Au/CeO₂ Catalysts: Structure and CO Oxidation Activity

Abstract

In this comprehensive review, the main aspects of using Au/CeO₂ catalysts in oxidation reactions are considered. The influence of the preparation methods and synthetic parameters, as well as the characteristics of the ceria support (presence of doping cations, oxygen vacancies concentration, surface area, redox properties, etc.) in the dispersion and chemical state of gold are revised. The proposed review provides a detailed analysis of the literature data concerning the state of the art and the applications of gold–ceria systems in oxidation reactions.

1. Introduction

Carbon monoxide (CO) is one of the extended pollutants of our environment, thus requiring prevention and control to insure the adequate protection of public health [1]. Carbon monoxide is hardly detectable, colorless, tasteless, odorless, non-irritating very toxic gas for humans and animals because of its high affinity to hemoglobin and the cell’s oxygen transport blocking. Exposure to CO must never exceed 25 ppm in 8 h or 50 ppm in 4 h because, above these limits, important detrimental effects are observed up to the lethal concentrations of around 650–700 ppm [2]. Moreover, carbon monoxide is also a precursor of ground-level ozone, which can generate serious respiratory problems. All of these environmental issues render CO oxidation reaction the preferred option for CO depletion in polluted atmospheres.

In 1987, almost 30 years ago, the discovery of Haruta and co-workers revolutionized the concept of CO oxidation for environmental proposes [3]. They demonstrated that very small gold nanoparticles are extremely active for carbon monoxide oxidation at sub-ambient temperatures. This achievement provoked a great scientific interest converting gold in the most popular metal for the catalytic oxidation of CO [4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28]. Figure 1 shows the growing trend in a number of publications per year involving CO oxidation using gold-based catalysts. A several fold increment in the last decade of the publications with search strings “CO oxidation” and “gold-based catalysts” was observed indicating the importance of this catalytic process in heterogeneous catalysis.

A number of other possible applications for gold-based catalysts able to perform CO oxidation at room temperature appear such as carbon dioxide lasers, gas sensors, respirators for protecting firefighters, and miners from CO poisoning, air-cleaning devices, etc. [29].

Out of the environment protection issue, the CO abatement is also a key reaction in many industrial processes. The water gas shift reaction (CO + H₂O ↔ CO₂ + H₂) is one of these reactions. Despite the fact that the reaction was described at the end of the 18th century, its main applicability was not found before the 1920s, when the Haber–Bosh process for ammonia synthesis called for pure hydrogen production. From this moment on, the water gas shift (WGS) reaction gained importance and became an inseparable part of various industrial processes, such as ammonia and methanol production, hydrogen supplies for the petrochemical industry, Fisher–Tropsch reactions, and numerous reforming reactions.

Presently, a renewed interest in the WGS reaction emerges associated to the hydrogen fuel cell (FC) technologies. Hydrogen as an energy carrier is usually produced from hydrogen-containing molecules, such as hydrocarbons, ammonia, or chemical hydrides [30]. The best short-term option seems to be the hydrogen production via hydrocarbons reforming. However, the resulting products are a mixture mainly constituted by hydrogen and carbon oxides [31]. The presence of CO in the hydrogen flows, even in small amounts at the ppm level, inactivates the fuel cell electrodes. Therefore, to avoid fuel cell anode poisoning, several fuel processing steps, including CO removal by means of a WGS and CO preferential oxidation reactions (PROX) or methanation must be applied. The development of efficient catalysts for these processes is fundamental to guarantee the success of the so-called “hydrogen economy”.

Among all the reported catalysts, gold-based materials have been broadly employed for the WGS reaction showing promising results in many cases [32,33,34]. Some of the main benefits of using gold-based catalysts include good activity in the low-temperature range, the absence of pre-treatment, and pyrophoricity. Nevertheless, these systems also show drawbacks including extreme sensibility to the preparation procedure (existence or not of gold nanoparticles), the importance of a support nature (especially development of an active gold–support interface), and relatively low catalyst stability attributed to support deactivation. However, when these parameters are carefully controlled and the support is prudently selected, gold-based materials are outstanding for the WGS reaction.

The production of hydrogen pure enough to feed polymer electrolyte membrane fuel cell (PEMFC) requires further clean-up processes after the WGS unit in order to reduce the CO concentration to compatible levels, typically 10–50 ppm depending on the cell anode [10]. The preferential CO oxidation in rich-H₂ streams (Scheme 1) has been proposed as cheap and efficient alternative for the elimination of the final traces of CO [35,36,37,38]. The oxidation is rapid and responds quickly to changes in operating conditions, but it is essential to choose catalysts and operating conditions that minimize the oxidation of hydrogen. In this regard, control of temperature is found to be particularly important [10].

Attending the above-mentioned scheme, the PROX reaction requires selective catalysts able to abate CO without oxidizing H₂, an undesirable side reaction. In other words, high CO oxidation activities coupled with low hydrogen ones are essential requirements for the PROX catalysts [38,39]. Various studies have confirmed that the rate of CO oxidation over supported Au catalysts exceeds that of H₂ oxidation [40,41,42,43]. Moreover, gold-based catalysts show extraordinarily high activity in the low temperature range, which is appropriate for fuel cell applications. The gold-based catalysts often result after combining this noble metal with support with the intrinsic ability for the CO abatement. Therefore, the final result is the combination of both the ability of the gold nanoparticles and that of the support for the CO oxidation. However, cooperative effects have also been observed between both components of the catalyst, promoting a catalytic performance that is something more that the mere sum of their separate abilities to oxidize CO.

Within the studied supports, usually metal oxides, there are reports about the use of MnO_x [44], Fe₂O₃ [45], and Co₂O₃ [46], among others. Their intrinsic ability for the CO oxidation is due to some oxygen species that are especially activated through electronic transfer between them and the metallic cations, allowing a successful activation of the CO and O₂ molecules for CO₂ production. For that reason, when these materials are employed as catalytic supports, they are known as active supports. CeO₂ is one of the most studied active supports due to its redox behavior achieved by the easy and reversible transformation of Ce⁴⁺ cations into Ce³⁺ cations. This redox property strongly depends on the mechanism followed by this solid for the oxygen vacancy formation. Oxygen vacancies are punctual defects specially established at the surface of the material [47,48] and are responsible for the ability of the CeO₂ to exchange oxygen species with the environment. In this sense, CeO₂ can be considered as an oxygen buffer due to its oxygen storage capacity [49]. It has been widely reported that the doping of the CeO₂ structure by the addition of one or more cations with a different nature, allows achieving mixed systems with improved oxygen exchange ability. Many different cations have been successfully used: Zr [11,50], Zn [17,51], Fe [52,53], Co [54], La [55,56], Eu [57], Gd [56], and Sm [58], among others. Taking into consideration the wide range of possible combinations between ceria and different cations, the different synthesis procedures that can be applied to obtain the mixed oxides, and the strict control of the variables required during the synthesis procedures, studies in the near future will be devoted to understanding the structural properties achieved for the obtained solids and to correlating them with their catalytic activity in the CO oxidation reactions, despite the large number of yet published papers in this topic many works. The transition metals are interesting and potential modifying agents of the CeO₂ lattice because they have d electrons that can interact with the unoccupied energy levels of the cerium cations [59], promoting the ionic mobility into the mixed structure.

Thus, considering the intrinsic properties of cerium oxide and that the deposition of small amounts of gold in the surface of this material allows for the obtainment of much more active systems in the CO oxidation reactions [20,23,25,60], including total oxidation, WGS, and PROX, this review presents Au/CeO₂ catalysts as a case study and develops relevant aspects regarding the preparation, characterization, and catalytic activity evaluation of these materials, reported in the literature.

