. Exploiting Hansen Solubility Parameters to Tune Porosity and Function in Conjugated Microporous Polymers

Here we expand our recently reported Bristol – Xi'an Jiaotong (BXJ) approach using simple salts to ﬁ ne-tune the porosity of conjugated microporous materials synthesized by various reaction approaches, including Buchwald – Hartwig (BH), Sonogashira – Hagihara, oxidative coupling and Suzuki cross-coupling. The surface area and the porosity of the produced conjugated microporous polyanilines (CMPAs) acquired from the non-salt-added BH coupling are optimized by the addition of inorganic salts. BXJ-salt addition provides a facile route to radically improve the BET surface area from 28 to 901 m 2 g (cid:1) 1 for PTAPA and from 723 m 2 g (cid:1) 1 to 1378 m 2 g (cid:1) 1 for PAPA in a controllable manner. In addition, the surface area shows a gradual decrease with an increase in the ionic radius of salts. We furthermore show high compatibility of this approach in the synthesis of typical CMPs, further increasing the surface area from 886 to 1148 m 2 g (cid:1) 1 , 981 to 1263 m 2 g (cid:1) 1 , and 35 to 215 m 2 g (cid:1) 1 for CMP-1 , PTCT and p -PPF , respectively. More importantly, the BXJ approach also allows the broad PSD of the CMPs to be narrowed to the microporous range only, mimicking COFs and MOFs. With the porosity optimized, CO 2 uptakes are dramatically improved by >300% from 0.75 mmol g (cid:1) 1 to 2.59 mmol g (cid:1) 1 for PTAPA and from 2.41 mmol g (cid:1) 1 to 2.93 mmol g (cid:1) 1 for PAPA . Careful addressing of Hansen solubility parameters (HSPs) of solvents and resulting polymers through salt addition has the potential to become an important design tool for the preparation of fully tuneable porous materials. We are currently exploring further methods to tune both structure and function in a wide range of organic porous materials.


Introduction
Rapid consumption of fossil fuels has caused excessive and uncontrolled CO 2 emissions, resulting in global climate change and wide-ranging environmental challenges. 1 As such, methodologies focusing on CO 2 capture are of great interest to meet these growing global environmental problems. 2,3Increasing effort has been focused on the search for materials for CO 2 sequestration, and a wide variety of advanced porous materials have been developed.Such materials include crystalline polymers (covalent organic frameworks (COFs), 4 metal-organic frameworks (MOFs), 5 covalent triazine frameworks (CTFs) 6 ) as well as amorphous organic polymers (porous organic cages (POCs), 7 polymers of intrinsic microporosity (PIMs), 8 porous coordination polymers (PCPs), 9 hyper-crosslinked polymers (HCPs), 10 porous aromatic frameworks (PAFs), 11 and conjugated microporous polymers (CMPs) 3,[12][13][14][15] ).Among them, CMPs possess attractive intrinsic properties, including low skeleton density, high chemical stability, rich microporosity and large surface area, exhibiting considerable advantage over other organic polymers, especially over MOFs and COFs, in terms of the stability. 16,17Additionally, they exhibit unique extended pconjugation and rich heteroatom content throughout the porous 3D networks. 17These intrinsic properties enable CMPs to be suitable for application in many areas of interest, especially in CO 2 capture, in which the micropores and heteroatoms contribute to their high affinity towards CO 2 , thus increasing their adsorption capacity and selectivity. 15 wide range of coupling chemistries, such as Sonogashira-Hagihara, 12 Yamamoto, 18 Suzuki, 19 Friedel-Cras, 20 and oxidative coupling, 21 have been devoted to construct novel CMPs.Recently, Buchwald-Hartwig (BH) coupling has been applied in CMP design and synthesis, where carbon-nitrogen bonds can be formed in a facile manner using a Pd catalyst and a base. 15A simple route was therefore developed to form N-rich redoxactive CMPs, i.e., 3D analogues of the well-known redox-active polymer poly(aniline) (PANi).However, using BH coupling generally resulted in products with low porosity. 14,15Moreover, such materials possessed unexpectedly low micropore volumes and broad pore size distributions (PSDs) when compared, for example, with structurally similar poly(aryleneethynylene)s (PAEs). 14,15To address this unexpected challenge, we have shown in a recent paper 22 that the surface area of a model BHsynthesized CMP, polytriphenylamine (PTPA), could be netuned and optimized by the addition of inorganic salts to the reaction mixtures.This approach, termed the Bristol-Xi'an Jiaotong (BXJ) approach, yielded PTPA with a 20 times higher surface area than the standard reaction.The initially broad PSD of PTPA could additionally be tuned by the BXJ approach to a narrow micropore range distribution, closely mimicking that of COFs and MOFs.Our developed approaches therefore present an opportunity to tune surface area and porosity, potentially for a range of amorphous polymers.However, this potentially impactful methodology has only been applied in one specic instance in our rst published study, 22 and its general applicability to create a design tool for wider application remains to be fully explored.
Thus, we describe here a series of BXJ-assisted conjugated microporous polyanilines (CMPAs) synthesized by BH coupling chemistry. 23,24We explored different linkers with longer length or with more polymerization nodes (i.e., A 3 + B 3 , as depicted in Scheme 1) compared to that used in PTPA to fully conrm the generality of our BXJ approach in the synthesis of amorphous CMPs using BH coupling.Additionally, application of the BXJ approach to other typical coupling reactions, including Sonogashira-Hagihara, Suzuki and oxidative coupling, were studied to further optimize the porosities of these typical model materials (i.e., CMP-1, p-PPF and PTCT).

