• Tidak ada hasil yang ditemukan

state-of-the-art, challenges, and opportunities Zeolitic imidazolate framework membranes for gas separations:Current Journal of Industrial and Engineering Chemistry

N/A
N/A
Protected

Academic year: 2024

Membagikan "state-of-the-art, challenges, and opportunities Zeolitic imidazolate framework membranes for gas separations:Current Journal of Industrial and Engineering Chemistry"

Copied!
38
0
0

Teks penuh

(1)

Review

Zeolitic imidazolate framework membranes for gas separations:

Current state-of-the-art, challenges, and opportunities

Mohamad Rezi Abdul Hamid

a,

*, Thomas Choong Shean Yaw

a

,

Mohd Zahirasri Mohd Tohir

a

, Wan Azlina Wan Abdul Karim Ghani

a

, Putu Doddy Sutrisna

b

, Hae-Kwon Jeong

c,

*

aDepartmentofChemicalandEnvironmentalEngineering,FacultyofEngineering,UniversitiPutraMalaysia,Serdang,Selangor43400,Malaysia

bDepartmentofChemicalEngineering,FacultyofEngineering,UniversityofSurabaya(UBAYA),Surabaya,Indonesia

cArtieMcFerrinDepartmentofChemicalEngineeringandDepartmentofMaterialsScienceandEngineering,TexasA&MUniversity,CollegeStation,TX 77843-3122,UnitedStates

ARTICLE INFO

Articlehistory:

Received6March2021

Receivedinrevisedform27March2021 Accepted30March2021

Availableonline7April2021

Keywords:

Metalorganicframeworks Zeoliticimidazolateframeworks Membranes

Gasseparations Hollowfibers

ABSTRACT

Advancedporousmaterialswithuniformmolecularscaleaperturesarehighly desirableforenergy related separations.Among them, zeolitic imidazolateframeworks(ZIFs) have received significant interestasseparationmembranesowingtotheirnarrowaperturesthatcantargetsmallgases.ZIFs exhibit excellent gas transport properties, such as unprecedentedly high C3H6 permeability and remarkable C3H6/C3H8 selectivity, inaccessible byother porous inorganic materials and polymers.

DepositionofultrathinZIFsonporoussubstratestoformgas-selectivebarriershasbeenamajorfocusin thisarea.TherehasbeenasignificantdevelopmentinthesynthesisofultrathinZIFmembranesforgas separations.Inthisreview,wepresentasummaryofcurrentstate-of-the-artinZIFmembraneprocessing andhighlightuniquemicrostructuralfeaturesofthepreparedmembranes.Followingthis,wediscuss levelofseparationperformancesoftheseadvancedmembranesfocusingonthreeemerging/unsolved applications.Finally,weprovideourperspectivesonfutureresearchdirectionsinthearea.

©2021TheKoreanSocietyofIndustrialandEngineeringChemistry.PublishedbyElsevierB.V.Allrights reserved.

Contents

Introduction ... 18

Zeoliticimidazolateframework(ZIF)membranes ... 20

AscalableZIFmembraneconcept ... 20

Generalsynthesisstrategy(in-situvs.secondarygrowth) ... 21

State-of-the-artinZIFmembranesynthesisforgasseparations ... 22

ReducingZIFmembranesthickness.Whatisthelimit? ... 22

Interfacialsynthesis ... 23

Continuousmicrofluidicflowprocessing ... 24

Solvent-freevaporphasesynthesis ... 24

Electrochemicalbasedsynthesis ... 25

ApplicationareasofZIFmembranes ... 26

H2purifications ... 27

CO2separations ... 30

Condensablegasseparations ... 31

C3olefin/paraffinseparations ... 31

C2olefin/paraffinandC4isomersseparations ... 34

Futureperspectives ... 35

*Correspondingauthors.

E-mailaddresses:[email protected](M.R.AbdulHamid),[email protected](H.-K.Jeong).

https://doi.org/10.1016/j.jiec.2021.03.047

1226-086X/©2021TheKoreanSocietyofIndustrialandEngineeringChemistry.PublishedbyElsevierB.V.Allrightsreserved.

ContentslistsavailableatScienceDirect

Journal of Industrial and Engineering Chemistry

j o u r n a l h o m ep a g e: w w w . e l s e v i e r . c o m / l o c at e / j i e c

(2)

SynthesisofpolycrystallinemembranesotherthanZIF-8 ... 35

HybridSODZIFcontainingnon-isostructuralsecondarylinkers ... 35

ZIF-basedmixed-matrixmembranes ... 36

Realistictestconditionandothertechnicalchallenges ... 36

Finalremarks ... 37

Declarationofinterest ... 37

Acknowledgements ... 37

References ... 37

Introduction

Separationandpurificationofchemicalmixturesareimportant stepsinmodernchemicalindustries,accountingforroughly10– 15%oftheglobalenergyconsumption[1,2].Developmentofmore energyefficientchemicalseparationprocessescouldsavebillions ofdollarsofenergycostandreducegreenhousegasemission[1].

Membrane-based technology is a promising energy efficient alternative overtheconventional thermally-drivenprocessesto purifyindustriallyimportantgases.Gasseparationsbymembranes require less energy and investment cost compared to other competing technologies including distillation, adsorption, and absorption[3,4].SchollandLively[1] estimatethat membrane- basedseparationsuse90%lessenergycomparedtothedistillation.

Amongotheradvantagesofmembranesarelowenergyconsump- tion,operationalflexibility(meaningthatthemembranecanwork asastandaloneunitorretrofittedintoexistingprocessunit),small carbon footprint, and linear scale up (applicable for small and mediumscaleprocesses)[5–9].

Membranetechnologyisafairlymaturetechnologybutisstill expanding.Membranetechnologyhasaprojectedmarketvalueof

$2.6billionin2022,86%higherthanthemarketvaluebackin 2018[6,10].Themajorityofcommercialgasseparationmembranes are polymerbased, inwhich 90% of themare intended forthe separation of non-condensable gases such as hydrogen (H2) purification, nitrogen(N2)production fromair,and natural gas treatment[4,10,11].Polymersareaversatileclassofmaterialwith diversefunctionalitiesandexcellentprocessabilities[12].Theyare inexpensiveandcanbeformedasflatsheetsorhollowfiberswith effectiveskinlayerthicknessesoflessthan100nmusingphase separationtechniquesandtheirvariance[11].Polymershowever suffer from the permeability-selectivity trade-off[13]. In other

words,highlypermeablepolymerswillpossesslowselectivityand viceversa.Thetrade-offrelationshiprepresentsachallengeamong membranescientists todeveloppolymermembraneswithhigh permeabilities and selectivities. While there have been some improvementsmadeovertherecentyears,onlyamarginalshiftin theupper boundwas observed [14].Polymers alsosuffer from swellingandplasticizationissueswhichtypicallyoccuratelevated pressurescompromisingmembraneseparationperformances[15– 17].It isalsoworthyofmentioningthat90%ofthecommercial polymer membranes are made from less than 10 polymer materials(e.g.,celluloseacetate,polyimide,polysulfone,polycar- bonate,andsiliconerubber)whichmosthavebeenusedforthe lastfourdecades[18].

Otherthan the bigfour commercialmembrane applications (i.e.,H2purification,N2productionfromair,naturalgastreatment, and vapor recovery), a much larger potential market for membranesareintheseparationsofcondensable gasessuchas methane(CH4)recoveryfromheavier hydrocarbons and olefin/

paraffin separations (e.g., ethylene/ethane, propylene/propane, andn-butane/i-butane)[4].Forinstance,separationofolefinfrom theirrespectiveparaffinisoneofthelargestgas-phaseseparations inchemicalindustriestypicallyachievedthroughlowtemperature distillation. Ethylene (C2H4) and propylene (C3H6) are the two largestchemicalcommoditiesin chemical/petrochemicalindus- trieswithacombinedannualproductionof2.9108metrictons [19]. C2H4 and C3H6 are important chemical feedstocks for the productionofawidevarietyofintermediatesandpolymers.For production of polymers(e.g., polyethylene and polypropylene), C2H4andC3H6purityof99.9mol%and99.5mol%,respectively,are required[20,21].Olefin/paraffinseparationsusingpolymermem- branes are challenging due to similar physical and chemical propertiesofthepenetrants.Moreover,poorlydefinedfreevolume

Fig.1.(a)ExampleofcrystalstructuresofZIFs.AdaptedwithpermissionfromRef.[38]Copyright2009AmericanChemicalSociety(b)Comparisonoftheporeaperturesof ZIFs(XRD-derivedoreffectiveapertures)withmolecularsizesofpenetrants.Inthisfigure,hybridmoleculardimensions(kineticdiameterandVanderWaals)areusedto representmolecularsizesofpenetrants.PenetrantsizesaretakenfromRef.[39]and*symbolrepresentseffectiveapertureofZIFs.

