Thierry Muller
*a and
Stefan Bräse
*bc
aCompetence Center Colorants & Functional Chemicals, Chemical Research, Group Technology & Innovation, Clariant Innovation Center G860, Industriepark Höchst, 65926 Frankfurt, Germany. E-mail: thierry.muller@clariant.com; Fax: +49 69 305 38470; Tel: +49 69 305 3858
bInstitute of Organic Chemistry, KIT-Campus South. Fritz-Haber Weg 6, 76131 Karlsruhe, Germany. E-mail: braese@kit.edu; Fax: +49 721-608-48581; Tel: +49 721-608-42902
cInstitute of Toxicology and Genetics, KIT, Campus North, Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany
First published on 19th December 2013
The preference of material chemists for specific organic building blocks for the generation of porous crystalline or amorphous covalently linked materials or supramolecular architectures, held together by hydrogen bonding or metal coordination, is reviewed. Tetrakisphenylmethane and adamantane cores are readily to hand and easily endowed with various functional groups suitable for network generation. Besides, these structures fulfil the stiffness requirements in order to generate permanently porous frameworks. By reviewing the major types of porous networks through selected examples, the authors intend to give a concise overview to the specialist in the field and to provide the non-specialist with a tool box of possibilities.
First of all, one might wonder why the field of molecular assemblies – discrete or endless, crystalline or amorphous ones – has raised such interest within the scientific community. The on-going enchantment is due to the applications of such constructs, mainly to the playground for host–guest chemistry they provide. The adsorption, storage and desorption of guest molecules within a given structure is indeed of tremendous economic importance. The development of save gas reservoirs for the automotive sector is only one out of numerous expected several billions euro markets. In order to understand the interest in organic based porous structures, one has to know that for a long time; nearly all porous resources have been limited to carbonaceous (charcoal) or inorganic (zeolites) materials. Only recently, organo-metallic networks, as well as purely organic materials featuring permanent porosity, could be achieved by imposing stringent prerequisites on the building blocks.1 These new materials offer several advantages over their inorganic counterparts.2 Metal–organic and especially purely organic structures are essentially composed of light elements (C, H, O, N, B, S). As such, they have unreached low densities and are very appealing to the automobile industry where weight loss is an essential issue. Furthermore, they can easily be tuned by design leading to adjustable parameters and defined chemical functionalization within the pore walls.
This brings us to the second pillar insuring the popularity of such materials which is the facile modularity of the organic component. It is easy for an organic chemist to modify a given organic tecton; formally elongating it for varying the pore size or introducing additional functional groups for further applications. When it comes to the tecton itself, tetrahedral structures I–III (Fig. 1) fulfil the stiffness requirements combined with an easy synthetic access. Last but not least, they present a good propensity to efficiently assemble to higher molecular assemblies. All this makes these molecules preferred targets for materials chemists.
There are a lot of different kinds of assemblies and networks – discrete or endless, crystalline or amorphous ones – which are either held together by supramolecular interactions (hydrogen bonding or metal-coordination) or through covalent bonds. The frontier between these structures is not always clear cut. The same is true for the naming of these constructs. Among the crystalline supramolecular architectures one may cite purely organic Molecular Tectonics,3 essentially hydrogen bonded networks and so-called Metal Organic Frameworks (MOFs) or Coordination Polymers (CPs) composed of organo-metallic subunits – held together by coordination bonds. The covalent purely organic networks may be divided into amorphous and crystalline structures. At least for the polymeric frameworks, several groups in the field have developed their own terminology. For the purpose of this review, this type of structures will be gathered under Hyper Crosslinked Polymers (HCPs). There are far less crystalline covalent networks than amorphous ones. The crystalline architectures and the ones presenting a highly ordered structure will be grouped under Covalent Organic Frameworks (COFs).
Applications deriving from guest adsorption and release have been the major diving force for the evolution of the whole sector. The building blocks and the corresponding architectures have been improved in order to optimize the overall surface area, the pore size and the pore size distribution. In order to make it easier to follow the evolution in the field for the non-specialist, basic concepts concerning porous materials will be discussed first. Then, synthetic access to core structures I–III will be talk over followed by the presentations of each class of architectures through selected examples. Applications will, majorly but not exclusively, deal with host–guest chemistry. Finally, future trends in the sector will be discussed.
The research dealing with the generation of porous frameworks and their applications is multidisciplinary by nature and lies at the interface of crystal engineering, physical chemistry, polymer sciences and synthetic organic chemistry. Some essential concepts and notions concerning porous frameworks will be discussed in detail in Sections 2.1–2.3.
Insolubility of extended 2D and 3D metal–organic and purely organic structures most of the time precludes the stepwise synthesis of such frameworks, leaving only self-/co-condensation as a viable approach for their generation.6,7 Although 2D structures are relatively straightforward to obtain, extending the concept to 3D networks is not trivial as any combination of building blocks could potentially give rise to an enormous variety of products.5 Hard and soft templates – so-called porogens or auxiliary structure-directing agents – were thus used to obtain defined open frameworks with a unique arrangement of the building units.8,9 This template approach is however too complex and cost intensive for industrial purposes.10 Yaghi was the first to generate a 3D crystalline porous COF using a template free approach by postulating “that the most symmetric nets are the most likely to result in an unbiased system and that those with just one kind of link will be preferred and are thus the best to target.”6,11
In order to obtain crystalline structures, reactions must occur under thermodynamic control which limits to hydro- or ionothermal syntheses.18 It is however often tedious to find the optimum reaction conditions leading to crystalline networks; besides most of these procedures suffer from poor reproducibility. Yet more reliable and easy to process polymerization reactions are kinetically controlled and thus lead exclusively to non-crystalline, amorphous polymers.18 This dilemma could recently be solved by showing that crystallinity is not a prerequisite for molecular control over pore size in rigid frameworks.19,20 Narrow pore size distributions – although not mono disperse – have been obtained in 3D porous organic polymers showing that long range order is not necessary for obtaining uniform pore sizes.21 One has yet to bear in mind that complete conversion of all functional groups – generation of a perfect framework without defects – is not achievable under kinetic control.5 The essence is however that both – thermodynamically and kinetically controlled – processes can be used to generate valuable crystalline and amorphous open organic frameworks.
