Poisoning Effects of Alkali and Alkaline Earth Metal Doping on Selective Catalytic Reduction of NO with NH₃ over the Nb-Ce/Zr-PILC Catalysts

Abstract

The poisoning effects of alkali metals (K and Na) and alkaline earth metals (Ca and Mg) on catalytic performance of the 2Nb4Ce/Zr-PILC catalyst for the selective catalytic reduction of NO_x with NH₃ (NH₃-SCR) were investigated, and physicochemical properties of the catalysts were characterized by means of the X-ray diffraction XRD (XRD), Brunner−Emmet−Teller (BET), hydrogen temperature-programmed reduction (H₂-TPR), X-ray Photoelectron Spectroscopy (XPS), ammonia temperature-programmed desorption (NH₃-TPD), and in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) techniques. Doping of M (M = K, Na, Ca, and Mg) deactivated the 2Nb4Ce/Zr-PILC catalyst according to the sequence of 0.8 K > 0.8 Na > 0.8 Ca > 0.8 Mg (M/Ce molar ratio = 0.8). The characterization results showed that the decreases in redox ability, NH₃ adsorption, Ce³⁺/Ce⁴⁺ atomic ratio, and amount of the chemisorbed oxygen (O_β) were the important factors influencing catalytic activities of the alkali metal-and alkaline earth metal-doped samples. Consequently, compared with the Mg- and Ca-doped samples, doping of K caused the 2Nb4Ce/Zr-PILC sample to possess the lowest redox ability, NH₃ adsorption, and amount of the O_β species, which resulted in an obvious deactivation effect.

1. Introduction

It is well known that the selective catalytic reduction of NO_x with NH₃ (NH₃-SCR) is one of the most effective technologies to remove NO_x from stationary source flue gas, but poor resistance to poisoning induced by the alkali metals and alkaline earth metals is still an urgent problem to be solved [1]. Fly ash is a major problem for the SCR catalysts because it may clog the pores of the catalysts and react with the active components [1,2,3]. However, the alkali or alkaline earth metals are the main components of fly ash, and have a strong toxic effect on the SCR catalysts. Effects of alkali or alkaline earth metals on catalytic activity of the well-known vanadium-titanium-based catalysts utilized in industry have been widely reported [4,5]. Especially for the tungsten-free catalysts, alkali and alkaline earth metal (mainly sodium, potassium, calcium, and magnesium) salts are generally considered to reduce acidity, and may partially react with V₂O₅, giving rise to the deactivation of the SCR catalysts [1,6].

In recent years, ceria as the NH₃-SCR catalyst has attracted much more attention, owing to its high oxygen storage capacity, excellent redox property, and nontoxicity [7,8]. Therefore, there have been many studies on the ceria-based NH₃-SCR catalysts in recent years, such as Ce–Ti [9], Ce–W [10], Ce–Nb [11], and Ce–Mo [12]. By introducing W, Nb, P or other solid acid component, SCR performance of the Ce-based catalysts could be remarkably improved. Among them, the Ce-Nb-based catalysts have broad applications in the flue gas post-treatment system due to their excellent performance [11,13]. In addition, a support is also considered as an important component of a NH₃-SCR catalyst because it can provide acidic sites and good dispersion of the active sites. Pillared interlayer clay (PILC) is a two-dimensional zeolite-like material. It is synthesized by replacing charge-compensating cations in the clay layers with macropolymers formed by hydrolysis of oligohydroxy metal cations or salts, thus separating these layers and finally forming porous networks with two-dimensional channels [14,15,16]. Compared with the zeolite-like support, PILC is a low-cost material with a large pore volume, a high surface area, good thermal stability, and high acidity. Hence, PILC could have specific applications if the column size and composition were modified.

Although SCR performance of the Nb-Ce-based catalysts has been reported, few studies have been carried out on the Zr-pillared clay-supported Nb-Ce catalysts, especially effects of the alkali or alkaline earth metals on NH₃-SCR performance, furthermore the interaction between the poison and cerium is still unclear. Therefore, the main purpose of this study is to investigate effects of doping alkali and alkaline earth metals on catalytic performance of the 2Nb4Ce/Zr-PILC sample and elucidate their deactivation mechanisms.

2. Results

2.1. SCR Performance

According to one of our previous studies, 2Nb4Ce/Zr-PILC showed excellent catalytic activity [17]. Therefore, the poisoning effects of alkali and alkaline earth metals on catalytic performance of 2Nb4Ce/Zr-PILC were investigated.

