It is well reported that supported bi-component Pt-Sn catalysts can be employed in many reactions, including alkane dehydrogenation [1-3], hydrocarbon rearrangement [4], low temperature carbon monoxide oxidation [5], and alcohol electro-oxidation [6]. Among these applications, propane dehydrogenation is considered to be one of the most important petrochemical processes because of the rapidly growing demand for propylene [2, 7-9]. Supported Pt-Sn catalysts have been widely used because of their high activities and environmentally friendly characteristics [1, 2]. Aluminium oxide (Al2O3) is the most common support for these materials, owing to its high surface area and acidity. However, Al2O3-supported Pt-Sn catalysts are deficient in terms of poor stability and lack of durability during recovery[1, 10, 11].Thus, the development of novel supports to improve the stability of these catalysts is not only highly desirable but also timely.
In recent years, zeolites such as SUZ-4 and ZSM-5 have been found to be good supports because of their high surface areas, good thermal stabilities, large pore volumes and tunable acidity [12-14]. SUZ-4 is a new type of synthetic zeolite patented by the British Petroleum Company in 1992. In the three dimensional topological structure of this material, straight ten-membered channels intersect with two eight-membered channels at an angle of approximately 74°, which is similar to the structure of ZSM-5 zeolite [15, 16]. SUZ-4-supported catalysts have been widely used in many processes, such as the conversion of n-hexane[16], the synthesis of dimethyl ether from methanol [17], and the elimination of nitrogen oxides [18]. However, there have not yet been any reports concerning the application of SUZ-4-supported catalysts to propane dehydrogenation, while ZSM-5-supported catalysts have attracted significant attention in this regard [13, 19]. In contrast to γ-Al2O3, the three-dimensional microporous ZSM-5 zeolite has a well-defined, ten-membered, ring-crossed channel system that prevents the formation of large hydrocarbon molecules, thus improving the catalyst's stability [13]. Recently, Zhou's group [20] investigated propane dehydrogenation over ZSM-5-supported Pt-Sn catalysts and found that the propylene selectivity was significantly improved by introducing promoters to neutralize the support acidity. The addition of hydrogen to the reaction system effectively inhibited the cracking of propane to C1 and C2 products and also reduced carbon deposition on the catalyst surface, thus improving both the dehydrogenation selectivity and catalytic stability [21]. Despite this, the ZSM-5-based catalysts were still easily deactivated by carbon deposition under the chosen reaction conditions. To resolve this issue, our own group developed the SUZ-4-supported catalyst PtSnNa/SUZ-4, which afforded a 20% propylene yield. Although propylene can be obtained in similar 18%-23% yields by increasing the Sn loading when using a PtSn/ZSM-5 catalyst [22], our catalyst has the advantage of being more robust and undergoing very little deactivation due to carbon deposition.
An SUZ-4 zeolite with the molar ratio SiO2/Al2O3=21 and a ZSM-5 zeolite with the molar ratio SiO2/Al2O3=20 were prepared by methods previously described in the literature [23, 24].In each case, the resulting solid phase was filtered, washed with distilled water several times, dried at 110 ℃ for 12 h and then calcined at 550 ℃ for 4 h. This was followed by NH4+ exchange in aqueous NH4Cl (1 mol/L). H-SUZ-4 and H-ZSM-5 were obtained by calcining the ammonium forms of SUZ-4 and ZSM-5 at 550 ℃ for 4 h. PtSnNa catalysts supported on either the SUZ-4 or the ZSM-5 zeolite (Pt=0.5%, Sn=2.0%, Na=1.0%) were prepared by sequentially impregnating the H-SUZ-4 or H-ZSM-5 with an aqueous mixture of H2PtCl6 and SnCl4 (H2PtCl6=5 mg/mL, SnCl4=5.85 mg/mL) and with aqueous NaCl (0.5 mol/L). The impregnated samples were dried at 110 ℃ for 4 h, calcined at 520 ℃ for 4 h, and then dechlorinated in air containing water vapor at 530 ℃ for 4 h.
