NH 3 -Selective Catalytic Reduction of NO x to N 2 over Ceria Supported WO x Based Catalysts: Inﬂuence of Tungsten Content

: A series of HPW/CeO 2 catalysts generated from 12-tungstophosphoric acid, H 3 PW 12 O 40 (HPW), supported on ceria and presenting different tungsten loadings (2, 4.5, 9, 16, and 40 wt% W) were prepared and characterized by N 2 physisorption, XRD, IR, Raman, and UV-Vis. The different characterization techniques suggested that low loading of tungsten resulted in mainly isolated sites, while high tungsten loading produced polymeric or tungsten clusters. Those materials exhibited high activity in NH 3 -SCR of NO x into N 2 . Moreover, the series of experiments indicated that low loading in tungsten (2% HPW/CeO 2 ) displayed the highest activity with a remarkable N 2 selectivity (99%) at medium-high temperature (300–515 ◦ C), owing to the high amount of monomeric tungstate coverage on the catalyst surface.


Introduction
In the last decades, growing interest has been given to developing novel technologies for reducing the harmful gases in environment. Among these technologies are denitration processes for the removal of nitrogen oxides (NO and NO 2 or NO x ). Emitted NO x is mainly originated from combustion reactions at fairly high temperatures, typically in hydrocarbon fuel combustion engines, industrial plants, and ammonia combustion power plants [1,2]. Nonetheless, the modifications made to the combustor type or in the combustion processes effectively reduce NO x emissions [3,4]. For example, in currently operated power plants it is possible to improve combustion efficiency up to 99% and reduce pollutant emissions to concentrations lower than 100 ppm just by optimization of injection angles and temperatures at the inlet of combustors [5]. The corresponding optimization of mobile combustion engines has also been conducted. However, due to pursued restriction in NO x emission for mobile transport application, such engineering optimization remains insufficient to satisfy upcoming regulations. Hence, the urgency to develop more efficient NOx abatement technologies is still essential. NOx removal includes three approaches [6]: (i) direct decomposition of NO x into O 2 and N 2 ; (ii) NO x storage reduction (NSR); and (iii) NO x selective catalytic reduction (SCR). NO x selective catalytic reduction (SCR) technology is classified according to the reducing agents used, such as hydrocarbons [7], oxygenated hydrocarbons [8,9], hydrogen [10], or ammonia (or urea) [11,12]. The latter is regarded as an efficient approach to reduce NO x emissions due to its high selectivity and relatively simple implementation (without compromising the engine efficiency).
Commercial NH 3 -SCR applications typically use V 2 O 5 -WO 3 /TiO 2 as a catalytic converter due to its high performance in a broad temperature window, low cost, and an excellent resistance to SO 2 [13]. Nevertheless, the N 2 selectivity decreases at high temperatures. Furthermore, the release of toxic V 2 O 5 at high temperatures creates another environmental issue [14]. Therefore, developing high performance and eco-friendly catalysts to perform effective SCR denitration turns out to be a key step for the success of this process.
Tungsten based catalysts have attracted more interest owing to their superior acidity, high capacity to suppress undesirable reactions (such as SO 2 oxidation) and ability to lead to a better dispersion of the deNO x active sites [15,16]. In recent investigations, the incorporation of the tungsten to metal oxides such as MnO x , CuO x , TiO x , ZrO x , and FeO x greatly increased the NH 3 -SCR activity [17]. Shibo et al. reported that the addition of a small amount of tungsten to Cu 0.02 Fe 0.2 TiO x improves the active surface acidity and increases both the specific surface area and the number of total active sites, which all promote catalytic activity [18]. Additionally, Wang et al. found that the presence of W with MnO x -TiO 2 forms the active intermediates species responsible for SCR performance [19]. NH 3 -SCR on supported catalysts showed high performance with ceria support in association with tungsten, and the redox behavior of ceria and its high capacity to store oxygen were remarkably improved during the SCR process [20]. Several research investigations indicate that the synergic effect is created between the tungsten and Ce 4+/ Ce 3+ on the ceria support surface, resulting in a better catalytic conversion [21][22][23]. In order to further improve and optimize this system, it is crucial to investigate the relationship between the structure of tungsten and the reactivity in NH 3 -SCR reactions.
