Metal–support interactions of Ru-based catalysts under conditions of CO and CO₂ hydrogenation Free

Catalysts used for numerous practical industrial applications normally consist of metal or metal oxide nanoparticles dispersed onto the surface of an insulator or a semiconductor oxide, serving as the support material.¹  Catalytic performance for several heterogeneous reactions is often influenced by interactions between the active metal nanoparticles and the oxide support. The nature of such interactions may significantly affect, either directly or indirectly the chemisorptive and catalytic properties.² 

The direct role of the support, i.e. its participation in elementary reaction steps involves bifunctional effects, which are based on the creation of new active sites located at the metal–support interface or at the point of contact between the metal particles and the carrier.^1,3,4  Both, the nature and strength of these catalytic sites are defined, to a large extent, by the type of the support and their creation is able to exceptionally improve catalytic activity, selectivity and stability for several oxidation and hydrogenation reactions. For example, oxidation of CO, hydrogenation of several compounds (CO, CO₂, acetylene, crotonaldehyde) and methanol oxidation are well known examples for the direct role of the support on the reaction pathway.^2,3,5–10  This synergistic effect often involves spillover of adsorbed reactive molecules from the metal and/or the support surface to the periphery between the metal particle and the oxide support, where the reaction takes place resulting in high product yields.

Regarding the indirect role of the support, this is based on influences on crucial physicochemical properties which are considered to define catalytic activity, including the morphology, dispersion and size of metal crystallites, their surface and bulk electronic properties, as well as their resistance to sintering and poisoning.¹ 

Metal–support interactions are classified as weak, medium and strong.^3,11,12  Weak metal–support interactions are often observed over metal nanoparticles dispersed on non-reducible supports (e.g. Al₂O₃, SiO₂ etc.), whereas strong metal–support interactions (SMSI) are associated with metal nanoparticles supported on reducible oxides (e.g. TiO₂, CeO₂ etc.), giving rise mainly to geometric effects.³  Strong metal–support interactions were first introduced by Tauster et al.^13,14  over Group VIII metal particles supported on titania and other reducible oxides (e.g. CeO₂), who found that H₂ and CO chemisorptive properties of such catalysts are strongly suppressed following catalyst reduction at elevated temperatures (>500 °C).

Although electronic interaction may also operate concomitantly, the mechanism of the SMSI phenomenon has been interpreted by most researchers based on the geometric decoration model.^1,2,15–20  This model involves the formation of intermetallic bonds with localized charge transfer from the reduced oxide support to the metal crystallites, followed by migration of the so formed suboxides onto the metal surface forming thin decoration layers,²¹  thus suppressing the chemisorption ability of the metallic particles.

Geometric effects depend on various parameters, including the surface energy and the work function, Φ, of the metal, as well as the subsurface non-stoichiometry of the oxide support and the existence of oxygen vacancies in its lattice.^3,22  It has been found that decoration of the metal particles by reduced oxide layers is favored for metals characterized by higher surface energy compared to that of the oxide support, and higher work function.^23,24  Metals with low Φ values are preferentially oxidized rather than encapsulated on the oxide surface. Moreover, it has been found that diffusion of oxygen vacancies, from the bulk to the surface, promotes the reduction of the support, resulting in suboxide formation.²⁵  Therefore, strong-metal support interactions have been found to be enhanced over reducible metal oxides characterized by high oxygen vacancy defects such as TiO₂, V₂O₅ and CeO₂. SMSI effects have been investigated over many reactions, including CO and CO₂ hydrogenation to CH₄, CO oxidation and Fischer–Tropsch synthesis.^13,26–28 

Besides geometric effects induced by SMSI, it has been argued that electronic metal–support interactions may also occur when a metal is dispersed onto the surface of an oxide support. Such electronic effects are induced mainly by the work function differences between the metal particles and the semiconducting support material and have been proposed to explain a large variety of catalytic phenomena.^3,21  The origin of electronic type of interactions could be due to collective interactions between the metal particle and the support, which are defined by the bulk electronic structure of the two phases, resulting in short- or long-range charge transport to or from the metal nanoparticles.^1,29  Alternatively, the electronic interactions may originate from the interactions of metal clusters with local sites of the support characterized by specific properties including acidity or basicity.

The former concept is based on the metal-semiconductor boundary-layer theory, according to which the equilibrium requires that the Fermi energy level of electrons of the two phases is at the same height.^1,12,29,30  This can be achieved by charge transfer between the metal crystallites and the support until the Fermi level at the interface is equilibrated, processing significant alterations in physicochemical properties of both the metallic phase and the support. The electron transfer takes place from the semiconductor to the metal in cases of metals characterized by higher work function of electrons than the work function of the semiconductor. The inverse direction is operable in the opposite case.

