“Ship-in-Bottle” Platinum Metals Clusters in NaY Micropores

Zeolites, namely faujasites (NaY, NaX), mordenite, ZSM-5 and ALPO-5, are aluminosilicate and aluminophosphate crystals consisting of microporous cages and channels of molecular dimensions (5 to 12 Å), see Table I, interconnected by smaller windows, see Figure 2.

Such micropores and ordered channels can be regarded as “nanometre-sized microreactors”, offering a templating space for the synthesis of selected metal complexes, such as platinum metals clusters. While some robust metal carbonyl clusters, such as Rh₆(CO)₁₆, (van der Waal radius of 10 Å) and [Pt₁₂(CO)₂₄]^2- (8 × 12 Å) are unable to enter directly into the NaY (12 Å) pores through the small window (6 Å), they can, however, be synthesised and encapsulated in the cages by successive carbonylation reactions of the precursor rhodium and platinum ions with CO + H₂ or CO + H₂O, which act as building blocks, entering through the window. The metal ions are introduced into the cages by conventional ion-exchange, while some volatile subcarbonyls and metal complexes are introduced into the cages by the solid-state dispersion technique (8). This in situ intrazeolitic preparation of nanomaterials, in this case Rh₆(CO)₁₆ and [Pt₁₂(CO)₂₄]^2- in micropores, is the “ship-in-bottle” synthesis, see Figure 3.

To produce Rh₆(CO)₁₆ in NaY, ion-exchanged Rh³⁺ in NaY zeolite was reduced with CO + H₂ or CO + H₂O at 393 to 473 K to give the mononuclear dicarbonyl species, [Rh(CO)₂(-O-)₂] in Figure 3, where O are shared with the zeolite wall. This then migrated through the zeolite channels and underwent a subsequent cluster-oligomerisation to Rh₄(CO)₁₂ (6 Å) and Rh₆(CO)₁₆ (10 Å), the latter just fitting in the interior of NaY (12 Å) (9, 10). This is analogous chemistry to that occurring in aqueous alkaline solution (11). Rh₆(CO)₁₆ was selectively produced with a minor contribution from residual mononuclear Rh carbonyls inside NaY. Rh₆(CO)₁₆ encapsulated in a NaY zeolite cage by the “ship-in-bottle” technique was well characterised by IR bands for linear and face-bridging carbonyls, which are different from those of Rh₆(CO)₁₆ adsorbed on NaY surfaces. Additionally, EXAFS (Extended X-ray Absorption Fine Structure) data provided evidence for the stoichiometric formation of a hexanuclear rhodium carbonyl cluster, in good agreement with a free molecule, in terms of coordination numbers and atomic distance.

Similarly, Ir₆(CO)₁₆ in NaY cages was synthesised using ion-exchanged Ir⁴⁺/NaY with CO + H₂ at 1 atm and 323 K via the intermediate formation of Ir(CO)₂ and Ir₄(CO)₁₂, in analogous processes to those for [Rh₆(CO)₁₆]/NaY (9, 12). Gates and colleagues reported that Ir(CO)₂(acac) impregnated with NaY was converted, using CO + H₂ at 20 atm and 623 K, to Ir₆(CO)₁₆, whereas [Ir₆(CO)₁₅]^2- was formed in basic NaX (SiO₂:Al₂O₃ = 3.2) (13). In this case, the zeolite micropores acted as a “common templating microreactor” to selectively synthesise the homologous hexanuclear clusters in the confined reaction space.

