Sb-Containing Metal Oxide Catalysts for the Selective Catalytic Reduction of NO x with NH 3

: Sb-containing catalysts (SbZrO x (SbZr), SbCeO x (SbCe), SbCeZrO x (SbCeZr)) were prepared by citric acid method and investigated for the selective catalytic reduction (SCR) of NO x with NH 3 (NH 3 -SCR). SbCeZr outperformed SbZr and SbCe and exhibited the highest activity with 80% NO conversion in the temperature window of 202–422 ◦ C. Meanwhile, it also had good thermal stability and resistance against H 2 O and SO 2 . Various characterization methods, such as XRD, XPS, H 2 -TPR, NH 3 -TPD, and in situ di ﬀ use reﬂectance infrared Fourier transform (DRIFT), were applied to understand their di ﬀ erent behavior in NO x removal. The presence of Sb in the metal oxides led to the di ﬀ erence in acid distribution and redox property, which closely related with the NH 3 adsorption and NO oxidation. Brønsted acid and Lewis acid were evenly distributed on SbCe, while Brønsted acid dominated on SbCeZr. Compared with Brønsted acid, Lewis acid was slightly active in NH 3 -SCR. The competition between NH 3 adsorption and NO oxidation was dependent on SbO x and metal oxides, which were found on SbCe while not on SbCeZr.


Introduction
Nitrogen oxide (NO x ), one of the main environmental contaminants emitted from stationary sources and mobile sources, causes photochemical smog, acid rain, and the depletion of the ozone layer [1,2]. Several technologies have been proposed for NO x removal, such as direct decomposition of NO x , NO x storage reduction (NSR), and the selective catalytic reduction (SCR) process with ammonia (NH 3 -SCR) or hydrocarbons (HC-SCR). The NH 3 -SCR process has been proven to be the efficient method to satisfy the restrict regulations of NO x emissions [3]. In general, the NH 3 -SCR reaction is a competitive process between NH 3 oxidation and NO x reduction, which are closely affected by reaction temperature. The enhanced oxidation activity at low temperature promotes NO x selective reduction, while it also takes a risk for the depletion of NH 3 at high temperature leading to the decrease in NO reduction. Meanwhile, the reaction thermal shock and exhaust components (H 2 O and SO 2 ) put forward strict requirements on the design of NH 3 -SCR catalysts.
CeO 2 possesses unique oxygen storage-release capability by the formation of Ce 3+ /Ce 4+ redox couples [4], but the maximum NO conversion on CeO 2 was about 30% in the whole temperature The apparent activation energy of the NH3-SCR reaction is an important factor to evaluate the role and efficiency of the catalyst in the reaction [25]. By calculating the reaction rate constant of the NH3-SCR reaction over SbCeZr and SbCe, the apparent activation energy of the catalysts were obtained. The reaction activation energy over the SbCeZr catalyst (33.3 KJ/mol) was lower than that of the SbCe catalyst (42.1 KJ/mol), which further confirmed the excellent NH3-SCR reaction performance of SbCeZr.
In order to investigate the thermal stability of SbCeZr and SbCe, the catalysts were aged at 700 °C for 4 h in the air before testing. As shown in Figure 1b, aging had a side effect on catalytic activity, which led to the loss in activity and shrink from the operation temperature window. The presence of Zr improved the thermal resistance of thermal shock. T50 of aged SbCeZr turned out to be 247 °C and the operation temperature window narrowed to 300-413 °C, resulting from the sintering. The maximum NOx conversion over aged SbCe was only 73% at 373 °C, i.e., aging led to the significant deactivation.

Effects of H2O and SO2
SO2 and H2O are unavoidable in the feed gas, so the effect on the catalytic activities over SbCeZr and SbCe were investigated. As shown in Figure 2a, the NH3-SCR performance of SbCeZr was slightly affected by SO2 or H2O at temperatures below 220 °C. However, the presence of H2O significantly reduced the catalytic activity at temperatures above 420 °C, which could be attributed to the competitive adsorption between H2O and NH3 [26,27]. In the presence of O2 and SO2, the deposited sulfates on the catalysts led to the decrease of the specific surface area (SSA) [28] and blocked the active sites at low temperature [29]. With increased temperature, the decomposition of sulfates led to the restoration of active sites [30]. The presence of H2O facilitated SO2 oxidation at low temperature and inhibited sulfate decomposition at high temperature. Therefore, the presence of H2O and SO2 had a more detrimental effect than H2O or SO2. The apparent activation energy of the NH 3 -SCR reaction is an important factor to evaluate the role and efficiency of the catalyst in the reaction [25]. By calculating the reaction rate constant of the NH 3 -SCR reaction over SbCeZr and SbCe, the apparent activation energy of the catalysts were obtained. The reaction activation energy over the SbCeZr catalyst (33.3 kJ/mol) was lower than that of the SbCe catalyst (42.1 kJ/mol), which further confirmed the excellent NH 3 -SCR reaction performance of SbCeZr.
In order to investigate the thermal stability of SbCeZr and SbCe, the catalysts were aged at 700 • C for 4 h in the air before testing. As shown in Figure 1b, aging had a side effect on catalytic activity, which led to the loss in activity and shrink from the operation temperature window. The presence of Zr improved the thermal resistance of thermal shock. T 50 of aged SbCeZr turned out to be 247 • C and the operation temperature window narrowed to 300-413 • C, resulting from the sintering. The maximum NO x conversion over aged SbCe was only 73% at 373 • C, i.e., aging led to the significant deactivation.

