An important issue in selective catalytic reduction of NO_x by NH₃ (NH₃-SCR) is the deactivation effects of alkali metals, alkaline-earth metals, and halogens on SCR catalysts [1-5]. Commercial SCR catalysts can be significantly deactivated by alkali and alkaline-earth salts obtained from fly ash. The effects of alkali and alkaline-earth metals on V₂O₅-WO₃/TiO₂ catalysts have been studied extensively [3, 6-9]. It was believed that the surface acidity of SCR catalysts decreased significantly upon K, Na, and Ca poisoning [4, 8, 10]. Other researchers proposed that suppression of reducibility was more important in the deactivation of V₂O₅/TiO₂ catalysts by Na⁺ and Ca²⁺ ions [11]. It was revealed that the strength of the V=O bond was reduced and the oxidizing capability of the V-based catalyst was weakened by loading of alkali metal on the catalyst surface [12]. Du et al. [13] investigated the effect of alkali metal species on SCR catalysts using theoretical and experimental methods, and found that the poisoning effect of Na⁺ depended on the type of anion used. The SCR catalyst could be also deactivated due to pore blocking by alkali and calcium sulfate deposition [14].

Although the poisoning of SCR catalysts by alkali metals has been extensively researched, there are still many issues to be studied. For instance, the deactivation mechanism of Ca on SCR catalyst is controversial. Yang et al. [15] studied the adsorption of NH₃ on CaO and sulfated CaO and found that NH₃ could be dissociated to produce NH₂ on both the CaO and sulfated CaO catalysts. The intermediate NH₂ could then react with surface oxygen to form NO on CaO, and possibly reduce NO over the CaSO₄ catalyst. It was reported that the reducibility of vanadium species and the intensity of surface acidity was weakened by CaO and CaCO₃, whereas the concentrations of the surface-active oxygen and the Brnsted acid sites were improved by loading of CaSO₄ on the SCR catalyst [9]. Furthermore, it should be emphasized that calcium salts have the strongest Lewis acidity among the alkaline-earth metals [16].

Halogens (especially Cl and Br) play important roles in the removal of NO_x and Hg⁰ by SCR catalysts [17]. The addition of calcium bromide to the fuel is believed to be an effective means for oxidizing Hg⁰ in coal-fired boilers [18]. CaBr₂ injection is widely applied to improve Hg⁰ oxidation and capture in power plants [19-21]. Bromides are thought to donate the most electron density to metals. An increase in electron density at the metal center facilitates oxidative processes [22, 23]. Although the emergence of quantitative acid sites (which are very important in the NH₃-SCR reaction) could be observed upon Cl addition, the deactivation effects of Cl on SCR catalysts are serious and irreversible [2, 24]. In spite of these observations, the correlation between bromides and SCR catalysts has been ignored. The effects of bromides on the catalytic performance and structure of SCR catalysts is not clear.

In this study, the commercial V₂O₅-WO₃/TiO₂ catalyst was poisoned by different bromides (NH₄Br, NaBr, KBr and CaBr₂). The fresh and poisoned catalysts were studied by means of Brunauer-Emmett-Teller (BET) surface area analysis, X-ray photoelectron spectroscopy (XPS), hydrogen temperature-programmed reduction (H₂-TPR), oxygen temperature-programmed oxidation (O₂-TPO), and carbon dioxide temperature-programmed desorption (CO₂-TPD), aiming at a better understanding of the effects of bromide-poisoning on the structure and NH₃-SCR activity of the commercial SCR system.

The commercial V₂O₅-WO₃/TiO₂ catalyst (hereafter denoted as L catalyst) was obtained from China Guodian Corporation. The bromide-poisoned catalysts were prepared by impregnating the L catalyst with aqueous solutions of NH₄Br, NaBr, KBr and CaBr₂ (the mass of loaded bromides was equal). Then the samples were dried at 110 ℃ and calcined at 450 ℃ for 4 h (except for the NH₄Br-poisoned sample, because the decomposition temperature of NH₄Br is less than 450 ℃). The poisoned samples were denoted as L-NHBr, L-NaBr, L-KBr and L-CaBr, respectively.