2. Preparation of Au/CeO₂ Catalysts

The last 20 years saw the pursuit of method perfection for preparing gold-based systems. Gold nanoparticles production, supported or not on ceria, is very sensitive to the preparation parameters, such as pH, concentration, use of bases, temperature, or even the presence of light. Most information regarding the advances on gold nanoparticle production and the subsequent supporting process is summarized in book chapters [40,61] and various reviews [5,8,29,32,60,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93]. A well-established paradigm in the gold catalysis field is that the smaller the gold particles size, the higher their activity in numerous oxidation reactions. The deposition–precipitation method using either urea or bases such as sodium or potassium hydroxide or carbonate is the most preferred method to support gold nanoparticles on mineral supports [70,94,95,96]. The use of direct anionic exchange including an ammonia-washing step is also well represented [97,98,99]. Recent reports devote a substantial part of the fairly old colloidal route [100] with all its variations mainly concerning the reduction agents. This latter route starts to become important with the development of carbon-supported gold nanoparticles for selective oxidations in liquid phase [101,102,103]. The modification of the colloidal methods arises from the choice of the reduction agent or for the moment of its application. Besides these mostly employed conventional methods of preparation of Au/CeO₂ catalysts, new reports presenting alternative routes deserve interest. We review some of them without the sake of exhaustiveness. The approach proposed by Lakshmanan et al. [104] is one of these examples. In their proposal, glycerol acts as a reducing agent of the already preformed Au/CeO₂ precipitate. The particularity of this work is that they used glycerol not only as a gold reducer but also as a participant in the catalytic reaction. The catalyst precursor to glycerol weight ratio affects the average gold particle size and the product distribution in the selective oxidation of glycerol to lactic acid.

Another interesting approach is the use of supercritical CO₂ as an anti-solvent for the preparation of Au/CeO₂ catalyst [105,106]. The catalyst precipitates on adding the precursor solution to supercritical CO₂. This Au/CeO₂ catalyst demonstrates an enhanced activity in comparison to conventionally prepared systems, this behavior being assigned to the variation of the nucleation velocity during the catalyst preparation. Very recently, a novel method based on the photo-thermal effect generated from localized surface gold plasmon resonance (LSPR) has been proposed for preparing Au/CeO₂ core–shell structures [107]. The method benefits from the heat generated by LSPR on Au nanoparticles, which induces the polymerization reaction of the ceria precursor sol and results in ceria formation just on the Au nanoparticles surface. Mitsudome et al. [108] obtain similar Au/CeO₂ core shell structures by mixing reverse micelle solutions of Au(III), Ce(III), and NaOH. In this mixture, a redox reaction between Au(0) and Ce(IV) occurs yielding the Au/CeO₂ core shell structure. The interfacial cooperative effect of the structure results in maximal selectivity to alkenes in the reaction of selective semi-hydrogenation of various alkynes.

Bera and Hedge [109] proposed the solution combustion method to prepare Au/CeO₂ catalysts. In this method, Au remains in ionic form until a thermal treatment at 800 °C is applied. Yang et al. [110] proposed the vapor phase synthesis of Au/CeO₂ catalyst. The method consisted of vaporizing the metal target by using a second harmonic generation neodymium-doped yttrium aluminium garnet (Nd–YAG) laser on the cold top plate where the oxide support is placed. This preparation method appears to be very versatile, i.e., a variety of efficient catalysts can be designed by using simple and flexible control of different synthetic parameters.

The decomposition of a single-source precursor was also reported as a versatile method for gold catalysts preparation [111]. The method normally uses the double complex salts (DCS) of transition metals, ionic compounds for which cation and anion are coordination compounds, and appears well suited for the synthesis of bimetallic catalysts. The bimetallic catalysts may be formed either by sequential impregnation of the monometallic salts onto the support or by the impregnation of an organic solution of a DCS. This method allows for the control of the bimetallic phase formed and hence of the catalyst activity. Another proposed method to control Au particle sizes consists in the deposition of thiol protected Au_n(SR)_m nanoclusters, where n and m are the number of Au atoms and SR the thiolate ligands, respectively, where R is an alkyl group [112]. The main difference of this method with the traditional deposition precipitation one resides in the precise size of the gold thiolate clusters resulting in monomodal particle size distribution in the final material. In general, particle sizes below 2 nm, or having less than 200 atoms can be conceived. Although the activity of these systems was not satisfactory, their possible use as model catalyst for mechanistic studies could be easily envisaged.

Besides the above-mentioned preparation methods, a common strategy, very popular in the last decade, is to disperse Au/CeO₂ catalysts on high surface area supports. Generally, this strategy leads to very performant materials either with enhanced Au dispersion on the support surface or a higher surface-to-bulk ratio and therefore complete Au/CeO₂ interface exposure to the reactants. The latter effect is observed when the size of both gold and CeO₂ nanoparticles is similar. Ramírez Reina et al. [14,113] performed an extensive study over the deposition of bare or transition metal-doped Au/CeO₂ catalyst over alumina. On the other hand, Escamilla-Perea et al. [114] proposed the use of SBA-15, and Hernandez et al. [115] the use of wormhole hexagonal mesoporous silica. The neutral surfactant templating route was used to prepare the latter, and its modification with cerium oxide was performed either by direct synthesis or by impregnation methods. Gold was finally deposited via deposition–precipitation.

Carabineiro and coworkers have synthesized cerium oxides by an exotemplating procedure [116], using activated carbon or carbon xerogel as templates. The resulting Au/CeO₂ catalyst was produced via the double impregnation method (a modification of the deposition–precipitation method). Furthermore, they have applied the solvothermal method for obtaining ultrafine cubic CeO₂, using methanol as a solvent and studying other variables during the synthesis [117]. The obtained cerium oxides were used as supports for the deposition of gold through the double impregnation method and these materials were compared with a reference gold catalyst synthesized with a commercial ceria. Although the use of the solvothermal methodology resulted in the slight decrease in the catalytic performance of the gold catalysts during the CO oxidation reaction due to the sintering of the gold nanoparticles, for these materials, a superior contribution of the oxidic gold species was observed; for the reference catalyst, the gold species were in the metallic state. This result remarks that the structural properties of the materials are strongly influenced by the different variables involved in the synthesis process. In this sense, Carabineiro et al. [118] have also studied the effect of the nature of the gold precursors used for obtaining Au/CeO₂ catalysts, observing the promotion of gold sintering when chlorine remains in the solids.

Moreover, Liu et al. [119] proposed three-dimensional ordered macroporous (3DOM) Au/CeO₂–Co₃O₄ system with well-defined macroporous skeleton and mesoporous walls created via a precursor thermal decomposition-assisted colloidal crystal templating method. They found long-term stability of the material under CO–PROX reaction conditions and attributed it to the 3DOM interconnected mesoporous wall structure, which can efficiently transfer reactant species.

Equal or even more variation of the synthetic methods concerning ceria can be found in the literature. The morphology-dependent catalysis reported in the case of ceria-based systems excites a variety of publications dealing with the controlled synthesis and self-assembling of the ceria structures [120,121]. As the Au/CeO₂ interaction is of paramount importance in order to design and fabricate well performing catalytic materials, the control of the ceria morphology, and the gold particle size will result in a fine control of Au/CeO₂ interface. Huang et al. [122] compare the catalytic properties of Au nanoclusters deposited on one-dimensional CeO₂ (1-D) nanorods to conventional nanoparticulated CeO₂. This study reveals that the predominantly exposed (100)/(110) surface of the ceria nanorod structures notably affects Au dispersion, which in turn leads to a higher redox activity, catalyst reducibility, and superior activity in the CO oxidation reaction. The effect of the ceria morphology on the catalytic activity of gold-based catalyst in the CO oxidation reaction was also reported [123]. Uniformly distributed crystalline nanorods, nanocubes, and nanopolyhedra of ceria were prepared via a hydrothermal method, and Au was deposited via colloidal one. Ceria materials exposing other than (100) faces were more active in the oxidation reaction. The Au/CeO₂ nanorods (110 + 100) and Au/CeO₂ nanopolyhedra (111 + 100) showed higher activity than Au nanocube (100) catalysts. A one-step hydrochloric acid-assisted solvothermal method for the preparation of well-dispersed CeO₂ hollow nanospheres with high surface area and its decoration by colloidal Au nanoparticles were reported by Liu et al. [124]. According to the authors, this method leads to the existence of different Au species, both metallic and positively charged, which strengthened the interface interactions with the support. Zhu et al. [125] proposed a method for in situ growth of Au/CeO₂ core-shell nanoparticles and CeO₂ nanotubes by mixing gold and ceria precursors. The particularity of this method is that the used ceria precursor was already preformed Ce(OH)CO₃ nanorods. The formation of the Au/CeO₂ core-shell nanocomposite was obtained through an interfacial oxidation–reduction reaction between HAuCl₄ and Ce(OH)CO₃, where Au(III) was reduced to Au(0) by Ce(III) that was oxidized to Ce(IV). A final hydrolysis step generates CeO₂ nanoparticles. The advantage of this synthetic strategy was reported to be the independence of the method of foreign reducing agents and surface modifications. Hierarchical structures based on gold nanoparticles embedded into hollow ceria spheres and a mesoporous silica layer were proposed by Zhang et al. [126]. Compared with the CeO₂/Au/SiO₂ catalyst, the obtained hierarchical composite exhibits superior catalytic activity in the hydrogenation of 4-nitrophenol at room temperature, behavior assigned to the synergistic effect between the hollow structure and the mesoporous shell.

