Stepwise Synthesis of Metal−Organic FrameworksMathieu Bosch, Shuai Yuan, William Rutledge, and Hong-Cai Zhou*

Department of Chemistry, Texas A&M University, College Station, Texas 77843, United States

CONSPECTUS: Metal−organic frameworks (MOFs) are acategory of porous materials that offer unparalleled controlover their surface areas (demonstrated as higher than for anyother material), pore characteristics, and functionalization.This allows them to be customized for exceptional perform-ance in a wide variety of applications, most commonlyincluding gas storage and separation, drug delivery, lumines-cence, or heterogeneous catalysis. In order to optimizebiomimicry, controlled separations and storage of smallmolecules, and detailed testing of structure−property relation-ships, one major goal of MOF research is “rational design” or“pore engineering”, or precise control of the placement ofmultiple functional groups in pores of chosen sizes and shapes.MOF crystal growth can be controlled through judicious design of stepwise synthetic routes, which can also allowfunctionalization of MOFs in ways that were previously synthetically inaccessible. Organic chemists have developed a library ofpowerful techniques over the last century, allowing the total synthesis and detailed customization of complex molecules.Our hypothesis is that total synthesis is also possible for customized porous materials, through the development of similarmultistep techniques. This will enable the rational design of MOFs, which is a major goal of many researchers in the field. Wehave begun developing a library of stepwise synthetic techniques for MOFs, allowing the synthesis of ultrastable MOFs withmultiple crystallographically ordered and customizable functional groups at controlled locations within the pores. In order todesign MOFs with precise control over pore size and shape, stability, and the placement of multiple different functional groupswithin the pores at tunable distances from one another, we have concentrated on methods which allow us to circumvent the lackof control inherent to one-pot MOF crystallization.Kinetically tuned dimensional augmentation (KTDA) is an approach using preformed metal clusters as starting materials andmonotopic carboxylates as equilibrium shifting agents to make single crystals of ultrastable MOFs. Postsynthetic metathesis andoxidation (PSMO) takes advantage of the fast ligand exchange rate of a metal ion at the low oxidation state as well as the kineticinertness of the same metal at high oxidation state to make ultrastable and highly crystalline MOFs. Multiple similar strategieshave been successful for the metathesis of Fe-based MOFs to Cr3+. Several highly crystalline Ti-MOFs have also been prepared.Kinetically controlled linker installation and cluster metalation methods utilize a stable MOF with inherent coordinativelyunsaturated sites as matrix and postsynthetically install linkers or grow clusters on the matrix, so that a robust MOF withprecisely placed functionalities is realized. This method has diverse applications especially when specific functional groups ormetals having synergistic effects are desired in the proper proximity.Exceptional porosity and stability are required for many potential applications. We have demonstrated several of these, includingentrapment of nanoscaled functional moieties such as enzymes. We have developed a series of metal−organic frameworks (PCN-333) with rationally designed ultralarge mesoporous cages as single-molecule traps for enzyme encapsulation. We successfullyincorporated metalloporphyrins, well-known biofunctional moieties, into robust MOFs for biomimetic catalytic applications. Byrationally tuning the synthetic conditions, we obtained several different porphyrinic Zr-, Fe-, and Ti-MOFs with distinct pore sizeand concentrated acid or base stability, which offer eligible candidates for different applications. These and other stepwise kinetictuning and catalyst incorporation methods are small steps toward achieving the grand challenge of detailed control of theplacement of matter on an atomic and molecular level.

1. INTRODUCTION

One-pot metal−organic framework (MOF) syntheses are oftendeveloped through trial and error. The Yaghi group hasdescribed the customization of MOF structures through the“reticular approach” of functionalizing and lengthening thelinkers of known structures, and mixed-ligand MOFs are often“multivariate”, or distributed randomly. This allows custom-ization of the pore sizes of known structures, and the placement

of many functional groups within them, but it can also makedetailed control of the shapes of the pores and the placement ofthe functional groups within the pores difficult.1,2 Many earlyMOFs made from divalent metals show exceptional porosityand promise for a wide variety of applications, but some proved

Received: December 5, 2016Published: March 28, 2017

Article

pubs.acs.org/accounts

© 2017 American Chemical Society 857 DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

unsuitable for many applications because of their lack of long-term stability to ambient moisture.3 Therefore, researchers suchas Ferey et al. and Lillerud et al. developed stable MOFs basedon trivalent and tetravalent metals (M3+ and M4+) which tendto possess much greater chemical stability due to the decreasedlability of the M−O bonds.4,5 The exceptional stability of M3+

and M4+ based MOFs have made them feasible for use in wide-scale practical applications. However, the stability of the high-valent metal−ligand bonds also causes challenges in MOFcrystal growth and functionalization.In order to produce MOFs that are chemically robust, a

prerequisite for most applications, a few basic approaches havebeen developed over the last several years, which can berationalized by the application of hard−soft acid−base theory.MOFs based on copper and zinc with carboxylate ligands areboth easy to crystallize as large single crystals and arechemically unstable for the same reason: their metal−ligandbonds are relatively labile. Those bonds can be made less labile,increasing MOF chemical stability, by using softer, bettermatching ligating moieties such as those based on nitrogenheterocycles, or by using carboxylate linkers modified to protectthe SBU or increase bond strength, or by using harder higher-valence metals that naturally have less labile bonds withrelatively hard carboxylates.6 The first two approaches oftenrequire complicated or expensive organic synthesis that limitsthe large-scale synthesis of the resulting MOFs, so our grouphas focused mostly on the use of high-valence metals. However,this often results either in noncrystalline, nonporousamorphous products, or MOFs that could not be grown assingle crystals for characterization, with other characterizationmethods taking far more time and effort overall. To overcomethis difficulty, we have in the past few years developed several

