Nitrous oxide (N2O) is a greenhouse gas and major stratospheric source of NOx contributing to ozone destruction [1, 2]. Although N2O is not a major contributor to global warming (~6%), its effect is much more potent than that of CO2 and CH4. Anthropogenic practices have led to a rapid increase in atmospheric N2O concentrations, with an annual growth rate of 0.2%-0.3% [1]. Therefore, control of N2O emissions from combustion and chemical processes has become a significant concern.
To control N2O emissions from chemical processes, catalysts for N2O decomposition have been widely studied in the last three decades. Among them, Fe/ZSM-5 [3, 4, 5, 6, 7, 8, 9] has been extensively studied owing to its remarkable activity and stability for N2O decomposition even in the presence of relatively large amounts of O2, NO, SO2, and H2O that are typical poisoning gases for other N2O abatement catalysts. Fe/ZSM-5 has been traditionally prepared by aqueous ion exchange, solid-state ion exchange, and isomorphously substituted methods. Steaming [10, 11, 12, 13, 14] and thermal treatments [15, 16, 17, 18, 19] are beneficial for solid-state ion-exchanged and isomorphously substituted Fe/ZSM-5 for N2O decomposition, and this has led to systematic investigations on the effect of steaming conditions on isomorphously substituted Fe/ZSM-5 [20, 21, 22, 23]. Such research studies reported the key role of oligonuclear Fe oxo clusters or small intra-zeolitic Fe species on the external framework of the zeolite in N2O decomposition, whereas Lewis and Brönsted acidic sites only play a minor role. Kaucký et al. [24] reported that both the short Fe-Fe distances and the presence of Al-Lewis sites in the vicinity of Fe sites resulted in higher N2O decomposition activity. Additionally, Dubkov[13] and Hensen et al. [18, 19] determined that thermal treatment of solid-state ion-exchanged Fe/ZSM-5 in inert gas resulted in the reduction of some Fe3+ species to Fe2+ that promoted N2O decomposition activity. However, rationalization of the advantages of such Fe/ZSM-5 treatments on N2O-decomposition activity remains contr oversial.
Because of its convenient synthesis, aqueous ion-exchanged Fe/ZSM-5 has been considered one of the most promising candidates for the decomposition of N2O emitted from industries. Certain types of iron species, such as iron ions and iron oxyhydroxide species, are prone to exchange with protons of zeolites. Moreover, (FeO)x-Al on the external framework can form in channels following calcination of Fe/ZSM-5 [25, 26, 27]. Previous research studies reported on the performance optimization of aqueous ion-exchanged Fe/ZSM-5 using varying solutions and iron sources [25, 27]. The effect of thermal treatment on solid-state ion-exchanged and isomorphously substituted Fe-ZSM-5 has been studied by Pérez-Ramírez and Hensen et al. However, the influence of thermal treatment on aqueous ion-exchanged Fe/ZSM-5 relating to N2O decomposition activity is not well understood.
Herein, Fe/ZSM-5 catalysts were thermally treated in the temperature range of 600 to 900 ℃ in Ar. The effects of thermal treatment on the structure of Fe oxo species and textural properties of Fe/ZSM-5 were considered and characterizations were performed accordingly. To achieve varying types of iron species during thermal treatment, Fe/ZSM-5 with 2.6 wt% Fe (Fe/Al molar ratio = 1) was prepared by the aqueous ion- exchange method.
Fe/ZSM-5 catalysts were prepared by ion-exchange from aqueous solution. Briefly, HZSM-5 zeolite, with Si/Al molar ratio = 36, was added to Fe(III) nitrate aqueous solution at 90 ℃. After vigorous stirring, the sample was washed and dried in air at 100 ℃ for one night (referred to as non-calcined Fe/ZSM-5). The non-calcined Fe/ZSM-5 was then calcined at 550 ℃ in air for 3 h (referred to as Original). Fe/ZSM-5 prepared with 2.6 wt% Fe (Fe/Al molar ratio = 1) was treated in the temperature range of 600-900 ℃ in pure Ar. The resulting thermally treated catalysts are referred to as FZT (where T refers to the temperature employed). X-ray diffractometry measurements of the samples, as described below, showed that the MFI zeolite structure was retained after the high-temperature treatments.
