Ordered mesoporous materials

Ordered mesoporous materials

(Parte 1 de 3)

Review Ordered mesoporous materials

Ulrike Ciesla a,b, Ferdi Schuth a,* a Johann Wolfgang Goethe-Universitat Frankfurt, Institut fur Anorganische und Analytische Chemie,

Marie-Curie-Str. 1, D-60439 Frankfurt-am-Main, Germany b Department of Chemistry, University of California–Santa Barbara, Santa Barbara, CA, USA

Received 5 January 1998; received in revised form 13 March 1998; accepted 15 March 1998


About six years after the first publication, ordered mesoporous oxides can be prepared by a variety of procedures and over a wide range of compositions using various diVerent surfactant templates. The mechanisms of formation, although still a matter of discussion, are understood in principle, and the macroscopic morphology as well as the orientation of the pores can be controlled in fortunate cases. However, still lacking are groundbreaking developments in the field of applications, either in catalysis or other areas. The state-of-the-art in the synthesis, characterization and application of ordered mesoporous oxides will be covered in this review. © 1999 Elsevier Science B.V. All rights reserved.

Keywords: Review; Mesoporous materials; MCM-41

1. Introduction the main aspects in zeolite chemistry. Larger pores are present in porous glasses and porous gels,

Porous solids are used technically as adsorbents, which were known as mesoporous materials at the catalysts and catalyst supports owing to their high time of the discovery of MCM-41 [2]. However, surface areas. According to the IUPAC definition they show disordered pore systems with broad [1], porous materials are divided into three classes: pore-size distributions. Other mesoporous solids microporous (<2 nm), mesoporous (2–50 nm) and were synthesized via intercalation of layered matemacroporous (>50 nm). Well-known members of rials, such as double hydroxides, metal (titanium, the microporous class are the zeolites, which pro- zirconium) phosphates and clays. These also have vide excellent catalytic properties by virtue of their very broad mesopore-size distributions, as well as crystalline aluminosilicate network. However, their additional micropores. applications are limited by the relatively small pore With MCM (Mobil Composition of Matter) 41 openings; therefore, pore enlargement was one of the first mesoporous solid was synthesized that showed a regularly ordered pore arrangement and a very narrow pore-size distribution. After the * Corresponding author. Present address: MPI fur

e-mail: schueth@mpi-muelheim.mpg.de (1) characterization; (2) the mechanism of forma-

tion; (3) the synthesis of new materials based on the MCM-41 synthesis concept; (4) morphology control; and (5) the technical applications of MCM-41 and related mesoporous materials. These fields will be surveyed in this review.

2. Characterization

A procedure for the preparation of low-density silica was already described in a patent filed in 1969 [3]. Di Renzo et al. [4] reproduced the synthesis reported in the patent and found that it leads to a material identical to mesoporous MCM-41, which scientists of the Mobil Oil Corporation patented in 1991 [5]. However, in the original patent only few of the remarkable proper- Fig. 1. X-ray diVraction pattern of high-quality calcined ties of the material were actually discovered. It MCM-41 made by Huo and Margolese [7]. was the Mobil scientists who really recognized the spectactular features of these ordered mesoporous oxides. The question thus arises: why is the charac- to contain substantial amounts of MCM-41. Corma recently attributed the apparently ‘‘worse’’terization of these mesoporous materials so diY- cult? As the most investigated member of the XRD pattern to the formation of smaller although no less ordered MCM-41 crystallites [10]. In ourM41S family, MCM-41 provides an excellent example of the problems in characterizing meso- group the diVraction patterns of MCM-41 with diVerent degrees of defects were recently simulatedporous materials. MCM-41 has a honeycomb structure that is the result of hexagonal packing [1]. It was found that even when the hexagonal pore structure contains a large number of defects,of unidimensional cylindrical pores. Reliable characterization of the porous hexagonal structure a hexagonally indexable three-reflection pattern can be calculated. Decreasing domain size, how-requires the use of three independent techniques [6]: X-ray diVraction (XRD), transmission ever, leads to loss of high-order reflections. To elucidate the pore structure of MCM-41electron microscopy (TEM) and adsorption analysis. transmission electron microscopy is usually used.

Fig. 2 shows a TEM image of the hexagonalThe XRD pattern of MCM-41 shows typically three to five reflections between 2h=2° and 5° arrangment of uniform, 4 nm sized pores in a sample of MCM-41. However, the exact analysis(Fig. 1), although samples with more reflections have also been reported [8,9]. The reflections are of pore sizes and thickness of the pore walls is very diYcult and not possible without additionaldue to the ordered hexagonal array of parallel silica tubes and can be indexed assuming a hexago- simulations because of the focus problem. Chen et al. [12] showed for MCM-41 that the thicknessnal unit cell as (100), (110), (200), (210) and (300). Since the materials are not crystalline at the of the features—the pore sizes and wall thicknesses—depend strongly on the focus conditions,atomic level, no reflections at higher angles are observed. Moreover, these reflections would only and careful modeling is necessary for precise analysis. Moreover, most MCM-41 samples not onlybe very weak in any case, owing to the strong decrease of the structure factor at high angles. By show ordered regions but also disordered regions, lamellar and fingerprint-like structures [13]. Themeans of X-ray diVraction it is not possible to quantify the purity of the material. Samples with existence of a lamellar phase after calciniation is unlikely, because silicate layers are too distantonly one distinct reflection have also been found the adsorption and desorption branches (Fig. 3).

