Introduction

Emissions of hazardous air pollutants from coal-fired power plants and industrial production processes, including fine particulate matters (PM2.5), volatile organic compounds (VOCs), SO2, etc, have led to a wide range of adverse health effects and environmental issues1,2. The simultaneous capture of trace VOCs (such as benzene) and SO2, coexisting in related industrial exhaust gases such as coke oven emissions, has arouse interest for efficient hazards remediation3,4. Adsorption technologies based on porous physisorbents are attractive for removing airborne pollutants due to their relatively low energy footprint and generation of less secondary waste4,5. However, physisorption driven by non-covalent binding typically offers weak interaction with VOCs, especially for benzene at trace concentrations6,7. The situation with SO2 is even more challenging due to its corrosive and acidic nature. Additionally, co-adsorption of moisture can decrease both benzene and SO2 capacity or induce framework decomposition through reactions with H2SO3 in the latter case5. While porous materials hold potential for utility in benzene and SO2 capture, the differing physical and chemical properties of benzene and SO2 present a significant challenge in designing compounds capable of effectively capturing both hazards simultaneously8.

Metal-organic frameworks (MOFs) offer promise in addressing this issue due to their bespoke pore size and tunable environments5,9,10. Open metal sites (OMSs) in MOFs, usually serving as accessible Lewis acid sites, represent strong binding sites for a large variety of guest molecules, such as H2, CO2, olefins, and alkynes11. HKUST-1 and MOF-74 have been demonstrated as typical sorbents characteristic of active OMSs for efficient gas adsorption/separation12,13. By forming strong M–guest interactions, they can also provide high affinity with both benzene and SO2, holding potential in their simultaneous capture. Unfortunately, their performance would deteriorate at low concentration, and most OMSs are susceptible to moisture14. This means a great compromise of adsorption capacity or even degradation of frameworks. MOFs composed of OMSs exhibiting preferential adsorption to benzene and SO2 over water, are thus highly desired, but their design and synthesis is demanding.

Herein, we report the reticular assembly of a pyrazolate MOF, BUT-78 (BUT = Beijing University of Technology) and its application in the capture of trace benzene and SO2. Featuring rich accessible Ni(II) sites and suitable pores, BUT-78 achieves a high benzene uptake of 5.08 mmol g–1 at 10 Pa, outperforming the previous benchmarks. Particularly, an adsorption capacity of ~0.51 mmol g–1 for 10 ppm benzene and ~1.21 mmol g–1 for 250 ppm SO2 under 30% RH have been demonstrated, suggesting that BUT-78 could serve as a powerful platform for simultaneous capture of trace benzene and SO2, regardless of the presence of moisture. In addition, the high chemical stability of BUT-78 enables it to survive after consecutive breakthrough experiments.

