computational study of the effect of organic linkers on natural gas upgrading in metal–organic frameworks

1、Computational study of the effect of organic linkers on natural gas upgrading in metalorganic frameworksComputational study of the effect of organic linkers on natural gas upgradingin metalorganic frameworksWei Mu, Dahuan Liu, Qingyuan Yang*, Chongli ZhongLab of Computational Chemistry, Department o

2、f Chemical Engineering, Beijing University of Chemical Technology, Beijing 100029, Chinaa r t i c l ei n f oArticle history:Received 7 April 2021Received in revised form 2 September 2021Accepted 21 October 2021Available online 23 October 2021Keywords:SeparationMolecular simulationDensity functional

3、theoryMetalorganic frameworksa b s t r a c tIn this work a hierarchical multiscale approach combining grand canonical Monte Carlo simulation anddensity functional theory calculation was performed to study the effect of the chemical properties of nineorganic linkers on CO2/CH4mixture separation in me

4、talorganic frameworks (MOFs). The computationalresults show that the organic linkers decorated with the electron-donating groups can strengthen the dis-tribution of the electrostatic field in the pores of MOFs, and greatly enhance the adsorption selectivity ofCO2/CH4mixture in MOFs. This enhancement

5、 becomes stronger with the increase of the electron-donat-ing ability of groups. In addition, this work also demonstrates that the negative steric hindrance effects onthe separation behavior should be considered when the organic linkers are modified with multiple sub-stitutions in designing new mate

6、rials. The knowledge obtained is expected to provide useful informationfor tailoring the electrostatic properties of MOFs for separation of various gas mixture systems of practi-cal importance.? 2021 Elsevier Inc. All rights reserved.1. IntroductionThe combustion of vast amounts of carbon-based foss

7、il fuels hasled to one of the most serious global environmental problems. Nat-ural gas mainly composed of methane (CH4) is thought as one ofthe cleanest carbon fuels due to the low emission of carbonaceousgreenhouse gas. However, the presence of carbon dioxide (CO2) innatural gas reduces its calorie

8、 content and also induces pipelinecorrosion. A widely recognized strategy for the feasible CO2cap-ture technology is that minimal environmental impact and lowcosts should be achieved. The currently commercial amine-basedsystems used for CO2removal bear several drawbacks such as cor-rosion control an

9、d considerable energy needed for solvent regener-ation 1. Pressure swing adsorption (PSA) technology based on theporous adsorbents is known to be one of the most efficient andaffordable processes for CO2removal from natural gas 2. Manyresearch efforts have been carried out to identify and to develop

10、a suitable porous material with high CO2 affinity and capacity.Although the conventionally involved zeolite materials like zeo-lites 13X and NaY have been claimed to have high performancefor separation of CO2from mixtures 3,4, it is difficult to regener-ate them without significant heating which lea

11、ds to low productiv-ity and great expense 5.Owing to their flexibility to design through control of the archi-tecture and chemical functionality of the pores, metalorganicframeworks (MOFs) have emerged as a new family of nanoporousmaterials that offer promising applications for gas storage and sep-a

12、ration 6,7. Up to date, many studies have been carried out toinvestigate the removal of CO2from natural gas in MOFs. For exam-ple, experiments have been carried out to study the combination of the grand canonical Monte Carlo simulations(GCMC) and density functional theory (DFT) calculations, a syste

13、m-atic computational study was performed in this work to study theseparation of CO2/CH4mixture in MOFs by introducing four typesof functional groups (CH3, F, OH and NH2) into the organic lin-ker in IRMOF-1. The knowledge obtained is expected to provideuseful information for tailoring the electrostat

14、ic properties ofnew MOFs for separation of various gas mixture systems of practi-cal importance.2. Models and computational details2.1. MOF structuresIn the present work, IRMOF-1, a representative of the high por-ous MOFs, was selected as the host material, which has beenwidely studied both experime

15、ntally 19,20 and theoretically1416,18,21. The guest-free framework structure of IRMOF-1was constructed from its experimental single-crystal X-ray diffrac-tion (XRD) data 22, and the unit cell crystal structure is shown inFig. 1. As can be seen from this figure, the crystal structure of IR-MOF-1 is v

16、ery simple that consists of each oxide-centered Zn4Otetrahedron connected by six 1,4-benzenedicarboxylate (BDC)linkers to form a three-dimensional porous cubic framework.In order to understand how the electron-combining abilities ofthe organic affect the separation of CO2/CH4mixture in MOFs, fourtyp

