Ethylene is one of the most important chemicals with a large demand and production. The manufacture of ethylene in industry currently is by the thermal cracking of liquefied petroleum gas or naphtha ^[1]. This is, however, confronted with the challenges of decreasing petroleum resources and severe environmental problems. As is known, bio-ethanol is a renewable material and it is now produced on a large scale through fermentation processes ^[1-4]. Therefore, the catalytic conversion of ethanol offers a green route to the production of ethylene and has been paid more attention ^[1-14].

Currently, many catalysts have been developed for ethanol dehydration to ethylene, including oxides ^{[11, 12, 15]}, heteropolyacids ^{[8, 16]} and zeolites ^{[5, 17-21]}. A 98.7% ethylene yield was obtained at 460 ℃ by a TiO₂/g-Al₂O₃ catalyst ^[12]. Dogu’s group ^[16] developed heteropolyacid catalysts for use at low reaction temperature; however, the ethylene selectivity was only 77.0% at 250 ℃. Zeolites have been widely used in many fields, such as environm ental decontamination, the detergent industry, and catalysis. Since the 1980s, researchers have used zeolites for the dehydration of ethanol to ethylene ^[1]. The zeolites used were ZSM-5, b zeolite, Si-Al-phosphate (SAPO) zeolite, A type zeolite, and AM-11 type zeolite ^[1]. Among these zeolite catalysts, the most promising is ZSM-5 zeolite ^{[1, 2, 4, 18, 22-24]}, which has many obvious merits, such as a regular pore structure, outstanding thermal and hydrothermal stability, low formation of coke, high surface area, and adjustable acidity ^[25]. Guo’s group ^[4] studied a nano-scale HZSM-5 z eolite that gave >98.43% ethylene selectivity at >98.40% conversion at 240 ℃ for 630 h. Zhao’s group ^[2] used a steam treated HZSM-5 zeolite as the catalyst for the ethanol hydration to ethylene and obtained > 90% ethylene yield at 275 ℃; this excellent performance could be maintained for 350 h. Besides these, metal-modified HZSM-5 zeolite catalysts were also studied ^{[18, 24]}. Le Van Mao et al. ^[24] investigated bioethanol dehydration to ethylene using a ZSM-5/Zn-Mn catalyst. The result showed that when the reaction temperature reached 400 ℃, the conversion of ethanol was 96% and the selectivity to ethylene was 49%. Ouyang et al. ^[18] performed ethanol dehydration to ethylene over La-doped ZSM-5 zeolite. The stability measurement in the laboratorial bioreactor displayed that both the conversion of ethanol and selectivity to ethylene over this catalyst was maintained at >98% for >950 h at 260 ℃.

Iron doped ZSM-5 was verified to be active for the transformation of ethanol into hydrocarbons ^{[14, 26-30]}. The hydrocarbon product consisted of predominantly C₁-C₉ hydrocarbons, i.e., alkanes, olefins and aromatics. However, in the literature, a high iron loading (up to 10 wt% ^[30]), ZSM-5 with a relatively high Si/Al ratio (>40 ^{[26, 31]}), and high reaction temperature (up to 400 ℃ ^{[27, 30]}) were used. As a result, in the conversion of ethanol, C³⁺ olefins and aromatic compounds were the main products ^{[30, 31]}.

In this paper, the dehydration of ethanol to ethylene was carried out over an iron exchanged ZSM-5 with the Si/Al ratio ranging from 25 to 300. The iron exchanged ZSM-5 catalyst with a low Si/Al ratio exhibited a high catalytic performance for ethanol dehydration to ethylene at a low reaction temperature. The structure and properties of H-ZSM-5(25) and Fe-ZSM-5(25) catalysts were also studied.