Sachtler and co-workers reported a new type of palladium carbonyl cluster [Pd₁₃(CO)_x] (x is unknown) in the supercage of a NaY zeolite (12 Å) and [Pd₆(CO)_x] in a NaA zeolite (5 Å), although they were not fully characterised by IR and EXAFS (14). Ichikawa and colleagues have recently extended the “ship-in-bottle” synthesis to some other metal clusters, such as H₄Ru₄(CO)₁₂/NaY (15), [HRu₆(CO)₁₈]^-/NaY (16), [Ru₆(CO)₁₈]^2-/NaX (17), [Pt₉(CO)₁₈]^2-/NaY (18, 19) and [Pt₁₂(CO)₂₄]^2-/NaY (20). In particular, to produce [Pt₁₂(CO)₂₄]^2-/NaY, calcined Pt²⁺/NaY was heated from 298 to 373 K under CO with a trace of H₂O. This resulted in the successive formation of different carbonyl species with characteristic carbonyl IR bands. By analogy with IR data from known platinum carbonyl complexes, it was suggested that Pt²⁺/NaY reacts with CO to form PtO(CO) in NaY and a proposed Pt trigonal intermediate species [Pt₃(CO)₃(μ₂-CO)₃]. These are eventually converted, by stacking, to the dark-green Chini complex of [Pt₁₂(CO)₂₄]^2- (2080s and 1824m cm^-1) (20).

By contrast, [Pt(NH₃)₄]²⁺ ion-exchanged with NaY selectively gave the smaller, orange-brown Chini cluster, [Pt₉(CO)₁₈]^2-, which has intense IR bands at 2056s and 1798m cm^-1 (19, 20).

The spatial restrictions of the micropores in NaY and NaX may hinder cluster migration and interactions, and promote cluster isolation and stability. “Ship-in-bottle” syntheses may provide opportunities for the “rational” design of discrete platinum metals clusters of uniform size and metal composition. Clusters already made have sufficient stability against sintering and leaching under prevailing reaction conditions. Some encapsulated metal clusters are listed in Table II.

Robust Platinum Clusters in the Channels of Mesoporous FSM-16

A new family of mesoporous sieve materials, such as FSM-16 and MCM-41, has been synthesised by using micelle surfactant templates, such as alkyltrimefhyl ammonium salts (21, 22). These materials consist of ordered mesoporous channels of diameter 20 to 100 Å, larger than those of conventional zeolites. The mesoporous sieve materials are potential hosts for robust metal clusters aligned in the ordered channels and accessible to larger substrates in catalytic reactions.

Chini Cluster Synthesis

Robust rod-like platinum clusters [Pt₃(CO)₆]_n^2- n = 5 and 6; 6 × 15 to 18 Å) (the Chini complex) have been successfully synthesised using the mesoporous FSM-16 channels (28 and 48 Å) as a host reactor (23, 24). This was achieved by exposing FSM-16 (28 Å) impregnated with H₂PtCl₆ to CO at 323 K (22) and gave cis -Pt(CO)₂Cl₂ and [Pt(CO)Cl₃]^-. These were then converted with CO + H₂O vapour at 323 K to an olive-green complex (23). The Chini complex extracted from FSM-16 by cation metathesis was identified as [Pt₁₅(CO)₃₀]^2- by FTIR and UV-vis spectroscopy (v_co = 2086s and 1882m cm^-1 and UV-vis reflectance at λ_max = 452 and 805 nm). The Pt₁₅ cluster anions formed in FSM-16 were isolated and relatively stabilised by using quaternary alkyl ammonium cations NR₄⁺ as spacing molecules.

Similarly, a larger trigonal prismatic Pt₁₈ cluster anion has been synthesised uniformly (24, 29) by the reductive carbonylation of H₂PtCl₆ in FSM-16 of channel size 48 Å. This was characterised by IR, UV and EXAFS spectroscopy by analogy to the reference complex, [NEt₄]₂[Pt₃(CO)₆]₆.

