A fl uorene based covalent triazine framework with high CO 2 and H 2 capture and storage capacities †

Department of Chemistry, Ludwig-Maximil 5-13, 81377 Munich, Germany. E-mail: be +49 89 2180 77440; Tel: +49 89 2180 7742 Max Planck Institute for Solid State Rese Germany. E-mail: b.lotsch@f.mpg.de; Fax: Nanosystems Initiative Munich (NIM) and 80799 Munich, Germany Inorganic Chemistry III, Universität Bayreu Germany Max Planck Institute for Intelligent Syste Germany † Electronic supplementary information elemental analysis, XRD patterns, C tris(9H-uoren-2-yl)-1,3,5-triazine, CPPI a plots, Ar-QSDFT ttings, CO2-NLDFT dis at 298 and 313 K, H2 isotherms at 298 a 298 K, Henry and IAST calculation 9H-uorene-2-carbonitrile and 9H-uo measurements. See DOI: 10.1039/c3ta154 Cite this: J. Mater. Chem. A, 2014, 2, 5928


Introduction
0][11][12][13][14][15][16] Recently, low temperature syntheses of CTFs have set the stage for further applications such as for chemo-and sizeselective membranes 17 and in optoelectronics. 10he development of potent gas storage systems has been fueled by the need for highly selective gas capture materials apt to selectively lter out or enrich relevant gases such as methane or carbon dioxide.The anthropogenic emission of CO 2 is known to be a major source of global warming.The global emission of CO 2 by power plants and public transportation has risen remarkably in the last decades. 18According to the International Energy Agency (IEA), appropriate capture and storage of postcombusted CO 2 (CCS) has the potential of decreasing the emissions up to 20%. 19Typically, ue gas of a coal-red power plant consists of approximately 15% CO 2 , 5% H 2 O, 5% O 2 and 75% N 2 and is emitted at 40-80 C and 1 bar. 20,21Therefore, materials suitable for CCS require a high preference for adsorption of CO 2 under these conditions.Amine scrubbing and cryogenic cooling are the only established technologies for CO 2 capture up to date, which, however, have the disadvantage of increasing the energy requirements of a power plant by 25-40%. 20,22Especially amine scrubbing needs large volumes of solvents, high temperatures for regeneration and costly disposal when expired. 22,23The disposal and the formation of toxic byproducts during the regeneration step additionally raise environmental concerns about this technology.5][26][27] Main advantages of solid adsorbents are their long lifetimes and recovery at moderate temperatures.Especially POPs are promising because of their high chemical and thermal stabilities as well as synthetic versatility, giving rise to a large variety of functional and structural designs.
In this work we present the synthesis and characterization of a new uorene-based covalent triazine framework (-CTF) at different temperatures.We tested the material properties regarding CO 2 adsorption at 273 K, 298 K and 313 K, along with the CO 2 over N 2 selectivity, showing high capture capacities and good selectivities.Additionally, we measured high-pressure adsorption of H 2 at 77 K and 298 K, yielding uptakes at the forefront of polymeric hydrogen storage systems.
Argon, carbon dioxide and nitrogen adsorption/desorption measurements were performed at 87, 273, 298 and 313 K with an Autosorb iQ surface analyzer (Quantachrome Instruments, USA).Samples were outgassed in a vacuum (10 À7 mbar) at 200-300 C for 6 h to remove all guests.Pore-size distributions were determined using the calculation model for Ar at 87 K on carbon (slit pore, QSDFT equilibrium model) or for CO 2 at 273 K on carbon (NLDFT model) of the ASiQwin soware (v2.0) from Quantachrome.For BET calculations pressure ranges of the Ar isotherms were chosen with the help of the BET Assistant in the ASiQwin soware.In accordance with the ISO recommendations multipoint BET tags equal or below the maximum in V(1 À P/P 0 ) were chosen.The isosteric heats of adsorption were calculated from the CO 2 adsorption isotherms using the Quantachrome soware ASiQwin (v2.0) based on the Clausius-Clapeyron equation (see ESI, † section 6c).
