1,2-Dibromo-methyl benzene, n-vinylimidazole, DVB, azodiisobutyronitrile (AIBN), bis(triphenylphosphine)palladium(II) dichloride [Pd(PPh₃)₂Cl₂], palladium(II) acetate [Pd(OAc)₂], cuprous iodide (CuI), iodobenzene, 2-methyl-3-butyn-2-amine, etc. were purchased from Innochem in Beijing, China. Potassium carbonate (KHCO₃) and silver nitrate (AgNO₃) were procured from Xilong Science Co., LTD. Solvents such as toluene, ethyl acetate (EA), dimethyl sulfoxide (DMSO), tetrahydrofuran (THF) and methanol (MeOH), sulfuric acid (H₂SO₄) are purchased from local suppliers.

The monomer (NHC-1) was synthesized by quaternary ammonium reaction of 20.0 mmol 1-vinylimidazole and 10.0 mmol halogenated aromatic hydrocarbons in 20 ml toluene at 90 °C for 24 h^[24]. Secondly, NHC-HCO₃ was synthesized by mixing 2 mmol NHC-1 and 4.2 mmol KHCO₃ in MeOH at room temperature for 48 h to achieve anion conversion, stirring (400 rpm) at room temperature under shade for 12 h for metal pre-coordination synthesis Ag@NHC-HCO₃. Finally, the polymer Ag@POP-HCO₃ was obtained by polymerization in 3 mL DMSO under high pressure of 100 °C for 24 h with 0.10 mmol Ag@NHC-HCO₃ as the polymerization unit, 2.0 mmol DVB as the crosslinker, and 33 mg AIBN as the initiator [Scheme 1].

The functional groups and structures of the catalysts were investigated by various analytical techniques. These included 1H/13C nuclear magnetic resonance (NMR) spectroscopy (Bruker SBAVance III 400 M, Germany), solid-state ¹³C NMR (Bruker Avance Neo 400WB 400M, Germany) and Fourier infrared spectroscopy (FT-IR, Thermo Scientific Nicolet iS20, United States). Additionally, the elemental composition of the nanoparticles was further analyzed by X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific K-Alpha, United States). The morphology and pore distribution of the catalyst were observed and determined by a scanning electron microscope (SEM, ZEISS sigma300, Germany), a transmission electron microscope (TEM, FEI Tecnai F20, United States), and a specific surface area and pore volume analyzer (BET, Kubo-X1000, China). Finally, the active center type of the catalyst was analyzed by in situ Raman (Horiba HR Evolution, Japan) and the thermal stability was studied by thermogravimetric analysis (TGA, NETZSCH STA2500, Germany).

The carboxylation cyclization reactions between propargyl amines and CO₂ were employed to evaluate the performance of the catalysts proposed by this work [Scheme 2]. The brand of pharmaceutical reagent used in this system is listed in Supplementary Materials. In particular, an appropriate amount of catalyst was carefully weighed and added to a 10 mL branched glass reaction tube. Subsequently, 0.2 mmol of the reaction substrate and a specific quantity of solvent were introduced into the tube. The tube was then connected to CO₂ atmosphere (0-100 vol.% CO₂ /N₂) provided by a balloon. Finally, the reaction tube containing the reaction mixture was then placed in a dark environment and subjected to magnetic stirring (400 rpm) for a certain time at a certain reaction temperature. After the reaction, the yield was determined using high performance liquid chromatography (HPLC Dionex UltiMate 3000) with biphenyl as the internal standard. The structure of the reaction products was elucidated by analyzing ¹H/¹³C NMR spectra obtained using a Bruker AVANCE-III 400 spectrometer.

The polymer material Ag@POP-HCO₃ was employed as a catalyst to catalyze the model carboxylation cyclization reaction between 2-methyl-4-phenylbut-3-yn-2-amine and CO₂. The reaction conditions, including the choice of solvent, reaction temperature, catalyst dosage, CO₂ concentration and reaction time, were optimized using a univariate method. The optimization conditions resulted in the following: when utilizing 15 vol.% CO₂/N₂ simulated gas as the carbon source and 0.2 mmol propargyl amines as the substrate, it was observed that employing 1 mL of DMSO at room temperature for a reaction time of 12 h resulted in an optimal yield of 77% using 10 mg of Ag@POP-HCO₃ catalyst [Figure 2A-C].