The chemical composition (including the concentration of Fe) of the catalysts was determined by atomic absorption spectrometry (AAS; Shimadzu, AA6300).
Catalysts were characterized by X-ray diffraction (XRD) using a computerized Rigaku D/max-RB diffractometer (Japan, Cu Kα radiation, 0.154056 nm). Scans were taken from 2θ = 10° to 2θ = 90° at a speed of 4°/min. The accelerating voltage and the applied current were 40 kV and 300 mA, respectively.
The nitrogen adsorption-desorption isotherms were obtained at -196 ℃ on a Quantasorb-18 automatic instrument (Quanta Chrome Instrument Co.). Prior to measurements, samples were evacuated at 300 ℃ for 10 h. Specific areas were computed from the sorption isotherms using the Brunauer-Emmett-Teller (BET) method.
UV-Vis diffuse reflectance spectroscopy (DRS) was conducted on aU-3010 spectrophotometer (Hitachi) equipped with a standard diffuse reflectance unit under ambient conditions. The scanning range was 190-800 nm and the scan rate was 300 nm/min. HZSM-5 was used as a reference material. The measured spectra were converted into Kubelka-Munk functions and deconvoluted into Gaussian sub-bands that could be quantitatively assigned to the different iron species present in the prepared samples.
Fe K-edge X-ray absorption fine structure (XAFS) of the samples was measured on a BL14W1 beamline at the Shanghai Synchrotron Radiation Facility (China). The storage ring was operated at 3.5 GeV with a ring current of 200 mA. A single Si (111) crystal was used to generate a monochromated X-ray beam. The spectra were recorded with sampling steps of 0.5 eV in the X-ray absorption near edge structure (XANES) region and 1-5 eV in the extended X-ray absorption fine structure (EXAFS) region. The Fe K-edge spectra were recorded in fluorescence mode under ambient conditions and the effect of fluorescence saturation was assessed using α-Fe2O3 as a reference. The EXAFS data were extracted from the measured absorption spectra with the XDAP code. The pre-edge was subtracted using a modified Victoreen curve, and the background was subtracted using cubic spline routines. Finally, normalization was performed by dividing the subtracted absorption spectra by the intensity of the absorption spectrum at 50 eV above the Fe K-edge. Structural information was determined by multi-shell fitting in R space [28]. The fits were assessed by k1 and k3 weighting. Errors were estimated at ±10% for the coordination numbers (N), ±1% for the coordination distances (R), ± 5% for the Debye-Waller factor (σ2), and ±10% for the inner-potential correction (E0).
Activity measurements were carried out in a fixed-bed quartz flow reactor (internal diameter, 4 mm), containing ~50 mg catalyst (particle size, 125-200 μm) in all experiments. The reactor was heated by a temperature-controlled furnace. A thermocouple was placed on the external surface of the reactor tube. A N2O (0.15 vol%) reaction mixture was introduced into the reactor at a gas hourly space velocity (GHSV) of 35 000 h-1. Analysis of the reaction products was carried out by gas chromatography (Agilent 6890N equipped with Porapak Q for the analysis of N2O and CO2, and molecular sieve 5A columns for the analysis of N2, O2, and NO). The reaction system was maintained at the selected reaction temperature for 1 h to reach steady state prior to product analysis. In all tests, N2 and O2 were the only gaseous products observed.
The apparent activation energy (Ea) and the turnover frequency (TOF) were calculated [26] from the following equations:
where k is the first-order rate constant and nFe is the number of active iron obtained from the O2-temperature-programmed desorption (O2-TPD) studies, as described in section 2.4.
The O2-TPD experiments were performed under a He (50 cm3/min) flow over 300 mg catalyst and a heating rate of 30 ℃/min. Prior to the experiment, the catalysts were pretreated under a flow of 2% N2O/Ar for 1 h, followed by cooling to 150 ℃ under the same flow. The adsorbed oxygen species remained on the surface after N2O decomposition over the catalyst, and were detected using a mass spectrometer (Hiden).
For the N2O pulse-response experiments, a six-port switching valve was installed in front of the reactor to periodically change between 2 vol% N2O/He and pure He atmospheres. Prior to the experiment, the decomposition of N2O over the catalyst was maintained at 500 ℃ for 1 h to reach steady state. The periodical switching between 2% N2O/He and pure He was performed at intervals of 4 min. A similar procedure was adopted in the air pulse-response experiments to compare the diffusion rate of N2 with that of O2 in the catalyst bed.