DeNO_x performance of the different alkali metal- and alkaline earth metal-poisoned 2Nb4Ce/Zr-PILC samples were tested in a simulating gas system, as shown in Figure 1. When the molar ratio of Na/Ce, K/Ce, Ca/Ce or Mg/Ce was equal to 0.8, different alkali metals and alkaline earth metal oxides exhibited different degrees of inhibition on NH₃-SCR activity, among which the poisoning effect of Mg was the weakest: NO conversion over 0.8 Mg-2Nb4Ce/Zr-PILC decreased only slightly (about 20%) from 150 to 350 °C, but the activity was still high in the range of 400−450 °C. The degree of Ca-poisoning was almost the same as that of Mg-poisoning. The activity of Ca-doped sample decreased slightly in the entire temperature range, and the maximum NO conversion was 80% at 450 °C. However, when K and Na were doped to the 2Nb4Ce/Zr-PILC sample, their NO conversions decreased significantly. Especially, K-poisoning was the most serious (NO conversion was only 65% at 450 °C). According to the above results, doping of alkali and alkali earth metals on poisoning degree of the 2Nb4Ce/Zr-PILC sample decreased in the order of 0.8 K > 0.8 Na > 0.8 Ca > 0.8 Mg, indicating that inhibition of the alkali and alkaline earth metals increased proportionally to the alkalinity. The 0.8 K-2Nb4Ce/Zr-PILC sample was deactivated most seriously among all of samples doped with equal molar amount of alkali and alkali earth metals.

Figure 2 shows deNO_x activities of the 2Nb4Ce/Zr-PILC samples with different amounts of K at different temperatures. The 2Nb4Ce/Zr-PILC sample showed high activity in the whole test temperature range (especially at 300–450 °C, where NO conversion reached 96% at 400 °C). However, K doping exhibited a strong inhibitory effect on deNO_x performance of the sample. When the K/Ce molar ratio was 0.3, activity of the sample decreased by 10–30% in the whole temperature range. With increasing the K/Ce molar ratio to 0.8, the activity inhibition became gradually serious, and the activity was less than 65% at 150–450 °C. These results demonstrate that deNO_x performance of the sample was greatly inhibited and the inhibition effect increased with a rise in K/Ce molar ratio.

2.2. NH₃ Oxidation

To clarify effects of the alkali and alkaline earth metals on NH₃ oxidation, NH₃ oxidation experiments were performed. Figure 3A shows catalytic activities of the fresh and M-poisoned samples for NH₃ oxidation. Over all of the samples, NH₃ began to be oxidized at 300 °C, and NH₃ conversion increased with a rise in temperature. However, NH₃ oxidation activity over each of the M-poisoned samples showed a downward trend as compared with the fresh sample, indicating that doping of the alkali and alkaline earth metals inhibited ammonia oxidation. It was reported that a strong redox ability could lead to over-oxidation of NH₃ [18,19]. Therefore, the decrease in ammonia oxidation activity might be caused by the decrease in redox ability (discussed below). As shown in Figure 3B, during the NH₃ oxidation process, N₂O formation over all of the samples increased with a rise in temperature. The poisoning of the alkali or alkaline earth metal progressively led to formation of the undesired N₂O product, especially over the 0.8 K-2Nb4Ce/Zr-PILC sample.

2.3. XRD

Crystal phases of the samples before and after M-poisoning were determined by the XRD technique, and their patterns are presented in Figure 4. The XRD peaks belonging to the two-dimensional hk reflections (at 2θ = 19.8° and 34.9°) and the quartz and cristobalite impurities (at 2θ = 26.6° and 28°) were clearly observed in all of the samples [15]. These peaks were characteristic of montmorillonite. No peaks assignable to the alkali metal oxides were detected, which was amorphous and dispersed evenly on the surface of 2Nb4Ce/Zr-PILC.

2.4. Surface Area and N₂ Adsorption-Desorption Isotherm

Textural parameters of the samples before and after M-poisoning are summarized in

Table 1. BET surface areas and pore volumes of the samples.