Regarding the concept of electronic interactions between the metal clusters and the local sites of the support, it has been proposed that protons of an acidic support are delocalized on the metal particles forming a proton adduct that withdraws electron density from metal atoms.^31–34  The formation of such electron-deficient metal clusters has been reported to occur by metals (Rh, Ru, Au, Pt, Ir) supported on zeolite (Y, L, ZSM5) supports.^32,35–38  In case of metal clusters dispersed on basic supports, the opposite effect is operable resulting in negative charging of metal due to enhancement of the electron density of the oxygen atom of the support with increasing alkalininty.^33,39 

It is of interest to note that electronic interactions are strongly affected by the nature of both the metallic phase and the oxide support leading to different dominating mechanisms. The electron density transferred to the metal depends on the electronic structure of the support and the structural and morphological properties of the metal, and has been proposed to be inversely proportional to the electronegativity of the support and the size of the metal crystallites.^22,24,40  Moreover, electronic interactions may be enhanced by the presence of promoters (e.g. alkali or alkaline earth metals), resulting in modification of electron density of promoted metal surfaces. This is favored by doping of semiconductor support materials with cations of valence lower or higher than that of the host cation and has been proposed to affect the chemisorptive properties of metal surfaces toward reactive molecules, leading to improvement of catalytic activity and/or selectivity of certain reactions.^1,2,41  The effect of dopant-induced-metal–support interactions was first introduced by Verykios and co-workers,^42,43  who demonstrated that small Pt particles supported on TiO₂ doped with higher valence cations (Sb⁵⁺, Ta⁵⁺, W⁶⁺) exhibit significantly reduced chemisorption capacity for H₂, O₂ and CO. This was attributed to long-range electronic interactions at the metal–support interface, involving electron transfer from the doped support to the metal particles.

Well known examples of important catalytic reactions affected by metal–support interactions are Fischer–Tropsch synthesis of hydrocarbons and ethanol,^44–46  Water-gas Shift (WGS) reaction,^4,9,47–49  steam reforming of methane,⁵⁰  dry reforming of methane,⁵¹  methanol synthesis,^9,52  CO oxidation,^4,9,27,53,54  as well as CO and CO₂ methanation reactions.^7,26,55 

In the present report, the effect of metal–support interactions of Ru based catalysts has been focused on the CO and CO₂ hydrogenation activity and is discussed with respect to physicochemical properties (Ru particle size, support nature) of catalysts as well as catalysts synthesis and operating conditions.

Supported Ru catalysts have been proved to be promising materials for both the activation and interaction of several reactant molecules in a number of environmental and energy related applications. The beneficial effect of Ru has been proposed to be related to metal–support interactions, which are strongly affected by the physicochemical properties of catalysts, including Ru particle size and oxidation state, as well as the nature of the support employed.

When a metal is deposited on the surface of an oxide carrier, metal–support interactions may occur, the extent of which is strongly influenced by the nature of the metallic phase and, mainly, by the size of metal particles.^1,2,12,13  It has been suggested that the effect is more pronounced for metal particles smaller than 4–5 nm, whereas larger particles are less sensitive to interaction with the support. Results are in agreement with those reported by Haller et al.,²  who suggested that an electron transfer is more noticeable on smaller particles for all Group VIII metals supported on TiO₂. In terms of electronic effects, this was attributed either to the different electronic structures of small particles, or to the ability of small particles to stabilize some ions (e.g. Ru^δ+) against reduction and/or to the electron-deficiency of small particles, the unoccupied energy levels of which are more than those of large particles. The electron-deficiency assignment has been accepted by many researchers and was further proved by XPS experiments, which showed that the binding energy of core-electrons increases with decreasing metal particle size (Fig. 1).¹² 

In terms of bifunctional effects, the observed trend between metal particle size and metal–support interactions can be correlated with the length of the interfacial perimeter around the metal particle, which increases with decreasing the metal particle size.^2,56,57  Therefore, the higher the interfacial length the higher the number of active sites at the metal–support interface and, therefore, the higher the reaction rate.

The morphology of metal nanoparticles may be modified due to interactions with the carrier, leading to variations in the fraction of (a) the exposed crystallographic planes and (b) the coordination number of surface metal atoms.¹²  In the case of Ru based catalysts, it has been found that “corrugated” flat, highly disordered particles, created following reduction of the catalyst, exhibit exceptional activity for the hydrogenolysis of several organic compounds (ethane, propane, n-butane and 2,2,3,3-tetramethybutane), which was suppressed after catalyst oxidation.⁵⁸  Similarly, Tauster¹³  reported that the strong metal–support interactions strongly influence the morphology of metals dispersed on TiO₂ or similar oxides in a manner which depends on the size of metal particles. In the case of small metal particles, a flattening of particles forming thin, raftlike structures has been observed, whereas in the case of large metal particles the oxide inundates the metallic surface and the function of the support is suppressed.