Figure 1 shows the steady-state conversions of N2O as a function of temperature over the original and thermally treated Fe/ZSM-5 catalysts. As observed, thermal treatment significantly improved the activity of Fe/ZSM-5 catalysts. Among the prepared catalysts, FZ750, treated at 750 ℃, achieved 100% N2O conversion at the lowest reaction temperature studied. Thus, 750 ℃ was regarded as the optimal treatment temperature. FZ850 featured considerably higher activity than FZ800 and FZ750 in the low-temperature reaction region, whereas FZ750 exhibited the best activity for N2O decomposition in the high-temperature reaction region. This suggests the occurrence of complex transformations in the active species structure during the thermal treatment process.
N2 adsorption was performed to investigate the textural properties of HZSM-5 and Fe/ZSM-5 catalysts. The N2 adsorption-desorption isotherms exhibited high N2 uptakes at low relative pressures and a plateau at high relative pressures (Fig. 2). The isotherms were classified as Type I according to IUPAC classification for microporous materials [29] that is typical of HZSM-5, as investigated in a previous research study [20]. The surface area and total pore volume of the original and thermally treated Fe/ZSM-5 catalysts were calculated from the corresponding N2 adsorption-desorption isotherms, and are listed in Table 1. Upon Fe species exchange with protons, N2 uptake of the non-calcined Fe/ZSM-5 sample decreased (Fig. 2). Accordingly, the surface area and total pore volume of non-calcined Fe/ZSM-5 decreased (Table 1). It was likely that some Fe oxide species incorporated into the zeolite channel during the ion-exchange process, thus reducing N2 uptake on non-calcined Fe/ZSM-5 [19]. Treatment temperatures below 750 ℃ instigated partial migration of small Fe species from the micropore to the external surface of the zeolite, thereby increasing the catalyst pore volume. In contrast, treatment temperatures above 750 ℃ resulted in the aggregation of FeOx clusters and collapse of the ZSM-5 zeolite framework, subsequently blocking the channels in FZ800, FZ850, and FZ900 that was responsible for the low total pore volume and surface area [19].
The XRD patterns of HZSM-5 and thermally treated Fe/ZSM-5 catalysts are shown in Fig. 3. All catalysts displayed highly crystalline ZSM-5 features, as indicated by the strong intensity of the characteristic peaks and the weak background noise in the XRD patterns. These results suggest that the different treatment temperatures employed did not degrade the crystalline structure of the ZSM-5 zeolite.
Hensen et al. [19] reported that severe calcination conditions could induce growth and ordering of the Fe oxide aggregates in Fe/ZSM-5, as prepared by chemical vapor deposition of FeCl3. As observed in Fig. 3, distinct hematite peaks were not observed in FZ600 and FZ700, indicating that the Fe oxides on the catalysts were mainly amorphous. Crystalline Fe2O3 was observed in catalysts FZ750, FZ800, FZ850, and FZ900. To further assess the distribution of the FeOx species, the following experiments were conducted.
The nature and distribution of the Fe species on the catalysts were investigated by UV-Vis DRS spectroscopy. The obtained spectra were converted into Kubelka-Munk functions and assigned to Fe3+←O charge-transfer bands, as shown in Fig. 4. Bands between 190 and 305 nm were attributed to isolated Fe3+ species (Fe1), which are tetrahedrally coordinated within the zeolite framework (band below 250 nm) and with higher coordination (band within 244-305 nm). Octahedral Fe in small oligonuclear Fe oxide species generated bands between 305 and 400 nm (Fe2). Bands above 400 nm were attributed to characteristic asymmetric peaks of large Fe oxide particles. Bands between 400 and 500 nm were attributed to polynuclear Fe oxide (Fe3) and bands above 500 nm were characteristic of large Fe oxide particles without N2O decomposition activity [19, 23, 30]. Based on the work of Pérez-Ramirez et al. [23, 30], a semiquantitative estimation of the distribution of the Fe species in the zeolite was performed by deconvoluting spectra into Gaussian sub-bands, as shown in Table 2 and Fig. 5.
