Metal-free CO₂ fixation: conversion of epoxides to cyclic carbonates by a highly efficient bifunctional boron-tethered N-heterocyclic nitrenium catalyst

Abstract

The development of sustainable, metal-free catalytic platforms for the chemical fixation of carbon dioxide (CO₂) into value-added products remains a formidable challenge. Herein, we report the rational design of a highly efficient bifunctional organoboron catalyst, featuring a Lewis acidic boron center tethered to an N-heterocyclic nitrenium ion, for the solvent-free cycloaddition of CO₂ to epoxides. This system exhibits enhanced catalytic activity and recyclability, delivering a broad range of cyclic carbonates in excellent yields under neat conditions. Thorough mechanistic studies reveal the key transient intermediates in the reaction pathway. Notably, our catalyst demonstrates the highest activity among the metal-free systems reported to date for the coupling of CO₂ to propylene oxide under ambient conditions (TOF = 12.4 h⁻¹, 25 °C, 1 bar), establishing this organoboron framework as a remarkably robust and green catalyst.

A metal-free, boron-tethered N-heterocyclic nitrenium (NHN) catalyst enables synergistic CO₂ fixation into epoxides at 25 °C and 1 bar, efficiently yielding cyclic carbonates to advance sustainable main-group catalysis.

Introduction

The escalating atmospheric concentration of carbon dioxide (CO₂), a primary greenhouse gas, is a leading reason for global climate change. This environmental challenge necessitates the development of efficient chemical methodologies to valorize CO₂ into value-added products.^1–4 As an abundant, inexpensive, and renewable C1 feedstock, CO₂ can be converted into various essential industrial chemicals, including methanol, syngas, and formic acid.^5–8 Among the established utilization strategies, the coupling of CO₂ with epoxides to yield cyclic carbonates stands out as a highly atom-economical and sustainable pathway.^9–12 These cyclic carbonates are crucial industrial commodities, widely utilized as electrolyte solvents for lithium-ion batteries, environmentally benign aprotic solvents, medicinal degreasers, and fundamental building blocks for the synthesis of degradable polymers such as polyurethanes and polycarbonates.^13–15 However, overcoming the inherent thermodynamic stability and kinetic inertness of the CO₂ molecule requires highly effective catalytic systems. While numerous metal-based catalysts have been successfully developed for this transformation,^16–19 the inevitable issue of trace metal contamination in the final products remains a significant drawback. Consequently, the rational design of robust, metal-free organocatalysts has emerged as a sustainable alternative,^20–22 making the pursuit of novel metal-free catalytic platforms for CO₂ cycloaddition a highly desirable objective.

Due to its high natural abundance and low cost, boron has become a key element in modern organic synthesis. Building on this widespread utility, recent advances in main-group catalysis have led to the development of various boron-based systems tailored for the development of various boron-based systems tailored for the conversion of CO₂ and epoxides into cyclic carbonates (Scheme 1A).²³

Scheme 1. (A) Selected examples of organoboron catalysts for CO₂ cycloaddition; (B) this work: CO₂ fixation by boron-tethered NHN catalyst, Cat 1.

