ARTICLEReceived3Jan2013|Accepted5Mar2013|Published16Apr2013Newmaterialsformethanecapturefromdiluteandmedium-concentrationsourcesJihanKim1, AmiteshMaiti2, Li-ChiangLin3, JoshuahK. Stolaroff2, BerendSmit1,3,4&RogerD. Aines2Methane (CH4) is an important greenhouse gas, second only to CO2, and is emitted into theatmosphere at different concentrations from a variety of sources. However, unlike CO2, whichhas a quadrupole moment and can be captured both physically and chemically in a variety ofsolventsand poroussolids, methaneis completelynon-polar and interacts very weakly withmost materials. Thus, methane capture poses a challenge that can only be addressed throughextensive material screening and ingenious molecular-level designs. Here we reportsystematicinsilicostudiesonthemethanecaptureeffectivenessoftwodifferentmaterialssystems, thatis, liquidsolvents(includingionicliquids)andnanoporouszeolites. Althoughnoneoftheliquidsolventsappearseffectiveasmethanesorbents, systematicscreeningofover 87,000 zeolite structures led to the discovery of a handful of candidates thathave sufcient methane sorption capacity as well as appropriate CH4/CO2and/orCH4/N2selectivitytobetechnologicallypromising.DOI:10.1038/ncomms26971MaterialsSciencesDivision, LawrenceBerkeleyLaboratory, Berkeley, California94720, USA.2PhysicalandLifeSciencesDivision, LawrenceLivermoreNational Laboratory, Livermore, California 94550, USA.3Department of Chemical and Biomolecular Engineering, University of California, Berkeley, California94720, USA.4DepartmentofChemistry, UniversityofCalifornia, Berkeley, California94720,USA. CorrespondenceandrequestsformaterialsshouldbeaddressedtoA.M. (email:amaiti@llnl.gov).NATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications 1&2013MacmillanPublishersLimited.Allrightsreserved.Methane(CH4)isasubstantial driverof global climatechange, contributing 30% of current net climateforcing1. In addition, concern over methane ismounting due to leaks associated with rapidly expandingunconventional oil and gas extraction and the potential forlarge-scale release of methane from the Arctic2. At the same time,methane is a growing source of energy3, and aggressive methanemitigation is increasingly recognized as a key to avoidingdangerous levels of global warming4,5.Methaneisemittedatawiderangeofconcentrationsfromavariety of sources, including natural gas systems, entericfermentation(livestock), landlls, coal mining, manure manage-ment, wastewatertreatment, ricecultivationandafewcombus-tion processes. We can generally group the methaneconcentrations of sources into three categories: high purity(490%), medium purity (575%) and dilute (o5%). High-puritymethanecanbesoldtothecommoditynatural gas market orconverted to other chemicals (for example, methanol and carbonblack) bycurrent industrial practices. Medium-puritymethaneincludeslandll gas, coal-minedrainagegas, anaerobicdigestergas and low-quality gas fromfossil formations. Avariety oftechnologies has been developed for generating electricity orhigh-grade process heat from medium-purity methane, includingtechnologies suchas thehomogeneous charge gas enginethatoperatesjustabovethemethaneammabilitylimitinair(5%).Small or inconvenient ows of medium-purity methane are oftensimply ared2. Treatable dilute methane sources are some of thelargest intotal emissions, including coal-mine ventilationair,manurestorageheadspaceandanimalfeedinghouseventilationair. Sometechnologieshavebeendevelopedforoxidizingdilutemethane, such as the thermal ow-reversal reactor, but theygenerally yield only low-grade heat or small amounts of electricityas a co-benet.It is highly desirable to be able to concentrate a dilute methanestreamtomediumpuritytoeffectivelyutilizetheenergy, ortoconcentrate a medium-purity stream to high purity to convert itto a liquid or sell it. Conversion is especially attractive for manysmall or remote sources. Purication of some mediumandhigher-puritynatural gasesiscurrentlypracticedindustriallybysorptionof thenon-methanecomponents(CO2andH2S). Thepracticebecomes uneconomicor impractical belowabout 40%methane6. For methane concentrations o40%, or for separationof methane fromair rather thanacidgases, we wouldlike asorbent for methane itself.In this contribution, we explore the possibility of using asorbent for methane purication. We consider the followinggeneral problems: (1)concentratingamedium-puritystreamtothe high-purity range and (2) concentrating a dilute stream to themedium-purityrange. Forpurposesofanalysis, ourproxiesforthesecasesare(1)alow-qualitynatural gas(simpliedas30%CH4 and 70% CO2 by mole fraction at a total pressure of 70 bar)and (2) coal-mine ventilation air (simplied as 1% CH4, 1% CO2and98%N2at a total pressure of 1 bar). Bothof these arepotential large-scale, high-impact applicationsforasorbent formethane. The example cases illustrate two general characteristicsof methanepurication. First, becauseprocesses relatedtotheones that produce methane also produce CO2, almost all