The main methods for the design and preparation of single-site heterogeneous catalysts on inorganic oxide supports are described and reviewed. Catalytically active metal sites can be either introduced into the framework of porous materials via direct synthesis or added to a pre-existing support by post-synthesis techniques. Particular attention is paid to selected examples where the geometry, the nature and the chemical surroundings of the active single site is a key factor to obtain catalytic systems with enhanced chemo-, regio- and stereoselectivity. The ever-increasing capabilities of ‘nanoarchitecture’ at molecular level enable chemists to build ideal catalysts for the sustainable transformation of bulky and high added-value molecules.
In the last decade of the twentieth century and at the beginning of the twenty-first century decade, a large scientific effort was focused on the design, synthesis and characterization of a wide variety of nanoporous inorganic oxides with very large channels for catalytic purposes (Thomas 1999; Davis 2001; Maschmeyer & van de Water 2006; Climent et al. 2011; Xuereb & Raja 2011). Microporous zeolites and zeotypes are optimal catalysts for the transformation of relatively small molecules and have been largely applied for industrial catalytic processes. However, microporous solids are unsuitable when bulkier and more functionalized reactants are involved in catalytic processes. For these purposes, mesoporous molecular sieves (porous materials whose pore diameters fall in the range 2–50 nm) could represent an ideal compromise to retain the advantages of a porous structure, while minimizing mass-transfer limitations of large molecules, such as those typically found in the fine chemicals industry (Guisnet & Guidotti 2006). The development of nanostructured mesoporous inorganic oxides paved the way to the use of oxidic supports with very high specific surface areas as supports for isolated metal centres (Thomas 1994), and led to the expansion of single-site heterogeneous catalysis (Copéret et al. 2003; Thomas et al. 2005, 2009).
In a single-site heterogeneous catalyst (SSHC), active sites are well defined and evenly distributed entities (single sites). They have distinct chemical surroundings, as in conventional homogeneous systems, but they also show all the advantages of heterogeneous systems, in terms of easy separation, recover and recyclability. Single sites are usually (although not necessarily) located over solid supports with high surface area, and they show the following general peculiarities: (i) they consist of a limited and defined number of atomic species, i.e. one atom (as in truly ‘single’ sites) or few atoms (as in the case of chemically defined metal clusters), (ii) they are spatially isolated from each other, (iii) they all have identical energy of interaction between the site itself and a reactant, and (iv) they are structurally well characterized.
In the design of a new SSHC, by inserting (or depositing) metal single sites into (or onto) inorganic oxides and by modifying the first-neighbour chemical surroundings of the active centres, it is possible, in principle, to obtain the inorganic equivalent of an enzyme and tune the properties of the catalyst according to its applications (Lillerud et al. 2010; Xuereb et al. 2010). The robust inorganic framework can indeed mimic the polypeptidic structure in the enzyme protecting the catalytically active site and limiting the transfer of reactants and products. Any modification of the internal surface of the pores, in terms of acid/base, hydrophilic/hydrophobic or polar/apolar properties, recalls the role of ancillary aminoacids in enhancing the activity and/or the selectivity of the catalytic transformation. Then, the network of wide mesopores can accommodate the bulky and complex intermediates found in the production of fine chemicals, following bioinspired synthetic pathways.
The development of SSHCs, however, is still in its expansion phase, and several aspects related to the preparation and the catalytic application of these systems deserve deeper investigation and exploitation. The ever-increasing capabilities of ‘nanoarchitecture’ at a molecular level will provide chemists, in the next decades, with new tools to design and build ideal catalysts for desired reactions rather than adapt existing catalysts according to a conventional trial-and-error approach. In this scenario, the most relevant synthesis techniques and some case studies of SSHCs are described in the following sections.
2. Preparation of single-site heterogeneous catalysts on inorganic supports
Redox-active zeolitic structures, such as titanosilicalite-1 (TS-1), or transition-metal-containing open-structure aluminophosphates (AlPOs) appeared in the 1980s and can be considered as the first examples of SSHC, as they fulfil the main requirements of isolation, uniform distribution and controlled chemical environment of the active sites (Wilson et al. 1982; Weisz et al. 1984; Chen et al. 1994; Marchese et al. 1994; Thomas et al. 1994; Corma 1997). During the design and engineering steps that led to these catalytic systems, a major role was played by in situ characterization methods for solid catalysts (Thomas 1997). Then, starting from the 1990s, owing to the development of silica-based mesoporous molecular sieves, larger reactant molecules, which can enter the pore network, react there and leave it, can be transformed within the inner space of porous solids (Xiao 2005; Yang et al. 2009). Mesoporous silica materials, such as MCM-41, MCM-48, HMS, KIT-6, SBA-15 or SBA-16, are, indeed, ideal supports for SSHCs, since they exhibit high surface area (600–1300 m2 g−1) and their inner and outer surfaces have a profusion of pendant silanol groups (generally from 1 to 3 OH groups per square nanometre) that are optimal loci for immobilizing transition metal catalytic centres (Rigutto et al. 2007). Among them, materials with a good hydrothermal stability, such as HMS or SBA-15, owing to a large pore wall thickness and highly condensed frameworks with limited connectivity defects, are preferable, as they can more easily withstand the strong conditions occasionally experienced during synthesis or catalysis (Zhang et al. 2005).