17、es of functional groups were adopted in this work to substitutethe H atoms in the organic linkers of IRMOF-1, resulting in eightdifferent organic linkers as shown in Fig. 2. The eight designedMOF materials with these organic linkers were denoted as a gen-eral formula, IRMOF-(XX)n, where XX and n den

18、ote the type andnumber of the functional groups used to substitute the H atomsin the BDC linkers of IRMOF-1, respectively. It should be pointedout that the designed IRMOF-NH2is identical to the available IR-MOF-3 22.2.2. Computational methods2.2.1. Geometry optimization of the full periodic MOF stru

19、cturesIn order to study the structural stabilities of the designed MOFs,DFT calculations with periodic boundary conditions were per-formed to optimize the structures of their unit cells. In these calcu-lations, the gradient corrected (GGA) correlation functional ofPerdew and Wang (PW91) was used wit

20、h the precise numericalbasis set DNP (double numerical plus polarization). All of the calcu-lations were carried out using the DMol3code as implemented inthe Materials Studio package 23. The above DFT methodologyhas been successfully employed to study the structures of MOFs24,25.2.2.2. Atomic partia

21、l charge calculationsIn all simulations, atomic partial charges for the MOFs are re-quired as input parameters. Similar to the method in the early pio-neering works by Ganz et al. 21 and others 26, cluster modelswere cut from the optimized structures of the eight designed MOFsand the experimental on

22、e of IRMOF-1 as shown in Fig. 3, where thecluster models of IRMOF-1 and IRMOF-(NH2)3together with thespecified atomic types are illustrated as the examples. For all ofthe cleaved clusters, the terminations of them were saturated withmethyl (CH3) groups. Prior to calculating the atomic partialcharges

23、, these cluster models were firstly optimized by DFT meth-od. The PerdewBurkeErnzerhof (PBE) exchange-correlation func-tional along with the polarized diploid-f valence basis set (DZVPP)is applied in these calculations 27. After the optimizations, thesecluster models were used to calculate the atomi

24、c partial charges inthe frameworks of the MOFs. As done in previous works16,17,21,26, DFT calculations with B3LYP functional were carriedout to compute the charge distributions, and the basis set LANL2DZwas used for metal atom Zn, while 6 ? 31 + G? was used for theremaining atoms. The electrostatic

25、charges were used as the atom-ic partial charges, and the ChelpG method was adopted, which hasbeen recognized as the most popular and reliable electrostaticcharge calculation method 28. Details of the calculations can befound elsewhere 16,17,21. All the calculations were performedusing the GAUSSIAN

26、03 suite of programs 29, and the calculatedresults are given in Tables 1 and 2.2.2.3. Force fieldsIn this study, a single site model with the LennardJones (LJ)interaction was used to describe a CH4molecule, and the potentialparameters (rCH4= 3.73 ? and eCH4/kB= 148.0 K) were taken fromthe TraPPE for

27、ce field 30. CO2was modeled as a rigid linear mol-ecule with three charged LJ sites located on each atom. A combina-tion of the sitesite LJ and coulombic potentials was used tocalculate the CO2CO2intermolecular interaction. The LJ potentialparameters for atom O (rO= 3.011 ? and eO/kB= 82.96 K) and a

28、tomC (rC= 2.789 ? and eC/kB= 29.66 K) in CO2molecule were takenfrom the work of Babarao and Jiang 11. The CO bond length is1.18 ?. In this model, the intrinsic quadrupole moment is approx-imately described by the partial point charges centered at each LJsite (qO= ?0.288 e and qC= 0.576 e).For the MO

29、Fs studied in this work, an atomistic representationwas used to model all of them. Similarly, the CO2-adsorbent inter-actions were described by a combination of the sitesite LJ andCoulombic potential, while the CH4-adsorbent interactions wereonly considered by the sitesite LJ potential. The LJ poten

30、tialparameters for the framework atoms in MOFs were taken fromthe universal force field (UFF) 31, as listed in Table 3. In our sim-ulations, all the LJ cross-interaction parameters were determinedby the LorentzBerthelot mixing rules.2.2.4. Simulation techniquesGrand canonical Monte Carlo (GCMC) simu

31、lations were em-ployed to calculate the adsorption of pure components and theirmixtures with equimolar composition in MOFs at 298 K. Detailson the method can be found in our previous work 16,17. Similarto previous works 1018, all the MOFs were treated as rigidframeworks, since the effects of the dyn

32、amics of MOFs become sig-nificant only when the guests are large and/or strong guesthostinteractions exist in the system at room temperature. The numberof the unit cells of MOFs adopted in the simulation cell is 2 ? 2 ? 2.At very low pressures, the molecules adsorbed in MOFs are verysmall, thus, sim

