DOI: 10.1002/cssc.201000080

A Bio-Metal–Organic Framework for Highly Selective CO2Capture: A Molecular Simulation StudyYifei Chen and Jianwen Jiang*[a]

Introduction

The atmospheric CO2 concentration has increased from 280 to385 ppm since the industrial revolution, largely due to fossilfuel combustion and deforestation.[1] It is generally believedthat anthropogenic CO2 emissions have caused a global sur-face temperature rise of 0.74 8C over the past century. Current-ly carbon capture and sequestration (CCS) is not only of scien-tific interest, but also a societal issue for environmental protec-tion. A key process in CCS is the capture of CO2 from pre-com-bustion shifted syngas and post-combustion flue gas. A hand-ful of approaches have been proposed for CO2 captureincluding cryogenic distillation, amine scrubbing, membraneseparation, and sorbent adsorption. Cryogenic distillation is en-ergetically intensive because a phase transition occurs. Aminescrubbing has a deficiency in solvent regeneration and causesequipment corrosion. Membrane separation is bound by a per-meability–selectivity trade off. Compared to other methods,CO2 capture by adsorption in porous materials has a considera-bly higher energy efficiency, low capital cost, large separationcapability, and can easily be scaled up.

To achieve high efficiency separation of a gas mixture by ad-sorption, the selection of a specialized adsorbent is critical. Inthe past, a large number of experimental and simulation stud-ies have been reported on CO2 capture for shifted syngas andflue gas. In most cases, shifted syngas was mimicked by a CO2/H2 mixture and flue gas by a CO2/N2 mixture. For example, theselectivities of CO2/H2 and CO2/N2 mixtures in dehydrated Na-4A zeolite were predicted to decrease with increasing pressureat room temperature.[2] The gases CO2 and N2, as single com-ponents and as a binary mixture, were simulated in three zeo-lites with identical chemical composition but different porestructures.[3] The effects of various operating conditions were

investigated for CO2/N2 separation in MFI and FAU mem-branes.[4, 5] A recent review summarized various zeolites andother adsorbent materials for CO2 capture.[6]

MOFs have emerged as an important class of hybrid porousmaterials.[7] Composed of metal–oxide clusters and organiclinkers, MOFs possess extremely large surface areas (up to5000 m2 g�1) and some of the high porosities (up to 90 %) re-corded for crystalline materials. The variation of metal oxidesand the judicious choice of organic linkers allow the pore size,volume, and functionality to be tailored in a rational mannerfor designable architectures. There has been an increasing in-terest in using MOFs as sorbents for the separation of CO2/H2

and CO2/N2 mixtures towards CO2 capture. Simulation was per-formed for a CO2/N2 mixture in Cu-BTC, and CO2/H2 mixture incatenated isoreticular (IR) MOFs.[8, 9] MOFs, including IRMOF-1and Cu-BTC, were examined for CO2 capture by predicting mix-ture adsorption and diffusion properties.[10, 11] The selectivity ofCO2/N2 was found to increase through the coordination of H2Omolecules to the open metal sites in Cu-BTC, and by the pres-ence of highly polar ligands in a zinc-paddlewheel MOF.[12, 13] Acomparative study was reported for the separation of CO2/N2

in a series of MOFs and zeolites.[14] The adsorption of gas mix-tures (shifted syngas, flue gas, and natural gas) was simulated

A recently synthesized bio-metal–organic framework (bio-MOF-11) is investigated for CO2 capture by molecular simulation.The adenine biomolecular linkers in bio-MOF-11 contain Lewisbasic amino and pyrimidine groups as the preferential adsorp-tion sites. The simulated and experimental adsorption iso-therms of pure CO2, H2, and N2 are in perfect agreement. Bio-MOF-11 exhibits larger adsorption capacities compared to nu-merous zeolites, activated carbons, and MOFs, which is attrib-uted to the presence of multiple Lewis basic sites and nano-sized channels. The results for the adsorption of CO2/H2 andCO2/N2 mixtures in bio-MOF-11 show that CO2 is more domi-nantly adsorbed than H2 and N2. With increasing pressure, the

selectivity of CO2/H2 initially increases owing to the strong in-teractions between CO2 and the framework, and then decreas-es as a consequence of the entropy effect. However, the selec-tivity of CO2/N2 monotonically increases with increasing pres-sure and finally reaches a constant. The selectivities in bio-MOF-11 are higher than in many nanoporous materials. Thesimulation results also reveal that a small amount of H2O has anegligible effect on the separation of CO2/H2 and CO2/N2 mix-tures. The simulation study provides quantitative microscopicinsight into the adsorption mechanism in bio-MOF-11 and sug-gests that bio-MOF-11 may be interesting for pre- and post-combustion CO2 capture.