Effects of H 2 O and SO 2
SO 2 and H 2 O are unavoidable in the feed gas, so the effect on the catalytic activities over SbCeZr and SbCe were investigated. As shown in Figure 2a, the NH 3 -SCR performance of SbCeZr was slightly affected by SO 2 or H 2 O at temperatures below 220 • C. However, the presence of H 2 O significantly reduced the catalytic activity at temperatures above 420 • C, which could be attributed to the competitive adsorption between H 2 O and NH 3 [26,27]. In the presence of O 2 and SO 2 , the deposited sulfates on the catalysts led to the decrease of the specific surface area (SSA) [28] and blocked the active sites at low temperature [29]. With increased temperature, the decomposition of sulfates led to the restoration of active sites [30]. The presence of H 2 O facilitated SO 2 oxidation at low temperature and inhibited sulfate decomposition at high temperature. Therefore, the presence of H 2 O and SO 2 had a more detrimental effect than H 2 O or SO 2 .
The presence of SO 2 on SbCe had an influence similar to that on SbCeZr. Unfortunately, the presence of H 2 O had a detrimental effect on NH 3 -SCR performance over SbCe. The light-off temperature (T 50 ) shifted to 284 • C, and the maximum NO x conversion was only 66% at 370 • C. When both H 2 O and SO 2 were introduced simultaneously, the synergy effect made the NH 3 -SCR performance drop even more. T 50 shifted to high temperature and the maximum NO x conversion also decreased accordingly. The decrease of NH 3 -SCR performance was not only related to the formation of ammonium salt on the catalyst surface but also connected with the fact that the active phase was sulfated and then formed stable sulfate species [31]. The presence of SO2 on SbCe had an influence similar to that on SbCeZr. Unfortunately, the presence of H2O had a detrimental effect on NH3-SCR performance over SbCe. The light-off temperature (T50) shifted to 284 °C, and the maximum NOx conversion was only 66% at 370 °C. When both H2O and SO2 were introduced simultaneously, the synergy effect made the NH3-SCR performance drop even more. T50 shifted to high temperature and the maximum NOx conversion also decreased accordingly. The decrease of NH3-SCR performance was not only related to the formation of ammonium salt on the catalyst surface but also connected with the fact that the active phase was sulfated and then formed stable sulfate species [31].
Furthermore, repeated temperature ramps (three times) containing SO2 over the SbCeZr catalyst and the SbCe catalyst were also conducted. For SbCeZr, the second time was consistent with the third time, and slightly less active than the first time below 250 °C. It suggests that SbCeZr had a better performance of resistance to SO2. For SbCe, during the three test ramps, the T80 kept shifting to higher temperatures, which narrowed the operation window.

XRD
The XRD patterns of the fresh and aging catalysts are shown in Figure 3. The XRD patterns of CeO2 and ZrO2 showed sharp diffraction peaks corresponding to CeO2 with a cubic fluorite structure ( Furthermore, repeated temperature ramps (three times) containing SO 2 over the SbCeZr catalyst and the SbCe catalyst were also conducted. For SbCeZr, the second time was consistent with the third time, and slightly less active than the first time below 250 • C. It suggests that SbCeZr had a better performance of resistance to SO 2 . For SbCe, during the three test ramps, the T 80 kept shifting to higher temperatures, which narrowed the operation window.

