Extensive Screening of Solvent‐Linked Porous Polymers through Friedel–Crafts Reaction for Gas Adsorption

Porous solids with high surface areas, thermal stability, and chemically robust nature have attracted great attention because of their applications in gas capture and storage, water treatment, and catalysis. In addition to the commonly used porous materials in the industry such as zeolites, activated carbons, and molecular sieves, recent developments show that various classes of new porous solids have emerged, such as metal organic frameworks (MOFs), covalent organic frameworks (COFs), covalent organic polymers (COPs), hypercross-linked polymers (HCPs), microporous organic polymers (MOPs), porous aromatic frameworks (PAFs), porous polymer networks (PPNs), and benzimidazole-linked polymers (BILPs). To be used in industrial scale, these materials are expected to be highly porous, thermally and chemically robust, and easily synthesized from inexpensive precursors. Although numerous materials have been reported so far, most have drawbacks such as difficult synthetic procedures, expensive and complex starting monomers, or require rare earth metals. So far, only a handful of these porous materials have proven relatively scalable with affordable synthesis. Considering the availability of the precursors and reagents, Friedel–Crafts (FC) alkylation is one of the most promising

synthetic methodologies for making highly porous networks with stable C─C-bonded frameworks. Since the early works by Davankov et al. [14] to make porous resins, the method was widely adapted to offer HCPs and MOPs. The applications vary from simple gas capture and storage [12d] to catalysis. [15] In FC alkylation reactions, inexpensive Lewis acids such as aluminum chloride (AlCl 3 ) and iron chloride (FeCl 3 ) and chlorinated solvents such as 1,2-dichloroethane (DCE), dichloromethane (DCM), and chloroform (CHCl 3 ) are commonly used. Although FeCl 3 works fine, using highly reactive AlCl 3 with chlorinated solvents must be controversial as there is a great chance of participation in the FC reaction. [16] Despite these reactivity concerns, several reports used AlCl 3 as a Lewis acid and chlorinated solvents as a solvent. [17] Previously, we showed that chlorinated solvents not only readily participate in FC reactions, but they can produce highly porous network polymers with aromatic monomers in the presence of AlCl 3 . [18] In principle, common chlorinated solvents serve as both linkers and solvent. Surprisingly, excess linkers produced network structures, an observation that can only be explained by the formation of highly reactive intermediates. Here, we extend the scope of this chemistry and studied three classes of 18 monomers under the solvent-mediated FC conditions. We then analyzed their gas adsorption properties in great detail. The porous networks show high uptake for carbon dioxide (CO 2 ), hydrogen (H 2 ), and methane (CH 4 ) gases. The method developed here is one of the most affordable FC reactions so far and one of the simplest for the formation of highly porous polymers. Because the reaction takes place rapidly at ambient conditions and uses inexpensive precursors, they will ultimately become the most promising candidates for the gas capture and separation at industrial scales.