Sorption performance is however not only dependant on surface area, pore volume and the chemical nature of the pore walls. Rapid access all the way through the pore reticulation system has to be established too.31 This is where organic porous frameworks outplay their inorganic and organo-metallic counterparts as they allow the generation of hierarchical structures. Purely organic frameworks can indeed be prepared as multimodal micro- and mesoporous material.21 Thus, in contrast to purely microporous materials like zeolites and MOFs, these materials combine the advantages of micro- and mesoporosity. They have an extremely high surface area but much fewer transport and diffusion restrictions. The micropores are in fact packed and unpacked via the interconnected mesopores thus accelerating the diffusion by orders of magnitude.21 There is no more pore blocking at the surface – a critical disadvantage of zeolites.21 Another advantage of purely organic networks compared to inorganic or hybrid structures is their really low density. As these architectures are entirely made up from light elements, they are capable of storing guest molecules in high weight percentages.15
Gas sorption of these novel purely organic compounds is however not yet completely understood. As these materials are rather soft compared to their inorganic and organo-metallic counterparts, swelling and elastic deformations – so-called breathing effects – have to be considered.32 This renders most evaluation methods for pore size distribution like nonlinear density functional theory (NLDFT) unreliable and there is a real need for efficient methods to be developed.33
In general, pore metrics are readily controlled by careful design of the building units and choice of the reaction conditions. Under template-free conditions, the solvent usually plays the role of porogen or auxiliary structure-directing agent. Therefore, the nature of the solvent as well as the reaction concentration directly influences the pore metrics.18 Only a few exceptions are known where the porosity is mostly independent of the initial building block concentration.34 Other factors which have to be considered when dealing with really small – microporous or ultra-microporous – voids are the increasing capillary pressure and high interfacial energies which tend to close the pores.35
Compound ICa is then typically converted to the tetrabromo- ICb or tetraiodo derivative ICc through electrophilic substitution using bromine (Br2)40 or iodine and bistrifluoroacetoxyiodobenzene (PIFA) in tetrachloromethane (CCl4)41 respectively (Scheme 1). Both substances constitute common intermediates for further functionalization.
The silane, germane, stannane and plumbane derivatives are typically prepared by fourfold nucleophilic addition of an organometallic aryl halide to the corresponding tetrachloride.42 As optimal conditions vary however for each compound, each synthesis will be discussed in more detail.
Tetrakis(4-bromophenyl)silane (ISib) is readily to hand. Nearly all procedures involve a monolithiation of a 1,4-dihalobenzene using butyllithium (BuLi) as lithiating agent.43,44 The generated lithium intermediate is then reacted with tetrachlorosilane (SiCl4) to yield ISib in four consecutive nucleophilic substitutions. There might however be problems due to a side reaction of remaining excess of butyllithium which reacts with tetrachlorosilane. The best procedure involves 1-bromo-4-iodobenzene (3) and a slight excess of n-butyllithium in pentane as non-coordinating solvent (Scheme 2).45 Through repeated washing cycles with pentane, p-bromophenyllithium (4) can be separated from remaining n-BuLi and the formed bromobutane. Finally a suspension of pure 4 in pentane reacts with SiCl4 to yield target compound ISib in 85%.45
Tetrakis(4-bromophenyl)germanium (IGeb) and stannane (ISnb) are obtained following a similar procedure as for the siliane derivative ISib although for IGeb and ISnb purification of the generated bromophenyllithium does not seem necessary. Reaction of 1,4-dibromobenzene (5) with tetrachlorogermanium (GeCl4) or tetrachlorostannane (SnCl4) respectively in the presence of butlylithium in diethyl ether (Et2O) leads to IGeb (ref. 46) in 67% and to ISnb (ref. 47) in 89% yield respectively (Scheme 3).
To the best of our knowledge, a tetrakis(4-halophenyl)plumbane derivative has not yet been described, it should however be assessable via a similar route as for the previous compounds. Comparable plumbane derivatives have already been prepared according to fourfold nucleophilic additions to lead(II) chloride (PbCl2).42
Scheme 4 Synthesis of 1,3,5,7-tetrakis(4-bromophenyl)adamantane (IIb) and 1,3,5,7-tetrakis(4-idophenyl)adamantane (IIc). |
Gutiérrez has developed a synthetic approach towards 1,3,5,7-tetracyanoadamantane which gives in term access to the tetrakis(aminomethyl) and the tetrakiscarboxylates particularly interesting for the generation of coordination polymers.51
Double lithiation of dibromo compound 7 and subsequent addition to ester 8 yields carbinol 9 (Scheme 5).52 The latter is then converted into spirobi[fluorene] IIId which finally yields common precursor 9,9′-spirobi[fluorene]-3,3′,6,6′-tetraol (IIIe).
With all three monomer types in hand, we will now turn to their utilisation in supramolecular architectures, and in covalent assemblies, networks and polymers.
Hydrogen bonds are interesting as they allow for an easy tuning of strength of the interaction by varying the heteroatoms involved and the number of hydrogen bonds for a given interaction. Furthermore, they allow introducing some directionality into the interaction by varying the angle between donor and acceptor. Last but not least, nature uses numerous hydrogen bonding interactions whose motives have gone through evolutionary optimization. Some of these have been used to assemble artificial supramolecular structures. As hydrogen bonds are reversible, they allow for thermodynamic control of the self-assembly process and thus lead in general to crystalline or at least highly ordered materials.
Wuest was among the first to use tectons of type I to generate porous hydrogen-bonded networks. Formally grafting four 2,4-diaminotriazine groups onto IC (Fig. 2, left, R1 = R2 = H) resulted in a three dimensional framework in which each monomer is held in position by 16 intertectonic hydrogen bonds (Fig. 2, right).3 The accumulation of the individually weak (≤7 kcal mol−1) hydrogen bonds within this structure, makes the latter so robust that it withstands guest exchange or partial guest removal without structural alternations (up to 63% of the original solvent content of the pores) or loss of crystallinity.3
Fig. 2 Left tecton ICf with 2,4-diaminotriazine (R1 = R2 = H) sticky sites. Right schematic representation of the H-bonding (red) within the network. |
Wuest's ICf based network has recently been used by Chen in gas sorption studies.53 It is one of the rare hydrogen bonded networks whose guest molecules could be completely removed (high vacuum at 100 °C for 24 h) without collapsing of the structure, thus allowing for gas sorption studies. The network revealed a preferential uptake of acetylene (C2H2) compared to ethylene (C2H4) at ambient temperature making it a very interesting material for C2H2/C2H4 separation.53
Such networks even allow cleaving parts of the individual tectons and thereby increasing the guest volume of the initial structure, again all without loss of crystallinity.
In order to do so, Wuest designed a network based on tecton ICg and held together by triaminotriazine hydrogen-bonding interactions (Fig. 3, R1 = R2 = H).54 In this particular case, amino groups not used in intertectonic hydrogen bonding are oriented into the channels of the network and should thus be available for further chemical modifications. To test this hypothesis, Wuest derivatized these amino functions with an ester moiety (methyl-3-aminopropanoate, Fig. 3, R1 = CH2CH2OCOCH3, R2 = H) and submitted the resulting crystals to mild hydrolysis conditions. Within two weeks, 40–60% of the original acetate groups had been cleaved as measured by 1H NMR of the dissolved samples. Single-crystal X-ray diffraction confirmed the NMR results as it showed structural integrity of the network and established that hydrolysis occurred isostructurally within the network.54
Fig. 3 Schematic drawing of a central tecton IC bearing triaminotriazine sticky sites surrounded by 4 neighbours (grey) and their H-bonding interactions (red). |
Functionalization of the triaminotriazine group with alkyl chains resulted in amorphous tectons (Fig. 3, R1 = R2 = (CH2)nCH3 with n = 3–11).55 It was found that increasing the chain length lowered the viscosity and the glass transition temperatures (Tg) of these materials. Melts of these constructs do not behave like conventional linear polymeric or dendrimeric materials but rather form a novel class of material.55
Wuest took also advantage of the tendency of aryl boronic acids to assemble to cyclic hydrogen bonded dimers (Fig. 4A).56 Tetrabronic acids ICh and ISih produced isostructural diamondoid networks with up to five-fold interpenetrating frameworks (Fig. 4B). Nevertheless, 60% and 64% respectively of the volumes, remain accessible for guest inclusion.
