Purely Covalent Molecular Cages and Containers for Guest Encapsulation

1Molecular encapsulation is a key step for many processes found in Nature. For example, endocytosis is a crucial process for cells to internalize macromolecules and particles using transport vesicles that isolate a substance from surroundings by encapsulation. Moreover, enzyme-binding pockets formed by amino acid residues result in selective binding and catalytic performance. Nonpolar residues in pockets provide a hydrophobic environment to the binding cavity, which also contains buried polar and charged units to minimize solvation from water molecules. Supramolecular architectures, particularly molecular cages, provide synthetic cavities to effectively bind guest molecules by multiple noncovalent interactions with the cage-building blocks surrounding the cavity., The inclusion of guest molecules in suitable host-cage derivatives results in restricted systems that confer entrapped substrates novel properties. With their discovery, the possibility of preparing cage structures with specific cavities has drawn the attention of supramolecular chemists, who have developed numerous systems for the encapsulation of guest molecules for different applications, such as stabilization of species, gas separation, catalysis,− among many others.,

2Sometimes cage compounds offer a unique binding pocket compared to enzyme-binding sites, which can be tailored in terms of the size, shape, and functional groups pointing toward the cavity, and many other parameters by customizing the structure of the cage. In this review, we explore and analyze the properties of different families of fully organic cages and containers by focusing on the properties of the cavity, including information like cavity size and volume, wherever possible, and encapsulated guests as well as main applications (Figure 1). For this purpose, we divide the review into different sections that group distinct families of cages by their properties or the methodology employed for their synthesis. The first section describes the general strategies widely used to prepare organic cages, including irreversible bonds, reversible bonds, and templates. After this, the review includes two sections that group cages according to their solubility in organic solvents and their solubility in water.

3Despite the wide scope of this area, most of the reviewed work in the field has to date referred to very specific cage families and applications. We intend to herein provide a broader overview by emphasizing the synthesis and properties of the cage cavity for the encapsulation of guest molecules. In addition, we have attempted to bring together different families of molecules with a central cavity, including hemicarcerands, carcerands, cryptands, cavitands, capsules, and cages. This approach aims to provide a general conceptual view of this area. We hope that the compiled examples provide a comprehensive guide to understand the field that fuels the development of novel molecular structures with new or enhanced properties and applications.

4The nomenclature of the cage compounds that we describe below follows the convention adopted in the literature., As such, we consider hemicarcerands, carcerands, cryptands, cavitands, capsules, and cages. Whenever appropriate, the size of the cavity (volume and diameter) described in the original reference is included. For those cages that do not have this information available, we prepared molecular models using the wave function Spartan software and determined the volume and diameter of the cavity using pywindow.,

5This review explores different families of purely organic cages. The reported examples reveal the wide diversity of cage systems that can be prepared by using building blocks of different shapes and sizes. We highlight distinct possible synthetic protocols to prepare organic cages, the properties of the cavity of the cage, and their applications. The review does not include, in general, cages with small cavities to encapsulate inorganic anions, and the reader can find information elsewhere. The review is not focused on coordination cages containing metals and organic ligands. In this area, the reader can find seminal works from Fujita, Stang, Nitschke, Raymond, and others in comprehensive reviews covering these families of cages.−

6The examples included in the review are mainly from the past decade, with a large number of examples from the last five years. Literature search has been included up to the first semester of 2021.

7This section describes the main strategies used to prepare molecular cages, including examples that explore the effect of different reaction conditions on the outcome of the cage formation reaction. We also include representative examples that show the effect of the preorganization and rigidity of building blocks on the cage formation reaction, which highlights the possible geometries and variety of cages that can be obtained. This section also includes a description of the main water-solubilizing groups used to obtain water-soluble cages that are further explained in detail in the examples described in Section 4. The last part of this section includes the molecular modeling strategies adopted to simulate and predict the properties of molecular systems, including the porosity and the outcome of the cage self-assembly reaction from a set of building blocks.

8Cage synthesis involves the assembly of convergent building blocks to form a hollow structure with a central cavity. The design of cage-building blocks is a key issue for successful cage synthesis. The most important factors are the shape, conformational preferences, and the reactive functional groups that will link together the building blocks within the cage framework. This section describes the main synthetic strategies used to prepare organic cages and is divided into two main categories depending on the bond reaction used to assemble the cage from building blocks, i.e., irreversible and reversible bond formation. Whereas the first examples of cages were obtained using irreversible bonds, for instance SN2 reactions to form C–O bonds, most recent cage formation methods are based on reversible bonds. In fact, the recent advances made in the field of covalent molecular cages can be attributed to important progress in dynamic covalent chemistry, allowing the synthesis of cage compounds in a few steps and usually with good yields. Most of the cages described in the literature based on reversible bonds contain imine bonds built up from amine- and aldehyde-containing blocks. Other also widely employed reactions for the construction of cages based on reversible bonds employ hydrogen, chalcogen, boronate ester, and disulfide bonds.

9The main synthetic advantage of reversible bond formation in the synthesis of cages is the self-correction of the mistakes made in the multistep self-assembly reaction pathway, which finally allow obtaining of a thermodynamic more stable cage product., In contrast, irreversible bonds usually result in low cage formation yields because they normally favor oligomers or polymers instead of cages. While self-assembly kinetics of metal coordination organic cages has been extensively studied, related studies for purely organic cages remain mainly unexplored with relatively very little available information.

10A key aspect in the cage formation kinetics is the effective molarity (EM) of the reactive groups, which can be increased by employing templates. These concepts have been widely used in the synthesis of macrocycles, and similar conceptual ideas have been followed for the formation of cages. EM is a key factor that defines the cooperativity of self-assembled systems; in particular, the product of the equilibrium constant K of the building block self-assembly reaction by EM determines the extent to which the self-assembled structure is formed. If product K·EM > 1, the cooperative intramolecular interaction between the self-assembled structure building blocks is favored, and this structure will be the main self-assembled product. These concepts are valid for any self-assembled structure held together by cooperative intramolecular interactions regardless of its architecture, including molecular cages.

11Despite the versatility and the large number of irreversible organic synthesis reactions, their applications in the synthesis of organic cages are quite limited. As stated above, irreversible bonds do not allow the mistakes made during cage formation to be corrected, which limits the synthesis of large cage structures because cage assembly process efficiency decreases with the required number of bonds to be formed during cage assembly. Back in 1969, Lehn, Sauvage, and Dietrich reported one of the first examples of molecular cages (i.e., macrobicycles) by irreversible bond-forming reactions. They obtained overall yields of ca. 25%. However, when cage complexity increases, the formation of large assemblies through irreversible bonds becomes inefficient and requires additional protection/deprotection steps to minimize side reactions. This is exemplified, for instance, in the reported carceplex synthesis by Sherman and co-workers. Synthesis involves four steps: transforming the starting material 1 into intermediates 2–4 with yields of 26%, 16%, 58%, and 35%, which results in an overall carceplex 5 formation yield of 0.8%. A low overall yield limits the preparation of large quantities (Figure 2).

12For smaller assemblies that involve the assembly of two building blocks, overall cage formation yields are much higher and allow, for instance, the use of carbon–carbon formation reactions to prepare different cage structures. Chen and co-workers proved the efficient use of Eglington–Glaser coupling for cage synthesis from precursor 6 by reporting the synthesis of an all-carbon triptycene-based molecular cage 7 with a 58% yield (Figure 3 top). Following a similar approach, Doonan, Sumby, and co-workers prepared precursor 8, which allowed the synthesis of the organic cage 10. Whereas cage synthesis using the same synthetic strategy from precursor 8 has only a 20% yield, that of synthesis form intermediate 9 is 52–53% (Figure 3 bottom)., Cage 10a has a modeled structure with a cavity shape of a distorted triangular prism and internal vertical and horizontal diameters of 13.5 and 12 Å, respectively. The authors found that the gas adsorption properties of this material very much depended on the crystallization polymorph. Slow crystallization yielded 10aα, which is nonporous to N2 gas. In contrast, rapid precipitation led to the polymorph 10aβ, which displayed a remarkably large BET surface area of 1153 m2/g, which is rare for molecular cages. The calculated pore size distributions showed two voids: one with a diameter of 11 Å associated with the internal cavity and another with a diameter of 6 Å associated with extrinsic voids. Six-pyridine functionalization (cage 10b) brings about cage structure distortion that results in different packing and gas absorption performance and is nonporous to N2. The authors predicted that the precise control of the internal functionality of the cage cavity would help in tuning gas absorption performance.

13The formation of carbon cages with strained structures is also possible using building blocks with the appropriate geometry. Itami, Kamada, and co-workers reported the first synthesis of an all-benzene [6.6.6] carbon nanocage 11, which resembled the junction unit of branched carbon nanotubes. These structures can be obtained by means of specific synthetic protocols that involve L-shaped cis-diphenylcyclohexane units and linear phenylene groups coupled using Suzuki–Miyaura coupling or nickel-mediated homocouplings to afford the nonstrained cage intermediate that is subsequently transformed into the strained cage structure (Figure 4)., By this methodology, it was possible to prepare carbon nanocages 12 [5.5.5],13 [4.4.4], and 14 [2.2.2]. The synthesis of cage 11 from precursor 15 through strained cage intermediate 16 is described in Figure 4. The absorption wavelength of cages are red-shifted by increasing cage size as a result of a combination of factors associated with the special HOMO and LUMO properties of these fully conjugated systems.

14The success of the above method is based on the cage formation step from building block precursors with the appropriate geometry to yielding a strain-free intermediate cage. Yamago and co-workers used cage precursor 17 to synthesize carbon nanocage 18 by a different strategy but one also based on preparing an intermediate strain-free cage that was later transformed into the final cage product. The developed method was based on employing a hexanuclear platinum octahedral cage 17 consisting of four building blocks of a stannylated trisubstituted benzene derivative. The reductive elimination of platinum yielded the target carbon nanocage 18 with a 19% yield (Figure 5).

15Toyota and co-workers prepared from precursor 19 a fully aromatic hydrocarbon cage 20 that features two triptycene bridge units using Suzuki–Miyaura coupling in the cage-formation step with a 10% yield. Cage 20 has an approximately spherical and rigid cavity surrounded by multiple hydrogen atoms pointing toward the cavity with good complementarity for fullerenes. In particular, the cage encapsulates guests C60 or C70 in toluene with association constants of 1.3 × 104 M–1 and 3.3 × 105 M–1, respectively. The selectivity toward C70 over C60 defined as Kassoc(C70)/Kassoc(C60) is 25, which highlights the better size, shape, and electronic properties of C70 to be coordinated in the cage cavity. These authors also did DFT calculations, which showed that several CH···π contacts played an important role in complex formation and predicted higher interaction energy with C70 compared to C60 (Figure 6).

16Li and co-workers reported the one-pot synthesis of covalent organic cages from 1,3,5-tris(2,4-dimethoxyphenyl)benzene 21 monomer and paraformaldehyde or isobutyraldehyde to yield the dimeric cage 22 with a 52% yield and the tetrameric cage 23 with a 46% yield, respectively. The [2]cage 22 has a cavity composed of three nearly identical buckets with a size of 10.5 Å × 9.4 Å and an additional central cavity of 9.4 Å × 5.7 Å. The [2]cage 22 was used as the stationary phase in gas chromatography, which allowed the separation of benzene/cyclohexane and toluene/methylcyclohexane mixtures (Figure 7).

17At this point, the different examples of molecular cages presented above in this section consist of sp2 or sp2–sp hybridized carbons. In addition to these structures, Wu and co-workers prepared a cage structure with 3D global aromaticity in an entirely conjugated diradicaloid. The cage-formation step involved a Yamamoto homocoupling of precursor 24 to yield nonaromatic cage 25 with a yield of 21%, which was oxidized using DDQ to the final conjugated cage 26 with a yield of 21%. The obtained cage showed 3D global aromaticity/antiaromaticity behavior depending on the number of delocalized π electrons and the spin state. For the prepared cages, the neutral compound displayed an open-shell singlet ground state with the 38π monocyclic conjugation pathway showing aromaticity; the dication had a triplet ground state with a dominant 36π monocyclic conjugation pathway displaying aromaticity, the tetracation had 52 π-electrons (open-shell singlet) delocalized along the 3D rigid framework showing 3D global antiaromaticity, and the hexacation possesses D3 symmetry with 50 delocalized π-electrons presenting global aromaticity (Figure 8).

18By a similar approach, Wu and co-workers reported the first π-conjugated diradical molecular cage 27 in which two polychlorotriphenylmethyl (PTM) radicals were linked by three bis(3,6-carbazolyl) bridges. The two carbon-centered PTM radicals, which were separated by a long distance, were weakly coupled via electronic interactions through the carbazole spacers (see the resonant structures 27a–27c in Figure 9).

19Flood and co-workers used copper(I)-catalyzed alkyne–azide cycloaddition of precursors 28 and 29 to prepare cage 30 with a 15% yield. After purification, the X-ray crystal structure of 30 showed a complex with NaCl, and the chloride anion located in the cavity of the cage stabilized by nine hydrogen bonds, which indicates very high strong affinity (Figure 10a). Note that NaCl was not intentionally added, and the authors suggest that it was most probably scavenged from the silica used in the chromatographic purification of the cage. It was possible to remove 90% of the chloride from the cage by performing six extractions with deionized water. Host–guest experiments revealed nanomolar affinity toward chloride (Kassoc = ∼108 M–1) and lower affinity toward larger anions (Kassoc bromide > nitrate > iodide) and anti-Hofmeister behavior. These exceptional properties allowed the authors to use the cage for both salt extraction and corrosion inhibition. Santoyo-Gonzalez and co-workers performed the copper(I)-catalyzed alkyne–azide cycloaddition of precursors 31 and 32 in the presence of toluene as a template to obtain cages 33 with high yields (>75%), which highlights the efficiency of the followed synthetic approach (Figure 10b).

20Dynamic covalent chemistry (DCC) allows thermodynamic equilibrium, which is a concept that has been key to developing cage structures. The use of reversible bonds allows cage self-assembly from simple binary mixtures of building blocks, and it is also possible to prepare more complex structures by employing the self-sorting of complex mixtures containing three building blocks or more. If self-sorting takes place by a narcissistic pathway, no complexity is introduced into the cage system, and the same products obtained from binary mixtures are generated. In contrast, if self-sorting occurs with a social or integrative pathway, it is possible to achieve complex cage structures. In social self-sorting, different building blocks with the same reactive groups are incorporated into the cage structure. In integrative self-sorting, building blocks with distinct reactive groups are incorporated into the cage structure. These different self-sorting methods are still very limited to date (Figure 11). Besides, once the cage is formed, the dynamic nature of cages can also generate dimerization in interlocked cages, transformation into distinct cage structures (transforming a cage into a another cage structure containing the same building blocks), exchange of components (transforming the cage structure to another cage with the same topology), and cage disassembly (breaking up a cage into its building block components).

21Of the various reversible bonds, imine bond formation is one of the most widely used in cage formation for a thermodynamically stable cage among the different possible cage derivatives to be prepared. Mastalerz and co-workers carefully studied imine bond dynamics in organic cage synthesis (see below). The literature generally describes that imine cages are thermodynamically controlled rather than kinetically controlled products. This criterion has been generally used to predict cage stability by molecular modeling, which assumes that self-correction mechanisms will take place until the thermodynamic equilibrium is reached and, therefore, the stablest cage structure is formed. This assumption is generally true because imine bonds are chemically labile. In some cases, however, the outcome of the reaction is the kinetic product of the reaction.

22On the one hand, the solvent in which the self-assembly reaction is performed plays an important role. For instance, an important key parameter is cage solubility, which is highly solvent-dependent. The reaction of 34 and 35 results in cage 36, which precipitates from the reaction mixture in MeOH, EtOH, MeCN, CHCl3, THF, and dioxane, whereas it remains in solution in CH2Cl2 and toluene (Table 1). In contrast, cage 37 is more soluble and only precipitates in MeCN, THF, and dioxane. The same reactions (Figure 12 top) were also performed in the presence of 2 mol % trifluoroacetic acid (TFA) as a catalyst. The obtained results are comparable to those obtained without acid, with higher yields except for methanol. To determine imine cage formation dynamics, the scrambling of [2 + 3] imine cages with their deuterated analogues under different conditions like those reported in Table 1 and Table 2 was studied (Figure 12 bottom). For cage 36, an exchange was observed only in CH2Cl2 after 7 days. Even after adding water or catalytic amounts of acid to the reaction mixture to facilitate imine hydrolysis in solvents other than CH2Cl2, no scrambling was observed. So it can be concluded that reasonable cage solubility, as observed in CH2Cl2, is necessary for dynamic exchange. Moreover, although the cage is soluble in toluene, no dynamic exchange was observed in this solvent, just as no exchange was observed in MeOH, EtOH, MeCN, CHCl3, THF, and dioxane, for which the cage had limited solubility. For cage 37, no exchange was observed in any solvent under the same conditions, and it can be concluded that the aliphatic imine bonds in cage 37 were more stable than the aromatic imine bonds in cage 36. It is necessary to add acid to produce the scrambling of cage 37 in MeCN or EtOH. Exchange experiments of cages with deuterated aldehyde building blocks and opposite amines provide further information systems’ dynamics and stability (Figure 12).

23To study the dynamics of imine cage formation for larger cage systems, it is necessary to use solubilizing groups to solubilize cage and reaction intermediates. Along these lines, Mastalerz and co-workers introduced three deuterated and nondeuterated solubilizing n-hexyloxy groups into triptycene triamine 38, which was reacted with dialdehydes 39–42 and resulted in cages 43–46 (Figure 13). Cages and intermediates were fully soluble, which allowed cage formation kinetics to be explored. Whereas the [4 + 6] endo cages 45 and 46 quickly formed (approximately 1 h), the [4 + 4] cage 43 formed slowly, even at 150 °C. The authors found in the reaction mixture that larger species (oligomers and polymers) were formed, which were converted into the thermodynamically controlled [4 + 4] cage product. These results fall in line with scrambling experiments, which revealed that the cubic [4 + 4] cage 43 and its deuterated analogue did not undergo exchange. In contrast, the [4 + 6] exofunctionalized cage 44 scrambled in the presence of catalytic amounts of TFA or p-toluidine. The [4 + 6] endo cages 45 and 46 showed only significant exchange with p-toluidine and TFA and slow exchange in the presence of TFA and water (Figure 13).

24Another key aspect that defines the success or failure in the synthesis of organic cages is the geometry and rigidity of building blocks, which must have the appropriate conformation and preorganization. Mastalerz and co-workers studied the influence of building block conformational rigidity on the formation of a [4 + 4] imine cage with a truncated tetrahedral geometry over time (Figure 14). The rotational freedom of the Csp2–Csp3 bonds in at least one of the precursors needs to be highly restricted for the formation of [4 + 4] imine cage. The preorganization of reacting groups is crucial if cage formation is kinetically controlled. The stablest conformation of 35a has alternating substituents by placing all the methylamines on one side of the benzene ring with a very large methylamine Csp2–Csp3 rotational barrier (+227.7 kcal/mol). In contrast, 35b has a very small rotation barrier (+3.5 kcal/mol), which results in fast rotation under reaction conditions and, therefore, amine groups are not preorganized. For the 47a building block, the most stable conformation contains aldehyde groups slightly out of plane (dihedral angle ∼40°) by the steric strain of ethyl moieties. However, this preorganization is not favorable because C=O bonds are not orthogonally oriented to the molecular plane. The rotational barrier for 47a is +34.5 kcal/mol. The most stable conformation for 47b is that with all the aldehyde groups in the molecular plane. The rotational barrier is +42.4 kcal/mol.

25As a consequence of the non-appropriate preorganization of building blocks, it is not surprising that the reaction between 35b and 47b does not produce the corresponding 48d cage and yields polymeric byproducts. The reaction of 35b and 47a also produces polymeric byproducts rather than cage 48c. In contrast, the reaction of the favorable preorganized 35a building block with 47b yields cage 48b (27% yield) in acetonitrile at room temperature in 3 days in the absence of catalytic trifluoroacetic acid. This scenario suggests that cage formation is kinetically controlled, and thermodynamic products are the least structurally ordered polymers to form in the presence of catalytic acid. The reaction of 35a and 47a also yields the corresponding 48a cage with a 46% yield. The solid-state structures of 48a and 48b show that cage windows are blocked by alkyl chains. Activation of solids by the exchange first with n-pentane and then with liquid ethane results in porous solids with BET surface areas of 443 m2/g for cage 48b and 71 m2/g for cage 48a (Figure 14).

26Subtle differences in building block size/geometry can result in completely different cage structures, as shown in the following example. Stoddart, Cram, and co-workers demonstrated that the reaction of tetraformylcavitand 49 with 1,3-phenylenediamine under thermodynamic control quantitatively yielded the corresponding hemicarcerand formed by two cavitands connected together with four diamines through eight imine bonds., On the basis of this cage motif, Warmuth and co-workers reported the synthesis of octahedral nanocage 50 by the condensation of a tetraformylcavitand and ethylene-1,2-diamine catalyzed by trifluoroacetic acid in CHCl3 with an 82% yield. This cage structure was formed by six cavitands, connected together with 12 diamino bridging units through 24 imine bonds, and its cavity size was approximately 1700 Å3 (Figure 15). When the authors performed the reaction using 1,3-diaminopropane or 1,4-diaminobutane, they obtained hemicarcerand 51 with a yield over 95%. It has the same topology as that obtained by Cram when performing the same reaction with 1,3-phenylenediamine.