(3)

ofpolymersresultinlowdiffusionalselectivity[5].Toaddressthe shortcoming of polymers, researchershave begun investigating othertypesofmembranematerialssuchaszeolites[22],carbon molecularsieves[23],graphene[24],andmetals[25].Generally speaking, inorganicmicroporousmembranespossessbettergas transportpropertiesthanthoseofpolymermembranesinaddition tohavingsuperiorchemicalandthermalstabilities.

Metalorganicframeworks(MOFs)arehybridorganic-inorganic porousmaterialsdesignedbybridginginorganicmetalsormetal- containing clusters and organic linkers through coordination bonding [26–28]. MOFs are well known for having uniform molecular scale apertures, large internal surface areas (1,000– 10,000m2g1), high porosities (up to 90% pore volume), and tailorable pore sizes and internal surfaces [29,30]. Unique propertiesofMOFshavedrawnconsiderableattentionforawide variety of technologicalapplicationsincludinggas storage[31], chemicalseparations[32],catalysis[33],gassensing[34],anddrug delivery[35].Zeoliticimidazolateframeworks(ZIFs),asubsetof MOFs,areconstructedbybridgingdivalentmetalions(e.g.,Zn2+, Co2+, Cd2+, etc.) with imidazole-based linkers to form three- dimensionalnetworkswithzeolitetopologies(e.g.,SOD,MER,LTA, RHO,etc.)asdepictedinFig.1(a)[36,37].ZIFsarethermallystable up to550C and chemically inertinvarious solventsincluding alkalinewaterandorganicsolvents[37].

MostZIFshaveporesizesoflessthan5.0Åwhichareinthesize range of small molecules. Among the ZIF structures that are available,SOD-typeZIFssuchasZIF-7(0.30Å)[37],ZIF-8(0.34Å) [37],ZIF-9(<0.34Å)[37],ZIF-67(0.33Å)[40],andZIF-90(0.35Å) [41]isthemostrelevantZIFtopologyforsmallgassieving.TheSOD cageofZIFsisshowninFig. 1(a).Narrowsix-memberedringsofthe cageprovidehighgasdiffusionselectivitywhilelargeopencavities guaranteehighgasdiffusionthroughtheframeworks.Thermally activatedflip-floppingmotionsofimidazoleligandsenlargesthe pore aperturesof ZIFsbeyond itsXRD-derived values. Effective apertureofZIF-8wasfoundtobemuchlarger(4.0–4.2Å)thanits XRD-derived apertureof3.4Å, attributabletotherotationof 2- methylimidazole linkers [39]. Meanwhile, ZIF-90 has effective apertureof5.0Å,muchlargerthanitsnominalapertureof3.5Å [42].Beingcrystallinematerials,ZIF sufferfromhaving discreet and limited available apertures as shown in Fig.1(b). In other words, ZIFsusabilityarelimitedtoaparticulargasmixtureand cannotbeusedtoefficientlyseparateothergases.

Researcherstypicallygoaroundthisfundamentallimitationof ZIFsbyadoptinga hybridapproach. Incorporationof secondary metalsorlinkersintoparentZIFstructureseitherthroughdenovo approach or post-synthetic modification (PSM) modifies the transport properties ofZIFstotarget othergasmixtures which previously inaccessible when using pure ZIFs [42–48]. Prior researchonZIFshasmostlyfocusedonthediscovery,characteri- zation,andutilizationofthematerialsintheirpowder/bulkform.

However, over the recent years there have been particularly significantdevelopmentsinZIFmembranesynthesisfor separa- tionapplicationswithvariousinnovativestrategiesproposed[49– 54].AsillustratedinFig.2,thenumberofpublicationsonZIF-based membranesforgasseparationshavesteadilyrisen.ThelistofZIFs beinginvestigatedasseparationmembranesincludesZIF-7,ZIF-8, ZIF-9, ZIF-22,ZIF-67, ZIF-78,ZIF-90,ZIF-93,ZIF-95,and ZIF-100 amongothers,withZIF-8beingthemostwidelyinvestigatedZIFs.

TherearetwoapproachesadoptedbyresearcherstoutilizeZIFs forseparationapplications.Firststrategyistoembedpreformed ZIFnanocrystalsintoacontinuouspolymermatrixtoformwhatis knownas mixed-matrixmembranes(MMMs). MMMs arecom- posite materials consisting of zero-, one- or two-dimensional inorganicnanofillers(typicallymoreselective),dispersedhomo- genously in a continuouspolymer phase [5].Unlike thepurely inorganiczeolites,organiclinkersofZIFsprovidebetterinteraction

with polymer matrix, resulting in a more intimate interface.

MMMs combine desirable properties of both polymer and inorganicphases[55].Openframeworksandsize-selectivenature of ZIF fillers allow for improvement in composite membrane permeability and selectivity while maintaining low cost and processingconvenienceof polymer.ForZIF-based MMMs,there havebeenseveralinterestingdiscoveriesmade.Forinstance,Park etal.[56]managedtoin-situtransformsub-1

m

mthick(750nm) premade 6FDA-DAM (4,40-(hexafluoroisopropylidene) diphthalic anhydride-2,4,6-trimethyl-1,3-diaminobenzene) polyimide coat- ingoncommercialpolyethersulfonehollowfibersandα-alumina discsintoZIF-8/6FDA-DAMMMMsusingwhattheyrefertoasthe polymer-modification-metal-organic-framework-formation (PMMOF)technique.ThePMMOFedMMMsshowedimpressively highpropylene/propane(C3H6/C3H8)separationfactorashighas 38.07.1 satisfying the commercial requirement despite using lowervolumefractionofZIF-8(32.9vol%).Thoughinteresting,the subjectofMOF/ZIF-basedMMMsforgasseparationsisbetterleft foranotherdiscussion.Moreover,therearemanyexcellentreviews alreadyavailabletokeepthereaderupdatedonthesubjectmatter [57–61].

SecondstrategytoutilizeZIFsforseparationapplicationsisto grow a continuous and defect-free polycrystalline ZIF layer.

PolycrystallineZIFmembranesareusuallyconsideredasinorganic membranesbecausetheirstructures,synthesis,andgastransport propertiesaresimilartothoseofinorganiczeolites[62].Sincefree standing ZIFs possess low mechanical strength, the molecular sievelayersareusuallygrownonporoussupports(e.g.,ceramic, polymer,andmetal)toprovidemechanicalstrength[63].Unlike ZIF-based MMMs which show moderate improvement in both permeabilityandselectivity,separationperformancesofpureZIF membranesaresignificantlyhigher.However,amajorchallengeis tofabricatelargeareaanddefect-freeZIFpolycrystallinelayerson porous substrates in cost-effective manner. There have been significantdevelopmentsinthisareainthepastyears.Despitethe risinginterest,thenumberofpublishedworksonZIFmembranes onindustriallyrelevantsubstrates(i.e.,hollowfiber,capillary,and tube)isstilllow.Thisisnosurpriseconsideringthedifficultiesin preparing high quality ZIF membranes on these substrates, especially onthelumen side of thefibers, compared toplanar substrates.

Herein,wepresentasystematicreviewofcurrentstate-of-the- artinpolycrystallineZIFmembranessynthesis,theirapplications ingasseparations,andthereportedseparationperformances.A selectionofresearchworksreportedwithinthelastfiveyearshas Fig. 2.Number of publications per year with phrase “zeolitic imidazolate framework(ZIF)membranes”and“zeoliticimidazolateframework(ZIF)hollow fibermembranes”since2009.DataobtainedfromWebofScience(December2020).

(4)

become the main focus of this review. The applications of ZIF membranes included in the discussion are limited to three unsolved applications in gas separation membranes (e.g., H2

purification,CO2removal,andolefin/paraffinseparations).There are several noteworthy reviews on ZIF membranes for gas separations already available in literature [49–54]. To set our workapart,wedecidedtoputmoreemphasisonthepreparation of ZIF membranes on high packing substrates, which is a less popularsubject inthefield.Wethenprovidecommentarieson several technical challenges in ZIF membrane synthesis and conclude by highlighting future research direction on the developmentofpure/hybridZIF membranestargetingothergas mixtures. We hopethatthecurrent contributioncouldprovide meaningfulinsightsandideasforthedesignofhighperformance gas separationmembranes.We alsohopetoinspiremembrane scientiststocomeoutwithinnovativesolutionstofabricatehigh performanceZIFmembranesinacost-effectivemanner.

