Selective Catalytic Reduction of Nitric Oxide with Propylene over Fe / Beta Catalysts Under Lean-Burn Conditions

Fe/Beta catalysts were used for the selective catalytic reduction of nitric oxide with propylene (C3H6-SCR) under lean-burn conditions, which were prepared by liquid ion-exchange (LIE), solid-state ion-exchange (SIE), and incipient wet-impregnation (IWI) methods. The iron species on Fe/Beta were characterized and identified by a combination of several characterization techniques. The results showed preparation methods had a significant influence on the composition and distribution of iron species, LIE method inclined to produce more isolated Fe3+ ions at ion-exchanged sites than IWI and SIE method. C3H6-SCR activity tests demonstrated Fe/Beta(LIE) possessed remarkable catalytic activity and N2 selectivity at temperature 300–450 ◦C. Kinetic studies of C3H6-SCR reaction suggested that isolated Fe3+ species were more active for NO reduction, whereas Fe2O3 nanoparticles enhanced the hydrocarbon combustion in excess of oxygen. According to the results of in situ DRIFTS, more isolated Fe3+ sites on Fe/Beta(LIE) would promote the formation of the key intermediates, i.e., NO2 adspecies and formate species, then led to the superior C3H6-SCR activity. The slight decrease of SCR activity after hydrothermal aging of Fe/Beta(LIE) catalyst might be due to the migration of isolated Fe3+ ions into oligomeric clusters and/or Fe2O3 nanoparticles.


Introduction
Nitric oxides (NO x ) from combustion exhaust gas are the major air pollutants drive to the formation of photochemical smog and haze [1,2].Up to now, the removal of NO x in the exhaust of diesel engine under lean-burn conditions is still a significant challenge.Selective catalytic reduction of NO x by hydrocarbons (HC-SCR) is a promising technology for reducing NO x , which can achieve the elimination of both hydrocarbons and nitric oxides in pollution source at the same time [3].Transition metal (e.g., Cu, Fe) modified zeolites exhibit remarkable catalytic activity in the HC-SCR process, many studies have been conducted on zeolite structure, metals nature and preparation methods of zeolite catalysts [4][5][6].However, the poor hydrothermal stability still limits their practical application [7,8].
Recently, zeolites with Chabazite (CHA) structure, such as SSZ-13 [9,10] and SAPO-34 [11,12], have received much attention due to their outstanding hydrothermal durability and activity in the NH 3 -SCR.However, for the HC-SCR reaction, the pore radius of CHA structure (3.8 Å) may be a little smaller for hydrocarbon molecules (e.g., C 3 H 8 , C 3 H 6 ) through the zeolite channels.Then our work will focus on the metal nature and preparation method of large or medium pore zeolites catalysts.Fe-based zeolites (e.g., Fe/ZSM-5 and Fe/Beta) are more attractive for HC-SCR because of their appropriate activity at high temperature and better hydrothermal stability [13].Chen et al. [4] first prepared Fe/ZSM-5 catalysts using the sublimation method for SCR of NO by iso-butane, which exhibited durable activity and high selectivity in the presence of water vapor.Janas et al. [5] reported that FeSiBEA catalyst with low iron content prepared in acidic conditions are active for NO reduction with C 2 reducing agents.Pan et al. [6] compared the support effects on NO x adsorption and reduction by propylene over a series of Fe/zeolites catalysts, and found that catalytic activity decreases following the order: Fe/Beta > Fe/FER > Fe/ZSM-5 > Fe/MOR at high temperatures (>300 • C).Moreover, Fe/Beta catalysts are more stable against hydrothermal aging than Fe/ZSM-5 due to the difference in the zeolite structures [14].
In general, zeolite-based catalysts are prepared using liquid-phase or solid-state ion-exchange, wet impregnation, and chemical vapor deposition, by which Fe species are introduced into parent zeolite.Fe/zeolite with controllable iron loading can be obtained using wet impregnation, but the zeolite pores are easily blocked by the agglomeration of the Fe species.Fe/zeolite catalyst prepared by the ion-exchange usually exhibits higher activity, however, the activity is also dependent on the exchange level.Furthermore, the complex coexistence of extra-framework Fe species are usually observed in Fe/zeolite catalysts prepared by different methods [15][16][17].These Fe species may be present as: (i) isolated and/or binuclear ferric ions anchored to the zeolite framework; (ii) oligomeric clusters formed inside and/or outside the pores; (iii) Fe 2 O 3 nanoparticles on the external surface of the zeolite.For NH 3 -SCR, previous studies [17,18] had shown isolated iron ions and/or oligomeric Fe x O y clusters may be the favorable active sites in Fe/ZSM-5 catalysts.Zhang et al. [19] studied the selective oxidation of methane over FeO x /SBA-15, and proposed isolated ferric ions as the active sites for the selective oxidation of CH 4 to HCHO, whereas Fe x O y clusters are less active and Fe 2 O 3 nanoparticles mainly contribute the complete oxidation of CH 4 .To best of our knowledge, there has been no systematic analysis and comparison of the correlation of the Fe species and HC-SCR performance over Fe/Beta catalysts.
In this work, Fe/Beta catalysts were prepared by different methods used for SCR of NO by C 3 H 6 under lean-burn conditions.To identify and quantify active Fe species in C 3 H 6 -SCR of iron species, a combination of characterization techniques was applied, such as X-ray diffraction (XRD), high-resolution transmission electron microscope (HRTEM), X-ray photoelectron spectra (XPS) temperature-programmed reduction with hydrogen (H 2 -TPR), UV-vis diffuse reflectance spectra (DRS), and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS).Especially, the evolution of iron species after the hydrothermal aging and their roles in C 3 H 6 -SCR were also discussed.

Physical Properties
The N 2 adsorption-desorption isotherms of Fe/Beta catalysts prepared by different methods alone with H-Beta support are shown in Figure 1.The isotherms of Fe/Beta samples were similar to H-Beta suggested that the incorporation of Fe species with low content exerted little influence on the textural properties of parent material.In addition, the BET results and Fe loading of samples are summarized in Table 1.The ICP-AES analysis showed that Fe/Beta catalysts prepared by three methods had almost the same Fe loading (about 1.5 wt.%).Compared to parent H-Beta, Fe/Beta sample prepared by LIE method had a slight decrease in the specific surface area (from 519.9 to 512.1 m 2 /g).However, the Fe/Beta prepared by IWI and SIE methods showed the obvious decrease of specific surface area (from 519.9 to 480.5 and 486.5 m 2 /g, respectively).
In general, Fe/zeolite prepared by wet impregnation could easily cause the aggregation of iron species and partially blocked some zeolite channels [20].Therefore, it could be assumed that the ferric ions at ion-exchange sites, which might be the main product with LIE method, would little affect the textural properties of Beta zeolite.On the contrary, other iron species (e.g., iron oxide nanoparticles)  In general, Fe/zeolite prepared by wet impregnation could easily cause the aggregation of iron species and partially blocked some zeolite channels [20].Therefore, it could be assumed that the ferric ions at ion-exchange sites, which might be the main product with LIE method, would little affect the textural properties of Beta zeolite.On the contrary, other iron species (e.g.iron oxide nanoparticles) might be the by-products with IWI and SIE methods, resulting in the decrease of surface area and pore volume of Fe/Beta catalysts.