MOF synthesis, crystallization, and postsynthetic modification(PSM) methods, building on years of advancements in thesetechniques by a large number of research groups.7−9

When high-valence metal salts such as FeCl3 andpolycarboxylate linkers are mixed under traditional solvother-mal conditions, the carboxylate linkers quickly bind to the ironto form noncrystalline, amorphous metal−ligand coordinationpolymers. “Modulated synthesis” using a monocarboxylic acidwas developed by the Kitagawa group to alter crystal growthand morphology, and was used by the Behrens group to growsingle-crystals of high-valence Zr-MOFs that had previouslyonly been characterized using nanocrystalline powder.8,9 Thismethod slows nucleation both through lowering solutionpH(slowing and reducing ligand deprotonation) and throughcompetitive ligation of the metals in solution. Depending onthe metal and ligand used, the exact concentration ofmodulating reagent can be increased over several trial reactionsto allow the synthesis of amorphous products, to polycrystallinepowders, typically in high yield, to large single crystals of thesame MOF in lower yield, and finally to clear solutionsproducing no products at a high concentration of acid.10 Thesemethods allow the synthesis of high-valence, chemically robustMOFs with tunable crystallinity and yield. However, they stilltypically require the use of acidic solutions at temperaturesexceeding 100 °C, which limits direct one-pot incorporation ofsome functional groups. Our research has focused on enablingthe rational design of MOFs through finding methods toovercome these and similar limitations. By using these stepwisepre- and postsynthetic modification methods, multiple func-tional groups can be placed at controlled positions within thestructure in order to customize materials for biomimicry, gasseparation, catalysis, and other new applications.

Figure 1. (a) Four different connecting modes of the [Fe2M(μ3-O)] cluster. Carboxylates on ligands and terminal acetates are represented by blackand purple, respectively. (b) Thirty different ligands and two types of mixed ligands used in constructing Fe-MOFs.10 Reprinted by permission fromMacmillan Publishers Ltd.: Nature Communications copyright 2014.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

858

2. KINETICALLY TUNED DIMENSIONALAUGMENTATION

While the modulated synthesis method alone seemed to besufficient for the synthesis of single-crystals of a variety of Zr-MOFs, Fe- and Ti-MOFs remained difficult to synthesize orcharacterize.Growing single crystals, as opposed to powders, isstill difficult in these systems. This is likely because crystalnucleation is too fast and widespread, creating many micro-crystalline powder particles which do not have the opportunityto continue to grow into single crystals. It may also beinfluenced by the lesser reversibility in the metal−ligandbond.The modulating acid concentration zone betweenpolycrystalline powders and clear solutions was small ornonexistent. At a very high modulating reagent concentration,starting materials such as solvated metal salts may be favored, asthe pH would be too low for a resulting MOF framework to bestable. Soluble metal-carboxylate clusters may also be favoredwhen the concentration of nonlinking monocarboxylates is toomuch higher than that of linkers. Inspired by concepts like“scale chemistry” introduced by Ferey,11 we decided not to usehighly soluble metal salts as a starting material. Instead, westarted from soluble metal-acetate preformed clusters, [Fe2M-(μ3O) (CH3COO)6] (M = Fe2+/3+, Co2+, Ni2+, Mn2+, Zn2+).10

This alters the equilibria in favor of slowing nucleation andallowing larger crystal growth in several ways. First, the lesslabile cluster-carboxylate bond is far slower to be displaced byan energetically very similar linking-carboxylate bond. Entropiceffects, with the displacement into solution of multiplemonocarboxylates per linker, play a major role in thecrystallization, and the amount of modulating acetic acid canbe tuned to slow linker ligation and framework growth evenfurther. This was validated experimentally with the synthesisand single-crystal X-ray diffraction (SCXRD) characterizationof 34 different Fe-MOFs based on this cluster and 30 distinctligands (Figure 1).The trends observed in this work provide key insight into