Results

Design, synthesis and structure

There are general and facile strategies developed for the synthesis of carboxylate-based MOFs15. By contrast, it is difficult to obtain pyrazolate (Pz for short) MOFs, especially single crystals. The strong M(II)-N(Pz) coordination makes their structure determination and diversification challenging16. Compared with the open Cu(II) sites as found in carboxylate MOFs, square coordinated Ni(II) sites in isoreticular pyrazolate MOFs could present comparable affinity toward benzene and SO2, while they are less susceptible to water. This is supported by the fact that some Ni4 cluster based compounds were synthesized without terminal water on the square coordinated Ni(II) sites17,18. Integrating such Ni4 clusters into pyrazolate MOFs is expected to render good performance in the simultaneous capture of trace VOCs and SO2 from polluted air. As continuation of interest and a step forward in this regard, we synthesized a Pz-MOF BUT-78 through solvothermal reaction between the tetrahedral pyrazole ligand 3,3’,5,5’-tetra(1H-pyrazol-4-yl)−1,1’-biphenyl (H4BPTP, Supplementary Fig. 1) and Ni(OAc)2·4H2O in N,N-dimethylformamide (DMF) under vigorous stirring at 423 K for 3 h. With trace acetic acid as modulator, BUT-78 was obtained as yellow microcrystalline powder (Supplementary Fig. 2). The formation of microcrystals was validated by powder X-ray diffraction (PXRD) analysis (Supplementary Fig. 3a) and scanning electron microscopy (SEM) imaging (Supplementary Fig. 3b). The structure of BUT-78 was determined via structure simulation based on a topological analysis strategy (Fig. 1). We started from the simplest scenario, where the structure composed of a single type of secondary building unit (SBU), organic linker, and connecting edge, known as a binodal edge-transitive network, was focused on. Among various Ni(II)-Pz SBUs, the Ni4Pz8 cluster featuring rich accessible Ni(II) sites, is probable as it shares similar synthesis conditions with two representative Ni(II)-Pz MOFs, Ni3(BTP)217 and BUT-3318, both of which feature the Ni4Pz8 cluster. The Ni4Pz8 cluster, possessing D4h symmetry and being symmetrically and geometrically equivalent to the classical Zr6(μ3-O)4(μ3-OH)4(OH)4(H2O)4 cluster19 (Supplementary Fig. 3c), plays the role of a cubic 8-connected node in binodal edge-transitive nets (Fig. 1d). Sequentially, analysis of the ligand revealed that in related MOFs, the 4-connected biphenyl ligand tends to adopt a Td symmetry due to the rotation of the C − C bond within the inner biphenyl group (Fig. 1b, e and Supplementary Fig. 3d). Consequently, a flu network constructed by cubic 8-connected Ni4Pz8 clusters and tetrahedral 4-connected ligands was selected (Fig. 1f). The structural modeling of BUT-78 was carried out by referring to a reported Zr-MOF with flu topology, PCN-60520 (Supplementary Fig. 3e), followed by geometric optimization and Rietveld refinement. The refined BUT-78 [Ni4(BPTP)2] crystallizes in the space group of Fmmm (a = 18.065 Å, b = 25.584 Å, and c = 19.560 Å) with satisfactory residual values (Rp = 3.78% and Rwp = 4.65%) against the experimental PXRD pattern (Supplementary Fig. 3a and Supplementary Table 1). Moreover, the molar ratio of metal ions to ligands in the activated BUT-78 sample was determined to be 2.03:1, which supports the validity of the calculated structure (see Section 3 in Supplementary Methods). The development of BUT-78 has diversified pyrazolate MOFs by appending the flu topology into the pyrazolate MOFs library. In this structure, large cavities are observed, which can accommodate a sphere of diameter ~16.0 Å with square window of edge ~8.2 Å (Fig. 1c). The total solvent-accessible volume in BUT-78 is 63% of the unit volume as estimated by PLATON. BUT-78 and PCN-605 share the same topology, while BUT-78 exhibits a reduced cavity and pore size (Fig. 1c and Supplementary Fig. 3e) relative to PCN-605, which is likely to enhance the material stability and increase its binding affinity towards guest molecules.

Fig. 1: Construction and characterization of BUT-78.
figure 1

a Ni4Pz8 cluster. b BPTP4– ligand. c Framework structure (color code: H, white; C, black; N, blue; and Ni, green). Topological representation of (d) 8-connected node, (e) tetrahedral linker, and (f) flu net. g N2 adsorption/desorption isotherms measured at 77 K and (h) PXRD patterns after different sample treatments.

Full size image

Porosity and stability

To assess the porosity of BUT-78, N2 sorption measurements were performed at 77 K. Before gas sorption experiments, the as-synthesized sample was exchanged with methanol by using Soxhlet extractor followed by evacuation under vacuum at 373 K for 6 h. The adsorption isotherm shows a N2 uptake of 21.39 mmol g–1 (STP) at the inflection point of P/P0 = ~ 0.03 (close to the theoretical amount of 23.30 mmol g–1), giving an evaluated Brunauer–Emmett–Teller (BET) area of 2031 m2 g–1 (Fig. 1g, Supplementary Fig. 4 and Supplementary Table 2). Pore size distribution was calculated from the N2 adsorption data using a nonlocal density functional theory (NLDFT) model, which reaches the maximum at around 1.50 nm, agreeing with its crystal structure (Supplementary Fig. 5).