Sample	BET Surface Area (m2/g)	Pore Volume (cm3/g)
2Nb4Ce/Zr-PILC	271	0.164
0.3K-2Nb4Ce/Zr-PILC	231	0.150
0.8K-2Nb4Ce/Zr-PILC	215	0.134
0.8Na-2Nb4Ce/Zr-PILC	215	0.140
0.8Ca-2Nb4Ce/Zr-PILC	217	0.148
0.8Mg-2Nb4Ce/Zr-PILC	224	0.149

. Surface area and pore volume of 2Nb4Ce/Zr-PILC were 271 m²/g and 0.164 cm³/g, respectively. When the M was doped, surface area and pore volume dropped obviously. For example, surface area and pore volume of 0.8 K-2Nb4Ce/Zr-PILC decreased to 215 m²/g and 0.134 cm³/g, respectively. These results indicate that the doped M caused the partial blocking of pores, which explains part of the reasons why activity of the M-poisoned sample declined.

Figure 5 shows N₂ adsorption–desorption isotherms of the samples before and after M-poisoning. Each isotherm was nearly type I in the lower relative pressure range, which were characteristic of microporous materials. Nevertheless, the hysteresis loop corresponding to type H3 appeared in the region of the higher relative pressure range, indicating that the layered structure was preserved and the typical pores were slit-like [20]. It can be seen from Figure 5 that the adsorption volume of each M-poisoned sample decreased at a lower relative pressure, which proves that the amount of micropores in the sample decreased. The reason for such a phenomenon was due to the partial coverage of pores by the doped alkali and alkaline earth metal oxides on the samples. It should be noted that since the pillared montmorillonite formed mostly by a large amount of micropores and a few amount of mesopores, surface areas of the Zr-pillared samples were mainly contributed by the micropores and the contribution of mesopores to surface area could be ignored.

2.5. Reducibility

It is known that the redox property usually plays an important role in the NH₃-SCR reaction [21,22,23]. Therefore, redox properties of the samples before and after M-poisoning were evaluated using the H₂-TPR technique, and their profiles are shown in Figure 6. For each sample, four reduction peaks were observed at 473–489, 570–580, 671–706, and 779–815 °C, respectively. The first peak at 473–489 °C corresponded to reduction of the surface Ce⁴⁺ to Ce³⁺ [22], the second one at 570–580 °C was ascribed to reduction of the iron oxide species in montmorillonite, the third one at 671–706 °C was attributed to reduction of the bulk CeO₂, and the last one at 779–783 °C was ascribed to reduction of niobium oxide [21,22]. It is generally believed that the reduction peak temperature represented the reduction ability. A lower reduction temperature indicates a stronger reduction ability [23]. For the K-poisoned sample, the reduction peak at 483 °C was slightly shifted to a higher temperature than that at 476 °C of the fresh 2Nb4Ce/Zr-PILC sample. In order to accurately compare redox ability of the fresh and M-poisoned samples, H₂ consumption was calculated and presented in

Table 2. The reduction peak temperatures and H₂ consumption of the samples.

Sample	Reduction Peak Temperature (°C)				H2 Consumption (mmol/g)
	Peak 1	Peak 2	Peak 3	Peak 4
2Nb4Ce/Zr-PILC	473	570	671	779	0.390
0.3K-2Nb4Ce/Zr-PILC	476	576	677	815	0.291
0.8K-2Nb4Ce/Zr-PILC	483	575	701	806	0.236
0.8Na-2Nb4Ce/Zr-PILC	479	572	694	808	0.255
0.8Ca-2Nb4Ce/Zr-PILC	480	570	692	800	0.262
0.8Mg-2Nb4Ce/Zr-PILC	489	580	706	812	0.265

. H₂ consumption of the K-poisoned sample decreased significantly (from 0.390 mmol/g for the fresh sample to 0.236 mmol/g for the 0.8 K-2Nb4Ce/Zr-PILC sample). This result indicates that there was a synergistic effect between Ce and Nb, which decreased amount of the reducible species on the surface of 0.8 K-2Nb4Ce/Zr-PILC. Similar results were also obtained for the other samples after poisoning of Na, Ca, and Mg, in which all of the reduction peaks were shifted slightly to higher temperatures, and H₂ consumption decreased significantly. For example, H₂ consumption decreased to 0.255, 0.262, and 0.265 mmol/g for the 2Nb4Ce/Zr-PILC samples doped with 0.8 Na, 0.8 Ca, and 0.8 Mg, respectively. The results show that doping of the alkali metals to 2Nb4Ce/Zr-PILC could lead to stabilization of the active component and make it more difficult to be reduced. In addition, it is also revealed that effects of K- and Na-poisoning on redox ability of the fresh samples were more serious than those of Mg- and Ca-poisoning. Redox ability of these samples was also in good agreement with their SCR activities, which indicates that redox ability of the samples played an important role in the SCR reaction.

