The selective catalytic reduction of NO_x by NH₃ (NH₃ SCR) has been widely investigated because of its ability to control NO_x pollutant emissions from mobile and stationary sources. Among NH₃-SCR catalysts, CeO₂-based materials have received increasing attention because they are considered as potential environmentally friendly non-vanadium catalysts for use in place of commercial V₂O₅-WO₃/TiO₂ catalysts, which contain toxic V₂O₅ [1, 2]. It is well known that CeO₂ has unique redox ability and high oxygen storage capacity, which can easily result in a synergistic catalytic effect with many other metals or metal oxides in catalytic processes [3, 4]. Although pure CeO₂ itself is unsuitable as a catalyst for NH₃ SCR, CeO₂-based catalysts including CeO₂-MnO_x [5-7], CeO₂-MoO₃ [8], CeO₂-SnO₂ [9], CeO₂-TiO₂ [10-12], and CeO₂-WO₃ [13, 14] exhibit desirable catalytic activity in the temperature range from 100 to 450 ℃. In particular, the CeO₂-WO₃ catalyst displays excellent NO conversion and N₂ selectivity over a wide temperature range from 200 to 450 ℃ and retains its catalytic activity in the presence of H₂O and SO₂ [14]. Thus, CeO₂-WO₃-based catalysts are a promising choice for industrial NO_x degradation by NH₃ SCR at both low and medium-high temperature.

A recent study found that adsorbed NH₃ could be oxidized by the surface oxygen of CeO₂ at high temperature, forming surface NO_x species [15]. This resulted in poor N₂ selectivity of pure CeO₂ at high temperature because of the competitive reactions of NO and O₂ with NH₃. It has been reported that WO₃ could improve the catalytic activity and N₂ selectivity of CeO₂-based catalysts at high temperature by suppressing the nonselective catalytic oxidation of NH₃ to NO_x and enhancing NH₃ adsorption [16]. In situ infrared (IR) and Raman spectroscopy results suggested that a CeO₂-WO₃ catalyst prepared via a co-precipitation method involved a reaction mechanism consisting of two independent cycles for the NH₃ SCR reaction [17]. These cycles included a redox cycle induced by the excellent oxygen storage capability and reducibility of cubic fluorite CeO₂ and an acid site cycle resulting from Brönsted acid sites provided by the W-O-W species of Ce₂(WO₄)₃. The addition of WO₃ to CeO₂ can increase its surface acidity [13, 17], which plays an important role in promoting the NH₃ SCR performance of CeO₂. However, it has been reported that when MoO₃, which is related to WO₃, was supported on ZrO₂, the amount of nitrate species adsorbed obviously decreased with increasing coverage of MoO₃ [18]. The influence of the coverage of WO₃ on CeO₂ on NO adsorption and further effects on NH₃ adsorption in the NH₃ SCR reaction have not been elucidated.

In this work, a series of WO₃/CeO₂ supported catalysts are prepared via impregnation and characterized by X-ray diffraction (XRD), Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), H₂ temperature programmed reduction (H₂-TPR), NH₃ temperature programmed desorption (NH₃-TPD), and in situ diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy. We systemically investigate how the WO₃ loading affects the catalytic performance of WO₃/CeO₂ in NH₃ SCR, mainly focusing on (1) studying the relationships between the states of WO₃ on the surface of CeO₂ and the catalytic activity of WO₃/CeO₂, and (2) further exploring how NH₃ and/or NO adsorption on CeO₂ is influenced by surface modification with WO₃.

CeO₂ as a support was prepared by thermal decomposition of Ce(NO₃)₃·6H₂O at 550 ℃ for 4 h in flowing air. WO₃/CeO₂ was prepared by wet impregnation. The requisite quantities of ammonium tungstate and oxalic acid (molar ratio of 1:1.5) was added dropwise into water under stirring and then an appropriate amount of CeO₂ was added to the mixture. The solvent was evaporated at 100 ℃, and then the remaining material was dried at 110 ℃ for 3 h before being calcined at 500 ℃ for 4 h in flowing air. For simplicity, xWO₃/CeO₂ samples are denoted as xW/Ce, where x is the loading amount of WO₃. For example, 0.8W/Ce represents 0.8 mmol WO₃ per 100 m² CeO₂ (0.8 mmol/(100 m²)). WO₃ and CeO₂ were mechanically mixed with the same components as 0.8W/Ce; the mechanically mixed sample is denoted as 0.8W/Ce-MM. In addition, the 1.6W/Ce sample was immersed in 25 wt% ammonium hydroxide at 70 ℃ for 2 h, filtered, dried at 110 ℃ for 3 h, and then calcined at 500 ℃ for 4 h in flowing air to give a sample denoted as 1.6W/Ce-N (N represents ammonium hydroxide washing).

