Porous Molecular Solids and Liquids

Abstract

Until recently, porous molecular solids were isolated curiosities with properties that were eclipsed by porous frameworks, such as metal–organic frameworks. Now molecules have emerged as a functional materials platform that can have high levels of porosity, good chemical stability, and, uniquely, solution processability. The lack of intermolecular bonding in these materials has also led to new, counterintuitive states of matter, such as porous liquids. Our ability to design these materials has improved significantly due to advances in computational prediction methods.

Short abstract

Porous molecular materials lack intermolecular bonds between their building blocks. Hence, unlike porous extended frameworks, they can be processed in solution to form crystalline porous solids, amorphous porous solids, or even porous liquids.

Porous molecular materials have been featured in previous reviews,¹⁻¹⁶ some of which have also given histories of this field,^5,15 which dates to work in the 1970s by Barrer and Shanson on organic zeolites.¹⁷⁻²⁴ This Outlook will not repeat those reviews but will focus instead on the rapidly emerging opportunities in this area.

Terminology

For the purposes of this review, “porous molecular materials” or “porous molecules” are porous solids or liquids where the molecular subunits are not connected by strong intermolecular bonds, such as coordination bonds or covalent bonds. We further define “molecule” as a discrete molecule with a specific molar mass, as opposed to a polymer, although we do touch briefly on polymers of intrinsic microporosity (PIMs), which are in a sense also “porous molecular materials”. “Porous organic cages” (POCs), “hydrogen-bonded organic frameworks” (HOFs), and “supramolecular organic frameworks” (SOFs) are hence all subclasses of porous molecular materials. “Extrinsic” and “intrinsic” porosity, respectively, refer to pores that are located between molecules or within molecules, typically inside a cage. For example, “organic molecules of intrinsic microporosity” (OMIMs) are molecular solids that are extrinsically porous because the rigid molecules pack badly; they do not have intrinsic porosity built into the molecule, as in a porous organic cage. “Extended frameworks” refers to crystalline solids comprising strong intermolecular coordination or covalent bonds, such as metal–organic frameworks (MOFs) or covalent organic frameworks (COFs), unless specified; for example, where discussing “amorphous MOFs”. “Porous polymer networks” are extended covalent networks that, unlike COFs, are amorphous rather than crystalline; these have extended bonding throughout the network and are hence insoluble, unlike PIMs.

Why Porous Molecules?

Porous molecular materials are composed of discrete molecules, unlike MOFs,^25,26 COFs,^16,27 and porous polymer networks,^16,28 all of which are connected by intermolecular bonds. The field of extended porous frameworks and networks is burgeoning: the area of porous MOFs alone is vast. Why, then, do we need another flavor of porous material? We believe that porous molecular materials are complementary to more established porous solids because of their distinguishing features, which include the following.

Noncovalent Synthesis

MOFs are synthesized directly as insoluble solids by metal–organic bond formation; COFs are generated using reversible covalent bonds. By contrast, porous molecular crystals are produced by crystallization, in which no new bonds are formed. It is hence relatively easy to grow large, high-quality porous molecular crystals—for example by slow solvent evaporation (Figure 1a)²⁹—although in some cases, specific desolvation protocols are required to retain order and porosity.^30,31 By comparison, erroneous side reactions may become locked into MOFs and COFs, even when the chemistry is reversible, although this can also be viewed as a tool to engineer properties.³² Defects can certainly be prevalent in porous polymer networks formed by irreversible routes.³³ The ability to grow high-quality porous molecular crystals is a potential advantage in applications that require long-range order. It also aids characterization: for example, most of our porous organic cage structures were solved by single-crystal X-ray diffraction, whereas there are very few examples of single-crystalline COFs.^34,35

Figure 1.

Porous molecular materials can be produced as (a) crystalline solids,⁶⁴ (b) amorphous solids,^36,42 or (c) liquids.⁴⁸ For scale, the coin diameter in panel a is 22.5 mm. The amorphous thin film in panel b is made from a scrambled cage mixture, composition denoted by the chromatogram. Panel c, lower, shows xenon gas being displaced from a porous liquid by chloroform. In all cases, cycloimine cages are depicted, formed from the [4 + 6] condensation of 1,3,5-triformylbenzene with various 1,2-substituted vicinal diamines.