naturalmethanestreamscontainsignicant levelsof CO2. Second, fordilutestreams, N2 isthe dominantcomponenttoexclude. O2isalso abundant, but for combustion applications, it is desirable tocarry at least as much O2 as CH4 through the process.For goingfrommedium- tohigh-puritymethane, asorbentselectivity of CH4over CO2greater than 1 is required andgenerally, the higher the selectivity, the fewer absorptiondesorptioncycles requiredto reachdesiredpurity. For goingfromdilutetomediumpurity, theselectivityofCH4overN2ismore important for determining the number of cycles. Theselectivity of CH4 over CO2 may also be important, depending ontheapplication. Forcombustionatlowconcentrations(e.g., 5%CH4), selectivity is not relevant. However, if the gas is tobefurther puried, or is to be used in an advanced (e.g., biologicallybased) conversion process, then selectivity close to or greater than1 is desirable.In this contribution, we carry out thorough exploration of twodifferent classes of capture materials for their effectiveness inmethane capture: (1) liquid solvents and (2) nanoporous zeolites.We show that although none of the common solvents (includingionicliquids (ILs)) appears topossess enoughafnitytowardsmethanetobeof practical use, systematicscreeningof aroundone hundredthousandzeolite structures has uncovereda fewnanoporous candidates that appear technologically promising.We use free-energy proling andgeometric analysis inthesecandidate zeolites to understand how the distribution andconnectivity of pore structures andbinding sites canleadtoenhanced sorption of methane while being competitive with CO2sorption at the same time.ResultsLiquid solvents for methane capture. Liquids have traditionallybeen known as poor absorbers of methane, however, theexperimental literature has limiteddatafor methanesolubilityboth in conventional solvents as well as inILs. Athoroughinvestigation of liquid solvents is important for two reasons: (1) agoodliquidsolvent for methane couldnot onlybe usedas acapture agent but also could be injected to break down methanehydrate (clathrate), andextract methane fromthis potentiallyhugeuntappedenergysource79and(2)therehavebeenrecentexperimental attempts of exploring novel solvents (ILs) forefcient absorption10andcatalyticconversion11of methane; asystematic in silico study could contribute greatly to suchprogress.To compute methane solubility in liquids, we adopted animplicitsolventmethod, thatis, COSMO-RS12,13, inwhichonerepresents both the solute and solvent molecules by the histogramof their surface screening charges called the s-prole. Allinteractions, including Coulombic, van der Waals andhydrogen-bondinteractions, arethendenedintermsof theses-proles. One can use this formalism to compute the partitionfunction, the Gibbs free energy and many other thermodynamicquantities, including pseudo-chemical potentials (m*)14. Thepseudo-chemical potential of any solute present in x molefraction in a solution (msolvent*) is dened as its chemical potential(m)withouttheidealentropyofmixingtermkBTln(x); inotherwords, m msolvent* kBTln(x). mself* is the pseudo-chemicalpotential of the solute in its own pure liquid state (x 1),where the chemical potential is m mself* kBTln(1) mself*.Mole-fractionsolubility(x) canbe computedbyequatingthetwochemicalpotentials,msolvent* kBTln(x) mself*, whichleadsto the following equation:x expfmselfmsolvent=kBTg: 1Note that by its very denitionthe pseudo-chemical potentialincludes a contribution fromthe activity coefcient, which,however, becomes negligible in the limit of innite dilution.Whenthesoluteisdissolvingfromthegasphase, oneneedstorelate the quantity m*self to the chemical potential in the gas phase,which can be expressed as a function of the gas fugacity15.However, because our liquid exploration involved a xed pressure(1 bar) and a xed temperature (300 K), we found it moreconvenient to use m*self as a tting parameter so as to match theexperimental solubility of all relevant gases, that is, CH4ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms26972 NATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications&2013MacmillanPublishersLimited.Allrightsreserved.(refs 9,16), N2(refs 17,18) and CO2(ref. 19) in one specicstandard solvent, that is, ethanol. This led to the values ofm*self 26.8, 27.6 and 17.6 kJ mol1for CH4, N2andCO2, respectively, to be used with COSMO-RS-computed msolventunder the parameter settings described in the Methods section.First, we exploredthe feasibilityof usingliquidsolvents toconcentrate coal-mine ventilationair, representedby(inmolefraction) 1%CH4, 1%CO2and98%N2at atotal pressureof1 bar. Theaimwastocreateanoutputstreamwith45%CH4.This translates to a desiredCH4/N2selectivity 45. Here wedeneselectivityastheratioofHenrysconstants, forexample,CH4/N2selectivity KH(CH4)/KH(N2). Roughly speaking, aCH4/N2selectivityof B5(alongwithCO2/CH4selectivitynotmuch larger than 1) leads to an output stream of 5% CH4 withinone cycle (because the starting concentration