The adsorption at very low relative pressure, p/p0, is due to monolayer adsorption of N2 on the walls of the mesopores and does not represent the pres- ence of any micropores [21,2]. However, in the case of materials with pores larger than 4.0 nm

[23] or using O2 or Ar as adsorbate [17], the isotherm is still type IV but also exhibits well-

defined hysteresis loops of the HI type [20]. The presence and size of the hysteresis loops depend on the adsorbate [17], pore size [23] and temperature [24]. Non-local density functional theory (NLDFT) provides an accurate description of the thermodynamics of nitrogen confined in pores of this size and predicts the thermodynamic limits for the adsorption–desorption hysteresis loops [25]. Comparison of the theoretically predicted thermo- dynamic hysteresis loops with experimental dataFig. 2. Transmission electron micrograph of MCM-41 featuring 4.0 nm sized pores, hexagonally arranged. We should like to on MCM-41 supports the classical physical sce- thank BASF AG, Ludwigshafen for recording the image. nario of capillary condensation in an open cylindrical capillary. In contrast, there is still no clear explanation for those cases where the hysteresisfrom one another to preserve the spacing in the silicate organic phase and collapse without addi- loops are absent. Calculations by Ravikovitch et al. [2,25] showed that the absence cannot betional post-treatments. Chenite et al. [14] showed that the equidistant parallel lines observed in the explained by the capillary critical temperature being achieved, which was previously assumedmicrographs are related to the hexagonal repeat between tubules. The honeycomb structure is [26]. To determine the pore-size distributions in cylin-suYciently regular to give fringes in projection under proper orientation of the specimen. Feng drical pores, several methods are known based on geometrical considerations [27], thermodynamicset al. [15] investigated the curved, fingerprint-like structures in more detail, and observed two dislocation and two disclination defect structures which are similar to those detected in pure liquid-crystal phases. They proposed that since the mesoporous structure cannot shear without fracturing, these defects must have formed in the unpolymerized liquid-crystal-like silicate precursor phase.

Adsorption of probe molecules has been widely used to determine the surface area and to characterize the pore-size distribution of solid catalysts. Soon after the preparation of MCM-41 the phy- sisorption of gases such as N2,O 2 and Ar had been studied to characterize the porosity [16–19].

The nitrogen adsorption isotherm for MCM-41 with pores of around 4.0 nm, which is type IV in the IUPAC classification [20], shows two distinct features: a sharp capillary condensation step at a Fig. 3. Adsorption isotherm of nitrogen on MCM-41 with 4.0 nm pores at 7 K [16].relative pressure of 0.4 and no hysteresis between

[20] or a statistical thermodynamic approach [25]. independent studies showed in the case of the original MCM-41 synthesis that the wall thicknessIn addition, freezing point depression can be used of around 1.0 nm remains constant if the poreand was followed by nuclear magnetic resonance sizes are varied in the range from 2.5 to 10.0 nm(NMR) spectroscopy in the case of MCM-41 [28]. [32,3].The traditional method for analyzing pore-size The characterization of the pore walls focusesdistributions in the mesopore range is the on two aspects: (1) structural properties; and (2)Barrett–Joyner–Halenda (BJH) method [29,30] the surface chemistry. From X-ray diVraction atwhich is based on the Kelvin equation and, thus, high angles it is known that mesoporous MCM-41has a thermodynamic origin. However, compared materials are amorphous. Therefore, the walls arewith new methods that rely on more localized best analyzed in terms of local atomic ordering. Adescriptions such as density functional theory very powerful method for the characterization of(DFT) [2,25] and Monte Carlo simulation (MC) framework locations is solid-state NMR spectro-[31], the thermodynamically based methods overscopy. Uncalcined MCM-41 samples show threeestimate the relative pressure at desorption and diVerent 29Si-NMR peaks which can be assignedtherefore underestimate the calculated pore diameto Q2,Q 3 and Q4 silicon species [21]. After calcina-ters by ca. 1.0 nm. Moreover, the theoretical basis tion Q4 environments are formed at the expensefor the BJH analysis becomes fairly weak if the of Q3 and/or Q2. The degree of this transformationstep for N2 at 7 K lies below p/p0=0.42, because depends on the severity of reaction conditionsthis is considered to be the stability limit of the applied to the condensation of Si–OH groups.meniscus. Pore sizes calculated in such cases are Because of the formation of catalytically activestill probably in the right range, but a sound acid sites, the incorporation of aluminum intotheoretical foundation for such values is missing. silica frameworks is of special interest. The alumi-So far, the method of determination of pore-size num-containing as-made samples show tetra-distributions in cylindrical pores based on the hedrally coordinated aluminum as well as octaheresults of NLDFT calculations seems to be the drally coordinated atoms [34,35]. However, some most accurate one [25]. Maddox et al. [31] recently authors could achieve the incorporation of only suggested a new ‘‘heterogeneous interaction exclusively tetrahedrally coordinated aluminum by model’’ based on MC simulations. This method varying the aluminum source [36–38]. Janicke considers that the capillary condensation pressure et al. [39] observed that the octahedral aluminum used to calculate the pore size is sensitive to can be converted into species with solid–fluid interactions. According to the authors, tetrahedral coordination upon calcination. other methods, including the density functional Tetrahedrally coordinated aluminum is of special theory, seem to underpredict the pore size by a interest and the desirable species, because it is significant amount because the approaches assume assumed that only these are incorporated into the a homogeneous surface of the wall, which under- framework and therefore are responsible for the estimates the binding energy of the submonolayer formation of the acid sites, while the octahedral atoms adsorbed at low pressures. aluminum species are occluded in the pores or