In both networks, each tecton is involved in a total of eight hydrogen bonds with four neighbours. Besides, each diamondoid network is connected to two adjacent interpenetrating networks by additional hydrogen bonds between the cyclic dimers (Fig. 4C). Channels in the network based on ICh were significantly smaller than in the ISih based construct. This was assumed to be due to the longer Si–C (1.889 Å) bond compared to the C–C (1.519 Å) bond.56 Thus, replacing the central atom of core I can have significant effects on the resulting networks. Further examples of replacing the central carbon atom in I by silicon, germanium, tin or lead will be discussed in Sections 6 and 7.
Another way to enhance the channel volume or the porosity of a network is to reinforce the rigidity of its tectons. Whereas in core I, the four phenyl groups are free to rotate around the bond with the central core atom, this rotation is prohibited in the spirocyclic core III, and therefore prevents conformational changes required to obtain a close packing in the solid state. Wuest exemplified this by comparing hydrogen bonded networks of tetrakis(4-hydroxyphenyl)methane (ICi) and 3,3′,6,6′-tetrahydroxy-9,9′-spirofluorene (IIIi) (Fig. 1, R = OH).52 In the ICi based structure, 28% of the volume is available for guest inclusion compared to 43% in the IIIi based crystals.
Wuest furthermore exploited the chemistry of urethanes and ureas to act as sticky sites in IC and ISi based suprastructures (Fig. 5).57 Porous crystalline networks were obtained with up to 66% of volume accessible to guest molecules. Most interesting however is the fact that enantiomerically pure tetraurethane ICj and tetraurea ICk could be prepared and that their crystallisation yielded porous chiral hydrogen-bonded networks (Fig. 5, right). The latter are very promising candidates for enantiospecific inclusion of guest molecules leading to applications such as racemic resolution, and asymmetric catalysis. As a matter of fact, enantiomerically pure cores of type I are rather rare58 and this is one of the few examples of hydrogen bonded networks offering enantiopure nanospace.57
Fig. 5 (A) Hydrogen bonding between urethanes. (B) Hydrogen-bonding between ureas. Right tetraurethane ICj and tetraurea ICk. |
Kabe hydrolysed (methanetetrayltetrakis(benzene-4,1-diyl))tetrakis(diisopropylsilanol) (ICl) in order to generate polyhedral organosilanols (Fig. 6).59 No interpenetrated networks were observed and the resulting silanol frameworks were used for selective hydrocarbon inclusion.
Fig. 6 Dia network of ICl hexane based on TOPOS analysis. For the SiOH based tetrahedral structure, isopropyl groups are omitted for clarity. |
Richert60 and Weber61 nearly simultaneously reported about hydrogen bonding structures of tecton IC held together by nucleobase pairing. Weber used one base per sticky site whereas Richert employed dinucleotide DNA arms to assemble a three dimensional network. Later on, Richert extended this concept to the adamantane core II and experimented different linkers to attach the DNA strand to the core.62 He was able to form a solid material based on the adamantane core from micromolar aqueous solution at 95 °C. Both groups have however until now not been able to produce suitable single crystals of their material for an X-ray structural study.
It is also possible to prepare host–guest structures based on nitro and cyano hydrogen bonding interactions. Pure tetrakis(4-nitrophenyl)methane ICm (ref. 63) and combinations with tetrakis(4-cyanophenyl)methane ICn (ref. 64) yielded various crystalline structures depending on the crystallisation solvent. However, nitro–nitro as well as nitro–cyano interactions are generally too weak and unspecific to generate highly ordered suprastructures.65
The conjugated base of a carboxylic acid, a so-called carboxylate, is a very efficient metal coordination group. Yaghi was the first to use the corresponding 4,4′,4′′,4′′′-methanetetrayltetrabenzoic acid ICo and the equivalent adamantane tetraacid IIo as three dimensional organic linker offering four coordination sites to produce MOFs.66
Fig. 7 shows three different views of a IIo – zinc MOF.
Based on this work, Lambert compared MOFs based on ICo, ISio and IGeo.46 All three structures were found to be quite different. The zinc portion of the ICo MOF is a dinuclear zinc-oxo cluster with square-planar geometry, whereas for ISio it consists of zinc-oxo chains with two zinc atoms having different geometries and for IGeo, linear zinc-oxo clusters with three zinc atoms bearing two different geometries were observed. These differences in morphology were attributed to the varying bonding length of the central atom with the four phenyl rings.46
Lin generated two ICo based copper MOFs, [Cu2(ICo)(H2O)2]·(DEF)6·(H2O)2 which crystallizes in tetragonal space group P42/mmc and [Cu2(ICo)(H2O)2]·(DMF)14·(H2O)5 which crystallizes in tetragonal space group P2Ic (DEF = diethylformamide, DMF = dimethylformamide).67 Both MOFs exhibit a Brunauer–Emmett–Teller (BET) surface of 526 m2 g−1 and 791 m2 g−1 respectively. When these structures were suspended in benzene with a subsequent freeze-drying step, the BET surfaces increased to 1560 m2 g−1 and 1020 m2 g−1 respectively. This constitutes a threefold enhancement for the first and an augmentation of 29% for the second MOF. The authors explain this drastic surface enhancement through the bypassing of the liquid phase which is thought to induce unwanted mesopore collapsing due to surface tension. Thus freeze-drying not only enhances the surface area but should also lead to a better structure preservation.67 Lin used this freeze-drying method to prepare two more MOFs containing tetraacid derivatives of core structure I.68 Using yet another tetra carboxylic acid derivative of I, namely 4′,4′′′,4′′′′′,4′′′′′′′-methanetetrayltetrakis(([1,1′-biphenyl]-4-carboxylic acid)) (ICp), Lin reported about a solvent induced single-crystal to single crystal transformation (SC → SC) of a 2D CP into a 3D MOF with enhanced porosity and hydrogen uptake capacity.69 Such a transformation involves cooperative movement of atoms as well as breaking and forming coordination or even covalent bonds. In this special case, exposure to dichloromethane (CH2Cl2) at room temperature induces a 2D (A) to 3D (B) SC → SC transformation as a result of dimerization of the meta-connecting points (Fig. 8).69
Fig. 8 Schematic representation of a SC → SC transformation of 2D CP A into 3D MOF B through dimerization of Zn connecting points. |
This topological transformation goes along with significant framework stability and an enhancement of porosity and hydrogen uptake. CP A has a BET surface of 177 m2 g−1 and a hydrogen uptake capacity of 0.3 wt% compared to a BET surface of 1170 m2 g−1 and 1.75 wt% H2 capacity for MOF B.69
Suh used tetraacid ICo to prepare a cobalt containing MOF, [CoII4(μ-OH2)4(ICo)2·(H2O)4]n·13nDMF·11nH2O.70 The resulting desolvated solid [CoII4(μ-OH2)4(ICo)2]n exhibits selective gas uptake for hydrogen and oxygen over nitrogen at 77 K and a preference of carbon dioxide over methane at 195 and 273 K. Zhou prepared the silane based tetraacid ISio and studied the corresponding MOFs.71
Schröder generated a MOF based on octaacid 5,5′,5′′-((3-carboxy-5-methylphenyl)methanetriyl)triisophthalic acid (ICq).72 The polyhedral copper complex [Cu4(ICq)(H2O)]solv has a 4,8-conntected structure of rare scu topology bearing octahedral and cuboctahedral cages and shows a total carbon dioxide (CO2) uptake of 314.6 cm3 (STP)/cm3 at 20 bar, 293 K and a total hydrogen (H2) uptake of 6.0 wt% at 20 bar, 77 K (Fig. 9).