The steady-state NH₃-SCR performance and NH₃ oxidation over the catalysts were evaluated in a fixed-bed quartz reactor. The reaction conditions were as follows: 0.15 g catalyst, 500 ppm NO, 500 ppm NH₃, 5% O₂, and N₂ balance with a total gas flow rate of 200 mL/min, gas hourly space velocity (GHSV) = 7.0 × 10⁴ h^–1 (STP). The temperature-programmed desorption of NH₃ (NH₃-TPD) and NO_x (NO + O₂-TPD) were also carried out on the same system.

The BET specific surface area was estimated by N₂ physisorption at –196 ℃ using a Quantachrome Nova Automated Gas Sorption System. The XPS spectra were obtained using a PHI-5300/ESCA system. The H₂-TPR, O₂-TPO and CO₂-TPD experiments were accomplished by means of a Micromeritics AutoChem Ⅱ 2920 device. In a typical O₂-TPO experiment, about 50 mg of catalyst was used. The catalyst was first pretreated with helium (30 mL/min) at 300 ℃ for 1 h, and then exposed to 500 ppm NH₃ at the same temperature for 1 h. After cooling down to room temperature in helium (30 mL/min), the sample was purged by O₂ until the baseline was stable. Finally, the sample was heated to 1000 ℃ at a rate of 10 ℃ mL/min in a specific O₂ content. The procedures for NH₃-TPD, H₂-TPR and CO₂-TPD were similar to those used in our previous work [2].

In order to study the effect of bromides on the SCR catalyst, the commercial L catalyst was sampled and impregnated by four kinds of Br-containing salts (NH₄Br, NaBr, KBr, and CaBr₂). The effects of bromides on SCR activity are shown in Fig. 1.

The addition of NH₄Br had a negligible effect on SCR activity over the L catalyst. At 400–450 ℃, the NO_x conversion over L-NaBr and L-CaBr decreased to a certain extent compared with the fresh sample. However, the NO_x conversion decreased significantly over L-KBr catalyst in the test temperature range (especially at high temperature). At 450 ℃, only 24% of NO_x conversion could be obtained over the L-KBr catalyst. No change in N₂ selectivity was observed over the NH₄Br-poisoned catalyst compared to the L catalyst (see Fig. 1(b)). The N₂ selectivity at high temperatures decreased over L-NaBr, L-KBr and L-CaBr catalysts. In the case of the L-KBr catalyst, the N₂ selectivity was much lower than that of the fresh sample at 450 ℃. This indicated that the poisoning effects of Br-containing salts on the SCR activity were very different due to the various kinds of cations. The results demonstrated that the counter-anions played a crucial role in determining the activity of bromide-poisoned catalysts. For discussing the poisoning effect of bromides on structure and activity, the fresh and poisoned L catalysts were characterized by means of BET, XPS, TPR, and TPD.

The addition of bromides influences the surface physicochemical properties and consequently affects the catalytic activity. The specific BET surface area was tested and the results are shown in Table 1. It is shown that the addition of Br-containing salts had a negligible effect on the specific surface area. The biggest change was found in the L-NHBr sample, where the BET surface area increased to 47.9 m²/g (from 46.0 of m²/g of the L catalyst).

Surface-chemisorbed oxygen was believed to be the most active oxygen species, which contributes to excellent SCR activity [3]. XPS analysis was carried out and the results of O 1s analysis of bromide-poisoned samples are shown in Table 1 and Fig. 2. Generally, the O 1s peaks could be fitted into three bands. The sub-band at ~533.2 eV was assigned to H₂O on the catalyst surface (hereafter, denoted as O_w) [25]. The sub-band at 531.5–532 eV was attributed to the surface-chemisorbed oxygen (such as O₂²⁻ pertaining to defect-oxide or hydroxyl oxygen, hereafter, denoted as O_α). The sub-band at ~530.4 eV was attributed to the lattice oxygen O²⁻ (hereafter, denoted as O_β) [26]. Three kinds of oxygen existed on the fresh L catalyst. Compared with the fresh sample, the O_α/(O_α + O_β + O_w) ratio increased over the L-CaBr catalyst. The increasing ratio of O_α indicated the formation of new surface-chemisorbed oxygen due to the addition of CaBr₂. On the contrary, the ratio decreased over other catalysts, especially the L-NaBr (22.6%) and L-KBr (22.8%) catalysts. The decrease of surface-chemisorbed oxygen might be ascribable to the overlay of active sites by Br-containing salts. The lower ratio of O_α/(O_α + O_β + O_w) is likely to contribute to poor reducibility and ultimately affect the SCR activity. In the literature, the decrease in the ratio of O_α after impregnation of alkali metals was attributed to the formation of strong bonds between surface oxygen and the doped metal [3]. The order of O_α/(O_α + O_β + O_w) ratios was as follows: L-CaBr > L > L-NHBr > L-KBr > L-NaBr, and the trend was almost consistent with that of SCR activity. The low concentration of O_α might result in a less reactive oxidation procedure, which would diminish the SCR activity [3].

