In this study, we used a high-throughput computational screening approach to examine the potential of metal–organic frameworks (MOFs) for capturing propane (C3H8) from different gas mixtures. We focused on Quantum MOF (QMOF) database composed of both synthesized and hypothetical MOFs and performed Grand Canonical Monte Carlo (GCMC) simulations to compute C3H8/N2/O2/Ar and C3H8/C2H6/CH4 mixture adsorption properties of MOFs. The separation of C3H8 from air mixture and the simultaneous separation of C3H8 and C2H6 from CH4 were studied for six different adsorption-based processes at various temperatures and pressures, including vacuum-swing adsorption (VSA), pressure-swing adsorption (PSA), vacuum–temperature swing adsorption (VTSA), and pressure-temperature swing adsorption (PTSA). The results of molecular simulations were used to evaluate the MOF adsorbents and the type of separation processes based on selectivity, working capacity, adsorbent performance score, and regenerability. Our results showed that VTSA is the most effective process since many MOFs offer high regenerability (>90%) combined with high C3H8 selectivity (>7 × 103) and high C2H6 + C3H8 selectivity (>100) for C3H8 capture from air and natural gas mixtures, respectively. Analysis of the top MOFs revealed that materials with narrow pores (<10 Å) and low porosities (<0.7), having aromatic ring linkers, alumina or zinc metal nodes, typically exhibit a superior C3H8 separation performance. The top MOFs were shown to outperform commercial zeolite, MFI for C3H8 capture from air, and several well-known MOFs for C3H8 capture from natural gas stream. These results will direct the experimental efforts to the most efficient C3H8 capture processes by providing key molecular insights into selecting the most useful adsorbents.

Propane (C3H8) is a highly valuable hydrocarbon with a wide range of uses, such as a fuel,1 an environmentally friendly cooling agent,2 and a raw material for the production of olefins.3 Propane generally coexists with other light hydrocarbons, such as ethane (C2H6) and n-butane (n-C4H10), in the natural gas mixture. Separation of propane from natural gas is important since it diminishes the energy content of the natural gas, which is mainly composed of methane (CH4). Propane also exists in air as a volatile hydrocarbon,4 and its capture is necessary for the benefit of public health and to reduce global warming. Separation of propane from natural gas and air mixtures using conventional separation methods, such as cryogenic separation, is an energy-intensive process.5 Adsorption-based separation by using porous materials has emerged as a powerful alternative for C3H8 separation.6 Zeolites and activated carbons have been utilized for this process, but they generally suffer from low C3H8 adsorption capacities, leading to low selectivities due to their limited pore volumes and surface areas.7–9 Therefore, there is ongoing research to find the optimal adsorbent materials for efficient C3H8 capture from both natural gas and air mixtures.

Metal–organic frameworks (MOFs), which are formed in a wide variety of topologies by the combination of metal nodes and organic linkers, are an emerging class of crystalline porous materials.10–13 MOFs have distinct advantages over traditional porous materials, such as tunable physical and chemical properties, tailorable pore sizes, record surface areas, and ultrahigh porosities. These features make MOFs promising candidates for several gas adsorption and separation applications, such as H2 and CH4 storage,14–17 CO2 capture and sequestration,18–20 and CH4/N2 and CH4/H2 separations.21,22 Several experimental studies have recently focused on MOF adsorbents for C3H8 capture from light hydrocarbons, such as CH4 and C2H6, by measuring the single-component gas adsorption isotherms and calculating the selectivities of materials using the ideal adsorbed solution theory (IAST).23–29 For example, C3H8 and CH4 uptakes of two MOFs, FIR-125 and FIR-126, were experimentally measured and their C3H8/CH4 selectivities were predicted as 255 and 232, respectively, for the separation of equimolar mixtures at 1 bar, 298 K.25 C3H8, C2H6, and CH4 uptakes of five isoreticular mesoporous MOFs (NIIC-20-G) were measured, and the selectivities for C3H8/C2H6:50/50, C3H8/CH4:50/50, and C3H8/C2H6/CH4:10/10/80 mixtures were calculated to be in the ranges of 10.4–29, 345–1110, and 261–1234, respectively, at 1 bar, 298 K.27 C3H8 and CH4 uptakes of JLU-MOF66 and JLU-MOF67 were measured, and their C3H8/CH4 selectivities were calculated as 308 and 297, respectively, for equimolar mixture separation at 1 bar, 298 K.28 

These results demonstrate that MOFs achieve high C3H8 selectivities, and they have the potential to be used as adsorbents in C3H8 capture applications. However, these experimental studies have been limited to a few different types of MOFs. There can be much more promising materials among many other existing MOFs. The number of synthesized MOFs has been rapidly increasing, and 122 738 MOFs (retrieved by the ConQuest software30 on 27 November 2023) have already been deposited into the CSD (Cambridge Structural Database).31,32 Examining all these synthesized MOFs using purely experimental methods is not possible due to time and resource limitations. The high-throughput computational screening (HTCS) approach based on molecular simulations has played an important role in screening large numbers of MOFs but mostly for CO2 capture.33–36 Generation of several MOF databases, including Computation-Ready and Experimental MOF (CoRE MOF)37,38 and computer-generated hypothetical MOF (hMOF) databases,39–42 accelerated the computational screening of MOFs to identify the best adsorbent candidates and to direct the experimental studies to these materials for target applications. For C3H8 capture, HTCS studies mostly focused on C3H8/C3H6 (propylene) separation.43–46 For example, Yeo et al.44 studied MOFs, zeolite imidazolate frameworks (ZIFs), and zeolites available in the CSD and Inorganic Crystal Structure Database (ICSD) using Grand Canonical Monte Carlo (GCMC) simulations for the separation of equimolar C3H8/C3H6 mixture at 0.5 and 5 bar, 400 K. Their results showed that the structures having high C3H8 selectivities (>2) suffer from low working capacities (<1 mol/kg). Solanki and Borah47 studied the same separation using GCMC simulations at 0.1 and 1 bar, 298 K, and examined 6888 CoRE MOFs. They identified the top 20 MOFs having C3H8 selectivities in between 1.35 and 1.45 and working capacities in between 8.88 and 11.55 mol/kg. The CoRE MOF database was also studied for C3H8/C2H6/CH4:3/7/90 separation by GCMC simulations at 0.1 and 1 bar, 298 K, and the best MOFs were identified to have C2H6 + C3H8 selectivities and working capacities in the ranges of 52.6–125.4 and 2.15–5.99 mol/kg, respectively.48 Yuan et al.49 performed GCMC simulations to study hydrophobic, hypothetical MOFs for the capture of non-methane hydrocarbons, including C3H8, C4H10, n-pentane (C5H12), and n-hexane (C6H14), from air at 1 bar, 298 K. Their results showed that MOF adsorbents can achieve very high C3H8 selectivities up to 7.3 × 105. This literature review suggests that there is a strong need for the evaluation of both synthesized and hypothetical MOFs for C3H8 separation from air and for the simultaneous separation of C3H8 and C2H6 from the natural gas mixture to fully unlock the adsorbent potential of this very large material space.

In this work, we proposed a computational screening approach to evaluate C3H8/N2/O2/Ar and C3H8/C2H6/CH4 mixture separation potentials of both synthesized, real MOFs and computer-generated, hypothetical MOFs using molecular simulations. We focused on a recently established hybrid Quantum MOF (QMOF) database,50,51 composed of 20 375 different types of MOFs. GCMC simulations were first performed to produce C3H8/N2/O2/Ar and C3H8/C2H6/CH4 mixture adsorption data of MOFs at three different pressures, 0.1, 1, and 10 bar, and two different temperatures, 298 and 343 K. These data were then used to calculate the adsorbent performance evaluation metrics of MOFs: selectivity, working capacity, adsorbent performance score, and regenerability of each material for six distinct adsorption-based separation processes, including vacuum-swing adsorption (VSA) at two different temperatures, pressure-swing adsorption (PSA) at two different temperatures, vacuum–temperature swing adsorption (VTSA), and pressure–temperature swing adsorption (PTSA). We then ranked the adsorbents based on the calculated metrics and identified the best materials for each process in addition to selecting the optimal separation process for C3H8 separation. Our detailed analysis for the structure–performance relations of the promising adsorbents revealed the most important molecular features of the top-performing materials for efficient C3H8 capture. These results will not only direct the experimental efforts to the most useful MOF adsorbents but also will be useful for the future design of new MOFs at the molecular level.