2.6. Surface Elemental Composition

In order to further explore the chemical valence distributions and surface composition changes on the surface of the samples after M-poisoning, the fresh and M-poisoned samples were analyzed using the XPS technique, and their Ce 3d and O 1s XPS spectra are shown in Figure 7 and Figure 8, respectively.

As shown in Figure 7, the peaks denoted as u, u^II, u^III, v, v^II, and v^III represented the 3d¹⁰4f⁰ state of the surface Ce⁴⁺ species, while those denoted as u^I and v^I represented the 3d¹⁰4f¹ initial electronic state of the surface Ce³⁺ species [24,25]. It can be seen from the XPS spectra that the Ce³⁺ and Ce⁴⁺ species co-existed on the sample surface, indicating that Ce was not completely oxidized. According to the calculated results using the relative areas of the corresponding peaks (

Table 3. Surface element compositions of the samples.

Sample	Composition of Cerium Species (at%)			Composition of Oxygen Species (at%)
	Ce3+	Ce4+	Ce3+/Ce4+ Atomic Ratio	Oα	Oβ	Oγ
2Nb4Ce/Zr-PILC	45.1	54.9	0.82	12.5	42.3	45.2
0.3K-2Nb4Ce/Zr-PILC	33.4	66.6	0.50	12.4	28.3	59.3
0.8K-2Nb4Ce/Zr-PILC	29.0	71.0	0.41	13.1	22.6	64.3
0.8Na-2Nb4Ce/Zr-PILC	33.7	66.3	0.51	12.8	25.1	62.1
0.8Ca-2Nb4Ce/Zr-PILC	35.8	64.2	0.56	14.7	31.2	54.1
0.8Mg-2Nb4Ce/Zr-PILC	37.0	63.0	0.59	12.7	36.2	51.1

), the Ce³⁺/Ce⁴⁺ atomic ratio dropped in the order of 2Nb4Ce/Zr-PILC (0.82) > 0.8 Mg-2Nb4Ce/Zr-PILC (0.59) > 0.8 Ca-2Nb4Ce/Zr- PILC (0.56) > 0.8 Na-2Nb4Ce/Zr-PILC (0.51) > 0.8 K-2Nb4Ce/Zr-PILC (0.41). For the K-poisoned samples, with the rise in K doping, the Ce³⁺/Ce⁴⁺ atomic ratio decreased sharply in the order of 2Nb4Ce/Zr-PILC (0.82) > 0.3 K-2Nb4Ce/Zr-PILC (0.50) > 0.8 K-2Nb4Ce/Zr-PILC (0.41). After doping of the M to 2Nb4Ce/Zr-PILC, the Ce³⁺/Ce⁴⁺ atomic ratio also decreased, especially on the Na- and K-doped samples. This result indicates that poisoning of the M led to a decrease in amount of the Ce³⁺ species and an increase in amount of the Ce⁴⁺ species on the sample. It has been reported that the Ce³⁺ species originated from structural defects of CeO₂ and were accompanied by formation of oxygen vacancies. A higher Ce³⁺/Ce⁴⁺ atomic ratio could bring about charge imbalance and unsaturated chemical bond, which increased amount of the chemisorbed oxygen species on the sample surface, thereby promoting the repeatable Ce³⁺/Ce⁴⁺ redox cycles [25]. In contrast, the redox cycle of Ce³⁺/Ce⁴⁺ was greatly inhibited once the Ce³⁺ species disappeared. Therefore, the change in Ce³⁺ species concentration on the surface might be one of the reasons affecting the SCR activity.