Catalytically active single sites can have a structural role, when they are introduced into the framework of spatially ordered materials, via direct synthesis (figure 1). Otherwise, single sites can be added to a pre-existing support by post-synthesis techniques.
Following a direct-synthesis approach, SSHCs with a microporous and/or mesoporous inorganic structure can be prepared. In this type of preparation, precursors of the metal active sites (e.g. salts and alkoxides) are added to the synthesis gel of the inorganic support and part of the framework atoms are isomorphously substituted by metal species. Beside zeolite-based catalysts, one-pot direct synthesis was largely used to develop microporous AlPOs, silicoaluminophosphates (SAPOs) and framework-substituted variants, in which different metals (such as Co, Mg, Zn, Ti, etc.) are isomorphously substituted to framework Al, Si or P atoms. At the end, solid catalysts with an ordered array of pores and isolated acid and/or redox-active functionalities are obtained.
Different methods used to produce SSHCs via post-synthesis modification are schematized in figure 2. In such approaches, the isolated sites can be deposited, heterogenized, tethered or linked to a pristine matrix.
Then, with regard to the methods of creation of single-site centres, two main approaches can be considered: post-synthesis covalent deposition and post-synthesis non-covalent deposition.
In the first case, the active centre is added to the support (which can also be commercially available) as a precursor that can be deposited irreversibly (more precisely, anchored or grafted) onto the surface as it is, by the formation of covalent bonds, or after a functionalization with a side chain (a tether; Kolasinski 2007).
In the second case, the homogeneous precursors of the single-site active centres are immobilized onto the surface of solid supports by non-covalent interactions, such as hydrogen bonds, weak van der Waals interactions, which allow their confinement (trapping or encapsulation). In particular, encapsulation covers a large selection of methods for immobilizing catalytically active species within the pores of a solid, and it allows one to keep the optimal performances of the original homogeneous catalysts unaffected. In some cases, owing to a positive cooperative effect, it is also possible to have a final system with improved characteristics with respect to the parent precursor.
3. Selected examples and applications of single-site heterogeneous catalysts
(a) Grafted Ti single-site transition metals supported onto high surface area silica as epoxidation catalysts
Titanium-containing mesoporous silica has been largely investigated in the literature and synthesized following different synthetic approaches. Among the different preparation methodologies proposed in the literature, the grafting reaction has largely been exploited to introduce the metal centres in the siliceous framework, obtaining efficient SSHCs. For example, Ti centres were successfully grafted onto the inner walls of MCM-41 mesoporous silica by using an organometallic precursor, titanocene dichloride [Ti(Cp)2Cl2] (Cp=C5H5) (Maschmeyer et al. 1995), and tested as catalysts in the olefin epoxidation. Such materials proved to be active in the epoxidation of model substrates (i.e. cyclohexene) to the related epoxides.
Mesoporous titanosilicate molecular sieves can also be suitable catalysts for the selective epoxidation of bulky and richly functionalized alkenes, such as cyclic terpenes of interest for the flavour and fragrance industry, which often cannot be easily carried out for microporous titanosilicate molecular sieves (Sheldon et al. 1998; Ratnasamy et al. 2004).
Data in the literature reported that mesoporous titanosilicate samples can be suitable catalysts for the selective epoxidation of this class of compounds, and the resulting epoxides are useful intermediates for the synthesis of a broad series of valuable and high added-value products (van der Waal et al. 1998; Ravasio et al. 2004). For instance, a series of cyclic unsaturated alcoholic terpenes, namely α-terpineol (1), terpinen-4-ol (2), carvotanacetol (3), isopulegol (4), carveol (5) and limonene (6) (scheme 1), was selectively epoxidized with tert-butylhydroperoxide (TBHP) over Ti-containing mesoporous silica catalysts with both an ordered and non-ordered porous array. High conversions (up to 90%) and good selectivity to the desired epoxides (60–80%) were achieved in the liquid phase, in polar aprotic solvents (acetonitrile and ethyl acetate) under reflux conditions.
The catalyst activity is mainly governed by the hydroxyl-function position, and the closer the hydroxyl group to the unsaturation, the higher the conversion rate of the terpene. In addition, a direct comparison between the turn-over frequencies obtained over grafted Ti–MCM-41 (where Ti is deposited by post-synthesis treatments) and in-framework Ti–MCM-41 (where Ti is inserted in the catalyst's matrix during synthesis) shows that the epoxidation rate over the former is up to 10 times higher than over the latter, owing to the better exposition and availability of Ti centres in the grafted solid (Sankar et al. 1994; Oldroyd et al. 1996; Berlini et al. 2000; Thomas & Raja 2004). Stereoselectivity can be high on such reactants, and a diastereospecific epoxidation (100% diastereoisomeric excess (d.e.)) was obtained on homoallylic and bishomoallylic terpenes, 1 and 2, respectively, owing to the role of the OH function in directing the oxygen transfer from the oxidant to one specific side of the six-membered ring. In fact, when the OH group in 1 is replaced by an acetyl moiety (α-terpinylacetate), the d.e. value decreases abruptly from 100 per cent down to 66 per cent (Guidotti et al. 2002).
