15327Ordered porous materials for emerging applications

15327Ordered porous materials for emerging applications

(Parte 1 de 2)

Orderedporous materials for emergingapplications

Mark E. Davis

Chemical Engineering, California Institute of Technology, Pasadena, California 91125, USA

“Space—the final frontier.” This preamble to a well-known television series captures the challenge encountered not only in space travel adventures, but also in the field of porous materials, which aims to control the size, shape and uniformity of the porous space and the atoms and molecules that define it. The past decade has seen significant advances in the ability to fabricate new porous solids with ordered structures from a wide range of different materials. This has resulted in materials with unusual properties and broadened their application range beyond the traditional use as catalysts and adsorbents. In fact, porous materials now seem set to contribute to developments in areas ranging from microelectronics to medical diagnosis.

orous solids are of scientific and technological interest because of their ability to interact with atoms, ions and molecules not only at their surfaces, but throughout the bulk of the material. Not surprisingly, traditional applications of porous materials thus involve ion exchange, adsorption (for separation) and catalysis, and many of these benefit from the high order that can be achieved in solids such as zeolites.

The pores of solids are classified according to size: pore sizes in the range of 2nm and below are called micropores, those in the range of 2nm to 50nm are denoted mesopores, and those above 50 nm are macropores. The distribution of sizes, shapes and volumes of the void spaces in porous materials directly relates to their ability to perform the desired function in a particular application. The need to create uniformity within the pore size, shape and volume has steadily increased over recent years because it can lead to superior applications properties. For example, a material with uniform micropores, such as a zeolite, can separate molecules on the basis of their size by selectively adsorbing a small molecule from a mixture containing molecules too large to enter its pores. Clearly, a distribution of pore sizes would limit the ability of the solid to separate molecules of differing sizes.

In addition to the pore space, the atoms in the solid creating that space can be important. For example, molecular sieves comprising pure silica are hydrophobic and can adsorb organic components from water, whereas molecular sieves comprising aluminosilicate are hydrophilic and can thus adsorb water from organic solvents. Because the control over the uniformity of pore space and the composition of the solid that creates the space is of importance, this review concentrates on porous materials with more ordered structures. Here, I will cover significant issues in microporous materials that have occurred over the last decade or so. In addition to microporous materials, the well-ordered mesoporous solids will be discussed, as they have recently received much attention. The type of materials described here are generally synthesized at low temperatures (less than 473K) and pressures (a few atmospheres) and in aqueous media. The syntheses do not normally give the thermodynamically most stable product but rather a material that is the result of a kinetically controlled synthetic pathway. Although the starting reagents and compositions and the precise procedures will differ for each type of material, the syntheses are reproducible and thus allow for use in applications. A brief review on zeolites and molecular sieve syntheses that discusses these issues is available1,a s are reviews on microporous and ordered mesoporous materials syntheses and their applications as catalysts2.

In 1988, the first report of a crystalline microporous material with uniform pores larger than 1.0nm appeared3, the ability to syn- thesize well-ordered mesoporous materials was about to be announced4, and scaffolding-lik e (but non-porous) structures built from three-dimensional, linked modular molecular units5 were appearing in the literature. Moreover, zeolite-based membranes with ‘molecular sieving’ properties were discussed6, and the inclusion of electro- and photo-active guest molecules in porous host materials7 gave rise to interesting phenomena. But despite their promise for advanced materials applications, none of these materials led to practical successes. Today, the situation is rather different: our understanding of the structure of ordered porous materials and how to control and tune it has increased considerably and advances have led to practical applications. In fact, the future

Figure 1 Pore characteristics in the aluminophospates AlPO4-1, AlPO4-5 and VPI-5.

a, Representations of the pores in the aluminophosphates AlPO4-1, AlPO4-5 and VPI- 5. The line segments represent oxygen atoms that bridge between two tetrahedral atoms (intersection points) that are in this case either Al3 or P5 with strict alternation to give a composition of AlPO4. Rhomboids indicate the unit cell. b, Measured pore sizes by argon adsorption techniques. Note that VPI-5 shows a pore diameter above

1.0 nm. (DW is the change in adsorption mass and Dr is the change in radius of the pore size.) review article uses of ever more sophisticated versions of these materials look very promising.