Zeoliticimidazolateframework(ZIF)membranes

AscalableZIFmembraneconcept

Fig.3summarizesimportantcriteriaforwhatwebelievetobea scalable and economically attractive ZIF membrane concept.

Selectingmembranematerialswithpropergaspermeabilityand selectivityisthefirststeptowardsindustrialscaledeploymentof gasseparationmembranes.Economicsofthemembranesystem relyheavilyonmembranetransportproperties.Highlyselective andhighlypermeablemembranesguaranteehighpurityproduct withhighrecoveryrateusinglessmembranearea[12].ZIFsbeing crystallinematerialsoffergoodselectivityandpermeabilityfora specificgasmixture.Forinstance,ZIF-8offershighintrinsicC3H6

permeability of390Barrerandpermselectivityof 130;anideal candidate for C3H6/C3H8 separations (1 Barrer=3.3481016 molmm2s1Pa1) [39]. While this aspect of gas separation membranesislargelyresolvedthroughaproperZIFselection,one shouldnotethatnomatterhowpermeabletheselectedZIFis,it must befabricatedintoathin (sub-1

m

mthick)and defect-free layertoachievehighgasfluxes[18].Incommercialgasseparation plants, the required membrane area is in the order of several hundred thousands of square meter (1,000–500,000m2) [18].

Increasingmembraneproductivity iscriticallyimportantforZIF membranesasthemembranesarelikelytohavehigherfabrication

costcomparedtopolymermembranes.Highlyproductivemem- branesrequire smallermembraneareatoperformtherequired separation. Capital investment for membrane installations can thereforebereducedtoapointwhereitcanbeaseconomically attractiveasthetraditionalpolymermembranes.

Membraneproductivity,whichisthecoreideaofthisreviewis definedasamolarflowrateoffeedgasesthatamembranecan processperunittime(mols1).Theproductivityofamembrane canbeexpressedusingthefollowingequation:

Qi¼PiA

D

pi

l   ð1Þ

where Pi,A,l and

D

piarepermeability ofgas i(molmm2s1 Pa1),membranearea(m2),membranethickness(m),andpartial pressuredifferenceofgasibetweenfeedandpermeatesides(Pa), respectively.Qithatistheamountofgasesthatamembranecan processperunittime(mols1)dependsonmembranethickness and surface area. Theoretically, one can achieve a ten-fold membraneproductivityincreasebyreducingmembranethickness from1

m

mto0.1

m

m.Productivityincreasecanalsobeachievedby increasingmembranesurfacearea.Inotherwords,synthesizing themembraneonhighsurface-to-volumemodulessuchashollow fibers.Hollowfiberscanprovidehighpackingdensityupto13,000 m2m3.FabricatingZIFmembraneonhighpackingmodulesand/

orreducingmembranethickness,ideallytosub-0.1

m

mrangeare amongtheareascurrentlyaddressedbymembraneresearchers.

However,itshouldberemindedthatneitherofthesestrategiesis straightforwardwithonlyafewlabscalesuccessesreported.

ScalableZIFmembranesmust haveexcellentreproducibility, robustsynthesis, andability tobefabricatedintomodulesin a cost-effective manner. In addition, the membranes must have excellentchemicalandthermalstabilitiesespeciallyforthosethat willbeusedunderchemicallyaggressive andhightemperature conditions.Thecostofinorganicmembranesarestillexpensive.

AccordingtoLinetal.[62],thecostofinorganicmembranesgrown on ceramic microfiltration membranes is anywhere between

$1,000–5,000perm2.Forcomparison,thecostofpolymermem- branesis$5and$10perm2 forhollowfiberand flatgeometry, respectively[64].In somestudies,polymershavebeen usedto replace inorganic supports in the synthesis of ZIF membranes aimingtoreducemembranecost.However,thesupportsgenerally do not have good thermal and chemical stabilities. Inorganic supportsincludingalumina,titania,zirconia,andmetalarefairly

Fig.3.SchematicillustrationandcriteriaforascalableandcommerciallyattractiveZIFmembraneconcept.

(5)

expensive,contributingtoroughly80%oftheoverallmembrane cost [65]. Nevertheless, thelong term cost saving from having chemically and thermally resilient supports might favor the selectionofinorganicsupportsoverorganicsupports[64].Finally, industriallyrelevantZIFmembranesmust havelongmembrane lifetimeofatleastthreetofiveyears[18].Atpresent,difficultyin preparingdefect-freeultrathinlargeareamembranes,lowmodule packingdensity,highmanufacturingcost,andmodestreproduc- ibilityareamongthebarriersthatpreventedthecommercialscale deploymentofZIFmembranes[62].

Generalsynthesisstrategy(in-situvs.secondarygrowth)

FabricationprotocolsofZIFthinfilmsandmembranesgenerally followsthatofzeolitemembranesasbothareporouscrystalline materials[9].ZIFmembranescanbepreparedviain-situmethod orsecondary(seeded)growthmethod.In-situsynthesisisamore popular choice among researchers to prepare polycrystalline membranes as the preparation steps is less complicated [50].

Meanwhile, secondary growth can provide better control over membranemicrostructures(e.g.,crystalorientation,crystalsizes, thickness, grain boundary structures, etc.) which consequently affectthemembranetransportproperties[66].

As illustrated in Fig. 4(a), in-situ synthesis is a one-step solvothermal or hydrothermal membrane formation onporous substrateswithoutpre-attachedseedcrystals.Unlikethesecond- arygrowthstrategy,in-situsynthesisdoesnotrequirecomplicated preparations,makingitattractiveforscaleup.Traditionally,in-situ membrane synthesis is achieved by directly immersing bare, surface modified, or metal saturated supports in a synthesis solution for a predeterminedperiod. To date,various synthesis protocolshavebeendevisedforthepreparationofpolycrystalline ZIFmembranesincludingdirectsolvothermalgrowth[67],layer- by-layerdeposition[68,69],counter-diffusion[70,71],microwave synthesis [43,72], electrospray deposition [73,74], interfacial growth[75,76],andsolvent/solvent-freezincoxidetransformation [77–79]along withothertechniques.In-situsynthesis hasbeen appliedtoprepareawidevarietyofZIFstructuresincludingZIF-7 [74,80],ZIF-8[67,81],ZIF-9[82],ZIF-22[83],ZIF-67[84],ZIF-78 [85],ZIF-90[86,87],ZIF-93[88],ZIF-95[89],ZIF-100[90],mixed- metalZIFs(e.g.,Co–Zn–ZIF-8[91,92]),andmixed-ligandZIFs(e.g., 2-ethylimidazole-ZIF-8 [93],benzimidazole-ZIF-8 (ZIF-7-8)[94], and2-imidazolecarboxaldehyde-ZIF-8(ZIF-8-90)[95]).

Inorganicceramics,metals,andpolymersareamongthemost commonly used substrates for in-situ ZIF membrane synthesis.

They,however,arerelativelyinertprovidinglowheterogeneous nucleationdensity[96–99].Physicalorchemicalfunctionalization are often introduced to promote and direct heterogeneous nucleationand growthofZIFsonpreferredsurfaces.Decorating support surfaces with reactive groups suchas amino [82,100], hydroxyl[96,101,102],andimidazoline[69,103]groupsimproves heterogeneousnucleationofZIFsowingtotheirability toform complexeswithmetalions.Thesesurfacegroupsareexpectedto promotebetteradhesionbetweenZIFlayersandsupportsthrough covalent (e.g., Zn–N) or noncovalent bonds which result in mechanicallystrongmembranes[83,87,104].3-aminopropyltrie- thoxysilane(APTES)surfacemodificationshavebeenemployedto fabricate tightly packed ZIF polycrystalline layers [82,100].

Following a similar concept, Jiang et al. [101] and Ruan et al.

[102]functionalizedsurfaceofanodized aluminumoxide(AAO) andα-Al2O3discs,respectively,withpolydopaminebio-adhesives tofabricatehighlypermselectiveZIF-8membranes.