HZSM-5 powder (purchased from Fuxu Zeolite Corporation, China) with different molar ratios of Si/Al = 25-300 was first calcined in air at 550 ℃ for 4 h. Iron (Fe) exchanged ZSM-5 catalyst was obtained by three consecutive ion exchanges of HZSM-5 with aqueous solutions of iron nitrate (Fe(NO₃)₂×9H₂O). First, the calcined powder of HZSM-5 was dispersed into a 0.3 mol/L aqueous solution of metal nitrate salt using 10 mL solution per gram of zeolite. The pH of the solution was controlled at 1.5 in accordance with previous papers ^{[26, 31]}. After that, the resulting mixture was magnetically stirred at 25 ℃ for 12 h. After filtration, rinsing and drying at 120 ℃, the catalyst was calcined at 550 ℃ for 4 h. The as-prepared catalyst was named as Fe-ZSM-5(X), where X refers to the Si/Al molar ratio of the ZSM-5 zeolite. Before it was used in the reaction, the catalyst was pressed, crushed and sieved into the 20-30 mesh fraction. The iron content of Fe-ZSM-5(25) was determined by flame atomic absorption spectrometry (performed by a Varian AA240 apparatus). The result showed that the Fe-ZSM-5(25) catalyst possessed an iron content of 0.46 wt%.

X-ray diffraction (XRD) was carried out with a Bruker D8-Advance X-ray diffraction equipment with Cu K_α radiation operated on 40 kV and 40 mA. The pattern was recorded over the 2q angle ranging from 5° to 80° at a scan rate of 5°/min. NH₃ temperature-programmed desorption (NH₃-TPD) was conducted by employing a Quantachrome Autosorb-1 analyzer with a TCD. The catalyst was first heated at a rate of 10 ℃/min from room temperature to 450 ℃ and pretreated at that temperature for 0.5 h under argon. After it was cooled to 100 ℃, the catalyst was subjected to NH₃-saturation in a stream of 5% NH₃/He with a flow rate of 40 mL/min. After purging with helium at 100 ℃ for 0.5 h, NH₃ was desorbed by heating the catalyst to 650 ℃ at a rate of 15 ℃ /min.

Fourier transform infrared (FT-IR) spectroscopy of pyridine adsorption was recorded using a Varian 3100 spectrometer equipped with a liquid nitrogen cooled MCT detector. The resolution and scan number were 4 cm^-1 and 32, respectively. First, the catalyst was ground and dispersed in KBr powders (mass ratio = 33/100). Then, it was pressed into a disc and pre-heated at 400 ℃ for 4 h under N₂ in a Harrick in situ IR cell. After it was cooled to room temperature, the catalyst was exposed to pyridine vapor for 1.5 h. The catalyst was heated at 250 ℃ for 1.5 h in a N₂ atmosphere, followed by exposure to 0.1 MPa vacuum at room temperature for 3 h for removing gaseous and physisorbed pyridine molecules. Then the catalyst was used for the FT-IR measurement. Diffuse reflectance spectroscopy (DRS UV-VIS) was conducted by using a Perkin Elmer Lambda 35 spectrometer equipped with a 60 mm integrating sphere using BaSO₄ as standard.

The catalytic reaction was performed in an electrically heated fixed bed reactor (i.d. = 10 mm, length = 400 mm). Catalyst granules (3 g) were charged into the middle of reactor. Before reaction, the temperature was raised to a preset one in the range of 200-320 ℃. Bio-ethanol (60 wt% aqueous solution) was directly pumped into the reactor by a micro-pump without prior gasification or any diluent gas. The liquid hourly space velocity (LHSV) was controlled at 0.84 h^-1. Gaseous and liquid products from the bottom of the reactor were passed through a gas-liquid separator. The gaseous products were analyzed by a TCD gas chromatograph (PE Clarus 500) with a Porapak Q column. The liquid products were analyzed by an FID gas chromatograph (Shandong Lunan SP-2000B GC) with a SE-54 column. The conversion (denoted as c) and product selectivities (denoted as S_i) were calculated in accordance with a previous paper ^[20].

Fig. 1 shows the XRD patterns of HZSM-5(25) and Fe-ZSM-5(25). Both patterns present a set of diffraction peaks at 2θ = 7.9°, 8.8°, 23.0°, 23.2°, 23.6°, 23.9° and 24.4°, indicating that these two catalysts possess the same ZSM-5 zeolite structure. For the Fe-ZSM-5 catalyst, besides the peaks assigned to ZSM-5, no other peak was observed. These observations showed that the ion exchange process neither deteriorated the crystal structure of ZSM-5 zeolite nor generated a new phase detectable by XRD.