Structural Characteristics
To further investigate the influence of preparation method on iron species of Fe/Beta zeolite, the XRD patterns of different Fe/Beta catalysts alone with H-Beta support were tested and depicted in Figure 2. It shows that Beta zeolite diffraction peaks (2θ = 7.7°, 21.6°, 22.5°, 25.5°, 29.7°, 33.6°, and 43.9°) mainly appeared on all Fe/Beta catalysts in Figure 2(a), indicated that the introduction of iron species did not destroy the parent zeolite structure.The results were consistent with the N2 adsorption/desorption results.In Figure 2(b), the 2θ region between 31 and 36° was enlarged to observe the characteristic peaks of the Fe2O3 phase (JCPDS file No. 79-1741).Although three Fe/Beta samples had almost the same Fe loading (about 1.5 wt.%), the diffraction peak at 33.2° assigned to Fe2O3 only appeared on Fe/Beta(SIE) sample.Fe/Beta(LIE) and Fe/Beta(IWI) samples did not show diffraction peaks for the bulk Fe2O3, suggesting that iron species were dispersed well or mainly

Structural Characteristics
To further investigate the influence of preparation method on iron species of Fe/Beta zeolite, the XRD patterns of different Fe/Beta catalysts alone with H-Beta support were tested and depicted in Figure 2. It shows that Beta zeolite diffraction peaks (2θ = 7.7 • , 21.6 • , 22.5 • , 25.5 • , 29.7 • , 33.6 • , and 43.9 • ) mainly appeared on all Fe/Beta catalysts in Figure 2a, indicated that the introduction of iron species did not destroy the parent zeolite structure.The results were consistent with the N 2 adsorption/desorption results.In Figure 2b, the 2θ region between 31 and 36 • was enlarged to observe the characteristic peaks of the Fe 2 O 3 phase (JCPDS file No. 79-1741).Although three Fe/Beta samples had almost the same Fe loading (about 1.5 wt.%), the diffraction peak at 33.2 • assigned to Fe 2 O 3 only appeared on Fe/Beta(SIE) sample.Fe/Beta(LIE) and Fe/Beta(IWI) samples did not show diffraction peaks for the bulk Fe 2 O 3 , suggesting that iron species were dispersed well or mainly migrated to the ion-exchange sites on these samples.It should be noted that the characteristic peaks at about 33.4 • appeared on H-Beta and Fe/Beta samples, which belong to the Beta zeolite diffraction peaks.The characteristic diffraction peak assigned to Fe 2 O 3 phase only appeared on Fe/Beta(SIE) sample, which illustrated the Fe 2 O 3 nanoparticles were not the dominant iron species, especially when preparing Fe/Beta catalyst with low iron content.These observations imply that the preparation method would lead to some differences of iron species on the Fe/Beta catalysts.migrated to the ion-exchange sites on these samples.It should be noted that the characteristic peaks at about 33.4° appeared on H-Beta and Fe/Beta samples, which belong to the Beta zeolite diffraction peaks.The characteristic diffraction peak assigned to Fe2O3 phase only appeared on Fe/Beta(SIE) sample, which illustrated the Fe2O3 nanoparticles were not the dominant iron species, especially when preparing Fe/Beta catalyst with low iron content.These observations imply that the preparation method would lead to some differences of iron species on the Fe/Beta catalysts.

Morphologies
Through the characterization of N2 adsorption/desorption and XRD, it is not clear whether iron species was incorporated into the framework or it was just dispersed well over the zeolite surface.This needs to be confirmed by means of certain characterization methods.Then the morphologies of Fe/Beta samples were observed by TEM, as shown in Figure 3.It is seen that iron-containing crystalline species (e.g.bulk Fe2O3) could not be observed on the surface of Beta support in all samples.In fact, three Fe/Beta samples appeared to be more similar to pure Beta zeolite (not shown).Although the TEM result could not be the direct evidence of the types of iron species, it strongly supported that there was no bulk Fe2O3 species existed in Fe/Beta in this case.

Morphologies
Through the characterization of N 2 adsorption/desorption and XRD, it is not clear whether iron species was incorporated into the framework or it was just dispersed well over the zeolite surface.This needs to be confirmed by means of certain characterization methods.Then the morphologies of Fe/Beta samples were observed by TEM, as shown in Figure 3.It is seen that iron-containing crystalline species (e.g., bulk Fe 2 O 3 ) could not be observed on the surface of Beta support in all samples.In fact, three Fe/Beta samples appeared to be more similar to pure Beta zeolite (not shown).Although the TEM result could not be the direct evidence of the types of iron species, it strongly supported that there was no bulk Fe 2 O 3 species existed in Fe/Beta in this case.
at about 33.4° appeared on H-Beta and Fe/Beta samples, which belong to the Beta zeolite diffraction peaks.The characteristic diffraction peak assigned to Fe2O3 phase only appeared on Fe/Beta(SIE) sample, which illustrated the Fe2O3 nanoparticles were not the dominant iron species, especially when preparing Fe/Beta catalyst with low iron content.These observations imply that the preparation method would lead to some differences of iron species on the Fe/Beta catalysts.

Morphologies
Through the characterization of N2 adsorption/desorption and XRD, it is not clear whether iron species was incorporated into the framework or it was just dispersed well over the zeolite surface.This needs to be confirmed by means of certain characterization methods.Then the morphologies of Fe/Beta samples were observed by TEM, as shown in Figure 3.It is seen that iron-containing crystalline species (e.g.bulk Fe2O3) could not be observed on the surface of Beta support in all samples.In fact, three Fe/Beta samples appeared to be more similar to pure Beta zeolite (not shown).Although the TEM result could not be the direct evidence of the types of iron species, it strongly supported that there was no bulk Fe2O3 species existed in Fe/Beta in this case.

XPS
XPS analysis was carried out to identify the chemical state of surface iron species.The Fe 2p spectra of Fe/Beta catalysts are shown in Figure S1.The low resolution of Fe 2p spectra is probably due to the low iron content on the surface of Fe/Beta catalysts.It is seen that all Fe/Beta catalysts exhibited two peaks of Fe 2P spectra, the first at 711.3 eV for the Fe 2p3/2 and the second at 725.3 eV for the Fe 2p1/2, with a satellite peak located at 718.9 eV.The binding energies of these peaks can be ascribed to Fe 3+ ions in the literatures [21,22].The results imply that iron species on all samples surface primarily existed in the +3 valence state.Among Fe/Beta catalysts prepared by different methods, the highest intensity of peaks assigned to Fe 3+ could be observed on Fe/Beta(LIE) sample.Considering its excellent C3H6-SCR performance, it is considered that the higher amount of Fe 3+ species will facilitate NO reaction with C3H6.

H2-TPR
Usually the reduction properties of metal species is closely related with their chemical environment, then the H2-TPR technique was carried out to speculate iron species on Fe/Beta catalysts.The TPR profiles of different Fe/Beta samples and a comparison with Fe2O3 sample (analytically pure, Sinopharm) are showed in Figure 4.As expected, the profile of Fe2O3 exhibited a low-temperature peak at 455 C and a high-temperature peak centered at about 750 C, which typically attributed to the reduction sequence as Fe2O3→Fe3O4→FeO→Fe [23,24].Different from the TPR profile of Fe2O3, the main reduction peaks of all Fe/Beta samples appeared above 600 C, which were obvious higher than that of Fe2O3.According to our previous study [2,25], the TPR peaks of iron supported metal oxides (e.g., Al2O3, TiO2, and SiO2) at temperature below 500 C correspond to the reduction of Fe 3+ to Fe 2+ in nanoparticles, as well as the reduction of Fe2O3 to Fe3O4, which indicated

XPS
XPS analysis was carried out to identify the chemical state of surface iron species.The Fe 2p spectra of Fe/Beta catalysts are shown in Figure S1.The low resolution of Fe 2p spectra is probably due to the low iron content on the surface of Fe/Beta catalysts.It is seen that all Fe/Beta catalysts exhibited two peaks of Fe 2P spectra, the first at 711.3 eV for the Fe 2p3/2 and the second at 725.3 eV for the Fe 2p1/2, with a satellite peak located at 718.9 eV.The binding energies of these peaks can be ascribed to Fe 3+ ions in the literatures [21,22].The results imply that iron species on all samples surface primarily existed in the +3 valence state.Among Fe/Beta catalysts prepared by different methods, the highest intensity of peaks assigned to Fe 3+ could be observed on Fe/Beta(LIE) sample.Considering its excellent C 3 H 6 -SCR performance, it is considered that the higher amount of Fe 3+ species will facilitate NO reaction with C 3 H 6 .