In the traditional in-situ synthesis, crystal nucleation and growth occur simultaneously, providing small window for independent optimizations. It is therefore quite challenging to controlmicrostructuresoftheZIFmembranespreparedusingthis approach. Take solution-based-counter-diffusion method as an example.Ideally,‘metal-ligandreactionzone’shouldoccuratthe vicinity of the substrates [67]. Slow precursor diffusion rates relativetoreactionratesleadtoaformationofthickmembranes insidethesupports,whichreducethemembranefluxes.Onthe otherhand,highprecursordiffusionratesrelativetoreactionrates lead to homogenous nucleation resulting in discontinuous membraneswith macroscopicvoids [105]. Nucleationdensities ofZIFonsupportsurfacesdetermineminimumthicknessofthe resultingmembranes(i.e.,lownucleationdensitiesonsupports producethickmembranesandviceversa).

Fromourreview,wenoticedthatresearchershavebegun to moveawayfromthetraditionalsolution-basedin-situsynthesis.A fewofthemhavebegunconsideringsolvent-lessroutestoprepare ZIFmembranes.Thesenewmethodsnotonlysimplerandfaster, butalsoabletoproduceZIFmembraneswiththicknessoflessthan 100nm.Wewillreviewtheseinnovativestrategiesindetailinthe followingsection.

Secondarygrowthisanapproachwidelyusedinthesynthesis of inorganic zeolite membranes which later applied to ZIF membranesynthesis[108–110].Asopposedtothein-situmethod, secondary growth is a two-steps process which involves the depositionofhighqualityseedcrystalsonsubstratesfollowedby hydrothermal or solvothermal growth where the seed crystals

Fig.4.Comparisonbetween(a)in-situgrowthand(b)secondaryseededgrowthforZIFmembranesynthesis.InsetelectronimageshowsdenselypackedZIF-8seedlayers depositedonα-Al2O3substratesusingmicrowaveheating.ModifiedwithpermissionfromRef.[106]Copyright2010andRef.[107]Copyright2015AmericanChemical Society.

(6)

growandlaterformcoherentfilms.Ingeneral,highqualityseed layers consist of densely-packed seeds with uniform surface coverageandstronganchoragetosupportsurfaces(seeFig.4(b), inset). Theseedlayerscanbeobtainedthroughmanualrubbing [111], dip/slip coating [44,92,112], reactive seeding [85,113], microwave seeding [91,107,114], and thermal seeding [115]. A denselypackedZIFseedlayerdepositedonsubstrateenablesthe formationofathinmembrane.Inmostcases,themajorityofZIF membranes prepared using the secondary growth method are relatively thin withthicknessesof 2–3

m

mat most[91,107,116– 120].Despitemorecomplicated,nucleationandgrowth stepsin secondary growth are decoupled, allowing each step to be optimized independently. Therefore, one has the freedoms to manipulateimportantnucleationandgrowthparameterstoobtain membraneswithdesiredmicrostructures.Forinstance,Buxetal.

[121]managedtogrow12

m

mthickZIF-8membraneswith(110) planepreferredorientationbyvaryingthesecondarygrowthtime.

Liuetal.[122]manipulatedsecondarygrowthsynthesisrecipeto favorcrystalgrowth alongc-directiontoformhighlyc-oriented ZIF-69membranesonporousα-Al2O3discs.

State-of-the-artinZIFmembranesynthesisforgasseparations

ReducingZIFmembranesthickness.Whatisthelimit?

Oneofthepressingissuesinzeolitemembranes,MOFandZIF membranes included, is high cost stemming from the use of expensiveinorganicsupports[3,123].Cheapersubstratessuchas polymers can reduce membrane manufacturing cost to some extent, but then again, thermal and chemical stability of the polymersgiverisetonewproblems.Expertsbelievethatatrue solution behind this matteris productivityincrease toa point where the capital cost of setting up inorganic membranes is economically as attractive to polymer membranes [124]. For practical purposes, increasing membrane flux (by decreasing membranethickness)islikelytheonlywayforwardtojustifythe relativelyhighcostofinorganicmembranes[95].Oneofthemajor goalsininorganicmembranecommunityistosynthesizehighly selectiveZIFmembraneswiththicknessinthesub-100nmrange.

Tsapatsis[124]estimatedthatforcertainapplications,membrane thicknessneedtobeevenfurtherdecreasedto50nm.Whilethis goalmayseemoutofourreachaboutadecadeago,severalrecently

publishedworkshavedemonstratedthatsynthesizingsub-100nm thickZIFmembranesisnolongerunattainable.

Fig. 5 shows important milestones in the development of ultrathinZIFmembranesmadeinthepastdecade.Tothebestof ourrecollection,thefirsteverZIFmembranesforgasseparationis a38

m

mthickZIF-8membranepreparedviamicrowavesynthesis in2009[72].Theapplicationofsecondarygrowthmethodinthe synthesisofZIFmembranesdrasticallyreducemembranethick- nessto1–3

m

m[107,114,120].Developmentsofsub-1

m

mthickZIF membranes appear in the latter half of the decades which consequently shifting membrane productivity toward higher values.Shamsaeietal.[131]preparedultrathinZIF-8membranes withthicknessof200nmonbromomethylatedpoly(2,6-dimethyl- 1,4-phenyleneoxide)substratesusingachemicalvapormodifica- tionmethod.Kwonet al.[126] tookadvantageoftheOstwald- ripeninglikeprocessoftheZIF-8nanocrystalsinaligandvapor environmenttofabricate300–400nmthickmembranesforC3H6/ C3H8 separations.Recently publishedwork by Qiao etal. [129]

reportedultrathinZIF-8membraneswiththicknessofonly45nm, exhibiting unprecedentedlyhighC3H6 permeanceof 3,000GPU (1GPU=3.3481010molm2s1Pa1). For comparison, C3H6

permeanceof 1–2

m

mthickZIF-8membranes arein therange of30–60GPUonly.TherecordforthinnestZIFmembranesforgas separationswasa17nmthickZIF-8membranesynthesizedbyLi etal.[130].Basedonthecurrenttrend,itisnotunrealistictoexpect ZIF membranes in the coming years tohave thickness of only severalunitcells.

Fromourreview,weobservedadeparturefromconventional solvent-basedmembranesynthesistosolvent-freerouteswhich currentlygaining popularity. Room temperature,openenviron- ment,andflowsynthesishavebeguntoslowlydisplacethehigh temperature autoclave-based synthesis. There are also a few scientific works attempting to synthesize ZIF membranes on hollowfibersandtubes,especiallyontheborespaceofthefibers [76,86,116,117,132]. In addition to thickness reduction, we also observedsignificantreductioninZIFsynthesistime.ZIFmembrane synthesiswhichadecadeagotakesafewhourstocompletecan nowbesynthesizedinjustunder10min[133].Hillmanetal.[43]

reported1

m

mthickhybridZIF-7-8membranespreparedinjust under90s,thefastestZIFmembranepreparationuptodate.The above-mentionedadvancesareaninterestingdevelopmentinthe area.

Fig.5.ImportantmilestonesinthedevelopmentofultrathinZIFmembranesmadesince2009.Notethatthemajoradvancesinthesynthesisofsub-500nmthickZIF membranesweremostlyreportedinlatterhalfofthedecades(lastfiveyears).ReprintedwithpermissionfromRefs.[72]and[67]Copyright2009andCopyright2013 AmericanChemicalSociety;reprintedwithpermissionfromRefs.[119]and[125]Copyright2011andCopyright2021ElsevierB.V;reprintedwithpermissionfromRef.[126]

Copyright2016RoyalSocietyofChemistry;reproducedwithpermissionfromRefs.[71,127,128,129],Copyright2017,Copyright2018,Copyright2019,andCopyright2020 WileyVCH;reproducedwithpermissionfromRef.[130]Copyright2017SpringerNature.

(7)

Inthissection,wereviewcurrentstate-of-the-artinultrathin (sub-1

m

m thick) ZIF membrane synthesis for gas separations.

Emphasiswillbeputtothosegrownonhighsurface-to-volume substrates (e.g., hollow fibers). The state-of-the-art in ZIF membranesynthesisisdividedintofourcategories:(i)interfacial synthesis,(ii)continuousflowprocessing,(iii)solvent-freevapor phase synthesis, and (iv) current-driven synthesis. We should emphasize that the ZIF membranes prepared usingthe above- mentioned methodsnotonlyinnovativebutalsoshowstate-of- the-artseparationperformancessurpassingmajorityofreported resultsprior.