Fig. 9 (A) Connectivity of ICq8– to 8 [Cu2(O2CR)4] paddlewheel units. (B) 4–8 connected scu topology. |
Pyridine units are also known as good metal coordinators. Van der Boom studied the linear versus exponential formation of supramolecular assemblies out of pyridine containing cores 10–14, ICr with palladium (Pd) (Fig. 10).73 The cores were chemically attached to different surfaces and the supramolecular assemblies were grown in a two-step deposition process (step 1: addition of metal complex, step 2: addition of core). The growth process was observed by UV-vis spectroscopy and ellipsometry.
Assemblies based on cores 10–14 showed exponential growth whereas for tectons 15 and ICr, a linear assembly formation was observed. These findings were attributed to the fact that cationic complexes 10–14 form porous structures which allow for incorporation of palladium. Cores 15 and ICr form closely packed assemblies which are unable to store palladium. As an exponential growth conducted in a two-step deposition process needs an excess of one of the components to be present, only cores 10–14 allow for an exponential formation of molecular based assemblies with an excess of palladium being present throughout the growing process.73
Loeb has prepared tetra-(4(4-pyridyl)phenyl)methane (ICs) bearing a completely rigid scaffold and used it to generate two copper(II) based MOFs, {(CuCl2(ICs))}x and {(Cu3Cl6(ICs)2)}x with PtS-type topology (Fig. 11).74
Fig. 11 Space-filling representation of the X-ray structure of MOF {(Cu3Cl6(ICs)2)}x. (A) Down the a-axis. (B) Down the a-axis showing the two interwoven networks. |
Long generated two tetrakis(4-tetrazolylphenyl)methane (ICt) based MOFs.75 Although Mn6(ICt)3·5DMF·3H2O collapses upon guest removal, [(Cu4Cl)(ICt)2]2·5DMF·11H2O not only reveals permanent porosity but shows an even more interesting property. Upon Soxhlet extraction with methanol, copper chloride (CuCl2) is extracted from the crystalline structure, leaving chloride deficient Cu4 squares to produce Cu4(ICt)2·0.7CuCl2 where the central Cl− anion from ca. 50% of the {Cu4Cl}7+ clusters has been eliminated. Exposed metal coordination sites are potent guest molecule binding sites and render the generated MOF highly interesting for hydrogen uptake for instance. The newly prepared MOF revealed a BET surface area of 2506 m2 g−1 and an H2 uptake of 4.1 wt% at 20 bar and 77 K.
Terpyridine units (tpy) are known to be very strong metal complexing agents. Nguyen assembled tetrakis[(4′-phenyl-2,2′:6′,2′′-terpyridine)phenyl]methane (ICu) as organic linker with metal (II) species (M = Zn, Fe, Ni and Ru).76 The covalent entrapment of [Ru(tpy)2]2+ into the backbone of the supra structure enables the Ru–MOF to exhibit luminescence at room temperature.
Grimsdale also studied type-I based MOFs with potential optical properties held together by terpyridine units.77 He showed that a two component system made of zinc mediated assembly of cores ICv bearing electron accepting dialkoxybenzothiadiazoles (n-type) and ICw containing electron rich dialkoxybenzenes (p-type) shows interesting optical properties which can be tuned by the selection of the units (Fig. 12).
At this stage, one should also mention hetroaryl cores 1,3,5,7-tetrakis(4-phenyltetrazol-5-yl)-78 and 1,3,5,7-tetrakis(1,2,4-triazol-4-yl)adamantane79 as well as phosphorous based 1,3,5,7-tetrakis(4-phenylphosphonic acid)adamantane78 which have all three served for the generation of MOFs.
Covalent yet reversible bond formation between aromatic boronic acids and 1,2- or 1,3-diols leading to 5- and 6-membered cyclic boronate esters through condensation, is the most widespread thermodynamically controlled solvothermal self-assembly process.1,82 Boronate-linked networks are assembled with great ease and high efficiency. They are relatively stable to moisture and air and exhibit reversible self-repair capabilities.83
Lavigne4 and Yaghi14 developed this approach almost simultaneously, although Yaghi was the first to actually generate 3D crystalline COFs based on (methanetetrayltetrakis(benzene-4,1-diyl))tetraboronic acid (ICx) and (silanetetrayltetrakis-(benzene-4,1-diyl))tetraboronic acid (ISix). Self-condensation of ICx or ISix delivered a network based on tetrahedral nodes (bor) (Fig. 13, COF-102) and co-condensation with triphenylene-2,3,6,7,10,11-hexaol yielded frameworks based on triangular nodes (ctn) (Fig. 13, COF-108).6 These porous crystalline structures exhibit high thermal stability, up to 500 °C, combined with high surface areas (3472 m2 g−1 BET for COF-102) and extremely low densities (down to 0.17 g cm−3 for COF-108). They also possess excellent gas uptake capacities as exemplified for COF-102 at 35 bar: hydrogen (H2): 72 mg g−1 at 77 K; methane (CH4): 187 mg g−1 at 298 K; carbon dioxide (CO2): 1180 mg g−1 at 298 K.12
Yaghi also studied the adsorption mechanism and uptake of methane in COFs.84 They were able to observe multilayer formation and proved that for 3D COFs assembled through boronic acids, the adsorption sites can be located on the surface of aromatic and boroxine rings. According to these studies, COF-102 has one of the best delivery amounts (difference between the volume adsorbed at 100 and 5 bar).
Zhou studied the structural stability and elastic properties of COF-102 and COF-108.85 Both were found to possess very modest bulk and shear moduli which are typical for MOFs. COF-108 was found close to its structural collapse due to its high porosity and low elastic stiffness. Interesting in this context is that no activation of COF-108 has been reported in the literature. Thus it is unclear if COF-108 retains its crystallinity upon complete removal of guests.