Solubility

Unlike extended frameworks, many porous molecules are soluble in organic solvents. They can be solution processed into a variety of forms, such as thin films and membranes. This applies to molecular cages (Figure 1b),³⁶ polymers of intrinsic microporosity (PIMs),^21,37 and organic–organic mixed-matrix membrane (MMM) composites,^38,39 where a crystalline porous molecular filler reduces membrane aging.

Amorphous Porous Solids and Porous Liquids

Porous molecules do not have to be crystalline solids. PIMs, a large and growing class of materials, are amorphous.^21,37,40 Likewise, smaller molecules can form amorphous porous phases (Figure 1b),⁴¹⁻⁴⁴ and certain porous organic cages can be processed deliberately into either amorphous or crystalline forms: the amorphous forms retain the porosity imparted by the cage, but have different physical properties (Figure 2).²⁹ Amorphous MOFs⁴⁵ are a related, emerging class of material, but these are typically formed by amorphization of a parent crystalline framework, for example by ball milling. Amorphous porous molecular solids can be prepared directly by using processing tricks, such as rapid solvent removal,⁴⁶ or by making molecules unsymmetrical so that they cannot crystallize.⁴² Amorphous MOFs are typically less porous than their crystalline porous counterparts, whereas some cages, such as CC3, a [4 + 6] cycloimine cage, are substantially more porous as amorphous solids (Figure 2).²⁹ The most radical extension of this amorphization concept is the idea of porous molecular liquids (Figure 1c); these can be accessed by designing porous molecules that melt⁴⁷ or that are highly soluble in size-excluded solvents.^48,49

Figure 2.

Some amorphous molecular cage solids can be more porous than their crystalline counterparts;²⁹ typically, the reverse is true for MOFs. The molecule shown is a tetrahedral [4 + 6] cycloimine cage with (R,R)-1,2-cyclohexanediamine vertices, packed (a) as a crystalline porous solid and (b) as an amorphous porous solid.

Structural Flexibility

Since we first surveyed the field of porous organic molecules,² the area of flexible MOFs has advanced significantly.⁵⁰⁻⁵² Nevertheless, our original comments² still hold: molecular crystals are not interconnected by strong covalent or coordination bonds, and they can undergo large rearrangements in the solid state. This has led to molecular crystals with “on/off” porosity switching behavior (Figure 3)⁵³ and, recently, macrocyclic pillar[n]arenes that show extremely high selectivity for styrene over ethylbenzene via a guest-induced restructuring process.⁵⁴

Figure 3.

On/off porosity switching in a porous organic cage crystal.⁵³ Guest 1 transforms the molecule into a porous polymorph; guest 2 transforms it into a nonporous polymorph. This suggests a potential guest trapping mechanism for hazardous or valuable molecules. The cage molecule is a [4 + 6] cycloimine cage with flexible 1,2-diaminoethane vertices.

Stability

Intuitively, porous molecular solids might be expected to be less stable than extended MOFs and COFs. This is not always the case. For one thing, it is not essential to use reversible, dynamic chemistry to make crystalline porous molecules. Rigid molecules that pack badly to create so-called “extrinsic” porosity date back to Barrer’s early studies on Dianin’s compound.^17,19,30 Likewise, macrocyclic calixarenes²² are not prone to the chemical instability that can be associated with highly reversible MOF and COF chemistries. There are fewer examples of intrinsically porous cage molecules formed using irreversible coupling routes, but the work of Doonan on alkyne-linked cages points the way there, too.^55,56 Chemical stability and crystal stability are not the same thing, but they are frequently interlinked, and certain chemically stable porous molecular crystals show stabilities that exceed those of most MOFs and COFs. For example, we reported a highly crystalline [4 + 6] porous organic aminal cage with good chemical and crystal stability in water for 12 days at pH values ranging from 1.7 to 12.3 (Figure 4).⁵⁷ Likewise, Banerjee has shown that porous organic cages can be stable under both acidic and basic conditions because of keto–enol tautomerism.⁵⁸ Few MOFs or COFs are stable over such broad pH ranges; indeed, most zeolites are unstable under either acidic or basic conditions. To our knowledge, these cage molecules^57,58 are among the most chemically resistant crystalline microporous solids known, although they are currently isolated examples. The dream of molecular organic zeolites¹⁷ lives on.