Synthesis
The alkyl-chlorinated solvents and highly reactive AlCl 3 have readily been used in a number of reports that feature FC reactions or Scholl-type aromatic couplings. [17] We suspected that the solvent may not stay inert and will react with aromatics. Hence, benzene and anhydrous AlCl 3 were reacted at room temperature in 1,2-dichloroethane (DCE), dichloromethane (DCM), and chloroform (CHCl 3 ) solvents. The precipitates formed in all three solvents and they were insoluble in both water and organic solvents. Getting insoluble products was surprising, as chlorinated solvents were used in excess and the products were expected to be multisubstituted benzenes that are readily soluble in organic solvents. However, the reactions yielded high-molecular-weight network polymers rather than soluble molecules, an observation that can only be explained by the formation of highly reactive intermediates ( Figure 1a). In situ-formed intermediates are more reactive than starting monomers and chlorinated solvents, hence self-polymerize to give high-molecular-weight polymers rather than multisubstituted aromatics ( Figure S1, Supporting Information). This finding is contrary to some of the reports in the literature, particularly those that claim Scholl's reaction (aromatic-aromatic self-coupling) in chlorinated solvents in the presence of AlCl 3 ( Figure S2, Supporting Information). [17a,cÀe,g] (a) (b) (c)  To further confirm the reactivity of solvents or aromatic coupling, we conducted model reactions of DCM, CHCl 3 , and DCE in excess benzene (solvent) and in the presence of anhydrous AlCl 3 (Figure 1b). Despite using excess aromatic benzene, no aromatic coupling (e.g., biphenyl, terphenyl) was observed, but FC products, diphenyl methane, triphenyl methane, and 1,2-diphenylethane, were formed (see Figure S5-S10, Supporting Information, for detailed information). In principle, here, alkyl-chlorinated solvents serve as both linkers and solvent. DCM and DCE act as bidentate linkers to give diphenylmethane-and 1,2-diphenylethane-based polymers, whereas chloroform acts as tridentate linker to give triphenylmethane-based polymer (Figure 1c). Although FC reactions are established to run over [carbocation] þ [MetalCl x ] À as active species, Lewis acids such as FeCl 3 were recently reported to produce FeCl 2 þ in a disproportionation mechanism. [19] When studying this possibility, our trials with FeCl 3 in chlorinated solvents failed to produce any insoluble network polymers. At the same time, AlCl 3 with stoichiometric chlorinated reagents in nonchlorinated solvents like DMF also did not yield any network polymers, showing that disproportionation paths do not affect the outcome of the network formations in this work.

Reactivity of Aromatic Monomers and Solvents
We, then, extended this discovery into a range of total 18 different aromatic hydrocarbons in DCM, CHCl 3 , and DCE solvents, proving that a wide substrate scope is indeed possible ( Table 1). Among three solvents, CHCl 3 formed insoluble polymers with 13 monomers, whereas DCE and DCM formed 9 and 7 polymers, respectively. This can be explained by, due to the electron-withdrawing chlorines, the higher reactivity of CHCl 3 and its benzal chloride (benzyl dichloride)-based intermediates. While DCM and benzyl chloride intermediates are more reactive than DCE and (2-chloroethyl)benzene, the longer and flexible DCE can avoid steric hindrance and react with more monomers than DCM.
Benzene formed network structures with all three solvents. We believe that inefficient packing and six possible extension positions contribute to this variation. In contrast, methyl-substituted benzenes were not as successful. The mesitylene (1,3,5-trimethylbenzene), for example, did not give polymer with any of the solvents, and xylene (mixture of isomers) formed a network only with CHCl 3 . Steric hindrance by methyl groups proves to be a detrimental factor in the formation of a crosslinked structure. For instance, toluene did not yield a polymer in DCM, but longer DCE, where steric hindrance could be avoided, produced an insoluble polymer.
Most heteroatom-containing aromatics are known to be less likely to favor an FC reaction over the aromatic ring. [20] The heteroatom usually chelates with the Lewis acid and the electron-poor adduct deactivates the aromatic rings ( Figure S3, Supporting Information). Therefore, pyridine, phenol, aniline, and triphenylphosphine did not give polymer with any of the solvents. Diphenyl ether, diphenyl amine, and triphenylamine formed polymers with highly reactive CHCl 3 . It is interesting to note that triphenyl amine formed polymer, whereas triphenylphosphine did not. We think that is because of the planar (119.6 ) geometry of triphenylamine, leading to a weaker interaction between nitrogen and AlCl 3 , and the higher reactivity of CHCl 3 compared with other solvents. On the contrary, this is not true for the triphenylphosphine (TPP) because of its trigonal pyramidal (102.8 ) layout, which causes unhindered, strong interaction with Lewis acids ( Figure S3, Supporting Information). Hence, no network forms from TPP and halo-organics.
Polycyclic aromatics yielded network structures with all tested solvents. This is in accordance with most of the reported literature; whether it is crystalline MOFs and COFs, or amorphous porous polymers, polycyclic aromatics monomers have been widely used in the formation of porous structures.