Fig. 2 shows the FT-IR spectra of pyridine adsorption on HZSM-5(25) and Fe-ZSM-5(25). Bands at 1540 and 1450 cm^-1 were present. These can be assigned to Brönsted and Lewis acid sites, respectively ^[32]. One can see that both the bands for the Brönsted and Lewis acid sites showed changes in their integrated areas with the catalyst, indicating variations in the concentrations of the Brönsted and Lewis acid sites. The ratio of the integrated areas of the bands at 1540 and 1445 cm^-1 can be employed to evaluate the relative concentration of Brönsted and Lewis acid sites ^[33]. These ratios, denoted as B/L, for the different catalysts are summarized in Table 1. One can see that the Fe-ZSM-5(25) catalyst has an obviously lower B/L value than the HZSM-5(25) zeolite.

Fig. 3 shows the NH₃-TPD profiles of HZSM-5(25) and Fe-ZSM-5(25). The peaks in these desorption profiles can be classified into two groups, low-temperature peaks (LT; < 300 ℃) and high-temperature ones (HT; > 300 ℃). The concentration of acid sites (C) and temperature for the maximum of the peaks (T_m) are listed in Table 1. The distribution of acid sites was also calculated from the TPD profiles by the deconvolution method. The ratios of the concentration of HT acid sites to LT acid sites, denoted as C_H/C_L, were also calculated. HZSM-5(25) exhibited a LT peak at 202 ℃ and a HT peak at 469 ℃, suggesting that two types of acid sites were present. Fe-ZSM-5 showed a clear decrease in the total acidity relative to the pristine HZSM-5 zeolite. Compared to HZSM-5(25), for Fe-ZSM-5(25), the LT peak increased to the higher temperature at 215 ℃. However, the HT peak decreased to a lower temperature (438 ℃).

Fig. 4 shows the UV-VIS spectra of HZSM-5(25) and Fe-ZSM-5(25). The electronic spectrum in the UV-visible region is a useful technique for studying the electronic state of isolated transition metal ions. According to the literature ^{[34, 35]}, the UV/VIS spectrum is especially sensitive to the charge transfer bands of the iron ion. The wavelength of the bands reflects the coordination number and the degree of aggregation. One can see that HZSM-5(25) showed a strong band at 240 nm, which is related to the Al units of the T₁ transition charge transfer processes ^[36]. In comparison, Fe-ZSM-5(25) exhibited three characteristic absorptions at 232, 277, and 357 nm. In Brückner’s work ^[34], the bands below 300 nm were ascribed to isolated Fe³⁺ sites. Their position depends on the number of ligands. In the present work, the band at 232 nm was attributed to isomorphously incorporated Fe³⁺ ions in the zeolite framework with tetrahedral or higher coordination ^{[34, 36]}, and the band at 277 nm to isolated extraframework Fe³⁺ sites in octahedral coordination ^[37]. Thus, we concluded that framework and extraframework Fe³⁺ were present simultaneously in Fe-ZSM-5(25). The small band at 357 nm was associated with the iron ions of small oligomeric Fe_xO_y clusters ^{[34, 38]}. There was hardly any band that can be attributed to large Fe₂O₃ particles, which would show bands above 450 nm ^{[34, 35]}, observed. These results suggested that the Fe species was distributed over the Fe-ZSM-5 catalyst predominantly as isolated framework and extraframework Fe³⁺, and a small amount of oligomeric Fe_xO_y clusters. However, hardly any large Fe₂O₃ particles existed.

The Si/Al ratio of the zeolites remarkably affected the distribution and strength of the surface acid sites which played a vital role in the dehydration of ethanol to ethylene ^[10]. Fig. 5(a) shows the dehydration of ethanol to ethylene over HZSM-5(X) as a function of the X (Si/Al ratio) value. The reaction temperature was 230 ℃. One can see that the conversion of ethanol increased first, and reached a maximum (79.2%) at the Si/Al ratio of 100, and then it decreased with the further increase of the Si/Al ratio value. However, the yield of ethylene decreased monotonically with the Si/Al ratio. HZSM-5(25) zeolite gave the highest yield of ethylene (47%).