Figure 4.

Porous molecular crystals can have good chemical stability. This cage crystal is stable over a pH range of 1.7–12.3, and neither the X-ray diffraction pattern (a) nor the nitrogen sorption isotherm (b) is affected by acid or base treatment.⁵⁷ Inset shows predicted (red) and observed (blue) conformation for the cage. The cage molecule is made stable by reduction of the imine functionalities and conversion to an aminal. The lower panel shows the stability range for the cage, mapped onto the pH scale as depicted by universal indicator colors.

Porous Molecular Crystals

Notwithstanding the differences summarized above, there are also many similarities between porous molecular crystals and extended MOFs and COFs. Extended frameworks are synthetically diverse because an almost unlimited array of organic linkers is conceivable. It is also possible to carry out postsynthetic modifications on extended porous frameworks⁵⁹ and to mix different linkers,⁶⁰ further increasing this diversity. The synthetic space for porous molecular materials has the potential to be equally large. As for extended frameworks, multiple building blocks can be combined within a single porous solid^42,61,62 (or porous liquid).^48,49 Also, like MOFs and COFs, porous molecules can be postfunctionalized to vary the physical properties.^57,63

One difference between porous molecular crystals and extended porous frameworks is the density and surface area range that can be accessed. When the first porous organic cages were reported in 2009, the most porous molecules had a Brunauer–Emmett–Teller (BET) surface area of around 600 m² g^–1, corresponding to a crystal density of 0.973 g cm^–3.⁶⁴ By comparison, the most porous MOF at that time, UMCM-2,⁶⁵ had a surface area of 5200 m² g^–1 and a crystal density of 0.401 g cm^–3; the corresponding values for COF-103 were 4210 m² g^–1 and 0.38 g cm^–3.⁶⁶ This gulf between porous molecular crystals and extended frameworks has since narrowed.⁶⁷ Mastalerz has led the way here, producing an intrinsically porous organic boronate ester cage (3758 m² g^–1/desolvated density not reported)³¹ and an extrinsically porous hydrogen-bonded benzimidazolone framework (2796 m² g^–1/0.755 g cm^–3),³⁰ both with porosity levels that are remarkable for molecular solids. The least dense desolvated molecular solid to date has a density of just 0.41 g cm^–3.⁶⁸ For most practical purposes, then, porous molecular crystals can be considered somewhat competitive with MOFs and COF in terms of available porosity. A caution, though, is that physical and chemical stability is a likely concern for any crystalline solid with a very low density (e.g., <0.5 g cm^–3), irrespective of material subclass. As commented before,⁶⁹ robust activated carbons have been commercially available since the 1990s with surface areas of more than 3000 m² g^–1. Functional added value will be needed to propel porous molecular crystals, and other porous frameworks, toward applications. Some of the more interesting targets might be small-pore, low surface area solids that can perform challenging separations.^54,70,71 Taking that argument further, “zero-dimensional”⁷² porous molecular crystals may be interesting for gas storage and release.

