Introduction

After the industrial revolution, the concentration of CO2 in the atmosphere has been steadily increasing because of the continuous consumption of fossil fuels. This can cause global warming and climate change, and consequently lead to environmental disaster1. Carbon capture and sequestration technologies have been implemented to deal with the massive CO2 emissions from coal-fired power plants2,3,4. Among capture technologies, alkanolamine solution scrubbers are one of the most developed systems; however, they are affected by corrosiveness and amine evaporation and have a high energy penalty for regeneration5,6. To address these limitations, noncorrosive solid materials with lower heat capacity have attracted considerable attention as alternative adsorbents7,8,9.

Metal–organic frameworks (MOFs) with high surface area and tunable pores have been shown to possess superior CO2 adsorption capacities compared with ordinary solid adsorbents such as zeolites, carbons, and amine-incorporated porous materials10,11,12,13,14,15. In particular, expanded MOF-74 type frameworks, M2(dobpdc) (M = Mg, Mn, Fe, Co, Ni, Zn; dobpdc4− = 4,4′-dioxidobiphenyl-3,3′-dicarboxylate), exhibit exceptional CO2 adsorption performances because of their high-density open metal sites that are functionalized by diamines, alcoholamines, and alkoxyalkylamines16,17,18,19,20,21. Among the series, diamine-functionalized Mg2(dobpdc) with record-high adsorption capacities has been proven to be one of the most promising adsorbents for the post-combustion CO2 capture process, while the shape of the isotherm and the ease with which the CO2 can be removed play a much important role from a process point of view22,23. Its adsorption and desorption behaviors can be modulated by selecting the appropriate amine structure24,25,26,27,28. To improve the structural durability of the MOFs under humidification conditions, cyclic diamines or polyamines have been grafted onto the open metal sites of the Mg2(dobpdc) platform29,30. Coating the platforms with hydrophobic polymers also enhanced the structural stability in the presence of moisture31. Currently, diamine-grafted Mg2(dobpdc) materials are prepared in the powered form, which is not suitable for CO2 capture processes because of disadvantages such as pressure drop, poor reusability and processability. In fact, MOF powders are shaped for practical applications via mechanical shaping including granulation, extrusion, pressing, and spray drying, template shaping including sol–gel method and electrospinning, and other shapings32,33. Particularly, spray drying provides a cost-effective, rapid, and scalable method to continuously fabricate the shaped materials34. Because diamine-grafted Mg2(dobpdc) compounds have been extensively explored due to their exceptional CO2 adsorption capacities for post-combustion capture processes16,17,18,19,20,24,25,27,28,29,30,31,35,36, shaping of the MOF powders is important to increase their processibility and mechanical strength for real-world CO2 capture applications37,38. Furthermore, to overcome the intrinsic limitations of pristine shaped MOFs such as CO2 adsorption capacity and structural stability under humid conditions, their postsynthetic modifications are required but nearly unexplored.

In the present work, we report the scalable synthesis of Mg2(dobpdc) as well as Mg2(dobpdc)-alumina composite microbeads (MOF/Al) via the spray-dried shaping of MOF with alumina sol. The attrition index of the beads used was <0.6%, which meets the desirable mechanical property for the CO2 capture process using solid sorbents. The shaped MOFs were further functionalized by N-ethylethylenediamine (een) and then coated by silanes to obtain een–MOF/Al–Si. The post-functionalized microbeads exhibited a substantial CO2 capacity, >11 wt%, and improved long-term material stability under humid conditions.