27An analysis of the molecular models of the octahedral cage 50 structure shows that the ethylenediamine units are forced into a gauche conformation. In contrast, they are in an anti conformation in hemicarcerand 51 by minimizing the linker strain. This is supported by the fact that the same reaction with (R,R)-1,2-diaminocyclohexane, which has the amino groups in a trans conformation, only produces polymeric products and no hemicarcerand is detected, which is expected given the unfavorable conformation of this diamine. The authors also found that using CHCl3 as the solvent is key for obtaining the octahedral cage. Thus by performing the reaction in CH2ClCH2Cl or CHCl2CHCl2, it resulted in a cage formation yield below 20%. Overall, this simple efficient methodology demonstrates that it is possible to synthesize nanosized cage molecules with almost 20 components in a single step.

28Further studies by Warmuth and co-workers allowed determining the solvent effect in the self-assembly of the tetraformylcavitand 49 and 1,2-ethylenediamine building blocks. The authors found that the final outcome of the reaction could be fine-tuned by changing the solvent to predominately obtain tetrahedral cage 52 in THF, octahedral cage 50 in CHCl3 and square antiprismatic cage 51 in CH2Cl2. These results highlight the key role of the solvent in the cage self-assembly process, and the importance of solvent optimization for maximizing cage formation yield (Figure 16).

29In another example, solvent effects are also important and define the CC1 cage formation outcome. Cooper and co-workers described an improved multigram synthesis for the CC1 [4 + 6] imine cage by obtaining the cage with yields close to 100%, whereas the original synthesis reported a yield of approximately 35% (Figure 17)., The first reported synthesis of cage CC1 consisted in performing the reaction by direct crystallization via the slow vapor diffusion of solid aldehyde 47b in a concentrated solution containing ethylenediamine, whereas the aldehyde solution is slowly added to the amine solution in the new approach. Solvent screening allowed CH2Cl2 to be selected as the optimal solvent (Table 3). In addition, optimizing the concentration of reagents allowed isolation of the cage with a 96% yield. Above a critical benzene-1,3,5-tricarboxaldehyde (23.13 mM) concentration, conversion decreases from 100% to 98%, and product isolation becomes more difficult. This synthetic strategy allows the synthesis of cages with different properties by the modification of building blocks to obtain self-assembled porous molecular materials for selective guest binding in either the crystalline or amorphous solid state.

30Besides imine bond formation, other reversible reactions also allow to obtain cage structures. For example, alkyne metathesis is a reversible reaction that can be used to prepare cages, which generally provides thermodynamically stable products upon equilibrium. Moore and co-workers used precursor 54 for the synthesis of tetrahedral cage 55 using alkyne metathesis with high yields. The authors suggest that the formed cages are thermodynamically very stable and create large kinetic barriers to prevent cage disassembly. Using tetrahedral cage 55, Moore, Zhang, and co-workers made postsynthetic modifications to obtain organic cages with alkynyl (55), alkenyl (56), and alkyl (57) edges. This allowed them to study the effects of building block shape persistence, as well as the building blocks’ rigidity/flexibility effects, on cage porosity. Once cages had been prepared, both the rapidly crystallized and slowly crystallized solids were studied by nitrogen adsorption experiments, which illustrated a general porosity trend: alkynyl (55) > alkenyl (56) > alkyl (57). Additional molecular dynamics calculations demonstrated that the microporosity of the molecular cages directly correlated with their relative shape persistence (Figure 18).

31Alkyne metathesis is also compatible with imine bonds by allowing complex cage structures to be prepared. Moore and co-workers reported the first system to combine orthogonal alkyne and imine bonds to prepare a molecular cage with a reversibly removable vertex. Monitoring the cage formation reaction from precursor 58 with [Mo] catalyst by GPC showed that higher-molecular-weight oligomeric products are initially formed, which reversibly corrected to form tetrahedral cage 59 as the favored reaction product. The imine-linked vertex can be removed through an Sc(III)-catalyzed transamination to produce cage disassembly by forming compounds 60 and 61. Cage reassembly can be performed by removing volatile methylamine and redissolving the residue in the presence of 5 Å molecular sieves to yield the initial organic cage. The authors pointed out that the developed method offered a strategy for preparing cages of altered symmetry and one that provides new possibilities to prepare cages that can be selectively modified (Figure 19).

32The reversible reaction between resorcinol and α,ω-alkanedials can also be used to prepare stable cages. The first example using this reaction for cage formation was reported by Nishikubo and co-workers, who prepared a cage structure from the condensation of resorcinol and 1,5-pentanedial whose structure (62) resembles a molecular waterwheel. Cage 62 is obtained with an 83% yield through a mechanism that involves the reaction of resorcinol and 1,5-pentanedial to initially yield oligomeric and polymeric species, which evolve to the final cage product with time. The cage showed high selectivity for the complexation of Rb+ versus Na+, K+, and Cs+, which was presumably encapsulated in the central hydrophobic cavity (Figure 20).

33Cages self-assembled by reversible bonds may have limited stability that can reduce their applications. This can be overcome by transforming the final reversible bonds formed in the cage into irreversible bonds., A typical example is the transformation by the reduction of reversible imine bonds to amino groups. In a more sophisticated study, Mastalerz and co-workers reported the transformation of imine cages 63 in only three steps involving cages 64 and 65 to yield the final hydrocarbon cages 66. This robust methodology allowed the preparation of an even larger cage structure C72H72, which was predicted but was impossible to synthesize until this methodology was developed. The main advantage of this methodology was the use of high-yielding imine bond formation in the first step to generate the cage scaffold, unlike the direct synthesis of hydrocarbon cages using an irreversible bond formation that normally results in low yields (Figure 21).

34Otte and co-workers described a robust synthetic protocol using amine and aldehyde building blocks to prepare heteroleptic cages. One example is described in Figure 22. It started from reactant 67, which was obtained by the transformation of one of the amino groups in triamine building block 35a into an azide moiety, which was further condensed with dialdehyde 34b to obtain intermediate 68, in which azide groups were further converted into amines to give intermediate 69. Condensation with another dialdehyde resulted in the formation of the final cage 70. Following a similar strategy, the authors prepared cages 71 and 72 (Figure 22).

35Whereas the synthesis of cages requires building blocks with the appropriate geometry and preorganization, in some instances the expected cage product is not obtained. As an alternative, different research groups have developed interesting methods using guest templates to obtain the target cage structure. With this general approach, Ballester and co-workers prepared chiral polyimine molecular capsules with polar interiors using N-oxides as templates for the efficient self-assembly of tetraaldehyde calix[4]pyrrole 73 with diamines in capsule 74 (Figure 23). Initial attempts to assemble capsules from their building blocks yielded insoluble precipitate materials that precluded capsule formation. The same results were obtained when N-oxides 75 or 76 were used as templates. In contrast, the addition of one equivalent of bispyridyl-N-oxide templates 77a or 77b resulted in a quantitative assembly of capsule 74. The template formed hydrogen bonds with two tetraaldehyde calix[4]pyrrole units in 73 and brought them to close proximity by favoring capsule formation by the reaction with the diamine and by avoiding undesired oligomerization reactions (Figure 23). Further examples include the templated formation of hydrogen-bonded capsules, which are described in detail in section 3.2.3.

36Using the same templating bispyridyl-N-oxide motifs, Ballester and his team used tetraurea calix[4]pyrrole 78 to form the corresponding dimeric tetraurea calix[4]pyrrole capsules. The capsule was formed reversibly by a cyclic array of 16 hydrogen bonds in nonpolar solvents in the presence of an appropriate template (4,4′-bipyridine-bis-N-oxide 76). The template was required to obtain the capsule because no evidence of cage formation was observed in the absence of the template. The template significantly stabilized capsule assembly by the formation of hydrogen bonds between N-oxide oxygen atoms and the calixpyrrole (Figure 24)., By changing one hydrogen atom for a methyl group at each benzylic carbon, four stereogenic centers with the same chirality were created. These chiral groups produced cyclochiral diastereoisomers, designated as S,P-78b and S,M-78b or R,P-78b and R,M-78b, depending on the configuration of the chiral group. Using an enantiomerically pure calixpyrrole in the presence of 4,4′-bipyridine-bis-N-oxide 76, capsule self-assembly only took place with conformationally cyclochiral enantiomers ((S,P·S,M)-(78b)2 or (R,P·R,M)-(78b)2) with urea groups rotated 180° to favor hydrogen bonding. In contrast, when the assembly was performed using racemic calixpyrrole in the presence of 76, a mixture of kinetically trapped capsule products following an approximate statistical mixture was obtained. Interestingly, the mixture evolved to the equilibrium by the addition of a small amount of hydrogen-bonding solvent THF to yield a single heterodimeric capsule (S,M·R,P)-(78b)2. When self-assembly was performed in the presence of trimethylamine-N-oxide, the single heterodimeric capsule (S,M·R,P)-(78b)2 was obtained even in the absence of THF.

37On the basis of these systems, a light-controlled cage assembly/disassembly of the above-described hydrogen-bonded cages was possible by introducing photoactive chemical units into the calix[4]pyrrole building block. Functionalization of the calix[4]pyrrole with azobenzene groups allowed azobenzene light-induced trans-to-cis-isomerization upon irradiation, which resulted in the disassembly of cage 76⊂(79)2. The authors suggested that the isomerization of only one azo group probably sufficed to prompt capsule disruption. Capsule reassembly was achieved by reverse cis-to-trans relaxation in the dark. As described above in the parent system, the presence of template bipyridine bis-N-oxide 76 was also neceassary here to observe capsule formation (Figure 25).

38Besides templated hydrogen bonding, templated imine chemistry has also been used to favor the synthesis of certain cages. For instance, Warmuth and co-workers used cavitand 80 and ethylene diamine to form water-soluble dynamic hexaimine cryptophane 81, which was assembled in aqueous media in the presence of different templates. The cage formation mechanism takes advantage of the binding of the cryptophane derivative block to a range of guests, with van der Waals volumes ranging from 57 Å3 (CH2Cl2) to 118 Å3 (choline). The binding versatility of cryptophane was attributed to its ability to adapt to the size of the guest by wrapping/unwrapping motion around the polar C3 axis. The maximum binding strength was observed for CHCl3 (Kassoc = 8000 M–1). The guests that are too large in one dimension or more (e.g., acetophenone, toluene or [N(n-Bu)4]+) or are too reactive (i.e., 2,3-cyclohexenone) did not template the formation of cryptophane 81 (Figure 26).

39The postassembly modification of metal–organic assemblies provides an efficient strategy for synthesizing pure covalent organic cages that are otherwise difficult to synthesize by conventional direct methods. Han, Sun, and co-workers used macrocyclic precursor 82 to prepare cage 83, which cannot be obtained by a direct reaction using a metal template strategy (Figure 27a). Synthesis involves the formation of a silver(I) carbene cage 82·Ag from 82 with a 89% yield. Complex 82·Ag is irradiated with a high-pressure mercury lamp to produce a quantitative photochemical, a [2 + 2] cycloaddition reaction, to yield metal–organic cage 83·Ag. The demetallization of the complex generates covalent cage 83 with a 78% yield. The authors proved the versatility of this method for the synthesis of multiple cage structures, including cages 84 and 85 as representative examples (Figure 27b).,

40Cage structures usually contain a large number of nonpolar groups that result in hydrophobic structures, which are soluble in organic solvents, but are not water-soluble in most cases. However, water insolubility limits the potential application of cages in aqueous media. One possibility to overcome this problem is to incorporate water-solubilizing groups, which are usually anchored to outer cage walls.

41Following this approach, Verboom, Reinhoudt, and co-workers synthesized and determined the solubility of cavitand 86 by incorporating a wide range of water-solubilizing groups, including a series of charged groups and neutral hydrophilic tetraethylene glycol chains. The solubility of neutral cavitands with ethylene glycol increases with the number of ethylene glycol chains and decreases with temperature. The cavitands containing ammonium groups and lower-rim pentyl groups have the highest water solubility as a result of an amphiphilic character. The cavitands with pyridinium substituents and lower-rim methyl, hydroxypropyl, and propyl acetate substituents are still soluble in water but to a much lesser extent (Figure 28). Scherman, Bare, and Nikan achieved water solubility in the same cavitand motif structure by using phosphate groups in the R1 position and produced octa-anionic species.

42Gibb and co-workers employed different cationic, anionic, and uncharged water-solubilizing groups to solubilize in water deep cavitand 87 (Figure 29). Additional examples of water-solubilizing groups (including carboxylates, ammonium groups, pyridinium groups, etc.) used in cavitands are described in Section 4 of this review.

43Computational approaches and molecular modeling have been used to predict cage structure and topology, cavity size, and related properties of large data sets of porous organic cages to guide synthetic researchers toward materials with the required properties. In fact the integration of computational and experimental fields is a promising tool that can significantly accelerate the discovery of new cage systems., In particular, rapid screening can provide a methodology in organic cages that could significantly help to develop materials with specific characteristics, whereas evolutionary algorithms can be used to determine the required size of precursors to achieve a given cage cavity size and properties.

44Of the different parameters that can be modeled, the shape persistence of organic cages can be predicted by modeling. To achieve this goal, Jelfs and co-workers prepared a large data set of 63472 organic cages with many topologies by employing an automated computational assembly from building blocks through several chemical reactions. The authors found that the cages obtained by imine condensation of trialdehydes and diamines in a [4 + 6] reaction were the most likely ones to form shape persistent cages. In contrast, thiol reactions were most likely to give collapsed cages. Using this library and machine-learning models, the authors were able to predict shape persistence with an accuracy of up to 93%.

45One of the main applications of porous materials based on organic cages is the separation of gas molecules. During this process, the diffusion of gases through the material is a key aspect that can also be modeled computationally based on the packing of molecular cages. In fact the solid-state porosity of organic cages is directly related to the packing of individual cages and the connectivity between cages because interconnected channels are necessary for guest molecules to pass through the material.

46On the basis of the core structure of cage CC1, it is possible to obtain a large family of isostructural cages (see structures CCn, 88–93 in Figure 30) by changing the diamine building block. On the basis of this concept, by means of density functional theory (DFT) calculations and molecular dynamics (MD) simulations, Haranczyk, Bernabei, and co-workers investigated the thermodynamic stability of cages CCn, which have been synthesized and their crystallographic data are summarized in Table 4. These cages have pore-limiting diameters within the 1.67–3.97 Å range, whose porosity is defined mainly by the diamine linker, and by the solvent used in the crystallization because it can yield different crystal polymorphs with a distinct porosity. By searching the PubChem3D database, which has ca. 80 million structures, the authors selected a set of six vicinal diamines that form porous organic cages with benzene-1,3,5-tricarboxaldehyde (88–93). The computational calculation allowed the determination of the solid-state crystal structure of cages 88–93 by theoretical calculations (note that cages 88–93 were predicted computation structures and were not, therefore, synthesized in a chemistry laboratory). The analysis of the resulting energy–structure–porosity maps showed structures with unusually high porosity. This computational approach constitutes the first step toward the high-throughput screening of porous organic cages for the assembly of functional materials (Figure 30).

47Cooper and co-workers used molecular dynamics simulations to calculate the diffusion of six small gas molecules (H2, N2, CO2, CH4, Kr, and Xe) in flexible crystalline porous organic cage CC3. The experimental determination of the porosity of CC3 (Figure 31) revealed a BET surface area of 624 m2/g (N2, 77 K). The modeling showed that the cage in the solid state has a dynamic cavity that undergoes size and shape fluctuations by playing a key role in the diffusion of molecules through solid material. This would explain that large gas molecules Kr and Xe, whose size was larger than the cage pore-limiting diameter (3.62 Å), could pass through. The simulations also predicted that the larger gas molecule SF6 was unable to diffuse in this materials’ pores (Figure 31). This method illustrates the consequences of cage flexibility, particularly the cage’s flexibility in the solid state, for the system’s porosity, rather than intermolecular displacements.

48Whereas most of the materials based on organic cages, i.e., solid materials formed by organic cages, are built using cages with the same chirality. The materials built with cages with a different chirality are much less known. Sholl and co-workers performed molecular simulations of the CC3-racemic material to determine its structural properties because it is challenging to experimentally obtain the structure of this racemic crystal. They determined from the calculations that the composition of CC3-racemic material is made up of four distinct molecular species as so: two homochiral cages (CC3-R and CC3-S) and two heterochiral cages (CC3-RS and CC3-SR) in a minor proportion that varied from 8% to 28%. This computational information is valuable because it is not experimentally possible to obtain this information due to the differences that the distinct cage isomers produce in X-ray powder diffraction are too small to be quantified (Figure 32).

49Cooper, Jelfs and co-workers developed a hybrid discovery workflow that fuses computation with robotic synthesis to find new organic cage molecules. These authors studied 78 building block combinations (see structures 34, 35, 47b, and 94–103 in Figure 33), which allowed 33 cages (32 of which were new) to be synthesized using a robot synthesizer in a one-pot reaction with no work up (see structures 104–106 in Figure 33). To increase the chances of obtaining a target cage predicted by the calculation, the cage must be strongly favored energetically. If cage formation is favored by a small energy quantity, other experimental factors, such as solvent stabilization, can change the outcome predicted by simulation and allow cages to be experimentally prepared that are not favored according to the simulation. This method is powerful because it helps to focus experiments by identifying linkers that are more reliable for cage formation. One example of building blocks and the corresponding cages is described in Figure 33. Cage structure 105 can be tuned by replacing terephthalaldehyde at any given ratio with electron-deficient and highly reactive tetrafluoroterephthalaldehyde. This substitution increases the stability and crystallinity of the obtained material in a stepwise manner without influencing crystal packing. This methodology can be used to fine-tune the cavity and to precisely control guest encapsulation selectivity.

50Jelfs, Cooper, Day, and co-workers reported a powerful hybrid workflow that merges computations with experiments to guide the targeted discovery of new sophisticated supramolecular materials. This methodology allows the social self-sorting of a mixture of three components (triamine, diamine, trialdehyde; see Figure 34) to be predicted, unlike the methodology described in Figure 33, which involves only two components. The authors used this methodology to prepare organic cage pots 107 from three different components without exploiting orthogonal reactivities to direct the assembly. For this purpose, initial computational modeling was used to predict the synthetic viability of the organic cage pots by comparing the formation energies of the socially self-sorted organic cage pots and the formation energies of the separate binary organic cages, which would be obtained if narcissistic self-sorting happened (Figure 34). The authors indicated that these chiral organic cage pots have analogies to hemicryptophanes and, therefore, suggested that they have the potential to form guest–host complexes to encapsulate target guests.

51Slater and co-workers used computational modeling to rationalize the outcomes of the self-assembly of 1,2-cyclohexanediamine with a series of tetra-aldehydes 108–112 of different sizes and shapes. Experimentally, aldehyde 108a did not self-assemble into high-symmetry cages but generated low-symmetry socially self-sorted heteroleptic cages [111 + 108a] (cage 113) and [111 + 108b] and [111 + 109] (cage 114). The computational modeling results agreed with the experiments because they predicted that these heteroleptic cages were more stable than the corresponding homoleptic cages (Figure 35).

52Although organic cages have been used as building blocks to prepare mechanically interlocked molecules, cage⊂cage complexes have not yet been synthesized. These interesting systems would have a central cavity whose size would correspond to the size of the guest cage cavity, which would allow confined space applications. The power of the molecular modeling methodologies developed by Jelfs and co-workers permitted them to model complexes in which one guest cage encapsulated into a host-cage cavity yielded a nested organic cage complex. The authors believed that the predicted feasible structures would help synthetic chemists to obtain the first nested cage structure.

53This section describes examples of cages and containers soluble in organic solvents and also includes solid materials based on molecular cages. The section is divided into different subsections based on the type of container and on the type of reaction employed to self-assemble the cage. Cage properties are described in detail by focusing on the several geometries with the distinct cavity sizes that can be obtained by using slightly different building blocks. Most of the examples described in this section have been used as host systems for the encapsulation of different guest molecules, and their host–guest properties are also described. In some examples, cages have been employed to host gas molecules. The corresponding gas sorption experiments are also described, especially examples that link together porosity in the solid state with the flexibility and rigidity of cage molecules; i.e., rigid cages display solid-state porosity and flexible cages are not porous in the solid state.

54The term cavitand was formulated by Cram in 1982 to define molecules with a cavity for hosting guest molecules. Since then, the expansion of this field has given rise to the development of numerous cavitands types.− Such systems have allowed the preparation of carcerands and hemicarcerands by putting the convergent geometry of cavitands to good use, which makes them ideal building blocks to prepare hollow molecular structures., Of the different structures, calixarenes are one of the most widely used cavitand motifs. One of the simplest carcerand structures (115) can be obtained by the condensation of two calixarenes 2 with two difunctional spacers (Figure 36).

55Hemicarcerand structures based on resorcinarenes present a combination of rigid resorcinarene moiety and linker building blocks, which can have different degrees of flexibility. Dyker and co-workers showed that resorcinarene hemicarcerand 116 can adapt its shape and size according to guests’ steric demand by aligning its guests to match the complementary electrostatic potential. By way of example, the volume of the cavity in 116 changes from 197 to 154 Å3 when toluene and chloroform guests are, respectively, encapsulated (Figure 37).

56Martínez, Saaidi, and co-workers showed that subtle differences in the functional groups that envelope the cavity of cages 117 and 118 play a key role in defining affinity to encapsulate a certain guest. This is nicely demonstrated in the encapsulation of organochlorine insecticide chlordecone 119 by 117 and 118. Whereas hemicryptophane 117, which contains amide groups, displays a moderate binding constant of 126 M–1, receptor 118, which contains amine groups, shows a much higher association constant of 2.1 × 104 M–1. The structure of the latter complex is stabilized by the hydrogen bonding between the geminal diol of 119b and the nitrogen atom of cage amino groups and by the halogen−π interactions between the guest chlorine atoms and hemicryptophane aromatic rings (Figure 38).