[a] Y. Chen, Prof. J. JiangDepartment of Chemical and Biomolecular EngineeringNational University of Singapore, 117576 (Singapore)Fax: (+ 65) 67791936E-mail : chejj@nus.edu.sg

Supporting Information for this article is available on the WWW underhttp://dx.doi.org/10.1002/cssc.201000080.

982 � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 982 – 988

in cationic soc-MOF and anionic rho-zeolite-MOF (ZMOF), andunprecedented high selectivities were predicted.[15, 16]

Rosi and co-workers recently developed a unique set of bio-MOFs based on adenine.[17, 19] Adenine, a purine nucleobase, isan important natural nitrogen-based heterocycle consisting offour imino nitrogen atoms and one exocyclic amino nitrogenatom. Interestingly, all the five nitrogen atoms can coordinatewith metals. Adenine possesses multiple binding modes and isan ideal biomolecular building block for bio-MOFs. Among sev-eral synthesized bio-MOFs, it was demonstrated that bio-MOF-1 could serve as a host for adsorbing a cationic drug,[17] a zinc–adeninate macrocycle was found to have large cavities for gassorption,[18] and bio-MOF-11 showed exceptional ability toadsorb CO2 by measuring the uptakes of pure CO2, H2, andN2.[19] It is relatively straightforward to experimentally deter-mine the adsorption of pure gases, however, quantitative mea-surement on the adsorption of gas mixtures is challenging.Currently, there is no reported study on mixture adsorption inbio-MOFs.

To further develop bio-MOFs as ideal sorbents requires afundamental understanding of their adsorption propertiesfrom a molecular level. In this work, we report the first molecu-lar simulation study for mixture adsorption in a bio-MOF; morespecifically, the adsorption of CO2/H2 and CO2/N2 mixtures inbio-MOF-11. First, the models and force-fields are validated bycomparing simulated and experimental isotherms of pure CO2,H2, and N2. Then the favorable adsorption sites are identified inbio-MOF-11. Finally, the adsorption and separation of CO2/H2

and CO2/N2 mixtures are examined and compared with thosein other porous materials. A gas mixture usually contains mois-ture. Hence, the effect of H2O on the separation is also investi-gated.

Results and Discussion

Pure gases

Figure 1 shows the adsorption isotherms of pure CO2, H2, andN2 in bio-MOF-11. All isotherms belong to type I, which is thecharacteristic feature of adsorption in a microporous adsorb-ent. Almost perfect agreement is found between the simulatedand experimental results for all three gases, indicating the ac-curacy of the models and force-fields used in the study. Onthis basis, we envision that the simulated selectivities of themixtures shown below are reliable, although no experimentaldata is available for comparison.

After a closer look at Figure 1, we found that the simulationslightly underestimates the experimental results, particularlyfor H2, at low pressures, and overestimates at high pressures.The deviations might be attributed to the fact that the struc-ture used in the simulation is a perfect crystal, whereas experi-mental samples usually contain impurities. The impurities dis-persed in the channels cause a stronger interaction with theadsorbate and consequently enhance adsorption at low pres-sures. The impurities also block the channels and decrease thefree volume, which reduces adsorption capacity at high pres-sures. We also examined the effect the framework charges on

CO2 adsorption. As shown in Figure S1 (Supporting Informa-tion), the extent of CO2 adsorption is largely underestimated inthe absence of the framework charges. This indicates that theframework charges play an important role in the accurate pre-diction of CO2 adsorption in bio-MOF-11, which is remarkablydifferent to IRMOFs.[20] Furthermore, the simulated isothermswere fitted to the dual-site Langmuir–Freundlich equation,using the parameters listed in Table S1. These parameters werethen used to predict the adsorption of binary mixtures by theideal adsorbed solution theory (IAST).[21]