In 2019, Andrea and Kerton demonstrated that triphenylborane and tris(pentafluorophenyl)borane, when combined with a suitable cocatalyst, could promote the synthesis of cyclic carbonates and polycarbonates from CO₂ and epoxides (Table 1, entry 1).²⁴ That same year, Lu and co-workers achieved this transformation utilizing B₂O₃ in conjunction with tetrabutylammonium bromide (TBAB) as the cocatalyst (entry 2).²⁵ However, both systems suffered from the necessity of high temperatures and elevated pressures. Subsequently, various other binary systems comprising a boron-based catalyst and a nucleophilic cocatalyst have been reported (entries 3–6).^26–29 More recently, Zhang et al. introduced a novel class of organoboron catalysts capable of selectively producing cyclic carbonates under solvent-free conditions (entries 7 and 8).³⁰ Nevertheless, achieving quantitative yields still required harsh parameters; under ambient conditions, their system yielded only 32% conversion with a turnover frequency (TOF) of 4 h⁻¹. Addressing these persistent limitations, we have developed a highly sustainable, metal-free, boron-tethered N-heterocyclic nitrenium (NHN) ion as a catalytic platform for the efficient cycloaddition of CO₂ to a broad range of epoxides (Scheme 1B). Recently, NHNs, the isoelectronic and isostructural analogs of N-heterocyclic carbenes (NHCs), have garnered significant attention among chemists.^31,32 Building upon the seminal work of Gandelman's group, which demonstrated the tunable Lewis acidity of NHNs through the isolation of stable phosphine adducts, the field has experienced rapid growth.³³ The reactivity of the NHN framework is exploited across diverse catalytic applications, ranging from main-group photocatalysis to Lewis acid-mediated transformations.^34–38 We hypothesized that the NHN core appended to a highly Lewis acidic boron motif, together with a suitable counteranion, can be harnessed to facilitate the challenging ring-opening of epoxides under ambient conditions. Notably, this catalyst (Cat 1) works smoothly at 25 °C and 1 bar of CO₂ pressure, without the need for any solvent or additional co-catalyst, to yield cyclic carbonates. To the best of our knowledge, our rationally designed system delivers the highest catalytic activity (TOF = 12.4 h⁻¹) reported to date for any metal-free catalyst operating under ambient conditions (entry 9).

Table 1. Comparison of selected organoboron catalysts with our catalytic system.

Entry	Cat. (mol%)	Co-cat (mol%)	Epoxide	T (°C)	CO2 pressure (bar)	Time (h)	Solvent	Yield (%)	TON	TOF (h−1)
1 (ref. 24)	BPh3 (0.1)	PPNCl (0.1)	PO	100	20	3	Neat	92	920.0	307.0
2 (ref. 25)	B2O3 (0.5)	TBAB (1.0)	PO	100	20	2	Neat	95	190.0	95.0
3 (ref. 26)	H4B2O4 (9.0)	TBAI (13.5)	PO	rt	1	21	Neat	95	10.6	0.5
4 (ref. 27)	TEOAB (0.05)	TBAB (0.25)	ECH	120	17	3	Neat	99	1980.0	660.0
5 (ref. 28)	Boronic acid (10)	TBAI (5.0)	PO	50	10	7	H2O	98	9.8	1.4
6 (ref. 29)	Boron–pyridine catalyst (1.0)	TBAI (4.0)	PO	50	10	21	Neat	97	97.0	4.6
7 (ref. 30)	Boron-based quaternary ammonium salt (0.2)	—	PO	25	20	16	Neat	>99.9	499.5	31.2
8 (ref. 30)	Boron-based quaternary ammonium salt (1.0)	—	PO	25	1	8	Neat	32	32.0	4.0
9 (this work)	Cat 1 (1.0)	—	PO	25	1	8	Neat	>99	99.0	12.4

Results and discussion

Our investigation commenced with the synthesis and screening of a library of boron-appended NHN ions to evaluate their catalytic efficacy for CO₂ cycloaddition. The prepared catalysts were well characterized using nuclear magnetic resonance (NMR) and high-resolution mass spectrometry (HRMS) (see SI). Furthermore, single crystals suitable for X-ray diffraction (SC-XRD) analysis were obtained via the slow evaporation of Cat 1 in CDCl₃. The solid-state structure revealed the coordination of a water molecule, originating from trace solvent-moisture to the boron center of the catalyst (Fig. 1). While tricoordinate boranes readily form adducts with many Lewis bases, stable borane–water adducts are rare, as highly electron-deficient boron centers typically undergo rapid hydrolysis rather than simple coordination.³⁹ The coordination of H₂O is confirmed by a B2–O1 bond length of 1.666(8) Å. Within the nitrenium core, the N4–N5–N6 bond angle is 106.6(4)°, with N4–N5 and N5–N6 bond distances of 1.324(6) Å and 1.312(7) Å, respectively.

Fig. 1. Molecular structure of Cat 1 with thermal ellipsoids at 50% probability level. Selected bond lengths [Å] and angles [°]: N4–N5: 1.324(6), N5–N6: 1.312(7), B2–O1: 1.666(8), N4–N5–N6: 106.6(4).

