The conventional technique for the control of NO_x emissions is selective catalytic reduction (SCR) of NO_x by NH₃ ^[1] using various commercial catalysts containing V₂O₅, WO₃, and TiO₂. However, the use of such commercial catalysts is limited because of their low activities (300-400 ℃, 85%) ^[2]. Cheap transition-metal-doped catalysts with exceptional catalytic performances are now widely used in several industrial processes ^[3]. Previous studies have shown that TiO₂-ZrO₂ supports have higher thermal stabilities and amounts of acid sites than do pure TiO₂ and ZrO₂ ^[4]. In this study, we investigated and compared the NH₃-SCR mechanisms (above 300 ℃) over V₂O₅-x%WO₃/TiO₂-ZrO₂ and V₂O₅/9%WO₃-TiO₂ catalysts.

Although the NH₃-SCR mechanism has been widely studied, it is still the subject of much debate. This is because (1) different catalysts have been used, resulting in different active sites and (2) different temperatures have been used in NH₃-SCR reactions over different catalysts. Recently, the reactions NH_3(ads) → NO_(g) + O_2(g) and NO_(ads) + O_2(ads) → NH_3(g) over different catalysts at different temperatures have been studied and possible reaction pathways have been proposed ^[5-8]. However, the adsorption behaviors of NH₃ and NO at different temperatures were not clear, e.g., whether and how NH_3(g), NO_(g), and NO_(g) + O_2(g) are adsorbed on the catalyst surface above 300 ℃. The differences among the adsorption intensities of these gases are also unclear. Most previous theories have indicated that the reaction involves sequential H abstraction (or oxidation) of NH₃ species to give the coordinated form of NH₃ or the protonated form, i.e., NH₄⁺, followed by partial oxidation by metal oxides (V₂O₅, WO₃, and TiO₂) ^[5-8]. The catalyst surface could therefore be reduced by NH₃ or NH₄⁺. However, temperature-programmed reduction by NH₃ (NH₃-TPR) of the catalysts has not been investigated, and the proposed catalyst surface reduction by NH₃ is controversial. All these points have important implications for the NH₃-SCR mechanism, and intensive research to clarify these issues is needed.

In this study, V₂O₅-WO₃/TiO₂-ZrO₂ catalysts with various WO₃ loadings were synthesized and used in NH₃-SCR. The effects of the amounts of WO₃ and ZrO₂ on the catalytic performances and reaction mechanisms over V₂O₅-WO₃/TiO₂-ZrO₂ were investigated using the Brunauer-Emmett-Teller (BET) method, X-ray diffraction (XRD), TPR, and in situ diffuse-reflectance infrared Fourier-transform spectroscopy (DRIFTS).

The V₂O₅-x%WO₃/TiO₂-ZrO₂ catalysts were prepared by impregnation of a TiO₂-ZrO₂ solid solution with ammonium tungstate hydrate and ammonium vanadate.

A TiO₂-ZrO₂ solid solution (Ti:Zr molar ratio 1:1) was prepared using a precipitation method. ZrOCl₂·8H₂O and TiCl₄ (Sinophram Chemical Reagent) were used as precursors. Ammonia water (25%-28%) was added dropwise to the metal precursor solution at 25 ℃ until the pH was between 9 and 10. The obtained milky slurry was aged in the solution at 25 ℃ for 24 h. The hydroxide was filtered off, washed with deionized water, and dried at 110 ℃ for 12 h. The samples were calcined at 450 ℃ in air for 3 h before milling and screening to 60 mesh.

An incipient wetness impregnation method was used to synthesize x%WO₃/TiO₂-ZrO₂ (x = 0, 6, 9, 12, weight ratio). (NH₄)₁₀W₁₂O₄₁·xH₂O (Sinophram Chemical Reagent) was added to the TiO₂-ZrO₂ carrier and the mixture was stirred at 35 ℃ for 2 h. The residue obtained by evaporation at 80 ℃ was dried at 110 ℃ for 12 h. The samples were calcined at 450 ℃ in air for 3 h, and milled and screened to 60 mesh. The same procedure was used to synthesize 9%WO₃/TiO₂ with anatase TiO₂ as the carrier.