CO2 capacity in bio-MOF-11 is 4.1 mmol g�1 at 100 kPa,which is about 1–3 times higher than that in most MOFs (e.g. ,MOF-2, MOF-177, MOF-505, IRMOF-1,�3,�6, and�11).[22] Bio-MOF-11 also outperforms many zeolites and activated car-bons,[6] amine-functionalized and imidazole-based MOFs interms of CO2 capacity.[19] H2 adsorption in bio-MOF-11 is nota-bly high with a capacity of 1.5 wt % at 77 K and 1 bar, which ishigher than that in many MOFs (e.g. , MOF-177, ZIF-8, MIL-100,IRMOF-2,�3,�9,�18, and�20).[23] The high capacities ob-served in bio-MOF-11 are attributed to the narrow channelsand the Lewis basic sites present in the framework. The chan-nels have narrow diameters of 4.2–5.2 �, leading to a substan-tial overlap of the potential energy fields for adsorbate mole-cules. The adenine linker consists of two Lewis basic sites; oneis an amino group and the other a pyrimidine nitrogen atom.A total of four amino groups and four pyrimidine nitrogenatoms are directly exposed to each cavity.[19] These electron-rich sites strongly interact with the adsorbate, particularly CO2

that has a quadrupole moment.To identify the favorable adsorption sites of CO2 in bio-MOF-

11, Figure 2 shows the radial distribution functions, g(r), of CO2

around N1, N6, and Co atoms at 298 K and 10 kPa. The atomsN1 and N6 are located in the pyrimidine and amino groups, re-spectively. From simulation, g(r) was calculated by

gijðrÞ ¼DNij V

4pr2Dr NiNjð1Þ

Figure 1. Adsorption isotherms of pure CO2 and N2 at 298 K and H2 at 77 K,respectively. The open symbols are simulation results and the filled symbolsare experimental results.[19] The lines are fits of the dual-site Langmuir–Freundlich equation to the simulation data.

ChemSusChem 2010, 3, 982 – 988 � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 983

A Bio-Metal–Organic Framework for Highly Selective CO2 Capture

in which r is the distance between species i and j, DNij is thenumber of species j around i within a shell from r to r+Dr, V isthe volume, and Ni and Nj are the numbers of species i and j. Apronounced peak is centered at r�4.0 � in the g(r) aroundboth N1 and N6. The g(r) around Co is essentially zero at r<4.0 � and exhibits a lower peak at r�8.0 �. This structural anal-ysis reveals that CO2 molecules are preferentially adsorbedonto the Lewis basic sites in adenine linkers with the shortestdistance approximately 2.7 �, rather than proximal to the Coatoms. Such a behavior is remarkably different from commonMOFs, for example, the preferentially sites in IRMOF-1 werefound to be close to metal clusters.[24] Nevertheless, the behav-ior is similar to ZIF-8 in which the adsorption sites are near 2-methylimidazolate linkers.[25] We therefore infer that the bio-molecular linkers in bio-MOF-11 should be tuned to further en-hance the adsorption capacity of CO2.

Figure 3 shows the simulation snapshot and density contourof CO2 adsorption in bio-MOF-11 at 298 K and 10 kPa. For clari-ty, only one cavity in the framework is illustrated. CO2 mole-cules are dominantly located in the cavity and adsorbed ontothe amino and pyrimidine groups. A few CO2 moleculesappear to be located in the diagonal regions of the cavity.However, they are indeed in a cavity along the perpendiculardimension, which can be seen by rotating Figure 3 by 908.

CO2/H2 mixture

The adsorption of a CO2/H2 mixture with a composition of15:85 was simulated in bio-MOF-11 to represent pre-combus-tion CO2 capture. The composition is typically encountered fora CO2/H2 mixture in H2 production. As shown in Figure S2,both CO2 and H2 in the mixture are preferentially located inthe cavities. Figure 4 a shows the adsorption isotherm at 298 Kas a function of total pressure. CO2 uptake sharply increaseswith increasing pressure and then approaches a plateau,whereas H2 uptake is vanishingly small over the entire range.There are three reasons that CO2 is predominantly adsorbedover H2: (1) CO2 is a three-site molecule and has a stronger in-teraction with the framework than two-site H2 ; (2) the temper-ature 298 K is considered subcritical for CO2 (Tc = 304.4 K), but

supercritical for H2 (Tc = 33.2 K); that is, CO2 is more condensa-ble than H2 at 298 K; (3) the presence of Lewis basic sites inadenine significantly enhances the interaction with CO2.