12-tungstophosphoric acid H 3 PW 12 O 40 (HPW) with a Keggin structure is a cheap and readily available tungsten source and is widely used for both acid and redox-catalyzed reactions [24]. For the NH 3 -SCR, HPW has been used as precursor on ceria (CeO 2 ) to improve the NH 3 adsorption ability after calcination [25]. However, this catalyst shows a poor performance that may be due to low surface area, low dispersion of the tungsten species, and uncontrolled synthesis method. In other studies, Song et al. [26] and Geng et al. [22] described the higher performance of ceria grafted HPW which the authors claimed to be related to better NH 3 adsorption. Nevertheless, for these systems, the active species controlling the activity are not precisely determined.
In the present study, a series of HPW/CeO 2 (2, 4.5, 9, 16 and 40 wt% W) catalysts were synthesized and characterized. The intention is to evaluate the nature of tungsten supported on ceria in NH 3 -SCR efficiency. A series of different tungsten loadings allows determining the effect of the dispersion. Simultaneously, this study will also reveal the amount of tungsten required to obtain the optimal catalyst. Special attention is addressed to the structure analysis of the materials using X-ray diffraction, FTIR, and Raman spectroscopy, to the surface characterization by means of N 2 adsorption, in relationship with NH 3 -SCR performance.

Characterization of the Catalysts
The catalysts textural properties and tungsten surface density (Equation (1)) are presented in Table 1. HPW is a non-porous solid and possesses only a low specific surface area. On the other hand, when deposited on a high surface area ceria (S BET = 247 m 2 g −1 , surface area determined by the BET approach), a tungsten functionalized material with high specific surface area can be obtained. The surface area and the average pore volume decrease with increasing W loading ( Figure S1). This can be explained by a pore blocking mechanism when larger tungsten content is introduced. Thus, the surface is partially inhibited [27]. Changes in surface area and pore volume may affect the catalytic activity of the materials. The HPW shows diffraction peaks of the Keggin structure, assigned to the triclinic space group P-1 (PDF#00-050-0656). The cubic CeO 2 phase (PDF#04-013-4361) is successfully identified for all of the HPW/CeO 2 catalysts ( Figure 1). The peaks at 2θ = 28.  (400), respectively [28] (PDF#04-013-4361). In addition, no characteristic peaks related to the Keggin structure are detected even at higher tungsten contents, which indicates that the Keggin structure is either decomposed during calcination into tungsten oxides with an amorphous form or is highly dispersed on CeO 2 . Based on the Scherrer equation, the average crystallite sizes (Table 1) remain more or less in the same order.
Catalysts 2021, 11, x FOR PEER REVIEW 3 of 12 mechanism when larger tungsten content is introduced. Thus, the surface is partially inhibited [27]. Changes in surface area and pore volume may affect the catalytic activity of the materials. The HPW shows diffraction peaks of the Keggin structure, assigned to the triclinic space group P-1 (PDF#00-050-0656). The cubic CeO2 phase (PDF#04-013-4361) is successfully identified for all of the HPW/CeO2 catalysts ( Figure 1). The peaks at 2θ = 28.1°, 33.2°, 47.4°, 52.2°, 56.3°, and 69.5° are ascribed to the diffraction lines (111), (200), (220), (311), (222), and (400), respectively [28] (PDF#04-013-4361). In addition, no characteristic peaks related to the Keggin structure are detected even at higher tungsten contents, which indicates that the Keggin structure is either decomposed during calcination into tungsten oxides with an amorphous form or is highly dispersed on CeO2. Based on the Scherrer equation, the average crystallite sizes (Table 1) remain more or less in the same order. FT-IR spectra of the catalysts are illustrated in Figure 2 and Figure S2. The prepared samples display a peak at 493 cm −1 , attributed to stretching vibrations of Ce-O [29]. The rather weak IR signals observed around 947 and 682 cm −1 are assigned to νs(W = O) fragment and W-O-W moiety, respectively. For the catalysts containing 16 and 40 wt% W, a novel band appeared at 820 cm −1 assigned to asymmetric stretching vibrations of W = O [30]. Compared with the tungstophosphoric acid precursor [31], the HPW in the catalysts did not preserve its Keggin framework, also suggesting that the HPW is decomposed into tungsten oxides after calcination [32]. FT-IR spectra of the catalysts are illustrated in Figure 2 and Figure S2. The prepared samples display a peak at 493 cm −1 , attributed to stretching vibrations of Ce-O [29]. The rather weak IR signals observed around 947 and 682 cm −1 are assigned to ν s (W = O) fragment and W-O-W moiety, respectively. For the catalysts containing 16 and 40 wt% W, a novel band appeared at 820 cm −1 assigned to