Specific surface areas of samples were measured by nitrogen adsorption at –196 ℃ on a Micrometrics ASAP-2020 analyzer (Micromeritics, USA) by the Brunauer-Emmet-Teller method. Before each adsorption measurement, the catalyst sample (about 0.1 g) was degassed in a N₂/He mixture at 300 ℃ for 4 h.

Powder XRD patterns were collected using a Philips X'pert Pro diffractometer (APL, Switzerland) with Ni-filtered Cu K_α1 radiation (0.15408 nm). The X-ray tube was operated at 40 kV and 40 mA.

Raman spectroscopy analysis was performed on a LabRAM Aramis (Horiba, Japan) laser Raman spectrometer using an Ar⁺ laser beam. Raman spectra were recorded with an excitation wavelength of 532 nm and laser power of 10 mW.

XPS experiments were performed on a PHI 5000 Versa Probe high-performance electron spectrometer (ULVAC-PHI, Japan) using monochromatic Al K_α radiation (1486.6 eV) and operating at an accelerating power of 15 kW. Before each measurement, the sample was outgassed at room temperature in an ultra-high vacuum chamber ( < 5×10⁻⁷ Pa). Sample charging effects were compensated for by calibrating all binding energies (BE) with the C 1s peak of adventitious carbon at 284.6 eV. This reference gave BE values with an accuracy of ± 0.1 eV.

H₂-TPR measurements were carried out in a quartz U-tube reactor connected to a thermal conductivity detector (TCD) using a 7%H₂/93%Ar mixture as the reductant. Before switching to the H₂/Ar stream, each sample (50 mg) was pretreated in a N₂ stream at 200 ℃ for 1 h. TPR was performed from room temperature to 850 ℃ at a rate of 10 ℃/min.

NH₃-TPD experiments were carried out on a multifunction chemisorption analyzer (Tianjin Pengxiang, China) with a quartz U-tube reactor and detected by a TCD. Each sample (about 0.1 g) was pretreated in high-purity N₂ (40 mL/min) at 450 ℃ for 1 h. After pretreatment, each sample was saturated with 1 vol% NH₃ (10 mL/min) at 100 ℃ for 1 h and subsequently flushed with high-purity N₂ at the same temperature for 1 h to remove gaseous and weakly adsorbed NH₃. The sample was then heated to 600 ℃ at a rate of 10 ℃/mi in flowing high-purity N₂ (40 mL/min).

In situ DRIFT experiments were performed on a Nicolet Nexus 5700 FTIR spectrometer (Nicolet, USA) using a diffuse reflectance attachment (Harrick) equipped with a reaction cell with ZnSe windows. Thirty-two scans at a resolution of 4 cm^–1 were obtained and spectra were presented as Kubelka-Munk functions with reference to background spectra recorded for the catalyst in N₂. Powder samples were pretreated in N₂ at 400 ℃ for 1 h prior to the collection of background spectra and adsorption experiments. The reaction conditions were: 1000 ppm NH₃, 1000 ppm NO, 5 vol% O₂, and N₂ balance. All adsorptio-desorption experiments were conducted at 200 ℃. To achieve saturated adsorption, samples were preadsorbed by introduction of NH₃ and/or NO and O₂ for 1 h, purged with N₂ for 15 min, and then relative spectra were collected.