The separation factor of a gas mixture is quantified by selec-tivity, which is defined by

Si=j ¼ ðxi=xjÞðyj=yiÞ ð2Þ

in which xi and yi are the mole fractions of component i in ad-sorbed and bulk phases, respectively. Figure 4 b shows the se-lectivity of CO2 over H2 as a function of total pressure. With in-creasing pressure, the selectivity sharply increases, passes amaximum of 375 at 400 kPa, and then decreases. It is expectedthat the selectivity will reach a constant upon further increas-ing the pressure. The initial increase at low pressures is causedby the strong interactions between CO2 molecules and themultiple adsorption sites in bio-MOF-11, and further promotedby the cooperative intermolecular interactions of adsorbedCO2 molecules. The decrease in selectivity is primarily attribut-ed to the entropy (packing) effect at high pressures. The H2

molecule is smaller in size and can fit into the channels moreeffectively. In the pressure range studied, the selectivity of theCO2/H2 mixture in bio-MOF-11 is between 230 and 375. Numer-

Figure 2. Radial distribution functions of CO2 around N1, N6, and Co atomsin bio-MOF-11 at 298 K and 10 kPa.

Figure 3. a) Simulation snapshot and b) density contour of CO2 in bio-MOF-11 at 298 K and 10 kPa. CO2 molecules are represented by sticks. The densityhas a unit of 1 �3 and a brighter color indicates a higher density. (Co, pink;O, red; C, grey; H, white; N1, green; N6, blue; N3, N7, and N9, cyan).

984 www.chemsuschem.org � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 982 – 988

J. Jiang et al.

ous experiments and simulations have examined the separa-tion of a CO2/H2 mixture in porous materials, and are listed inTable S2 for comparison. Whereas bio-MOF-11 has a lower CO2

capacity than other materials, it exhibits a higher selectivitycompared to zeolites and non-ionic MOFs.

The predictions from IAST are shown in Figure S3 for a CO2/H2 mixture using the parameters given in Table S1. The predict-ed adsorption isotherms agree well with the simulation results.The predicted selectivity is qualitatively consistent with thesimulation, however, the magnitude of selectivity is underesti-mated by IAST. In an IAST prediction, a small inaccuracy in themole fractions would lead to a large deviation in the selectivi-ty.

A gas mixture usually contains moisture, which would affectadsorption and separation. For example, in the presence ofH2O, the selectivity of CO2 over H2 in soc-MOF increases at lowpressures due to the promoted adsorption of CO2 by H2Obound to metal atoms, but decreases at high pressures as aresult of the competitive adsorption of H2O over CO2.[15] In Na-rho-ZMOF, the interaction between CO2 and Na+ is substantial-ly reduced by a trace amount of H2O added into the CO2/CH4

mixture; consequently, CO2 adsorption drops and the selectivi-ty decreases by an order of magnitude.[26] To examine theeffect of H2O in this study, the adsorption of a CO2/H2 mixturewas simulated in the presence of 0.1 % H2O (mole fraction). As

seen in Figure 4, H2O has a negligible effect on the adsorptionof both CO2 and H2 and the CO2/H2 selectivity is slightly en-hanced due to the co-adsorption between CO2 and adsorbedH2O.