Figure 1 represents the computational methodology that we proposed to study MOF adsorbents for C3H8 capture from natural gas and air mixtures. Crystallographic structures of MOFs were obtained from the QMOF database.50,51 This database offers a diverse set of MOFs with different chemical and physical features and previously studied for predicting bandgaps51,52 and heat capacities of materials.53 The QMOF database contains 20 375 distinct types of non-disordered, unbounded solvent-free structures, of which 16 884 are synthesized and 3491 are hypothetical. The structural features of all MOFs, such as the accessible surface area (Sacc), the largest cavity diameter (LCD), and the pore limiting diameter (PLD), were calculated using the Zeo++ as reported before.54 A probe with a diameter of 3.72 Å, corresponding to the size of a nitrogen (N2) molecule, was employed to determine the surface area, while the geometric pore volume was calculated with a probe size of zero. C3H8 is the largest molecule among all the gas molecules that we studied in this work; therefore, we limited the set of MOFs to those having PLDs greater than the kinetic diameter of C3H8 (5 Å)55 and ended up with 3496 materials, including 671 synthesized and 2825 hypothetical materials.

FIG. 1.

Schematic of our computational screening methodology together with the details of the six separation processes that we examined throughout this work.

FIG. 1.

Schematic of our computational screening methodology together with the details of the six separation processes that we examined throughout this work.

Close modal

To evaluate the affinity of MOFs toward different gas molecules, we first utilized Widom’s particle insertion method as implemented in the RASPA56 simulation package to compute Henry’s constant (KH,i) and heat of adsorption (Q0st,i) of each gas species (i) in MOFs at 298 K. The ratios of Henry’s constants were used to compute the infinite dilution C3H8 selectivities of MOFs in air mixture (denoted as “C3H8,Air”), SC3H8,Air0=KH,C3H8/(KH,N2+KH,O2+KH,Ar), and C2H6+C3H8 selectivities of MOFs in natural gas mixture (denoted as “C2+C3,NG”), SC2+C3,NG0=(KH,C2H6+KH,C3H8)/KH,CH4. We used the ratio of Henry’s constants of gases as an initial metric to evaluate 3496 MOFs for their separation performance for both gas mixtures. Our objective in this approach is to pinpoint the MOFs that show promise for achieving high mixture selectivities. This is based on the observation that MOFs demonstrating high ideal selectivities at infinite dilution typically show high mixture selectivities. This approach has widely been used in high-throughput computational screening studies of MOFs.46,57 Figure S1 of the supplementary material shows the relation between the PLDs of MOFs and their (a) C3H8 and (b) C2H6 + C3H8 selectivities under the infinite dilution condition, and the highlighted region above the dashed line shows the first 1000 MOFs computed to have the highest selectivities. As a result of this prescreening, we identified two different subsets of 1000 promising MOFs for C3H8 capture from air and natural gas mixtures, and 982 MOFs were found to be common in both sets. These 1000 MOFs were processed to the next stage of computations for adsorption simulations.

GCMC simulations were performed on these 1000 MOFs to compute the adsorption of two different gas mixtures: a four-component mixture representing ambient air with trace amounts of C3H8 (77.922% N2, 20.979% O2, 0.999% Ar, and 0.1% C3H8)58 and a three-component natural gas mixture (85% CH4, 10% C2H6, and 5% C3H8).59,60 5 × 104 initialization and 5 × 104 production cycles were used during the simulations of both multi-component mixtures to minimize uncertainties in our calculations. For example, at 1 bar, 298 K, reported uncertainties for C3H8, N2, Ar, and O2 uptakes are 1.9%, 1.6%, 6.9%, and 2.0% in air mixture, respectively. For natural gas mixture, uptakes of C3H8, C2H6, and CH4 have uncertainties of 0.6%, 1.7%, and 2.4% under the same condition, respectively. The Lennard-Jones (LJ) 12-6 potentials were used for all gas–gas and gas–MOF interactions, while additional Coulomb interactions were defined for N2 and O2 molecules. Transferable potentials for phase equilibria (TraPPE) force field were used to describe the gases C3H8, C2H6, CH4, N2, and O2.61 C3H8 and C2H6 were modeled as three- and two-site molecules, respectively, while CH4 was modeled as a single-site molecule.61 N2 and O2 were modeled using rigid, three-site models, including one site at the center of the molecule and one site on each side of the molecule.62–64 Ar was modeled as a single LJ site.65 Force field parameters were taken from Dreiding66 for the MOF atoms, and for the metal atoms, which are not available in Dreiding, Universal Force Field (UFF) parameters were used.67 The Lorentz–Berthelot mixing rule was employed to obtain the cross-interaction parameters. The cutoff distance for LJ interactions was set to 12 Å. The Ewald summation was utilized to compute the Coulombic interactions for N2 and O2. Density Derived Electrostatic and Chemical (DDEC6)68,69 point charges available in the QMOF database were used to compute the Coulombic interactions. The coordinates of all framework atoms were kept fixed during simulations following the literature to save computational time.35 

Molecular simulations were performed at pressures of 0.1, 1, and 10 bar and at temperatures of 298 and 343 K to mimic the VSA, VTSA, PSA, and PTSA processes as represented in Fig. 1. The Peng–Robinson equation of state was used to convert the fugacity to pressure. The accuracy of our GCMC simulations has been validated by showing a good agreement between our simulated C2H6, CH4, N2, O2, and Ar adsorption data and the experimentally measured adsorption data under a wide range of pressure and temperature conditions in our previous studies.70–73 In addition, we compared the simulated single-component C3H8 uptakes of several MOFs at various pressures with the available experimental data collected from the literature.74–77 The good agreement between simulations and experiments shown in Fig. S2 indicates the accuracy of molecular models used in the simulations.

Gas uptake results obtained from molecular simulations were used to compute several adsorbent performance evaluation metrics of MOFs for the selective capture of desired molecules, which are C3H8 for air mixture (denoted as “C3H8,Air”) and C2H6 and C3H8 for natural gas mixture (denoted as “C2+C3,NG”). The adsorption selectivity (SC3H8,Air and SC2+C3,NG), working capacity (ΔNC3H8,Air and ΔNC2+C3,NG), adsorbent performance score (APSC3H8,Air and APSC2+C3,NG), and percent regenerability (R%C3H8,Air and R%C2+C3,NG) were calculated using the expressions given in Table SI. The selectivity was calculated by dividing the uptake of the desired molecules by the uptake of the undesired molecules, normalized by the bulk mixture composition. The working capacity was defined as the difference between the adsorbed amounts of desired molecule under the adsorption and desorption conditions of each process. The adsorbent performance score was described as the product of working capacity and selectivity. The percent regenerability was calculated as the ratio of working capacity to the uptake of the desired gas under the adsorption condition. These metrics were calculated for all MOFs, for two different gas mixtures, and for six different separation processes.

The top 20 MOFs, which exhibit the highest APS with R% > 80%, were identified separately for each process condition. The structural and chemical properties of the best adsorbents were examined in detail to reveal the materials’ essential features for efficient C3H8 capture. The MOFseek software78 was used to identify the linker subunits and metal types of the top MOFs after the removal of functional groups. For 40 high-performing MOFs identified for two different gas separations, 19 different linker subunits were listed with their 2D structures in Table SII.

We first investigated the potential of MOFs for C3H8 capture from air. Figure 2 shows the relations between selectivities and working capacities and the relations between regenerabilities and APSs calculated for 1000 MOFs at three different vacuum-based separation processes, VSA298, VSA343, and VTSA. For the VSA process at 298 (343) K, SC3H8,Air and ΔNC3H8,Air values were computed in the ranges of 280–4.1 × 104 (79–1.3 × 104) and 0.01–2.43 (2.5 × 10−3–1.57) mol/kg, respectively, as shown in Figs. 2(a) and 2(b). For the VTSA process, ΔNC3H8,Air values were computed to be generally higher, 0.01–3.60 mol/kg, with the same selectivity range of VSA298 process, as illustrated in Fig. 2(c). In each figure, the dashed lines show the medians of each performance metric calculated for 1000 MOFs, and the high-performing adsorbents are expected to exceed these lines. The common trend in all three separation processes is that MOFs with high C3H8 selectivities (>103) achieve high C3H8 working capacities (>1 mol/kg). This is more pronounced for VTSA since the integration of temperature-swing to the desorption step results in low C3H8 capacities, leading to very high working capacities. We also observed that MOFs tend to have low ΔNC3H8,Air and SC3H8,Air at VSA343 condition, compared to other processes. This can be explained by the weaker interactions between C3H8 molecules and frameworks at 343 K, resulting in lower uptakes compared to those calculated at 298 K.

FIG. 2.