The powder X-ray diffraction (XRD) patterns of all samples were obtained with a Philips X'pert pro diffractometer using Cu Kα radiation at 40 kV and 40 mA, from 5° to 50°. Surface areas were calculated by the BET method based on N2 adsorption isotherms recorded at the temperature of liquid nitrogen using a Micromeritics ASAP2010 analyzer. The samples were degassed at 300 ℃ and 0.133 Pa prior to analysis, after which isotherms were acquired at-196 ℃. NH3-TPD profiles of the specimens were obtained in a flow-type fixed-bed reactor at ambient pressure. The catalysts were pre-treated at 500 ℃ for 2 h under an Ar flow. The NH3 adsorption temperature was 100 ℃, and the temperature was raised at a rate of 10 ℃/min. The desorbed NH3 was detected by a gas chromatograph (GC) equipped with a thermal conductivity detector (TCD).
H2 chemisorption on the supported PtSnNa catalysts was assessed both before and after the propane dehydrogenation reaction according to a previously described procedure [25]. Each of the catalysts was reduced in a H2 flow at 500 ℃ for 2 h and then out-gassed in an Ar flow at 540 ℃ for 2 h before H2 chemisorption measurements.
The amount of carbonaceous material deposited on each catalyst during the propane dehydrogenation reaction was measured using thermo-gravimetric (TG) analysis (STA 449C-Thermal star 300 TA-MS apparatus). Catalyst samples (each approximately 0.02 g) obtained after a 10-h reaction were heated from room temperature to 900 ℃ in O2 (at 25 mL/min) at a heating rate of 10 ℃/min, and the amounts of coke on the specimens was calculated from the resulting TG curves.
Temperature-programmed oxidation (TPO) was determined with the same apparatus as used for the H2 chemisorption experiments. An approximately 0.1-g sample was placed in a quartz reactor and then heated to 800 ℃ in a mixture of O2 (at 3.0 mL/min) and Ar (at 30 mL/min) at a heating rate of 10 ℃/min. Temperature-programmed reduction (TPR) was performed using the same apparatus employed during the TPO assessments. In these trials, approximately 0.1-g samples were placed in a quartz reactor and subsequently heated in a flow of 5% H2-95% Ar (at 20 mL/min) at a heating rate of 10 ℃/min.
Purposely poisoned catalysts (containing 0.02% S) were prepared by impregnating the reduced PtSnNa/SUZ-4 or PtSnNa/ZSM-5 catalysts with an ethanol solution of dibenzothiophene (0.2 mg/mL), followed by flushing with nitrogen for 1 to 2 h to evaporate residual ethanol and drying at 110 ℃ for 4 h.
All catalysts were reduced in H2 at 500 ℃ for 2 h before catalytic evaluation. The propane dehydrogenation reaction was carried out in a quartz tubular micro-reactor under reaction conditions of 0.1 MPa, 590 ℃, C3H8/H2=1/3 (molar ratio) and a propane weight hourly space velocity (WHSV) of 3.0-1. The reaction products (C3H6, CH4, C2H6 and C2H4) were analyzed by an online GC (Shimadzu GC-14A, Japan) with a TCD and an Al2O3 packed column.
A series of catalysts with different sodium ion contents were prepared and tested for their activity during propane dehydrogenation. The addition of sodium ions improved both the propane conversion and the propylene selectivity (entries 2, 3 vs. 1), as shown in Table 1. A sodium content of 1.0% was found to be optimal (entries 3 vs. 2, 4) and so was used in the following experiments. It is known that the by-products of this reaction are methane and ethane due to the cracking of propane or propylene on catalyst acid sites. As shown in Fig. 1, the pure zeolite catalysts H-SUZ-4 and H-ZSM-5 exhibited high initial reaction activities but very poor stability, as reflected in the rapidly decreased propane conversions (Fig. 1(a), curves (1) and (2)). In addition, the H-SUZ-4 and H-ZSM-5 both showed very low propylene selectivity (Fig. 1(b), curves (1) and (2)), which is an even more crucial property for industrial production. The evident deactivation and low propylene selectivity were likely due to the excessive concentrations of acid sites on these materials, leading to carbon deposits and undesirable cracking reactions, respectively. The PtSnNa-based catalysts, including both PtSnNa/SUZ-4 and PtSnNa/ZSM-5, were much more advanced. Between these two, the PtSnNa/SUZ-4 was found to be superior based on its higher stability and propylene selectivity (Fig. 1(a), curves (3) vs. (4); Fig. 1(b), curves (3) vs. (4)). Moreover, the PtSnNa/SUZ-4 afforded the highest propylene yields of up to 20% (Fig. 1(c), curves (3) vs. (1, 2) and (4.5)). It should be noted that, even without a catalyst, some small amount of propane was still transformed, although with poor selectivity, because of cracking reactions (Figs. 1(a-c), curves (5)).