Thermogravimetric analysis indicates that BUT-78 can be thermally stable up to 703 K (Supplementary Fig. 6). The chemical stability of BUT-78 was also investigated by immersing samples in NaOH aqueous solution (5 M) and HCl aqueous solution (pH = 2) at room temperature, as well as in boiling water for 24 h, respectively. The almost unchanged mass of samples, N2 sorption isotherms (Fig. 1g), and PXRD patterns (Fig. 1h) suggest that BUT-78 could retain structure integrity after all these treatments. In comparison, PCN-605 has negligible N2 uptake after activation, indicating framework collapse upon guest removal (Supplementary Figs. 7 and 8). Such discrepancy highlights the strong Ni(II)-N bonding of BUT-78, making it promising for applications under harsh environmental conditions.

Gas sorption

BUT-78 was examined for its VOCs adsorption properties at 298 K (Fig. 2 and source data). It exhibits a benzene uptake of 10.71 mmol g–1 at 298 K and 11.3 kPa (Fig. 2a). This represents a high benzene uptake at saturated loading pressure. It’s notable that a sharp increase in the uptake was observed at low pressure region from the benzene adsorption profile. As a result, BUT-78 shows an uptake of 5.08 mmol g–1 at 10 Pa (Fig. 2a inset), outperforming the previous benchmark benzene adsorbents (Fig. 2b)21,22,23,24,25. Adsorption isotherms of co-existing VOCs including n-hexane, cyclohexane, ethylbenzene, p-xylene, o-xylene and ethanol were also recorded for BUT-78. It exhibits high adsorption uptake at 10 Pa for these compounds: 3.80 mmol g–1 for n-hexane, 3.59 mmol g–1 for cyclohexane, 2.36 mmol g–1 for ethylbenzene, 4.58 mmol g–1 for p-xylene, 2.43 mmol g–1 for o-xylene and 0.54 mmol g–1 for ethanol, respectively (Supplementary Fig. 9). The high VOCs adsorption capacities of BUT-78 at low partial pressure suggest its potential as a smart sorbent for trace VOCs removal.

Fig. 2: Gas adsorption properties of BUT-78.
figure 2

a Benzene adsorption isotherms at 298 K (inset illustrates the 0~1.0 kPa region). b Benzene uptakes at 10 Pa compared with representative porous sorbents at low relative pressure (see also Supplementary Table 5). c SO2 and CO2 sorption isotherms at 298 K. d IAST calculation of binary adsorption isotherms for SO2/CO2 gas mixtures (10/90).

Full size image

In terms of harmful and corrosive gas adsorption, BUT-78 demonstrates a remarkable SO2 uptake of 13.8 mmol g–1 at 298 K and 101.4 kPa (Fig. 2c), which exceeds some leading adsorbents such as MFM-190(NO2) (12.7 mmol g–1)26, MFM-601 (12.3 mmol g–1)27, SIFSIX-1-Cu (11.0 mmol g−1)28 and Mg-gallate (5.8 mmol g–1)29. More importantly, it shows a SO2 uptake of 1.10 mmol g–1 at a particularly low pressure of 0.4 kPa and 298 K, indicating a high capacity of SO2 capture can be achieved at pressure as low as the atmospheric level. CO2 is a competing adsorbate in air with SO2. BUT-78 shows a CO2 uptake of 3.81 mmol g–1 at 298 K and 100.1 kPa (Fig. 2c). An ideal adsorbed solution theory (IAST) SO2/CO2 (10/90) selectivity of 101 could be obtained at 298 K and 100 kPa (Fig. 2d, Supplementary Fig. 10 and Supplementary Table 3), which is also superior to some leading adsorbents (Supplementary Table 4)30,31,32,33. These results suggest that BUT-78 holds the potential of selective SO2 capture from industrial effluent streams.

The isosteric heat (Qst) for benzene and SO2 adsorption in BUT-78 was estimated by fitting the Clausius-Clapeyron equation to the adsorption isotherm of each gas at different temperatures (Supplementary Figs. 11–13). The Qst was calculated to be 33.2 and 39.0 kJ mol–1 for benzene and SO2 at the loading of 1.60 mmol g–1 (Supplementary Table 5 and 6, Supplementary Figs. 14 and 15), suggesting a strong affinity between BUT-78 and benzene/SO2.

