1. Introduction

Concerns about the environment mean that new, increasingly demanding emissions standards have been gradually introduced around the world to improve urban air quality. California has the most demanding emissions legislation in the world. Its super ultra low emission vehicle (SULEV₂) standard mandates durability over 120,000 miles and the zero emission vehicle (ZEV₂) limits require up to 150,000 miles durability (1, 2). The typical exhaust gas temperatures of gasoline engines (≤800–850°C) are higher than those of diesel engines, but with very low start-up temperatures (≤250°C) that severely hinder hydrocarbon emissions control (3). In stoichiometric engines, close-coupled catalysts are exposed to bed-temperatures reaching 1050°C, resulting in difficulties in terms of the durability. Therefore, both the catalytic activity and stability of the TWC must be carefully designed to meet both the stringent emissions norms and durability requirements.

There are many factors, for instance phase changes, poison adsorption, coking, mechanical and thermal lash, that can seriously affect the catalyst’s structural and textural properties, metal dispersion, oxygen storage capacity (OSC) and redox activity (4, 5). The key deactivation mechanisms of catalysts are basically three-fold: chemical, mechanical and thermal. Thermal deactivation remains a fundamental problem to be resolved in TWC deactivation (6).

Compared with the base metals, precious metals are more commonly employed in automotive catalysts because of their intrinsic reactivity, preferable poisoning resistance and higher thermal stability that are required for automotive applications (7). However, precious metals also have intrinsic defects, for example particle growth and coalescence will occur under the complex reaction environment, resulting in rapid reduction of catalytic activity. This review is especially focused on providing an overview of the deactivation mechanisms in rare earth-based TWC and the corresponding strategies for improving durability.

2. Oxide Washcoat Materials

Alumina is employed as the most common washcoat material in TWC owing to its excellent chemical and thermal stability. Ceria-based additives play a critical role in providing oxygen mobility and oxygen storage in TWC due to their recognised ability to undergo rapid reduction/oxidation cycles, according to Equation (i) (8):

4CeO₂ ⇌ 2Ce₂O₃ + O₂ (i)

However, Al₂O₃ and CeO₂-based materials, as the main components of washcoats, are subjected to significant sintering as presented in Figure 1. Sintering is seen as one of the major deactivation mechanisms (4, 9). Washcoat sintering typically occurs at very high temperatures with coalescence and growth of particles as well as collapse and elimination of pores leading to a decrease of the specific surface area and encapsulation of active metal nanoparticles by washcoat, as well as a sharp loss of active surface area.

In the case of Al₂O₃, sintering is accompanied by thermally induced phase transformation. According to Arai and coworkers, γ-Al₂O₃ (100–200 m² g^–1 surface area) undergoes consecutive and irreversible phase transformation at high temperatures (10). The porous γ-Al₂O₃ starts to turn gradually to δ-Al₂O₃ at an approximate temperature of 900°C, which successively changes to θ-Al₂O₃ at approximately 1000°C and finally to non-porous α-Al₂O₃ (only ~5 m² g^–1 surface area) above 1200°C (7, 11), which results from the coalescence of particles and the elimination of pore structures.

Compared to other washcoat materials for TWC, Al₂O₃ has a higher specific surface area and better thermal and chemical stability that is particularly significant for metal dispersion. Hence, improving skeleton stability and suppressing phase transformation of γ-Al₂O₃ are necessary. These can be achieved through skeletal doping with specific dopants (i.e. lanthanum oxide (La₂O₃), zirconia (ZrO₂) and yttrium(III) oxide (Y₂O₃)). Inserting metal ions (especially La³⁺ with a large ionic radius) into the skeleton of transitional Al₂O₃ to prevent the diffusion of O²⁻ and Al³⁺ ions should be the most effective strategy for improving thermal stability. However, atomic scale insertion is still a technical challenge and remains an interesting research field.

CeO₂ readily crystallises and sinters with growth of particles and loss of surface area, leading to rapid reduction of the oxygen storage and release properties that are the most critical feature of TWC. Similarly, the formation of CeO₂-ZrO₂ solid solution achieved through introduction of Zr⁴⁺ ions into the CeO₂ skeleton not only stabilises the host lattice of crystalline CeO₂ at high temperatures, but also enhances the catalytic activities of water-gas shift reactions and hydrocarbon steam-reforming reactions in catalytic converters (12, 13).

CeO₂-ZrO₂ mixed oxides have a surface area below 2 m² g^–1 and the particles undergo significant coalescence and growth. One study found a crystallite size of 19.6 nm after thermal treatment at 1000°C for 5 hours under static air (14). Although total oxygen storage capacity (OSC-cI) and the amount of O₂ adsorption/desorption per mol of CeO₂ (OSC-cII) showed no obvious changes, the oxygen release rate (OSC-r) suffered a sharp decrease as shown in Table I.