XRD
The XRD patterns of the fresh and aging catalysts are shown in  With the introduction of Sb, the diffraction peaks were broadened and weakened, which led to the absence of obvious peaks. No diffraction peaks related to SbOx were detected, and a completely amorphous structure was formed on the Sb-containing catalysts. Compared with the fresh samples, the diffraction peaks of aged catalysts (aging at 700 °C in air for 4 h) intensified, indicating the increase in the grain size.
According to Table 1, the specific surface area (SSA) of SbCeZr (73 m 2 /g) was slightly higher than that of SbCe (68 m 2 /g). Combining the value of specific surface area (Table 1) with activity showed that specific surface area was not the decisive factor in catalyst activity. It thus suggested that other x 5 x 5 x 2 x 2  With the introduction of Sb, the diffraction peaks were broadened and weakened, which led to the absence of obvious peaks. No diffraction peaks related to SbO x were detected, and a completely amorphous structure was formed on the Sb-containing catalysts. Compared with the fresh samples, the diffraction peaks of aged catalysts (aging at 700 • C in air for 4 h) intensified, indicating the increase in the grain size.
According to Table 1, the specific surface area (SSA) of SbCeZr (73 m 2 /g) was slightly higher than that of SbCe (68 m 2 /g). Combining the value of specific surface area (Table 1) with activity showed that specific surface area was not the decisive factor in catalyst activity. It thus suggested that other factors were affecting activity. The loss of SSA was found after aging, especially on SbCe. The SSA of SbCe-700 • C was only 9 m 2 /g, which was only 13% of fresh SbCe. However, 57% SSA of SbCeZr was lost after aging. Combining the results of XRD and SSA showed that SbCeZr possessed the higher thermal stability as compared to SbCe.  Figure 4 presented the XPS spectra of Sb 3d, Ce 3d, and O 1s for Sb-containing catalysts, and the surface atom concentration and chemical state are summarized in Table 1. For the SbCeZr catalyst, the bulk phase mole ratio of Sb/(Ce + Zr) was 0.57, slightly lower than the surface atom concentration ratio 0.66. However, as for the SbCe catalyst and SbZr catalyst, the bulk phase mole ratio was much lower than the surface concentration. These results indicated that more Sb existed on the surface of CeO 2 and ZrO 2 over SbCe and SbZr, while more Sb was incorporated into the CeZrO x .   Figure 4a showed the Sb 3d3/2 spectra of Sb-containing catalysts and SbOx. Because of the O 1s and Sb 3d5/2 photoemission line overlapping around 530.0 eV, the Sb3d3/2 spectra at higher binding energy were used. This peak could be separated into two contributions-the Gaussian line centered at 540.4 eV corresponds to antimony with an oxidation state of Sb 5+ , while the peak at 539.6 eV corresponds to Sb 3+ [9,16,32] . As shown in Table 1, the proportion of Sb 5+ increase followed the order of SbOx < SbZr < SbCe < SbCeZr. The ratio of Sb 5+ over SbOx was 50.3%, meaning an equal proportion of Sb 5+ and Sb 3+ , i.e., Sb2O4 was the main species in SbOx, which was in line with the XRD results. It was obvious that the concentration of Sb 5+ over the SbCeZr was significantly improved to 64.3%. Figure 4b presents the Ce 3d spectra of the catalysts. The Ce 3d peaks could be deconvoluted into four pairs of spin-orbit doublets. Four main peaks in 3d5/2 at 881.9, 884.9, 887.8, and 898.0 eV were labeled as V, V', V'', and V''', respectively. The four main peaks in 3d3/2 at 900. 8 Binding Energy (eV)  Figure 4a showed the Sb 3d 3/2 spectra of Sb-containing catalysts and SbO x . Because of the O 1s and Sb 3d 5/2 photoemission line overlapping around 530.0 eV, the Sb3d 3/2 spectra at higher binding energy were used. This peak could be separated into two contributions-the Gaussian line centered at 540.4 eV corresponds to antimony with an oxidation state of Sb 5+ , while the peak at 539.6 eV corresponds to Sb 3+ [9,16,32]. As shown in Table 1, the proportion of Sb 5+ increase followed the order of SbO x < SbZr < SbCe < SbCeZr. The ratio of Sb 5+ over SbO x was 50.3%, meaning an equal proportion of Sb 5+ and Catalysts 2020, 10, 1154 6 of 17 Sb 3+ , i.e., Sb 2 O 4 was the main species in SbO x , which was in line with the XRD results. It was obvious that the concentration of Sb 5+ over the SbCeZr was significantly improved to 64.3%. Figure 4b presents the Ce 3d spectra of the catalysts. The Ce 3d peaks could be deconvoluted into four pairs of spin-orbit doublets. Four main peaks in 3d 5/2 at 881.9, 884.9, 887.8, and 898.0 eV were labeled as V, V , V", and V ", respectively. The four main peaks in 3d 3/2 at 900.8, 903.0, 907.4, and 916.2 eV were labeled as U, U , U", and U ", respectively. The peaks labeled as V' and U' corresponded to Ce 3+ and the other peaks labeled as V, V", V ", U, U", and U " corresponded to Ce 4+ [12,33]. From the Ce 3d XPS spectra, it could be concluded that the Ce 3+ and Ce 4+ species coexisted in all the samples, and the ratios of Ce 3+ /(Ce 4+ + Ce 3+ ) of all the catalysts are listed in Table 1. The Ce 3+ /(Ce 4+ + Ce 3+ ) ratio over SbCeZr (42.1%) was higher than that over SbCe (32.9%). The presence of Sb 5+ led to the increase of the Ce 3+ concentration due to the redox reaction Sb 3+ + Ce 4+ ↔ Sb 5+ + Ce 3+ . The transformation of Ce 4+ to Ce 3+ created the charge imbalance and oxygen vacancy, facilitating the adsorption of oxygen species or activation of reactants in NH 3 -SCR [34].

XPS
The O 1s spectra of all the catalysts are displayed in Figure 4c. The spectra were deconvoluted into two peaks. The peak at 530.2 eV was attributed to lattice oxygen (O β ), and the other one at around 531.2 eV was denoted to surface chemisorbed oxygen (O α ). O α species usually correlated with oxygen defects or surface oxygen, such as O 2 2− by peak deconvolution over the samples were 47.8%, 40.6%, and 33.4% over SbCeZr, SbCe, and SbZr, respectively, suggesting more surface chemisorbed oxygen formed over SbCeZr.