Results and discussion
To explore the wider applicability of our BXJ approach in the synthesis of BH-prepared CMPs, salts with varying anion (i.e., NaF, NaCl, NaBr, NaI) or cation (i.e., LiNO 3 , NaNO 3 , KNO 3 , Ba(NO 3 ) 2 ) sizes were applied in the synthesis of CMPA networks.As an initial comparison we used tris(4-bromophenyl)amine as the core, and compared 1,4-phenyldiamine (used in our previous studies) with 4,4 0 -diaminodiphenyl sulde (i.e., longer diphenyl linker) as linkers, leading to the formation of PTPA, and PTAPA, respectively, as shown in Scheme 1.The polymers acquired were chemically stable (i.e., were insoluble in common organic solvents and 3 M HCl or NaOH solutions), suggesting the hyper-crosslinked nature of the polymers. 1This heteroatomcontaining BXJ CMP also exhibited similar multiple oxidation states as found for PTPA described in our previous work.The materials were initially a light brown colour under a N 2 atmosphere during the polymerization, but gradually turned dark with increasing exposure to air, conrming the successful construction of C-N bonds and a PANi-type structure by BH coupling. 22Additionally, their molecular structures were conrmed by Fourier transform infrared (FTIR) and solid-state 13 C cross-polarization magic angle spinning nuclear magnetic resonance (SS 13 C CP/MAS NMR).As shown in Fig. 1 and S1, † apart from the characteristic peaks ascribed to the functional groups of the core and linkers, bands assigned to C-Br bonds (at 710, 1004, and 1070 cm À1 ) and those ascribed to NH 2 stretching (peaks at 3400 and 3300 cm À1 ) are absent from the spectra of the products. 15The SS 13 C CP/MAS NMR spectrum of PTAPA depicted in Fig. 2 showed the expected two main resonances at 127 and 141 ppm (assigned to the unsubstituted phenyl carbons and substituted phenyl carbons, respectively), Scheme 1 Synthetic route for the formation of salt-tuneable CMPA networks.as well as an additional peak (at 120 ppm) owing to the presence of the sulfur heteroatom. 14These results tted well with the simulated results shown in the ESI, Fig. S2, † further conrming the successful BH coupling reactions.
Interestingly, the addition of salts during polymerization did not change the molecular structure (as shown in FTIR and SS 13 C CP/MAS NMR), but led to macroscopic gelation of PTAPA during polymerization (rather than the usual precipitation when no salt was added), as was also observed in PTPA synthesis. 22These changes were also reected in the SEM micrographs of these materials (shown in Fig. S4 †), where the normally particulate-like morphology of CMPAs became almost foam-like aer the addition of salts.These changes were early indicators of the positive effect the BXJ salts had on the physical properties of CMPAs produced by BH coupling.
We further investigated this inuence of salts on the PTPA structures and, potentially, the porosity through detailed BET surface area analysis.PTAPA, with a longer linker than PTPA, showed a very low BET surface area of 28 m 2 g À1 under standard BH cross-coupling conditions (as shown Table 1).A remarkable improvement of surface area of more than 30 times to 901 m 2 g À1 was obtained with the aid of NaF.The surface area was then systematically decreased to 256 m 2 g À1 by increasing the anion size of the employed sodium salts.This tuneable trend was also reected in the change of total pore volume, with an almost 10 times increase from the initial value of 0.078 cm 3 g À1 to 0.81 cm 3 g À1 with the use of NaF, and gradual reduction to 0.21 cm 3 g À1 (when NaI was applied).This delicate and ne control over the physical properties shows the suitability of the BXJ approach for overall structural design and optimization.General applicability was also conrmed with our BXJ nitrate salts on ne-tuning the surface area of PTAPA as shown in Table 1.A surface area of 652 m 2 g À1 was effectively reached by LiNO 3 , and then tuned to 556 m 2 g À1 by adding cations of increasing size.Note that these results were highly reproducible, with selected syntheses repeated three times (see Experimental section).However, unlike the surface area, there was no obvious trend in tuning the pore volume of PTAPA when applying salts with increasing cation radius.This outcome demonstrates the formation and control of pores with different volumes are more complex, especially with nitrate salts, as they could not be as nely controlled by the salt as the surface area.It is noteworthy that similar trends were also observed for the control of PTPA.See Table S1 † for full details.
In an effort to continue to further validate the applicability of our BXJ approach, we investigated the inuence of the BXJ method on the BET surface area of another CMPA: PAPA, with one additional polymerization node in the linker than PTPA (see a Surface area calculated from the N 2 adsorption isotherm using the Brunauer-Emmett-Teller method.b The total pore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method.c The micropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for micropore (r < 2 nm) volume.d The ultramicropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for ultramicropore (r < 0.7 nm) volume.
Scheme 1).Again, the addition of salts during polymerization did not change the molecular structure (as shown in FTIR and SS 13 C CP/MAS NMR, Fig. 1 and 2, respectively), but led to macroscopic gelation (as found for the other systems).
Interestingly, an unexpected high surface area of 723 m 2 g À1 , as shown in Table 2, was acquired before salt tuning, indicating that the chosen reaction conditions were already a very good starting point for our investigations.In addition, earlier studies showed that higher connectivity in monomers oen lead to higher surface area. 29,30In order to explore whether these already high values could be tuned, and further control exerted over other physical characteristics, we varied the salts using our BXJ approach.Here a very well-dened trend in the surface area was observed when sodium salts with varying anion radius were applied: the surface area was almost doubled to 1378 m 2 g À1 by NaF, and then decreased with increasing anionic radius of the salts.This obtained surface is, to the best of our knowledge, one of the highest surface areas reached for BH-synthesized polymers.Nitrate salts (where the cation radius was varied) also showed regular inuence on the surface area of PAPA as shown in Table 2.The initial surface area was increased slightly to 818 m 2 g À1 with the decrease of cation radius.However, this inuence was much less pronounced for PAPA (compared with that of PTPA and PTAPA).
To gain insight into the inuence of BXJ salts on the microporosity of these CMPAs, optimized dosages, as determined from the study of salt dosage inuence on the porosity as shown in Tables 1, 2 and Fig. S6, † of various sodium salts (NaF, NaCl, NaBr, NaI) and nitrate salts (LiNO 3 , NaNO 3 , KNO 3 , Ba(NO 3 ) 2 ) were now employed in the BXJ synthesis of PTAPA and PAPA.Specically, salt concentrations of 0.75 mmol (for PTAPA) or 0.5 mmol (for PAPA) were used and the porosities of polymers acquired were investigated by N 2 adsorption and desorption isotherms, as shown in Fig. 3.For PTAPA a typical Type II isotherm was observed, as shown in Fig. 3(a) and (c), while a very broad PSD was obtained when using the standard reaction without salt as control (as shown in Fig. 3(b) and (d)).Such a broad PSD most likely results from interparticular voids or interparticulate porosity rather than from the intrinsic porosity. 13,15Type I isotherms were acquired for BXJ-PTAPA aer adding both types of salt (Fig. 3(a) and (c)), with the resulting PSD becoming very narrow, showing only micropores.This result indicates that large numbers of micropores of dened size were produced by adding salts to the reaction mixture. 12,31s for PAPA, even though its isotherm showed intrinsic Type-I a Surface area calculated from the N 2 adsorption isotherm using the Brunauer-Emmett-Teller method.b The total pore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method.c The micropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for micropore (r < 2 nm) volume.d The ultramicropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for ultramicropore (r < 0.7 nm) volume.
shape before salt tuning, the PSD was signicantly narrowed aer salt addition.As such, tuning the pores of amorphous CMPAs to resemble COF-or MOF-like distributions is therefore within reach with our BXJ approach (see the relevant XRD diffractograms in Fig. S7, † and transmission electron microscope (TEM) images in Fig. S8 and S9 †).