For the further improvement of stability of the CeO₂-ZrO₂ composite, ternary CeO₂-based composite materials were introduced. As Lee’s group reported, La³⁺ and Sm³⁺ can effectively stabilise the cubic Ce_0.40Zr_0.60O₂ phase that plays a critical role in oxygen mobility and OSC, while other rare earth elements have the opposite effect (15). The ternary component content also affects the OSC stability according to Chen et al. (16): a high neodymium content will gradually convert the CeO₂-ZrO₂ cubic phase into a Nd_0.50Zr_0.50O_1.75 cubic phase which leads to a decline in the OSC.

Another strategy for improving the stability of CeO₂-ZrO₂ materials is the use of Al₂O₃ particles as diffusion barriers, introduced to form Al₂O₃-CeO₂-ZrO₂ hybrid oxides as shown in Figure 2. Compared to CeO₂-ZrO₂ mixed oxides, the Al₂O₃-CeO₂-ZrO₂ hybrid oxides still have a high surface area and OSC after durability ageing under the same conditions (14). There is however difficulty in controlling the metal-support interfaces which may have a negative effect on particular reactions occurring in the catalytic converter (17). The most negative influence for automotive exhaust catalysts is the formation of rhodium aluminate compounds at the Rh-Al₂O₃ interface (18).

As mentioned above, strategies that simultaneously take into account catalyst stability and control of the metal-support interface should be developed, to ensure high activity and sufficient durability, which are then able to meet upcoming emission standards. This is a promising research field but with some great technological challenges.

3. Active Precious Metals

Besides a sufficient washcoat surface area, two other key contributors to high catalytic activity are the surface area and dispersion of the precious metal nanoparticles. Over time in operation, the active precious metal nanoparticles will be subjected to crystallite growth due to sintering, resulting in reduced activity, even to the point of total deactivation.

Atomic and crystallite migration are two hypotheses for sintering of supported catalysts (19) as shown in Figure 3. In these cases, particle migration and coalescence (PMC) and Ostwald ripening (OR) are the two main sintering mechanisms according to Bowker (20). Both PMC and OR are dynamic processes which involve atomic exchange. The PMC sintering mechanism occurs when two or more particles touch or collide and then merge to form a larger particle. In contrast, the OR sintering mechanism occurs when material evaporates from the surface of a smaller particle with higher surface energy, and transfers onto the surface of a larger particle.

Sintering of active precious metal nanoparticles is closely related to reaction temperature and the size and composition of the nanoparticles (17). When quasi-liquid layers form on the crystal surface, migration of surface atoms and crystallites will occur at the Hüttig or Tamman temperatures of the bulk phase that are often used to semi-empirically describe the temperatures at which atomic migration and crystallite migration can occur. The ratio of the melting point (T_melting) of unique particles with radius, r, and heat of fusion, L, to that of the bulk phase (T₀) can be given by Equation (ii) (21):

where γ and ρ represent the surface free energy and density of the solid and liquid, respectively. Table II exhibits T₀, T_Tamman and T_Hüttig of platinum, palladium, Rh and their compounds. Metallic Pt, Pd and Rh have higher Hüttig and Tamman temperatures than the corresponding compounds, demonstrating that higher valency metals may have poorer thermal stability. Equation (ii) highlights that the melting point of unique particles depends on their composition, size and also surface free energy.

The experimental observations show that temperature is a dominant factor for sintering (22). The sintering behaviours of active precious metals can also be affected by any other conditions that lead to changes in the surface energy of the particles, such as reaction atmosphere (23), shape and composition of metal nanoparticles (24), metal dispersion (25), the loading of the metal on supports (26) and the interactions between the supports and metals (27).

Higher durability requires a lower sintering rate to maintain enough metal surface for long-term use and this is often accomplished through stabilising the active metal nanoparticles. Figure 4 shows the three typical strategies for precious metal sintering suppression (28–30): (a) support anchoring for metal nanoparticles, (b) alloying of active metals and (c) inhibition of atomic and crystallite migration via support encapsulation technology, for instance the construction of core-shell nanostructures and atomic layer deposition. In TWC, precious metal sintering is difficult to suppress by support anchoring due to the fluctuations of exhaust gas compositions and temperatures under operating conditions. In addition, support anchoring effects only occur at given temperature windows (31). Alloying by the introduction of a second metal with a higher melting point is a simpler and more controllable way to control metal sintering and improve thermal durability. Hence, additional composition and content optimisation of bimetallic catalysts for better self-stabilisation effect may be of great importance. However, according to our years of experience, alloying is both difficult to control and to avoid the negative metal-support interactions for industrial applications remains a challenge.

