CHAPTER 1: Stabilizing Gold Nanoparticles by Solid Supports

Catalysis by nanostructured materials has attracted tremendous interest recently.^1–4  Nanostructured catalysts may have interesting catalytic properties associated with their small sizes and geometric/electronic structures. In particular, Haruta and co-workers found that gold nanoparticles finely dispersed on some metal oxide supports have excellent activities in low-temperature CO oxidation.^5–11  This finding has been followed by thousands of studies on supported gold catalysts and their catalytic applications in environmental catalysis and chemical synthesis.^12–26 

Gold nanoparticles may be synthesized via a traditional colloidal chemistry approach,^27–30  in which AuCl₄⁻ ions are reduced by sodium citrate, tetrakis(hydroxymethyl)phosphonium chloride (THPC) or sodium borohydride (NaBH₄), and the formed gold nanoparticles can be stabilized by polyvinylalcohol (PVA), polyvinylpyrrolidone (PVP), polydiallyldimethylammonium chloride (PDDA), or cetyltrimethylammonium bromide (CTAB).^13,31  These gold nanoparticles (colloids) can be used either directly as catalysts in liquid phase or deposited onto solid supports.

Alternatively, gold nanoparticles can be formed on solid supports by loading a gold precursor (usually a gold salt or complex) onto solid supports followed by reduction or calcination.¹³  During the reduction or calcination process, the gold cations are reduced into gold atoms that aggregate into gold nanoparticles. The extent of agglomeration during this stage depends on many factors such as the ambient, temperature, and duration of the process as well as the nature of solid supports.

Supported metal catalysts are usually composed of metal nanoparticles and solid supports. Solid supports may provide a platform for dispersing and stabilizing gold nanoparticles so as to expose more surface gold atoms to the reactants, thus increasing catalytic activity. They may tune the oxidation state of gold by charge transferring or by mediating the reducing degree of gold precursors upon calcination or reduction. Some supports may undergo phase transformation or structural collapse under high temperatures, thus aggravating the sintering of gold nanoparticles on these supports or leading to the encapsulation of gold nanoparticles by these supports. Figure 1.1(c) shows a schematic diagram illustrating the phase transformation of a support at high temperatures.³²  Besides the facets related to metal–support interactions mentioned above, solid supports may participate in catalysis by adsorbing and activating reactants as well as supplying active oxygen. They may also, of course, strongly adsorb some reaction intermediates or products, leading to catalyst deactivation.

Although solid supports can disperse gold nanoparticles, the sintering of gold nanoparticles at elevated temperatures is often inevitable because of their low melting points and high surface free energies.^13,33  Figure 1.1(a) and Figure 1.1(b) show two models (crystalline migration and atom migration) proposed for the sintering of metal nanoparticles on supports.³²  In the first model, entire metal crystallites migrate, collide, and coalesce on the support surface. In the second model, metal atoms migrate from one crystallite to another via the surface or the gas phase, making big crystallite bigger and small crystallites smaller. The phase transformation or structural collapse of supports under elevated temperatures, as shown in Figure 1.1(c), may exacerbate the sintering or encapsulation of gold nanoparticles.³² 

Because the catalytic activities of supported gold catalysts often decrease sharply as the gold nanoparticle agglomerate under elevated temperatures but a high temperatures is often encountered during the calcination, operation, and regeneration of catalysts,³⁴  it is necessary to enhance the thermal stability of supported gold nanoparticles. This can be achieved by improving synthesis details, e.g., thoroughly washing away residual chloride that may facilitate sintering.^35,36  However, a more versatile way is to tune the structural environment surrounding gold nanoparticles, e.g., by strengthening the metal–support interaction and by designing sturdy inorganic shells that encapsulate gold nanoparticles.^19–21,33  These strategies rely on synthesis and modification of catalytic materials.