However, in the epoxidation of these terpenes, the use of materials with an ordered array of mesopores is not strictly necessary in order to have an efficient heterogeneous epoxidation catalyst. Provided that the pore diameter of the solid is large enough to accommodate the reactant molecules and that the titanium active sites are evenly dispersed and available on the silica support surface, the morphology of the surroundings of the titanium atoms does not considerably affect the catalytic performance. Indeed, when the pore diameters are too large and no proper ‘sieving’ effect occurs, the siliceous materials act as simple supports for the active sites, and an ordered array of pores does not seem essential (Berlini et al. 2001). The performances obtained over Ti-containing mesoporous solids are often fully comparable to those obtained over Ti-containing amorphous non-ordered silicas, in terms of conversion, selectivity and activity (Fraile et al. 2001a; Guidotti et al. 2003a).
In general, the hydrophilic nature of MCM-41-based solids can be a drawback if relatively non-polar olefins have to be epoxidized: if a more hydrophobic catalyst is needed, the silica surface silanols can be silylated. Catalysts whose silanols were silylated with hexamethyldisilazane or triethoxyfluorosilane show a higher activity in the epoxidation of simple alkenes, such as cyclohexene, cyclooctene or linear alkenes, compared with unsilylated ones (Corma et al. 1998; Peña et al. 2001; Kamegawa et al. 2011). Nevertheless, in the case of substrates with intermediate polarity and with functional groups that are able to interact with the catalyst surface, a hydrophilic environment around Ti(IV) sites can lead to good results. Thus, for instance, silylated Ti–MCM-41 catalysts were more active than non-silylated catalysts in limonene epoxidation. On the contrary, in the epoxidation of α-terpineol, the surface silanols played a relevant role in enhancing the chemoselective formation of the epoxide (Guidotti et al. 2008a). Analogously, in the full epoxidation of methyl linoleate, the positive interaction of the already formed epoxide rings with the hydrophilic catalyst surface favoured the transformation of methyl epoxyoleate (the monoepoxide) into methyl diepoxystearate (the diepoxide). A direct correlation between the polar character (expressed as surface density of silanols per gram of catalyst) and the selectivity to diepoxides after 24 h was indeed found in the epoxidation of soy-bean oil fatty acid methyl ester (FAME) mixtures over titanosilicate catalysts with various textural properties (Guidotti et al. 2006).
Each pair of reactant/catalyst, however, should be studied case by case and, in some situations, a partial silylation of the solid can also be envisaged, in order to tune accurately the hydrophilic/hydrophobic character of the catalyst, in order to maximize the yield of the desired product (Guidotti et al. 2011a).
The grafting of Ti(IV) species over a mesoporous silica can give rise to a bifunctional acid and redox-active oxidation catalyst (the Tiδ+ site with free coordination sites) (Bhaumik & Tatsumi 1999; Bisio et al. 2011). Isopulegol epoxide (a compound with fungicidal and insect-repellent activity) was synthesized over Ti-grafted MCM-41 under one-pot two-step conditions (Guidotti et al. 2000; Ravasio et al. 2004). Citronellal was first converted into isopulegol, by acid-catalysed cyclization (ene reaction), which was then epoxidized to isopulegol epoxide, without isolation of the intermediates (scheme 2).
In order to avoid the direct epoxidation of the starting unsaturated aldehyde (to citronellal epoxide), the oxidant TBHP was added only after the complete cyclization of citronellal into isopulegol. With this approach, a global yield of 68 per cent in isopulegol epoxide was obtained.
Titanocene-grafted silicas showed promising results too in the epoxidation with TBHP of a wide variety of unsaturated FAMEs: methyl oleate, methyl elaidate, methyl linoleate and mixtures of methyl esters obtained from high-oleic sunflower, castor, coriander and soy-bean oils (Guidotti et al. 2003b, 2006). In particular, in the comparison between methyl oleate and methyl elaidate, with cisand trans configurations, respectively, it was found that the structural features of an ordered solid, such as Ti–MCM-41, are necessary to attain very high yields into trans-epoxystearate. So, with these substrates, by choosing titanosilicate mesoporous materials with an ordered hexagonal array of pores, such as Ti–MCM-41, it was possible to obtain very high yields in epoxidized FAMEs (up to 95% yield of epoxidized methyl linoleate or 96% yield in epoxidised castor oil methyl ester in 24 h) with a relatively small excess of TBHP oxidant (only 10% more than the stoichiometric amount needed) (Guidotti et al. 2007, 2008b). As a main advantage, by this method, unsaturated FAMEs can be epoxidized with no use of peracids (that are instead used industrially in the Prilezhaev epoxidation process), hence under completely acid-free reaction conditions.