57Cavitands that incorporate water-solubilizing groups have been widely used as molecular hosts in water to put the hydrophobic nature of the resulting cavity to good use. Numerous examples are included in section 4.1 of this review.

58Cryptands are also an important class of organic cages, which show high binding affinity to guest molecules, especially paraquat derivatives. These host–guest properties allow the formation of pseudorotaxanes,− rotaxanes, catenanes,, and polymers. The complexation of paraquat 120a by cryptand 121a results in a [3]pseudorotaxane host–guest complex, where the guest is encapsulated in the cavities of two cryptand molecules stabilized by multiple face-to-face π-stacking interactions. Using the same cryptand 121a and bisparaquat guest 120b, a [3]pseudorotaxane is formed by cooperative complexation with association constants K1 = 1.2 × 103 M–1 and K2 = 2.0 × 104 M–1 (Figure 39a)., Introducing a pyridine moiety into one of the bridges of 121a yields cryptand structure 121b. This cryptand forms a 1:1 host–guest complex with paraquat 120a (Figure 39b).− Cryptand 121c is obtained by introducing an azobenzene into the bridge. Trans and cis conformations are controlled by irradiation with UV light (to form the cis isomer) or visible light or by heating (to form the trans isomer). This isomerism allows switching the binding with 2,7-diazapyrenium 120c between OFF for the trans isomer and ON for the cis isomer (Figure 39c). The more elaborated geometry of host 122, which has two cryptand cavities, is able to encapsulate two paraquat molecules 120a, as determined by 1H NMR and the single-crystal X-ray structure, which highlights the versatility of cryptands in sophisticated host structures (Figure 39d).−

59As described in section 2.1, imine chemistry is one of the most well-established reversible reactions in cage synthesis. The versatility of imine bond formation from amines and aldehydes has allowed the preparation of numerous cage structures since the first reported example in 1991 by Cram and Quan.,, In this section, we describe representative examples of different families of organic cages based on imine chemistry as reported in the literature.

60Cages can be prepared with customized cavity volumes, from small to large, by changing the size of the building blocks that preserve the same cage structure type; for example, [2 + 3] cages from difunctional and trifunctional building blocks with complementary reactivity. Cages with very small cavities have been reported by Banerjee and co-workers, who performed a one-pot synthesis of amine-linked organic cages 125 from 123 and a diamine through cage intermediate 124 (Figure 40). The procedure additionally removed the complications associated with the postsynthetic modifications of imine-linked porous organic cages to obtain amine cages by reduction. In the prepared cage structures, an increasing chain length resulted in an odd–even alternation between the eclipsed and gauche conformations of cages by controlling cage conformation and cavity size. Synthesized cages 125 in the solid state had moderate porosity for 125d (202 m2 g–1) and 125c (87 m2 g–1), whereas 125a (17 m2 g–1) and 125b (23 m2 g–1) were nonporous. These cages had excellent and much better chemical stability in water, acids, and bases than imine-linked porous organic cages.

61Cages whose size was slightly larger had enough space to allocate small solvent molecules in their cavity. He and co-workers reported a [2 + 6] one-pot synthesis that involved m-phthalaldehyde 34c and hexakis-amine 126 building blocks to yield a superphane cage 127 by imine bond formation. The synthesis required heating to favor the formation of the cage product in DMSO. The cage cavity was able to encapsulate two water molecules to form water dimers stabilized by multiple hydrogen bonds with the cage (Figure 41).

62In contrast to the above examples, which show a relatively small cage volume, the use of larger building blocks allowed bigger cavity sizes to be acquired. For instance, Chen and co-workers used building blocks 35a and 128 to prepare cage 129 by imine bond formation. Reduction of imine bonds, deprotection of phenol hydroxyl groups and methylation of amine groups yielded the final cage 130 with three endohedral phenol hydroxyl groups. Cage 130 was used as catalyst in a Friedel–Crafts reaction and displayed a moderate catalytic effect (Figure 42).

63In addition to the selected examples of the above-described [2 + 3] cages, tris-primary amines, aliphatic and aromatic, have been extensively used to prepare cage structures. Some of these triamines are commercially available and others can be prepared by following the synthetic protocols in the literature. Zhang and co-workers showed that a complementary geometry between triamine 131 (planar) or 132 (pyramidal) and dialdehydes 133–138 is key for obtaining cage structures. These authors suggested that the angle and directionality of reactive groups are essential for achieving an alignment of the aldehyde and amino moieties in the building blocks with an appropriate angle. Additionally, rotational freedom in building blocks can reduce the angle strain established by cage formation and, hence, facilitate the synthesis of cages 139 and 140 (Figure 43).

64Further examples of [2 + 3] cages were prepared by Lehn and co-workers, who demonstrated that triamine “tren” (tris(2-aminoethyl)-amine) with different dialdehyde building blocks (34c, 34d, 95a, 97a, 141, 142, 143) showed significant self-sorting behavior in the formation of imine-based macrobicyclic cryptand-type organic cages (144–148). Using eight different dialdehyde building blocks, the authors prepared the corresponding cages. Afterward, self-sorting experiments run with one aldehyde building block or several revealed high selectivity toward the formation of some cages over others, which highlights the influence of the aldehyde structure. The presence of a heteroatom kinetically and thermodynamically favors the formation of cages 144. This is highlighted in a competitive self-sorting experiment of dialdehydes 34c and 34d with “tren”, which exclusively yields cage 144 with a 92% yield (Figure 44). The flexibility of components also plays a key role in self-sorting, as highlighted in the self-sorting of a mixture containing 34c, 143, and “tren”, which yielded the homoleptic cage 145 in a roughly quantitative yield. For other combinations of aldehydes, selective cage formation has not been observed; for example, in the case of a mixture of either 34d, 141a, and “tren” or 95a, 142, and “tren”, the formation of a mixture of two cages has been observed. More complex self-sorting experiments using a mixture of six dialdehyde components has pointed out that the presence of a heteroatom in the aldehyde aromatic ring favors cage formation. Additionally, intramolecular hydrogen-bonding interactions, and the antiparallel orientation of dipole groups, also play a key role in cage formation (Figure 44).

65By following the same approach based on using “tren” and different dialdehyde building blocks, Mukherjee and co-workers showed that intramolecular H-bonding plays a key role in the selective formation of imine organic cages. In particular, they used dialdehyde 149 and “tren” to prepare cage 150, and dialdehyde 151 and “tren” to prepare cage 152. The treatment of cage 150 with aldehyde 151, which contains phenolic groups, displaced aldehyde 149 from the original cage to yield cage 152. The authors found that the reaction was driven by intramolecular H-bonding stabilization (Figure 45).

66Mukherjee and co-workers employed related systems to report the selective formation of a single isomer of a [3 + 2] self-assembled organic cage from an unsymmetrical dialdehyde and the flexible triamine “tren”. The reaction of “tren” with aldehyde 153 exclusively formed single isomer 154b. In contrast, when “tren” reacted with aldehyde 155 under the same reaction conditions, a mixture of isomeric cages 156a and 156b was obtained. The same results were found if the reaction was performed from intermediate 157. The only difference between aldehydes 153 and 155 was one hydroxy group and, therefore, intramolecular hydrogen bonding with the nearby formyl or imine groups played a key role in the observed isomer selection differences. The experimental results indicate that the formation of a single cage isomer depends on the shape and size of the building aldehyde, while theoretical calculations suggest that the energy difference between isomers is the key factor in isomer selection (Figure 46).

67The previous examples show the formation of [3 + 2] cages by the reaction of di- and trifunctional building blocks of complementary reactivity. In addition to this topology, [4 + 4] tetrahedral cages can be obtained from trifunctional building blocks with complementary reactivity. The reaction of “tren” with trialdehyde building blocks with the appropriate geometry results in the formation of tetrahedral cages, with “tren” units placed in corners and trialdehyde moieties on the tetrahedron faces. By following this concept, Li and co-workers prepared tetrahedral cages and triangular prisms using a trialdehyde and “tren” and found that the self-assembly process was driven toward the products that maximized the intramolecular π–π and CH−π interactions. The conformational preferences of trialdehyde building blocks 102 and 158 also impacted the self-assembly outcome. While a twisted conformation (102b) favored the formation of tetrahedron 159, rigid and planar building blocks 158 favored the formation of prism 160. For trialdehyde 102a, which had more conformational freedom and could adopt both twisted and planar conformations, both tetrahedral 159 and prismatic 160 cages were obtained with good yields (Figure 47). Tetrahedral cage 159 efficiently encapsulated white phosphorus (P4) by reducing its intrinsic reactivity toward oxygen. In this complex, each phosphorus atom in P4 pointed toward the adjacent central 102b phenyl ring to produce a host–guest orbital overlap. This interaction could be responsible for the stability of the supramolecular complex.

68Li, Cao, Jiao, and co-workers adopted the same strategy to obtain a tetrahedral cage structure. The synthetic route used building block 161 to prepare polychlorotriphenylmethyl (PTM) radical units 162. This building block was reacted with “tren” to obtain the [4 + 4] tetrahedral cage 163. Cage formation was favored by the intramolecular CH···π and hydrogen-bonding interactions in the cage structure, which also favored the stability of the cage framework’s homochirality. In each cage structure, the PTM subunits adopted the same helical conformation (all P or all M), and the “tren” vertices all had the same clockwise (C) or anticlockwise (A) orientation. This resulted in chiral structures P4A4 or M4C4 cages. The magnetic measurements showed that each cage had 3.58 spins and the four PTM radicals present in the cage were weakly coupled through space with a distance between the two central PTM radical carbon atoms of 9.74 Å (Figure 48).

69Replacing “tren” with 1,3,5-tris(aminomethyl)-benzene 35a and performing the reaction with trialdehyde building blocks 47a–47c resulted in the formation of the [4 + 4] tetrahedral cages 48 (see Figure 14). Following this approach, Mastalerz and co-workers prepared shape-persistent tetrahedral imine cages 48, which had the same structure but different substituents, and all with similar cavity volumes and window sizes. They used these cages to study the encapsulation of tetra-n-alkylammonium cations., The thermodynamic and kinetic results showed that guest uptake was highly solvent-dependent and suggested that it was favored by a squeezing mechanism, in which the cage remained intact and the guest entered the cage cavity through the window rather than by a gate-opening mechanism through imine bond cleavage. Guest encapsulation involved folded chain conformations, and guest packing coefficients of >65%, led to restricted guest movement in the cage cavity (Figure 49).

70The different properties of “tren” and 1,3,5-tris(aminomethyl)-benzene 35c have been used to selectively self-sort cage formation with dialdehyde building blocks 164 and 165 by Mukherjee and co-workers. Of a mixture of triamines “tren” and 35c, and dialdehydes 164 and 165, only two cages formed (166 and 167). In contrast, when components were separately assembled, it was possible to obtain the four corresponding different cages. The two nonpreferred cages could be transformed into the two preferred combinations by reacting them with the appropriate triamine or dialdehyde (Figure 50 shows an example of the transformation of cage 168). The authors indicated that the behavior of these dynamic systems could hardly be rationalized from the individual component’s behavior (Figure 50).

71On the basis of the [2 + 3] cage shapes described in Figure 50, with a suitable cavity to encapsulate guest molecules, Mukherjee and co-workers prepared fluorescent cage 169 to detect explosive picric acid (170). Sensing of picric acid took place by quenching the cage’s fluorescence emission, which was reduced by 94%. Picric acid encapsulation allowed this explosive to be detected at a concentration as low as 6.4 ppb. Guest binding took place by the secondary amine groups of the cage that interacted with the phenol group of picric acid. This interaction resulted in the proton transfer from the acidic phenolic OH group to the basic amino group to form a strong ion pair picrate–protonated cage. The probe was selective for picric acid over other nitro-aromatic compounds, such as 2,4-dinitrophenol, 4-nitrophenol, trinitrotoluene, 2,4-dinitrotoluene, 3,4-dinitrotoluene, 4-nitrotoluene, 4-nitrobenzoic acid, 1,3-dinitrobenzene, and nitrobenzene (Figure 51).

72Several authors have obtained chiral cages using appropriate chiral building blocks. For instance, Šolomek, Coskun, and co-workers reported the synthesis of chiral [2 + 3] imine cages using three built-in rylene units by means of imine bonds from precursors 171. The naphthalene-1,4:5,8-bis(dicarboximide) 172 (NDI) and pyromellitic diimide 173 (PMDI) cages showed good selectivity for CO2 adsorption over N2 and CH4, whereas the cage containing perylene-3,4:9,10-bis(dicarboximide) (PDI) could not be synthesized on a large scale to perform gas adsorption studies. In addition to these experiments, the authors found that cage 174 displayed efficient delayed fluorescence emission, which was attributed to the precise spatial arrangement of the PDI units in the cage structure. The spectroscopic and electrochemical data suggested that a rapid (pseudo)equilibrium was struck between the singlet excited state of 174 and the intracage charge-separated state. The repopulation of the bright singlet excited state by radical ions recombination produced the observed delayed fluorescence (Figure 52). This work demonstrates that rylene cages provide a precise spatial arrangement of PDI moieties, which is an important requirement for electronic materials by allowing electronic communication between them. The authors point out that these cage systems can be used in the future to develop new porous materials with unique photochemical and electronic properties.

73On the basis of the same amino chiral building blocks, but by changing trialdehyde to tetraaldehyde, Zhang, Ke, and co-workers prepared enantiopure [4 + 2] imine cages using perylene diimide 171c and electron-rich TFBE building block 175 to yield cage 176 with a cavity size of ca. 1.5 × 0.9 × 0.8 nm3. Cage (R,R,R,R)-176 encapsulated from planar polycyclic aromatic hydrocarbons (PAHs) to fullerene guest molecules (Figure 53). Molecular modeling indicated that PAHs molecules were near the cage’s PDI cores, and each PDI-TFBE cage could accommodate 10 guest perylene molecules. The visible-light-driven Smiles rearrangement of 2-aryloxybenzoic acids to aryl salicylates with good efficiency was achieved in the presence of the imine cage that was used as a catalyst. The authors suggested that the reaction took place by a single-electron transfer (SET) between the excited (R,R,R,R)-176* cage and the carboxylate group from the substrate (Figure 53).

74Warmuth and co-workers developed an approach using imine bond formation to prepare multicomponent nanometer-sized octahedrons. Octahedral capsules (Cy6Lx12 and Cy6Tx8) products were obtained from tetra-formyl resorcinarene cavitands 177 and amine derivatives (ethylenediamine, p-benzenediamine and compounds 178–180). To obtain this cage structure, it was crucial that the cage building blocks had a high degree of conformational rigidity with the appropriate geometry to exclude other possible capsule products to maximize selectivity. Variation in diamine length for octahedrons or the size of the triamine for rhombicuboctahedrons allowed cavity size to be customized to obtain large cavity volumes ranging from 7400 Å3 to 13000 Å3. These volumes were on a scale of sizes of smaller globular proteins or other biomacromolecules, and the authors suggested that these cages could find biomedical, biochemical, or materials science applications (Figure 54)., One advantage of these systems was their robust synthetic methodology, which allowed scaling up synthesis to a multigram scale. For example, the synthesis of the octahedral nanocontainer with an ethylenediamine spacer was obtained on a 5 g scale. In addition to the symmetric capsules described in Figure 54, the same research group prepared a chiral capsule using a cavitand with less symmetry. The obtained chiral capsule had a large enantiomerization free energy barrier of 21.5 kcal/mol.

75By means of the resorcinarene core motif 177b, Yuan and co-workers prepared different imine cage structures with cavity volumes ranging from 360 Å3 to 11 200 Å3. Synthesis involved tetraformylresorcin[4]arene cavitand 177b with different diamine linkers (1,2-ethylenediamine, 1,5-pentanediamine, 4-methyl-1,3-phenylenediamine, 4,4′′-diaminoterphenyl, etc.). The authors prepared the [2 + 4] dimeric cages 181 and 182, the [3 + 6] trimeric triangular prism cage 183, and the [6 + 12] hexameric octahedral cage 184 (Figure 55), as well as several isostructural cages (not shown in Figure 55). The cavity of these systems was formed by π-rich aromatic rings with surrounding phenol groups along with imine groups. The gas adsorption measurements in the solid state suggested that the prepared cages were all porous materials with BET values above 1000 m2 g–1. These solids exhibited solvatomorphism, which resulted from changes in their crystallographic packing in the solid state. These changes in solid-state packing modified the materials’ gas sorption properties, which have been used to separate ethane from an ethane/ethylene mixture using the cage formed by cavitand 177b and p-phenylenediamine.

76Kim, Baik, and co-workers used tetramino-porphyrin 185 and dialdehyde building block 186 to prepare the gigantic porphyrinic organic cage structure 187. This cage has a P12L24 topology cavity with a diameter of 4.3 nm. The authors managed to synthesize the cage after numerous attempts, and the main challenges were the solubility of systems, and the formation of both polymeric byproducts and small-sized cages that were entropically favored. The X-ray crystal structure showed that 187 has the largest cavity in a pure organic cage reported to date (Figure 56).

77Pan, Huang, and co-workers described the [3 + 6] condensation of tetraaldehyde 188 with 1,2-cyclohexanediamine to form imine cage 189. Crystallization of the cage in the presence of excess NaN3 produced the coordination of Na+ to the cage and resulted in a hierarchically porous structure with interconnected channels and excellent CO2 storage capacity (Figure 57 top). Jiang, Chen, Wang, and co-workers adopted a similar synthetic strategy with tetraaldehyde 190 and 1,2-cyclohexanediamine to prepare porphyrin cage 191 with good efficiency. This cage has been used as a high-efficacy singlet oxygen generation system (Figure 57 bottom).

78Mastalerz and co-workers applied a [3 + 6] imine condensation of triptycene tetraaldehydes 192 with 1,2-cyclohexyl-diamines to prepare pumpkin-shaped cages 193 and 194 with concave walls, named cucurbitimines, for resembling cucurbiturils. Besides the [3 + 6] 193 cage, which was obtained with a ca. 50% yield, the [4 + 8] cage 194 was also formed with only a 5% yield for both enantiopure (R,R)- and (S,S)-1,2-cyclohexyldiamines. The [3 + 6] cage 193 encapsulated neutral guests, such as CH2Cl2, THF, or toluene. Guest exchange was also studied, and the authors found that THF was exchanged by toluene on more than 2 days at room temperature or in less than 1 h at 100 °C (Figure 58).

79Zhang, Liu X., Liu Z., and co-workers employed precursors 47b and 195 to prepare catenanes 196 and 197. Catenane 198 is formed by two dissymmetric cages by means of a space-discriminative postassembly modification strategy using a voluminous reductant like NaBH(OAc)3 to selectively reduce the outer imine bonds. The authors calculated that the two possible direct syntheses offered a very low theoretical probability of cage formation after considering all the possible cages that could be obtained by the random assembly of building blocks. In particular, the theoretical probabilities of cage formation were 6.5% for the scrambling approach and 25% for the partially directed approach. In contrast, the postmodification approach was much more efficient given its 100% theoretical probability (Figure 59a). This method is based on the different environments around the imine bonds within catenane framework 196. The imine bonds in the inner cavity are protected by steric congestion, and the outer imine bonds are exposed to the solvent and are, therefore, accessible to react with voluminous reductant NaBH(OAc)3. In contrast, smaller reductant NaBH4 produces nonselective reduction by reducing both the inner and outer imine bonds. The authors point out that this synthetic strategy can be used to develop catenated cages with sophisticated topologies to obtain materials with advanced functionalities (Figure 59).

80Greenway and co-workers performed the social-sorting of mixtures containing tritopic amine 35c, ditopic aldehyde 34c, and tetra-topic aldehyde 112d, which resulted in a mixture of dumbbell cage 198 and cage 199 (Figure 60). This work provides a proof-of-concept for preparing complex systems containing many covalently bonded cages.

81Several authors have used cages to anchor nanoparticles in the internal cage and to obtain nanoparticles with a controlled narrow size distribution. For instance, Dong, Chen, and co-workers employed 2-hydroxy-1,3,5-triformylbenzene 200 and 1,2-cyclohexanediamine to prepare cage 201, which was loaded with Ag nanoparticles. This cage has the core structure of cage CC3 but changed the 1,3,5-triformylbenzene building block used to prepare cage CC3 by 2-hydroxy-1,3,5-triformylbenzene 200. The authors found that the N and O atoms in the cage templated the formation of the Ag nanoparticles in the cavity. The formation of nanoparticles in the cage was studied by HRTEM, which showed comparable homogeneous nanoparticle distribution to the cage shell shape. The formation of Ag nanoparticles in the cage cavity was also supported by loss of porosity from the gas adsorption–desorption experiments (Figure 61). Beaumont, Jiang, and co-workers prepared palladium nanoparticles using cage CC3. They demonstrated that the initial Pd(NO3)2 encapsulation in the cage cavity was a key factor for the formation of the ultrasmall nanoparticles obtained by the reduction with H2, with no chemical damage to the cage structure. In contrast, if the larger sized Pd(acac)2 metal precursor was used, no control in nanoparticle size was observed, and this was attributed to the impossibility of the precursor fitting inside the cage cavity. The selective binding of palladium ions in an imine cage with a tubular structure has also been used for the encapsulation of Pd nanoparticles inside the cavity of a cage obtained by the condensation of a benzo[c][1,2,5]thiadiazole tetraaldehyde derivative and 1,2-cyclohexanediamine.