Most of the publications relevant to gold catalysis deal with the conventional synthesis, characterization, and applications of supported gold nanoparticles, as well as the elucidation of the nature of active sites and reaction mechanisms.^13,16,26  Only a small portion of publications have addressed the thermal stability and stabilization of gold nanoparticles on solid supports. The Dai group at the US Oak Ridge National Laboratory has been interested in designing new-structured gold nanocatalysts with enhanced properties, including catalytic activity, stability on stream, and thermal stability.^37–70  It is known from these studies that catalytic performance and thermal stability of supported gold catalysts depend critically on their composition and catalyst structure. Below we first summarize some recent advances in the stabilization of gold nanoparticles by solid supports, and then furnish our perspectives on future development.

Supported gold catalysts are usually prepared by loading gold onto supports (e.g., TiO₂, ZrO₂, SiO₂, Al₂O₃, Fe₂O₃, and CeO₂) via impregnation, deposition–precipitation, co-precipitation, and colloidal deposition.¹³  When supported gold catalysts are prepared by impregnation, deposition–precipitation, or co-precipitation methods, the gold precursors (gold cations) reduce to metallic gold atoms that migrate and aggregate into gold nanoparticles upon reduction or calcination. When gold catalysts are prepared by loading pre-formed gold nanoparticles (colloids) onto supports, it is not necessary to reduce the metallic gold nanoparticles again, but a calcination or pre-treatment step may still be needed to remove organic capping agents that may influence catalytic activity and to enhance the metal–support interaction.

Regardless of preparation methods, the formed supported gold catalysts usually have simple metal–support interfaces (e.g., Au-TiO₂). The sintering of gold nanoparticles on these neat metal oxide supports is a common problem under elevated temperatures. Attempts have been made to build up more complex interfaces for enhancing the thermal stability. For instance, Figure 1.2 illustrates the structural feature of Au/TiO₂/SiO₂ (i.e., gold nanoparticles supported on TiO₂-modified SiO₂ support), highlighting the additive–support (TiO₂-SiO₂) and metal–additive (Au-TiO₂) interfaces in addition to the metal–support (Au-SiO₂) interface.⁷¹  The presence of additional interfaces or complex structures may mitigate the sintering of gold nanoparticles due to the enhanced metal–support interaction.

Au/SiO₂ catalysts are usually not very active for CO oxidation and gold nanoparticles on SiO₂ can sinter easily, unless the preparation method is carefully chosen.^48,49,72  To enhance the thermal stability, one idea is to modify the SiO₂ support by another metal oxide before loading gold. For instance, Tai and co-workers developed Au/TiO₂/SiO₂ catalysts.⁷³  SiO₂ wet-gel was prepared by the hydrolysis of Si(OCH₃)₄ in the presence of NH₄OH, and was subsequently soaked in a toluene solution of Ti(iso-OC₃H₇)₄. Dodecanethiol-capped gold nanoparticles (2.1 nm) were then deposited onto TiO₂/SiO₂. For comparison, a TiO₂ support was prepared using Ti(iso-OC₃H₇)₄ as the precursor, and was used to load gold nanoparticles. As shown in Figure 1.3,⁷³  gold nanoparticles in Au/TiO₂/SiO₂ were still small (average diameter 2.2 nm) after calcination in air at 400 °C, whereas those in Au/TiO₂ grew obviously (average diameter 4.0 nm). Although the authors did not show the sintering behavior of gold nanoparticles on a neat SiO₂ support, this study nicely showed that gold nanoparticles exhibit high thermal stability on TiO₂-modified SiO₂ gel.

In another work, Yan et al. developed Au/TiO₂/mesoporous SiO₂.³⁹  The mesoporous SiO₂ (SBA-15) surface was functionalized by amorphous TiO₂ via a surface–sol-gel method, using Ti(OC₄H₉)₄ as the precursor. Gold was then loaded onto the support via a deposition–precipitation method. For comparison, Au/P25 TiO₂ was also prepared. Here the ‘P25 TiO₂’ refers to a commercial TiO₂ furnished by Degussa. Although the as-synthesized Au/P25 TiO₂ showed high activity in CO oxidation when the reaction temperature was below −20 °C, the Au/P25 TiO₂ calcined at 300 °C was much less active due to the aggregation of gold nanoparticles. For comparison, the activities of the as-synthesized and 300 °C calcined Au/TiO₂/mesoporous SiO₂ were similar, due to the preservation of small gold nanoparticles at 300 °C.³⁹  The authors additionally showed that it was difficult to load gold onto mesoporous SiO₂ via deposition–precipitation due to the low isoelectric point of SiO₂, and the obtained gold nanoparticles were usually big.