Results and discussion

Synthesis and characterizations of MOF/Al microbeads

The dimethylformamide (DMF)-coordinated Mg2(dobpdc) was synthesized by H4dobpdc and MgCl2·6H2O in DMF/EtOH under a solvothermal reaction using a 300 mL reactor. We also successfully employed a 35 L steel bomb reactor to produce the analytically pure solid product in a kg scale. This was used for the MOF shaping. The Mg2(dobpdc) sample was characterized by powder X-ray diffraction (PXRD), N2 isotherms, and scanning electron microscopy (SEM) (for PXRD, N2 isotherms, and SEM see Supplementary Fig. 1). Fully dried Mg2(dobpdc) rod-shaped particles were ground to ensure the uniform distribution of the particles. The ground sample was mixed with 10 wt% alumina sol as a binder and was further treated with the ball-milling process for 30 min to obtain an aqueous slurry. The slurry of Mg2(dobpdc) with alumina sol was injected into a spray drier to obtain DMF–MOF/Al spherical beads in the particle size range 30–70 µm (Fig. 1 and for SEM image see Supplementary Fig. 2). The SEM images and SEM-energy dispersive X-ray spectroscopy (SEM-EDS) data exhibited that the MOF rods and alumina particles were well distributed over the surface and the interior of the DMF–MOF/Al bead (Fig. 2c–f and for SEM-EDS mapping see Supplementary Fig. 3). The PXRD data revealed that the structure of the DMF–MOF/Al beads was consistent with that of Mg2(dobpdc) (Fig. 2a and for PXRD see Supplementary Fig. 4). The infrared (IR) spectrum of the DMF–MOF/Al sample showed additional peaks at 1070 cm−1, which are characteristic of Al–O stretching, and at 3330 and 3100 cm−1, which are attributed to O–H stretching. This indicates the existence of alumina in the bead (for IR spectra see Supplementary Fig. 5). ICP-AES analysis revealed that the ratio of Mg to Al in the DMF–MOF/Al was 1:0.363 (for ICP-AES results see Supplementary Table 1). The binding energies of Al 2p and 2s orbitals in the XPS data of the DMF–MOF/Al beads also indicated the successful incorporation of the alumina binder into the spherical bead (for XPS data see Supplementary Fig. 6)39. For practical applications, the shaped material should have the appropriate mechanical strength because the collisions of the beads with each other and the surface of the adsorption bed during the capture process will break the shaped particles into the powdered form. An attrition test was conducted to examine the mechanical properties of the microbeads. The attrition index was determined to be ~0.6%, indicating that alumina-bound spheres are sufficiently robust for application in the CO2 capture process40,41.

Fig. 1: Schematic of the spray-drying process of MOF/Al microbeads.
figure 1

a Scale-up synthesis of Mg2(dobpdc) using 300 mL and 35 L steel bomb reactors. b Mixing Mg2(dobpdc) and alumina sol binders through a ball-milling process. c Shaping of MOF/Al microbeads via spray dry method. d Functionalization of MOF/Al microbeads by N-ethylethylenediamine (een).

Fig. 2: Characterizations of MOF/Al microbeads and diamine-grafted MOF/Al microbeads.
figure 2

a PXRD data of DMF–MOF, DMF–MOF/Al, and een–MOF/Al. b N2 isotherms for Mg2(dobpdc), MOF/Al, and een–MOF/Al at 77 K. c SEM image of the surface of DMF–MOF/Al. d Enlarged SEM image of the surface of DMF–MOF/Al. e SEM image of the inside of MOF-MOF/Al. f Enlarged SEM image of the inside of DMF–MOF/Al.

To confirm the porous properties of DMF–MOF/Al beads, N2 sorption was recorded at 77 K. The Brunauer–Emmett–Teller surface areas of Mg2(dobpdc) and MOF/Al were 3178 and 1656 m2 g−1, respectively (Fig. 2b and for porous properties see Supplementary Table 2). The reduction in the surface area of MOF/Al indicates that the pore channel is partially blocked by the alumina binders. This is supported by the decreased pore volume and the unchanged pore channel size (for pore size distribution see Supplementary Fig. 7).

Preparation and characterizations of diamine-grafted MOF/Al microbeads

The functionalization of Mg2(dobpdc) with diverse diamines enhances CO2 adsorption characteristics that are suitable for post-combustion CO2 capture applications35. To increase the CO2 uptake, we grafted MOF/Al with N-ethylethylenediamine (een). Before the functionalization of een, DMF–MOF/Al was soaked in MeOH for 3 days at room temperature to replace the DMF molecules coordinated with the open metal sites with the readily removable MeOH. The absence of the C=O stretching vibration of DMF at 1663 cm−1 in the IR spectrum suggests the complete removal of DMF from the bead sample (for IR spectra see Supplementary Fig. 5). The SEM images and XPS data show that the shape and composition of MOF/Al were maintained after MeOH soaking (for SEM and XPS data see Supplementary Figs. 2 and 6). We treated the MeOH-soaked sample (MeOH-MOF/Al) with 50 equivalents of een in n-hexane for 12 h at 50 °C without stirring to obtain an een-functionalized phase, een–MOF/Al. After een grafting, the structure of Mg2(dobpdc) remained intact, and the morphology of the beads was identical to that of MOF/Al (for SEM image see Supplementary Fig. 2). N–H stretching vibration at 3300 cm−1 was observed in the IR spectrum, and the N1s peak was observed at 400 eV in the XPS profile. This indicates the existence of een in the sample (for IR and XPS data see Supplementary Figs. 5 and 6).