Fig. 5(b) shows the ethanol dehydration to ethylene over Fe-ZSM-5(X) as a function of the X (Si/Al ratio) value at 230 ℃. The results showed that both the conversion of ethanol and yield of ethylene decreased uniformly with increasing Si/Al ratio. As a result, the highest yield of ethylene (84%) was obtained at the Si/Al ratio of 25.

Fig. 5(c) shows the effect of reaction temperature on the dehydration of ethanol to ethylene over Fe-ZSM-5(25). When the reaction temperature increased, the conversion of ethanol increased from 53.3% (at 200 ℃) to above 99.8% (above 260 ℃), indicating that this reaction is accelerated by increasing the reaction temperature. However, the yield of ethylene increased first from 47.8% (at 200 ℃) to 98.5% (at 260 ℃) and then it decreased with further increase in reaction temperature. The highest ethylene yield (98.5%) was obtained at the reaction temperature of 260 ℃.

Fig. 5(d) shows the effect of time-on-stream on ethanol dehydration to ethylene over HZSM-5(25) and Fe-ZSM-5(25). At 260 ℃ reaction temperature and 0.81 h^-1 of LHSV, >70% of ethylene yield at >95% ethanol conversion was obtained over HZSM-5(25), which was maintained for about 70 h. Comparatively, at the same reaction conditions, the better catalytic performance (>97% ethylene yield at > 98% ethanol conversion) was obtained over Fe-ZSM-5(25). The high catalytic activity was maintained for as long as 1440 h. The outstanding stability of Fe-ZSM-5(25) indicated a very promising potential for further development.

Table 2 shows the detailed results of the conversion of ethanol over the various catalysts. According to previous work ^{[6, 9, 39, 40]}, the catalytic activity for ethanol dehydration is correlated with the acid centers of the catalysts ^[6]. As is known, the surface acidity of a zeolite is associated with the Si/Al ratio of zeolite. HZSM-5(25) provided a larger acid concentration than HZSM-5(300) due to the lower Si/Al ratio ^[41]. One can see that HZSM-5(25) (Entry 1) exhibited a similar conversion of ethanol to HZSM-5(300) (Entry 2). However, the former catalyst produced large numbers of C³⁺ hydrocarbons, while the latter one only gave ethylene (C₂H₄) and diethyl ether ((C₂H₅)₂O). Takahara et al. ^[39] studied the dehydration of ethanol over various solid acid catalysts and suggested that diethyl ether is the reaction intermediate for the formation of ethylene. This was supported by Domen et al. ^[10] using FT-IR. Inaba et al. ^{[21, 30]} and Calsavara et al. ^[31] studied the conversion of ethanol to hydrocarbons ^{[21, 30, 31]} and proposed that ethanol first dehydrates to generate ethylene, which was further transformed to higher hydrocarbons and aromatics. Therefore, for HZSM-5(300), the low surface acidity led to the formation of diethyl ether as a reaction intermediate. However, for HZSM-5(25), the high surface acidity resulted in the further transformation of ethylene to C³⁺ hydrocarbons as byproducts.

One can see in Table 2 that after ion exchange, Fe-HZSM-5(25) (Entry 3) exhibited a higher yield of ethylene (98.52%) than HZSM-5(25) (73.30%). This would be due to the modification of the acidity of HZSM-5. According to Aguayo et al. ^[40], for the conversion of ethanol to hydrocarbons, the dehydration of ethanol to ethylene occurs over weak acid sites, whereas the subsequent reaction steps which generate the higher hydrocarbons and aromatics require sites of a higher acid strength. The NH₃-TPD results (Fig. 3 and Table 1) displayed that the ion exchange of iron decreased the total acidity of ZSM-5 zeolite, especially the HT acid sites. As a result, the C_H/C_L value of Fe-ZSM-5(25) (0.29) was lower than that of HZSM-5(25) (0.7). Therefore, it is reasonable that the significant reduction of the HT acid sites of HZSM-5 by the ion exchange of iron benefited ethanol dehydration to ethylene.