H 2 -TPR
Usually the reduction properties of metal species is closely related with their chemical environment, then the H 2 -TPR technique was carried out to speculate iron species on Fe/Beta catalysts.The TPR profiles of different Fe/Beta samples and a comparison with Fe 2 O 3 sample (analytically pure, Sinopharm) are showed in Figure 4.As expected, the profile of Fe 2 O 3 exhibited a low-temperature peak at 455 • C and a high-temperature peak centered at about 750 • C, which typically attributed to the reduction sequence as Fe 2 O 3 →Fe 3 O 4 →FeO→Fe [23,24].Different from the TPR profile of Fe 2 O 3 , the main reduction peaks of all Fe/Beta samples appeared above 600 • C, which were obvious higher than that of Fe 2 O 3 .According to our previous study [2,25], the TPR peaks of iron supported metal oxides (e.g., Al 2 O 3 , TiO 2 , and SiO 2 ) at temperature below 500 • C correspond to the reduction of Fe 3+ to Fe 2+ in nanoparticles, as well as the reduction of Fe 2 O 3 to Fe 3 O 4 , which indicated the presence of crystalline iron oxide phases.The difficulty in reducing the iron species implied that the chemical environment of the iron species of Fe/Beta was different from that of supported metal oxides.We conclude the crystalline phases of Fe 2 O 3 were not the main iron species on Fe/Beta samples due to the reduction attributed to iron oxide species did not occur in the H 2 -TPR.Wei et al. [26] had also investigated the reduction properties of Fe containing SAPO-34 catalysts, and found that the supported Fe species of impregnated FeAPSO-34 began to reduce at temperature above 300 • C, while no iron species reduction occurred even until 900 • C for synthesized FeAPSO-34.
Catalysts 2019, 9, x FOR PEER REVIEW 6 of 17 the presence of crystalline iron oxide phases.The difficulty in reducing the iron species implied that the chemical environment of the iron species of Fe/Beta was different from that of supported metal oxides.We conclude the crystalline phases of Fe2O3 were not the main iron species on Fe/Beta samples due to the reduction attributed to iron oxide species did not occur in the H2-TPR.Wei et al. [26] had also investigated the reduction properties of Fe containing SAPO-34 catalysts, and found that the supported Fe species of impregnated FeAPSO-34 began to reduce at temperature above 300 C, while no iron species reduction occurred even until 900 C for synthesized FeAPSO-34.Notice that all Fe/Beta samples presented a weak reduction peak at 142 C, which was much lower than that of ferric oxide.According to the literature [27,28], there may be three different sites for isolated metal ions located in BEA zeolites: the straight channel (α site), a planar deformed sixmember ring of the hexagonal cage (β site), and the hexagonal cage (γ site).When the iron species are introduced into the zeolites by liquid ion-exchange, the predominant type of the ferric ions is located in the β site, and a minor amount of ferric ions anchored to α and γ site are formed at Fe/Al >0.1.Moreover, it was found that metal ions in α site are most easily reduced, while ions in γ site are the most difficult reduced site.Ma et al. [20] had also reported the similar reduction peak appeared in Fe/Beta(LIE) samples with low iron content and suggested that such iron species are most likely isolated Fe 3+ in α site.Based on the above analysis, the main Fe species in Fe/Beta samples with low Fe content were isolated Fe 3+ ions located in zeolites.The more easily reduced iron species might be related to isolated Fe 3+ in α site, while much more stabilized isolated Fe 3+ in β site would be dominant on Fe/Beta samples.

UV-Vis
UV-vis diffuse reflectance spectra was further applied to characterize Fe species on these Fe/Beta catalysts, and the results are displayed in Figure 5.It is obvious that there was no distinct characteristic bands over H-Beta support, then the bands of Fe/Beta were attributed to Fe-O chargetransfer of different iron species.Schwidder et al. [29] and Feng et al. [30] assigned the bands below 290 nm to isolated Fe 3+ in tetrahedral and octahedral coordination, the bands in the range of 300-400 nm to oligomeric FexOy clusters, and the bands above 400 nm to Fe2O3 nanoparticles on the external surface of the zeolite.The UV-vis spectra of Fe2O3 (as shown in Figure S2) also showed that the primary bands appeared above 400 nm, which could be attributed to Fe2O3 crystallites species.To obtain the detailed information of Fe species, the UV-vis spectra was de-convoluted into several subbands by Gaussian method and the results are shown in Figure S3.Moreover, the semi-quantitative analysis of different Fe species by the area ratios of the sub-bands was summarized in Figure 6.Notice that all Fe/Beta samples presented a weak reduction peak at 142 • C, which was much lower than that of ferric oxide.According to the literature [27,28], there may be three different sites for isolated metal ions located in BEA zeolites: the straight channel (α site), a planar deformed six-member ring of the hexagonal cage (β site), and the hexagonal cage (γ site).When the iron species are introduced into the zeolites by liquid ion-exchange, the predominant type of the ferric ions is located in the β site, and a minor amount of ferric ions anchored to α and γ site are formed at Fe/Al >0.1.Moreover, it was found that metal ions in α site are most easily reduced, while ions in γ site are the most difficult reduced site.Ma et al. [20] had also reported the similar reduction peak appeared in Fe/Beta(LIE) samples with low iron content and suggested that such iron species are most likely isolated Fe 3+ in α site.Based on the above analysis, the main Fe species in Fe/Beta samples with low Fe content were isolated Fe 3+ ions located in zeolites.The more easily reduced iron species might be related to isolated Fe 3+ in α site, while much more stabilized isolated Fe 3+ in β site would be dominant on Fe/Beta samples.