High-pressure hydrogen adsorption/desorption measurements were performed on an automated Sievert's type apparatus (PCTPro-2000) with a so-called micro-doser (MD) from HyEnergy.The original setup was upgraded by a heating and cooling device to regulate the sample temperature.The adsorption and desorption isotherms (0-20 bar) were measured at various temperatures (77 to 298 K) in a sample cell volume of z1.3 mL using ultra high purity hydrogen gas (99.999%).Samples were outgassed in a vacuum (4.5 Â 10 À6 mbar) at 200 C for 6 h to remove all guests.The isosteric heat of adsorption is calculated from the absolute adsorbed hydrogen according to a variant of the Clausius-Clapeyron equation (see ESI †).
Cryogenic hydrogen adsorption/desorption measurements at 19.5 K were measured with laboratory-designed volumetric adsorption equipment with a temperature controlled cryostat.The experimental set-up has been described in detail elsewhere. 28,29mples were activated under vacuum (10 À4 mbar) at 150 C for 12 h, prior to each measurement.For the laboratory-designed cryostat, the temperature control was calibrated by measuring the liquefaction pressure for hydrogen and nitrogen in the empty sample chamber at various temperatures.
Infrared (IR) spectroscopy measurements were carried out on a Perkin Elmer Spectrum BX II (Perkin Elmer, USA) with an attenuated total reectance unit.
Powder X-ray diffraction (XRD) was measured on a BRUKER D8 Advance (Bruker AXS, Madison, Wisconsin, USA) in the Bragg-Brentano geometry or on a HUBER G670 (HUBER Diffraktionstechnik, Rimsting, Germany) in the Guinier geometry equipped with an imaging plate detector.
Elemental analysis (EA) was carried out with an Elementar vario EL (Elementar Analysensysteme, Germany).
Magic angle spinning (MAS) solid-state nuclear magnetic resonance (ssNMR) spectra were recorded at ambient temperature on a BRUKER DSX500 Avance NMR spectrometer or a BRUKER AvanceIII HD 400 NMR spectrometer (Bruker Biospin, Rheinstetten, Germany) with an external magnetic eld of 11.75 T and 9.4 T, respectively.The operating frequencies are 500.1 MHz, 125.7 MHz and 400.1 MHz, 100.6 MHz for 1 H and 13 C, respectively and the spectra were referenced relative to TMS ( 1 H, 13 C).The samples were contained either in 2.4, 3.2 mm or 4 mm ZrO 2 rotors.The 1D 1 H 13 C cross-polarization (CP) MAS spectra were acquired with a ramped-amplitude (RAMP) CP sequence and contact times between 2 and 10 ms.CPPI (crosspolarization combined with polarization inversion) experiments were carried out to get information about the number of hydrogen atoms directly attached to the carbon.An initial contact time of 2 ms was used and spectra with inversion times up to 400 ms were measured.The measurements were carried out using spinning frequencies of 10 kHz and 15 kHz for CP, and 5.1 kHz for CPPI measurements, respectively.During acquisition broadband proton decoupling using SPINAL64 or TPPM was carried out.
High resolution mass spectroscopy (HRMS) was carried out on a JEOL MS700 (JEOL, Japan), by direct electron ionization (DEI).
Differential thermal analysis and thermogravimetry (DTA/ TG) were measured on a SETARAM TG-DTA92-2400 combined DTA-TG-thermobalance (SETARAM Instrumentation, France) in aluminum oxide crucibles.Heating was performed from room temperature to 800 C with a heating rate of 5 C min À1 under a helium atmosphere.
Microwave reactions were carried out in a Biotage Initiator (Biotage AB, Sweden) in 10-20 mL microwave vials from Biotage.

