Nitrogen oxides (NO_x) are major air pollution emission in industrial processes and human activities, which need to be tightly controlled under stringent environmental regulations. Selective catalytic reduction of NO with NH₃ (NH₃-SCR) over V₂O₅-WO₃/TiO₂ catalyst is an efficient and cost-effective technology for eliminating NO_x from stationary sources [1-3]. This process, however, can be operated only in a rather high and narrow temperature range of 300–400 ℃ [4, 5]. Unlike the case in a coal-fired power plant, the temperature of NO_x-containing exhaust gases from other sources such as glass furnace, metallurgy sintering furnace and cement plant is often lower than 300 ℃ [6-9]. Moreover, the biological toxicity caused by vanadium species also limits the application of vanadium-based DeNO_x catalysts. Hence, the design of vanadium-free SCR catalysts that are sufficiently active at low temperature is highly desired.

In the past few years, Ce-based oxides emerged as a promising candidate for vanadium-based NH₃-SCR catalysts because of the high oxygen storage capacity (OSC) and facile Ce⁴⁺/Ce³⁺ redox cycle of CeO₂ [6, 10-13]. Among them, Ce-Ti mixed oxide catalyst has drawn much attention because of its large surface area and good redox ability [14-16]. For instance, CeO₂-TiO₂ catalyst for NH₃-SCR reaction was first synthesized by Xu et al. [17] and admirable catalytic performance was observed at 275–400 ℃. Afterwards, Gao et al. [18] explored the influence of preparation methods on the NH₃-SCR performance of CeO₂-TiO₂ catalyst and found that the catalyst prepared by single step sol-gel showed better SCR activity. However, the poor low-temperature SCR activity and SO₂ durability of these Ce-Ti-based catalysts still limit their practical application. Much effort to improve the low-temperature SCR catalytic performance of Ce-Ti-based catalysts has been made over the last several years. Among them, the introduction of other metal oxides to enhance their SCR activity has received considerable attention. The studies have proved that CeO₂-TiO₂ incorporated with various metals, such as Mn, W, Cu, Sn and Fe, achieved better NH₃-SCR catalytic performance [19-23]. MoO₃ was employed to increase the activity of traditional V-based deNO_x catalyst [24]. Recently, Jiang et al. [25] prepared Ce-Mo-Ti oxide catalysts by a sol-gel method and found that the addition of Mo could improve the SCR activity of CeO₂-TiO₂ catalyst and its resistance to SO₂ and H₂O. Our study found that the incorporation of MoO₃ into TiO₂ could enhance the low-temperature activity and SO₂/H₂O resistance of CeO₂/TiO₂ catalyst by improving the properties of carrier [26]. Li et al. [27, 28] also revealed that the addition of MoO₃ has a promoting effect on the activity and arsenic resistance of supported CeO₂/TiO₂ catalyst. Nevertheless, few studies have focused on supported MoO₃/CeTiO_x catalysts for NH₃-SCR reaction, which were obtained only by deposition of MoO₃ on CeTiO_x support. Based on these findings, we considered that not only the supported MoO₃/CeTiO_x could be a novel promising NH₃-SCR catalyst with high low-temperature activity and good durability of H₂O/SO₂, but also surface Mo modification alone is a good strategy to study the promoting effect of Mo on CeTiO_x catalyst. Therefore, in this work, we attempted to develop a novel non-vanadium low-temperature NH₃-SCR catalyst by only loading MoO₃ on CeTiO_x support. It was found that the addition of MoO₃ has a noticeable promoting effect on the activity of CeO₂-TiO₂ for the NH₃-SCR reaction. On the basis of the characterization results, the reason for the distinct promoting effect of MoO₃ has been elucidated.

