1. Introduction

Nitrogen oxide (NO_x, including NO and NO₂) is still one of main atmospheric pollutants at present. Its large emission could result in environmental pollution like acid rain and photochemical smog [1-8]. In addition, it can affect human health and cause some respiratory diseases [9,10]. To control the emission of NO_x from stationary or mobile source, many NO_x removal (deNO_x) technologies have been developed. Selective catalytic reduction of NO_x by NH₃ (NH₃-SCR) for both mobile and stationary sources is the most effective technology to control NO_x emissions [11-18]. Catalyst is the core of this technology, scientists have been working on to develop and improve more effective NH₃-SCR catalysts in recent years, which mainly be classified as metal oxide catalysts, noble metal catalysts, transition metal catalysts and molecular sieve catalysts (Fig. 1) [19-21].

Among various kinds of SCR catalysts, molecular sieve catalysts are always one of hotspots in research. Molecular sieve catalyst is a kind of crystalline silicate or silicon aluminate, with regular pore structures, mainly formed by oxygen-bridge bonds connected silicon oxygen tetrahedron or aluminum oxygen tetrahedron [22-24]. According to the different structure of molecular sieve catalysts, it can be divided into CHA type [25,26], LTA type [27,28], FAU type [29-31], MFI type [32-36], AEI type [37-41] and so on. Molecular sieve catalysts can ion-exchange with metal ions due to its unique structure and abundant Brønsted acid sites so that active components are dispersed highly in aluminosilicate. Because of its large specific surface area, molecular sieve catalysts are also exhibited its huge potential to act as a promising catalyst carrier. Therefore, they have been widely used in the study of NH₃-SCR catalysts. Table S1 (Supporting information) enumerates some typical NH₃-SCR molecular sieve catalysts. Some molecular sieve and metal oxide composite catalysts that reported in NH₃-SCR [38,42-45], are also listed in the Table S1. Among them, commercial catalysts like Cu-SSZ-13 and Cu-SAPO-34 are widely used in mobile sources, and researchers have conducted a lot of research on two catalysts [46-50]. Raquel et al. found that Cu-SSZ-13 can be synthesized by the one-pot method using copper‐tetraethyl pentamethylene complex and N, N, N‐trimethyl‐1‐adamant ammonium as organic structure directing agent (OSDAs) [51]. Xiao et al. found that Cu-SAPO-34 catalysts synthesized by direct ion exchange method (DIE) can separate more content of isolated Cu²⁺ and surface Cu, so its activity under low temperature is better than that of Cu-SAPO-34 synthesized by the conventional multistep ion exchange method [52]. And different types of copper species on Cu-SAPO-34 and Cu-SSZ-13 can be identified by combining IR spectroscopy and theoretical calculations [53]. In addition, there are many other types of molecular sieve catalysts used in NH₃-SCR. Yuan et al. developed a direct hydrothermal method by using a ferric complex, i.e. ethylenediaminetetraacetic acid ferric sodium (EDTA-FeNa), as both an iron source and a structure-directing agent, in which Fe/ZSM-5 molecular sieves showed significant NH₃-SCR activity. Under simulated industrial conditions, the conversion of NO_x is more than 99% in a wide temperature range of 573–693 K. At the same time, Fe/ZSM-5 molecular sieve also has good stability and tolerance to water vapor and sulfur dioxide, making these Fe/ZSM-5 zeolites promising candidates for practical application [54]. Naraki et al. proposed that Fe-BEA catalysts obtained by doping Fe into the beta molecular sieve framework had good performance for NH₃-SCR [55]. Zhu et al. believed that better NH₃-SCR performance of Fe-Beta was not affected by the specific structure, but because the layered Fe-Beta molecular sieve catalysts had more Fe active sites and could better disperse Fe species between layers [56]. Fully copper‐exchanged high‐silica LTA zeolites not only have good NH₃-SCR performance, but also own high hydrothermal stability [28,57]. Catalysts supported by molecular sieve catalysts or bimetal exchanged molecular sieve catalysts, such as MnO_x/SAPO-34 [58], CuNi/SSZ-13 [59], CuFe/ZSM-5 [60,61], also have good SCR activity.

Although molecular sieve catalysts have been reported to possess many advantages that guarantee their excellent SCR performance, there are still many factors that will cause the deactivation of catalysts in actual working conditions. Like metal oxidation catalysts, many researchers have thoroughly studied the poisoning mechanisms of alkali metals, heavy metals, phosphorus and SO₂, and even the multiple poisoning of K&Pb and Cd&SO₂ on common SCR catalysts, and put forward some corresponding anti-deactivation strategies to address these poisoning problems [21,62-65]. Similarly, hydrothermal aging and deactivation caused by sulfur poisoning are two difficult problems in the application of molecular sieve catalysts [45,66-74]. In addition, toxic deactivation caused by alkali metals, alkaline earth metals, phosphorus and HCl, etc. will also reduce the activity of molecular sieve catalysts and affect their service life [75-77]. Therefore, in addition to the catalytic activity of molecular sieve catalysts, their deactivation mechanism is also worthy of attention and research. Only when the deactivation mechanism of catalysts is clear, more effective anti-deactivation strategies can be proposed. Here we summarize the deactivation mechanisms of molecular sieve catalysts and some anti-deactivation strategies proposed by predecessors, hoping to give readers a better understanding in this research field (Fig. 2).