Figure 8 shows O 1s XPS spectra of the fresh and M-poisoned samples. Each spectrum could be decomposed into three components. The component at binding energy (BE) = 530.1–530.3 eV was assigned to the surface lattice oxygen (labeled as O_α) species, the one at BE = 531.3–531.8 eV was attributed to the surface chemisorbed oxygen (labeled as O_β) species, and the strong and broad one at BE = 532.3–532.6 eV was ascribed to the surface oxygen species in the Si−O bonds of SiO₂ (labeled as O_γ) [26]. It has been widely reported that the surface chemisorbed oxygen (O_β) exhibited a high activity in the SCR reaction since it was more mobile than lattice oxygen [2,24] and the lower concentration of the former would lead to a less reactivity. The relative concentrations of three oxygen species on the sample surface were estimated from area ratios of the corresponding characteristic peaks, and the results are listed in Table 3. The order in amount of the O_β species decreased in the sequence of 2Nb4Ce/Zr-PILC (42.3%) > 0.8 Mg-2Nb4Ce/Zr-PILC (36.2%) > 0.8 Ca-2Nb4Ce/Zr-PILC (31.2%) > 0.8 Na-2Nb4Ce/Zr-PILC (25.1%) > 0.8 K-2Nb4Ce/Zr-PILC (22.6%), which was in accordance with their SCR performance and amounts of oxygen vacancies. For the K-poisoned samples, with the rise in K concentration, the atomic ratio of the O_β species decreased sharply in the order of 2Nb4Ce/Zr-PILC (42.3%) > 0.3 K-2Nb4Ce/Zr-PILC (28.3%) > 0.8 K-2Nb4Ce/Zr-PILC (22.6%). According to the literature, the change in O 1s signal after M-doping could be explained as formation of the strong bond between the doped M and surface oxygen center, which made reducibility of the surface species decrease (H₂-TPR results) [2]. Therefore, doping of the M induced an inhibition effect on formation of the chemisorbed oxygen species.

2.7. Surface Acidity

2.7.1. NH₃-TPD

The adsorption capacity of NH₃ exerted an important effect on NH₃-SCR performance, which was strongly related to surface acidity of a catalyst [27,28]. In order to investigate effect of the alkali metal- and alkaline earth metal-doping on surface acidity of 2Nb4Ce/Zr-PILC, NH₃-TPD experiments were performed and the results are illustrated in Figure 9. It can be seen that NH₃ could be desorbed at 180, 230, and 300 °C, respectively. According to the literature [27], the three desorption peaks of each sample were attributed to desorption of the physisorbed ammonia and some NH₄⁺ bounded to the weak Brønsted acid sites (weak acid sites), the NH₄⁺ from the strong Brønsted acid sites (medium acid sites), and the coordinated NH₃ from the Lewis acid sites (strong acid sites), respectively. The corresponding desorption peak temperatures and acidity are listed in

Table 4. NH₃ desorption temperatures and NH₃ desorption amounts from the samples.

Sample	Temperature (°C)			Acidity (mmolNH3/g)			Total Desorption Amount (mmol/g)
	Weak Peak	Medium Peak	Strong Peak	Weak Peak	Medium Peak	Strong Peak
2Nb4Ce/Zr-PILC	175	218	287	0.048	0.092	0.119	0.259
0.3K-2Nb4Ce/Zr-PILC	184	229	306	0.023	0.062	0.064	0.149
0.8K-2Nb4Ce/Zr-PILC	181	225	293	0.030	0.046	0.054	0.130
0.8Na-2Nb4Ce/Zr-PILC	183	229	295	0.033	0.057	0.069	0.159
0.8Ca-2Nb4Ce/Zr-PILC	182	228	303	0.032	0.065	0.080	0.177
0.8Mg-2Nb4Ce/Zr-PILC	180	227	304	0.037	0.071	0.074	0.182

. For the M-poisoned samples, the desorption amount of NH₃ decreased, as compared with that of the fresh sample. It is well known that the Brønsted acid sites were favorable for the adsorption of NH₃, thus improving the low-temperature SCR activity [28]. The order in amount of the Brønsted acid sites was as follows: 2Nb4Ce/Zr-PILC (0.092 mmol/g) > 0.8 Mg-2Nb4Ce/Zr-PILC (0.071 mmol/g) > 0.8 Ca-2Nb4Ce/Zr-PILC (0.065 mmol/g) > 0.8 Na-2Nb4Ce/Zr-PILC (0.057 mmol/g) > 0.8 K-2Nb4Ce/Zr-PILC (0.046 mmol/g); and the total acid amount decreased in the sequence of 2Nb4Ce/Zr-PILC (0.259 mmol/g) > 0.8 Mg-2Nb4Ce/Zr-PILC (0.182 mmol/g) > 0.8 Ca-2Nb4Ce/Zr-PILC (0.177 mmol/g) > 0.8 Na-2Nb4Ce/Zr-PILC (0.159 mmol/g) > 0.8 K-2Nb4Ce/Zr-PILC (0.130 mmol/g). The NH₃-SCR activity of the M-poisoned samples decreased in the following order: 0.8 K-2Nb4Ce/Zr-PILC > 0.8 Na-2Nb4Ce/Zr-PILC > 0.8 Ca-2Nb4Ce/Zr-PILC > 0.8 Mg-2Nb4Ce/Zr-PILC, which was consistent with those in amount of the Brønsted acid sites and total acid amount of the acid sites. It is worth noting that NH₃ desorption amount of the 0.8 K-2Nb4Ce/Zr-PILC sample was much lower than those of the other samples in the whole temperature range. This result indicates that K-doping induced a more negative effect on surface acidity of 2Nb4Ce/Zr-PILC, in good consistency with its poor SCR performance.