Microporous materials with large pores

The preparation of the aluminophosphate VPI-5—the above-mentioned crystalline microporous material with uniform pores larger than 1.0nm—opened up the area of extra-large pore crystalline materials. Extra-large pores are obtained if more than 12 oxygen atoms span the circumference of the pore (see also Fig. 1), and the resultant pore size allows practical applications for which materials with smaller pores are not suitable. In VPI-5, the pores are circular one-dimensional channels that, owing to the crystallinity of the material, have an absolutely uniform diameter of 1.2nm and endow the material with an approximately 30% void fraction8. Following the discovery of VPI-5, numerous extra-large pore materials were synthesized. Table 1 lists typical (though not all) representatives of such materials, all having structures that contain rings made of more than 12 oxygen atoms and most of them phosphate-based. Except for the silicas, these extra-large pore materials exhibit in their as-synthesized form at least one of the following features: (1) mixed metal-ion coordination (such as aluminium in octahedral and tetrahedral coordination), (2) terminal OH groups, and (3) the presence of other non-tetrahedral framework species (such as OH,

H2O, F). These featureslower the framework stability relative to that of fully tetrahedrally coordinated materials such as zeolites and their pure-silica analogues. For example, JDF-20 (see Table 1) decomposes upon removal of occluded organic species9, and exchange of the 1,2-diaminocyclohexane molecules from as-synthesized ND-1 by alkali-metal cations and small alkylammonium cations transforms the material to other products10.

The diversity in porosity encountered in extra-large pore crystalline materials is strikingly illustrated by the phosphate-based materials listed in Table 1. For example, the pore shape of cloverite resembles a cloverleaf (hence the origin of the name)1: the 20- membered ring enclosing the pore contains four terminal OH groups that intrude into the opening to provide the cloverleaf shape. Cloverite is thus unlikely to accommodate molecules of a certain size and shape that can be accommodated in the circular pores of VPI-5. In fact, the role of pore size as well as pore shape in effecting the separation of molecules has recently been demonstrated with ETS-4 (ref. 12): as illustrated schematically in Fig. 2, the pore shape of this material is crucial to discriminate between unsymmetrical N2 and symmetrical CH4. This principle, which has led to separation devices that are nowcommercially available (as

Molecular Gate Technology; details available from molecular.gate@engelhard.com), might be extendable to materials with larger pores and the separation of larger molecules.

In addition to pore shape variations, the materials listed in Table 1 also exhibit pronounced differences in the shape and size of their void space. For example, the pore system of cloverite is threedimensional, containing 3-nm cages that are each accessible through six clover-shaped pores; in contrast, VPI-5 has one-dimensional and uniform pore channels. The fact that cloverite has large cages to which smaller pores provide access allows for the construction of inclusion species and ‘ship-in-a-bottle’ syntheses, where individual reagents are sufficiently small to enter through the pores, but the assembled inclusion complex is then too large to escape the cages.

The practical value of phosphate-based extra-large pore materials is limited by their relatively poor thermal and hydrothermal stabilities as compared to those of silica-based molecular sieves. Although some of the phosphate materials are sufficiently stable for certain (low-temperature) application areas, concern was raised over whether all extra-large pore materials would lack stability, owing to the presence of extra-large rings in their structures. In the case of VPI-5, the lack of stability has been attributed to the nature of the structural units, rather than to the presence of the extra-large ring (ref. 13 and references therein). This was inferred from VPI-5 and the aluminophosphate AlPO4-H2 being made of the same basic structural units and exhibiting similar instabilities, even though the latter has one-dimensional pores comprising only 10-membered rings. More direct evidence for this interpretation was provided by the successful synthesis of the two extra-large pore silicas UTD-1 (ref. 14) and CIT-5 (ref. 15), which are both crystalline silicas containing 14-membered pore rings (see Table 1). The thermal and hydrothermal stabilities of these materials are comparable to the stabilities of other zeolites containing smaller rings within their structures, thus confirming the view that the presence of extra-large rings does not itself result in destabilization. Rather, the lack of stability seen in phosphate-based materials is due to other structural features, for example, mixed metal-ion coordination, terminal OH groups and the presence of non-tetrahedral framework species such as OH, H2O or F (as mentioned above). Given the limited stability of phosphate-based materials, the synthesis of extra-large pore crystalline silicas is a promising advance. Both UTD-1 and CIT-5 (see Table 1) have one-dimensional pore systems, with elliptical (0.75 £ 1.0nm) and circular (0.75nm) pore shapes, respectively. These materials provide the precedent for the ability to synthesize extra-large pore crystalline solids and apply them to a greater variety of commercially relevant applications.