UV-Vis
UV-vis diffuse reflectance spectra was further applied to characterize Fe species on these Fe/Beta catalysts, and the results are displayed in Figure 5.It is obvious that there was no distinct characteristic bands over H-Beta support, then the bands of Fe/Beta were attributed to Fe-O charge-transfer of different iron species.Schwidder et al. [29] and Feng et al. [30] assigned the bands below 290 nm to isolated Fe 3+ in tetrahedral and octahedral coordination, the bands in the range of 300-400 nm to oligomeric Fe x O y clusters, and the bands above 400 nm to Fe 2 O 3 nanoparticles on the external surface of the zeolite.The UV-vis spectra of Fe 2 O 3 (as shown in Figure S2) also showed that the primary bands appeared above 400 nm, which could be attributed to Fe 2 O 3 crystallites species.To obtain the detailed information of Fe species, the UV-vis spectra was de-convoluted into several sub-bands by Gaussian method and the results are shown in Figure S3.Moreover, the semi-quantitative analysis of different Fe species by the area ratios of the sub-bands was summarized in Figure 6.Obviously, the distribution of three iron species (e.g.isolated Fe 3+ , oligomeric FexOy clusters, and iron oxide nanoparticles) significantly depended on preparation methods.It can be seen that isolated Fe 3+ ions located at ion-exchanged sites were the main species (more than 50%) on all Fe/Beta samples.The LIE sample showed the strongest characteristic peak (as shown in Figure S3 and Figure 5) around 270 nm, which indicated the formation of high dispersion Fe 3+ ions with H-Beta interface.The concentration of isolated Fe 3+ ions could be observed: Fe/beta(LIE) > Fe/beta(IWI) > Fe/beta(SIE).On the contrary, the concentration of Fe2O3 nanoparticles followed the order: Fe/beta(SIE) > Fe/beta(IWI) > Fe/beta(LIE).There were more Fe2O3 nanoparticles (more than 30%) on the surface of SIE and IMI samples.These results agree well with the BET and XRD results, indicated SIE and IMI methods inclined to produce more iron oxide nanoparticles.However, the concentration of oligomeric FexOy clusters on three Fe/Beta samples was relatively low (less than 16%).Gao et al. [31] reported that two main Cu species present on Cu/SAPO-34 catalysts prepared by SIE method, and found that isolated Cu 2+ ions are highly active and selective in NH3-SCR, while CuO particles mainly contribute to the non-selective oxidation of NH3 above 350 C.The role of the different iron species in C3H6-SCR reaction will be further discussed later with the results of catalytic activity and in situ DRIFTS studies.Obviously, the distribution of three iron species (e.g.isolated Fe 3+ , oligomeric FexOy clusters, and iron oxide nanoparticles) significantly depended on preparation methods.It can be seen that isolated Fe 3+ ions located at ion-exchanged sites were the main species (more than 50%) on all Fe/Beta samples.The LIE sample showed the strongest characteristic peak (as shown in Figure S3 and Figure 5) around 270 nm, which indicated the formation of high dispersion Fe 3+ ions with H-Beta interface.The concentration of isolated Fe 3+ ions could be observed: Fe/beta(LIE) > Fe/beta(IWI) > Fe/beta(SIE).On the contrary, the concentration of Fe2O3 nanoparticles followed the order: Fe/beta(SIE) > Fe/beta(IWI) > Fe/beta(LIE).There were more Fe2O3 nanoparticles (more than 30%) on the surface of SIE and IMI samples.These results agree well with the BET and XRD results, indicated SIE and IMI methods inclined to produce more iron oxide nanoparticles.However, the concentration of oligomeric FexOy clusters on three Fe/Beta samples was relatively low (less than 16%).Gao et al. [31] reported that two main Cu species present on Cu/SAPO-34 catalysts prepared by SIE method, and found that isolated Cu 2+ ions are highly active and selective in NH3-SCR, while CuO particles mainly contribute to the non-selective oxidation of NH3 above 350 C.The role of the different iron species in C3H6-SCR reaction will be further discussed later with the results of catalytic activity and in situ DRIFTS studies.Obviously, the distribution of three iron species (e.g., isolated Fe 3+ , oligomeric Fe x O y clusters, and iron oxide nanoparticles) significantly depended on preparation methods.It can be seen that isolated Fe 3+ ions located at ion-exchanged sites were the main species (more than 50%) on all Fe/Beta samples.The LIE sample showed the strongest characteristic peak (as shown in Figure S3 and Figure 5) around 270 nm, which indicated the formation of high dispersion Fe 3+ ions with H-Beta interface.The concentration of isolated Fe 3+ ions could be observed: Fe/beta(LIE) > Fe/beta(IWI) > Fe/beta(SIE).On the contrary, the concentration of Fe 2 O 3 nanoparticles followed the order: Fe/beta(SIE) > Fe/beta(IWI) > Fe/beta(LIE).There were more Fe 2 O 3 nanoparticles (more than 30%) on the surface of SIE and IMI samples.These results agree well with the BET and XRD results, indicated SIE and IMI methods inclined to produce more iron oxide nanoparticles.However, the concentration of oligomeric Fe x O y clusters on three Fe/Beta samples was relatively low (less than 16%).Gao et al. [31] reported that two main Cu species present on Cu/SAPO-34 catalysts prepared by SIE method, and found that isolated Cu 2+ ions are highly active and selective in NH 3 -SCR, while CuO particles mainly contribute to the non-selective oxidation of NH 3 above 350 • C. The role of the different iron species in C 3 H 6 -SCR reaction will be further discussed later with the results of catalytic activity and in situ DRIFTS studies.