Above, we have highlighted two examples for the pre-modification of SiO₂ supports by TiO₂ before loading gold. The presence of TiO₂ species not only stabilizes gold nanoparticles on the modified supports, but also increases the isoelectric point of supports (note that the isoelectric point of TiO₂ is higher than that of SiO₂), thus increasing the gold loading when the deposition–precipitation method is used to load gold. In addition, the additional Au-TiO₂ interface leads to high activity in CO oxidation.^74–81  SiO₂ supports can also be modified by other metal oxides (e.g., CoO_x,⁸²  ZnO,⁸³  CeO₂,^83,84  CuO⁶⁹ ) to increase the dispersion of gold nanoparticles on supports. The role of these metal oxide additives is similar to that of the TiO₂ additive mentioned above.

Au/TiO₂ is the most studied gold catalyst. If prepared properly, it should be active for low-temperature CO oxidation, but the sintering of gold nanoparticles on TiO₂ is still a problem. Yan et al. developed a new catalyst, Au/Al₂O₃/TiO₂, for CO oxidation.⁴⁴  First, the P25 TiO₂ support was modified by amorphous Al₂O₃ via surface–sol-gel processing of Al(sec-OC₄H₉)₃ followed by controlled hydrolysis. Gold was then loaded onto Al₂O₃/TiO₂ via deposition–precipitation using HAuCl₄ as the precursor. Interestingly, the gold nanoparticles on the Al₂O₃/TiO₂ support showed excellent thermal stability upon aging at 500 °C, whereas gold nanoparticles on neat TiO₂ sintered significantly.⁴⁴ 

Ma et al. subsequently prepared Au/M_xO_y/TiO₂ catalysts.⁵¹  In the preparation, the surface–sol-gel method was not used. Instead, a traditional impregnation method was used to load metal nitrates onto TiO₂. The metal nitrates/TiO₂ were calcined to form M_xO_y/TiO₂ supports, and gold was loaded via deposition–precipitation.⁵¹  The use of impregnation instead of the surface–sol-gel method to functionalize TiO₂ support was based on several considerations. First, it was thought that a catalyst (e.g., Au/Al₂O₃/TiO₂) prepared involving decomposing a metal nitrate (e.g., Al(NO₃)₃) on TiO₂ support followed by loading gold should exhibit a similar performance compared with its counterpart prepared using a surface–sol-gel method. Second, the nitrate decomposition method is less demanding and therefore suitable for large-scale preparation. Third, the synthesis via the surface–sol-gel method is constrained by the availability, expensiveness, and storage stability of metal alkoxide precursors.

It was found that Au/M_xO_y/TiO₂ (M=Ca, Ni, Zn, Ga, Y, Zr, La, Pr, Nd, Sm, Eu, Gd, Dy, Ho, Er, Yb) retained significant activity in CO oxidation even after thermal treatment at 500 °C.⁵¹  This was explained by the enhanced thermal stability of gold nanoparticles caused by the surface modification of TiO₂ support by certain metal oxides, as demonstrated by X-ray diffraction (XRD) and transmission electron microscopy (TEM) data.^51,63  In addition, it was speculated that the presence of metal oxide additives with redox properties adjacent to gold nanoparticles may change the oxidation state of gold and the redox property of the support. More experiments should be performed to better understand the promotional effects of the metal oxide additives on TiO₂ support.^85–88 

The empirical observations mentioned above have led to further fundamental research. Liu and co-workers shed light on the stabilizing effect of the amorphous Al₂O₃ by means of density functional theory (DFT) calculations.⁸⁹  Figure 1.4 shows the models proposed for the local structures of a gold atom or a two-layer gold strip on an Al₂O₃/TiO₂ surface.⁸⁹  These models were used for their theoretical calculations. The authors found that the binding of gold on Al₂O₃/TiO₂ was much stronger than that on TiO₂, and the stronger binding was valid for some other metals (i.e., Ag, Cu, Pt, Pd, Ir) on Al₂O₃/TiO₂ compared with these metals on TiO₂. This finding explained the enhanced thermal stability of gold nanoparticles on Al₂O₃/TiO₂ versus TiO₂. A further idea would be to extend this DFT approach to the Au/M_xO_y/TiO₂ system to obtain deeper insights.