2. Deactivation mechanism and anti-deactivation strategies for hydrothermal aging 2.1. Deactivation mechanism for hydrothermal aging

Molecular sieve catalysts are often applied in motor vehicles, and the temperature of motor vehicle exhaust can be as high as 800 ℃ above, and the exhaust always has indefinite quantity water. Under this condition, the catalyst have to undergo hydrothermal aging, leading to the collapse of the framework and the agglomeration of active components, thus its activity will be reduced, and also its service life. At present, the stability of hydrothermal aging of molecular sieve catalysts is one of the important criteria for commercial application. Over the past decade, there were many reports on the hydrothermal aging of molecular sieve catalysts. Kwak et al. thermally aged four kinds of Cu-zeolite catalysts (Cu-beta, Cu-ZSM5, Cu-SSZ-13, Cu-Y) at 800 ℃ for 16 h. It was found that the aluminum coordination environment of Cu-ZSM-5 and Cu-β changed obviously, accompanied with Cu-aluminate species and CuO were found in two catalysts. And Cu-SSZ-13 is mainly reduced in redox capacity [78]. Young et al. analyzed the hydrothermal aging and deactivation mechanism of Cu-SSZ-13 under different Cu/Al and Si/Al conditions. It was found that the higher the Cu/Al, the more serious decrease of activity caused by hydrothermal aging [72]. Since the D6R (double-6-ring subunit) site was first occupied by the Cu²⁺ ion, the CHA site also began to be occupied as Cu/Al ratio gradually increased. However, because of the unstable nature of Cu²⁺ on CHA sites, condensation was more likely to occur than Cu²⁺, which led to the formation of CuO_x. Therefore, it was proposed that the main reason for the hydrothermal aging deactivation of Cu-SSZ-13 was the growth of CuO_x destroyed zeolite cages and pores, which led to the structure collapse. Other researchers believed that hydrothermal aging will lead to dealumination of Cu-SSZ-13 catalysts, resulting in the serious reduction of Brønsted acid sites and the crystallinity decrease. At the same time, the isolated Al(OH)₃ ion and Cu²⁺ ions combined to form CuAlO_x species, which deactivated the Cu species [79-82]. However, there are different opinions on Cu-SAPO-34 about the hydrothermal aging. Su et al. showed that vacancies in the framework of SAPO-34 caused by desilication can be repaired by migration of Al and P atoms outside the framework to defects. After aging, active Cu species in Cu/SAPO-34 zeolites tend to accumulate, which enhanced the redox ability and maintained its acidity [83]. But some studies have shown that dealumination, Cu²⁺ migration and weakening of acidity can occur both in Cu-SAPO-34 and Cu-SSZ-13 after hydrothermal aging, leading to the decline in NH₃-SCR performance [79]. Fan et al. found that after aging at 950 ℃ for 3 h, the crystallinity of SAPO-34 began to decrease. After 6 h, the zeolite crystals began to transition to SiO₂ and AlPO₄ phases. Meanwhile, the Si-OH-Al bond was broken, the Brønsted acid site and Cu²⁺ species were reduced. And the higher the Cu loading, the more obvious the Si-OH-Al bond breakage [84,85]. Furthermore, Wang et al. found that after hydrothermal aging (at 700 ℃ in 10 vol% H₂O/air for 48 h), CuO₂ species on the Cu-SAPO-34 surface would be transferred to the ion exchange site, thus improving the activity of the catalyst at 200–260 ℃. But in general, hydrothermal aging usually causes Cu-SAPO-34 to lose a large amount of acidity [86,87]. Peter et al. showed the change process of catalyst structure after aging of Cu-ZSM-5 and Cu-IM-5 catalysts: (1) partial dealuminization of molecular sieve, (2) reversible migration of Cu species, (3) forming some inactive or stable Cu-Al clusters (Fig. S1 in Supporting information). They also indicated that the stability of Al in the framework had great influence on the properties of Cu-ZSM-5 and Cu-IM-5 catalysts after hydrothermal aging [88].

The causes of hydrothermal aging deactivation of Fe/zeolite catalysts have also been investigated: (Ⅰ) the facile dealumination of Al-OH-Si sites, (Ⅱ) the fast loss of dimeric iron active sites, and (Ⅲ) tardy migration of isolated iron ions after dealumination [89]. Fig. 3 showed the general scheme for dealumination, migration and aggregation of Fe-ZSM-5 catalysts after the process of hydrothermal aging. The hydrothermal aging deactivation of Fe/BEA, Fe/SSZ-13 and other Fe/zeolite catalysts also follows mechanisms such as dealumination and Fe species change [90-92]. Iwasaki et al. also found that the decreasing order of SCR performance of several Fe molecular sieve catalysts is: Fe-BEA > Fe-MFI > Fe-FER > Fe-LTL > Fe-MOR (Fig. S2 in Supporting information). The activity of each catalyst decreased after hydrothermal aging, and the order of activity was as follows: Fe-MFL > Fe-BEA > Fe-FER > Fe-LTL > Fe-MOR. Therefore, the degree of deactivation caused by hydrothermal aging also depends on the ratio of silicon to aluminum [93]. For dual-active molecular sieve catalysts such as CuFe/BEA, hydrothermal aging can also inhibit the formation of active oxygen vacancies and charge transfer between Cu and Fe [94].