As seen from the above-mentioned examples, more and more complex structures are proposed every day. Nevertheless, no matter what the employed method of ceria preparation or gold deposition is, only a fine control of size and morphology of both components is required for their successful application in a variety of reactions.

3. Characterization of Au/CeO₂ Catalysts Involved in Oxidation Reactions

Since the discovery in 1987 by Haruta and coworkers of the catalytic activity of very small gold particles (<5 nm) supported on different oxides in the CO oxidation, even at temperatures below room temperature [3], a huge number of studies have been devoted to the application of gold nanoparticles, not only for CO oxidation but also for different oxidation reactions. There are evidences that gold may be involved in the adsorption or the activation of the reactants during the oxidation reactions, and this strongly depends on its interaction with the support by means of electronic transfer, which is also dependent of the particle size and physical structure of the gold clusters [60]. Chen and Goodman [6] have demonstrated that the CO oxidation ability increases until a maximum value by decreasing the gold nanoparticles to 3 nm; below this size, the CO conversion decreases. These results are consistent with several works in which the higher catalytic activity in CO oxidation reactions is achieved by means of depositing gold on reducible supports [127,128].

Generally, excellent catalytic properties in the oxidation reactions [15,28,129,130,131,132,133,134,135] and WGS reaction [136,137,138,139] can be obtained by combining gold with bare ceria or by using modified ceria oxides. Upon ceria modification, the enhancement of some structural properties has resulted in a relevant improvement of the catalytic activity. The easiness of Ce⁴⁺ cations to Ce³⁺ reducibility results in the creation of oxygen vacancies and allows considerable mobility of the surface oxide ions of the ceria lattice [60]. The determinant role of ceria’s punctual defects in oxidation reactions has been widely proven. In fact, this material is considered as an active support when employed to prepare catalysts for oxidation reactions.

Moreover, the dispersion of gold over support surfaces has been shown to be dependent on the oxygen vacancies population within the oxide [11,15,17,20,51,57,127,140,141,142,143,144]. Rodríguez et al. [139] evidenced via Scanning Tunneling Microscopy (STM) the presence of oxygen vacancies in a model catalyst formed by CeO₂ nanoparticles dispersed on the (111) surface of a gold single crystals. Similarly, Au nanoparticles deposited via Physical Vapor Deposition (PVD) on CeO₂ (111) were also studied via STM. This study reveals that gold preferentially nucleates on surface defects. Moreover, the generation-annihilation of defects during the CO oxidation reaction is intimately connected with the reactivity of ceria surfaces [145]. This dynamic of gold clusters on the ceria surface has been corroborated by Density Functional Theory (DFT) calculations on model systems. These calculations emphasize the role of surface oxygen vacancies on the sintering-breakdown of gold nanoparticles as a function of temperature and reaction atmosphere [135]. Gold nucleation on Eu-modified ceria surface vacancies has been assessed by Raman spectroscopy and X-ray diffraction (XRD). In this study, Hernández et al. [20] show that oxygen vacancy concentration and gold dispersion is a function of the employed atmosphere and temperature. The redox processes occurring at the ceria surface during reaction determines the gold reversible sintering-breakdown processes. Moreover, this reversible transformation is strongly affected by the hydroxylation of the reducible oxide surface that, at high temperatures, is involved in the re-oxidation of the cerium sites in a mechanism that implies the competition of hydroxyl groups and gold atoms for oxygen surface vacancies [25]. DFT calculations of the role of water on CO oxidation over Au/TiO₂ catalysts support this H₂O–Au competition for surface oxygen vacancies [12,18].

Consequently, the catalytic performance of this sort of material during the oxidation reactions is closely related to the interaction between gold and these punctual defects. This has driven numerous efforts for understanding the behavior of this kind of defects in order to pretend control their influence on the particle size distribution of gold. Nevertheless, surprisingly, there is not a general agreement on the nature of the active sites in this kind of catalyst during oxidation reactions. This is why different conclusions about the chemical nature of the more active gold species may be found in the literature [146], and evidences of a determinant catalytic role of Au⁺ species [34,147] or even Au³⁺ species [128] have been presented.

For instance, Casalleto et al. [148] observed the presence of Au⁰, Au⁺, and Au³⁺ species in Au/CeO₂ catalysts by means of X-ray photoelectron spectroscopy (XPS) analyses and proposed Au⁺ species as effective in promoting CO oxidation at low temperature. Despite this, Corma et al. [128] presented spectroscopic evidences of oxygen supply during CO oxidation catalyzed by gold supported on nanocrystalline ceria. The authors established a correlation (Figure 2) between the specific rate for CO oxidation and the ratio between the intensities of the infrared signals associated with the adsorption of CO over Au³⁺ (band at 2148 cm⁻¹ representing Au³⁺–CO) and Au⁰ species (band at 2014 cm⁻¹ representing Au⁰–CO).

The authors also proposed that nanocrystalline ceria supplies reactive oxygen in the form of surface superoxide η¹ species and peroxide adspecies at a one-electron defect site to the supported active gold species to oxidize CO. The presence of the cationic species of gold was confirmed by means of XPS measurements, and the presence of superoxide and peroxide species on ceria throughout Raman spectroscopy. The formation of peroxide species upon oxygen adsorption on electron-rich support vacancies has also been proposed for other Au catalysts [21].

Rodríguez et al. [149] studied the decomposition of SO₂ over Au/CeO₂ (111) and AuO_x/CeO₂ species, analyzing the adsorption of this molecule over their surfaces by means of synchrotron-based high-resolution photoemission. These authors observed a complete decomposition of SO₂ achieved only after the introduction of oxygen in the ceria support. Therefore, they associated the active sites of the [Au + Au_x]/CeO₂ system to pure gold nanoparticles in contact with oxygen vacancies, demonstrating the activating role of such defects over the clusters of noble metal.

All of these models only coincide on the importance of the Au/CeO₂ interface for achieving excellent catalytic activity of those systems.

The relevance of the interaction between gold and oxygen vacancies has driven other interesting studies on the structural properties of this kind of catalyst: These properties can be resumed in different categories such as textural properties, reducibility, oxygen mobility, oxygen storage capacity (OSC), and morphology, among others. Experimental evidences have shown that all of these properties may be modulated by modifying the synthesis procedure of Au/CeO₂ systems. For instance, Si and Flytzani-Stephanopoulos in order to establish the possible effects of the preferential exposition of some crystal-planes of the ceria surface on the WGS reaction studied synthesis procedures allowing for the obtainment of morphology-controlled ceria nanoparticles [138]. Figure 3 shows micrographs of the obtained ceria nanoparticles morphologies and the corresponding catalytic performance of 1% Au/CeO₂ catalysts. Rod-like ceria nanoparticles enclosed by (110) and (100) planes resulted in higher activity for the WGS reaction, demonstrating the strong influence of ceria morphology on Au/CeO₂ catalytic behavior.