CeTiO_x mixed oxides with different Ce/Ti mole ratios (4:1, 1:1, 1:4) were prepared by an inverse co-precipitation method. In a typical synthesis, requisite amounts of ammonium cerium(Ⅳ) nitrate and titanium(Ⅳ) tetrachloride were dissolved in distilled water at room temperature and stirred for 1 h, and then the obtained slurry was added dropwise into excess ammonia (25%) under stirring until pH = 10. The milky solution was kept stirring for another 2 h, followed by aging for 24 h. The obtained precipitate was filtered and washed with deionized water until no pH change or Cl^‒ residual (detected by the solution of AgNO₃) was detected. The obtained powder was dried at 110 ℃ overnight and then calcined in flowing air at 550 ℃ for 3 h. The MoO₃/CeTiO_x catalysts were prepared by incipient-wetness impregnating the obtained CeTiO_x supports with a proper amount of ammonium molybdate ((NH₄)₆Mo₇O₂₄·4H₂O) solution, then dried at 110 ℃ and calcined at 500 ℃ for 3 h. For simplicity, the obtained MoO₃/CeTiO_x catalysts were denoted as yMo/CT, where y is the mass fraction of the loaded Mo.

Powder X-ray diffraction (XRD) was performed on a Shimadzu XRD-6000 power X-ray diffractometer with Cu K_α radiation (λ = 1.5418 ). The data were collected at 2θ = 10°–80° with 8°/min scanning rate. Nitrogen adsorption-desorption isotherms were measured on a Micrometrics ASAP-2020 adsorption analyzer via the Brunauer-Emmett-Teller (BET) method at –196 ℃. The samples (0.1 g) were activated for 4 h at 300 ℃ before each adsorption measurement. X-ray photoelectron spectroscopy (XPS) analysis was implemented on a PHI 5000 Versa Probe system at an accelerating power of 15 kW, equipped with monochromatic Al K (1486.6 eV) radiation. The adventitious C 1s peak at 284.6 eV was used as calibration to compensate the charging effect of the binding energies (BE, ± 0.1 eV error) of the samples. H₂-temperature-programmed reduction (H₂-TPR) was carried out using a H₂-Ar mixture (7 vol.% H₂) as reducing agent. Typically, 50 mg sample was used in a quartz U-tube reactor for each measurement. NH₃-temperature-programmed desorption (NH₃-TPD) experiments were performed on a multifunction chemisorption analyzer with a quartz U-tube reactor, detected by a thermal conductivity detector (TCD). Raman spectra were collected on a Spex 1877 D triplemate spectrograph with 2 cm^-1 resolution. A 532 nm DPSS diode-pump solid semiconductor laser was used as the excitation source with power output of ca. 5 mW. The in situ DRIFT spectra were collected on a Nicolet Nexus 5700 FTIR spectrometer with a scanning number of 32 at a resolution of 4 cm^‒1, and a diffuse reflectance reaction cell (HARRICK) was used. For the experiments of NH₃ or NO + O₂ adsorption-desorption, the IR cell with sample was saturated with NH₃ (1 vol.%) or NO + O₂ (500 ppm NO and 5 vol.% O₂) for 1 h at room temperature, then the sample was purged with N₂ for 30 min, and the spectra were recorded at target temperature by raising the temperature from 50 ℃ to 400 ℃.

The activity tests were carried out on a fixed-bed quartz reactor using a 0.25 g catalyst of 40–60 mesh. The feed gas mixture contained 500 ppm NO, 500 ppm NH₃, 5% O₂, 6% H₂O (when used), 200 ppm SO₂ (when used), and Ar as the balance gas. The total flow rate of the feed gas was 200 mL/min, corresponding to a GHSV of 48, 000 mL g^–1 h^–1. The reaction temperature was increased from 100 to 400 ℃. The composition of the gas in the inlet and outlet streams was measured with a quadrupole mass spectrometer (Dycor Dymaxion DM300M, AMETEK) when the catalytic reaction reached steady-state condition at each test temperature point.