H 2 -TPR
The redox property of catalysts plays an important role in the NH 3 -SCR reaction [37,38]. Figure 5 shows the H 2 -TPR profiles of the catalysts. CeO 2 showed two reduction peaks centered at 474 and 781 • C, which were attributed to the reduction of surface oxygen and bulk oxygen of CeO 2 , respectively [22,39]. SbO x also showed two reduction peaks centered at 616 and 701 • C, which could be ascribed to the reduction of Sb 5+ to Sb 3+ and Sb 3+ to Sb 0 , respectively [9]. The reduction profiles of SbZr were similar to those of SbO x with the reduction temperature shifting to lower temperature. The reduction behavior of SbCe and SbCeZr was different from that of CeO2 and SbOx. A broad peak was observed on SbCe and SbCeZr centered at 539 °C and 494 °C, respectively, which could be attributed to the combination of reduction peaks of SbOx and CeO2 [40]. The broad reduction peak of SbCeZr shifted to lower Tonset (the onset reduction temperature of active oxygen) compared to the other catalysts (SbCe and SbZr). The lower reduction temperature of SbCeZr indicated the increase in redox and the generation of easily reducible species.
H2-TPR is usually used as a descriptor for the relative strength of the interaction among active components. The similar reduction profiles of SbOx and SbZr suggested the rather weak interaction between SbOx and ZrO2. However, a strong interaction was found on SbCe due to the co-reduction of SbOx and CeO2, which also led to the higher reduction temperature. By contrast, with the help of ZrO2, the strong interaction between SbOx and CeO2 was weakened, resulting in the lower onset reduction temperature of active oxygen. The reduction behavior of SbCe and SbCeZr was different from that of CeO 2 and SbO x . A broad peak was observed on SbCe and SbCeZr centered at 539 • C and 494 • C, respectively, which could be attributed to the combination of reduction peaks of SbO x and CeO 2 [40]. The broad reduction peak of SbCeZr shifted to lower T onset (the onset reduction temperature of active oxygen) compared to the other catalysts (SbCe and SbZr). The lower reduction temperature of SbCeZr indicated the increase in redox and the generation of easily reducible species.
H 2 -TPR is usually used as a descriptor for the relative strength of the interaction among active components. The similar reduction profiles of SbO x and SbZr suggested the rather weak interaction between SbO x and ZrO 2 . However, a strong interaction was found on SbCe due to the co-reduction of SbO x and CeO 2 , which also led to the higher reduction temperature. By contrast, with the help of ZrO 2 , the strong interaction between SbO x and CeO 2 was weakened, resulting in the lower onset reduction temperature of active oxygen.

NO-TPD
NO-TPD experiments were performed to investigate the adsorption and oxidation properties of NO x on different catalysts. NO-TPD profiles and the calculated NO x adsorption capacity are presented in Figure 6 and Table 2, respectively. The NO desorption peak below 100 • C was attributed to physical adsorbed NO, and the one at about 260 • C was attributed to chemical adsorbed NO [34]. Since the strength of NO bonding to the ceria surface depends on the reduction degree of support (unsaturated sites) [41], it could be concluded that the Ce 3+ content on the SbCeZr surface was much higher than that of SbCe, which was consistent with the XPS result.  The reduction behavior of SbCe and SbCeZr was different from that of CeO2 and SbOx. A broad peak was observed on SbCe and SbCeZr centered at 539 °C and 494 °C, respectively, which could be attributed to the combination of reduction peaks of SbOx and CeO2 [40]. The broad reduction peak of SbCeZr shifted to lower Tonset (the onset reduction temperature of active oxygen) compared to the other catalysts (SbCe and SbZr). The lower reduction temperature of SbCeZr indicated the increase in redox and the generation of easily reducible species.
H2-TPR is usually used as a descriptor for the relative strength of the interaction among active components. The similar reduction profiles of SbOx and SbZr suggested the rather weak interaction between SbOx and ZrO2. However, a strong interaction was found on SbCe due to the co-reduction of SbOx and CeO2, which also led to the higher reduction temperature. By contrast, with the help of ZrO2, the strong interaction between SbOx and CeO2 was weakened, resulting in the lower onset reduction temperature of active oxygen.

NO-TPD
NO-TPD experiments were performed to investigate the adsorption and oxidation properties of NOx on different catalysts. NO-TPD profiles and the calculated NOx adsorption capacity are presented in Figure 6 and Table 2, respectively. The NO desorption peak below 100 °C was attributed to physical adsorbed NO, and the one at about 260 °C was attributed to chemical adsorbed NO [34]. Since the strength of NO bonding to the ceria surface depends on the reduction degree of support (unsaturated sites) [41], it could be concluded that the Ce 3+ content on the SbCeZr surface was much higher than that of SbCe, which was consistent with the XPS result.  NO 2 came from the oxidation of adsorbed NO by active oxygen on the catalyst surface. The desorption temperature and peak area correlated with the adsorption strength and oxidation ability of NO. The ability of the catalyst for oxidizing NO to NO 2 plays an important role in improving the catalytic activities since it can promote the occurrence of the "fast SCR" reaction [42,43]. Considering the temperature range of the NH 3 -SCR reaction, the desorbed peaks in the temperature range of 150-400 • C were mainly discussed.
The main NO x desorption peak of SbCe was at 245 • C, slightly lower than those of SbCeZr (258 • C) and SbZr (273 • C). The calculated NO x amount was in the order of the following sequence: SbCeZr > SbZr > SbCe. There were no obvious differences in the NO x desorption amount between SbCeZr and SbZr, but the amount of NO x desorption on SbCeZr was almost 2 times higher than that of SbCe. It is worth mentioning that 31.7% NO converted to NO 2 on SbCeZr, while only 13.5% NO did so on SbCe. With the help of NO 2 , the "fast SCR" occurred and facilitated NO x removal.