MOF kinetics and crystallization equilibria. We found thathigher-connected ligands of similar size always required moreacetic acid to produce large single crystals, with in some cases10 times higher concentrations of acetic acid required to getsingle crystals of a MOF with the same SBU, when using atetracarboxylate linker instead of a dicarboxylate linker. Thisimplies that the larger coordination number of the linker,displacing more acetate from the starting clusters, was a highlyimportant factor in the MOF crystallization, though higherconnectivity may increase the thermodynamic stability of theframework as well. When starting from carboxylate clustersinstead of metal salts such as FeCl3 less acetic acid or othernucleation-disfavoring changes were required to grow singlecrystals. Starting from a preformed cluster allows the chemist toincorporate other metals such as Co2+ into the (μ3O) clusterand thus tune the SBU for other properties, and acts as a leverfor crystallization control.However, the preassembled inorganic clusters are not always

preserved during the MOF formation process. The clustershould be kinetically stable under the synthetic conditions used.We aimed at using the KTDA method for the synthesis of aporphyrinic titanium MOF. A preformed Ti6O6(O

iPr)6(abz)6(abz = 4-aminobenzoate) cluster was selected as startingmaterial which is expected to form a she topology whencombined with tetratopic porphyrinic linker (TCPP). Startingfrom a preformed Ti6O6 cluster, a single crystalline porphyrinic

titanium MOF, PCN-22, was obtained. A large excess ofbenzoic acid was required to produce single crystals, as istypical for tetracarboxylate linkers with high-valence metals. Asfar as we know, this was the first report of a single-crystallineTi-carboxylate MOF and one of very few Ti-MOFs everreported.12 However, the Ti6O6 cluster was transformed into aTi7O6 cluster in PCN-22 (Figure 2). This shows how a robust

cluster under the MOF synthetic conditions is required for theKTDA method, but as starting from a titanium salt did notproduce the MOF products, it shows that even attemptedKTDA can be a useful kinetic modulation method. This issimilar to other examples found by our group where attemptedKTDA resulted in new MOFs with different SBUs than thatwhich were added, under conditions where the attemptedoriginal SBU was not completely stable.13

It is worth noting that use of different amounts and types ofmodulating reagents can produce effects both during and aftercrystallization. For example, lower pKa modulators such astrifluoroacetic acid have been reported to increase the Lewisacidity, pore volume and/or surface area, and proportion ofmesoporous defects in UiO-66-type Zr MOFs when comparedto use of higher temperatures and higher pKa modulators,which can produce UiO-66 with few defects.14,15 Some MOFscan even be synthesized under basic conditions in order toinfluence their porosity.16 It is possible to tune the properties ofUiO-66 and many other MOFs by using different modulatingreagents of different solubility and pKa. This should be triedwhen attempting to grow single crystals, as each solvothermalsystem only crystallizes in a certain range of modulatorconcentration. The pKa of the modulating reagent is anotherkinetic parameter that we can tune as part of our crystallizationtoolbox. However, certain systems and metals, such as Cr(III)-based MOFs which possess even less labile metal−ligand bondsthan Fe(III) or Zr(IV)-based MOFs, are still difficult tocrystallize directly, even with these tools. Furthermore, thismethod does not directly allow the rational design of new

Figure 2. (a) Ti6O6(OiPr)6(abz)6 and TCPP as starting materials; (b)

cluster and structure of PCN-22; (c) cluster and structure of predictedMOF with she topology.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

859

MOFs. However, KTDA is still a useful part of the synthetictoolbox, even if it does not directly provide structural control.This is because the empirical “kinetic control” methodsdeveloped as part of KTDA are used in the later structuralcontrol methods. In methods such as postsyntheic metathesisand oxidation or sequential linker installation, a MOF with aknown structure is modified using the kinetic methodsdeveloped in KTDA.

3. POSTSYNTHETIC METATHESIS AND OXIDATIONSome extremely stable high-valence MOFs such as Ti- and Cr-based MOFs are extremely rare, due to the aforementioneddifficulties in their synthesis and characterization. Building on alarge variety of postsynthetic modification techniques devel-oped by many researchers in the MOF field,7 our group hasdeveloped a variety of approaches based on metal metathesis inorder to produce MOFs based on these metals, which thendemonstrate much higher chemical stability than theisoreticular parent frameworks they are based on. Postsyntheticmetathesis and oxidation (PSMO) refers to the synthesis of atargeted MOF based on metals such as Sc3+, Zn2+, or Mg2+. Theuse of scandium3+ here, along with example such as Zr4+ MOFs,is actually a useful example of how “high-valence” and “low-valence” MOF is a shorthand that is too simplistic to becompletely accurate. Sc3+-based MOFs, as d0 metals, actuallytend to be less or similarly chemically stable than, for example,d9 Cu2+-based MOFs, despite their increased valence, possiblybecause ligand exchange is not energetically destabilized by theligand field stabilization energy (LFSE) of the metal. Similarly,d0 Zr4+ and Ti4+ MOFs tend to be less stable than d3 Cr3+

MOFs. There are also many other factors affecting stability,such as SBU and ligand connectivity, porosity, and thefunctionalization and ligating moieties of the linker.Because the original MOFs are made with labile metals, an

excess of less labile metal in solution after synthesis should beentropically driven to exchange. The speed and completenessof this metal metathesis depends on both the metalconcentration and the relative thermodynamic stability ofeach metal in the MOF SBU’s coordination environment.Building on previous work by the Dinca and Cohen groups,17,18

