Furukawa and Yaghi (2009) began the work into COFs for CO₂ sequestration by the examining the gas storage capability of boron COFs for the storage of CO₂ at both high and low pressures and found that the ability to adsorb CO₂ was directly related to the internal surface area. In this study by Furukawa et al. the COFs were divided into three groups: The first two groups were 2D structures that had 1D pores and the third group consisted of 3D structures with 3D pores. The pores of group 1 were 9Å (COF-1 and COF-6), while group 2 materials had larger pores (27,16, and 32 Å for COF-5, COF-8, and COF-10, respectively). The 3D COFs in group 3 had medium-sized pores (12 Å for COF-102 and COF-103 both). The 3D COFs in the third group had the best gas uptake capacities. For instance, COF-102 had gas uptake of 1,180 mg g⁻¹ at 298 K for carbon dioxide measured at 35 bar.

The synthesis of the boron linked COF for the purpose of CO₂ capture was continued by Kahveci et al. with the development of TDCOF-5 which was synthesized from a mixture of 1,4-benzene diboronic acid and hexahydroxytriptycene heated in a mesitylene–dioxane solvent mixture for 5 days at 120°C (Kahveci et al., 2013). The calculated Brunauer–Emmett–Teller (BET) surface area was found to be 2,497 m²g⁻¹ while the Langmuir model gave a surface area of 3,832 m²g⁻¹ that is similar to the calculated Connolly surface (4,973 m²g⁻¹). TDCOF-5 can store up to 9.2 wt% of CO₂ (2.1 mmol g⁻¹) with a Q_st value of 21.8. The use of the triptycene core in TDCOF-5 leads to more accessible boron sites that can facilitate interaction with gas molecules. Greater accessibility to these boron sites increased the CO₂ adsorption performance of TDCOF-5 at low pressures.

Synthesis of COFs capable of competitive CO₂ adsorption was expanded to the design of imine linked COFs by Rabbani et al. they reported the synthesis of ILCOF-1 with an internal surface area (Braunner-Emmet-Teller) of 2,723 m² g⁻¹ and a Langmuir surface area of 3,453 m² g⁻¹ (Rabbani et al., 2013). ILCOF-1 was shown to be exceptionally well suited for high pressure H₂, CH₄, and CO₂ storage. ILCOF-1 was synthesized by the condensation reaction between 1,3,6,8-tetrakis(p-formylphenyl)pyrene and p-phenylenediamine in 1,4-dioxane in the presence of aqueous acetic acid at 120°C for 3 days (Figure 2). The use of expanded pyrene cores in the construction of imine-linked COFs represented a breakthrough in imine COF synthesis as previous work such as the 3D COF-300 had a surface area of SA_BET = 1,360 m² g⁻¹ (Uribe-Romo et al., 2009). Previous to this work by Rabbani et al. the use of imine-linked COFs in gas storage had been only investigated theoretically (Mendoza-Cortes et al., 2012). ILCOF-1 was stable up to 400°C as determined by thermogravimetric analysis (TGA). The calculated surface area (SA_BET) and major pore-size distribution are much larger than other crystalline imine-linked COFs including COF-300 (1,360 m² g⁻¹, 7.2 Å) (Uribe-Romo et al., 2009); COF-366 (735 m² g⁻¹, 17.6 Å) (Wan et al., 2011); and COF-LZU1 (410 m² g⁻¹, 12 Å) (Ding et al., 2011). ILCOF-1 showed significant CO₂ uptake at high pressure (29.3 mmol g⁻¹ at 298 K and 40 bar) these values are greater than those for 2D and 3D COFs (5.2–27.3 mmol g⁻¹ at 298 K and 55 bar). ILFOC-1 also outperforms hyper-branched conjugated polymers (HCPs; 10.6–13.3 mmol g⁻¹ at 298 K and 35 bar) (Martín et al., 2011) and PAF-1(29.6 mmol g⁻¹ at 298 K and 40 bar) (Ben et al., 2009). An explanation for the high measured CO₂ uptake of ILCOF-1 is the enhancement of CO₂ binding created by the N···CO₂ interactions in the nitrogen-containing pore walls ILCOF-1 which do not exist in the nitrogen-free PAF and HCP structures.

Ge et al. (2016) synthesized an azo (N = N) based covalent organic framework (COF-TpAzo) via a Schiff base condensation reaction. COF-TpAzo was synthesized by the condensation reaction between 1,3,5-triformylphloroglucinol (Chong et al., 2003) (0.3 mmol, 63.0 mg) and 4,4′-azodianiline (Santurri et al., 2003) (0.45 mmol, 95.4 mg) in 1,4-dioxane (3 mL). The maximum CO₂ adsorption of COF-TpAzo was 105.6 mg·g⁻¹ (Ge et al., 2016). Yet COF-TpAzo was also 39 times more selective to CO₂ over CH₄ and 127 times more selective of CO₂ over N₂ (Table 1 entry 69). At 298 K the CO₂/N₂ and CO₂/CH₄ selectivity of COF-TpAzo was 145 and 43, respectively. The increase at in CO₂ selectivity at a higher temperature was explained by conventional CO₂ affinities and the N₂ phobicity of azo group (Patel et al., 2013).

Work by Li Z. et al. toward the development of stable COFs with C-N linkages for CO₂ specific adsorption involved the synthesis of a new azine-linked covalent organic framework, ACOF-1, by condensation of hydrazine hydrate and 1,3,5-triformylbenzene under solvothermal conditions (Li Z. et al., 2014). ACOF-1 is with high selectivity toward CO₂ over N₂ and CH₄. ACOF-1 can store up to 177 mg g⁻¹ of CO₂ at 1 bar. Dalapati et al. (2013) studied the Py-Azine COF, because it can be engineered with smaller pore sizes due to the short structural length of the azine unit. The CO₂ isotherms of ACOF-1 at 273 K (Figure 3), showed 17.7 wt% CO₂ uptake at 273 K and 1 bar. This values is greater than CO₂ uptake reported for COF-5 (5.9 wt%, SA_BET = 1,670 m² g⁻¹) (Furukawa and Yaghi, 2009), COF-103 (7.6 wt%, SA_BET = 3,530 m²g⁻¹) (Furukawa and Yaghi, 2009), TDCOF-5 (9.2 wt%, SA_BET = 2,497 m² g⁻¹) (Kahveci et al., 2013), and ILCOF-1 (6.0 wt%, SA_BET = 2,723 m² g⁻¹) (Rabbani et al., 2013). This is despite the fact that the aforementioned COFs have larger internal surface areas. Dalapati et al. account for the higher CO₂ uptake capacity of ACOF-1 by considering the great number of nitrogen sites at the pore walls and excellent porosity as well as the high specific surface area, large pore volume and small pore size of the material (Shen et al., 2013).

Li et al. continued their work on azine linked COFs for the capture of CO₂ by synthesizing COF-JLU2 this time utilizing 1,3,5-triformylphloroglucinol and hydrazine hydrate as starting materials (Li et al., 2015). COF-JLU2 (Figure 4A). The CO₂ adsorption isotherms were measured as high as 1 bar at 273 and 298 K. COF-JLU2 shows a CO₂ uptake of 21.7 wt % at 273 K and 1 bar (Figure 4B, Table 1 entry 10). The isosteric heats of adsorption (Q_st) of COF-JLU2 were estimated for CO₂ by analyzing adsorption data collected at various temperatures so that better understanding of the CO₂ uptake properties could be obtained (Reid et al., 1998). COF- JLU2 shows a Q_st of 31 kJ mol⁻¹ (Table 1 entry 10) at lower adsorption values, this is greater than the values reported for COFs (Côté et al., 2007; El-Kaderi et al., 2007; Tilford et al., 2008; Furukawa and Yaghi, 2009; Kahveci et al., 2013), imine-linked organic cages (Jin et al., 2011), and diimide polymer (Farha et al., 2009). Li et al. attribute the excellent CO₂ uptake capacity with high adsorption to the inherent microporosity of framework and the strong dipole-quadrupole interactions between the nitrogen and oxygen atoms in COF-JLU2 and CO₂.