The wall thickness can be calculated by deter- exist as an amorphous byproduct. Two-dimenmining the diVerence between the lattice parame- sional (2D) NMR experiments show clearly that ter (a=2d(100)/E3) determined by X-ray diVraction interfacial framework aluminum is incorporated and the pore size obtained by nitrogen adsorption into the MCM-41 material and that the framework analysis. However, one should bear in mind that aluminum is preserved following calcination treatthe values are only estimates, because so far no ment [40]. Moreover, by using 2D NMR, Janicke reliable means for pore-size analysis exists. et al. [40] observed that the surfactant species are Moreover, the lattice parameters are often calcu- dipole–dipole coupled to both tetrahedrally coorlated from quite broad reflections and therefore dinated aluminum atoms and adsorbed water molecules. Therefore, it is suggested that bothdo not correspond to an exact value. However, four- and six-coordinated aluminum species are showing either cylindrical [46] or hexagonal [47] pores.present in the framework of the hydrothermally synthesized MCM-41 sample.

The surface properties of the pore walls were 3. Formation mechanismstudied by adsorption of molecules on the surface and by using Fourier transform infrared (FTIR) The original MCM-41 synthesis was carried outanalysis. By adsorbing polar or unpolar molecules in water under alkaline conditions 1. Similar toon surfaces it is possible to measure the hydrophilic zeolite syntheses, organic molecules—surfac-or hydrophobic properties of surfaces. Therefore, tants—function as templates forming an orderedby carrying out the adsorption of cyclohexane organic–inorganic composite material [49]. Via[21], benzene [32] and water [41], the relatively calcination the surfactant is removed, leaving thehydrophobic character of siliceous MCM-41 has porous silicate network. However, in contrastbeen clearly demonstrated. The water adsorption to zeolites, the templates are not single organicisotherm [23] corresponds to a type V in the molecules but liquid-crystalline self-assembledIUPAC classification [20], which is an indication surfactant molecules. The formation of the inor-of the hydrophobic character in the low-pressure ganic–organic composites is based on electrostaticregion of the adsorption isotherm. Moreover, a interactions between the positively charged surfac-small amount of surface OH sites is detected and tants and the negatively charged silicate species.at least three diVerent silanol groups [42,42] can Several studies have investigated the buildingbe distinguished by using pyridine adsorption: mechanism of MCM-41 and seem to be at firstsingle, hydrogen-bonded and geminal silanol inconsistent. However, the ‘‘liquid-crystal templat-groups. A fourth possible silanol group has also ing’’ (LCT) mechanism suggested by Beck et al.been reported [43]. The presence of silanol groups

[32] early after the discovery of MCM-41 seemsis important for surface modifications like silylato include all these proposed mechanisms, tion for hydrophobization of the surface [32]. although the details were not specified in that

Even aluminum-containing MCM-41 is fairly publication (Fig. 4). They proposed two main hydrophobic as shown in a comprehensive sorption pathways, in which either the liquid-crystal phase study [4]. Because of its high sorption capacity is intact before the silicate species are added (pathfor hydrocarbons, MCM-41 might thus be used way 1), or the addition of the silicate results in for removing hydrocarbons from moist gas the ordering of the subsequent silicate-encased streams. surfactant micelles (pathway 2). The reason for

Aluminum-containing MCM-41 materials were the apparently diVerent reaction pathways results investigated for their acidic sites on the surface, from changes in surfactant properties, depending which is important for catalytic reactivity. on the surfactant concentration in water and the Therefore, adsorption studies using bases such as presence of other ions [50,51]. MCM-41 can be ammonia [21,45] and pyridine [45] were carried synthesized with surfactant concentrations as low out. Based on temperature-programmed desorp- as the critical micelle concentration (CMC) up to tion (TPD) and FTIR results, aluminum-contain- concentrations where liquid crystals are formed ing MCM-41 possesses an acidity similar to that [52]. In very dilute aqueous solutions (~10−3 to of amorphous aluminosilicate. This result is in 10−2 mol l−1 surfactant concentration) the existing good agreement with the absence of X-ray reflec- species are spherical micelles. Monnier et al. [53] tions at high angles and NMR results indicating

