Читать книгу Heterogeneous Catalysts - Группа авторов - Страница 69

Having made, purified, and fully characterized chemically synthesized, atomically precise clusters, one can start using these species in catalysis. There are a limited number of fundamental studies focused on chemically made clusters as model homogeneous catalytic systems in the gas phase by employing mass spectrometry approach [49, 50]. There are studies that used ligated clusters as homogeneous catalysts in the liquid phase [41, 51–53]. Lewis proposed an interesting criterion with respect to catalysis by clusters: “unusual or enhanced reactivity of a catalyst composed of two or more metal atoms which differs from the individual metal components indicates cluster catalysis” [41]. However, studies comparing mono‐atomic catalysts and cluster catalysts are rare. Recently, selective hydrogenation of α,β‐unsaturated ketones and aldehydes was demonstrated using thiol‐protected cluster Au₂₅(SR)₁₈ as a homogeneous catalyst, whereby 100% selectivity toward unsaturated alcohols was observed [54]. Classical use of chiral ligands to induce stereoselectivity employed in homogeneous catalysis with monometallic organometallic complexes was recently applied to metal cluster catalysts with some success [55]. In the case of larger metal clusters, the chirality of the system may arise from the specific geometry of both the metal core and non‐chiral ligands bonded to the core – an unexplored opportunity with respect to achieving stereoselective control using clusters [56].

However, care must be taken not to assign observed activity to intact clusters as the results of a recent study show that “very small gold clusters (3 to 10 atoms) formed from conventional gold salts and complexes can catalyze various organic reactions at room temperature, even when present at concentrations of parts per billion” and that such clusters formed, in situ, “give reaction turnover numbers of 10⁷ at room temperature” [57]. This is an important issue since ligands bound to the cluster core stabilize clusters, help to solubilize them in an appropriate solvent, and, at the same time, should be blocking reagents from accessing metal core (as space‐filled images in Figure 5.4). Removal of ligands is likely to compromise the stability of the original clusters and could lead to formation of ultrasmall clusters, which could be responsible for the observed activity. A potential solution could be the use of stable, atomically precise nanoclusters with uncoordinated sites at the metal core, of which synthesis was only very recently developed [58]. Finally, the recovery and recycling of valuable catalyst is a major problem of homogeneous catalysis; clusters could be potentially recovered by extraction if the ligand periphery is appropriately modified (charge, functional groups, etc.) [59]. In some cases, clusters could be tethered to the support by flexible molecular linkers, which would form strong bonds with the support and the cluster ligand periphery, keeping the cluster intact and flexible enough to act as homogeneous catalyst; this approach effectively heterogenizes (anchors) homogeneous cluster‐based active site, enabling easy recycling [60].

The use of metal clusters as precursors in fabrication of heterogeneous catalysts was pioneered (among others) by Ichikawa [61], Gates [62, 63], Johnson, and Thomas [60, 64–66]. Chemically synthesized, purified, and fully characterized clusters can be deposited onto various supports (activated carbon, oxides, etc.) using impregnation and adsorption methods. In most cases, researchers have excellent control over the metal weight loading (by using known desired amounts of support and cluster precursor). In such “site‐isolated” catalysts, each cluster will form a single active site at the surface of support. In this way, atomically precise control over composition of the active site could be achieved so that the nature of the cluster chosen as a precursor in catalyst fabrication will fully define the nature of the active site.