A wet impregnation method was used to synthesize 1%V₂O₅-x%WO₃/TiO₂-ZrO₂ and 1%V₂O₅-9%WO₃/TiO₂ as follows. NH₄VO₃ (Sinophram Chemical Reagents) was added to x%WO₃/TiO₂-ZrO₂ powder and 9%WO₃/TiO₂ powder; the rest of the process was the same as that used to prepare x%WO₃/TiO₂-ZrO₂.

We denote 1%V₂O₅-x%WO₃/TiO₂-ZrO₂ by x%WO₃, where x represents the amount of WO₃ in the catalyst.

XRD patterns of the catalysts were obtained using an XD-3 diffractometer (Beijing PEPSEE, China) with Cu K_α radiation (λ = 0.15406 nm). The specific surface areas of the catalysts were measured based on N₂ adsorption-desorption at liquid N₂ temperature (-196 ℃) using a Micromeritics F-sorb 3400 adsorption apparatus (Beijing Jinaipu, China). NH₃-TPR was performed in a quartz U-tube reactor connected to a thermal conductivity detector, using an NH₃-He mixture (0.503 vol% NH₃) as a probe. TPR was performed from 25 to 700 ℃ at a rate of 10 ℃/min.

In situ (DRIFTS) was performed using a Nicolet 6700 FT-IR instrument (Thermo Electron Corporation, USA) with a Harrick accessory (HTC-3, Harrick Scientific Corporation, USA) in the wavenumber range 400-4000 cm^-1 at a 4 cm^-1 resolution. NH₃-adsorbed in situ DRIFTS was performed as follows. (1) A thin, intact, self-supporting wafer of the catalyst (10 mg) was prepared and mounted inside a high-temperature cell (HTC-3, Harrick Scientific Corporation, USA) and subjected to a high-purity N₂ stream at 400 ℃ for 1 h. (2) After cooling to room temperature, the wafer was exposed to an NH₃ (99.999%) stream at a flow rate of 5 mL/min for 30 min. (3) The wafer was flushed with a high-purity N₂ stream for 30 min. (4) IR spectra were recorded at various target temperatures in a high-purity N₂ stream; the temperature was increased at a rate of 10 ℃/min.

The catalytic activity was investigated in a fixed-bed stainless-steel reactor (i.d. 6 mm) at 200-450 ℃, using 300 mg of catalyst. A gas mixture of 0.08% NO, 0.08% NH₃, and 5% O₂ (by volume) was fed into the reactor and N₂ was used to give a total flow rate of 100 mL/min (gas hourly space velocity (GHSV) = 20000 h^-1). The NO and NO₂ concentrations were continually monitored using a flue gas analyzer (Testo 350-XL).

The specific surface areas and pore size distributions of the catalysts are shown in Table 1 and Fig. 1. The pore size distributions clearly show that the amounts of pores of sizes 384.2-8.3 nm and 3.1-1.7 nm were unchanged by WO₃ addition, but the amounts of pores of size 8.3-3.9 nm increased greatly. This indicates that WO₃ addition enhanced the thermal stability of the pore structure, especially of pores of size 8.3-3.9 nm. The BET surface areas (S_BET) decreased with increasing WO₃ loading. The XRD patterns (Fig. 2) show that no crystalline tungsten oxide was present on the catalyst surfaces. This suggests that amorphous WO₃ was dispersed on the surface of the TiO₂-ZrO₂ support and occupied the pores, decreasing the surface area. WO₃ addition increased the BET surface area of V₂O₅/TiO₂-ZrO₂. This suggests that WO₃ acts as a structural promoter by improving the structural properties of the catalysts ^{[9, 10]}. The BET surface area of the four-component catalyst obtained by ZrO₂ addition was considerably higher than that of the triple-oxide catalyst. This is because of the high surface area of the TiO₂-ZrO₂ solid-solution support. However, the results of experiments using the V₂O₅/TiO₂-ZrO₂ and V₂O₅-9%WO₃/TiO₂ catalysts suggest that the surface area may not determine the catalytic performance.