82Gu and co-workers prepared diarylethene photochromic cage 202 (with an adjustable cavity size by irradiation) by the condensation of tren with a diarylethene-dialdehyde, followed by imine reduction. Cavity reduction took place with 254 nm UV light to yield the cage closed form, and cavity enlargement occurred with 550 nm visible light to yield the cage open form as described in Figure 62. The diameters of the cavity sizes of the open and closed forms were 1.17 and 0.91 nm, respectively. In addition, the authors found that the cage could coordinate Au(III), and after reduction with NaBH4, the Au NPs inside the cage cavity formed. The different microenvironments of the open/closed cages containing Au NPs had a direct effect on the catalytic activity of Au on the reduction of 4-nitrophenol with NaBH4. The observed apparent reaction kinetic constants were 4.1 × 10–2 s–1 for the open cage containing Au NPs and 1.7 × 10–2 s–1 for the closed cage containing Au NPs. This reduction in the kinetic constant is related to the structure of cages, which modifies the accessibility of reactants to the nanoparticle. The closed form of the cage has a flatter structure, which makes it more difficult for the 4-nitrophenol to reach Au NPs and results in a lower the reaction rate. Conversely in the open form, the cage diameter is larger and 4-nitrophenol can easily reach Au NPs (Figure 62).

83Cooper and co-workers loaded halogen and organometallic guest molecules into the porous structure of a CC3 crystalline organic cage. To do so, the desolvated crystals of CC3 were placed in a sealed vessel in the presence of solid iodine. Sublimed iodine at room temperature condensates into the pores of the CC3 crystals, producing a change of color from white to almost black. Iodine atoms were organized in a diamondoid arrangement that formed I5– in the channels of the CC3 structure. Cage CC3 was also able to load OsO4 as a guest. Moreover, the authors envisaged the potential delivery of the encapsulated guests upon cage dissolution in common solvents (Figure 63 top). Similarly, Uemura and co-workers developed a method to perform radical polymerization of vinyl monomers in the porous structure of organic cage CC3. For this purpose, the authors adsorbed styrene monomer units in the pores of solid CC3, which resulted in a suitable monomer arrangement for polymerization. The flexibility of the organic cage packing structure allows polymerization to take place in the pores of the structure, unlike the rigid porous hosts, such as zeolites and MOFs, where polymerization does not take place (Figure 63 bottom).

84Sun, Yang, Wei, and co-workers hierarchically assembled porous colloids based on CC3 cages and surfactants with a positive or negative charge (Figure 64). The resulting materials have porous networks that can encapsulate guest molecules such as fluorescent dyes. In these materials, when the cationic rhodamine 6G and anionic Congo Red dye molecules are encapsulated in the CC3–surfactant porous colloid, charge selective uptake/release behavior is observed, which suggests that the electrostatic interaction between the surfactants and dyes play a key role. The authors encapsulated cationic rhodamine in the negatively charged CC3-SDS colloid and anionic Congo Red in the positively charged CC3-DTAB colloid. When using amphoteric surfactants, the polar head of the surfactants acts as a pH responsive electrostatic gate for the selective release of encapsulated guests. The authors used amphoteric surfactants lauryl dimethyl amine oxide (LDAO) and lauryl dihydroxyethyl betaine (LDB), which allowed the opening of the gate at a basic and an acidic pH, respectively. The properties of these materials open up new avenues to develop electrostatic gates based on surfactants by incorporating additional properties, such as pH-switchable behavior. The authors also used these systems to encapsulate an enzyme in the CC3–surfactant porous colloid to display greater catalytic activity compared to the free enzyme.

85Postsynthetic modification of organic cages is a practical strategy for increasing their functions and applications because it allows the synthesis of structures that would otherwise not be possible. Mastalerz used dialdehyde 42 and triamine 203 to prepare the first shape-persistent adamantoid nanocage 204 with a cavity volume of 678 Å3. The cage was synthesized by reversible imine condensation in an effective and simple one-pot synthesis. To increase cage robustness, the simplest postsynthetic transformation of imine cages is to reduce imines to amines. However, reduction induces the transformation of sp2 into sp3-hybridized atoms, which results in a more flexible cage that may lose its shape persistency and, thus, its porosity. Hence the reduction of 204 to give the reduced 205 led to structure collapse in the solid state that did not allow gas penetration. Cage 205 displays a BET surface area of less than 1 m2/g, unlike parent imine cage 204 with a BET surface area of 1377 m2/g in the solid state. Besides, cage 206 (obtained from 40 and 203) has a BET specific surface area of 1037 m2/g (N2, 77 K). To maintain the rigidity of imine cage 204 while increasing chemical robustness, Mastalerz and co-workers reported the conversion of 204 into quinoline cage 207 through a Povarov reaction. In this case, the obtained material had a BET specific surface area of 698 m2/g. Cage 207 was stable to a wide range of pH values, including very acidic conditions (concentrated sulfuric acid, pH −1.9) and very basic conditions (concentrated NaOH, pH 15.2) (Figure 65).

86Mastalerz and co-workers used imine transformation in cage 208 to carbamate (cage 210) through intermediate cage 209 to increase stability while maintaining cage structure and rigidity. Cage structure rigidity after the corresponding reaction was confirmed by the BET surface areas (N2, 77 K) of 744, 12, and 105 m2/g for cages 208, 209, and 210, respectively. Thus, whereas amine cage 209 displayed a small BET surface area as expected for its increased flexibility, carbamate cage 210 and especially imine cage 208 displayed a much bigger BET surface area (Figure 66). Cage 210 was stable under very acidic conditions and porosity remained after treatment with 10 M HCl for 16 h at room temperature. The cage was also stable in concentrated hydrochloric acid (pH = −1) at room temperature, 1 M hydrochloric acid (pH = 0) at 100 °C, or 1 M aqueous NaOH (pH = 14) (Figure 66).

87In general, amide-based cages also have good chemical stability, which is, therefore, a suitable bond to obtain robust organic cages. However, direct synthesis through amide bond formation is an unsuitable procedure to obtain cages as shown in earlier sections of this review, where the cage assembly limitations through irreversible bonds were analyzed. To overcome these synthetic limitations, Mastalerz and co-workers used imine transformation to amide by the Pinnick reaction. This reaction was run to transform imine cage 204 into the corresponding amide cage 211, which was impossible to obtain by the direct reaction of 203 and 212 (Figure 67). Amide cages 211 are porous in the solid-state with a BET surface area of 275 m2/g (N2, 77 K). The obtained amide cage is chemically robust at acidic (pH 1) and basic media (pH 14.5), which are one of the reported cages with the widest pH range stability. Only concentrated H2SO4 (36 M) produced cage decomposition. This exceptional robustness allowed the postfunctionalization of cage 211 with reactions like bromination and nitration, which require severe conditions. The resulting cages have electron-withdrawing substituents Br and NO2 that modify the polarity of amide bonds. Cage 204 was reacted with MeI to yield cage 213, which can also be obtained by the reaction of 203 and 214 (Figure 67).

88Cooper, Day, and co-workers used computational crystal structure calculations to predict the packing preferences of homochiral methylated cages CC14-R and CC15-R (Figure 68). This methodology allowed cage cavity size and shape, along with pore connectivity, to be calculated, which are both key parameters to predict the gas sorption properties of porous materials based on molecular cages. The authors foresee that future design strategies based on the developed methodology will enable access to a bigger family of cages with specific physical properties for given applications.

89With CCn tetrahedral imine organic cages (see the structures in Figure 30), Cooper and co-workers showed the possibility of designing and preparing a solid-state solution ternary crystal with the interactions that occur between the porous cages that involve chiral recognition. In this solid, cage CC1 holds together cages CC3-R and CC4-R without requiring stoichiometric ratios. The authors state that finely tuning composition to optimize porosity may be useful for molecular separations or catalysis (Figure 69). By making minor changes in the molecular structure and modifying the crystallization conditions, it is possible to modify the crystal packing of cages, which allows uncommon ternary cocrystals to be prepared. More complex crystal structures based on core–shell crystal packing have also been prepared, which permit synergistic combinations to fine-tune chemical shell composition that enables the control of properties like surface hydrophobicity or CO2/CH4 selectivity.

90Kim, Roh, and co-workers used a shape-persistent hydrophobic organic cage as a synthetic ion channel. By means of the condensation of triamine 215 and tetra(aldehyde) porphyrin 216 in CHCl3 in the presence of catalytic amounts of trifluoroacetic acid, the authors prepared porphyrin-based covalent organic cage 217 in 95% yield. The obtained cage has a cavity diameter of ∼1.95 nm and 12 windows with average diameters of ∼3.7 Å. The authors used the cage embedded in a lipid bilayer to study anion transport through the membrane. The anion transport selectivity across the lipidic membrane follows the Hofmeister series: I– > NO3– > Br– > Cl– > SO42–. This correlation between transport efficacy and anion hydrophobicity implies that anion dehydration is crucial for their transport through cage 217 (Figure 70).

91Zao and co-workers also described organic cages as water channels across a lipid bilayer. To do so, they used Cooper’s CCn tetrahedral imine organic cages containing four triangular windows and one central cavity. These cages had already been used by Cooper and co-workers for reversible water uptake, which proved their ability to host the water molecules in the cavity. The authors employed six different cages (i.e., CC1, CC3, RCC3, FT-RCC3, CC5, CC19) to study water permeability and salt rejection properties. Water transport experiments were performed by embedding cages in liposomes formed by a lipid bilayer. The authors found that cages prevented the transport of small cations and anions but allowed fast water permeation at approximately 109 water molecules per second. The most important factors to determine water and ion permeation are the pore window size, cage structural rigidity, and hydrophilicity, and their ability to form interconnected channel networks. As cage systems are highly symmetric, employing them as water channels did not depend on their orientation, which facilitates their use. As potential applications, the authors suggested using these cages in composite materials like membranes for water desalination (Figure 71).

92Like reversible imine bond formation, hydrazone bond formation is reversible under acidic conditions, such as trifluoroacetic acid, and allows error correction mechanisms during the self-assembly process of binding blocks to finally obtain the stablest thermodynamic cage product. Compared to imine bonds, which have relatively limited stability, hydrazones are thermodynamically and kinetically more stable. With this concept, Warmuth and co-workers replaced the amino groups in their previously described building blocks (see Figure 15, Figure 16, and Figure 54) with a hydrazide group (C(O)NHNH2) and prepared the [2 + 4]-, [4 + 8]-, and [6 + 12]-poly acylhydrazone capsules 218, 219, and 220. The reaction of tetraformylcavitand 177 with hydrazides 221–225 resulted in cage formation yields above 95% for some reported cages. Cages had cavity diameters ranging from 2 to 4.5 nm. The authors found that carefully choosing building blocks is crucial for obtaining the desired capsule structure and minimizing the formation of other capsules as side products. For this purpose, the linker strain in the cage structure must be as low as possible (Figure 72).

93Szumna, Potrzebowski, and co-workers used the cavitand 226 and cystine-based ligand 227 to prepare capsule 228. This capsule was used for mechanochemical encapsulation in the solid state of fullerenes. Despite capsules not possessing large enough entrance windows, the disulfide and hydrazone bonds present in the structure of the cage became sensitive to mechanical stress and were activated by a mechanism in which mechanical stress energy facilitated the breaking, exchange, and reforming of the reversible chemical bonds to transform mechanical energy into chemical energy. This activation is specific for flexible porous cage structures because the activation of disulfide and hydrazone bonds does not take place in small molecules, nonporous or conformationally deformable capsules. Under solvent-free conditions, capsule pores are empty, and the cage is less stable and prone to collapse when subjected to mechanical stress if the cage structure is not rigid. The authors point out that neat milling is a promising encapsulation method that can be especially useful in the future in those cases that require breaking capsule covalent bonds for encapsulation or to perform guest encapsulation under solvent-free conditions to avoid the solvent competition that reduces guest affinity (Figure 73).

94From tetraformylresorcin[4]arene 226 and hydrazine, Szumna and co-workers were not able to obtain the target cage 230. However, the modification of the aldehyde groups in tetraformylresorcin[4]arene 226 with hydrazine, to yield calixarene 229, and the further reaction with unmodified tetraformylresorcin[4]arene 226 resulted in cage 230 formation in quantitative yields. The obtained cage has a cavity volume of 2283 Å3, and the structure is intrinsically chiral due to the directional arrangement of the hydrazone groups and the rigidity of the structure stabilized by hydrogen bonds. On the basis of this property, partial separation of enantiomers was achieved by chiral HPLC using 230 (Figure 74).

95Hydrogen-bonding interactions can be used as the driving force to self-assemble organic cages from building blocks with an appropriate concave conformation (see structures 231–233 in Figure 75). Pioneering work on hydrogen-bonded capsules was done by the group of Rebek, who back in 1993 developed a molecular capsule (widely known as the “tennis ball”) held by complementary hydrogen bonding (231)2 and later more robust assemblies like dimeric capsules (232)2 and softballs (233)2, among others. These capsules have a hydrophobic cavity that allows encapsulating organic molecules, such as benzene, toluene, adamantane derivatives, dicyclohexyl carbodiimide, etc. This groundbreaking strategy has been further developed over decades, allowing the preparation of numerous and imaginative hydrogen-bonded capsules.,

96On the basis of this strategy, Mendoza and co-workers reported the self-assembly of resorcinarene 234 containing 2-benzimidazolone bridges that are held together by a cyclic array of complementary hydrogen bonds to result in capsule (234)2 with a cavity size of 5.7 Å × 14.7 Å. The cage behaves similarly to Rebek’s parent cage, and it encapsulates propionic, pivalic, cyclohexanecarboxylic, or 1-adamantanecarboxylic acids (Figure 76).

97Larger assemblies can be obtained by increasing the angle of the hydrogen-bonding motifs in the calixarene building block. In 1997, Atwood and MacGillivray reported the ability of cavitand 235 to self-assemble in spherical hexameric structure (235)6 in nonpolar solvents to form the largest molecular cage reported until that time. The structure is stabilized by 60 cooperative hydrogen bonds, which involves eight water molecules and the six calixarene motifs that form the structure. The capsule has a central cavity with a diameter of 17.7 Å and a cavity volume of 1375 Å3 (Figure 77). The structure can also be stabilized by replacing some water molecules with 2-ethylhexanol, which results in a cage with a cavity volume of 1290 Å3, 1-butanol (cavity volume of 1186 Å3), and 1-propanol (cavity volume of 1769 Å3). The last one corresponds to a 25% cavity volume increase caused by the 1-propanol molecules that separate the cavitand building blocks of the hexamer (Figure 57 bottom).

98Besides the direct hydrogen-bonding interactions between the constituent building blocks, the cage cavity can also be closed by the combination of the calixarene organic moiety and inorganic anions. Following this approach, Atwood and co-workers reported ion pair recognition in a molecular capsule (236) by summing the electrostatic and hydrogen-bonding interactions to form 1:1 host–guest complexes with tetramethylammonium salts and with halides Cl–, Br–, and I– as counterions. The Me4N+ cation is fully encapsulated inside the cavity and the halide anion seals the calixarene cavity by hydrogen bonding (Figure 78). Encapsulation takes place in solvents like CDCl3 and CDCl3/MeOD 9:1, whereas cation release occurs in DMSO.

99By using the same idea of putting the hydrogen-bonding interactions between calixarene and inorganic anions to good use, Szumna and co-workers prepared exceptionally large capsules with a dodecahedron geometry and a diameter of 30 Å by the self-assembly of 12 units of 5-fold symmetry calixarene (pyrogallol[5]arene 237a or resorcin[5]arene 237b) and 60 anions (chloride or bromide). The structure is stabilized by hydrogen bonds involving phenolic OH groups from the calixarene derivatives and anions. Titration curves of different 237a:salt ratios suggest the presence of distinct cage structures, such as 238a and 238b (see Figure 79 bottom). Cage stability very much depends on the solvent. So whereas capsules are formed in THF and benzene, they do not exist in chloroform because the solvation of anions is stronger in this medium, which disfavors the interaction of anions with the calixarene building blocks (Figure 79).

100Conversely to the hydrophobic interior of calixarenes, calixpyrroles have a polar cavity that can efficiently bind to anions to fix the calixpyrrole in a cone conformation. Ballester and co-workers reported the use of a calix[4]pyrrole derivative 239 containing a urea group to generate a dimeric capsular assembly (239)2 with multiple ordered coencapsulated guests (including polar guests trimethylamine-N-oxide, 2-(trimethylammonio)acetate, and 3-(trimethylammonio)propanoate, Figure 80) when the stoichiometry of components is set at the 2:1:1 molar ratio (calix[4]pyrrole/polar guest/methyltrioctylammonium chloride (MTOA+Cl–)). In contrast, a 1:1:1 molar ratio exclusively produced the encapsulation of the polar guest in the cavitand with no dimeric capsular assembly formation (see the example described in Figure 80). Of the different combinations of assayed guests, the authors concluded that to achieve capsule formation, using a chloride counterion in MTOA+ salt was necessary. In addition, the sum of the volumes of all the encapsulated guests needed to come closer to the optimum 55% capsule cavity occupancy value (Figure 80).

101More complex assemblies, such as heterodimeric assemblies, can also be prepared using two different cavitand molecules and the appropriate template. By this strategy, in another work, Ballester and co-workers reported the self-assembly of tetraurea-calix[4]arene 240 and tetraurea-calix[4]pyrrole 241 in dimeric capsules 240·241 with a cavity volume of 280 Å3. Self-sorting requires the presence of a trimethylamine-N-oxide guest that can modulate the outcome of the reaction to exclusively yield the heterodimeric capsule using CD2Cl2 or CDCl3 as the solvent (see structure (Me3NO·CDCl3)⊂240·241 in Figure 81). The different properties of the cage cavity, hydrophobic calixarene and polar calixpyrrole drive molecular recognition and the position of guests in the cavity. The polar part of the trimethylamine-N-oxide guest forms hydrogen bonds with calixpyrrole moiety, and the nonpolar part interacts with the calixarene site. In addition to the trimethylamine-N-oxide guest, one solvent molecule is coencapsulated at the calixarene site (see structure in Figure 81).

102Ballester and co-workers also reported that responsive capsules were obtained by incorporating light-responsive groups into a calix[4]pyrrole building block. In particular, the authors synthesized tetraureacalix[4]arenes 242, which are functionalized with terminal azobenzene groups in the trans conformation. These trans calixarenes quantitatively dimerize in CD2Cl2 in the presence of the Me4P+ cation, which is additionally encapsulated in the cavitand (Me4P+⊂(242)2). When samples are irradiated with light, the trans azobenzene groups isomerize to the cis isomer, which results in a reduced cavity size that leads to a partial Me4P+ release with no cage degradation. The inclusion of bulky groups in the terminal phenyl group in azobenzene moiety played a key role in controlling the amount of released cargo (12–70%), which was larger when two tertbutoxy carboxyl groups were included. (Figure 82). The authors suggested that a further increase in the bulkiness of the substituents in the terminal phenylazo groups could result in a complete cargo release by trans–cis photoisomerization.

103Szumna and co-workers prepared capsules with pockets that exploited the inherent functionality of peptides. To this end, the authors functionalized aldehyde groups in cavitand 226 with peptide chains by the Mannich reaction (cavitand 243, Figure 83, top), with imine bonds with short peptides 244 (cavitand 245, Figure 83, bottom) or with semicarbazone bonds with peptides 246 (cavitand 247, Figure 84)., In these systems, the complementarity between peptidic chains is key for the outcome of the self-assembled structure. The cages assembled through the Mannich reaction (Figure 83, top) produce heterochiral capsules obtained by self-sorting. Selectivity arises from the steric repulsions of homochiral cages, which destabilize the capsule assembly. This is evidenced by the quantitative self-sorting of a mixture of (d-243)2 and (l-243)2 to yield the formation of the heterochiral dimeric capsule (d-243)(l-243) by performing a few dissolve–evaporate cycles or treating it with 10% MeOH. This protocol is necessary because capsule dimers are kinetically stable and the mixture of (l-243)2 and (d-243)2 in CDCl3 remains unchanged for >14 days at room temperature or while heating for 48 h at 60 °C. The capsule has a cavity volume of 310 Å3 and encapsulates small polar molecules, such as EtOH or two nitromethane and four water molecules, as determined by X-ray crystallography. The reversible attachment of short peptides 244 to resorcinarene scaffold 246 (to give 245) by imine bonds allows chiral self-sorting experiments to be run using a mixture of racemic peptides 244. Depending on peptidic sequence 244, the process was driven by self-assembly to yield homochiral (l-245b)2 and (d-245b)2 or heterochiral capsules (l-245b)(d-245b) (see the scheme in Figure 83, bottom).

104The same group found that in semicarbazone compound 247, obtained from peptides 246 and tetraformylresorcin[4]arene 226, the backbones in 247 are placed in an appropriate position for the self-assembly of dimeric capsules (247)2 to form a structure that resembles the eight-stranded β-barrels in noncompetitive solvent CDCl3. Unlike the crowded interiors of natural β-barrels, the capsule structure is not perfectly sealed because of the imperfect complementarity of peptide-binding motifs, and they, therefore, present windows for guest uptake and release. Capsules have cavities in the order of 800 Å3 (see the example in Figure 84 displaying a cavity with 765 Å3). After optimizing the capsule structure, the authors found that the peptide chains 246 containing three amino acids in an (L, D, D) sequence gave well-defined dimeric capsules, with their side chains positioned inside the capsule cavity. Additionally, bulky polar side chains inside the capsule were avoided, and the only methyl group from Ala was accommodated in the cavity. The reversibility of the semicarbazone bonds, obtained by the reaction between azapeptides and aldehydes, allowed an efficient chiral self-sorting reaction to be performed in situ based on cavitand peptide chains. The authors point out that the elongation of peptides and changes in the position of side chains can open up numerous possibilities to obtain systems that mimic natural catalytic sites (Figure 84). Using the (247)2 isostructural semicarbazone-based peptidic capsules obtained from cavitand 226 and different acylhydrazone linkers, which also have cavities in the order of 800 Å3, it is possible to achieve the quantitative encapsulation of fullerene C60 or C70 using the reversible character of the semicarbazone bonds by following mechanochemical methods in the solid state.