Concerning the textural properties, it seems that gold nanoparticles deposition over ceria or doped ceria systems does not produce dramatic alterations. Only a slight decrease in Brunauer Emmet Teller (BET)-specific surface area can be observed, often attributed to the calcination process after gold deposition that may promote surface area drop. Higher stability of the textural properties and, in some cases, an increase in the specific surface area is achieved in doped ceria systems. This effect is more frequently associated to the inclusion of doping agents rather to the presence of gold.

Nevertheless, not all heteroatoms modify the cerium oxide in the same way. For instance doping ceria with Zr, Zn, and Fe results in supports for gold catalysts with different properties [51]. Solid solution formation was achieved only in the case of Zr and Fe doping, whereas the use of Zn resulted in ceria and zinc oxide segregation. Only the introduction of Zr into the framework of ceria resulted in textural property promotion (higher BET surface area). After gold deposition, no further differences in the BET area occurred; however, the average pore size increased. This phenomenon has been also previously reported by Domínguez et al. [150] over Au/Al₂O₃ and Au/CeO₂ catalysts deposited on reticulated ceramic foams issued from stainless steel wastes and employed for the 2-propanol oxidation. In this work, the change of the pore structure upon gold particles incorporation was associated with the enlargement of the pores by gold, as proposed previously by Somorjai et al. [151]. According to this, the insertion of gold nanoparticles may modify the porous structure of the supports, including ceria and doped ceria, by increasing their pore sizes.

As for the reducibility studies, they are usually carried out by means of temperature programmed reduction experiments (TPR) using H₂. The temperature of reduction of ceria or doped ceria systems is considerably decreased in the presence of gold [60], especially the reduction of the surface Ce⁴⁺ cations, promoted by the existence of close contact between oxide and noble metal nanoparticles [152].

Closely related to the reducibility feature of Au/CeO₂ systems is their OSC. According to Duprez et al. [153], Ce-based oxides with cationic valences and oxygen vacancies may be considered as materials with high OSC. These properties allow for the storage of active oxygen species (O²⁻, superoxide, etc.) in O₂ excess and their release when the partial pressure of oxygen decreases. The experimental procedure for OSC measurements implies the use of reductant molecule, often CO and less H₂ [153]. According to Yao and Yu Yao [154], the CO₂ produced during the first CO pulse allows for the calculation of the OSC, whereas the total amount of CO₂ formed after several CO pulses gives the complete oxygen storage capacity of the material (OSCC).

Fonseca et al. [155] studied Au/CeO₂, CuO_x/CeO₂, and Au–CuO_x/CeO₂ catalysts for the CO–PROX reaction and observed by means of OSC measurements that about one monolayer of ceria can be reduced at 400 °C. The surface reducibility of the systems with gold (Au/CeO₂ and Au–CuO_x/CeO₂) is always superior at lower temperatures compared with the initial supports (CeO₂, CuO_x/CeO₂). This seems to be closely related to their catalytic behavior, with all gold-based materials clearly superior in the CO oxidation at low temperatures. Furthermore, a cooperative gold–copper effect was established for the Au–CuO_x/CeO₂ resulting in moderate and totally reversible inhibition by CO₂ presence and absence of inhibition by H₂O.

The OSC over gold-based catalysts prepared with doped ceria systems as supports (Au/M–CeO₂/Al₂O₃; M = Fe, Cu, and Zn) has been also studied by Reina et al. [142]. In this case, like in all previously reported, the enhancement of the OSC by the presence of noble metal was observed at a low temperature (150 °C). The authors described the role of the noble metal as a “channel”, allowing the oxygen migration from the gaseous phase to the structure or in the opposite way via the storage of oxygen species in the framework of the support.

4. Au/CeO₂ Catalysts: Structure and Oxidation Activity

4.1. Oxidation Reactions Catalyzed by Au/CeO₂

A well-known property of bulk metallic gold is its high inertness and non-changing character being ideal for jewelry purposes. Its low chemical and catalytic activity converts gold in the least reactive transition metal, often referred also as “coinage metal” [70,146,156]. However, gold becomes extremely active when subdivided into nanoscale (usually less than 5 nm) and dispersed over an adequate metal oxide or over activated carbon supports [15,29,64,71,138,157]. Among the different options, ceria results frequently in an optimum choice as support for gold nanoparticles. Due to its excellent redox properties, ceria is probably the most extended support employed in oxidation catalysis, and Au/CeO₂ combinations result in a perfect synergy. Here, the attention is driven to a number of reactions in which Au/CeO₂-based catalysts have an outstanding behavior.

4.2. CO Oxidation

The activity of gold-based catalysts towards CO oxidation is known to be sensitive to (i) the preparation method; (ii) the nature of the supports; (iii) the size of the gold particles; (iv) the pre-treatment conditions; and (v) the gold–support interactions [10]. Among all the support choice is of primordial importance. The support usually plays multiple roles, such as mechanical functionality, i.e., to disperse and stabilize the supported nanoparticles against agglomeration and at the same time it could participate in the catalytic process.

Conventionally, the supports for CO oxidation can be classified depending on their reducibility or ability to supply reactive oxygen: reducible materials such as CeO₂, Fe₂O₃, CoO_x, and non-reducible supports such as MgO and Al₂O₃. Currently, gold supported on a reducible transition metal oxide always exhibits better activity for CO oxidation than gold supported on non-reducible oxides [6,158]. However, non-reducible materials with electron-rich oxygen vacancies may also enhance the catalytic activity of gold by activating gaseous oxygen [16,21]. Ceria, due to its chemical properties, has become one of the most promising supports for oxidation reactions and has been broadly studied [7,8,9,10,15,17,20,26,28,51,57,129,143,159,160,161]. The relevance of cerium oxide, as a support for gold nanoparticles, is mainly based on its high oxygen storage capacity resulting from oxygen mobility in its lattice [47]. The latter is directly correlated to the creation, stabilization, and diffusion of oxygen vacancies, especially on oxide surfaces, due to the reversible redox behavior of the Ce⁴⁺/Ce³⁺ couple and leading to the formation of non-stoichiometric oxides CeO_2−x (0 < x < 0.5). The concentration of these structural defects is usually correlated with the catalytic activity in CO elimination reactions [51,127,146,162,163,164,165]. In addition, these defects may act also as activation points of the CO and O₂ molecules, facilitating the reaction. Indeed, the participation of ceria in the reaction mechanism is vital since the main criticism towards CO oxidation mechanisms involving only Au atoms is related to the poor capacity of gold for oxygen chemisorption and activation [60]. Figure 4 shows a simplified reaction mechanism for the CO oxidation over Au/CeO₂. The rapid Ce⁴⁺/Ce³⁺ interconversion leads to the generation of oxygen vacancies (Equation (1)), where O₂ can be chemisorbed (Equation (2)). In parallel, carbon monoxide sits on metallic gold particles (Equation (3)), leading to adsorbed CO that would further react on the metal/oxide interface with O activated in the support vacancy producing the final CO₂ (Equation (4)).

$$({\text{CeO}}_2)\to \frac{1}{2}{\mathrm{O}}_2\uparrow +{\mathrm{V}}_{\mathrm{O}}^{••}+2e^{\prime }\to \frac{1}{2}{\mathrm{O}}_2\uparrow +{\mathrm{V}}_{\mathrm{O}}^{••}+2{\text{Ce}}_{\text{Ce}}^{\prime };$$

(1)

$${\mathrm{V}}_{\mathrm{O}}^{••}+{\mathrm{O}}_2\to {\mathrm{V}}_{\mathrm{O}}^{••}-{\mathrm{O}}_{2(\text{ads})};$$

(2)

$$\text{Au}+{\text{CO}}_{(\mathrm{g})}\to \text{Au}-{\text{CO}}_{(\text{ads})};$$

(3)

$$\text{Au}-{\text{CO}}_{(\text{ads})}+{\mathrm{V}}_{\mathrm{O}}^{••}-{\mathrm{O}}_{2(\text{ads})}\to {\text{CO}}_2\uparrow +\text{Au}+{\mathrm{V}}_{\mathrm{O}}^{••}.$$

(4)

Some authors have suggested that the presence of gold clusters weakens the oxygen species bonding on CeO₂ lattice, thus facilitating the formation of more reactive species [34,136]. In some cases, these reactive sites can participate in the reaction mechanism by creating activated species directly from the reactants mixture (e.g., peroxi-like and superoxi-like species), potentiating the oxidation reactions [166,167,168,169]. Guzman et al. [170] clearly demonstrated the role of these oxygen activated species in the CO oxidation process using in situ Raman spectroscopy. They obtained Raman spectra of gold supported on nanocrystalline ceria before and during the CO oxidation test, detecting bands at 1123 and 966 cm⁻¹ ascribed to superoxide and peroxide species, respectively [171]. These bands virtually disappear during CO oxidation, indicating the direct participation of these oxygen species in the reaction. In this study, it is also stated that the formation of defect sites on nanocrystalline ceria is promoted by the gold presence underlying the importance of the Au/CeO₂ contact and their synergistic effect [170]. Figure 5 schematizes the CO oxidation process picturing the direct participation of reactive oxygen species in the process. This synergistic effect has also been demonstrated for ZnO-doped CeO₂ systems where the band at 1127 cm⁻¹ ascribed to superoxide-like species has been observed by FTIR [172].

