Na-ZSM-5 with SiO₂/Al₂O₃ = 25 was obtained from Nankai University. The Na-ZSM-5 was mixed with 0.5 mol/L NH₄NO₃ aqueous solution and heated in a water bath at 80 °C for 24 h, followed by washing and filtration. The resultant NH₄-ZSM-5 was dried overnight at 110 °C and then calcined at 550 °C for 6 h in air to obtain H-ZSM-5. Fe-ZSM-5 catalysts were prepared by liquid ion exchange. Typically, 10 g ZSM-5 precursor was added to FeCl₂·4H₂O solution (17.8 g/L) with constant stirring under a continuous flow of N₂ (200 mL/min) in a water bath at 80 °C. After 24 h, the mixture was filtered and washed with deionized water. The resulting powder was dried overnight at 110 °C and calcined at 600 °C for 6 h in air to obtain Fe-Na-ZSM-5 and Fe-H-ZSM-5, which are hereafter denoted Fe-NaZ and Fe-HZ, respectively. The hydrothermal aging of Fe-NaZ and Fe-HZ was carried out at 750 °C in 10% H₂O/air for 12 h.

H₂-TPR experiments were performed with a Micromeritics AutoChem 2920 chemisorption analyzer equipped with a thermal conductivity detector (TCD). For the analysis, 100 mg of powdered sample was pre-treated in 20% O₂ at 550 °C for 20 min. After cooling to 30 °C in Ar, the H₂-TPR measurements were recorded in 10% H₂/Ar (50 mL/min), with a heating rate of 20 °C/min and a final temperature of 900 °C.

UV-Vis DRS spectra were recorded with a Hitachi UV3010 (Japan). BaSO₄ was used as the reference material. The Fe-ZSM-5 catalysts were diluted with BaSO₄ at a ratio of 1:4. The spectra were converted with the Kubelka-Munk (KM) function for comparison.

In situ DRIFTS experiments of NH₃ adsorption over the Fe-ZSM-5 catalysts were performed on a Fourier transform infrared spectrometer (FTIR; Nicolet Nexus 670). Each sample was pretreated at 550 °C in a flow of 20% O₂/N₂ for 20 min, then purged with pure N₂ for 20 min and cooled to 140 °C in N₂. Then, the samples were exposed to a flow of 1% NH₃/N₂ and the spectra were recorded until NH₃ adsorption saturation.

Powder XRD measurements were carried out on a Panalytical X-ray diffractometer (Netherlands, X’Pert PRO MPD) with a Cu K_α radiation source. The 2θ data from 5° to 60° were collected with a step size of 0.02°.

Solid state ²⁷Al MAS NMR spectra were collected on a Brucker AVANCE III 400MHz WB Solid-State NMR spectrometer. All ²⁷Al MAS NMR spectra were externally referenced to an aqueous solution of Al(NO₃)₃ at 0 ppm, and were obtained at a sample spinning rate of 10 kHz using 15° pulses and a 0.5 s recycle delay, a total of 12000 pulses being accumulated.

The SCR reaction conditions were controlled as follows: 500 ppm NO, 500 ppm NH₃, 5% O₂, 5% H₂O (when used), 100 ppm SO₂ (when used), 500 ppm C₃H₆ (when used), N₂ balance; 50 mg catalyst, total flow rate of 500 mL/min and gas hourly space velocity (GHSV) of 320 000 h⁻¹. The effluent gas was analyzed using an FTIR spectrometer (Nicolet Nexus 670). NO_x conversion (X_NOx) was calculated as follows:

X_NOx = (1 − ([NO] + [NO₂])_out/([NO] + [NO₂])_in) × 100%

The elemental composition of the samples was analyzed using an inductively coupled plasma optical emission spectrometer (ICP-OES). Although the two catalysts were prepared by exactly the same procedure, the Fe loading of Fe-NaZ (1.62 wt%) was higher than that of Fe-HZ (1.27 wt%). Similar results have been reported and explained to be caused by the fact that the change in solution pH during the ion exchange process might be different for different precursors ^[9]. The residual Na content in Fe-NaZ was 0.61 wt%.