we reasoned that care must be taken to select MOFs with SBUsthat the new metal will be stable in. Furthermore, directexchange of low-valence to low-valence metals, while fast, didnot produce an increase in MOF chemical stability, while directexchange from low to high-valence metals was incomplete andslow. One of the solutions was to select MOFs with SBUs andcoordination environments that were highly stable with the newmetal, exchange for a lower-valence, unstable ion of that metalunder an anoxic environment, and then allow the air to oxidizethe metal to its stable higher oxidation state. This wasdemonstrated in PCN-426, where Mg(μ3-O) clusters serve asan SBU which can be easily exchanged for Fe(II) and Cr(II),which remain labile in those oxidation states. After oxidation,this process produced a MOF which was to our knowledge thefirst example of a Cr-MOF of sufficient crystal size and qualityfor characterization on a nonsynchrotron single-crystaldiffractometer.19 This is an example of how even materialsthat cannot be directly crystallized can be synthesized andcharacterized through stepwise methods that change environ-mental conditions to “climb” over an energetic barrier into alocal energy minima.PSMO was also applied to the synthesis of multiple titanium

frameworks based on appropriate SBUs. MIL-100(Sc), PCN-

333(Sc), MOF-74(Zn), and MOF-74(Mg) were all quicklyexchanged with Ti(III), (Figure 3) which was then oxidized in

air to Ti(IV) with varying rates of metathesis ranging from 35%in MOF-74(Mg−Ti) to 100% in MOF-74(Zn−Ti), combinedwith promising photocatalytic activity measurements.20

We soon worked to extend this process in order to stabilizeand functionalize a framework, PCN-333, that was useful due toits extremely large 5.5 nm mesopores, but had relatively highchemical stability, but slightly lower than other MOFssynthesized with the same SBU, presumably due to thosemesopores.21,22 Modulated synthesis using 4,4′,4″-s-triazine-2,4,6-triyl-tribenzoate (H3TATB) with MCl3 (M = Al, Fe, V,Sc, In) and trifluoroacetic acid produced frameworks whichwere isoreticular with MIL-100.22 Due to the many-sidedshapes of the larger pores in their MTN topology, isoreticularexpansion produces greater pore size increases than it does inframeworks with pores that have fewer “sides” or windows.These large pores proved useful in the encapsulation andprotection of several enzymes, which retained catalytic activitywithin the pores under conditions that unprotected enzymeswere denatured under. The chemical stability of this framework,remaining intact under conditions of pH 3−9, is high for such ahighly porous material, but needed to be improved for certaincatalytic applications requiring more concentrated acid or basicsolutions.One metal produces MOFs which typically have higher

chemical stability than that of Fe(III)−Cr(III). However,growth of PCN-333(Fe) already pushed the limits of our earlierkinetic tuning and crystallization methods, with only powdersynthesized, which had to be characterized using synchrotronpowder X-ray diffraction methods. Attempts to directlysynthesize PCN-333(Cr) produced only amorphous, non-porous products. Furthermore, directly synthesized Cr-MOFssuch as MIL-101 have proven highly difficult to functionalize, asthe high-temperature, high-pressure conditions necessary forthe synthesis of even powders of these MOFs decompose manyfunctional groups that have been used in other MOFs.23 Bothof these problems were solved simultaneously by the “dualexchange” of PCN-333(Fe) to PCN-333-N3(Cr). Postsyntheticligand exchange of PCN-333(Fe) allowed the functionalizationof the framework with moieties that are too fragile to survivesolvothermal synthesis, or that formed another framework

Figure 3. Stepwise PSMO synthetic method allows high-valencemetals to be used in a MOF structure that was not found through one-pot methods. Ref 20. Published by The Royal Society of Chemistry.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

860

structure when direct synthesis with a similar linker wasattempted (Figure 4).21,24 Subsequent metal metathesis allowed

such functionalities into the Cr-MOF, which was too kineticallyinert for successful postsynthetic ligand exchange. Interestingly,PCN-333(Sc) showed only 64.8% exchange for Cr underconditions that produced a 99.8% exchange ratio when startingfrom PCN-333(Fe) at 150 °C over two 30 min incubations inCrCl3 DMF solution, due to framework collapse when startingfrom PCN-333(Sc); it is not stable enough to survive theexchange conditions, while PCN-333(Fe) was. Unlike in thecase of PCN-426, replacing Fe(III) with Cr(III) via metathesisof Fe(II) to Cr(II) followed by oxidation was not necessary,due to the more similar ligand exchange rate constants ofFe(III) and Cr(III).Another method that takes advantage of redox changes to

control the metathesis reaction, “reductive labilization,” wasdemonstrated through an alternative preparation of PCN-333(Cr). In this case, anoxic Cr(II) solutions were used for anouter-sphere reduction of Fe(III) in PCN-333(Fe) to Fe(II),which quickly exchanged with the Cr(III) that was produced bythis reaction. PCN-333(Cr) was also shown to be much morechemically robust than PCN-333(Fe), showing stability over 24h in solutions of pH 0−11, and surviving uptake of alkylamines,used to increase the CO2 uptake of the material throughchemisorption, while PCN-333(Fe) was degraded by thealkylamine molecules.