In summary, the effect of hydrothermal aging on molecular sieve catalysts can be divided into three parts: (1) affect the framework of molecular sieve catalysts; (2) affect active components; and (3) affect the surface properties of catalyst such as acidity and redox ability. The active sites, acidity and redox ability of catalysts are closely related to the NO_x reduction ability, hence, hydrothermal aging will lead to the decrease of the activity of molecular sieve catalysts.

2.2. Anti-deactivation strategies for hydrothermal aging

In order to solve the aging deactivation problem of molecular sieve catalysts, some anti-deactivation strategies have been proposed and shown as following.

(1) Improve the stability of the catalysts by adding additives. Iwasaki et al. proposed earlier that rare earth elements (rare earth metals with trivalent cation radii of 1.05–1.15 Å) can inhibit dealumination and lead to the formation of terminal silanol and iron oligomer, so that Fe/BEA catalyst can maintain good NH₃-SCR activity after hydrothermal aging (at 973 K for 5 h under 3% H₂O/air condition) [95]. Subsequently, it was found that introducing a small amount of Ce or Y rare metals into Cu-SSZ-13 or Cu/SAPO-34 molecular sieve catalysts by using ion exchange method can improve the stability of the framework and active centers, so as to ensure the SCR activity and anti-aging ability [96-99]. And also, Deng et al. believed that Ce doping was not only conducive to the formation of aluminum framework, but also helped stabilize the framework and Cu active centers, and improved the hydrothermal stability of catalysts. Chen et al. concluded that La ions incorporating into Cu-SSZ-13 can improve its hydrothermal stability when the optimal La content for Cu₄-La-SSZ-13 is ≤0.15 wt% [100]. However, with the La content increased, the framework aluminum and Cu active centers of Cu/SSZ-13 are reduced after hydrothermal aging, and even the CHA framework structure is destroyed, which is very unfavorable to the hydrothermal stability [101]. Similarly, Guan et al. also investigated that Ce species protected the copper ions and prevented the desalinization during aging over Cu/SAPO-34. After Ce doping, active copper sites and single dentate nitrate adsorption sites were greatly increased, thus improving SCR performance and stability (Fig. 4) [102].

(2) Optimize the preparation or treatment methods of catalysts. Jiang et al. proposed that Cu-SSZ-13 prepared by ion exchange method had higher resistance to hydrothermal aging than that prepared by one-pot method [108]. Ma et al. reported that Cu-SSZ-13 and Cu-SAPO-34 had a strong synergistic effect and the existence of Cu-SAPO-34 could enhance the hydrothermal stability of Cu-SSZ-13 (Fig. 5) [109]. Moreover, it can be found that the Cu-zeolite catalysts prepared by SAPO-34 instead of the traditional SSZ-13 microcrystal can yield higher performance as the crystal seed was inhibited during hydrothermal aging process, and the isolated Cu²⁺ site remained almost unchanged [110]. Wang et al. obtained mesoporous MCM-48 zeolite catalysts with better hydrothermal stability by carefully adjusting the synthesis conditions [111]. Therefore, the selection of the best preparation method and the best crystal seed can improve the possibility of obtaining the catalyst with high hydrothermal stability. Some optimized treatment of zeolite catalysts can also improve the hydrothermal stability. Huang et al. made Cu-SAPO-34 catalysts using the same composition but different crystal seed sizes (1–13 µm) by one-pot synthesis. It was found that smaller crystal seed size of the fresh catalyst induces less Si-O-Al structures on the surface, and less affected by water vapor hydrolysis, and less damage to the molecular sieve structure after long-term hydrothermal aging at low temperature, that is, higher the hydrothermal stability (Fig. S5 in Supporting information) [112]. Zhang et al. used ammonium hexafluoro silicate (AHFS) to modify Cu-SSZ-13 and reported it possessed the enhanced hydrothermal stability. After hydrothermal treatment at high temperature, the AHFS treated Cu-SSZ-13 still kept the CHA structure stably and the original activity was remained to a large extent, while the CHA structure of the contrast sample had collapsed and some of the active Cu²⁺ ions had been transformed into low active CuO, which greatly reduced the activity [113]. Zhang et al. improved the hydrothermal stability of Cu-ZSM-5 by modification of surface [114]. Chen et al. applied co-templates with tetraethylammonium hydroxide and copper-tetraethylenepentamine to synthesis the nanosized Cu/SAPO-34. Compared with conventional Cu/SAPO-34 catalysts, the as-prepared catalysts had the enhanced NH₃-SCR activity and the hydrothermal stability [115]. Wang et al. discussed that the increase of P/Al ratio can not only improve the crystallinity of the catalyst, but also increase the grain size, which is beneficial to the stability of the zeolite framework. Cu-SAPO-34 with large grain size owns the advantage of promoting the transformation of CuO into isolated Cu²⁺ ions and inhibits the aggregation of Cu species. In addition, the catalyst with a larger grain size also provides more acid sites, so that the catalyst possesses the good hydrothermal stability (Fig. S6 in Supporting information) [116].