Given the importance of the oxygen defects in ceria to achieve high performances in CO oxidation, there has been a research trend in the last decade aiming to potentiate oxygen vacancy formation on ceria lattice. The O²⁻ removal from the oxide surface or oxide lattice can be achieved via several methods such as thermal treatments [173], electron irradiation [174], X-rays exposition [175], or chemical reduction [51]. However, for catalytic applications, the most popular strategy is to dope ceria with lower valence metallic cations [20,176,177,178]. The formation of oxygen vacancies in CeO₂–MO_x support is directly related to the formation of a Ce–M solid solution. The dopant ions with electronegativity and ionic radius close to those of cerium cations are considered the most appropriate modifiers of ceria structural and chemical properties. The presence of these dopants can provoke structural distortions such as ceria lattice contraction, thus favoring the formation of oxygen vacancies by mechanical stress.

Extensive research carried in our laboratory show the positive effect of a number of dopants including Eu [20,57], Zn [15,17,51], Fe [51,53,174], Cu [37,160], Zr [11,51,179,180], Gd [56], Ni [143], Cr [143], and Co [14], which potentiate ceria defect chemistry resulting in highly effective catalysts for oxidation reactions. Several situations may be envisaged as a function of the dopant cation. In the case of solid solutions formation on doping ceria with aliovalent cations, several mechanisms may occur for compensating the differences in the formal charge of cations. All possible compensation mechanisms are summarized in the following defect equations:

$$x{\mathrm{M}}_2{\mathrm{O}}_3({\text{CeO}}_2)\to 2x{\mathrm{M}}_{\text{Ce}}^{\prime }+x{\mathrm{V}}_{\mathrm{O}}^{••}+3x{\mathrm{O}}_{\mathrm{O}}^x;$$

(5)

$$x{\mathrm{M}}_2{\mathrm{O}}_3({\text{CeO}}_2)\to \frac{x}{2}{\mathrm{M}}_i^{•••}+\frac{3x}{2}{\mathrm{M}}_{\text{Ce}}^{\prime }+3x{\mathrm{O}}_{\mathrm{O}}^x;$$

(6)

$$x{\mathrm{M}}_2{\mathrm{O}}_3({\text{CeO}}_2)\to 2x{\mathrm{M}}_{\text{Ce}}^{\prime }+\frac{x}{2}{\text{Ce}}_i^{4•}+3x{\mathrm{O}}_{\mathrm{O}}^x;$$

(7)

$$x\text{MO}({\text{CeO}}_2)\to x{\mathrm{M}}_{\text{Ce}}^{″}+x{\mathrm{V}}_{\mathrm{O}}^{••}+x{\mathrm{O}}_{\mathrm{O}}^x.$$

(8)

The vacancy compensation mechanism is schematized in Equation (5). In this situation, two M³⁺ cations replace two Ce⁴⁺, leading to the formation of one oxygen defect. The dopant interstitial compensation, Equation (6), and the cerium interstitial Equation (7) mechanisms do not result in vacancy formation. Furthermore, for the case of divalent cations, Equation (8), one M²⁺ ion replaces one Ce⁴⁺ species, also resulting in one oxygen defect. In particular, the interstitial compensation mechanism and oxygen vacancies creation using divalent cation are very frequent. Empirical calculations carried out for M³⁺ cations show that the oxygen vacancy compensation mechanism is undoubtedly the preferred route, especially for large dopants cations (radius > 0.8 Å) [48]. Solid solutions with isovalent cations result in a slightly different situation. For example, the use of Zr⁴⁺ as a dopant potentiates the creation of oxygen vacancies, but this effect is mainly due to the distortion of ceria lattice and mechanical stresses induced by the differences in ionic radii between cerium and zirconium [11,51]. However, even if solid solutions cannot be formed, as in the case of Zn [15,17,51], the formation of oxygen vacancies under CO atmosphere is enhanced by the close contact between ZnO crystallites and CeO₂ nanoparticles [172] according to the model summarized in Equations (9)–(14):

$$x\text{ZnO}\stackrel{\text{CO}}{\to }x{\mathrm{V}}_{\mathrm{O}}^{2•}(\text{ZnO})+x{\text{Zn}}_{\text{Zn}}^x-{\text{CO}}_2(\text{ads});$$

(9)

$${\text{Zn}}_{\text{Zn}}^x-{\text{CO}}_2(\text{ads})\to {\text{Zn}}_{\text{Zn}}^x+{\text{CO}}_2\uparrow ;$$

(10)

$${\mathrm{V}}_{\mathrm{O}}^{2•}(\text{ZnO})+{\mathrm{O}}_{\mathrm{O}}^x({\text{CeO}}_2)\to {\mathrm{V}}_{\mathrm{O}}^{2•}({\text{CeO}}_2)+{\mathrm{O}}_{\mathrm{O}}^x(\text{ZnO});$$

(11)

$${\mathrm{V}}_{\mathrm{O}}^{2•}({\text{CeO}}_2)+{\text{Ce}}_{\text{Ce}}^{\prime }\to {\text{Ce}}_{\text{Ce}}^x+{\mathrm{V}}_{\mathrm{O}}^{•}({\text{CeO}}_2);$$

(12)

$${\mathrm{O}}_2+{\mathrm{V}}_{\mathrm{O}}^{•}({\text{CeO}}_2)\to {\mathrm{V}}_{\mathrm{O}}^{2•}({\text{CeO}}_2)-{\mathrm{O}}_2^{-}(\text{ads});$$

(13)

$${\text{Ce}}_{\text{Ce}}^x+{\mathrm{V}}_{\mathrm{O}}^{2•}({\text{CeO}}_2)-{\mathrm{O}}_2^{-}(\text{ads})\stackrel{\text{CO}}{\to }{\text{Ce}}_{\text{Ce}}^{\prime }-{\text{CO}}_3^{2-}(\text{ads})+{\mathrm{V}}_{\mathrm{O}}^{•}({\text{CeO}}_2).$$

(14)

In this model, the limiting reaction is the formation of oxygen vacancies in ZnO. This ZnO_1−x phase acts as an oxygen vacancy generator in the ceria phase. These vacancies are the active sites for activating oxygen as superoxide ions.

Figure 6 provides an example of CO oxidation activity improvement achieved when Fe or Zn are employed as CeO₂ dopants. The addition of tiny amounts of Zn and Fe (1 and 2 wt %) notably boost the CO oxidation activity of the Au/CeO₂-based materials. The results are especially impressive given the fact that all doped catalysts reached full CO conversion at sub-ambient temperatures, while the unmodified Au/CeO₂/Al₂O₃ achieved complete CO abatement at around 40 °C. In this specific case, the main difference between Fe and Zn as ceria promoters is that, while Fe³⁺ cations can replace Ce⁴⁺ forming a Ce–Fe solid solution [36,48], Zn²⁺ does not enter into ceria lattice. However, in both situations, a higher population of oxygen vacancies is identified. The immediate consequence of Ce–Fe solution is the formation of oxygen vacancies to balance the lattice charges upon substitution. Zn instead promotes surface ceria reducibility, generating oxygen defects at the ZnO–CeO₂ interface as described above [14].