of CH4B1%). Theactual CH4 concentration in the output stream will depend on thenumber of stages and other processing parameters. However, foreconomic reasons, the idea is to look for materials that can lead tothe desired CH4 concentration in as little cycles as possible. ThepresenceofO2wasnotconsideredexplicitlyjustforsimplicity,and it was assumed that the output stream will have sufcient O2to be able to combust the methane. For some applications(further purication of the gas or some chemical conversions), werequire CH4/CO2selectivity to be not much less than 1 asotherwise, CO2 becomes the major fraction in the output stream.With this in mind, we computed the solubilities (usingequation (1)) of CH4, N2and CO2in 73 common room-temperature liquidsolvents at T300 KandP 1 bar, whichalso include10commonILs (seeSupplementaryTableS1). Theconventional solvents were chosen from a wide variety of classes,including alkanes, cyclo-alkanes, alkenes, alkynes, aromatics,alcohols, aldehydes, ether, ketones, amines and thiols.Figure 1 plots the CH4/N2selectivity (dened as the ratioof Henrys constants KH(CH4)/KH(N2)) on the y axis andthe corresponding CH4Henrys constant (KH(CH4) inmol kg1bar1)onthexaxis. Ingeneral, thesolubilityintheILs is relatively low in the relevant conditions such that one canreliablycomputeselectivityfromtheratioofHenrysconstants.Interestingly, there are manysolvents withCH4/N2selectivity45, with1-octanamineand1-octanethiol havingvalues 410.The solvents withthe highest CH4solubility (that is, highestKH(CH4)) appear tobe the small alkanes like pentane, cyclo-pentane, hexane and so on, as well as carbon disulphide (CS2). Allof these solvents also have desirable CH4/N2 selectivity values ofB8. Unfortunately, the Henrys constant is too low, beingB0.044 mol kg1bar1even for the best solvent, that is,pentane. Foraninlet partial methanepressureof 0.01 bar, thiswould amount to a loading of only 4.4 104mol kg1, whichdoesnotmakeiteconomicallyviable. Inaddition, Fig. 2showsthat evenfor the alkanes the highest value of the CH4/CO2selectivityislow, beingonly0.36forcyclo-decaneand0.34forpentane. This means that inthe output streamthere will beroughly three times more CO2 as compared to CH4.The 10 ILs screened in Figures 1 and 2 (lled symbols) includecationic classes imidazolium, ammonium, phosphoniumandpyridinium, and anions BF4, PF6 and Tf2N (refs 15, 20). We ndthat the highest CH4 solubility in these classes of ILs occurs in the[ammonium][Tf2N] systems with a Henrys constant of0.005 mol kg1bar1or less, that is, almost an order ofmagnitudesmallerthanthatinpentane. Thisisconsistentwiththe limited available experimental data10,21. The CH4/CO2selectivityinILsisalsomuchworsethanregularalkanes, beingtypically o0.15, which might be expected, given the ionicenvironment.Theaboveanalysisclearlydemonstratesthat liquidsolvents,including ILs, are not suitable for concentrating andutilizingmethanefromlowtomediumemissionsourcesthemethanesolubility is too low, and the CH4/CO2selectivity is notfavourable. Thus, we turned to another systemwith a largenumber of possible structural and compositional varieties, that is,nanoporous solids, like zeolites.Zeolites for methane capture. Zeolites are porous materialscommonly used as adsorbents. Because of their diverse topologyresulting from various networks of the framework atoms, zeolitescanbe usedfor many different types of gas separations andstorage applications. In this work, we analysed 190 experimentallyrealizedInternational Zeolite Association(IZA) structures andover 87,000 predicted crystallography open database (PCOD)structures from Deems hypothetical zeolite database22to searchfor materials suitable for methane capture. Because of thequadruple moment of CO2, the interactionbetweenthe CO2molecules and the framework atoms are generally much strongerthan for the CH4molecules. Accordingly, nding zeolite0.00 0.01 0.02 0.03 0.04 0.05024681012DiethyletherDimethyl-diazenePentaneCyclo-pentane CS2Hexane1-OctanethiolCH4/N2 selectivity (molar)CH4 Henry's constant (mol kg1 bar1)1-OctanamineFigure 1|CH4/N2 molar selectivity versus CH4 Henrys constant in liquidsolvents. Selectivity is dened as the ratio of Henrys constants KH(CH4)/KH(N2). Results are shown for 73 common liquid solvents (including 10 ILs).AllcalculationswereperformedatT 300KandP 1 bar.The namesofthemostinterestingsolventsare indicated.TheILsareindicatedbylledsymbols.ThefullsolventlistisprovidedinSupplementaryTableS1.0.00 0.01 0.02 0.03 0.04 0.050.00.10.20.30.4Cyclo-hexaneDiethyletherPentaneCyclo-pentaneCS2HexaneCH4/CO2 selectivity (molar)CH4 Henry's constant (mol kg1 bar1)Cyclo-decaneFigure2|CH4/CO2molarselectivityversusCH4Henrysconstantinliquidsolvents.SelectivityisdenedastheratioofHenrysconstantsKH(CH4)/KH(CO2). Resultsareshownfor73commonliquidsolvents(including10ILs). AllcalculationswereperformedatT 300KandP 1 bar. The names of the most interesting solvents are indicated. The