Selectivities and working capacities of selected 1000 MOFs for C3H8/N2/O2/Ar mixture separation for (a) VSA298, (b) VSA343, and (c) VTSA processes. Their regenerabilities and adsorbent performance scores for (d) VSA298, (e) VSA343, and (f) VTSA processes.

FIG. 2.

Selectivities and working capacities of selected 1000 MOFs for C3H8/N2/O2/Ar mixture separation for (a) VSA298, (b) VSA343, and (c) VTSA processes. Their regenerabilities and adsorbent performance scores for (d) VSA298, (e) VSA343, and (f) VTSA processes.

Close modal

To identify the most promising materials, we investigated the APSC3H8,Air and R%C3H8,Air of each MOF in Figs. 2(d)2(f). MOFs that were not able to exceed the medians for selectivity and working capacity were shown as gray points. High R%C3H8,Air is required for economic efficiency separation processes; thus, we set a limit of 80% for this metric, shown as a dashed line in Figs. 2(d)2(f). For VSA298, VSA343, and VTSA processes, among the 1000 MOFs that we considered, 879, 991, and 998 MOFs surpassed R% > 80% limit, respectively. The APSC3H8,Air of 1000 MOFs was computed in the ranges of 3.4–3.3 × 104, 0.3–1.3 × 104, and 3.7–7.0 × 104 mol/kg, respectively. For the VSA298 process, MOFs with high APSC3H8,Air (>103 mol/kg) suffer from low R% (<80%) in Fig. 2(d). This trend is less observable for VSA343 process in Fig. 2(e) due to a threefold decrease in APSs compared to the values calculated for VSA298 process. Figure 2(f) illustrates that large increases in working capacities due to temperature-swing are reflected on both APS and R% of MOFs under VTSA condition, as MOFs achieve the highest values of both metrics in this process. Overall, our results indicate that (i) MOFs achieve higher C3H8 selectivities and working capacities for vacuum-swing adsorption processes at 298 K than at 343 K, and (ii) VTSA is the optimal process condition among these three processes for capturing C3H8 from air as MOFs offer the highest APSC3H8,Air with the highest R%C3H8,Air values.

Figure 3 shows the calculated metrics of MOFs for C3H8 capture from air at three different pressure swing-based separation processes: PSA298, PSA343, and PTSA. Figure 3(a) shows that SC3H8,Air and ΔNC3H8,Air of MOFs are in the ranges of 158–1.6 × 104 and 0.05–3.46 mol/kg, respectively, for PSA298 process. MOFs have lower selectivities, 63.6–6.6 × 103, and lower working capacities, 0.02–1.76 mol/kg, for PSA343 process as shown in Fig. 3(b) due to the dramatic decrease in C3H8 uptakes at 1 bar, 343 K (1 × 10−3–1.97 mol/kg), compared to uptakes at 1 bar, 298 K (0.01–3.74 mol/kg). Figure 3(c) represents that the ΔNC3H8,Air values of MOFs are in the ranges of 0.09–3.96 mol/kg for PTSA process, with the same selectivities calculated for PSA298 process as their adsorption conditions are the same. MOFs with high selectivities (>103) tend to have low working capacities (<0.8 mol/kg) for PSA298 process. On the other hand, MOFs can achieve both high selectivities (>103) and high working capacities (>1.2 mol/kg) for PTSA process.

FIG. 3.

Selectivities and working capacities of selected 1000 MOFs for C3H8/N2/O2/Ar mixture separation for (a) PSA298, (b) PSA343, and (c) PTSA processes. Their regenerabilities and adsorbent performance scores for (d) PSA298, (e) PSA343, and (f) PTSA processes.

FIG. 3.

Selectivities and working capacities of selected 1000 MOFs for C3H8/N2/O2/Ar mixture separation for (a) PSA298, (b) PSA343, and (c) PTSA processes. Their regenerabilities and adsorbent performance scores for (d) PSA298, (e) PSA343, and (f) PTSA processes.

Close modal

Figures 3(d)3(f) show that many MOFs (930 out of 1000) surpassed R% > 80% limit under PTSA condition than under PSA298 and PSA343 conditions (392 and 793 MOFs, respectively). MOFs also achieve very high APSC3H8,Air, 24.5–1.7 × 104 mol/kg, for PTSA process compared to PSA298 and PSA343 processes (21.6–5.5 × 103 and 2.2–4.6 × 103 mol/kg, respectively), thanks to the temperature-swing effect. Overall, our findings suggest that (i) MOFs achieve high C3H8 selectivities and working capacities at low temperature processes, and (ii) PTSA is the best option among the three pressure-swing based separation processes, as many MOFs offer both high APSC3H8,Air (>103 mol/kg) and high R%C3H8,Air (>80%).

We then compared the performances of MOFs in vacuum-swing- and pressure-swing-based separation processes in Fig. 4. The ranges of calculated performance metrics under different process conditions are also given in Table SIII. Figure 4(a) shows that MOFs offer very high R%C3H8,Air (>80%) for VTSA and PTSA processes (998 and 930 materials among 1000 MOFs, respectively) compared to VSA298 and PSA298 processes (879 and 392 MOFs, respectively). This can be explained by the effect of temperature-swing, which results in more efficient desorption of C3H8, leading to very high R%C3H8,Air values. Figure 4(b) represents the distribution of APSC3H8,Air of MOFs having R%C3H8,Air >80%. Twenty-one MOFs offer very high APSC3H8,Air (>2 × 104 mol/kg) for VTSA process, while there is no such MOF for PTSA, VSA298, or PSA298 processes. The reason is that the addition of temperature-swing enhances C3H8 working capacities much more under vacuum-swing condition than under pressure-swing condition. This increase results in very high APSC3H8,Air under VTSA condition than under PTSA condition. At the VSA298 process, there are MOFs with high APSC3H8,Air, but they suffer from low R%C3H8,Air, and these materials may not find place in applications due to their low regenerabilities. For example, qmof-a9be673 has a very high APSC3H8,Air of 1.7 × 104 mol/kg, but it has a very low R%C3H8,Air of 23.9% under VSA298 condition. The addition of temperature-swing results in almost fourfold increase in both APSC3H8,Air and R%C3H8,Air (6.3 × 104 mol/kg and 87.6%, respectively) and makes this MOF the top material under VTSA condition. All in all, VTSA process is the most efficient process to use MOFs as adsorbents for achieving high APSC3H8,Air and R%C3H8,Air.

FIG. 4.

(a) The distribution of 1000 MOFs with respect to their R%C3H8,Air. (b) The distribution of MOFs having R%C3H8,Air > 80% with respect to their APSC3H8,Air. (c) The relation between APSC3H8,Air, pore sizes, and heat of adsorption of C3H8 under VTSA condition.

FIG. 4.

(a) The distribution of 1000 MOFs with respect to their R%C3H8,Air. (b) The distribution of MOFs having R%C3H8,Air > 80% with respect to their APSC3H8,Air. (c) The relation between APSC3H8,Air, pore sizes, and heat of adsorption of C3H8 under VTSA condition.

Close modal

We then examined the impact of structural features of MOFs on their separation performance under VTSA condition since it was identified as the most effective separation process. Figure 4(c) shows the distribution of APSC3H8,Air with respect to PLD and Q0st,C3H8 of 998 MOFs having high R%C3H8,Air (>80%). 268 MOFs with narrow pores (PLD < 8 Å) achieve high APSC3H8,Air (>103 mol/kg). Due to the differences between the kinetic diameters of C3H8 (5.0 Å) and other gas molecules in the mixture (N2:3.6 Å, O2:3.5 Å, and Ar:3.4 Å), C3H8 molecules can confine more strongly to these narrow pores, resulting in high SC3H8,Air (>103) and APSC3H8,Air (>103 mol/kg). We also observed that MOFs exhibiting high APSC3H8,Air (>103 mol/kg) have high Q0st,C3H8 (>30 kJ/mol) values, indicating the high affinities of these MOFs toward C3H8 molecules.