Each of the specimens was assessed by XRD. As shown in Fig. 2, no crystalline phase signals attributable to Pt, Sn or Na were observed after the impregnation, calcination or hydrogen reduction steps, possibly due to the low concentrations and/or the high dispersion states of these metals (Fig. 2, curves (2) and (5)) [18].Moreover, compared with the original structures of the SUZ-4 and ZSM-5 zeolite supports, the XRD patterns of the supported catalysts do not show any characteristic changes (Fig. 2, curves (1) vs. (2); (4) vs. (5)). After the propane dehydrogenation reaction was performed for 10 h, the signals of the SUZ-4 and ZSM-5 zeolites are also observed in the XRD patterns of the catalysts, as shown in Fig. 2, curves (3) and (6), indicating that the supports were stable during the reaction.
Fig. 3 presents the nitrogen adsorption-desorption isotherms of the catalysts. The H-SUZ-4 and H-ZSM-5 generated types Ⅰ and Ⅳ adsorption-desorption isotherms, respectively (curves (2) and (4)), corresponding to microporous and mesoporous materials. Furthermore, the adsorption-desorption isotherms of the supported catalysts are similar to those of the supports (curves (1) vs. (2), (3) vs. (4)). This result indicates that the uniform pore structures of the supports were maintained after loading of the Pt, Sn and Na components.
Table 2 summarizes the characterization data for different catalysts. It is evident that the pore volumes (Vp) and surface areas of the supported catalysts were decreased (entry 2 vs. 1; entry 4 vs. 3) and that these values were further reduced following the reaction (entries 2 and 4). These results can possibly be attributed to blocking of the support pores by the Pt or by coke. Compared with the PtSnNa/ZSM-5, the Vp of the PtSnNa/SUZ-4 after the reaction was decreased by a smaller amount, clearly showing that the pores in this new material were less likely to be blocked by coke deposition. The amounts of coke deposited on each catalyst during the reaction could be determined from TG analysis and are presented in Table 2. The pore sizes of both catalysts were increased following the reaction. In addition, the H2 uptake of the fresh PtSnNa/SUZ-4 catalyst was close to that of the fresh PtSnNa/ZSM-5 catalyst, which is consistent with their similar catalyst activities in the initial reaction stage (Fig. 1(a)). Compared with the PtSnNa/SUZ-4, the H2 uptake of the PtSnNa/ZSM-5 catalyst was decreased to a greater extent (ca. 36%) after coke deposition (9 wt.%). However, the differences in the performance of these two catalysts during propane dehydrogenation cannot be attributed to the insignificant difference in the deposited coke amounts. These results indicate that there might be some other reasons for the rapid deactivation of the PtSnNa/ZSM-5 catalyst.