C 3 H 6 -SCR Performance
The C 3 H 6 -SCR catalytic behavior of Fe/Beta catalysts were examined in excess of oxygen, and the results are displayed in Figure 7a.For H-beta sample, the maximum NO conversion was only 30% within the whole temperature region.After the introduction of iron species, the NO reduction efficiency over Fe/Beta catalysts increased significantly at the temperature ranges of 250-450 • C. For Fe/Beta(LIE) sample, the maximum NO conversion reached >90% and N 2 selectivity achieved >95% in the range of 300-450 • C. It is found that the C 3 H 6 -SCR activity decreased in the following order: Fe/Beta(LIE) > Fe/Beta(IWI) > Fe/Beta(SIE) > H-beta.At nearly the same iron content, Fe/Beta(LIE) catalyst exhibited superior catalytic activity than other catalysts, indicating that the preparation method might lead to the different distribution of iron species, which was an important factor affecting C 3 H 6 -SCR activity.Furthermore, Fe/Beta(LIE) catalyst in the present case has shown higher HC-SCR activity than other Fe-based zeolite catalysts reported in the literatures (e.g., Fe/ZSM-5 [15], Fe/MFI [32], Fe/SBA-15 [33]), probably due to the Beta zeolite had better adsorption property of hydrocarbons than other zeolites [34].
the results are displayed in Figure 7a.For H-beta sample, the maximum NO conversion was only 30% within the whole temperature region.After the introduction of iron species, the NO reduction efficiency over Fe/Beta catalysts increased significantly at the temperature ranges of 250-450 C.For Fe/Beta(LIE) sample, the maximum NO conversion reached >90% and N2 selectivity achieved >95% in the range of 300-450 C.It is found that the C3H6-SCR activity decreased in the following order: Fe/Beta(LIE) > Fe/Beta(IWI) > Fe/Beta(SIE) > H-beta.At nearly the same iron content, Fe/Beta(LIE) catalyst exhibited superior catalytic activity than other catalysts, indicating that the preparation method might lead to the different distribution of iron species, which was an important factor affecting C3H6-SCR activity.Furthermore, Fe/Beta(LIE) catalyst in the present case has shown higher HC-SCR activity than other Fe-based zeolite catalysts reported in the literatures (e.g., Fe/ZSM-5 [15], Fe/MFI [32], Fe/SBA-15 [33]), probably due to the Beta zeolite had better adsorption property of hydrocarbons than other zeolites [34].
The propylene conversions of Fe/Beta catalysts during the reduction of NO were compared in Figure 7b.The highest C3H6 conversion was obtained over Fe/beta(LIE) catalyst when temperature was below 300 C, and the other Fe/Beta samples exhibited the lower conversion but better than that over H-beta sample, which were consistent with the order of C3H6-SCR activity.When the temperature was above 350 C, the C3H6 conversion over all Fe/Beta catalysts achieved more than 90%.However, the propylene conversion remained stable with a further increase of temperature, while NO conversion dropped sharply above 450 C.It implies that the propylene combustion would dominate at high temperatures in C3H6-SCR reaction, resulting in the low selectivity of propylene toward NO reduction.Taking into account the co-existence of H2O and SO2 in the gas emission form vehicles, it is of particular interest to investigate the influence of H2O/SO2 on DeNOx performance of Fe/Beta catalysts.The corresponding results concerning the effect of H2O/SO2 on C3H6-SCR activity over Fe/beta(LIE) catalyst are depicted in Figure 8.It is found that similar DeNOx performance over Fe/beta(LIE) in the presence of 5% H2O could be observed with the active window (e.g.temperature for 50% conversion) shifting a little to lower temperature.On the other hand, the DeNOx performance was slightly suppressed below 300 C and the active window shifted by 50 C to high temperature in the presence of 0.02% SO2.Furthermore, the coexistence of 5% H2O and 0.02% SO2 led to more serious activity decline when the reaction temperature was below 350 C.However, Fe/beta(LIE) still maintained high activity in the temperature range of 350-450 C in the coexistence of H2O and SO2.Kucherov et al. [35] proposed that the ferric ions in tetrahedral and distorted tetrahedral sites are responsible for the H2O tolerance of Fe/Zeolite because water vapor do not chemically adsorb on those ferric ions above 200 C.The above results suggest that Fe/beta(LIE) catalysts had moderate resistance to SO2 and preferable insensitivity to H2O.The propylene conversions of Fe/Beta catalysts during the reduction of NO were compared in Figure 7b.The highest C 3 H 6 conversion was obtained over Fe/beta(LIE) catalyst when temperature was below 300 • C, and the other Fe/Beta samples exhibited the lower conversion but better than that over H-beta sample, which were consistent with the order of C 3 H 6 -SCR activity.When the temperature was above 350 • C, the C 3 H 6 conversion over all Fe/Beta catalysts achieved more than 90%.However, the propylene conversion remained stable with a further increase of temperature, while NO conversion dropped sharply above 450 • C. It implies that the propylene combustion would dominate at high temperatures in C 3 H 6 -SCR reaction, resulting in the low selectivity of propylene toward NO reduction.
Taking into account the co-existence of H 2 O and SO 2 in the gas emission form vehicles, it is of particular interest to investigate the influence of H 2 O/SO 2 on DeNO x performance of Fe/Beta catalysts.The corresponding results concerning the effect of H 2 O/SO 2 on C 3 H 6 -SCR activity over Fe/beta(LIE) catalyst are depicted in Figure 8.It is found that similar DeNO x performance over Fe/beta(LIE) in the presence of 5% H 2 O could be observed with the active window (e.g., temperature for 50% conversion) shifting a little to lower temperature.On the other hand, the DeNO x performance was slightly suppressed below 300 • C and the active window shifted by 50 • C to high temperature in the presence of 0.02% SO 2 .Furthermore, the coexistence of 5% H 2 O and 0.02% SO 2 led to more serious activity decline when the reaction temperature was below 350 • C.However, Fe/beta(LIE) still maintained high activity in the temperature range of 350-450 • C in the coexistence of H 2 O and SO 2 .Kucherov et al. [35] proposed that the ferric ions in tetrahedral and distorted tetrahedral sites are responsible for the H 2 O tolerance of Fe/Zeolite because water vapor do not chemically adsorb on those ferric ions above 200 In addition, hydrothermal durability is a key criterion for SCR catalyst of diesel engine exhaust due to the periodical process of particulate filter regeneration (temperature up to ∼700 C).In this work, a harsh hydrothermal treatment (700 C for 24 h in 10 vol.%H2O) was performed on Fe/beta(LIE) catalyst.The C3H6-SCR performance of the fresh and aged Fe/Beta(LIE) samples was compared in Figure 9.When the reaction temperature was below 300 C, the aged sample displayed the similar catalytic activity to that of the fresh sample.Notably the aged sample still maintained high activity with NO conversion >90% in the temperature range of 300-400 C.However, the activity of aged sample was obviously less than the fresh sample above 400 C, indicating that hydrothermal treatment had a slight inhibition effect at high temperatures on C3H6-SCR.9. NO conversion over the fresh and aged Fe/Beta(LIE) catalysts.The hydrothermal aging was performed at 700 C for 24 h in the presence of 10% H2O.Reaction conditions are same as those in Figure 7.

In Situ DRIFTS
The adsorbed intermediates on the surface of different Fe/Beta catalysts were further investigated by in situ DRIFTS.Figure 10 shows the FT-IR spectra for different exposure times for NO + O2 + C3H6 adsorption at 200 °C.In addition, hydrothermal durability is a key criterion for SCR catalyst of diesel engine exhaust due to the periodical process of particulate filter regeneration (temperature up to ∼700 • C).In this work, a harsh hydrothermal treatment (700 • C for 24 h in 10 vol.%H 2 O) was performed on Fe/beta(LIE) catalyst.The C 3 H 6 -SCR performance of the fresh and aged Fe/Beta(LIE) samples was compared in Figure 9.When the reaction temperature was below 300 • C, the aged sample displayed the similar catalytic activity to that of the fresh sample.Notably the aged sample still maintained high activity with NO conversion >90% in the temperature range of 300-400 • C.However, the activity of aged sample was obviously less than the fresh sample above 400 • C, indicating that hydrothermal treatment had a slight inhibition effect at high temperatures on C 3 H 6 -SCR.In addition, hydrothermal durability is a key criterion for SCR catalyst of diesel engine exhaust due to the periodical process of particulate filter regeneration (temperature up to ∼700 C).In this work, a harsh hydrothermal treatment (700 C for 24 h in 10 vol.%H2O) was performed on Fe/beta(LIE) catalyst.The C3H6-SCR performance of the fresh and aged Fe/Beta(LIE) samples was compared in Figure 9.When the reaction temperature was below 300 C, the aged sample displayed the similar catalytic activity to that of the fresh sample.Notably the aged sample still maintained high activity with NO conversion >90% in the temperature range of 300-400 C.However, the activity of aged sample was obviously less than the fresh sample above 400 C, indicating that hydrothermal treatment had a slight inhibition effect at high temperatures on C3H6-SCR.
Figure 9. NO conversion over the fresh and aged Fe/Beta(LIE) catalysts.The hydrothermal aging was performed at 700 C for 24 h in the presence of 10% H2O.Reaction conditions are same as those in Figure 7.

In Situ DRIFTS
The adsorbed intermediates on the surface of different Fe/Beta catalysts were further investigated by in situ DRIFTS.Figure 10 shows the FT-IR spectra for different exposure times for NO + O2 + C3H6 adsorption at 200 °C.

In Situ DRIFTS
The adsorbed intermediates on the surface of different Fe/Beta catalysts were further investigated by in situ DRIFTS.Figure 10 shows the FT-IR spectra for different exposure times for NO + O 2 + C 3 H 6 adsorption at 200 • C.
The weak vibrations located at 3128 and 2952 cm −1 could correspond to N 2 O 4 adsorption products [38].Some negative bands around 3682 cm −1 were attributed to the vibrations of surface hydroxyl O-H stretching [39].Among three tested samples, Fe/beta(LIE) exhibited much stronger adsorption toward nitrate species, especially for NO 2 adspecies (nitro or adsorbed NO 2 molecule) on Fe 3+ sites [40] at 1628 cm −1 , which was hardly observed on the surface of Fe/beta(IWI) and Fe/beta(SIE).In addition, the intensity of NO 2 adspecies decreased significantly after evacuation, indicating that a fraction of NO 2 (probably from the oxidation of NO) physically adsorbed on iron sites.After NO + O2 co-adsorption for 60 min, the FT-IR spectra of Fe/beta(IWI) and Fe/beta(SIE) samples were quite similar.It can be seen that the bands assigned to bidentate nitrate [36] , unidentate nitrate [37], and monodentate nitrite appeared at 1589 cm -1 , 1261 cm -1 , and 1157 cm -1 , respectively.The weak vibrations located at 3128 and 2952 cm -1 could correspond to N2O4 adsorption products [38].Some negative bands around 3682 cm −1 were attributed to the vibrations of surface hydroxyl O-H stretching [39].Among three tested samples, Fe/beta(LIE) exhibited much stronger adsorption toward nitrate species, especially for NO2 adspecies (nitro or adsorbed NO2 molecule) on Fe 3+ sites After introduction of C 3 H 6 for 30 min over Fe/beta(LIE), some oxidation products from propylene appeared in the 1700-1300 cm −1 range.The new band at 1396 cm −1 corresponded to formate species [41], and the band at 1675 cm −1 was assigned to amide or organic nitro-compounds [42].The 1628 cm −1 band ascribed to NO 2 adspecies disappeared in company with the intensity of nitrate species decreased gradually.However, Fe/beta(IWI) and Fe/beta(SIE) exhibited weak ability for propylene selective activation, formate species could hardly be perceived in the FT-IR spectra, only the stretching vibration of =C-H assigned to adsorbed propylene [42] appeared around 2962 cm −1 during the exposure of C 3 H 6 at 200 • C. Liu et al. [43] studied the formation of acetate/formate species by in situ DRIFTS and DFT calculations, which considered as the prerequisite for the generation of -NCO and -CN species.Considering C 3 H 6 -SCR activity results, we speculate that formate species generated from the selective oxidation of adsorbed propylene would react readily with the adsorbed NO 2 /NO 3 − species over Fe/beta(LIE), whereas adsorbed propylene over Fe/beta(IWI) and Fe/beta(SIE) might be less active for NO reduction at low temperatures.