Closing this section, we finally investigated the variety of the top 20 MOFs that we identified among the 1000 MOFs for C3H8 capture from air and compared their properties with our original dataset. The list of the top 20 MOFs with their calculated performance metrics and structural and chemical features is given in Table SIV. Top-performing materials for the separation of air mixture have narrow-pores (PLD: 5.0–6.0 Å and LCD: 5.5–6.6 Å) and less porous structures (ϕ: 0.42–0.60). Figure 5(a) illustrates that the subset of MOFs mostly consists of materials from the hypothetical MOF databases (BoydWoo39,40 and GMOF41) rather than the synthesized MOF databases. This is reflected on the distribution of the top materials as well as they mostly originated from the hypothetical GMOF database. There are 12 different linker subunits (L) observed in the top 20 MOFs as shown in Fig. 5(b). Four out of the top 20 MOFs consist of L-6 (silicon hexafluoride), which is important since L-6 containing MOFs only constitutes 0.8% of the full dataset. The common trend among linker subunits of the top materials is that they mostly consist of aromatic rings, which are generally reported as the strong interaction sites for alkane molecules (C3H8 and C2H6).49,79–81 Analysis of the metal types for the top MOFs in Fig. 5(c) showed that alumina (Al) and zinc (Zn) are the most common ones, as they also often appear in our full dataset. These results are also supported by the literature, as Zn-based MOFs are proved to be high-performing for alkane capture applications.82 We finally compared the top-performing MOFs that we identified from the QMOF database with a commercial zeolite MFI. For a more balanced comparison, reported gas uptake values in the literature58 were used to calculate the C3H8 selectivity of MFI (642.7 at 1 bar, 298 K, and 689.4 at 10 bar, 298 K) for a mixture of C3H8/N2/O2/Ar (0.1% C3H8, 77.922% N2, 20.979% O2, and 0.999% Ar). All the top MOFs listed in Table SIV outperform MFI in terms of selectivity, and this emphasizes the high potential of MOF adsorbents for C3H8 capture from air.

FIG. 5.

Distribution of number of MOFs based on (a) databases, (b) linker subunit types, and (c) metal types for the separation of C3H8/N2/O2/Ar mixture.

FIG. 5.

Distribution of number of MOFs based on (a) databases, (b) linker subunit types, and (c) metal types for the separation of C3H8/N2/O2/Ar mixture.

Close modal

In this section, we focused on assessing MOF adsorbents’ potential for the simultaneous separation of C3H8 and C2H6 from the natural gas mixture. Figure 6 shows the computed selectivities (SC2+C3,NG), working capacities (ΔNC2+C3,NG), regenerabilities (R%C2+C3,NG), and adsorbent performance scores (APSC2+C3,NG) of 1000 MOFs for the capture of C3H8 and C2H6 from CH4 under three different vacuum swing-based separation processes. For VSA298 (VSA343) process, SC2+C3,NG and ΔNC2+C3,NG were calculated in the ranges of 10.4–789 (5.4–43.8) and 0.04–6.49 (0.13–4.12) mol/kg, respectively, as shown in Figs. 6(a) and 6(b). For VSA298 process, MOFs having high SC2+C3,NG (>100) suffer from low ΔNC2+C3,NG (<1.62 mol/kg). Only 14 out of 180 MOFs with high SC2+C3,NG (>100) surpassed the upper region designated by the median of ΔNC2+C3,NG as depicted in Fig. 6(a). For VSA343 process, the SC2+C3,NG and ΔNC2+C3,NG values of MOFs were dramatically lower than those calculated for VSA298 process, as shown in Fig. 6(b), due to decreased uptakes of C3H8 and C2H6 with increasing temperature. For VTSA process, working capacities were computed in between 0.22 and 7.59 mol/kg, as depicted in Fig. 6(c). Similar to VSA298 process, MOFs with high SC2+C3,NG (>100) tend to have low ΔNC2+C3,NG (<2.86 mol/kg). The number of MOFs achieving both high SC2+C3,NG and ΔNC2+C3,NG under VTSA condition (39 MOFs) is higher than the number of MOFs under VSA298 condition (14 MOFs) since temperature-swing leads to an increase in ΔNC2+C3,NG of MOFs.

FIG. 6.

Selectivities and working capacities of selected 1000 MOFs for C3H8/C2H6/CH4 mixture separation for (a) VSA298, (b) VSA343, and (c) VTSA processes. Their regenerabilities and adsorbent performance scores for (d) VSA298, (e) VSA343, and (f) VTSA processes.

FIG. 6.

Selectivities and working capacities of selected 1000 MOFs for C3H8/C2H6/CH4 mixture separation for (a) VSA298, (b) VSA343, and (c) VTSA processes. Their regenerabilities and adsorbent performance scores for (d) VSA298, (e) VSA343, and (f) VTSA processes.

Close modal

We aimed to identify the MOFs with the highest adsorbent performance scores among the ones surpassing the regenerability limit of 80%. Figures 6(d)6(f) show that 115, 575, and 764 out of 1000 MOFs surpassed 80% limit and the APSC2+C3,NG values of these materials were computed in the ranges of 3.1–467.1, 0.93–132.2, and 19.8–802.7 mol/kg for VSA298, VSA343, and VTSA processes, respectively. We observed that MOFs offering the highest APS achieve R% > 80% for each process. The highest APS and R% values were observed under VTSA condition, as the increase in the desorption temperature leads to a large increase in ΔNC2+C3,NG. Overall, (i) additional temperature-swing over the vacuum-swing separation process significantly increases the working capacity and APS values of MOF adsorbents, and (ii) VTSA is the optimal process among vacuum-swing separation processes because MOFs having high R%C2+C3,NG (>80%) achieve very high APSC2+C3,NG (>100 mol/kg) as well.

Figure 7 shows the selectivities, working capacities, regenerabilities, and APSs calculated for 1000 MOFs for C3H8 and C2H6 capture from the natural gas mixture at three different pressure-swing-based separation processes: PSA298, PSA343, and PTSA. SC2+C3,NG and ΔNC2+C3,NG of MOFs were computed to be in the ranges of 4.5–765 and 0.01–8.18 mol/kg for PSA298 process in Fig. 7(a). For PSA343 process, we calculated relatively lower SC2+C3,NG and ΔNC2+C3,NG values of 6.2–179 and 0.05–5.71 mol/kg, respectively, as shown in Fig. 7(b). These results showed that the increase in adsorption temperature resulted in a major decrease in both selectivities and working capacities of MOFs, which is similar to the trend we observed in vacuum-swing separation processes. For PTSA process, SC2+C3,NG and ΔNC2+C3,NG values were calculated to be in the ranges of 4.5–765 and 0.11–8.95 mol/kg, as represented in Fig. 7(c). For all three processes, MOFs offering high SC2+C3,NG (>100) suffer from low ΔNC2+C3,NG. Only 192, 170, and 219 MOFs can surpass the median limits of SC2+C3,NG and ΔNC2+C3,NG under PSA298, PSA343, and PTSA conditions, respectively. APSC2+C3,NG of MOFs were in the ranges of 0.8–178.3, 1.3–132.6, and 1.9–363.0 mol/kg for PSA298, PSA343, and PTSA processes, respectively, as shown in Figs. 7(d)7(f). Seventy-four MOFs surpassed the R% limit for PTSA process, while there were only 3 MOFs exceeding this limit under both PSA298 and PSA343 conditions. Similar to our findings in the vacuum-swing separation process, additional temperature-swing results in an increase in R% values obtained under pressure-swing process conditions. As a result, MOFs achieve higher selectivities (>25), working capacities (>4 mol/kg), APSs (>100 mol/kg), and R% (>80%) under PTSA condition compared to other pressure-swing processes.

FIG. 7.

Selectivities and working capacities of selected 1000 MOFs for C3H8/C2H6/CH4 mixture separation for (a) PSA298, (b) PSA343, and (c) PTSA processes. Their regenerabilities and adsorbent performance scores for (d) PSA298, (e) PSA343, and (f) PTSA processes.

FIG. 7.

Selectivities and working capacities of selected 1000 MOFs for C3H8/C2H6/CH4 mixture separation for (a) PSA298, (b) PSA343, and (c) PTSA processes. Their regenerabilities and adsorbent performance scores for (d) PSA298, (e) PSA343, and (f) PTSA processes.

Close modal

We then compared the performances of MOFs in vacuum-swing and pressure-swing separation processes for the co-separation of C3H8 and C2H6 from CH4. Figure 8 shows that MOFs can achieve high R% values only for VTSA and PTSA processes. Figure 8(a) depicts that there are 516 and 3 MOFs with very high R% (>90%) for these two processes, while under VSA298 and PSA298 conditions, there is no such MOF. Figure 8(b) illustrates that 24 of the highly regenerable MOFs also have high APSC2+C3,NG (>500 kg/mol) for VTSA process, while there is no such MOF under other conditions. The temperature-swing is vital for MOFs with high APSC2+C3,NG but low R%C2+C3,NG. For instance, the MOF, qmof-ee6fe4f having an APSC2+C3,NG of 399.1 mol/kg and a low R%C2+C3,NG of 46.5% under VSA condition, became the best material with a two-fold increase in APSC2+C3,NG and R%C2+C3,NG (799 mol/kg and 93.15%, respectively) under VTSA condition.