Characterization
The COPs were characterized by Fourier-transform infrared spectroscopy (FTIR), nuclear magnetic resonance (NMR) spectroscopy, elemental composition analysis, scanning electron microscope (SEM), and thermogravimetric analysis (TGA). The characteristic FTIR bands of aromatic CÀH bending at 600À900 cm À1 , aliphatic CÀH bending at 1000À1200 cm À1 , aromatic CÀC stretching at 1400À1600 cm À1 , aliphatic CÀH stretching below 3000 cm À1 , and aromatic CÀH stretching above 3000 cm À1 support the proposed structures ( Figure 2). When the spectra of three different solvents were compared, DCE-linked polymers exhibited the strongest aliphatic CÀH stretching vibrations, and DCM-linked polymers possess stronger aliphatic CÀH stretching vibrations than CHCl 3 -linked polymers, showing the characteristics of each linker. The elemental analysis (EA) of COPs was measured after repetitive Soxhlet washings and drying at 120 C. Apart from high content of carbon and hydrogen, EA also revealed oxygen ( Table 2). The oxygen was likely introduced through hydrolysis and oxidation of surface-tethered alkyl chlorides during postsynthetic steps ( Figure S4, Supporting Information). The CHCl 3 -linked COPs contain higher oxygen content (7À10 wt%) than DCM-linked COPs (2À3 wt%) and DCE-linked COPs (1À2 wt%), correlating with the reactivity of surface chlorides (benzal chloride, benzyl chloride, and (2-chloroethyl)benzene). Hydrolysis and oxidation products were also confirmed by characteristic broad OÀH stretching bands at 3300À3500 cm À1 and carbonyl C═O stretching at 1690 cm À1 in CHCl 3 -linked COPs. There is also a residual aluminum from the catalyst, which is found to be very low (confirmed by Inductively coupled plasma mass spectrometry (ICP-MS)). Solid-state 13 C (CP/MAS) NMR spectra of three COPs (one representative from each solvent) are shown in Figure 2c. The spectrum shows mainly two types of carbons, aromatic and aliphatic carbons. The peak at around 140 ppm belongs to substituted aromatic carbons and the one at 127 ppm can be assigned to unsubstituted aromatic carbons. Peaks at 34, 36, and 55 ppm can be attributed to aliphatic carbons     Figure S5-S10, Supporting Information, for NMR spectra). In CHCl 3 -linked COPs, there are oxidized carbon peaks at 82 ppm (C─O) and 190 ppm (C═O), further supporting the hydrolyzed and oxidized surface species. The TGA of the solvent-linked COPs shows thermal stabilities up to the 400 C, thanks to their stable C─C-bonded framework ( Figure S16-S20, Supporting Information). In CHCl 3 -linked COPs, due to their hydrophilic surface, there is a 4À10 wt% mass loss up to 120 C, coming from the desorption of adsorbed water. SEM surface images reveal that COP grains vary from the spherical to flat structure, and the size of spheres varies among COPs ( Figure S21-S27, Supporting Information).

Porosity Analysis
Porosity of COPs was analyzed from Ar adsorption and desorption isotherms at 87 K. The BET surface areas of these materials ranged from nonporous to as high as 1685 m 2 g À1 (Table 2). Among 29 COPs, ten of them, mostly from polycyclic aromatic-based structures, exhibited highly porous (>100 m 2 g À1 ) morphology and those isotherms are shown in Figure 3. Tetraphenyladamantane (TPA) consistently gave high porosities with all three solvents. We articulate on the fact that TPA has larger spacing between the phenyl legs of the tetrahedral monomer with rigid, locked, and a saturated hydrocarbon core. This observation is in line with our previous work on Tröger's base networks, where the specific geometry of TPA dictates a highly porous network. [21] Naphthalene and biphenyl also followed suit, producing networks with DCM and CHCl 3 . The main exception was at tetraphenylmethane and triphenylmethane mainly due to the intramolecular crosslinking possibilities between freely rotating benzene rings. Surprisingly, 1,3,5-triphenylbenzene led to two of the largest BET surface areas with DCM (COP-130, 1685 m 2 g À1 ) and with CHCl 3 (COP-140, 1219 m 2 g À1 ). In general, flat aromatic monomers tend to pack πÀπ stacking, a common mechanism in the formation of ordered COFs. [22] In FC conditions, one would expect this mechanism leading to lower porosity. Most of the DCE-linked COPs showed nonporous structures because of the freely rotating aliphatic dimethylene linkers.