Discussion
C 3 H 6 -SCR activity tests showed that the catalytic behavior of Fe/Beta catalysts prepared by different methods was quite different.At the same support and iron loading, Fe/Beta(LIE) catalyst showed the most active performance for C 3 H 6 -SCR reaction, which might be related to the maximum relative content of Fe 3+ ions.In order to further reveal the correlation of the preparation methods on the iron species and C 3 H 6 -SCR performance, some kinetic studies of the reaction were carried out over Fe/Beta catalysts.According to the percentage of Fe species and NO conversion over Fe/Beta catalysts (Figures 6 and 7), we correlated the content of isolated Fe 3+ with the NO conversion rate at 200 • C, and the results are showed in Figure 11.It is interesting to see that NO conversion rate increased monotonically with increasing the content of isolated Fe 3+ species on Fe/Beta.These results implied that the isolated Fe 3+ ions might be the immediate factor affecting the C 3 H 6 -SCR activity.Furthermore, we calculated the apparent turnover frequencies (TOF, number of NO molecules converted per Fe per second) of Fe/Beta with total iron, and the TOF function at 200 • C is showed Figure S4.The Fe/Beta(LIE) sample with more isolated Fe 3+ ions showed a higher frequencies than other samples with the same iron loading, which also indicated isolated Fe 3+ might be more active for NO reduction.
[40] at 1628 cm -1 , which was hardly observed on the surface of Fe/beta(IWI) and Fe/beta(SIE).In addition, the intensity of NO2 adspecies decreased significantly after evacuation, indicating that a fraction of NO2 (probably from the oxidation of NO) physically adsorbed on iron sites.
After introduction of C3H6 for 30 min over Fe/beta(LIE), some oxidation products from propylene appeared in the 1700-1300 cm -1 range.The new band at 1396 cm −1 corresponded to formate species [41], and the band at 1675 cm -1 was assigned to amide or organic nitro-compounds [42].The 1628 cm −1 band ascribed to NO2 adspecies disappeared in company with the intensity of nitrate species decreased gradually.However, Fe/beta(IWI) and Fe/beta(SIE) exhibited weak ability for propylene selective activation, formate species could hardly be perceived in the FT-IR spectra, only the stretching vibration of =C-H assigned to adsorbed propylene [42] appeared around 2962 cm -1 during the exposure of C3H6 at 200 °C.Liu et al. [43] studied the formation of acetate/formate species by in situ DRIFTS and DFT calculations, which considered as the prerequisite for the generation of -NCO and -CN species.Considering C3H6-SCR activity results, we speculate that formate species generated from the oxidation of adsorbed propylene would react readily with the adsorbed NO2/NO3 − species over Fe/beta(LIE), whereas adsorbed propylene over Fe/beta(IWI) and Fe/beta(SIE) might be less active for NO reduction at low temperatures.