Synthesis and characterization
The standard procedure for preparing CTFs is by ionothermal synthesis using ZnCl 2 both as a "solvent" and as a Lewis acidic catalyst at temperatures above 300 C. [1][2][3][4] CTFs synthesized at low temperatures (300-400 C) may show little crystallinity, yet at the same time may have well-dened local structures. 1,5,9,32ynthesis at higher temperatures typically leads to dramatic increases in the surface areas accompanied by the loss of structural elements, especially the triazine moieties. 4,5In this work we followed this procedure using 9H-uorene-2,7-dicarbonitrile as a precursor.The precursor was synthesized by a Negishi crosscoupling reaction from 2,7-dibromo-9H-uorene in the microwave with just 5 min reaction time (Scheme 1).This straightforward method to produce aromatic dinitriles is analogous to the synthesis described by us previously. 5Work-up was done by recrystallisation and gave good yields of 76%.The syntheses of the CTFs were performed at temperatures between 300 and 600 C, yielding uorene-CTFs at 300 C (-CTF300), 350 C (-CTF350), 400 C (-CTF400), 500 C (-CTF500) and 600 C (-CTF600).The reaction times were 48 h for all samples, except for -CTF300 and -CTF350, where the reaction times were raised to 96 h aer very low yields had been obtained due to unreacted monomers or oligomers.Reaction times of 168 h did not give higher yields or changes in the material properties.CTF-0, CTF-1 and CTF-2 are the only three examples of CTFs which show moderate crystallinity. 1,12,32Therefore, it was little surprising that our materials were found to be largely amorphous in the XRD measurements, except for broad peaks at 14.8, 22.7 and 33.6 2q.The peak at 22.7 2q is attributed to a 00l reection indicating a "graphitic" layer stacking with an interlayer distance of $3.9 Å, which is rather large and consistent with only weak interlayer coupling (Fig. S1, ESI †).
To probe whether trimerization in these samples was completed with the uorene units still being intact, we used IR spectroscopy and ssNMR measurements and compared the results with the model compound tris(9H-uoren-2-yl)-1,3,5triazine.Fig. 1 depicts the IR and Fig. 2 and S2 † the 13 C ssNMR spectra of the as-synthesized materials.-CTF300 shows well resolved signals in the IR spectrum, whereas the signals weaken at higher synthesis temperatures and are attened out to a large extent for -CTF600, indicating graphitization of the networks.This is conrmed by the 13 C ssNMR measurements, which show a broadening and weakening of the signals with rising temperatures.Remarkably, the NMR spectrum of -CTF600 could not be measured owing to the high degree of graphitization.The IR and ssNMR spectra of -CTF300 and -CTF350 show the presence of a nitrile band (2223 cm À1 , 109.5 ppm), indicating that there are still unreacted nitrile groups in the polymer.Those signals are not visible at higher synthesis temperatures.The strong IR band in the -CTF300 spectrum at 1511 cm À1 can be assigned to the C-N stretching mode of the triazine ring, 32,33 whereas the band at 1352 cm À1 is due to inplane triazine ring stretching vibrations. 3,34This and the corresponding NMR shi at 168.6 ppm evidence successful formation of the triazine moieties (Fig. 2, bottom le).The signals between 144 and 118 ppm and the peak at 36.9 ppm belong to the uorene unit, which accordingly stays intact at 300 C. The signal at 50 ppm is assigned to a sp 3 CH group, which was corroborated by a CPPI measurement (Fig. S4 †).This resonance suggests the formation of 9,9 0 -biuorenyl units via crosslinking.A broadening of the bands between 1500 and 1100 cm À1 and the decrease of the triazine bands in the IR  This journal is © The Royal Society of Chemistry 2014 spectrum of -CTF350 indicates commencing degradation of the system.Again, ssNMR measurements conrm these results.The bands at 1607 and 814 cm À1 appear both in the precursor and in -CTF300 and can hence likely be assigned to the uorene species.These bands are still retained in the spectra of -CTF350 and -CTF400, indicating intact uorene units in those samples.The spectra of -CTF500 and -CTF600 do not show the uorene signature anymore, which conrms the ongoing degradation of the system at elevated temperatures.