NH 3 -TPD
Surface acidity of the catalysts plays a key role in the adsorption and activation of NH 3 [44,45]. NH 3 -TPD experiments were conducted to characterize the acid amount and strength according to peak area and desorption temperature, and the profiles are shown in Figure 7. Four broad peaks could be observed in the NH 3 -TPD profiles in the temperature range of 100-450 • C. NH 3 desorption peaks at low temperature (<200 • C) were associated with NH 3 desorption from weak acid sites derived from coordinatively unsaturated Ce n+ or Zr n+ sites [46,47], and the peaks in the temperature range of 200-300 • C were assigned to the dissociation of NH 4 + on Brønsted acid sites associated with surface hydroxyl groups. The peaks at temperatures higher than 300 • C were denoted as NH 3 desorption from strong Lewis acid [48][49][50]. The original weak Lewis acidity of CeO 2 and ZrO 2 was also increased by SbO x loading, which could be attributed to the electron withdrawing of SbO x species in proximity of Ce n+ /Zr n+ Lewis acid sites. The NH 3 species desorbed from acid sites (>200 • C) played a decisive role in the NH 3 -SCR reaction [51,52]. The nature of the active sites could be assessed based on the quantity of sites and their intrinsic activity [49]. Quantitative analysis was employed to compare the amount of acid sites, and the ratios of different species based on the deconvoluted results are also shown in Table 2. The amount of acid sites had the following sequence: SbCeZr > SbCe > SbZr. The large total acid amount was found over SbCeZr (1.46 mL/g), which was nearly 2 times as much as that on SbZr (0.78 mL/g). It suggested that the interaction among the catalyst components facilitated the increase in acid amount. The nature of the active sites could be assessed based on the quantity of sites and their intrinsic activity [49]. Quantitative analysis was employed to compare the amount of acid sites, and the ratios of different species based on the deconvoluted results are also shown in Table 2. The amount of acid sites had the following sequence: SbCeZr > SbCe > SbZr. The large total acid amount was found over SbCeZr (1.46 mL/g), which was nearly 2 times as much as that on SbZr (0.78 mL/g). It suggested that the interaction among the catalyst components facilitated the increase in acid amount.

Adsorption of NH3
The adsorption and activation of NH3 play an important role in the NH3-SCR reaction [53]. The adsorption of NH3 was investigated by in situ DRIFT shown in Figure 8. The bands at 3166, 3241, and 3357 cm −1 were assigned to N-H stretching vibration modes of the coordinated NH3 [54,55]. The bands at 1198 and 1599 cm −1 were attributed to coordinated NH3 linked to Lewis acid sites (NH3-L) [34,56], and the bands at 1440 and 1689 cm −1 were assigned to asymmetric and symmetric bending vibrations of NH4 + bounded to Brønsted acid sites (NH4 + -B) [15,40,50]. The band at 1258 cm −1 was assigned to the deformation vibration of NH3 bonded to Lewis acid [57].

Adsorption of NH 3
The adsorption and activation of NH 3 play an important role in the NH 3 -SCR reaction [53]. The adsorption of NH 3 was investigated by in situ DRIFT shown in Figure 8. The bands at 3166, 3241, and 3357 cm −1 were assigned to N-H stretching vibration modes of the coordinated NH 3 [54,55]. The bands at 1198 and 1599 cm −1 were attributed to coordinated NH 3 linked to Lewis acid sites (NH 3 -L) [34,56], and the bands at 1440 and 1689 cm −1 were assigned to asymmetric and symmetric bending vibrations of NH 4 + bounded to Brønsted acid sites (NH 4 + -B) [15,40,50]. The band at 1258 cm −1 was assigned to the deformation vibration of NH 3 bonded to Lewis acid [57].
Catalysts 2020, 10, 1154 9 of 17 3357 cm −1 were assigned to N-H stretching vibration modes of the coordinated NH3 [54,55]. The bands at 1198 and 1599 cm −1 were attributed to coordinated NH3 linked to Lewis acid sites (NH3-L) [34,56], and the bands at 1440 and 1689 cm −1 were assigned to asymmetric and symmetric bending vibrations of NH4 + bounded to Brønsted acid sites (NH4 + -B) [15,40,50]. The band at 1258 cm −1 was assigned to the deformation vibration of NH3 bonded to Lewis acid [57].   The presence of abundant Brønsted and Lewis acid sites on the SbCeZr catalyst should be responsible for the improvement in SCR activities over a wide temperature range [58].

Adsorption of NH 3 Followed by Introduction of NO + O 2
The spectra of introducing NO + O 2 into the catalysts after NH 3 adsorption as a function of time were recorded and are shown in Figure 9. The decrease of NH 4 + -B and NH 3 -L species on SbCeZr with the introduction of NO + O 2 presented the reaction between NO x and adsorbed NH 3 species (Figure 9a). The intensities of the bands ascribed to NH 3 -L and NH 4 + -B species were reduced after NH 3 [60]. The recovery of the negative broad band of surface hydroxyl groups (3441 cm −1 ) was attributed to the formation of water, which further confirmed the occurrence of the reaction.
was essential. The further investigation on the consumption rate of NH3-L and NH4 + -B species showed the slightly faster disappearance of NH3-L species than NH4 + -B species in the reaction. Extending the exposure time led to the appearance of nitrate species, such as bridged nitrate (1605 and 1247 cm −1 ) [59], bidentate nitrate (1579, 1559, and 1516 cm −1 ) [55], and monodentate nitrate (1541 cm −1 ) [60]. The recovery of the negative broad band of surface hydroxyl groups (3441 cm −1 ) was attributed to the formation of water, which further confirmed the occurrence of the reaction. As shown in Figure 9b, the phenomenon that occurred on SbCe was similar to that on SbCeZr, i.e., the appearance of nitrate species at the expense of the disappearance of adsorbed NH3. The disappearance of NH3-L species confirmed that NH3-L species could be involved in the reaction, and NH3-L species were slightly active compared with NH4 + -B species. The adsorbed NH3 on the catalyst surface reacted with NO/NO2 in the gas phase, following the Eley-Rideal (E-R) mechanism.