CO2/N2 mixture

To examine post-combustion CO2 capture by bio-MOF-11, theadsorption of a CO2/N2 mixture with a composition of 15:85was simulated. Figure 5 a shows the adsorption isotherm of

the mixture at 298 K as a function of the total pressure. Similarto Figure 4 a, CO2 uptake sharply increases with increasingpressures and reaches saturation at high pressures. In contrast,N2 uptake only slightly increases at low pressures. The adsorp-tion of CO2 is substantially greater than N2, because of thethree reasons mentioned above for the CO2/H2 mixture. Never-theless, the selectivity of CO2/N2 versus pressure is qualitativelydifferent from that of CO2/H2. As shown in Figure 5 b, the selec-tivity monotonically increases with increasing pressure and ap-proaches a constant at high pressures. The increase in selectivi-ty is due to the stronger CO2–CO2 intermolecular interactionsat higher pressures. The decrease in selectivity seen in Fig-ure 4 b for CO2/H2 is not observed for CO2/N2. This is becausethe molecular sizes of CO2 and N2 are not significantly differ-

Figure 4. a) Adsorption isotherm and b) selectivity of the CO2/H2 mixture(15:85) in bio-MOF-11 as a function of total pressure at 298 K in the absenceand presence of 0.1 % H2O.

Figure 5. a) Adsorption isotherm and b) selectivity of the CO2/N2 mixture(15:85) in bio-MOF-11 as a function of total pressure at 298 K in the absenceand presence of 0.1 % H2O.

ChemSusChem 2010, 3, 982 – 988 � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 985

A Bio-Metal–Organic Framework for Highly Selective CO2 Capture

ent, unlike CO2 and H2 in which H2 is much smaller than CO2 ;therefore, the entropy effect is not dominant for the CO2/N2

mixture at high pressures.CO2/N2 separation has been investigated for other nanopo-

rous materials, as listed in Table S3. Bio-MOF-11 has a selectivi-ty of 30–77, exhibiting a better separation capability than zeo-lites and non-ionic MOFs. We note that the selectivities of bothCO2/H2 and CO2/N2 mixtures in bio-MOF-11 are smaller than inrho-ZMOF.[16] This is not unexpected because rho-ZMOF con-sists of a highly ionic framework and extra-framework cations,which induce substantially strong interactions with CO2 mole-cules and exceptionally high selectivities.

The performance of IAST for the CO2/N2 mixture shown inFigure S4 is similar to that for the CO2/H2 mixture. The adsorp-tion isotherms are well predicted, however, the selectivity isonly in qualitative agreement with the simulation. The effect ofH2O on the adsorption of the CO2/N2 mixture was also simulat-ed. As seen in Figure 5, the addition of H2O slightly increasesthe selectivity of CO2/N2 ; this is more obviously at high pres-sures.

Conclusions

We have reported a molecular simulation study for CO2 cap-ture in a bio-metal–organic framework (bio-MOF-11). The bio-MOF consists of paddle-wheel cobalt–adeninate–acetate clus-ters and interlacing narrow channels. The amino and pyrimi-dine groups are Lewis basic sites, which are preferential for ad-sorption. This is in contrast to most MOFs in which the metalclusters are the favorable adsorption sites. To further facilitateCO2 adsorption in bio-MOF-11, we suggest that the adeninelinkers rather than the metals should be functionalized. Be-cause of the Lewis basic sites and the narrow channels in theframework, bio-MOF-11 exhibits large capacities of CO2 and H2,and indeed larger than many zeolites, activated carbons, andMOFs. The simulated adsorption isotherms of CO2, H2, and N2

agree very well with experimental data, which demonstratesthe accuracy of the models and potentials used.

In an attempt to evaluate the capability of bio-MOF-11 forpre- and post-combustion CO2 capture, it was found that CO2

is more favorably adsorbed than H2 and N2. The selectivities ofthe CO2/N2 and CO2/H2 mixtures behave quantitatively differentas a function of pressure. Due to the strong interactions be-tween CO2 molecules and the framework, the selectivity ofCO2/H2 at low pressures increases with increasing pressure. Athigh pressures, however, it decreases because H2 has a smallermolecular size and the entropy effect comes into play. The se-lectivity of CO2/N2 monotonically increases at low pressuresand gradually approaches a constant. The selectivities of bothmixtures in bio-MOF-11 are higher than in most zeolites andnon-ionic MOFs. In the presence of H2O (0.1 % in mole frac-tion), the selectivities are slightly enhanced, particularly for theCO2/N2 mixture at high pressures. This study suggests that ade-nine-based bio-MOFs might be useful for CO2 capture. The mi-croscopic insight from molecular simulation is important forthe quantitative understanding of the adsorption mechanism

and the rational design of new bio-MOFs for emerging applica-tions.