While several COFs show the ability to capture CO₂, industrial applications of the COFs are hampered by their intolerance to harsh conditions especially water (Lanni et al., 2011; Uribe-Romo et al., 2011; Li et al., 2018; Zhang et al., 2018). The crystallinity of COF materials is dependent upon of the reversible reactions that comprise their synthesis; with the kinetic, amorphous products initially being the dominant species but the eventual formation of the lower energy thermodynamic product leading to COF crystals as the major product (Zhu and Zhang, 2017). The four reactions by which most COFs are synthesized, boronic trimerization, boronate ester formation, trimerization of nitriles, and Schiff base condensation are all reversible and so any crystal formed by these reactions will suffer degradation even in mild aqueous environments (Lanni et al., 2011; Spitler et al., 2011). Making COFs from non-reversible reactions is a challenging but effective alternative (Zhang et al., 2018). In a breakthrough work by Kandambeth et al. (2012), a robust COF was synthesized from an irreversible reaction in two steps; the first being a reversible Schiff base reaction and the second being an irreversible keto-enol tautomerization. The new COFs (TpPa-1 and TpPa-2) (Kandambeth et al., 2012) were tested by Kandambet et al. to determine the CO₂ uptake capacity. The CO₂ uptake of TpPa-1 was found to be 153 mg g⁻¹ at 273 K which makes it as effective as COF-6 (Furukawa and Yaghi, 2009) but with a great increased stability. TpPa-1 and TpPa-2 were soaked in water for 7 days after which pXRD peaks showed no change in position and FTIR analysis gave no new peaks to indicate the COF returning to starting material. Even after treatment in 9N HCl for 3 days the positions and intensities of pXRD peaks of the two COFs remained the same. TpPa-2 was able to withstand 9N NaOH for 7 days without a change in surface area, FTIR peaks, or pXRD pattern. This impressive stability is a result of the irreversible enol-to-keto tautomerism. Later work by Zou et al. used a mechanochemical synthesis of TpPa-COF nanosheet clusters and effectively applied the nanosheet clusters for CO₂ capture (Zou et al., 2017). TpPa-COF was also synthesized by use of microwave radiation in 1 h (Wei et al., 2015).

Recently EL-Mahdy et al. (2018) reported the synthesis and CO₂ adsorption of triphenylamine COFs (TPA-3NH₂) and (triphenyltriazine COFs) (TPT-3NH₂). TPA-3NH₂ was synthesized by the polycondensation of tris(4-aminophenyl)amine with various triarylaldehydes while TPT-3NH₂ was synthesized by the polycondensation of 2,4,6-tris(4-aminophenyl)triazine with various triarylaldehydes (Figure 5). The triacrylaldehydes differed in planarity, symmetry, and nitrogen content. By considering the symmetry and planarity of the monomers the highly crystallin COFs can be custom designed for specific applications. El-Mahdy et al. noted that the use of symmetrical monomer units for the synthesis of COFs with hexagonal structure was successfully achieved by other researchers (Haase et al., 2017; Zhai et al., 2017; Gao et al., 2018) but these materials have limited application in gas storage. El-Mahdy et al. sought to control the crystallinity of synthesized COF and elucidate the effect of crystallinity on the CO₂ uptake. The pXRD patterns of the COFs confirmed the crystallinity of the synthesized materials and pointed to an increasing degree of crystallinity upon increased planarity of the monomers. El-Mahdy et al. related the surface are of the COF to the symmetry and planarity of their monomers by determining the Brunauer–Emmett–Teller (BET) and Langmuir surface areas of the synthesized COFs (Table 1).

The BET surface area of TPT-COF-6 which was made from the most symmetrical and planar linkers was 1,535 m² g⁻¹. This value was more than double that of the TPA–TPA-COF (TPA-COF-1) (714 m² g⁻¹), synthesized from a mixture of the symmetrical and least planar building blocks. This was not the case for unsymmetrical monomers. The TPA–TPP-COF (TPA-COF-2), synthesized from the least planar building blocks (TPA-3CHP), had a BET surface area of 478 m² g⁻¹ and TPA-COF-3 had a BET surface area of 557 m² g⁻¹ prepared from the most planar TPT-3CHO monomer (Table 1 entry 63). In the same vein, TPT-COF-4 had a BET surface area of 1,132 m² g⁻¹ which was markedly less than that of TPT-COF-5 (1,747 m² g⁻¹) (Table 1 entries 64–65). EL-Mahdy et al. (2018) summarize that the symmetry and planarity are the two main factors that affect the COF's total surface area. Synthesizing COFs with less planar starting materials reduced the surface area for both the symmetric and unsymmetrical COFs.

The CO₂ adsorption of TPA-COFs and TPT-COFs at temperatures of 298 and 273 K (Table 1 entries 61–66) were examined by El-Mahdy et al. The CO₂ uptake efficiency was heavily influenced by the nitrogen content and nitrogen nucleophilicity of the COFs. The TPA-TRF-COF (TPA-COF-3) had the highest CO₂ uptake capacity (63.94 and 91.15 mg g⁻¹ at 298 and 273 K) and had the highest nitrogen content of all the TPA-COFs synthesized in this study. The TPA-TPP-COF (TPA-COF-2) showed CO₂ uptake 1.2-fold higher than that of the TPA-TPA-COF (TPA-COF-1), despite the fact that the former had three triarylamine and three triarylpyridine unites in each pore while the latter had 6. It should be added that the nitrogen content for these COFs is similar. El Mahdy et al. deduced that the better uptake efficiency of TPA-COF-2 was caused by the steric hindrance around the nitrogen atoms in the non-planar triarylamine unit. The nucleophilicity of the nitrogen atom decreased as the steric hindrance around the nitrogen increased; which decreases the number of carbon dioxide molecules adsorbed (Kortunov et al., 2016). The same concepts can be applied to the TPT-COF series for which TPT-COF-6 had the highest CO₂ uptake capacities 65.65 and 92.38 mg g⁻¹ at 298 and 273 K, respectively, TPT-COF-6 had six triaryltraizine units. TPT-COF-5 showed CO₂ uptake capacity 1.1 times higher than TPT-COF-4 (Table 1 entries 64–66).