1 A large number ofdiVerent synthesis procedures exist.It seemsthe presence of amorphous walls in the case of that almost every single group has developed their special prepa-pure silica as well as aluminosilicate. Combining ration method. Since such a large synthesis variety leads toall results from the diVerent characterization meth- MCM-41 and a comparison between these diVerent method is ods, two structural models with amorphous walls almost impossible, we did not want to discuss the synthesis in detail. An overview was recently given in [48].have been constructed for the hexagonal MCM-41

Fig. 4. Liquid-crystal templating (LCT) mechanism proposed by Beck et al. [32] showing two possible pathways for the formation of MCM-41: (1) liquid-crystal-initiated and (2) silicate-initiated.

studied this situation in detail using concentrations tion was investigated by Chen et al. [56] using 14N-NMR spectroscopy. The randomly orderedof around 1 wt%. They proposed that three steps are involved in the formation of the surfactant– rod-like micelles interact with the silicate species to yield tubular silica arranged around the externalsilica composite. First, the oligomeric silicate polyanions act as multidentate ligands for the cationic surface of the micelles. These composite species spontaneously form the long-range order indica-surfactant head groups, leading to a strongly interacting surfactant/silica interface with lamellar tive of MCM-41. A similar mechanism was proposed on the basis of in situ electron paramagneticphase. In the second step preferential polymerization of the silicate in the region of the interface resonance (EPR) measurements with surfactant concentrations below 2 wt% [57]. The first step isoccurs, which leads to the reduction of the negative charge at the interface. The following charge- the formation of domains, which consist of micellar rods encapsulated with silicate ions showingdensity matching between the surfactant and the silicate leads to a phase transformation, forming hexagonal order. Second, the silicate ions polymerize at the interface, resulting in hardening of thethe hexagonal surfactant–silicate composite. In the following, it was recognized that the layered inter- inorganic phase. At high concentrations surfactants form lyo-mediate is not always necessary, but that the charge-density matching could directly lead to a tropic mesophases [50,51]. Depending on the nature of the surfactant, the concentration and thehexagonal or a cubic phase. In particular, the work of Chmelka, Stucky and co-workers [54] with temperature, these mesophases show diVerent phases with, e.g., hexagonal, cubic or lamellarcubic octamers of silica and thus decoupled selfassembly and silica polymerization showed that structure. Attard et al. [58] used this approach to synthesize MCM-41 as monoliths. However, inunder conditions when no condensation occurs (low temperature) the surfactant self-assembly contrast to the original MCM-41 synthesis, nonionic instead of cationic surfactants were used.governs the system, but as soon as the silica polymerizes, the resulting structure is controlled Therefore, in this case not ionic interactions but hydrogen bonding and hydrophilic/hydrophobicby the inorganic framework. Also, an initially formed layered structure observed by Steel et al. interactions are responsible for the formation of the inorganic–organic composite. However, the[5] was proposed as the ‘‘puckering layer model’’. The silicate species in aqueous solution form a use of non-ionic surfactants and their perspectives in mesophase preparation will be discussed in morelayered structure, further ordering resulting in puckering of the silicate layers and the formation detail in the following section. Almost at the same time as the synthesis ofof the hexagonal channels.

Increasing the surfactant concentration results MCM-41 was published, Yanagisawa et al. [59] described an alternative synthesis pathway forin the self-assembly of surfactant rods. This situa- preparing mesoporous silicate from a layered sili- syntheses based on ionic interactions, the liquidcrystal approach has been further extended tocate, kanemite, which consists of single layers of

SiO4 tetrahedra. This material is designated as include two additional pathways showing organic– inorganic interaction other than ionic. Under neut-FSM-16 (Folded Sheet Mesoporous Materials)2.

The preparation is similar to MCM-41 using a ral conditions mesostructures are formed by using neutral (S0) [6] or non-ionic surfactants (N0)layered silicate as silica source. Instead of a liquidcrystal templating mechanism a ‘‘folded sheet’’ [67]. In this approach (S0/N0I0) hydrogen bonding is considered to be the driving force for the forma-mechanism [60] is proposed: the layered organic– inorganic composites are formed by intercalation tion of the mesophase. In the so-called ligandassisted liquid-crystal templating mechanism,of the layered silicate using surfactants [61]. The transformation to the hexagonal phase occurs covalent bonds are formed between the inorganic precursor species and the organic surfactant mole-during the hydrothermal treatment by condensation of the silanol groups. However, some authors cules followed by self-assembling of the surfactant [68]. Interestingly, Zhao et al. [69] showed thatproposed dissolution of the kanemite under the reaction conditions [62]. The dissolution rate of the liquid-crystal templating approach is not limited to amphiphilic molecules. By use of non-kanemite is dependent on the quality of the starting material, and such deviations might explain vary- amphiphilic mesogens, which show the ability to form liquid-crystal structures, lamellar meso-ing results from diVerent groups. In this case the kanemite would act as a silica source comparable structured inorganic–organic composites were synthesized. Following the principle of charge match-to the MCM-41 synthesis. MCM-41 and FSM-16 are similar but show slightly diVerent properties ing, the mechanism is based on the (S−I+) pathway since the template molecules were nega-in adsorption [62] and surface chemistry [63]. tively charged.