The H₂-TPR profiles obtained from fresh and hydrothermally aged Fe-HZ and Fe-NaZ are shown in Fig. 1(a). A mechanical mixture of 5 wt% α-Fe₂O₃ with H-ZSM-5 was prepared and tested for comparison. The H₂-TPR profile of 5 wt% α-Fe₂O₃/H-ZSM-5 showed broad reduction features between 325-650 °C. This suggests that the reduction of iron oxide particles completed below 650°C, by the reduction of Fe₂O₃ to Fe₃O₄ and then reduction of FeO_x to Fe⁰ via FeO ^{[10, 12, 13, 14, 15, 16]}. According to previous work on the H₂-TPR of Fe-ZSM-5 catalysts, the first reduction peak corresponds to the reduction of Fe³⁺ to Fe^{2+ [10, 12, 13, 14, 15, 16]}. When Fe³⁺ is located at ion-exchanged positions, further reduction of Fe²⁺ to Fe⁰ occurs at temperatures above 1000 °C owing to the collapse of the zeolite framework ^{[10, 12]}.

As shown in Fig. 1(a), the H₂-TPR profile of fresh Fe-HZ exhibited one main reduction peak at 370 °C, which can be assigned to the reduction of Fe³⁺ to Fe²⁺. No significant reduction peaks were observed from 550 °C to 900 °C in the profile of Fe-HZ, which indicates that most of the iron in Fe-HZ was located at exchange sites ^{[13, 14]}. The first peak in the H₂-TPR profile of fresh Fe-NaZ can be also attributed to the reduction of Fe³⁺ at exchanged sites to Fe²⁺. The main peaks at 484 °C and the small shoulder at 435 °C should correspond to reduction of larger FeO_x particles ^{[10, 12, 13, 14, 15, 16]}. Fe species located at the same sites should show similar H₂ reduction temperatures. Therefore, the first reduction peak for Fe-NaZ could be assigned to the reduction of Fe species located in ion exchange sites, similar to that seen for the dominating Fe species in Fe-HZ.

After hydrothermal deactivation, the H₂ consumption peaks of the aged samples were shifted to higher temperatures, implying a decease in the reducibility of active iron species in Fe-ZSM-5. Aged Fe-HZ showed a main peak at 405 °C while aged Fe-NaZ showed a main peak at 512 °C with a shoulder at 405 °C. According to Brandenberger et al. ^[10], the reduction peak of iron oxide clusters can shift to higher temperature with their size. The present H₂-TPR results indicate that the migration of Fe ³⁺ from ion exchange sites and the formation of small FeO_x particles took place during the hydrothermal aging. Deconvolution of the H₂ consumption peaks (Fig. 1(b)) revealed that the percentage area of the first peak with respect to the total area of the H₂ consumption peaks was around 34% for both fresh and aged Fe-NaZ. This suggests that the degree of migration of Fe ions in Fe-ZSM-5 is mainly decided by their location in the zeolites.

Fig. 1. (a) H2-TPR profiles of fresh and hydrothermally aged Fe-HZ and Fe-NaZ; (b) Deconvoluted sub-bands of the H2-TPR profiles.

UV-vis DRS was used to study the distribution of Fe species in the Fe-ZSM-5 samples. The UV-Vis spectra of the fresh and aged Fe-ZSM-5 are compared in Fig. 2. Generally, the bands below 300 nm are assignable to isolated Fe³⁺ species located on ion exchange sites, while bands above 300 nm are assignable to oligomeric clusters (between 300-400 nm) and Fe₂O₃ particles (above 400 nm) ^{[17, 18]}. As shown in Fig. 2, the spectra of Fe-NaZ exhibited a significant intensity of bands above 400 nm. In contrast, most of the peak bands of Fe-HZ were below 400 nm. To estimate the Fe species distribution in the catalyst, the UV-vis spectrum of each sample was deconvoluted with r² > 0.999, and the results are summarized in Table 1. From these data we obtained a rough estimation of the percentage of isolated Fe³⁺ species, oligomeric clusters, and Fe₂O₃ particles compared with the total Fe present in Fe-ZSM-5. The relative concentration of isolated Fe³⁺ species in Fe-HZ was found to be higher than that in Fe-NaZ, whereas Fe-NaZ contained more oligomeric Fe_xO_y clusters and Fe₂O₃ particles. After hydrothermal aging, the intensity of the bands above 300 nm increased, which indicates that hydrothermal aging caused the migration of isolated Fe³⁺ species from iron exchange sites and the formation of small clusters of iron oxides, which confirms the H₂-TPR results.