The XRD patterns of the catalysts are shown in Fig. 2. The tetragonal phase of vanadium oxide was identified in the V₂O₅/TiO₂-ZrO₂ catalyst; the peak at 27.5° was ascribed to rutile TiO₂ and those at 25.3° and 53.97° were ascribed to anatase TiO₂ ^[11]. Sharp peaks ascribable to zirconium oxide were clearly observed. The V₂O₅-x%WO₃/TiO₂-ZrO₂ catalysts had similar XRD patterns; no crystalline-phase peaks were observed, suggesting that these metal oxides did not form large crystals, indicating that addition of WO₃ improved dispersion of metal oxides on the support. A comparison of the XRD pattern of 9%WO₃ with that of the V₂O₅-9%WO₃/TiO₂ catalyst suggests that ZrO₂ doping facilitated dispersion of all metal oxides, with no crystalline peaks being observed for titanium oxide (anatase or rutile). These results show that addition of WO₃ and ZrO₂ gave well-dispersed four-component catalysts.

The surface acidities of the catalysts were investigated using in situ DRIFTS analysis of NH₃ adsorption. The bands at 1155-1222 and 1600-1624 cm^-1 in the spectra can be assigned to symmetric and asymmetric vibrations, respectively, of the N-H bonds in NH₃ coordinated to Lewis acid sites ^{[12, 13]}. The bands at 1450 and 1680 cm^-1 correspond to the asymmetric and symmetric vibrations, respectively, of NH₄⁺ ions formed by NH₃ chemisorption at Brönsted acid sites ^[14]. All the samples were exposed to NH₃ (10 mL/min), followed by N₂ purging at 25 to 400 ℃.

For the V₂O₅/TiO₂-ZrO₂ catalyst (Fig. 3(a)), bands appeared at 1180, 1450, and 1680 cm^-1 after N₂ purging; this implies that the catalyst has both Brönsted and Lewis acid sites. The bands at 1450 and 1680 cm^-1 completely disappeared at 200 ℃, but the band at 1180 cm^-1 was still present at 400 ℃. This suggests that the NH₄⁺-Brönsted acid species was not very stable ^[15].

For the WO₃-modified counterpart (Fig. 3(b)-(d)), the band at 1450 cm^-1 did not vanish until the temperature was 250 ℃, suggesting that WO₃ addition improved the stability of the Brönsted acid species ^[14]. Fig. 4 shows that the intensity of the peak at 1180 cm^-1 was not affected by the amount of WO₃ added, but the intensity of the band at 1454 cm^-1 was: V₂O₅-9%WO₃/TiO₂-ZrO₂ > V₂O₅-12%WO₃/TiO₂-ZrO₂ > V₂O₅-6%WO₃/TiO₂-ZrO₂ > V₂O₅/TiO₂-ZrO₂. This implies that WO₃ addition increases the number of Brönsted acid sites.

Fig. 3 shows that at 250 ℃ the amount of NH₄⁺ (1450 cm^-1) decreased significantly at 250 ℃, but coordinated NH₃ could still be observed above 250 ℃. The residence time of coordinated NH₃ was therefore longer than that of adsorbed NH₄⁺. NO_x reduction is a surface reaction, therefore a longer residence time benefits NO_x reduction. For the V₂O₅-9%WO₃/TiO₂-ZrO₂ sample (Fig. 4), the intensity of the Lewis acid peak was weaker than that for V₂O₅-12%WO₃/TiO₂-ZrO₂, but the intensity of the Brönsted acid peak was much stronger. The large number of Brönsted acid sites enabled reduction of a large amount of NH₄⁺ in a short residence time (Fig. 6(a)). This is one of the reasons why NO_x conversion over V₂O₅-9%WO₃/TiO₂-ZrO₂ was higher than that over V₂O₅-12%WO₃/TiO₂-ZrO₂.