Catalytic reactions were performed in a fixed-bed quartz reactor tube. The reaction conditions were: 500 ppm NO, 500 ppm NH₃, 5 vol% O₂, 200 ppm SO₂ (when used), and 5 vol% H₂O (when used) with the balance N₂, gas flow rate of 100 mL/min, and 0.1 g of catalyst. Prior to catalytic testing, samples were pretreated in N₂ at 200 ℃ for 1 h. The effluent gases including NO, NH₃, NO₂, and N₂O were continuously analyzed by an online Nicolet IS10 IR spectrometer (Nicolet, USA) equipped with a gas cell. IR data were collected at the required temperature to keep the reaction at a steady state for 15 min. The NO conversion and N₂ selectivity were calculated using the following equations:

NO conversion (%) = 100 × ([NO]_in – [NO]_out)/NO_in

N₂ selectivity (%) = 100 × ([NO]_in – [NO]_out + [NH₃]_in – [NH₃]_out – [NO₂]_out – 2[N₂O]_out)/([NO]_in – [NO]_out + [NH₃]_in – [NH₃]_out)

The NH₃ SCR performance of W/Ce samples was evaluated as a function of temperature. As shown in Fig. 1(a), both pure CeO₂ and WO₃ displayed low catalytic activity. However, when WO₃ was supported on CeO₂, the supported W/Ce samples exhibited much higher NO conversion than that of pure CeO₂. Overall, NO conversion over W/Ce samples increased first and then decreased with increasing WO₃ loading. 0.8W/Ce displayed the highest catalytic activity of the samples, with more than 88% NO conversion from 200 to 450 ℃. As shown in Fig. 1(b), the N₂ selectivity of W/Ce increased with the WO₃ loading. More than 95% N₂ selectivity was obtained over the whole temperature range when the WO₃ loading was higher than 0.4 mmol/(100 m²). Thus, it can be concluded that the catalytic performance of W/Ce depends on WO₃ loading. Interestingly, when 1.6W/Ce was washed with 25 wt% ammonium hydroxide, as illustrated in Fig. 1(c), NO conversion over 1.6W/Ce-N was higher than that of 1.6W/Ce and almost at the same level as that of 0.65W/Ce. The effect of washing with ammonium hydroxide on the catalytic activity of 1.6W/Ce needs to be investigated further to be fully understood. In addition, the catalytic activity of 0.8W/Ce in the presence of H₂O and SO₂ was investigated at 300 ℃. As presented in Fig. 1(d), NO conversion remained unchanged at ca. 94% during 25 h of continuous testing, indicating that the supported WO₃/CeO₂ catalyst had the same excellent ability to tolerate H₂O and SO₂ as a reported WO₃-CeO₂ catalyst prepared via co-precipitation [11].

Fig. 2 shows XRD patterns of the W/Ce samples. When the WO₃ loading was lower than 0.8 mmol/(100 m²), all peaks were assigned to the cubic fluorite-type CeO₂ (PDF-ICDD 34-0394), which indicates that WO₃ was highly dispersed on the CeO₂ surface. However, when the WO₃ loading was higher than 0.8 mmol/(100 m²), new peaks appeared at 2θ = 23.2°, 23.6°, and 24.4° originating from monoclinic WO₃ (PDF-ICDD 83-0591). This result is consistent with the reported experimental dispersion capacity of 4.8 W⁶⁺ per nm² over CeO₂ (equal to 0.8 mmol/(100 m²)) [19, 20]. Interestingly, 0.8W/Ce-MM also displayed peaks from WO₃.

Meanwhile, Raman spectroscopy effectively discriminated the states of WO₃ supported on CeO₂. As shown in Fig. 3, the band at 463 cm⁻¹ was the characteristic F_2g vibration of CeO₂ and that at 259 cm⁻¹ could be assigned to the tetrahedral displacement of oxygen from the ideal fluorite lattice [21]. Pure WO₃ exhibited five major bands at 270, 325, 712, 806, and 900–970 cm⁻¹ originating from W–O–W deformation (F_2g mode), W–O–W bending vibration (E mode), W–O–W stretching vibrations (A_1g/E_g modes), and W=O stretching, respectively [22, 23]. With regard to W/Ce samples, the band at 900–1000 cm⁻¹ could be assigned to the stretching of terminal W=O, and shifted to higher wavenumber with increasing WO₃ loading because of the weaker interaction between WO₃ and CeO₂ at higher WO₃ loading [24]. When the WO₃ loading was lower than 0.4 mmol/(100 m²), only bands at 900–1000 cm⁻¹ arising from W=O stretching could be detected, which indicates that isolated tungstate species formed. However, when the WO₃ loading was higher than 0.4 mmol/(100 m²), a new band attributed to W–O–W deformation at 806 cm⁻¹ appeared, indicating the formation of polytungstate species. The intensity of the band at 806 cm⁻¹ strengthened when the WO₃ loading exceeded 0.8 mmol/(100 m²), which was caused by the formation of crystalline WO₃, as confirmed by XRD. Increasing WO₃ surface density on a ZrO₂ support caused the states of WO₃ to change from a submonolayer of isolated tungstate to polytungstate and then to crystalline WO₃ [25]. Thus, WO₃ loading should decide the states of WO₃ on both CeO₂ and ZrO₂. Isolated WO₃ existed below its dispersion capacity on CeO₂ and polytungstate/crystalline WO₃ formed above this dispersion capacity.