asymmetric stretching vibrations of W = O [30]. Compared with the tungstophosphoric acid precursor [31], the HPW in the catalysts did not preserve its Keggin framework, also suggesting that the HPW is decomposed into tungsten oxides after calcination [32]. H3PW12O40 Raman spectrum ( Figure 3a, lower spectrum) displays the Keggin skeleton characteristic peaks ascribed to symmetric-asymmetric stretching modes of W I = Od (1011 cm −1 ), W II = Od (994 cm −1 ), P-Oa (906 cm −1 ), W-Ob-W (522 cm −1 ), and W-Oc-W (237 cm −1 ) bonds [33]. For the ceria, an intense symmetrical band at 459 cm −1 is ascribed to the F2g vibration mode in the cubic fluorite phase [34]. Additionally, another band observed at 590 cm −1 can probably be instigated by Ce-O symmetry perturbation due to the oxygen vacancies presence in the CeO2 structure (figure S3) [35]. All the supported catalysts show the ceria bands with a slight shift. The IOv/IF2g values (where IF2g and IOv were the intensity of F2g (459 cm −1 ) and Ov (590 cm −1 ), respectively) reveal the oxygen vacancies contents in the catalysts, which are determined by Raman deconvolution of Ov and F2g peaks. These fairly low values are similar for all catalysts even at higher immobilized W amount ( Figure  S4, Table S1), indicating that the ceria preserves its surface morphology [36]. In order to further explore the W effect on the ceria support, Raman analysis of tungsten species are shown in 650-1000 cm −1 range. It has been reported that the WO3 bulk, monomeric and polymeric tungstate configurations appeared respectively at 810 cm −1 , 930 cm −1 and 960 cm −1 [23]. In this series of HPW/CeO2 catalyst, peaks centered around 805-815, 915-930, and 945-960 cm −1 are observed. Moreover, for high W content, the mono (915-930 cm −1 ) and poly-tungstate (945-960 cm −1 ) appear as one broad peak, which is submitted to a deconvolution operation in order to calculate the contributions of both forms ( Figure S4, Table 2). The 2 and 4.5 wt% W/CeO2 catalysts (0.3 and 0.95 W at/nm 2 ) show only monomeric tungsten species. When increasing the tungsten loading, the monomeric form would decrease and the ceria surface should gradually be covered by the poly-tungstate oligomeric structure and WO3 nanoparticles.  [33]. For the ceria, an intense symmetrical band at 459 cm −1 is ascribed to the F 2g vibration mode in the cubic fluorite phase [34]. Additionally, another band observed at 590 cm −1 can probably be instigated by Ce-O symmetry perturbation due to the oxygen vacancies presence in the CeO 2 structure ( Figure S3) [35]. All the supported catalysts show the ceria bands with a slight shift. The I Ov /I F2g values (where I F2g and I Ov were the intensity of F 2g (459 cm −1 ) and O v (590 cm −1 ), respectively) reveal the oxygen vacancies contents in the catalysts, which are determined by Raman deconvolution of O v and F 2g peaks. These fairly low values are similar for all catalysts even at higher immobilized W amount ( Figure S4, Table S1), indicating that the ceria preserves its surface morphology [36]. In order to further explore the W effect on the ceria support, Raman analysis of tungsten species are shown in 650-1000 cm −1 range. It has been reported that the WO 3 bulk, monomeric and polymeric tungstate configurations appeared respectively at 810 cm −1 , 930 cm −1 and 960 cm −1 [23]. In this series of HPW/CeO 2 catalyst, peaks centered around 805-815, 915-930, and 945-960 cm −1 are observed. Moreover, for high W content, the mono (915-930 cm −1 ) and poly-tungstate (945-960 cm −1 ) appear as one broad peak, which is submitted to a deconvolution operation in order to calculate the contributions of both forms ( Figure S4, Table 2). The 2 and 4.5 wt% W/CeO 2 catalysts (0.3 and 0.95 W at/nm 2 ) show only monomeric tungsten species. When increasing the tungsten loading, the monomeric form would decrease and the ceria surface should gradually be covered by the poly-tungstate oligomeric structure and WO 3 nanoparticles.  The evolution of the bandgap as a function of tungsten loading is determined by UVvisible diffuse reflectance spectroscopy, using the O 2-to metal charge transfer band ( Figure  4). The HPW polytungstate cluster gave an edge energy value (Eg = 3.03 eV) between the isolated tungsten (Na2WO4: Eg = 4.89 eV) and WOx crystals (Eg = 2.3 eV) [37]. The W-O-W bonds number and the electronic delocalization through the WOx polyhedral neighbors can alter the bandgaps. Nevertheless, the nature of the support oxides also affects the Eg values. CeO2 itself has a bandgap at 2.7 eV and will inevitably prevent a reliable determination of the bandgap of the catalysts. Hence, it is difficult to probe the tungsten forms on the surface based on UV-Vis. Nevertheless, this investigation clearly shows that higher loading of tungsten gradually decreases the edge energy (Figure 4b), indicating the presence of more oligomeric/polymeric tungsten species.
