The field of microporous materials has recently seen remarkable development owing to the potential applications of these materials in gas storage, separation and catalysis [1-4]. Compared with traditional porous materials, organic microporous materials have the advantages of high specific surface areas, good solubility, diverse synthetic methods and wide applicability [2]. With the ever increasing sophistication of synthetic methods available for the construction of organic porous materials, porous materials with various functions can now be synthesized according to researchers' diverse needs.

Covalent organic frameworks (COFs) are a unique class of crystalline porous organic polymers with inherent porosity and periodic molecular ordering [5-7]. It is, however, difficult to construct crystalline COFs because their syntheses require reversible condensation reactions that have to be carried out under thermodynamic rather than kinetic control [8].

Covalent triazine-based frameworks (CTFs) are a special and emerging class of COFs [9] that are produced by ionothermal trimerization of aromatic polynitriles in molten ZnCl₂ to give triazine rings [10]. For example, the first reported CTF, CTF-1 [10], was prepared by trimerization of p-dicyanobenzene. CTFs are gaining an increasing amount of attention because of their excellent chemical and thermal stabilities, ultra-high Brunauer-Emmett-Teller (BET) surface areas and porosities, as well as the high amount of nitrogen functionalities in their networks. Owing to these remarkable properties, CTFs have been applied in many fields, including catalysis [11-13], gas adsorption and storage [14-16], sorbent materials [17, 18], organic semiconductors [19], electrical materials [20, 21] and others [22-24].

To synthesize CTFs on a large scale for potential commercialization, an economical synthetic pathway is required. The key monomers used in CTFs are multiple cyano-substituted benzenes, but these are not available on a large scale. The traditional synthetic routes to cyano-substituted benzenes include Rosenmund von Braun reactions [25-27] from aryl halides, diazotization of anilines and the Sandmeyer reaction [28]. However, these synthetic methodologies are associated with the use of overstoichiometric amounts of toxic cyanides, such as Zn (CN)₂[29], CuCN [30], KCN [31], NaCN [32] and trimethylsilyl cyanide (TMSCN) [33], as the cyanide source, which is a serious environmental concern. Recently, nontoxic potassium hexacyanoferrate (Ⅱ) was introduced as a cyanation agent by Beller's group [34-36]. They demonstrated that aryl halides can react with potassium hexacyanoferrate (Ⅱ) to give cyano compounds in the presence of catalytic amounts of palladium catalysts and ligands such as 1, 5-bis (diphenylphosphino) pentane (dpppe), 1, 1′-bis (diphenylphosphino) ferrocene (dppf), triphenylphosphine (PPh₃), tricyclohexylphosphine (PCy₃) and 2, 2′-bis (diphenylphosphino) diphenyl ether (Bpephos). A drawback of this procedure is the need for expensive ligands, which not only greatly increases the cost of the experiments, but also limits the scale of the reaction. Recently, several triazine-functionalized porous polymers were synthesized using imide-functionalized 1, 3, 5-triazine frameworks and showed high CO₂-adsorption capacities [37, 38]. Modak et al. [37] reported a one-pot bottom-up synthetic strategy for the triazine-functionalized porphyrin-based porous organic polymer TPOP-1, which showed a CO₂ uptake of up to 6.2 mmol/g at 3 × 10⁵ Pa and 0 ℃. Gomes et al. [38] reported the synthesis of the polymer TRITER-1 by condensation polymerization of 1, 3, 5-tris-(4-aminophenyl) triazine (TAPT) and terephthaldehyde.

In this work, we have developed a new synthetic pathway for CTFs. First, multiple cyano-substituted aromatics were synthesized using inexpensive and nontoxic K₄[Fe (CN)₆]·3H₂O as the cyanide source for cyanation of aryl bromides. This reaction was catalyzed by Pd (OAc)₂ without addition of other ligands. Second, CTFs were synthesized by polymerization of the multiple cyano-substituted aromatics in molten ZnCl₂, which acts as both a catalyst and a solvent at temperatures of 400-600 ℃ [10]. The obtained CTFs exhibited excellent CO₂-adsorption properties.

All reagents were purchased from Alfa Aesar, J & K Scientific Ltd., Aladdin Reagent and Sigma Aldrich, and were used without further purification. 1, 4-Dicyanobenzene (1b), 2, 6-dicyanopyridine (2b) and 1, 4-dicyanonaphthalene (3b) were prepared from their corresponding aryl bromides with K₄[Fe (CN)₆]·3H₂O in the presence of Pd (OAc)₂ according to a previously reported method [34, 39]. 1b-3b were isolated from the reaction mixtures and characterized by ¹H NMR and ¹³C NMR spectroscopy.