Despite the encouraging results reported for Au supported on promoted ceria, there have been some examples where the use of a dopant does not benefit the activity and could lead to detrimental results instead. For instance, in a recent paper from Reddy’s group, some potential ceria promoters such as Fe³⁺, Zr⁴⁺, and La³⁺ for CO oxidation were analyzed [178]. Fe and especially Zr greatly boosted the CO oxidation capacity of the original Au/CeO₂ catalysts, while La shifted the light off curve towards higher temperatures. Furthermore, the La-doped material was not able to reach a full conversion in the studied temperature range. Apparently, while Zr and Fe favor the formation of ceria oxygen vacancies and improve ceria textural properties, the La incorporation caused a negative effect due to the presence of carbonate species blocking the active sites essential for CO oxidation. Thermogravimetric Analyses (TGA) revealed that the carbonates species formed on La-doped ceria are stable and remained anchored on the catalyst surface even at high temperatures. According to these authors, further tuning of the acidity of the support seems to be a sensible way to control carbonates formation [178].

In summary, Au/CeO₂-based catalysts represent an excellent option for direct oxidation of carbon monoxide in contaminated atmospheres with potential application in environmental catalysis. A number of factors can influence the catalytic properties of this type of material, such as gold particle size, preparation method, and Au/CeO₂ contact. Most of these parameters are broadly discussed in the literature and are not the subject of this section. However, there is general agreement on the important role played by ceria. With all other factors optimized (nanogold particle size, metallic dispersion, etc.), highly effective CO oxidation catalysts can be produced only when ceria electronic properties and defect concentration is tailored with an adequate dopant. Vast efforts has been made in recent years aiming to achieve an optimum performance, but there is still some room for further improvements.

4.3. WGS

As for the CO oxidation, the most popular supports for gold nanoparticles in WGS reactions are CeO₂-, TiO₂-, Al₂O₃-, Fe₂O₃-, ZrO₂-, and carbon-based materials, and combinations thereof [146]. However, ceria has taken the edge over all other supports and has recently become the most used material in gold-catalyzed WGS reactions. Figure 7 reflects this issue. Most of the works carried out in the WGS reaction using gold-based catalysts have included ceria as a part of the catalytic formulation.

The success of CeO₂ in the WGS is mainly due to the combination of an elevated OSC and a high ability to shift easily between reduced and oxidized states (Ce³⁺/Ce⁴⁺). A fundamental study of the reaction provides a deeper understanding of the benefits of ceria in this particular process. For instance, a theoretical investigation applying DFT methods to an ideal Au/CeO₂ system clearly illustrated that the water dissociation step into OH and H fragments is the rate limiting step for the WGS [146]. Figure 8 shows results from Rodriguez’s lab clarifying these aspects. Au (111) or Au nanoparticles have poor catalytic performance due to their inability to carry out the O–H bonds breaking in the water dissociation process (H₂O → OH + H). As indicated in Figure 8A, the biggest energy gap for gold is associated to this water activation transition state. However, the addition of ceria greatly improves the hydrogen production. The relatively large energetic barrier can be overcome by using ceria and specifically ceria oxygen vacancies as nucleation points to activate water. For comparison other metals, as for example Cu, can perform the water splitting by itself without support assistance since they present a much smaller energy barrier (Figure 8B).

It is worth mentioning that a perfect (111) CeO₂ surface is not able to dissociate water either. The water splitting ability exists only when O vacancies are present [146].

It is well accepted that the WGS reaction on an Au/CeO₂ catalyst proceeds as follows: CO is absorbed on Au sites, water splitting occurs on oxygen vacancies of the support, and the subsequent reaction takes place on the metal-oxide interface [165]. On these premises, the design of efficient catalysts for the WGS reactions involves the dispersion of very small gold nanoparticles on a highly defective ceria carrier.

The importance of oxygen vacancies and Au/CeO₂ synergy to achieve high performance is inherently imposed by the reaction mechanism. Conventionally, two main mechanism are proposed for the WGS: (i) the so-called redox mechanism (or surface mechanism), where CO adsorbs on gold particles and reacts with oxygen from the support, which, in turn, is reoxidized by water; and (ii) the associative mechanism, also called formate mechanism, where formate-like species are generated from the interaction of CO with OH groups of the support. The decomposition of these intermediates yields to the reaction products, CO₂ and H₂ [32]. Irrespectively of the mechanism ruling the process, the participation of both gold nanoparticles and ceria acting synergistically is fundamental. Figure 9 schematizes the redox mechanism reflecting the indispensable role played by the mobile oxygen from ceria and ceria oxygen defects.

In this context, intensive research has been carried out aiming to potentiate ceria oxygen mobility and oxygen vacancies population. Similar to the CO oxidation, the most extended strategy to enhance Au/CeO₂ performance is the use of promoters. Rare earth metals are a common option and, when added to ceria, provide improved thermal stability and increased activity [181]. Wang et al. [182] investigated a ceria/zirconia catalyst doped with rare earth metals (La, Nd, Pr, Sm, and Y) and found that all dopants increased activity and selectivity, with La, Nd, and Pr performing best. The results with lanthanum as a dopant of ceria are particularly interesting [183]. Liang et al. reported WGS activity of Au-supported La_xCe_1−xO_y nanorods (x = 0–0.5) with a uniform aspect ratio. They established that the reducibility of La_xCe_1−xO_y nanorods is determined by the preferential exposure of highly active (110) planes and by the La-to-Ce ratio. The La doping was found to act as an inhibitor of La_xCe_1−xO_y nanorods growth along the [110] direction and resulting in preferential exposure of (110) crystal planes for CeO₂ nanorods, a lower reduction temperature, and an increased number of reducible sites in the WGS reaction temperature range. Fu et al. [184] reported as well an enhanced activity for Au/CeO₂-based catalysts doped with lanthanum and gadolinium.

The only serious drawback of these systems is its stability issue and activity drop due to the formation of cerium/lanthanum hydroxycarbonate in shutdown conditions. Some authors have dealt with the carbonate formation problem and suggested that the tendency of ceria to stabilize carbonate on its surface could be inhibited through the addition of another metal, which does not stabilize these species [185]. Vindigni et al. [137] reported that Ce₅₀Zr₅₀ and Ce₈₀Zr₂₀ mixtures present similar acidities, both being two times higher than that of pure ceria. Therefore, zirconia addition modifies the acid–base properties of the support and less stable carbonate-like species are formed. This feature, along with the presence of high gold dispersion, makes Au/Ce₅₀Zr₅₀ the best catalyst in terms of catalytic activity and stability within the studied systems. At the same time, the addition of zirconium has a positive impact on the OSC of ceria [186]. Zirconium facilitates lower-energy bonding between oxygen molecules when compared to pure ceria. These weakly bonded oxygen molecules allow higher reducibility and thereby higher OSC, which in turns facilitates the WGS reaction [187].

The impact of the OSC in the WGS performance of a series of transition-doped Au/CeO₂/Al₂O₃ catalysts was recently addressed [142]. As shown in Figure 10A, all dopants (Cu, Fe, and Zn) sensibly boost the oxygen storage capacity of the reference Au/CeO₂/Al₂O₃ catalysts, an effect especially important at the highest studied temperature (350 °C) where the stability of the reaction intermediates is weakened; most likely, the redox mechanism dominates the process. Bearing in mind that the OSC informs about the most reactive and most available oxygen atoms on catalyst surfaces directly involved in the redox process, there must be a direct correlation between OSC and catalytic activities in redox reactions as WGS. Indeed, the catalytic activity Figure 10B follows the same trend as the OSC at high temperature: Fe-doped Au/CeO₂/Al₂O₃ > Zn-doped Au/CeO₂/Al₂O₃ > Cu-doped Au/CeO₂/Al₂O₃ > Au/CeO₂/Al₂O₃.