The pre-modification strategy is not limited to the modification of oxide supports by another metal oxide. For instance, TiO₂ was treated by an aqueous H₃PO₄ solution before loading gold.⁵²  That treatment was found to enhance the thermal stability of gold nanoparticles. A similar stabilization effect was found with Au/H₃PO₄-Al₂O₃⁹⁰  and Ag/H₃PO₄-TiO₂.⁹¹  However, the presence of phosphorus species also suppressed the catalytic activities. In addition, it is not clear why the H₃PO₄ treatment can help stabilize gold or silver nanoparticles.

Supported gold catalysts are usually prepared by loading gold onto neat metal oxide supports such as TiO₂. However, few attempts have been made to additionally modify supported gold catalysts when gold was already loaded onto solid supports.^92–95  Ma et al. developed PO₄³⁻/Au/TiO₂ catalysts.⁵²  In the synthesis, Au/TiO₂ was prepared by deposition–precipitation, reduced at 150 °C in H₂–Ar, and soaked in a diluted H₃PO₄ solution, followed by washing and drying. This treatment can stabilize gold nanoparticles on TiO₂, as evidenced by the fact that PO₄³⁻/Au/TiO₂ calcined at 500 °C still had small gold nanoparticles and appreciable activity in CO oxidation at room temperature. However, overloaded phosphate ions may lead to low catalytic activity due to the blockage of active sites. In addition, it is not clear why the H₃PO₄ treatment can stabilize gold nanoparticles.

Zhu et al. developed SiO₂/Au/TiO₂ catalysts.⁵⁰  In the synthesis, pre-formed Au/TiO₂ was treated in Si(OCH₃)₄, H₂N(CH₂)₃Si(OC₂H₅)₃, or ((CH₃)₃CO)₃SiOH solution, followed by calcination (Figure 1.5).⁵⁰  The silicon precursors decomposed to amorphous SiO₂ near gold nanoparticles upon calcination, therefore mitigating the sintering of gold nanoparticles and maintaining high catalytic activity. Although XRD and TEM data demonstrated the maintenance of small gold nanoparticles upon thermal aging,^19,50  there is no direct information on the location and morphology of the amorphous SiO₂ additive.

Ma et al. developed SiO₂/Au/TiO₂ catalysts using an alternative method.⁵⁶  In the synthesis, SiO₂ was loaded onto Au/TiO₂ via atomic layer deposition (ALD). Si(OCH₃)₄ vapor was used as the precursor to the amorphous SiO₂ overlayer, and the SiO₂ content could be adjusted by varying the number of ALD cycles or Si(OCH₃)₄ exposures. High-resolution TEM images showed the presence of amorphous SiO₂ adjacent to the gold nanoparticles and TiO₂ support. The amorphous SiO₂ may mitigate the sintering of gold nanoparticles.⁵⁶  Although the ALD method is expected to provide more uniform coating of gold nanoparticles and TiO₂ support, it is more demanding in terms of the apparatus and control of the experimental conditions. In addition, the uniform and dense coating of gold nanoparticles makes the reactants less accessible to the active sites, thus leading to lower catalytic activities in CO oxidation.

The stabilization effect of SiO₂ coatings that physically isolate and encapsulate gold nanoparticles is understandable. But sometimes the loading of SiO₂ was too low to allow for complete coating, whereas the stabilization effect of the amorphous SiO₂ was still observed.⁵⁰  Rashkeev and co-workers conducted DFT calculations, and concluded that the deposition of SiO₂ on TiO₂ support may cause lattice-mismatch instabilities and lead to the formation of strong anchoring sites for gold nanoparticles, even if the coverage of SiO₂ is below monolayer.⁹⁶  Moreno and co-workers provided direct imaging evidence for the stabilization effect caused by the SiO₂ decoration attached to gold nanoparticles (Figure 1.6).⁹⁷  Note that the SiO₂ decoration was formed by the transformation of HS(CH₂)₃Si(CH₃)₃ modified gold nanoparticles supported on SiO₂ substrate but not by post-modification.