Porous metal–organic frameworks

A different approach to preparing microporous solids involves the coordination of metal ions to organic ‘linker’moieties, thus yielding open framework structures. In fact, these materials have a long history, and examples include transition metal cyanide compounds (early examples are Hofmann-type clathrates, Prussian-Blue type structuresandWernercomplexes)andthediamond-likeframework bis(adiponitrilo)copper (I) nitrate16. Open frameworks comprising metal–organic units gained renewed inerest in the 1990s, but the inability of these solids to maintain permanent porosity and avoid structural rearrangements upon guest removal or guest exchange (some leading to complete collapse of the framework) has been an obvious shortcoming. However, metal–organic frameworks (MOFs) that exhibit permanent porosity have now been prepared17–19. The first such solid was MOF-5, which consists of

Table 1 Representative examples of crystalline materials with ring sizes above 12

Material Year reported Main framework composition Ring size (oxygen atoms) Pore size (nm)* Reference

NTHU-1 2001 GaPO 24 ND 136 ND, Not determined. *By adsorption. †Structural collapse upon removal of organic.

review article

Zn2 and 1,4-benzenedicarboxylate and has a microporous volume larger than any known zeolite17. Another example is a coppercontaining framework with pores of 1.64nm diameter18. Recently developed MOF-type solids exhibit the lowest densities reported for any crystalline material (0.41–0.21 g cm23)a sw ella sah igh methane storage capacity19.

Porous metal–organic frameworks are unlikely to compete with zeolites and other oxide-based porous materials in high-temperature applications owing to their limited long-term stability under such conditions and high cost. But the ability to prepare such solids, including frameworks with extra-large pores and high pore volumes, is nevertheless likely to open up many application possibilities in niche areas. For example, the methane adsorption capacity of solids based on copper dicarboxylates and triethylenediamine20 and of MOF-type solids19, exceeds that of any other known crystalline material. The functionalization of the organic component of the framework, or incorporation of functional organic groups directly into the framework, may yield porous solids that contain different groups capable of binding guests and/or catalysing chemical reactions involving adsorbed guests. However, organic-functionalized porous oxides are now also available21 and are likely to compete successfully with metal–organic frameworks in such applications.

Unique application possibilities may arise from the ability to exploit the metal component and/or its interaction with guest molecules to design porousmaterials with unusual physicochemical properties, such as redox potentials, light absorption properties or magnetic moments. Moreover, the metal–organic framework approach seems particularly suited to constructing a wide range of chiral porous materials22,23 for applications requiring enantioselective adsorption and catalysis. In contrast, the formation of chiral porous structures from oxides is proving very difficult, despite one example of enantioselective catalysis and separation using a partially chiral zeolite1.

Ordered mesoporous materials

In parallel to the above work on crystalline microporous materials with extra-large pores and on crystalline metal–organic porous solids, the discovery and development of well-ordered, mesoporous materials occurred. Intense focus on ordered mesoporous materials waslargely initiated byareport4fromworkersat Mobil in 1992, who described the successful preparation of mesoporous silicas with hexagonal and cubic symmetry and pore sizes ranging from 2 to 10nm, through the use of surfactants as organizing agents. It was clear that extra-large pore, crystalline materials like VPI-5, AlPO4-8 and cloverite influenced this work, and at that time, no conclusions were drawn as to whether or not the walls of the ordered mesoporous materials had precise atomic ordering4. In fact, the hexagonal variant of the material was speculated to be a larger-pore version of VPI-5 (see Fig. 3).

The 29Si nuclear magnetic resonance (NMR) spectrum of Mobil’s mesoporous material MCM-41 suggested a disordered wall structure. But this evidence is not conclusive, given that crystalline puresilica materials with *BEA topology exhibit similar 29Si NMR spectra, owing to a large number of framework defect sites. Proof for the lack of crystallinity of MCM-41 was the subsequent observation24 that dehydrated samples gave Raman bands indicative of planar 3-membered ring stretch vibrations; crystalline materials containing 3-membered rings do not reveal these vibrations25 because the rings are no longer isolated when contained within a continuous three-dimensional framework26. Moreover, the spectra matched those of amorphous oxides and glasses with 3-membered rings attheir surfaces.Thelackofpreciseatomic positioning in these well-ordered mesoporous materials thus suggests that analogous,

Figure 2 Pore size and shape of ETS-10. a, Circular pore shape does not discriminate between nitrogen and methane. b, Elliptical shape of pore, obtained upon heat treatment, allows nitrogen adsorption only.