The oxidation state of ruthenium catalysts seems to be crucial for the initiation of the SMSI state. Metallic Ru (Ru⁰) is considered to enhance the attainment of the SMSI state, while the partially oxidized Ru (Ruⁿ⁺) is considered responsible for the inhibition of SMSI state.⁵⁵  An explanation is offered by Li et al.⁵⁵  who found that electrons transferred from the TiO₂ support to the Ru surface are captured by Ruⁿ⁺ species, which may be present on the catalyst surface, thus hindering the occurrence of SMSI. This inhibition can be completed, following complete reduction of RuO_x species to Ru⁰. Contrary to the above results, Guo et al.²⁶  demonstrated that, in the case of Ru/CeO₂ catalysts, SMSI were more intensive over the most positive charged Ru atoms and the least positive charged Ce atoms located next to Ru. This is in agreement with the results reported by Wang et al.⁵⁹  over Ru/CeO₂ catalysts, who found that the presence of Ruⁿ⁺ species at the interface between Ru and CeO₂ induces metal–support interactions involving electron transfer from Ru to the support surface. This resulted in a significant increase of the concentration of oxygen vacancies, which were found to favor CO₂ activation and conversion to methane. The observed differences in the literature are most probably related to the different nature of the support employed, which, as will be discussed below, play a key role in metal–support interactions.

Metal–support interactions have been found to be significantly affected by the physicochemical properties of the support, including specific surface area, crystallite size, reducibility/oxygen storage capacity and Lewis acidity. Among these properties, the reducibility of the oxide carrier and its ability to store and release oxygen, which is usually referred to as oxygen storage capacity, is generally recognized as one of the main factors in determining SMSI effect.^60–62  The reducibility of an oxide carrier can be described as the tendency of the oxide to lose oxygen atoms under a reducing environment, resulting in the formation of oxygen vacancies and in modification of the oxide surface composition, from M_nO_m to M_nO_m−x.⁴  When H₂ is used as the reducing agent, reduction of the oxide carrier probably involves spillover of atomic hydrogen from the metal surface and, therefore, the SMSI effects is favored on metal systems that are able to dissociate H₂.² 

Metal oxide supports characterized by high specific surface area, usually accompanied by small crystallite size, are more reducible in the presence of dispersed metal particles.⁶³  For example, in the case of TiO₂ supported catalysts, it is well known that titanium dioxide is partially reduced to TiO_2−x by hydrogen at high temperatures, according to the following equation:

The process is promoted by the presence of dispersed metal crystallites and is believed to be the origin of the SMSI effect.^{12–17,63–65}  The reducibility of titanium dioxide was found to increase significantly with decreasing the primary crystallite size of titania, thus favoring metal–support interactions. Similarly, it has been proposed that ceria reduction is structure sensitive and that smaller crystals are easier to reduce.⁶⁶  This is also the case for Ru catalysts supported on composite CeO₂–Al₂O₃ carriers.⁶⁷  Decreasing the concentration of CeO₂ from 98 to 30 wt.% resulted in an increase of the specific surface area and a decrease of ceria particle size, which were accompanied by an increase of ceria reducibility, which was further enhanced by the presence of Ru.

Recently, Puigdollers et al.⁴  reported that the removal of oxygen atoms from the metal–support interface is more favorable compared to other regions on the catalyst surface. This implies that the metal–oxide interface is characterized by higher reducibility than the bulk oxide carrier, and, therefore, the reactivity of interface sites is higher, resulting in improved catalytic performance for reactions occurring at the metal–support interface (e.g. WGS, CO oxidation etc.).

Regarding the effect of the acidic nature of the support, many researchers agree that metal–support interactions weaken by decreasing the Lewis acidity of the oxide carrier.^68,69  Li et al.⁷⁰  and Wang et al.⁷¹  agreed that using an Al₂O₃ support with less surface acidic hydroxyl groups resulted in reduced metal–support interactions. This is in accordance with results of Zhang et al.⁷²  who showed that Al₂O₃ supports characterized by smaller specific surface area, fewer hydroxyl groups and less Lewis acid sites are beneficial to weaken metal–support interactions. Moreover, Prieto et al.⁷³  demonstrated that strong Lewis acid oxides (e.g. WO_x, TaO_x) are characterized by electron withdrawing metal cations in high oxidation states. Thus, such oxides are considered to be more reducible, favoring the SMSI effect.⁷³ 

The onset of metal–support interactions has been proposed to take place during the procedure of catalysts synthesis. In the case of using transition metal ions as metal precursors, such as Ru(NO)(NO₃)₃, and RuCl₃, they interact with the support surface in a manner which depends on the preparation method employed leading to variations in the physicochemical properties of the catalysts thus prepared.¹²  It has been suggested that strong ion-support interactions result in well-dispersed metal particles, and based on the above discussion, in enhancement of SMSI effect. Coq et al.¹²  reported that among the preparation methods of (a) impregnation, (b) ion exchange and (c) ligand exchange or chemical grafting, the latter one involves strong ion-support interactions able to lead to nano-sized particles of 1–2 nm diameter. Moreover, Haller et al.²  showed that high metal dispersions can be achieved by using an ion-exchange preparation technique thus, resulting in a high degree of metal–support interactions.