The structures of SiO₂/Au/TiO₂ highlighted above are relatively straightforward. Further attempts have been made to design catalysts with more complex architectures.^98,99  For instance, Yin and co-workers prepared core- satellite nanocomposite catalysts protected by a porous SiO₂ shell (Figure 1.7).⁹⁸  In the synthesis, a superparamagnetic Fe₃O₄ core was coated by a SiO₂ shell, and gold nanoparticles were loaded onto SiO₂/Fe₃O₄. The resulting material was coated with another layer of SiO₂ to fix the position of the gold nanoparticles, and a ‘surface-protected etching’ technique was used to make the outer SiO₂ shell porous. The SiO₂/Au/SiO₂/Fe₃O₄ catalyst was still active after running the catalytic reduction of 4-nitrophenol for several times, whereas Au/SiO₂/Fe₃O₄ without the porous SiO₂ shell suffered from agglomeration of gold nanoparticles and loss of activity during the reaction.

To put this work in perspective, it should be mentioned that the post-modification strategy was not limited to the modification of supported gold catalysts. Other modified catalysts reported include SiO₂/Pt/zeolite,¹⁰⁰  SiO₂/Pt/C,^101–104  SiO₂/Pt/Fe₂O₃,¹⁰⁵  SiO₂/Pt/SiO₂,^106,107  TiO₂/Pt/SiO₂,¹⁰⁸  Al₂O₃/Pd/SiO₂,¹⁰⁹  Al₂O₃/Pd/Al₂O₃/SiO₂,¹¹⁰  Al₂O₃/Pd/Al₂O₃,¹¹¹  and HfO_x/Pd/SiO₂.¹¹²  The thermal stability of metal nanoparticles in these catalysts has been significantly enhanced. The modifier of choice was SiO₂ due to the relative easiness of preparation and high thermal stability of SiO₂, but post-modification by metal oxides has not been demonstrated widely. It is expected that post-modification of supported metal catalysts by metal oxides (especially transition metal oxides) will bring more functionalities to the catalyst systems. However, the synthesis of metal oxide coatings is more difficult and the thermal stability of metal oxide coatings may not be that good.

Delicate materials synthesis has lead to new materials with interesting structures or morphologies, thus providing new opportunities for designing better catalysts. The dumbbell-like Au-Fe₃O₄ composite is such a material.^113–115  The dumbbell structure is composed of a small gold nanoparticle at one end and a bigger Fe₃O₄ nanoparticle at the other end. These two components interact with each other through an Au-Fe₃O₄ interface. Yin et al. prepared Au-Fe₃O₄ dumbbell nanoparticles via in situ decomposition of HAuCl₄ and Fe(CO)₅, followed by oxidation of the Fe component, and then dispersed the Au-Fe₃O₄ dumbbell nanoparticles on several supports via colloidal deposition (Figure 1.8).⁵⁹  Figure 1.9 shows the TEM images of unsupported and SiO₂-supported Au-Fe₃O₄ dumbells.⁵⁹  The presence of Fe₃O₄ attached to gold nanoparticles was proposed to be responsible for the high activity in CO oxidation and thermal stability of the resulting Au-Fe₃O₄/SiO₂ catalyst.

Zheng and co-workers studied supported Au-Fe₃O₄ dumbbell catalysts in more detail.¹¹⁶  Gold nanoparticles (6.7 nm) were epitaxially grown on Fe₃O₄ nanoparticles (15.2 nm) to form dumbbell particles, and the dumbbell particles were then dispersed onto TiO₂. For comparison, Au/TiO₂ was prepared by dispersing gold nanoparticles (6.7 nm) onto TiO₂. The thermal stability was studied by calcining these catalysts at 350, 450, and 550 °C and recording TEM images afterwards. The sintering of gold nanoparticles in Au-Fe₃O₄/TiO₂ was less obvious than that in Au/TiO₂.