From the N2 isotherms of een–MOF/Al, the surface area was calculated as 611 m2 g−1 (Fig. 2b and for surface area see Supplementary Table 2). Because the diamine was grafted onto the open metal sites of the framework, the surface area of een–MOF/Al was smaller than that of MOF/Al, and its pore size decreased from 18.4 to 14.8 Å (for pore size distribution see Supplementary Fig. 7). The CO2 isotherms of een–MOF/Al had a CO2 uptake of 3.48 mmol g−1 at 150 mbar CO2, which is similar to that of een–MOF (for CO2 isotherms see Supplementary Figs. 8 and 9)28. This indicates that using alumina nanoparticles for shaping binds the MOF rods, while keeping the pores open for CO2 diffusion. The CO2 uptake of the sample was negligible at 1 bar CO2 at 140 °C, implying that CO2 can be desorbed even under 100% CO2 at that temperature. The complete desorption of the adsorbed CO2 is important for the repeated reuse of the adsorbent for real-world post-combustion applications. The step pressure in the isotherms of the MOF beads suggests the mechanism of operation of CO2 insertion, which was observed in diamine-functionalized Mg2(dobpdc) series35. We calculated the isosteric heat of the adsorption (Qst) using the dual-site Langmuir–Freundlich (L–F) and Clausius–Clapeyron equations (for isosteric heats of CO2 adsorption see Supplementary Fig. 10). The −Qst of een–MOF/Al was in the range 69–82 kJ mol−1, attributed to the chemisorption.

Introducing hydrophobic silane into the MOF/Al

Flue gas emitted from coal-fired power plants contains 5–7% water vapor2. Hydrophilic materials can readily adsorb water vapors; however, the desorption of the adsorbed water molecules requires high energy input42. Thus, a potential adsorbent for CO2 capture should also be hydrophobic to preclude the access of water vapors to the pores of the material. To examine the surface properties of DMF–MOF/Al and een–MOF/Al, we measured the contact angles using a water droplet. No contact angle was observed with MOF/Al and een–MOF/Al, indicating that the samples were hydrophilic (Fig. 3a, b and for contact angles see Supplementary Fig. 11). To enhance hydrophobicity, silanes H–(CH2)n–Si(OMe)3 with chains of different numbers of carbon atoms were introduced into the een–MOF/Al beads. When n = 3, there was no contact angle, and the hydrophilic characteristic remained (for contact angles see Supplementary Fig. 12). By increasing n, a high contact angle (>105°) was obtained; however, it disappeared quickly. Remarkably, when octadecyltrimethoxysilane with n = 18 was used to obtain the silane-coated product, een–MOF/Al–Si, a contact angle >100° was maintained for 30 min (Fig. 3c, d). It is clear that a longer carbon chain with silane increases the contact angle of the sample because of the increased bulkiness and hydrophobicity. The PXRD pattern of een–MOF/Al–Si agreed with that of een–MOF/Al, indicating that the MOF structure was well-maintained during the silane coating (for PXRD data see Supplementary Fig. 13). In the IR spectra, the C–H stretching vibrations at 2850 and 2950 cm−1 were assigned to the long carbon chains of the silane moiety (for IR spectra see Supplementary Fig. 14). Both the SEM images and SEM-EDS data showed that the shape of the spherical beads was maintained and that the silanes were well distributed over the surface and interior of een–MOF/Al–Si (Fig. 3e–h and for SEM-EDS mapping see Supplementary Fig. 15). This is also corroborated by the XPS results in which Al and Si peaks were visible (for XPS data see Supplementary Fig. 16). The N2 isotherm of een–MOF/Al–Si at 77 K exhibited a surface area of 17 m2 g−1 (for N2 isotherm see Supplementary Fig. 17). The silane coating of een–MOF/Al can obstruct the diffusion of N2 gas by blocking the pore channels with the long carbon chain. This significantly reduces the surface area of een–MOF/Al–Si. Furthermore, the CO2 isotherm was collected at 195 K (for CO2 isotherm see Supplementary Fig. 18). The isotherm showed that een–MOF/Al–Si adsorbs CO2 at that low temperature. This result suggests that CO2 can be diffused into the MOF channel despite the interference of the long carbon chain on silane, which is due to the smaller size of CO2 than N2.