Most of the Fe species in non-calcined Fe/ZSM-5 were octahedral Fe3+ and oligonuclear Fe oxide [30], most probably derived from the ion-exchange process (Fig. 4). In contrast, bulk Fe oxide species were observed in the high- temperature- treated catalysts, as evidenced by the band at ~400 nm. It is interesting to note that the intensity of bands between 400 and 500 nm (I3) initially increased and then decreased with increasing treatment temperatures (Table 2). As previously mentioned, Fe oxide aggregates on the external surface during high-temperature treatments [19, 24, 27]. Based on the XRD results, the presence of the inflexion points was due to several reasons: thermal treatment below 750 ℃ promoted the aggregation of amorphous Fe oxide; the transformation of FeOx species from the amorphous to the crystalline state within the 750-850 ℃ treatment resulted in a reduced volume of FeOx particles; and the aggregation of crystalline Fe oxide particles increased I3 observed for FZ900. The current findings suggested that thermal treatment likely instigated aggregation in both the amorphous and crystalline Fe oxides. The trend observed for I3 that was in accordance with catalyst activity suggested that the polynuclear Fe oxide species, among all Fe oxide species identified, were the most active towards N2O decomposition.
The original and thermally treated Fe/ZSM-5 catalysts, and Fe2O3, FeO, and Fe foil reference compounds were thoroughly examined by XAFS, including XANES and EXAFS. XANES is a fingerprint technique that allows identification of the oxidation state and coordination environment provided availability of a suitable model compound [25, 31]. The normalized Fe K-edge XANES profiles are shown in Fig. 6(a). The absence of a pre-edge absorption was due to the forbidden 1s→3d transition in octahedral coordinations [19]. The weak peak at ~7140 eV, corresponding to 1s→4pz transition, suggests that only the distorted octahedral symmetry of the oxide of Fe3+ was present in the samples [19]. Fig. 6(b) shows the first derivative of the Fe K-edge XANES profiles of the original and thermally treated Fe/ZSM-5 catalysts, and Fe2O3, FeO, and Fe foil. The similarity between the spectra of the Fe/ZSM-5 catalysts and Fe2O3 indicates that the Fe species in the Fe/ZSM-5 catalysts are not reduced during thermal treatment in Ar.
Fe K-edge EXAFS analysis provides information on the structural environment of the Fe species in the original and thermally treated Fe/ZSM-5 catalysts [31]. Fig. 7 shows the k3-weighted Fourier transforms of EXAFS oscillations into R space. The spectra of the original and thermally treated Fe/ZSM-5 catalysts were obviously and significantly different from each other, including their peak position and intensity. In R space, an asymmetric peak was apparent at ~0.16 nm, which was attributed to the six oxygen atoms in the first Fe-O coordination shell [31]. The contribution at 0.25 nm was ascribed to considerable scatter and fitted with the Fe-Fe shell. Multi-shell analysis of the EXAFS data was performed using the models proposed by Battiston et al. [31]. The Fe-O and Fe-Fe references were calibrated for the EXAFS data obtained from hematite Fe2O3 by fitting in R space (see Table 3). In particular, the first oxygen-coordination shell was fitted at two different distances, 0.191 nm (Fe-O)1 and 0.20 nm (Fe-O)2. However, during the curve fitting process, the number of oxygen atoms in the first coordination shell was not always fitted to six, suggesting the existence of Fe-OH [31]. The second coordination shell can be fitted with two Fe-Fe. The fitting results are listed in Table 4. For ease of comparison between the experimental data and curve fitting results, the Fourier transforms of k3-weighted EXAFS oscillations and the corresponding fitting curves are presented in Fig. 8.
Table 4 shows that thermal treatment greatly influenced the coordination number and slightly changed the bond lengths of Fe-O and Fe-Fe. Thermal treatment induced agglomeration and ordering of iron oxide, as reflected by the increasing Fe-Fe coordination numbers. The (Fe-O)2 bond coordination number showed a similar changing trend to that related to the catalysis activity with increasing thermal treatment temperatures, indicating that (Fe-O)2 was the active species for this reaction.