To summarize, dumbbell-structured Au-Fe₃O₄ particles have been used to make supported gold catalysts. The advantage of the catalyst design is that active interfaces can be installed on sturdy supports, thus mitigating sintering and enhancing catalytic activity. In addition, the sizes of gold particles and Fe₃O₄ can be tuned, and the solid supports can be chosen according to desired needs. However, the types of catalysts are limited by the availability of dumbbell structures; many other dumbbell structures have not been developed. In addition, the local contact structures of the catalysts were not made clear in previous studies. High angle annular dark field (HAADF) imaging may provide more detailed information.

Gold-containing bimetallic particles have been used to prepare supported catalysts.^117–119  In most cases, the supported particles are assumed to be metallic before and after thermal treatments or catalytic reactions, especially considering that it is difficult to oxidize some noble metal components (e.g., Pt) with oxygen. However, the transformation of supported bimetallic particles under certain conditions is often inevitable.^{67,70,120–126}  In particular, some metals (e.g., Cu, Ni, Fe) can be oxidized at elevated temperatures. This situation has led to the development of catalysts with enhanced performance.

Zhou et al. developed Au-NiO/SiO₂ catalysts.^60,61  First, NiAu alloy nanoparticles were prepared via the reduction of mixed Ni and Au salts by butyl-lithium. The NiAu nanoparticles were then deposited onto SiO₂ support. Upon treating NiAu/SiO₂ in H₂ at 600 or 720 °C and then in O₂ at 300 °C, the catalyst transformed into Au-NiO/SiO₂, the working catalyst. The amorphous NiO was demonstrated to locate near gold nanoparticles, thus mitigating the sintering of gold nanoparticles. As shown in Figure 1.10, gold nanoparticles in Au/SiO₂ (a reference catalyst prepared by colloidal deposition) sintered obviously after thermal treatment, whereas those in Au-NiO/SiO₂ were quite small.⁶⁰  This in situ transformation strategy may prompt the development of new advanced catalysts with better chemical and structural properties.

So far, we have highlighted some gold catalysts with complex metal–support, metal–modifier, and modifier–support interactions. The thermal stability of gold nanoparticles can be enhanced in this way. The question now arises as whether gold catalysts with single metal–support interfaces can still have good thermal stability.

Li and co-workers designed mesoporous multicomponent nanocomposite colloidal spheres (MMNCSs) in four steps: (1) synthesis of gold nanoparticles and CeO₂ nanoparticles; (2) mixing of these nanoparticles in a solvent at a certain ratio; (3) assembly of the mixed colloidal nanoparticles according to an emulsion-based bottom-up self-assembly strategy; and (4) calcination at 200 °C to form mesoporous structures.¹²⁷  Figure 1.11 shows the N₂ adsorption–desorption isotherm, pore size distribution, and the corresponding TEM image of Au-CeO₂ MMNCs.¹²⁷  The resulting Au-CeO₂ composite catalyst showed higher activity and better thermal stability than a conventional Au/CeO₂ catalyst. Similar observations were obtained for Au-TiO₂ MMNCSs.¹²⁷ 

Murray and co-workers prepared Au-Fe₃O₄ binary nanoparticle superlattices (BNSLs) through self-assembly of gold and Fe₃O₄ nanocrystals at a liquid–air interface, and then transferred the BNSLs onto Si wafers.¹²⁸  Au-Fe₃O₄ BNSLs with different compositions and geometric arrangements can be fabricated in this way. In addition, the size of gold nanoparticles as well as the thickness of Au-Fe₃O₄ BNSLs on Si wafers can be selected. This type of catalysts provides an excellent model for studying the contribution of interfacial sites for catalysis as well as thermal stability of gold catalysts. The authors demonstrated a strategy of keeping gold nanocrystals separated to enhance the thermal stability of gold catalysts.