Fig. 3: Characterizations of een–MOF/Al–Si.
figure 3

a Contact angle of een–MOF/Al. b Photograph of een–MOF/Al after dropping water. c Contact angle of een–MOF/Al–Si. d Photograph of een–MOF/Al–Si after dropping water. e SEM image of een–MOF/Al–Si. fh SEM-EDS mappings of een–MOF/Al–Si.

We supposed that silanes can react with the hydroxy groups on the alumina binders and with the non-coordinating functional groups of the ligands on the surface of MOF in een–MOF/Al via condensation among methoxy, hydroxyl, and carboxylic groups43. To prove this hypothesis, we introduced octadecyltrimethoxysilane to een–MOF and alumina under the same conditions applied for bead coating. The silane-coated een–MOF (een–MOF–Si) and alumina (alumina–Si) showed hydrophobicity with contact angles of 63° and 141°, respectively (for contact angles see Supplementary Fig. 19). The characteristic peaks were positioned at 2850 and 2950 cm−1 in the IR spectra of silane-coated alumina and een–MOF, which can be assigned to the C–H stretchings in silane (for IR spectra see Supplementary Fig. 20). Furthermore, the XPS analysis revealed that Si peaks were observed in een–MOF–Si and alumina–Si, indicating the presence of silane moiety in the samples (for XPS data see Supplementary Figs. 21 and 22). This demonstrates that the silanes are coated on the alumina surface as well as the MOF surface in een–MOF/Al–Si.

CO2 adsorption–desorption behaviors of een–MOF/Al and een–MOF/Al–Si

The CO2 adsorption curves were measured at different temperatures to evaluate the CO2 adsorption performance of een–MOF/Al–Si (Fig. 4a). The CO2 uptake was 3.15 mmol g−1 at 150 mbar, which is still significant and only slightly lower than that of een–MOF/Al. The step pressure implies that the CO2 insertion mechanism is operative even after the introduction of silane. For een–MOF/Al–Si, the step moved to a higher pressure because long carbon chains on silane slightly disturb CO2 diffusion into the MOF channel at low partial CO2 pressures (for CO2 isotherms see Supplementary Fig. 9). To confirm a selectivity of CO2/N2, N2 isotherms were obtained at the same temperature ranges as the CO2 isotherms (for N2 isotherms see Supplementary Fig. 23). We calculated the selectivity using formula, \(S = ( {{q}_{\mathrm{co}_2}/{q}_{\mathrm{N}_2}} )/( {P_{\mathrm{co}_2}/P_{\mathrm{N}_2}})\), where qi and pi are the adsorption capacity and the partial pressure, respectively, of component i. The selectivity at 15% CO2 and 75% N2 corresponded to 53,182 at 40 °C, 1260 at 60 °C, 595 at 80 °C, 523 at 100 °C, 5 at 120 °C, and 3 at 140 °C (for selectivity see Supplementary Table 3). The tendency shows that the selectivity decreased as the temperature increased. The selectivity of 595 at the adsorption temperature of 80 °C indicates that the presence of N2 could be ignorable during adsorption of CO2. The calculated Qst of een–MOF/Al–Si was in the range 71–73 kJ mol−1, which is similar to that of een–MOF/Al (for isosteric heats of CO2 adsorption see Supplementary Fig. 10). For een–MOF/Al–Si, we observed the change in ΔT = Tdes − Tads with the adsorption temperature (Tads) at the desorption temperature (Tdes) of 130 and 140 °C, respectively, under 2.5 and 15% CO2 (for correlation diagram see Supplementary Fig. 24). At Tdes = 140 °C, the working capacity increased from 11.1 to 12.6 wt%, even under a low pressure of 2.5% CO2 as ΔT increased from 50 to 100 °C. The observed working capacity was in the upper bound of the CO2 capacity observed for solid adsorbents7,35. We investigated the adsorption capacities of een–MOF/Al–Si using Thermogravimetric analyses (TGA) at different CO2 concentrations and temperatures as a function of time (for adsorption capacities see Supplementary Fig. 25). To display adsorption rate clearly, we differentiated CO2 uptake curve with respect to time (for adsorption rate see Supplementary Fig. 26). From the differentiated data, CO2 adsorption rate increased as the CO2 partial pressure increased and the temperature decreased. For example, adsorption rates at 40 °C changed from 4.57 wt% min−1 at 2.5% CO2 to 22.6 wt% min−1 at 15% CO2 to 53.72 wt% min−1 at 100% CO2. At 15% CO2, the adsorption rate of 22.61 wt% min−1 at 40 °C was significantly reduced to 0.15 wt% min−1 at 140 °C. These results unveiled that een–MOF/Al–Si has a high adsorption rate even in the form of beads28,36,44. We also performed temperature swing adsorption (TSA) cycle tests of een–MOF/Al–Si to evaluate the reusability (for TSA cycles see Supplementary Fig. 27). Adsorption was conducted at 15% CO2 and 80 °C for 5 min and desorption took place at 100% CO2 and 140 °C for 1 min. After 100 cycles, the adsorption amount of een–MOF/Al–Si was maintained without a capacity loss. Thus, een–MOF/Al–Si is a promising adsorbent with high working capacity, high adsorption rate, and reusability.