In summary, both IR spectroscopy and ssNMR measurements indicate degradation of the networks at temperatures higher than 300 C, which is in line with observations made for all synthesized CTFs so far.Temperatures in excess of 400 C are not tolerated by the triazine rings and likely induce retro-trimerization reactions with subsequent rearrangements, giving rise to irreversible formation of C-C bonds accompanied by a loss of N 2 .
To gain more information about the degree and course of degradation, we utilized EA measurements.EA data shown in Table S2 † reveal a trend which is very similar to many synthesized CTFs: by raising the synthesis temperature, the nitrogen and hydrogen content decreases and the carbon content increases.The loss of nitrogen is very signicant with a nitrogen loss of more than 40% for -CTF600 compared to -CTF300.Notably, the nitrogen content of -CTF300 is already lower than the calculated value, thus conrming the IR results, which reveal smaller amounts of triazine compared to the model compound.
To conrm the thermal stability of the materials we carried out DTA/TG measurements on the samples -CTF300 and -CTF600.The results (Fig. S25 and S26 †) show degradation temperatures close to the synthesis temperatures and a rather low weight loss, especially for the -CTF600 sample.

Porosity measurements
Although an increase in the synthesis temperature leads to materials with less local order and nitrogen content, it typically entails materials with higher surface areas (SAs). 2,3,5The porosities of the -CTFs were determined by argon and carbon dioxide physisorption measurements.Fig. 3 (le) shows the argon isotherms of the samples -CTF350, -CTF400, -CTF500 and -CTF600.It should be mentioned that -CTF300 (Fig. S6 †) adsorbed only small amounts of argon and therefore showed overall poor porosity.The isotherm of -CTF350 is typical for microporous materials in that it shows rapid argon uptake at low relative pressures (p p 0 À1 < 0.05).Nevertheless, it cannot be described as a type I isotherm because of continuous adsorption of argon at higher pressures, indicating additional mesoporosity with a broad mesopore size distribution and a hysteresis almost spanning the entire range of p p 0 À1 ¼ 0-0.9.
The hysteresis resembles type H4, typical for materials with slitshaped pores.The low pressure character of the hysteresis indicates pores within the size range of the adsorbate, causing delayed desorption. 35The isotherm of -CTF400 shows rapid argon adsorption at low relative pressures as well.The continuous adsorption at higher relative pressures is higher than in -CTF350, which should be ascribed to an increased fraction of mesopores.The isotherms of -CTF500 and -CTF600 show similar shapes and can be described as a combination of isotherms of type I and IV.The micropore lling is followed by mesopore lling showing a type H2 hysteresis around p p 0 À1 ¼ 0.4.H2 hystereses describe systems with rather ill-dened pore sizes and shapes and are oen observed for amorphous materials.The SAs of the materials were calculated from the isotherms based on the BET model and are listed in Table 1. 36The highest SA of 2862 m 2 g À1 is found for -CTF400, which is the third highest for all CTFs aer CTF-1 (400/600 C, 3270 m 2 g À1 ) 2 and bipy-CTF (700 C, 3219 m 2 g À1 ). 5 For -CTF500 and -CTF600 the values decrease, which is unexpected, since for all published CTFs a rise in synthesis temperature showed an increase in the SA, accompanied by a continuous decomposition of the materials ("foaming").This phenomenon can be explained by the pore size distribution of the materials which was calculated by QSDFT methods (Fig. 3, right).For -CTF350 mainly micropores are observed with two main peaks at 1.05 and 0.5 nm.The latter indicates pores smaller than 0.5 nm (ultramicropores), which are not observable by argon measurements.For -CTF400 the peak at 1.05 nm increases and additionally a wide distribution of pores from 1.2 to 3 nm is observed, with mainly pores in the micropore region.-CTF500 and -CTF600 show comparable distributions, with an additional peak around 2.5 nm, which can be assigned to small mesopores.Overall, an increasingly mesoporous character with rising synthesis temperature is typical for CTFs, yet normally the increasing amount of mesopores is not accompanied by a signicant decrease of the micropores. 2,5he amount of micropores compared to mesopores can be calculated by the ratio of the microporous volume V mic to the total pore volume V tot .These pore volumes can be calculated from the amount of vapor adsorbed at chosen relative pressures assuming the pores are lled with a liquid adsorbate, and from DFT calculations.The results for the -CTFs are listed in Tables 1 and S4 † and conrm the previous observations.-CTF400 features the largest absolute amount of micropores followed by -CTF500 and -CTF600.Remarkably, -CTF350 has the highest fraction of micropores with respect to the total pore volume (85%), which is clearly higher than for -CTF500 and -CTF600 (both 60%).The widening of the pores can favorably be explained by the degradation of the networks at elevated synthesis temperatures, leading to a "swelling" ("foaming") of the materials.