The characterization studies carried out for the different catalysts have shown that for low W loadings (2.5 and 4 wt % W), tungsten is mainly present on the ceria surface as monomeric species ( Figure 5) while increasing the W loading also leads to oligomeric supported species and some WO3 bulk. The estimated proportion of monomeric W species decreasing from 100% (2 wt% W) to 96 (4.5 wt% W), 64 (9 wt% W), 44 (16 wt% W), and 36% (40 wt % W).  The evolution of the bandgap as a function of tungsten loading is determined by UV-visible diffuse reflectance spectroscopy, using the O 2− to metal charge transfer band ( Figure 4). The HPW polytungstate cluster gave an edge energy value (Eg = 3.03 eV) between the isolated tungsten (Na 2 WO 4 : Eg = 4.89 eV) and WO x crystals (Eg = 2.3 eV) [37]. The W-O-W bonds number and the electronic delocalization through the WO x polyhedral neighbors can alter the bandgaps. Nevertheless, the nature of the support oxides also affects the Eg values. CeO 2 itself has a bandgap at 2.7 eV and will inevitably prevent a reliable determination of the bandgap of the catalysts. Hence, it is difficult to probe the tungsten forms on the surface based on UV-Vis. Nevertheless, this investigation clearly shows that higher loading of tungsten gradually decreases the edge energy (Figure 4b), indicating the presence of more oligomeric/polymeric tungsten species.

Catalytic Activity
The NO conversion and N2 selectivity of HPW/CeO2 with various W contents and 2% HPW/SiO2 (used for comparison) are displayed in Figures 6 and S6. The NOx reduction performance of the catalysts decreased in the order 2% HPW/CeO2 = 4.5% HPW/CeO2 > 9% HPW/CeO2 > 16% HPW/CeO2 > 40% HPW/CeO2 > 2% HPW/SiO2 > CeO2 > HPW. Pure HPW showed almost no activity under the experimental conditions of the reaction due to its high non selective NH3 oxidation. Modest NH3-SCR catalytic activity is observed on the pure ceria support, with a NO conversion less than 30% at 400 °C. It is observed that when the HPW is loaded on the ceria, the activity is significantly improved (NO conversion up to 98%). This is due probably to the synergistic effect created between tungsten species and CeO2. The Ce 4+ ions on CeO2 had a good NOx oxidation ability but a poor NH3 adsorption [22]. The 2-4.5% HPW/CeO2 show an outstanding de-NOx activity, reaching a maximum NO conversion of 98% across the temperature range 300-515 °C. Meanwhile, the catalysts maintained a higher N2 selectivity, particularly above 515 °C. Such performance is comparable with existing conventional catalysts [38]. When the tungsten loading was increased to 9%, the catalyst became less efficient with a NOx removal of 98% at the 296-420 °C interval. The operation temperature interval is also reduced with the insertion of higher amounts of tungsten into the ceria. The 16% HPW/CeO2 and 40% HPW/CeO2 catalysts showed NOX conversion of 98% at 286-416 °C and 274-405 °C, respectively. based on the obtained results, it is worthwhile to note that the NO efficiency is gradually reduced with increasing tungsten loadings to 9, 16, and 40 wt%. Interestingly, the SCR activity is closely dependent on the tungsten loading. It is known for these catalysts types that the modification of ceria with a high amount of tungsten (from HPW acid) enhances the acidity and promotes the NH3 adsorption at low temperature [22]. This can explain   The characterization studies carried out for the different catalysts have shown that for low W loadings (2.5 and 4 wt% W), tungsten is mainly present on the ceria surface as monomeric species ( Figure 5) while increasing the W loading also leads to oligomeric supported species and some WO 3 bulk. The estimated proportion of monomeric W species decreasing from 100% (2 wt% W) to 96 (4.5 wt% W), 64 (9 wt% W), 44 (16 wt% W), and 36% (40 wt% W).