There is one important issue pertinent to this approach – activation of the catalysts. Activation of the “as‐deposited” cluster‐based catalysts requires the removal of ligands and expose the metal core surface, as well as to establish a direct chemical interaction between the cluster core and support – effectively grafting the cluster core onto the support. Such ligand removal could be justified in light of the steric protection of the metal core by the ligands (as illustrated in Figure 5.4) that prevent the access of the reagents to the catalytically active cluster metal core. Care must be taken to preserve the composition of the cluster during such activation (as opposed to fragmentation or aggregation of the clusters). Ichikawa suggested oxidation of the supported metal clusters under mild conditions, followed by the reduction of resulting oxide particles [61]. An alternative milder approach was developed by Johnson and Thomas [60, 64–66], where mild heat treatment of the supported metal carbonyl clusters under vacuum resulted in the formation of the naked cluster cores immobilized on the support. This approach is based on the understanding that an equilibrium exists between free CO and CO ligands bound to the cluster core, which could be shifted (according to Le Chatelier's principle) by choice of conditions; and the removal of free gaseous CO by pumping could be quite efficient. By drawing an analogy between homogeneous and heterogeneous catalysis, Gates highlighted the interaction between the metal cluster core and the ligands in the as‐synthesized cluster [62, 63]. In classical homogeneous catalysis, where monometallic complexes are used, both the steric and electronic factors of the ligands can be fine‐tuned to achieve optimal catalytic performance. An example of tuning steric factors in cluster‐based catalysis was mentioned above with respect to the use of clusters with chiral ligands in stereoselective homogeneous catalysis. In the case of heterogeneous catalysts, tweaking the interaction between the metal clusters and the support material presents an opportunity for tuning of the electronic factors via, for example, charge transfer between support and the cluster‐based active site.

Some supports are known to enhance catalytic activity of metal particles, and in particular, reducible supports such as titania (TiO₂) and ceria (CeO₂) are known to belong to this category and exhibit the so‐called strong metal–support interactions. Other supports, such as silica (SiO₂) and boron nitride (BN), are known to be inert – they do not boost catalytic activity of supported particles.

Gold metal clusters on inert supports were shown to exhibit size‐dependent activity in the oxidation of styrene oxidation by air alone (in many other cases peroxides are used as oxidants or initiators) [67]. The authors used phosphine‐stabilized ∼1.5 nm Au clusters (“Au₅₅”) deposited from solution onto inert SiO₂ and BN supports and activated by heat treatment under vacuum at 200 °C. The activity in the partial oxidation of styrene (as contrary to the complete oxidation to CO₂) was observed only when particle size remained below 2 nm in the activated catalysts. Noteworthy, overloading of support with the same precursor (use of 6 wt% cf. 0.6 wt%) resulted in pronounced aggregation and loss of activity, while catalysts with low loading of sub‐2 nm particles were stable and recyclable (Table 5.1). The control over the metal loading is elegantly simple in this case because the calculated amount of cluster is dissolved in the appropriate solvent (dichloromethane) and stirred with determined amount of support (BN or SiO₂), followed by the removal of solvent under vacuum and thermal activation at 200 °C.

Table 5.1 Catalytic results of the partial oxidation of styrene using O₂ alone for supported Au₅₅ and comparison catalysts prepared by various techniques.

Source: Reprinted with permission from Turner et al. [67]. Copyright 2008, Springer Nature.

Catalyst	Preparationa)	Exact loadingb) (wt%)	Mean Au sizec) (nm)	Conversion (%)	Selectivity (%)
0.6‐wt% Au55/BN	Au55	0.63	1.6	19.2	82.3	14.0	3.9
0.6‐wt% Au55/SiO2	Au55	0.67	1.5	25.8	82.1	12.0	5.7
0.6‐wt% Au55/SiO2	Au55	0.67	—	21.4	69.2	23.7	7.1
Recycled 1
0.6‐wt% Au55/SiO2	Au55	0.67	—	15.9	63.7	27.1	9.2
Recycled 2
6‐wt% Au/SiO2	Au55	6.35	3.0	Trace	Trace	—	—
0.6‐wt% Au/SiO2	PVP	—	4.0	Trace	Trace	—	—
l‐wt% Au/C	ME	—	17.0	No reaction	—	—	—
5‐wt% Au/SiO2	IW	—	>30	No reaction	—	—	—

PVP, poly‐n‐vinylpyrrolidone; ME, microemulsion; IW, incipient wetness.