It is well known that the nature of metal precursor used for catalyst preparation plays a key role in both the physicochemical properties of catalysts and their activity for certain reactions. In the case of chlorine-containing metal precursors, chlorine ions are incorporated into the support lattice during reduction of the catalyst, resulting in a rapid alteration of the redox properties of reducible oxides (e.g. CeO₂).⁷⁴  In particular, the hydrogen chemisorption properties of the support vary and, as a result, the support becomes heavily and irreversibly reduced even at temperatures as low as 350 °C. The inability of the support to be re-oxidized modifies its reducibility, which, as discussed above, is relevant to the onset of the SMSI effect.

It is generally accepted that the metal–support interactions are strongly affected by the reduction temperature of the catalyst, in a manner which depends on the metal–support combination.⁷⁴  In the case of Pt/CeO₂ catalysts, H₂ reduction temperatures up to 700 °C form “decoration” CeO_x layers on the surface of Pt particles, which can be reversibly removed upon catalyst re-oxidation at the same temperature. However, higher reduction temperatures suppress the SMSI effect, resulting in the formation of CePt₅ alloys.⁷⁵  Decoration of Pt particles with TiO_x has been proposed to occur in a wider reduction temperature range (500–825 °C) over Pt/TiO₂ catalysts.^12,65,76  In general, for noble metal catalysts supported on TiO₂, surface oxygen vacancies has been proposed to be created as the reduction temperature increases, causing the reduction of Ti⁴⁺ to Ti³⁺.²  Moreover, Wang et al.⁵⁴  reported that the electrical conductivity, considering to be indicative for the formation of oxygen defects, increases significantly with reducing Au/TiO₂ catalyst under 10% CO/N₂ flow at 400 °C. According to the authors, the formation of titania defects not only on the surface but also in the bulk of TiO₂ induces electronic metal–support interactions. Results agree well with those reported by Akubuiro et al.,⁴³  who reported that the electrical conductivity of platinized titania is a strong function of the temperature and the environment in which the sample exists, and increases significantly in a H₂-rich reducing environment. The authors found that the electrical conductivity of doped (Sb⁵⁺, Ta⁵⁺, W⁶⁺) Pt/TiO₂ catalysts was five orders of magnitude higher under CO hydrogenation conditions than that measured under CO oxidation condition.^42,77  The reduced electrical conductivity in an oxygen environment is due to scavenging of conduction electrons according to eqn (2), indicating that the Schottky barrier height decreases and the amount of charge transferred into the Pt crystallites is smaller.

where V_O²⁻ represents doubly charged oxygen vacancies.

Komaya et al.⁷⁸  also found that increasing the reduction temperature of Ru/TiO₂ catalysts from 300 to 500 °C results in an increase of the extent of titania layer migration on the surface of Ru particles. Similarly, Reyes et al.⁷⁹  reported that reducing Ir/TiO₂ catalyst at high temperature (500 °C) results in surface decoration of the Ir crystallites by reduced titania suboxides, resulting in the creation of Ir–TiO₂ interfacial sites and bifunctional effects in the SMSI state. However, reduction of Ir/TiO₂ catalyst at low temperatures (300 °C) could not result in the development of dual active sites, suppressing the SMSI effect.

Operating conditions, including reaction temperature and the nature of reactant molecules, are expected to influence the degree of metal–support interactions. For example, high reaction temperatures and/or strongly reducing atmospheres (e.g. during hydrogenation reactions) favor metal–support interactions. Furthermore, electronic interactions vary depending on the type of reactant/product molecules, thus resulting in modification of their adsorption properties and reactivity. The optimum surface coverage of reactant/products depends on their heat of adsorption, which is strongly influenced by electronic effects. For example, the rate of ethylene hydrogenation over metal catalysts goes through a maximum for an intermediate value of ethylene heat of adsorption, which corresponds to Rh catalyst and provides the optimum ethylene surface coverage.¹² 

Moreover, the SMSI occurring over group VIII metals supported on reducible oxide supports suppresses the adsorption of both H₂ and CO on the metal surface.³  This can be seen in Fig. 2, where the hydrogen uptake (H/metal) following reduction at high temperature (500 °C), where SMSI effect is enhanced, is plotted as a function of hydrogen uptake following reduction at low temperature (200 °C) over a series of metals supported on TiO₂.¹²  The suppression of both H₂ and CO uptake under conditions where SMSI are favored has been related to inhibition of electron transfer from the metal atom to the adsorbed molecule due to the electrostatic interactions between the support's cations and the metal's electrons. Therefore, in cases where CO is the reactant, the SMSI effect could facilitate or not the CO conversion to reaction products in a manner which depends on the reaction pathway. For example, the SMSI effect is not expected to be beneficial for reactions where a CO dissociation step (2CO→C+CO₂) is required. However, it is possible that geometric effects may cause an increase of CO conversion, through the participation of the lattice oxygen of the oxide carrier on the reaction mechanism.²⁷  Moreover, for reactions taking place at the metal support interface, such as the WGS, the creation of interfacial active sites induced by the bifunctional effect results in enhancement of catalytic activity.^48,63  Thus, it can be suggested that besides electronic interactions, the contribution of geometric and bifunctional effects is significant and may be advantageous or not, depending on the metal–support combination and the reaction pathway.