Since the liquid-crystal structures of the surfactant serve as organic template, rather than single4. Alternative synthesis pathways and new materials molecules commonly proposed as templates in zeolite chemistry3, the behavior of the surfactant in binary surfactant/water systems is the key forThe liquid-crystal-controlled synthesis of

MCM-41 opened a wide variety of synthesis the controlled preparation of silica mesostructures [54,70]. According to a microscopic model intro-approaches for developing new materials. The successful preparation of new porous materials duced by Israelachvili et al. [71], the relative stabilities of diVerent aggregate shapes and thecould be achieved by developing new synthesis pathways and by taking advantage of the liquid- corresponding mesophase structures can be predicted. The preferred shape of self-assembled sur-crystal chemistry provided by the surfactant.

So far, we have described the organization of factant molecules above the CMC depends on the eVective mean molecular parameters that establishcationic quaternary ammonium surfactants and anionic silicate species (S+I−) to produce three- the value of a dimensionless packing parameter g, which is defined as: g¬V/a0lc, where V is thedimensional periodic biphase arrays. However, cooperative interactions between inorganic and eVective volume of the hydrophobic chain, a0 is the mean aggregate surface area per hydrophilicorganic species based on charge interaction can also be achieved by using reverse charge matching head group and lc is the critical hydrophobic chain (S−I+) [64] or by mediated combinations of cat- ionic or anionic surfactants and corresponding 3 Although the label ‘‘template’’ is commonly used in zeoliteinorganic species (S+X−I+,X −=halides; or chemistry, most ‘‘templates’’ do not act in the sense of a mold,S−M+I−,M +=alkali metal ion) [65]. Besides the like the liquid-crystalline surfactants, but rather direct the syn- thesis to a particular structure by mostly unknown mechanisms. 2 The number depends on the length of the surfactant alkyl It would thus be more appropriate to use the term ‘‘structure director’’.chain, in this case a C16 surfactant was used.

length [71]. The parameter g depends on the the preparation of mesoporous oxides other than silica. The following sections discuss how suchmolecular geometry of the surfactant molecules, such as the number of carbon atoms in the hydro- materials can be prepared in line with the concepts described above.phobic chain, the degree of chain saturation and the size and charge of the polar head group. In addition, the eVects of solution conditions, includ- 4.1. Silica polymorphs from ionic synthesis pathwaysing ionic strength, pH, co-surfactant concentration and temperature, are included implicitly in V, a0 and lc [72]. In classical micelle chemistry, meso- Similar to the original MCM-41 synthesis by using cationic surfactants and anionic silicatephase transition occurs when the g value is increased above critical values (Table 1). species (S+I−), MCM-48, the cubic member of the M41S group of mesoporous materials, wasMoreover, the phase transitions also reflect a decrease in surface curvature from the cubic first synthesized by scientists of the Mobil Oil Corporation [32]. MCM-48 can be synthesized(Pm3n) over the hexagonal to the lamellar phase. Spherical aggregates are preferentially formed by either by adjusting the silica/CTAB ratio and varying synthesis conditions [75,76], or by using geminisurfactants possessing large polar head groups. On the other hand, if the head groups are small and surfactants [7]. It has been suggested [53] and later confirmed [7] that the structure contains apacked tightly, the aggregation number increases, resulting in rod-like or lamellar aggregates. By three-dimensional channel network with channels running along [1] and [100] directions.including the inorganic component, Stucky et al. [73,74] expanded the model to the ternary NaOH/ Calculations from X-ray diVraction and TEM show that the structure of the so-called bicontinu-cetyltrimethylammoniumbromide (CTAB)/tetraethoxysilane (TEOS) system and created a synthe- ous phase has the space group Ia3d, which has also been found in the binary water/CTAB systemsis-space diagram of mesophase structures. By using XRD, NMR and polarized microscopy, the [78]. Moreover, Alfredsson et al. [79] studied MCM-48 samples by both scanning electronbinary and ternary systems were recently investigated in great detail by Firouzi et al. [54]. microscopy (SEM) and TEM, and realized that the particles have a crystal or ‘‘cubosome’’ shapeAdditionally, Vartuli et al. [75] studied the eVect of surfactant/silica ratios on the formation of associated with them. Although all the particles are not perfect polyhedra, they noticed that amesostructures. Obviously, the ratio is a critical variable in the mesophase formation and by vary- prevailing shape is the truncated octahedron. Furthermore, the M41S family has been extendeding the surfactant/silica molar ratio from 0.5 to 2.0 products with hexagonal, cubic (Ia3d), lamellar by the lamellar mesostructure MCM-50 [80]. The lamellar phase can be represented by sheets orand uncondensed cubic octamer composite structures were obtained. bilayers of surfactant molecules with the hydrophilic heads pointed towards the silicate at the inter-The use of these new synthetic concepts has focused on two main areas: (1) the synthesis of face. Removal of the surfactant from between the silicate sheets results in structure collapse and losssilicates with new structural properties; and (2) of porosity. However, by means of a post-treat- ment with TEOS, it is possible to produce ther-Table 1 Surfactant packing parameter g, expected structure and exam- mally stable lamellar materials. So far, the ples for such structures structure has not been solved successfully, but two structural models are proposed: first, the structureg Expected structure Example is like the classical pillared layer silicate, or second,