105Szumna and co-workers also studied the effect of solvent polarity on the self-assembly of peptidic capsules, originally developed in nonpolar solvents (vide ante) like CDCl3. The authors prepared cavitands 249 that contained the more polar residues histidine (248a) and glutamine (248b) in the peptidic chain. However, this did not form sufficiently strong hydrogen bonding interactions to yield the cage in DMSO or methanol. This is not surprising because these solvents are often used to unfold natural proteins or breaking hydrogen-bonded aggregates. The authors also used C60 as a template to, in this case, allow complexes to be obtained in which C60 was wrapped by cavitands to form dimeric capsules C60⊂(249a)2 and C60⊂(249b)2. The cages containing C60 could be obtained by complexation during cavitand synthesis and also by heating for long time periods or by solid-state mechanochemical activation. In all cases, the resulting products contained a mixture of free cavitand 249 and the dimeric capsule with an encapsulated fullerene molecule (C60⊂(249)2). By using methanol, it was possible to isolate the C60-containing dimeric capsules due to solubility differences with the free cavitand (Figure 85).

106Mastalerz and co-workers reported the synthesis of octameric hydrogen-bonded capsule (250)8 from enantiomerically pure building block (+)-(P)-250. The chiral building block was designed to self-assemble in a single geometric orientation by reducing the formation of polymeric byproducts and producing a capsule with a cavity volume of 2300 Å3. The capsule was able to host different tetraalkylammonium salts with significant shape and size selectivity. For example, the association constant of n-C14H29(CH3)3NBr with the cage was 100-fold higher than that of (n-C16H33)4NBr (Figure 86).

107Tiefenbacher and co-workers reported the hexameric cage (251)6 based on the self-assembly of calixarene building blocks 251 through intermolecular amide–amide interactions. The resulting structure was stabilized by 24 intermolecular hydrogen bonds and had a cavity volume of 2800 Å3, which is the largest cavity size reported to date for a hydrogen-bonded capsule or cage. Spontaneous cage self-assembly takes place in chloroform and other chlorinated solvents, and cage formation is favored at higher concentrations of monomers. As a reference, at the 1 mM concentration, the monomer is mainly observed, while at 50 mM concentration, the cage was the predominant species. The cage was demonstrated to encapsulate spherical C60 and C70 fullerenes through favorable dispersive and π–π-interactions between the guest and the aromatic rings in the cage cavity (Figure 87).

108Markiewicz, Jenczak, and co-workers prepared a robust enantiopure octameric cage (252)8 held together by 48 cooperative hydrogen bonds between the 252 monomers. The capsule structure was retained in both the solid state, where crystals were obtained by the slow evaporation of a chloroform solution, and also in solution in nonpolar solvents, such as deuterated 1,1,2,2-tetrachloroethane (TCE-d2). Polar solvents like DMSO-d6 disrupted the hydrogen-bonding network of the cage and resulted in the free monomer. The large cavity size (Vcavity = 1719 Å3) allowed C60 and C70 to be encapsulated with preferential C70 binding due to structural and electronic complementarity (Figure 88).

109Diederich and co-workers reported the first dimeric capsule 253·254 stabilized by halogen bonding (XB) from resorcin[4]arene cavitands containing tetrafluorohalophenyl motifs halogen bond donor groups (254) and lutidyl halogen bond acceptor groups (253). The X-ray solid-state structure of dimeric capsule 253·254 showed two encapsulated benzene molecules (Figure 89 top). Capsule assembly was observed for X = I and Br, but no capsule was found for X = F and Cl. Capsule formation was enthalpy-driven and entropically disfavored., The authors also reported the first chalcogen-bonded dimeric capsule (255)2 obtained from resorcin[4]arene cavitands 255 containing either tellurium or sulfur (Figure 89, bottom). Later Yu, Rebek, and co-workers prepared analogous chalcogen-bonded cages (255)2 containing selenium and water-solubilizing groups that formed stable dimeric capsules in water (Figure 89, bottom).

110The use of reversible boronate ester bonds has also been employed for the self-assembly of cage structures. In particular, building blocks based on catechol and aryl-boronic acids have allowed the preparation of a wide range of cage structures. With boronate ester bonds, Beuerle and co-workers used catechol-functionalized tribenzotriquinacenes 256 and 1,4-phenylene diboronic acids 257 to self-assemble cube-shaped cages 258 (Figure 90). Cages 258 are the first examples of purely organic cubes with the highest possible cubic symmetry Oh. The one-pot process involves the cross-linking of 20 individual components through dynamic boronate ester bonds. The authors predict that this strategy will help to prepare lower-symmetry cubic cages analogues. The authors also explored the effect on solubility by changing the alkyl substituents in the cage structure to find that solubility in organic solvents decreases gradually by shortening the alkyl chains at vertices. This structural modification modulated the size of the square pore windows to provide three different porous systems in the solid state: cage windows, extrinsic pores, and intrinsic cage cavities (see packing of 258d, 258e, and 258f in Figure 90). The resulting materials are a rare example of organic cages that alternate micropores and mesopores with very large BET surface areas (N2, 77 K) of up to 3426 m2/g for 258f (Figure 90 bottom). The authors point out that further modifications of systems, such as extending aromatic linkers or adding functional side chains, will allow the fine-tuning of materials’ porosity.

111In addition to this family of cages, Beuerle and Klotzbach demonstrated that the bite angle of ditopic diboronic acid building blocks (259, 60°; 256, 90°; 260, 120°; and p-benzenediboronic acid, 180°) predictably defines the topology of the cage obtained by a reaction with the catechol-functionalized tribenzotriquinacene 256 (TBTQ) building block, with 90° between reactive sites (Figure 91 top). With these building blocks, it is possible to prepare cages with a trigonal-bipyramidal 261, tetrahedral 262, or cubic 258 geometry. Additionally, with building blocks 256, 260, and p-benzenediboronic acid, it is also possible to obtain a social self-sorted three-component cage 263, which corresponds to the first example of this cage type. This methodology offers a controlled synthetic protocol to prepare cages with a predefined topology based on building blocks’ geometry (Figure 91). Additionally, the transformation of covalent organic frameworks based on 256 and p-benzenediboronic acid into molecular cage 261 is possible by linker exchange using ortho ligand 259. Cage 261 efficiently encapsulates fullerenes C60 and C70. Besides, the cage acts as a masking template in the Prato reaction using C60⊂261 by forcing reactions to happen at the guest positions, which implies that the windows of the cage provide selectivity for the unfavorable N-methylfulleropyrrolidine tris-adduct (Figure 91 bottom).

112Mastalerz and co-workers used triptycene tetraol (264) and triboronic acid 265 to synthesize cage 266. The self-assembly reaction involved 12 triptycene tetraol molecules 264 and eight triboronic acid molecules 265 via the formation of 24 boronic esters. The cage has minimum and maximum diameters of 2.6 and 3.1 nm, respectively. In solid-state, the material has a very high specific BET surface area of 3758 m2/g (N2, 77 K) (Figure 92 top)., Using this strategy, the authors used triptycene hexaol 267 and p-benzenediboronic acid 268 derivatives to prepare cages 269. The self-assembly reaction involves four triptycene hexaol molecules 267 and six p-benzenediboronic acid derivatives 268. The obtained cages 269 are porous in the solid-state with a specific BET surface area of up to 511 m2/g (N2, 77 K) (Figure 92 bottom).

113Severin and co-workers reported the first example of cages containing boronate esters and dative B–N bonds. The multicomponent assembly of tripyridyl building blocks 270, diboronic acid 271, and catechol 272 yielded cages 273. The obtained cage 273a has a cavity size with a top–bottom wall distance of 6.8 Å, perfectly sized to encapsulate one molecule of coronene or one molecule of triphenylene (see the structures of the host–guest complexes in Figure 93). Extending the cavity by using 4,4′-biphenyldiboronic acid instead of p-benzenediboronic acid (top-bottom wall distance of 11.0 Å in cage 273b) allowed the encapsulation of two triphenylene molecules. Despite these cages’ good synthetic availability, the authors point out that the low stability and limited solubility in organic solvents are significant limitations, although solid-state applications could be feasible (Figure 93).

114Beuerle and co-workers described another example of a supramolecular cage assembled through reversible boron–nitrogen dative bonds. To prepare the cage, an initial reaction of 261 and 274 yielded tribenzotriquinacene trisboronate ester 275. The self-assembly reaction of 275 with 1,4-diazabicyclo[2.2.2]octane quantitatively yielded cage 276 in a concerted cooperative assembly pathway. Despite efficient cage formation, cage stability was limited. Thus, heating a solution containing the cage at a temperature around 360 K, produced cage disassembly. The disassembly was fully reversed by cooling down the mixture to room temperature. Besides, acidification produced irreversible cage decomposition (Figure 94).

115Chiral cages have drawn attention for having the potential to mimic the properties of enzyme-binding sites. As such, many cage structures include chiral building blocks, besides previous sections of this review also describing different examples of chiral cages, including CCn cages, and cages containing: cyclohexanediamine (74, 107, 113, 114, 172–174, 176, 189, 191, 193, 194), building blocks with P–M chirality (163, (250)8), and chiral amino acids (228, (243)2, (245b)2, (247)2, (249)2, and (252)8).

116As stated above, a common method to introduce chirality into cage systems is to use chiral groups on the building blocks employed to assemble the cage structure. Following this approach, Cooper and co-workers reported numerous chiral cages prepared by imine condensation reactions of (R,R)-1,2-cyclohexanediamine and different aldehydes (see further examples in the gas separation in section 3.3.1 of this review). Cooper and co-workers prepared porous organic nanoparticles by means of chiral recognition between different organic cages. Cages were self-assembled in a modular fashion that can be predicted by molecular modeling calculations. The authors also prepared enantiopure cages CC7 and CC8 using tris(4-formylphenyl)amine 101 and chiral diamines (R,R)-1,2-cyclohexanediamine and (R,R)-1,2-cyclohex-4-enediamine. Both cages had inner diameters of 1.2 nm and a cavity volume of approximately 1500 Å3 (Figure 95).

117Mastalerz and co-workers employed chiral tris(salicylaldehyde) 277 to prepare the large cubic cage 278. The reported synthesis involved the chiral self-sorting of eight chiral tris(salicylaldehyde) 277 and 12 p-phenylenediamine molecules. The cage structure has an internal diameter within the range 3.3–3.5 nm and constitute one of the largest reported organic cages. When the cage formation reaction was performed using racemic tris(salicylaldehyde) 278, 23 cage isomers were possible, but only enantiopure and mesocage isomers were observed. This high selective chiral self-sorting toward the isomers with the highest degree of symmetry was the result of enthalpic and entropic factors. Despite the small energy differences of enantiopure and mesocage isomers, a careful selection of solvent allowed preferential meso cage formation, which can be achieved with an 83% yield. Gas adsorption measurements of the racemic cages in the solid state revealed a large BET specific surface areas of up to 1487 m2/g (N2, 77 K), which is one of the highest values obtained for organic cages (Figure 96).

118Mastalerz and co-workers performed the condensation of triaminotribenzotriquinacene 279 with bis(salicylaldehyde) 280 to obtain three unstrained [2 + 3] cages (281) (Figure 97). The authors investigated if salicylimine cages 281 could be efficiently obtained from racemic building blocks by means of chiral self-sorting. An analysis of the reaction outcome with thermodynamic equilibrium evidenced narcissistic self-sorting that led to a racemic mixture of homochiral cage compounds (M,M)-281 and (P,P)-281 with the minor formation of heterochiral cage isomer (M,P)-281. These cages were in rapid equilibrium in the presence of TFA/D2O, which allowed an analysis by variable temperature 1H NMR in THF-d8 to estimate the enthalpy and entropy energies involved in the equilibrium. Enthalpy favored the formation of (M,P)-281 over racemic-281 (i.e., a racemic mixture of (M,M)-281 and (P,P)-281) by 6.5 kJ/mol. In contrast, entropy favored the formation of racemic-281 over (M,P)-281 by 38 J/(K mol). The BET surface areas (N2, 77 K) were 918 m2/g for (M,M)-281, 550 m2/g for (M,P)-281, and 211 m2/g for racemic-281. Despite the large difference in surface areas, the calculated Henry’s selectivity, defined as the ratio between Henry’s constant of the strongly and weakly adsorbed components, were similar for all the compounds (CO2/CH4 = 3.75–4.93; CO2/N2 = 17.5–25.3; CH4/N2 = 4.32–5.52) (Figure 97).

119Enantiopure hemicryptophanes are generally obtained by the chromatographic separation of diastereomers or by chiral HPLC resolution of racemic mixtures. Kuck, Chow, and co-workers developed a method using racemic tribenzobenzotriquinacene (TBTQ) (±)-282 and enantiopure (1S,2S)-diaminocyclohexane to obtain a mixture of three diastereoisomeric imine cages (283) that can be isolated by silica gel chromatography. After isolation, the individual cryptophanes were hydrolyzed to yield the corresponding enantiopure trialdehydes (Figure 98). The authors also developed a method to obtain enantiomerically pure TBTQ (282) from diastereomeric TBTQ triamides using Boc-d- and Boc-l-phenylglycine as chiral auxiliaries. This building block was used to construct enantiomerically pure carbon cages.

120Jarosz, Szyszka, and co-workers reported the synthesis of fluorescent diastereoisomeric molecular cages 284 and 285 containing cyclotriveratrylene and sucrose. Cages are efficient receptors for acetylcholine and choline. Cages (P)-285 and (M)-285 have a similar affinity for choline than for acetylcholine, i.e., no selectivity is observed. In contrast, cage (P)-284 has a better affinity for choline over acetylcholine, while cage (M)-284 has a better affinity for acetylcholine over choline. The opposite selectivity seems to be associated with the complementarity of each guest, with the cavity size and shape of the chiral twisted ((P)-285 and (M)-285) and non twisted ((P)-284 and (M)-284) geometry of each diastereoisomer (Figure 99).

121Alonso-Gómez, Cid, and co-workers synthesized the two enantiomers of a helical prism-like cage (286a) to show 10-fold chiroptical amplification compared to its isolated building blocks. Initial studies have revealed that the cage could encapsulate an Fc+ molecule as determined by 1H NMR and HR-ESIMS, but no encapsulation was observed for Fc (Figure 100). To improve affinity toward electron-poor guests, the authors prepared a modified version of cage (286b) by introducing six methyl groups into phenyl rings to increase electron density and to also obtain a more defined cage cavity (Figure 100 bottom). They performed a series of host–guest studies, which were carried out with structurally similar guests to Fc+ (see structures 287–291 in Figure 100), with volumes within the 168–283 Å3 range, which fits into the cylindrical cage cavity of radius 4.3 Å, height 8.3 Å, and volume 482 Å3. The presence of methyl groups significantly enhanced binding affinities from, for example, 50 M–1 to 900 M–1 for 290+ in CD2Cl2. An analysis of guest volumes and electronic properties revealed that the observed affinity was associated mainly with guest electronic properties and electron-poor guests had a better affinity.

122Jiang, Schalley, and co-workers reported the synthesis of naphthol-based cage 292 (naphthocage). The chiral cage, whose enantiomers were not separated in that study, is flexible and can adopt a self-inclusion conformation. Cage 292 has a high binding affinity to different monocharged organic cations, including methyl and ethylammonium salts, methylpyridinium, tropylium, ferrocenium (288+), and cobaltocenium (287+) (Figure 101) with Kassoc > 107 M–1 in CD2Cl2/CD3CN 1:1. The extremely stable ferrocenium⊂naphthocage (288+⊂292) complex (Kassoc = 1010 M–1) can be electrochemically switched by reducing ferrocenium 288+ to ferrocene 288 (Figure 101).

123Qiu, Shu, Shen, and co-workers reported employing enantiopure helicine building blocks 293 to prepare covalent organic cages (294) via imine condensation of enantiopure 293 and “tren”. The obtained cages had three helicene units aligned in a propeller shape to provide a triple-stranded helical architecture. This unique architecture was used for the enantioselective recognition of 1-phenylethylamine, 1-phenylethanol, and 1,1′-binaphthalene-2,2′-diol, which is promising for future asymmetric catalysis uses (Figure 102).

124Porous materials have unique properties for gas separation, gas sorption, catalysis, and other exceptional applications. Molecular cages also form part of the family of porous materials by forming solids with both intrinsic porosity (associated with cage cavity) and extrinsic porosity (associated with voids between individual cage molecules). Molecular organic cage applications involve mainly the encapsulation of guest species in their cavity. Some applications have already been described in the review for selected cages, and this section describes in more detail the use of organic cages to obtain porous materials for gas encapsulation and other applications. In particular, different representative examples that highlight materials’ porous nature based on organic cages are included in this section. Typically, these materials’ porosity is obtained by measuring the BET surface area using N2 at 77 K. Most examples also include the uptake of CO2 measurements, and determination of selectivity for, e.g., CO2/methane. Note that we have also included information on porosity in some examples described in previous sections of this review because this parameter is related to cage rigidity/flexibility and is, therefore, useful in cage designs.

125Separating similar sized and shaped molecules is a challenging task that usually involves cryogenic methods. As an alternative, materials based on the properties of cage cavities can be used for this purpose with excellent performance. To finely tune porous materials’ properties, it is necessary to control pore size, shape, and connectivity. Increasing cage cavity volumes and cage structure rigidity allows porous materials to be obtained whose surface areas approach 4000 m2/g. A material’s overall porosity includes the intrinsic porosity from the cage cavity and the extrinsic porosity between cage molecules. Several methods have been followed to increase porosity in cage materials. By way of example, by means of the bulky directing groups in CC9 and CC10, it is possible to hinder crystal packing from generating voids between cage molecules and to, hence, generate extrinsic porosity that is affected by cage packing in the solid state. Solid materials formed by cages CC9 and CC10 show solid-state packing differences and, therefore, differences in both extrinsic porosity and pore connectivity due to heteroatomic and aromatic intermolecular interactions of the fluorine atoms in CC10 (Figure 103). Additionally, the flexibility of cage structures strongly impacts adsorption properties, which contributes to porosity into three different classes: static porosity, dynamic porosity, and cooperative porosity. Static porosity is associated with the connected voids in solid material. Dynamic porosity is related to the flexibility of the connected void network when voids are empty. Cooperative porosity is associated when it is required to place guests in the cavity to transport guests along the void network because voids are disconnected in the empty host.

126Cage rigidity is a key factor to obtain porous materials in the solid state. Mastalerz and co-workers studied the effect of building blocks’ rigidity on the formation of [2 + 3] cages and the effect on gas sorption properties in the solid state. For this purpose, the authors used triptycene triamine 203 and two different bis-salicylaldehydes (280 and 295) to prepare the [2 + 3] cages 296 and 297. Rigidity directly impacted cage formation, and using more rigid aldehyde (i.e., 280) formed a cage with a 69% yield, whereas more flexible aldehyde (i.e., 295) produced a cage with a 33% yield, presumably because flexibility allowed more oligomeric and polymeric byproducts to be obtained. Significant differences were also found in solid-state porosity because the cage with the more flexible linker underwent a phase change during nitrogen gas evacuation after gas sorption that yielded permanent low porous material. So whereas rigid cage 296 had a BET surface area of 744 m2/g (N2, 77 K), the more flexible cage 297 had one of only 30 m2/g, which indicates that pores were very small and the adsorption was kinetically hindered. Cages 296 and 297 showed a good CO2 uptake at 298 K by adsorbing 2.7 and 2.3 mmol/g, respectively. Additionally, good CO2/methane selectivity was achieved; i.e., 4 for cage 296 and 10 for cage 297 (Figure 104).

127Cooper and co-workers used organic cage CC3 for the solid-state separation of rare gases with unprecedented performance. This cage had a cavity with a diameter of 4.4 Å that came very close to the diameters of xenon (4.10 Å) and radon (4.17 Å). From the static view of the 3D structure in the solid state, the narrowest point in pore channels was placed at the cage windows between adjacent cages, with a diameter of only 3.6 Å. This diameter was slightly smaller than the diameter of Kr (3.69 Å) and was, in theory, too narrow to permit the diffusion of either xenon or radon. However, the cage structure vibrations associated with cage flexibility led to an enlarged pore diameter for a small fraction of time. This phenomenon was quantified by molecular dynamics simulations and showed a time-averaged pore-limiting envelope that was broad enough to allow the diffusion of both xenon and radon as experimentally observed. In fact, experimental gas adsorption isotherms demonstrated the substantial selective uptake of both krypton and xenon in CC3. Promising results for chiral separations were observed for the homochiral CC3-R and CC3-S crystals with the preferential adsorption of the 1-phenylethanol enantiomer with the opposite chirality, which suggests future applications in enantioselective separations (Figure 105).

128One key aspect for determining solid materials’ properties (e.g., guest uptake kinetics) is surface morphology. To understand the properties of the surface of these porous solids, Cooper and co-workers took AFM measurements, which allowed the surface in crystals of cage CC3-R to be studied. The results showed no distortion from the expected surfaces from their single-crystal X-ray structure, which proves that the surface does not relax or deviate from bulk crystal packing.

129In addition, the solid-state structure can be altered by changing the solvent because it modifies the crystal packing preferences of cages, and also by the interconversion in the solid state by the response to specific chemical triggers, which are associated with cage flexibility and the presence of different conformers in the solid state.