Extrinsically porous molecular crystals^17,19,21,22 predate the first intrinsically porous organic cages,^64,73 but until recently, the levels of porosity in extrinsically porous crystals were modest, and their potential applications largely hinted at. That picture is changing; a range of porous hydrogen-bonded organic frameworks (HOFs)⁷⁴—also referred to as supramolecular organic frameworks (SOFs)⁷⁵—has now been reported.^{30,68,74−93} The most porous of these so far are Mastalerz’s triptycene benzimidazolone and analogues,^30,68 but multiple properties have been studied in addition to surface area, such as fullerene capture,⁷⁶ uranium capture,⁸⁶ fluorocarbon capture,⁹⁴ ion absorption,⁸⁹ proton conduction,⁷⁸ thermoelasticity,⁷⁹ sensing,^81,84,87 enantioseparations,⁹¹ CO₂ separation,^{78,80,82,85,90,95} and hydrocarbon separation.^68,74,92 This is a rapidly developing area, and solution processability is a potential advantage for HOFs and SOFs, too. Based on these developments, intrinsically and extrinsically porous molecules look set to be equally valid strategies for the future.

The initial focus for porous molecular crystals was on gas adsorption and molecular separations, although recent studies suggest that they might also be relevant in “second generation” framework applications, such as proton conductivity.^78,96 An obvious exception here would be conjugated COFs,⁹⁷ where there are as yet no molecular analogues; a worthy future target. For some applications, such as the separation of krypton and xenon, molecular crystals were demonstrated to have better selectivity than extended frameworks,⁷⁰ at least transiently.⁹⁸ However, since porous molecular crystals and porous extended frameworks can now be thought of somewhat interchangeably, the most profitable areas for porous molecular materials in the future might exploit one or more of the differentiating features listed above. There are several underexplored areas. For example, hydrolytically stable “molecular molecular sieves”^57,99 could have applications in water purification¹⁰⁰ and, potentially, desalination, especially if they could be solution-cast as crystalline thin films with adequate permeance. Likewise, there is just one report so far of solution-cast organic–organic MMM composites, which involved a polymer of intrinsic microporosity (PIM-1) and an organic imine cage (CC3),³⁸ and this is an unexplored playground for the future.^39,101

Intrinsically Porous Framework Linkers

Porous molecules can also be used as linkers in extended frameworks to obtain additional functionality. As yet, there are few reports of “cage MOFs”¹⁰² or “cage polymers”¹⁰³ (and no reports of crystalline “cage COFs”), but in principle this should be a fertile area, noting that MOFs were already synthesized from macrocycles, such as cyclodextrins.¹⁰⁴ To give one example, MOFs can be prepared from amine cages, where the amines act as metal coordination ligands and the cage linker provides a (limited) degree of intrinsic porosity.¹⁰²

It ought to be possible to create a wide range of intrinsically porous cage linkers for MOFs, particularly as we extend the range of cage topologies that are available (e.g., 1-D tubular cages¹⁰⁵). To achieve this, it might be necessary to develop porous molecules with extra peripheral functionality^106,107 to allow for metal chelation or for COF formation. It might even be possible to create mechanically bonded cage linkers for MOFs or for COFs.^108−110

Amorphous Porous Molecular Solids

There are fewer well-characterized examples of amorphous porous molecular solids in comparison with porous molecular crystals, at least for smaller molecules; PIMs is a more developed area.^21,37,40 One of the first detailed studies on porous small-molecule amorphous solids was carried out by Atwood and co-workers.⁴¹ Subsequently, much more porous amorphous molecular solids were reported,²⁹ including rigid organic molecules of intrinsic microporosity (OMIMs)^44,111 and “scrambled” organic imine cages that are purposefully desymmetrized such that they cannot crystallize.⁴² Amorphous porous molecular solids present interesting opportunities: like PIMs, they are solution processable and amorphous, but unlike PIMs, they can have prefabricated cavities and window sizes with precisely defined size and geometry, although these are typically interconnected by less well-defined extrinsic pores (Figure 2b).⁴³ These molecular solids can be cast, like PIMs, using standard procedures such as spin-coating to give uniform conformal thin films (Figure 1b).³⁶ The amorphous imine cage films tested in our preliminary studies had, in fact, worse aging characteristics than PIMs,³⁶ but if that problem could be solved, then similar systems might be useful for gas separations. For example, the inherent size,^70,112 shape,^71,112−114 or chiral^70,113 selectivity of porous organic cages might be exploited in thin, amorphous layers supported on a more permeable polymer support, or within a separation column.^71,113