The aforementioned works make it clear that controlling the stacking and connectivity of COF materials will lead to an increase in CO₂ uptake. However, while synthetic procedures for controlling the structure in MOFs (Stock and Biswas, 2012; Furukawa et al., 2013) have been well studied less is known about altering synthetic methods to realize structural differences in a COF. Gao et al. (2018) set out to understand if different structural COF variant could be made from the same flexible building blocks but by using different solvents for synthesis. Gao et al. found the mixture of tetraphenylethane (Ascherl et al., 2016; Pang et al., 2016) building blocks; which contain 4 amine or aldehyde units, heated for 7 days in a o-dichlorobenzene and n-butanol solvent mixture will synthesize [4+4] TPE-COF-I but when the building blocks were heated in 1,4-dioxane the new [2+4]. TPE-COF-II was synthesized (Figure 6). TPE-COF-II had exceptional CO₂ adsorption performance of 118.8 cm³g⁻¹ (23.2 wt% CO₂ at 1 atm, 273 K). TPE-COF-I was less effective for CO₂ uptake with 68.6 cm³ g⁻¹ (13.4 wt%, 1 atm, 273 K) and 37.8 cm³ g⁻¹ (7.38 wt%, 1 atm, 298 K). The BET surface area of TPE-COF-I and TPE-COF-II were found to be 1,535 and 2,168 m²g⁻¹. The pore volume of TPE-COF-II was 2.14 cm³ g⁻¹ (P/P₀ = 0.984) this was larger than the pore size of TPE-COF-I (1.65 cm³g⁻¹ P/P₀ = 0.983) (Table 1 entries 67, 68).

Molecules with perfluorinated moieties were developed by Zhao et al. as starting materials for COFs capable of effectively adsorbing CO₂ in environments with high water levels (Zhao et al., 2013). Zhao et al. devised a perfluorinated covalent triazine-based framework (FCTF-1) that could capture CO₂ with high selectivity. The fluorine (F) groups were crucial for increasing FCTF-1′s CO₂ adsorption and selectivity because the very polar C–F bonds encourage CO₂ adsorption by inducing electrostatic interactions. These electrostatic interactions become more important at low pressures. At 273 K and 0.1 bar, FCTF-1's CO₂ uptake was 1.76 mmol g⁻¹ through equilibrium adsorption. It is known that perfluorinated alkenes have a remarkable affinity toward CO₂ (DeSimone et al., 1994; Eastoe et al., 2006) motivating the incorporation of perfluorinated alkene linkers into a COF framework. Zhao et al. utilized the known CTF-1 triazine framework as the host for perfluorinated molecules. The combination of trazine frameworks with perfluorinated moieties is beneficial for the following reasons:

FCTF-1 was synthesized by Zhao et al. following the same procedure as developed for CTF-1 (Kuhn et al., 2008) except that tetra-fluoroterephthalonitrile was used as the precursor. The incorporation of F groups into CTF not only enhanced the CO₂ adsorption capacity but also led to more preferential adsorption of CO₂ over N₂, i.e., higher CO₂-N₂ selectivity. FCTF-1, however, lacked the crystallinity of CTF-1.

Later work on the synthesis of COFs for the adsorption of CO₂ sought to increase the affinity of the porous materials for CO₂ through the introduction of CO₂ binding moieties onto the channel walls of the COF while retaining crystallinity (Huang et al., 2015a,b; Ge et al., 2016; Wang et al., 2018). Huang et al. showcased this strategy by fixing COOH moieties into the channel walls of a imine-linked 2D COF. This COF consists of porphyrins linked by phenol containing bridges with phenol groups on the pore walls (Figure 7); this 2D COF has a low carbon dioxide adsorption capacity prior to installation of the COOH functional groups (Huang et al., 2015a). The carboxylic acid content in the COF pores was controlled by altering the ratio of the 2,5-dihydroxyterephthalaldehyde (DHTA) and 1,4-phthalaldehyde (PA) bridging linkers.

After addition of the COOH moieties in the pores of [HO₂C]_x%-H₂P-COFs, the COF becomes microporous, this is made evident by the typical type I sorption curve in Figure 8. As the percent of bridging linkers containing COOH groups increased from 25 to 100% the BET surface area decreased from 786 to 364 m² g⁻¹ whereas the pore size decreased from 2.2 to 1.4 nm (Table 1 entries 31–34). It has been previously reported that carboxylic acid groups engage in dipolar interactions with carbon dioxide (Rochelle, 2009; Dawson et al., 2011b). The carboxylic acid functional groups in [HO₂C]_x%−H₂P-COFs are positioned at the termini and have the relatively same pK_a value (5.86) as free carboxylic acids. Huang et al. show that [HO]_x%-H₂P-COFs exhibit low CO₂ uptake capacities at 1 bar and 273 K when compared to the [HO₂C]_x%-H₂P-COFs. The HO]_x%-H₂P-COFs had capacities of 31–35 mg g⁻¹ at 273 K and 46–63 mg g⁻¹ at 298 K whereas the [HO₂C]_x%-H₂P-COFs had capacities of 180 mg g⁻¹ at 273 K and 76 mg g⁻¹ at 298 K. The adsorption capacities of the [HO₂C]_x%-H₂P-COFs was directly correlated to the carboxylic acid content as shown in Table 1, entries 31-34. This is strong evidence for the validity of channel wall COOH functionalization as a method to increase CO₂ adsorption.

Huang et al. (2015b) expanded the study of increasing CO₂ uptake in COFs through channel wall optimization by using click reactions to install a litany of functional groups in the pores of an imine linked porphyrin COF. Huang et al. synthesized four novel COFs each with a different percentage of ethynyl functional groups installed in the pore walls. This was accomplished through condensation between 5,10,15,20-tetrakis(p-tetraphenylamino)porphyrin and a variable mixture of 2,5-bis(2-propynyloxy)terephthalaldehyde (BPTA) and 2,5-dihydroxyterephthalaldehyde (DHTA). The resultant [HC≡C]_x-H₂P- COFs had ethynyl contents of X = 25, 50, 75, and 100 where X = [BPTA]/([BPTA] + [DHTA]) ×100. After isolation of the COFs, Huang et al. used click reactions to produce 20 new COFs with ethyl, acetate, hydroxyl, carboxylic acid, and amino groups; ranging from hydrophobic to hydrophilic and from basic to acidic (Figure 9).

Huang et al. noted that CO₂ uptake was influenced by the functional groups that had been installed. The introduction of ester units onto the pore walls of the COF to make ([MeOAc]_x-H₂P-COFs) resulted in CO₂ uptake ability that far surpassed COFs made through the introduction of ethyl units ([Et]_x-H₂P-COFs) x = 25, 50, 75, 100 (Table 1 entries 39–47). [MeOAc]₅₀-H₂P-COFs exhibited CO₂ adsorption capacities of 47 and 88 mg g⁻¹ at 298 and 273 K, respectively (Table 1 entries 44–47). This value is 1.6-fold greater than that of the best-performing [Et]_x-H₂P-COFs (Table 1 entries 40–43). As was determined in the previous work Huang et al. found that the introduction of carboxylic acid groups improved the CO₂ uptake ability of the COF greatly. Another well performing functional group was the hydroxyl group (Table 1 entries 52–55). For instance, [EtOH]₅₀-H₂P-COF exhibited a capacity of 71 mg g⁻¹at 298 K and 124 mg g⁻¹ at 273 K, these values are 2.3–2.4 times higher than those found for [HC≡C]₂₅-H₂P-COF. The CO₂ adsorption was markedly improved upon installation of amino groups on the pore surface (Table 1 entries 56–59). The capacities of [EtNH₂]₂₅-H₂P-COF, [EtNH₂]₅₀-H₂P-COF, [EtNH₂]₇₅- H₂P-COF, and [EtNH₂]₁₀₀-H₂P-COF were 60, 82, 67, and 52 mg g⁻¹ at 298 K and 116, 157, 133, and 97 mg g⁻¹ at 273 K, respectively. The highest adsorption capacity of [EtNH₂]₅₀-H₂P-COF showed an increase nearly three times higher than the adsorption capacity of [Et]₅₀-H₂P-COF and [HC≡C]₅₀-H₂P-COF.