Finally, the use of aqueous hydrogen peroxide is typically difficult over most mesoporous titanosilicates (Clerici 2009). Actually, specific experiments on the interaction of TBHP and H2O2 with Ti(IV) centres have clarified that a 1-aliquot addition of aqueous H2O2 leads to a rapid and irreversible transformation of Ti(IV) centres into TiO2-like domains, which are then catalytically inactive in epoxidation (Gianotti et al. 2007). Nevertheless, Mayoral and co-workers reported the synthesis of Ti(IV)–silica catalysts by grafting Ti(OiPr)4 onto a commercial non-ordered mesoporous silica. This catalyst showed promising results in the epoxidation of non-functionalized alkenes, dienes and allylic alcohols, not only using TBHP as oxidant (Cativiela et al. 1996), but also with dilute aqueous hydrogen peroxide under mild conditions (23% epoxide yield after 1 h), owing to a slow drop-wise addition of oxidant (Fraile et al. 2000). In the latter case, the contribution of direct epoxidation to the productive H2O2 consumption increases considerably with respect to one-shot addition (from 21% to 56%), improving the selectivity to epoxide. The role of the slow addition of H2O2 is crucial to avoid, or at least minimize, the useless decomposition of the oxidant and to keep the local water concentration as low as possible. The side production of allylic oxidation side-products (cyclohexenol and cyclohexenone) is thus negligible and the selectivity to the desired epoxide is very high (Fraile et al. 2001b, 2003). In addition, specific immersion calorimetry studies to evaluate the interaction between the reactant and the catalyst evidenced a very good affinity of cyclohexene towards titanosilicate, which is a crucial factor in the epoxidation reaction (Vernimmen et al. 2011).
The protocol of the controlled drop-wise addition of aqueous H2O2 was applied to titanocene-grafted silica catalysts, and it was thus possible to attain excellent selectivity (greater than 98%) and interesting yields (from 44% to 63%) in cyclohexene epoxide (Guidotti et al. 2009). Even higher yields in epoxide were found over Ti–MCM-48 than over Ti–MCM-41 or Ti–SiO2 owing to better isolation, dispersion and stability of Ti(IV) sites. Recently, the drop-wise addition of H2O2 has also been extended successfully to the epoxidation of methyl oleate (Guidotti et al. 2011b). All the tested titanosilicate molecular sieves showed good catalytic activity and comparable behaviour in terms of conversion, selectivity and yield of methyl 9,10-epoxystearate, notwithstanding the morphology and the texture of the silica support. By optimizing the reaction conditions (especially, the catalyst to reactant ratio), high yields of up to 91 per cent in epoxide were obtained for Ti–MCM-41, with a high stereoselectivity (80%) towards the cis-epoxide.
Anyway, even in the presence of aqueous H2O2, if the reaction is rapid enough, the oxidation can occur before the aggregation and deactivation of Ti sites takes place. This is the case with the oxidation of substituted phenols into benzoquinones and, for instance, very good yields (96%) in 2,3,5-trimethyl-1,4-benzoquinone, a vitamin E precursor, were obtained from 2,3,6-trimethylphenol after 1 h, with no remarkable deactivation of the catalyst and a good recyclability (Kholdeeva et al. 2007, 2009a). In this class of reactions, the catalysts prepared by grafting titanocene dichloride revealed a strong dependence of the product selectivity on the surface concentration of titanium active centres. Mesoporous materials with titanium surface concentration in the range 0.6–1.0 Ti nm−2 were identified as optimal catalysts for the transformation of alkylphenols into benzoquinones, whereas catalysts having less than 0.6 Ti nm−2 produced a mixture of benzoquinones and dimeric by-products. In fact, several spectroscopic evidences showed that dinuclear (Ti–O–Ti-like species) and/or subnanometre-size clusters evenly distributed on the silica surface are the best active sites for this reaction. This phenomenon can be explained by the particularly favourable interaction between the phenolic substrate and (at least) two contiguous Ti hydroperoxo groups, as depicted in the suggested oxidation mechanism (scheme 3).
Over such catalysts, the target of 100 per cent selectivity was attained in the oxidation of both trimethylphenol and dimethylphenol, and this is the first example that clearly demonstrates the advantages of Ti cluster-site catalysts over Ti single-site catalysts in hydrogen peroxide-based oxidation reaction. Moreover, by using several mesoporous silica supports, it was possible to find a correlation displaying the dependence of the selectivity on surface Ti density and to use it as a predictive tool for the preparation of catalysts with optimized performance for the chemospecific oxidation of substituted phenols to related quinones with hydrogen peroxide (Kholdeeva et al. 2009b).
(b) Anchored organometallic complexes as single-site heterogeneous catalysts
Different approaches have been introduced to immobilize on solid surfaces the active organometallic complexes that are designed and optimized to work under homogeneous conditions in the liquid phase.
One possible method is based on the surface organometallic chemistry, in which most of the know-how developed for homogeneous organometallic complexes can be transferred to the study of novel organometallic species where the solid surface itself acts as a ‘special’ ligand to the metal. Owing to this methodological approach, Basset and co-workers have produced a number of organometallic complexes anchored onto non-porous silica surfaces (Basset & Choplin 1983; Basset & Ugo 2009). The organometallic complex is attached directly to the oxide support (silica, alumina, etc.) via either a covalent/ionic bond or a Lewis–acid–Lewis–base interaction. In these systems, the coordination sphere of the metal involves not only the ancillary ligands that influence the stability, the activity and the selectivity as in homogeneous catalysis, but also the surface.
Another strategy of immobilization is to link the transition metal complex catalyst to a mesoporous material via a spacer chain (a tether). This immobilization can be carried out by substituting one of the ligands of the homogeneous catalyst by a similar moiety that contains a linker and which is able to anchor, typically via a covalent bond, to the support/carrier (oxide, polymer and dendrimer). According to this approach, most of the synthetic effort is devoted to preserve as much as possible, the molecular structure and the chemical environment after anchoring (Copéret & Basset 2007).