We evaluated the catalytic performance of Cat 1 using propylene oxide (1a) as a model substrate. Utilizing 1a, optimization studies revealed Cat 1 to be highly active, selectively delivering propylene carbonate (2a) in >99% NMR yield after 8 hours under ambient conditions (room temperature, 1 bar CO₂) (Table 2, entry 1). The indispensable role of the nucleophile was demonstrated by exchanging the catalyst's iodide (I⁻) counteranion with a non-coordinating hexafluorophosphate (PF₆⁻), which completely suppressed product formation (entry 2). Furthermore, altering the structural spacer between the NHN core and the boron moiety proved detrimental to the catalytic efficiency, resulting in diminished yields of 2a (entries 3 and 4). The lower catalytic activity upon increasing the alkyl spacer length between NHN and BBN may arise due to the increase in the conformational degrees of freedom which renders the epoxide activation and subsequent iodide-mediated ring opening steps entropically more demanding. Reducing the reaction time to 6 or 4 hours resulted in a slight reduction in yields to 93% and 85%, respectively (entries 5 and 6). Finally, lowering the catalyst loading to 0.8 mol% still afforded an 85% yield of 2a (turnover number, TON = 106.2), whereas a further reduction to 0.5 mol% diminished the yield to 67% (entries 7 and 8).

Table 2. Optimization of reaction conditions^a.

Entry	Variation from standard reaction conditions	Yieldb (%)
1	None	>99
2	Cat 2 instead of Cat 1	n.d.c
3	Cat 3 instead of Cat 1	92
4	Cat 4 instead of Cat 1	82
5	6 h instead of 8 h	93
6	4 h instead of 8 h	85
7	0.8 mol% loading instead of 1 mol%	85
8	0.5 mol% loading instead of 1 mol%	67

^aStandard reaction conditions: propylene oxide 1a (1.43 mmol), Cat 1 (1 mol%), CO₂ (1 bar), neat at RT for 8 hours.

^bYields were determined by ¹H NMR analysis using mesitylene as the internal standard.

^cNot detected.

With the optimized conditions in hand, we evaluated the generality of our catalytic system across a diverse array of epoxides (Table 3). The model substrate 1a afforded the corresponding cyclic carbonate (2a) in >99% NMR yield. Extending the alkyl chain length of the terminal epoxides was well-tolerated, furnishing the desired products (2b–2d) in excellent yields. Furthermore, substrates bearing versatile chloro- and hydroxy-functional groups (1e–1f) underwent smooth conversion under standard conditions. The protocol also proved highly effective for aryl-substituted substrate, with styrene oxide (1g) delivering 98% of 2g. Exploring a series of glycidyl ethers revealed that primary, secondary, and even sterically demanding tert-butyl substituents (2h–2j) were all seamlessly accommodated. Importantly, glycidyl ethers containing unsaturated alkene and alkyne linkages were compatible, delivering the respective cyclic carbonates (2k–2l) in very good yields. Aromatic and heteroaromatic epoxides (1m–1n) also exhibited excellent reactivity, underscoring the broad functional group tolerance of the methodology. Finally, extending this protocol to more challenging bis-epoxides successfully produced the corresponding bis-cyclic carbonates (2o–2q) in moderate to excellent yields. Internal epoxide, cyclopentene oxide (1r), gave a trace amount of product under standard reaction conditions. However, upon increasing the catalyst loading to 2 mol% and extending the reaction time to 40 hours, the corresponding cyclic carbonate (2r) was isolated in 27% yield. This limited reactivity may arise from the considerable steric bulk of the 9-BBN framework, which hinders the initial coordination to the sterically crowded internal epoxide. Furthermore, the subsequent S_N2-type ring-opening step is sterically restricted at the substituted internal carbon atoms.

Table 3. Substrate scope^a,b,c.

^aReaction conditions: epoxide (1.43 mmol), Cat 1 (1 mol%), CO₂ (1 bar), neat at RT for 8 hours.

^bYields were determined by ¹H NMR analysis using mesitylene as the internal standard.

^cIsolated yields are given in parentheses.

^dCat 1 (2 mol%), CO₂ (1 bar), neat at RT for 40 hours.