The pore size distributions of the -CTFs indicate ultramicropores, which cannot be detected by argon measurements, but by carbon dioxide physisorption measurements, which allow us to probe pore sizes down to z0.35 nm.The isotherms shown in Fig. 4 have comparable shapes, which is consistent with the literature.In contrast to its non-porous character established by Ar physisorption, -CTF300 adsorbs moderate amounts of CO 2 , which can be rationalized by the presence of pores that are accessible to CO 2 but not to Ar.A comparable phenomenon is observed for -CTF350, which adsorbs more CO 2 than all other -CTFs, although having around half as much accessible SA as determined by argon physisorption.The accessible CO 2 SAs were calculated with DFT simulations and are shown in Table S6.† The values are much lower than those for argon measurements, since the calculations only take pores smaller than 1.45 nm into account.Nevertheless, -CTF300 shows a SA of 297 m 2 g À1 which is nearly half as much as that of -CTF500 and -CTF600.The highest SA is observed for -CTF400, followed by -CTF350.All samples show a broad pore size distribution ranging from 0.35-1.00nm (Fig. S16 †).A hydrogen physisorption measurement at 20 K validated the SA of -CTF400 giving a BET SA of 2829 m 2 g À1 (Fig. S12 and S13 †).

Gas storage and selectivity studies
The high capacities of the -CTFs for CO 2 sorption at 273 K are promising for the usage as CCS materials.Therefore, temperature-dependent sorption studies at 298 and 313 K were carried out and the heats of adsorption were calculated (Fig. 5).
The uptakes at 1 bar and the heats of adsorption at zero coverage are summarized in Table 1.With 4.28 mmol g À1 at 273 K -CTF350 shows the highest uptake of all materials.This value is lower than that of recently reported FCTF-1 (4.67-5.53mmol g À1 ), 16 but higher than the reported values for all other CTFs such as CTF-0 (4.22 mmol g À1 ), 9 CTF-1 (2.47-3.82mmol g À1 ), 16 CTF-P2-P6 (1.88-3.39mmol g À1 ), 10 CTF-P1M-P6M (0.94-4.20 mmol g À1 ), 10 MCTF300-500 (2.25-3.16mmol g À1 ), 12 PCTF-1-7 (1.84-3.23 mmol g À1 ) 11,15 and TPI-1-7 (0.68-2.45 mmol g À1 ). 34In addition, they are higher than the uptake capacities of numerous POPs such as covalent organic  frameworks (COFs; 1.21-3.86mmol g À1 ), 37,38 microporous polyimides (MPIs; 2.25-3.81mmol g À1 ), 39 and hyper-crosslinked organic polymers (HCPs; 1.91-3.92mmol g À1 ) 40,41 (for an expanded list see Table S7 †).Higher values were reported so far for benzimidazole-linked polymers (BILPs; 2.91-5.34mmol g À1 ), [42][43][44] and porous polymer frameworks (PPFs; 2.09-6.12mmol g À1 ). 45For the amounts adsorbed at 298 K similar results were found.As mentioned above, CO 2 uptakes depend on the surface area, the pore sizes and the interaction energy between a sorbent and a sorbate.The latter can be expressed by the heats of adsorption Q st which we calculated for our materials (Fig. 5, bottom right).The values at the limit of zero coverage are shown in Table 1 and are the highest for -CTF300 and almost the same for the other -CTFs, ranging in the upper eld compared to other POPs (see Table S7 For the use of -CTFs as potential ue gas sorbents the selectivities of CO 2 over N 2 need to be examined.Therefore, the Henry equation and the ideal adsorbed solution theory (IAST) were used (ESI, chapter 8 †).The calculated values are listed in Table 1.As expected, -CTF300 and -CTF350 show the highest selectivities, due to the low adsorption potential towards argon compared to CO 2 .The selectivities range in the upper level of POPs and, remarkably, -CTF300 shows a higher selectivity than FCTF-1. 