Fig. 2. UV-vis spectra of fresh and hydrothermally aged Fe-HZ and Fe-NaZ.

Table 1 Area percentages of sub-bands derived by deconvolution of the UV-vis DRS spectra. Catalyst I1 (%) I2 (%) I3 (%) Fresh Fe-HZ 50.2 31.7 18.1 Fresh Fe-NaZ 44.0 32.4 23.6 Aged Fe-HZ 46.9 33.7 19.4 Aged Fe-NaZ 40.6 34.1 25.3 I1, I2, and I3 represent the percentage of isolated Fe3+ species (subbands at λ ≤ 300 nm), oligmeric clusters (subbands at 300 nm < λ < 400 nm), and Fe2O3 particles (subbands at λ > 400 nm), respectively.

Table 1 Area percentages of sub-bands derived by deconvolution of the UV-vis DRS spectra.

The in situ DRIFTS spectra of various samples measured after saturated NH₃ adsorption are shown in Fig. 3. The bands related to NH₃-adsorbed species appear at 3500-1400 cm⁻¹ after NH₃ adsorption over Fe-ZSM-5. The bands related to NH₃ adsorbed species appearing at 3500-250 cm⁻¹can be assigned to the N-H stretching vibration of adsorbed NH₃ ^[14]. The band at 1480 cm⁻¹is assigned to the symmetric bending vibration of NH₄⁺ chemisorbed onto Brønsted acid sites ^{[14, 17, 18, 19, 20]}. The negative band at 3610 cm⁻¹ can be assigned to the OH stretch of Brønsted acid sites resulting from the interaction of surface hydroxyls with NH₃, which may reflect the relative concentration of Brønsted acid sites on the Fe-ZSM-5 ^{[10, 19, 20, 21]}. The intensity of the band at 3610 cm⁻¹ was higher for Fe-HZ than Fe-NaZ. A magnification of the3610 cm⁻¹ band is inset in Fig. 3. No IR bands corresponding to Brønsted acid sites were observed for Na-ZSM-5. Protonic sites can be introduced into Na-ZSM-5 by the ion exchange process ^{[10, 11]}. The same bands were observed for fresh Fe-NaZ after NH₃ adsorption, but the intensity was lower compared with those observed for fresh Fe-HZ. After hydrothermal aging, the intensity of the negative band at 3610 cm⁻¹and the intensity of the bands of NH₃ adsorption species over Fe-HZ and Fe-NaZ were significantly decreased.

Fig. 3. DRIFTS spectra of NH3 adsorbed over Fe-ZSM-5 catalysts (inset: magnification of the 3610 cm-1 band).

Figure 4 shows the XRD patterns of the fresh and hydrothermally aged Fe-HZ and Fe-NaZ catalysts, and of a mechanical mixture of 5% α-Fe₂O₃ with H-ZSM-5 for comparison. The aged Fe-HZ and Fe-NaZ exhibited little change compared with those of the fresh samples. This indicates that the crystalline zeolite structure was retained after hydrothermal aging. The typical XRD peaks of α-Fe₂O₃ were observed at 33.1°, 35.6°, 40.9°, and 49.5° ^{[14, 22, 23]}. Low intensity peaks of α-Fe₂O₃ were detected in the XRD patterns of fresh and aged Fe-NaZ. This confirmed the presence of Fe₂O₃ particles in Fe-NaZ.

Fig. 4. XRD patterns of the different catalysts.