Owing to the anchoring strategy, large and bulky moieties can be covalently attached onto the surface of mesoporous silicas, whose mesopores are large enough to accommodate this kind of guest species. As an example, Ti-containing polyhedral oligomeric silsesquioxane (Ti–POSS), which is an excellent model of isolated and well-dispersed Ti sites in the SiO2 matrix, was first successfully deposited (Krijnen et al. 1998; Smet et al. 2000; Zhang et al. 2007) and then recently anchored via covalent bonds to the surface of an ordered (SBA-15) and a non-ordered silica support (Carniato et al. 2008, 2011) (figure 3). In the latter example, Ti–POSS moieties were, therefore, mainly accommodated as dinuclear dimer species on the external surface of the ordered mesoporous silica and in the large mesopores of the non-ordered silica support. In both cases, isolated Ti(IV) single sites were deposited on the silica surface, but they were not located in close proximity of silanol groups. These solids can thus be considered as model systems to study the behaviour of isolated Ti centres and the extent of interaction between the silanol-rich silica surface and the metal species. The final Ti–POSS–SBA-15 catalyst revealed a good dispersion of Ti sites and an interesting catalytic activity. Epoxidation tests on unsaturated terpenes (limonene, carveol and α-pinene) with TBHP as oxidant were performed to monitor the activity and accessibility of Ti(IV) sites in the anchored samples and compared with conventional titanocene-grafted Ti/SBA-15 and Ti/SiO2 systems. In terms of activity, anchored catalysts displayed comparable conversion and turn-over number values almost identical to grafted ones. On the contrary, in terms of selectivity to epoxide products, the anchored Ti–POSS-containing materials showed slightly better results than the grafted ones owing to the absence of proximal silanols surrounding the Ti centres and, hence, to the less marked formation of acid-catalysed side products (Carniato et al. 2011).
(c) ‘Supported hydrogen-bonded’ complexes as hydrogenation catalysts
Even though most of the heterogenized anchored catalyst are obtained via a covalent-linking approach, some examples of non-covalently bound homogeneous catalysts show specific advantages, such as remarkably mild grafting procedures (anchoring at room temperature, with no need of high-temperature treatments), with minimal or no ligand modification of the parent homogeneous catalyst and the possibility to easily recover the anchored complex for further characterization studies by standard liquid-phase techniques.
In this field, the groups of Bianchini & Psaro introduced, in 1999, a non-covalent method of preparing silica-immobilized metal catalysts for use in solid–liquid reactions in apolar solvents (Bianchini et al. 1999, 2000). The procedure involved the formation of hydrogen bonds between the terminal silanols of the silica surface and the oxygen atoms of sulphonate groups from phosphine and/or from trifluoromethansulphonate counteranion. This heterogenization protocol gives rise to supported hydrogen-bonded (SHB) catalysts and is alternative to the supported liquid-phase (SLP) method developed by Davis that involves the dissolution of a homogeneous catalyst in a hydrophilic solvent adsorbed as a thin layer onto a porous material with a hydrophilic surface (e.g. silica or glass) (Davis 1992).
The immobilization of the zwitterionic Rh(I) catalyst [(sulphos)Rh(cod)] (; cod=cycloocta-1,5-diene) was achi- eved by a straightforward solvent impregnation method using anhydrous dichloromethane solutions of the catalyst precursors that are then stirred in the presence of the support material until all the metal complex has been chemisorbed onto the silica surface. Reproducible 1–2 wt.% metal loadings were generally obtained. Once grafted to silica, the Rh complex was not extracted back into CH2Cl2, whereas stirring the SHB complex in alcohols at room temperature resulted in its complete desorption from silica and its dissolution into the solvent.
The immobilized catalyst (sulphos)Rh(cod)/SiO2 (1/SiO2 in scheme 4) is active for the hydrogenation of alkenes in either flow reactors (ethene and propene) or batch reactors (styrene) in hydrocarbon solvents. Moreover, in all the cases investigated, there was no evidence whatsoever of the formation of contiguous Rh–Rh sites, indicating that the catalytic active sites are truly isolated Rh atoms (i.e. real SSHCs), as in the homogeneous phase. A further interesting feature of SHB catalysis is provided by the facile and quantitative extraction of the grafted metal products with alcohols. This allows the study of metal products of single-site catalytic reactions by applying standard spectroscopic techniques in solution, as occurs when the heterogenization of organometallic complexes is achieved by the ion-pairing technique.
The SHB methodology has been extended to cationic metal complexes, of which very few heterogenized examples, obtained via the ion-pairing technique, are known (Beckler & White 1986; Scott et al. 1994). For this purpose, the Ru(II) complex [(sulphos)Ru(NCMe)3](OSO2CF3) can be mentioned. On the basis of spectroscopic characterization, a structure can be proposed for Ru(II)/SiO2, in which both the complex cations and the triflate anions are tethered to silica by hydrogen-bond interactions involving the oxygens from the groups and the −OH groups from the terminal silanols on the silica surface (2/SiO2 in scheme 4) (Bianchini et al. 2000). Obviously, the sketch is only a schematic representation of the reality, as there will be cations and anions hydrogen bonded to silica via two or even one ≡Si–OH units.