Fig. 4: Gas sorption behaviors and long-term stability test under humid condition.
figure 4

a Adsorption isotherms of CO2 for een–MOF/Al–Si at various temperatures. b Water vapor isotherms of een–MOF, een–MOF/Al, and een–MOF/Al–Si. c CO2 adsorption capacity of een–MOF, een–MOF/Al, and een–MOF/Al–Si after exposure to ambient air by using TGA. d CO2 adsorption capacity of een–MOF, een–MOF/Al, and een–MOF/Al–Si after exposure to 10% H2O and 90% CO2 at 140 °C by using TGA. The flow rate was 50 cc min−1.

CO2 adsorption mechanisms of een–MOF/Al and een–MOF/Al–Si

To examine the CO2 adsorption mechanism of een–MOF/Al and een–MOF/Al–Si, we performed in situ IR spectroscopy under CO2 and plotted the results as a function of time (for IR spectra see Supplementary Figs. 28 and 29). The samples were activated at 140 °C for 2 h under N2 and then exposed to a stream of 100% CO2 at 40 °C. For both samples, new peaks emerged at 3429, 3408, 1651, and 1319 cm−1 after exposure for 1 min. The first two peaks (3429 and 3408 cm−1) were attributed to the N–H stretchings, and the other two (1651 and 1319 cm−1) were assigned to C–O and C–N vibrations, respectively. This reveals that carbamate species (R–NH–CO2) is generated in een–MOF/Al and een–MOF/Al–Si, which is in good agreement with the CO2 insertion mechanism observed in diamine-functionalized Mg2(dobpdc) frameworks17,28. Furthermore, we acquired the solid-state 13C NMR spectra of een–MOF/Al and een–MOF/Al–Si before and after CO2 uptake. In the spectra of een–MOF/Al–Si, the carbons on een were observed in the range 35–56 ppm, carbons on octadecylsilane at 31 ppm, benzene carbons of dobpdc4− in the range 120–133 ppm, benzene carbon adjacent to oxygen at 167 ppm, and carboxylate carbon at 174 ppm. After CO2 adsorption, the positions of the carbons in the carboxylate group and een groups changed slightly, indicating that these groups were affected by CO2 insertion due to the proximity to the insertion site. Moreover, a new peak appeared at 163 ppm, which is attributed to the carbamate carbon generated after CO2 adsorption (for solid-state 13C NMR spectra see Supplementary Fig. 30)45. Meanwhile, the peaks related to the carbons of the silane were mostly unchanged, suggesting that the silane does not affect the CO2 adsorption mechanism. In the solid-state 15N NMR spectra of een–MOF/Al and een–MOF/Al–Si, two peaks were observed at 15.1 and 33.7 ppm; these were assigned to the two nitrogen atoms of een. After CO2 uptake, the peaks shifted to 40.3 and 74.6 ppm, corresponding to ammonium and carbamate nitrogen atoms, respectively. The nuclear magnetic resonance (NMR) results revealed that the adsorption mechanism involves even after the MOF shaping and silane coating (for solid-state 15N NMR spectra see Supplementary Fig. 31).