The catalytic N2O decomposition reaction involves the adsorption of N2O at the active catalytic site followed by decomposition, resulting in the formation of N2 and surface oxygen. The surface oxygen can be desorbed upon reaction with another oxygen atom or another N2O. The four steps are shown in the following Eqs.:
The rate of O2 desorption is known to influence the activity of the catalyst [18]. The rate-limiting step is either that shown in Eq. (5) or in Eq. (6) depending on the catalyst.
In this study, the characterization and activity tests demonstrated that the long bond length (Fe-O)2 in the Fe oxide particles was likely advantageous to the activity of the catalysts. To examine the relationship between the intrinsic mechanism of N2O decomposition and the role of (Fe-O)2 in Fe oxides, pulse-response and O2-TPD experiments were conducted.
N2O decomposition is initiated upon activation of N2O on the active site, leading to adsorbed O species (α-O), Eq. (4). Two different mechanisms are postulated for the regeneration of the active site: reaction between another N2O molecule and the oxidized site, Eq. (5), and the recombination of adsorbed oxygen atoms, Eq. (6). To discriminate between these two alternative reactions pathways, pulse-response experiments were performed by pulsing either N2O or air.
Fig. 9(a) shows the response of the N2O pulse over catalyst FZ750. Other samples showed similar behaviors to FZ750. As clearly observed, the evolution of molecular oxygen was delayed when compared with that of N2. As shown in Fig. 9(b), N2 and O2 evolved at the same time in the air pulse study, indicating the absence of different diffusion rates of N2 and O2 in the bed of catalysts. These results demonstrate that the generated rate of diffusion of O2 is lower than that of N2 in N2O decomposition. According to Sun et al. [15], the recombination of adsorbed oxygen atoms, Eq. (6), is the limiting step of the catalytic N2O decomposition process.
Fig. 10 shows the O2-TPD profiles of the thermally treated Fe/ZSM-5 catalysts pretreated with 2% N2O at 485 ℃ for 1 h. The broad peak obtained at 400-800 ℃ was ascribed to the desorption of α-O, which was formed upon decomposition of N2O. As observed in Fig. 10, the trend of the O2 desorption temperature was consistent with that of the activity of the catalysts, confirming the importance of O2 desorption to the activity of the catalysts. For better comparison, the number of effective Fe sites (CFe) calculated according to the amount of the desorbed O2 and turnover frequency (TOF) of N2O decomposition over each active site are listed in Table 5. FZ750 displayed the highest concentration of active iron (CFe), whereas FZ900 featured the highest TOF among all Fe/ZSM-5 catalysts. The trend of CFe was proportionally correlated to the activity of the catalysts, indicating that the number of active sites (nFe) determined the activity of catalysts. The high activity of FZ750 was likely because of the high amount of (Fe-O)2. The crystalline Fe oxide likely promoted N2O decomposition. In contrast, smaller values of CFe reduced the apparent activity of the catalysts.
The nature of Fe oxide strongly influenced the catalytic activity of Fe/ZSM-5 towards N2O decomposition. Based on the XRD data, two different typologies of FeOx present in the catalysts were considered: (1) amorphous state and (2) crystalline state. The UV-Vis DRS results showed that aggregation of the two FeOx species increased with increasing treatment temperatures. The FeOx aggregation phenomenon was in accordance with earlier observations of Fe/ZSM-5 treated under high temperatures [19, 25, 27].
In this study, aggregation of amorphous and crystalline FeOx showed different influence on the activity of the thermally treated Fe/ZSM-5 for direct N2O decomposition. At treatment temperatures below 750 ℃, FeOx was amorphous. The activity of the catalysts improved with increasing treated temperatures, indicating that aggregation of amorphous FeOx improved the activity of the catalysts. Based on the UV-Vis DRS spectra, XAFS, and O2-TPD results of the thermally treated Fe/ZSM-5 catalyst, the long (Fe-O)2 bond length in the amorphous polynuclear FeOx species promoted the activity of catalysts, indicating that (Fe-O)2 was the active species. At treatment temperatures above 750 ℃, crystalline FeOx in the Fe/ZSM-5 catalysts aggregated with increasing treated temperatures. Although the intrinsic activity of crystalline FeOx was higher than that of the amorphous FeOx, as observed from the TOF data, increasing amounts of active Fe-O bonds were sealed in the bulk of crystalline FeOx with increasing treated temperatures, thereby leading to reduced CFe values and decreased catalyst activity.