In the isotherms of porous COPs, there is a rapid increase in the low-pressure range (P P o À1 < 0.01) and small hysteresis between adsorption and desorption, indicating both microporous and mesoporous morphology. [23] In particular, the DCM-linked COP-128 and COP-130 showed strong hysteresis. The hysteresis in adsorption and desorption isotherms is widely observed in porous organic polymers such as polymers of intrinsic microporosity (PIMs) [24] and polyimide porous polymers. [25] The origin of hysteresis is associated with swelling of materials, [25a,26] the restricted access to pores, [27] and mesoporous morphology. [28] The differential and cumulative pore size distributions (PSDs) were obtained using nonlocal density functional theory (NLDFT) model ( Figure 3). The PSD profile shows that COP-128 and COP-130 possess the highest mesoporous volumes of 51% and 52%, respectively. The CHCl 3 -linked COPs showed higher microporosity (up to 83%) compared with DCM and DCE (%50%). We believe that microporosity was formed as a result of higher crosslinking of three dentate CHCl 3 compared www.advancedsciencenews.com www.advenergysustres.com with two dentate linkers (DCM and DCE). In addition, tetraphenyladamantane-based COPs show rapid increase near saturation pressures because of their macroporosity (Figure 3 and Figure S29f, Supporting Information). We believe that macroporosity comes from the interparticle voids of small particles. [29] As tetraphenyladamantane has low solubility in chlorinated solvents, it led to smaller grains. This is also evident from SEM images ( Figure S21 and S27, Supporting Information).

The CO 2 Adsorption
Due to their high surface areas and low energy regeneration, solid porous adsorbents are being extensively investigated as potential CO 2 sorbents. Herein, we studied CO 2 adsorption properties of COPs at 273 and 298 K up to 1.1 bar and isotherms for porous COPs (>100 m 2 g À1 ) are shown in Figure 4a-d. As expected, CO 2 adsorption and desorption isotherms showed no hysteresis indicating physisorptive CO 2 binding. At 1.1 bar and 273 K, 1,3,5-triphenylbenzene-based COP-130 (DCM) and COP-140 (CHCl 3 ) exhibited the highest CO 2 uptake performances, with capacities of 4.71 and 4.35 mmol g À1 , respectively (Table 3). These COPs outperform other FC-based and hydrocarbon-based polymers (Table S2, Supporting Information) such as HCP-1 (2.16 mmol g À1 , 1970 m 2 g À1 ) [12d] and PAF-1 (2.05 mmol g À1 , 5600 m 2 g À1 ). [30] These values are even competing with some oxygen-/nitrogen-rich structures such as BILP-6 [11a] (4.79 mmol g À1 , 1261 m 2 g À1 ). Here, unlike other polymerization methods, the reaction of excess solvent linkers leads to richer surface chemistry with dangling species. Unreacted halogens may contribute to CO 2 uptake, [31] although lesser than the remnant oxygens that are introduced by their hydrolysis. We believe that the oxygenated surfaces of COPs produced here are responsible for their superior CO 2    (Figure 4e,f ). Indeed, as expected these numbers are higher than FC-based HCPs [12d] (20À24 kj mol À1 ) and aromatic PAFs [30] (15À20 kj mol À1 ), explaining the better CO 2 uptake performance of these COPs. When different solvent linkers were compared, CHCl 3 -linked COPs (30À37 kj mol À1 ) showed higher Q st values than DCM and DCE (27À34 kj mol À1 ). This is due to the higher oxygen content of CHCl 3 -linked COPs, which result in stronger CO 2 binding. Apart from surface chemistry, porosity of sorbents also plays an important role in dictating gas adsorption properties. Here, whether they are porous or nonporous, CO 2 adsorption performances of all 29 COPs were measured up to 1.1 bar at 273 K and