Combined with the results of catalytic activity testing, the NO_x degradation efficiency of the supported W/Ce catalysts gradually increased with WO₃ loading below the dispersion capacity of WO₃ on CeO₂, and then decreased above the dispersion capacity. Thus, the catalytic performance of W/Ce depended on the state of WO₃ on the CeO₂ surface. It is worth noting that the XRD and Raman results of 1.6W/Ce-N were different from those of 1.6W/Ce. As illustrated in Fig. 2, only the peaks of CeO₂ were observed in the XRD pattern of 1.6W/Ce-N. Meanwhile, Fig. 3 shows that the bands at 325, 712, and 806 cm⁻¹ originating from the crystalline WO₃ disappeared and only a band at 965 cm⁻¹ assigned to dispersed WO₃ was detected. Both XRD and Raman results indicate that crystalline WO₃ in 1.6W/Ce was removed by ammonium hydroxide so that only dispersed WO₃ was left on the CeO₂ surface. Combined with the results of catalytic activity testing, crystalline WO₃ had the lowest catalytic activity, with less than 20% NO conversion over the whole temperature range. Once the crystalline WO₃ was removed from 1.6W/Ce, NO conversion improved. In addition, although 0.8W/Ce-MM had the same components as those of 0.8W/Ce, as depicted in Fig. 1(a), 0.8W/Ce-MM displayed much lower NO conversion than 0.8W/Ce, and XRD analysis also revealed that crystalline WO₃ formed over 0.8W/Ce-MM. Therefore, we conclude that crystalline WO₃ inhibited the catalytic activity of the W/Ce catalysts.

XPS was carried out to determine the surface components and chemical states of elements of the W/Ce samples. As shown in Fig. 4(a), Ce 3d spectra were composed of two multiplets (labelled as v and u) assigned to 3d_3/2 and 3d_5/2, respectively, because of spin-orbit coupling. Through Gaussian-Lorentz fitting, the bands labelled u' and v' were ascribed to the primary photoemission of Ce³⁺ and the other six bands (u''', v''', u″, v″, u, and v) were assigned to Ce⁴⁺ [26-28]. Ce³⁺ content was calculated by the formula:

Ce(Ⅲ) (%) = 100 × [S(u) + S(v)]/∑[S(u) + S(v)]

As listed in Table 1, the relative content of Ce³⁺ did not obviously change with increasing WO₃ loading. In addition, as depicted in Fig. 4(b), for the W 4f spectra of W/Ce samples, the BE of W 4f_5/2 and W 4f_7/2 were fixed at 37.4 and 35.2 eV, respectively, indicating W was in the +6 oxidation state in all the W/Ce samples [29]. Therefore, the electronic interaction between CeO₂ and WO₃ can be ignored in the W/Ce samples.

It has been reported that the XPS peak intensity ratio I_W4f/I_Ce3d is proportional to WO₃ loading in WO₃/CeO₂ samples when the WO₃ loading is lower than the dispersion capacity of WO₃ on CeO₂ [20]. To calculate the amount of WO₃ in 1.6W/Ce-N, the linear fitting between I_W4f/I_Ce3d for (0.2, 0.4, and 0.8) W/Ce (denoted as y) and the corresponding WO₃ loading (denoted as x) was carried out by the least squares method, which gave the linear equation y = 0.336x + 0.034 (Fig. 4(c)). Table 1 reveals that I_W4f/I_Ce3d of 1.6W/Ce-N was 0.252, thus its WO₃ loading was 0.65 mmol/(100 m²), which could explain why the NO conversion over 1.6W/Ce-N was as good as that over 0.65W/Ce. This analysis reveals that crystalline WO₃ inhibited the catalytic activity of the W/Ce samples.