The oxidation state of vanadium was also acquired and the V 2p XPS spectra for different catalysts are presented in Fig. 3. The presence of two oxidation states (V⁵⁺, V⁴⁺) for fresh and poisoned catalysts was established by peaks at 517.1–517.6 and 516.4–516.7 eV. V⁴⁺ and V⁵⁺ appeared on all of catalysts and the main contribution corresponded to V⁴⁺ species. The ratios of V⁴⁺ / (V⁴⁺ + V⁵⁺) are listed in Table 1. It is noteworthy that the V⁴⁺/(V⁴⁺ + V⁵⁺) ratio of the L-CaBr and L-NaBr catalysts increased dramatically compared to that of the L catalyst. This revealed that the addition of CaBr and NaBr might alter the electronic density distribution around V. According to literature, the increased amount of V⁴⁺ would favor the SCR reaction due to the redox cycle of V⁵⁺/V⁴⁺ [27]. It was also reported that the formation of V⁴⁺ was promoted by fluorine doping due to charge compensation, which enhanced the amount of surface superoxide ions correspondingly [28].

In order to investigate the effect of bromides on the reducibility of L catalyst, H₂-TPR was carried out with different catalysts and the resultant profiles are shown in Fig. 4. For the fresh L catalyst, two main broad peaks were found at 586 and 857 ℃, which could be attributed to the reduction of V⁵⁺→V³⁺ and W⁶⁺→W⁰, respectively [3, 7]. The impregnation of NH₄Br and CaBr₂ enhanced the reducibility of L catalyst and both peaks shifted to lower temperatures, especially for the L-CaBr catalyst, where the two peaks were found at 538 and 773 ℃. The relatively low reduction temperature on L-CaBr is possibly due to the strong interaction between CaBr₂ and the catalyst surface. The superior reducibility might be correlated with the high ratio of O_α and V⁴⁺ on CaBr₂-poisoned catalysts. An improvement in the reducibility on Ca-doped catalyst due to enhanced oxygen mobility was reported by other researchers [29]. It was also reported that the addition of Ca could improve the reducibility of Pt oxides supported on Al₂O₃ by reducing the Pt-Al interaction [30]. However, it was confirmed that the addition of CaSO₄ resulted in more N₂O production in the SCR reaction due to its excellent reducibility [9].

For the L-NaBr and L-KBr catalysts, the peaks ascribed to the reduction of V⁵⁺→V³⁺ and W⁶⁺→W⁰ also appeared but the intensity of all peaks decreased. More importantly, the peaks corresponding to V⁵⁺ reduction shifted to higher temperatures, and the peaks ascribed to W⁶⁺ reduction shifted to lower temperatures, for both catalysts. Similar changes were also observed after addition of Na and K to V-W/Ti catalysts. Chen et al. [3] attributed this to the diminution in the reduction degree of V species and the influence of the chemical environment around W species by Na, K doping. The TPR results displayed that the reducibility of surface species decreased over L-KBr and L-NaBr catalysts.

According to the literature, regeneration of the V⁵⁺=O species could be achieved through two steps, which involve the generation of V⁴⁺ and oxidation of the reduced species by gaseous O₂ [31]. For the purpose of investigating the redox properties of bromide-poisoned L catalysts, the samples were pretreated in NH₃ and followed by the O₂-TPO experiment.