The ²⁷Al MAS NMR spectra of fresh and hydrothermally aged Fe-HZ and Fe-NaZ are shown in Fig. 5. The fresh Fe-ZSM-5 samples showed a strong signal at around 55 ppm, attributed to tetrahedrally coordinated aluminum, and a weak signal at around 0 ppm, assignable to octahedrally-coordinated aluminum (extra-framework) ^{[10, 24, 25]}. After hydrothermal aging, the intensity of the peak at 55 ppm was decreased owing to the removal of the aluminum from the tetrahedral zeolite framework. The intensity of the octahedral aluminum peak at 0 ppm was also decreased in the spectra of the aged samples. These results indicate that the aluminum removed from the tetrahedral zeolite framework was converted into distorted “NMR invisible” Al species ^{[10, 24]}. The intensity of the tetrahedral aluminum peak in aged Fe-HZ more significantly was reduced than that of aged Fe-NaZ. The integrated peak area of the tetrahedral Al³⁺ ions in the aged samples was around 53% in Fe-NaZ and 26% in Fe-HZ. This indicates that the zeolite framework of Fe-NaZ was more stable during hydrothermal aging than that of Fe-HZ.

Fig. 5. Solid state 27Al-NMR spectra of fresh and hydrothermally aged Fe-HZ (a) and Fe-NaZ (b).

We conducted a comparative study of the effect of hydrothermal deactivation, and the presence of SO₂ and C₃H₆ on the SCR activity of the Fe-HZ and Fe-NaZ catalysts. Figure 6(a) shows NO_x conversion as a function of reaction temperature at 190-550 °C over fresh and hydrothermally aged Fe-HZ and Fe-NaZ catalysts. The fresh Fe-HZ showed higher SCR activity than the fresh Fe-NaZ within the whole reaction temperature range. Hydrothermal aging resulted in a significant decrease in NO_x conversion over the Fe-ZSM-5 catalysts, especially at low temperatures. The aged Fe-HZ also showed higher SCR activity than the aged Fe-NaZ.

The effect of SO₂ on the Fe-ZSM-5 catalysts is shown in Fig. 6(b). When 100 ppm SO₂ was added into the flue gas, a decline in NO_x conversion at low temperatures (below 350 °C) and a slight increase in NO_x conversion at high temperatures (>350 °C) were observed. The SO₂ durability (100 ppm) of Fe-HZ and Fe-NaZ was investigated at 325 °C with the copresence of 5% water (Fig. 6(d)). The SCR reaction was left to stabilize for 1 h before adding SO₂ and water. A sharp decline in the NO_x conversion of both the Fe-HZ and Fe-NaZ catalysts was observed upon addition of the SO₂ and water to the gas flow. The NO_x conversion of Fe-HZ was decreased from ca. 80% to ca. 55%, while that of Fe-NaZ was decreased from ca. 65% to ca. 45% after 20 h of reaction.

Fig. 6. SCR performance of Fe-ZSM-5 catalysts. (a) Hydrothermally deactivated; (b) Effect of SO2; (c) Effect of C3H6; (d) SO2 durability; (e) C3H6 durability.

Figure 6(c) shows the NO_x and C₃H₆ conversions over Fe-HZ and Fe-NaZ in the presence of 500 ppm C₃H₆. The presence of C₃H₆ in the feed gas stream was found to resulted an apparent decrease in the NO_x conversion ability of both of the catalysts. Fe-NaZ exhibited higher NO_x conversion than Fe-HZ in the presence of C₃H₆. Fig. 6(e) shows the C₃H₆ durability of Fe-HZ and Fe-NaZ at 350°C. After the SCR reaction had been stable for about 1 h, 500 ppm C₃H₆ was added to the reaction gas flow. The decrease in SCR activity caused by adding C₃H₆ was rapid in the first few minutes and then became slower in the time that followed. After 5 h reaction, the NO_x conversion of Fe-HZ decreased from 82% to 40% while that of Fe-NaZ decreased from 79% to 44%. It seems that the impact of C₃H₆ on the SCR activity was more significant for Fe-HZ than Fe-NaZ. When C₃H₆ was removed from the feed gas, the NO_x conversion of Fe-NaZ recovered within 20 min, while the NO_x conversion of Fe-HZ recovered slowly.

The SO₂-poisoned and C₃H₆-poisoned samples were regenerated after the durability tests by heating at 550 °C in air for 30 min. The regenerated catalysts were tested in the standard SCR reaction and the results were compared with those obtained for the fresh samples. As shown in Fig. 7, the NO_x conversions of the poisoned samples were partly recovered after the heat treatment. This indicates that the SO₂ and C₃H₆ poisoning of Fe-ZSM-5 was reversible. A better recovery of activity was observed for Fe-NaZ under the present regeneration conditions than Fe-HZ.