To obtain a fundamental understanding of the inuences of our BXJ approach on these CMPAs, we further explored the theory of Hansen solubility parameters (HSPs) 32 in the context of our synthetic approach.According to the HSP theory (see detailed guide in Section S2 in ESI †), solvents are designated as "good solvents" for a polymer when the difference in their total solubility parameter |d T | < 1 (where |d T | ¼ |d T, solvent À d T, polymer |).Solvents with a d T difference of 1 < |d T | < 3 would be classied as "intermediate solvent", and would still be able to perform well during a polymerization.When the |d T | > 3, solvents are indicated as "poor solvents" for the resulting polymer and is not suitable for polymer synthesis. 33,34As a general guide, the early phase separation of our polymer networks (and the resulting large average diameter pores and low BET surface areas) could result from weak matching of d T, solvent with that of the growing polymer.Solvents with matching HSPs and good compatibilities (difference in |d T | < 1) with the (growing) polymer networks would, in contrast, contribute to a much later-stage phase separation, yielding polymers with uniform PSD and rich micropores (see our recent publication and the ESI † for detailed discussion). 22,33,34With this background in mind, we calculated 35 the HSPs of PAPA and PTAPA, with a detailed step-by-step methodology and results presented in the ESI † (Section S3, results shown in Fig. S10, Tables S2 and  S3 †).The estimated total solubility parameter (d T ) of PTAPA and PAPA, listed alongside that of PTPA and literature values of solvents, are summarized in Table 3.
Results in Table 3 clearly shows the reason for the BXJ tuning performance on different CMPAs.For PTAPA, the |d T | between THF and PTAPA was 3.9, indicating THF is a poor solvent for PTAPA synthesis.This explains why a low surface area of 28 m 2 g À1 was acquired in THF without any salt addition.However, salts contribute to the permanent dipole interactions (d P ) and the hydrogen-bonding interactions (d H ) of the solvent, resulting in a decrease of |d T | and increase of compatibility of the solvent for polymer during the polymerization.A high surface area was therefore acquired for PTAPA aer exploiting the BXJ approach.Additionally, the electronegativity of the ions decreases with the increase of salt ionic radius.As such, ions with larger radii should have less inuence on the d P and d H components of the solvent (by adjusting the hydrogen bonding and polarity of the solvents), thus resulting in a less effective adjustment (and therefore matching) of the HSPs.The surface area of PTAPA could be therefore ne-tuned by the radii of the salts, as shown in Table 1.However, for PAPA, an intrinsic low d T was estimated, and the |d T | between THF and PAPA was only 1.4, suggesting that THF was an intermediate solvent suitable for PAPA to polymerize.This observation is consistent with the surface area data shown in Table 2, in that no or much less salts were needed to tune the solvent to be compatible for PAPA polymerization; a high surface area of 723 m 2 g À1 could be reached by simply using THF, but then increased dramatically to almost double (1378 m 2 g À1 ) by salt tuning.
For the case of PTAPA, a much larger d T of 23.4 was obtained, compared with the values of 22.6 and 20.9 for PTPA and PAPA, respectively.This obtained value validated the observations that signicantly higher salt concentrations (1.00 mmol) were needed to acquire matching solvent quality for PTAPA, as observed in dosage-dependent results shown in Tables 1 and 2. These observations also explain the fact that it was easier to overtune the solvent for PAPA (with the lowest d T ) compared with PTPA and PTAPA.These results further support the applicability of our BXJ approach and HSP mechanism for netuning the surface area of CMPAs in a quantitative manner.With this principle fully established, we expanded our investigations and switched solvents to dioxane to prepare PAPA (d T, dioxane ¼ 20.5, and thus a very close match of (|d T | ¼ 0.4) to the calculated value for PAPA).As shown in Fig. S11(d-f), † Tables 1 and 2, an increased surface area of 838 m 2 g À1 was achieved by dioxane even when using the standard BH reaction without salts (compared with a value of 723 m 2 g À1 acquired with THF as solvent under such conditions).Exploring properties with salt additions of 0.5 mmol Na 2 SO 4 led to overtuning the HSPs of dioxane, and consequently to a decrease in the surface area of PAPA from 838 m 2 g À1 to 614 m 2 g À1 .
We further expanded our BXJ approach to other typical CMPforming reactions, including Sonogashira-Hagihara, 12 oxidative, 21 and Suzuki 36 couplings, to attempt tuning and improving the surface area of the prototypical examples from such reactions.As a starting point we added 0.5 mmol of NaF (based on our experience for CMPA optimization as outlined above) to these reaction mixtures.Results, and comparisons to published data are shown in Fig. 4 and Table 4, clearly demonstrating the successful application of the BXJ approach in these reactions too.The surface area of poly(aryleneethynylene)s (CMP-1) acquired from Sonogashira-Hagihara reaction was increased from 886 to 1148 m 2 g À1 .In the case of PTCT, produced by oxidative coupling, the surface area increased from 981 to 1263 m 2 g À1 , while for p-PPF, synthesized by Suzuki coupling, a signicant increase in surface area from 35 m 2 g À1 to 215 m 2 g À1 was observed, despite the fact that inorganic salts, such as CuI, FeCl 3 and K 2 CO 3 were already utilized as catalysts or bases in the original reactions.More interestingly, the PSDs of these products were generally narrowed, as shown in Fig. 4(b), (e) and (h).Improved values for micropore volumes were also observed: 0.58 to 0.80 cm 3 g À1 (CMP-1), from 0.60 to 0.86 cm 3 g À1 (PTCPT), and from 0.081 to 0.14 cm 3 g À1 (p-PPF), respectively.
These results further reinforced the adaptability of our BXJ approach to facilitate the creation of well-dened micropores and high surface area for polymers acquired by non-BH reaction routes.It is noteworthy that the porosity was enhanced by BXJ salts in a similar way as for CMPAs by inuencing the phase separation and the polymerization degrees, as evidenced in the very different  a Porosity data calculated from N 2 adsorption and desorption were unavailable from the original paper published by authors. 35The authors acquired surface area from CO 2 sorption isotherm due to the low porosity, and the data was not suitable for comparison.b The total pore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method.c The micropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for micropore (r < 2 nm) volume.d The ultramicropore volume calculated from the desorption branch of the N 2 isotherm using the NL-DFT method for ultramicropore (r < 0.7 nm) volume.b c d SEM morphologies obtained for these model networks (Fig. S12 †) as well as the enhanced thermal stability (Fig. S13 †) before and aer salt addition.As shown for CMPAs, the molecular structures of these samples were not changed by our BXJ method either, and conrmed by FTIR study (Fig. 1, S14 and S15 †). 39s the materials synthesized through the BXJ approach showed very promising microporosity, the inuence on the functionality (i.e., these materials' ability to adsorb CO 2 at 273 K) was investigated.Data are shown in Fig. 5, with the capacity at 1 atm summarized in Tables 1 and 2. Signicant increases in CO 2 sequestration capacities were acquired with our BXJ approach.CO 2 uptake was dramatically improved by >300% for PTAPA from 0.75 mmol g À1 to 2.59 mmol g À1 ; a small enhancement from 2.41 mmol g À1 to 2.93 mmol g À1 was observed for PAPA.The inuence was also clearly observed for the polymers produced by non-BH reactions, as shown in Table 4, where a maximum improvement of 47% in CO 2 capacity was achieved for CMP-1.
However, unlike the surface area, there was no obvious trend for the CO 2 uptake ability when increasing the ionic radius of anions or cations in the applied salts.The CO 2 uptake shown above for PTAPA and PAPA samples seems to be unrelated to other porosity parameters such as total pore volume, micropore volume and ultramicropore volume.These observations further support the conclusion that the CO 2 adsorption of such porous systems is related to a number of factors, not only surface area. 37,38By comparing the CO 2 uptake ability with other polymers in Table S4, † our CMPAs, powered by the BXJ approach, exhibit very attractive CO 2 sequestration capabilities, owing to the combination of their rich micropores, high surface area and signicant N content.