Long-term stability test under humid conditions

The water vapor isotherms of een–MOF, een–MOF/Al, and een–MOF/Al–Si were collected at 25 °C (Fig. 4b). The water uptake of een–MOF/Al–Si was lower than that of een–MOF and een–MOF/Al, indicating that the silane coating makes the material more hydrophobic by impeding the access of water molecules. To examine the effect of coating on the structural stability and CO2 performance31, we exposed the samples to ambient air for 15 days. The CO2 adsorption amount decreased almost linearly for een–MOF and een–MOF/Al as the exposure time increased, reaching 83% (11.33 wt%) and 78% (11.95 wt%) of the initial capacity, respectively. In contrast, for the silane-coated een–MOF/Al–Si, the CO2 uptake remained almost constant with maintaining a capacity of >12 wt% after exposure for 15 days (Fig. 4c). The PXRD data revealed that the silane-coated sample retained its structural integrity, while the structures of the other samples partially collapsed (for PXRD data see Supplementary Fig. 32). To further investigate the material’s stability in the presence of water vapor, we exposed it to harsh conditions of 10% H2O and 90% CO2 at a desorption temperature of 140 °C, which are more practical environments for real-world post-combustion CO2 capture applications (Fig. 4d). The adsorption capacity of een–MOF/Al linearly decreased to 76% (11.07 wt%) of the original capacity after 3 days. The capacity loss was due to the structural collapse (for PXRD data see Supplementary Fig. 33). It is noted that the capacity of een–MOF/Al–Si was almost retained with a capacity of ~12 wt% after 3 days of exposure (Fig. 4d). This result demonstrates that the silane coating significantly enhances structural stability and performance durability (for CO2 adsorption capacities see Supplementary Fig. 33).

Furthermore, to investigate CO2 adsorption in the presence of water vapor, we collected CO2 uptake at dry conditions of 80 °C under 15% CO2, 85% He, and a flow rate of 10 cc min−1. The CO2 adsorption was recorded at wet conditions of 80 °C under 15% CO2, 3.75% H2O, 81.25% He, and a flow rate of 10 cc min−1. The water vapor was generated from a bubbler and was removed at a trap after passing through the sample before reaching the thermal conductivity detector (TCD). The CO2 adsorption capacity and rate for een–MOF/Al–Si were similar to those of TGA results. The amount of CO2 uptake was 10.24 wt% in the dry conditions and 10.88 wt% in the wet conditions. The CO2 adsorption capacity was a little higher at wet conditions than dry conditions, which was commonly observed in diamine-grafted Mg2(dobpdc) frameworks24,28,31. Thus, een–MOF/Al–Si can selectively adsorb CO2 even in the presence of water vapor.

Since amines can be degraded in the presence of O2 from the flue gas, potential adsorbents are required to be stable under O212,13. To confirm the O2 stability of een–MOF/Al–Si, adsorption were carried out under a mixed gas of 15% CO2 and 5% O2 and at 80 °C for 5 min, while desorption occurred at 100% CO2 and 140 °C for 1 min (for TSA cycles see Supplementary Fig. 35). After 50 cycles, it was confirmed the CO2 adsorption capacity was almost maintained at 9.52 wt% when compared to 9.96 wt% in the first cycle.

Conclusion

We demonstrated the scalable synthesis of Mg2(dobpdc) as well as shaped MOF microbeads via a spray dry method. The microbeads exhibited excellent mechanical properties, which are important for practical applications. We also performed the sequential postsynthetic functionalizations such as diamine grafting and silane coating of the beads to obtain a hydrophobic material with a significant working capacity and enhanced structural stability under humid conditions. The post-functionalized MOF microbead materials with the remarkable long-term performance can be applied as adsorbents in real-world CO2 capture processes.

Methods

Preparation

H4dobpdc was synthesized according to the previous literature16. MgCl2·6H2O (≥98.0%), N-ethylethylenediamine (een, ≥98.0%) and octadecyltrimethoxysilane were obtained from Sigma-Aldrich. All solvents were reagent grade (≥99.0%) and used as received. Alumina sol (10 wt%, pH = 4, particle size = 10–20 nm) was purchased from Alutec company.

Synthesis of Mg2(dobpdc) in a g scale

Mg2(dobpdc) was synthesized through a previous synthesis method, and synthesized through a 300 mL high-pressure reactor. MgCl2·6H2O (19.7 g, 970 mmol) was dissolved in EtOH (100 mL) and H4dobpdc (7.60 g, 277 mmol) was dissolved in DMF (100 mL). The resulting solutions were poured into a 300 mL autoclave. The mixture was reacted for 72 h at 130 °C. After the reaction, the resulting powder was separated via filtration and soaked in DMF.

Synthesis of Mg2(dobpdc) in a kg scale

Scale-up Mg2(dobpdc) was synthesized in the following method. MgCl2·6H2O (1.50 kg, 7.29 mol) was dissolved in EtOH (10.0 L) and H4dobpdc (1.00 kg, 3.64 mol) was dissolved in DMF (10.0 L). The resulting solutions were poured into a 35 L autoclave. The mixture was reacted for 72 h at 140 °C. After the reaction, the resulting powder was separated via filtration and soaked in mother liquid.