The SHB Ru(II)/SiO2 system is a selective catalyst for the hydrogenation of benzonitrile to benzylidenebenzylamine in n-octane (35% conversion and 99% selectivity after 1.5 h). No appreciable ruthenium leaching into the hydrocarbon phase was observed, even after three recycles. Notably, the reductive transformation of an organic nitrile into a secondary imine has never been reported to occur in a heterogeneous fashion. The results obtained in this work are thus quite interesting, as they outline a new and clean route to the preparation of secondary bis(aryl)imines, which are important organic synthons.
The Ru(II) complex immobilized on silica has been also successfully applied in hydrodenitrogenation/hydrodesulphurization catalysis of model hetero- and polyaromatics (Bianchini et al. 2003). The immobilised Ru(II) catalyst was active and highly selective in the hydrogenation of indole to indoline (turn-over frequency of 14 h−1, 100% selectivity to indoline), even in the absence of acids. It is worth noting that for this substrate, almost no hydrogenation was observed using the homogeneous catalyst in either CH2Cl2 or tetrahydrofuran, even for reaction times as long as 5 h (Barbaro et al. 2002). The presence of a protic acid is necessary to generate an equilibrium concentration of the 3H-indolium cation that contains a localized, hence reducible, C=N bond. The use of protic acids to promote the hydrogenation of indole is a common procedure in heterogeneous catalysis, for example, with Raney Ni or copper chromate catalysts (Adkins & Coonradt 1941; Smith & Utley 1965).
The formation of 3H-indolium cation is very unlikely to occur with the SHB Ru(II)/SiO2 system, as it needs strong acids: even trifluoroacetic acid and p-toluenesulphonic acid give incomplete protonation of indole, leading to an equilibrium concentration of 3H-indolium (Jackson & Smith 1964). Therefore, neither isolated silanols of silica (pKa>9) nor hydrogen-bonded silanols (pKa 5–7) should be able to protonate indole.
The hydrogen-bond interactions between the N–H proton of indole and the oxygen atoms of the silanols groups (either isolated or, much more likely, engaged in hydrogen bonding to the sulphonate groups) may activate the heterocyclic ring to expose a more localized, hence reducible, C2−C3 bond to ruthenium. Strong support for the positive influence of N−H…O(H)−Si<bonds on indole hydrogenation by Ru(II)/SiO2 was provided by the use of methylindole in the place of indole. The protection of the nitrogen atom with a methyl group did not allow for substrate hydrogenation. Silica surface-mediated reactions are a relatively new synthetic protocol, leading to interesting results in terms of activity and selectivity, as well as environmentally friendliness (Kabalka & Pagni 1997). The reactions can be conducted by simple mixing of silica, reagent and substrate in the dry state followed by heating, while the products are generally extracted from the adsorbent with an appropriate solvent. Mediation by silica was found to involve either isolated or associated silanols. Similar examples of silica surface-mediated reactions include the oxidation of sulphides and sulphoxides by TBHP (Kropp et al. 2000).
Like the SLP immobilization (Fache et al. 2000; Bianchini & Barbaro 2002), the SHB technique has been extended to chiral metal complexes with chelating phosphines and, as such, it is the simplest procedure ever reported for the synthesis of chiral heterogeneous catalysts, as commercially available chiral ligands may be employed with no modification of their structure (Wan & Davis 1994).
Two different types of SHB chiral catalysts, all containing rhodium, have been prepared: neutral complexes, such as [Rh(nbd)((R)-(R)-BDPBzPSO3)] ((R)-(R)-BDPBzPSO3 = (R)-(R)- 3 -(4-sulphonate)benzyl-2,4-bis(diphenylphos- phino)pentane, nbd=norbornadiene), and cationic complexes containing triflate counter-anions, such as [Rh(nbd)((+)-DIOP)]OTf and [Rh(nbd)((S)-BINAP)] OTf (3/SiO2 in scheme 4), where (+)-DIOP = (+)-2,3-O-isopropylidene-2,3-dihydroxy-1,4-bis(diphenylphosphino)butane and (S)-BINAP = (S)-(−)-1,1′-binaphthalene-2,2′-diyl)bis(diphenylphosphine).
The immobilization of the neutral precursors implies a direct hydrogen-bonding interaction between the metal complex and surface silanols and, therefore, the sulphonate group must be previously introduced into the chiral ligand.
In contrast, the cationic complexes reside close to the silica surface by electrostatic interaction with the triflate ions that, in turn, are immobilized on the support via hydrogen bonds. Consistently, no immobilization whatsoever was observed when the triflate anion was replaced by counter-anions such as or BF− that are not capable of hydrogen-bonding interaction with silica.
Chiral SHB rhodium catalysts have been used in the enantioselective hydrogenation of prochiral olefins (Bianchini et al. 2001), particularly itaconates and 2-acetamido acrylates, in n-heptane or n-hexane. As a general trend, the immobilization of the chiral precursors on silica did not reduce their enantioselectivity when compared with analogous homogeneous reactions in methanol, a beneficial effect was observed in some cases (cf. hydrogenation of dimethylitaconate, homogeneous versus SHB [((S)-BINAP)Rh(nbd)]OTf, which show 100% conversion, with 25% of enantiomeric excess (e.e.) (S) and 100% conversion, with 30% of e.e. (S), respectively). Moreover, the conversion values were generally comparable with or higher than those in the homogeneous phase.