By co-condensation of (methanetetrayltetrakis(benzene-4,1-diyl))tetraboronic acid (ICx) with 1,2,4,5-tetrahydroxybenzene, Ren and Zhu obtained a 3D microporous crystalline network, MCOF-1 featuring a BET surface of 874 m2 g−1 and a uniform pore size of 6.4 Å, by far the smallest among all COFs.86 In contrast to meso- and macroporous materials, the pore size of microporous networks enables an enhancing of the molecular binding between the overlapping atoms in the cavities due to van der Waals forces. This leads in turn to stronger interactions between the framework and the guest molecules. The latter resulted in the present case in exceedingly high selectivity for different mixtures of methane (CH4), ethane (C2H6), propane (C3H8) and ethylene (C2H4). MCOF-1 exhibits a C3H8/CH4 selectivity larger than 1800, and C2H6/CH4 and C2H4/CH4 of 80 and 20 respectively. The selectivity of MCOF-1 can be attributed to both a pore effect and a preferential adsorption. The pore size does not allow two molecules to enter the cavity simultaneously and the apparent aversion for CH4 over the other gases makes the rest.86
Despite their exceptionally high surface area and unreached low densities, COFs have in general no well-developed application.87 Thus, some theoretical and practical efforts to functionalize the interior of these networks in order to target precise applications have been made.
Wang modelled lithium doped borosilicate frameworks for hydrogen storage.88 Kang used ab initio calculations to define COF structures for reversible carbon dioxide uptake at room temperature.89 Goddard employed canonical Monte Carlo (GCMC) simulations to study the hydrogen uptake in lithium, sodium and potassium metallated COFs.90 Zhang performed theoretical investigations on the hydrogen and acetylene storage of calcium and magnesium doped COFs.91 The same group investigated the hydrogen chemisorption, diffusion and associative desorption on COF surface.92
Fischer prepared three inclusion compounds of COF-102 by introducing metallocenes using the solvent free gas phase infiltration method.93 Inclusion of highly stable ferrocene (FeCp2) gave a network with approximately 4 FeCp2 per COF formula unit, (FeCp2)4@COF-102. The COF framework kept its structural integrity and crystallinity. The presence of FeCp2 crystals within the COF structure could be excluded, FeCp adapts however an arrangement which replicates the host structure (Fig. 14). Its uptake was found to be highly reversible. After inclusion of highly reactive cobaltocene (CoCp2), the COF-102 structure was again found intact which proves its robustness.
Fig. 14 Schematic representation of the inclusion structure (FeCp2)4@COF-102. Ferrocene has a closed packing arrangement. |
Uptake of [Ru(cod)(cot)] (cod = 1,5-cyclooctadiene, cot = 1,3,5-cyclooctatriene) revealed strong interaction of the Ru atom with the COF-structure. As [Ru(cod)(cot)] possesses labile ligands, one can imagine substituting them by network interactions. All three metallocenes contain π-electron rich ligands and are thought to interact with the more electron deficient boryl-substituted phenyl rings of the network through π–π donor–acceptor interactions.93
Dichtel functionalized COF-105 with alkyl [–(CH2)11CH3] and allyl [–CH2CHCH2] chains via a monomer truncation strategy.94 By preparing monomers of type I containing only three aryl boronic acids and one aryl alkyl or allyl group, and mixing these monomers with ICx, functionalized COF-102-C12 and COF-102-allyl were obtained respectively. Crystallinity, permanent porosity and high surface area of the original COF-102 were preserved in both cases. The same is true for the original network structure. The truncated monomers were incorporated throughout the lattice rather than on the crystallite surface. Although boroxine formation is reversible, its hydrolysis is thought too slow to allow the liberation of the truncated monomers from the COF-102 interior. COF-102 and COF-102-allyl were both exposed to osmium tetraoxide (OsO4) vapours and then evacuated. COF-102 could be completely emptied whereas evacuated COF-102-allyl revealed the presence of Os atoms throughout the network as a result of OsO4 having reacted with the allylic double bond.94
Zhu prepared the germanium analogues of COF-102 and COF-105.95,96 Both networks are crystalline porous structures with ctn topology. COF-102-Ge has a BET surface of 1288 m2 g−1 and COF-105-Ge exhibits a BET surface of 747 m2 g−1. Both frameworks are composed of luminescent monomer (germanetetrayltetrakis(benzene-4,1-diyl))tetraboronic acid (IGex). Introducing germanium into a network should result in a low reduction potential and a low lying LUMO (lowest unoccupied molecular orbital) due to σ*–π* conjugation initiated by the σ*(Ge)–π*(Ph) interaction. The crystalline backbone of the network largely favours electron delocalization and should increase the electrostatic interaction of the network with guest molecules. Such electron donating frameworks are expected to attract electronegative groups such as nitro functions through coulombic interactions. Guest inclusion is thought to go along with a decrease of the luminescence of the network. COF-102-Ge and COF-105-Ge have thus been tested for the sensitive detection of 2,4-DNT (2,4-dinitrotoluene) and TNT (trinitrotoluene), two common explosives and shown very promising results. Introduction of nitrobenzene, 2,4-DNT and TNT at different concentrations resulted indeed in a significant luminescence quenching of COF-105-Ge.96
Severin extended the scope of the reversible boroxine ring formation by using additional dative B–N bonds in multicomponent assemblies which bear a high potential for extended network generation.82,97
Imine or Schiff base formation, yet another reversible condensation reaction, has also been exploited for the generation of crystalline COFs. Yaghi was again the first to use the reversible imine formation under solvothermal conditions to generate a crystalline network.80 His group used tetrahedral tetraamine ICy and linear terephthalaldehyde (16) to produce porous COF-300 structure bearing diamond topology (Scheme 6). Thermal gravimetric analysis (N2) revealed that COF-300 is stable up to 490 °C, its crystallinity was confirmed by powder X-ray analysis.98 Argon adsorption at 87 K indicated permanent porosity of COF-300 with a BET surface of 1306 m2 g−1.80 The identical network under its amorphous form will be discussed in Section 7.
Scheme 6 Synthesis of COF-300. (A) Single framework of COF-300. (B) Representation of the dia-c5 topology. Contains elements taken with authorization from a JACS paper. |
Maly performed postsynthetic modifications on COF-300.99 Treatment with borane reduced the imine groups to amines and resulted in a basically non reversible network connection motive. The amine network has similar thermal resistance than COF-300, it loses however its permanent porosity (BET surface of 10 m2 g−1) probably because of the enhanced flexibility engendering a structural collapse upon guest removal. Nevertheless, the amine network is still able to take up guest molecules. This new material proved to be much more stable against acidic hydrolysis than COF-300, as the amine connective linkage has no reversible character. The latter could be further derivatized. Adding acetic anhydride transformed the amines into the corresponding amide groups. Evidence for the acetylation was given by infra-red and 13C CP-MAS NMR analysis.99 The authors did not make any statement about the crystallinity of these two newly formed materials.