Fig. 5 shows that a weak band appeared at 1550 cm^-1 after NH₃ adsorption on the V₂O₅-x%WO₃/TiO₂-ZrO₂ samples. Similar bands from intermediate species have previously been detected over vanadium-based catalysts after reaction with NH₃, and were assigned to -NH₂ amide species ^{[12, 16]}. This result confirms that WO₃ addition promoted NH₃ activation.

Fig. 6(a) shows the SCR activities of the V₂O₅-x%WO₃/TiO₂-ZrO₂ catalysts. The highest NO_x conversion was achieved over V₂O₅-9%WO₃/TiO₂-ZrO₂ between 300 and 450 ℃. The reasons might be as follows. (1) It had a large number of Brönsted acid sites (Fig. 4); (2) the pores of size 8.3-3.9 nm (mesopores: 50-2 nm) were stable (Fig. 1); (3) the good dispersion obtained by addition of 9%WO₃ increased strong interactions among the metal oxides (Fig. 2). Although the BET surface area of V₂O₅/TiO₂-ZrO₂ was much higher than that of V₂O₅-9%WO₃/TiO₂, the catalytic performance of the latter sample was better in the required temperature window, implying that the surface area was not the major factor in the catalytic performance. It can be concluded that a combination of the above-mentioned properties of the V₂O₅-9%WO₃/TiO₂-ZrO₂ sample was the main reason for its higher catalytic performance.

Fig. 6(b) shows that without O₂ (at the same GHSV), NO_x conversion decreased greatly, indicating that the presence of O₂ accelerates NO_x conversion. The role of O₂ will be discussed in detail in Section 3.6.

The effects of SO₂ and H₂O on NO_x reduction were also studied, using 6% H₂O and various amounts of SO₂ in the gas mixture at the same GHSV. Fig. 7 shows that NO_x conversion decreased significantly in the presence of H₂O and SO₂. When the H₂O and SO₂ feeds were stopped, the catalytic activity was not completely recovered, which implies that H₂O and SO₂ poisoning of the catalysts was partly irreversible. The poisoning effect was more apparent with increasing SO₂ concentration.

The adsorption properties of the SCR reactants over V₂O₅-x%WO₃/TiO₂-ZrO₂ catalysts were investigated extensively. Figs. 8(a) and 9(a) show DRIFT spectra of the catalysts after exposure to NH₃/N₂ at for 30 min at 300 and 350 ℃. The bands at 1610, 1214, and 1321 cm^-1 are attributed to asymmetric and symmetric deformation, respectively, of NH₃ coordinated with Lewis acid sites and adsorbed NH₃ gas. Coordinated NH₃ is clearly the dominant species during NH₃ adsorption. Figs. 8(b) and 9(b) show that a weak band appeared at 1621 cm^-1 (ν_as(NO₂)), indicating that SCR involved gaseous NO or weakly adsorbed NO₂; these results are consistent with those previously reported ^{[17, 18]}.

The IR spectra of NO + O₂ co-adsorbed at 300 ℃ are shown in Fig. 10. After introduction of NO + O₂ into the cell for 30 min, the same band appeared at 1621 cm^-1, assigned to ν_as(NO₂) ^[19-22]; the intensity of the NO band was the same regardless of whether or not O₂ was introduced. NO and O₂ readily react to produce NO₂, but NO₂ is not easily adsorbed on the catalyst surface. A comparison of the adsorption states and intensities of NH₃ and NO showed the following. (1) Above 300 ℃, NH₃-SCR involved strongly adsorbed NH₃ and gaseous or weak NO₂ species, and NO could only be adsorbed onto the catalyst surface as NO₂. (2) Above 300 ℃, the catalytic reaction over V₂O₅-x%WO₃/TiO₂-ZrO₂ could only follow the Eley-Ridel mechanism (the catalytic reaction occurred between adsorbed molecules A and B; molecule B was weakly adsorbed or not adsorbed). (3) For the V₂O₅-x%WO₃/TiO₂-ZrO₂ samples, it can be concluded that the catalyst activated NH₃ and could not activate NO. This conclusion is supported by the suggestion that NO reacts in the gas phase and is almost unable to be adsorbed and then activate NO.