4. SEQUENTIAL LINKER INSTALLATION ANDCLUSTER METALATION

MOFs have been very widely studied due to their combinationof ease of crystalline synthesis, high surface area, easyfunctionalization, and high chemical stability. They arepromising for many applications due to these properties,especially as they have been demonstrated to be functionalizedsimultaneously with up to 8 different moieties in controlledratios, as in multivariate MOFs.25 One limitation has been thatplacement of multiple functional groups at controlled positions,important for applications such as biomimetic catalysis, hasbeen limited, though some postsynthetic modificationapproaches, such as those recently published by the Yaghi

group, have been promising.26 The synthesis of mixed-ligandMOFs offers another solution to this limitation, as each ligandcan be functionalized separately, and the distance andorientation of the functional groups from each other will beknown and controlled by the MOF structure. Mixed ligandMOFs are rare, have had limited chemical stability, and havemostly been confined to two types of linkers, with mixed phasesor domains forming when more symmetrically distinct linkersare mixed in one-pot reactions.27,28 In our experiments, one-potsolvothermal synthesis using linear linkers of different lengthalways produced a mixture of different MOF phases, and so akinetically controlled, stepwise approach was used to produceseveral new MOFs through single-crystal to single-crystaltransformations.To overcome these difficulties, we targeted the Zr6 SBU,

which possesses tunable connectivity, from 6 to 12 connectionsdepending on the linker used in the MOF. In the UiO-66 typestructure, otherwise known as the fcu topology, thecarboxylates on the linear linkers must be coplanar in orderfor the Zr6 SBUs to form 12-connected nodes. By using a linkerthat twists at a 90° angle, i.e., dimethyl biphenyl dicarboxylate,instead of the unhindered, nonfunctionalized biphenyldicarboxylate of UiO-67, another structure, PCN-700, isproduced at lower temperatures (Figure 5).29 The bcutopology of PCN-700 is similar to fcu, except that 4 equatoriallinkers on the SBU are missing, producing an 8-connectednode. (Figure 6).The missing coordination areas on the Zr6 SBU in PCN-700

consist of terminal oxygens that can be, in turn, displaced by a

Figure 4. Stepwise synthesis was used to functionalize PCN-333(Cr).Reprinted with permission from ref 21. Copyright 2015 AmericanChemical Society.

Figure 5. Temperature control in the synthesis of PCN-700, which hasunsaturated SBUs suitable for later linker installation: (a) Linkerconformation in fcu nets; (b) linker conformation in bcu nets; (c)UiO-67 and (d) PCN-700 structures viewed along a-axis; cages of (e)UiO-67 and (f) PCN-700 structures. Reprinted with permission fromref 29. Copyright 2015 American Chemical Society.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

861

postsynthetic ligation reaction. Two of the sites have differentlengths than the other two, which allows the stepwiseplacement of two different types of linear linkers, which canbe separately functionalized (Figure 6). Furthermore, all thistakes place in a crystallographically ordered fashion, and so theprecise distances of each of the three possible linkers and anyfunctional groups attached to them can be measured, which wasdemonstrated through the synthesis and crystallographiccharacterization of PCN-701, -702, -703, and -704, withdifferent ordered linkers. PCN-700 shows a high degree offlexibility, which allows the installation of linkers with differentlengths and a combination thereof.30 Guided by geometricalanalysis, 11 new MOFs were obtained by linker installation,each bearing up to three different functional groups inpredefined positions (Figure 7). Systematic variation of thepore volume and decoration of the pore environment wasrealized by linker installation, which resulted in synergisticeffects including an enhancement of H2 adsorption capacities ofup to 57%. In addition, a size-selective catalytic system foraerobic alcohol oxidation reaction was built in PCN-700through linker installation, which shows high activity and

tunable size-selectivity. Altogether, these results exemplify thecapability of the linker installation method in pore environmentengineering of stable MOFs with multiple functional groups,giving an unparalleled level of control.31

In addition to multifunctional mixed-linker MOFs beingmade synthetically accessible through the control afforded bystepwise synthesis, mixed-metal Zr MOFs have also beenproduced. PCN-700s terminal oxygens on its equatorial planeare postsynthetically solvothermally reactive with other metals,which over time grow onto the Zr6 SBUs, forming decanuclearZr6M4 (M = Ni, Co) clusters, which can be shown stepwisewith successive SCXRD “snapshots” if the reaction isinterrupted every several hours (Figure 8).32

5. CONCLUSIONSSince the late 1990s, research interest in MOFs has beengrowing continuously. In a recent review article, Yaghi et al.proposed that the new materials should have multiple kinds ofbuilding units, and the arrangement of these building unitswithin crystals should have specific sequences. Advancedfunctionalities in MOFs require more complex structures and

Figure 6. (a) Chemical equation and (b) kinetic control of the stepwise linker installation process.

Figure 7. Single crystal structures of 11 geometrically predicted MOFs resulted from linker installation of PCN-700. Reprinted with permissionsfrom ref 31. Copyright 2016 American Chemical Society.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