In order to enhance hydrogen uptake at 298 K, Goddard modified the design of his COF-300 replacing terephthalaldehyde (16) by 2,5-dihydroxyterephtalaldehyde (17) (Scheme 7).100
COF-301 offers a bidentate linker for metal-chelation. The idea was to provide metal sites within the framework and thus enhance hydrogen (H2) uptake capacities. Calculation indicated that COF-301-PdCl2 would reach 60 g total H2 per L at 100 bar which corresponds to 1.5 times the DOE (U.S. Department of Energy) 2015 target. COF-301 could however not be fully metalated. Maybe solvation of the palladium salt prevents it from entering the network.100
Mastalerz followed a similar concept. Reacting 5,5′,5′′,5′′′-methanetetrayltetrakis(2-hydroxybenzaldehyde) (ICz) with benzene-1,2-diamine (18) in the presence of zinc, nickel or palladium acetate resulted in the corresponding polymeric porous metal assisted salphen organic frameworks (MaSOFs) (Scheme 8).101 The metal ion seems to have an essential part during the formation of these materials as in the absence of metal acetate, no network is obtained not even in the presence of trifluoroacetic acid. This suggests that the metal ion rather acts as a template than a Lewis acid and makes this a one pot three component reaction. The nickel containing material MaSOF-1, proved to be a microporous network with a BET surface of 647 m2 g−1, the zinc doped material MaSOF-2 had a similar BET surface of 630 m2 g−1.102 Although, MaSOF-1 is generated through a reversible imine formation and possesses a narrow pore size distribution, MaSOF-1 is an essentially amorphous material. The authors believe that the metal salphen formation is an irreversible process or that the reverse reaction is too slow to allow for a generally reversible imine formation, necessary to obtain a crystalline network. First results concerning gas adsorption indicate that the latter are influenced by the nature of the metal ion present in the network.102
Up to now, all COFs, although crystalline materials have been obtained as microcrystalline powders, limiting their structural determination to modelling supported analysis of powder X-ray diffractograms. Wuest was the first to prepare a monocrystalline COF by reversible self-addition polymerization of tetrahedrally orientated nitroso groups.81 Scheme 9 A shows that reversible dimerization of nitroso derivative 19 leads to cis-20a and trans-azodioxides 20b. Using tetrakis(4-nitrosophenyl)methane (ICaa), tetrakis(4-nitrosophenyl)silane (ISiaa) and 1,3,4,5-tetrakis(4-nitrosophenyl)adamantine (IIaa) respectively, the spontaneous generation of the corresponding diamondoid networks via trans-azodioxy linkages was expected. After the syntheses of the cores ICaa, ISiaa and IIaa, filtration of the mixtures, yellow crystals formed upon standing of the solutions due to spontaneous azodioxide formation with yields ranging from 44 to 63%. The obtained crystals were all uniform in morphology with dimensions approaching 0.5 mm in the case of the IIaa based COF (Scheme 9B).
Scheme 9 (A) Reversible dimerization of two nitroso groups 19 to cis-20a and trans-azodioxides 20b. (B) Crystals of IIaa based COF grown from 4:4:1 (v/v) mesitylene/ethanol/tetrahydrofuran. |
The structure was determined by single-crystal X-ray diffraction and the uniformity of the bulk samples was verified by powder X-ray diffraction. The COF based on IIaa crystallizes in the tetragonal space group P42/n yielding a diamondoid network linked through NN bonds of normal length (1.305 Å) with 35% of the volume available for guest inclusion.81
This selectivity could be enhanced by doping HCP-1 with lithium.105 Two levels of Li doping were tested, 0.35 (HCP-1Li0.35) and 0.55 (HCP-1Li0.55) lithium atoms per naphthalene diimide linker respectively. Analysis with inductively coupled plasma atomic emission spectroscopy (ICP-AES) confirmed the doping amount. Although the overall surface area for CO2 at 273 K slightly decreases with increasing Li doping, probably due to partial pore blocking, the CO2/CH4 selectivity clearly increases with increasing amounts of lithium, cumulating in a CO2/CH4 selectivity of about 170 for HCP-1Li0.55 at low pressures.105 HCP-1 also exhibited substantial removal capabilities of toxic industrial chemicals ammonia, cyanogen chloride and sulfur dioxide with capacities as high as or better than activated carbon.106 Two further HCPs based on imide linkages and tetraamine ICy as monomer have been prepared. One uses perylene dianhydride as linear linker. The resulting network features permanent porosity with a BET surface of 2213 m2 g−1 and micropores of an average size of 5.4 Å.107 The second HCP employs pyromellitic dianhydride as linear connection unit, presents a BET surface of 1454 m2 g−1 and bears pores with an average size of 5.9 Å.108 Wang also prepared the corresponding adamantane polyimide HCP based tetraamine IIy and pyromellitic dianhydride.109 This HCP presents a BET surface of 868 m2 g−1 and shows good adsorption capacities for carbon dioxide, hydrogen, organic and water vapors. Hupp and Nuguyen prepared a catalytically active polyimide HCP based on ICy and bis(phtalic acid) porphyrin as linear linker.110 Treatment with iron chloride (FeCl2) and magnesium chloride (MnCl2·4H2O) produced the metallated HCPs respectively. Metals are believed to be exclusively located within the prophyrin units. Both metallated materials showed catalytic activity in oxidation reactions. Both heterogeneous catalysts showed better turnovers than homogenous porphyrin complexes but much slower initial rates than their homogeneous counterparts. Furthermore, although both HCPs catalysts can be recycled, their catalytic activities are greatly reduced.110
Another example of an HCP acting as heterogeneous catalyst is again based on a tetraamine ICy network, connected with terephthalaldehyde (16) to form a polyimine based HCP.98 Although we have already encountered exactly the same network under its crystalline form COF-300 (see Section 6),80 this time, the obtained material is essentially amorphous as observed by wide angle X-ray scattering (WAXS). Metal doping is obtained be reacting the material with the corresponding metal salts. Thus, three networks bearing respectively copper(I) (HCP-2Cu(I)), copper(II) (HCP-2Cu(II)) and iridium (HCP-2Ir) were prepared (Fig. 15).
ICP analysis revealed a metal uptake of 2.73 wt% of copper(II) and 2.70 wt% of iridium respectively. HCP-2Cu(I) showed a remarkable diastereoselectivity in the cyclopropanation of cyclic olefins. In the case of dihydropyrane (22) for example, a 100:0 trans:cis ratio of 3-oxabicyclo[4.1.0]heptane (24) was obtained using ethyl diazoacetate (23) (Scheme 11A). No metal leaching into the solution was observed after the fourth cycle and the catalyst proved stable with no loss of either reactivity or selectivity. HCP-2Cu(I) was tested in the hydrogenation reaction of alkenes to alkanes. Reactions were run in ethanol at 40 °C, 2 bar H2 and gave generally like for the reduction of 1-octene (25) to octane (26), excellent yields (Scheme 11B). In this case, ICP analysis indicated a slight metal leaching into the solution (0.02%).
El-Kaderi used tetraamine ICy to prepared porous borazine-linked HCP-3 by in situ thermolysis of the generated tetraamino-(-4-aminophenyl)methane borane adducts in monoglyme (Scheme 12).111 HCP-3 features thermal stability up to 400 °C, a BET surface of 2244 m2 g−1 and can store significant amounts of H2 (1.93 wt%) and CO2 (12.8 wt%) at 77 and 273 K respectively.