Fig. 4(d) displays the O 1s spectra fitted by Gaussian-Lorentz curves. Bands labelled O' and O'' were attributed to surface oxygen species and lattice oxygen, respectively [21, 30]. The O'' bands of 1.6W/Ce and 1.2W/Ce were shifted to higher BE compared with those of the other samples, reflecting the contribution of the lattice oxygen of crystalline WO₃ [29]. In fact, compared with that of the 1.6W/Ce sample, the O'' band of 1.6W/Ce-N shifted to lower BE because of the removal of crystalline WO₃. As listed in Table 1, the relative content of surface oxygen species decreased with increasing WO₃ loading. It has been reported that superoxide and peroxide species could form on the surface of CeO₂ [31, 32]. Thus, it is likely that the adsorption of active oxygen species on CeO₂ was inhibited when the CeO₂ surface was covered with dispersed WO₃.

The influence of WO₃ loading on the redox performance of the catalysts was investigated by H₂-TPR measurements. As shown in Fig. 5, pure WO₃ was reduced by H₂ above 700 ℃. Pure CeO₂ and W/Ce samples displayed three reduction peaks: α at ca. 400 ℃, β at 500–640 ℃, and γ at 760 ℃, which were ascribed to the reduction of surface active oxygen species, surface or subsurface layers of CeO₂, and bulk CeO₂, respectively [33, 34]. With increasing WO₃ loading, the intensity of α peaks obviously decreased compared with those for pure CeO₂, indicating that the amount of surface active oxygen species decreased with increasing WO₃ loading, which was in accordance with the XPS results. Meanwhile, β peaks shifted to higher temperature with increasing WO₃ loading. Considering that WO₃ was not reduced below 700 ℃ and inhibited the adsorption of active oxygen species, it can be inferred that WO₃ supported on CeO₂ would prevent H₂ from reducing CeO₂, which weakened the redox activity of the W/Ce catalysts compared with that of CeO₂. In addition, compared with 1.6W/Ce, the reduction temperature of β peaks in 1.6W/Ce-N shifted to lower temperature by ca. 70 ℃, which further indicates that crystalline WO₃ could inhibit the redox ability of the W/Ce catalysts. This is one of the reasons why 1.6W/Ce exhibited lower catalytic activity than that of 1.6W/Ce-N.

NH₃-TPD experiments were conducted to investigate the intensity and amount of acid sites on pure CeO₂ and W/Ce samples. As illustrated in Fig. 6, pure CeO₂ displayed a peak from 100 to 300 ℃, indicating that adsorbed NH₃ was totally desorbed at ca. 300 ℃. Meanwhile, the W/Ce samples exhibited a broader peak and adsorbed NH₃ was totally desorbed at ca. 370 ℃. Thus, modification with WO₃ could increase the adsorption strength of NH₃ on the W/Ce catalysts. Peak areas in NH₃-TPD curves were calculated and are listed in Table 1. Compared with that of pure CeO₂, the number of acid sites increased considerably in W/Ce samples, which was caused by the new acid sites provided by dispersed WO₃, as reported elsewhere [35]. For the W/Ce samples, the content of acid sites first increased and then decreased with increasing WO₃ loading. Interestingly, 0.8W/Ce-MM had the lowest content of acid sites in all the samples. Combined with the XRD results, these findings suggest that crystalline WO₃ covered the surface of W/Ce and could block the adsorption of NH₃, which inhibited the catalytic activity of the W/Ce samples.

NH₃ adsorption was measured by in situ DRIFT spectroscopy to further investigate the influence of WO₃ loading on the content of surface acid sites of W/Ce samples. As shown in Fig. 7, for pure CeO₂, bands at 1123 and 1570 cm⁻¹ were ascribed to the adsorption of NH₃ on Lewis acid sites of CeO₂ [36]. For 0.2W/Ce, as well as bands at 1123 and 1570 cm⁻¹, new bands at 1175 and 1600 cm⁻¹ were attributed to the adsorption of NH₃ at Lewis acid sites provided by W⁶⁺ of isolated tungstate species [16]. When the WO₃ loading exceeded 0.4 mmol/(100 m²), new bands at 1423 and 1664 cm⁻¹ assigned to the adsorption of NH₃ on Brönsted acid sites were formed. The Brönsted acid sites were mainly provided by the W–OH–W motif of polytungstate species [16, 37]. The peak intensity of Lewis acid sites at 1123 and 1570 cm⁻¹ decreased with increasing WO₃ loading and totally disappeared when the WO₃ loading was higher than 0.8 mmol/(100 m²); this was because the dispersed WO₃ prevented NH₃ adsorption on the CeO₂ surface. At the same time, the ratio of Brönsted to Lewis acid sites provided by WO₃ species increased with WO₃ loading, consistent with the results reported for WO₃/TiO₂ and WO₃/ZrO₂ [35, 37]. Considering these findings along with XRD and Raman results, isolated WO₃ species mainly provided Lewis acid sites and polytungstate species provided Brönsted acid sites over WO₃/CeO₂. Thus, it can be concluded that the states of WO₃ played an important role in determining the surface acid properties of the W/Ce samples.