First, a series of Mo/C_fT samples with different Ce/Ti mole ratios (f) were prepared to explore the influence of Ce/Ti molar ratio on NO conversion over Mo/C_fT catalysts. Herein, the loading amount of Mo was fixed at 4 wt.%. As shown in Fig. S1, the order of NH₃-SCR catalytic activity followed Mo/Ce₄TiO_x > Mo/Ce₁TiO_x > Mo/Ce_0.25TiO_x. The results of the catalytic activities indicated that Mo/CeTiO_x catalyst exhibited the optimal NO conversion when the molar ratio of Ce/Ti was 4:1. Then, a series of yMo/Ce₄TiO_x samples with different Mo loadings were prepared and the NH₃-SCR activity was tested. As shown in Fig. 1, activity test results showed that the pristine Ce₄TiO_x support had rather poor SCR activity in the whole operation temperature range (100–400 ℃) with the maximum NO conversion of 71% at 350 ℃. In contrast, the yMo/Ce₄TiO_x catalysts displayed much higher SCR activity in the temperature range of 200–400 ℃. Among which, 4Mo/Ce₄TiO_x sample exhibited a NO conversion of more than 90% within a broad operation temperature window (200–400  ℃) under the GHSV of 48000 mL g^–1 h^–1. Most notably, the addition of Mo significantly increased the activity of Ce₄TiO_x catalyst in the low-temperature range, obtaining 4-fold increase in NO conversion compared to Ce₄TiO_x at 150 ℃ (from ca. 20% to ca. 80%). These results indicated that loading Mo on Ce₄TiO_x surface had a remarkable promoting effect on NH₃-SCR activity especially at low temperature and the optimal loading amount was 4 wt.% Mo. Therefore, comparative studies between 4Mo/Ce₄TiO_x catalyst (Mo/CT for short in the following chapters) and Ce₄TiO_x support (CT for short) were conducted to elucidate the effect of MoO₃ loading on NH₃-SCR performance.

The existence of sulfur dioxide and water vapor was unavoidable in the industrial application of NH₃-SCR catalysts, so it is important to evaluate the influence of SO₂ and water on the catalytic performance of our deNO_x catalyst. Thus, the SO₂ and H₂O resistance of the Mo/CT catalyst during NH₃-SCR reaction at 250 ℃ was evaluated through a long-time test, and the corresponding results are presented in Fig. 2. When 200 ppm SO₂ was injected at 250 ℃ (Fig. 2A), the NO conversion of the Mo/CT catalyst decreased slightly at first and then was kept at about 94% during the test period. Meanwhile, the NH₃-SCR activity of the CT sample was also measured in the presence of 200 ppm SO₂ as a comparison. The results showed that the NO conversion over CT slowly decreased with an increase in time and was kept at ca. 40% after 24 h. Moreover, when 200 ppm SO₂ and 5 vol% H₂O were introduced into the reaction gas at the same time (Fig. 2B), the deactivation of the CT sample was much faster than that of Mo/CT. After 24 h, the NO conversion on the CT sample decreased from ca. 56% to ca. 29%, and the decline rate of the activity was 27%. However, the NO conversion over the Mo/CT sample still maintained at about 92%. The above results indicated that the Mo/CT catalyst exhibited above 90% NO conversion in the range of 200–400 ℃ and good H₂O/SO₂ resistance at 250 ℃ for NH₃-SCR reaction. Therefore, the Mo/CT catalyst can be used to remove NO_x from stationary pollution sources containing a certain amount of SO₂ and H₂O vapor because of its great catalytic performance.

Fig. 3A presents the XRD patterns of CT and Mo/CT catalysts. Both samples exhibited the fluorite type crystalline structure without any phase segregation, characteristics of CeO₂ (PDF-ICDD 34-0394) [29]. Furthermore, no diffraction peak attributed to MoO₃ was detected in the XRD pattern of Mo/CT catalyst, which indicated that the molybdenum oxide was highly dispersed or presented as amorphous entities on the surface of CT [30]. To get further information of Mo species, Raman characterization was operated. It is well known that in comparison with XRD, Raman is more sensitive to distinguish the dispersion state of Mo. Fig. 3B displays Raman spectra of CT and Mo/CT samples. It shows that the two Raman spectra all exhibited a main band ascribed to the triply degenerate F_2g mode of fluorite ceria and no Raman bands corresponding to MoO₃ were detected, indicating that MoO₃ was indeed present in the form of highly dispersed state on the surface of the CT support [31, 32]. Moreover, the Raman band intensity of the Mo/CT catalyst was weaker than that of the CT support, which also suggested that the highly dispersed MoO₃ species on the surface of CT covered the support and weakened the signal [33]. These Raman results are in good agreement with XRD. We also noted that the F_2g band of the Mo/CT catalyst shifted slightly from 451 to 444 cm^‒1 owing to electronic interactions between MoO₃ and CT support.