FIG. 8.

(a) The distribution of 1000 MOFs with respect to their R%C2+C3,NG. (b) The distribution of MOFs with R%C2+C3,NG > 80% with respect to their APSC2+C3,NG. (c) The relation between APSC2+C3,NG, pore sizes, and porosities of MOFs under VTSA condition.

FIG. 8.

(a) The distribution of 1000 MOFs with respect to their R%C2+C3,NG. (b) The distribution of MOFs with R%C2+C3,NG > 80% with respect to their APSC2+C3,NG. (c) The relation between APSC2+C3,NG, pore sizes, and porosities of MOFs under VTSA condition.

Close modal

Table SV shows the ranges of performance metrics calculated for all processes, and VTSA is the best process for separating C3H8 and C2H6 from CH4 using MOF adsorbents. Therefore, we explored the relation between APSC2+C3,NG of MOFs under VTSA condition with the pore sizes and porosities of materials in Fig. 8(c). Although there is no clear trend, an optimal performance window (APSC2+C3,NG > 500 mol/kg) is observable when porosities and pore sizes of MOFs are in the ranges of 0.6–0.7 and 5–10 Å, respectively. This analysis suggests that narrow-pored MOFs tend to become top performers for C3H8 and C2H6 capture.

We then analyzed the distribution of the top 20 MOFs that we identified from 1000 MOFs and the full set of MOFs according to their original databases, linker subunits, and metal types. The list of the top materials with their calculated performance metrics, structural and chemical features is given in Table SVI. For the separation of natural gas mixture, top materials tend to have narrow-pored structures (PLD: 5.1–9.7 Å and LCD: 6.4–10.5 Å), with mediocre porosities (ϕ: 0.61–0.70). Figure 9(a) shows that most of the top materials originated from the hypothetical GMOF and BoydWoo databases. Figure 9(b) represents that there are 11 different linker subunits observed in the top 20 MOFs. Fifteen out of 20 top MOFs constitute L-1 (thieno[3,2-b]thiophene-2,5-dicarboxylic acid) or L-3 ([1,1′-biphenyl]-4,4′-dicarboxylic acid) linkers, which are the most predominant linker subunits among the full set. These linkers mostly include aromatic rings having strong interaction sites for C3H8 and C2H6 molecules.49,79–81 Four of the top 20 MOFs were found to be anion-pillared in agreement with the literature emphasizing their potential for alkane separation. Analysis of the metal types in Fig. 9(c) showed that Al-based MOFs are the most common ones in both the top material set and full set.

FIG. 9.

Distribution of number of MOFs based on (a) databases, (b) linker subunit types, and (c) metal types for the separation of C3H8/C2H6/CH4 mixture.

FIG. 9.

Distribution of number of MOFs based on (a) databases, (b) linker subunit types, and (c) metal types for the separation of C3H8/C2H6/CH4 mixture.

Close modal

As the final part of our work, we compared the performances of the top materials with those of several other adsorbents for C3H8 separation. Studies in the literature generally have focused on binary (C3H8/CH4) or ternary (C3H8/C2H6/CH4) mixture separations. To make a comparison with the literature, we computed binary C3H8/C2H6 and C3H8/CH4 selectivities for the top 20 MOFs identified in our work using the gas uptakes acquired from our ternary mixture simulations. C3H8/CH4 and C3H8/C2H6 selectivities of the top 20 MOFs were calculated to be in the ranges of 27.1–2241 (8.9–2044) and 3.3–51.5 (1.7–35) at 1 (10) bar and 298 K, respectively. These values are higher than the selectivities reported for various MOFs under the same conditions listed in Table SVII, underlying the promise of QMOFs. For the simultaneous capture of C3H8 and C2H6 from C3H8/C2H6/CH4 mixtures, CoRE MOFs were evaluated under VSA condition.48 The top 10 CoRE MOFs demonstrated C2H6 + C3H8 selectivities in between 52.6 and 125.4 and adsorbent performance scores in between 223 and 390 mol/kg for the separation of C3H8/C2H6/CH4:3/7/90 mixture. Under VTSA conditions the top MOFs we identified from the QMOF database for the separation of a similar mixture, C3H8/C2H6/CH4:5/10/85, achieve higher selectivities (89.3–144.1) and adsorbent performance scores (516–799 mol/kg). Thus, we concluded that hypothetical and synthesized MOFs available in the QMOF database also have great potential for this separation.

In this study, we utilized a high-throughput computational screening approach and examined two different subsets of 1000 MOFs retrieved from the QMOF database for two different separation applications: C3H8 capture from air and simultaneous capture of C3H8 and C2H6 from CH4. Considering the types of different separation processes we investigated, such as PSA, VSA, PTSA, and VTSA processes, this work represents the most extensive study on assessing MOF adsorbents for C3H8 capture, to the best of our knowledge. Our results showed that VTSA is the optimal separation process for both gas separations: For C3H8/N2/O2/Ar separation, 998 out of 1000 MOFs that we studied achieved high regenerabilities (>80%) and adsorbent performance scores up to 7 × 104 mol/kg. For C3H8/C2H6/CH4 separation, 764 out of 1000 MOFs surpassed the same regenerability limit and exhibited high adsorbent performance scores up to 799 mol/kg. For C3H8/N2/O2/Ar separation, all the top-performing MOFs exhibit superior C3H8 selectivities up to 3.5 × 104 (1.6 × 104) compared to commercial zeolite, MFI, at 1(10) bar and 298 K. For C3H8/C2H6/CH4 separation, the top MOFs identified from the QMOF database demonstrated very high C2H6+C3H8 selectivities, 82–144 (38.2–194) at 1(10) bar, 298 K, outperforming previously reported selectivities of synthesized MOFs in the literature. Detailed structural and chemical analyses of the top MOFs revealed that Al- and Zn-based MOFs having aromatic linkers with narrow pores and low porosities exhibit a superior C3H8 separation performance. We believe that these results will be useful for selecting the best adsorbent materials and process conditions for adsorption-based capture of propane from air and natural gas mixtures.

The supplementary material contains the distribution of Henry’s selectivity and selection of 1000 MOFs for C3H8 capture from air and C2H6 + C3H8 separation from CH4; comparison of the simulated and experimentally reported C3H8 isotherms of MOFs; expressions for the performance metrics; illustration of the linker subunits obtained from the top 20 MOFs; ranges of the performance metrics for 1000 MOFs for C3H8 capture from air and C2H6 + C3H8 capture from the natural gas mixture under different process conditions; properties of the top 20 MOFs for C3H8/C2H6/CH4 and C3H8/N2/O2/Ar separations; and the list of reported selectivity values of synthesized MOFs for C3H8 capture from C2H6 and CH4.

S.K. acknowledges ERC-2017-Starting Grant. This study received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (ERC-2017-Starting Grant, Grant Agreement No. 756489-COSMOS). This work was also supported by the Scientific and Technological Research Council of Turkey (TUBITAK) under the 1001-Scientific and Technological Research Projects Funding Program (Project No: 122Z536). The authors declare no competing financial interest. The authors thank Dr. Cigdem Altintas for her fruitful discussions.

The authors have no conflicts to disclose.

Goktug Ercakir: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Gokhan Onder Aksu: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Seda Keskin: Conceptualization (lead); Funding acquisition (lead); Investigation (lead); Methodology (lead); Project administration (lead); Supervision (lead); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available within the article and its supplementary material.