Synthesis of Mg2(dobpdc)/alumina beads (DMF–MOF/Al)

DMF–MOF/Al was manufactured in the following method. DMF–MOF was dried in an oven for 24 h at 100 °C. The resultant sample (100 g) was grounded with water (340 mL) and alumina sol (111 g, alumina content 10 wt%) using 3 mm ZrO2 ball (130 g) in a ball-mill instrument for 30 min under 300 rpm. After grinding, the obtained MOF/Al slurry was injected into a spray drier using a feeding machine. The MOF/Al slurry, which passed through the atomizer, was formed into spherical particles. When the high-temperature gas was blown to the spherical particles, it hardened immediately (input gas temperature = 290 °C, output gas temperature = 130 °C). The dried particles were collected in a dry particle collector to yield DMF–MOF/Al beads (58 g).

Synthesis of een-Mg2(dobpdc)/alumina beads (een–MOF/Al)

DMF–MOF/Al was soaked in MeOH for 72 h and refreshed everyday with MeOH. The MeOH-soaked sample was placed in a 100-mL Schlenk flask and dried for 24 h at 150 °C under vacuum. The dried sample (300 mg, 0.78 mmol) was replaced in a 100-mL one-neck round flask. N-ethylethylenediamine (4.12 mL, 39.1 mmol) and hexane (20 mL) were put into the flask using cannula. The mixture was soaked for 12 h at 50 °C. After the reaction, the mixture was filtered and washed with hexane. Light purple products were dried under vacuum for 2 h.

Synthesis of een-Mg2(dobpdc)/alumina–silane beads (een–MOF/Al–Si)

Dried een–MOF/Al (300 mg) was placed in a 100-mL one-neck round flask. Octadecyltrimethoxysilane (1.5 g) and hexane (40 mL) were poured into the flask. The mixture was soaked for 72 h at 50 °C. After the reaction, the mixture was filtered and washed three times with hexane. Gray products were dried under vacuum for 2 h.

Synthesis of een–MOF-Si and alumina–Si

een–MOF–Si was synthesized in the following method. Dried een–MOF (300 mg) was placed in a 100-mL one-neck round flask. Octadecyltrimethoxysilane (1.5 g) and hexane (40 mL) were put into the flask. The mixture was soaked for 72 h at 50 °C. After the reaction, the mixture was filtered and then soaked in fresh hexane for 3 days in order to eliminate unreacted octadecylsilane. After 3 days, een–MOF–Si was collected by filtration. Alumina–Si was synthesized in the similar reaction using alumina. After heating of alumina sol at 100 °C for 6 h under vacuum, alumina in a solid state was obtained. Obtained alumina was grounded by a mortar to make fine powders. Ground alumina (300 mg) was placed in a 100-mL one-neck round flask. Octadecyltrimethoxysilane (1.5 g) and hexane (40 mL) were put into the flask. The mixture was soaked for 72 h at 50 °C. After the reaction, the mixture was filtered and then soaked in fresh hexane for 3 days in order to eliminate unreacted octadecylsilane. After 3 days, alumina–Si was collected by filtration.

Physical measurements

PXRD patterns were recorded using Cu Kα radiation (λ = 1.5406 Å) with a Rigaku Ultima III diffractometer with a scan rate of 2° min−1 and a step interval of 0.02°. Contact angle was measured by Phoenix-MT(T). X-ray photoelectron spectroscopy (XPS, K-Alpha+) data were taken at Hanyang University. SEM images were collected using S-4600 at Korea Basic Science Institute (KBSI).

Attrition test

Attrition test was carried out by a crafted device at Korea Research Institute of Chemical Technology based on ASTM D5757 standard method46. This device is composed of distributor, attrition tube, setting chamber, and fine collection assembly. The distributor and attrition tube have a tube with 35 mm inside diameter, 710 mm in length, and three 0.397 mm holes on the distributor. The attrition tube is linked to a setting chamber and a fines collector. The sample was injected to the distributor by blowing air in a rate of 10 L min−1. The sample that was passed through the three holes on the distributor collided with each other in the attrition tube. The crashed sample in the attrition tube was transferred to the setting chamber, and the finely ground sample was transported through the internal filter of the chamber to the fines collector. This process was performed for 5 h under dry air. The attrition index was calculated by following formula:

$${\mathrm{Attrition}}\;{\mathrm{index}}\;\left( {{\mathrm{AI}}} \right) = \frac{{{\mathrm{Weight}}\;{\mathrm{of}}\;{\mathrm{collected}}\;{\mathrm{sample}}\;{\mathrm{on}}\;{\mathrm{fine}}\;{\mathrm{collection}}\;{\mathrm{assembly}}}}{{{\mathrm{Weight}}\;{\mathrm{of}}\;{\mathrm{initial}}\;{\mathrm{sample}}}}\; \times 100$$
(1)