(3) Adjust the pore structure of molecular sieve catalysts. Pu et al. found that MSAMS-2 and MSAMS-4 which are belong to SBA-15-type mesoporous aluminosilicates had small, stable five-element ring subunits that effectively prevented the attack of water molecules, so this type of mesoporous zeolite catalyst had good hydrothermal stability [117]. In addition, mesoporous ZrMCM-48 zeolite catalysts, mesoporous V-MCM-41 molecular sieve catalysts and mesoporous MCM-48 molecular sieve catalysts have high hydrothermal stability [118-120]. Xu et al. obtained highly hydrothermal stable mesoporous molecular sieves by ZSM-5 desilication [121]. Luo et al. prepared mesoporous aluminosilicate molecular sieve catalysts with high hydrothermal stability by diluted ZSM-5-type precursor with mesoporous MCM-41 [122]. In addition, the stable ordered mesoporous aluminosilicate catalysts (MASMS-1) can be obtained in high temperature steam by using ZSM-5 as precursor assembled with mesoporous MCM-41 [123]. Moreover, some researchers propose that the most promising method to strengthen the hydrothermal stability of mesoporous aluminosilicate is to construct mesoporous structures by using nano-clusters of aluminosilicate zeolite as raw materials [124]. Therefore, it is believed that the hydrothermal stability can be modulated via adjusting the pore structure of molecular sieve catalysts.

(4) Search for a new molecular sieve catalyst with high hydrothermal stability. There are many kinds of molecular sieve catalysts. At present, molecular sieve catalysts commonly used in SCR are only a small part of them, and there are many other types of molecular sieve catalysts waiting for researchers to explore. Moliner et al. synthesized the copper-exchanged SSZ-39 molecular sieve for the first time [125]. The main component of SSZ-39 (AEI) is silico-aluminate and it has a three-way pore system (8-R) with large cavities of D6R in the structure. The molecular sieve catalyst can keep Cu²⁺ stable in the test, avoid dealuminization and copper migration, and had good catalytic performance. Shan et al. compared the NO_x reduction performance of Cu-SSZ-39 and Cu-SSZ-13 before and after hydrothermal aging (Fig. S7 in Supporting information). It was found that Cu-SSZ-39 catalysts still maintained considerable activity after aging, which was due to the more stable framework aluminum and Cu²⁺, resulting in high hydrothermal stability [126]. There are some other types of molecular sieve catalysts, such as high-silica LTA zeolites, high-silica ERI zeolite, EU-7 zeolite, USY zeolite and nano-CHA zeolite, which have also been proved to have high hydrothermal stability [57,127-132]. Therefore, the design for new type of molecular sieve catalysts is also a promising direction for enhancing the hydrothermal stability.

3. Deactivation mechanism and anti-deactivation strategies for SO₂ poisoning 3.1. Deactivation mechanism for SO₂ poisoning

Sulfurous compound is often contained in fuel and engine oil, leading to the production of SO₂ in the exhaust gasses [133]. Therefore, the introduction of gaseous SO₂ is mainly used to simulate sulfur poisoning in the research. Sulfur poisoning mainly affects the NO_x reduction activity and durability of catalysts, especially Cu molecular sieve catalysts. Peter et al. exposed Cu-CHA catalysts to SO₂ and found that SO₂ can cause irreversible deactivation, thus affecting the service life of catalysts [134]. William et al. treated Cu-CHA catalysts by sulfurization and found the specific surface area after sulfurization reduced by 20%. Moreover, active Cu sites in catalysts could be severely hindered, leading to the decreased absorption of NO_x [135]. Besides, the presence of ammonium sulfate was found in the sulfurized samples, and ammonia sulfate covered the surface of the catalyst, which also affected the activity. Su et al. have studied the sulfur species on the surface of sulfur-poisoned Cu-CHA catalysts and found that H₂SO₄, CuSO₄ and Al₂(SO₄)₃ species deposited on the surface, which indicates that sulfur has an effect on both the active center and the framework of molecular sieve catalysts [136]. It had also been shown that copper bisulfate species would be produced in the sulfur poisoned Cu-CHA catalysts [137]. Hammersh and Shih et al. studied the specific sites of Cu-CHA who interacted with SO₂, and found that SO₂ mainly interacted with Z-CuOH sites which contained a framework aluminum center, that is, Z-CuOH site was more prone to SO₂ deactivation than Z₂Cu site (Fig. S8 in Supporting information) [138,139].

In order to further explore the toxicity of sulfur poisoning to molecular sieve catalysts and clarify the deactivation mechanism of sulfur poisoning, a single molecular sieve catalyst was selected for more detailed study. Sandra et al. explored the impact of SO₂ on Cu-SSZ-13 at different temperature and found that SO₂ caused serious deactivation of the catalyst at lower temperature (220 ℃), while SO₂ had little effect on Cu-SSZ-13 catalysts at higher temperature (400 ℃) and SO₂ would also reduce the selectivity of catalyst for N₂ [140]. Ammonium sulfate was mainly formed at about 200 ℃, while copper sulfate was mainly produced after sulfur poisoning at 400 ℃, which indicated that ammonia sulfate species had a great influence on the NO_x reduction ability of molecular sieve catalysts [141]. Kurnia et al. found that under SCR conditions (NH₃ + NO + O₂ + H₂O), the available copper sites in the redox cycle were reduced after SO₂ poisoning due to occupation by ammonium sulfur species, indicating that sulfur poisoning could significantly reduce the catalytic activity of molecular sieve catalysts by reducing the redox capacity [142]. For another commercial catalyst of Cu-SAPO-34, after sulfur poisoning, sulfates formed after sulfur poisoning also result in greatly reduced specific surface area and pore volume. In more details, some pores among Cu-SAPO-34 are blocked, active Cu sites that can participate in the redox cycle were rapidly decreased which had significant effect on the standard SCR reaction [143-145]. However, a slight difference with Cu-SSZ-13 is that SO₂ is difficult to interact with H-SAPO-34, so SO₂-poisoning hardly affected the crystallinity and characteristic structure [146].

