substantial amounts of CO, CO₂, and CH₄, which can be produced from a variety of carbon resources such as biomass and industrial off-gases. The operational simplicity of the present method should allow the construction of combined processes involving PSA and/or membranes. Moreover, this work demonstrates a new aspect of main group catalysis beyond its application as a simple alternative to well-established transition metal–catalyzed processes, i.e., the main group–catalyzed hydrogenation of unsaturated molecules under mixed gas conditions.

MATERIALS AND METHODS

General considerations

Unless otherwise noted, all manipulations were conducted under a nitrogen atmosphere using standard Schlenk line or grove box techniques. MSs (4 Å) were activated by heating with a heat gun in vacuo (ca. 0.2 mmHg) for 5 min. ¹H, ¹¹B, ¹³C, ¹⁹F, and ³¹P NMR spectra were recorded on Bruker Avance III 400 or JEOL JNM-400 spectrometers at 25°C. The chemical shifts in the ¹H NMR spectra were recorded relative to Me₄Si or residual protonated solvent [C₆D₅H (δ 7.16), CHCl₃ (δ 7.26), C₇D₇H (δ 2.08), and CDHCl₂ (δ 5.32)]. The chemical shifts in the ¹¹B NMR spectra were recorded relative to BF₃. The chemical shifts in the ¹³C NMR spectra were recorded relative to Me₄Si or deuterated solvent [C₆D₆ (δ 128.06), CDCl₃ (δ 77.16), and CD₂Cl₂ (δ 53.84)]. The chemical shifts in the ¹⁹F NMR spectra were recorded relative to α,α,α-trifluorotoluene [δ −65.64]. The chemical shifts in the ³¹P NMR spectra were recorded relative to 85% H₃PO₄ as an external standard. Assignment of the resonances in ¹H and ¹³C NMR spectra was based on ¹H-¹H correlation spectroscopy, heteronuclear multiple-quantum coherence, and/or heteronuclear multiple-bond correlation experiments. High-resolution mass spectrometry was performed at the Instrumental Analysis Center, Faculty of Engineering, Osaka University. A single-crystal x-ray diffraction analysis was carried out using the Rigaku XtaLAB Synergy equipped with the HyPix-6000HE detector. Analytical GC was carried out on a Shimadzu GC-2025 gas chromatograph, equipped with a flame ionization detector, or a Shimadzu GC-2010 gas chromatograph, equipped with a barrier discharge ionization detector.

Materials

All commercially available reagents including superdehydrated solvents (n-hexane, toluene, tetrahydrofuran, and diethyl ether) were used as received. Benzene-d₆ and toluene-d₈ were distilled from sodium benzophenone ketyl before use. CDCl₃ and CD₂Cl₂ were stored inside the grove box over MSs (4 Å) after several freeze-pump-thaw cycles. Qin, H₄-Qin, and Lut were used after distillation over CaH₂. Triaryl boranes (B² to B⁶), (26, 39–42) potassium (2,6-dichlorophenyl)trifluoroborate, (41) 1,5-dichloro-2,4-difluoro-3-iodobenzene (50), 1-chloro-2,4-difluoro-3-iodobenzene (50), and 1,5-dibromo-2,4-difluoro-3-iodobenzene (50) were prepared by following the reported procedures. Gaseous chemicals including H₂, CO, CO₂, CH₄, H₂/CO (a 1/1 molar ratio), H₂/CO₂ (a 1/1 molar ratio), and H₂/CO/CO₂ (a 1/1/1 molar ratio) were purchased and used as received otherwise noted.

Synthesis of B⁹

A solution of 1,5-dibromo-2,4-difluoro-3-iodobenzene (3.45 g, 8.67 mmol, 0.29 M in Et₂O) was slowly treated with ⁱPrMgCl (8.7 ml, 8.7 mmol, 1.0 M in Et₂O). After stirring at room temperature for 1 hour, the resultant solution was transferred into a suspension of potassium (2,6-dichlorophenyl)trifluoroborate (1.05 g, 4.16 mmol, 0.42 M in Et₂O) at 0°C. The reaction mixture was then allowed to warm to room temperature, where it was stirred for another 14 hours. After the removal of all volatiles in vacuo, the residue was extracted with α,α,α-trifluorotoluene (50 ml for three times; warmed to 70°C before use). The combined organic layers were concentrated in vacuo and washed with hexane (cooled to −20°C before use) to afford B⁹ as a white solid (2.56 g, 3.66 mmol, 88%).

H₂ purification from H₂/CO/CO₂ via a B⁹-catalyzed hydrogenation/dehydrogenation sequence

A 30-ml autoclave was charged with Qin (145 mg, 1.01 mmol), B⁹ (13.9 mg, 0.199 mmol), and tetradecane (59.5 mg; internal standard). Once sealed, the autoclave was pressurized with H₂/CO/CO₂ (4 atm each; 5.4 mmol each) and heated to 100°C for 2 hours. After cooling to room temperature, all volatiles were removed in vacuo, and GC analysis showed the production of H₄-Qin in 95%. Then, the reaction mixture was transferred into a 10-ml two-neck flask. During this manipulation, the residue inside the autoclave reactor was extracted with toluene to minimize the loss of reagents. The toluene was then removed in vacuo; however, this extraction step is not essential for the following dehydrogenation. The reaction mixture was then stirred at 200°C for 3 hours. The volume of collected gas was measured using a graduated cylinder to calculate the yield of H₂ (39 ml, 1.7 mmol). The conversion of H₄-Qin (94%), the yield of Qin (91%), and the purity of the collected H₂ gas were determined using GC analysis.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S44

Tables S1 to S32

NMR spectra

References