33、ulation cells with 4 ? 4 ? 4 size were used at theseFig. 1. Unit cell crystal structure of IRMOF-1 viewed along the 1 0 0 direction (Zn,blue; O, red; C, gray; H, white). (For interpretation of the references to colour in thisfigure legend, the reader is referred to the web version of this article.)W

34、. Mu et al./Microporous and Mesoporous Materials 130 (2021) 768277state points so that enough molecules are accommodated to guar-antee the simulation accuracy. A cutoff radius of 12.8 ? was ap-plied to the LJ interactions, and the long-range electrostaticinteractions were handled using the Ewald sum

35、mation technique.Periodic boundary conditions were applied in all three dimensions.Gas-phase fugacities used to perform GCMC simulations were cal-culated with PengRobinson equation of state. For each state point,GCMC simulation consisted of 10,000,000 steps to guarantee equil-ibration followed by 10

36、,000,000 steps to sample the desired ther-modynamic properties.3. Results and discussion3.1. Stability and atomic partial charges of MOFsPeriodic DFT calculations were performed to validate whetherthe structures of the eight designed MOFs are stable. From the cal-culation results, we found that the

37、unit cell length of IRMOF-NH2(i.e., IRMOF-3) is 25.703 ?, which is in good agreement with theexperimental value (25.7465 ?) 22. In addition, the DFT calcula-tion results also show that the unit cell lengths of other sevenFig. 2. Eight different organic linkers adopted in this work.Fig. 3. Model clus

38、ters used for the calculations of atomic partial charges: (a)IRMOF-1 and (b) IRMOF-(NH2)3.Table 1Atomic partial charges of IRMOF-1 and IRMOF-XX with single substitution (unit of e).Atom typesIRMOF-1IRMOF-NH2IRMOF-OHIRMOF-FIRMOF-CH3O1CZnO1C1C2C3C3XC3YC4XC4YC5H3O3CH3OC3CH3CFN3H3N?1.9961.637?0.7570.671

39、0.079?0.122?1.8611.501?0.7070.623?0.022?1.8721.503?0.6880.5980.043?1.8681.502?0.6880.6050.034?1.8471.487?0.6840.5940.0270.358?0.128?0.306?0.2140.1570.1500.333?0.138?0.342?0.2150.1930.149?0.5640.4190.287?0.130?0.270?0.1890.1670.1550.162?0.1210.273?0.1920.1610.1440.125?0.2021.062?0.202?0.8000.353Table

40、 2Atomic partial charges of IRMOF-XX with multiple substitutions (unit of e).AtomtypesIRMOF-(OH)2IRMOF-(NH2)2IRMOF-(NH2)3IRMOF-(NH2)4O1CZnO1C1C2C3C3XC3YC4C4XC4YC5H3O3CH3ON3H3N?1.8721.500?0.6900.6330.0830.265?1.8761.513?0.8000.8320.1660.531?1.8711.532?0.8000.837?0.302?1.8781.545?0.9161.076?0.4210.328

41、0.3540.364?0.327?0.3250.264?0.394?0.0970.2180.181?0.5450.4020.211?1.120.473?0.9060.403?0.9940.44878W. Mu et al./Microporous and Mesoporous Materials 130 (2021) 7682new designed MOF materials lie in the range from 25.61 to 25.68 ?.Thus, the unit cell lengths of the eight designed MOFs are veryclose t

42、o that of IRMOF-1 (25.832 ?). Fig. 4 shows the optimizedunit cell crystal structures of IRMOF-(OH)2and IRMOF-(NH2)4asexamples, where the unit cell lengths are 25.673 and 25.612 ?,respectively. As can be seen from this figure, the optimized crystalstructures do not collapse, and keep the similar thre

43、e-dimensionalporous cubic framework as that of IRMOF-1. These results indicatethat the structures of these eight MOFs are stable, and can be fur-ther used for the following investigations in this work.Based on the DFT calculations, the atomic partial charges ineach MOF studied in this work were calc

44、ulated, and the resultsare presented in Tables 1 and 2, where Table 1 shows the resultsfor IRMOF-1 and IRMOF-XX (XX = NH2, OH, F and CH3) withsingle substitution in the organic linker, while Table 2 shows theresults for IRMOF-(XX)n(XX = OH and NH2) with multiple substi-tutions. As shown in Tables 1

45、and 2, for each IRMOF-(NH2)nwithdifferent number of substituent group NH2, although the chemi-cal environments for the atoms N are somewhat different, theiratomic partial charges are very close. Thus, the atoms N in each IR-MOF-(NH2)nmaterial are classified as the same atomic types withequal atomic