In general, the SO₂ deactivation mechanism of Cu-SAPO-34 can be summarized as follows: (1) SO₂ may attack isolated Cu²⁺ to form copper sulfate/sulfite species, causing the active Cu²⁺ species to lose the ability to participate in the reaction; (2) competitive adsorption of SO₂ and NO_x, and (3) ammonium sulfate covers active sites and pores are blocked [147-150]. Other types of molecular sieve catalysts, such as Cu-SSZ-39 with high hydrothermal stability, also cannot avoid the reduction in NH₃-SCR activity caused by sulfate coverage of active sites [151]. Therefore, the improvement of sulfur resistance of molecular sieve catalysts is also an important direction in the research of catalysts.

3.2. Anti-deactivation strategies for SO₂ poisoning

At present, some effective anti-deactivation strategies for sulfur poisoning of molecular sieve catalysts have been proposed. Liu et al. reported, the Cu-SAPO-34 catalysts synthesized by ion exchange method had better sulfur resistance than that prepared by impregnation method, so it is believed that the preparation method can enhance the sulfur tolerance of molecular sieve catalysts [152].

He et al. studied SO₄²⁻-coordinated Cu-SAPO-34 catalyst which can sustain the integral SCR reaction cycle after SO₂ poisoning. It is due to the reserved reduction and re-oxidation process of Cu species (Fig. 6) [153]. In addition, since Ce is easy to react with SO₂ to form hard-to-decompose CeSO₄ species, the researchers proposed to add Ce to molecular sieve catalysts as a sacrifice site to protect the active site and reduce the inactivation caused by sulfur poisoning. Fan et al. introduced Ce into the framework of SAPO-34 and found that Ce suppress the reaction of SO₂ over Mn-Ce/CeSAPO-34 catalyst to form NH₄HSO₄, and Ce is easier to interact with SO₂, thus retaining the activity of Mn centers [154]. Li et al. also observed a similar enhancement effect by doping Ce into Fe/Beta catalyst [155]. Li et al. coated CeO₂ over Cu-SAPO-18 catalysts and found that it could also achieve the purpose of improving sulfur resistance [156]. In addition to Ce, another rare earth element Pr had also been found to prevent the sulfurization of MnO_x/SAPO-34 catalyst and suppress the coating of ammonium sulfate upon catalysts (Fig. S9 in Supporting information) [157]. Yu et al. designed Cu-SSZ-13-ZnTi₁₀O_x catalyst and it showed the great NH₃-SCR performance and had better anti-SO₂ poisoning ability compared to Cu-SSZ-13. ZnTi₁₀O_x, as a sacrificial component, may be preferentially poisoned by SO₂, resulting in the migration of Cu²⁺ from 8MR to 6MR which enhanced interaction with SSZ-13 support, thus improving SO₂ resistance [45]. These results indicated that adding additives can improve the tolerance of molecular sieve catalysts to SO₂.

On the other hand, in order to solve the problem of insufficient anti-SO₂ ability for Cu-ZSM-5 microporous molecular sieve catalysts, Peng et al. synthesized multistage porous Cu-ZSM-5-meso catalyst with both micropore and mesoporous structure. In contrast to the conventional Cu-ZSM-5 molecular sieve catalysts, the new molecular sieve catalyst had higher mesoporous mass transfer capacity, larger specific surface area, stronger acidity and NO adsorption ability, and had better NH₃-SCR activity and anti-SO₂ performance at medium and low temperature [158]. He et al. also recently found that constructing phobic Cu sites can alleviate SO₂ poisoning of Cu-SAPO-34 in NH₃-SCR via a hydrothermal aging (HA) method. After HA treatment, the adsorption capacity of SO₂ and deposition amount of ABS were reduced, and the overoxidation and polymerization of Cu species induced were also inhibited, thus maintaining the redox balance in NH₃-SCR reaction [159].

4. Deactivation mechanism and anti-deactivation strategies for alkali-metal poisoning 4.1. Deactivation mechanism for alkali-metal poisoning

Due to the increasing demand for energy and the decreasing number of fossil fuels, researchers have turned their attention to renewable energy. As a new type of green power plant fuel, biomass is widely distributed and has low carbon emission. Unlike typical fossil fuels, however, biomass contains much higher contents of alkali metals [160]. Though SCR units can be placed after the electrostatic precipitator of power plant, the exhaust gasses still contain certain contents of fly ash involving alkali and alkaline earth metals, which will produce toxic effects on SCR catalysts [161].

The exhaust gasses from vehicle sources is also reported to contain different content of alkali metals as generated from the combustion of oil, etc. Thus, molecular sieve catalysts applied both among stationery and vehicle sources are also inevitably affected by these toxic substances. Due to its larger specific surface area and stronger acidity, Cu-based molecular sieve catalysts have promoted tolerance to alkali/alkaline earth metal than commercial VW/Ti catalysts [162]. However, the SCR performance of Cu-based molecular sieve catalysts will be significantly affected when the toxic substance content is too high.