Models and Methods

Rosi and coworkers[19] synthesized bio-MOF-11 by a solvothermalreaction. It has a formula of Co2(ad)2(CO2CH3)2 and is thermallystable up to 200 8C. The framework consists of paddle-wheelcobalt–adeninate–acetate cluster (Figure 6 a) as the secondary

building unit (SBU). There are five types of nitrogen atoms; N1 andN6 are Lewis basic sites and N3, N7, and N9 are covalently bondedto cobalt atoms. A three-dimensional structure with augmented lvtnetwork topology is generated by the SBUs. The structure is tet-ragonal with a lattice length of a = b = 15.4355 � and c = 22.775 �.The cavities with diameter of 5.8 � are periodically distributedthroughout the structure (Figure 6 b). These cavities form interlac-ing narrow channels along the crystallographic a and b dimen-sions.[19]

To estimate the charges of bio-MOF-11 framework atoms, a frag-mental cluster (Figure S5) was cleaved and saturated by hydrogenatoms. The electrostatic potentials around the cluster were calcu-lated by density functional theory (DFT). It has been widely recog-nized that first principles-derived charges appreciably fluctuatewhen a small basis set is used; however, they tend to convergebeyond the 6-31G(d) basis set.[27] Consequently, the 6-31G(d) basis

Figure 6. a) X-ray crystallographic image of the cobalt–adeninate–acetatecluster. N1 and N6 are the Lewis basic pyrimidine and amino groups; N3,N7, and N9 are bonded to cobalt atoms. b) A unit cell of bio-MOF-11. Thecavities are indicated by the green circles. (Co, pink; O, red; C, grey; H,white; N1, green; N6, blue; N3, N7, and N9, cyan).

986 www.chemsuschem.org � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 982 – 988

J. Jiang et al.

set was used in the DFT calculation for all the atoms except Coatoms, for which the LANL2DZ basis set was used with effectivepseudo-potentials. The DFT calculation used the Lee–Yang–Parrcorrelation functional (B3LYP) and was carried out with Gaussi-an 03.[28] The concept of atomic charges is solely an approximationand no unique straightforward method is currently available to rig-orously determine atomic charges. In this study, the atomic charg-es (Table S4) were estimated by fitting to the electrostatic potentialusing the CHelpG scheme.[29] In addition to the Coulombic interac-tions, the dispersion interactions of the framework atoms werepresented by Lennard–Jones (LJ) potential with parameters adopt-ed from the universal force-field (UFF).[30] A number of simulationstudies have demonstrated that UFF can accurately predict gas ad-sorption in various MOFs, for example, Ar in Cu-BTC,[31] CO2 andCH4 in IRMOF-1,[20, 32, 33] H2 in Zn(bdc)(ted)0.5,[34] and CO2 and CH4 inMIL-101.[35]

Adsorbate CO2 was represented by the elementary physical model(EPM), which was fitted to the experimental vapor–liquid equilibri-um data of bulk CO2.[36] The C�O bond was assumed to be rigid at1.161 �, whereas the ffOCO bond was flexible and governed by aharmonic potential 1=2kq (q-q0)2 with force constant kq/kB =

153 355.79 (K rad�2) and equilibrium angle q0 = 1808. The intermo-lecular CO2–CO2 interactions were modeled by the additive pair-wise site-site LJ and Coulombic potentials

uij ðrÞ ¼X

a2ib2j

4eab

sab

rab

� �12

� sab

rab

� �6� �þ qaqb

4pe0rab

� �

ð3Þ

in which sab and eab are the collision diameter and well depth, re-spectively, e0 = 8.8542 � 10�12 C2 N�1 m�2 is the permittivity of thevacuum, and qa is the charge on atom a. Both H2 and N2 weremimicked by two-site models. The bond length was 0.74 � forH�H and 1.10 � for N�N. The LJ potential parameters for H2 andN2 were fitted, respectively, to their experimental bulk proper-ties.[37, 38] H2O was represented by the three-point transferable inter-action potentials (TIP3P) model, in which the O�H bond lengthwas 0.9572 � and the HOH angle was 104.528.[39] Table 1 lists the LJ

potential parameters and charges for CO2, H2, N2, and H2O. Thecross LJ parameters were evaluated by the Lorentz–Berthelot com-bining rules.