Fig. 7. SCR performance of fresh and regenerated Fe-ZSM-5. (a) After SO2 durability test; (b) After C3H6 durability test.

The acidity and redox properties of Fe-ZSM-5 are two important parameters for the SCR reaction. The Brønsted acid sites of Fe-ZSM-5 are important for activating NH₃ species and are assumed to have a promoting effect for the SCR reaction at low temperatures ^{[21, 26]}. Fe active sites are involved in the NO oxidation to NO₂, which is considered the slow step of the SCR reaction ^{[22, 27]}. Among the various types of Fe species coexisting in the Fe-ZSM-5 catalysts, the isolated Fe³⁺ species are considered to contribute to the SCR reaction at low temperatures (< 300 °C) ^{[23, 28, 29]}. Therefore, the higher SCR activity of Fe-HZ might be mainly attributed to the higher amount of active Fe³⁺ ions and the greater number of Brønsted acid sites in this catalyst compared with those in Fe-NaZ.

The characterization results showed that hydrothermal aging of Fe-ZSM-5 catalysts led to the dealumination of the framework, the migration of Fe ions to form Fe_xO_y clusters, and a significant decrease in acidity, in accordance with previous reports ^{[10, 30]}. Protons remaining on zeolites are well known to accelerate the process of steam-induced aluminum loss ^{[10, 11, 15]}. Fe-NaZ contained fewer protonic sites than Fe-HZ, so the dealumination of the zeolite framework by hydrothermal deactivation occurred to a much lesser extent for Fe-NaZ. The Brønsted acid sites of aged-Fe-HZ and aged-Fe-NaZ were significantly decreased to the point of almost completely disappearing. Therefore, the higher SCR activity of aged Fe-HZ should be attributed to the greater number of residual Fe³⁺ species located in the ion-exchanged sites of this catalyst. The loss of redox active Fe sites, rather than dealumination of zeolite framework, should be responsible for the decrease in SCR activity.

The impact of SO₂ on the SCR activity of Fe-ZSM-5 was different at low and high temperatures, in agreement with previous studies ^{[31, 32, 33]}. Long et al. ^[33] suggested that the formation of surface iron sulfate on Fe-ZSM-5 increased the surface acidity of the catalyst and resulted in enhancement of its high temperature activity. At low temperatures (< 325 °C), the formation of ammonium sulfite/sulfate might occur and cover some iron active sites, resulting in a decrease in SCR activity ^[34]. The recovery of low temperature activity of Fe-ZSM-5 after regeneration can be explained as caused by the decomposition of these surface ammonium sulfite/sulfate species at high temperature.

Unburned hydrocarbons are always present in real exhaust gas. Zeolite SCR catalyst can be deactivated by hydrocarbons as a result of active site poisoning and pore blockage ^{[35, 36, 37, 38, 39]}. Pore blockage is known to mainly take place on acid sites ^{[35, 36, 37, 38, 39]}. The present Fe-HZ and Fe-NaZ catalysts were treated at 200 °C in 1000 ppm C₃H₆ in air for 4 h to obtain coked samples. The color of the coked Fe-HZ was dark brown, while that of the coked Fe-NaZ was a lighter brown (Fig. 8). Thermal analysis was carried out on the fresh and coked samples. When compared with those of the fresh catalysts, the mass loss of the HC-poisoned Fe-HZ was 4.7% higher, while that of the Fe-NaZ 1.2% was higher (Fig. 8). This indicates that more coke was formed on Fe-HZ compared with Fe-NaZ. Additionally, Fe-NaZ showed higher C₃H₆ conversion than Fe-HZ during SCR in the presence of C₃H₆ (Fig. 6(c)). Therefore, the better C₃H₆ poisoning resistance of Fe-NaZ should be attributed to less coking on the Fe-NaZ and the higher activity of C₃H₆ oxidation for this catalyst.

Fig. 8. Thermal analysis of Fe-HZ (a) and Fe-NaZ (b) as-prepared and after poisoning in 1000 ppm C3H6 at 200 °C.