ILsareindicatedbylledsymbols.ThefullsolventlistisprovidedinSupplementaryTableS1.NATURECOMMUNICATIONS|DOI:10.1038/ncomms2697 ARTICLENATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications 3&2013MacmillanPublishersLimited.Allrightsreserved.structures that simultaneously lead to a large adsorbed CH4concentration(relativetoadsorbedCO2)andhighCH4loadingposes a signicant challenge even with the large number ofdiverse zeolite structures at our disposal. For mixtures thatcontain methane at relatively low pressure, the binding energy ofmethane is the primary factor that determines the performance ofthestructure. Ontheotherhand, forseparationsthat occurathigher pressures, the CH4CH4interactioncouldalso play asignicant role. Thus, weexpect that thetotal pressureof theinitial mixture gas will largely dictate the type of materialsoptimal for methane capture.First, weperformedthezeolitescreeningfortheapplicationinvolving a low-quality natural gas mixture feed consisting of 30%CH4and 70%CO2(in mole fraction) at 70 bar and 300 K.Figure 3 indicates the adsorbed CH4concentration (molefraction) in the adsorbed gas mixture as a function of theadsorbed amount of CH4(that is, CH4loading) for all ofthe IZA and the predicted zeolite structures. The resultsimmediately reveal an IZA structure, SBN, that stands out in itsperformanceroughly2.75(mol kg1)adsorbedCH4and45%adsorbed CH4concentration. We also utilized the recentlypublished force eld D2FF, which is developed based ondispersion-corrected density functional theory (DFT), fromSholl andco-workers23to compute the SBNCO2adsorptionisotherms. AdsorptionisothermscalculatedviaD2FFhavebeenshowntoaccuratelyreproducetheexperimental isotherms forCHA, MFI and DDR silicate zeolite structures23. In our work, theSBNCO2isotherms obtainedfromD2FFpredict evenbetteroverall performance (see Fig. 3) with reduced CO2 uptake for allof the pressure values. To further test the robustness of ournding, sensitivity analysis was conducted by changing (in stepsof 1%) the SBNlattice parameters from3to 3%of theoriginal optimized value of a 14.37 , b 12.45 andc 13.85 . For eachperturbedstructure, thelatticeconstantswere kept xedwhile the atomic coordinates were optimizedusing DFT with the quantum code SIESTA (ref. 24). As shown inFig. 3, all of the perturbedSBNdata points are close totheoriginal SBN data, indicating that in spite of a small dependenceof its performance on force eld parameters and lattice constant,SBN is clearly one of the best structures for methane capture.To further explore the reason behind the exceptionalperformanceofSBN, we plotinFig. 4thesimulatedadsorptionisotherm curves and the free-energy landscapes in this zeolite forbothCO2andCH4. Withinthe idealizedSBNstructure, bluerepresents low-energy regions andredrepresents high-energyregions, with the rest indicating inaccessible regions. To help withthe analysis, the adsorption isotherm curves are divided into threeregimes based on the total pressure (that is, Po1 bar,1 baroPo100 bar and P4100 bar), representing the Henry0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.00.10.20.30.40.50.6SBN(G-P)IZA zeolitesPredicted zeolitesAdsorbed CH4 concentrationAdsorbed CH4 loading (mol kg1)CO2/CH4 = (70/30)(P,T) = (70 bar, 300 K)SBN(D2FF)Figure 3|Molar concentration of CH4 in the adsorbed phase versus CH4loading in zeolites. Results are shown for 190 IZA (blue) and over 87,000predicted (red) zeolite structures with input gas-phase mixture compositionof30%CO2and70%CH4atpressureof70bar andT 300K. TheSBNthatshowsexceptionalperformanceforthisseparationisplottedfortwodifferentforceelds, Garcia-Perez(bluesolid)andD2FF(pinksolid).The unitcellsoftheSBNwerechangedfrom 3to3%oftheir originalsizes, andrelaxedusingDFT forGarcia-Perez(bluenoll).1021011001011021031041050369121518aLoading (molecules per unit cell)Pressure (bar)CO2CH4abcbabccFigure 4|Zeolite SBN CO2/CH4 adsorption isotherm curves and unit-cellfree-energyproles.(a)CO2(red)andCH4(green)adsorptionisothermdataatT 300K wherethetwocurvesintersectoneanotheratP 5 bar andP 105bar.The dashedlinesdividetheisothermsintothreeimportantregions: (1)Henryregime,(2)strongCH4CH4interactionregime and(3)saturationregime. Free-energylandscapeinsideSBNunitcell for (b) CH4 and (c) CO2 molecule with lattice parameters a 14.374,b 12.448andc 13.846. Darkbluerepresentslow-energystrongbindingsites,andredrepresentshighenergies.Thethreeyellowarrowsseparateonebindingsitetoanother, andthedistance valuesrangefrom4to4.6.ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms26974 NATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications&2013MacmillanPublishersLimited.Allrightsreserved.regime, strong CH4CH4interaction regime and saturationregime, respectively (see Fig. 4). In the Henry regime(Po1 bar), theguest particleframeworkinteractiondominatestheadsorptionpropertieswheretheHenrysconstantKH0.86and 0.52 mol kg1bar1for CO2 and CH4, respectively, leadingtohigher uptakeof CO2inthis region. Thedifferencecanbeexplained by comparing the binding energy values (CH4:22.09 kJ mol1and CO2: 28.57 kJ mol1) of the twomolecules. At PB5 bar, thetwoisothermcurvesintersect eachother, beyond which the CH4uptake becomes larger. Tounderstandthe increaseduptake of CH4inSBNat P45 bar,we analyse in Fig. 4 the free-energy proles of CO2 and CH4. Wend that the CH4 adsorption sites are highly localized with eachsite connected to three nearest neighbour adsorption sites(indicatedbyyellowarrowsinFig. 