In addition, in the case of different K doping to the fresh sample, ammonia desorption amount also decreased to some extent. The calculated ammonia desorption amount of the 2Nb4Ce/Zr-PILC, 0.3 K-2Nb4Ce/Zr-PILC, and 0.8 K-2Nb4Ce/Zr-PILC samples were 0.259, 0.149, and 0.130 mmol/g, respectively, indicating that the increase in K concentration made ammonia adsorption ability of the sample decrease sharply. Therefore, M-doping could reduce the adsorption of ammonia, which was confirmed by the results obtained in the following in situ DRIFT experiments.

2.7.2. DRIFTS Study of NH₃ Adsorption

The in situ DRIFTS spectra were used to investigate the NH₃ species adsorbed on the sample surface to further distinguish the different acid types [29,30,31,32]. Figure 10 shows the in situ DRIFTS spectra of the adsorbed NH₃ species on the fresh and M-poisoned samples. The absorption bands at 1685 and 1440 cm⁻¹ were attributed to the NH₄⁺ coordinated to the Brønsted acid sites [22,29], the ones at 1180, 1250, and 1601 cm⁻¹ were assigned to the coordinated NH₃ linked to the Lewis acid sites, the ones at 3360 and 3265 cm⁻¹ were attributed to the N–H stretching vibration modes of the coordinated NH₃, and the one at 965 cm⁻¹ was ascribed to the gas-phase or weakly adsorbed NH₃ [30]. The above results indicate that both the Lewis acid sites and the Brønsted acid sites existed on the sample surface, which was consistent with the NH₃-TPD results. As shown in Figure 10A, after doping the M to 2Nb4Ce/Zr-PILC, both the NH₃ coordinated to the Lewis acid sites (1601, 1250, and 1180 cm⁻¹) and the NH₄⁺ coordinated to the Brønsted acid sites (1685 and 1440 cm⁻¹) were also detected on all of the M-poisoned samples. Nevertheless, the adsorption of NH₄⁺ and NH₃ on the M-poisoned samples also showed a different degree of decline in band intensity. This result shows that all of the doped M exhibited a strong inhibitory effect on NH₃ adsorption at the acid sites, especially at the Brønsted acid sites. The order of adsorption strength at the Brønsted acid sites was as follows: 2Nb4Ce/Zr-PILC > 0.8 Mg-2Nb4Ce/Zr-PILC > 0.8 Ca-2Nb4Ce/Zr-PILC > 0.8 Na-2Nb4Ce/Zr-PILC > 0.8 K-2Nb4Ce/Zr-PILC, which agreed well with the NH₃-TPD results. Moreover, as shown in Figure 10B, the absorption band intensity decreased in the order of 2Nb4Ce/Zr-PILC > 0.3 K-2Nb4Ce/Zr-PILC > 0.8 K-2Nb4Ce/Zr-PILC, suggesting that the acid sites of the sample also decreased with a rise in K content. Compared with the other poisoned samples, the absorption band at the acid sites of the 0.8 K-2Nb4Ce/Zr-PILC sample was the weakest. This result indicates that the acid sites on the surface of the K-doped sample were the most seriously destroyed (in consistency with the above TPD results), resulting in the lowest deNO_x activity.