All the COFs mentioned so far possess neutral frameworks. Incorporation of ionic building blocks would introduce a myriad of tunable and novel functions to a COF framework but would interrupt the π-π stacking which is essential to 2D COF crystallinity (Du et al., 2015; Huang et al., 2016; Ma et al., 2016; Diercks and Yaghi, 2017). Huang et al. (2017) were able to design an ionic COF (PyTTA-BFBIm-iCOF) which was able to achieve high crystallinity despite the ionic interfaces on the pore walls due to an atypical reverse AA-stacking mode. PyTTA-BFBIm-iCOF was synthesized by 4,4′,4″,4^‴–(pyrene-1,3,6,8-tetrayl) tetraaniline (PyTTA) as the neutral knot and 5,6-bis(4-formylbenzyl)- 1,3-dimethyl-benzimidazolium bromide as the cationic linker; the two moieties were linked by an imine bond. The two building blocks were mixed in o-dichlorobenzene and n-butanol in the presence of acetic acid catalyst and heated to 120°C for 3 days under solvothermal conditions and produced a crystallin powder as shown by the pXRD pattern Figure 10. The pXRD of PyTTA-BFBIm-iCOF after 24-h immersions in 3M HCl, 3M NaOH, or DMF were still crystalline as determined by pXRD.

The BET surface area was determined to be up to 1,532 m² g⁻¹; only one 2.3 nm pore was determined by DFT with the total pore volume estimated to be 0.7 cm³g⁻¹. PyTTA-BFBIm-iCOF uptakes CO₂ with capacities of 93 mg g⁻¹ at 298 K and 177 mg g⁻¹ at 273 K and 1 bar (Table 1 entry 60). This was determined to be almost three times the performance of the analogous neutral COF (PyTTA-TPhA-COF) which was 65 mg g⁻¹ at 273 K and 36 mg g⁻¹ at 298 K and 1bar. PyTTA-BFBIm-iCOF had a Q_st value of 30.2 kJ mol⁻¹ as calculated from the isotherms.

Babarao and Jiang (2008) performed atomistic simulation studies on 3D, 2D, and 1D COFs. Simulations were performed with a Gibbs ensemble Monte Carlo (GEMC) simulations at set pressure (Jiang and Sandler, 2003, 2006). One simulation box was used for the adsorbent and another for bulk adsorbate. Molecules could move from one box to the other, but the total number of adsorbent molecules remained constant.

In these simulations, the adsorbent volume remained constant, but the bulk phase was allowed to change at a fixed pressure. This method allows the uptake to be determined at the desired bulk pressure, density, and enthalpy. By simulating the aggregation of centers-of-mass for 100 equilibrium CO₂ configurations, Babarao and Jiang (2008) were able to illustrate the density distribution contours for COF-106, COF-6, and COF_NT at 1,000 kPa, as seen in Figure 11. This allows the favorable locations of CO₂ adsorption could be estimated accurately. Babarao and Jiang (2008) found that CO₂ adsorption in COF-108 takes place close to the carbon-oxygen-boron rings, as the pressure increases adsorption will take place in the central cavity. In COF-6 CO₂ molecules accumulate at the cramped central pores; this is because the inter-layer distance in COF-6 is small. In the 1D COF_NT, the CO₂ molecules are adsorbed in the center of the nanotubes as well as the small spaces amongst the tubes.

Babarao and Jiang (2008) developed a function of the framework density ρf, free volume V_free, porosity ϕ, and accessible surface area A_surf as a method to quantitatively determine the CO₂ storage capacity. The framework density (ρf) and gravimetric capacity (Nexgra) are inversely related. The volumetric capacity (Nexvol) decreases linearly. Expressions determined by Babarao and Jiang (2008) for Nexgra and Nexvol (1, 2) can be seen below.

As free volume (V_free) rises, Nexgra rises as well. Nexvol has a small drop after an initial small rises. Expressions determined by Babarao and Jiang (2008) for the relation between Nexgra and Nexvol with V_free are shown below (3, 4).

Nexgra and Nexvol rise with the porosity guided by the expressions (5, 6) determined by Babarao and Jiang (2008)

Nexgra and Nexvol can be expressed as functions of accessible surface area (A_surf) as seen in expressions (7, 8) as determined by Babarao and Jiang (2008) As A_surf increases so do Nexgra and Nexvol.

There is a useful interrelationship between Nexgra and Nexvol with the framework density, free volume, porosity, and accessible surface area. Nexgra and Nexvol would be complemented through a decrease in the framework density or by increasing the free volume, porosity or accessible surface area. These insights elucidated by Babarao and Jiang (2008) are useful for predicting the CO₂ adsorption uptake prior to a COF synthesis.

Computational studies aiming to develop new strategies for chemists to optimize the CO₂ adsorption of new COF materials utilize simulations of known COFs. Tong et al. (2014) investigated the effects of different π-π stacking motifs as they related to the ability for CO₂ capture by simulating 46 COFs. 2D-COFs with hexagonal and tetragonal topologies and 3D-COFs with ctn, bor, and dia topologies. This set of COFs included the majority of main group light elements to yet be synthesized (C, O, B, N, H, F, Si, etc.), porphyrinic COFs which incorporated heavy metals like Zn and Ni were also in studied. The comprehensive analysis of elements provided a solid basis for obtaining relevant and accurate results. Tong et al. calculated charges of the atomic particles for the 46 COFs using the DFT calculations for the purpose of modeling guest–host intermolecular interactions. Subsequently grand canonical Monte Carlo (GCMC) simulations were employed to investigate the adsorption behaviors of the CO₂-N₂ mixture in COFs at 298 K. This was done using CADSS (Complex Adsorption and Diffusion Simulation Suite); a code developed by Tong et al. The simulation consisted of five types of trials for the molecules: attempts (i) to randomly displace a molecule (translation or rotation), (ii) to re-grow a molecule at a random position, (iii) to create a new molecule, (iv) to delete an existing molecule, and (v) to exchange the molecular identity. Three properties of the COF were studied by GCMC simulations: (i) structure–selectivity relationship, (ii) structure–working capacity relationship, (iii) structure–sorbent selection parameter relationship. In prior work, Wu et al. had it developed a concept dubbed “adsorbility (AD)” for MOFs (Wu et al., 2012a) and found a good correlation between the selectivity and difference in adsorbility of adsorbates (ΔAD = ΔQst0/φ) in MOFs (Wu et al., 2012a,b). The good correlation between selectivity and ΔAD makes ΔAD a valuable indicator for selectivity. This means COFs should be designed with a large ΔAD to improve the separation selectivity. Two broad routines were suggested by Tong et al. to boost ΔAD for adsorbent COFs:

An important discovery of Tong et al. (2014) was the crucial role that the staggered stacking of COF layers plays in increasing the adsorption of CO₂ in 2D COFs. As can be observed from the snapshots shown in Figure 12, CO₂ is adsorbed in the pores of eclipsed CTF-2 at random positions. Meanwhile, in staggered CTF-2, the adsorbed CO₂ molecules reside at the slit-shaped pores between the staggered layers and most of the CO₂ molecules are in the center of these pores. While located at the center of these slit-shaped pores the CO₂ molecules are interacting with two COF layers. Tong et al. note that in this simulation more CO₂ is adsorbed on the staggered CTF-2 then the eclipsed CTF-2.