Hydrogenation of CO is one of the most important chemical processes leading to high added value chemicals and fuels, including methanol, methane, long-chain hydrocarbons as well as long-chain n-aldehydes, alcohols, olefins and n-paraffins.⁸⁰  Among various processes, methanation of CO (eqn (3)) has been widely studied not only due to the significant practical importance of methane production, but also because it is a simple process, which can be used as probe reaction to study various catalytic phenomena, including metal–support interactions.

The reaction is catalyzed by Ru, Rh and Ni based catalysts, the activity of which can be improved with increasing metal loading or particle size and/or by appropriate selection of metal oxide support.^8,81,82  It is generally accepted that methanation of CO proceeds via dissociation of carbon monoxide to C and O atoms, followed by their hydrogenation into CH₄ and H₂O, respectively. The nature of active species depends on the metal–support combination employed and has been proposed to be governed by interactions between the metal and the support. In our study over Ru/TiO₂ catalysts, it was found that CO species adsorbed linearly on reduced Ru sites (Ru_x–CO) are active for the reaction, with their population and reactivity being increased with increasing metal loading or particle size.^83,84 

The role and influence of metal–support interactions on CO methanation was studied by Abdel-Mageed et al.⁷  over Ru/TiO₂ catalysts. It was found that catalytic activity depends strongly on the specific surface area of TiO₂ (ranging from 20 to 235 m² g⁻¹), exhibiting optimum results for an intermediate value of 121 m² g⁻¹ (Fig. 3). FTIR studies showed that the adsorption strength of CO on Ru surface increases with increasing TiO₂ specific surface area up to the above value and decreases for higher specific surface areas due to a partial overgrowth of Ru particles by TiO_x species, leading to lower catalytic activity. Taking into account that Ru particle size decreases with increasing surface area, the latter observation opposed to previous studies over supported Ru catalysts, where the CO adsorption energy was found to decrease with increasing Ru dispersion. The authors concluded that the observed increase of Ru nanoparticles for the highest surface area catalyst followed by the CO adsorption energy increase is caused by strong metal–support interactions. This effect prevails the effect of Ru particle size, which has been considered to be the key factor determining the CO hydrogenation activity over Ru based catalysts.

In addition to the specific surface area, the nature and the electronegativity of the support have been found to be correlated with metal–support interactions and affect catalytic performance of Ru based catalysts for the CO methanation reaction. For example, the conversion of CO toward CH₄ can be improved by controlling the electronic state of Ru by metal–support interactions, e.g. with the addition of acidic or basic reagents and/or by the appropriate selection of metal oxide support.⁸⁵  The degree of charge transfer between the metal and the support was correlated with the electronegativity of the support, which is a measure of its electron affinity. In particular, the electron affinity of the oxide was found to control both catalytic activity and selectivity of Ru for CO hydrogenation. Using oxide carriers of high electronegativity resulted in enhancement of both the turnover number and the chain growth probability in CO hydrogenation reaction (Fig. 4). This was attributed to weakening of CO adsorption and increasing of hydrogen coverage on the catalyst surface caused by the deficient electron density of Ru dispersed on highly electronegative supports.

Regarding the effect of the nature of the support on the CO hydrogenation activity, we have found that the specific reaction rate of CO conversion, under conditions of selective methanation of CO, is appreciably affected by the support upon which Ru is dispersed.⁸²  This can be clearly seen in the Arrhenius-type diagram of Fig. 5, where it is observed that activity decreases in the order of TiO₂>Al₂O₃∼CeO₂>YSZ>SiO₂ with TOF at 200 °C being ca two orders of magnitude higher when Ru is dispersed on TiO₂, compared to SiO₂. The beneficial effect of TiO₂ support for CO methanation was attributed to interaction between the Ru crystallites and titania, which may lead to electron donation from Ru into an antibonding n orbital of CO. As a result, the Ru–C bond becomes stronger, whereas C–O bond becomes weaker, favoring its dissociation, which has been considered as the rate-determining step for CO methanation reaction.⁸³  Such metal–support interactions have been proposed to be smaller or do not exist at all for metal catalysts dispersed on supports characterized by low oxygen storage capacity (e.g. Al₂O₃, MgO, SiO₂).