Co-Adsorption of NO+O2
The adsorption of NO+O2 was investigated by in situ DRIFT. As shown in Figure 10a, bridged nitrate (1245, 1606 cm −1 ), bidentate nitrate (1579, 1559, 1519 cm −1 ), and monodentate nitrate (1539 cm −1 ) were formed on SbCeZr after exposure to NO+O2 for 30 min. The NH3 adsorption rate was much faster than the NOxgeneration rate. With the exposure time prolonged, the bands of nitrate As shown in Figure 9b, the phenomenon that occurred on SbCe was similar to that on SbCeZr, i.e., the appearance of nitrate species at the expense of the disappearance of adsorbed NH 3 . The disappearance of NH 3 -L species confirmed that NH 3 -L species could be involved in the reaction, and NH 3 -L species were slightly active compared with NH 4 + -B species. The adsorbed NH 3 on the catalyst surface reacted with NO/NO 2 in the gas phase, following the Eley-Rideal (E-R) mechanism.

Co-Adsorption of NO + O 2
The adsorption of NO + O 2 was investigated by in situ DRIFT. As shown in Figure 10a, bridged nitrate (1245, 1606 cm −1 ), bidentate nitrate (1579, 1559, 1519 cm −1 ), and monodentate nitrate (1539 cm −1 ) were formed on SbCeZr after exposure to NO + O 2 for 30 min. The NH 3 adsorption rate was much faster than the NO x − generation rate. With the exposure time prolonged, the bands of nitrate intensified accordingly. Bridged nitrate seemed dominant in all the kinds of nitrates. In the case of SbCe (Figure 10b), rather weak bands of nitrates were observed after the exposure to the same feed gas for the same time, and the adsorption intensity of bridged nitrate was rather weaker than that of the monodentate and bidentate nitrates. It could be seen that Sb interacted with different metals (Ce or CeZr), resulting in the difference in the NO 3 species distribution on the catalyst surface.
Catalysts 2020, 10, x FOR PEER REVIEW 11 of 17 intensified accordingly. Bridged nitrate seemed dominant in all the kinds of nitrates. In the case of SbCe (Figure 10b), rather weak bands of nitrates were observed after the exposure to the same feed gas for the same time, and the adsorption intensity of bridged nitrate was rather weaker than that of the monodentate and bidentate nitrates. It could be seen that Sb interacted with different metals (Ce or CeZr), resulting in the difference in the NO3 -species distribution on the catalyst surface.

Adsorption of NO + O 2 Followed by Introduction of NH 3
Compared with the results of NO + O 2 reacting with preadsorbed NH 3 (Figure 9), the effects of adding NH 3 to preadsorbed NO + O 2 ( Figure 11) on activity were not so obvious. It is worth mentioning that a new band linked to NH 4 + -B species (1440 cm −1 ) gradually formed over SbCeZr as the exposure time increased. However, the bands linked to NH 3 -L (1599 and 1198 cm −1 ) were not detected. Therefore, the formed nitrate species had detrimental effects on NH 3 adsorption on Lewis sites, but negligent effects on NH 3 adsorption on Brønsted acid sites.

Adsorption of NO+O2 Followed by Introduction of NH3
Compared with the results of NO+O2 reacting with preadsorbed NH3 (Figure 9), the effects of adding NH3 to preadsorbed NO+O2 ( Figure 11) on activity were not so obvious. It is worth mentioning that a new band linked to NH4 + -B species (1440 cm −1 ) gradually formed over SbCeZr as the exposure time increased. However, the bands linked to NH3-L (1599 and 1198 cm −1 ) were not detected. Therefore, the formed nitrate species had detrimental effects on NH3 adsorption on Lewis sites, but negligent effects on NH3 adsorption on Brønsted acid sites. On the contrary, there were no bands related to NH3 adsorption over SbCe even after prolonging the exposure time up to 30 min. The phenomena illustrated that the formed nitrate species significantly inhibited NH3 adsorption. Combined with the results of Figure 11, this indicated that NH3 adsorption and NOx oxidation could be competitive on SbCe, but not on SbCeZr.

Discussion
NOx conversions over CeO2 and SbOx were very low, but it increased significantly when they were combined together. Compared with CeO2, the reason for the improvement of catalytic activity over SbCeOx was due to alleviating nitrate adsorption and leaving more active sites for NH3 adsorption and activation [40]. Considering the competitive adsorption between NH3 and nitrate species, we prepared SbCeZr and investigated it for NH3-SCR performance. SbCeZr outperformed SbCe not only in activity and stability but also in SO2 and water tolerance. SbCeZr exhibited high activity with NOx conversion over 80% in the temperature range of 202-422 °C. The introduction of H2O and SO2 had little influence on the activity of SbCeZr, but an adverse effect on the catalytic activity of SbCe was found.
The presence of Zr influenced the distribution of Sb on the surface or bulk. According to ICP-AES and XPS results (Table 1), the surface amount of Sb on SbCe was much higher than that on SbCeZr, displaying the enrichment of Sb on the SbCe surface. However, H2-TPR indicated the strong interaction between SbOx and CeO2. Unlike SbCe, the presence of Zr made the distribution of Sb, Ce, and Zr uniform and weakened the interaction between SbOx and CeO2. The weakened interaction also brought about abundant Sb 5+ species on the catalyst surface (Table 1) and the lower onset reduction temperature of active oxygen ( Figure 5). Furthermore, the generation of NO2 during NO-TPD ( Figure 6) also illustrated the enhancement in reducibility, which facilitated the "fast SCR" reaction during NH3-SCR [32].
The adsorption of NH3 on the catalyst surface was considered as the initial step of the NH3-SCR reaction [15], so NH3 adsorption played an important role during NOx removal. Compared with SbCe, NH3 adsorption was enhanced on SbCeZr, which was attributed to the increase of Brønsted and Lewis acid amounts (Figure 7). The acid distribution on the catalysts was different. The amount On the contrary, there were no bands related to NH 3 adsorption over SbCe even after prolonging the exposure time up to 30 min. The phenomena illustrated that the formed nitrate species significantly inhibited NH 3 adsorption. Combined with the results of Figure 11, this indicated that NH 3 adsorption and NO x oxidation could be competitive on SbCe, but not on SbCeZr.