Interfacialsynthesis

Interfacial synthesis takes advantageof theimmiscibility of solventstoconfineZIFcrystallizationselectivelyattheinterface betweenthetwosolvents.Akeyrequirementbehindthismethod is a selection of solvents with appropriate solubility towards organic and inorganicprecursors. Uponcontact, theprecursors inter-diffuse towards the respective immiscible phases and subsequentlycrystallizeattheliquid–liquidinterface.Continuous ZIF layer itself later becomes a diffusion barrier between the solventsthuslimittheliquid–liquidinterfaceonlyatlargeinter- crystallinegaps[76].Todate,anumberofZIFmembranes(e.g.,ZIF- 8[75,76,132,134,135]andZIF-90[86])havebeenpreparedusing this strategy. Different solvent interfacial systems have been investigated and in most cases the membranes wereprepared underrelativelymildsynthesisconditions(50–70C).

Interfacial microfluidic membrane processing (IMMP) devel- oped by Brown et al. [76] utilized 1-octanol/water interfacial systemtoprepareZIF-8membranesonTorlon1polyamide-imide hollow fibers for H2 purification and C3H6/C3H8 separation. Zn (dissolved in 1-octanol) and 2-methylimidazole (dissolved in water)solutionswerepassedthroughtheboreandshellsideofthe hollowfibershousedinsideacustom-madereactor.Twosynthesis variables were found important to prepare defect-free ZIF-8 membranes:flowconfigurations(static,continuous,orcombina- tionofboth)andinitialsynthesisconditions.Forexample,static flowofZn/1-octanolsolutiondidnotproducecontinuous ZIF-8 filmsduetoinsufficientZnionsinborespaceofthehollowfibers [76]. The ZIF-8 membranes supported on hollow fibers, after

optimizationsteps, werewell-adheredto substrates and had a uniform thickness of 8.8

m

m. Control over membrane micro- structuresandoptimizationofIMMPsynthesisconditionsintheir follow up works resulted in ZIF-8 membranes with similar thicknessbutwithmuchimprovedC3H6/C3H8separationperform- ances[132,135].Recently, IMMPwasusedtofabricateallnano- porousMFI/ZIF-8MMMs(8.0

m

mthick)ontheborespaceofpoly (amide-imide) hollow fibers for C3H6/C3H8 separations (see Fig.6(a))[136].CrystallineMFIzeoliteswithanaverageparticle sizeof141nmweredispersedincrystallineZIF-8matrixtoprepare all nanoporous crystalline MMMs. Energy dispersive X-ray elemental mapping was used to determine dispersion of the zeolite particles. Incorporation of medium pore MFI zeolites (channelporesizeof5.5Å)incrystallineZIF-8matrixboostedthe membranepermeabilitieswithoutsacrificingtheintrinsicselec- tivityofZIF-8[137].Thesilicalite-1particleswerefirsttreatedwith 1,3-diaminopropanetofunctionalizesilanoldefectsofthecrystals [138].TheparticleswerethendispersedinZn/1-octanolsolution andlaterpassedthroughboreofthehollowfiberataflowrateof 10

m

Lh1(aqueous2-methylimidazoleonshellside)toformthe hybridMFI/ZIF-8membranes.

Ina differentstudy,Biswal etal. [75] tookadvantageof the liquid–liquid interface between isobutyl-alcohol and water to prepare ZIF-8andCu-BTC MOF(BTC,1,3,5-benzenetricarboxylic acid) membranes on porous PBI-BuI hollow fibers for helium separation.Simplemanipulationofthelocationofmetalandlinker precursors inside the synthesis module enabled formation of membranesoneithersideofthehollowfibers.Theas-synthesized ZIF-8membraneswererelatively thick:10

m

mthickfor ZIF-8 membranegrownontheboresideand 20

m

mthickforZIF-8 membranegrownontheshellside.Itisnotedthatisobutyl-alcohol withlowerboilingtemperaturethanthatof1-octanol(108vs. 195 C) allows the solventto be removedfrom ZIF-8cavities using milderactivationtemperature.

Forgasseparationapplications,ZIF-8isnottheonlymaterialof interest here. There are a range of ZIFs for gas separation applications.However,solubilityproblemsbetweenZIFprecursors andsolventsmaylimitnotonlythetypesofZIFmembranestobe preparedbutalsolimittypesofmaterialstobeusedassubstrates [140].Forinstance,ZIF-7synthesisrequirespolaraproticsolvents (e.g.,N-methyl-pyrrolidinoneanddimetylformamide).Underthis

Fig.6.Interfacialsynthesisstrategytoprepare(a)allnanoporousMFI/ZIF-8MMMsonTorlon1hollowfibers(b)ZIF-90membranesonmacroporouscarbonhollowfibers.

ContinuousmicrofluidicflowprocessingstrategytoprepareZIF-8membraneson(c)Torlon1hollowfibersviain-situgrowthand(d)Matrimid15218hollowfibersvia secondarygrowth.ReproducedwithpermissionfromRef.[136]Copyright2019WileyVCH,Ref.[86]Copyright2017WileyVCH,Ref.[139]Copyright2017AmericanChemical Society,andRef.[117]Copyright2018ElsevierB.V.

(8)

circumstance,theuseofpolymericsubstratesisnotrecommended asmostofthemhavepoorchemicalstabilityinthesesolvents.In an attempt to grow ZIF-90 membranes on cross-linked PVDF hollowfibers,Eumetal.[86]observeddetachment/delamination ofZIF-90filmsfromthePVDFsubstrates(Fig.6(b)).Swellingofthe hollow fibers in dimetylformamide and later contraction toits originaldimensionupondryingcreatedinterfacialstrainstothe ZIF-90layerwhichledtodetachmentandcrumplingofthelayer.

Replacing PVDFhollow fibers with ‘inert’ carbon hollow fibers eliminated the swelling issues, resulting in defect-free ZIF-90 membraneswiththicknessof3.1

m

m.

Continuousmicrofluidicflowprocessing

Continuous microfluidic flow processing has become an increasingly popular method to synthesize ZIF membranes especially on bore spaceof hollow fibers. Polycrystalline ZIF-7, ZIF-8, ZIF-90, and ZIF-93 membranes have been successfully fabricated using this strategy [88,100,115,117,139,141,142]. In a typicalsynthesis,ZIFprecursorsarecontinuouslyfedthroughbore orshellsideofhollowfibers(usingperistalticorsyringepumps) fora predeterminedperiodtoobtaindefect-freemembranes.In mostcases,thesynthesisisperformedunderroomtemperature conditions. Membrane grown on bore of hollow fibers uses a relativelysmallamountofprecursorsolution(duetosmallbore spaceofthefibers),contributingtocostsaving.

Several flow protocols have been developed to prepare ZIF membranesonhollowfibers.Cacho-Bailoetal.[142]continuously injectedamixtureofmetalandlinkerprecursorstoborespaceof polysulfonehollowfibers(innerdiameter,ID315

m

m)ata flow rateof 100

m

Lmin1 for75mintoin-situformZIF-7and ZIF-8 membranes.ThepreparedZIF-8(3.6

m

mthick)andZIF-7(2.7

m

m thick) membraneswerethentestedfor H2purificationandCO2

separations.Inanotherwork,Martietal.[139]separatelyinjected 2-methylimidazoleandZnprecursorsolutionsthroughshelland boresideofTorlon1hollowfibers(outerdiameter,OD400

m

m), respectively, toprepare ZIF-8 membraneson thehollow fibers (Fig.6(c)).The8.8

m

mthickZIF-8membraneswerethentested post-combustion CO2captureapplication.ThelocationsofZIF-8

layersonthehollowfibersweremanipulatedbysimplyswitching theflowofprecursorsolutionsfromtheshellsidetotheboreside andviceversa.

Jeongetal.[115–117]andHuangetal.[100]utilizedcontinuous microfluidicsecondarygrowthtogrowZIF-8seedlayersonboreof polymer(e.g.,Matrimid15218,ID344

m

mandPVDF,ID1400

m

m) andα-Al2O3(ID 1000

m

m)hollowfibers previouslyobtainedvia solvothermal or microwaveheating. In both studies,Znand 2- methylimidazoleprecursorswerefirstmixedandagedforseveral minutes.Thegrowthsolutionwasthenfedthroughtheborespace ofthehollowfibersataspecifiedflowrate/durationtoallowthe seed crystalsto grow and form coherentfilms (Fig. 6(d)).One advantage of this approach is flexibility to choose duration of secondarygrowthandflowrateofgrowthsolution,whichenables a formation of thinner ZIF membranes. Continuous and well- intergrownZIF-8membranesasthinas 800nmthickonbore sideofMatrimid15218hollowfibersweredemonstratedbyour group[117].It is important tonotethat duringsynthesis, bulk crystallization of ZIFs still exists [143]. That is to say, if these homogenously grown ZIFs were not utilized for example as adsorbents,itwillleadtoawasteofexpensiveprecursors.