GC-MS analysis was performed on a GCMS-QP2010SE spectrometer. n-Hexadecane was selected as the internal standard for GC-MS after optimization. The NMR spectra (¹H, 400 MHz; ¹³C, 100 MHz) were recorded on a Bruker Mercury Plus 400-MHz spectrometer. CDCl₃ was used as the solvent with tetramethylsilane (TMS) as the internal standard. The synthesis of CTFs was carried out in a muffle burner (TSX1200). The specific surface area of the CTFs was measured on an adsorption instrument (Quantachrome Quadrasorb SI).

Aryl bromide (6.0 mmol), K₄[Fe (CN)₆]·3H₂O (2.4 mmol), Na₂CO₃ (6.0 mmol) and Pd (OAc)₂ catalyst (0.2 mol% relative to the aryl bromide) were mixed in NMP (12 mL) in a 25-mL flame-dried double-neck round-bottom flask and heated with stirring at 130 ℃ under argon. The progress of the reaction was monitored by GC-MS. At the end of the reaction, the mixture was cooled to room temperature. Samples were quenched with H₂O (or with sat. NH₄Cl when pyridines were used as substrates), and extracted with EtOAc (25 mL × 3). The organic layer was washed with water (15 mL × 3) and 5% NH₃·H₂O (15 mL × 2), and then dried over Na₂SO₄. Evaporation of the solvent left the crude product, which was further purified by column chromatography over silica gel (60-120 mesh) eluting with petroleum ether/ethyl acetate to afford the pure aryl nitrile.

1, 4-Dicyanobenzene (1b, DCB). ¹H NMR (CDCl₃, 400 MHz): δ=7.80 (s, 4H); ¹³C NMR (101 MHz, CDCl₃): δ=132.81, 117.00, 116.76.

2, 6-Dicyanopyridine (2b, DCP). ¹H NMR (CDCl₃, 400 MHz): δ=8.15 (dd, J=8.4, 7.4 Hz, 1H), 7.99 (d, J=7.8 Hz, 2H); ¹³C NMR (101 MHz, CDCl₃):δ=139.13, 135.22, 131.29, 115.53.

1, 4-Dicyanonaphthalene (3b, DCN). ¹H NMR (CDCl₃, 400 MHz): δ=8.35 (dd, J=6.4, 3.2 Hz, 2H), 7.98 (s, 2H), 7.87 (dd, J=6.4, 3.2 Hz, 2H); ¹³C NMR (101 MHz, CDCl₃): δ=132.00, 131.09, 130.30, 126.03, 116.27, 115.20.

Traditional CTF synthesis has been described elsewhere [10]. In this work, aryl nitriles (7.8 mmol) and anhydrous ZnCl₂ (39.0 mmol) were transferred into a quartz ampoule (15 cm³) under an inert atmosphere. The ampoule was evacuated, sealed and heated to 400 ℃ for 20 h and then to 600 ℃ for another 20 h. The ampoule was then cooled to room temperature and opened carefully (Caution: the ampoule is under pressure, which is released on opening). The reaction mixture was then ground and washed thoroughly with large amounts of water to remove most of the ZnCl₂. It was then stirred in dilute HCl for 15 h to remove the residual salt. After this purification step, the resulting black powder was filtered, washed successively with water and THF and dried in vacuum at 120 ℃. The yield of the reaction was high ( > 91%) to quantitative.

We used 1, 4-dibromobenzene as a model substrate to screen suitable reaction conditions for the cyanation of aryl bromides (Table 1). We obtained similar yields using N, N-dimethylformamide (DMF) and 1-methyl-2-pyrrolidinone (NMP) as the solvent (around 60%, Table 1, Entries 1 and 3); however, the yield decreased to only 41% after replacing the solvent with N, N-dimethylacetamide (DMAc) (Table 1, Entry 2). To our surprise, a high yield of 1, 4-dicyanobenzene was obtained at a relatively low catalyst concentration when the reaction time was prolonged (Table 1, Entries 5 and 6). However, increasing the amount of Pd (OAc)₂ led to a decrease in the yield (Table 1, Entries 7-9). This might have been caused by excessive catalyst aggregation and transformation to insoluble palladium black [34, 40]. In practice, the use of (0.1-0.2) mol% ligand-free Pd (OAc)₂ gives good yields in the cyanation reaction of 1, 4-dibromobenzene with K₄[Fe (CN)₆]·3H₂O (NMP, Na₂CO₃, 130 ℃). No desired product was observed when the cyanation was performed in DMSO, toluene or ethylene glycol owing to the poor solubility of 1, 4-dibromobenzene and K₄[Fe (CN)₆] (Table 1, Entries 10-12). The best reaction conditions were thus determined to be 0.2 mol% Pd (OAc)₂ and NMP as the solvent, with a reaction time of 2 h ( > 99%) (Table 1, Entry 4).