As shown in Fig. 5, a small peak at 393 ℃ and a strong peak at 752 ℃ were found on the L catalyst. The low-temperature peak could be attributed to reoxidation of surface V (such as V⁴⁺) after NH₃ reduction. The peaks shifted to lower temperatures after poisoning by KBr. On the L-CaBr catalyst, a small peak was found at 450 ℃. The reoxidation temperature shifts to a higher value than the corresponding temperature on L catalyst. Thus, the reoxidation of active sites was retarded by CaBr₂ addition, which meant that the stability of active sites was improved upon modification. Combining the H₂-TPR and O₂-TPO data, it could be concluded that the redox properties were affected by addition of bromides. The reducibility of surface V species decreased over L-KBr and L-NaBr catalysts, while the reoxidation ability was retarded on the L-CaBr catalyst. For commercial V-W/Ti catalyst, the redox cycle of V⁵⁺ and V⁴⁺ (V⁵⁺ ↔ V⁴⁺) plays an important role in the NH₃-SCR reaction. Not only the reducibility, but also the ability of re-oxidation, is important for NH₃-SCR. The results indicated that the active sites (such as V⁵⁺) might be reduced easily; however, the re-oxidation of V⁴⁺ was difficult after CaBr₂ addition. This may be an important reason for the lower activity of these catalysts.

One of the most significant steps in the NH₃-SCR reaction is the NH₃ adsorption on the surface acid sites of the catalysts. The surface acidity of bromide-poisoned catalysts was characterized by NH₃-TPD. As shown in Fig. 6, all curves present a wide NH₃ desorption peak at 150 – 350 ℃. After poisoning by CaBr₂, the peak shifted to a higher temperature and the amount of desorbed NH₃ increased significantly compared to that of fresh L sample. This indicated the appearance of new acid sites and enhancement of the surface acidity of L-CaBr catalyst. The amounts of desorbed NH₃ were calculated from NH₃-TPD and the results are displayed in Table 1. The capacity of L-CaBr for NH₃ adsorption was about 1.5-times that of the fresh L sample. For the L-NHBr and L-NaBr-loaded catalysts, the amount of desorbed NH₃ decreased at different degrees compared to that of the fresh L catalyst. It is worth noting that the amount of desorbed NH₃ was much lower over L-KBr catalyst (~39% of acidity of L catalyst), and the peak shifted to lower temperatures, which demonstrated that the surface acidity was weakened by KBr [3]. In the literature, alkali poisoning was studied by a density functional theory-based approach and the results indicated that the decrease in surface acidity is due to electron transfer from the K atom to the terminal O atom [7]. These results reveal that the adsorption of NH₃ on SCR catalyst is facilitated after CaBr₂ poisoning, while the NH₃ adsorption is restrained on other catalysts. However, strong acidity might affect selective catalytic oxidation of NH₃ over L-CaBr catalyst. This will be discussed later.

Meanwhile, the NO adsorption capacity of bromide-poisoned catalysts was characterized by NO + O₂-TPD and the results are also displayed in Table 2 and Fig. 7. It is shown that small amounts of NO and NO₂ were desorbed on the L catalyst (2.00 μmol/g of NO, 3.24 μmol/g of NO₂). For the L-CaBr catalyst, both NO and NO₂ could be found and the amount of NO₂ increased to 4.60 μmol/g. However, for the L-KBr catalyst, NO₂ could hardly be detected in the test temperature range. The formation of NO₂ was related to the surface oxidized species after NO adsorption. As indicated in H₂-TPR, the reducibility of surface species decreased over L-KBr catalyst. The oxidation of NO adsorbed species would be affected by this catalyst.

The quantity of adsorbed NH₃ on the surface is related to the surface acidity, and the H-abstraction degree of NH₃(ads) is related to the redox properties of SCR catalysts [32]. The catalytic oxidation of NH₃ plays an important role in the NH₃-SCR reaction. In order to discuss the effect of bromide-poisoning on SCR performance, the NH₃ oxidation performance was tested over these catalysts and the results are displayed in Fig. 8. The fresh and bromide-poisoned catalysts exhibited nearly no NH₃ conversion at temperatures lower than 300 ℃. On the fresh L catalyst, the NH₃ conversion rapidly increased to 48% as the temperature increasing to 450 ℃. The NH₃ oxidation performance was improved over L-CaBr catalyst and the NH₃ conversion increased to 58% at 450 ℃. The addition of NH₄Br showed positive effects on NH₃ oxidation and the NH₃ conversion increased at high temperatures over L-NHBr catalyst, whereas the oxidation of NH₃ was restrained over L-KBr catalyst, which was in agreement with the SCR activity at high temperatures.