1/3 cubic (Pm3n) SBA-1 the structure may be composed of a variation in 1/2 hexagonal (p6) MCM-41, FSM-16, SBA-3 the stacking of surfactant rods. Huo et al. [7,81]

1/2–2/3 cubic (Ia3d) MCM-48 made great progress in preparinghigh-quality meso- 1 lamellar MCM-50 structures by using surfactants other than CTAB in the S+I− system. Sophisticated modification of 4.2. Metal oxide mesophases from ionic synthesis approachthe surfactant by varying the chain or the head group made it possible to prove the qualitative Monnier et al. [53] suggested the possibility ofmodel of Israelachvili [71] for silica/liquid crystal substituting the silicate by metal oxides that aresystems. Moreover, gemini surfactants, with two able to form polyoxoanions, because the charge-quaternary ammonium head groups separated by matching model is not specific for silica but appli-a methylene chain of variable length and with each cable to all other systems in which condensablehead group attached to a hydrophobic tail, can be polyions are possible. Subsequently, mesostruct-used to prepare a mesophase that has three-dimen- ured surfactant composites of tungsten oxide,sional hexagonal (P63/mmc) symmetry, showing molybdenum oxide and antimony oxide have beenregular supercages that can be dimensionally tai- synthesized [64,65]. In the case of tungsten oxidelored and a large inner surface. This mesostructure and antimony oxide hexagonal structures could beis called SBA-2 (Santa Barbara No. 2) [81].

obtained, whereas molybdenum oxide only formedInterestingly, the P63/mmc mesophase is so far the lamellar structures. Furthermore, the approachonly one that has no lyotropic surfactant/water could be extended to the charge reversed systemmesophase analog. A new, very fascinating phase (S−I+) by using polyoxocations and to the medi-could be developed by McGrath et al. [82] which ated combinations of S+X−I+ and S−M+I−.is based on the surfactant L3 phase by using However, besides the two hexagonal phases andcetylpyridinium chloride as surfactant. The silicate an impure hexagonal phase in the case of leadL3 phase is similar to aerogels or xerogels, but oxide, only lamellar phases could be obtained atpossesses two distinct, continuous, interpenetrating

Since vanadium oxides are catalytically very inter-because the channels contain water, which can be esting, several groups investigated vanadium-based removed at low temperatures. Moreover, in conmesostructures using the S+I− approach. Lamellar trast to other mesoporous materials described so [85] as well as hexagonal [86,87] mesostructured far, the pore sizes are determined by the relative phases could be successfully prepared. Moreover, concentration of the surfactant rather than the Janauer et al. [8] reported the crystal structure surfactant chain length. of the first lamellar vanadium oxide phase contain-

In a diVerent approach, using the mediated ing discrete vanadium oxide clusters and not a synthesis pathway S+X−I+, the cooperative continuous transition metal oxides lattice. In analassembly of cationic silicate species with cationic ogy to the extension of concepts from zeolite surfactant mediated by chloride ions in strongly chemistry to mesoporous alumosilicates, another acidic solutions leads to hexagonal MCM-41 meso- main interest is focused on the preparation of structure [65]. However, since the material pre- mesostructured aluminophosphates. Using catpared from the acid synthesis is not identical with ionic surfactants, lamellar [89], hexagonal [90,91] MCM-41 synthesized in alkaline medium, it is and even the cubic [91] mesostructured aluminooften labeled SBA-3 [7,83] or APM [84] (acid- phosphate phase and, in the case of silicoaluminoprepared mesostructure). Additionally, a large- phosphates (SAPO), the hexagonal phase could cage (~30 A ) cubic silica structure with the space be synthesized [92]. Also, Rao et al. [93] recently group Pm3n has been obtained using the acid succeeded in the preparation of mesostructured synthesis approach. The globular surfactant lamellar, hexagonal and cubic aluminoborates by micelle has the largest surface curvature of all using the (S−I+) approach. However, a main lyotropic liquid crystals, and the micelle surface problem of all the mesostructures mentioned so has the lowest charge density [65,83]. So far, the far is thermal stability and, therefore, the removal so-called SBA-1 mesophase can only be obtained of the surfactant. None of these structures could be obtained as MCM-41 or MCM-48 analogousin the acid synthesis approach.