the data are shown in Table 3. To understand the relation between adsorption capacities and porosity parameters, we plotted CO 2 uptake capacities at 1.1 bar and 273 K against BET surface areas (Figure 4g), Langmuir surface areas (Figure 4h), and micropore volumes (Figure 4i), and they revealed three important findings. First, all plots showed similar trends; no difference was observed among BET surface area, Langmuir surface area, and micropore volume plots. Second, highly porous COP-130 and COP-140, although having best uptake capacities, underperformed with respect to their porosities. This can be explained by their lower number (per surface area) and uneven distribution of surface functional groups on pore surfaces. This is because the higher surface area will result in lower number of oxygenated species per surface area. Third, although porous COPs exhibited relatively a linear trend with porosity parameters, surprisingly, some nonporous polymers showed higher CO 2 adsorption capacities, even competing with higher surface area structures. For instance, nonporous COP-148 (2.5 m 2 g À1 ) exhibited 1.71 mmol g À1 CO 2 uptake at 1.1 bar and 273 K, surpassing porous COP-129 (1.06 mmol g À1 , 533 m 2 g À1 ), COP-138 (1.54 mmol g À1 , 409 m 2 g À1 ), and COP-150 (1.67 mmol g À1 , 120 m 2 g À1 ). We believe that these COPs possess ultramicroporous cavities inaccessible to larger Ar, but accessible to smaller CO 2 through a quadrupole interaction with surface. [32] To better evaluate the porosities of COPs, BET surface areas were also calculated from CO 2 adsorption isotherms at 273 K and the results are shown in Table 3. While nonporous COPs exhibited higher BET surface areas from CO 2 adsorption than Ar, highly porous COPs showed lower BET surface areas. For instance, less-porous COP-148 (2.5 m 2 g À1 ) exhibited 300 m 2 g À1 using CO 2 , but COP-130 (1685 m 2 g À1 ) showed 878 m 2 g À1 (using CO 2 ), almost half of its surface area when Ar was used.

The CH 4 and H 2 Adsorption
Solid porous adsorbents are viable candidates for safely storing and transporting hydrogen (H 2 ) and methane (CH 4 ) gases. Therefore, to test our materials, we then switched the gas probe to H 2 and CH 4 . The H 2 adsorption and desorption isotherms were measured at 77 K ( Figure 5). The COPs exhibited high uptake capacities toward H 2 adsorption (Table 3). As in CO 2 adsorption at 1.1 bar, COP-130 and COP-140 showed highest capacities with values 2.02 and 1.70 wt%, respectively. These numbers are among the leading materials in H 2 adsorption (Table S3, Supporting Information). For example, COP-130 stores more H 2 than higher surface area materials such as COF-102 (3530 m 2 g À1 , 1.20 wt%) [1b] and PAF-1 (5600 m 2 g À1 , 1.50 wt%). [30] To investigate the relation of H 2 uptake and porosity, we plotted the H 2 adsorption capacities of COPs (at 1.1 bar) against their BET surface areas from Ar and CO 2 gases (Figure 5c,d). The plot against Ar surface areas showed similar trend as in Figure 4g (CO 2 plots), with nonporous COPs having higher H 2 uptake capacities than expected. In contrast, most of the H 2 uptake capacities showed fairly linear trend when plotted against CO 2 surface areas. Similar to CO 2 , smaller H 2 can access ultra-microporous pores where Ar cannot adsorb. Therefore, it is not surprising to observe a linear trend between CO 2 and H 2 adsorption properties. The CH 4 isotherms were measured at 273 K, up to 1.1 bar (Figure 6). Similarly, COPs exhibited good performance toward CH 4 adsorption, with uptake capacities as high as 1.31 mmol g À1 (COP-140) at 273 K, 1.1 bar (Table 3). These values are comparable with important porous materials such as PAF-1 (5600 m 2 g À1 , 0.80 mmol g À1 ) [30] and BILP-6 (1261 m 2 g À1 , 1.68 mmol g À1 ) (see Table S4, Supporting Information, for the comparison table). As we did in H 2 adsorption, CH 4 adsorption capacities at 1.1 bar were correlated with BET surface areas from Ar and CO 2 ( Figure 6). The CH 4 uptake exhibited similar behavior as in H 2 ; nonporous COPs showed higher CH 4 capacities when plotted against Ar surface areas but exhibited linear trend against CO 2 surface areas. This is surprising, as CH 4 has a larger kinetic diameter (0.38 nm) than Ar (0.34 nm); hence, one should expect www.advancedsciencenews.com www.advenergysustres.com linear correlation between CH 4 capacities and BET surface areas from Ar. However, linear trend with CO 2 surface areas indicates that CH 4 can access ultra-micropores where only H 2 and CO 2 can fit. One possible explanation is that these COPs possess flexible frameworks with affinity toward CH 4 adsorption. Several studies reported CH 4 -specific pore expansion of porous materials to allow higher uptake capacity. [18a,33] Previously, we showed that at high pressures, flexible COP-150 and COP-148 can expand and store high amount of CH 4 . [18a] These COPs feature methanophilic aromatic and aliphatic hydrocarbon framework with alkyl bridges to allow rotational freedom enabling structural expansion with external stimuli. Although the COPs studied here show exceptional promise toward CH 4 and H 2 uptake, they were only tested up to 1.1 bar. As CH 4 and H 2 are often stored at high pressures (up to 100 bar), high-pressure gas uptake analysis is needed for better assessment of materials performance.

Conclusion
In conclusion, we have discovered that chlorinated solvents can be used as linkers in the formation of porous network polymers with aromatic compounds in the presence of AlCl 3 . Using the new method, a series of 29 different porous polymers were prepared by varying 18 different aromatic monomers in dichloromethane, chloroform, 1,2-dichloroethane. Dichloromethane and 1,2-dichloroethane act as bidentate linkers to produce diphenylmethane and 1,2-diphenylethane-based polymers, respectively, but chloroform acts as tridentate linker to yield triphenylmethane-based polymers. Produced COPs featured from nonporous to highly porous structures. Two of the best performing polymers are 1,3,5-triphenylbenzene-based COP-130 and COP-140 with BET surface areas of 1685 and 1219 m 2 g À1 , respectively. The highly porous COPs showed excellent gas uptake properties toward CO 2 (up to 4.71 mmol g À1 , 273 K, 1.1 bar), CH 4 (up to 1.31 mmol g À1 , 273 K, 1.1 bar), and H 2 (up to 2.02 wt%, 77 K, 1.1 bar). The H 2 and CH 4 adsorption capacities showed linear correlation with BET surface areas using CO 2 , rather than Ar gas surface areas. This easy and cheap method makes these polymers potential candidates for gas separation and storage applications in industrial scale. Considering the scalable and inexpensive synthesis, COPs studied here show great promise for CO 2 , H 2 , and CH 4 uptake.  www.advancedsciencenews.com www.advenergysustres.com (99.0%), triphenylamine (98%), triphenylphosphine (99.0%), and deuterated chloroform (99.96% deuterium and contains tetramethylsilane) were purchased from Sigma Aldrich, USA. Triphenylmethane, tetraphenylmethane (98%), and 1,3,5-triphenylbenzene (99%) were purchased from Alfa Aesar, Korea. Mesitylene (97%) and biphenyl (99%) were purchased from Acros Organics, Korea. 