The rate of N2O decomposition and correlations of thermodynamic parameters are given as follows [18, 26, 32]:
where r is the rate of N2O consumption, k is the rate coefficient, kB is the Boltzmann constant (1.38 × 10-23 J/K), h is the Plank constant (6.626 × 10-34 J∙s), T (K) is temperature, A is the pre-exponential factor, Ea is the apparent activation energy, n is the order of reaction, and Δr‡Smθ is the standard molar entropy of activation. The linear relationship between the reaction rate (r) and N2O partial pressure (pN2O) in Fig. 11 indicates that the decomposition of N2O over FZ750 catalyst is a first-order reaction. Findings from previous research studies [18, 26] also indicated that N2O decomposition is a first-order reaction (n = 1). Arrhenius plots obtained from the steady-state N2O decomposition (below 20% conversion efficiency) over the thermally treated Fe/ZSM-5 catalysts are shown in Fig. 12. Kinetic parameters for the catalysts were calculated from Arrhenius plots under the assumption of first-order disappearance of nitrous oxide, as shown in Table 6. The accuracy of Ea was ±10 kJ/mol.
From Table 6, the apparent activation energy Ea and apparent pre-exponential factor A of the catalysts were substantially different. As seen, Ea increased from 139 kJ/mol for FZ600 to 209 kJ/mol for FZ750, and then decreased to 105 kJ/mol for FZ900. Simultaneously, A increased from 3.0 × 105 to 7.9 × 1010, and then declined to 5.3 × 103 s-1∙Pa-1. The values of Ea were proportional to lnA values of the catalysts (Fig. 13) that can be well explained by the compensation effect [33]. Based on the steady-state activity of the catalysts towards N2O decomposition, A was a better determining factor than Ea to the activity of the catalysts. Equation (6) shows that Δr‡Smθ of the various transition state species (e.g., α-O) plays an important role in determining the value of A. Based on these analyses, the standard molar entropy Δr‡Smθ of α-O and the active sites were the major factors influencing the reaction rate.
The structure of the active site likely influenced the value of active entropy of the transition state of α-O. When α-O deviates from the Fe active sites, prior to rupture of the bond (Fe-O)2, the latter donates electron to α-O because the bond energy of (Fe-O)2 species is lower than that of (Fe-O)1 species, thereby confirming the hypothesis that (Fe-O)2 species are the active species.
Based on the value of A, (Fe-O)2 species promoted the activity of the catalysts by increasing the entropy of the transition state of α-O. The higher the amount of active sites on the catalyst, the larger the total value of active entropy. Therefore, FZ750 that possessed the highest amount of (Fe-O)2 had the highest activity towards N2O decomposition.
The polynuclear Fe oxide species facilitate the recombination of adsorbed oxygen atoms in contrast to the oligonuclear Fe oxide species. (Fe-O)2 existed in both the oligonuclear and polynuclear Fe oxide species. However, the higher amount of (Fe-O)2 species in the polynuclear Fe oxide species was beneficial to enhancing the activity of the catalysts. For example, CN of (Fe-O)2 and the content of the oligonuclear Fe oxide species in the original Fe/ZSM-5 catalyst were both high. However, CFe was low. Therefore, the number of (Fe-O)2 in the polynuclear FeOx species determined the CFe of the catalysts, which was estimated by multiplying CN of (Fe-O)2 by the content of the amorphous polynuclear FeOx species.
The aggregation of FeOx species on the external framework of ZSM-5 zeolite during thermal treatment played a dominant role in the activity of Fe/ZSM-5 towards N2O decomposition. The optimum treatment temperature was determined as 750 ℃. Fe-O with long bond lengths in the polynuclear amorphous FeOx species was the active species for N2O decomposition. The amount of active FeOx species was a key factor influencing the apparent activity of the Fe/ZSM-5 catalysts. When compared with the apparent activation energy Ea, the apparent pre- exponential factor A was more important in determining the reaction rate, where A depends on the total value of active entropy. Acknowledgments
We thank Prof. Yu-ying Huang at Shanghai Synchrotron Radiation Facility (China) for making available synchrotron beam time for us to carry out XAFS measurement.
2014, Vol. 35 Issue (12): 1972-1981
PDF (849 KB)
2014, Vol. 35 Issue (12): 1972-1981
PDF (849 KB)