To evaluate the stability of the catalytic system, an in situ recyclability study was conducted under standard reaction conditions (Fig. 2). Consecutive cycles were performed by sequentially charging the reaction vessel with fresh batches of 1a and pressurizing with 1 bar CO₂, without any intermediate recovery steps. The catalyst, Cat 1, demonstrated excellent initial activity, affording a >99% yield in the first cycle. In the subsequent runs, the yields exhibited a gradual decrease to 91%, 82%, and 75% for the second, third, and fourth cycles, respectively. This demonstrated that the catalyst remained active for over four cycles. To gain mechanistic insights into the reaction pathway, in situ NMR studies were conducted. The ¹¹B NMR spectrum of Cat 1 exhibits a broad singlet at 88.2 ppm. Upon the addition of 1a, an upfield chemical shift of 1.6 ppm (to 86.6 ppm) was observed (Scheme 2A). This indicates a weak coordination between the epoxide oxygen atom and the Lewis acidic boron center of Cat 1, which increases the electron density at the boron nucleus. This interaction was further corroborated by variable-temperature ¹H NMR, which revealed corresponding downfield shifts for the protons of 1a (see Fig. S2). Stirring this mixture at room temperature for 2 hours led to the complete disappearance of the 86.6 ppm signal, concomitant with the emergence of a new peak at 4.7 ppm (Scheme 2B). This profound upfield shift is highly characteristic of a tetra-coordinated boron species, signifying the successful ring-opening of the epoxide. Subsequently, upon the introduction of 1 bar of CO₂ and further stirring for 2 hours, low-temperature (−50 °C) ¹³C NMR analysis revealed a distinct resonance at 158.1 ppm (see Fig. S4). This signal might correspond to the carbonyl carbon of the transient acyclic hemicarbonate intermediate formed prior to intramolecular cyclization. Based on these experimental findings and the mechanistic precedents,³⁰ a plausible pathway for the CO₂ cycloaddition reaction to achieve the cyclic carbonate is proposed (Scheme 2C). The catalytic cycle initiates with the weak coordination of propylene oxide (1a) to the boron center, generating intermediate IM 1 (¹¹B NMR: δ = 86.6 ppm). This electrophilic activation facilitates the nucleophilic attack of the counterion iodide (I⁻) at the less sterically hindered carbon of 1a, promoting ring-opening to yield the tetra-coordinated alkoxide intermediate IM 2 (¹¹B NMR: δ = 4.7 ppm). Following this, a molecule of CO₂ readily inserts into the newly formed boron–alkoxide bond to generate the key acyclic hemicarbonate species IM 3 (¹³C NMR: δ = 158.1 ppm), which could be stabilized electrostatically by the positively charged nitrenium nitrogen. Finally, an intramolecular nucleophilic attack by the carbonate oxygen onto the methylene carbon drives the ring-closure step, releasing the product propylene carbonate (2a) and regenerating Cat 1 for the next cycle.

Fig. 2. Recyclability test of Cat 1.

Scheme 2. (A and B) Mechanistic investigations using ¹¹B NMR (C) proposed catalytic cycle.

Conclusions

In summary, we have developed a highly efficient class of metal-free, bifunctional organoboron catalysts that selectively convert epoxides to cyclic carbonates via the solvent-free cycloaddition of CO₂. This reaction proceeds smoothly at room temperature and under 1 bar of CO₂ pressure to produce a diverse array of cyclic carbonates in excellent yields, while also exhibiting robust catalyst recyclability. To the best of our knowledge, our rationally designed system delivers the highest catalytic activity (TOF = 12.4 h⁻¹) reported to date for any metal-free catalyst operating under ambient conditions. Thorough in situ NMR studies successfully captured the key transient acyclic hemicarbonate and tetra-coordinated boron intermediates, providing experimental validation for the proposed catalytic cycle. This structurally defined, highly active, and recyclable main-group catalytic system represents a significant advancement in sustainable chemistry, paving the way for the next generation of rational catalyst design for greenhouse gas valorization. Efforts to expand the synthetic utility of this versatile catalytic platform toward other challenging transformations are currently underway in our laboratory.

Author contributions

K. G., R. M. and R. R. M. conceived and designed the project. R. M and R. R. M. carried out the optimization, characterization, substrate scope, and mechanistic studies. K. G., R. M. and R. R. M. wrote the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

CCDC 2563179 contains the supplementary crystallographic data for this paper.⁴⁰

The datasets supporting this article are available as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6sc05761f.