The adsorption of pure CO2, H2, and N2 as well as the CO2/H2 andCO2/N2 mixtures were simulated by the grand canonical MonteCarlo (GCMC) method. The chemical potentials of adsorbate in ad-sorbed and bulk phases are identical at thermodynamic equilibri-um and can be connected with the pressure of bulk phase; there-fore, GCMC has been widely used for the simulation of adsorption.The bio-MOF-11 framework was assumed to be rigid and the po-tential energies between framework and adsorbate atoms werepre-tabulated. The rationale of assuming a rigid framework is that

low-energy equilibrium configurations are involved in adsorptionand the flexibility of the framework only has a marginal effect. Asdemonstrated in a recent simulation study on the adsorption ofnoble gases in IRMOF-1, rigid and semi-flexible frameworks gaveclose results at both low and room temperatures.[40] The LJ interac-tions were evaluated with a spherical cut off of 13 � with the long-range corrections added; the Coulombic interactions were calculat-ed using the Ewald sum. The real/reciprocal space partition param-eter and the cut off for reciprocal lattice vectors were chosen to be0.2 ��1 and 8, respectively, to ensure the convergence of the Ewaldsum. The number of trial moves in a typical GCMC simulation was2 � 107, though additional trial moves were used at high pressures.The first 107 moves were used for equilibration and the second 107

moves for ensemble averages. Five types of trial moves were ran-domly attempted in the GCMC simulation: displacement, rotation,partial regrowth at a neighboring position, entire regrowth at anew position, and swap with reservoir including creation and dele-tion at equal probability. For mixtures, another type of trial move,the exchange of molecular identity, was also included. Unless oth-erwise mentioned, the simulation uncertainties were smaller thanthe symbol sizes presented in the Figures.

Acknowledgements

The authors gratefully acknowledge support from the NationalUniversity of Singapore (R-279-000-297-112).

Keywords: adenine · adsorption · carbon capture · metal-organic frameworks · computational chemistry

[1] D. Normile, Science 2009, 325, 1642 – 1643.[2] E. D. Akten, R. Siriwardane, D. S. Sholl, Energy Fuels 2003, 17, 977 – 983.[3] A. Goj, D. S. Sholl, E. D. Akten, D. Kohen, J. Phys. Chem. B 2002, 106,

8367 – 8375.[4] M. P. Bernal, J. Coronas, M. Menendez, J. Santamaria, AIChE J. 2004, 50,

127 – 135.[5] T. Seike, M. Matsuda, M. Miyake, J. Mater. Chem. 2002, 12, 366 – 368.[6] S. H. Choi, J. H. Drese, C. W. Jones, ChemSusChem 2009, 2, 796 – 854.[7] M. Eddaoudi, J. Kim, N. Rosi, D. Vodak, J. Wachter, M. O’Keefe, O. M.

Yaghi, Science 2002, 295, 469 – 472.[8] Q. Y. Yang, C. Y. Xue, C. L. Zhong, J. F. Chen, AIChE J. 2007, 53, 2832 –

2840.[9] Q. Y. Yang, Q. Xu, B. Liu, C. L. Zhong, B. Smit, Chin. J. Chem. Eng. 2009,

17, 781 – 790.[10] S. Keskin, D. S. Sholl, Langmuir 2009, 25, 11786 – 11795.[11] S. Keskin, J. K. Johnson, D. S. Sholl, Microporous Mesoporous Mater.

2009, 125, 101 – 106.[12] A. O. Yazaydin, A. I. Benin, S. A. Faheem, P. Jakubczak, J. J. Low, R. R.