4)withaseparationbarrier.The distances betweenthese adsorptionsites range from4to4.6 withtheaveragebeingataround4.33 . Suchseparationvalues aligncloselywiththeminimumenergydistanceof twocentre-of-massCH4molecules, B4.2 . Ontheotherhand, theadsorptionsitesforCO2arelessdistinctfromoneanother, anddo not in general correspond to optimal CO2CO2 distances. As aresult, the CH4 uptake in SBN is higher than that of CO2 uptakefor P up to a fewthousand bars, well above the pressuresconsidered for low-quality natural gas separations. At PZ100 bar,the CH4 loading saturates to 16 molecules per unit cell, equal tothe number of distinct CH4 adsorption sites that can be countedfrom the free-energy prole in Fig. 4.Forcoal-mineventilationaircomprisedof1%CH4, 1%CO2and98%N2at a total pressure of 1 bar, similar analysis wasconducted to evaluate structures suitable for this separation. Thethree-component Ideal Adsorbed Solution Theory was utilized toobtainthe mixture loading values fromthe pure componentisotherms. Figure 5a shows the results for the IZAand thepredictedzeolitestructures. Atthiscondition, theuptakevaluesof these three gases tend to lie within the linear region in the purecomponent isotherms for most zeolite structures, andaccordingly, a strong correlation exists between Henrysconstant and uptake values at this pressure. Thus, it is notsurprising that in our analysis, we found that the top 20 structures(where the metric was takentobe simplythe product of thesolubility and selectivity) in the predicted zeolite database possessa CH4Henrys constant 41.36 mol kg1bar1at 300 K,whichputsthemwithinthetop0.1%of thelargest KHinthedatabase. Zeolite SBN, which has a relatively smallerKH0.52 mol kg1bar1, is predicted to have poor perfor-mance for this separation. The analysis of the free-energylandscaperevealsthatmostofthetopstructuresarecomprisedof one-dimensional channels for the CH4 molecules. In addition,geometry analysis utilizing Zeo 25indicates that themaximumincludedspherealongthesefreepathshasdiametersbetween56 . Thus, these channels canbe characterized asnarrowandnotcage-like. Ingeneral, thezeolitestructureswithnarrow channels can form strong CH4binding sites asthe number of frameworkoxygenatoms locatedwithinclose(i.e., 34 ) vicinitytothe centre of the bindingsites canbemaximizedwithinthistopology. Ontheotherhand, acage-likeenvironment tends to provide fewer framework oxygen atoms atoptimumdistances fromthe CH4, and thereby leads to lowHenrys constant values.In this separation, we further focused on zeolite structures thatalso possess high adsorbed CH4/CO2 ratio. For this low-pressurecondition, theadsorbedCH4/CO2ratiocanberegardedasthepreviouslydenedselectivityastheratioof Henrysconstants.Figure 5b shows the CH4/CO2 selectivity of all structures plottedinFig. 5aas afunctionof adsorbedCH4amount, withopensymbols indicating the structures with adsorbed CH4 molar ratio40.08 in Fig. 5a. The IZA structures with the largestadsorbedCH4/CO2ratiowereidentiedtobeZONandFER,with KH(CH4) 1.29 and 1.12 mol kg1bar1, respectively.Moreover, we have identied many PCOD structures thathave a very large adsorbed CH4/CO2ratio (42.00), whichcanbe promisingfor the separation(e.g., PCOD8301873andPCOD8307399). Although IZA structures such as UFI have evenlarger KH(CH4) 1.39 mol kg1bar1thanwhat is computedfor ZON and FER, the CH4/CO2 selectivity is only 0.44, which ismuch smaller compared to ZON (0.85) and FER (0.94). Analysisof these three structures based on the free-energy proles revealsthat the number of low-energy adsorption sites for CO2 is largercompared with CH4, which might partially be responsible for therelativelylowCH4/CO2selectivityvalueinUFI. Ingeneral, itisdifcult to nd common characteristics among zeolite structuresthat possess both large KH(CH4) and large CH4/CO2 selectivity asintricate andsubtle differences inthe frameworkcompositionseem to make large contributions.DiscussionWiththe aimof discoveringmaterials capable of isolating orconcentrating methane at minimum energy costs, we have carriedout extensiveinsilicoscreeningof alargenumber of conven-tional liquidsandILs, aswell asover87,000zeolitestructures.0.000 0.005 0.010 0.015 0.020 0.0250.000.020.040.060.080.100.120.14 aIZA zeolitesPredicted zeolitesAdsorbedCH4molarratioAdsorbed CH4amount (mol kg1)CO2/CH4/N2= (1/1/98)(P,T) = (1 bar, 300 K)All Selected IZA zeolitesAll Selected predicted zeolites0.000 0.005 0.010 0.015 0.020 0.0250.00.51.01.52.02.5bFERAdsorbedCH4/CO2ratioAdsorbed CH4amount (mol kg1)CO2/CH4/N2 = (1/1/98)(P,T )= (1 bar, 300 K)ZONUFIFigure5|CH4captureusingzeolitesforcoal-mine ventilatedair.