Dynamic breakthrough experiments

To evaluate the performance of BUT-78 in capturing trace benzene and SO2, we conducted dynamic gas breakthrough experiments under ambient conditions (Supplementary Fig. 16, Fig. 3 and source data). First, benzene breakthrough experiments were carried out at 298 K with a mixture gas stream (10 ppm benzene) passing through a column packed with ~10 mg BUT-78 at a gas flow rate of 10 mL min–1. As shown in Fig. 3a, benzene broke through the BUT-78 column after ~44 h (4400 h g–1), corresponding to a benzene capture capacity of ~1.18 mmol g–1. Regeneration experiments revealed that the adsorbed benzene molecules in BUT-78 could be efficiently removed at 353 K within one hour (Supplementary Fig. 17). Practically, trace benzene capture is implemented in the presence of humidity, which may impact the performance of most adsorbents. Breakthrough experiments were also performed on BUT-78 under humid conditions22. Under the relative humidity (RH) of 30% and 50%, BUT-78 exhibited a breakthrough time of ~3000 h g–1 (~0.80 mmol g–1) and ~1050 h g–1 (~0.28 mmol g–1), suggesting the feasibility of trace benzene capture by BUT-78 under humid conditions (Fig. 3a and Supplementary Fig. 18).

Fig. 3: Dynamic breakthrough experiments over BUT-78 at 298 K.
figure 3

a Benzene (10 ppm) breakthrough curves under dry and 30% RH conditions. b Breakthrough curves of simulated flue gas (containing 2500 ppm SO2) stream under dry and 30% RH conditions. c Simultaneous benzene (10 ppm) and SO2 (250 ppm) breakthrough curves at 30% RH. d Comparison of dynamic benzene and SO2 capture performance between BUT-78 and some leading sorbents at 30% RH.

Full size image

SO2 breakthrough experiments were conducted with a mixture gas stream containing 2500 ppm SO2. BUT-78 revealed a SO2 capture capacity of ~16 h g–1 (~0.86 mmol g–1) (Supplementary Fig. 19), which is close to the static SO2 uptake of ~0.94 mmol g–1 at ~0.25 kPa as observed from the isotherm (Fig. 2c). Next, we simulated the model desulfurization process by conducting breakthrough experiments with a quadruple simulated flue gas mixture containing 2500 ppm of SO2, 13.5% CO2, 4% O2 and 82.25% N234. CO2, O2 and N2 broke through the column immediately, and the retention of SO2 was calculated to be ~14.5 h g–1 (~0.78 mmol g–1) (Fig. 3b). After the breakthrough of SO2, complete regeneration of BUT-78 was achieved within 1.5 h by heating the packed bed at 353 K (Supplementary Fig. 20). We also investigated desulfurization of wet simulated flue gas (Fig. 3b and Supplementary Fig. 21). When the RH increased from 30% to 50% and 80%, a SO2 capture capacity of ~13.5 h g–1 (~0.72 mmol g–1), ~13.0 h g–1 (~0.70 mmol g–1) and ~13.6 h g–1 (~0.73 mmol g–1) was recorded on BUT-78, respectively. Recycling experiments revealed that BUT-78 showed no obvious capacity loss after three successive breakthrough cycles at 30% RH (Supplementary Fig. 22).