mesoporous materials. The thermal instability is probably due to the diVerent oxo chemistry of the metals in comparison with silicon. It was assumed that the existence of several relatively stable oxidation states of the metal centers results in oxidation and/or reduction during calcination [64]. In addition, incomplete condensation of the species inside the pore walls could also be responsible for the thermal instability as was assumed in the case of antimony oxide [65]. Furthermore, Stein et al. [94] showed in the case of mesostructured tungsten oxide that the walls consist of uncondensed Keggin ions and a complete condensation does not seem possible. The first porous transition metal oxide

reported by Antonelli and Ying [95] was TiO2 showing a hexagonal structure. The mesostructure was prepared by using an anionic surfactant with phosphate head groups and titaniumalkoxy precursors, which were stabilized with bidentate ligands such as acetylacetone. After calcination at

350°C, porous TiO2 with surface areas of around 200 m2 g−1 was obtained. The formation mecha- nism has not been well studied yet. Froba et al. [96] suggested in a recent study that the mechanism is analogous to the mediated S−M+I− approach since potassium ions are proposed to be the mediating ions. However, the existence of anionic tita- nium precursors under the reaction conditions of Fig. 5. Transmission electron micrograph of calcined zirconium oxo phosphate showing the hexagonal pore arrangement ana-pH 5–6 is not likely according to titanium sol–gel logous to MCM-41 materials. We should like to thank Drchemistry [97]. Parallel to the work of Antonelli

Ushida and Prof. Dr Schlogl from the Fritz Haber-Institut,and Ying, we investigated the preparation of meso- Berlin for recording this image.

structured zirconias in our group. By using zirco- nium sulfate as precursor, which is able to interact with cationic surfactant molecules, mesostructured lization is well known for sulfated [100] and phosphated [101] zirconias due to the anions. Thesurfactant composites with hexagonal phase could be prepared [98]. By a post-synthesis treatment new zirconia-based materials show high surface areas and pore sizes in the range between micro-with phosphoric acid or by reducing the amount of sulfate in the composite, two porous materials and mesopores. A similar approach, stabilization using the phosphoric acid treatment, has also beenwith diVerent compositions, thermally stable up to 500°C, were obtained. The new materials are called successfully used in the case of hafnium oxide and leads to microporous hafnium oxide, which iszirconium oxide sulfate and zirconium oxo phosphate, and show MCM-41 analogous structures as structurally analogous to MCM-41 [102]. Additionally, Kim et al. [103] treated zirconiumcan be clearly seen in the TEM images (Fig. 5). The high thermal stability is due to a crystallization sulfate surfactant mesostructures with chromate solutions to obtain a binary transition metaldelay caused by the sulfate or phosphate groups in the zirconia structures so that the disordered (Zr–Cr) oxide framework. After calcination, the porous product exhibits surface areas of up towall structure, which is favorable for these mesoporous materials, is retained [9]. Delay of crystal- 374 m2 g−1. Moreover, Stein and co-workers [104] used the post-synthesis treatment with phosphate work wall thickness (1.7–3.0 nm) but less long- range hexagonal order [1,112]. Extendingions to link Al13 clusters of mesostructured Al13 cluster–surfactant salts. During the linking reac- the neutral route to a biomimetic templating approach by using neutral ‘‘bola-amphiphiles’’tion the clusters break up to form a network of

AlOh–O–PTd linkages, as well as a fraction of H2N(CH2)nNH2, lamellar silicas were prepared that have vesicular particle morphology, denotedtetrahedral aluminum sites. The procedure can also be used for the preparation of galloalumi- MSU-V (Michigan State University) [113,114]. The MSU-V materials are thermally stablenophosphates.