Catalytic Activity
The NO conversion and N2 selectivity of HPW/CeO2 with various W contents and 2% HPW/SiO2 (used for comparison) are displayed in Figures 6 and S6. The NOx reduction performance of the catalysts decreased in the order 2% HPW/CeO2 = 4.5% HPW/CeO2 > 9% HPW/CeO2 > 16% HPW/CeO2 > 40% HPW/CeO2 > 2% HPW/SiO2 > CeO2 > HPW. Pure HPW showed almost no activity under the experimental conditions of the reaction due to its high non selective NH3 oxidation. Modest NH3-SCR catalytic activity is observed on the pure ceria support, with a NO conversion less than 30% at 400 °C. It is observed that when the HPW is loaded on the ceria, the activity is significantly improved (NO conversion up to 98%). This is due probably to the synergistic effect created between tungsten species and CeO2. The Ce 4+ ions on CeO2 had a good NOx oxidation ability but a poor NH3 adsorption [22]. The 2-4.5% HPW/CeO2 show an outstanding de-NOx activity, reaching a maximum NO conversion of 98% across the temperature range 300-515 °C. Meanwhile, the catalysts maintained a higher N2 selectivity, particularly above 515 °C. Such performance is comparable with existing conventional catalysts [38]. When the tungsten loading was increased to 9%, the catalyst became less efficient with a NOx removal of 98% at the 296-420 °C interval. The operation temperature interval is also reduced with the insertion of higher amounts of tungsten into the ceria. The 16% HPW/CeO2 and 40% HPW/CeO2 catalysts showed NOX conversion of 98% at 286-416 °C and 274-405 °C, respectively. based on the obtained results, it is worthwhile to note that the NO efficiency is gradually reduced with increasing tungsten loadings to 9, 16, and 40 wt%. Interestingly, the SCR

Catalytic Activity
The NO conversion and N 2 selectivity of HPW/CeO 2 with various W contents and 2% HPW/SiO 2 (used for comparison) are displayed in Figure 6 and Figure S6. The NO x reduction performance of the catalysts decreased in the order 2% HPW/CeO 2 = 4.5% HPW/CeO 2 > 9% HPW/CeO 2 > 16% HPW/CeO 2 > 40% HPW/CeO 2 > 2% HPW/SiO 2 > CeO 2 > HPW. Pure HPW showed almost no activity under the experimental conditions of the reaction due to its high non selective NH 3 oxidation. Modest NH 3 -SCR catalytic activity is observed on the pure ceria support, with a NO conversion less than 30% at 400 • C. It is observed that when the HPW is loaded on the ceria, the activity is significantly improved (NO conversion up to 98%). This is due probably to the synergistic effect created between tungsten species and CeO 2 . The Ce 4+ ions on CeO 2 had a good NO x oxidation ability but a poor NH 3 adsorption [22]. The 2-4.5% HPW/CeO 2 show an outstanding de-NO x activity, reaching a maximum NO conversion of 98% across the temperature range 300-515 • C. Meanwhile, the catalysts maintained a higher N 2 selectivity, particularly above 515 • C. Such performance is comparable with existing conventional catalysts [38]. When the tungsten loading was increased to 9%, the catalyst became less efficient with a NO x removal of 98% at the 296-420 • C interval. The operation temperature interval is also reduced with the insertion of higher amounts of tungsten into the ceria. The 16% HPW/CeO 2 and 40% HPW/CeO 2 catalysts showed NO X conversion of 98% at 286-416 • C and 274-405 • C, respectively. based on the obtained results, it is worthwhile to note that the NO efficiency is gradually reduced with increasing tungsten loadings to 9, 16, and 40 wt%. Interestingly, the SCR activity is closely dependent on the tungsten loading. It is known for these catalysts types that the modification of ceria with a high amount of tungsten (from HPW acid) enhances the acidity and promotes the NH 3 adsorption at low temperature [22]. This can explain the higher activity observed for highly loaded tungsten catalysts at low temperature, as shown in Figure 6. However, the opposite trend is observed in the high temperature region. It was found that at higher degree of tungsten loadings, an intense agglomeration of particles was produced, which would cover the active sites. Furthermore, the presence of WO 3 nanoparticles on the ceria surface may affect the nature of these sites. The catalytic performance decreased then from 9% tungsten loading.