For the third strategy, support encapsulation technology can form additional metal-support interfaces, but both the stability of shells in core-shell nanostructure catalysts and the cost of atomic layer deposition technology still require further technical breakthroughs.

4. Redispersion and Regeneration

The metal dispersion and sintering rates of supported metal catalysts strongly depend on ageing atmosphere, as shown in Table III (32), indicating that the atmosphere found in exhaust gases can accelerate the deactivation rate over that seen in inert atmosphere. It is noted that redispersion is the opposite process to sintering. The reactant-induced disintegration of the active phase is of great importance in redispersion of active precious metals for long-term use. The disintegration strongly depends on the formation rate of adatom complexes such as PtO_x , Pd(CO)_x and Rh(NO)_x . These adatoms will form when metal-metal or metal-support interactions are replaced by metal-reactant bonds (33). Newton et al. found that Pd nanoparticle redispersion occurs during CO/NO cycling, which points to the formation of Pd(CO)_x and Pd(NO)_x adatom complexes (34).

Sintering and redispersion occur under reducing and oxidising atmospheres, respectively (35). Similar phenomena can be observed in Rh-based catalysts as well as Pt-based catalysts (36, 37). Thus, redispersion of the active phase in TWC might be interpreted as shown in Figure 5. The redispersion of precious metal particles in TWC can be achieved in two ways: disintegration or oxidation. The changes of chemical states that affect oxidation reactions largely depend on air-to-fuel ratio fluctuations and operating temperatures (38).

Grunwaldt’s group found that regeneration of deactivated bimetallic catalysts could be achieved by hydrogen reduction due to the formation of alloyed particles leading to enhanced oxidation activity under lean combustion conditions (39). This regeneration mechanism entirely differs from reduction-induced sintering because oxidation improves the precious metal dispersion for monometallic catalysts while the opposite is seen during the reduction process.

As reported by Birgersson’s group, regeneration of TWC can also be achieved through a thermal treatment under an atmosphere containing chlorine (40). However, it should be noted that oxy-chlorination may cause a narrow temperature window for oxidation of metallic species and the loss of precious metals due to the formation of volatile MO_xCl_y compounds. In the case of Pt, Pt(OH)_xCl_y and α-PtO₂ will transform into PtO_xCl_y above 500°C, resulting in Pt loss, as reported by Lieske et al. (41). Christou and coworkers reported a new ‘oxalic acid’ regeneration methodology for aged commercial TWC without any negative effects, which may be a promising method for further investigation to improve durability (42). In our experience, however, artificial regeneration of TWC is difficult to achieve for industrial applications. Thus, these regeneration methods are considered undesirable for improvement of durability of TWC and more attention should be given to other strategies.

The composition of gasoline engine tail-pipe emissions fluctuates quickly and dramatically between oxidation and reduction environments during operation. This makes ‘on-board’ redispersion and regeneration of TWC highly unsuitable and hardly controllable. CeO₂, used as the most common additive in TWC, provides oxygen mobility capability from the surface of CeO₂ to precious metal particles and promotes the dispersion of active metals (43). Nagai et al. reported that the stability of oxidation states depends on the precious metal oxide-support interaction (44). In our opinion, improving the oxygen activation properties of CeO₂ and optimising the interaction between precious metals and CeO₂ should be investigated to enhance the redispersion and regeneration capability of the active phase.

5. Conclusions

Thermally induced deactivation mechanisms are attributed to metal surface loss and the sharp decline of washcoat specific surface area as well as the collapse of pore structures caused by sintering. Redispersion and regeneration phenomena of TWC are only experimental observations and there is still a long development path towards industrial application due to real-world challenges and exhaust gas composition that can actually accelerate deactivation. To improve the durability of TWC, further attention should be given to understanding the metal-support interactions and the diffusion inhibition of atoms and crystallites in order to reduce the sintering rate of the precious metal and stabilise the structure of washcoats. Inserting specific ions into the skeleton of Al₂O₃ and CeO₂-ZrO₂ mixed oxides is an effective method for improving structure stability, but atomic scale insertion still presents a challenge in manufacture. In our opinion, the use of alloying effects to stabilise precious metal nanoparticles is very helpful for the improvement of thermal durability. Nevertheless, further optimising the composition and content of the alloys and reducing undesirable metal-support interactions still pose great challenges in TWC application and manufacture.

Acknowledgements

The authors are grateful for support by the National Natural Science Foundation of China (21463015) and the Provincial Natural Science Foundation of Yunnan, China (2014FA045).