Interlayer slipping between the π-π stacked layers of 2D COFs is a controllable feature that is shown to be exploitable when optimizing a COF material for the increased adsorption of gasses (Uribe-Romo et al., 2011; Feng et al., 2012b; Kandambeth et al., 2012; Biswal et al., 2013; Chandra et al., 2013; Fang et al., 2014). Sharma et al. (2017) studied the effect of interlayer slipping on CO₂ adsorption and CO₂/N₂ separation in microporous TpPa1 and mesoporous TpBD and PI-COFs using quantum mechanical and grand canonical Monte Carlo simulations. In this study, Babarao et al. modeled CO₂ bonding sites in the slipped COF structures focusing on bilayer slipping only. Three COFs were simulated in this work: TpPa1 (Kandambeth et al., 2012), TpBD (Biswal et al., 2013), and PI-COFs (Fang et al., 2014). When modeling the slipped structures of TpPa1 and TpBD COFs calculations were simplified by only considering two COF layer supercells with shifting only in the 90 and 60° directions, as shown in Figure 13. Sharma et al. (2017) acknowledge that slipping in other directions are a possibility but base their models on the classification scheme of Koo et al. (2012). Ab initio DFT was carried out in Vienna Ab Initio Simulation Package (VASP) to model the COF structures with at various modes of slipping (Kresse and Furthmüller, 1996). Sharma et al. also calculated binding energy of CO₂ by modeling the lone gas molecule in a 2 × 2 × 7 supercell. The position of the CO₂ molecule in the supercell was determined by simulated annealing. CO₂-CO₂ interactions and CO₂-COF interactions were simulated by Sharma et al. with the Lennard–Jones (LJ) potential model with Coulombic interactions. Using the RASPA code to implement Monte Carlo (MC) simulation in the grand canonical ensemble, Babarao et al. calculated the single component adsorption isotherms of CO₂ and N₂ (Dubbeldam et al., 2013). To obtain the mixture (CO₂:N₂ in a 15:85 ratio) adsorption selectivity, Sharma et al. (2017) performed IAST calculations (Myers and Prausnitz, 1965). Sharma et al. found that changes in the alignment of COF layers during slipping changes the CO₂ binding sites.

Sharma et al. (2017) found that CO₂ molecules in both AA and slipped COFs show an inclination to absorb close to oxygen atoms between the linkers because the oxygen atoms are more negatively charged than nitrogen and more prevalent in the pores. The CO₂ molecules absorbed in the AA will bind at a site between two connector oxygens from two different COF layers, being the same distance from both oxygens, this has been noted in other studies (Dinga and Wang, 2013; Xu et al., 2017). In the slipped COF structures CO₂ binds adjoining to a single COF layer and is interacting with three oxygen atoms and with a smaller distance between layers compared to the AA. These two aforementioned factors increase the binding energy to −44.62 kJ mol⁻¹ (E_vdw ~ −23.86 kJ mol⁻¹) for 90-AA1 TpPa1 COFs and −43.57 kJ mol⁻¹ (E_vdw~ −23.71 kJ mol⁻¹) for 60-AA1 TpPa1 COFs. An important point emphasized by Sharma et al. (2017) is that all oxygen atoms in the AA structure are degenerate. This is not true for the 90-AA1 slipped structure because there is a loss of symmetry. This results in the AA structure having six binding sites per unit cell while the 90-AA1 slipped structure has only one. So even though the slipped structure has higher binding energy the CO₂ uptake is reduced. The 60-AB structure of TpPa1 had the greatest CO₂ uptake capacity followed by the 90-AA1 structure of the PI-COF-2. Sharma et al. conclude that premeditated design of a slipped COF with exceptional CO₂ uptake ability hinges on various factors, such as the chemical structure of building blocks, the energetic interaction between neighboring COF layers and the synthesis protocol.

Computation studies aiming to determine the optimal makeup of CO₂ binding functional groups on the channel walls of COF materials have also been undertaken (Dash, 2018; Wang et al., 2018). Work by Lu et al. used grand canonical Monte Carlo (GCMC) to investigate the CO₂ adsorption and separation behaviors in four alkyl anime functionalized triphenylamine-based COFs (TPA-xC-NH₂, x = 1–4) (Wang et al., 2018). It is reasoned that the CO₂ adsorption capacity at 273 K is greater than CO₂ adsorption capacity at 298 K due to the exothermic process (Zhou et al., 2017). The reduction of CO₂ adsorption capacity when−1C-NH₂ is introduced is rationalized by the loss of surface area in TPA-1C-NH₂ in relation to the unmodified TPA-COF. This trend does not continue through as the length of the alkylamine chain increases; as evident by the high CO₂ adsorption capacities of TPA-4C-NH₂ which had the longest alkylamine chain. TPA-4C-NH₂ had a CO₂ adsorption capacity of 3.41 mmol/g at 298 K and 5.26 mmol/g at 273 K. This represents a 50.2% increase in CO₂ adsorption capacity at 298 K and a 30.5% increase at 273 K. TPA-4C-NH₂ adsorbs 1.58 mmol/g CO₂ at 0.15 bad and 298 K making it viable for CO₂ capture at power plants (Hu et al., 2016).

Lu et al. also investigate the selectivity of CO₂ over N₂ by TPA-COFs. Lu et al. used S= qCO2/qN2pCO2/pN2 to define selectivity where q_CO2 is the adsorption capacity of CO₂ and q_N2 is the adsorption capacity of N₂ (Wang et al., 2018). The bulk phase ratio of CO₂ and N₂ (p_CO2:p_N2) was set to 15:85. Since N₂ has very little polarity the addition of longer alkyl amine chains only reduced N₂ uptake as internal surface area decreases. This explains why the ability of the TPA-COFs to adsorb CO₂ is greatest for the pristine TPA-COF and decreases as the length of the alkylamine chain increases. Since CO₂ affinity increases as the alkylamine chain increases, by increasing the length of the alkylamine chains so does the selectivity of CO₂/N₂. This is affirmed by the analysis of the CO₂ selectivity wherein the TPA-COF with−4C-NH₂ shows a CO₂/N₂ selectivity of 11.7 at 298 K and 1 bar, representing a 121% increase in selectivity from the unmodified TPA-COF.

The study of CO₂ adsorption into COFs synthesized with reticular N containing molecules was studied through the computational work of Bibek Dash; in this study, the CO₂ adsorption capabilities of COFs with N-containing linkers (Dash, 2018) were calculated. Bibek Dash used the Gaussian 09 package to perform the DFT calculations. CO₂, N₂ molecules on the COF models were optimized with all atoms fully relaxed. The binding affinity of CO₂ to the azo group was compared to the affinity for N₂ to the azo group using an azobenzene model. Electrostatic interactions between the N of the ring and the CO₂ molecule were designated as dipole-quadrupole and dipole-induced dipole interactions (Vogiatzis et al., 2009; Altarawneh et al., 2014).

The synthesis of COF-505 consisting of helical, intertwined threads to make a weaving crystal structure by Liu et al. was an advancement in the COF chemistry (Liu et al., 2016). COF-505 consists of imine linked threads are woven around copper ions; the flexible threads allow for many degrees of freedom so that the structure can stretch and bend without collapsing the general framework. The copper ions can also be removed and subsequently replaced with no loss of crystallinity. Chen et al. performed computational studies to examine the adsorption of CO₂ onto COF-505 as well as the potential of COF-505 as a photocatalyst (Chen et al., 2019). Chen et al. used Grimme's DFT-D3 dispersion correction scheme with the Becke–Johnson damping function to perform the dispersion correction to DFT calculations (Grimme, 2011; Grimme et al., 2011). Chen et al. determined that the CO₂ molecule could be adsorbed onto both the metal ions and the organic linkers in COF-505. The adsorption energy of CO₂ onto COF-505 was found to be −0.16 eV for Cu-COF-505 and −0.18 for Pd-COF-505.