Conclusions
Full synthetic control over the surface area and porosity of BHproduced CMPAs through our BXJ approach is reported in this paper.Using HSP theory, we showed tuneable surface area and porosity for two new CMPAs by adjusting the ionic radius of the used salts, following on our initial work with PTPA.Surface areas were dramatically increased from 28 m 2 g À1 to 901 m 2 g À1 for PTAPA and from 723 m 2 g À1 to 1378 m 2 g À1 for PAPA.With a clear understanding of the factors inuencing the materials properties established, the general applicability of our approach was extended to other non-BH coupling reactions for the rst time in a detailed fashion.The BXJ approach shows high adaptability for typical metal-catalyzed reactions used for the preparation of porous materials, increasing the surface area from 886 to 1148 m 2 g À1 , from 981 to 1263 m 2 g À1 , and from 35 to 215 m 2 g À1 for CMP-1, PTCT, p-PPF, respectively.More importantly, the BXJ approach also allows the broad PSD of the polymers to be tuned to approach those found for COFs and MOFs.With the porosity optimized, CO 2 sequestration capacities were dramatically improved by more than 300% in some cases.Salt tuning is therefore an important rst step to increase our understanding and ability to tune the HSPs of solvents to adjust polymer properties and function.Our BXJ approach not only opens a facile and simple avenue for optimizing CMP properties and functionality, but also provides a rational and generalized route towards the design of novel porous materials.We are currently exploring further effective methodologies to establish design rules to optimize and tune HSPs with the aim to obtain porous materials with fully tuneable functionality for wide application.