46、partial charges. In addition, Tables 1 and 2 also showthat, the atomic partial charges in the eight designed MOFs aremuch different from those in the host material IRMOF-1, especiallyin the MOFs modified by NH2and OH groups. Since the organiclinkers have been identified as the secondary adsorption s

47、ites forthe adsorption of adsorbate molecules in MOFs 1018, variationsof the chemical properties of the organic linkers by modificationshould have great impacts on the separation of the mixture consid-ered in this work.3.2. Validation of the force fieldTo confirm the reliability of the force field a

48、dopted in this work,the adsorption isotherms of pure CO2and CH4in IRMOF-1 weresimulated at 298 K and compared with their experimental data,as shown in Fig. 5a. It can be found in this figure that the simulatedisotherms are both in good agreement with the correspondingexperimental data of pure CO2and

49、 CH4in IRMOF-1 32,33. Forthe other eight MOF materials, to the best of our knowledge, thereare no experimental data for the adsorption isotherms of CO2andCH4in them. In order to further validate whether the above forcefield also works for them, the adsorption of H2at 77 K in IRMOF-NH2(i.e., IRMOF-3)

50、 was simulated and compared with the avail-able experimental data 34. The potential model for H2moleculeused here was the same as that in our previous work 35. The re-sults shown in Fig. 5b indicate that the simulations enable goodreproduction of the experimental isotherm. Therefore, it could bereas

51、onably thought that the force fields adopted in this work arereliable to investigate the adsorption behaviors of CO2/CH4mixturein IRMOF-1 and other eight derivative MOF materials.3.3. Adsorption selectivity in MOFsIn separation processes, a good indication of the separation abil-ity is the selectivi

52、ty of a porous material for different componentsin mixtures. The selectivity for component A relative to componentB is defined by S = (xA/xB)(yB/yA), where xAand xBare the mole frac-tions of components A and B in the adsorbed phase, respectively,while yAand yBare the mole fractions of components A a

53、nd B inthe bulk phase, respectively.Fig. 6a compares the simulated selectivities of CO2from equi-molar CO2/CH4 mixture in IRMOF-1 and IRMOF-XX with singlesubstitutions (XX = NH2, OH, F and CH3) at 298 K, where theerror bars shown on the simulated data are indicative of the statis-tical uncertainties

54、, and the heights of them are comparable to thesizes of the data symbols. In these MOFs, functional groups NH2, OH and CH3are the electron-donating groups, while F is theelectron-withdrawing group. Among them, the electron-donatingability follows the order NH2> OH > CH3, and this abili

55、ty of CH3 group is very weak. Compared with IRMOF-1, the resultsshown in Fig. 6a clearly show that the selectivities in IRMOF-NH2and IRMOF-OH are the highest, and the selectivities in IRMOF-Fand IRMOF-CH3are comparative to those in IRMOF-1 with onlyslightly larger values. These observations indicate

56、 that the intro-Table 3LJ potential parameters for the framework atoms in MOFs studied in this work.LJ parametersMOF_ZnMOF_CMOF_OMOF_HMOF_NMOF_Fr (? Ae=kB0)2.463.433.122.573.263.0062.4052.8430.1922.1434.6925.14Fig. 4. Optimized unit cell crystal structures of: (a) IRMOF-(OH)2and (b) IRMOF-(NH2)4.0.0

57、0.51.01.52.02.5051015202530Uptake mmol/gPressure MPa CO2 (Exp.) CO2 (Simulated) CH4 (Exp.) CH4 (Simulated)(a)0.000.020.04Pressure MPa0.060.080.100.00.51.01.52.0Hydrogen Uptake wt% H2 (Exp.) H2 (Simulated)(b)Fig. 5. Comparison of the simulated and experimental 3234 adsorption isotherms of: (a) pure C

58、O2and CH4in IRMOF-1 at 298 K and (b) H2in IRMOF-NH2(i.e., IRMOF-3) at77 K.W. Mu et al./Microporous and Mesoporous Materials 130 (2021) 768279duction of functional groups with stronger electron-donating abil-ity into the organic linkers can enhance the adsorption selectivityof CO2/CH4mixture in MOFs,

59、 while a very weak effect occurs forthe introduction of the electron-withdrawing group.To compare the effects of the electron-donating abilities of thefunctional groups NH2and OH on the selectivities of CO2fromCO2/CH4mixture, Fig. 6b shows the selectivities in IRMOF-(XX)2with double substitutions (XX = NH2and OH), and the error barsare also presented to indicate the statistical uncertainties. Obvi-ously, the selectivities in IRMOF-(NH2)2are higher