Catalysts 2021, 11, x FOR PEER REVIEW 7 of 12 the higher activity observed for highly loaded tungsten catalysts at low temperature, as shown in Figure 6. However, the opposite trend is observed in the high temperature region. It was found that at higher degree of tungsten loadings, an intense agglomeration of particles was produced, which would cover the active sites. Furthermore, the presence of WO3 nanoparticles on the ceria surface may affect the nature of these sites. The catalytic performance decreased then from 9% tungsten loading. The performance of 2% HPW/CeO2 catalyst is greatly superior to that of 2% HPW/SiO2 (80% NO conversion is achieved at 420 °C), indicating the favorable synergistic effect created between tungsten and the ceria. In the 2% HPW/CeO2, the interaction between Ce atom and isolated tungsten (monomeric) (revealed by Raman) produces W(=O)2(O-Ce)2 and W(=O)(O-Ce)4 species [23,39]. This may explain the best activity and the highest stability at high temperature obtained for this 2% HPW/CeO2 catalyst, where the highest proportion of isolated species is observed. Isolated tungsten species on ceria surface, prepared by an organometallic approach, has recently been demonstrated to be extremely active for NH3-SCR [40].  The turnover frequencies TOF (calculated based on the converted NO over the total amount of tungsten) were determined to compare the catalytic activity as a function of the tungsten loadings at low (219-253 °C) and high temperatures (525-560 °C). As suggested above, materials with high tungsten loading contain mainly polymeric tungsten species, while samples with low tungsten content comprise mostly isolated species. Hence, a detailed TOF analysis (Figure 7) may provide insights to associate the nature of the active sites with the activity. The TOF is higher at the low temperature region for highly loaded tungsten on ceria, demonstrating that 40% HPW/CeO2 is more active catalyst. Conversely, in the high temperature region, an increase in the TOF values is observed with lower tungsten contents. These findings indicate that the polymeric tungsten species can readily perform SCR of NO in the low temperature region, while the isolated tungsten species are more active in the high temperature region. Indeed, the isolate tungstate sites in 2% HPW/CeO2 would be more accessible to reactants. Moreover, this material shows an acidic behavior, as determined by NH3-TPD (33.4 µmol/g), a parameter known to improve the catalytic performances even with low W concentration ( Figure S6). For this reason, the use of only a small amount of tungsten is enough to achieve the best activity for NO reduction. The performance of 2% HPW/CeO 2 catalyst is greatly superior to that of 2% HPW/ SiO 2 (80% NO conversion is achieved at 420 • C), indicating the favorable synergistic effect created between tungsten and the ceria. In the 2% HPW/CeO 2 , the interaction between Ce atom and isolated tungsten (monomeric) (revealed by Raman) produces W(=O) 2 (O-Ce) 2 and W(=O)(O-Ce) 4 species [23,39]. This may explain the best activity and the highest stability at high temperature obtained for this 2% HPW/CeO 2 catalyst, where the highest proportion of isolated species is observed. Isolated tungsten species on ceria surface, prepared by an organometallic approach, has recently been demonstrated to be extremely active for NH 3 -SCR [40].
The turnover frequencies TOF (calculated based on the converted NO over the total amount of tungsten) were determined to compare the catalytic activity as a function of the tungsten loadings at low (219-253 • C) and high temperatures (525-560 • C). As suggested above, materials with high tungsten loading contain mainly polymeric tungsten species, while samples with low tungsten content comprise mostly isolated species. Hence, a detailed TOF analysis (Figure 7) may provide insights to associate the nature of the active sites with the activity. The TOF is higher at the low temperature region for highly loaded tungsten on ceria, demonstrating that 40% HPW/CeO 2 is more active catalyst. Conversely, in the high temperature region, an increase in the TOF values is observed with lower tungsten contents. These findings indicate that the polymeric tungsten species can readily perform SCR of NO in the low temperature region, while the isolated tungsten species are more active in the high temperature region. Indeed, the isolate tungstate sites in 2% HPW/CeO 2 would be more accessible to reactants. Moreover, this material shows an acidic behavior, as determined by NH 3 -TPD (33.4 µmol/g), a parameter known to improve the catalytic performances even with low W concentration ( Figure S6). For this reason, the use of only a small amount of tungsten is enough to achieve the best activity for NO reduction. Additional catalysis experiment has been investigated for 2% HPW/CeO2 catalyst in the presence of 5% water ( Figure S8), reflecting more realistic conditions in combustion engines. The conversion slightly decreased in the presence of water, which is typical. This could be attributed to agglomeration of the tungsten species accelerated by the presence of water. However, the activity and selectivity remained high.