With respect to the selectivity toward N₂, KBr and CaBr₂ poisoning showed negative effects on the NH₃ oxidation reaction (see Fig. 8(b)). For L-CaBr catalyst, the strong interaction between NH₃ and surface acid sites might affect the selective catalytic oxidation of NH₃. The adsorbed NH₃ species, either in the protonated NH₄⁺ form or in the coordinated NH₃ form, would be partly oxidized by the adsorption sites first [32]. The decrease in N₂ selectivity was possible owing to overoxidation of NH₃ at high temperatures, which is believed to be one important reason for the decline of activity in NH₃-SCR.

CO₂-TPD was performed to characterize the surface basicity of fresh and bromide-poisoned L catalysts. The intensity and distribution of surface basic sites could be characterized by the desorption temperature and the amount of CO₂. The CO₂-TPD profiles are shown in Fig. 9.

The peaks at lower temperature (located in the range of 450–600 ℃) are assigned to –OH basic sites on the catalysts [2]. The peaks at higher temperature (located in the range of 650–850 ℃) could be assigned to basic O^2– attached to metal ions, which exhibited a strong interaction with CO₂ [2]. The fresh L catalyst exhibited two desorption peaks in the test temperature range. The peaks could be assigned to –OH basic sites (at 463 ℃) and strong basic sites O^2– (at 745 ℃). The strength of the basic sites could be reflected by the desorption temperatures of CO₂ [33]. After bromide poisoning, the peaks assigned to –OH basic sites shifted to higher temperatures, suggesting that the intensity of –OH basic sites increased. The halogens are the most electronegative elements. The bromides are supposed to donate the most electron density to the metal [22]. Besides, the Na⁺, K⁺, and Ca²⁺ cations might enhance the electron donation effects on the surface. This might be an important reason for the enhancement of surface basicity of bromide-poisoned catalysts. For the L-KBr catalyst, a new peak appeared at 588 ℃, which implied that an additional basic site might be produced after loading of KBr. It was reported that the surface basicity could be proved by the binding energy (BE) of O1s in the XPS spectra [34]. In this work, the BE value of O 1s decreased to 530.3 on L-NaBr catalyst, indicating that the surface basicity was improved.

The enhancement of surface basicity might affect the adsorption of NH₃ over L-KBr and L-NaBr catalysts. This was in agreement with NH₃-TPD results. The enhancement of basicity should also affect the interaction between NO adsorbed species (such as NO₂^– and NO₃^–) and the catalyst surface. From NO + O₂-TPD, it could be seen that the desorption temperatures of NO increased on the L-KBr catalyst. This demonstrated that the stability of NO adsorbed species was improved. Meanwhile, the amount of adsorbed NO on L-KBr catalyst, and the amount of adsorbed NO₂ on L-CaBr catalyst both increased. The strengthening of basicity by KBr and CaBr₂ was an important factor. On the other hand, the redox property was suppressed significantly on L-KBr catalyst. Thus, the NH₃-SCR catalytic activity decreased, even though more NO could be adsorbed on this catalyst.

In summary, the SCR activity of a commercial catalyst was affected significantly by bromide poisoning. The counter-anions played a crucial role in determining the activity of bromide-poisoned catalysts. The decrease in the ratio of surface chemisorbed oxygen, reducibility, surface acidity, and amount of adsorbed NO_x contributed to the poor activity and N₂ selectivity over the KBr-poisoned catalyst. The inhibition of reoxidation ability, overoxidation of NH₃ and poor N₂ selectivity in NH₃ oxidation retards the SCR activity at high temperature over L-CaBr catalyst. Additional in situ work to investigate the structure-activity relationships over bromide-poisoned catalysts is being carried out in our lab.

Acknowledgments

We thank Dr. Lu from China Guodian Corporation for providing the commercial V₂O₅-WO₃/TiO₂ catalyst. We appreciate the support from Beijing Engineering Research Center of Biogas Centralized Utilization.