N₂ adsorption/desorption isotherms of the samples were collected to understand the textural properties of CT and Mo/CT. It can be observed from Fig. 4A that both samples exibit type-IV isotherms according to IUPAC classification, showing the presence of mesoporous structure [34]. Besides, the H3-type hysteresis loops (P/P₀ of 0.4–1.0) manifested the presence of irregular pores, which was revealed by the BJH pore distribution curves (Fig. 4B). Table 1 shows BET surface area and pore structure results of the two catalysts. The specific surface area and total pore volume of CT was 135 m²/g and 0.303 cm³/g, while that of Mo/CT was 101 m²/g and 0.273 cm³/g, respectively. The specific surface area and total pore volume of Mo/CT decreased owing to the blocking effect of the CT support by the deposited molybdenum oxide. However, the A_BET values of CT and Mo/CT followed the opposite trend of the corresponding SCR activity. Therefore, we believe that the improvement of the SCR activity over Mo/CT catalyst cannot be attributed to the changes of specific surface area caused by molybdenum addition.

XPS experiments were conducted to explore the chemical states of surface species and the types of oxygen species over CT and Mo/CT samples, as presented in Fig. 5. For the Ce 3d XPS spectra in Fig. 5A, the bands labeled μ, μ', μ'', ν, ν' and ν'' represented the 3d¹⁰4f⁰ state of Ce⁴⁺, whereas u' and v' represented the 3d¹⁰4f¹ state, corresponding to Ce³⁺ [6, 14]. It can be seen that the redox pair of Ce³⁺/Ce⁴⁺ was observed in the Mo/CT and CT catalysts. According to the area ratio of the peaks representing Ce³⁺ and Ce⁴⁺, the percentage of Ce³⁺ in the total Ce (Ce³⁺ and Ce⁴⁺) could be calculated. The addition of molybdenum to CT led to the slight increase in the percentage of Ce³⁺/(Ce³⁺ + Ce⁴⁺), which indicated there was interaction between the oxides of Mo and Ce. With regard to Mo 3d XPS spectra in Fig. 5B, two main peaks owing to Mo 3d_5/2 and Mo 3d_3/2 were observed at 232.2 and 235.6 eV, which were attributed to MoO₃ or Mo⁶⁺ oxide [35, 36]. The O 1s XPS information of these samples was presented in Fig. 5C. The peak at 532.1 eV was considered as surface oxygen species (donated as O_α), and the band of O_β at 529.2 eV was ascribed to lattice oxygen species. The O_α/(O_α + O_β) ratio of all the samples were calculated by the area integral of O_α and O_β. It was found that the relative concentration ratio of O_α over Mo/CT catalyst was lower than that of CT (Table 1), suggesting that the number of surface oxygen species decreased after the introduction of molybdenum.

The redox properties of an NH₃-SCR catalyst always play an important role in the NH₃-SCR catalytic performance [37-39]. Thus, H₂-TPR was carried out to evaluate the redox properties of CT and Mo/CT catalysts. As shown in Fig. 6, three reduction peaks were observed for the two samples, which were assigned to the reduction of the surface oxygen species of ceria (α), the reduction of surface Ce⁴⁺ (β), and the reduction of bulk of Ce⁴⁺ (γ), respectively [29, 40, 41]. In addition, the reduction of the well dispersed Mo species and surface Ti⁴⁺ was also likely to be included in peak β [28]. It is worth noting that loading Mo could slightly affect the reduction temperature for peak β. The β peak moved to higher temperature, from 570 ℃ for CT to 582 ℃ for Mo/CT, which was caused by the coverage of molybdenum oxides. Similar results were also obtained for vanadium and tungsten dispersed on CeO₂ [42]. At the same time, the area of peak α over the Mo/CT sample attributed to the reduction of surface oxygen species obviously declined because of the loading of Mo. This phenomenon illustrated that Mo loaded on CT could decrease the number of labile surface oxygen and defect sites, leading to low reducibility of the catalysts. This deduction can be supported by the results of XPS spectra shown in Fig. 5. It is clear that the O_α signal representing surface chemisorbed oxygen was much higher for CT than for Mo/CT. These results were also in line with the observations of Li et al. [35]. In brief, the H₂-TPR result manifested that the reducibility of Mo/CT was restrained compared with CT. According to the above characterization results and NH₃-SCR activity in Fig. 1, the experiments of NH₃-TPD and in situ DRIFTS were carried out to explore the reason for better catalytic activity of Mo/CT. The results may be more relevant for assessing the catalytic performance of the NH₃-SCR catalysts because they were obtained under conditions very close to or even identical with those of the catalytic reaction.