BUT-78 was then evaluated for its practical performance in simultaneously capturing these two pollutants from industrial exhausts. The capture capacity of ~0.51 mmol g–1 (~1900 h g–1) for 10 ppm benzene and ~1.21 mmol g–1 (~180 h g–1) for 250 ppm SO2 were achieved under 30% RH (Fig. 3c). Under 50% RH, BUT-78 revealed a lower benzene capture capacity of 0.20 mmol g–1 (~750 h g–1) and a higher SO2 capacity of ~4.19 mmol g–1 (~625 h g–1) (Supplementary Fig. 23). BUT-78 maintains the performance of simultaneous benzene and SO2 capture under humid conditions, and the humidity is even able to improve the capacity of SO2 capture by partially forming H2SO335. This is supported by the characteristic FT-IR peak of O–H at 3235 cm–1 and S–O at 970 cm–1 in sulfite of the post-breakthrough sample (Supplementary Fig. 24). Recycling breakthrough tests revealed that BUT-78 showed no loss of both benzene and SO2 capture capacity after five successive cycles (Supplementary Fig. 25). Moreover, the structure of BUT-78 remained intact after breakthrough experiments as evidenced by unaltered N2 sorption isotherms and PXRD patterns against the pristine sample (Supplementary Fig. 26). Reports on simultaneous VOCs and SO2 capture are rare. Sun and co-workers have investigated activated cokes for their combined adsorption of toluene and SO2, with the optimized hierarchical material showing high capacity, although its performance at low concentrations has not been thoroughly studied8. It’s worth noting that most SO2 physisorbents have been examined by breakthrough experiments where gas streams usually contain 2500 ppm of SO2. High uptake at lower concentration of 250 ppm, especially in the presence of moisture, requires stronger binding with SO2. The only report on MOF sorbents is KAUST-7, which exhibits low capacity (0.11 mmol g–1)36. In the present research, BUT-78 has demonstrated good performance (1.21 mmol g–1) for 250 ppm of SO2 even under 30% RH. It’s also capable of simultaneous capture of other VOCs and SO2, such as cyclohexane and ethanol, with a capacity of ~0.62 mmol g–1 (~2300 h g–1) for 10 ppm cyclohexane and ~0.59 mmol g–1 (~88 h g–1) for 250 ppm SO2 (Supplementary Fig. 27), and ~0.30 mmol g–1 (~1100 h g–1) for 10 ppm ethanol and ~1.14 mmol g–1 (~170 h g–1) for 250 ppm SO2 (Supplementary Fig. 28).

In order to benchmark the performance of BUT-78, we compared it with other representative porous materials featuring high benzene uptakes including BUT-5522, BUT-6623, HKUST-137, microporous activated carbon (AC)23, and MCM-4123, and those with high SO2 uptakes, such as SIFSIX-1-Cu28. These samples were reproduced according to the reported procedures, and then tested (10 ppm benzene and 250 ppm SO2) for dynamic breakthrough experiments at 30% RH. They exhibit trace benzene and SO2 capture capacity as follows: BUT-55 (0.48 and 0.15 mmol g–1), BUT-66 (0.03 and 0.66 mmol g–1), HKUST-1 (0.02 and 0.01 mmol g–1), microporous AC (0.41 and 0.01 mmol g–1), MCM-41 (0.01 and 0.01 mmol g–1) and SIFSIX-1-Cu (0.03 and 0.13 mmol g–1) (Supplementary Figs. 29–34). BUT-78 (0.51 and 1.21 mmol g–1) outperforms these materials with record-breaking performance (Fig. 3d). In addition, N2 sorption isotherms suggest decreased porosity of BUT-55 and HKUST-1 after SO2 exposure during breakthrough experiments, where BUT-66 and SIFSIX-1-Cu almost lose their porosity (Supplementary Figs. 35–38). BUT-78 therefore sets a benchmark for trace binary pollutants removal from industrial exhaust gas, with high capacity and good recyclability.