Alkali/alkaline earth metals that lead to molecular sieve catalysts poisoning generally include K, Na, Ca, Mg, etc. Putluru et al. exposed Cu-HZSM-5 and Cu-HMOR catalysts to KCl aerosol for a long time, and found that the catalysts were deactivated by 58% and 48% respectively [163]. Ma et al. found that when potassium content was high (> 0.5%), the NO_x reduction ability of Cu-SAPO-34 significantly affected, and proposed that potassium poisoning would cause Cu²⁺ (the highest activity) to be transformed into square-plane CuO/Cu₂O clusters (the lowest activity), causing the deactivation of Cu-SAPO-34 [164]. Liu et al. systematically investigated the toxic effects of various potassium salts such as K₂CO₃, K₃PO₄ and K₂SO₄ on Cu-SSZ-13. It was found that the diffusion of K⁺ into molecular sieve catalysts and then can form Si-O (K)-Al by exchanging with H⁺ and separated Cu²⁺ (Fig. S10 in Supporting information), leading to the decline of Brønsted acid sites and isolated Cu²⁺, thus inhibited the NH₃ adsorption on the surface and destructed the catalytic activity of molecular sieve catalysts [165]. The toxic effect of Na⁺ on Cu-zeolite catalyst is similar to that of K⁺, and Na⁺ will also conduct ion exchange with H⁺ and Cu²⁺in catalysts, resulting in the decrease of Cu²⁺ and acid center [166]. The toxic effect of alkaline earth metal Ca on catalyst is slightly different from that of alkali metal. Xue et al. loaded CaO on Cu/ZSM-5 and caused the decline of specific surface area and acid sites. Meanwhile, the interaction between CaO and CuO weakened the CuO reducing capacity, which will affect the activity of the catalyst [167]. The effects of K, Na, Ca and Mg on molecular sieve catalysts were compared. It was found that the toxic order of K, Na, Ca and Mg to Cu-SSZ-13 was Mg > Ca > Na > K, to Cu-SSZ-39 was K > Mg > Ca > Na, and to Cu-SAPO-18 was K > Na > Ca > Mg [168-170]. This shows that different kinds of molecular sieve catalysts have different degrees of tolerance to alkali/alkaline earth metals.

The alkali deactivation mechanism of Fe molecular sieve catalysts is mainly for the decrease of acidity and NH₃ storage capacity, but more important deactivation mechanism is that the pores are blocked by free poisons and/or the micropores are narrowed due to the adsorption of alkaline metals at cation sites [171]. Otherwise, alkali metals can generate the reduction of active Fe species in Fe zeolites. Liu et al. reported that more K deposition (> 0.5%) occupies the Brønsted acid site of Fe/Beta catalyst, resulting in the accumulation of active Fe species, increasing the NO oxidation activation energy at Lewis acid sites, but also inducing the formation of inert nitrate, blocking pores of beta zeolite, and significantly reducing active Fe species [172]. The deactivation mechanism of alkali/alkaline earth metals on molecular sieve catalysts can also be summarized as: (1) the effects on the active species; (2) the effects on catalyst structure; and (3) the influence on acidity and oxidation reducibility of catalysts.

4.2. Anti-deactivation strategies for alkali-metal poisoning

According to the alkali-metal deactivation mechanism of molecular sieve catalysts, alumina-rich zeolites with rich Brønsted acids was developed to improve the exchange capacity of K⁺ and protect isolated Cu²⁺ ions [173]. Moreover, the preparation of molecular sieve catalysts with core-shell structure is also a good method to improve the resistance to alkali metals. Liu et al. prepared Fe/Beta@meso-CeO₂ catalyst with core-shell structure, and found that this catalyst not only had excellent alkali resistance, but also had good sulfur toxicity resistance [174]. Chen et al. synthesized Cu/SSZ-13 by a simple method to form a core-shell like structure. First of all, the crystal size of Na/SSZ-13 is controlled by adding different number of seed crystals in the process of synthesizing SSZ-13. Following with the template removed from Na/SSZ-13, NH₄/SSZ-13 was obtained by twice ion-exchanging with 0.1 mol/L NH₄NO₃. Then NH₄/SSZ-13 was ion-exchanged by 0.1 mol/L Cu (NO₃)₂ to prepare Cu/SSZ-13.The results showed that larger the crystal size of molecular sieve catalysts, more favorable the distribution of Cu in the crystal core, and reduce the possibility to be replaced by Na to form CuO [175]. Zhang et al. found an MnO_x octahedral molecular sieve catalyst which had an alkali metal capture site can possess excellent alkali metal resistance (Fig. 7) [176,177]. Therefore, the method of separating the activated sites from the alkali metal sites can also be used to enhance the ability to resist alkali metal poisoning of molecular sieve catalysts.

5. Deactivation mechanism and anti-deactivation strategies for hydrocarbon poisoning 5.1. Deactivation mechanism for hydrocarbon poisoning

Because of the incomplete combustion of fuel, there are some hydrocarbons (HC) in the diesel exhaust, which will also influence the NO_x reduction ability of molecular sieve catalysts. Kim et al. proposed that C₃H₆ can inhibit the reaction between NO_x and feed NH₃, thus reducing the conversion rate of NO_x [178]. Some studies have shown that under standard SCR conditions, C₃H₆ has a great inhibition on the catalytic performance of Cu-zeolite catalysts. This deactivation mechanism is not raised from the competitive adsorption of C₃H₆ and NH₃, but from some intermediate species formed on surface during C₃H₆ oxidation [179]. Acrolein and coke, for example, would significantly mask the copper sites, thereby reducing catalyst performance. In addition, C₃H₆ will quickly reduce NO₂ to NO, affecting the process of fast SCR reaction. Some researchers proposed that the influence of hydrocarbons on Fe-zeolite catalysts was greater than that of Cu-zeolites [180]. He et al. found that the introduction of C₃H₆ would not only lead to coke deposition on Fe/Beta catalyst, but also result in partial reduction of Fe²⁺, thereby reducing NH₃-SCR activity of Fe/Beta [181]. In addition, the blockage of channels and Fe sites caused by the deposition of intermediate products, the competitive adsorption with ammonia and the unnecessary consumption of NH₃ caused by secondary reactions are all reasons for the decreased SCR activity after hydrocarbon poisoning of Fe-zeolite catalyst [182-184].