Surface acid properties of NH₃-SCR catalysts are another crucial aspect affecting the catalytic activity apart from the redox property. Accordingly, NH₃-TPD and in situ NH₃ adsorption experiments were employed to evaluate the adsorption behavior of NH₃ on the surface of CT and Mo/CT catalysts. First, the NH₃-TPD experiment was performed to probe the number of acid sites in the CT and Mo/CT catalysts, and the corresponding results are shown in Fig. 7. It can be seen that the NH₃-TPD profiles of CT and Mo/CT catalysts both exhibited four desorption peaks in the temperature range of measurements, which are labeled as Ⅰ, Ⅱ, Ⅲ, and Ⅳ. The Peak Ⅰ around 150 ℃ was assigned to the desorption of physisorbed NH₃ species. Peak Ⅱ was attributed to weak acid sites, and the peaks Ⅲ and Ⅳ were ascribed to strong acid sites [14, 43]. The quantitative analysis results of NH₃-TPD are summarized in Table 1. It shows that the total acidity of CT and Mo/CT was 0.49 and 0.68 mg/g, respectively. Evidently, the Mo/CT catalyst exhibited a remarkably increased total acid amount compared with CT, which suggested that Mo surface modification can effectively improve the surface acidity of Mo/CT catalyst to promote the adsorption of NH₃ molecules, and finally enhance the NH₃-SCR activity.

NH₃-TPD can determine the total number of acid sites, but it fails to distinguish Brönsted acid sites and Lewis acid sites on catalyst surface. To ascertain the nature of acid sites and acquire more information of the increased acidity, NH₃ adsorption in situ DRIFTS was carried out, and the corresponding results are depicted in Fig. 8. For the CT catalyst in Fig. 8A, after NH₃ adsorption and N₂ purge, several bands were detected in the range of 1000–2000 cm^‒1. The bands at 1579 and 1181 cm^‒1 were ascribed to coordinated NH₃ bound to Lewis acid sites, and the weak band at 1440 cm^‒1 could be related to the NH₄⁺ species on Brönsted acid sites [19, 28, 44]. For the Mo/CT catalyst in Fig. 8B, the in situ DRIFT spectra of NH₃ adsorption were obviously different from that of the CT catalyst. In addition to the band at 1183 cm^-1 attributed to the coordinated NH₃ linked to Lewis acid sites, a new band at 1672 cm^‒1 assigned to NH₄⁺ species bound to Brönsted acid sites was observed [18, 45, 46]. Moreover, the intensity of the band at 1433 cm^‒1 ascribed to the NH₄⁺ species bound to Br nsted acid sites was much stronger than that of the CT catalyst. With the temperature rising, all bands became weaker and the band intensity of Brönsted acid sites (1440, 1433, 1672 cm^‒1) decreased faster than that of Lewis acid sites. This phenomenon indicated that NH₃ bonded to Lewis acid sites was more stable than that bonded to Brönsted acid sites on catalyst surface. Combining with the NH₃-TPD results, the weak acid should be related to the Brönsted acid and the strong acid was related to the Lewis acid. This meant that the loading of Mo could result in more Brönsted acid sites on the Mo/CT catalyst surface, and similar results were found on CeO₂-MoO₃ catalyst in Li's work [35]. Moreover, we can find that the area ratio of Brönsted acid sites to total acid sites over Mo/CT was much higher than that on CT in the low-temperature range (< 300 ℃) (Fig. S2). And, Mo/CT displayed the higher NO conversion in the same temperature range (Fig. 1). These phenomena suggested that the larger number of Brönsted acid sites caused by Mo addition played an important role in the improvement of catalytic activity in the low-temperature range over the Mo/CT catalyst for NH₃-SCR reaction in the present work.