862

pore environments, accompanied by heightened challengeswith their geometric design. Constructing metal−organicframeworks from multiple components is a pathway to enablesophisticated applications which require high complexity inhighly ordered crystalline materials. The above-discussedresearch provides possible methods to increase MOF complex-ity while maintaining structural regularity to the maximumextent. We have begun developing a library of stepwisesynthetic techniques for MOFs, allowing the synthesis ofultrastable MOFs with multiple crystallographically ordered andcustomizable functional groups at controlled locations withinthe pores. In the long term, we expect to realize total synthesisin MOFs for customized structures and functionalities, throughthe development of stepwise synthetic techniques. Althoughcurrently not possible, the high designability and diversesynthetic methodologies of MOFs discovered in recent yearssuggest that such a feat is feasible in the future, and representsone path toward “rational design” of MOFs that is a major goalwithin the field.The kinetic tuning, crystallization, and postsynthetic

modification methods described here are part of a growingsynthetic toolbox for MOF researchers. Each of these methodscan be used to achieve the synthesis and characterization ofnew, targeted MOFs that would not be possible in a simpleone-pot solvothermal crystallization process. Eventually, wehope to help realize the dream of rational, stepwise design ofmetal−organic frameworks with precise control over pore sizeand shape, stability, and the placement of multiple differentfunctional groups within the pores at tunable distances fromone another. This precise control will be necessary to producematerials optimized for energy storage, catalysis, or otherapplications that can benefit from the engineering of poreenvironments.We have developed and demonstrated the KTDA, PSMO,

sequential linker installation, structure-assisted functionalanchor implantation, single-enzyme encapsulation, and othermethods for this purpose. Each of these methods, more thansimply allowing the synthesis of the particular MOFs publishedwith them, can be used to synthesize a limitless variety ofMOFs customized for many different applications. Thestepwise methods we have described allow researchers to

overcome the harsh conditions necessary for MOF crystal-lization and the randomness of functional group placement thathave previously limited biomimicry.Small molecule separations are governed by selecting a

separation material that has pore size, shape, and functionaliza-tion that select very preferentially between molecules that maybe very similar to one another, such as racemic mixtures ordifferent isomers of a molecule. Stepwise synthesis offers newopportunities to design MOFs for this purpose. Smallmolecular characterization can also be undertaken in this way,for example through the crystalline sponge or coordinativealignment methods.33,34 Using the PSMO and sequential linkerinstallation methods to produce materials customized forpreferential interaction by multiple aligned functional groupswith small molecules of interest will be a rich avenue of futureinvestigation. The KTDA method can be used for isoreticularmanipulation of the SBU of known MOFs, and modulation oftheir surface area and catalytic activity, as well as synthesis ofMOFs not synthetically accessible through in situ SBUassembly. Immobilization of enzymes and homogeneousmolecular catalysts can be used to produce robust, reusableheterogeneous catalysts. Finally, development of new stepwisesynthesis and modification techniques is vital to allow moreprecise pore engineering and to work toward the rational designof MOFs.

■ AUTHOR INFORMATIONCorresponding Author

*E-mail: zhou@chem.tamu.edu. Web site: http://www.chem.tamu.edu/rgroup/zhou/.

ORCID

Hong-Cai Zhou: 0000-0002-9029-3788Notes

The authors declare no competing financial interest.

Biographies

Mathieu Bosch graduated from the University of Houston in 2013with a B.S. in chemistry, and immediately thereafter joined Dr. Hong-Cai Zhou’s group as a Ph.D. student at Texas A&M University. His

Figure 8. Cluster transformation during the cluster metalation and ligand migration; the yellow arrows illustrate the metal migration, and greenarrows the ligand migrations.32

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

863

research focuses on pore engineering in MOFs and porous polymers,and the development of MOF synthesis and postsyntheticmodification techniques.

Shuai Yuan graduated from Shandong University in 2013, where hestudied the design and synthesis of luminescent MOFs. He became aPh.D. student in Dr. Hong-Cai Zhou’s group at Texas A&MUniversity in 2013, and he focuses his research on the design andsynthesis of metal−organic frameworks for gas storage, gas separationand catalysis.

William Rutledge graduated from Millersville University in 2014, afterresearching iridium N-heterocyclic carbene-based transfer hydro-genation catalysts, and became a member of Dr. Hong-Cai Zhou’sgroup at Texas A&M University as a Ph.D. student in 2015. Hisresearch is on the synthesis of new MOF materials specialized towardsclean energy technologies.

Hong-Cai “Joe” Zhou graduated from Texas A&M University with hisPh.D. in 2000 under the supervision of Prof. F. A. Cotton. He thencompleted postdoctoral research with Prof. R. H. Holm at HarvardUniversity in 2002, and became a faculty member at Miami University,Oxford. In 2008, he became a full professor at Texas A&M University,a Davidson Professor of Science in 2014, and a Robert A. Welch Chairin Chemistry in 2015. His research focuses on synthetic methods toobtain extremely stable framework and other porous materials withinteresting and useful catalytic or clean-energy related properties.

■ ACKNOWLEDGMENTSM.B. acknowledges the Texas Space Grant and the Texas A&MGraduate Merit Fellowship. S.Y. acknowledges the Texas A&MEnergy Institute Graduate Fellowship funded by ConocoPhil-lips and Dow Chemical Graduate Fellowship. W.R. is supportedby the Center for Gas Separations Relevant to Clean EnergyTechnologies, an Energy Frontier Research Center funded bythe U.S. Department of Energy, Office of Science, Office ofBasic Energy Sciences under Award Number DE-SC0001015.M.B. and H.-C.Z. were supported by the Welch Foundationthrough a Robert A. Welch chair in Chemistry (A-0030).