Mean Au particle size, calculated by counting particles in high‐resolution TEM images of many different regions, is the crucial factor in determining activity. Results from recycled Au₅₅‐based catalysts confirm that deactivation does not occur. All reactions were carried out at 100 °C in toluene.

^a Catalyst preparation: Au₅₅, Au₅₅ preparation.

^b Exact loading was determined using inductively coupled plasma mass spectroscopy.

^c For full statistical breakdown of particle size distributions.

The importance of efficient activation of cluster‐based catalysts for CO oxidation was highlighted by a study that compared ligand removal using ozone exposure against a rapid thermal treatment for catalysts containing Au₁₃[PPh₃]₄[S(CH₂)₁₁CH₃]₄ (8 Å diameter) ligand‐protected clusters on titania (anatase phase) [68]. The authors demonstrated that ozone treatment, followed by calcination at 400 °C for two hours in air, resulted in the most active catalyst, which had smaller gold particles compared with the less active catalyst treated by calcination alone.

The team of Professor Tsukuda used [Au₁₁(PPh₃)₈Cl₂]Cl, mentioned earlier [27], to prepare a series of mesoporous silica‐supported catalysts for oxidation of alcohols. The authors reported the importance of using a mixture of solvents (80% dichloromethane and 20% ethanol) in order to achieve homogeneous dispersion of clusters over a large surface area by optimizing the solvent‐mediated interaction of cluster with support [69]. Heat treatment at 200 °C produced active catalysts with Au clusters of ∼1 nm confined within a high surface area mesoporous silica support (SBA‐15) [69]. In another excellent study, the same team used hydroxyapatite (Ca₁₀(PO₄)₆(OH)₂) as a support for a range of thiol‐protected gold clusters Au_n (n = 10, 18, 25, 39). The authors pinpointed a maximum of activity with an impressive turnover frequency (TOF) of 18 500 h⁻¹ per Au atom at n = 39 in solvent‐free peroxide‐initiated aerobic selective (∼99%) oxidation of cyclohexane to cyclohexanol and cyclohexanone [70]. The use of hydroxyapatite prevented cluster aggregation due to strong interaction between the Au clusters and PO₄³⁻ moieties. However, it proved impossible to differentiate catalysts with respect to cluster sizes since, in all samples, these appeared to be around 1 ± 0.5 nm in size even when sophisticated high‐angle annular dark field‐scanning TEM was used at moderate to high magnifications. This observation was explained by formation of “an ensemble of structural isomers produced in the calcination of Au_n(SG)_m,” which “will give rise to polydisperse TEM images.” Aberration‐corrected electron microscopy studies of pure Au₁₄₄ supported on ultrathin carbon film, followed by reconstruction of obtained data, allowed construction of 3D electron density maps of such clusters [71]. Atomically resolved structures of Au₉ clusters supported on TiO₂ nanosheets (Figure 5.6) obtained using aberration‐corrected electron microscopy were successfully matched with structural isomers of supported clusters predicted by high‐level quantum mechanical calculations [72].

Figure 5.6 High‐resolution electron microscopy images (a1–c1) and (a2–c2) of Au₉ clusters deposited on titania nanosheets under optimized imaging conditions. The scale bar is 0.5 nm. Panels I, III, and IV show the density functional theory (DFT) models for three isomers of Au₉.

Source: Al Qahtani et al. 2016 [72]. Reproduced with permission of American Institute of Physics. (See online version for color figure).