Synthesis of conjugated microporous polyanilines without salts (control)
A Schlenk tube was charged with tris(4-bromophenyl)amine (0.5 mmol) as core, and the following linkers: 4,4 0 - diaminodiphenyl sulde (0.5 mmol to achieve a 1 : 1 core to linker ratio, yielding PTAPA), or, tris(4-aminophenyl)amine (0.5 mmol to achieve a 1 : 1 core to linker ratio, yielding PAPA), XPhos (0.045 mmol, 4.5 mol%), Pd(dba) 2 (0.03 mmol, 3 mol%) and NaOtBu (3.5 mmol, 7 equiv.).Note that a 1 : 1 core to linker ratio led to higher surfaces area for our CMPAs.However, when we used a 1 : 1.5 stoichiometry of core to linker, low surface area CMPAs were acquired, even though salts were added.This phenomenon was also observed in the synthesis of CMP-1&2, which are structurally related to our CMPAs by using 1,3,5-triethynylbenzene as the core and 1,4-diiodobenzene/4,4 0diiododiphenyl as the linker.We inferred that excessive bromo functionality would promote linkage of Br-NH 2 through the BH coupling, leading to networks with higher surface area of CMPAs.Anhydrous THF (30 mL) was thereaer added under a nitrogen atmosphere.The reaction mixture was then heated under stirring and kept at 65 C. Aer 48 hours, the mixture was cooled to room temperature and the product collected by centrifugation.The products were puried with CHCl 3 , ethanol, methanol, and boiling water (200 mL each), respectively, to remove the catalyst, impurities, and oligomers.The products were further puried by a Soxhlet extraction with methanol (24 h), THF (24 h) and chloroform (24 h), respectively, before drying under vacuum at 60 C for 24 h.The average yields of PTAPA and PAPA were 56.87% and 95.24%, respectively.We conrmed that the results optimized by BXJ method were highly reproducible, with similar results for BET surface area and porosity when same synthetic procedures were repeated.For example, preparation of PTAPA (tuned by Ba(NO 3 ) 2 ) was performed 3 times, and yielded the following values for surface area and porosity shown in Fig. S16 and Table S5.† The very similar surface areas, pore volumes and pore size distributions clearly conrmed the high reproducibility of our BXJ-BH method for CMPA synthesis.