The thermal stability of gold nanoparticles may be improved by encapsulating gold nanoparticles in oxide shells. The inorganic shells can physically confine the gold nanoparticles, thus mitigating their sintering. In addition, they may provide interfacial sites active for catalysis. Normally there is one gold nanoparticle in one shell, not multiple gold nanoparticles in one shell. Depending on whether there is any space between gold nanoparticle and the shell, core-shell and yolk-shell structures can be distinguished.^129–133  These catalysts have metal–support interfaces without other additives, but the catalyst structures are different from conventional supported metal catalysts. To make the catalyst structure more complex, supported core-shell structures may be prepared,⁶⁶  and another component (e.g., a metal oxide) can be added into the oxide shell.¹³⁴ 

Yin and co-workers developed the Au@porous SiO₂ catalyst.¹³⁵  Au@SiO₂ core-shell particles were synthesized first via a modified sol-gel procedure using gold sol as the core and tetraethylorthosilicate as the silicon source for the formation of SiO₂ shells. The surface of SiO₂ shells was then functionalized by PVP, and etched by NaOH solution. The interior of the Au@SiO₂ was selectively etched to create porous structures, whereas gold nanoparticles were still encapsulated into the SiO₂ shells. The Au@porous SiO₂ catalyst demonstrated excellent stability in catalytic reduction of 4-nitrophenol. In contrast, uncoated gold nanoparticles aggregated immediately in the reaction mixture.

Schüth and co-workers developed a yolk-shell ‘Au,@ZrO₂’ catalyst.¹³⁶  Gold nanoparticles (15–17 nm) were prepared by sodium citrate reduction, and were coated with SiO₂ shells. The Au@SiO₂ spheres were then coated by ZrO₂ shells, and the SiO₂ in between the gold nanoparticles and the ZrO₂ shells was removed by NaOH solution. The ZrO₂ shells not only protected the gold cores, but also created Au-ZrO₂ interfaces active for CO oxidation. In addition, the thin (20 nm) porous ZrO₂ shells make the encapsulated gold nanoparticles accessible to the reactants.

Schüth and co-workers also developed a yolk-shell ‘Au,@ZrO₂’ catalyst with smaller gold particle sizes (Figure 1.12).¹³⁷  Gold nanoparticles (15 nm) were coated by SiO₂ shells. The size of the encapsulated gold cores was decreased to 10 nm by treating Au@SiO₂ core-shell particles with NaCN solution, during which some metallic gold atoms reacted with NaCN to form soluble [Au(CN)₂]⁻. The material was then coated by ZrO₂ shells (thickness 7–10 nm), and the SiO₂ shells in between the gold cores and ZrO₂ shells were leached by aqueous NaOH. The resulting catalyst could withstand calcination at 800 °C, with no loss of catalytic activity in CO oxidation. Nevertheless, it was difficult to reduce the size of gold particles by NaCN further, and an alternative aqua regia treatment allowed fine-tuning of gold particle sizes down to 5 nm but the catalyst was deactivated by residual chlorine.

Zheng and co-workers prepared Au@hollow mesoporous-ZrO₂ (Au@hm-ZrO₂).¹³⁸  Hydrophobic dodecanethiol-capped gold nanoparticles (6.3 nm) underwent ligand exchange with mercaptoundecanoic acid, and were treated with ammonia to be water soluble. The water-soluble gold nanoparticles were then treated by Si(OC₂H₅)₄ to form SiO₂ shells (120 nm). The Au@SiO₂ spheres were coated with ZrO₂ via the hydrolysis of Zr(OC₄H₉)₄ in the presence of Brij 30 surfactant and calcined at 850 °C. The SiO₂ shells were removed by leaching with NaOH solution. The catalyst was active for the reduction of 4-nitrophenol, and showed excellent thermal stability.

Tang and co-workers prepared Au@CeO₂ via a facile ‘self-templating’ method.¹³⁹  HAuCl₄ and CeCl₃ were mixed with aqueous glucose and urea, and the solution was subjected to hydrothermal treatment to generate a dark brown solution. The suspension was then calcined at 600 °C to obtain Au@CeO₂. The role of glucose is to reduce cationic gold and to form amorphous carbon submicrospheres encapsulating the formed gold nanoparticles and adsorbing Ce³⁺ ions. The final calcination step removes the carbon submicrospheres and forms CeO₂ shells. Au@CeO₂ showed much higher activity than Au/CeO₂ due to the better stabilization of gold nanoparticles and more contact areas between gold nanoparticles and CeO₂.