Alternatively, instead of modifying ceria support, gold can be promoted by another metal. For instance, Cu [188] and Co [189] were successfully employed to produce highly active Au–Cu/CeO₂ and Au–CoO_x/CeO₂ for the WGS reaction. In the case of cobalt, the authors conclude that the bimetallic catalysts adsorb more CO molecules, since it could be adsorbed on both cobalt and gold surfaces. The facilitated CO adsorption together with the improved OSC of cobalt containing catalysts explain the excellent WGS activity [189]. In the case of bimetallic Au–Cu, the addition of Au over CuO/CeO₂ promotes its capacity to release lattice oxygen and makes the oxidation of CO at low temperatures easier [188]. A more exhaustive study regarding bimetallic systems was presented by Juan et al. [190]. It was demonstrated that the addition of Pt, Pd, W, or Ni to Au–ceria provoke a synergistic effect and the resultant bimetallic catalysts are much more active than the parent Au–ceria system, Au–Pt–Ceria being the most active one. The authors expressed the activity in terms of CO fractional conversion and not in normalized specific activities (activity per metal site), since the interactions in the bimetallic catalysts are not sufficiently clear to differentiate between CO adsorption on one metal or another. The total number of metal sites per gram of the catalyst (in mmoles per gram), evaluated via CO chemisorption with the assumption of 1:1 CO to metal site ratio, was 2.7 on Au, 3.8 on Pt, and 2.5 on the Au–Pt sample [190]. Further details of the activity normalization are given in [190].

Alternatively, instead of modifying the catalysts composition, the WGS activity of Au/CeO₂-based systems can be influenced by adjusting the reaction conditions (i.e., co-feeding small amount of molecular oxygen in the gas mixture—the so-called “oxygen assisted water gas shift” (OWGS). Indeed, the OWGS is a good approach not only to improve the CO conversion rate but also to mitigate the deactivation effects related to carbonaceous species deposition or ceria over-reduction [190]. Flytzani-Stephanopoulos group’s presented a comparison using powder Au and Pt catalysts for OWGS, where small amounts of gaseous oxygen (air) stabilizes nano-structured Au–ceria and Pt–ceria catalysts in H₂-rich gases and start/stop operation [191]. Recently, González-Castaño et al. established some key differences between Au/CeO₂ and Pt/CeO₂ structured catalysts in the OWGS [192]. Apparently, only the gold-based system benefits for an actual “O₂-assisted WGS” reaction; meanwhile, the CO conversion increment observed for the Pt monolith results from the parallel WGS and CO oxidation reaction.

Going back to the stability of Au/CeO₂-based catalysts in the WGS process, their performance under discontinuous operation is especially relevant. For instance, under on-board operation (i.e., hydrogen alimented vehicles), all catalysts involved in the fuel processor (the reforming catalysts, the WGS catalysts, the CO–PROX, or the methanation catalyst) must tolerate the start-up/shut-down cycles of the engine. In fact, that this type of start/stop stability tests are the most demanding proves it can be imagined for a shift catalyst since, during the stop stages, the reactor is cooled down at room temperature with the reactive mixture flowing through the catalytic bed. The latter involves liquid water condensation on the catalysts’ porous structure, eventually deactivating the system. Despite the fact that this kind of test seems essential for practical reasons, they are not yet extended enough and few reports can be found in the literature considering these situations [192,193,194,195]. In a recent paper, multicomponent Au/CeO₂–CuO/Al₂O₃ catalysts were submitted to a series of start/stop cycles, and its behavior was compared to that presented by an Au/CuO/Al₂O₃ catalyst (Figure 11A) [144]. The Au/CuO/Al₂O₃ catalyst does not tolerate the start/stop sequences suffering clear deactivation after a series of consecutive cycles. On the contrary, the ceria-containing catalyst (Au/Ce2Cu8/Al) successfully withstands the start/stop operations with no evidence of activity depletion, an encouraging result for realistic application. The presence of ceria seems therefore to be indispensable. The spent sample were analyzed via temperature programmed desorption (TPD), and the CO₂ desorption profiles provide some clues of the Au/CeO₂–CuO system behavior (Figure 11B). The carbonaceous species formed under reaction conditions for the Au/CuO/Al₂O₃ catalyst formulation seem to be strongly attached to the catalyst’s surface and requires higher temperatures of removal. In contrast, the high oxygen mobility introduced by ceria facilitate carbonate-like species decomposition at lower temperatures [195].

In summary, Au/CeO₂-based systems are among the preferred materials for low- and medium-temperature WGS reaction. Their activity and stability can be considerably upgraded by altering the catalysts’ composition (use of ceria dopants, bimetallic formulations, etc.) or by adjusting the reaction conditions (for instance, introducing small amounts of air in the so-called oxygen-assisted WGS). The improved Au/CeO₂ formulations may present enough stability towards start/stop situations, a vital requisite for an intended realistic application.

4.4. Preferential CO Oxidation (CO–PROX)

The production of hydrogen pure enough to feed PEMFC requires further clean-up processes after the WGS unit in order to reduce the CO concentration to compatible levels with the PEMFC (typically 10–50 ppm) [10]. The preferential CO oxidation in rich-H₂ streams (Equation (15)) has been proposed as cheap and efficient alternative for the elimination of the final traces of CO [35,36,37,38]. The oxidation reaction is rapid and responds quickly to changes induced in the operating conditions. It is then essential to choose catalysts and operating conditions that minimize the oxidation of hydrogen. In this regard, the control of the reactor temperature is found to be particularly important [10].

CO + 1/2O₂ → CO₂ (desired reaction);

(15)

H₂ + 1/2O₂ → H₂O (undesired reaction).

(16)

Attending the above-mentioned equations, PROX reaction requires selective catalysts able to abate CO without oxidizing H₂ (Equation (16)). In other words, high CO oxidation activities coupled with low hydrogen ones are essential requirements for the PROX catalysts [38,39]. Various studies have confirmed that the rate of CO oxidation over supported Au catalysts exceeds that of H₂ oxidation [40,41,42,43]. Moreover, gold-based catalysts show extraordinarily high activity in the low-temperature range, which is appropriate for fuel cell applications. Earlier studies on Au/CeO₂ catalysts found high activity and good selectivity of those systems in the 70–120 °C temperature range, caused by the active participation of ceria in the oxidation process governed by gold [38,39]. The strong influence of the catalysts’ preparation method is a general statement in catalysis by gold, but it seems that for an Au/CeO₂-based catalyst for a PROX reaction, this becomes particularly relevant. For instance, the work of Luengnaruemitchai et al. [196] reflects this issue, showing in their case that the co-precipitation method is by far the best option. The superior activity of the catalyst prepared by co-precipitation seems to be related to structural and morphological reasons. For example, the ceria crystallite size for the sample prepared by co-precipitation was about 4–5 nm, which is much smaller than those prepared by impregnation and sol–gel methods. The smaller the ceria crystals size, the higher the concentration of oxygen defects and the higher the catalytic activity.

Apart from the preparation method influence, there is still much debate in the literature regarding the stability of gold-based catalysts in the PROX reaction. In an fuel processor, hydrogen would reach the PROX unit together with CO₂ and H₂O; thus, a suitable PROX catalyst must also tolerate high amounts of CO₂ and H₂O present in the reformate stream [196]. In this sense, Au/CeO₂ systems demonstrated sufficient stability with time on stream, in contrast to other gold catalysts, such as Au/Fe₂O₃ [197] and Au/TiO₂ [198], which lost a significant part of their initial activity during the first hours of reaction. A good example of high stability is extracted from the paper of Deng et al. [147], where an Au/CeO₂-based catalyst with low gold content preserves its CO conversion and selectivity during 50 h of continuous operation under the presence of CO₂ and H₂O.