Gas adsorption measurements

Gas sorptions with CO2 (99.999%) and N2 (99.999%) were performed using a Micromeritics 3flex instrument after the pretreatment of each sample. Density functional theory pore size distributions for Mg2(dobpdc) (MOF), MOF/Al, een–MOF/Al, and een–MOF/Al–Si were estimated from N2 isotherms at 77 K using the model of cylindrical pores with oxide surface. The water vapor isotherms were recorded by ASAP 2020 instrument after the treatment of samples at 298 K. To investigate the CO2 adsorption capacity in the presence of water vapor, we used modified temperature-programed desorption (TPD) processes in the BELCAT-II instrument. We obtained TCD signals for CO2 under dry and wet conditions during TPD processes and quantized the amount of CO2 uptake versus time through the area of TPD signals for CO2.

Thermogravimetric analyses and water vapor exposure test

TGA were obtained at a ramp speed of 20 °C min−1 in a stream of 15% CO2 and 100% CO2 using a TA instrument Discovery TGA with a flow rate of 70 mL min−1 for all gases. The water vapor exposure test was conducted by a homemade setup. The water vapor (about 10% H2O) was formed from a round-bottomed flask in which water was warmed at 50 °C. CO2 (99.999%) gas was flowed through the flask and then passed into the sample chamber heated at 140 °C. After exposure test, the samples were recollected to measure adsorption capacity using a TGA instrument under 15% CO2 for 30 min after activation under N2 atmosphere.

Exposure tests

For air exposure, each sample (100 mg) was put into a 20 mL vial and exposed to an ambient air. Then, we measured adsorption uptake of the sample using a TGA instrument at 15% CO2 for 30 min after activation under N2. For humidity exposure, water vapor (about 10% H2O) was generated from a round-bottomed flask by warming water at 50 °C. CO2 (99.999%) was passed through the flask and then injected into a homemade sample chamber heated at 140 °C. After the exposure tests, samples were recollected to measure adsorption capacity using a TGA instrument under 15% CO2 for 30 min after activation under N2.

Spectroscopic measurements

IR spectra were recorded using a Nicolet iS10 FTIR spectrometer. In situ IR spectra were taken with an air‐tight homemade IR cell containing KBr windows. Prior to the IR measurements, high‐purity N2 (99.999%) was flowed into the sample chamber for removal of atmospheric CO2 from the IR instrument. Solid-state NMR spectroscopy was collected with a Bruker AVANCE II + 400 MHz NMR system at KBSI Seoul Western Center.

Isosteric heats of adsorption calculations

We employed the dual-site L–F equation (Eq. (2)) to analyze the CO2 adsorption characteristics in the whole range of the isotherm for een–MOF/Al and een–MOF/Al–Si:

$$q = \frac{{q_{\rm{sat,A}}b_{\rm{A}}p^{\alpha {\rm{A}}}}}{{1 + b_{\rm{A}}p^{\alpha {\rm{A}}}}} + \frac{{q_{{\rm{sat,B}}}b_{\rm{B}}p^{\alpha {\rm{B}}}}}{{1 + b_{\rm{B}}p^{\alpha {\rm{B}}}}}$$
(2)

here, q represents the amount of CO2 (mmol g−1), p the pressure (bar), qsat the saturation amount (mmol g−1), b the L–F parameter (bar−α), and α the L–F exponent (dimensionless) toward two adsorption sites A and B:

$$(\ln p)_q = \left( {\frac{{Q_{{\rm{st}}}}}{R}} \right)\left( {\frac{1}{T}} \right) + C$$
(3)

The isotherm fits were utilized to determine the exact pressures, p, corresponding to constant amounts of CO2 adsorbed, q, at three temperatures. The Clausius–Clapeyron equation (Eq. (3)) was applied to attain the isosteric heats of adsorption (Qst).