The higher concentration of CO₂ in the carbon source gas leads to a greater yield within the same reaction time (12 h) [Figure 2D]. With the increase of CO₂ concentration to 50 vol.% CO₂/N₂ and 100 vol.% CO₂/N₂, the yield can reach 87% and 93%, respectively. When using air as the carbon source, only a 1% yield of oxazolidinone compounds was obtained. However, no oxazolidinone was produced in the absence of CO_2, such as under a pure N₂ atmosphere. These findings suggested that the carboxylation cyclization reaction between propargyl amines and CO₂ was dependent on the nucleophilic addition and cyclization steps of CO₂.

Under the optimal reaction conditions, POP-HCO₃ and Ag@POP-Br (these synthesis methods are shown in Supplementary Schemes 1 and 2) were also utilized as catalysts to facilitate the conversion of low concentration CO₂ through the model reaction, respectively. Both materials are components of the structure of Ag@POP-HCO₃. The results showed that POP-HCO₃ did not yield any target products, while Ag@POP-Br resulted in a relatively low yield of only 24% [Figure 3A]. However, Ag@POP-HCO₃ outperformed both catalysts with a resulting yield of 77%. These findings indicate that both silver and HCO₃^- active sites in the polymer materials are essential for catalyzing the carboxylation cyclization reactions.

To investigate the impact of HCO₃^- on the catalytic performance, different Ag@POP-HCO₃ materials with varying HCO₃^- content were synthesized by adjusting the reaction ratios of NHC and KHCO₃ to establish the anionic environment of NHC-1 in the Ag@POP-HCO₃ synthesis process. The catalytic performance for the model reaction was then evaluated [Figure 3B]. When the reaction ratios of NHC-1 and KHCO₃ ranged from 1:0 to 1:2, the yield increased up to 77% due to rising HCO₃^- content in the Ag@POP-HCO₃ material. However, there was no significant increase in yield when changing the ratio of NHC and KHCO₃ to 1:3. This is because the theoretical reaction ratio of NHC and KHCO₃ is 1:2; increasing it to 1:3 did not raise the HCO₃^- content in the Ag@POP-HCO₃. These findings further confirm that HCO₃^- is an effective active site in the Ag@POP-HCO₃ catalyst.

In order to further verify the experimental results, the specific surface area and microporous fusion of NHC-1 and HCO₃^- proportional polymers were determined by N₂ adsorption and desorption curves [Figure 4, Supplementary Figures 1 and 2]. The higher the HCO₃^- content, the larger the specific surface area and microporous melting of the catalyst. When NHC-1: HCO₃^- = 1:1.5, the specific surface area and micropore porosity are the largest. The specific surface area is 366.37m²/g, and the micropore volume is 0.139 m³/g. However, due to the small number of active sites of HCO₃^- compared with the NHC-1: HCO₃^- = 1:2 polymer, the catalytic conversion rate of HCO₃^- to oxazolidinone was only 52%. When NHC-1: HCO₃^- = 1:3, the specific surface area and micropore volume of the catalyst remain basically unchanged. Furthermore, increasing the ratio of NHC-1 to HCO₃^- to 1:3 does not increase the HCO₃^- content in Ag@POP-HCO₃. Based on the above experimental results, it is proved that HCO₃^- is an important influencing factor of the catalytic system.

Increasing the usage of DVB in the polymerization process for the materials will increase both the specific surface area and micropore volume, potentially benefiting the catalytic performance^[30,31]. However, it will decrease both the silver and HCO₃^- active sites, negatively affecting the catalytic performance. Therefore, an optimum polymerization ratio needs to be sought to balance both sides of the conflict. Experimental data revealed that the best catalytic conversion of low concentration CO₂ to oxazolidinone occurs when Ag@NHC-HCO₃: DVB = 1:20 [Figure 5]. According to Figure 6, Supplementary Figures 3 and 4, the presence of DVB in the polymer benefits the increase in specific surface area and micropore volume of the polymer. As shown in Figure 7, Ag@POP-HCO₃ has significant selectivity for CO₂ adsorption at both 0 °C and 25 °C. Comparison of the CO₂ adsorption isotherms of Ag@POP-Br and Ag@POP-HCO₃ (1:20) showed that introduction of HCO₃^- anions into the catalyst led to enhancement of CO₂ adsorption capacity. Additionally, there was a positive correlation between adsorption performance for CO₂ and the increase in DVB usage [Figure 8]. Although the maximum CO₂ capacity belonged to the polymer material Ag@POP-HCO₃ (1:30), optimal catalytic performance occurred with a polymerization ratio of 1:20 for Ag@POP-HCO₃, indicating that an ideal equilibrium state was achieved in terms of distribution of active center, pore structure and adsorption capacity of the catalyst when using a ratio of 1:20.