Discussion
NO x conversions over CeO 2 and SbO x were very low, but it increased significantly when they were combined together. Compared with CeO 2 , the reason for the improvement of catalytic activity over SbCeO x was due to alleviating nitrate adsorption and leaving more active sites for NH 3 adsorption and activation [40]. Considering the competitive adsorption between NH 3 and nitrate species, we prepared SbCeZr and investigated it for NH 3 -SCR performance. SbCeZr outperformed SbCe not only in activity and stability but also in SO 2 and water tolerance. SbCeZr exhibited high activity with NO x conversion over 80% in the temperature range of 202-422 • C. The introduction of H 2 O and SO 2 had little influence on the activity of SbCeZr, but an adverse effect on the catalytic activity of SbCe was found.
The presence of Zr influenced the distribution of Sb on the surface or bulk. According to ICP-AES and XPS results (Table 1), the surface amount of Sb on SbCe was much higher than that on SbCeZr, displaying the enrichment of Sb on the SbCe surface. However, H 2 -TPR indicated the strong interaction between SbO x and CeO 2 . Unlike SbCe, the presence of Zr made the distribution of Sb, Ce, and Zr uniform and weakened the interaction between SbO x and CeO 2 . The weakened interaction also brought about abundant Sb 5+ species on the catalyst surface (Table 1) and the lower onset reduction temperature of active oxygen ( Figure 5). Furthermore, the generation of NO 2 during NO-TPD ( Figure 6) also illustrated the enhancement in reducibility, which facilitated the "fast SCR" reaction during NH 3 -SCR [32].
The adsorption of NH 3 on the catalyst surface was considered as the initial step of the NH 3 -SCR reaction [15], so NH 3 adsorption played an important role during NO x removal. Compared with SbCe, NH 3 adsorption was enhanced on SbCeZr, which was attributed to the increase of Brønsted and Lewis acid amounts (Figure 7). The acid distribution on the catalysts was different. The amount of Brønsted acid was higher than the amount of Lewis acid on SbCeZr, whereas the amount of Brønsted acid was similar to that of Lewis acid on SbCe. Meanwhile, in situ DRIFT experiments of NH 3 adsorption also confirmed the strong adsorption of NH 3 and the faster adsorption rate on SbCeZr (Figure 8). The reaction rate of NO x with NH 3 adsorbed on Lewis acid was slightly faster than that with NH 4 + adsorbed on Brønsted acid (Figure 9) (E-R mechanism).
It is worth mentioning that the formed nitrates had negligible effects on the adsorption of NH 3 over SbCeZr; however, detrimental impacts were found over SbCe. It meant that NO oxidation and NH 3 adsorption were both around Ce n+ sites, so there was competition between them. Amorphous ZrO 2 possessed abundant hydroxyl [61], which showed obvious acidity properties. Then, the strong withdrawing effects of SbO x further increased the acid amount and strength of ZrO 2 , which was confirmed by the high NH 3 desorption amount (Figure 7). Therefore, the separation of active sites alleviated the competition effects between NO oxidation and NH 3 adsorption, thus promoting NO x removal.
Given the above discussion, it was shown that the strong acid and weak reducibility of SbZr made it less active in NH 3 -SCR. However, a strengthened interaction was found between SbO x and CeO 2 , favoring NO x removal. Unfortunately, the competition between NO oxidation and NH 3 adsorption on the same active sites still needed to be alleviated to improve its thermostability and activity in the presence of H 2 O and SO 2 . In the presence of ZrO 2 , NH 3 adsorption and NO oxidation on different active sites widened the operation window.

Catalyst Preparation
The SbCe, SbCeZr, and SbZr catalysts, where the molar ratio of Sb/Ce and Zr/Ce was 1.0, were prepared by the citric acid method. First, Sb(CH 3 COO) 3 was dissolved in a solution of citric acid, then Zr(NO 3 ) 4 and Ce(NO 3 ) 4 were added. The molar ratio of citric acid to total metal components (the sum of Ce, Zr, and Sb) was 1.5. Then, the resultant mixture was stirred at 80 • C in a water bath until forming a gel. The gel was dried at 110 • C for 12 h to form a porous, foamlike solid. Finally, the precursor was calcined at 500 • C for 3 h in a muffle furnace (SX2, Shanghai Pudong Rong-Feng Scientific Instrument Co. Ltd., Shanghai, China). After calcining, the sample was crushed and sieved to 40-60 mesh prior to evaluation. CeO 2 , SbO x , and ZrO 2 were also prepared by the same method.