The V₂O₅-x%WO₃/TiO₂-ZrO₂ catalysts had similar adsorption properties, therefore V₂O₅-9%WO₃/TiO₂-ZrO₂ was selected to study the NH₃-SCR mechanism in-depth because it gave the highest NO_x conversion. The investigation was performed at 300 ℃ using NO + O₂ as the reactants. Figs. 11 and 12 show the in situ DRIFT spectra over V₂O₅-9%WO₃/TiO₂-ZrO₂ pretreated by exposure to NH₃/N₂ (5 mL/min), followed by NO (5 mL/min) + O₂ (10 mL/min) at 300 ℃. During NH₃ adsorption and N₂ purging (Figs. 11(a) and 12(a)), bands were detected at 1204, 1550, 1602, 3156, 3264, and 3347 cm^-1. The band at 1550 cm^-1 was ascribed to the -NH₂ scissoring mode of an inorganic amide species, and those at 1602, and 1204 and 3347 cm^-1 were ascribed to asymmetric and symmetric deformations, respectively, of coordinated NH₃. The bands at 3156 and 3264 cm^-1 were ascribed to the overtone of the asymmetric NH₃ deformation as a result of Fermi resonance ^[23]. It is worth noting that all the bands remained during N₂ purging. These results show that Lewis acid sites were the active acid sites above 300 ℃. When the NH₃-loaded catalyst was exposed to NO + O₂ (Figs. 11(b) and 12(b)), all the band intensities decreased continuously in the period from 30 to 50 min, possibly because NO_(g) reacted with the -NH_2(ads) ^[24] to form the intermediate [NH₂-NO]; this activated complex then decomposed to yield N₂ and H₂O. During the NO + O₂ reaction, a new band appeared at 1595 cm^-1, ascribed to the bending vibration of H₂O ^[25]. When NO + O ₂ was introduced onto the NH₃-loaded catalyst for 30 min (Fig. 12(c)), a new peak emerged at 3657 cm^-1 (symmetric deformation of H₂O). These results prove that H₂O was formed during the NO + O₂ reaction.

Figs. 13 and 14(a) show in situ DRIFT spectra of the V₂O₅-9%WO₃/TiO₂ catalysts exposed to NO (Fig. 13(a)), NO + O₂ (Fig. 13(b)), and NH₃ (Fig. 14(a)) at 300 ℃. Fig. 13 shows two bands at 1365 and 1621 cm^-1, which were assigned to the symmetric and asymmetric deformations, respectively, of NO₂ ^[25]; the intensities of these two bands were close to zero regardless of whether or not O₂ was introduced. The bands at 1248, 1325, 1437, and 1668 cm^-1 in Fig. 14(a) were ascribed to the symmetric stretching vibrations of NH₃, gas-adsorbed NH₃, and symmetric and asymmetric deformations of NH₄^{+ [18]}. The following conclusions can be drawn from Figs. 13 and 14(a). (1) NO can only be adsorbed on the surface in the form of NO₂. (2) Adsorbed NH₄⁺ and coordinated NH₃ were the dominant species during NH₃ adsorption and both of them were stronger adsorbates than NO; these results are consistent with those of research on vanadia-based catalysts ^[17]. (3) Brönsted and Lewis acid sites were the active sites on the V₂O₅-9%WO₃/TiO₂ catalyst above 300 ℃. (4) The Brönsted acid sites were not always weaker than the Lewis acid sites. Ramis reported that NH₄⁺ adsorbed on Brönsted acid sites was less thermally stable than coordinated NH₃, indicating less involvement of NH₄⁺ in SCR under the conditions used ^[5]. The adsorption states of NH₃ and NO suggest that the catalytic reaction followed the Eley-Ridel mechanism in the case of V₂O₅-9%WO₃/TiO₂.