(a)AdsorbedCH4concentrationasfunctionofCH4loadingforIZA(blue) and predicted (PCOD) (red) structures. (b) The same data points areplottedwithadsorbedCH4/CO2ratioasafunctionofadsorbedCH4amount.Thenoll datapointsrepresentzeolitestructuresfromathathaveadsorbedCH4molarratio 40.08.Thus, thebeststructuresarestructures within the set of no ll data points that also have large adsorbedCH4amountandlargeCH4/CO2ratiovalues.NATURECOMMUNICATIONS|DOI:10.1038/ncomms2697 ARTICLENATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications 5&2013MacmillanPublishersLimited.Allrightsreserved.Cutting-edge computational tools were employed, including aDFT-basedquantum-chemical implicit solvent formalism13forthe liquid solvents and a recently developed, highly efcientsorptioncode26,27for the zeolites that employs classical forceelds with well-validated parameter sets. Two specic applicationareas were targeted, that is, concentrating methane fromamedium-concentration source to a high concentration (forexample, purifyingalow-qualitynatural gas)andconcentratinga very dilute methane stream into one of the moderateconcentrations (for example, enablingenergyproductionfromcoal-mine ventilation air). Both these applications warrantmaterialsthat needtohavehigherafnitytomethanethantoCO2,achallengingpropositiongivenmethanesessentiallynon-polar character. In this regard, all the liquids in our investigationfall shortthe best overall liquid, pentane, has the requiredCH4/N2selectivity (B8), but low Henrys constant(0.044 mol kg1bar1) and low CH4/CO2selectivity (0.34).However, some of the zeolites show considerable promise both interms of CH4uptake capacity and CH4/CO2selectivity.Particularly noteworthy is the zeolite SBN, whichhas a largenumber of binding sites that are formedinsucha way thatmaximizes the CH4CH4 interactions, resulting in extraordinarilyhighperformancefor concentratingmethanefromamedium-concentration source to a high concentration. For dilute methane,onthe other hand, thereare zeolites like ZONandFERthatpossess large KH(CH4) as well as high CH4/CO2selectivity,making them excellent candidates for concentrating dilutemethane streaminto moderate concentration. For the latterseparation, we have also identied other structures in thepredicted zeolite database that could potentially outperformZONand FER. All these structures consist of narrowone-dimensional channels that create strong binding sites for CH4 byhaving a signicant number of framework oxygen atoms atoptimaldistancesfromtheCH4centre. Forfurtherinformationabout theseIZAzeolitestructures (e.g. latticeconstant, atomicpositions and space group), please refer to http://www.iza-structure.org/databases/.Besidespuresilicazeolitestructures, wehavealsoconductedlarge-scale screening on aluminosilicate zeolite structures inwhich some of the silicon atoms are substituted with aluminiumatomswiththeadditionof cationstoensurechargeneutrality.Similar workhas beenconductedinthe past toanalyse CO2adsorption in aluminosilicate zeolite structures28. In our analysis,we found aluminosilicate zeolite structures to be, in general, sub-optimal formethanereduction; thepresenceof cationscreatesstrong binding sites for the CO2 molecule, and subsequently leadstoinferiorCH4/CO2selectivityvalues. However, morein-depthstudy needs to be conducted to determine whether we cancompletelyruleout aluminosilicatezeolitesasaviableclassofmaterials for methane capture. One can also attempt to screen alarge database of porous materials such as metal-organicframeworks, zeolitic imidazolate frameworks or covalentorganic frameworks29to identify structures optimal formethane capture. Unfortunately, many of the synthesizedstructurespossessstronginteractionsbetweenthemetal atomsandtheCO2molecules, andthusaremorepromisingforCO2capture. Finally, there have been recent attempts in the literatureto utilize activated carbon30and graphene31for methane capture.Activatedcarboncouldbeagoodstoragemediumforrelativepure methane30, but is probably not an ideal system topreferentially adsorb CH4over CO2. Graphene, with properdoping and inter-layer spacing could potentially isolate CH4(ref. 31), but moreworkneeds tobedonetooptimizesuchasystem for selective methane capture with high loading capacity.A large-scale screening approach such as ours could beappropriate for such an exploration.MethodsSolubilitycalculationsinliquidsolventsandILs. As in our previouswork32,the surface screening charges (and the corresponding s-proles) were computedusing the DFT code Turbomole33, the BeckePerdewexchange-correlationfunctional34,35and an all-electron representation using the triple-zeta valence basisset with polarization36,37. These s-proles were then used to compute thepseudo-chemical potentials msolvent* using the commercial COSMO-RS codeCOSMOTherm (version C2.1, Release 01.10), available from Cosmologic Inc.