1-Bromoadamantane and 2-bromo-2-methylpropane were purchased from Tokyo Chemical Industry (TCI), Japan. Aluminum chloride (95%) and phosphorus pentoxide (P 2 O 5 ) were purchased from Junsei Chemicals, Korea. Dichloromethane, chloroform, 1,2-dichloroethane, and 1,2-dichlorobenzene were distilled over P 2 O 5 under nitrogen atmosphere (where they are referred to as a dry solvent). All the solvents were used without purification when they were used for washing purposes or as Soxhlet extraction solvents. Physical properties and safety information of all the chemicals used in this study are shown in Table S1, Supporting Information. Characterization: FTIR with attenuated total reflectance (FTIR-ATR) was done using Shimadzu IRTracer, Gladi-ATR 10 model Fourier-transform infrared spectrometer. Solid-state CP/MAS 13 C NMR spectra of COPs were measured by Bruker Avance III 400 WB NMR spectrometer. The elemental composition of polymers was measured using a combustion (C/H/N/S and O) analyzer and Al content was obtained from ICP-MS measurements after digesting polymers in solution. Prior to ICP-MS measurements, the polymers were digested in concentrated 60 wt% HNO 3 solution. About 5 mg of polymer sample was added to 1 mL of nitric acid and the mixture was heated to 80 C and sonicated for 30 min. TGA analyses were conducted on a Shimadzu DTG-60A by heating the samples from 30 C up to 800 C at a rate of 10 C min À1 under N 2 . Scanning electron microscope (SEM) images were taken by field-emission SEM, Magellan400, FEI Company, USA. Porosity analysis of the solvent-linked COPs was conducted with the Micromeritics ASAP 2020-accelerated surface area and porosimetry analyzer. Samples were degassed at 150 C for 6 h under vacuum. The Ar adsorption and desorption isotherms at 87 K were obtained to give pore parameters, including BET (at P P o À1 ¼ 0.01À0.25) surface area and Langmuir (at P P o À1 ¼ 0.01À0.25) surface area, PSD, and pore volume. The PSDs of solvent-linked COPs were calculated from Ar adsorption isotherms by the NLDFT method using a slit pore model. Total pore volumes of COPs were calculated at P P o À1 ¼ 0.99. Micropore volumes of COPs were extracted from NLDFT cumulative PSDs under 2 nm. The CO 2 adsorption and desorption isotherms were measured up to 1.1 bar at 273 and 298 K, using a volumetric system (ASAP 2020, Micromeritics Inc., USA). The CH 4 and H 2 adsorption and desorption isotherms were measured up to 1.1 bar at 273 and 77 K, respectively, using a volumetric system (ASAP 2020, Micromeritics Inc., USA). The temperature during adsorption and desorption was kept constant using a circulator. Before the gas adsorption measurements, all samples were properly dried at 120 C under vacuum for 12 h and degassed under vacuum at 150 C for 6 h.

Materials
Synthesis: Tetraphenyladamantane was prepared by a reported procedure, [34] and all other chemicals were commercially available. Dichloromethane, chloroform, and 1,2-dichloroethane solvents were distilled over P 2 O 5 under inert gas when they were referred to as dry. Other chemicals were used without further purification. Solvent-linked COPs were synthesized under inert conditions by mixing %1 g (or 1 mL) aromatic monomer with 2-4 molar equivalent (depending on the monomer) of AlCl 3 in 20 mL of dry chlorinated solvent in the 30 mL glass vial (Caution: HCl might build up pressure in the vial). The reaction mixture was stirred for 48 h at room temperature under inert conditions. Throughout the synthesis, the color of the solution mixture slowly turned to red and then to black (Caution: AlCl 3 and methanol reaction is highly exothermic). The obtained solid was filtered and washed with methanol and chloroform in Soxhlet extractor. The resulting solid was dried under vacuum at 120 C. For the detailed procedures of each COP polymer, refer to the Supplementary Information.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.