Willis, R. Q. Snurr, Chem. Mater. 2009, 21, 1425 – 1430.[13] Y. S. Bae, O. K. Farha, J. T. Hupp, R. Q. Snurr, J. Mater. Chem. 2009, 19,

2131 – 2134.[14] B. Liu, B. Smit, Langmuir 2009, 25, 5918 – 5926.[15] J. W. Jiang, AIChE J. 2009, 55, 2422 – 2432.[16] R. Babarao, J. W. Jiang, J. Am. Chem. Soc. 2009, 131, 11417 – 11425.[17] J. An, S. J. Geib, N. L. Ros, J. Am. Chem. Soc. 2009, 131, 8376 – 8377.[18] J. An, R. P. Fiorella, S. J. Geib, N. L. Ros, J. Am. Chem. Soc. 2009, 131,

8401 – 8403.[19] J. An, S. J. Geib, N. L. Ros, J. Am. Chem. Soc. 2010, 132, 38 – 39.[20] R. Babarao, J. W. Jiang, Langmuir 2008, 24, 6270 – 6278.[21] A. L. Myers, J. M. Prausnitz, AIChE J. 1965, 11, 121 – 127.[22] A. R. Millward, O. M. Yaghi, J. Am. Chem. Soc. 2005, 127, 17998 – 17999.[23] D. J. Collins, H. C. Zhou, J. Mater. Chem. 2007, 17, 3154 – 3160.[24] D. Dubbeldam, H. Frost, K. S. Walton, R. Q. Snurr, Fluid Phase Equilib.

2007, 261, 152 – 161.[25] H. Wu, W. Zhou, T. Yildirim, J. Am. Chem. Soc. 2007, 129, 5314 – 5315.

Table 1. LJ potential parameters and charges for CO2, H2, N2, and H2O.

Adsorbate Site s [�] e/kB [K] q(e)

CO2

C 2.785 28.999 +0.6645O 3.064 82.997 �0.33225

H2 H 2.50 14.5 0N2 N 3.32 36.4 0

H2OH 0 0 +0.417O 3.151 76.47 �0.834

ChemSusChem 2010, 3, 982 – 988 � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org 987

A Bio-Metal–Organic Framework for Highly Selective CO2 Capture

[26] R. Babarao, J. W. Jiang, Energy Environ. Sci. 2009, 2, 1088 – 1093.[27] P. C. Hariharan, J. A. Pople, Chem. Phys. Lett. 1972, 16, 217 – 219.[28] Gaussian 03, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.

Robb, J. R. Cheeseman, J. A. Montgomery, Jr. , T. Vreven, K. N. Kudin, J. C.Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M.Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M.Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B.Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala,K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski,S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D.Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul,S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I.Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A.Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W.Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc. , Wallingford CT, 2004.

[29] C. M. Breneman, K. B. Wiberg, J. Comput. Chem. 1990, 11, 361 – 373.[30] A. K. Rappe, C. J. Casewit, K. S. Colwell, W. A. Goddard, W. M. Skiff, J. Am.

Chem. Soc. 1992, 114, 10024 – 10035.[31] A. Vishnyakov, P. I. Ravikovitch, A. V. Neimark, M. Bulow, Q. M. Wang,

Nano Lett. 2003, 3, 713 – 718.

[32] G. Garberoglio, A. I. Skoulidas, J. K. Johnson, J. Phys. Chem. B 2005, 109,13094 – 13103.

[33] A. I. Skoulidas, D. S. Sholl, J. Phys. Chem. B 2005, 109, 15760 – 15768.[34] J. Liu, J. Y. Lee, L. Pan, R. T. Obermyer, S. Simizu, B. Zande, J. Li, S. G.

Sankar, J. K. Johnson, J. Phys. Chem. C 2008, 112, 2911 – 2917.[35] Y. F. Chen, R. Babarao, S. I. Sandler, J. W. Jiang, Langmuir 2010, 26,

8743 – 8750 .[36] J. G. Harris, K. H. Yung, J. Phys. Chem. 1995, 99, 12021 – 12024.[37] E. Pantatosaki, G. K. Papadopoulos, H. Jobic, D. N. Theodorou, J. Phys.

Chem. B 2008, 112, 11708 – 11715.[38] C. S. Murthy, K. Singer, M. L. Klein, I. R. McDonald, Mol. Phys. 1980, 41,

1387 – 1399.[39] W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, M. L. Klein,

J. Chem. Phys. 1983, 79, 926 – 935.[40] J. A. Greathouse, T. L. Kinnibrugh, M. D. Allendorf, Ind. Eng. Chem. Res.

2009, 48, 3425 – 3431.

Received: March 12, 2010

Revised: May 24, 2010

Published online on July 9, 2010

988 www.chemsuschem.org � 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 2010, 3, 982 – 988

J. Jiang et al.