Zhang generated microporous polyamide HCPs by reacting 4,4′,4′′,4′′′-methanetetrayltetrabenzoyl chloride (ICac) and 4,4′,4′′,4′′′-silanetetrayltetrabenzoyl chloride (ISiac) respectively with piperazine (Scheme 13).112
Resulting HCP-4C and HCP-4Si showed a CO2 uptake of up to 9.81 wt% and a good CO2/N2 selectivity of up to 51 at 1 bar and 273 K. Through an interfacial polymerization on the surface of a porous polyacrylonitrile (PAN) substrate, ultrathin microporous membranes with thicknesses of about 100 nm could be generated and tested for nanofiltration purposes. The latter membranes had a high calcium chloride (CaCl2) rejection and followed the salt rejection sequence of CaCl2 > NaCl > Na2SO4. Solution-based molecular layer deposition of tetra acid chloride ICac and tetraamine ICy yielded HCP-5 which showed similar properties (Scheme 13).113 The rejection sequence of the different salts was found to be dependent on the membrane termination (NH2 or CO2H groups).
Yet another interesting connection motive bringing three monomers together is the trimerization of cyano groups to the corresponding triazines. Kuhn and Thomas turned this normally irreversible reaction into a dynamic and thus reversible process by running it in ionothermal conditions (molten zinc chloride at around 400 °C) and could thus obtain 2D crystalline structures.114 All their attempts to reproduce these results in three dimensions with 4,4′,4′′,4′′′-(adamantane-1,3,5,7-tetrayl)tetrabenzonitrile (IIad) resulted in amorphous HCP-6 (Fig. 16).115
Kuhn extensively studied the possibilities to tune the porosity of their system.115 They found that rather than building block size and geometry, the reaction conditions were essential to control the porosity. Higher temperature for example enabled them – probably via irreversible carbonisation reactions – to generate amorphous bimodal micro- and meso-porous material with control over both pore size and size distribution.10,21 The high temperatures and long reaction times might however cause practical problems as there are only a handful of monomers which withstand such harsh conditions. Zhang could significantly shorten the reaction time using microwave assisted polymerization.116 Cooper went a step further and replaced zinc chloride acting as Lewis acid by trifluoromethanesulfonic acid (TFMS), a strong Brønsted acid.117 4,4′,4′′,4′′′-methanetetrayltetrabenzonitrile (ICad) could thus be polymerized either at room temperature or under microwave assisted synthesis. The resulting HCP has a BET surface of 1152 m2 g−1 and a carbon dioxide uptake of 4.17 mmol g−1 at 1 bar at 273 K.117 Janiak used a triazine adamantane based HPC for carbon dioxide sorption and separation.118
Click chemistry has also found its way into porous organic networks.119 The Husigen 1,3-dipolar cycloaddition reaction has almost simultaneously been employed by three groups. Cooper,120 Nguyen121 and ourselves122 reported independently about core type I based click porous networks. The first two groups clicked tetrakis(4-ethynylphenyl)methane (ICae) and tetrakis(4-azidophenyl)methane (ICaf) under slightly different conditions in order to obtain HCP-7 with a BET surface of 1128 (ref. 120) and 1440 m2 g−1 (ref. 121) respectively (Scheme 14).
Our approach clicked tetraalkyne ICaf and the corresponding adamantane derivative IIaf with the linear diazide 1,4-diazidobenzene respectively.122 The adamantane based HCP proved to be very efficient for CO2 capture at low pressure (155 mg g−1 at relative p/p0 pressure of 0.97 at 195 K).
El-Kaderi reported about benzimidazole-linked HCP-8 based on 4,4′,4′′,4′′′-methanetetrayltetrabenzaldehyde (ICag) and 1,2,4,5-benzenetetraamine tetrahydrochloride (27) (Scheme 15).123 HCP-8 featured a BET surface of 1135 m2 g−1 and remarkable selectivity for methane (CH4), nitrogen (N2) and carbon dioxide (CO2) at 1 bar and 273 K: (CO2/N2): 79 and (CO2/CH4): 10.
We have recently reported about a reversible homopolymer formed through disulphide linkages, its post-functionalization and its controlled depolymerisation under very mild conditions (Scheme 16).124 This was the first time that a 3D HCP based on tetrakis-(4-thiylphenyl)methane (ICah) had been generated through disulphide bridges. Although HCP-9 turned out to be non-porous, its easy synthesis, post-functionalization and most importantly its controlled depolymerisation under mild conditions with nearly quantitative recovery of monomer ICah, promises new applications.
Scheme 16 Synthesis and controlled depolymerisation of HCP-9 (AcOET = ethyl acetate; DTT = dithiothreitol; TCEP = tris(2-carboxyethyl)phosphine; PBS = phosphate buffered saline. |
There is however at least one HCP where the carbon–carbon bond forming reaction is not metal induced. This is the case of the thionyl chloride-catalyzed aldol self-condensation of 1,1′,1′′,1′′′-(methanetetrayltetrakis(benzene-4,1-diyl))tetraethanone (ICai) (Scheme 17).125 Thionyl chloride (SOCl2) generates hydrochloric acid (HCl) in situ which catalyses the dimerization of two acetyl moities into the corresponding β-hydroxyl ketone which in turn may eliminate one water molecule to form an α,β-unsaturated ketone. Under these reaction conditions, there is however yet another reaction taking place, namely the cyclotrimerization of three acetyl units leading to a 1,3,5-trisubstituted benzene. Han observed both linking motives in his microporous HCP-10.125
Scheme 17 Microporous HCP-10 generation through aldol condensation and cyclotrimerization reactions. |
All the following examples of HCP formation via carbon–carbon bond formation will be based on organo-metallic homo- or cross-coupling reactions.