Following the successful construction and optimization of the internal structure, the micro-structure of the optimized catalyst Ag@POP-HCO₃ (1:20) was characterized and verified. The specific surface area of Ag@POP-HCO₃ (1:20) material is 345.2 m²/g and the micropore volume is 0.1241 cm³/g. The adsorption and desorption curve of Ag@POP-HCO₃ exhibits type IV isotherm features [Figure 9]. The analysis of the pore size distribution curve indicated a pore size of approximately 3 nm for the polymer material, demonstrating characteristics of mesoporous materials.

The ¹³C NMR spectrum of NHC-HCO₃ revealed an increase in HCO₃^- peaks, indicating the successful replacement of original Br^- in NHC-1 by HCO₃^- through the reaction of NHC-1 with KHCO₃. The solid-state ¹³C NMR spectrum of Ag@POP-HCO₃ [Figure 10] shows vibration peaks at 29 ppm caused by methylene units (C3, C8) attached to the benzene ring, wide overlapping peaks at 40-60 ppm attributed to polyethylene (C4) vibration, vibration peaks at 128 ppm, 138 ppm, and 145 ppm associated with C signals from imidazole and aromatic rings, and a spike at 150 ppm due to HCO₃^-[32].

The FT-IR spectra of Ag@POP-HCO₃, Ag@POP-Br and POP-HCO₃ polymers are exhibited in Figure 11A. The peaks observed at 3,100-2,850 cm^-1 can be attributed to the stretching vibrations of aliphatic C-H bonds, while a bending vibration peak of around 710 cm^-1 is caused by aromatic C-H. These findings suggest DVB unit presence in the structure of the polymers^[33,34]. Peaks observed at 1,600 cm and 1,156 cm are separately induced by the vibration of the aromatic ring and imidazole ring skeleton (C-C)^[35], while peaks at 1,640 cm and 1,445 cm arise from C=N and C-N bonds^[36,37], indicating that imidazole framework of Ag@POP-HCO₃ remains undamaged after KHCO₃ modification and silver loading.

TGA [Figure 11B] in N₂ atmosphere shows that the catalyst Ag@POP-HCO₃ has excellent thermal stability. It did not exhibit significant mass loss until it reached a temperature above 400. However, a rapid increase of mass loss was observed between 400 °C and 500 °C, indicating substantial polymer decomposition. Further mass loss above 500 °C was attributed to imidazole ring decomposition^[8,9]. However, as the temperature increases, the polymer will undergo carbonization, eventually resulting in a non-volatile carbon skeleton and metal residues.

The SEM image reveals that the polymer Ag@POP-HCO₃ is formed through the accumulation and superposition of particles, resulting in irregular pore structures conducive to the enrichment of CO₂ in waste gas [Figure 12A]. TEM images demonstrate that the silver particles are uniformly dispersed within the polymer [Figure 12B and Supplementary Figure 5]. This can be attributed to the benefits of the method of metal pre-coordination, where the metal coordinated NHC-HCO₃ to form a metal complex monomer (Ag@NHC-HCO₃)_, and then copolymerizes with the crosslinking agent, ensuring the dispersion of metallic silver and preventing cluster phenomena^[10].

The XPS diagram [Figure 13 and Supplementary Figure 6] clearly indicates the binding energy of N 1s at 399.7 eV. A distinct O 1s electron binding energy is observed at 532.5 eV, providing further evidence that HCO₃^- has successfully replaced Br^-[38,39] although a minute quantity of Br^- remains. The peaks at 368.6 eV and 374.7 eV correspond to Ag 3d, representing the bimodal structure of Ag 3d_5/2 and Ag 3d_3/2, respectively, confirming the successful loading of metal Ag^[40,41].

In situ Raman spectra of Ag@POP-HCO₃ under CO₂ and N₂ atmosphere were depicted in Figure 14A. Peaks at 1,207 cm^-1 and 2,233 cm^-1 are clearly observed in the spectrum of Ag@POP-HCO₃ under CO₂, but not under N₂ atmosphere. The peak at 1,207 cm^-1 corresponds to the tensile vibration of CO₂ adsorbed on N, while the peak at 2,233 cm^-1 is derived from the antisymmetric stretching of the adsorbed CO₂ on the N-heterocyclic carbene^[42]. These findings demonstrate that N-heterocyclic carbene in Ag@POP-HCO₃ can effectively activate CO₂. In addition, elemental analysis and atomic absorption were performed on the catalyst, and the content of N and silver was measured to be 0.9% and 0.046 wt%, respectively [Supplementary Table 2].