In situ DRIFT spectra of the V₂O₅-9%WO₃/TiO₂ catalyst after adsorption of NH₃ and then flushing with NO + O₂ are shown in Fig. 14(b). During NO + O₂ purging, the bands at 1260 (δ_s(NH₃)), 1302 (adsorbed NH₃ gas), 1440 (δ_as(NH₄⁺)), 1680 (δ_s(NH₄⁺)), 3152 and 3264 (γ(NH₃)), and 3347 (γ_s(NH₃)) continually decreased as NO + O₂ was fed to the NH₃-saturated catalyst. Fig. 15(e) (partial enlarged drawing) shows that the band at 1543 (-NH₂) also disappeared gradually as NO + O₂ was introduced into the NH₃-saturated catalyst. The pathway of NO_x removal might therefore be as follows: NH_4(ads)⁺ reacts with NO_(g), NH_2(ads) reacts with NO_(g) and then the complexes [NH₄⁺-NO] and [NH₂-NO] decompose to N₂ and H₂O.

After feeding a gas mixture of NO and O₂, all the above bands vanished (Fig. 14(b)), which clearly shows that NH₃ can reduce NO all the way from NO_(g) to N₂. Two new bands, at 3657 and 3733 cm^-1, appeared simultaneously after purging with NO + O₂ for 30 min (Fig. 14(c)); these provide good evidence of the formation of adsorbed H₂O ^{[26, 27]}.

IR spectra of V₂O₅-9%WO₃/TiO₂-ZrO₂ and V₂O₅-9%WO₃/TiO₂ were recorded to enable further understanding of the role of O₂ in SCR. The samples were pretreated with NH₃ (5 mL/min) at 300 ℃, followed by NO (5 mL/min) and N₂ (10 mL/min). Fig. 15 shows that intensities of the bands at 1260 (δ_s(NH₃)), 1450 (δ_as(NH₄⁺)), and 1542 (γ(NH₂)) cm^-1 (Fig. 16(a)), 3156 and 3264 (γ(NH₃)), and 3347 (γ_s(NH₃)) cm^-1 (Fig. 15(a) and (b)) decreased slowly with NO feeding. The band at 3600-3800 cm^-1 (OH groups) was not observed after purging with NO (Fig. 16(b)). For V₂O₅-9%WO₃/TiO₂-ZrO₂ (Fig. 15(c) and (d)), similar phenomena were observed, i.e. the bands at 1204, 1542 (Fig. 16(c)), 3167, 3264, and 3347 cm^-1 slowly shrank when NO was introduced. The band at around 3370 cm^-1, corresponding to H₂O, did not appear (Fig. 16(d)).

Fig. 6(b) shows that the NO_x conversions over V₂O₅-9%WO₃/TiO₂-ZrO₂ and V₂O₅-9%WO₃/TiO₂ decreased significantly when O₂ was not present; this suggests that gaseous O₂ was crucial in the formation of H₂O. Fig. 17 also shows that NO was easily oxidized by O₂. Most researchers proposed ^{[14, 28, 29]} that the reaction of NH₃ with NO, resulting in partial reduction of O₂, can be explained by the need for the surface to be reoxidized by gaseous O₂ to complete the catalytic cycle. We used NH₃-TPR and H₂-TPR to confirm this. In the H₂-TPR profiles in Fig. 18, the peak at 385 ℃ was assigned to V₂O₅ → V₂O₃ ^[30], the peak at 697 ℃ was assigned to V₃O₁₃ → V₂O₄, the peak at 545 ℃ was assigned to TiO₂ → Ti_xO_y, and the peak at 750 ℃ was assigned to WO_2.9 → W. These results show that the catalyst could be reduced by H₂. However, the NH₃-TPR results suggested otherwise. A new peak appeared at 575 ℃, indicating that a new gas (with a different thermal conductivity coefficient) was formed, as a result of the design of the TPR instrument; above 575 ℃, the electrical signal suddenly changed significantly. It is possible that NH₃ decomposed to N₂ and other gases above 500 ℃. The NH₃-TPR results show that the metal oxides in the catalyst were not reduced by NH₃. The original hypothesis that the catalyst surface would be reoxidized by O₂ to complete the catalytic cycle is incorrect, and gaseous O₂ merely takes part in the formation of H₂O and NO₂.