By using cationic surfactants, but also cationic and, after calcination, porous lamellar silicas with high surface areas were obtained. Moreover,octamers as zirconium precursors, Hudson and Knowles [105,106] succeeded in the formation of Pinnavaia’s group was the first to use non-ionic surfactants, poly(ethylene oxide) monoethers ashigh-surface-area zirconia. Although first a scaVolding mechanism was assumed, it seems to well as Pluronic- and Tergitol-type surfactants in neutral aqueous media, which are able to formbe more likely that the high surface areas are due rather to the formation of very small particles than liquid crystals [64]. The mesoporous products, denoted MSU-X, show worm-like disordered meso-to a porous structure. Moreover, two other approaches using anionic [107] and amphoteric pores with pore sizes of 2.0–5.8 nm [115]. The synthesis could also be successfully used in the[108] surfactants lead to mesostructured hexagonal zirconia surfactant composites. However, so far it case of alumina, which shows the worm-like structure as well [116]. Attard et al. [58] improved thisdoes not seem to be clear whether these structures are also stable after calcination. synthesis approach by using non-ionic surfactant concentrations at which liquid crystals are formed.Furthermore, the surfactant-controlled synthesis approach is not limited to oxide-based materials This technique is termed ‘‘true liquid-crystal templating’’ (TLCT). Highly ordered hexagonal, cubicand recently novel mesostructures based on tin(IV) sulfide [109] and a lamellar mesostructure and lamellar mesoporous silica showing sharp pore-size distributions could be obtained. Withof thiogermanates [110] have been synthesized. However, these structures could only be obtained this approach they even recently succeeded in the synthesis of pure mesoporous metals of palladiumas surfactant composites and calcination leads to structural collapse of the mesophase. [117] and platinum [118,119] and mesoporous platinum films [120] by chemical reduction of metal salts dissolved in the aqueous domains of a4.3. N0/S0I0: silica polymorphs and metal oxide hexagonal lyotropic liquid-crystalline phase. Using a liquid-crystalline phase as template formed fromThe advantage of using neutral or non-ionic surfactants over the ionic route is the possibility non-ionic organic amphiphiles, Braun et al. [121] synthesized stable semiconductor–organic super-of recovery of the surfactant by extraction, which can be attributed to the relatively weak assembly lattices based on cadmium sulfide and cadmium selenide. They described the material as a compos-forces due to hydrogen bonding. Tanev and Pinnavaia [6] developed the neutral templating ite solid in which organic structures are molecularly dispersed in the inorganic lattice. Great progressroute for mesoporous materials which is based on hydrogen bonding and self-assembly between neut- in the preparation of mesoporous silicas was very recently made by Zhao et al. [122], who usedral amine surfactants and neutral inorganic precursors. The materials prepared by using the S0I0 triblock polyoxyalkylene copolymers for the synthesis of large-pore materials, so-called SBA-15.approach, so-called HMS (hexagonal mesoporous silica) materials, exhibit single d100 reflections However, the mechanism is proposed to be (S+X−I+) since the block copolymer is positivelyaccompanied by more or less pronounced diVuse scattering centered at ~1.8 nm. Compared with charged under the reaction conditions. SBA-15 can be prepared with pore sizes between 4.6 andthe electrostatically templated MCM-41 silicas, HMS materials show a consistently larger frame- 30.0 nm and wall thicknesses of 3.1 to 6.4 nm.

MCM-41 materials prepared by the cationic sur- combining niobium (or tantalum) alkoxides with one equivalent of long-chain amines, niobium (orfactant CTAB commonly have pore sizes of around 3.0 to 4.0 nm. In the conventional synthesis cosol- tantalum) alkoxide–amine complexes are formed and, by adding water, the condensation of niobiumvents such as trimethylbenzene (TMB) [32] and also some special synthesis variations [123,124] (tantalum) centers is forced. The hexagonal phase is formed by self-assembly of the metal alkoxide–have been used to expand the pore sizes up to around 10.0 nm. Even bigger pores of up to 50 nm amine complexes. After removal of the organic by extraction, the porous oxides show surface areaswith wall thicknesses of up to 10 nm, although not as regularly arranged as the ones prepared with of up to 500 m2 g−1. the triblock copolymers, were synthesized by

Goltner, Antonietti and co-workers [125,126] who used cationic as well as anionic block copolymers 5. Morphology control and alkoxysilanes as silica precursors. The silicas resulting after calcination are supposed to be exact The mesoporous silica-based materials obtained from the original MCM-41 synthesis by Beck et al.replicas of the mesophases. Inspired by the surfactant-controlled syntheses, even mondisperse [2] consist of aggregates and loose agglomerates of small particles. However, for many applicationsmacroporous materials could be prepared recently by Imhof and Pine [127] using an emulsion of in catalysis, chemical sensoring or as optical devices, defined morphologies are required. Theequally sized droplets as templates around which the inorganic material deposits through a sol–gel research has focused particularly on four developments: thin films, fibers, spheres and monoliths.process. Silica, titania and zirconia materials have been prepared. The organic can be removed while Yang et al. [131] showed the potential of acid synthesis (S+X−I+) in the preparation of a largepreserving the macroporous structure, although phase transformations into the crystalline oxide variety of diVerently curved, highly interesting morphologies. However, no details of how tophases have been observed. obtain desired morphologies or why these morphologies were obtained were given in that publica-4.4. S–I approach tion. Nevertheless, the acidic route pioneered by

Huo et al. [65] seems to be best suited for morphol-There are only a few studies concerning the preparation of covalently linked hybrid organic ogy control. Therefore, for most morphologically controlled syntheses, the (S+X−I+) approach wasnetworks, which, however, could be an interesting approach to a wide range of functionalized hybrid successfully used. Thin films have been prepared at the air/water [132,133] and oil/water [84] inter-materials with ordered mesopores. Huo et al. [7] and Burkett et al. [128] successfully prepared face as free-standing films, on both mica [134] and graphite [135,136] surfaces and by using dip- orhexagonal mesostructured silicas by this method. This approach has been more important in the spin-coating [137,138] techniques. Except for the spin- or dip-coated films and the film from thepreparation of mesoporous transition metal oxides. Ying’s group used it for the preparation of meso- oil/water interface, the grown films all show continuous hexagonal, mesoporous silica with poresporous hexagonal niobia [68] and tantalum oxide [129], which are called Nb- and Ta-TMS1 (trans- parallel to the film surface. Tolbert et al. [139] recently reported the synthesis of silicate surfactantition metal oxide mesoporous molecular sieve

(Parte 1 de 3)