A different mechanism for molecular separations is guest-induced restructuring, related to adductive or reactive crystallization processes,¹¹⁵ which can be applied to both crystalline and amorphous molecular phases alike.⁵⁴ Here, separation occurs because of selective conversion to a specific crystalline guest inclusion compound; that is, the host molecule chooses a specific guest, and is restructured by it. Such processes can be exquisitely selective,⁵⁴ but slow uptake kinetics and energy requirements for guest release are questions to address in terms of practical molecular separations. Again, the solution processability of these materials—perhaps onto or into a support material⁴⁶—might offer a way forward.

Another untapped area is amorphous, porous molecular composites. There is just one report of this so far,¹¹⁶ in which either guest molecules, such as pyrene, were blended into scrambled porous imine cage materials or, alternatively, the cages were blended into otherwise nonporous linear polymers. There are multiple opportunities here. For example, a cheap commercial polymer, such as polystyrene, might be rendered porous—or at least permeable—by simply blending it with molecular pores. A further opportunity might exist in catalysis, where a molecular catalyst is physically blended with molecular pores to create an amorphous microporous composite. As a proof of concept, we showed that molecular pyrene could be uniformly distributed throughout an amorphous cage matrix at loadings of up to around 10 wt %, without losing microporosity (Figure 6).¹¹⁶ Metal catalysts, or organocatalysts, could be blended into similar materials, which can also be chiral.⁷⁰ This would be a different paradigm from more conventional catalyst supports, such as MOFs,¹¹⁷ although at this stage the benefits and drawbacks are unclear; clearly, solubility of the cage support itself might preclude some applications.

Figure 6.

Molecular organic pores can be blended with guest molecules to make porous amorphous composites. Here, a scrambled porous imine cage (see Figure 1b) is blended with pyrene to make fluorescence composite powders (a, b).¹¹⁶ The composites retain microporosity up to around 10 wt % pyrene loading (c). This suggests applications such as catalysis, where a metal catalyst or organocatalyst is blended with molecular pores, rather than being attached to a preformed porous catalyst support.

Figure 5.

Porous molecular crystals can have remarkably high surface areas. Both intrinsic porosity (pores within the molecule)³¹ (a) and extrinsic porosity (pores between molecules)³⁰ (b) are valid routes for doing this.

Porous Molecular Liquids

Porous liquids are liquids with permanent rather than transient pores. The concept and technical challenges were first discussed by James in 2007.¹¹⁸ It proved challenging, however, to translate these ideas into reality. The “dam”¹¹⁹ for porous liquids has now burst; since 2012, six papers have been published that describe these materials and their properties,^{47−49,120−122} and five of these studies were based on shape-persistent [4 + 6] porous organic imine cages.^{47−49,120,122} The fundamental distinction between a porous liquid and a porous solid is fluidity, which has various potential process benefits; for example, liquids can be pumped around in a continuous system, which could facilitate guest loading and unloading steps. This assumes, of course, that the porous liquid viscosity is low enough to allow this. It has already been shown that porous liquids can have gas sorption capacities that are atypically high for liquids. In principle, other properties of porous solids should translate into porous liquids, such as guest selectivity.

There are several interesting research targets for porous liquids. First, these materials should be benchmarked against existing high free-volume liquids, such as ionic liquids and polyethers (e.g., Selexol),¹²³ though it may be noted that those solvents have been explored more for acidic gases, and that porous liquids are applicable to neutral gases, such as light hydrocarbons and xenon.^48,49 A second challenge is to create porous liquids with lower volatility or, ideally, zero volatility (e.g., a porous ionic liquid); in this respect, the crown ether or perchlorinated solvents used in our first systems^48,49 are not ideal. A third area to explore is guest adsorption and release kinetics: low- or zero-volatility porous liquids should facilitate pressure-swing or temperature-swing adsorption/desorption cycles. We also showed recently that less standard stimuli, such as ultrasonication,⁴⁹ can promote gas release from porous liquids, although it is unclear that this would be scalable or cost-effective in real processes.