The average interfacial distances (R_(pt-pt) = 3.07 Å) between adjacent triplatinum planes for these clusters in FSM-16, that is [Pt₃(CO)₆]₅^2-/NEt₄Cl/FSM-16 (28 Å) and [Pt₃(CO)₆]₆^2-/NEt₄Cl/FSM-16 (48 Å), are in good agreement with those of the reference salt [NEt₄]₂[Pt₃(CO)₆]₅ in boron nitride, within experimental error. By contrast, for Chini clusters in NaY micropores, for example, [Pt₃(CO)₆]₄^2-/NaY and [Pt₃(CO)₆]₃^2-/NaY (19, 20) the interfacial Pt-Pt distance of the Pt₃ and Pt₁₂ triangle is substantially shorter (R_(Pt-Pt)) = 2.98 to 2.99 Å) than the reference compounds both in crystalline form and in solution. This is because of compression and distortion of the bulky Chini clusters by the small (12 Å) NaY (and NaX) microcavities. Transmission electron microscopy (TEM) observation of [Pt₃(CO)₆]₅^2-/FSM-16 revealed that the platinum clusters were small specks of size 10 to 15 Å, uniformly scattered and aligned in the ordered FSM-16 channels (21, 22). No larger platinum particles or crystals were found on the external surface of FSM-16.

Alloy Clusters in NaY and MCM-41 Cages

Bimetallic carbonyl clusters, Rh_6-xIr_x(CO)₁₆/NaY (x = 2-4) (9-12), [Fe₂Rh₄(CO)₁₅]^2-/NaY (25) and [HRuCo₃(CO)₁₂]/NaY (17) have been synthesised by the “ship-in-bottle” technique and characterised by IR, EXAFS, UV and Raman spectroscopy, see Table II.

A series of [Rh_6-xIr_x(CO)₁₆]/NaY (x = 2-4) clusters was prepared by reductive carbonylation of double ion-exchanged [Rh³⁺ + xIr⁴⁺]/NaY in CO + H₂ at 1 arm and 393 to 473 K, in the same way as [Rh₆(CO)₁₆]/NaY (9, 10) and [lr₆(CO)₁₆]/NaY (9, 12) were prepared. These had linear and face-bridging CO bands characteristic of hexanuclear metal clusters. EXAFS studies revealed that the formation of the bimetallic Rhlr clusters was stoichiometric, with metal compositions similar to those of the starting ion-exchanged NaY. The reduced alloy clusters Rh_6-xIr_x/NaY (x = 2–4) were prepared by oxidising the carbonyl clusters, followed by reduction with hydrogen at 673–723 K. This novel preparation of hexanuclear Rhlr alloy clusters encapsulated in NaY is shown in Figure 4.

The RhFe carbonyl cluster [Fe₂Rh₄(CO)₁₅]^2- has also been prepared by the solid-state reaction of [HFe(CO)₁₁]/NaY with Rh₄(CO)₁₂ at 343 K (25), in an analogous reaction to that in solution between [Fe₃(CO)₁₁]^2- and Rh₄(CO)₁₂ or Rh₂(CO)₄Cl₂. An IR study revealed that the active Rh(CO)₂ species, generated by decomposition of precursor Rh₄(CO)₁₂, migrates inside the NaY zeolite framework.

This facile migration of monometal carbonyls results in the rebuilding of more stable bimetallic cluster anions. However, groups lead by Thomas and Johnson recently reported an elegant example of loading giant bimetal clusters, such as [Ru₁₂C₂(CO)₁₆Cu₄Cl₂]^2- and [Ag₃Ru₁₀C₂(CO)₂₈Cl]^2-, as the precursors in mesopores of MCM-41 (28 Å) (26). TEM and EXAFS data showed no evidence of segregation or aggregation after the pyrolysis of the RuCu and AgRu precursor clusters in MCM-41.

The bimetal catalysts produced are ∼ 15 Å in diameter, rosette-shaped with twelve exposed ruthenium atoms connected to a square base composed of partly concealed copper atoms. These RuCu clusters are anchored by copper ions bound to four oxygen bridging silanols of the mesopore lining. This anchoring model for RuCu clusters on MCM-41 is analogous to that for RhFe and RuCo bimetal clusters on silica, encapsulated in NaY cages (27), and immobilised by Fe-O and Co-O bonds, respectively, attached to silica or alumina surfaces.