Solvent-freevaporphasesynthesis

AnotherinnovativeroutetoprepareZIFmembranesonporous supports isvia solvent-freecrystallization.A solvent-freevapor phasesynthesisisabletoproducehighfluxandhighselectivity membranes. To the best of our knowledge, the thinnest ZIF membranes(17nmthick) reportedtodatearepreparedusing thismethod.Themethodisfairlysimple,thereforeamenablefor scaleup.Inatypicalprocess,metalprecursorssuchasmetal-based gelandmetaloxidesaredepositedonsubstratesviadip-coating [130],electrodeposition[144],atomiclayerdeposition[79],etc.

The metal precursors are then exposed to a ligand vapor environment to convert the metal precursors into ZIFs. An important factor to consider in this process is obtaining high ligandvaporconcentrationswithoutsignificantdecompositionof the ligands [145]. Therefore, the ability of solid ligands to be vaporizedwithoutthembeingdecomposedmaylimitthechoiceof

Fig.7.VaporphasestrategytosynthesizeultrathinZIF-8andhybridZIFmembranesonporoussupports.(a)ZngellayeronPVDFhollowfiberistransformedintoZIF-8 membranesundermImligandvaporenvironment.(b)ZnOlayeronα-Al2O3discissubjectedtomImvaporenvironmenttoformZIF-8membranes.Subsequentvaporphase 2abImligandtreatmentledtoincorporationoftheligandsintoparentZIF-8frameworks.(c)ScanningelectronmicrographofvariousCo-basedZIFmembranespreparedvia vaporphasetransformation.ReproducedwithpermissionfromRef.[130]Copyright2017SpringerNature,Ref.[147]Copyright2019JohnWileyandSonsandRef.[146]

Copyright2018ElsevierB.V.

(9)

linkers, hence, ZIF materials that can be fabricatedinto mem- branes.Abeneficialaspectofsolvent-freecrystallizationistheuse oflargevolumeofsolventduringsynthesisiseliminatedwhich makesitanenvironmentallyfriendlymethod.

Lietal.[130]fabricatedultrathinZIF-8membranesontheshell sideofPVDFhollowfibersforC3H6/C3H8separationsusingagel- vaportransformation.AsshowninFig.7(a),thinlayersofZngel werecoatedontheshellsideofthehollowfibers.Thehollowfibers werethenexposedto2-methylimidazole(mIm)vaporat150Cfor severalhourstocompletethecrystallizationprocess.Regulating Zn gel concentration and coating time were critical to obtain continuous gel layer with different thicknesses. As a result, ultrathinZIF-8membraneswiththicknessesrangingfrom17nm to 757nm wereprepared. Nian et al. [146] on the otherhand synthesized Co-based ZIF (Co2(bIm)4, bIm - benzimidazole) membranesontheboresideofceramictubularsupportsforH2/ CO2separations.UnlikeLi’swork[130]mentionedabove,Nian’s workinvolvesanintermediatestepwherethemetalgellayerwas firstheattreatedtoconverttheCogeltoCo3O4beforeexposingit tobImvapor.Itisimportanttopointoutthattheuseofsolid-state metalprecursorstoprepareultrathinmembranesusingmetalgel transformationcanbequitechallengingduetohighviscosityofthe gel.

Ma et al. [79] introduced a liquid/gel-free ligand-induced permselectivation(LIPS)methodtopreparehighperformanceZIF- 8membranesforC3H6/C3H8separations.Atomiclayerdeposition (ALD)was usedtoobtainzinc oxide(ZnO)depositsontopand inside

g

-alumina layer. TheALDcycles wererepeated upto50 cyclestoformanimpermeableZnObarrieronthesupports.The ZnO layerwasthentransformedintoaselectiveandpermeable ZIF-8 barrier under mIm ligand vapor environment. C3H6/C3H8

separationperformancesofthemembraneswereamongthebest that have been reported for ZIF-8 and ZIF-67 membranes.

Following this, they performed vapor phase ligand exchange (VPLT)of2-aminobenzimidazole(2abIm)topartiallyexchangethe mImligandsofZIF-8[147].AsdepictedinFig.7(b),incorporation ofbulkier2abImligandsreducesZIF-8effectiveporesizesthereby shiftingthemolecularcut-offtowardssmallermolecules(e.g.,CO2, O2,N2,andCH4).TheemergenceofnewbroadIRpeaksat1280 cm1confirmstheincorporationof2abImligandintoVPLT-LIPS- ZIF-8membranes.Thevaporphaseconcepthasalsobeenextended toallowpermselectivitytuningofZIF-8membranes.Intheirmost recentwork,Hayashietal.[148]performedafacilevaporphase treatmentofZIF-8membraneswithmanganese(II)acetylaceto- nate (Mn(acac)2). The membrane selectivity dramatically in- creasedduetoapresence ofMn(acac)2depositsonthesurface ofthemembranes.

A key benefit of ALD is ability to deposit a uniform and conformalthinfilmonsubstrateswithangstrom levelprecision [149,150].WhileALDhasmanydesirablefeatures,ALDisaslow processwithlowdepositionrates.Forexample,thegrowthrateof ZnO onvarious substrates by ALD is in the range of 0.5–4.0Å cycle1[151].Moreover,ALDisanenergy-intensiveprocessthat generatesahighrateofwaste[149].Nonetheless,theproposedall- vapor ZIF-8 membrane synthesis above-mentioned is ground- breaking and has the potential to effectively suppress several challengesthatconventionalsolution-basedmembranesynthesis currentlyfacing.

Electrochemicalbasedsynthesis

Electrochemical deposition is one of the promising and industriallyrelevantmethodstoin-situformMOF/ZIFthinfilms and membranes. Zhang et al. [152] provided a comprehensive reviewofMOFdepositionviaelectrochemicaldeposition.Electro- chemicaldepositionofMOF/ZIFsonsubstratescanbecategorized

into three types: (i) electrophoretic deposition, (ii) anodic deposition,and(iii)cathodicdeposition[152].

Electrophoreticdepositiontechniqueusesanexternalelectric fieldtomobilizechargednucleipresent inthebulk solutionto substrates positionedat theoppositelychargedelectrode[152].

Thefluxofthenucleidriventothesubstratesisproportionaltothe electricfieldstrength,nuclei concentrationinthesolution,and electrophoretic mobility [153]. Thedeposited ZIF nuclei onthe substrates,ifnecessary,canbesubjectedtofurthercrystalgrowth topromotebetterfilmintergrowth.Heetal.[127]introduceda novelelectrophoreticnucleiassemblyforcrystallizationofhighly intergrownthinfilmstechnique(ENACT)toprepareultrathinand defect-free ZIF-7 and ZIF-8 membranes on AAO discs for H2

purificationsandC3H6/C3H8separations.InENACT,ZIFsolswere firstagedforafewminutes(>3min).Then,aconstantelectricfield was appliedfor up to 4min to mobilize the ZIF nuclei to the substrates.Thesubstrateswerethenleftinsidethesynthesissolat constanttemperaturefor2htopromotefurthercrystalsgrowth, therebyformingcontinuousandwell-intergrownmembranes.The reportedZIF-7and ZIF-8membrane thicknesses were2.7

m

m and0.5

m

m,respectively.SynthesisofZIFmembranesonvarious porous substrates including ceramic, copper foil, carbon, and polyacrylonitrileusingENACTwerealsoreported.Anadvantageof ENACTisthesubstratesdonotnecessarilyhavetobeconductiveor requireanysurface modificationstodeposit ZIFcrystals.ENACT alsoisafacileandrapidmethodwhichholdsmeritforscaleup.

Anodicdepositiontechniqueusesametalanodetosupplythe building block of ZIFs (metal ions) to the synthesis solution throughanodicdissolutionoftheanode[154,155].Themetalions theninteractwithlinkersintheelectrolytesolutiontoformZIF crystalsonsubstrates.Incathodicdeposition,bothmetalionsand linkersarepresentinthesolvent.Electrochemicalreductionofa probasenearthecathodicelectroderegiongeneratesbase,which goes onto increase local pHof the solution [156]. A high pH environmentatthevicinityof theelectrodefacilitatesdeproto- nationofneutrallinkers,inducingZIFnucleationandgrowthon thesubstrates [128].One drawbackassociatedwiththe above- mentionedtechniques(i.e.,anodicandcathodicdeposition)isthey both require conductive surfaces to attract metal ions to the cathodesideandtopromotedeprotonationoflinkers.Commonly usednon-conductingsubstratessuchasceramicorpolymerneed toberendered conductive(bycoating themwith a conductive layer)forthemethodtowork.