Most research clearly supports the Brönsted-NH₄⁺mechanism. This mechanism was proposed by Topsǿe ^{[14, 31, 32]}, who suggested that adsorbed NH₃ was activated by transfer of an H atom to V⁵⁺=O, leading to reduction of gaseous or weakly adsorbed NO, with subsequent reaction with the activated NH₃, to yield N₂ and H₂O with release of V⁵⁺-OH and V⁴⁺-OH; the cycle was completed by oxidation of V⁴⁺-OH to V⁵⁺=O. Another mechanism (the Lewis acid mechanism) was proposed by Ramis et al. ^[33] and supported by Bagnasco et al. ^[34]. This mechanism also involves the transformation of Mⁿ⁺ to M^(n-1)+. The same conclusion was reached, i.e., that the catalyst was reduced by NH₃. However, the NH₃-TPR results show that the catalyst was not reduced by NH₃ during NH₃-SCR (V₂O₅-9%WO₃/TiO₂ and V₂O₅-9%WO₃/TiO₂-ZrO₂ gave similar NH₃-TPR results). The traditional mechanism is therefore not supported. V₂O₅ was the main active ingredient during NH₃-SCR; the NO_x conversions over TiO₂-ZrO₂ and V₂O₅/TiO₂-ZrO₂ provide the best proof (Fig. 19); WO₃ improved the stability of the catalyst structure (Fig. 1). The NO_x removal pathways via Lewis and Brönsted acid sites are shown in Fig. 20.

The pathways can be explained as follows. For Lewis acid sites, NH_3(g) is adsorbed on the catalyst surface in the form of a coordination compound. When the unshared electron pair enters the 3d orbital of V, the N-H bond is broken because of the strong electronegativity of the O atom (O 3.44 > N 3.04) and then intermediate 3 appears. Lastly, when gaseous NO is fed to the NH₃-adsorbed catalyst, the intermediate [NH₂-NO]* decomposes to N₂ and H₂O. For the Brönsted acid sites, the adsorbed NH₃ generates NH₄⁺, which reacts with NO_(gas) to form the intermediate [NH₄⁺-NO]*. Finally, this activated complex decomposes via [NH₄⁺-NO], presumably to N₂ and H₂O. During NH₃-SCR, the catalyst is not reduced by NH₃, unlike the traditional NH₃-SCR pathway.

Addition of sufficient WO₃ stabilized the pore structure and a combination of WO₃ and ZrO₂ improved dispersion of all the metal oxides. In particular, addition of 9%WO₃ addition significantly increased the amount of Brönsted acid sites and improved the thermally stability of the pore structure (3.9-8.3 nm); this contributed to the higher NO_x conversion over V₂O₅-9%WO₃/TiO₂-ZrO₂. ZrO₂ doping considerably increased the BET surface area of the triple-oxide catalyst (V₂O₅-9%WO₃/TiO₂) and improved dispersion of the active components on the surface. However, its main promotional role was to change the surface acidity of V₂O₅-WO₃/TiO₂. The pathway of NO_x reduction changed accordingly. There are only Lewis acid sites on the catalyst supported on TiO₂-ZrO₂, but there are both Brönsted and Lewis acid sites on the catalyst supported on TiO₂. The adsorption properties of NO (gaseous or weakly adsorbed) and NH₃ (strongly adsorbed) determine the catalytic mechanism. The catalysts studied follow the Eley-Ridel mechanism, but different dominant acid sites lead to different pathways for NO_x reduction, i.e., via [NH₄⁺-NO-Brönsted acid site] intermediates for V₂O₅-x%WO₃/TiO₂-ZrO₂ and via [NH₂-NO-Lewis acid site] intermediates for V₂O₅-9%WO₃/TiO₂. The metal oxides in the catalyst are not reduced by NH₃ and O₂ cannot reoxidize the catalyst surface, but participates in the formation of H₂O and NO₂.