(http://www.cosmologic.de). For an IL, a separate s-prole is constructedfor thecation and the anion, and the solvent represented as a 50:50 molar mixture of thetwo fragments. From extensive tests on the aqueous solubility of a large data set ofdrug molecules or organic solutes, it appears that COSMO-RS incurs an averageerror of the order of 0.30.5 log units38. Based on the above, an accuracy of thecomputed solubility to within a factor of 23 can be considered reasonable.Calculationofadsorptionisothermsinzeolites. The Henrys constant and thepure component adsorption isotherms for CO2, CH4 and N2 gas molecules werecomputed using our highly efcient graphics processing unit code26,27, and theIdeal Adsorbed Solution Theory was then applied to estimate the mixturecomponent uptake to reproduce the aforementioned conditions relevant tomethane separations39. In our simulations, all interactions between gas moleculesand the zeolite framework were described at the classical force eld level withatomic partial charges (for Coulombic interactions) and 12-6 LennardJonesparameters (for van der Waals interactions)taken from Garcia-Perez et al.40Theframeworkwas assumed to be rigid throughout the simulations, an assumptionthat is considered to be reasonable in zeolite structures41.References1. Barker, T. et al. (eds).In: Climate Change 2007: Mitigation. Contribution ofWorking Group III to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change (Cambridge UniversityPress, UK, 2007).2. Stolaroff, J. K., Bhattacharyya, S., Smith, C. A., Bourcier, W. L., Cameron-Smith, P. J. & Aines, R. D. Review of methane mitigation technologies withapplication to rapid release of methane from the Arctic. Environ. Sci. Technol.46, 64556469 (2012).3. Annual Energy Outlook 2011: With Projections to 2035, #DOE/EIA-0383 (U.S.Energy Information Administration, Washington, DC, 2011).4. Shindell, D. et al. Simultaneously mitigating near-term climate change andimproving human health and food security. Science 335, 183189 (2012).5. Near-term Climate Protection and Clean Air Benets: Actions for ControllingShort-Lived Climate Forcers (United Nations Environment Programme(UNEP), Nairobi, Kenya, 2011).6. Su, S., Beath, A., Guo, H. & Mallett, C. An assessment of mine methanemitigation and utilisation technologies. Prog. Energy Combustion Sci.,Pergamon-elsevier Science Ltd 31, 123170 (2005).7. Folger, P. Gas Hydrates: resource and hazard. CRS Report RS22990 (2010).8. Luo, Y. T., Zhu, J. H., Fan, S. S. & Chen, G. J. Study of the kinetics of hydrateformation in a bubble column. Chem. Eng. Sci. 62, 10001009 (2007).9. Ukai, T., Kodama, D., Miyazaki, J. & Kato, M. Solubility of methane in alcoholsand saturated density at 280.15 K. J. Chem. Eng. Data 47, 13201323 (2002).10. Kou, Y., Xiong, W., Tao, G., Liu, H. & Wang, T. Adsorption and capture ofmethane into ionic liquid.J. Nat. Gas Chem. 15, 282286 (2006).11. Yu, M., Zhai, L. Y., Zhou, Q., Li, C. P. & Zhang, X. L. Ionic liquids as novelcatalysts for methane conversion under a DC discharge plasma. Appl. Catal. A419, 5357 (2012).12. Klamt, A. Conductor-like screening model for real solvents: a new approachto the quantitativecalculation of solvation phenomena. J. Phys. Chem. 99,22242235 (1995).13. Klamt, A. COSMO-RS: From Quantum Chemistry to Fluid PhaseThermodynamics and Drug Design (Elsevier, 2005).14. Ben-Naim, A. Solvation Thermodynamics 1st edn (Plenum Press, New York,USA, 1987).15. Maiti, A. Theoretical screening of ionic liquid solvents for carbon capture.ChemSusChem 2, 628631 (2009).16. Brunner, E. & Hultenschmidt, W. Fluid mixtures at high pressures VIII.Isothermal phase equilibria in the binary mixtures: (ethanol hydrogen ormethane or ethane). J. Chem. Thermodyn. 22, 7384 (1990).17. Katayama, T. & Nitta, T. Solubilities of hydrogen and nitrogen in alcohols andn-hexene. J. Chem. Eng. Data 21, 194196 (1976).18. Kretschmer, C. B., Nowakowska, J. & Wiebe, R. Solubility of oxygen andnitrogen in organic solvents from 25oto 50oC. Ind. Eng. Chem. 38, 506509(1946).19. Dalmolin, I. et al. Solubility of carbon dioxide in binary and ternary mixtureswith ethanol and water. Fluid Phase Equilibria 245, 193200 (2006).20. Rogers, R. D., Seddon, K. R. & Volkov, S. (eds). Green industrial applications ofionic liquids. NATO Sci. Ser. (Kluwer, Dordrecht, 2002).21. Anderson, J. L., Dixon, J. K. & Brennecke, J. F. Solubility of CO2, CH4,C2H6, C2H4, O2, and N2 in 1-hexyl-3-methylpyridiniumARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms26976 NATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications&2013MacmillanPublishersLimited.Allrightsreserved.bis(triuoromethylsulfonyl)imide: comparison to other ionic liquids. Acc.Chem. Res. 40, 12081216 (2007).22. Pophale, R., Deem, M. W. & Cheeseman, M. W. A database of new zeolite-likematerials. Phys. Chem. Chem. Phys. 13, 1240712412 (2011).23. Fang, H. et al. Prediction of CO2 adsorption properties in zeolites using forceelds derived from periodic dispersion-correctedDFT calculations. J. Phys.Chem. C 116, 1069210701 (2012).24. Soler, J. M. et al. The SIESTA method for ab initio order-N materialssimulation. J. Phys. Cond. Matter 14, 27452779 (2002).25. Wilems, T. F., Rycroft, C. H., Kazi, M., Meza, J. C. & Haranczyk, M.Algorithms and tools for high-throughput geometry-based analysis ofcrystalline porous materials. Micropor. Mesopor. Mater. 