The existence of metal–support interactions during CO hydrogenation reaction was confirmed in our detailed mechanistic study over Ru/TiO₂ catalyst.⁸⁶  DRIFT experiments showed that the interaction of the catalyst surface with CO/H₂ mixture resulted in the development of a low-frequency carbonyl band at 2005 cm⁻¹, which extends down to 1900 cm⁻¹, assigned to CO adsorbed on Ru sites located at the metal–support interface ((TiO₂)Ru–CO) (Fig. 6A). The low frequency of this band was attributed to the strong electron-donating properties of ruthenium atoms located at the metal–support interface (e.g., Ru–Ti³⁺ sites), which originate from strong interaction with the reducible TiO₂ support. Interestingly, the intensity of the latter band increases significantly with increasing the concentration of H₂ in the gas stream, indicating that hydrogen may result in reduction of the TiO₂ surface in the vicinity of the metal crystallites leading to the creation of additional sites with enhanced electron-donating properties at the metal–support interface (e.g., Ru–Ti³⁺ sites) and, therefore, increased population of (TiO₂)Ru–CO species. Moreover, the low-frequency carbonyl band is red shifted with increasing H₂ concentration, implying weakening of the C–O bond strength, which is accompanied by enhancement of carbon hydrogenation to methane.

A similar correlation between the effect of the support type on the CO hydrogenation activity with strong metal–support interactions was made by Cattania et al.⁸⁷  Based on TPR and XPS measurements the authors found that when Ru is supported on Al₂O₃ both ruthenium oxidized and metallic states are present, whereas Ru exists only in its metallic state when it is supported on SiO₂. The oxidized state of Ru in the case of Ru/Al₂O₃ was generated by the migration of oxygen atoms from the Al₂O₃ lattice to the metal, indicating that strong metal–support interactions occur over this catalyst, leading to higher activity and selectivity for the CO hydrogenation reaction.

As discussed above, apart from the influence of the nature and physicochemical properties of the support, CO hydrogenation activity over Ru based catalysts is also affected by the size of Ru particles. The reaction is structure sensitive with respect to the metal and most researchers agree that both the reaction rate per surface metal atom (TOF) and CH₄ selectivity increase with increasing Ru crystallite size.^82,84,88,89  In certain cases, the structure sensitivity of CO hydrogenation has been correlated with metal–support interactions. For example, Kellnerand et al.⁸⁸  observed a decrease in specific activity for both methane and C₂ hydrocarbons production with decreasing Ru particle size, which was more pronounced for Ru dispersions higher than 0.7. The authors suggested that modifications of the electronic properties of small Ru particles caused by the size increase and/or interactions of Ru particles with the support may be responsible for the observed behavior. In particular, the electron density in the d orbitals protruding from the metal surface decreases over well dispersed Ru particles leading to a decrease of the degree of the charge back-donation from d orbitals to the π* antibonding orbitals of chemisorbed CO. As a result, the degree of C–O bond weakening is reduced. Taking into account that CO dissociation has been considered as the rate determining step in CO hydrogenation pathway and is favored by charge transfer to the π* orbital of adsorbed CO, the decrease in the back donation of d electrons could be responsible for the observed decrease in catalytic activity.

Based on the above studies, the existence of interactions between Ru particles and the oxide support have been considered to be responsible for variations in catalytic activity with respect to the metal particles or the support employed. However, in certain cases, although electronic metal–support interactions may be operable over Ru based catalysts, thus affecting catalysts chemisorptions properties, they are not able to define CO hydrogenation activity. For example, Taniguchi et al.⁹⁰  reported that reduction of Ru/TiO₂ catalysts at 773 K resulted in a drastic suppression of CO adsorption, which was assigned to SMSI since CO adsorption ability could be restored following catalyst oxidation and then reduction at 523 K. However, besides the observed variations in CO chemisorption properties under conditions where SMSI are favored, the activity of Ru/TiO₂ catalyst for the CO hydrogenation remains unaffected. This was explained considering that titania suboxides decorating metal surface in the SMSI state do not participate in the reaction mechanism and that Ru particles is exclusively responsible for the C–O bond cleavage. Similarly, Morris et al.⁹¹  reported that although substantial electron transfer may occur between TiO₂ and dispersed Ru particles due to the metal–support interaction effect, this is not directly responsible for the high activity and selectivity observed for the CO hydrogenation toward higher hydrocarbons. Interestingly, a negative effect of metal–support interactions on the CO hydrogenation activity was reported by Jackson et al.⁹²  over Ru based catalysts, who found that the turnover frequency for CO methanation reaction activity follows the order Ru/TiO₂<Ru/Al₂O₃<Ru/SiO₂, i.e. increases as the strength of metal–support interactions decreases. The authors stated that the rate determining step for the CO hydrogenation reaction was the addition of hydrogen to a partially hydrogenated carbon atom [CH_x(ads)+H(ads)→CH₄]. This indicated that the reaction rate depended not only on the ability of catalyst to provide CH_x species but also on the concentration of the adsorbed hydrogen atoms. This concentration was found to be enhanced following the same order with that of CO hydrogenation activity, implying that the ability of hydrogen adsorption on catalyst surface depended on the support type, the effect of which was reverse and prevailing related to the metal–support interaction effect.