Cluster Transformations to Nanoparticles in Micro/Mesopores

After the mild oxidation of [Rh₆(CO)₁₆]/NaY with dry O₂ by heating from 293 to 473 K, IR and EXAFS data suggest that the Rh carbonyl cluster is decomposed and converted to an oxide cluster. By successive reductions with hydrogen at 473 and 673 K, a skeletal rhodium cluster formed in the cavity (11). EXAFS analysis showed that the rhodium hexanuclear atomic unit had been retained in the zerovalent state. The coordination number (C.N.) is the same (C.N. = 3.2) as the original [Rh₆(CO)₁₆]/NaY and the atomic distances are close to those of metallic rhodium (R_(Rh-Rh) = 2.72 Å). Exposure of this reduced rhodium, Rh₆, to CO at 300 K results in the stoichiometric formation (CO:Rh_(total) = 2.6) of Rh₆(CO)₁₆. Thus the NaY micropores provide the preparative and catalytic cages to accommodate the metal tlusters through cyclic sequences of oxidation, reduction with H₂ and carbonylation with CO/H₂ to prevent cluster sintering in the micropores at higher temperatures.

Robust platinum clusters, such as [Pt₁₅(CO)₃₀]^2- in FSM-16 (28 Å) and [Pt₁₈(CO)₃₆]^2- in FSM-16 (48 Å) are similarly transformed to nanoparticles by gentle heating at 300 to 473 K in vacuum and were characterised by EXAFS and FTIR (27, 33). The platinum nanoparticles (size 15 to 20 Å), see Figure 5, were uniformly positioned, aligned and synthesised in the ordered mesoporous channels of FSM-16 by thermal evacuation at 573 K to remove CO from [Pt₁₅(CO)₃₀]^2- in FSM-16. The channels thus play a templating role in the synthesis of platinum metal nanoparticles by thermal pyrolysis of the Chini platinum complexes, see Figure 6.

Platinum Nanowires in Mesoporous Channels of FSM-16

We have recently reported the templated fabrication of platinum nanowires in mesoporous FSM-16 channels by the photoreduction of H₂PtCl₆/FSM-16 in the presence of propan-2-ol and water (28, 29). A typical TEM image of platinum nanowires in FSM-16 (28 Å) is shown in Figure 7. The nanowires have diameters of ∼ 3 nm, which is the pore size of FSM-16; their lengths can range from 50 to 200 nm. The nanowires are in the internal channels of FSM-16, not on the external surface. The TEM also shows that the nanowires have crystal faces, and the high resolution electron diffraction image gives a clear fringe pattern of Pt(110), implying that the platinum wires consist of a single crystal phase. The mesoporous channels of the FSM-16 thus act as a template to fabricate the platinum nanowires by a one-dimensional elongation of platinum crystals, see Figure 6. EXAFS characterisation shows that the platinum nano-wires and nanoparticles in FSM-16 have different parameters to platinum foils. For nanowires, the Pt-Pt distance and coordination number are 0.274 nm and 7.8, respectively, but 0.272 nm and 6.7, respectively, for nanoparticles (29).

However, XANES (X-ray Absorption Near Edge Structure) and XPS indicate that the nanowires in FSM-16 are slightly electron-deficient compared with platinum foil and nanoparticles in FSM-16. This may be due to the dominant interaction of the nanowires with the inner acidic surface of the FSM-16 pores. On exposure to UV-light, TEM and EXAFS studies show that platinum nanoparticles (∼ 1 to 2 nm diameter) are initially formed by the photoreduction of the platinum cations impregnated in the mesopores. [PtCl₆]^2- ions then migrate to the vicinity of the nanoparticles and are reduced on their surfaces to grow and elongate as nanowires in the confined FSM-16 channels, see Figure 8.

Ryoo has also reported the thermal preparation of platinum nanowires in MCM-41 by a stepwise hydrogen reduction of platinum ions which add to pre-reduced Pt/MCM-41 at above 823 K (30).