Figure 3 Transmission electron micrographs of VPI-5 and MCM-41. a, Micrograph of VPI-5; b, micrograph of MCM-41; note the difference in scale. Panel b is adapted from ref. 24, with permission from Elsevier Science.

review article well-ordered mesoporous materials can be prepared using element combinations found in other amorphous materials, such as aerogels, xerogels, organic–inorganic hybrid materials, metals and mixed-metal alloys. Indeed, some of these materials are now available in ordered, mesoporous forms27–29. Very recently, an ordered, hexagonal mesoporous benzene–silica composite was prepared30 from the assembly of 1-4-bis(triethoxysilyl)benzene

[(C2H5O)3Si–C6H4–Si(OC2H5)3] and alkytrimethylammonium surfactants. This material exhibits periodicity in the hexagonal arrangement of the mesopores obtained upon surfactant removal (5.25 nm lattice constant) and periodicity in the walls along the channel direction with a spacing of 0.76nm (silica and aromatic units alternate). It will be interesting to see whether this unique arrangement can be generalized to hybrid materials made of other combinations of organic and inorganic units.

Although the report on MCM-41 by Kresge et al.4 stimulated recent work on ordered materials with uniform mesoporosity, evidence that such materials could be prepared precedes their publication31,32. In fact, materials of the type reported by Kresge et al.4 may have already been made decades ago: in 1971, a material described as low-bulk density silica was obtained through hydrolysing and condensing tetraethylorthosilicate in the presence of cationic surfactants33. The porosity and structure of the low-bulk density silica were not reported, but attempts to reproduce the synthesis revealed that it may have yielded ordered mesoporous materials34. The first clear demonstration of the successful preparation of a material exhibiting uniform mesoporosity was achieved by Manton and Davidtz32, who obtained amorphous aluminosilicates with fairly uniform pores, by using quaternary ammonium cations such as tetrabutylammonium and tetrapropylammonium cations during the synthesis. Later, Yanagisawa et al. used cationic surfactants to synthesize silicates with uniform mesoporosity31.

The work by Kresge et al.4 was thus not in the absence of earlier claims of having prepared materials with uniform mesoporosity. But it nevertheless constituted a breakthrough because the high degree of order achieved in MCM-41 was unprecedented in the literature; there was little evidence of structural order in the materials reported before. (In fact, controlled-pore glass can have a very uniform mesopore size yet have no structural order in the glass.) Moreover, Kresge et al.4 identified the connection between the ordering observed in the mesoporous materials and the structure-directing aggregation properties of the surfactants used in their synthesis. (Block copolymers also form mesoscopic assemblies and are now used in the preparation of ordered mesoporous materials.)

Thecombination ofthesetwocontributions—thenoveltyofthenew materials, with their unprecedented high degree of ordering, and the introduction of a clear conceptual framework that could guide the design of new materials—stimulated an enormous amount of work. Today, a large number of mesoporous materials of varying composition, pore size and inorganic wall thickness are available.

With the exception of the ordered, hexagonal mesoporous benzene–silica composite prepared by Inagaki et al.30, ordered mesoporous materials are not crystalline, thus permitting their synthesis by many different routes, as illustrated by the examples reported in references 35–39. As might be expected, the degree of ordering can depend significantly on the method used to synthesize the mesoporous material, ranging from the high degree of ordering first reported by Kresge et al.4 to an almost complete lack of ordering. That is, while uniformly sized mesopores are produced, the degree to which they form ordered arrangements within the material shows tremendous variability.

It remains unclear why the vast majority of ordered mesoporous materials cannot be synthesized in crystalline form. The recent example ofstructural orderinthe wallsofthemesoporous benzene– silica composite30 may help shed light on this issue. Regarding the inorganic mesoporous materials, their syntheses occur under thermodynamic conditions comparable to those encountered during the syntheses of the more open zeolite frameworks40, suggesting that thermodynamic barriers in the assembly process are not causing the lack of crystallinity. Ordered mesoporous materials were first prepared using charged organic components, and charge matching between the organic and inorganic components should influence the assembly of ordered mesoporous solids. This aspect of their synthesis may be an impediment to crystallinity41.H owever,t he ordered mesoporous materials can now be prepared using non-ionic organic compontents. These assemblies would suggest that charge matching is not the only impediment to crystallinity. Likewise, it is now known that not only ordered mesoporous materials can form through intermediate phases that are layered, but also zeolites (examples include MCM-2 (ref. 42) and ferrierite43). The involvement of layered intermediates during formation thus does not seem to be why ordered mesoporous materials do not crystallize.