Discussion
C3H6-SCR activity tests showed that the catalytic behavior of Fe/Beta catalysts prepared by different methods was quite different.At the same support and iron loading, Fe/Beta(LIE) catalyst showed the most active performance for C3H6-SCR reaction, which might be related to the maximum relative content of Fe 3+ ions.In order to further reveal the correlation of the preparation methods on the iron species and C3H6-SCR performance, some kinetic studies of the reaction were carried out over Fe/Beta catalysts.According to the percentage of Fe species and NO conversion over Fe/Beta catalysts (Figure 6 and Figure 7), we correlated the content of isolated Fe 3+ with the NO conversion rate at 200 °C, and the results are showed in Figure 11.It is interesting to see that NO conversion rate increased monotonically with increasing the content of isolated Fe 3+ species on Fe/Beta.These results implied that the isolated Fe 3+ ions might be the immediate factor affecting the C3H6-SCR activity.Furthermore, we calculated the apparent turnover frequencies (TOF, number of NO molecules converted per Fe per second) of Fe/Beta with total iron, and the TOF function at 200 °C is showed Figure S4.The Fe/Beta(LIE) sample with more isolated Fe 3+ ions showed a higher frequencies than other samples with the same iron loading, which also indicated isolated Fe 3+ might be more active for NO reduction.From the results of Figure 7, it follows that the SCR reaction was predominant over the Fe/Beta(LIE) catalyst at temperatures below 400 °C but the propylene combustion would prevail at temperatures above 400 °C.Since the hydrocarbon combustion was typically accompanying the HC-SCR reaction and N2O was hardly observed in the products, there were two main reaction pathways in C3H6-SCR over Fe/Beta, which could be represented as: From the results of Figure 7, it follows that the SCR reaction was predominant over the Fe/Beta(LIE) catalyst at temperatures below 400 • C but the propylene combustion would prevail at temperatures above 400 • C. Since the hydrocarbon combustion was typically accompanying the HC-SCR reaction and N 2 O was hardly observed in the products, there were two main reaction pathways in C 3 H 6 -SCR over Fe/Beta, which could be represented as: (1) According to our previous work [25,44], Fe 2 O 3 nanorods dispersed on the surface of supports play a crucial role for NO reduction while enhance the complete oxidation of hydrocarbon, then a small amount of excess hydrocarbons is required for achieving a satisfactory NO conversion in the presence of oxygen.Fierro et al. [15] reported Fe x O y nanoparticles on over-exchanged Fe-ZSM-5 catalysts are not so effective for the HC-SCR reaction but enhance the propane combustion to CO 2 .Consequently, the isolated Fe 3+ species were contributed to the C 3 H 6 -SCR reaction, Fe 2 O 3 nanoparticles were less active for C 3 H 6 -SCR in presence of oxygen while enhanced the hydrocarbon combustion.
In situ DRIFTS further identified different iron species through the intermediates probably originated from specific iron sites.On the Fe/Beta(LIE) sample, the intensity of adsorbed intermediates (e.g., NO 2 adspecies and formate species) was obviously stronger than that of other samples, which indicated the formation of these intermediates mainly bonded to isolated Fe 3+ sites.On the contrary, the adsorbed propylene species on Fe/beta(IWI) and Fe/beta(SIE) were much more than Fe/beta(LIE), implied Fe 2 O 3 nanoparticles preferred to promote the formation of adsorbed propylene.M. Santhosh Kumar et al. [45] reported that the formation of nitrogen oxide intermediates in higher oxidation states requires the presence of trivalent active Fe sites, which undergo a periodic Fe 3+ /Fe 2+ redox cycle.However, Fe x O y clusters have a higher oxidizing power than isolated Fe 3+ sites, then lead to the unselective oxidation of the isobutane reductant.Based on our previous work [25,46], the HC-SCR reaction usually begins with the formation of NO 2 /NO 3 − and C x H y O z species.Thus more isolated Fe 3+ sites on Fe/Beta(LIE) would enhance the formation of the key intermediates, i.e., NO 2 adspecies and formate species, then led to the superior C 3 H 6 -SCR activity.Moreover, the catalytic activity decreased at high temperatures after hydrothermal treatment (Figure 9), which is another evidence that the distribution of iron species will affect C 3 H 6 -SCR performance.As can be seen from the XRD profiles (Figure S5), the Beta crystalline phase maintained well even after hydrothermal aging, implying that the inhibition of activity was not related to the deterioration of zeolite framework structure.Compared with the UV-vis spectra with deconvolution for the fresh and aged sample (Figure S6), the intensity of the band below 290 nm decreased while the intensity of the band above 400 nm increased after hydrothermal aging.The relative percentage of isolated Fe 3+ ions (the species more active for NO reduction) decreased from 65.2% to 51.3%, oligomeric Fe x O y clusters (the species may be less active for NO reduction) increased from 13.5% to 20.7%, and Fe 2 O 3 nanoparticles (the species more active for hydrocarbon combustion) increased from 21.3% to 28.0%, as shown in Table S1.Kucherov et al. [35] pointed out that isolated Fe 3+ can migrate out of position and be incorporated into Fe x O y clusters during high temperature calcination.Therefore, the slight decrease of catalysis activity at high temperatures after hydrothermal treatment might be due to the migration of isolated Fe 3+ ions into oligomeric clusters and/or Fe 2 O 3 nanoparticles.
On the other hand, the strong acidity of Beta zeolite may be one of important factors for HC-SCR reaction.Then Py-FTIR tests were further performed to investigate the nature of the acid sites on Fe/beta catalysts, and the results are showed in Figure 12.All Fe/Beta samples showed characteristic peaks around 1545, 1490, and 1450 cm −1 , corresponding to the pyridine coordinated on Brønsted acid sites, Lewis-Brønsted acid complex, and Lewis acid sites, respectively [47,48].Table 2 listed the concentration of acid sites on Fe/beta samples calculated according to Beer-Lambert law and the IMEC values.Under desorption conditions at 150 and 300 • C, the strength of Brønsted (229.23 and 197.24 µmol/g) and Lewis acidity (555.66 and 251.65 µmol/g) on Fe/Beta (LIE) sample was almost the lowest in three samples.It is general considered that Brønsted acid sites of zeolite are caused by the bridging hydroxyl protons in the vicinity of the tetrahedrally coordinated lattice-Al atoms.When iron was introduced to H-Beta, Fe 3+ ion could substitute the H proton of bridging hydroxyls on zeolite, therefore Fe/Beta(LIE) with more isolated Fe 3+ ions showed weaker acidity compared with Fe/Beta(IWI) and Fe/Beta(SIE) samples.Interestingly, Fe/Beta(LIE) catalyst with weaker acidity exhibited superior catalytic activity, which indicated that too many acid sites might not be necessary for C 3 H 6 -SCR reaction on Fe/Beta catalysts.The results are in accordance with previous reports [20,49], which showed that acidity of Fe-zeolite catalyst is not a crucial factor for high SCR activity, but Brønsted acid sites are necessary to bind and disperse the metal ions.
Catalysts 2019, 9, x FOR PEER REVIEW 13 of 17 exhibited superior catalytic activity, which indicated that too many acid sites might not be necessary for C3H6-SCR reaction on Fe/Beta catalysts.The results are in accordance with previous reports [20,49], which showed that acidity of Fe-zeolite catalyst is not a crucial factor for high SCR activity, but Brønsted acid sites are necessary to bind and disperse the metal ions.
For the LIE method, Fe was introduced into the H-Beta by liquid ion-exchange with Fe(NO3)3 solution (0.1 M, 1 g zeolite/10 ml solution).The slurry was stirred vigorously at room temperature for 24 h.Then the mixture was centrifuged and washed repeatedly with deionized water until the filtrate became neutral, followed by drying at 110 °C for 12 h and calcining at 500 °C in air for 5 h.The process including ion-exchange, centrifugation, rinsing and drying was repeated several times.For the SIE method, Fe was introduced into the H-Beta by solid-state ion-exchange with Fe(NO3)3·9H2O powder (0.11 g iron nitrate/1g zeolite), the mixture of fine milling powders was placed in a flow reactor, heated to 500 °C for 2 h under nitrogen atmosphere.The obtained sample was then dried at 110 °C for 12 h and calcined at 500 °C in air for 5h.For the IWI method, Fe(NO3)3 solution was added dropwise into H-Beta powder to give a 1.5% Fe loading on the catalyst.Then mixture was kept stationary at room temperature for 12 h, following the drying step for 12 h at 110 °C.Finally, the dried sample was calcined at 500 °C in air for 5 h.
The resulting catalysts with the Fe loading of about 1.5 wt.% were denoted as Fe/Beta(LIE), Fe/Beta(SIE), Fe/Beta(IWI), respectively.The hydrothermal treatment of Fe/Beta(LIE) sample was
For the LIE method, Fe was introduced into the H-Beta by liquid ion-exchange with Fe(NO 3 ) 3 solution (0.1 M, 1 g zeolite/10 mL solution).The slurry was stirred vigorously at room temperature for 24 h.Then the mixture was centrifuged and washed repeatedly with deionized water until the filtrate became neutral, followed by drying at 110 • C for 12 h and calcining at 500 • C in air for 5 h.The process including ion-exchange, centrifugation, rinsing and drying was repeated several times.For the SIE method, Fe was introduced into the H-Beta by solid-state ion-exchange with Fe(NO 3 ) 3 •9H 2 O powder (0.11 g iron nitrate/1g zeolite), the mixture of fine milling powders was placed in a flow reactor, heated to 500 • C for 2 h under nitrogen atmosphere.The obtained sample was then dried at 110 • C for 12 h and calcined at 500 • C in air for 5h.For the IWI method, Fe(NO 3 ) 3 solution was added dropwise into H-Beta powder to give a 1.5% Fe loading on the catalyst.Then mixture was kept stationary at room temperature for 12 h, following the drying step for 12 h at 110 • C. Finally, the dried sample was calcined at 500 • C in air for 5 h.
The resulting catalysts with the Fe loading of about 1.5 wt.% were denoted as Fe/Beta(LIE), Fe/Beta(SIE), Fe/Beta(IWI), respectively.The hydrothermal treatment of Fe/Beta(LIE) sample was carried out at 700 • C in 10 vol.%H 2 O for 24 h, and the aged catalyst was designated as Fe/Beta(LIE)-Aged.