Magnetic susceptibility measurements on samples of platinum nanowires and nanoparticles of ∼ 2 nm size encapsulated in FSM-16 were performed by varying the temperature from 5 to 300 K at 30 kOe using a SQUID magnetometer (28). The inverse susceptibility of the platinum nanoparticle/FSM-16 was found to obey the Curie-Weiss law with magnetisation ∼ 1.54 μ_b/gram of Pt and Curie temperature, θ = 0.05 K, see Figure 9. By contrast, for platinum nanowire/FSM-16, below 100 K, a Curie-Weiss law dependency was observed with magnetisation ∼ 0.131 μ_b/gram of Pt, and Curie temperature, θ = 0.48 K. There is a clear deviation from the Curie-Weiss law above 100 K. This unique superpara-like magnetism of the platinum nanowire/FSM-16 can be explained by the quantum size effect or spin-ordering due to the one-dimensional morphology characteristic of a nanowire, and may be based on a quantum confinement effect in the narrow wire and hetero-coupling between internal mesoporous channels of FSM-16.

Separating the platinum nanowires from the FSM-16 was attempted by dissolving FSM-16 with aqueous HF or NaOH solution, but the nanowires decomposed to form platinum aggregates. The platinum nanowires were successfully extracted when [NBu₄]Cl in a benzene/ethanol solution was used. TEM-EDX showed that the nanowires were pure platinum with no silicon, indicating successful removal from the FSM-16 support. The extracted wires are probably stabilised by the [NBu₄]Cl which covers their surface to form organosols.

Catalytic Reactions

The catalytic performances of encapsulated platinum nanomaterials have been investigated as functions of their morphology, structural confinement and cluster-support interactions.

Water Gas Shift Reaction

The water gas shift reaction (WGSR) has been recently studied on [Pt₁₅(CO)₃₀]^2-/[Et₄N]⁺/FSM-16 (23), platinum nanoparticles/FSM-16 (28 Å) (24), platinum nanowires/FSM-16 (28, 29), [Pt₁₂(CO)₂₄]^2-/NaY and on [Pt₉(CO)₁₈]^2-/NaY (12 Å) (31).

As shown in Table III, [Pt₁₅CO)₃₀]^2- cluster anions in FSM-16 exhibited remarkably high activities to form an equimolar mixture of CO₂ and H₂ at 298 to 323 K, compared to the Pt₁₂ and Pt₉ cluster anions in NaY. The higher activity of the Pt₁₅ cluster anions in FSM-16 is probably due to the flexible cluster frameworks, similarly seen in solution, compared with the Pt₁₂ and Pt₉ clusters confined in the NaY micropores. The turnover rates (TOF: mol(CO₂)/Pt surface atom/min) for the WGSR at 323 K for platinum nanowires (2.8 nm × 100-200 nm long)/FSM-16 (28 Å), were 60 to 90 times higher than TOF on the platinum nanoparticles (∼ 20 Å) in FSM-16 (28 Å) (29). XPS and EXAFS data suggest that the unusually high TOF for the platinum nanowires may be linked to the different morphology and to a larger interaction (or contact) with the acidic surfaces of FSM-16 channels, resulting in a bigger electron deficiency than for FSM-16-encapsulated platinum nanoparticles.

The NO + CO Reaction

The Pt₉ and Pt₁₂ carbonyl cluster dianions in NaY microcavities exhibit 10 to 15 times higher catalytic activity for NO reduction with CO to N₂ and N₂O at lower temperatures (300 to 473 K) (19, 20), compared to conventional Pt/γ-Al₂O₃ catalyst.

When [Pt₁₂(CO)₂₄]^2-/NaY is exposed to NO (150 torr), in situ FTIR studies show that NO breaks the intra- and intertrigonal Pt-Pt bonds on the Pt₁₂ carbonyl cluster – even at 298 K – to give a trigonal intermediate “Pt₃(CO)₆” and PtO(CO) during formation of N₂ (N₂O) and CO₂. Furthermore, [Pt₁₂(CO)₂₄]² can be reversibly regenerated from platinum carbonyl fragments, such as “Pt₃(CO)₆ and PtO(CO), inside NaY by reaction with CO and water at 300 to 353 K to produce CO₂. In situ FTIR and mass spectrometry studies suggest that NO reduction with CO proceeds catalytically to give N₂/N₂O and CO₂ through redox cycles of cluster-breaking-regeneration (Pt₁₂ ↔ Pt₃ + Pt₉) under the NO + CO atmosphere. In fact, the framework of the Pt₁₂ clusters in NaY is apparently retained under the stationary reaction.