Irrespective of the immobilization procedure of the metal complex (SHB or ion pairing to SHB counter-anion), no appreciable Rh leaching was observed within several consecutive heterogeneous runs, and an effective catalyst recycling with no loss of activity or enantioselectivity was accomplished by standard procedures.
When complexes are supported inside the pores (of the proper size) of mesoporous silicas, the resulting catalysts may show superior enantioselectivities, if compared with the homogeneous counterparts, owing to the confinement effect. Chiral Rh and Pd catalysts supported on commercial silicas, with different pore sizes, were tested in the asymmetric hydrogenation of methyl benzoylformate to methyl mandelate showing excellent e.e. values (i.e. [Rh(cod)PMP]CF3SO3/ SiO2 (PMP=(S)-(+)-1-(2-pyrrolidinylmethyl)-pyrrolidine) allows up to 94% e.e. to be obtained) (Raja et al. 2003). Analogous Rh complexes ([Rh(cod)L-L]CF3SO3, where L-L can be: (S,S,S)-[2-(4,5-dihydro-4′,5′-diphenyl- oxazol-2′-yl)ferrocenyl] phosphine ((S,S,S)-dipof); l-tryptophanbenzyl ester; (−)-(S)-2-(aminomethyl)-1-ethylpyrrolidine or (+)-(1R,2R)-1,2-diphenylethane-1,2-diamine) were immobilized on MCM-41 and used for the hydrogenation of (E)-(α)-phenylcinnamic acid to 2,3-diphenylpropionic acid showing similar good performances, up to 90 per cent conversion, 84 per cent selectivity and 97 per cent e.e. (Rouzaud et al. 2003).
SHB catalysts also found interesting applications in other reactions, different from the hydrogenation ones, such as cyclopropanation, Diels–Alder additions or epoxidation reactions.
Preformed polypyrazolylborate copper(I) complexes BpCu, Tp*Cu and [B(pz)4]Cu (Bp=dihydrobispyrazolylborate, H2B(pz)2; Tp*=hydrotris(3,5-dimethyl)pyrazolylborate, HB(pz*)3; B(pz)4=tetrakispyrazolylborate) were immobilized by classical and non-classical hydrogen-bond ligand-to-support interaction involving the hydroxyl groups of the silica surface and the borohydride BH group and/or the pyrazolyl nitrogen atoms on silica gel. The resulting catalysts were employed for the styrene cyclopropanation reaction. The catalytic activity is very similar to that observed under homogeneous conditions (i.e. 76% yield and 75:25 anti/syn ratio, under homogenous conditions versus 90%, yield and 54:46 anti/syn ratio, under heterogeneous conditions). The strength of such interactions lets the recovery and recycling of the supported catalysts for a number of cycles, even if some leaching was observed using solvents other than petroleum ether (Mar Díaz-Requejo et al. 2000).
Analogously, chiral bis(oxazoline) complexes of Cu(II), Zn(II) and Mg(II) immobilized on silica support via hydrogen-bonding interactions are very effective catalysts for the Diels–Alder reaction between 3-((E)-2-butenoyl)-1,3-oxazolin-2-one and cyclopentadiene allowing up to 93 per cent e.e. at room temperature to be obtained with the heterogeneous bis(oxazoline) complexes. Moreover, the catalysts can be recycled without significant loss of enantioselectivity (Wang et al. 2006).
Again, sulphonato-(salen)Mn(III) complex hydrogen bonded to silicas was revealed to be an effective and recyclable catalyst in the enatioselective epoxidation of styrene and α-methylstyrene (Tang et al. 2008). Enantiomeric excess values up to 100 per cent (with isolated epoxide yield of 87%) were obtained with a silica gel supported Mn complex in biphasic systems (BMImPF6/H2O).
Finally, a recent development of the SHB grafting methodology worth mentioning is the so-called ‘hybrid’ or transition complexes supported metal (TCSM) catalysts that consist of the combination, on the same support material, of dispersed metal nanoparticles (usually Pd) and grafted molecular complexes (mainly rhodium phosphine or amine complexes) (Dal Santo et al. 2010). These systems showed an enhanced activity, greater than the sum of the activities of the two single components, in aromatic hydrogenation reactions and for them, a synergistic effect of the single-site molecular metal complex on the metallic nanoparticles has been evoked.
(d) Single-site heterogeneous catalysts prepared by one-pot approaches
Microporous solids, such as zeolites and AlPOs, possess a well-organized inorganic structure composed of pores and channels with regular dimensions, where single-site catalytic entities can be introduced, aiming at the preparation of a wide range of catalysts for selective oxidation and acid-catalysed transformations (Hartmann & Kevan 1999; Albuquerque et al. 2006; Raja et al. 2008). The high performances in terms of activity and selectivity in oxidation reactions have been mainly attributed to the particular local structure and coordination geometry of the active metals when introduced in framework positions.