Benzene and SO2 binding sites

To gain insights into the underlying structure-property relationship, Grand Canonical Monte Carlo (GCMC) and Density Functional Theory (DFT) calculations were performed. Upon trapping benzene or SO2 molecules within the pores of BUT-78, analysis of the center of mass (COM) probability density distributions revealed a predominant localization in close proximity to the Ni4 clusters (Fig. 4a, b and Supplementary Fig. 39). Calculation of radial distribution functions (RDF) derived from averaging over the GCMC configurations helped determine the specific adsorption sites (Supplementary Fig. 40 and Supplementary Data 1). Benzene molecules mainly interact with the framework through hydrogen atoms (Hbenzene) (Fig. 4c). Such interaction is observed with two hydrogen acceptors: (i) the nitrogen atoms (NPz) of pyrazolate, evidenced by a relatively short mean distance of d(Hbenzene−NPz) = 2.98–3.17 Å, and (ii) the Ni(II) ions at an average distance of d(Hbenzene−Ni) = 3.40 Å. Meanwhile, the adsorption of SO2 molecules primarily occurs through electrostatic interactions between OSO2 and Ni(II) at a distance of 2.49 Å, as well as multiple dipole-dipole interactions between OSO2 and HPz at a distance of 2.61–2.84 Å (Fig. 4d). In situ FT-IR spectra of BUT-78 demonstrated the residing/leaving of SO2 in/from the framework during adsorption/desorption, evidenced by the characteristic peaks at 1330 and 1140 cm−1, respectively. A red shift was observed relative to free SO2 gases due to the interaction with accessible Ni(II) sites, which is in agreement with the previous reports (Supplementary Figs. 41 and 42)35. Furthermore, the identified adsorption sites for benzene and SO2 are calculated with binding energies of −64.5 kJ mol−1 and −70.6 kJ mol−1, respectively, aligning with their experimental Qst values.

Fig. 4: Benzene and SO2 binding sites exploration.
figure 4

Contour plot of center of mass (COM) probability density distributions for (a) benzene and (b) SO2 adsorbed in BUT-78 at 298 K and 101 kPa (view along the y axis). DFT calculated binding sites for (c) benzene and (d) SO2 (color code: H, white; C, black; N, blue; O, red; S, yellow; and Ni, green).

Full size image

The coordination geometry and electrostatic properties of accessible Ni(II) sites in BUT-78 are different from the typical OMSs as found in HKUST-138, MOF-7439, and MIL-100/10140,41. Because pyrazolates belong to strong field ligands, Ni(II) prefers to coordinate with pyrazolate ligands in a square planer dsp2 configuration (the thermally stable state). H2O, as a weak field ligand, is not favored to coordinate with Ni(II) ions to upset the preferential configuration17. This is supported by the single crystal structure of as-synthesized BUT-32 without coordinated water molecules to the Ni4 cluster observed and the hydrophobic nature of MOFs with Ni4 clusters at low RH revealed by their water vapor sorption isotherms (Supplementary Figs. 43−48)17,18. As vacant Lewis acid sites, their strong interactions with benzene and SO2 are maintained. These factors account for the weaker affinity of Ni4 clusters toward H2O but strong binding to benzene and SO2 of BUT-78. It should be noted that the competition on these sites will occur between benzene and SO2. Since SO2 exhibits slightly higher binding energy (−70.6 kJ mol−1) with the framework than benzene (−64.5 kJ mol−1), the adsorption of SO2 is preferential during their simultaneous adsorption. All these attributes have enabled the simultaneous capture of trace benzene and SO2 under humid conditions in this MOF.

Discussion

In summary, we present the reticular construction of a robust flu-topology pyrazolate MOF, BUT-78. Featuring rich accessible Ni(II) sites and suitable pores, BUT-78 advanced the benzene uptake to 5.08 mmol g–1 at 10 Pa, setting benchmark for trace benzene capture. Interestingly, the Ni(II) sites in this MOF feature strong affinity toward various guests but less vulnerable to water. By virtue of this, BUT-78 enables simultaneous capture of trace benzene and SO2 under humid conditions, which has not been achieved in porous sorbents yet. This research highlights the strength of reticular expansion of MOFs to access smart sorbents, whose desired structure and properties would offer facile solutions to the remediation of industrial effluent streams containing multiple hazards.

Methods

Chemicals and materials

All commercial chemicals were purchased without further extra purification. (3,5-dibromophenyl) boronic acid (98%), 1-(tetrahydro-2H-pyran-2-yl)-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1H-pyrazole (98%), Pd(PPh3)4 (98%) and maleic acid (99%) were purchased from Bide Pharmatech Co., Ltd. N,N-dimethylformamide (DMF, 99.9%), 1,4-dioxane (99.8%), K2CO3 (99%), acetic acid (99.5%) and Ni(OAc)2·4H2O (99.9%) were purchased from Beijing InnoChem Science & Technology Co.,Ltd. Benzene (99.7%), n-hexane (99.5%), cyclohexane (99.5%), ethylbenzene (99%), p-xylene (99%), o-xylene (99%), ethanol (99.9%) were purchased from J&K Scientific Ltd. Liquid nitrogen (99.99%), N2 (99.99%), CO2 (99.99%), SO2 (99.99%), calibration gas (5000 ppm of SO2, 27% of CO2, 8% O2 and 64.5% N2) and calibration gas (~20 ppm benzene balanced with dry air) were purchased from Beijing Beiyang United Gases Co.,Ltd.