Gas adsorption and separation are not the only possible applications for these new materials. Porous liquids might have unique properties for homogeneous catalysis: for example, they could provide reaction selectivity by confinement¹²⁴ or by chiral induction, though we showed recently that porous chiral cage liquids did not exhibit chiral selectivity,⁴⁹ unlike a crystalline homochiral porous solid formed from a similar cage molecule.⁷⁰ It is also possible that porous liquids could accelerate the rate of reactions involving gases (e.g., H₂, O₂, ethylene, etc.), perhaps allowing reactions to be carried out at lower gas pressures. More generally, porous liquids might stimulate new thinking in almost any area that uses liquid solvents; for example, one could envisage porous extraction media, porous cleaning fluids, or porous electrolytes.

This discussion has focused on porous molecular liquids. Of equal interest in the future might be porous liquids, or “fluidized porous solids”, that comprise a colloidal porous solid dispersed at a high loading in a suitable carrier fluid.¹²¹ In principle, this carrier fluid could itself have molecular, prefabricated porosity,⁴⁷⁻⁴⁹ for example to increase mass transport or to improve selectivity.

Designing Porous Molecular Materials

Exciting as they are, porous molecular materials are quite challenging to design. Indeed, the rarity of porosity in crystalline organic materials is illustrated by a recent study,¹²⁵ which found only a handful of porous molecular crystals from a sampling of more than 150,000 reported structures. Reticular synthesis¹²⁶ is a powerful tool for designing crystalline porous extended frameworks, but the (older) concept of crystal engineering¹²⁷ for porous molecular crystals has proved somewhat less generalizable. This is mostly because intermolecular interactions in molecular crystals are, in general, weaker and less directional than metal–organic bonding in MOFs or covalent bonding in COFs.²⁷ There are hence few isoreticular strategies for porous molecular solids. Heterochiral window interactions between cages have proved strong enough to serve as a platform for reliable crystal engineering,^{46,61,62,105,128−130} although in some cases, this interaction needs additional reinforcement by the inclusion of a shape-specific “directing solvent”,¹²⁸ and it does require the use of homochiral feedstocks.

One of the most attractive design strategies for porosity generation in molecular solids is to prefabricate the pores in the molecule itself, not least because molecular porosity and shape-persistence can be predicted, a priori, for candidate cage building blocks.^131−134 For example, the [4 + 6] organic imine cage, CC3, is porous in the crystalline state⁶⁴ by virtue of its molecular structure: few crystalline packings for CC3 are nonporous, even hypothetical ones. CC3 is also porous—more porous, in fact—as an amorphous solid (Figure 2b).²⁹ This strategy of molecular design can also succeed for porous liquids, where it is possible to pair shape-persistent porous cages with bulky solvents that are size-excluded from the cage windows.^48,49 Ultimately, though, molecular structure and solid-state structure cannot be disconnected. To give one example, crystalline CC3 shows perfect selectivity for C9 aromatic hydrocarbon separation whereas the amorphous form of the same molecule does not.⁷¹ The need to understand the crystal packing (or amorphous packing) is even more apparent for extrinsically porous molecules, where the likely physical properties may not be obvious at all from the building blocks in isolation.^17,19,30 There are two solutions here: the development of robust intermolecular synthons or “tectons” that will facilitate the intuitive design of porous molecular frameworks^{18,30,105,127,128} or the a priori prediction of structure from the molecular formula, irrespective of the intermolecular forces.¹³⁵ Some of the most powerful strategies might lie at the confluence of these two ideas.