DRIFT spectra were measured following NO + O₂ adsorption by the W/Ce samples to investigate the influence of WO₃ on the adsorption of nitrate species. As shown in Fig. 8, bands at 1606, 1583, and 1230 cm⁻¹ were assigned to bridging nitrate species, those at 1505 and 1270 cm⁻¹ were attributed to monodentate nitrate species, and a weak band at 1305 cm⁻¹ was attributed to nitrite species [18, 38]. The intensity of bands originating from nitrate species decreased and even disappeared with increasing WO₃ loading. Only nitrite species formed when the WO₃ loading exceeded 0.8 mmol/(100 m²). Supporting WO₃ inhibited the adsorption of nitrate on WO₃/ZrO₂ catalysts [18]. For W/Ce samples, considering the H₂-TPR results and O 1s spectra, increased coverage of WO₃ not only weakened the oxidation ability of the W/Ce catalysts but also diminished their content of surface active oxygen species. Thus, O₂ activation and NO oxidation were more difficult over W/Ce compared with the case over pure CeO₂. When WO₃ loading was higher than the dispersion capacity of WO₃ on CeO₂, the samples oxidized NO to nitrite instead of nitrate. Thus, increasing WO₃ loading can inhibit NO oxidation and nitrate species adsorption on the W/Ce samples.

DRIFT spectra were also measured following adsorption of NO, NH₃, and O₂ to study surface adsorption on W/Ce samples under real reaction conditions. As illustrated in Fig. 9, for pure CeO₂, bands at 1275 and 1256 cm⁻¹ were respectively assigned to monodentate and bidentate nitrate species, which indicates that nitrates were the stable adsorbed species in the NH₃ SCR reaction process. With regard to the W/Ce samples, bands observed at 1423, 1175, and 1123 cm⁻¹ were attributed to adsorbed NH₃ species, and no nitrate species were found (bands at 1275 and 1256 cm⁻¹ were very weak for 0.2W/Ce). These findings further confirm that supporting WO₃ could inhibit NO adsorption, and the stable adsorbed species on W/Ce was NH₃ under the reaction conditions. Both Fig. 7 and Fig. 9 show that only pure CeO₂, 0.2W/Ce, and 0.4W/Ce had a peak from Lewis acid sites at 1123 cm⁻¹ related to the adsorption of NH₃ at CeO₂ sites. Interestingly, these three samples displayed much lower NO conversion and N₂ selectivity than those of 0.8 W/Ce, following the order CeO₂ < 0.2W/Ce < 0.4W/Ce < 0.8W/Ce. The TPR and XPS results revealed that the redox ability and oxygen activation ability of the CeO₂ surface were inhibited by modification with WO₃. Thus, we infer that NH₃ adsorbed at CeO₂ sites would induce the nonselective oxidation of NH₃ more easily than that at WO₃ sites because of the higher redox activity of CeO₂ sites than that of WO₃ ones under real reaction conditions, which was unfavorable to improve catalytic performance.

In situ DRIFT spectroscopy is helpful to study the mechanism of surface reactions. In this experiment, NH₃ was preadsorbed on 0.4W/Ce by introduction of NH₃ for 1 h, purged with N₂ for 15 min, and then introducing NO and O₂. Spectra were collected as a function of time. As shown in Fig. 10(a), when NO and O₂ were introduced, with increasing reaction time from 0 to 6 min, bands from adsorbed NH₃ species (Lewis acid at 1173 cm⁻¹ and Brönsted acid at 1421 cm⁻¹) gradually decreased in intensity and finally disappeared. In the meantime, no nitrate species were formed. It can therefore be inferred that adsorbed NH₃ species were not replaced by nitrate species via competitive adsorption, but reacted with NO and O₂, which suggests that the NH₃ SCR reaction proceeded by the Eley-Rideal mechanism over the W/Ce samples.