Zhu produced an amorphous porous aromatic framework where tetrakis(4-bromo-phenyl)methane (ICb) is covalently linked together by a nickel(0)-catalysed Yamamoto-type Ullmann cross-coupling reaction (Scheme 18).126 The corresponding HCP-11 has an exceptionally high BET surface of 5640 m2 g−1 and an N2 uptake of 2000 cm3 g−1 at p/p0 = 0.94. HCP-11 is thermally stable up to 520 °C in air and its absolute H2 uptake can reach 10.7 wt% at 48 bar and 77 K,127 its CO2 uptake is 1300 mg g−1 at 40 bar and room temperature.128 A theoretical simulation of high pressure methane adsorption of HCP-11 predicted a total methane uptake as high as 728 mg g−1 at 53.8 bar at 298 K.129
Comotti studied the in situ solid-state polymerization of acrylonitrile in the cavities of HCP-11, proving that confined polymerization in porous frameworks leads to innovative nanostructured materials.130 Zhou showed that post-functionalization of HCP-11 leads to enhanced properties. The sulfonic acid grafted HCP-11-SO3H and its lithium salt HCP-11-SO3Li show increased isosteric heats of CO2 adsorption and uptake capacities.131 The same was true for HCP-11 tethered with different polyamines. The corresponding porous materials showed drastic increases in CO2 uptake capacities at low pressures and very high CO2/N2 adsorption selectivities.132 Qiu and Yao used HCP-11 as a template to nanoconfine ammonia borane (H3NBH3), a very interesting candidate for hydrogen storage due to its high stoichiometric hydrogen content (19.6 wt%) and moderate dehydrogenation temperature.133 The resulting network showed interesting results concerning the hydrogen systemic gravimetric capacity. HCP-12 and HCP-13 have been prepared from the siliane and germane based monomers ISib and IGeb respectively (Scheme 18) and show similar thermal and gas uptake capacities as HCP-11.134,135 Zhou prepared the adamantane based HCP-14 which features a H2 uptake of 4.28 wt% at 77 K (Scheme 18).136
The Suzuki cross-coupling reaction has also been employed to generate HCPs. Kaskel coupled tetrakis(4-bromo-phenyl)methane (ICb) and tetrakis(4-bromophenyl)silane (ISib) as well as (silanetetrayltetrakis(benzene-4,1-diyl))tetraboronic acid (ISix) with aryl halides and aryl boronic acids respectively to obtain the corresponding microporous HCP-15 and HCP-16 with BET surfaces of up to 1380 m2 g−1 and strong hydrophobic character (Scheme 19).137 The same group reacted tetraboronic acids ICx and ISix with linear imidazolium ligand 29 (Scheme 19) and explored the corresponding HCP-17 and HCP-18 as heterogeneous catalysts in N-heterocyclic carbene-catalysed conjugated Umpolung reactions of α,β-unsatured cinnamaldehyde.138
Chen and Han used monomer ICx in a similar approach to generate a benzimidazole containing HCP which was tested as heterogeneous catalyst in Knoevenagel condensation reactions.139 Two further HCPs generated via Suzuki cross-coupling reactions comprise an ICb based material featuring high adsorption capacities for organic molecules140 and a porous framework generated using ICx and heterocyclic linkers for highly selective CO2 capturing.141
Most of the HCPs entirely composed from aromatic units exhibit a strong hydrophobic character. Chen, Qi and Han pushed this concept a little bit further by generating fluorinated HCP-19 via direct C–H arylation polycondensation reaction of tetrabromo ICb and 1,2,4,5-tetrafluorobenzene (30) (Scheme 20).142
Scheme 20 Generation of fluorinated HCP-19, via direct C–H arylation polycondensation reaction. DMAc = dimethylacetamide. |
HCP-19 possesses a BET surface of up to 1170 m2 g−1 and a narrow pore size distribution of about 6.3 Å and due to its highly lipophilic character, a high adsorption ability for toluene (976 mg g−1 at saturated vapour pressure and room temperature).142
Besides tetrabromo and tetraboronic acid monomers, tetrakis(4-ethynylphenyl)methane (ICae) has also been used by a number of groups in organo-metallic homo- and cross-coupling reactions. Zhou homo-coupled ICae as well as the corresponding adamantane monomer IIae via oxidative Eglinton coupling.136 Several groups employed the octacarbonyldicobalt [Co2(CO)8]-catalysed alkyne trimerization reaction. Lin polymerised ICae with phosphorescent [Ru(bpy)3]2+ (byp = 2,2′-bipyridine) and [Ir(ppy)2(bpy)]+ (ppy = 2-phenylpyridine) respectively and used them for various photocatalytical reactions like, amongst others, aza-Henry reactions.143 Nguyen prepared catechol-functionalized HCP-20 (Scheme 21) which was post-synthetically metalated using copper, magnesium and manganese salts respectively.144 The density of catechol units can be tuned via the co-monomer stoichiometries during polymerization. Metalated HCP-20M shows enhanced heat of H2 adsorptions compared to non-metalated HCP-20.
Wang prepared a chiral HCP by embedding the Jørgensen–Hayashi catalyst and using the resulting porous HCP for heterogeneous catalysed asymmetric Michael addition reactions.145
Zhu used tetraalkyne ICae in Sonogashira–Hagihara cross-coupling polymerization with 2,4,6-tribromo-benzene-1,3,5-triol to produce a triol containing HCP which was subsequently lithiated.146 Latter HCP showed enhanced CO2/N2 selectivity and improved H2 storage capacity compared to the non-metalated network. Yu used the same coupling reaction but with the tetraiodo monomer ICc reacted with the corresponding alkyne functionalized porphyrin 32 (Scheme 22).147 The Ni-porphyrin units within HCP-21 are in part responsible for a CO2/N2 selectivity of HCP-21 and good isosteric heats of adsorption for H2 and methane.
Son cross-coupled tetraalkyne ICae with a metal complexing salene moieties during a Sonogashira reaction and studied their properties.148
Chen synthesized tetraphenylmethane based oxadiazolyl cores bearing interesting properties for light emitting devices.152 Zuilhof studied tetrahedral n-type materials based on tetra naphtalene imide derivatives.103
Tschierske prepared carbon and tin based cores ICal and ISnam bearing eight lipophilic alkane chains and proved that they did not show any birefringence property when they were explored as potential mesophases (Fig. 17).153
Corriu produced similar germane and tin core type I constructs bearing each four trisisoproxysilyl group. Sol–gel hydrolysis–polycondensation resulted in supramolecular self-organisation based on van der Waals interactions.154
Park generated solution processable microporous organic networks via organic sol–gel synthesis by reacting either tetrakisamine ICy with diisocyanate 33 or tetrakisisocyante ICan with diamine 34 to form the urea linked network 35 (Scheme 23).155
Kaskel prepared microporous hydrophibic polysilane 36 by reacting tetrakis(4-bromophenyl)silane ISib and tetraethylorthosilicate [Si(OC2H5)4] (Scheme 24).156 The resulting network 36 has a BET surface of 780 m2 g−1 and good gas adsorption properties.
Besides tetrakisphenylmethane cores, some groups have investigated building blocks composed of a central carbon atom bearing directly the connection motives without phenyl groups as spacers. Several groups have investigated tetraethynlymethane as building block.46,83,97,114,115 Venkatraman synthesised tetrapentenylmethane and investigated its network forming cabailities153 and Banert prepared and studied explosive tetraazidomethane.74
First of all there is the use of simply to handle, high yielding reactions generating the networks, that might fulfil the strict criteria of a click reaction or not. Although people have always been looking for such reactions, only a few of them have been exploited by now. Sulphur chemistry – disulphide formation, thiol–ene and thiol–yne reactions – bear a lot of potential. There are also a lot of reactions already commonly used in polymer sciences such as the nitroxide(-mediated) radical coupling or the nitrile imine-mediated tetrazole–ene cycloaddition (NITEC) which bear a lot of potential. Searching for such type of reactions, one should not forget post-functionalisation which is also a very powerful tool to access desired networks or drastically enhance/tune their properties. Again this research area is only in its infancy and there are lot more opportunities still waiting.
A second important development area is the controlled generation of networks on surfaces. This will not only enable the generation of completely new materials through a layer-by-layer approach but also contribute to the basic understanding of formation by joining surface analytical tools to the classical ones used for the bulk syntheses.
Last but not least there is the economic impact of these materials. Although some networks have been predicted to open billon euro markets, to date, the industrial applications are still limited. Some HCPs have found application in niche markets and a few global players are investing in MOFs but there has not yet been a real breakthrough. It seems as if this research field has some difficulties to spread out of the research laboratories. Let's hope it is just a matter of time.
This journal is © The Royal Society of Chemistry 2014 |