1.
M.
Rood Werpy
,
A.
Burnham
, and
K.
Bertram
, “
Propane vehicles: status, challenges, and opportunities
,”
ANL/ESD/10-2
(
US Department of Energy
,
2010
).
2.
P.
Purohit
,
L.
Höglund-Isaksson
,
N.
Borgford-Parnell
,
Z.
Klimont
, and
C. J.
Smith
, “
The key role of propane in a sustainable cooling sector
,”
Proc. Natl. Acad. Sci. U. S. A.
119
(
34
),
e2206131119
(
2022
).
3.
M.
Monai
,
M.
Gambino
,
S.
Wannakao
, and
B. M.
Weckhuysen
, “
Propane to olefins tandem catalysis: A selective route towards light olefins production
,”
Chem. Soc. Rev.
50
(
20
),
11503
(
2021
).
4.
G.
Etiope
and
P.
Ciccioli
, “
Earth’s degassing: A missing ethane and propane source
,”
Science
323
(
5913
),
478
(
2009
).
5.
S.
Picaud-Vannereux
,
F.
Lutin
,
E.
Favre
, and
D.
Roizard
, “
Energy efficiency of membrane vs hybrid membrane/cryogenic processes for propane recovery from nitrogen purging vents: A simulation study
,”
Sep. Purif. Technol.
240
,
116613
(
2020
).
6.
Q.
Ding
and
S.
Zhang
, “
Recent advances in the development of metal–organic frameworks for propylene and propane separation
,”
Energy Fuels
36
(
14
),
7337
(
2022
).
7.
Y.
Wu
and
B. M.
Weckhuysen
, “
Separation and purification of hydrocarbons with porous materials
,”
Angew. Chem., Int. Ed.
60
(
35
),
18930
(
2021
).
8.
P.
Pullumbi
,
F.
Brandani
, and
S.
Brandani
, “
Gas separation by adsorption: Technological drivers and opportunities for improvement
,”
Curr. Opin. Chem. Eng.
24
,
131
(
2019
).
9.
X.
Li
,
L.
Zhang
,
Z.
Yang
,
P.
Wang
,
Y.
Yan
, and
J.
Ran
, “
Adsorption materials for volatile organic compounds (VOCs) and the key factors for VOCs adsorption process: A review
,”
Sep. Purif. Technol.
235
,
116213
(
2020
).
10.
R.-B.
Lin
,
S.
Xiang
,
H.
Xing
,
W.
Zhou
, and
B.
Chen
, “
Exploration of porous metal–organic frameworks for gas separation and purification
,”
Coord. Chem. Rev.
378
,
87
(
2019
).
11.
J.
Yu
,
L.-H.
Xie
,
J.-R.
Li
,
Y.
Ma
,
J. M.
Seminario
, and
P. B.
Balbuena
, “
CO2 capture and separations using MOFs: Computational and experimental studies
,”
Chem. Rev.
117
(
14
),
9674
(
2017
).
12.
E. D.
Bloch
,
W. L.
Queen
,
R.
Krishna
,
J. M.
Zadrozny
,
C. M.
Brown
, and
J. R.
Long
, “
Hydrocarbon separations in a metal-organic framework with open iron(II) coordination sites
,”
Science
335
(
6076
),
1606
(
2012
).
13.
J.-R.
Li
,
Y.
Ma
,
M. C.
McCarthy
,
J.
Sculley
,
J.
Yu
,
H.-K.
Jeong
,
P. B.
Balbuena
, and
H.-C.
Zhou
, “
Carbon dioxide capture-related gas adsorption and separation in metal-organic frameworks
,”
Coord. Chem. Rev.
255
(
15–16
),
1791
(
2011
).
14.
S. E.
Bambalaza
,
H. W.
Langmi
,
R.
Mokaya
,
N. M.
Musyoka
, and
L. E.
Khotseng
, “
Experimental demonstration of dynamic temperature-dependent behavior of UiO-66 metal–organic framework: Compaction of hydroxylated and dehydroxylated forms of UiO-66 for high-pressure hydrogen storage
,”
ACS Appl. Mater. Interfaces
12
(
22
),
24883
(
2020
).
15.
G.
Orcajo
,
H.
Montes-Andres
,
J. A.
Villajos
,
C.
Martos
,
J. A.
Botas
, and
G.
Calleja
, “
Li-Crown ether complex inclusion in MOF materials for enhanced H2 volumetric storage capacity at room temperature
,”
Int. J. Hydrogen Energy
44
(
35
),
19285
(
2019
).
16.
D.
Alezi
,
Y.
Belmabkhout
,
M.
Suyetin
,
P. M.
Bhatt
,
Ł. J.
Weseliński
,
V.
Solovyeva
,
K.
Adil
,
I.
Spanopoulos
,
P. N.
Trikalitis
,
A.-H.
Emwas
, and
M.
Eddaoudi
, “
MOF crystal chemistry paving the way to gas storage needs: Aluminum-based soc-MOF for CH4, O2, and CO2 storage
,”
J. Am. Chem. Soc.
137
(
41
),
13308
(
2015
).
17.
D.
Wu
,
X.
Guo
,
H.
Sun
, and
A.
Navrotsky
, “
Thermodynamics of methane adsorption on copper HKUST-1 at low pressure
,”
J. Phys. Chem. Lett.
6
(
13
),
2439
(
2015
).
18.
S.
Tu
,
L.
Yu
,
J.
Liu
,
D.
Lin
,
Y.
Wu
,
Z.
Li
,
H.
Wang
, and
Q.
Xia
, “
Efficient CO2 capture under humid conditions on a novel amide-functionalized Fe-soc metal–organic framework
,”
ACS Appl. Mater. Interfaces
15
(
9
),
12240
(
2023
).
19.
J.
Dai
,
D.
Xie
,
Y.
Liu
,
Z.
Zhang
,
Y.
Yang
,
Q.
Yang
,
Q.
Ren
, and
Z.
Bao
, “
Supramolecular metal–organic framework for CO2/CH4 and CO2/N2 separation
,”
Ind. Eng. Chem. Res.
59
(
16
),
7866
(
2020
).
20.
M.
Feng
,
P.
Zhou
,
J.
Wang
,
X.
Wang
,
D.
Wang
, and
C.
Li
, “
Two solvent-induced in(III)-based metal–organic frameworks with controllable topology performing high-efficiency separation of C2H2/CH4 and CO2/CH4
,”
Inorg. Chem.
61
(
29
),
11057
(
2022
).
21.
W.
Dai
,
J.
Li
,
T.
Wu
,
X.-Q.
Chen
,
C.-H.
Wang
,
J.-H.
Tao
,
Q.-P.
Wang
,
C.-Y.
Wang
, and
M.-Q.
Chen
, “
Study on the adsorption properties of metal–organic frameworks (PCN-228’, PCN-230)
,”
Mater. Technol.
37
(
12
),
2032
(
2022
).
22.
M.
Chang
,
F.
Wang
,
Y.
Wei
,
Q.
Yang
,
J. X.
Wang
,
D.
Liu
, and
J. F.
Chen
, “
Separation of CH4/N2 by an ultra‐stable metal–organic framework with the highest breakthrough selectivity
,”
AIChE J.
68
(
9
),
e17794
(
2022
).
23.
B.
Zhou
,
T.
Zeng
,
Z. L.
Shi
,
G.
Zhang
,
S.
Horike
, and
Y. B.
Zhang
, “
An allosteric metal–organic framework that exhibits multiple pore configurations for the optimization of hydrocarbon separation
,”
Chem. - Asian J.
14
(
20
),
3552
(
2019
).
24.
Y.
Yuan
,
H.
Wu
,
Y.
Xu
,
D.
Lv
,
S.
Tu
,
Y.
Wu
,
Z.
Li
, and
Q.
Xia
, “
Selective extraction of methane from C1/C2/C3 on moisture-resistant MIL-142A with interpenetrated networks
,”
Chem. Eng. J.
395
,
125057
(
2020
).
25.
Y.
Sun
,
M.-Y.
Gao
,
Y.
Sun
,
D.-F.
Lu
,
F.
Wang
, and
J.
Zhang
, “
Two isostructural titanium metal–organic frameworks for light hydrocarbon separation
,”
Inorg. Chem.
60
(
18
),
13955
(
2021
).
26.
L.-Z.
Qin
,
X.-H.
Xiong
,
S.-H.
Wang
,
L.
Zhang
,
L.-L.
Meng
,
L.
Yan
,
Y.-N.
Fan
,
T.-A.
Yan
,
D.-H.
Liu
,
Z.-W.
Wei
, and
C. Y.
Su
, “
MIL-101-Cr/Fe/Fe–NH2 for efficient separation of CH4 and C3H8 from simulated natural gas
,”
ACS Appl. Mater. Interfaces
14
(
40
),
45444
(
2022
).
27.
A. A.
Lysova
,
K. A.
Kovalenko
,
A. S.
Nizovtsev
,
D. N.
Dybtsev
, and
V. P.
Fedin
, “
Efficient separation of methane, ethane and propane on mesoporous metal–organic frameworks
,”
Chem. Eng. J.
453
,
139642
(
2023
).
28.
L.
Kan
,
G.
Li
, and
Y.
Liu
, “
Highly selective separation of C3H8 and C2H2 from CH4 within two water-stable Zn5 cluster-based metal–organic frameworks
,”
ACS Appl. Mater. Interfaces
12
(