Energy–Structure–Function Maps

Crystal engineering has proved successful for porous molecular solids, but it works best within specific subclasses of materials, such as chiral organic cages.^{46,61,62,105,128} There is no synthon or tecton for porous molecular solids that is as versatile as, say, metal carboxylate bonding in isoreticular MOFs,¹³⁶ let alone a “magic bullet” for the reliable construction of porous molecular frameworks. In collaboration with colleagues at the University of Southampton, we have explored strategies for designing porous molecular solids using crystal structure prediction (CSP).^{62,68,105,129} This approach involves the application of space group symmetry followed by energy minimization and lattice energy calculation,¹³⁷ and it assumes no favored topology or specific intermolecular tectonic interaction. In principle, therefore, CSP is generalizable, although there still are difficult technical challenges, such as the computational expense of dealing with large or flexible molecules. We first showed that CSP can be used to predict the structures of porous organic cages.⁶² It was shown that the observed tendency for porous, chiral cage molecules such as CC3 to crystallize in a window-to-window fashion⁶⁴ was manifested in a large calculated lattice energy gap between the global minimum predicted structure and the bulk of the energy–structure landscape.⁶² For other, less directional cage molecules, such as the polymorphic, racemic cage CC1,⁵³ the calculated lattice energy differences between structures were much smaller.⁶⁴ Hence, these CSP methods provide us with a means to distinguish between molecules with deep-lying, unique global minima—that is, robust tectons—and molecules that are more likely to exhibit polymorphism. Based on such analyses, CC3 would be predicted from its energy–structure landscape to be a robust tecton,⁶² whereas CC1 would not.¹²⁹ More recently, the CSP approach was extended to extrinsically porous molecules,^68,138 such as the hydrogen-bonding frameworks developed by Mastalerz.³⁰ We showed that the energy–structure landscapes for such molecules can be complex: for example, a trigonal benzimidazolone was found to have at least four stable porous polymorphs with predicted densities in the broad range 0.4–1.25 g cm^–3 and predicted surface areas ranging from 442 to 3230 m² g^–1.⁶⁸ All four of these structures were located on the low-energy boundary, or “leading edge” of the calculated energy–structure landscape (Figure 7). By calculating physical properties for each structure on a predicted energy–structure landscape, we produced energy–structure–function (ESF) maps.⁶⁸ These color-coded maps can suggest, at a glance, which molecule in a candidate set might be most appropriate for a given application, such as methane storage⁶⁸ (Figure 7) or, by extension, any other physical property that can be calculated from structure. The use of ESF maps should extend beyond porous solids, and is in principle applicable to any physical property that can be calculated from crystal structure.

Figure 7.

Crystal engineering for porous molecular solids is challenging because the intermolecular interactions are typically weaker and less directional than for extended frameworks, such as MOFs and COFs. Energy–structure–function (ESF) maps can reveal the function–energy landscape for a candidate molecular building block.⁶⁸ Here, a benzimidazolone molecule (cf., Figure 5b) was shown to have at least four stable porous polymorphs (T2-α–T2-δ; red = predicted structures; blue = experimental crystal structures). T2-γ was predicted to have the most competitive methane delivery capacity (159 v STP/v), as confirmed by experiment.

Outlook

The area of porous molecular materials has matured significantly in the past few years through the combined efforts of several research groups worldwide. The field has evolved from a set of mostly isolated examples to include series of related materials that were designed in a broadly isoreticular fashion.^{46,61,62,105,128,129} Very recently, Diercks and Yaghi surveyed the area of COFs¹³⁹ and stated that, for crystalline COFs, “covalent bonds create robustness and directionality (necessary to control the spatial orientation of the building blocks)”. In this author’s view, the use of intermolecular covalent bonds is not always required for directionality in crystalline organic solids,^62,105 nor does it necessarily produce greater chemical robustness.^57,58,140 Certainly, the less predictable nature of noncovalent supramolecular assembly has often thwarted the purposeful design of molecular crystals, but that could be set to change.^68,141 This raises an interesting question: if MOF, COF, porous polymer, and porous molecular solid syntheses were all perfectly predictable, then which class of porous solid would one choose? The humorous answer is zeolites. Less frivolously, attention would undoubtedly turn to function, cost, and sustainability.¹¹ In that regard, solution-processable, robust porous molecular crystals might fill a unique space in the wide roster of applications for porous materials.

The author declares no competing financial interest.