149, 134141(2012).26. Kim, J., Martin, R. L., Rubel, O., Haranczyk, M. & Smit, B. High-throughputcharacterization of porous materials using graphics processing units.J. Chem.Theory Comput. 8, 16841693 (2012).27. Kim, J. & Smit, B. Efcient Monte Carlo simulations of gas molecules insideporous materials. J. Chem. Theory Comput. 8, 23362343 (2012).28. Kim, J., Lin, L. -C., Swisher, J. A., Haranczyk, M. & Smit, B. Predicting largeCO2 adsorption in aluminosilicate zeolites for postcombustion carbon dioxidecapture. J. Am. Chem. Soc. 134, 1894018943 (2012).29. Dawson, R., Cooper, A. I. & Adams, D. J. Nanoporous organic polymernetworks. Prog. Polym. Sci. 37, 530563 (2012).30. Bagheri, N. & Abedi, J. Adsorption of methane on corn cobs based activatedcarbon. Chem. Eng. Res. Des. 89, 20382043 (2011).31. Chen, J., Li, W., Li, X. & Yu, H. Improving biogas separation and methanestorage with multilayer graphene nanostructure via layer spacing optimizationand lithium doping: a molecular simulation investigation. Environ. Sci. Technol.46, 1034110348 (2012).32. Maiti, A., Kumar, A. & Rogers, R. D. Water-clustering in hygroscopic ionicliquidsanimplicit solvent analysis. Phys. Chem. Chem. Phys. 14, 51395146(2012).33. Schafer, A., Klamt, A., Sattel, D., Lohrenz, J. C. W. & Eckert, F. COSMOimplementation in TURBOMOLE: extension of an efcient quantumchemical code towards liquid systems. Phys. Chem. Chem. Phys. 2, 21872193(2000).34. Becke, A. D. Density-functionalexchange-energy approximation with correctasymptotic behavior.Phys. Rev. A 38, 30983100 (1988).35. Perdew, J. P. Density-functional approximation for the correlation energy of theinhomogeneous electron gas. Phys. Rev. B 33, 88228824 (1986).36. Eichkorn, K., Weigend,F., Treutler, O. & Ahlrichs, R. Auxiliarybasis sets formain row atoms and transition metals and their use to approximate coulombpotentials. Theor. Chim. Acta 97, 119124 (1997).37. Schafer, A., Huber, C. & Ahlrichs, R. Fully optimized contracted gaussianbasis sets of triple zeta valence quality for atoms Li to Kr. J. Chem. Phys. 100,58295835 (1994).38. Klamt, A., Eckert, F., Hornig, M., Beck, M. E. & Burger, T. Prediction ofaqueous solubility of drugs and pesticides with COSMO-RS. J. Comp. Chem. 23,275281 (2001).39. Myers, A. L. & Prausnitz, J. M. Thermodynamics of mixed-gas adsorption.AICHE 11, 121127 (1965).40. Garcia-Perez, E. et al. A computational study of CO2, N2, and CH4 adsorptionin zeolites. Adsorption 13, 469476 (2007).41. Smit, B. & Maesen, T. Molecular simulations of zeolites: adsorption, diffusion,and shape selectivity. Chem. Rev. 108, 41254184 (2008).AcknowledgementsThe work at LLNL was performed under the auspices of the US Department of Energy byLawrenceLivermore National Laboratory under Contract DE-AC52-07NA27344. J.K.was supported by the U.S. Departmentof Energy, Ofce of Basic Energy Sciences,Materials Sciences and Engineering Division under Contract No. DE-AC02-05CH11231.J.K. was also supported by the Assistant Secretary for Fossil Energy of the U.S.Department of Energy under Contract No. DE-AC02-05CH11231. The information,data, or work presented herein was funded in part by the Advanced Research ProjectsAgency - Energy (ARPA-E), U.S. Department of Energy. L.-C.L. was supported by theDeutsche Forschungsgemeinschaft (DFG, priority program SPP 1570). B.S. was sup-ported as part of the Center for Gas Separations Relevant to Clean Energy Technologies,an Energy Frontier Research Center funded by the US Department of Energy, Ofce ofScience, Ofce of Basic Energy Sciences under Award Number DE-SC0001015. Thisresearch used resources of the National Energy Research Scientic Computing Center,which is supported by the Ofce of Science of the US Department of Energy underContract No. DE-ACO2-05CH11231.AuthorcontributionsA.M. coordinated the project, led the effort in liquid/ionic liquid systems and wrote alarge part of the manuscript. J.K. and L.-C.L. performed large-scale simulations on thezeolites and wrote the zeolite section of the paper. J.K.S. contributed motivating scenariosand introductory text. R.D.A. and B.S. provided insightful comments and usefulguidance.AdditionalinformationSupplementary Informationaccompanies this paper at http://www.nature.com/naturecommunicationsCompeting nancial interests: The authors declare no competing nancial interests.Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/How to cite this article: Kim, J. et al. New materials for methane capture from dilute andmedium-concentration sources. Nat. Commun. 4:1694 doi: 10.1038/ncomms2697 (2013).NATURECOMMUNICATIONS|DOI:10.1038/ncomms2697 ARTICLENATURECOMMUNICATIONS | 4:1694| DOI: 10.1038/ncomms2697 | www.nature.com/naturecommunications 7&2013MacmillanPublishersLimited.Allrightsreserved.