Synthesis of model CMPs by non-BH reactions
These CMPs were synthesized according to the published procedures (including Sonogashira-Hagihara, 12 oxidative coupling 21 and Suzuki 36 ), and the BXJ-assisted polymers were obtained by an additional dosing of NaF (0.5 mmol). 22Specically, for CMP-1, 1,3,5-triethynylbenzene (2.0 mmol), 1,4-diiodobenzene (2.0 mmol), tetrakis-(triphenylphosphine)palladium (100 mg), copper iodide (30 mg) with/without NaF (0.5 mmol) were placed in the mixture of toluene (2.5 mL) and Et 3 N (2.5 mL) under a N 2 atmosphere in a Schlenk tube.The mixture was then heated to 80 C for 72 h, aer which it was cooled to RT and the resulting polymer were ltered and washed with chloroform, methanol and water (200 mL each), followed by a 72 h Soxhlet extraction with methanol (24 h), THF (24 h) and chloroform (24 h), respectively, before drying under vacuum at 60 C for 24 h.For PTCT, the carbazolesubstituted monomer TCT (1 mmol, 1 equiv.)with/without BXJ-NaF (0.5 mmol) were placed in CHCl 3 (50 mL).Anhydrous FeCl 3 (3 equiv.per carbazole unit) dissolved in CH 3 NO 2 (20 mL) was then added into the ash charged with monomer solution for a homogenous reaction.The mixture was vigorously stirred overnight, and the resulted polymer was ltered and washed with chloroform, methanol and water (200 mL each), followed by a 72 h Soxhlet extraction with methanol (24 h), THF (24 h) and chloroform (24 h), respectively, and, being dried under vacuum at 60 C for 24 h.For p-PPF, 1,3,5-triphenyl tribromide (0.46 mmol), 1,4-benzene diboronic acid (0.92 mmol) with/without BXJ-NaF (0.5 mmol) were dissolved in DMF (20 mL) aer been degassed by four freeze-pump-thaw cycles.Aer which, K 2 CO 3 (2 mmol) in water (4 mL) and tetrakis(triphenylphosphine)-palladium(0) (70 mg, 60.6 mmol) were added to this mixture, which was followed by another degassing step and purged with N 2 and stirred at 150 C in a Schlenk tube for 36 h.The mixture was cooled to RT and the resulted polymer was ltered and washed with chloroform, methanol, and water (200 mL each), followed by a 72 h Soxhlet extraction with methanol (24 h), THF (24 h) and chloroform (24 h), respectively.