The first successful use of a COF for the reduction of CO₂ to CO was reported by Lin et al. (2015). In this work the researchers synthesized two imine-linked COFs consisting of a cobalt porphyrin, 5,10, 15,20-tetrakis(4-aminophenyl)porphinato] cobalt [Co(TAP)] linked with bidentate aldehyde linkers. The two reticular COFs, COF-366-Co and COF-367-Co differed by the length of the aldehyde linker with the first using 1,4-benzenedicarboxaldehyde (BDA) and then using a 4,4'-Biphenyldicarbaldehyde (BPDA). The COFs were deposited on a conductive and porous carbon fiber for use in electrochemical experiments. Electrolysis of aqueous CO₂ saturated solutions with a sodium bicarbonate buffer with applied potentials between −0.57 and −0.97 V produced carbon monoxide.

Lin et al. (2015) hypothesized that due to the low solubility of CO₂ in H₂O that not all the electroactive cobalt porphyrin sites would fully participate in the electrocatalysis. The researchers elected to separate the active Co sites with catalytically inactive copper porphyrin struts which increased the turn over frequency per cobalt atom. The copper porphyrin was selected by screening the surface area of various COF-367-M (M = metal) compounds and selecting the metal porphyrin which provided the highest surface area. The new bimetallic COFs were denoted COF-366-Co(10%) and COF-366-Co(1%) to denote the percent of Co amongst metals. The bimetallic COFs had a higher TON than the cobalt COFs, as can be seen in Figure 14.

Work on Co-porphyrin containing COFs for the electrochemical reduction of CO₂ to CO was expanded into the realm of thin films by Diercks et al. (2018a). In this work the Co porphyrins were connected by imine linkages to struts modified with different electron withdrawing groups to synthesize 4 new COFs COF-366-Co, COF-366(OMe)₂-Co, COF-366-F-Co and COF-366-(F)₄-Co as shown in Figure 15. Yaghi et al. sought to determine the which functional groups provided the optimal electronic effect for the reduction of CO₂ to CO. To ensure that performance was a result of only the electrons the adsorption and Q_st for all the COFs were measured by Diercks et al. The CO₂ adsorption at 295 K was similar for all the four COFs (23.5–27.4 cm³ g⁻¹) and the Q_st of all COFs was between 24.6 and 24.1 kJ mol⁻¹. Cyclic voltammograms and X-ray absorbance measurements made clear the differences in electrochemical properties of the COFs as highlighted in Figure 16A.

For electrocatalysis, the COF was deposited onto a conductive electrode. Since it has been shown that deposition of a COF onto an electrode as a microcrystalline powder results in minimal electrochemical accessibility (DeBlase et al., 2018) the researches in this study elected to deposit COFs onto the surface of highly ordered pyrolytic graphite (HOPG) so as to increase accessibility of the cobalt centers. These oriented thin films outperformed the traditionally deposited COFs and bulk samples. Electrocatalysis took place in a 0.5 M sodium bicarbonate solution at pH 7.2 and an overpotential of −0.67 V (vs. RHE). Overall higher electron withdrawing character improves the catalytic activity of the COFs. As depicted in Figure 16B COF-366-F-Co has a current density for CO formation of 65 mA mg⁻¹ which is a 20 mA mg⁻¹ increase over COF-366-Co. COF-366-(F)₄-Co was an outlier as it had the lowest current density for CO formation but the was the second highest electron withdrawing character. Diercks et al. rationalize this observation by considering the hydrophobicity of the COF-366-(F)₄-Co framework which hinders electrolyte availability at the active site (Chen et al., 2013; Deria et al., 2013).

More recent work on cobalt containg COFs for the production of CO consists of three new COFs, Co-Pc-PBBA, Co-PcbPBBA, and Co-Pc-tPBBA; analyzed by Yao et al. (2018). In this computational work the cobalt atoms were held in the center of phthalocyanine linkers. Yao et al. used computational methods to first determine the CO₂ storage capacity of the COFs then identified the COFs as promising candidates for the electrochemical reduction of CO₂ to CO. The three reticular COFs are linked by boronic acid linkers between the phthalocyanine moieties with the difference being the length of the boronic acid linkers. Through this analysis, the effect of pore size could be investigated. Yao et al. first found that CO₂ molecules will be distributed at the corner sites of the COFs at normal pressure. Changing the pore size has little effect on this behavior because the 4 corner sites are equal in all three reticular COFs. This analysis, illustrated in Figure 17, describes the varying performance of the COFs at CO reduction at various pressures. At lower pressures, the COFs with smaller pore size are more economical. COFs with larger pore size can adsorb more CO₂ molecules on the surface of the walls at higher pressures while the COFs with smaller pores would begin to hold CO₂ in the empty center of their pores. This computational observation dictates that the large-pored COFs would be more efficient at higher pressures.

Yao et al. (2018) selected Co-Pc-bPBBA for their computational study. The reduction pathway from CO₂ to CO and various other hydrocarbons as elucidated by Yao et al. is depicted in Figure 18 with free energies of the reaction pathways are shown below in Table 2. CO₂ reduction begins by the evolution of a carboxyhydroxyl ^*COOH intermediate (Here the ^* signifies the adsorbed state); the energy of this ^*COOH formation is 0.30 eV on Co-Pc-PBBA, this is less than the what has been determined for the Co-porphine (0.46 eV) used in previous works. The ^*COOH then undergoes electron discharge and proton transfer; followed by loss of the hydroxyl group to form ^*CO. Yao et al. calculated the adsorption energy (E_ad) of CO on Co-Pc-PBBA to be −0.36 eV, which is < -0.26 eV on Co-porphine.

In 2018 Yang et al. reported the development of a 2D COF photocatalyst with an incorporated Re complex for the photocatalytic reduction of CO₂ to CO (Yang et al., 2018). To synthesize the so-named Re-COF, Huang et al. first mixed 2,2-bipyridyl-5,5-dialdehyde (BPDA) and 4,4′,4″-(1,3,5-triazine-2,4,6-triyl) trianiline (TTA) to create the COF scaffold by solvothermal reactions. Upon collection of the COF scaffold, the Re moieties were added by the reaction between bipyridine ligand and Re(CO)₃Cl to form Re-COF. The diffuse reflectance UV-vis of Re-COF shows an adsorption band at ~440 nm which is a product of the delocalized intramolecular charge transfer (ICT) band created by the π-conjugation of TTA and BPDA (Pop et al., 2013; Cho et al., 2016; Qian et al., 2017).

The collected gas phase of Re-COF photocatalysis was only 2% H₂ and 98% CO showing a 98% selectivity to CO production. The conversion of CO₂ to CO by the Re-COF has a TON 48 times greater than homogenous reaction with a Re(bpy)(CO)₃Cl catalyst. The mechanism of the CO₂ reduction by Re-COF was elucidated by Huang et al. through analysis of diffuse reflectance UV-visible spectra taken during catalysis. The proposed mechanism is shown in Figure 19.