Tothebestofourrecollection,allZIFmembranes(i.e.,ZIF-7,ZIF- 7-8,ZIF-8,andCo–Zn–ZIF-8)forseparationapplicationsreported inliterature weresynthesizedviacathodicdeposition.Theonly workthatreportsonZIFthinfilmformationviaanodicdeposition wasbyWorralletal.[157].Worralletal.[157]depositedavariety ofZIFthinfilmsincludingZIF-4,ZIF-7,ZIF-8,ZIF-14,andZIF-67on Cu or Znfoil for supercapacitor applications.Zhou et al. [133]

developed fast current-driven synthesis (FCDS) that enabled formation of ultrathin ZIF-8membranes for C3H6/C3H8 separa- tions. The membranes were synthesized in-situ inside an electrochemicalcellcontainingmethanolicsolutionofZn(COOH), 2-methylimidazolate (mIm), and (NBu4)PF6 modulator. 200nm thickZIF-8membranesonAAOdiscswerepreparedinjustunder 20min.ZIF-8thinfilmsonstainlesssteelnets,Nifoam,andporous stainlesssteeldiscswerealsodemonstrated.Exposuretoexternal directcurrentgeneratedZIF-8membraneswithanewlydiscov- ered ZIF-8_Cm phase with more suppressed linker mobility compared to the normal ZIF-8_I43m phase. Stiffer ZIF-8_Cm frameworksofthesynthesizedmembranesresultedinasuperior C3H6/C3H8molecularsievingcapability.Intheirfollowupwork, theysynthesizedmixed-linkerZIF-7-8membranesonAAOdiscs usingasimilarFCDStechniqueshowingunprecedentedCO2/CH4

separation performances [94]. In this case, a combination of

(10)

frameworksstiffening andporenarrowingfromtheadditionof bulkysecondarybenzimidazole(bIm)ligandssharpenedCO2/CH4

sievingpropertiesofthemembrane.Mostrecently,theyfabricated a series ofbimetallic (Co100-x–Znx–ZIF-8) withthicknessof less than 700nm in just under 20min [46]. The AAO discs were immersedinasynthesissolution(amixtureofZnandCosaltin methanolwithatotalmetalconcentrationof0.1MandmImin methanol with a concentration of 0.2M) which were then subjectedtoacurrentdensityof0.7mAcm2atroomtemperature forseveralminutes.

Wei et al. [128] used aqueous cathodic deposition (ACD) methodtosynthesizeZIF-8membranesonAAOdiscsforC3H6/ C3H8separations. Themethod used100%waterassolventand did not require addition of modulators. In setup shown in Fig.8(a),aconductiveAAOdiscwasusedasacathodicelectrode whilegraphitepaperwasusedasananodicelectrode.TheAAO disc was coated with platinum/palladium via sputtering to enhancediscelectricalconductivity.Anoptimalcurrentdensity of 0.13mAcm2 was chosen to prevent excessive water electrolysisandtoobtainhighZIFcrystaldepositionrates.High mImtoZnratio(Zn:mImratioof1:60)anddiluteZnsolution(Zn:

H2O of1:3889) wereused toensure completelinker deproto- nation andto slow downcrystal growth in the bulksolution, respectively.Fig.8(b)showstheevolutionoffilmmorphologies from10minto60min.Longersynthesistimeresultedinthicker ZIF-8 membranes with larger grain sizes. Among ZIF-8 mem- branesthatweresynthesized,the500nmthickZIF-8membranes displayedthehighestC3H6/C3H8separationfactor,outperform- ing the majority of reported ZIF-8 membranes for C3H6/C3H8

separations.

Fromourobservation,ENACT,FCDS,andACDabletoproduce ZIF-8 membranes withthickness of less than 500nm. Despite havingthinmembranes,C3H6/C3H8selectivityofthemembranes arestillgreaterthan100,indicatingthatthemembranespossess bettermicrostructuresandfewerdefects.Shortsynthesistimeand gap-filling mechanism of the exposed substrates are the likely reasonbehindthis.ZIF-8isnon-conductive.Oncethecrystalsform andcoversupportconductivesurfaces,theyactasinsulatinglayers that preventsfurthercrystalsgrowth [133].Therewillbefewer crystalformationontopofthealreadyformedcrystalsand‘piling up’ of crystalsare suppressed, forming single monolayer thick membranes [158]. Moreover, rapid membrane synthesis time providesanarrowwindowforcrystalgrowth.

At this moment,electrochemical-based membrane synthesis onlyfocusesonmechanicallyweakAAOflatsubstrateswhichlimit thegaspermeationmeasurementpressureto2.0baronly[128].

For practical applications, it is beneficial to demonstrate the applicability of the FCDS and ACD togrow ZIF membranes on strongersubstratessuchasα-Al2O3discsormetalplateswhichare abletotolerate highertransmembranepressuredifference(e.g., typicalpressureforC3H6/C3H8separationisaround18atm)[159].

Also,thesubstratesneedtoberenderedconductivefortheFCDS andACDtechniquetoworkbutonehastoadmitthatitisdifficult to render nonconductive polymer or ceramic to be conductive [133]. Electrochemical-based synthesis have the potential to prepareZIFmembranesonhighsurface-to-volumesubstratesas membraneformationisbelievedtobelesssensitivetosubstrate geometry but there are no successful demonstration reported.

Electrochemicaldepositionmethoddefinitelyhasthepotentialto produce ultrathin and high performance ZIF membrane for commercialapplications,but furthersubstantial researcheffort isrequired.

ApplicationareasofZIFmembranes

Applications of membrane technology in industrial gas separationsaresomewhatlimitedwherethemajorityofinstalled membrane modules, mostly polymers, are used in only four commonapplications:(i)H2recovery–H2/N2,H2/CH4,etc.(ii)N2

productionfromair–N2/O2(iii)naturalgastreatment–H2S/CH4, CO2/CH4,etc.(iv)vaporrecovery–C2H4/N2,C3H6/N2,etc.[6,10].In gas separation membranes, finding candidate materials with proper transport properties is critical to ensure high product purityandrecoveryrate.ThemajorityofZIFs,especiallySOD-type ZIFshaveaperturesinthesizerangeofimportantgases,therefore could be used as membrane materials for gas separations.

Preparationof ZIF membranes for gasseparations hasbeen an activeresearchareainthepastdecades.Therehasbeenanumber ofexcellentreviewsonZIFpolycrystallinemembranesavailablein literaturetokeepthereaderupdatedwiththerecentprogressin thearea[49–54].

Inthissection,wereviewthreeunsolvedmembraneapplica- tionsthat,ifsortedout,mightdisruptthecurrentgasseparation marketthatmostlydominatedbyconventionaltechnologies(e.g., distillation, adsorption, and absorption). Unsolved application areasthatwillbediscussedinthissectionareH2purification(H2/ CO2),CO2separations(CO2/N2andCO2/CH4),andcondensablegas separations(C2H4/C2H6,C3H6/C3H8,andn-C4H10/i-C4H10).Inthis section,wefirstpresentbenchmarktransportpropertiesrequired forthemembranestobecompetitivewiththeexistingtechnolo- gies. Then, we review majority of the examples of ZIF-based membranesforgasseparationsthatarecurrentlyavailable.Thegas

Fig.8. (a)SynthesisofultrathinZIF-8membraneusingaqueouscathodicdeposition.(b)MorphologicalevolutionofZIF-8membranessynthesizedusingmethodshownin(a) at10min,20min,30min,40min,and60minsynthesistime.ReproducedwithpermissionfromRef.[128]Copyright2019JohnWileyandSons.

(11)

separationperformancesofthemembranesaresummarizedina formoftableorpseudo-Robesonplot(permeancevs.selectivity).

H2purifications

H2isconsideredasa clean,efficient,and sustainableenergy carrierwithzerogreenhousegasemissionandnoenvironmental damage(oxidizationofH2producesonlywatervapor).Highpurity H2isusedinseveralimportantindustrialapplicationsincluding petroleumrefining,aerospaceapplications,ammoniaproduction, glasspurification,semiconductormanufacturing,etc.[160].Like any other important gases, H2 coexists along with other components(e.g.,N2,CO,CO2,andCH4)duringchemicalprocesses, therebyrequiringitsseparation[9].Largecapitalinvestmentsare requiredfortheinstallationofconventionalseparationtechnology toisolateH2fromlessdesirablespecies,whichconsequentlydrives theoverallH2costup[161].Efficientpurificationandrecoveryof H2 requiresa state-of-the-artseparation technologyand mem- brane technology is expected to play a key role in a cheaper productionofH2.