5.2. Anti-deactivation strategies for hydrocarbon poisoning

The effect of hydrocarbons on molecular sieve catalysts depends in part on the pore structure. Porous molecular sieve catalysts can prevent the diffusion of long chain HCs molecules to the internal active center, which can effectively reduce the deposition of coke and relieve the influence of HCs on catalyst [185]. Therefore, the problem of HCs poisoning can be dealt with by selecting molecular sieve catalysts with small pores. In addition, general strategies to hence the anti-poisoning ability of molecular sieve catalysts, like adding additives and constructing core-shell structure can also be used. Cao et al. introduced Y into Cu-SAPO-34 and found that the addition of Y not only efficiently promoted the copper dispersion, enhanced the acid density, increased separated copper ions, but also reduced the adsorption of C₃H₆ [186]. The addition of Ce can also play a role in strengthening the HCs resistance of molecular sieve catalysts [187,188]. The meso-Cu-SSZ-13@mesoporous aluminosilicate catalyst with a core-shell structure constructed by Zhang et al. also had the excellent propylene resistance (Fig. S11 in Supporting information) [189]. Recently, Zang et al. reported that Cu-ZK-5 catalysts with different copper loadings exhibited significantly different catalytic performance of NH₃-SCR and also expressed excellent resistance to propylene poisoning. The Cu_7.55-ZK-5 catalyst with appropriate copper loading owned a large amount of isolated Cu²⁺, which was conducive to the adsorption and activation of NO_x. In addition, the Cu_7.55-ZK-5 catalyst also exhibited a strong oxidation effect on propene, in which the propene can be completely oxidized at medium and high temperature [190].

6. Deactivation mechanism and anti-deactivation strategies for phosphorus poisoning

Phosphorus in mobile exhaust is mainly brought by fuel and lubricating oil, which will cause severe deactivation of molecular sieve catalysts. According to research, the activity of Fe/zeolite catalysts damaged severely when exposed to phosphate [171,191]. Of course, the Cu-zeolite catalyst will inevitably be affected by P species. In the early stage, Andonova et al. studied the influence of poisoning temperature and phosphorus concentration in the feed on the inactivation of Cu/BEA catalyst. It was observed that the pore condensation of H₃PO₄, the pore blockage and copper phosphate formation all resulted in the decrease of active copper species. These physical and chemical deactivation jointly affected the NH₃-SCR efficiency of molecular sieve catalysts [192]. The phosphorus poisoning mechanism on commercial Cu-SSZ-13 catalysts has been explored systematically. It is known that phosphate will be deposited in the pore, resulting in the blockage of the pore of the Cu-SSZ-13 catalyst, covering the acidic sites of the catalyst. And copper phosphate will be generated in the large cage of the molecular sieve, which will lead to the decrease of active Cu sites and the redox ability [193-196]. Recently, Guo et al. have also made some new discoveries about phosphorus deactivation mechanism of Cu-SSZ-13 (Fig. 8). They reported the strong interaction between Cu and P in the NH₃-SCR process at low temperature, which not only reduced the active site of Cu, but also restricted the dynamic movement of Cu. In addition, P poisoning can also significantly affect the redox cycling of Cu^Ⅱ ↔ Cu^Ⅰ at Cu sites, particularly the Cu^Ⅰ → Cu^Ⅱ reoxidation half-cycle, because P poisoning could inhibit the formation of vital dimer Cu intermediates [197].

Besides, phosphorus can also combine with aluminum in the molecular sieve framework to form AlPO_x species. Compared with fresh catalysts, the dealuminization of Cu-SSZ-13 poisoned by P was more serious after hydrothermal aging (Fig. S12 in Supporting information), indicating that P led to the decrease of the thermal stability [194,198].