The development of the reduction of CO₂ by COFs through the photocatalytic route was also pursued by Fu et al. in their development of the two azine based COFs for the reduction of CO₂ to methanol (Fu et al., 2018).

In 2018 work by Liu et al. (2018) used a COF framework to enhance the specificity of silver metal catalyzed reduction of CO₂ to CO through the use of COF thin film. Enhancing the selectivity of electrochemical reductions of CO₂ by silver is necessary due to the lower potential of the hydrogen evolution reaction (HER). Liu et al. constructed COF-300-AR; a highly stable amine linked COF made by reduction of its imine equivalent (COF-300). The reduction of extended network imine solids to their amine equivalents must overcome a limited diffusion of the reactants and the rigidity of the extended network (Swamy et al., 2010; Liu et al., 2014; Gui et al., 2015). To optimize the reduction of COF-300, Liu et al. studied the reduction of benzylidene aniline to benzyl aniline with NaBH₄. Having fine-tuned this procedure Deng et al. quantitively reduced the 3D framework of COF-300 into its amine analog maintaining the crystallinity of the COF. Deposition of COF-300-AR onto the surface of silver improved the selectivity of electrochemical reduction of CO₂ to CO from 13 to 53% and from 43 to 80% at −0.70 V and −0.85 V, respectively when compared to just the silver electrode. Addition of the COF thin film also inhibited the HER reaction from 80 to 22% at −0.70 V and 60 to 9% at −0.85V. Concomitantly, silver electrodes coated with a Nafion binder showed no improvement in CO evolution and glassy carbon coated with CO-300-AR showed no CO evolution either. Liu et al. deduced that the increased selectivity for CO production over HER is a result of carbamate formation at the COF-metal interface. The formation of this carbamate intermediate was confirmed through FTIR. Liu et al. summarized the mechanism of the CO reduction into three steps. First, amine functional groups close the surface of the silver electrode attracted CO₂ molecules. The second step is the formation of the carbamate close to the surface of the Ag electrode. Close proximity to the silver electrode allows for electron transfer to bypass the COF backbone. In the third step, the carbamate is reduced to CO and OH anions, regenerating the amine moieties of the COF. The third step is facilitated by the aqueous reaction environment.

A comprehensive study of photocatalytic CO₂ reduction at the single Ni sites of a COF was recently undertaken by Zhong et al. (2019). The synthesized material, so named Ni-TpBpy is a ketoenamine linked COF resultant from the solvothermal condensation of 1,3,5-triformylphloroglucinol and 5,5′-diamino-2,2′-bipyridin. The resultant COF with honeycomb channels was subsequently treated with Ni(ClO₄)₂ to make the metalated structure. BET analysis of the COF shows that post-metalation the material retains enough porosity to effectively admit CO₂ gas into the reactive channel sites, although the pore volume as determined by the N₂ adsorption/desorption measurements decreases from 0.6 cm⁻¹/g prior to metalation to 0.4 cm⁻¹/g. Despite this decrease in pore size Zou et al. discovered that the metalated Ni-TpBpy COF has a higher CO₂ adsorption and isosteric heat of adsorption than its unmetallized counterpart. The authors ascribe this observation to the increased Lewis acid-base interactions between the CO₂ and Ni sites in the COF. The Ni-TpBpy photocatalyst had an exceptional selectivity toward CO production over H₂ or other CO₂ reduction products with a 96% CO selectivity over H₂.

The first report of a COF catalyst for the synthesis of cyclic carbonates by CO₂ fixation was the 2012 work by Thomas and coworkers in which the researches synthesized three metal-free covalent triazine framework (CTF) cataysts (Roeser et al., 2012). CTFs are a subset of COFs synthesized by the trimerization of dicyanocompounds often at temperatures ≥400 C and with melted zinc chloride solvent (Kuhn et al., 2008; Bojdys et al., 2010). The CTFs prepared by Thomas and coworkers for their 2012 study were crystalline framework (CTF-1), fully amorphous analog of CTF-1, denoted as CTF-1-HSA, and fully amorphous framework (CTF-P-HSA). The synthesis of CTF-1 and the amorphous CTF-1-HSA differ in reaction temperature with CTF-1 being synthesized from 1,4-dicyanobenzene at 400°C and CTF-1-HSA being synthesized from 1,4-dicyanobenzene at 600°C. CTF-P-HSA was synthesized from heating of 2,6-dicyanopyrideine in ZnCl₂ at 600°C. Zinc chloride can potentially be catalytic itself so was removed by extensive washing with acidic water.

Thomas and coworkers confirmed the framework with more amorphous structure CTF-P-HSA to be the best choice for the cycloaddition of CO₂ to epoxides. The CTFs studied by Thomas and coworkers are highly active under mild conditions and had short reaction times even in the absence of solvent or co-catalyst. The authors explained the high catalytic activity of CTF-P-HSA was due to its amorphous structure resulting in a high surface area, hierarchical porosity, and additional mesopores which facilitated the diffusions to active sites when compared to crystalline frameworks with micropores. Different epoxides including propylene oxide (PO), styrene oxide (SO), cyclohexene oxide (CHO), and epichlorohydrin (CIPO) were used for the cycloaddition of CO₂ over the CTF-P-HSA to form cyclic carbonates. The catalysis was carried out at 130°C at times of 4, 8, and 12 h at 6.9 bar of CO₂ with high yields (Table 6). In the case of cycloaddition of CO₂ over the CTF-P-HSA, it was found that epichlorohydrin is more active than propylene oxide, styrene oxide, and cyclohexene oxide. CTF-P-HSA was recycled 6 times with no loss of activity. Thomas and coworkers also studied the catalytic formation of linear carbonates. Due to thermodynamic limitations, the formation of linear carbonates from CO₂ remained as a major issue. One of the synthetic routes starting from CO₂ was the one-pot synthesis of dimethyl carbonate from an epoxide, methanol, and CO₂. In that route, the cycloaddition to epoxide as well as the transesterification of a cyclic carbonate with methanol were performed in one single batch reaction. The researchers of this work used CTF-P-HSA catalyst for the direct formation of DMC. Between all the epoxides studies by Thomas and coworkers the maximum obtained selectivity of 17.5% was from the propylene oxide reaction under the 20 bar and 160°C.

In 2017 an interesting development was achieved by Buyukcakir et al. in the development of a series of cCTF materials, cCTF-400, CTF-450, and cCTF-500 which consisted of a cationic framework with a Cl⁻ counterion (Buyukcakir et al., 2017). The presence of this anion in the framework allows for the synthesis of cyclic carbonates via fixation of CO₂ to an epoxide without the addition of ionic cocatalyst (Table 3 entry 2). cCTF-500 showed an excellent CO₂ uptake at low pressure of 133 mg g⁻¹ at 273 K and 80 mg g⁻¹ at 298 K at 1 bar. This makes the charged cCTF-500 framework on par with other CTF's for CO₂ adsorption. Synthesis of the charged covalent triazine frameworks (cCTFs) differed only in the temperature of the reaction. Each cCTF was synthesized by placing 1,1′-bis(4-cyanophenyl0-[4,4′-bipyridine]-1,1′-diium dichloride into a sealed glass ampule with ZnCl₂ and heating for 48 h at 400, 450, and 500 C to afford the corresponding the cCTF.