16
),
18642
(
2020
).
29.
B.
Yuan
,
X.
Wang
,
X.
Zhou
,
J.
Xiao
, and
Z.
Li
, “
Novel room-temperature synthesis of MIL-100(Fe) and its excellent adsorption performances for separation of light hydrocarbons
,”
Chem. Eng. J.
355
,
679
(
2019
).
30.
I. J.
Bruno
,
J. C.
Cole
,
P. R.
Edgington
,
M.
Kessler
,
C. F.
Macrae
,
P.
McCabe
,
J.
Pearson
, and
R.
Taylor
, “
New software for searching the Cambridge structural database and visualizing crystal structures
,”
Acta Crystallogr., Sect. B: Struct. Sci.
58
(
3
),
389
(
2002
).
31.
C. R.
Groom
,
I. J.
Bruno
,
M. P.
Lightfoot
, and
S. C.
Ward
, “
The Cambridge structural database
,”
Acta Crystallogr., Sect. B: Struct. Sci., Cryst. Eng. Mater.
72
(
2
),
171
(
2016
).
32.
P. Z.
Moghadam
,
A.
Li
,
S. B.
Wiggin
,
A.
Tao
,
A. G. P.
Maloney
,
P. A.
Wood
,
S. C.
Ward
, and
D.
Fairen-Jimenez
, “
Development of a Cambridge structural database subset: A collection of metal–organic frameworks for past, present, and future
,”
Chem. Mater.
29
(
7
),
2618
(
2017
).
33.
R.
Anderson
and
D. A.
Gómez-Gualdrón
, “
Deep learning combined with IAST to screen thermodynamically feasible MOFs for adsorption-based separation of multiple binary mixtures
,”
J. Chem. Phys.
154
(
23
),
234102
(
2021
).
34.
E.
Ren
,
P.
Guilbaud
, and
F.-X.
Coudert
, “
High-throughput computational screening of nanoporous materials in targeted applications
,”
Digital Discovery
1
(
4
),
355
(
2022
).
35.
H.
Daglar
and
S.
Keskin
, “
Recent advances, opportunities, and challenges in high-throughput computational screening of MOFs for gas separations
,”
Coord. Chem. Rev.
422
,
213470
(
2020
).
36.
Z.
Li
,
B. J.
Bucior
,
H.
Chen
,
M.
Haranczyk
,
J. I.
Siepmann
, and
R. Q.
Snurr
, “
Machine learning using host/guest energy histograms to predict adsorption in metal–organic frameworks: Application to short alkanes and Xe/Kr mixtures
,”
J. Chem. Phys.
155
(
1
),
014701
(
2021
).
37.
Y. G.
Chung
,
J.
Camp
,
M.
Haranczyk
,
B. J.
Sikora
,
W.
Bury
,
V.
Krungleviciute
,
T.
Yildirim
,
O. K.
Farha
,
D. S.
Sholl
, and
R. Q.
Snurr
, “
Computation-ready, experimental metal–organic frameworks: A tool to enable high-throughput screening of nanoporous crystals
,”
Chem. Mater.
26
(
21
),
6185
(
2014
).
38.
Y. G.
Chung
,
E.
Haldoupis
,
B. J.
Bucior
,
M.
Haranczyk
,
S.
Lee
,
H. D.
Zhang
,
K. D.
Vogiatzis
,
M.
Milisavljevic
,
S. L.
Ling
,
J. S.
Camp
,
B.
Slater
,
J. I.
Siepmann
,
D. S.
Sholl
, and
R. Q.
Snurr
, “
Advances, updates, and analytics for the computation-ready, experimental metal–organic framework database: CoRE MOF 2019
,”
J. Chem. Eng. Data
64
(
12
),
5985
(
2019
).
39.
P. G.
Boyd
,
A.
Chidambaram
,
E.
Garcia-Diez
,
C. P.
Ireland
,
T. D.
Daff
,
R.
Bounds
,
A.
Gladysiak
,
P.
Schouwink
,
S. M.
Moosavi
,
M. M.
Maroto-Valer
,
J. A.
Reimer
,
J. A. R.
Navarro
,
T. K.
Woo
,
S.
Garcia
,
K. C.
Stylianou
, and
B.
Smit
, “
Data-driven design of metal–organic frameworks for wet flue gas CO2 capture
,”
Nature
576
(
7786
),
253
(
2019
).
40.
P. G.
Boyd
and
T. K.
Woo
, “
A generalized method for constructing hypothetical nanoporous materials of any net topology from graph theory
,”
CrystEngComm
18
(
21
),
3777
(
2016
).
41.
Y. S.
Lan
,
T.
Yan
,
M. M.
Tong
, and
C. L.
Zhong
, “
Large-scale computational assembly of ionic liquid/MOF composites: Synergistic effect in the wire-tube conformation for efficient CO2/CH4 separation
,”
J. Mater. Chem. A
7
(
20
),
12556
(
2019
).
42.
Y. J.
Colón
,
D. A.
Gomez-Gualdron
, and
R. Q.
Snurr
, “
Topologically guided, automated construction of metal–organic frameworks and their evaluation for energy-related applications
,”
Cryst. Growth Des.
17
(
11
),
5801
(
2017
).
43.
X.
Xue
,
M.
Cheng
,
S.
Wang
,
S.
Chen
,
L.
Zhou
,
C.
Liu
, and
X.
Ji
, “
High-throughput screening of metal–organic frameworks assisted by machine learning: Propane/propylene separation
,”
Ind. Eng. Chem. Res.
62
(
2
),
1073
(
2023
).
44.
B. C.
Yeo
,
D.
Kim
,
H.
Kim
, and
S. S.
Han
, “
High-throughput screening to investigate the relationship between the selectivity and working capacity of porous materials for propylene/propane adsorptive separation
,”
J. Phys. Chem. C
120
(
42
),
24224
(
2016
).
45.
H.
Tang
,
Q.
Xu
,
M.
Wang
, and
J.
Jiang
, “
Rapid screening of metal–organic frameworks for propane/propylene separation by synergizing molecular simulation and machine learning
,”
ACS Appl. Mater. Interfaces
13
(
45
),
53454
(
2021
).
46.
C.
Gu
,
J.
Liu
, and
D. S.
Sholl
, “
High-throughput screening of anion-pillared metal–organic frameworks for the separation of light hydrocarbons
,”
J. Phys. Chem. C
125
(
36
),
20076
(
2021
).
47.
V. A.
Solanki
and
B.
Borah
, “
Computational screening of metal–organic framework structures for separation of propane/propene mixture
,”
Mol. Simul.
46
(
17
),
1362
(
2020
).
48.
Y. K.
Ponraj
and
B.
Borah
, “
High-throughput computational screening of metal–organic frameworks for the separation of methane from ethane and propane
,”
J. Phys. Chem. C
125
(
3
),
1839
(
2021
).
49.
X.
Yuan
,
L.
Li
,
Z.
Shi
,
H.
Liang
,
S.
Li
, and
Z.
Qiao
, “
Molecular-fingerprint machine-learning-assisted design and prediction for high-performance MOFs for capture of NMHCs from air
,”
Adv. Powder Mater.
1
(
3
),
100026
(
2022
).
50.
A. S.
Rosen
,
V.
Fung
,
P.
Huck
,
C. T.
O’Donnell
,
M. K.
Horton
,
D. G.
Truhlar
,
K. A.
Persson
,
J. M.
Notestein
, and
R. Q.
Snurr
, “
High-throughput predictions of metal–organic framework electronic properties: Theoretical challenges, graph neural networks, and data exploration
,”
npj Comput. Mater.
8
(
1
),
112
(
2022
).
51.
A. S.
Rosen
,
S. M.
Iyer
,
D.
Ray
,
Z.
Yao
,
A.
Aspuru-Guzik
,
L.
Gagliardi
,
J. M.
Notestein
, and
R. Q.
Snurr
, “
Machine learning the quantum-chemical properties of metal–organic frameworks for accelerated materials discovery
,”
Matter
4
(
5
),
1578
(
2021
).
52.
Z.
Zhang
,
C.
Zhang
,
Y.
Zhang
,
S.
Deng
,
Y.-F.
Yang
,
A.
Su
, and
Y.-B.
She
, “
Predicting band gaps of MOFs on small data by deep transfer learning with data augmentation strategies
,”
RSC Adv.
13
(
25
),
16952
(
2023
).
53.
S. M.
Moosavi
,
B. Á.
Novotny
,
D.
Ongari
,
E.
Moubarak
,
M.
Asgari
,
Ö.
Kadioglu
,
C.
Charalambous
,
A.
Ortega-Guerrero
,
A. H.
Farmahini
,
L.
Sarkisov
et al, “