■ REFERENCES(1) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keeffe,M.; Yaghi, O. Systematic design of pore size and functionality inisoreticular MOFs and their application in methane storage. Science2002, 295, 469−472.(2) Furukawa, H.; Cordova, K. E.; O’Keeffe, M.; Yaghi, O. M. TheChemistry and Applications of Metal-Organic Frameworks. Science2013, 341, 1230444.(3) Zhou, H. C.; Long, J. R.; Yaghi, O. M. Introduction to Metal-Organic Frameworks. Chem. Rev. 2012, 112, 673−674.(4) Serre, C.; Millange, F.; Surble, S.; Ferey, G. A Route to theSynthesis of Trivalent Transition-Metal Porous Carboxylates withTrimeric Secondary Building Units. Angew. Chem., Int. Ed. 2004, 43,6285−6289.(5) Cavka, J. H.; Jakobsen, S.; Olsbye, U.; Guillou, N.; Lamberti, C.;Bordiga, S.; Lillerud, K. P. A New Zirconium Inorganic Building BrickForming Metal Organic Frameworks with Exceptional Stability. J. Am.Chem. Soc. 2008, 130, 13850−13851.(6) Bosch, M.; Zhang, M.; Zhou, H.-C. Increasing the Stability ofMetal−Organic Frameworks. Adv. Chem. 2014, 2014, 182327.(7) Tanabe, K. K.; Cohen, S. M. Postsynthetic modification of metal-organic frameworks-a progress report. Chem. Soc. Rev. 2011, 40, 498−519.(8) Schaate, A.; Roy, P.; Godt, A.; Lippke, J.; Waltz, F.; Wiebcke, M.;Behrens, P. Modulated Synthesis of Zr-Based Metal−OrganicFrameworks: From Nano to Single Crystals. Chem. - Eur. J. 2011,17, 6643−6651.

(9) Tsuruoka, T.; Furukawa, S.; Takashima, Y.; Yoshida, K.; Isoda, S.;Kitagawa, S. Nanoporous Nanorods Fabricated by CoordinationModulation and Oriented Attachment Growth. Angew. Chem., Int. Ed.2009, 48, 4739−4743.(10) Feng, D. W.; Wang, K. C.; Wei, Z. W.; Chen, Y. P.; Simon, C.M.; Arvapally, R. K.; Martin, R. L.; Bosch, M.; Liu, T. F.; Fordham, S.;Yuan, D. Q.; Omary, M. A.; Haranczyk, M.; Smit, B.; Zhou, H. C.Kinetically tuned dimensional augmentation as a versatile syntheticroute towards robust metal-organic frameworks. Nat. Commun. 2014,5, 5723.(11) Ferey, G. Building Units Design and Scale Chemistry. J. SolidState Chem. 2000, 152, 37−48.(12) Yuan, S.; Liu, T.; Feng, D.; Tian, J.; Wang, K.; Qin, J.; Zhang,Q.; Chen, Y.; Bosch, M.; Zou, L.; Teat, S.; Dalgarno, S.; Zhou, H. Asingle crystalline porphyrinic titanium metal-organic framework. Chem.Sci. 2015, 6, 3926−3930.(13) Bosch, M.; Sun, X.; Yuan, S.; Chen, Y.-P.; Wang, Q.; Wang, X.;Zhou, H.-C. Modulated Synthesis of Metal-Organic Frameworksthrough Tuning of the Initial Oxidation State of the Metal. Eur. J.Inorg. Chem. 2016, 2016, 4368−4372.(14) Shearer, G. C.; Chavan, S.; Bordiga, S.; Svelle, S.; Olsbye, U.;Lillerud, K. P. Defect Engineering: Tuning the Porosity andComposition of the Metal−Organic Framework UiO-66 viaModulated Synthesis. Chem. Mater. 2016, 28, 3749−3761.(15) Shearer, G. C.; Chavan, S.; Ethiraj, J.; Vitillo, J. G.; Svelle, S.;Olsbye, U.; Lamberti, C.; Bordiga, S.; Lillerud, K. P. Tuned toPerfection: Ironing Out the Defects in Metal-Organic FrameworkUiO-66. Chem. Mater. 2014, 26, 4068−4071.(16) Bosch, M.; Zhang, M.; Feng, D.; Yuan, S.; Wang, X.; Chen, Y.-P.; Zhou, H.-C. Lithium inclusion in indium metal-organic frameworksshowing increased surface area and hydrogen adsorption. APL Mater.2014, 2, 124103.(17) Brozek, C. K.; Dinca, M. Ti3+-, V2+/3+-, Cr2+/3+-, Mn2+-,and Fe2+-Substituted MOF-5 and Redox Reactivity in Cr- and Fe-MOF-5. J. Am. Chem. Soc. 2013, 135, 12886−12891.(18) Kim, M.; Cahill, J. F.; Fei, H.; Prather, K. A.; Cohen, S. M.Postsynthetic Ligand and Cation Exchange in Robust Metal−OrganicFrameworks. J. Am. Chem. Soc. 2012, 134, 18082−18088.(19) Liu, T. F.; Zou, L. F.; Feng, D. W.; Chen, Y. P.; Fordham, S.;Wang, X.; Liu, Y. Y.; Zhou, H. C. Stepwise Synthesis of Robust Metal-Organic Frameworks via Postsynthetic Metathesis and Oxidation ofMetal Nodes in a Single-Crystal to Single-Crystal Transformation. J.Am. Chem. Soc. 2014, 136, 7813−7816.(20) Zou, L.; Feng, D.; Liu, T.-F.; Chen, Y.-P.; Yuan, S.; Wang, K.;Wang, X.; Fordham, S.; Zhou, H.-C. A versatile synthetic route for thepreparation of titanium metal-organic frameworks. Chem. Sci. 2016, 7,1063−1069.(21) Park, J. Y.; Feng, D. W.; Zhou, H. C. Dual Exchange in PCN-333: A Facile Strategy to Chemically Robust Mesoporous ChromiumMetal-Organic Framework with Functional Groups. J. Am. Chem. Soc.2015, 137, 11801−11809.(22) Feng, D. W.; Liu, T. F.; Su, J.; Bosch, M.; Wei, Z. W.; Wan, W.;Yuan, D. Q.; Chen, Y. P.; Wang, X.; Wang, K. C.; Lian, X. Z.; Gu, Z.Y.; Park, J.; Zou, X. D.; Zhou, H. C. Stable metal-organic frameworkscontaining single-molecule traps for enzyme encapsulation. Nat.Commun. 2015, 6, 5979.(23) Lammert, M.; Bernt, S.; Vermoortele, F.; De Vos, D. E.; Stock,N. Single- and Mixed-Linker Cr-MIL-101 Derivatives: A High-Throughput Investigation. Inorg. Chem. 2013, 52, 8521−8528.(24) Park, J.; Feng, D.; Zhou, H.-C. Structure-Assisted FunctionalAnchor Implantation in Robust Metal−Organic Frameworks withUltralarge Pores. J. Am. Chem. Soc. 2015, 137, 1663−1672.(25) Deng, H. X.; Doonan, C. J.; Furukawa, H.; Ferreira, R. B.;Towne, J.; Knobler, C. B.; Wang, B.; Yaghi, O. M. Multiple FunctionalGroups of Varying Ratios in Metal-Organic Frameworks. Science 2010,327, 846−850.(26) Fracaroli, A. M.; Siman, P.; Nagib, D. A.; Suzuki, M.; Furukawa,H.; Toste, F. D.; Yaghi, O. M. Seven Post-synthetic CovalentReactions in Tandem Leading to Enzyme-like Complexity within