Various examples of single-site microporous catalysts prepared by the one-pot procedure and used for catalytic oxidation reactions are reported in the literature. Metal-containing AlPOs were extensively studied as catalysts for oxidation and oxidative hydrogenation reactions, obtaining interesting results (Thomas et al. 1999). In different cases, the catalytic performance of one-pot prepared materials was compared with that of solids containing the same active species in extra-framework positions (prepared by post-synthesis approaches, e.g. by ion-exchanging zeolites or zeotype materials) and proved to be better.
As an example, aluminophosphate Co–APO-5 solid was proposed as a catalyst for selective catalytic oxidation of organosulphur compounds (Chica et al. 2006). The catalyst was prepared by introducing the synthesis gel used for the AlPO preparation, Co(acetate), as a precursor of the redox site. A Co/H-Y zeolite material (containing a similar amount of metal species) was prepared by classical ion-exchange reactions to obtain a comparison.
The oxidation of different sulphur compounds (sulphides, benzothiophenes and thiophenes) with tert-butyl hydroperoxide occurs on both Co–APO-5 and Co/H-Y. Co–APO-5 gave the highest organosulphur oxidation rates (per Co atom), reflecting the role of tetrahedral Co framework cations that undergo facile redox cycles. Indeed, when Co3+ is in the framework position, it can be easily reduced to Co2+ with a parallel formation of OH surface groups. On the contrary, Co2+ cations at exchange sites in H-Y catalysts do not undergo redox cycles. The good conversion obtained with the largest organosulphur compounds (diphenyl sulphide and 4-methyldibenzothiophene) suggests that site accessibility is not a major problem within the channels of Co–APO-5, even for bulky substrates. In addition, this type of SSHC proved to be less prone to metal leaching under reaction conditions than ion-exchanged analogues.
In the recent literature, examples of the preparation of multi-functional microporous catalysts through one-pot reactions have been given. Raja and co-workers focused attention on bimetallic multi-functional solids, obtained by isomorphously replacing Al(III) and P(V) framework cations with tetrahedrally coordinated Co(III) and Ti(IV) ions, respectively (Paterson et al. 2011). The highly active catalysts exhibited an interesting synergistic effect between Co and Ti sites and an enhancement of the catalytic performance, when compared with results obtained from their corresponding monometallic analogues in the catalytic epoxidation of olefins. This behaviour was associated with the fact that the active metal ions are located not only in the AlPO framework, but also in close proximity with each other. It was also hypothesized that the simultaneous isomorphous substitution of Al(III) and P(V) in the AlPO-5 framework with Co(III) and Ti(IV) metal ions results in a higher fraction of Co(III) sites in the bimetallic catalyst with respect to the parent Co(III)–ALPO-5 (Barrett et al. 1996). These centres are supposed to be suitable sites for the generation of free-radical intermediates involved in the catalytic epoxidation reactions.
Other interesting examples of SSHCs were obtained by introducing redox metal centres in mesoporous solids through one-pot reactions. Although metal-containing mesoporous materials are often prepared by post-synthesis reactions, in some cases, the one-pot preparation approach is useful to obtain high-surface area solids with highly dispersed active species and with readily reducible/oxidizable metal centres. Very recently, the peculiar catalytic activity of one-pot prepared V–SBA-15 and V−SiO2 towards dichloromethane conversion with respect to impregnated solids was reported. This was associated with the presence of isolated V species incorporated inside silica walls produced during one-pot synthesis, while impregnated species especially contains poly-vanadate (or microcrystalline V2O5) phases that are not so active for gas-phase volatile organic compound oxidation purposes (Piumetti et al. 2010, 2011).
The examples presented in this brief overview show how SSHCs can be applied to different transformations of interest for the synthesis of richly functionalized fine chemicals. Redox-active metal-containing solid catalysts can be used as viable alternatives to stoichiometric non-catalytic processes (still widely used, especially in oxidation reactions) that are based on hazardous and environmentally unfriendly reactants. SSHCs proved, in many cases, to be robust, easily recyclable and potential candidates for developing new chemo-, regio- and stereoselective synthetic processes at a large-scale level. Nevertheless, the development of bioinspired inorganic equivalents of enzymes based on nanostructured molecular sieves with comparable activity and selectivity is still a challenge for researchers. This results in a main hurdle to an effective expansion of this class of catalysts at an industrial level, even if the high added value of the product could justify the use of valuable inorganic oxide-based SSHCs, often prepared by expensive and time-consuming synthetic routes.
Once again, a multi-disciplinary approach, based on a strict cooperation between chemists with different cultural backgrounds (in heterogeneous catalysis, organic synthesis, biochemistry, inorganic chemistry, materials sciences, molecular modelling, etc.) can be the only winning strategy to obtain successful results in this field in the next future.
The financial support of the European Community's 7th Framework Programme through the Marie Curie Initial Training Network NANO-HOST (grant agreement no. 215193) and of the Italian Ministry of Education, University and Research through the project ‘ItalNanoNet’ (Rete Nazionale di Ricerca sulle Nanoscienze; prot. no. RBPR05JH2P) is acknowledged.
One contribution of 14 to a Special feature ‘Recent advances in single-site heterogeneous catalysis’.
- Received January 16, 2012.
- Accepted February 14, 2012.
- This journal is © 2012 The Royal Society