Carbon dioxide hydrogenation is one of the potential paths proposed for the elimination and utilization of CO₂, which is considered as one of the main greenhouse gases resulting in global warming and major climate changes. Besides CO₂ reduction, the hydrogenation process is able to lead to the sustainable production of high added value chemicals and fuels, including carbon monoxide, methane, methanol, formic acid, higher hydrocarbons and higher alcohols provided that H₂ used is generated from renewable energy sources (e.g. water electrolysis, biomass etc.).^56,93 

The production of methane through interaction of CO₂ with H₂ (eqn (4)) has been extensively studied and found to be effectively catalyzed by supported Ru and Ni catalysts.^56,94–96 

Although Ru catalysts are more expensive, they are characterized by higher activity and stability compared to Ni catalysts, which suffers from deactivation at low temperatures due to the formation of mobile Ni subcarbonyls.^96,97 

Catalytic activity of Ru based catalysts depends strongly on the nature of the support and the size of metallic particles. Previous studies over Ru/Al₂O₃ and Ru/TiO₂ catalysts showed that the reaction is structure sensitive with respect to the metal i.e., both catalytic activity and methane selectivity increase significantly with increasing Ru crystallite size.^56,94  However, a few recent studies claimed that well dispersed Ru particles facilitate CO₂ hydrogenation to CH₄.^98,99  Differences may be related to metal–support interactions which seem to play an important role on methane selectivity. It has been proposed that catalytic properties of the active metallic phase, including morphology and dispersion, can be modified by interactions between metal particles and the support, thus, influencing the CO₂ hydrogenation activity.^95,96 

Regarding the mechanism of CO₂ hydrogenation over Ru based catalysts, it is believed that the reaction proceeds through intermediate formation of Ru-bonded carbonyls via the reverse water-gas shift (RWGS) reaction (eqn (5)) taking place at the metal–support interface, followed by hydrogenation of adsorbed CO to methane.^100–102 

Metal–support interactions and their effect on CO₂ hydrogenation activity and selectivity have been widely studied.^{55,97,98,103,104}  Trovarelli et al.¹⁰³  found that reduction of Ru/CeO₂ catalysts at high temperatures (500 °C) induces an interaction between Ru and CeO₂, which favors CO₂ hydrogenation to CH₄. Based on H₂-TPR and XPS measurements, the authors confirmed the existence of Ce³⁺ sites following high temperature reduction and suggested that CO₂ is activated by Ce³⁺ ions leading to its dissociation toward CO, which is further hydrogenated to CH₄. CO₂-TPD experiments showed that CO₂ dissociation provides oxygen surface species able to reoxidize Ce³⁺ to Ce⁴⁺, thus restoring catalytic properties. It was suggested that lattice oxygen vacancies play a key role on the reaction pathway and that the active sites for CO₂ hydrogenation are located at the perimeter between the metal and the CeO₂ support.

It is worth noticing that metal–support interactions under conditions of CO₂ hydrogenation may affect catalytic performance in parallel to other catalytic phenomena, like hydrogen spillover, and, thus the overall result being a contribution of several functions. Such a behavior was observed in a recent study dealing with the CO₂ hydrogenation reaction over Ru/CeO₂ catalysts.²⁶  Three different catalysts were tested: CeO₂ supported single Ru atoms, Ru nanoclusters (d_Ru=1.2 nm) and large Ru nanoparticles (d_Ru=4.0 nm). It was demonstrated that the reaction proceeds via the intermediate CO formation route, with the Ce³⁺ and Ru sites close to the metal–support interface being the active sites for the CO₂ dissociation and carbonyl hydrogenation, respectively.²⁶  Both the SMSI and hydrogen spillover effect were found to play important role on the activation of Ru carbonyls and the rate-determining step, exhibiting a balance over CeO₂ supported Ru nanoclusters, which also exhibited the optimum catalytic activity.

Based on the above, the creation of active sites at the metal–support boundary influences CO₂ hydrogenation and has been related, by many researchers, to the SMSI effect. However, in some cases, although the species adsorbed on sites located at the metal–support interface participate on the reaction pathway, the SMSI phenomenon may not be the dominant effect and/or may be also hindered due to the simultaneous occurrence of several catalytic phenomena. Such a behavior was reported by Li et al.,⁵⁵  who investigated the CO₂ hydrogenation reaction over Ru/TiO₂ prepared employing two different methods, spray reaction (SPR) and conventional impregnation (IMP). They showed that the former method resulted in higher catalytic activity, which was greatly favored following high temperature reduction (Fig. 7A).⁵⁵  Interestingly, this was attributed to the creation of new active sites located at the metal–support interface, which, however, was not related to the SMSI effect. In particular, the SMSI phenomenon was inhibited due to the presence of Ruⁿ⁺ (n<4) cations near the metal–support interface by capturing the electrons transferring from TiO₂ to metallic Ru. These Ruⁿ⁺ cations were formed following strong interaction of Ru atoms with oxygen atoms of TiO₂ and existed even after subjection to highly reducing environment. A proposed model of the active sites for the CO₂ hydrogenation reaction over Ru/TiO₂ catalysts prepared by the SPR method is presented in Fig. 7B.⁵⁵  Based on this model, the carbon end of the intermediate produced CO is bound to the metal and the oxygen end interacts with exposed Ti³⁺ or Ti⁴⁺ cations, facilitating the dissociation of the C–O bond and the hydrogenation of adsorbed carbon atoms to CH₄.⁵⁵ 