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

864

Metal-Organic Framework Crystals. J. Am. Chem. Soc. 2016, 138,8352−8355.(27) Lalonde, M. B.; Mondloch, J. E.; Deria, P.; Sarjeant, A. A.; Al-Juaid, S. S.; Osman, O. I.; Farha, O. K.; Hupp, J. T. Selective Solvent-Assisted Linker Exchange (SALE) in a Series of Zeolitic ImidazolateFrameworks. Inorg. Chem. 2015, 54, 7142−7144.(28) Deria, P.; Mondloch, J. E.; Karagiaridi, O.; Bury, W.; Hupp, J.T.; Farha, O. K. Beyond post-synthesis modification: evolution ofmetal-organic frameworks via building block replacement. Chem. Soc.Rev. 2014, 43, 5896−5912.(29) Yuan, S.; Lu, W. G.; Chen, Y. P.; Zhang, Q.; Liu, T. F.; Feng, D.W.; Wang, X.; Qin, J. S.; Zhou, H. C. Sequential Linker Installation:Precise Placement of Functional Groups in Multivariate Metal-OrganicFrameworks. J. Am. Chem. Soc. 2015, 137, 3177−3180.(30) Yuan, S.; Zou, L.; Li, H.; Chen, Y.-P.; Qin, J.; Zhang, Q.; Lu, W.;Hall, M. B.; Zhou, H.-C. Flexible Zirconium Metal-Organic Frame-works as Bioinspired Switchable Catalysts. Angew. Chem. 2016, 128,10934−10938.(31) Yuan, S.; Chen, Y.-P.; Qin, J.-S.; Lu, W.; Zou, L.; Zhang, Q.;Wang, X.; Sun, X.; Zhou, H.-C. Linker Installation: Engineering PoreEnvironment with Precisely Placed Functionalities in ZirconiumMOFs. J. Am. Chem. Soc. 2016, 138, 8912.(32) Yuan, S.; Chen, Y. P.; Qin, J. S.; Lu, W. G.; Wang, X.; Zhang, Q.;Bosch, M.; Liu, T. F.; Lian, X. Z.; Zhou, H. C. Cooperative ClusterMetalation and Ligand Migration in Zirconium Metal-OrganicFrameworks. Angew. Chem., Int. Ed. 2015, 54, 14696−14700.(33) Lee, S.; Kapustin, E. A.; Yaghi, O. M. Coordinative alignment ofmolecules in chiral metal-organic frameworks. Science 2016, 353, 808−811.(34) Inokuma, Y.; Yoshioka, S.; Ariyoshi, J.; Arai, T.; Hitora, Y.;Takada, K.; Matsunaga, S.; Rissanen, K.; Fujita, M. X-ray analysis onthe nanogram to microgram scale using porous complexes. Nature2013, 495, 461−466.

Accounts of Chemical Research Article

DOI: 10.1021/acs.accounts.6b00457Acc. Chem. Res. 2017, 50, 857−865

865