Progress in the development of CTFs for CO₂ fixation continued to 2018 when Yu et al. devised a bottom-up approach entailing deliberate linker selection for the synthesis of function-specific CTFs capable of cycloaddition of CO₂ to epoxides (Yu et al., 2018). Using precisely selected polymerization conditions, porous crystalline frameworks CTF-CSU1 and CTF-CSU19 with long-range structure were attained. To synthesize the CTFs, Yu et al. heated 3,6-dicyano-9H-carbazole and 3,6-dicyano-9-methylcarbazole in ZnCl₂ at 400°C for 48 h with the former producing CTF-CSU1 in 93% yield and the latter producing CTF-CSU19 in 91% yield. These CTFs have an exceptionally high nitrogen content due to the presence of both carbazole and triazine moieties. The high nitrogen content of the of extended polymers is known to be beneficial for the capture and catalytic chemical fixation of CO₂ (Talapaneni et al., 2015). Their results showed that in the CO₂ cycloaddition, CTF-CSU19 had higher contribution under atmospheric pressure and room temperature (0.1 MPa, 25°C, 48 h) with a TBAB co-catalyst. The elevation in catalytic efficiency was prescribed to the large surface area, pore volume and high nitrogen content. The cycloaddition reaction of CO₂ and epoxides having small-sized substituents can be catalyzed using CTF-CSU19 under mild conditions. In addition, CTF-CSU19 can catalyze the fixation of CO₂ to epichlorohydrin in high yields although the reacitivty of the larger styrene oxide substrate was more mild (Table 3 entry 3). On the other hand, substrates possessing large substituents like phenyl are relatively slow to react because they cannot easily diffuse into the channels for reactions. The scope of the molecules synthesized by Yu et al. in this study is presented in Table 4. For reusability test, the reaction of epibromohydrin with CO₂ catalyzed by CTF-CSUs/TBAB was studied and it was showed that both CTFs can be used for 5 times with only a slight decrease in activity.

In 2016, Saptal et al. synthesized crystalline catechol porphyrin COFs (2,3-DhaTph and 2,3-DmaTph) as a catalyst for the chemical CO₂ fixation to epoxides for the synthesis of oxazolidinones and cyclic carbonates (Saptal et al., 2016) 2,3-DmaTph showed remarkable thermal stability and was able to effect CO₂ fixation at 100 C (Table 3 entry 4). The molecular structures of COFs 2,3-DhaTph and 2,3-DmaTph are both depicted in Figure 20. The formation of cyclic carbonate under transition metal free and solvent-free conditions was notable.

During optimization, Saptal et al. began with the cycloaddition reaction of styrene oxide (SO) in order to synthesize styrene carbonate (SC) as a model reaction. Saptal et al. studied both polar and non-polar solvents and concluded that optimized catalysis could proceed without solvent. Also, Saptal et al. have studied different co-catalysts such as TBAB, TBAI, KI, and NaI, which serve to open the nucleophilic ring of epoxide. Their experiments show that using TBAI as co-catalyst results in a high yield of 94%. Sapatal and coworkers also reported the higher activity of 2,3-DhaTph, the catalyst with hydrogen bond donor capability, comparing to 2,3-DmaTph at 0.1MPa. Furthermore, Saptal et al. investigated the effect of pressure on the yield of the process and found that at high pressure (1MPa), styrene carbonate (SC) concentration was increased. Also, using co-catalyst (TBAI) was very important as a negligible amount of was SC obtained when no co-catalyst was used. Sapatal and coworkers have also found that high surface area of 2,3-DhaTph COF for cycloaddition reaction resulted in a high yield of the process.

Due to the presence of many active sites as well as porphyrin groups and well-oriented hydroxyl functional groups together, 2,3-DhaTph COF showed higher activity compared to 2,3-DmaTph COF for the chemical carbon dioxide fixation. Sapatal et al. state that this feature promoted interactions between active sites of the COF and incoming substrates to effectively transform the product. In the case of catalyst loading, temperature, and time Saptal and coworkers found that the optimum amount of catalyst for the highest conversion of SC is 0.02 mmol, the optimum temperature is 110°C, and optimum time is 12 h. Saptal et al. have also found that the epoxides with aromatic substituents worked better for possessing substituents such as F, Cl, and Br compared to the aliphatic epoxides. The molecules synthesized by Saptal et al. are laid out in Table 5.

Using 2,3-DhaTph COF to describe the mechanism, Saptal et al. note that vicinal alcohol groups of the catechol linker increase polarization of the C-O bond through hydrogen bonding. The polar C-O bond is then attacked by the anionic halogen of the co-catalyst facilitating cycloaddition as seen in Figure 21. As porous heterogeneous catalysts can be reused and recycled, cycloaddition reaction of SO to SC was used to investigate the recyclability of the COF catalyst. Sapatal et al. reported that the 2,3-DhaTph COF catalyst could be effectively reused with high catalytic activity.

In 2018 Zhi et al. identified two more COFs, so named COF-JLU6 and COF-JLU7; for the efficient synthesis of cyclic carbonates from CO₂ (Zhi et al., 2018). Rather than using a triazine linkage to construct the COF framework Zhi et al. used a triazine-based anilines linked with 2,5-dihydroxyterephaladehyde through imine bonds formed during solvothermal synthesis as depicted in Figure 22. In their study, Zhi et al. used the synthesized imine-linked COFs with hydroxy groups as heterogeneous catalyst for cycloaddition of CO₂ into cyclic carbonates under mild conditions. Zhi reported the average pore width of 3.1 and 3.3 nm for COF-JLU6 and COF-JLU7, respectively. The cycloaddition of CO₂ with epichlorohydrin and tetra-n-butylammonium bromide (TBAB) as co-catalyst was used as a model reaction and resulted in the yield of 86% and 92% for COF-JLU6 and COF-JLU7, respectively, at 40°C and 0.1MPa CO₂ pressure. The higher catalytic activity of COF-JLU7 was attributed to its greater CO₂ adsorption capacity. The scope of cyclic carbonates synthesized by Zhi et al. is shown in Table 6.

For the recyclability study, Zhi et al. used epichlorohydrin as the substrate and reported that COF-JLU7 could be recycled for five repeating cycles with high catalytic activity and selectivity. The proposed mechanism for CO₂ fixation by COF-JLU6 and COF-JLU7 can be seen in Figure 23 and is initiated by hydrogen bonding between the hydroxy groups on the pore walls of the COFs and the C-O bond in the epoxide accompanied by the Br⁻ attacking the epoxide carbon at the less sterically hindered side. This is followed by the formation of an oxyanion stabilized an adjacent hydroxy group from a COF layer. The oxyanion nucleophile then attacks CO₂ to form an alkylcarbonate anion which creates the cyclic carbonate upon ring closing while the COF is regenerated.

In 2017, Xu et al. synthesized and employed two dimensional (2D) metalloporphyrin-based covalent organic framework (COF) composites [COF-366-M and COF-SQ-M (M = Zn and Co)] in order to form cyclic carbonates through catalyzing the coupling of CO₂ and epoxides (Xu et al., 2017). Under mild conditions, COF-366-Zn was successfully employed as a co-catalyst for the synthesis of 1,2-butylene carbonate and 98.9% selectivity as well as 99.2% activity were obtained. The most pressing advantage of the heterogenous COF catalyst is the comparative ease with which the COF can be recycled when compared to the metalloporphyrin which is soluble in many cyclic carbonates. Xu et al. stated that the metalloporphyrin sites activation improvement was mainly because of the positive role of the microenvironment inside COF-366 frameworks. This could lead to an excellent catalytic performance. Due to their unique 2D COF structure, COF-366-Ms showed high adaptability for various epoxide substrates and solvent choice.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.