A plausible explanation for the lack of crystallinity may lie in the correlation between the framework density (FD, the number of atoms per nm3) and the structural features of porous materials. In 1989, Brunner and Meier revealed that crystalline structures containing tetrahedral atoms (T-atoms) exhibit a correlation between their FD and the parameter MINR, the minimum ring size for every T-atom44. (If the structure has T-atoms in rings of different sizes, then a “ ” is assigned. For example, MINR 4 denotes that some of the T-atoms are in 4-membered rings, while others are in rings of larger size.) All crystalline and ‘tetrahedral’ materials follow this relationship. In the case of oxide-based solids, most structures have MINR > 4, which would correlate with a maximum void fraction of about 0.5. The correlation does not hold for crystalline structures built from units containing non-tetrahedral atoms, which have indeed been shown to exceed this void fraction17–19. Regardless of the phenomena that are the origin of the correlation between the FD and MINR, they may be the reasons why the ordered mesoporous materials are not crystalline (void fractions violate this relationship).

In order to obey the correlation, ordered mesoporous materials would need to have MINR 3 ; that is, they would need to contain a large number of three-membered rings (3MRs). The recognition that 3MR building units are essential if very open framework materials are to be realized45 initially focused attention on beryllosilicate materials because many beryllosilicate minerals, albeit dense, have a large number of 3MR. In fact, the successful synthesis46 of an analogue of the mineral lovdarite illustrated that synthetic materials with a high number of 3MRs could be prepared. However, beryllium is highly toxic, and practically more useful implementations of the ‘3MR concept’ based on the FD-MINR relationship will therefore depend on the development of synthetic routes to 3MR structures using elements other than beryllium47.

Zinc seems a promising and non-toxic substitute for beryllium, given that some dense zincosilicates have isostructures to beryllosilicates that contain a large number density of 3MRs. Zincosilicate syntheses aimed specifically at the preparation of microporous crystalline solids containing 3MRs have therefore been developed48,49.

Table 2 Synthetic framework materials with numerous three-membered rings

Material Composition FD (atoms per nm ) Pore size Year Reference

OSB-2 Beryllosilicate 12.7 8 MR 2001 50 FD, framework density; MR, membered rings.

review article

Table 2 lists the synthetic crystalline materials that possess a significant number of 3MRs, which include zincosilicates, lithium silicates and a Ge2ZrO6F material. The materials listed illustrate the significant progress in the synthesis of MINR 3 materials, which now provides routes to new porous solids. Recently, an extra-large pore beryllosilicate with MINR 3 has been reported50, which constitutes a promising extension of the ‘3MR concept’ of Meier45. Although beryllium’s toxicity precludes practical use of this material, the conversion to a zincosilicate isostructure should be feasible.

As stated above, the reason that ordered mesoporous materials cannot be synthesized in a crystalline form is unknown. Materials with the pore volumes of the ordered mesoporous materials would require MINR 3 if they were to be tetrahedral frameworks, so it is possible that the appropriate oxide chemistry that wouldallow the realization of such materials has not yet been explored. A recently reported ordered mesoporous zincosilicate51 did not contain sufficient amounts of zinc to facilitate the formation of a significant number of 3MRs, and further exploration of zinc-based materials might thus allow the preparation of ordered mesoporous materials that are crystalline.

Hierarchical structures

Interest in hierarchical porous structures has burgeoned over the past decade. Zeolite particles have been used as building blocks to construct hierarchical porous structures. Pure, crystalline molecular sieves have been produced as free-standing films52, and films, spheres and fibres have been constructed from nanometre-sized crystals53–5. Techniques that use colloidal suspensions to shape ceramic structures are well established, and their extension to microporous colloidal particles for casting gels into designed shapes56 and for film casting57 provides flexible synthetic routes to a range of hierarchical microporous structures.

(Parte 1 de 2)

Comentários