130Cooper and co-workers modified cage CC3’s internal cavity by performing a series of organic reactions to tune its size (Figure 106). This method obtained hybrid materials capable of separating deuterium and hydrogen by kinetic quantum sieving (KQS). This methodology involves a series of protect–functionalize–deprotect that allows to tweak cavity size from 187 Å3 of CC3 to 30 Å3 of 6ET-RCC3. The fine-tuning of cavity size gave a very small difference between the volumes of the hydrogen molecule and the cavity. In fact the minimum molecular dimension of H2 was 2.2 Å and cage 6ET-RCC3 had a pore-limiting envelope centered at 1.95 Å and broad time-averaged size distribution, which is ideal for selective gas separation. In fact, the cage 6ET-RCC3 allows the separation of D2 and H2 with a selectivity of 8.0 with a high D2 uptake of 4.7 mmol/g. The authors also demonstrated that reversible SO2 capture can be also achieved using cage CC3 derivatives. Whereas cage CC3 displayed modest and reversible SO2 capture, the secondary amine RCC3 cage exhibited high irreversible SO2 capture, and tertiary amine 6FT-RCC3 showed an excellent and very high reversible SO2 capture (13.78 mmol/g; 16.4 SO2 molecules per cage). The high uptake at low partial pressures revealed the potential of cage 6FT-RCC3 for trace SO2 adsorption.

131Cooper, Briggs, and co-workers synthesized derivatives of trans-1,2-diamino-cyclohexane to prepare CC3 cage analogues. The introduction of two methyl groups into this building block resulted in cage CC16, which had frustrated packing with increased porosity compared to parent cage CC3. The BET surface area (N2, 77K) for these cages was 1023 m2/g (CC16) and 409 m2/g (CC3). The building block with two hydroxyl groups (cage CC17-R) maintained the porosity of parent cage CC3 with a BET surface area (N2, 77K) of 423 m2/g. In contrast, the incorporation of bulky dihydroethanoanthracene groups changed the self-assembly toward the formation of a larger non porous [8 + 12] cage CC18-S (BET surface area (N2, 77K) of 10 m2/g) (Figure 107). Scrambling reactions of cages containing 1,2-ethylenediamine (CC1) and R,R-1,2-cyclohexanediamine (CC3) lead to a distribution of molecules with different shapes that cannot effectively pack, which creates a on-crystalline porous amorphous solid. Changing the scrambling reaction ratio allows the fine-tuning of porous properties by allowing materials with high H2/N2 selectivity to be obtained. In some cases, H2/N2 selectivity is as low as 1–1.5 (mol/mol). By simply varying the composition of the scrambling reaction, it is possible to increase selectivity up to 5 as a result of the poor packing of the scrambled cage mixture.

132Banerjee and co-workers prepared a CC3 cage analogue using 2,4,6-trimethoxy-1,3,5-triformyl benzene 298 and 1,2-cyclohexanediamine as building blocks to form the imine-based cage 299 (Figure 108). In the cage structure, the OMe groups built significant steric hindrance and hydrophobicity around imine bonds and protected them from acid or basic hydrolysis, which resulted in enhanced cage stability. Additionally, the electronic effects of the OMe groups introduced into the cage aromatic rings resulted in electron-rich aromatic rings, which also improve cage stability. Cage 299 presents three different solid-state polymorphs, which are selectively obtained using different solvents. Of the three polymorphs, only one is porous to nitrogen. The porous polymorph (β) presents a window-to-window arrangement that results in a porous structure with a BET surface area (N2) of 370 m2/g (Figure 108).

133Cooper, Day, and co-workers used 1,2-cyclohexanediamine and tetraaldehyde building blocks (112a, 112b, and 112c) to obtain cage structures with a tubular geometry (TCC cages, Figure 109). At the ends of a tube, TCC cages had two roughly triangular windows that resembled the windows of cage CC3. However, window-to-window packing was observed between the homochiral TCCs unlike in cages CC3. In contrast, racemic TCC mixtures form 1D nanotubes structures in the solid state with a window-to-window arrangement of the cages along cages’ tube direction. The materials’ porosity depends on each cage structure, and the BET surface areas (N2, 77 K) for TCC1-R/TCC1-S are 881 m2/g and 1022 m2/g for TCC3-R/TCC3-S. In contrast, TCC2-R/TCC2-S does not display any microporosity and exhibits a BET surface of only 26 m2/g. The cocrystals of TCC2-R with CC3-S at the 1:2 molar ratio allowed a porous crystal structure to be obtained that was formed by the CC3-S layers pillared by CC3-S/TCC2 through window-to-window interactions. The obtained material had a BET surface area (N2, 77 K) of 1363 m2/g, which is bigger than the individual components. In TCC systems, the 2H solid-state echo NMR experiments were performed to determine the 13C T1 values to study the rotational dynamics of the para-phenylene rings that constitute the central cavity walls. The rotational rates fell within the 1.2–8 × 106 Hz range at 230 K with low activation energy barriers within the range 12–18 kJ/mol. The rotation rate significantly lowered to 5–10 × 104 Hz at 230 K by the encapsulation of iodine molecules. The authors pointed out that these results highlight the importance of the dynamics of cage systems in contrast to a simple analysis of static structures that can provide incomplete information. In light of this, studying cage dynamics could be important in competitive guest loading, molecular separation, and guest release (Figure 109).

134Zhang and co-workers prepared shape-persistent organic molecular cage 300 by the reaction of 135 and 180. Cage 300 in the solid state displayed completely reversible adsorption of CO2 and N2 with a selectivity of 73 for CO2 adsorption over N2 (at standard temperature and pressure (STP) of 20 °C, and 1 bar). The uptake value of CO2 was 4.46 cm3/g (amount of gas at STP), whereas N2 was practically not adsorbed at all (0.061 cm3/g at STP). The observed gas-adsorption selectivity can be ascribed to the formation of reversible carbamate bonds by CO2 interacting with the cage secondary amine groups, which are favored by the well-defined porous cage cavity structure (Figure 110).

135Sessler, Humphrey, Zhang, and co-workers prepared pyrrole-based organic cages 301 and 302 with an affinity toward CO2 gas associated with the polar cavity that provided pyrrole subunits. Cage 301 had a well-defined central cavity with a “floor” to “roof” distance of 11.3 Å. In contrast, cage 302 had a much less defined cavity. The gas sorption experiments using CO2, N2, H2, O2, and CH4 showed that both cages were selective toward CO2. The BET surface areas for CO2 were 279 m2/g for cage 302 and 111 m2/g for cage 301 (Figure 111).

136Schmidt and co-workers used trialdehyde 47b and triamine 303 to synthesize the first porous organic cage containing perfluorinated aromatic rings 304 (Figure 112). Solid-state packing from X-ray - crystal diffraction studies showed window-to-window channels connecting cages 304, which gave a BET surface area of 536 m2/g. The fluorinated aromatic ring increased CO2-philicity by allowing 19.0 wt % of CO2 to be captured (4.2 mmol/g, 273 K, 1 bar), which is higher than similar sized cages CC2 (13.2 wt % CO2) and CC3 (11.0 wt % CO2). The authors describe how the observed higher gas uptake is likely to be associated with the higher hydrophobicity of the cavity caused by fluorinated aromatic rings.

137Mastalerz and co-workers used trialdehyde 39 and triamine 203 to synthesize the [4 + 4] triptycene-based porous cube 305 with a large BET surface area of 1014 m2/g (N2, 77 K), high CO2 uptake (18.2 wt % at 273 K and 1 bar), and CO2/N2 selectivity of 33.7 (Figure 113).

138The concept of liquids with permanent porosity was theoretically proposed by James and co-workers in 2007 by defining three different types of porous liquids: type I (neat porous liquids), type II (porous hosts dissolved in sterically hindered solvents), and type III (microporous framework materials dispersed in hindered solvents).− Later in 2015, James, Copper, and co-workers used trialdehyde 47b and diamine 306 to prepare porous liquids (type II) based on organic cages 307, which is the first reported synthesis of this type of materials (Figure 114). For this purpose, they used the core structure of cage CC3, which has a cage pore diameter of 5 Å and windows of 4 Å diameter. They anchored a crown-ether in diamine building block 306 to provide the required solubility in the cavity-excluded solvents (Figure 114). To obtain the liquid cage phase, a crown ether was chosen as the solvent (15-crown-5) because it did not fit into cage pores and, therefore, cage pores remained empty. Using this solvent, it is possible to obtain a concentrated solution of the cage (44 wt %) that corresponds to only 12 solvent molecules per cage molecule. Different sets of experiments proved that the cage cavity was empty in this liquid, although the porous liquid was able to host gas molecules, such as N2, CH4, CO2, or Xe. Qiao and co-workers performed molecular simulations and determined that the intrinsic gas storage capacity of cage molecules followed the order of CH4 > CO2 > N2 and determined that each molecule had a preference for a different placement in cage cavity. CO2 is preferentially positioned in the center of the cavity, CH4 is located in both central and branched cavity regions, and N2 molecules are randomly located.

139By means of robotic high-throughput optimization strategies, Cooper and co-workers managed to prepare 29 CCn cage–solvent systems as type II porous liquids. Dynamic covalent scrambling can be used to increase the solubility of the CCn porous organic cages in different voluminous organic solvents to obtain type II porous liquids with empty cavities. Such porous liquids were found to adsorb large quantities of gases with a high percentage of cage occupancy: 72% for Xe (i.e., from 100 molecules of cage 72 containing a Xe molecule) and 74% for SF6.

140Cooper, Greenway, and co-workers prepared a series of CCn cage structures (Figure 115) with different window sizes that allowed the tuning of the gas selectivity of the type II porous liquids obtained by dissolving the cage with hexa-chloropropene. The methyl groups in cage CC15-R were located at the cage windows (see Figure 115, bottom), which brought about a decrease in window size to diameter 1.7 Å compared to 4.0 Å for the scrambled cage mixture CC33:133-R. A decrease in window size produces a significant reduction in Xe absorption by altering cage selectivity from Xe-selective (for cage CC33:133-R) to CH4-selective (for cage CC15-R).

141Dai and co-workers prepared a proof-of-concept type I porous liquid using anionic porous organic cages (ACC) and the K+/crown complex as the cation. To this end, they employed a cage system based on the analogous neutral CC1 cage by incorporating one carboxylic group per ACC cage molecule (Figure 116). To transform the porous ACC solid into a porous liquid, the supramolecular complexation of K+ with dicyclohexano-18-crown-6 (higher affinity) and 15-crown-5 (lower affinity) allowed a supramolecular cationic complex with a larger size to be obtained, which transformed the initial solids into viscous liquids. The addition of 15-crown-5 in a 2:1 cage/crown-ether ratio yielded solids, whereas excess 15-crown-5 afforded a type II porous liquid. In contrast, the addition of 18-crown-6 at the 3:1 cage/crown-ether ratio yielded a type I porous liquid. This system enhanced thermal stability as a consequence of the high dicyclohexano-18-crown-6/K+ binding strength and the high boiling point of dicyclohexano-18-crown-6. The authors suggest that this strategy can be extrapolated to prepare porous liquids with larger pores or anionic porous nanoparticles. The authors demonstrated that cages had empty cavities in the liquid phase by measuring CO2 sorption capacities. The values obtained for the porous liquid formed by 15-crown-5 and dicyclohexano-18-crown-6 were 0.375 and 0.429 mmol/g, respectively (Figure 116c). These adsorption capacities are lower than for solid ACC (1.062 mmol/g), which has been attributed to the intrinsic and extrinsic pores present in solid ACC, while the two porous liquids do not have extrinsic porosity and, therefore, the observed porosity is lower for the porous liquids than for the solid cage.

142Apart from using cages for gas adsorption, the development of materials based on molecular cages allowed the porous properties of cages to be employed in other different applications, such as manufacturing cage-based porous columnar materials as solid-stationary phases for chromatographic separations, porous membranes, and solid supports for catalysis. Composite materials that incorporate molecular cages’ porosity properties can be obtained by depositing molecular cages on surfaces to form coherent crystalline thin films and to obtain membranes that can be effective for molecular separations. Besides the possibility of manufacturing composite materials, cage stability plays a key role and, in this regard, the crystals of cage CC3, stable in water, yielded water encapsulation. All of this suggests that the practical applications of these materials in wet environments are possible. Hybrid quantum dots functionalized with cages have also been used for sensing nitrophenol isomers and enantiomers of phenylalaninol and phenylethanol by fluorescence measurements.

143Enantioselective potentiometric sensing is also possible using cage systems, as demonstrated by Yan and co-workers. By employing membrane electrodes based on cage CC9 (see structure in Figure 30), they were able to perform the enantioselective sensing of 2-aminobutanol. Cage CC3-R has been used for enantiomeric resolution by NMR, which illustrates the potential of cages for determining enantiomeric ratios. Further applications in sensing include the detection of vapors of aromatic solvents by guest encapsulation in thin films of cage molecules on quartz crystal microbalances, as well as the sensing of the vapor of the γ-butyrolactone (GBL) drug.

144Other applications include nanofiltration, which utilizes porous organic cage-based membranes for water purification, gas chromatography enantiomeric separation with homochiral porous organic cage 201 (see Figure 61) as the chiral stationary phase,− open tubular capillary electrochromatography, or gas storage.

145The cage structures presented in this review up to this point have a considerable hydrophobic character, which makes them soluble in organic solvents but insoluble in water. To achieve water solubility, employing water-solubilizing groups is a key aspect as described in section 2.2. This section describes examples of cavitands, capsules, and cages with water-solubilizing groups and their applications as hosts of hydrophobic guests. Guest encapsulation in an aqueous medium is driven by the hydrophobic effect, which plays a key role and is the main driving force of supramolecular complex formation. As the hydrophobic effect is driven by the exclusion of water molecules from a hydrophobic surface, the hydrophobic force increases with the hydrophobic surface area, i.e., larger surfaces have higher interaction energies.

146In 2003, Rebek and co-workers reported water-soluble cavitand 308 after incorporating four carboxylic groups to achieve water solubility. Moreover, 308 shows high affinity toward hydrophobic molecules (Figure 117). While alkanes adopted extended conformations in solution that minimized steric interactions and maximized the surface area, when encapsulated into cavitand 308 alkanes folded to reduce the amount of hydrophobic surface exposed to solvent. This resulted in unfavorable gauche interactions. The authors found that the alkyl chains of two common surfactants (sodium dodecyl sulfate, SDS, and dodecyl phosphatidyl choline, DPC) adopted helical conformations by encapsulation. The polar head groups placed in the cavitand opening exposed them to the aqueous solvent, and the filling of space and hydrophobic surface burial were the driving force. This behavior is also observed when binding n-alkanes (C5H12–C11H24), which coil into helices in the hydrophobic cavity while rapidly tumbling on the NMR time scale. Substituted hydrocarbons (C8H17OH, C8H17Br, C8H17SH, C8H17SCH3, C8H17CHO, Br(CH2)7–9Br, Br(CH2)8OH, HO(CH2)12OH, Cl(CH2)8Cl, I(CH2)8I) are bound in quite a different way because the nature of the headgroup determines guest orientation in the cavity. Polar oxygen substituents are found exclusively on the cavity rim, whereas less hydrophilic groups, such as halides and thiols, adopt conformations with the groups at the base and at the cavity rim by rapidly interconverting (see the up–down conformations in Figure 117). On the basis of water-soluble cavitand 308, Rebek and co-workers also prepared water-soluble cavitands 309a and 309b, which offer good (>1 mM) solubility in water. These hosts bind primary alkyl halides Cn–Y (n = 5, 6, and 9; Y = Cl, Br, and I) to form dynamic complexes where the host and guest adapt their shapes to accommodate one another. The guest undergoes yo–yo-like tumbling in the cavity with a preferred orientation toward the floor of the cavitand. The preferred orientation is consistent with their hydrophobicity, and the most hydrophobic iodide avoids coming into contact with aqueous medium more than bromide and chloride do. The larger iodide fraction in the down position indicates halogen bonding, even though the deconvolution of hydrophobic and halogen bonding was not possible (Figure 117). These findings generally suggest a general mode of binding for substrates by making the best of hydrophobic host–guest interactions.

147Hooley and co-workers used guests with NMR detectable nuclei (13C and 19F) to analyze how the external environment affected the recognition properties of water-soluble deep cavitand 308. The authors found that by inserting deep cavitand 308 into lipidic bilayer environments produced the compression of cavitand flexible walls at the same time as it provided an additional barrier to wall-opening. This compression pushed cavitand walls closer to the guest by forcing it to adopt unfavorable conformations and by modifying reaction equilibria by, for instance, favoring 4,4-difluorocyclohexanone hydration from 13% in the presence of the cavitand to 23% in the presence of both the cavitand and the lipidic bilayer (Figure 118). In deep cavitand 308, the encapsulation of trimethylammonium cations is driven by cation−π interactions between the surface of the guest and the polarized aromatic rings of the cavitand. This strong interaction (with binding affinities >104 M–1) has been used to prepare water-soluble nanoswitch systems based on derivatized Au nanoparticles with trimethylammonium that close the pores of mesoporous silica loaded with a fluorophore. Upon the addition of the cavitand, pores open and the guest is released.

148Rebek and co-workers also developed water-soluble tetrakis(β-d-glucosyl) deep cavitand 310a with covalently bound carbohydrates to provide solubility in water. Cavitand 310a was prepared by an azide–alkyne “click” reaction between the tetraazido cavitand and propargyl glycoside. When guests are lacking, in water the cavitand forms a homodimer, probably in a kite conformation that self-assembles into a velcraplex structure, to likely reduce the hydrophobic surface area that is exposed to the solvent. Adding water-soluble guests 311–315, which have the appropriate size to fill the cavitand cavity with hydrophobic alkane surfaces, induces caviplex formation through vase conformation (see the structures of caviplex, unbound velcrand, and the velcraplex structures in Figure 119a). The binding of hydrophobic small organic molecules 314+Cl– and 313 exhibits slow guest exchange on the NMR time scale at 600 MHz, with no cavity occupation by the water-solubilizing carbohydrate. The authors also observed complex formation with alkylammonium salts 314+ Cl– and I– in both human serum and spiked samples of human urine. These results reveal the potential of synthetic receptors for applications in biofluid diagnostics, but the authors point out that the binding affinity of the cavitand to guests needs to be increased for real applications (Figure 119b).,

149The same cavitand framework, but by incorporating four guanidine groups (310b) for water solubility, also presents two conformations in equilibrium: unbound velcrand and guest-induced vase stabilized by the formation of intramolecular hydrogen bonds (N···H···O) by the amide groups of the cavitand. In the host–guest complex, the guest polar groups are oriented toward the solvent-accessible open end of the receptor, while the hydrophobic parts remain inside the host’s electron-rich cavity. The highest affinities are for the most hydrophobic guests 316–318, whereas polar adamantanes 315 show the lowest binding affinity. Quinuclidinium guests 314 possess intermediate affinities, which suggests that they can participate in additional cation−π interactions in the complex. The binding observed for the different guests in water is generally modest. However, in the presence of dodecylphosphocholine (DPC) micelles, the cavitand is positioned at the DPC–water interface, which increases the binding strength, with the affinities typically observed for organic solvents. The increase in binding free energies in the presence of DPC (up to 3.2 kcal/mol) is enthalpic in nature and is attributed to the disruption of velcrand dimers (Figure 119b).

150Rebek and co-workers described the synthesis of water-soluble cavitand 319 by incorporating ammonium solubilizing groups. The cavitand has a kite conformation with C2v symmetry in water, whereas when using DMSO, THF, or methanol as cosolvents, the cavitand turns into a vase conformation with C4v symmetry. The authors found that the cationic compounds described in Figure 120 did not bind by tetrammonium cavitand 319, unlike cavitand 308 that binds these cationic guests with high affinity. Contrary to that trend, cavitand 319 is a very specific receptor for neutral or anionic adamantane derivatives and has similar affinities to cavitand 308 (see the association constants in Figure 120).

151Rebek and co-workers reported the synthesis of deep cavitand 320a–320b with pyridinium “feet” to show good solubility in aqueous media. This cavitand system with benzimidazolones on the upper “rim” 320c was reported in 2002 by de Mendoza and co-workers (Figure 76). Cavitand 320c dimerizes in organic solvents to form a capsule that encapsulates diverse guest molecules. In another work, Rebek and co-workers found that, despite the solvent’s hydrogen-bond competitive nature, water-soluble cavitand 320a also dimerizes in water in the presence of lengthy hydrophobic guests (n-heptane, n-octane, n-nonane, n-decane, n-undecane). Whereas shorter alkanes flip rapidly on the NMR time scale; n-decane shows a second set of signals, which indicates the exchange of n-decane between the two magnetic environments (i.e., yo-yo motion of the guest in the cage cavity) on the NMR time scale (Figure 121). To prevent dimer formation, Rebek and co-workers N-methylated the benzimidazolones on the upper rim to obtain cavitand 320b. Surprisingly, they found that long-chain n-alkanes, n-alcohols, and α,ω-diols adopt folded conformations to fit into the hydrophobic part in the cavity. Small n-alkanes (C6 to C11) tumble rapidly in the cavity and show time-averaged symmetric environments by NMR. In contrast, longer n-alkanes (C13 to C14) display different behavior with greater shielding on the central CH2 groups, which are placed near the cavitand floor in a folded conformation that also induces the widening of the open end. For n-alcohols and α,ω-diols, the position of CH2–OH is near the cavitand upper rim, which changes the alkyl chain conformation (Figure 121).