To summarize, gold containing core-shell structures have been synthesized and enhanced thermal stability has been demonstrated. The studies mentioned above aimed at demonstrating protocols, and the structures of the catalysts developed had many versions, but there is still no universal method to prepare a variety of core-shell or yolk-shell catalysts and it was not discussed in the literature which catalyst structure is ideal. In addition, the synthesis often involves leaching, which is not good for environmental and safety reasons. Convenient ways for large-scale production of catalysts are still needed.

Laursen et al. prepared zeolite-encapsulated gold nanoparticles.¹⁴⁰  This invention involves the immobilization of gold nanoparticles in an amorphous silica matrix, followed by hydrothermal treatment to crystallize the amorphous silica to a zeolite (silicalite-1) phase (Figure 1.13).¹⁴⁰  Most of the gold nanoparticles were trapped in the zeolite crystal. They exhibited high thermal stability and are resistant to acid leaching. On the other hand, some gold nanoparticles on the external surface of the zeolite crystal can still agglomerate upon heating, and they can be leached by aqua regia easily. In a subsequent work,¹⁴¹  the authors developed a two-step method in which gold nanoparticles were immobilized in amorphous SiO₂ followed by hydrothermal treatment to create gold-embedded silicalite-1 seeds. The resulting silicalite-1 seeds were then suspended in a ZSM-5 growth medium containing aluminum to form gold-embedded ZSM-5.

Here we have summarized some strategies and examples for the stabilization of gold nanoparticles on solid supports. These strategies include: (1) pre-modification of supports before loading gold; (2) post-modification of supported gold catalysts; (3) dispersion of Au-Fe₃O₄ dumbbell structures on supports; (4) in situ transformation of supported alloy catalysts into metal oxide modified supported gold catalysts; and (5) development of Au@oxide core-shell or yolk-shell structures. In these cases, the interfacial structures have been modified. The presence of additional additives may enhance the interaction between metal and support, thus mitigating the sintering of gold nanoparticles. These oxide additives (especially transition metal oxides) may also tune the oxidation state of gold and tune the redox properties of the support, thus enhancing catalytic activity. Alternatively, the encapsulation of gold nanoparticles by oxide shells may avoid the sintering of gold nanoparticles. Considering that the catalytic activities of supported gold catalysts depend critically on the size of the gold nanoparticles, the stabilization of gold nanoparticles is meaningful for maintaining their performance under high temperature conditions.

There are still some deficiencies in previous research, which may provide new opportunities for further investigation.

1. XRD and TEM data can provide direct evidence for the stabilization of gold nanoparticles (as judged by the sizes of gold nanoparticles), but they are not enough to pin down the location and morphologies of the additives that stabilize gold nanoparticles. High-resolution TEM and HAADF images may help address this issue and provide direct evidence.^{64,66,124,142}  For instance, Figure 1.14 shows an appealing high-resolution TEM image of the microscopic structure of Au/IrO₂/TiO₂ obtained from oxidative transformation of Au-Ir/TiO₂ at 400 °C.¹⁴² 

2. First principle DFT calculations may be able to tell us why the presence of certain additives may enhance the thermal stability of gold nanoparticles,^89,96,143  and to predict the compositions and structures of better catalysts. It is also worthwhile to use DFT methods to understand the nature of active sites and reaction mechanisms on new catalysts,¹⁴⁴  although care should be taken when interpreting the results because of different conditions in different systems.

3. Surface science studies, based on model catalyst systems, could be carried out to better understand the sintering behavior and support effect.^145–147 

4. The synthesis methodologies highlighted above have not been demonstrated systematically, and most of the studies rely on simple test reactions such as CO oxidation and reduction of 4-nitrophenol. It is worthwhile to synthesize a wide variety of catalysts following these synthesis methodologies, and to explore the applications of these catalysts in a wide variety of catalytic reactions.

Z. Ma thanks the National Natural Science Foundation of China (Grant Nos. 21007011 and 21177028), the PhD programs foundation of the Ministry of Education in China (Grant No. 20100071120012), and the overseas returnees start-up research fund of the Ministry of Education in China for the financial support. S. Dai thanks the Office of Basic Energy Sciences, US Department of Energy for financial support. Oak Ridge National Laboratory is managed by UT-Battelle, LLC for the US DOE under Contract DE-AC05-00OR22725.