16otivated by the high values of adsorbed CO 2 at 273 K and the large number of micropores, we tested -CTF400 as a hydrogen storage material by carrying out hydrogen physisorption measurements at 77 K and 298 K up to 20 bar.The isotherm at 77 K is displayed in Fig. 6 and shows adsorption of 4.36 wt% (45.2 mmol g À1 ), which is higher than that observed for 2D COFs (1.46-3.88wt%, sat.pressure) 38 and polymers of intrinsic microporosity (PIMs, 1.45-2.71wt%, 10 bar), 46-50 but lower than that of the highly porous 3D COFs (6.98-7.16wt%, sat.pressure), 38 PAFs (4.2-7.0 wt%) 51,52 and PPNs (8.34 wt%, 55 bar), 53 which however were measured at higher pressures.To date, only three CTFs were examined for their H 2 capacities and the measurements were done only up to 1 bar.CTF-1 adsorbs 1.55 wt%, 1 PCTF-1 1.86 wt% 11 and PCTF-2 0.9 wt%, 11 which is substantially lower than the uptake capacity of our material (1.95 wt%).
For hydrogen storage applications, knowledge of the heat of adsorption, along with the storage capacity, is of importance to better understand the microscopic host-guest interactions.Fig. S10 † shows the temperature variations of the absolute hydrogen adsorption curves (77-117 K), which provides the strength of the binding potential for hydrogen in -CTF400.In Fig. S11, † the isosteric heat of adsorption is shown as a function of the surface coverage normalized to saturation coverage (20 bar, 77 K).Analysis of the hydrogen adsorption enthalpy gives a maximum value of 6.25 kJ mol À1 at near zero surface coverage, decreasing to 3.65 kJ mol À1 with increasing H 2 loading.The overall average heat of adsorption equals 4.9 kJ mol À1 .It is worth noting that the average heat of adsorption of 4.9 kJ mol À1 is relatively high aer considering the rather large pore diameter (1.05-3 nm) and on comparison with most MOFs of similar pore sizes. 54This relatively high average heat of adsorption can be either due to stronger adsorption sites (possibly N sites of -CTF400) or due to the ultramicropores inside -CTF400 as detected by CO 2 NLDFT simulations (0.35-1.45 nm).Table S5 † summarizes the textural characteristics and hydrogen storage capacity of -CTF400.

Conclusions
The presented -CTFs were analyzed with respect to their local structure, porosity, and capacity for CO 2 capture.An intact uorene network connected by triazines could be established for -CTF300.The materials showed high SAs up to 2862 m 2 g À1 and high capacities for the adsorption of CO 2 , which can be rationalized by their high fraction of ultramicropores.We nd that -CTF350 with the highest fraction of micropores and moderate surface area shows the best CO 2 uptake (4.28 mmol g À1 at 273 K) and thus ranks in the upper level among all POPs.Additionally, hydrogen adsorption of -CTF400 shows comparable values to other POPs.Finally, the gas selectivities of CO 2 over N 2 of the -CTFs were tested and reveal high values for -CTF300 and -CTF350, thus rendering these materials promising candidates for gas capture and storage.

Table 1
BET surface areas, CO 2 and N 2 uptakes, heats of adsorption and CO 2 /N 2 selectivities of fl-CTFs a At 1 bar.b At 1 bar and 298 K. c At zero coverage.d At 298 K. e Pore volume for pores smaller than 2 nm calculated from the Ar QSDFT model.f Total pore volume from the Ar QSDFT model.