One can also see in Table 2 that a relatively low reaction temperature (230 ℃, Entry 4) is propitious for the formation of the reaction intermediate diethyl ether through inter- molecular dehydration ^[12]. However, a relatively high reaction temperature (290 ℃, Entry 5) led to the further transformation of ethylene to C³⁺ hydrocarbons. Besides, the influence of LHSV on the catalytic performance was also studied. At 230 ℃, 64.70% ethylene selectivity and 33.37% diethyl ether selectivity at 98.99% ethanol conversion was obtained at LHSV = 0.47 h^-1 (Entry 4). However, with the increase of the LHSV (Entries 6-8), the conversion of ethanol and selectivity to ethylene decreased while the selectivity to diethyl ether increased. This can be explicated by the fact that the contact time of ethanol molecules with the zeolite catalyst was shortened with the increase of LHSV, which resulted in the incomplete reaction of ethanol and led to the formation of the reaction intermediate diethyl ether ^[18].

Phillipes et al. ^[6] studied ethanol dehydration to ethylene on HZSM-5 zeolite. They found that although the Brönsted acid sites on HZSM-5 have high activity for ethanol conversion, these sites also can oligomerize ethylene to produce carbonaceous species, which rapidly cover the active sites and led to a dramatic reduction of HZSM-5 activity. Therefore, when used for ethanol dehydration to ethylene, the catalyst lifetime of a pristine HZSM-5 zeolite is usually short (below 60 h) ^[6]. In the present work, from the FT-IR of pyridine adsorption (Fig. 2), we can see that the B/L value of Fe-ZSM-5(25) was obvious lower than that of HZSM-5(25) (the former was 0.25, and the latter was 1.42). We also can see that in Fig. 5(d), Fe-ZSM-5(25) exhibited a lifetime as long as 1440 h time-on-steam, which was obvious longer than the pristine HZSM-5(25) zeolite ^{[4, 6, 39]}. It was reasoned that the decrease of the amount of Brönsted acid sites on HZSM-5 by the ion exchange of iron helped increase the catalyst lifetime due to the decrease of the generation of carbonaceous species catalyzed by the Brönsted acid sites of the zeolite.

In this work, we prepared the Fe-ZSM-5 catalyst by a method of three consecutive ion exchanges. As is known, the ion exchange method is usually employed to prepare a catalyst with a high dispersion of the metal component ^[42]. In Fig. 4, it was found that the Fe species was distributed over the Fe-ZSM-5 as predominantly isolated framework and extraframework Fe³⁺, and a small amount of oligomeric Fe_xO_y clusters. However, hardly any large Fe₂O₃ particles were detected. Guan et al. ^[37] studied the dehydration of ethanol to ethylene over FeSBA-15 synthesized by a direct synthesis method, and showed results that suggested that isolated Fe³⁺ species serve as the active sites for the formation of ethylene, whereas FeO_x clusters catalyzed the formation of ethylene and especially aldehyde. Therefore, in this work, the good catalytic performance of Fe-ZSM-5(25) was attributable to the large amounts of isolated framework and extraframework Fe³⁺ species and the low concentration of FeO_x clusters and Fe₂O₃ particles on the catalyst.

Iron exchanged ZSM-5 zeolite catalysts with Si/Al molar ratio ranging from 25 to 300 were used for ethanol dehydration to ethylene. Fe-ZSM-5 with a Si/Al ratio equal to 25 showed the best catalytic performance. At 260 ℃ and 0.81 h^-1 of LHSV, 97%-99% yield of ethylene at 98%-99% conversion of ethanol was obtained, which was maintained for 1440 h of time-on- steam. These catalysts were characterized by XRD, NH₃-TPD, FT-IR of pyridine adsorption and DRS UV-VIS. The results suggested that (1) ion exchange neither deteriorated the crystal structure of HZSM-5 zeolite nor generated a new phase detectable by XRD, (2) the ion exchange of iron significantly decreased the strong acid sites of the zeolite, which benefited ethanol dehydration to ethylene, (3) the ion exchange of iron decreased the Brönsted acid sites of the zeolite, which helped extend the catalyst lifetime due to the decrease of the generation of carbonaceous species catalyzed by the Brönsted acid sites of the zeolite, and (4) the doped Fe species on the Fe-ZSM-5 was in the form of predominantly framework and extraframework Fe³⁺, which increased the catalytic activity of the catalyst for ethanol dehydration to ethylene.