Catalysts Preparation
A series of HPW/CeO2 catalysts with increasing tungsten amounts (2, 4.5, 9, 16, and 40 wt %) were synthetized by wet impregnation method. The solution of tungstophosphoric acid (HPW) in ethanol was added to the ceria (CeO2) at room temperature. The solvent was removed by a rotary evaporator at 30 °C. Afterwards, the samples were calcined in air at 500 °C for 3 h, yielding the catalysts for further characterization and catalytic evaluation.

Catalysts Characterization
The surface areas were determined with an ASAP 2020 Micromeritics instrument (Norcross, GA, USA) and calculated from the Brunauer, Emmett, and Teller (BET) equation. The catalysts were outgassed at 200 °C for 4 h before the physisorption measurements. The tungsten surface density was calculated based on the following Equation (1): where Xw is the loading of the W in the catalyst, NA is the Avogadro number; MW is the atomic weight of W and SBET is the surface area of each catalyst. X-ray Diffraction (XRD) patterns were registered by Bruker D8 Advance diffractometer (Billerica, MA, USA) using Cu-Kα1 radiation at 40 KV and 40 mA. The refractory lights were collected by a Bruker Lynxeye XE detector for one hour. The diffractograms were analyzed by the X'pert high score software.
IR spectra were recorded in transmission mode in a JASCO FT/IR-4100 spectrometer (Tokyo, Japan) with 2 cm −1 resolution.
Raman spectroscopy was executed on Labram HR Evo spectrophotometer (from Horiba Jobin-Yvon, Kyoto, Japan) with a green laser diode (λ = 533 nm) using a CDD detector. The spectra were analyzed by LabSpec software with a spectral resolution < 0.5 cm −1 . Additional catalysis experiment has been investigated for 2% HPW/CeO 2 catalyst in the presence of 5% water ( Figure S8), reflecting more realistic conditions in combustion engines. The conversion slightly decreased in the presence of water, which is typical. This could be attributed to agglomeration of the tungsten species accelerated by the presence of water. However, the activity and selectivity remained high.

Catalysts Preparation
A series of HPW/CeO 2 catalysts with increasing tungsten amounts (2, 4.5, 9, 16, and 40 wt%) were synthetized by wet impregnation method. The solution of tungstophosphoric acid (HPW) in ethanol was added to the ceria (CeO 2 ) at room temperature. The solvent was removed by a rotary evaporator at 30 • C. Afterwards, the samples were calcined in air at 500 • C for 3 h, yielding the catalysts for further characterization and catalytic evaluation.

Catalysts Characterization
The surface areas were determined with an ASAP 2020 Micromeritics instrument (Norcross, GA, USA) and calculated from the Brunauer, Emmett, and Teller (BET) equation. The catalysts were outgassed at 200 • C for 4 h before the physisorption measurements. The tungsten surface density was calculated based on the following Equation (1): where X w is the loading of the W in the catalyst, N A is the Avogadro number; M W is the atomic weight of W and S BET is the surface area of each catalyst. X-ray Diffraction (XRD) patterns were registered by Bruker D8 Advance diffractometer (Billerica, MA, USA) using Cu-K α1 radiation at 40 KV and 40 mA. The refractory lights were collected by a Bruker Lynxeye XE detector for one hour. The diffractograms were analyzed by the X'pert high score software.
IR spectra were recorded in transmission mode in a JASCO FT/IR-4100 spectrometer (Tokyo, Japan) with 2 cm −1 resolution.
Raman spectroscopy was executed on Labram HR Evo spectrophotometer (from Horiba Jobin-Yvon, Kyoto, Japan) with a green laser diode (λ = 533 nm) using a CDD detector. The spectra were analyzed by LabSpec software with a spectral resolution < 0.5 cm −1 .