The reaction kinetics were investigated using a 15 vol.% CO₂/N₂ gas mixture as the carbon source [Figure 14B]. It was observed that after 12 h, the yield almost reached a plateau. Further extending the reaction time to 24 h resulted in only a 6% increase in yield. Additionally, when the catalyst was filtered out at 3 h and no longer participated in the subsequent CO₂ reaction, the reaction almost stopped. This confirms that this catalytic system operates through a heterophase catalytic mechanism.

Based on aforementioned experimental results, we propose a possible mechanism for Ag@POP-HCO₃ catalyzing the carboxylation cyclization of propargyl amines with low concentration CO₂ [Scheme 3]. Firstly, CO₂ is co-enriched by the pores and HCO₃^- anions of Ag@POP-HCO₃ catalyst. During the enrichment process, CO₂ is initially adsorbed by the pores and then captured by HCO₃^- anions. The solvent DMSO exhibits strong water absorption, which allows for species transformation of HCO₃^- and CO₂, thereby enhancing the capability of the catalyst Ag@POP-HCO₃ to enrich CO_2. Subsequently, the enriched CO₂ is activated by carbene N. The Ag in the catalyst interacts with the C≡C bond of propargyl amine through empty orbitals and π electrons to active propargyl amine. The activated CO₂ and propargyl amines are converted into carbamate intermediates 3. Carbamate intermediate 3 is converted to vinyl-silver intermediate 4 by intramolecular cyclization. Finally, oxazolidinone 2a was produced by in-situ demetallization, and the catalyst Ag@POP-HCO₃ was regenerated.

The carboxylation cyclization reactions were conducted under optimized reaction conditions using 15 vol.% CO₂/N₂ as a carbon source and various substituents of propargyl amines (For the synthesis procedure, see Supplementary Schemes 3 and 4)^[27,43] as substrate [Figure 15]. In addition to the model substrate (1a), benzene substituents, including electron-donating substituents (1b-1d), electron-withdrawing groups (1f-1h), all successfully participated in the carboxylation cyclization reaction in considerable yields. Additionally, successful completion of carboxylative cyclization reaction was also observed when the propargyl amines skeleton contained a heterocyclic structure (1i) or a terminal alkyne structure (1j). The study also investigated changes in substituents at R² and R³ positions (1k), resulting in a 60% yield.

In order to access the potential applications of the Ag@POP-HCO₃ catalytic system in actual flue gases, the impact of co-existing SO₂ and NO₂ on the model reaction was investigated using simulated flue gases as a carbon source [Table 1]. The results indicated that, with both SO₂ content of up to 428 mg/m³ and NO₂ content of up to 221 mg/m³, there was no significant negative impact on the reaction, which maintained a relatively stable yield [Figure 16A]. It is worth noting that SO₂ and NO₂ content in actual flue gases are usually less than 400 mg/m³ and 200 mg/m³, respectively. Therefore, it can be concluded that the catalytic system proposed in this work is also available for actual flue gases.

In practical application, a coal-fired flue gas (containing 8 vol.% CO₂) collected from a coal-fired power plant in Guilin was treated with Ag@POP-HCO₃ catalyst. This resulted in the successful conversion of CO₂ to oxazolidinone (2a), with a remarkable yield of 50%. When reaction time was extended to 24 h and 36 h, the yield reached 62% and 81%, respectively [Figure 16B]. When benzene substituents with electron-donating substituents (1b, 1d), halogen substituents (1e) and propargyl amine scaffolded with the heterocyclic structure (1i) or a terminal alkyne structure (1j), they could react with coal-fired flue gas, and form the corresponding oxazolidinones in 62%-77% yields [Figure 15]. Furthermore, the catalyst was used to catalyze the carboxylation cyclization reaction of propargyl amines with CO₂ in real air. Initially, the yield was only 1% during the 12 h reaction time; however, after extending the reaction time to 96 h, the yield increased to 9% [Figure 16B]. This experimental data strongly demonstrate that Ag@POP-HCO₃ catalyst exhibits high selectivity and catalytic activity for CO₂, even in actual flue gas environments containing pollutants such as SO₂ and NO₂.

The model reaction of carboxylation cyclization was conducted on a gram scale [Scheme 4] with a yield of 63% within 12 h reaction time. The Ag@POP-HCO₃ catalyst recovered from the previous reaction was utilized to react propargyl amines with CO₂ in simulated flue gas under optimal reaction conditions. It was observed that the catalyst could be recycled [Supplementary Figure 7]. These findings further provide strong evidence for the practicality of the Ag@POP-HCO₃ catalyst.