A data-science approach to predict the heat capacity of nanoporous materials
,”
Nat. Mater.
21
(
12
),
1419
(
2022
).
54.
T. F.
Willems
,
C.
Rycroft
,
M.
Kazi
,
J. C.
Meza
, and
M.
Haranczyk
, “
Algorithms and tools for high-throughput geometry-based analysis of crystalline porous materials
,”
Microporous Mesoporous Mater.
149
(
1
),
134
(
2012
).
55.
S.
Sircar
, “
Basic research needs for design of adsorptive gas separation processes
,”
Ind. Eng. Chem. Res.
45
(
16
),
5435
(
2006
).
56.
D.
Dubbeldam
,
S.
Calero
,
D. E.
Ellis
, and
R. Q.
Snurr
, “
RASPA: Molecular simulation software for adsorption and diffusion in flexible nanoporous materials
,”
Mol. Simul.
42
(
2
),
81
(
2016
).
57.
C.
Altintas
,
I.
Erucar
, and
S.
Keskin
, “
High-throughput computational screening of the metal organic framework database for CH4/H2 separations
,”
ACS Appl. Mater. Interfaces
10
(
4
),
3668
(
2018
).
58.
S.
Calero
and
P.
Gomez-Alvarez
, “
On the performance of FAU and MFI zeolites for the adsorptive removal of a series of volatile organic compounds from air using molecular simulation
,”
Phys. Chem. Chem. Phys.
17
(
39
),
26451
(
2015
).
59.
J.
Peng
,
J.
Zhong
,
Z.
Liu
,
H.
Xi
,
J.
Yan
,
F.
Xu
,
X.
Chen
,
X.
Wang
,
D.
Lv
, and
Z.
Li
, “
Multivariate metal–organic frameworks prepared by simultaneous metal/ligand exchange for enhanced C2–C3 selective recovery from natural gas
,”
ACS Appl. Mater. Interfaces
15
(
35
),
41466
(
2023
).
60.
L.
Lan
,
N.
Lu
,
J.-C.
Yin
,
Q.
Gao
,
F.
Lang
,
Y.-H.
Zhang
,
H.-X.
Nie
,
N.
Li
, and
X.-H.
Bu
, “
Simultaneous extraction of C3H8 and C2H6 from ternary C3H8/C2H6/CH4 mixtures in an ultra-microporous metal–organic framework
,”
Chem. Eng. J.
476
,
146750
(
2023
).
61.
M. G.
Martin
and
J. I.
Siepmann
, “
Transferable potentials for phase equilibria. 1. United-atom description of n-alkanes
,”
J. Phys. Chem. B
102
(
14
),
2569
(
1998
).
62.
J. J.
Potoff
and
J. I.
Siepmann
, “
Vapor-liquid equilibria of mixtures containing alkanes, carbon dioxide, and nitrogen
,”
AIChE J.
47
(
7
),
1676
(
2001
).
63.
L.
Zhang
and
J. I.
Siepmann
, “
Direct calculation of Henry’s law constants from Gibbs ensemble Monte Carlo simulations: Nitrogen, oxygen, carbon dioxide and methane in ethanol
,”
Theor. Chem. Acc.
115
(
5
),
391
(
2006
).
64.
B. L.
Eggimann
,
Y.
Sun
,
R. F.
DeJaco
,
R.
Singh
,
M.
Ahsan
,
T. R.
Josephson
, and
J. I.
Siepmann
, “
Assessing the quality of molecular simulations for vapor–liquid equilibria: An analysis of the TraPPE database
,”
J. Chem. Eng. Data
65
(
3
),
1330
(
2019
).
65.
E.
García-Pérez
,
J.
Parra
,
C. O.
Ania
,
D.
Dubbeldam
,
T. J.
Vlugt
,
J.
Castillo
,
P.
Merkling
, and
S.
Calero
, “
Unraveling the argon adsorption processes in MFI-type zeolite
,”
J. Phys. Chem. C
112
(
27
),
9976
(
2008
).
66.
S. L.
Mayo
,
B. D.
Olafson
, and
W. A.
Goddard
, “
DREIDING: A generic force field for molecular simulations
,”
J. Phys. Chem.
94
(
26
),
8897
(
1990
).
67.
A. K.
Rappé
,
C. J.
Casewit
,
K.
Colwell
,
W. A.
Goddard
III
, and
W. M.
Skiff
, “
UFF, A full periodic table force field for molecular mechanics and molecular dynamics simulations
,”
J. Am. Chem. Soc.
114
(
25
),
10024
(
1992
).
68.
T. A.
Manz
and
N. G.
Limas
, “
Introducing DDEC6 atomic population analysis: Part 1. Charge partitioning theory and methodology
,”
RSC Adv.
6
(
53
),
47771
(
2016
).
69.
N. G.
Limas
and
T. A.
Manz
, “
Introducing DDEC6 atomic population analysis: Part 2. Computed results for A wide range of periodic and nonperiodic materials
,”
RSC Adv.
6
(
51
),
45727
(
2016
).
70.
C.
Altintas
and
S.
Keskin
, “
Computational screening of MOFs for C2H6/C2H4 and C2H6/CH4 separations
,”
Chem. Eng. Sci.
139
,
49
(
2016
).
71.
Z.
Sumer
and
S.
Keskin
, “
Adsorption- and membrane-based CH4/N2 separation performances of MOFs
,”
Ind. Eng. Chem. Res.
56
(
30
),
8713
(
2017
).
72.
I. B.
Orhan
,
H.
Daglar
,
S.
Keskin
,
T. C.
Le
, and
R.
Babarao
, “
Prediction of O2/N2 selectivity in metal–organic frameworks via high-throughput computational screening and machine learning
,”
ACS Appl. Mater. Interfaces
14
(
1
),
736
(
2021
).
73.
Z.
Sumer
and
S.
Keskin
, “
Molecular simulations of MOF adsorbents and membranes for noble gas separations
,”
Chem. Eng. Sci.
164
,
108
(
2017
).
74.
U.
Böhme
,
B.
Barth
,
C.
Paula
,
A.
Kuhnt
,
W.
Schwieger
,
A.
Mundstock
,
J. R.
Caro
, and
M.
Hartmann
, “
Ethene/ethane and propene/propane separation via the olefin and paraffin selective metal–organic framework adsorbents CPO-27 and ZIF-8
,”
Langmuir
29
(
27
),
8592
(
2013
).
75.
M.
Hartmann
,
U.
Böhme
,
M.
Hovestadt
, and
C.
Paula
, “
Adsorptive separation of olefin/paraffin mixtures with ZIF-4
,”
Langmuir
31
(
45
),
12382
(
2015
).
76.
R.
Shi
,
D.
Lv
,
Y.
Chen
,
H.
Wu
,
B.
Liu
,
Q.
Xia
, and
Z.
Li
, “
Highly selective adsorption separation of light hydrocarbons with a porphyrinic zirconium metal-organic framework PCN-224
,”
Sep. Purif. Technol.
207
,
262
(
2018
).
77.
W. G.
Cui
,
T. L.
Hu
, and
X. H.
Bu
, “
Metal–organic framework materials for the separation and purification of light hydrocarbons
,”
Adv. Mater.
32
(
3
),
1806445
(
2020
).
78.
B.
Yang
,
D.
Hawley
,
J.
Yao
,
C.
May
,
J. E.
Mendez-Arroyo
, and
D. H.
Ess
, “
Demonstration of high-throughput building block and composition analysis of metal–organic frameworks
,”
J. Chem. Inf. Model.
62
(
19
),
4672
(
2022
).
79.
H.
Tang
and
J.
Jiang
, “
In silico screening and design strategies of ethane‐selective metal–organic frameworks for ethane/ethylene separation
,”
AIChE J.
67
(
3
),
e17025
(
2021
).
80.
G.
Ercakir
,
G. O.
Aksu
,
C.
Altintas
, and
S.
Keskin
, “
Hierarchical computational screening of quantum metal–organic framework database to identify metal–organic frameworks for volatile organic-compound capture from air
,”
ACS Eng. Au
3
(
6
),
488
497
(
2023
).
81.
S.-Q.
Yang
,
F.-Z.
Sun
,
R.
Krishna
,
Q.
Zhang
,
L.
Zhou
,
Y.-H.
Zhang
, and
T.-L.
Hu
, “
Propane-trapping ultramicroporous metal–organic framework in the low-pressure area toward the purification of propylene
,”
ACS Appl. Mater. Interfaces
13
(
30
),
35990
(
2021
).
82.
Y.
Chen
,
H.
Wu
,
D.
Lv
,
N.
Yuan
,
Q.
Xia
, and
Z.
Li
, “
A pillar-layer metal–organic framework for efficient adsorption separation of propylene over propane
,”
Sep. Purif. Technol.
204
,
75
(
2018
).

Supplementary Material