UV-visible spectroscopy was carried out on a PerkinElmer Lambda 900 (Waltham, MA, USA). The samples were measured in 190-850 nm range at room temperature using BaSO 4 as a reference. Tauc plot was used for determine the band gaps (Eg) by the Kubelka-Munk function in indirect transition, which allows the analysis of crystalline and amorphous compounds. The band gaps (Eg) are estimated from the plot of the tangent line of the (αhν) 1/2 = f (E) curve extrapolated to the abscissa.
Determination of the acidity of the 2%HPW/CeO 2 catalyst by NH 3 desorption: A typical mass of 50 mg of material was used during the experiments. The experiment included four steps: (1) degasification of the sample in Ar at 500 • C for 2 h; (2) adsorption of NH 3 at 120 • C for 1 h; (3) isothermal desorption with Ar at 120 • C until almost no NH 3 was detected; and (4) temperature programmed desorption in Ar at 10 • C/min up to 650 • C. The amount of NH 3 in the gas phase was obtained by using an Antaris IGS FTIR (Thermo-Fischer).

Catalysts Measurement
The NH 3 -SCR tests were carried out in a quartz continuous-flow reactor using 30 mg catalyst diluted in silicon carbide (SiC). The reactant mixture is composed by 350 ppm NH 3 , 10 vol. % O 2 , 300 ppm NO in helium with a 300 mL/min total gas flow rate at a velocity of 30,000 h −1 (GHSV). The reactor was heated from room temperature to 600 • C (heating rate of 10 • C min −1 ). The outlet gas concentrations (NH 3 , NO, N 2 O and NO 2 ) were analyzed by an Antaris IGS (Thermos-Fischer, Waltham, MA, USA) FT-IR equipped with a 200 mL gas cell and a DTGS detector. The NO conversion and N 2 selectivity were calculated as follows (Equations (2) and (3)):

Conclusions
HPW/CeO 2 catalysts are prepared with different tungsten loadings by wet impregnation method and characterized by N 2 physisorption, XRD, FT-IR, Raman and UV-Vis spectroscopies. The different characterization techniques suggested that low loading of tungsten resulted in mainly isolated sites, while high tungsten loading produced polymeric or tungsten clusters. Those materials exhibited high activity in NH 3 -SCR of NO x into N 2 . The materials were further studied for selective reduction reaction of NO x by ammonia. The series of catalysts with different loading allowed further structure-reactivity assignments. Polymeric tungsten species can readily perform SCR of NO in the low temperature region, while the isolated tungsten species are more active in the high temperature region. In terms of catalytic activity, at temperatures higher than 500 • C, the best NO conversion is obtained for the 2 wt% W HPW/CeO 2 catalyst and decreased when increasing the tungsten amount. The enhanced catalytic performance can be attributed to a synergetic interaction between the isolated W and CeO 2 , described by monomeric surface moiety expressed as W(=O)(O-Ce) 4 and W(=O) 2 (O-Ce) 2 on the interface of the 2 wt% W HPW/CeO 2 . Such species are believed to improve its stability at high temperature and facilitate the adsorption of reactants. Particularly with this catalyst, a high activity (NO conversion above 98%) is reached at temperature between 300-515 • C, which can be compared with the best catalysts described in the literature with high tungsten loadings [41]. Current findings reveal that the tungsten dispersion will highly affect the NH 3 -SCR of NO, in which high dispersion is favorable. The latter has recently been confirmed by a catalytic material based on isolated tungsten species supported on ceria prepared through a delicate organometallic approach. Nevertheless, current work presents a simple synthesis method to access highly dispersed and active catalyst with low loading of tungsten. These result shows that isolated atom materials are highly effective heterogeneous catalysts for future automotive applications [42].
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/catal11080950/s1, Figure S1: N 2 physisorption isotherms for each catalyst; Figure S2: IR spectra of ceria and HPW; Figure S3: Enlarged picture of the bands attributed to the LO vibration mode of CeO 2 at 500-700 cm −1 (Raman spectra); Figure S4: deconvolutions of Raman spectra for each catalyst in the range 300-650 cm −1 ; Figure S5: Deconvolutions of Raman spectra for each catalyst in the range 800-1050 cm −1 ; Figure S6: NO x conversion, over HPW as a function of temperature. Feed composition: 300 ppm NO, 350 ppm NH 3 , 5 vol. % H 2 O and 10 vol. % O 2 in He; Figure S7: Determination of the acidity of the catalysts 2%HPW/CeO 2 by NH 3 desorption; Figure S8: NO x conversion and N 2 selectivity obtained with 2%HPW/CeO 2 as a function of temperature with and without water.