Catalytic Activity Testing
The catalytic activities of the catalysts for the selective catalytic reduction of NO x with NH 3 in excess oxygen were conducted at atmospheric pressure in a fixed-bed continuous flow quartz reactor. In the evaluation, a 200 mg portion of catalyst (40-60 mesh) was applied, and the feed gas contained 500 ppm NO, 500 ppm NH 3 , 5% O 2 , 5% H 2 O (when used), 100 ppm SO 2 (when used), and Ar balance. The total flow was 300 mL/min and the gas hourly space velocity (GHSV) was about 150,000 h −1 . A Thermo Fisher NO-NO x -chemiluminescence analyzer (Waltham, MA, USA) was used to detect the concentration of NO and NO 2 before and after the reaction. The catalyst bed was heated at 2 • C/min, held for 45 min every 50 • C, and then the concentration of NO x was measured. NO x conversion was calculated using the following equation: where [NO x ] in and [NO x ] out represented the inlet and outlet concentrations of NO x , respectively.

Catalyst Characterization
The N 2 adsorption-desorption isotherms were measured on a Quantachrome NOVA1200 analyzer (Norcross, GA, USA) at −196 • C to obtain the surface area of the catalysts. Before measurement, the catalyst was degassed at 180 • C until it reached a stable vacuum of approximately 5 mTorr. After calculating the desorption data using the Brunauer-Emmett-Teller (BET) method, the specific surface area was obtained. The X-ray diffraction (XRD) patterns were carried out on a Bruker/D8 diffractometer (Karlsruhe, Germany) with Cu Kα radiation (λ = 0.154 nm). The X-ray photoelectron spectroscopy (XPS) spectra were obtained using a Thermo Scientific ESCALAB 250 spectrometer analyzer (Waltham, MA, USA), using a monochromatized Al-Kα source (1486.6 eV), and the binding energy of adventitious carbon (284.6 eV) was taken as a reference.
The temperature-programmed reduction of H 2 (H 2 -TPR) and the temperature-programmed desorption of NH 3 (NH 3 -TPD) were performed on a PX200 apparatus (Tianjin Pengxiang Technology Co. Ltd., Tianjin, China) with a thermal conductivity detector (TCD). For the H 2 -TPR experiment, a 100 mg portion of catalyst was pretreated in N 2 at 300 • C for 30 min, then cooled down to 30 • C. Finally, the catalyst was heated from 30 • C to 800 • C at a rate of 10 • C/min in 5% H 2 /Ar (40 mL/min). For the NH 3 -TPD, a 100 mg portion of catalyst was pretreated at 500 • C under a flow of Ar (50 mL/min) for 1 h, and the pretreated catalyst was exposed to a flow of 10% NH 3 /Ar (40 mL/min) for 1 h at 100 • C. Then, the sample was flushed by Ar (50 mL/min) for 1 h and heated from 100 • C to 500 • C at 10 • C/min.
The temperature-programmed desorption of NO (NO-TPD) experiments were conducted on custom-made equipment with a NO x analyzer. Before testing, a 200 mg portion of catalyst was pretreated at 500 • C under a flow of Ar (300 mL/min) for 1 h; after cooling to room temperature, the pretreated catalyst adsorbed NO with a flow of 500 ppm NO/Ar (300 mL/min) for 1 h. The catalyst was flushed by Ar (300 mL/min) for 1 h and then heated from room temperature to 500 • C at 10 • C/min.
In situ DRIFT spectra were collected using an FTIR spectrometer (Thermo Nicolet 6700, Waltham, MA, USA) with an MCT detector cooled by liquid nitrogen. Before each experiment, the catalyst was pretreated at 300 • C for 1 h in Ar (50 mL/min). After the temperature cooling to 200 • C, 1000 ppm NH 3 /Ar was introduced into the IR cell and the background was subtracted from the sample spectra. After adsorption saturation, 1000 ppm NO + 5% O 2 /Ar was introduced to replace NH 3 at 200 • C. The experiment of NO + O 2 adsorption and reaction with NH 3 was carried out following a similar process. First, the catalyst was pretreated at 300 • C for 1 h in Ar (50 mL/min), then 1000 ppm NO + 5% O 2 /Ar adsorption was carried out at 200 • C, and finally 1000 ppm NH 3 /Ar was introduced to replace NO + O 2 .

Conclusions
Sb-containing metal oxides (SbZr, SbCe, SbCeZr) were prepared by citric acid method and investigated for NH 3 -SCR. SbCeZr exhibited the higher performance in NH 3 -SCR than SbCe and SbZr, and NO x could be removed with 80% conversion in the temperature range of 202-422 • C. SbCeZr owned excellent activity even in the presence of water and SO 2 , and the thermostability was improved compared with SbCe. Hence, SbCeZr seemed to be a good candidate for the practical application in deNO x . The presence of Sb in the metal oxides led to the difference in acid distribution and redox property, which was closely related to the NH 3 adsorption and NO oxidation. Brønsted acid and Lewis acid distributed evenly on SbCe, while Brønsted acid dominated on SbCeZr. Although high amounts of acid were on SbZr, poor redox led to its negligent performance in NH 3 -SCR. The enhancement in redox ability promoted the generation of nitrates on SbCe. However, the competition between NO oxidation and NH 3 adsorption on the same active sites limited its application. In the presence of ZrO 2 , the interaction between Sb and Ce was weakened, which further increased the acid amount and redox property. The form of "dual active sites" favored the NH 3 adsorption and nitrate generation.