Characterization
A PerkinElmer Spectrum 100 spectrometer was used to acquire the FTIR spectra in the region between 4000 and 400 cm À1 .The SS 13 C CP/MAS NMR spectra were collected on a Varian VNMRS-600 spectrometer using a spin rate of 6800 Hz.A Quantachrome Quadrasorb instrument was applied for N 2 adsorption/ desorption isotherms (NAD) and CO 2 adsorption study aer degassing the samples under high vacuum at 150 C for 5 h.The specic surface areas were calculation from the Brunauer-Emmett-Teller (BET) model using the adsorption branches of the N 2 isotherms in the low pressure range from 0.05 to 0.20 at 77 K.The pore size distribution was calculated from the desorption branch of the N 2 isotherms using the nonlocal density functional theory (NL-DFT).Scanning electron microscope (SEM) images were acquired on a JEOL 5600LV SEM microscope.Powder X-ray diffraction (XRD) patterns were obtained with a Bruker D8 Advance diffractometer using Cu Ka radiation (2q ¼ 5 -80 , 40 kV, 30 Ma).Transmission electron microscopy (TEM) images were recorded on a JEM model 2100 electron microscope.

Fig. 1
Fig. 1 FTIR spectra of PTAPA: (a and b), before and after tuning by BXJ-Na salts; and PAPA: (c and d), before and after tuning by BXJ-NO 3 salts.

Fig. 3
Fig. 3 N 2 adsorption and desorption isotherm and nonlocal density functional theory-pore size distribution of PTAPA ((a and b), tuned by BXJ sodium salts; (c and d), tuned by BXJ-nitrate salts) and PAPA ((e and f), tuned by BXJ sodium salts; (g and h), tuned by BXJ nitrate salts).The pink rectangular strips highlight the microporous region.

Fig. 4
Fig. 4 N 2 adsorption and desorption isotherm, NL-DFT pore size distribution and cumulative pore volume of CMP-1 (a-c), PTCT (d-f) and p-PPF (g-i) tuned by 0.5 mmol BXJ NaF.The pink shaded areas highlight the microporous region.

Table 1
Porosity parameters and CO 2 uptake at 1 atm and 273 K of PTAPA networks produced by the BXJ route with different anions, cations, ion dosage and solvents

Table 2
Porosity parameters and CO 2 uptake at 1 atm and 273 K of PAPA networks produced by the BXJ route with different anions, cations, ion dosage and solvents

Table 3
Difference of total solubility parameter (d T ) between solvents and the PTAPA, PAPA, and PTPA

Table 4
Porosity parameters and CO 2 uptake at 1 atm and 273 K of typical CMPs produced with/without the BXJ approach