From a historical perspective, the very first commercial polymer-based hollow fiber membranes for H2 separations, namelyPRISM1membranes,weredevelopedbyMonsanto(part of Air Products and Chemicals Inc.) in the late 1970s. PRISM1 membranes(polysulfone)wereoriginallydesignedtoseparateH2

fromN2fromammoniareactorpurgestreamsbutlaterexpanded for wider applicationsincludingH2/COsyngasratioadjustment and H2 recovery from hydrocarbonand/or hydrotreater off-gas streams [18]. Following the commercial success of PRISM1 membranes,celluloseacetatespiral-woundmembranes(Separex, partofHoneywellUOP)andpolyimidemembranes(UbeIndus- tries) were developed for similar applications [18,161–163].

Membrane-based gas separation has expanded since then and newmembranematerials weredevelopedforotherapplications suchasCO2separationfromnaturalgas.AccordingtoGaliziaetal.

[10], H2 separations (i.e., H2/N2, H2/CO, and H2/CH4) were considered as solved problem in gas separation membranes.

ConsideringthelargescalesuccessofpolymermembranesforH2

separations,there islittleinteresttodevelopmore H2-selective membranesfortheabove-mentionedapplications.

ZIF membranes have the opportunity to be utilized for emerging applicationsof H2 separationsparticularly forH2/CO2

separations.ThemostcommonroutesofproducingH2arethrough hydrocarbonreforming andcoal gasification,producing roughly 96% of the global H2 supply [164]. The remaining fraction is producedthroughwaterelectrolysis[161].TheproductionofH2 viasteammethanereformingbeginswithinitialreformingstep reactionsat820C(2)[161,165].AdditionalH2isobtainedthrough water-gas-shift(3)andsteammethanereformingreactions(4).

CH4þH2O!COþ3H2  ð2Þ

COþH2O!CO2þH2  ð3Þ

CH4þ2H2O!CO2þ4H2  ð4Þ

Theresultingmixturesconsistof74%H2 and18%CO2 which requires a CO2 removalprocess toobtainhighpurity H2 [161].

MembranetechnologyisnotwidelyexploredforH2/CO2separa- tion due to low gas selectivity. As compiled by Robeson, the majorityofreportedpolymermembraneshaveH2/CO2selectivity of less than10 [13].Galizia et al. [10] argue that therequired properties for commerciallyattractiveH2/CO2 separation mem- branes are H2 permeance greater than 200GPU and H2/CO2 selectivitygreaterthan10.Forthisapplication,polycrystallineZIF

membranesmighthavethepotential tobeutilized.Thereisno shortageofscientificworksonZIFmembraneswithpromisingH2/ CO2separationperformances.H2/CO2separationperformancesof therecently reportedZIF membranesare presented inTable 1.

Additionally,H2/N2 and H2/CH4 separationperformances of the membranesarealsoincluded.

ZIF-8 is one of the most widely investigated ZIFs for H2

separations.Generallyspeaking,polycrystallineZIF-8membranes arenoteffectivetoisolateH2fromCO2duetolatticeflexibility.As illustratedinFig.9,H2permeancethroughZIF-8membranesis unprecedentedly high, reaching a value of 24,641GPU. Gas permeances through the membranes decreases following the orderofmoleculekineticdiameters:H2(2.9Å)>CO2(3.3Å)>N2

(3.6Å)>CH4 (3.8Å). The reported H2/CO2 selectivity of the membrane is higher than the Knudsen selectivity (7.0 vs 4.7), butfallshortof thecommercialrequirements.Inanotherwork, ultrathin ZIF-8 membranes supported on PVDF hollow fibers preparedbyLietal.[130]showedhighH2permeanceof10,454 GPU but low H2/CO2 ideal selectivity of 7.3. Despite showing impressivelyhighH2permeancesintheorderofseveralthousand GPUs,H2/CO2selectivitiesofZIF-8membraneswereunattractive (seeTable1).H2/CO2selectivityofZIF-67(Co-substitutedZIF-8) membranesisslightlyhigherthanthoseofZIF-8membranesdue to stiffer Co–N bonding, but still not attractive enough for commercialapplications. At25C and 1bar feedpressure, ZIF- 67membranes byZhouet al.[166] displayedH2permeanceof 4933GPUandidealH2/CO2selectivityof17.

AmongZIFsthatareavailable,ZIF-7mightbebettersuitedfor H2/CO2separationsduetoitsnarrowerapertures(3.0Å)thanZIF- 8. ZIF-7is constructed bylinking benzimidazoles withZn ions forming a cubic SOD zeolite topology [37].Unlike ZIF-8 mem- branes that can be easily preparedusing simple solvents (e.g., methanol or water), ZIF-7 membranes are synthesized under solvothermal conditions using dimethylformamide (DMF) as a solvent[80,98,112,167].ZIF-7membraneactivationcanbetrickyas DMFremovalfromZIF-7cavitiesmayleadtophasetransitionfrom ZIF-7-Itoahighly-distortedandlocally-strainedZIF-7-IIorlayered and non-porousstructure ofZIF-7-III phase [168].As shownin Table1,themajorityofthereportedZIF-7membraneshavehigher H2/CO2 selectivity than those of ZIF-8. Increasing permeation temperature improves separation efficiency as H2 transport throughthemembranesislessaffectedbytemperaturecompared toCO2.KimandLee [169]observedH2/CO2separationfactorof their ZIF-7 membranes increased to 10 when measurement temperature was increased to 150C. Selectivity increase with temperatureisbeneficialconsideringthepotentialapplicationof themembranewhichishightemperatureseparationofH2from CO2outputofWGSreactor[170].

ZIF-95 with POZ topology has a narrow aperture of 3.7Å estimatedfromsingle-crystalstructuredataandhasstrongaffinity towardCO2duetoquadrupolarinteractionbetweenCO2andZIF- 95 linkers [36,89]. Strong adsorption between CO2 and ZIF-95 framework limits the CO2 diffusion mobility through the framework. Ma et al. [172] synthesized ZIF-95 membranes on porousα-Al2O3discsviaaseededgrowth.Themembranesshowed highH2permeanceof507GPUandH2/CO2selectivityof42.Intheir followupwork,theysynthesizedc-orientedZIF-95membranesby secondarily growing the ZIF-95 nano-sheet seed layers using vapor-assisted in-planeepitaxialgrowth [171].Gaspermeances throughthemembranedecreasedfollowingtheorderofmolecular kineticdiameterasdepictedinFig.10(a).At50Cand1bar,the secondarily grown ZIF-95 membranes displayed H2 single gas permeance and H2/CO2 ideal selectivity of 1633 GPU and 29, respectively.H2permeanceandH2/CO2idealselectivityincreased to2882 GPU and 39, respectively, upon increasing permeation temperatureto225CasshowninFig.10(b).Meanwhile,50

m

m
(12)

Table1

Singleand/ormixedgaspermeancesandselectivitiesofvariousZIFmembranesforH2purifications.

ZIFs Substrates Synthesismethod Thickness(mm) H2permeance(GPU) Selectivity Testconditions Ref.

H2/CO2 H2/N2 H2/CH4

ZIF-7 PVDFhollowfiber Solvothermal synthesis

30 3,969 16.3 18.3 RT [98]

1bar Singlegas ZIF-7 Polysulfonehollow

fiber

Microfluidic synthesis

2.4 6 2.4 35.1 34.6 35C [142]

TMP:0–3bar Mixedgas(1:1) ZIF-7 Polypropyleneflat

sheet

Chelationassisted in-sit

Gambar

Fig. 3 summarizes important criteria for what we believe to be a scalable and economically attractive ZIF membrane concept.
Fig. 4. Comparison between (a) in-situ growth and (b) secondary seeded growth for ZIF membrane synthesis
Fig. 5 shows important milestones in the development of ultrathin ZIF membranes made in the past decade
Fig. 6. Interfacial synthesis strategy to prepare (a) all nanoporous MFI/ZIF-8 MMMs on Torlon 1 hollow fibers (b) ZIF-90 membranes on macroporous carbon hollow fibers.
+6

Referensi

Dokumen terkait