152Rebek and co-workers reported the binding and reactivity of long-chain ω-amino acids (C11 and C12) and α,ω-diamines (C11 to C18) in cavitand hosts 309 and 320 in water. Binding involves the folding of the long carbon chain to bury the hydrocarbon chain into the cavity to expose the polar amino and carboxylic groups to the aqueous medium. The cavitand acts as a template for the cyclization of ω-amino acids with EDC/Sulfo-NHS and also for diamine acylation with succinic acid with EDC. In both reactions, yields improved 4-fold compared to the nontemplated reaction (Figure 122)., Long-chain aliphatic guests were encapsulated in the hydrophobic cavity of the cavitand in the vase conformation. In contrast, when no guest was present, the cavitand self-assembled quantitatively into a velcrand dimer in D2O at millimolar concentrations. As the binding strength of the lactamization product is higher than the binding strength of reactants, classic product inhibition takes place. Therefore, the cavitand acts as a stoichiometric template, limiting the application as a catalytic template. The same cavitand structure templates the cyclization of medium- to large-sized rings through the conversion of long-chain diisocyanates to cyclic ureas in water. Hydrophobic forces drive the long chain of diisocyanates into cavitands in folded conformations, which are not observed in bulk solution, and the reacting ends in close proximity to favor and accelerate the cyclization reaction (Figure 122). Cavitand 309 templates selective cyclization through an aldol/dehydration reaction of long-chain α,ω-dialdehydes in water. Hydrophobic forces drive the encapsulation of dialdehydes into cavitands in folded conformations, which favors macrocyclization over the unwanted intermolecular reactions observed in bulk solution. The 11- to 17-membered ring macrocyclic aldol reaction products are isolated in 30–85% good yields using the cavitand template. One of the drawbacks is the need for a stoichiometric amount of the cavitand. The authors point out that efforts are being to develop a catalytic version. Overall, the development of templates based on container molecules opens up new catalysis avenues. This allows use of container concave surfaces, which induce folded conformations to template reactions, unlike conventional templates that wrap reacting components around to bring relevant functions together. The observed cavitand templated effect is encouraging for water-soluble cavitands applications to achieve other selective reactions, such as remote functionalization.

153Yu and co-workers prepared water-soluble cavitand 321, based on Rebek’s previous cavitand 309a (see Figure 122), which incorporates quinoxaline motifs by adding solvent-dependent conformational preferences (vide infra) to the cavitand structure (see Figure 123). This cavitand had better water solubility and a larger aromatic cavity compared to Rebek’s cavitand. Cavitand 321 in DMSO-d6 appeared only in the vase conformation, in D2O in the kite or dimeric kite conformation (velcrand), whereas a complex mixture of conformations was observed in MeOD-d4 as determined by NMR. Despite the unfavorable kite conformation observed in water, the cavitand was able to encapsulate hydrophobic (cycloalkanes) and amphiphilic (cycloalkyl carboxylic acids) guests, which suggests the presence of a deep aromatic pocket. In fact upon guest binding, the kite conformation is converted into the vase conformation to better accommodate the hydrophobic guest. The binding of cycloalkanes involved placing the molecule in the cavity, and with cycloalkyl carboxylic acids, the alkyl part was placed inside the cavity with the hydrophilic carboxylic acid group exposed to the aqueous environment. The authors suggest that the N-donor atoms near the top of the cavitand can be used for the coordination of metals for catalysis applications (Figure 123).

154Gibb and co-workers determined the structural factors for efficient guest encapsulation in water-soluble cavitands 322a and 322b, which differed only in the position of the carboxylic groups that resulted in subtle shaped differences of pocket surfaces by providing a distinct hydrophobicity that resulted in a pocket in a dry or wet state depending on the energy of water molecules in the cavity. These differences in energy produced weaker binding for the pockets in a wet state compared to the pockets in a dry state. This finding suggests a new perspective for the role of water molecules in nonpolar host cavities. To draw these conclusions, the authors determined the affinity of eight guests (Figure 124) with the two cavitands, with a lower affinity for cavitand 322b than for cavitand 322a in all cases. A plausible explanation for this observation was obtained by molecular dynamics simulations, which pointed to the increased wetting of the pocket of 322b, with stronger ion–dipole interactions between the host and bound water and, therefore, rendered it energetically unfavorable for the guest to displace cavity water. Another important determined aspect was the difference in affinity between positive and negative guests binding; negative guests binding was slightly stronger to 322a than the positively charged ones. The authors suggest that this is the result of complex ion–ion and ion–dipole interactions in the different host–guest complexes that cannot be simply modeled (Figure 124). The encapsulation of fatty acids by 322a shows dependence on hydrocarbon chain length, and is enthalpically driven and entropically neutral. While shorter chains favor 1:1 complex formation, longer chains prefer a 2:1 host–guest stoichiometry, i.e., encapsulation of the guest in a capsule structure. For the medium-sized chains of fatty acids, guest protonation/deprotonation plays a role in the formation of the 1:1 or 1:2 complex, which favors the 2:1 host–guest complex at a low pH where fatty acid is protonated. At basic pH values, the energy balance between the solvation of hydrophobic surfaces in both the host and guest, and the solvation of the guest carboxylate groups, favors the 1:1 complex (Figure 124 bottom).

155Tetra-endo-methyl octa-acid 322c (Figure 124) in the presence of n-alkanes (C17–C26) self-assembles via the hydrophobic effect into tetrameric (Vcavity = 1400–1500 Å3) and hexameric (Vcavity = 3200–3700 Å3) assemblies by encapsulating two and three guest molecules, respectively. For n-hexacosane (C26), the inner nanospace is over 5-fold the volume of the dimer of octa-acid 322c. The orientation of methyl groups can trigger the drying of nonpolar pockets in water and, therefore, the position of methyl groups triggers the dewetting process. In the absence of methyl groups (322a) or when they point upward (322d), the cavity remains wet. However, when methyl groups are placed inward (322c), the water evacuation thermodynamics from the host pocket are similar to capillary evaporation. In this way, the dewetting process of 322c leads to increased affinity to guest complexation, which is enthalpically and entropically favored (Figure 124). Studies using pyrene as the templating guest reveal that capsule formation involves the rapid (<1 ms) formation of a pyrene⊂cavitand inclusion complex (pyrene⊂322a), followed by the slower binding of a second cavitand molecule to yield the pyrene⊂capsule complex (pyrene⊂(322a)2) with a dissociation lifetime of 2.7 s.

156The molecular dynamics simulations of 322a and 322c show that the minor chemical modifications to the constituents of cavitands can significantly impact their structure and host–guest properties. Cavitand 322a forms host–guest complexes with alkanes (C1–C16, methane to hexadecane) with 1:1, 2:2, and 2:1 molar host–guest ratios. In contrast, 322c also forms host–guest complexes with alkanes with the same molar host–guest ratios (1:1 to 2:2 to 2:1) but, in this case, switching between the monomeric and the dimeric assembly with increasing chain length is not observed (see Figure 124, bottom). The extra endo-methyl substituents that differentiate 322c from 322a tend to crowd alkanes in dimers by reducing the conformational freedom of guests and destabilizing the complexes of 322c with guests containing 12 carbons or more. Additionally, this tightening increases the repulsive contributions in the 2:2 and 2:1 complex dimerization free energy for the shorter alkanes in 322c versus than observed for 322a (Figure 124). The authors point out that the developed methodology based on molecular dynamics simulations will be useful for understanding the impact of host chemical modifications on their host–guest properties to help with bottom-up designs.

157Ramamurthy and co-workers reported the formation of stable capsuplexes (guest⊂capsule) in aqueous solution using cavitand 322a and a range of guests that can undergo light-induced β-cleavage. This property allows capsuplexes to be opened to release guest molecules in aqueous media by using light. The studied guests are not soluble in water in the absence of cavitand 322a. This proof-of-principle example could result in a general supramolecular photochemical strategy to release chemicals to aqueous solution containing encapsulated hydrophobic precursors (Figure 125a). The authors also tested the photodimerization of encapsulated hydrophobic organic guests. For this purpose, they encapsulated in the hydrophobic cavity of water-soluble cavitand 322a the water-insoluble guests, p-methylstyrene, indene, and 4,4-dimethylcyclohex-2-enone, via the formation of 2:2 host–guest inclusion complexes (guest)2⊂(322a)2. Irradiation of the aqueous solutions of the inclusion complex with a mercury vapor lamp induced the corresponding dimerization reactions with good yields (Figure 125b). Encapsulation produces the preorganization of guest molecules by favoring dimer formation, which are not generally formed under conventional conditions. In particular, the templated photodimerization of indene proceeds with high selectivity to the antiproduct with >90% yield. This high efficiency was ascribed to guest cavity occupancy and the guest–cage noncovalent interactions with a reaction taking place by energy transfer from the host to the encapsulated indene guest. This energy transfer results in the formation of an excited triplet of indene, which further reacts to yield the final photodimer (Figure 125b). The presence of a benzoate anion in the top periphery is essential for the cavitand to be a triplet sensitizer. The authors envisage that this approach can be used in organic synthesis to perform photoreactions in the cavitand.

158The same water-soluble octa-acid cavitand 322a was used by Ramamurthy and co-workers to trap a high-energy conformer of a piperidine derivative by encapsulation (see structure guest⊂(322a)2 in Figure 125c). Whereas free propyloxy-substituted piperidine in water exclusively adopts a single conformation in which the alkoxy group is equatorially positioned, an additional less stable conformer is also observed by NMR upon encapsulation. This phenomenon was only found for the guest with an O-propyl chain, and the related guests with ethyl, butyl, pentyl, or hexyl exclusively adopt the favored conformation with the O-alkyl chain placed equatorially in the capsule (Figure 125c). The researchers suggest that this achievement can be used to trap high-energy conformations of other species in a capsule cavity.

159The same research group studied the encapsulation of phenyl-substituted hydrocarbons (alkanes, alkenes, alkynes) of different chain lengths in octa-acid cavitand 322a. Upon encapsulation inside the hydrophobic microenvironment of the octa-acid capsule (guest⊂(322a)2), small hydrocarbons (length <15 Å) adopted a linear conformation, but longer ones adopted folded conformations. Such folded conformations, which are unusual and inaccessible in solution, are stabilized by specific C–H/π and π–π interactions between the guest and the capsule.

160Ramamurthy and co-workers incorporated ammonium groups in the same cavitand structure (323b) to achieve water solubility and also maintaining the host’s ability to alter the guest molecules’ excited state behavior upon encapsulation. Cavitand 323a with amine groups is soluble in water under acidic conditions, which complements octa-acid capsule 322 that is soluble under basic conditions. Moreover, cavitand 323b with tetraalkyl ammonium groups is soluble in water under neutral conditions. Cavitands 323a and 323b also form capsules with a range of organic molecules in a similar way to parent octa-acid cavitand 322. The authors found that upon encapsulation, pyrene displayed only monomer emission with an I1/I3 ratio of 0.90. The absence of an excimer emission, which was observed for free pyrene in solution, supported the encapsulation of one single molecule in the cage cavity, while the observed I1/I3 ratio suggested an internal cavity polarity close to that of benzene (Figure 126). Cavitand (see the structures in Figure 124 and Figure 126) dimerization to form the corresponding capsule structure took place only in the presence of the templating guest and, additionally, both octa-acid capsule 322a and octa-amine capsule 323a exhibited pH-dependent capsule assembly disassembly in water. Octa-acid capsule 322a formed only under basic conditions, while octa-amine capsule 323a formed under acidic conditions. Once the capsule had formed, the capsule disassembly rate depends on pH, and also on the guest’s hydrophobicity, which is slower for the more hydrophobic guests.

161Dalcanale, Geremia, and co-workers used tetraphosphonate cavitands 324 for the molecular recognition of amino acids. To perform host–guest complexation studies in water, with the core structure of cavitand 324a, the authors synthesized a water-soluble cavitand by incorporating four pyridinium water-solubilizing groups 324b. The experiments in methanol were performed using an analogue cavitand containing propyl chains instead of pyridinium groups (324c). Whereas the complexation of the guest molecules described in Figure 127 in methanol was entropy-driven, the complexation was enthalpy disfavored in water and resulted in a drop in Kassoc of almost 3 orders of magnitude. Despite the reduction in binding affinity in water, a remarkable increase in selectivity was observed toward the N-methylated amino acids. An analysis of the X-ray solid-state structures indicated that cation−π interactions were key for encapsulation. A direct correlation was found between the Kassoc in methanol and the depth of the guest in the cavity, evaluated as the distance between the +N–CHx guest carbon atom and the cavity entrance plane defined by the oxygen atoms of the P=O groups (Figure 127).

162Sgarlata, Bonaccorso, and co-workers reported tetracationic calix[4]arene 325 as a receptor of several gemini organic dianions containing two aromatic rings and negatively charged sulfonic groups. The formation of homodimeric capsule (guest⊂(325)2) between tetracationic calix[4]arene 325 and guest molecules was due to hydrophobic and electrostatic interactions. The end of the gemini guest was inserted into the calixarene cavity stabilized by CH−π interactions and hydrogen bonds. The guests with an even number of methylene units acted as better templating agents. Isothermal titration calorimetry (ITC) measurements showed that the formation of the inclusion complex was both enthalpically and entropically favored in association with the release of the water molecules from inside the cavity, followed by the encapsulation of the guest (Figure 128).

163Although calixarene cavitands only provide a hydrophobic binding pocket, the use of calix[4]pyrrole hosts provides both hydrogen-bonding and hydrophobic interactions. Ballester and co-workers prepared octa-pyridinium superaryl-extended calix[4]pyrrole receptor 326, which binds neutral difunctional aliphatic guests in water, the hydrophobic effect, CH−π, NH−π, π–π, and hydrogen-bonding interactions playing an essential role in the binding process. Calix[4]pyrrole receptor 326 forms 1:1 inclusion complexes in water with bis-isonitriles, bis-formamides, and formamide-isonitriles, all with high affinity to display association constants in the order of 105 M–1. The cis-conformation of the formamide end group is preferentially included in the deep aromatic cavity to form four hydrogen bonds between the oxygen atom of the cis-formamide and the four pyrrole NH groups of the host. This host modifies the outcome of the acid-catalyzed hydrolysis of bis-isonitriles by favoring the formation of monoformamides because of the lowering hydrolysis reaction rate constants of the bound substrate. Protection is chain length-dependent, and the hydrolysis of the substrates with five methylene groups allows the corresponding monoformamide with 80% selectivity from the corresponding bis-isonitrile to be prepared (Figure 129).

164Cram and co-workers reported the first water-soluble hemicarcerand 327 back in 1997 by using eight carboxylate groups as water-solubilizing groups (Figure 130). The water-solubilizing groups overcome the lipophilic nature of hundreds of known hemicarceplexes at that time, which hampers them from being soluble in water. Guest encapsulation very much depends on guest and cavity solvation, and the fine balance of the energies involved is key in the molecular recognition process. The release of many host innerphase solvating water molecules and guest solvating bulkphase water molecules provides a substantial entropic driving force. A wide range of guests of different sizes, polarity, and water solubility can be encapsulated in 327 to form 1:1 complexes, which are stable at ambient temperature in a medium containing excess guest. The tested guests included different types of compounds (Figure 130), which varied from highly polar Me2SO, intermediate polarity-MeC6H4NH2 and 1,4-(Me)2C6H4, to relatively nonpolar 1,4-dimethoxybenzene and 1,3-dimethoxybenzene; and nonpolar naphthalene. For cationic guests (Me4NBr, PhNMe3Br, BnNMe3Br, and 3-MeC6H4CO2Na), the enthalpic water solvation energy of their charges was much higher in the bulk phase than in the interior of the hemicarcerand. Therefore, this factor is dominant in inhibiting complexation (Figure 130).

165Rebek and co-workers reported the dimerization of water-soluble cavitands around hydrophobic compounds (vide ante). As described in the previous section for water-soluble cavitands, capsules can also encapsulate n-alkanes and α,ω-diols (bolaamphiphiles). The authors described how n-alkanes are flexible and adapt their shape to the container (Figure 131). For capsule (320a)2, addition of 15% hexafluoroisopropanol to water favors the encapsulation of amphiphilic guests. Flexible guests adopt the conformation that best fills the available space in the cavity, even if high-energy conformations are required. In those cases in which the guest molecule size approaches the cavity volume, a fine balance of attraction and repulsion determines encapsulation performance. For this latter case, Ouari and co-workers used capsule (320a)2 to encapsulate several nitroxide spin probes. An EPR spectral-shape analysis provided guest rotational dynamics information inside the capsule. These authors found that minor deformations in either the capsule or the guest took place. Distortion of the ideal hydrogen-bonding pattern between the two cavitands enlarges the central part of the cavity, which allows the encapsulation of larger guests. This behavior is similar to the dissociation of host subunits by providing a “gap” that allows a guest to enter/exit.

166Szumna and co-workers prepared peptide-based cavitands 328a–328e that could encapsulate fullerene C60 (Figure 132). The NMR experiments showed that cavitands 328a and 328b in aqueous K2CO3 media and 328c–328e in water formed dimeric capsules by encapsulating one molecule of fullerene C60 in the central cavity. Small-angle X-ray scattering (SAXS) allowed the experimental core sizes and shapes to be obtained, which agreed with the molecular models and confirmed the encapsulation of C60. The formation of these dimeric capsules was ascribed to a combination of C60-cage hydrophobic forces and hydrogen-bonding interactions between the peptide chains of cavitands.

167Ras, Beyeh, and co-workers reported the synthesis of water-soluble N-ethanol ammonium resorcinarene chlorides 329 functionalized with terminal hydroxyl groups on the upper rim. Cavitands were monomeric in aqueous media and formed 1:1 host–guest complexes by encapsulating linear and cyclic alkanes, linear halogenated alkanes, and aromatic fluorophores via hydrophobic interactions with binding constants of up to 559 M–1 in D2O/MeOD (9:1, v/v) at 298 K. These cavitands were also able to form capsules in nonaqueous solvents with a cavity volume of 258 Å3, where, in this case, NH···halogen and OH···halogen interactions were responsible for capsule formation. The authors were able to obtain the crystal structure of the host–guest complex with 1,4-dioxane. The capsule encapsulated two 1,4-dioxane molecules (molecular volume of 1,4-dioxane = 94.1 Å3), which resulted in a packing coefficient of 72.9%. The encapsulation of 1,4-dioxane was driven by a combination of CH−π and hydrogen-bond interactions (Figure 133).

168Diederich and co-workers reported the synthesis of molecular baskets with a modification of a cleft-type resorcin[4]arene-based cavitand by introducing rigid bridges to generate 330 and 331 (Figure 134). In addition, the incorporation of PEG groups yielded water-soluble baskets 332a and 332b. These host molecules can encapsulate a variety of small heteroalicyclic guests, where dispersion, C–H··· π interactions, and polar interactions (C–O···C = O or S···π) stabilize host–guest complexes. Molecular basket 330 encapsulated the O-containing and S-containing guests in CDCl3 with modest Kassoc values following this tendency: 1,4-dithiane (non binding) < 1,4-thioxane (29 M–1) < 1,4-dioxane (43 M–1). For water-soluble host 332b, the same stabilizing interactions described for basket 330 (in addition to solvophobic effects) took place, and the host–guest interaction strength increased. In addition, the S···π interactions significantly contributed and resulted in the reverse tendency of the Kassoc values in D2O/CD3CN 2:1: 1,4-dithiane (3724 M–1) > 1,4-thioxane (2021 M–1) > 1,4-dioxane (1952 M–1).

169Berthault and co-workers developed water-soluble cryptophanes 333a–333c for the encapsulation of xenon for 129Xe magnetic resonance imaging (Figure 135)., To achieve water solubility, these cryptophanes incorporate six carboxylic acids (333b) or six phenol groups (333c). In addition to solubilizing properties, these groups interact with the environment, such as the local pH or the presence of cations by providing them with sensing properties. Cryptophane 333b is sensitive to the pH of the surrounding media and displays a change in the 129Xe NMR spectra of the encapsulated 129Xe (68 ppm at pH below 4.7 and 64 ppm at pH above 4.7), which provides a local pH measurement of around 4.7. For cryptophane 333b, the xenon exchanging rate is markedly affected by the nature of the counterions present in the media (e.g., Cs+, K+, Na+, or Li+), and for Cs+, xenon binding is inhibited (Figure 135). In fact the cage 333b system with Cs+ and Tl+ in basic media displays association constants as high as 5.3 × 108 M–1 for Cs+ and 2.9 × 109 M–1 for Tl+. The high observed association constant for Cs+ agrees with the fact that xenon does not enter the cavity in the presence of Cs+ by either blocking the entrance to the cavity or occupying the hydrophobic cryptophane cavity. On the basis of this system, Schröder, Haag, and co-workers incorporated dendronized water-solubilizing groups based on polyglycerol and also used the resulting systems 333d for 129Xe nuclear magnetic imaging (Figure 135). Non-ionic polyglycerol dendrons are well-known for their compatibility in biological applications because they reduce nonspecific binding to biological targets. The cavity of these cryptophanes is extremely hydrophobic with good affinity to Xe atoms, and with Kassoc = 2750 M–1 as determined by 129Xe NMR Hyper-CEST experiments (Figure 135). The authors believe that the dendronization of cryptophanes could be a useful and promising approach to develop new Xe-based biosensors.

170Warmuth and co-workers used cavitands 334 to prepare water-soluble octahedral nanocapsules 335a–335f and studied their molecular recognition features using NMR titrations (Figure 136). Nanocapsules were prepared by the reacting tetraformyl cavitands with several ethylene diamines, followed by reduction with NaBH4, to obtain cavities with volumes ranging from 1700 to 2000 Å3. NMR studies showed that nanocapsule 335 encapsulated small negatively charged and hydrophobic guests (p-toluensulfonic acid, 4-methylumbelliferyl phosphate, and Boc-protected aspartic acid) into the six identical and independent cavitands by forming 1:6 host–guest complexes. 335a and 335b also bind with nucleotides (ATP, dAMP, dGMP, and TTP) but on the outside of nanocapsules. The binding behavior of nanocapsules 335c–335f was not studied for its low solubility at a neutral pH and given the formation of aggregates.