Based on the above result, we extended the substrate scope to heteroaromatic compound 2, 6-dibromopyridine (2a) and polycyclic aromatic hydrocarbon 1, 4-dibromonaphthalene (3a), and the reactions were performed under the optimized reaction conditions (Table 2). Accordingly, 3a was almost completely converted to 1, 4-dicyanonaphthalene (3b). However, the yield of 2, 6-dicyanopyridine (2b) was less than 58%, even after 9 h of reaction time.

CTF_DCB, CTF_DCP [10] and CTF_DCN frameworks (Table 2) were synthesized by heating a mixture of the nitrile monomers DCB, DCP and DCN and ZnCl₂ in quartz ampules at 400-600 ℃ for 40 h. The molten ZnCl₂ acts as both solvent and catalyst for the self-polymerization of the triazine rings. The molar ratio of ZnCl₂ to the nitrile monomer was 5:1, which allowed the reaction to proceed smoothly.

The porosity, that is, the surface area and pore volume, and the pore size distribution of all CTFs were characterized by N₂ adsorption measurements (Fig. 1). All CTFs were outgassed in vacuum at 150 ℃ for 10 h, and the results are listed in Table 3. As shown in Fig. 1(a), the N₂ isotherms indicate that the three polymers are microporous; this is also demonstrated by the sharp adsorption/desorption step in the low relative pressure region (P/P₀=0-0.05). The isotherms for CTF_DCB and CTF_DCN also exhibit hysteresis loops, which indicate the presence of mesopores. The H2-type hysteresis loops present in the CTF_DCB and CTF_DCN desorption isotherms may be caused by the different mechanisms of the condensation and evaporation processes in the micro-and mesopores [41]. The surface areas of the CTFs, calculated using the BET equation, were 2404, 1650 and 780 m²/g for CTF_DCB, CTF_DCP and CTF_DCN, respectively. The pore size distributions of CTF_DCB, CTF_DCP and CTF_DCN were calculated by HK methods. As shown in Fig. 1(b), all three CTFs exhibit small micropore diameters centered around 0.5 nm; for example, CTF_DCB had a maximum pore size of 0.57 nm. Moreover, the overall micropore volume of CTF_DCB was 0.85 cm³/g, and those of CTF_DCP and CTF_DCN were 0.76 and 0.52 cm³/g, respectively (Table 3). The porosity and texture of the CTFs were homogeneous and the presence of small micropores was confirmed by the high-resolution transmission electron microscopy images shown in Fig. 2.

As shown in Fig. 3(a) and (b), CO₂ adsorption by CTFs was also investigated using volumetric methods at 0 and 25 ℃. All of the CTFs showed reversible CO₂ adsorption isotherms. The CO₂ uptake of CTF_DCP, which was synthesized from 2, 6-dicyanopyridine, reached 225 and 154 mg/g under 1 x 10⁵ Pa at 0 and 25 ℃, respectively. However, the CO₂ uptake of CTF_DCB, which had the highest BET surface area among the three CTFs studied, was only 148 mg/g at 0 ℃ and 99 mg/g at 25 ℃, close to the 129 mg/g at 0 ℃ and 97 mg/g at 25 ℃ measured for CTF_DCN. The high adsorption ability of CTF_DCP, compared with those of CTF_DCB and CTF_DCN, derives from the extra nitrogen atoms in the structure, which serve as active alkali sites and attract CO₂ molecules [42]. Comparison of the CO₂ uptake of these CTFs with those of metal organic frameworks (MOFs), conjugated microporous polymers (CMPs)[43] and zeolites [44], as well as those obtained under identical conditions for our previously reported CMPs, Co-CMP and Al-CMP [45], shows that the CTFs have good CO₂capture abilities at both 0 and 25 ℃ (Table 4). To better understand the adsorption properties, the isosteric heats of adsorption were calculated using a variant of the Clausius Clapeyron equation [46] from fits of the CO₂ isotherms at different temperatures. As shown in Fig. 3(c), the isosteric heats of CO₂ adsorption on CTF_DCB, CTF_DCN and CTF_DCP were calculated to be in the range 18.4-23.5, 20.7-24.1 and 22.5-25.3 kJ/mol, respectively. CTF_DCP had the highest isosteric heat of CO₂ adsorption, which is consistent with its large CO₂-uptake capacity.

We have developed an environmentally friendly method for the direct synthesis of multiple cyano-substituted aromatic compounds. This method uses palladium acetate without any additional ligands as the catalyst and nontoxic potassium hexacyanoferrate (Ⅱ) as the cyanide source, thus removing the need for special precautions. Notably, aromatic nitriles could be obtained in excellent yields using catalyst amounts of only (0.2-0.4) mol%. Moreover, three CTFs were synthesized from the obtained cyano-aromatic compounds. These CTFs exhibited high specific surface areas and excellent CO₂ absorption abilities. Furthermore, the synthesized porous polymers had good stability and multiple active sites for metal coordination; we thus expect them to find application in catalyst supports.