The effect of CO₂ and H₂O in the PROX reaction is a controversial matter. While there is general agreement in the literature that CO₂ negatively affects CO conversion, the effect of water has led to contradictory results. In principle, the presence of both CO₂ and H₂O leads to a competitive adsorption phenomenon, since these co-fed reactants would compete with CO and H₂ upon the arrival to the active sites, thus decreasing overall activity. CO₂ also has an extra effect, which is the formation of stable carbonate-like species on the catalysts’ surface, thus acting as site blocking and limiting the accessibility of CO and H₂. Herein, the nature of the support is thought to affect the catalyst behavior in the presence of CO₂; acidic supports seem to be more resistant to deactivation than the basic ones [64]. Based on the latter, most studies have found a decrease in CO oxidation when CO₂ is present [129,147,162,196,199]. In the case of water, some studies have reported a negative effect on the CO abatement ability due to the above-mentioned competitive effect [129]. However, other studies [147,199] found a positive influence of water in the reaction. For example, Liao and co-workers found an improved CO–PROX activity for an Au–Cu/CeO₂ catalyst in the presence of steam [199]. They attributed the positive effect of water to several reasons: (i) the promotion of the WGS reaction and, as a result, the CO conversion to CO₂; (ii) the formation of hydroxyl groups via the dissociative adsorption of H₂O on gold, behaving as reactive sites for CO oxidation; and (iii) the promotion of the decomposition of carbonates in the presence of water. The influence of water in the total CO oxidation (not PROX) over Au/CeO₂ samples is also a much-studied issue [23,28,200,201,202]. In general, a positive effect of water at low temperatures (less than 120 °C) is reported, generally ascribed to the oxidative capabilities of H₂O and to its ability to improve the mobility of oxygen species, favoring the control of the implied redox reactions and assuring the adequate oxidation degree of ceria and gold species. Moreover, the direct involvement of H₂O and OH groups in the activation of O₂ molecules and the transformation of the catalytic intermediates and inhibitors such as the carbonate species have been considered. Some authors also detect an optimal water concentration in the feed-stream, after which the catalytic activity decreases or remains constant.

In a similar way, as for total CO oxidation or WGS, the performance of the Au/CeO₂ catalyst for CO–PROX reaction can be upgraded via ceria doping. As proposed by Avgouropoulos et al., the use of dopant cations with an ionic radius and electronegativity close to those of a cerium cation is considered the most appropriate way to modify cerias’ structural and chemical properties [162]. This is related to heterocations' ability to cause structural distortions inside ceria, causing oxide lattice strain and favoring oxygen vacancy formation, which in turn leads to greater oxygen mobility and higher oxidation activity. In particular, Sm and Zn are reported to have a positive impact on the Au/CeO₂ catalyst activity, while La has the opposite effect. Additionally, the promoters also help to achieve greater resistance towards CO₂ poisoning [162]. Zr, Fe, and Zn were proposed by Laguna et al. as very effective ceria promoters, with each dopant causing a different effect on the ceria structural and electronic properties but always positively influencing the catalytic activity [51]. As stated in this work, the formation of the surface oxygen vacancies depends on the nature of the modifier, doping with Zr enhances its formation, while the use of Zn does not influence the vacancies number and Fe makes them disappear. Nevertheless, a high dispersion of gold nanoparticles is achieved in all cases. During the modification with Zn, the epitaxial interaction between CeO₂ and ZnO creates preferential sites for gold deposition on the interphase between the two oxides. Doping with iron results in a solid solution formation where Fe ions occupy interstitial positions in the CeO₂ structure. These low coordination positions result in electron-rich sites that replace oxygen vacancies for anchoring gold nanoparticles, thus increasing noble metal dispersion and overall activity. More recently, broader dopant screening for the CO–PROX reaction using Au/CeO₂–MO_x/Al₂O₃ catalysts with M = La, Ni, Cu, Fe, Cr, and Y has been reported [143]. In this report, Fe and especially Cu were detected as the most promising promoters. Furthermore, the use of alumina as a main support provides several advantages. Dispersing ceria over a high-surface carrier such as alumina results in higher surface-to-bulk ratios, which leads to enhanced OSC of the samples [140]. At the same time, this catalyst configuration only requires a small amount of ceria, which reduces the expense of cerium oxide considerably. The low amount of ceria as support for gold nanoparticles provides an innovative approach to design Au/CeO₂-based catalysts for the PROX reaction.

Although much progress has been made in Au/CeO₂-based catalysts for the CO–PROX reaction, these systems present an inherent drawback which is difficult to overcome. The CO oxidation selectivity of gold-based catalysts is poorer compared to that exhibited by the CuO–CeO₂ catalyst, a reference material in the PROX process [203]. Furthermore, the selectivity of Au/CeO₂-based catalysts decreases drastically upon increasing the temperature. This is a consequence of the high activity of gold nanoparticles to burn hydrogen, also confirmed by with kinetics of the processes (normally H₂ oxidation rate superposes that of CO oxidation at medium-high temperatures). As a result, all intended applications of Au/CeO₂ systems are questioned by the selectivity issue. However, the selectivity-to-stability ratio can be modulated by adjusting the reaction parameters, as for example space velocity and CO-to-O₂ ratio (lambda parameter). As recently reported elsewhere [129], for low lambda values, lower CO conversion is obtained, although with high selectivity due to the preferential reaction between the available oxygen and CO, especially at low temperatures. For high lambda values, greater activity for both CO and hydrogen oxidation was observed, thus decreasing the overall selectivity. The contact time is also a point to consider for improving the conversion-to-selectivity ratio. In this case, both conversion and selectivity follow the same trend: the higher the space velocity, the lower the conversion and selectivity. In other words, the conversion-to-selectivity balance can be improved by increasing the contact time between the reactive molecules and the catalyst [129]. Tuning these parameters, a good performance can be achieved by Au/CeO₂-based catalysts. For example, a Cu-doped Au/CeO₂/Al₂O₃ catalyst has exhibited high activity (95% of CO conversion) and good selectivity (55% towards CO₂) in a suitable temperature range (around 110 °C) [143]. On the other hand, the selectivity in PROX reactions is greatly influenced by the supports. In this sense, supports with suppressed activity for hydrogen oxidation are a logical alternative [161]. Finally, to improve the selectivity of the process, it is strongly recommended that hydrogen and carbon monoxide oxidation is split and that each reaction is studied separately [161]. This approach provides an in-depth understanding of the selectivity trends bringing fundamental knowledge to tailor the activity-to-selectivity balance towards more efficient CO–PROX catalysts.

5. Conclusions

All the bibliographic works described and discussed in this comprehensive review highlight the importance of Au/CeO₂-based catalysts for potential application in various oxidation reactions. The number of scientific papers considering such materials is continuously growing, and it is foreseen that this trend will remain in the next future. The extremely high activity of gold nanoparticles in almost all, selective, or total oxidation reactions is indisputable. Although the gold metal appears to be very versatile as a catalyst, its activity obeys certain rules, the presence of nanoparticles (2–3 nm) stabilized on “active support” being the most important. One of those “active” supports is cerium oxide due to its unique structure and redox behaviour. The most important feature of this oxide is the possibility for easy modification of its oxygen storage capacity and resulting ability to activate oxygen species.

On first sight, the combination of both phases should result only in benefit, and it is the case in a perfect situation. However, every reaction is a “brave new world”. The gold deposition method and support modification and formulation should be perfectly adapted to the characteristics demanded by the reaction, which is not an easy task. The CO total oxidation needs a maximal oxygen vacancy population in order to maintain an oxygen activation rate high enough to oxidize every single CO molecule, whereas for WGS and PROX an optimal number of defects is needed in order to suppress the causes of deactivation or decrease in selectivity. Not only is the catalytic system important, but the process parameters should also be adapted to the application requirements. It is possible in some cases to fulfill the requisites without applying the most performant system. An important approach for this is, for example, the use of CeO₂–Al₂O₃ system instead of bare CeO₂. The former presents higher a ceria surface-to-bulk ratio, thus optimizing ceria redox properties and presenting a cheaper and easier way to modify material.

Although in the last decade, Au/CeO₂-based catalysts are within the most popular gold-based systems for CO oxidation reactions, the complexity of the interaction between gold and ceria makes the control and elucidation of the specific active site that controls the reaction mechanism of every considered oxidation reaction a difficult task and still seeks general agreement. Moreover, a variety of new oxidation reactions in liquid phase will open new horizons for the Au/CeO₂ system. There is still much work to be done, but it could be helpful to remember that the reaction provides the requisites, and the system provides the possibilities.