CO Hydrogenation to Alcohols

Hydroformylations of ethene and propene and CO hydrogenation were conducted at 373 to 453 K on [Rh₆]/NaY and RhFe/NaY (33, 34). Acetaldehyde was selectively obtained as the hydroformylation product on [Rh₆]/NaY, while with RhFe/NaY higher activities and selectivities for alcohols were obtained. In particular with RhFe/NaX, selectivities of close to 83 mol% for linear alcohol were obtained in propene hydroformylation, perhaps due to being confined in the basic NaX pores.

A reduced sample of [Rh₆] + [Fe₃]/NaY was catalytically active for the hydroformylation of ethene and propene with almost the same specific rates and selectivities as those obtained on [Rh₆]/NaY, but there were no RhFe bimetallic promotion sites in the reaction. By contrast, bimetallic RhFe/NaY (and RhFe/NaX) provided a good yield of oxygenates, mainly ethanol and methanol, and decreased the hydrocarbon content in the CO + H₂ reaction at 623 K (33, 35). In fact, [Rh₆]/NaY and [Rh₆] + [Fe₃]/NaY preferentially formed methane and higher hydrocarbons with a minor yield of acetaldehyde. The results are accounted for by adjacent RhFe bimetal assemblies being active sites for the olefin hydroformylation and the CO + H₂ reaction to synthesise alcohols. It has been suggested that a two-site CO activation by bimetal clusters (for example, RhFe, RhCo and RuCo) on silica and alumina is responsible for the enhanced migratory insertion of CO into M-alkyl and M-H bonds (M is metal) to produce oxygenates from olefin carbonylation and CO hydrogenation, respectively (34).

Alkane Hydrogenolysis on RhIr Alloy Clusters in NaY

The hydrogenolyses of n -butane and ethane and benzene hydrogenation were used as probe reactions of Rhlr alloy clusters prepared from [Rh_6-xIr_x(CO)₁₆] in NaY pores (11–13). A dramatic decrease in hydrogenolysis activity, by four orders of magnitude, across all the Rhlr bimetallic clusters was observed as the number of iridium atoms in the rhodium assemblies was increased; the activity for benzene hydrogenation increased slightly. This remarkable suppression of activity was interpreted in terms of the rhodium assembly-size effect (rhodium clusters were broken by a modest iridium atom), and of decreasing electron deficiency of the rhodium-rich bimetal clusters in NaY. Alkane hydrogenolysis is classed as a “structure-sensitive” reaction in which relatively large ensembles of metal atoms are required to break the multi-bonds of reactant molecules such as CO, N₂ and alkanes. By contrast, benzene hydrogenation has a different response (it is structure-insensitive) to the metal composition of Rhlr clusters in NaY. These results demonstrate the important advantage of the “ship-in-bottle” technique for putting bimetallic clusters in micropores, giving an isolated site model for heterogeneous catalysis (3, 5).

Conclusions

Novel templating syntheses of nanomaterials using inorganic micro- and mesoporous cages and channels for the tailored design of platinum metals catalysts have been discussed. Extending these techniques could result in nanotechnology being used to produce nanoparticle alloy catalysts – of PtRe, PtIr and PtCo, for example, – in zeolites and textured carbon, with higher performances for oil refining and for hydrogen-powered fuel cells. New ways of synthesising fine chemicals, using the spatially confined chiral space of mesoporous material, implanted with platinum metals clusters, may also result. The novel “ship-in-bottle” nanotechnology can provide spatial ordering and alignment of nanoparticles and nanowires of metals and semiconductors, and for photosensitisers for microdevices, for electronics, magnetics and non-linear optics. Further research would help to explain the dependence of cluster catalysis and catalyst stability on structure and morphology and explain the micro-mechanisms of cluster-sintering and metal-support interactions.