Synthesis of [Ni4(BPTP)2] (BUT-78)

H4BPTP (30 mg, 0.072 mmol) was dissolved in 10 mL of N,N-dimethylformamide (DMF) under ultrasound in a 40 mL high-pressure vessel. To the solution acetic acid (30 μL) and Ni(OAc)2·4H2O (30 mg, 0.12 mmol) were added. The vessel was then tightly sealed and the mixture was heated in an oil bath of 150 °C for 3 h with stirring. After cooling to room temperature, the yellow solid (18.8 mg of activated sample, 40% yield based on the H4BPTP) was collected by filtration, and washed with DMF (2 × 10 mL), methanol (3 × 10 mL) and dried under reduced pressure at room temperature. Elemental analysis for BUT-78: Calc. C 54.20, H 2.63, N 21.08. Found. C 52.42, H 2.95, N 22.71.

Gas/vapor sorption measurements

MOF powder samples (about 100 mg for each) were soaked in 20 mL of DMF for 24 h at 60 °C. Next, DMF was decanted and the samples were transferred to the Soxhlet extractor containing 300 mL of methanol and then heated at 100 °C for two days. After solvent exchange, the samples were collected by centrifugation and dried on air. Before gas/vapor sorption measurements, MOF samples were loaded in a sample tube and degassed for 6 h at 100 °C on a BELPREP VAC II vacuum degasser. The N2 sorption measurements for BET analysis were conducted at 77 K in a liquid N2 bath. SO2, CO2 and vapor isotherms were measured using a temperature-controlled water bath. In cycling gas sorption measurements, the samples were degassed at an optimal temperature of 100 °C for 2 h for next run. Pore size distribution data were also calculated from the N2 adsorption isotherms at 77 K based on a nonlocal density functional theory (NLDFT) model in the BELMaster7 software package.

Breakthrough experiments

The breakthrough experiments were performed using a typical dynamic gas breakthrough equipment at ambient conditions. In a typical SO2 breakthrough experiment, degassed MOF powder sample was packed into a quartz glass column (the column was 80 mm in length with 4 mm internal diameter), which was activated in situ in the column under a flow of N2 (10 mL min–1) at 100 °C for 6 h. For trace SO2 capture experiments, a gas mixture was selected according to the declaration of the U.S. Department of Energy to simulate flue gas: 2500 ppm of SO2, 13.5% of CO2, 4% O2 and 82.25% N234. For the dry gas breakthrough experiment, a mixture of the calibration gas (5000 ppm of SO2, 27% of CO2, 8% O2 and 64.5% N2) and N2 (1:1 v/v) was passed through the packed tube at a flow rate of 8 mL min–1. The humid gas breakthrough experiment was similar to the dry gas breakthrough experiment except that the N2 flow was pre-saturated with water vapor by being bubbled through deionized water with desired temperature before being mixed with the calibration gas flow (Supplementary Fig. 16).

In a typical benzene breakthrough experiment, a calibration gas (~20 ppm benzene balanced with dry air) was used. 10 mg MOF powder sample was packed into a quartz glass column. Other testing methods are as the same as that for SO2. In a typical SO2 and benzene breakthrough experiment, a mixture of the calibration benzene gas (~20 ppm balanced with dry air), SO2 (~5000 ppm balanced with dry air) and air was passed through the packed tube at a flow rate of 10 mL min–1.

The flow rate of the gases was controlled by a mass flow controller (Alicat). The gas compositions at the outlet was determined continuously by mass spectrometry (MS, Hiden, HPR-20). After the contents of outlet gas reached the equilibrium, the adsorption bed was regenerated by He flow (10 mL min–1) for 2 h at 353 K.