3. Discussion

The poisoning effects of alkali metals (K, Na) and alkaline earth metals (Ca, Mg) on catalytic performance of the 2Nb4Ce/Zr-PILC sample were investigated. The poisoning extent of the M-doping decreased in the order of K > Na > Ca > Mg, and the results show that the poisoning effect increased proportionally with the alkalinity of the M. It is widely accepted that a higher Ce³⁺/Ce⁴⁺ atomic ratio can cause charge imbalance and generate unsaturated chemical bonds, which would lead to an increase in amount of the chemisorbed oxygen species on the sample surface, thus promoting the redox cycle of Ce³⁺/Ce⁴⁺. The XPS results (Figure 7) verify the existence of Ce³⁺ in the 2Nb4Ce/Zr-PILC sample and the Ce³⁺ content decreased in the M-poisoned samples. Under the actual preparation conditions, however, the nitrate of potassium, sodium, calcium or magnesium could form a molten salt flux on the surface of 2Nb4Ce/Zr-PILC, giving rise to the covering of the active sites [3]. Alkali and alkaline earth metal oxides with strong alkalinity could occupy the surface of 2Nb4Ce/Zr-PILC, thus reducing the amount of the chemisorbed oxygen species and the reducibility of cerium species. It can be found that the reduction peak was shifted to a higher temperature and the hydrogen consumption decreased (Figure 6) with the doping of the M. This result was a piece of important evidence that the reducibility of cerium species decreased. Surface acidity was another main factor suppressing SCR performance of the M-poisoned sample [4,33,34]. It is generally believed that NH₃ was firstly adsorbed at the acid sites via hydrogen abstraction or protonation, thereby forming the active NH₃ species [34], and then the active NH₃ species reacted with the adsorbed nitrate/nitrite intermediates (Langmuir−Hinshelwood mechanism) or gas-phase NO/NO₂ (Eley−Rideal mechanism) to produce H₂O and N₂. If the surface acidity was inhibited, the first step would become the rate-determining step and hence significantly decrease the reaction rate [33,34]. Therefore, the decrease in surface acidity of the M-poisoned sample (Figure 9 and Figure 10) exerted a negative impact on the SCR activity of 2Nb4Ce/Zr-PILC.

Therefore, the decreases in NH₃ adsorption, Ce³⁺/Ce⁴⁺ redox cycle (surface chemisorbed oxygen), and cerium species reducibility might be the main reasons why the 2Nb4Ce/Zr-PILC sample was poisoned by the M. Based on the above discussion, a deactivation mechanism of the M-doped 2Nb4Ce/Zr-PILC samples was proposed, as shown in Figure 11.

4. Experimental

4.1. Catalyst Preparation

The acid-leached montmorillonite was dispersed in deionized water at room temperature under stirring for 24 h, then mixed with the zirconium oxychloride aqueous solution. Afterwards, the mixed aqueous solution was stirred for 12 h and aged at room temperature for 12 h, and the product was finally dried at 100 °C for 12 h after being washed with deionized water three times and calcined at 400 °C for 2 h, thus obtaining the Zr-incorporated montmorillonite (denoted as Zr-PILC) support. The 2Nb/Zr-PILC sample with a Nb loading of 2 wt% was synthesized by impregnating Zr-PILC with a niobium oxalate aqueous solution, followed by stirring for 1 h, drying with a rotary evaporator, and calcining in air at 400 °C for 2 h. The 2Nb4Ce/Zr-PILC sample was prepared by impregnating 2Nb/Zr-PILC with a cerium nitrate aqueous solution. The preparation method was the same as that described above.

The alkali metal-doped 2Nb4Ce/Zr-PILC samples with different M/Ce molar ratios (labeled as x M, M = Na, K, Mg, and Ca; x = M/Ce molar ratio) were prepared by impregnating the 2Nb4Ce/Zr-PILC with KNO₃, NaNO₃, Mg(NO₃)₂ and Ca(NO₃)₂ aqueous solutions, respectively. The subsequent procedures were the same those stated above. The obtained samples were denoted as 0.8 M-2Nb4Ce/Zr-PILC.