The effect of phosphorus on NH₃-SCR catalysts is not absolute, and poisoning in the form of phosphate will lead to the decrease of catalyst activity. However, if phosphorylation is adopted, or phosphorus is introduced into the zeolite catalyst as a modifier, the performance of catalyst can be improved (Fig. S13 in Supporting information) [199-201]. In the actual working condition, phosphorus oxide (P₂O₅), partial phosphate (PO₃⁻) and phosphate (PO₄³⁻) are formed on the poisoning catalyst surface, among which partial phosphate is the main compound, so it will have a toxic effect on catalysts [194]. It has been concluded that the Cu-zeolite catalysts are more sensitive to phosphorus poisoning than to alkali and alkaline earth metals. However, there are still few strategies that can solve this problem. Chen et al. regenerated the phosphorus-poisoned Cu-SSZ-13 catalyst by hot water washing. In short, 1 g of the phosphated Cu-SSZ-13 catalyst was mixed with 100 g of deionized water. Then, the suspension was stirred and refluxed at 90 ℃ for a few days. After separation, washing with deionized water, and drying at 110 ℃, the regenerated Cu-SSZ-13 catalyst was obtained. It was found that after regeneration, the recovery of the decreased accessibility and porosity, the deactivated Brønsted acid site and the recovery of Cu²⁺ ions can restore the activity and hydrothermal stability of poisoned Cu-SSZ-13 catalysts, which provides an effective method for extending the lifetime of Cu-SSZ-13 catalysts in practical application [202]. Chen et al. revealed that phosphorus poisoning significantly decreased the low-temperature NH₃-SCR catalytic activity, however, such activity loss was alleviated by further hydrothermal aging treatment. It was found that Cu-P species produced by phosphorus poisoning decreased the redox ability of active copper species, resulting in the observed low-temperature deactivation. And Cu-P species could partly decompose with the formation of active CuO_x species and a release of active copper species after hydrothermal aging treatment [203]. It remains to be studied in the future to improve the phosphorus toxicity resistance of other molecular sieve catalysts.

7. Summary and outlook

In summary, molecular sieve catalysts have large specific surface area, unique and regular pore structure, and abundant exchangeable Brønsted acid sites. Industrial Cu exchange microporous molecular sieve catalysts, like Cu-SSZ-13 or Cu-SAPO-34 have been used for the NO_x reduction in mobile exhaust gasses for nearly ten years. These zeolite catalysts are active in a wide active temperature window and are resistant to high-temperature dealumination. However, they still faced many challenges including hydrothermal aging, SO₂ poisoning, alkali-metal poisoning, HCs poisoning and phosphorus poisoning deactivation [204,205]. Among them, hydrothermal aging is the most serious problem faced by molecular sieve catalysts at present, which needs to be solved urgently. In addition, other situations like HCl poisoning and Pt poisoning may also affect the activity of molecular sieve catalysts [206,207]. However, currently, the deactivation mechanism of molecular sieve catalysts is mainly studied in the above aspects. By summarizing the related researches, we can classify the deactivation mechanisms of hydrothermal aging or various toxic substances to catalysts as (1) the effect on catalyst structure, (2) the effect on active species, and (3) the effect on catalyst surface properties. These factors greatly damage the catalyst by reducing the adsorption of NH₃ or NO_x species, affecting or cutting off the reaction path, thus greatly impairing NO_x reduction performance. The toxic substances would also react with the feed gas to produce inert intermediate species, which covers the active sites of catalysts and lead to the useless consumption of NH₃. At present, the anti-deactivation strategies proposed to solve these problems of molecular sieve catalysts can also be concluded as follows: (1) adding additives, (2) constructing core-shell structure, (3) adjusting preparation methods and (4) exploring novel topological structures. These strategies can mitigate the deactivation of catalysts by establishing sacrifice sites and protective layer. However, these methods cannot solve all problems, such as P poisoning, and so far, there are few solutions other than hot water washing. These related studies show that molecular sieve catalysts are one of vital commercial catalysts with unique advantages and they are difficult to be replaced. Therefore, the research on mobile source catalysts is still based on further promote the NO_x reduction performance and achieve some deactivation issues of molecular sieve catalysts. In this paper, we briefly reviewed the deactivation mechanisms of molecular sieve catalysts, hoping to provide convenience for researchers to understand the various deactivation mechanisms of catalysts, so as to propose more effective anti-deactivation strategies.

In addition, the composition of exhaust gasses is complex, and there may be many factors affecting the activity of de-NO_x catalyst. For example, as for metal oxide catalysts, many reports have studied the deactivation mechanism of catalysts after multiple deactivation and anti-deactivation strategies [5,15,16,21,63,208]. Zou et al. unraveled the unanticipated differential compensated or aggravated Pb and SO₂ copoisoning effects over ceria-based catalysts for NO_x reduction and provided effective solutions to design ceria-based SCR catalysts with remarkable copoisoning resistance for the coexistence of heavy metals and SO₂. So molecular sieve catalyst in the application also face the same problem [5]. However, most studies on the deactivation of molecular sieve catalysts only consider one factor, and few studies report the influence of two and more factors together, such as SO₂&P, K&P or SO₂&hydrothermal aging [209]. There is a lack of research on the effects of the simultaneous existence of multiple poisons. There may be some interactions among various poisoning factors. The study and understanding of various poisoning mechanisms and the interaction between different toxic substance will be more conducive to the promotion of molecular sieve catalysts in the future. Therefore, the study on the simultaneous poisoning of multiple elements should be an important research filed of molecular sieve catalysts. Besides, for some unresolved poisoning problems, more efforts should be devoted to exploring innovative anti-deactivation strategies. Some researchers also mentioned that the high preparation cost of molecular sieve catalysts has greatly limited its wide application. Therefore, we need to synthesis molecular sieve catalysts from the perspective of low preparation cost, so as to maximize to advantages of molecular sieve catalysts and further reduce NO_x emissions.

Declaration of competing interest

All authors declare that there are no conflicts of interest, financial or otherwise in this work, and there are no other relationships or activities that can appear to have influenced the submitted work.

Acknowledgments

All authors acknowledge the support from the National Natural Science Foundation of China (No. 22125604), and the Chenguang Program supported by Shanghai Education Development Foundation and Shanghai Municipal Education Commission (No. 22Z00354).

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2023.108800.