The in situ DRIFT experiments of NO + O₂ co-adsorption were also employed to probe the surface adsorption the NO_x species over the two catalysts. For the CT catalyst (Fig. 9A), several vibration bands attributed to adsorbed nitrate species were observed when it was exposed to NO + O₂. According to the literature, the band at 1630 cm^‒1 was ascribed to bridging bidentate nitrates, the bands at 1605 and 1217 cm^‒1 were attributed to bridging monodentate nitrates, the band located at 1580 cm^‒1 was assigned to bidentate nitrates, and the bands at 1530 and 1278 cm^‒1 were assigned to monodentate nitrates. For the Mo/CT catalyst (Fig. 9B), when NO + O₂ were introduced into the DRIFTS cell at 150 ℃, some bands attributed to adsorbed NO_x species were also observed: bridged nitrates (1628, 1601, and 1227 cm^‒1), monodentate nitrates (1524 and 1271 cm^‒1), and bidentate nitrates (1577 cm^‒1) [47-52]. The bands at 1349 cm^‒1 attributed to ionic nitrate also appeared [53-55]. Compared with the NO + O₂ spectra of the CT and Mo/CT samples, it is worth noting that the intensity of adsorbed NO_x species on Mo/CT was significantly weaker than that on CT. Moreover, with increasing temperature, the band intensities of adsorbed NO_x species became weaker gradually. Especially, the adsorbed NO_x species on the Mo/CT catalyst disappeared completely at 300 ℃. The above results suggested that the highly dispersed molybdenum on Mo/CT not only resulted in more Brönsted acid sites formed on the catalyst surface, but also reduced the thermal stability of the inactive nitrate species, leaving more active sites available for the adsorption of NH₃, both of which were favorable for the promotion of SCR activity.

To investigate the reaction mechanism of NH₃-SCR reaction over the CT and Mo/CT catalysts, in situ DRIFTS of NH₃ + NO + O₂ co-adsorption was performed, and the results are displayed in Fig. 10. When the CT sample was exposed to the NH₃ + NO + O₂ mixed gas at room temperature, bidentate nitrate (1563 cm^‒1), monodentate nitrate (1520 and 1278 cm^‒1), and bridging nitrate (1624 and 1228 cm^‒1) were observed over the catalyst. Furthermore, combining the results of NH₃ adsorption in situ DRIFTS, the band at 1444 cm^‒1 related to NH₄⁺ ions bonded on Brönsted acid sites was also detected. Along with the rise of temperature, the band of Brönsted acid sites disappeared at 200 ℃ due to the desorption of NH₃ and the reaction between the adsorbed NH₃ species and NO_x species. More importantly, some interesting phenomena about adsorbed NO_x species were observed during the heating process. The band of monodentate nitrate (1278 cm^‒1) disappeared at 200 ℃ because of its poor thermal stability, and the band intensity of bidentate nitrate (1563 cm^‒1) and bridging nitrate (1624 and 1228 cm^‒1) increased with increasing temperature first and then decreased with a further increase in temperature, which was related to the reaction between the adsorbed NO_x and NH₃ species, as well as the transformation and dissociation of the adsorbed NO_x species. However, these bidentate and bridging nitrates did not disappear even at 400 ℃ owing to the strong adsorption. With regard to the Mo/CT catalyst, the bands of bridging nitrate (1626 and 1227 cm^‒1), bidentate nitrate (1600 and 1565 cm^‒1), monodentate nitrate (1515 and 1262 cm^‒1), and Brönsted acid (1433 cm^‒1) were also detected at corresponding positions. And the variation of these bands was similar to that of the CT catalyst with an increase in temperature. More significantly, however, a new band at 1342 cm^‒1 was observed, which was attributed to the intermediate species generated by the reaction of the adsorbed NH₃ and NO_x species [56]. This may be one reason that the Mo/CT catalyst displayed better catalytic performance than the CT catalyst for NH₃-SCR reaction.

A novel Mo-promoted CeO₂-TiO₂ catalyst has been developed for the selective catalytic reduction of NO by NH₃, which showed high deNO_x efficiency from 150 to 400 ℃. The redox properties and surface acidity of the MoO₃/CeO₂-TiO₂ catalyst were affected markedly by the addition of Mo species, thereby modified the adsorption of NH₃ and NO_x on the catalyst surface. The improvement of SCR activity over MoO₃/CeO₂-TiO₂ may be directly correlated to the increase in the number of Brönsted acid sites and weakly adsorbed NO_x species, as opposed to a decrease in the BET surface area and a change in the redox properties. Our results can provide a scientific reference for the development of CeO₂-based catalysts for practical applications.