In a different experiment, NO and O₂ were preadsorbed on 0.4W/Ce by introduction of NO and O₂ for 1 h, followed by purging with N₂ for 15 min. Figure 10(b) indicates that the main adsorbed species were bridging nitrate species (1606, 1583, and 1230 cm⁻¹) on the surface of the sample. When introducing NH₃, the intensity of bands from bridging nitrate species gradually decreased and those of bands from monodentate (1506 and 1264 cm⁻¹) and bidentate (1552 cm⁻¹) nitrate species gradually increased. In addition, bands from adsorbed NH₃ species (Brönsted acid at 1438 cm⁻¹ and Lewis acid at 1175 cm^–1) appeared. In our previous studies [39, 40], we found that bridging nitrate species adsorbed on Ce/TiSn and Ti-Cu/CeO₂ catalysts also disappeared and new monodentate and bidentate nitrate species formed when NH₃ was introduced. We propose that one NH₃ molecule could snatch an adsorption site from a bridging nitrate to form one Lewis acid, and then the bridging nitrate is transformed to a monodentate or bidentate nitrate. In terms of the present results, competitive adsorption between NH₃ gas and adsorbed bridging nitrates also occurred on 0.4W/Ce, similar to that observed on Ce/TiSn and Ti-Cu/CeO₂. Figure 10(b) shows that after introducing NH₃ for 10 min, the band intensity of adsorbed NH₃ and nitrate species was nearly unchanged, which indicates that adsorbed NH₃ and adsorbed nitrate species did not react with each other. Moreover, the DRIFT spectra for both NO and O₂ adsorption and NO, NH₃, and O₂ adsorption indicated that no nitrate species were adsorbed on 0.8W/Ce. However, 0.8W/Ce exhibited the highest catalytic activity in the samples. Therefore, we infer that the NH₃ SCR reaction over the W/Ce samples did not occur via the Langmuir-Hinshelwood mechanism. Our present results were not in accord with the literature finding that adsorbed nitrate species could react with adsorbed NH₃ species over WO₃-CeO₂ prepared by a coprecipitation method [41]. It is possible that the preparation method of the WO₃-CeO₂ catalyst influenced its reaction mechanism.

Based on the above in situ DRIFT spectral results, the adsorption behavior on the W/Ce samples included the following factors: (1) WO₃ loading affected surface acid properties, and the relative content of Brönsted acid sites rose with increasing WO₃ loading; (2) the adsorption ability of nitrate species was inhibited by the increased coverage of WO₃ and only nitrite formed when the WO₃ loading exceeded the dispersion capacity of WO₃ on CeO₂; (3) NH₃ was most stable adsorbed species in the NH₃ SCR reaction; (4) the NH₃ SCR reaction proceeded via the Eley-Rideal reaction route. Thus, when WO₃ loading was close to the dispersion capacity of WO₃ on CeO₂, the effect of NH₃ adsorption and activation was maximized to promote the Eley-Rideal reaction route. As a result, the 0.8W/Ce sample displayed the highest catalytic activity of those investigated.

We studied the influence of WO₃ loading on the performance of WO₃/CeO₂ catalysts in NO_x degradation. Several major conclusions were obtained. (1) Dispersed WO₃ was beneficial to improve catalytic activity while crystalline WO₃ inhibited catalytic activity. The 0.8W/Ce sample displayed the highest catalytic activity in the SCR reaction. (2) Crystalline WO₃ could be removed in 25% ammonium hydroxide to recover the catalytic activity of 1.6W/Ce. (3) Supporting WO₃ could inhibit the redox ability of CeO₂, which blocked NO oxidation and nitrate adsorption, while the modification of CeO₂ with WO₃ enhanced the content of surface acid sites. (4) In situ DRIFT spectra suggested that the NH₃ SCR reaction over the supported WO₃/CeO₂ catalysts mainly proceeded via the Eley-Rideal mechanism. As a result, when WO₃ loading was lower than the dispersion capacity of WO₃ on CeO₂, the oxidation-adsorption performance of NO was inhibited and the adsorption-activation performance of NH₃ was improved to promote the NH₃ SCR reaction via the Eley-Rideal route with increasing WO₃ loading.