171The same group used the water-soluble tetraformylcavitand 336 by incorporating R-OSO3– water-solubilizing groups to prepare capsule structures 337 upon the reaction with diamines. The thermodynamically controlled reactions of the cavitand with two equivalents of diamines NH2(CH2)nNH2 (n = 2–4) yielded only oligomeric aggregates. However, in the presence of a suitable templating guest (see the fit guests in Figure 137), the corresponding capsule was formed with >90% yield. The guests that did not fit in the cavity (have the wrong shape or too small/large) did not template cage formation. These capsules are dynamic, dissociate upon the addition of acid, and reform upon basification. Guest encapsulation is driven by the hydrophobic effect and capsules are able to exchange guests through the temporary hydrolysis of imine bonds. In nonaqueous solvents, guest exchange is very slow or may not even be possible (Figure 137).

172In 2011, Davis and Barwell prepared synthetic lectins 338a–338c and studied their recognition behavior toward carbohydrates (Figure 138).1H NMR studies, carried out in water (at pH 7.5 for 338a and 338b and at pH 12.0 for 338c), showed that synthetic lectins bound with carbohydrates (glucose, galactose, mannose, methyl β-d-glucoside, and methyl α-d-glucoside) to form 1:1 inclusion complexes with different strengths (through C–H···π interactions). In line with this, the higher binding constants were for 338b with all the equatorial carbohydrates glucose and methyl β-d-glucoside. The measured binding constants with cage 338a were lower because the presence of electron acceptor fluorine atoms decreased the strength of the C–H···π interactions. Lower binding constants were observed with cage 338c due to the presence of water molecules that cannot be easily displaced in the vicinity of the cavity around phenoxide moieties.

173Further modifications yielded cages 338d and 338e and also binding carbohydrates in water. As in previous studies, 1H NMR titrations in D2O were carried out to measure the stability constants of 338d with selected carbohydrates (d-glucose, methyl β-d-glucoside, d-mannose, d-galactose, d-xylose, d-lyxose, 2-deoxy-d-glucose, d-ribose, N-acetyl-d-glucosamine, d-cellobiose, d-lactose, d-maltose, d-gentiobiose, d-sucrose, and N-acetilneuraminic acid). Hexose monosaccharides were bonded with low affinities, and pentoses (i.e, xylose) were slightly more strongly coordinated. However, the higher binding constants were measured for disaccharides d-cellobiose, d-lactose, d-maltose, and d-gentiobiose. In all cases, 1:1 binding inclusion complexes (through C–H···π interactions) were formed. Cage 338e presented negligible binding constants with all of the tested carbohydrates. These lower affinities could be ascribed to weaker C–H···π interactions due to the electrodeficient character of the pyridyl ring.

174The binding of synthetic lectin 339 (Figure 138) to the selected carbohydrates (d-glucose, methyl β-d-glucoside, N-acetyl-d-glucosamine, d-mannose, d-galactose, d-xylose, d-lactose) was tested by 1H NMR titrations at pD = 13 in D2O. These titrations showed that cage 339 was able to selectively bind with d-glucose to form a 1:1 inclusion complex. Coordination included C–H···π interactions and hydrogen bonding with lateral cage chains.

175Increasing the length of aromatic spacers results in a larger cavity to suitably host larger carbohydrates. Tricyclic synthetic lectins 340a and 340b (Figure 139) were able to form inclusion complexes with equatorial disaccharides. Fluorescence titrations, carried out in water with cage 340a and the selected mono- and disaccharides (methyl β-d-cellobioside, d-cellobiose, d-xylobiose, N,N′-diacetyl-d-chitobiose, d-lactose, d-maltose, d-gentiobiose, d-trehalose, d-sucrose, d-cellotriose, d-glucose, N-acetyl-d-glucosamine, d-galactose) showed the formation of 1:1 inclusion complexes only with d-cellobiose (3140 M–1), d-lactose (230 M–1), and d-maltose (67 M–1). 1H NMR and isothermal titration calorimetry (ITC), carried out in water with cage 340b, and the same guests also led to the formation of 1:1 inclusion complexes for all the saccharides except d-trehalose, d-sucrose, and d-galactose. The higher association constants with cage 340b were measured for methyl β-d-cellobioside (4500 M–1) and d-cellobiose (3340 M–1). The disaccharide/monosaccharide selectivity achieved with cage 340b was remarkable and amounted to 1300:1 when comparing the association constants for d-cellobiose and d-glucose. NOESY experiments, carried out with 340a and methyl β-d-cellobioside, showed that disaccharide entered the cage cavity through C–H···π hydrophobic interactions and hydrogen bonding.

176By preserving a similar core structure, Davis and co-workers prepared synthetic lectin 341a, which was able to recognize maltodextrin among several mono- and oligosaccharides (Figure 140).1H NMR titrations, carried out in D2O, using 341a and several mono- and oligosaccharides (d-glucose, methyl β-d-glucoside, methyl α-d-glucoside, N-acetyl-d-glucosamine, methyl β-d-N-acetylglucosaminide, d-galactose, d-mannose, d-cellobiose, d-cellotriose, d-cellotetraose, d-lactose, d-maltose, d-maltotriose, d-maltotetraose), showed the formation of 1:1 inclusion complexes. However, cage 341a displayed marked selectivity toward α-linked maltodextrins (with association constants of 580, 1150, and 1620 M–1 for d-maltose, d-maltotriose, and d-maltotetraose, respectively) over the other tested saccharides. This selectivity pattern was ascribed to the presence of the eight methoxy groups located in the anthracene moieties of cage 341a.

177Davis and co-workers also prepared synthetic lectins 341b–341f (Figure 140) with different side chains and studied their effect on the affinity binding of carbohydrates (d-glucosamine, d-galactosamine, d-glucose, methyl β-d-glucoside, N-acetyl-d-glucosamine, d-galactose, d-mannose).1H NMR and fluorescence titrations, carried out with all of the cages and the selected saccharides in water, showed the formation of 1:1 inclusion complexes (through C–H···π hydrophobic interactions and hydrogen bonding). The higher association constants were measured for d-glucosamine for all five cages (160, 1400, 2000, 2400, and 7000 M–1 for 341c, 341d, 341e, 341b, and 341f, respectively). As seen for the association constants for d-glucosamine, as the dendrimer in the side chain expanded, the affinities for guests increased. NOESY experiments, carried out with cage 341f and d-glucosamine, indicated that both side chains contributed to guest binding through strain-free salt bridges (d-glucosamine was positively charged at a neutral pH, whereas dendrimers were negatively charged).

178Synthetic lectin 342 (Figure 141), composed of two different aromatic components (pyrene and biphenyl), also forms inclusion complexes with carbohydrates.1H NMR titrations of 342 with several carbohydrates (d-mannose, d-galactose, cellobiose, methyl β-d-glucoside, and methyl β-d-N-acetylglucosaminide) carried out in D2O showed the formation of 1:1 inclusion complexes (through hydrogen bonding and C–H···π interactions). Of all the tested carbohydrates, the stronger association constant was measured for cellobiose (260 M–1).

179In another work, Davis and co-workers also prepared synthetic lectin 343 for the enantioselective recognition of carbohydrates (Figure 141). After the synthesis of cage 343, attempts to separate both enantiomers were unsuccessful. However, binding 1H NMR studies in D2O with selected carbohydrates (N-acetyl-d-glucosamine, d-glucose, methyl β-d-glucoside, d-mannose, l-mannose) were carried out with the racemate, and the coordination features of both enantiomers were quite different. The best enantiomeric discrimination was observed for N-acetyl-d-glucosamine, for which stability constants of 1280 and 81 M–1 for the formation of 1:1 stoichiometry diastereomeric complexes (enantioselectivity of 16:1) were found.

180Biomimetic synthetic lectin 344 (Figure 141) was prepared. Its coordination behavior toward carbohydrates (d-glucose, methyl β-d-glucoside, methyl a-d-glucoside, d-glucuronic acid, d-xylose, 2-deoxy-d-glucose, d-galactose, d-mannose, d-ribose, d-fructose, d-cellobiose, N-acetyl-d-glucosamine, d-maltose, l-fucose, d-gluconic acid) and other biomolecules (d-mannitol, ascorbic acid, uracil, uric acid, cytosine, adenosine, l-phenylalanine, l-tryptophan, paracetamol) was tested in water using 1H NMR and ITC titrations. Of all the tested carbohydrates and biomolecules, the strongest inclusion complex (1:1 stoichiometry) was formed between cage 344 and d-glucose with an affinity constant of ca. 18000 M–1. The all-equatorial d-glucose bound with 344 by forming hydrogen bonds with the six urea groups on the side chains and through C–H···π hydrophobic interactions. Besides, methyl β-d-glucoside, glucuronic acid, and xylose (with pyranose structures and all-equatorial substitution patterns) showed affinities >5000 M–1. Finally, other tested carbohydrates were bound ca. 100-fold more weakly than d-glucose, while biomolecules did not form inclusion complexes with 344. Finally, cage 344 was used for d-glucose detection in the presence of l-glucose in complex biological settings (prefiltered human serum, free glucose cell culture medium, beer with an artificially reduced glucose concentration) using circular dichroism.

181Davis and co-workers prepared pyrene-containing synthetic lectins 345 and 346 capable of forming strong inclusion complexes with N-acetylglucosamine derivatives in water (Figure 142).1H NMR studies, carried out in D2O, showed that eclipsed receptor 345 only bound with methyl N-acetyl-β-d-glucosaminide to form a 1:1 host–guest complex with an association constant of 2100 M–1. However, staggered receptor 346 formed 1:1 host–guest inclusion complexes with N-acetyl-β-d-glucosaminide, N-acetyl-a-d-glucosaminide, N-acetyl-d-glucosamine, methyl, β-d-glucoside, and d-glucose, but with a marked selectivity toward N-acetyl-β-d-glucosaminide (association constant of 18200 M–1). NOESY, TOCSY, and COSY studies showed that carbohydrates were sandwiched between pyrene units through C–H···π interactions and formed hydrogen bonds with the amide moieties on side chains. Besides, cage 345 was able to bind glycopeptide 347 with a remarkable high association constant of 67000 M–1.

182Davis and co-workers prepared cage 348 and studied the formation of inclusion complexes with polysaccharides in water (Figure 142).1H NMR titrations, carried out in D2O with 348, monosaccharides (d-mannose, d-galactose, d-glucose, methyl β-d-glucoside, methyl N-acetyl-β-d-glucosaminide), and polysaccharides (d-cellobiose, d-cellotriose, d-cellotetraose, d-cellopentaose, d-cellohexaose, N.N′-diacetyl-d-chitobiose, N.N′,N′′-triacetyl-d-chitobiose, d-chitobiose), indicated the formation of weak 1:1 inclusion complexes with monosaccharides and the formation of 1:1 threaded complexes with polysaccharides with affinities within the 3600–19000 M–1 range.

183Introducing multiple positive charges into the cage structure is another strategy to achieve water solubility. Stoddart and co-workers prepared hexacationic BlueCage·6PF6349a, which contained six pyridinium rings fused with two central electron-deficient triazines, bridged by three p-xylene units (Figure 143). The six PF6– counterions interacted with BlueCage6+ through anion−π interaction, and one of them was located inside the cavity cage. BlueCage·6PF6349a bound pyrene in acetonitrile with a Kassoc value of 4.93 × 105 M–1, whereas, upon changing the counterion from hexafluorophosphate to the bulkier tetrakis(3,5-bis(trifluoromethylphenyl)borate, Kassoc was increased to 3.95 × 106 M–1. This enhancement in pyrene coordination was ascribed to the fact that all the bulky tetrakis(3,5-bis(trifluoromethylphenyl)borate counterions were located outside of the cage cavity, which facilitates guest binding through π–π stacking interactions. Hexacationic triangular covalent organic cage 349b presented a rigid cavity with an inner distance between the two 2,4,6-triphenyl-1,3,5-triazine groups of 10.97 Å, whereas the distance between two bipyridinium moieties was around 11.60 Å (Figure 143).1H NMR and 2D DOSY studies, carried out in D2O, showed that the inner cavity of 349b was able to accommodate two pyrene-1-carbaldehyde molecules. Both aromatic guests underwent π–π interactions with the 2,4,6-triphenyl-1,3,5-triazine groups in an A-D-D-A (A = acceptor, D = donor) fashion. Besides, there were dipole–cation or dipole–dipole interactions between the formyl group of the guest and the pyridinium rings in the spacers of the cage. Cage 349b was able to accommodate 1,5-bis[2-(2-(2-hydroxyethoxy)ethoxy)ethoxy)ethoxy]naphthalene in water by forming a 1:1 inclusion complex, in which a naphthalene moiety was located inside the cage cavity as a consequence of a strong hydrophobic effect.

184Cao and co-workers reported the synthesis and complexation behavior of tetraphenylethene-based octacationic cage 350 (Figure 144). The X-ray crystal structure of 350·8PF6 showed the presence of a cuboid internal cavity of 17 Å (length) × 11.6 Å (width) × 6.94 Å (height). 1H NMR and DOSY studies in CD3CN, showed that cage 350·8PF6 formed a 1:1 inclusion complex with coronene through multiple C–H···π interactions (between the hydrogen atoms of the tetraphenylethene benzene rings and the guest) and sandwich-type π–π interactions (between the π-electron deficient pyridinium rings and the π-electron-rich aromatic guest). Besides, 350·8Cl showed remarkable coronene adsorption features, i.e., a solid sample of 350·8Cl was able to extract coronene from an organic solution. Water-soluble 350·8Cl was able to form a 1:1 inclusion complex with sulforhodamine 101 as the 1H NMR, NOESY, and ITC measurements demonstrated. The formation of inclusion complexes was also followed by fluorescence titrations. Indeed, the broad emission band of 350·8Cl in water at 545 nm (excitation at 410 nm) was progressively quenched, and a new emission at 621 nm appeared after the progressive addition of sulforhodamine 101 as a consequence of inclusion complex formation.

185Li, Yang, and co-workers prepared organic prismatic cage 351 by self-assembly in water at 80 °C of an aldehyde derivative and a carbonyl dihydrazine through the formation of hydrazone bonds (Figure 145). NMR studies showed that this cage was able to form 1:2 cage–guest inclusion complexes with 1,5-dihydroxynaphthalene derivatives. Besides, the addition of naphthalene diimide derivatives yielded complexes in which one molecule of the dihydroxynaphthalene derivative and another of the naphthalene diimide derivatives (through charge-transfer interactions) were located inside the cage cavity.

186When equimolar amounts of 352 and 353 in chloroform–acetonitrile 4:1 v/v were reacted for 6–8 h, a white precipitate appeared. It was assigned to the formation of water-soluble interlocked cage 354 (Figure 146). The DFT calculations showed that the π–π interactions between triazine cores (separated by ca. 3.4 Å) stabilized the interlocked structure. The reaction (in the same solvent) of equimolar amounts of 355 and 356 yielded water-soluble non-interlocked cage 357. Meta isomers were unable to form an interlocked cage because the benzene rings connected to the triazine core rotated due to the steric hindrance between benzene and imidazolium hydrogens, as demonstrated by the DFT calculations. This loss of planarity of benzene rings twists triazine cores and leads to unfavorable π–π interactions.

187By the reaction of 359 with (1S,2S)-(−)-1,2-diaminocyclohexane or (1R,2R)-(−)-1,2-diaminocyclohexane at a 1:3 ratio in water at 80 °C, Li and co-workers prepared basket-shaped tris-cationic chiral cages S-358 and R-358 (Figure 147). Both cages did not decompose in water and remained kinetically inert in the presence of competitive amines (ethanolamine). These authors ascribed this stability to multivalence (aldehyde and amino precursor are connected through six imine bonds) and the preorganization of 1,2-diaminocyclohexane (imino units are immobilized in gauche positions). The ability of cage S-358 to recognize the selected guests was studied by 1H NMR in D2O. Cage S-358 formed 1:1 inclusion complexes with ethane, pentane, ethylene, 1-chloropentane, 1-bromopentane, 1-bromohexane. 1,2-dichloroethane, 1,5-dichloropentane, 1,2-dibromoethane, 1,3-dibromopropane, 1,4-dibromobutane, and 1,5-dibromopentane. The formation of inclusion complexes is ascribed to the dipole-cation and C–H···π hydrophobic interactions between methylene protons in guests and the phenyl moieties on the cage. Besides, cage S-358 was able to preferentially accommodate (R)-1,2-epoxibutane (association constant of 35 M–1) compared to (S)-1,2-epoxibutane (association constant was too small to be accurately determined). The values of the association constants for (S)-(−)-propylene oxide and (R)-(−)-propylene oxide with cage S-358 were 11 and 14 M–1, respectively.

188Stefankiewicz and co-workers prepared self-assembled fluorescent aromatic cages using tetraphenylethylene tetraaldehyde (360) and two cysteine hydrazides functionalized with hydrophilic (361) and hydrophobic (362) chains (Figure 148). LC-MS experiments, carried out in water (or in water–DMSO solutions) and, with a mixture of 360 and 361, showed the formation of cage 363 after 3 days at 50 °C. The same results, namely the formation of a similar, yet hydrophobic, cage, were obtained after 3 days at 50 °C for a mixture of 360 and 362 in chloroform–DMSO (9:1 v/v). The 1H NMR studies also confirmed the formation of both cages. The authors reported that hydrophilic cage 363 had a volume cavity of 4200 Å3, whereas this value for its hydrophobic counterpart came to 3200 Å3. Besides, the solutions of both cages presented broad emission bands at ca. 510 nm (excitation at 320 nm), which were ascribed to the presence of the tetraphenylethylene fluorophore.

189Sanders and co-workers used dynamic combinatorial chemistry for the template synthesis of water-soluble organic cages. The air oxidation of the solutions of 364 and 365 (1:2 molar ratio) in water at pH 8.0 for 7 days yielded a mixture of cyclic trimer 366a (26%), the tetramer of 366b (39%), and the dimeric capsule of 366c (35%), in which both compounds (364 and 365) were bound through disulfide bonds. Larger architectures (366d–366i) were obtained using polyamines as templates under the same experimental conditions (364 and 365 at a 1:2 molar ratio, water at pH 8.0, and a 7-day reaction). Using spermine, six different cage structures 366d–366i were generated and characterized using LC-MS. The most abundant cage 366i was formed by two molecules of 364 and seven of 365. The yields of the different formed cages could be modulated by using other linear polyamines, such as 1,4-butanediamine, spermidine, and triethylenetetramine (Figure 149).

190With a similar approach, Stefankiewicz and co-workers used pseudopeptide trithiol precursors 364, 367, and 368 to prepare water-soluble pseudopeptide molecular cages 369a–369e based on disulfide bonds. Precursors 364, 367, and 368 come in different sizes (small, medium, large), and the self-assembly reaction shows a degree of self-sorting related to the size of the trithiol. In a mixture containing small and large precursors (364 and 368), the reaction self-sorts into a pair of two homodimeric cages (369a and 369d). Besides, when the precursors of three different sizes are used (364, 367, and 368), the expected homodimeric cages form (369a, 369b, and 369d), as well with small-medium (369c) and medium-large (369d) heterodimeric cages (Figure 150). The developed cages display good stability in aqueous media, which opens the door of these hosts for drug delivery applications.

191This review offers an in-depth view of purely organic cages, including hemicarcerands, carcerands, cavitands, capsules, and cages. The systems described throughout the review show the wide scope achieved in this field by allowing preparation of molecules with cavities ranging from small to large volumes. In addition to size, flexibility/rigidity, size of windows, hydrophobicity, etc., all play a key role in defining their properties. From the synthetic point of view, clearly the self-assembly of cage molecules through reversible bonds is the most widespread methodology because its efficiency and versatility surpass the limitations of irreversible bond formation, which does not allow the correction of the mistakes made during the self-assembly reaction pathway.

192Imine cages dominate synthetic strategies due to their very efficient synthesis but have limitations, such as their stability, especially in aqueous environments. There is no doubt that using other bond formation strategies, which are presently minor, will contribute to extend the synthetic toolbox to prepare cages with custom properties. Indeed, the development of organic chemistry, particularly robust reversible reactions, will contribute to develop cage structures with properties that cannot be achieved with current synthetic methods.

193Although most organic cages are inherently insoluble in water because of their hydrophobic nature, the addition of water-solubilizing groups allows water solubility to be accomplished. Such groups mainly contain positive or negative charges, although long polyethylene glycol chains also enable water solubility. By this strategy, the cage’s hydrophobic nature is retained, which allows encapsulating hydrophobic guests with high affinity in stabilized water and other noncovalent interactions.

194Despite the vast progress made in the area of purely covalent molecular cages and containers, this field is clearly extending, and significant advances can be expected in the near future. We envisage that the field will continue to grow and offer structures with new features to improve efficiency in different applications, including synthesis, gas separation, gas storage, biomedical applications, etc. Finally, developing computational tools to predict the outcome of a cage formation reaction from the selected building blocks, as well as cage properties, is another essential objective to accelerate the discovery of cages with customized properties.

195The fine-tuning of cage structures will allow the numerous functions reported to date to extend. Cages with specific cavities will permit materials to be prepared for gas separation by targeting gases that cannot be operated in today’s state of the art and will achieve high absorption efficiency and selectivity in gas mixtures. Cages with big cavities, including water-soluble cages with hydrophobic cavities, will allow further guests to be encapsulated, chemical reactions in its cavity to be performed, or they can be used as delivery systems.