The smallest clusters are not always the most active species, where, in the case of benzyl alcohol oxidation, larger particles are more active: “unexpectedly, Au_∼144 and Au_∼330 on hierarchically porous carbon exhibited significantly higher turnover frequency than Au₂₅ and Au₃₈” [73]. In fact, use of Au₉ cluster‐based catalysts, which gradually evolved toward larger sizes, allowed us to establish a particle size effect in solvent‐free aerobic oxidation of cyclohexene – we showed that catalytic activity appeared only after Au⁰ particles larger than 2 nm had formed [74]. Such particle growth was observed only in the presence of trace amounts of peroxide species in the cyclohexene (stabilizer‐free reagent), highlighting the sensitivity of such catalysts in terms of particle size stability to the reaction media. This study also demonstrated that conventional bright‐field TEM is unsuited for the imaging of clusters on SiO₂ NPs due to the poor contrast with the supported sub‐1 nm Au clusters. X‐ray photoelectron spectroscopy can be used to pinpoint the presence of clusters on supports since the characteristic electron binding energy is shifted to higher values in the case of ultrasmall clusters (regardless of their origin – made under UHV or chemically synthesized) [39].

Interestingly, even for the seemingly similar in nature Au_n(SG)_m clusters (n = 10, 15, 18, 22, 25, 29, 33, 39), only some of the clusters (n = 10, 15, 18, 25, 39) proved to be stable with respect to aggregation during activation, while others (n = 22, 29, 33) produced larger aggregates under the same activation conditions [75]. Although catalysts with “as‐deposited” clusters had higher catalytic activity than the pure support, the removal of ligands resulted in about fourfold increase in the H₂ production rate. Similarly, about fourfold higher activity of a Au₁₀‐cluster‐based catalyst was observed in comparison with catalysts made using large Au NPs.

In cases of relatively weak affinity between clusters and surface of support, fabrication of catalysts with reasonably high metal loadings may require a special approach of adding a non‐solubilizing (for cluster) solvent to the deposition mixture. For example, in the case of deposition of the Au₉ on WO₃ for loadings greater than 0.1 wt%, n‐hexane was slowly added to the mixture of gold cluster and WO₃ in CH₂Cl₂ to ensure cluster deposition [76].

A final contemporary example is the use of cleverly designed Au₂₅‐loaded BaLa₄Ti₄O₁₅ water splitting photocatalyst with the cluster‐based active sites protected by chromium oxide shell for enhanced activity and stability [77]. The chromium oxide shell is impermeable to O₂ but permeable to H⁺, thus allowing the distinction of active sites for the evolutions of H₂ (by photoelectron reduction) and O₂ (by photohole oxidation). This resulted in a 19‐fold improvement in performance and excellent longevity of the catalyst due to the prevention of gold cluster sintering.

Mixed metal clusters offer the advantages of synergistic action and cooperativity by two different metals [53]; a recent highly comprehensive review covers the advances in the use of heterometallic cluster‐based catalysts [78]. As an example to illustrate that “each atom counts,” a series of chemically synthesized metal cluster catalysts were made using a family of clusters differing by an atom or two from each other – Ru₆Sn, Ru₅Pt, Ru₅PtGe, and Ru₅PtSn [36]. The obtained materials were tested (Figure 5.7) for the single‐step conversion of dimethyl terephthalate (DMT) into cyclohexanedimethanol (CHDM). The Ru₅PtSn‐based catalyst demonstrated superior activity, as evidenced by the highest conversion of the tested catalysts, as well as excellent selectivity toward the desired product CHDM. The only substantial by‐product is dimethyl hexahydroterephthalate (DMHT), which is an intermediate of the hydrogenation pathway. Further optimization of the catalyst design and catalytic test conditions could result in even better performance. For a detailed and critical update on specific examples of a wide range of cluster‐based catalysts and their performance in various catalytic processes, curious readers are referred to a monumental recent review by Liu and Corma [79].

Figure 5.7 Bar chart comparing the activity and selectivity of the Ru₅PtSn catalyst with those of bi‐ and trimetallic analogues for the hydrogenation of dimethyl terephthalate.

Source: Hungria et al. 2006 [36]. Reproduced with permission of John Wiley & Sons.