Catalysts Characterizations
The textural property of samples was characterized by adsorption/desorption of N 2 at −196 • C using Micromeritics ASAP-2460 apparatus (Norcross, CA, USA).The specific surface area and pore volume were calculated based on the BET and BJH equation, respectively.High-resolution transmission electron microscopy (HRTEM) was carried out in a JEM-2100F (JEOL, Tokyo, Japan) electron microscope with an accelerating voltage of 100 kV.The iron content in Fe/Beta was measured by inductively coupled plasma-atomic emission (ICP-OES, Agilent 725, Santa Clara, CA, USA).
The powder XRD was performed on a D/max-2550PC diffractometer (Tokyo, Japan) using Cu Kα radiation source (λ = 1.5418Å).The XRD patterns were collected in the range of 5-80 • at a scan rate of 2 • /min.XPS experiments were carried out on a ThermoFisher K-Alpha spectrometer (Waltham, MA, USA) using Al-Kα (1486.6 eV) radiation as the excitation source, the binding energies of Fe 2p, O 1s and Al 2p were referenced to the 284.6 eV C 1s.UV-vis diffuse reflectance spectra measurements were performed on a UV3600 spectrometer equipped with an ultraviolet-visible near-infrared spectrometer (KYOTO, Japan).The detection wavelength range was 190-800 nm (the integrating sphere was 190-2600 nm).
H 2 -TPR measurements were performed on an AutoChem II 2920 chemisorption analyzer (Newark, DE, USA).Prior to reduction, the sample (100 mg) was pretreated in He atmosphere at 300 • C for 30 min and then cooled down to 50 • C, and the reduction process was carried out up to 900 • C with a ramp rate of 10 • C/min in a stream of 10 vol.%H 2 /Ar.
The acid sites of the catalysts were measured by pyridine adsorption using FT-IR Frontier Infrared Spectrometer (Py-FTIR, Denver, CO, USA).The sample (100 mg) was pretreated at 300 • C in vacuum for 60 min, then cooled down to room temperature and adsorbed by pyridine for 30 min, finally infrared spectra were recorded when pyridine desorption occurred at 150 and 300 • C.

Catalytic Activity Tests
The activity of Fe/Beta catalysts in the SCR of NO with C 3 H 6 was conducted in a quartz tube microreactor, and the schematic diagram and experimental procedures were reported previously [2].In this study, the reactant gases containing 500 ppm NO, 500 ppm C 3 H 6 , 2% O 2 , 5% H 2 O (when needed), 200 ppm SO 2 (when needed), and the balance of N 2 were introduced at a flow rate of 100 mL/min (GHSV = 15,000 mLg −1 h −1 ).The concentrations of C 3 H 6 , NO, NO 2 , N 2 O and SO 2 were online measured by FTIR spectroscopy (Thermo Nicolet IS10) and the spectra were collected for a period of 30 min at the aimed temperature.NO conversion and N 2 selectivity were calculated according to the following equations:

In Situ DRIFTS Studies
The intermediates on the surface of catalysts during the reaction were investigated by in situ FTIR spectroscopy (Thermo Nicolet IS50), which equipped with an MCT/A detector using DRIFT cell.The scan number was 64 with a resolution of 8 cm −1 .Prior to each experiment, Fe/Beta sample (∼50 mg) was pretreated in N 2 at 500 • C for 2 h to remove impurities.The gas mixture consisted of 500 ppm NO, 2% O 2 with N 2 as the balance (20 mL/min) was introduced into the reaction cell, then saturated adsorption for 60 min followed by evacuation for 30 min, finally the gas mixture containing 0.1% C 3 H 6 /N 2 was introduced and a series of time-dependent DRIFT spectra were recorded.

Conclusions
In this work, we studied the selective catalytic reduction of NO with C 3 H 6 over Fe/Beta catalysts prepared by LIE, SIE, and IWI methods.Combined with the characterization results, isolated Fe 3+ ions, oligomeric Fe x O y clusters and iron oxide nanoparticles coexisted on Fe/Beta catalysts with low iron loading.The preparation methods had a significant influence on the composition and distribution of iron species, and LIE method inclined to produce more isolated Fe 3+ ions at ion-exchanged sites than IWI and SIE method.
C 3 H 6 -SCR activity showed Fe/Beta(LIE) sample exhibited the highest catalytic activity with NO conversion >90% and N 2 selectivity >95% at temperature 300-450 • C, and Fe/Beta(LIE) sample showed the appropriate resistance to SO 2 and H 2 O. Kinetic studies of C 3 H 6 -SCR reaction suggested that isolated Fe 3+ species were more active for NO reduction, whereas Fe 2 O 3 nanoparticles enhanced the hydrocarbon combustion in excess of oxygen.According to the results of in situ DRIFTS, more isolated Fe 3+ sites on Fe/Beta(LIE) would enhance the formation of the key intermediates, i.e., NO 2 adspecies and formate species, then led to the superior C 3 H 6 -SCR activity.The slight decrease of catalytic activity at high temperatures after hydrothermal treatment might be due to the migration of isolated Fe 3+ ions into oligomeric clusters and/or Fe 2 O 3 nanoparticles.Furthermore, Py-FTIR tests showed the excessive acid sites were not necessary for C 3 H 6 -SCR on Fe/Beta catalysts.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4344/9/2/205/s1, Figure S1: XPS spectra of Fe 2p of Fe/Beta catalysts prepared by different methods; Figure S2: UV-vis spectra of Fe 2 O 3 ; Figure S3: UV-vis spectra with deconvolution method of Fe/Bate catalysts; Figure S4: Apparent TOF with total Fe at 200 • C on Fe/Bate catalysts; Figure S5: XRD diffractograms of the fresh and aged Fe/Bate catalysts; Figure S6: UV-vis spectra with deconvolution method of the fresh and aged Fe/Bate catalysts; Table S1: Percentage of Fe species over the fresh and aged Fe/Bate catalysts.

Figure 2 .
Figure 2. XRD diffractograms of H-Beta and Fe/Beta catalysts, (a) a wide scan in the 2θ region of 5-50° and (b) an enlarged view in the 2θ region of 31-36°.

Figure 2 .
Figure 2. XRD diffractograms of H-Beta and Fe/Beta catalysts, (a) a wide scan in the 2θ region of 5-50 • and (b) an enlarged view in the 2θ region of 31-36 • .

Figure 2 .
Figure 2. XRD diffractograms of H-Beta and Fe/Beta catalysts, (a) a wide scan in the 2θ region of 5-50° and (b) an enlarged view in the 2θ region of 31-36°.

Figure 6 .
Figure 6.Percentage of various Fe species over Fe/Beta catalysts prepared by different methods.(a) Isolated Fe 3+ in tetrahedral and octahedral coordination, (b) Oligomeric Fe x O y clusters, (c) Fe 2 O 3 nanoparticles.

Figure 9 .
Figure 9. NO conversion over the fresh and aged Fe/Beta(LIE) catalysts.The hydrothermal aging was performed at 700 • C for 24 h in the presence of 10% H O. Reaction conditions are same as those in Figure 7.

Figure 11 .
Figure 11.Correlations of the NO conversion rate at 200 °C with the isolated Fe 3+ content on Fe/Beta catalysts.

Figure 11 .
Figure 11.Correlations of the NO conversion rate at 200 • C with the isolated Fe 3+ content on Fe/Beta catalysts.
Author Contributions: H.Z. (Hao Zhou) and Y.S. designed and administered the experiments.M.G. and M.L. performed experiments.H.Z. (Huishuang Zhao) and S.W. collected and analyzed data.H.Z. (Hao Zhou) wrote the manuscript.Funding: This research received no external funding.

Table 1 .
Properties of Fe/Beta catalysts prepared by different methods.LIE: liquid ion-exchange; IWI: incipient wet-impregnation; SIE: solid-state ion-exchange.a Determined by inductively coupled plasma (ICP) analysis.

Table 1 .
Properties of Fe/Beta catalysts prepared by different methods.LIE: liquid ion-exchange; IWI: incipient wet-impregnation; SIE: solid-state ion-exchange.a Determined by inductively coupled plasma (ICP) analysis.
• C. The above results suggest that Fe/beta(LIE) catalysts had moderate resistance to SO 2 and preferable insensitivity to H 2 O.

Table 2 .
Concentration of surface acid sites of Fe/Beta catalysts.

Table 2 .
Concentration of surface acid sites of Fe/Beta catalysts.