4.2. Catalyst Characterization

The X-ray diffraction measurement was performed on a Bruker D8 advance diffractometer (Bruker, Karlsruhe, Germany) equipped with a Cu Kα irradiation. N₂ adsorption−desorption isotherms were determined at −196 °C using a W-BK132F apparatus (JWGB, Beijing, China). X-ray photoelectron spectroscopy (XPS) was performed using a Thermo ESCALAB 250XI (Thermo Fisher, Waltham, MA, USA) to analyze the surface element compositions and metal chemical states of the samples, in which the operating pass energy (PE) was 50 eV with an X-ray source of 200 W and an Al Kα radiation (hv = 1486.6 eV). Before the analysis, all of the samples were degassed in vacuum to eliminate influence of the adsorbed gas on the surface of the samples. The binding energy of each XPS spectrum was calibrated against that (284.6 eV) of the standard C 1s signal of the contaminant carbon. The CASA XPS software was used to make the peak fitting of the XPS spectra and the background signals were deducted before analysis. Hydrogen temperature-programmed reduction of (H₂-TPR) and ammonia temperature-programmed desorption (NH₃-TPD) experiments were conducted on a PCA-1200 analyzer (Beijing Builder Electronic Technology, Beijing, China) equipped with a TCD detector using 100 mg of the sample. Before the H₂-TPR experiment, the sample was pretreated in a 5 vol% O₂/N₂ flow of 30 mL/min at 400 °C for 1 h and cooled to room temperature (RT). Then, the atmosphere was switched to a 5 vol% H₂/N₂ flow of 30 mL/min, and the temperature was raised to 950 °C at a heating rate of 10 °C/min. Prior to the NH₃-TPD experiment, the sample was pretreated in a He flow of 30 mL/min at 400 °C for 1 h, and then a NH₃ flow of 30 mL/min was passed through the sample for adsorption at 100 °C for 1 h. Subsequently, the sample was purged in a He flow of 30 mL/min for 30 min to remove the physically adsorbed NH₃. Finally, NH₃ desorption was carried out from 30 to 850 °C at a heating rate of 10 °C/min. The in situ diffuse reflectance Fourier transform infrared spectroscopic (in situ DRIFTS) experiments were performed on a Bruker 0 spectrometer (Bruker, Karlsruhe, Germany). The sample was pretreated at 400 °C for 1 h to remove the moisture. The background spectrum was recorded by exposing the sample to a N₂ flow of 100 mL/min, and switched to a (1100 ppm NH₃ + N₂ (balance)) flow of 100 mL/min to record the NH₃ absorption spectrum.

4.3. NH₃-SCR and NH₃ Oxidation

Catalytic activity tests of the samples for the NH₃-SCR reaction were carried out in a fixed-bed quartz tubular microreactor using 300 mg of the sample and 600 mg of quartz sand. The simulated flue gas consisted of (1100 ppm NH₃ + 1000 ppm NO + 4 vol% O₂ + 5 vol% H₂O + N₂ (balance)) and the total flow rate was 500 mL/min, giving a space velocity (SV) of 100,000 mL/(g h). The concentration of NO was measured by a MODEL1080 analyser (Beijing SDL Technology, Beijing, China). NO conversion was calculated according to the following equation:

$$\mathrm{NO}\mathrm{conversion}(\%)=\frac{{\left[\mathrm{NO}\right]}_{\mathrm{in}}-{\left[\mathrm{NO}\right]}_{\mathrm{out}}}{{\left[\mathrm{NO}\right]}_{\mathrm{in}}}\times 100\%$$

NH₃ oxidation activity was evaluated in a flow-through microreactor system equipped with a Fourier transform infrared spectroscopy (FT-IR) spectrometer. The total flow rate of the reactant mixture (1100 ppm NH₃ + 4 vol% O₂ + N₂ (balance)) was 500 mL/min and the SV was 100,000 mL/(g h). NH₃ conversion was calculated according to the following equation:

$${\mathrm{NH}}_3\mathrm{conversion}=\frac{{[{\mathrm{NH}}_3]}_{\mathrm{in}}-{[{\mathrm{NH}}_3]}_{\mathrm{out}}}{{[{\mathrm{NH}}_3]}_{\mathrm{in}}}\times 100\%$$

5. Conclusions

In this study, the deactivation of the 2Nb4Ce/Zr-PILC sample by the alkali and alkaline earth metals was investigated in detail. Doping of the M (M = K, Na, Ca, and Mg) deactivated the 2Nb4Ce/Zr-PILC sample, and their deactivation effects decreased in the sequence of K > Na > Ca > Mg. Through a series of characterization, it is shown that the M doping induced the decreases in NH₃ adsorption (surface acidity), Ce³⁺/Ce⁴⁺ atomic ratio (chemisorbed oxygen (O_β)), and redox ability (redox sites), which were accountable for such a deactivation.