Additional physicochemical parameters which have been found to be related with both metal–support interactions and CO₂ methanation activity are the structure and composition of TiO₂ carrier.^98,99,104  In particular, the interaction between Ru particles and TiO₂ was found to be varied with respect to the different facets of anatase TiO₂.¹⁰⁴  Metal–support interactions were stronger between Ru and (101) facet compared to (001) facet. This was explained based on DFT calculations, which demonstrated that the activation energy for the dissociation of the intermediate produced CO (the rate-determining step) was lower when Ru nanoparticles located onto the (101) plane, resulting in higher CO₂ hydrogenation activity. The effect of TiO₂ structure/composition on CO₂ methanation and its relation with metal–support interactions was also studied by Kim et al.,^98,99  who observed improved catalytic performance when Ru is supported on TiO₂ consisting of both rutile and anatase phases compared to pure rutile or anatase TiO₂ supports. This was assigned to the tendency of RuO₂ nanoparticles to migrate to the rutile phase, when both anatase and rutile phases co-existed, resulting in the formation of highly dispersed metallic Ru particles of superior methanation activity (Fig. 8).

As discussed in Section 2.1, the extent of metal–support interactions depends on the size of metal particles and was found to be more pronounced for metal particles smaller than 4–5 nm compared to larger particles, which are less sensitive to interaction with the support.^2,12  This seems to affect CO₂ hydrogenation activity as reported be Kowalczyk et al.,⁹⁷  who found that the performance of well dispersed Ru particles (d_Ru∼1.5 nm) is strongly influenced by metal–support interactions. They concluded that small electron-deficient ruthenium particles are more active in converting CO₂ to CH₄ compared to the electron-rich ones. Turnover frequency was found to follow the order Ru/Al₂O₃>Ru/MgAl₂O₄>Ru/MgO>Ru/C, which was consistent to that of electron-deficiencies of the metal as determined by the Lewis acidities of the supports.

In our study the effect of the nature of the support on the activity of Ru catalysts dispersed on commercial metal oxides (TiO₂, Al₂O₃, YSZ, CeO₂, SiO₂, ZrO₂) was investigated under conditions of selective methanation of CO in the presence of excess CO₂ in the gas stream. Results showed that TOF of CO₂ conversion is affected by the nature of the support, with Ru/TiO₂ being the most active catalyst among samples of this series (Fig. 5). A detailed mechanistic study was carried out over the later catalyst in order to explore the reaction pathway and identify the nature of active species, which drive selectivity for CO₂ hydrogenation reaction. DRIFT experiments showed that CO₂ hydrogenation over 5% Ru/TiO₂ catalyst proceeds through intermediate formation of Ru_x–CO (band at 2012 cm⁻¹) and (TiO₂)Ru–CO (band at 1970 cm⁻¹) species via the RWGS reaction at the metal–support interface according to (Fig. 6B):⁸⁶ 

The formation of Ru carbonyl bands is accompanied by evolution of CH₄ in the gas phase, which is enhanced with increasing H₂ concentration in the feed. Increase of reaction temperature resulted in a progressive decrease of the relative population of (TiO₂)Ru–CO species indicating that they are more reactive for CO₂ hydrogenation. As discussed above, the creation of Ru–Ti³⁺ sites originates from strong interaction of Ru atoms with the reducible TiO₂ support, indicating that metal–support interactions drive the activity and selectivity for the CO₂ hydrogenation reaction.

The interactions induced when metal nanoparticles are dispersed onto the surface of oxide supports are among fundamental issues in heterogeneous catalysis. Understanding the influence of electronic, geometric and bifunctional effects on catalytic activity and selectivity is of significant importance, and able to shed light into the determination of the reactions mechanism, assisting in the design and development of suitable catalyst formulations for a number of applications. Metal–support interactions are strongly influenced by the physicochemical properties of catalysts, including Ru structure and oxidation state, as well as the nature of the support. Additional parameters affecting the interaction between the metal and the support include catalysts synthesis (method, precursor's type, reduction temperature) and reaction (temperature, the nature of reactant molecules) conditions employed. Among catalytic reactions of special interest strongly influenced by metal–support interactions are CO and CO₂ hydrogenation, which are both efficiently catalyzed by Ru based catalysts. The effect of metal–support interactions is explained based on electron-donating properties of ruthenium atoms located at the metal–support interface resulting in the creation of new active sites able to weaken C–O bond and thus, hydrogenate CO or CO₂. However, in certain cases of metal–support combinations and/or operating conditions, the interaction between Ru particles and the support may be partially suppressed by additional effects, including hydrogen spillover, structure sensitivity, the presence of Ruⁿ⁺ species and/or support influences on hydrogen adsorption, which may dominate and thus, define CO and CO₂ hydrogenation activity.