To obtain more information regarding the reaction mechanism, the acidities of the different catalysts were examined by NH3 temperature-programmed desorption (TPD). The NH3-TPD profiles are displayed in Fig. 4 while the detailed data are given in Table 3. Here the acid amount is proportional to the corresponding desorption peak area. The different crystal structures of the H-ZSM-5 and H-SUZ-4 zeolite resulted in variations (both in strength and amount) in the surface acidity of the supports, and loading with PtSnNa led to a more obvious change. Two NH3 desorption peaks appear in these TPD profiles. The first signal, in the range of 190 to 200 ℃, is attributed to weak acid absorption, while the second peak, at 420 to 520 ℃, results from strong acid sites (Fig. 4) [26, 27]. The TPD profile of the PtSnNa/ZSM-5 is quite different from that of the H-ZSM-5. A broadened peak showing stronger NH3 desorption at low temperature appears, while the strong acid absorption signal is absent (Fig. 4, curves (1) vs. (2)), indicating that the addition of the Sn and Na promoters might preferentially neutralize the strong acid sites. Similar phenomena have been observed for PtSnNa/AlSBA-15[28]. The total number of acid sites was increased when metals were loaded on the ZSM-5-supported catalysts (Table 3, entries 2 vs. 1), but in the case of SUZ-4-supported catalysts, the acid sites decreased with the loading of PtSnNa (Table 3, entries 4 vs. 3). It should be noted that the acidity strength of the H-ZSM-5 catalyst was much weaker than that of the H-SUZ-4 catalyst. This is in agreement with a report that zeolites with mutually intersecting 10-ring channels have been found to possess lower acid strengths than those with intersecting 10-and 8-ring channels [29]. In addition, compared with the H-SUZ-4, the total acid sites of the PtSnNa/SUZ-4 decreased more severely, which benefitted the propane dehydrogenation [30]. Because excessive acidity can lead to the undesirable cracking reaction, the PtSnNa/ZSM-5 catalyst showed lower propylene selectivity than the PtSnNa/SUZ-4 (Fig. 1(b), curves (4) vs. (3)).
To investigate the detailed structures of these materials, experiments were performed using dibenzo[b, d]thiophene-poisoned catalysts. As can be seen from Fig. 5, both the PtSnNa/SUZ-4 and PtSnNa/ZSM-5 showed decreased activity after being treated with dibenzo[b, d]thiophene (curves (1) vs. (4) and (2) vs. (3)). However, the decrease in the activity of the PtSnNa/SUZ-4 was much more obvious than that of the PtSnNa/ZSM-5 (curves (1) vs. (2)). Because dibenzo[b, d]thiophene has a molecule diameter (ca. 0.8 nm) larger than the widths of the tunnels in both the SUZ-4 and ZSM-5 zeolite (ca. 0.46-0.56 nm) [23, 31, 32], the dibenzo[b, d]thiophene molecules could only contact the surface Pt, while the Pt particles inside the tunnels remained untouched. Thus, these experiments clearly show that a greater proportion of the Pt was dispersed on the zeolite surface of the PtSnNa/SUZ-4 than on the PtSnNa/ZSM-5. The strong acid sites of the SUZ-4 zeolite evidently prevented the impregnation of the Pt precursor H2PtCl6 into the zeolite. In contrast, the weak acid sites of the ZSM-5 zeolite (Table 3, entries 2 vs. 4) led to more of the precursor entering the zeolite tunnels, followed by transformation to highly dispersed Pt clusters during calcination [33]. Thus, more Pt clusters were dispersed inside the tunnels in the PtSnNa/ZSM-5 than in the PtSnNa/SUZ-4. This is in accordance with the observation that the Vp of the PtSnNa/ZSM-5 declined to a greater extent than that of the PtSnNa/SUZ-4 during the reaction (Table 2, entries 4 vs. 2).
The varying Pt distributions inside and outside the two supports could have generated the different carbon deposition distributions. The TPO profiles of the used catalysts are presented in Fig. 6(a). In the case of the PtSnNa/ZSM-5, two successive peaks representing different carbon deposits on the surface are observed (Fig. 6(a), curve (2)). Partial oxidation experiments (POE) that removed carbon deposits at 520 ℃ were found to reactivate the catalyst, leading to increased propylene selectivity while the propane conversion remained stable (Fig. 6(b), curves (1) vs. (2), (3) vs. (4)). These data indicate that the carbon deposits responsible for the 590 ℃ peak (Fig. 6(a)) did not affect the catalytic activity. The H2 uptake (Table 2, entry 4) of the PtSnNa/ZSM-5 after the POE was close to that of the fresh material, a result that is in good agreement with the approximate propane conversion values in Fig. 6(b). Thus, the peak at 520 ℃ is attributed to carbon deposits that covered the catalytic Pt, while the 590 ℃ peak is due to the carbon deposits located on the zeolite surface. These conclusions agree with reports in the literature [22]. Finally, based on the previous pre-treatment results, the 520 ℃ peak can be ascribed to the coke on the surfaces of Pt nanoparticles located in the zeolite tunnels. In the case of the PtSnNa/SUZ-4 (Fig. 6(a), curve (1)), the TPO profile shows only one type of coke (at 630 ℃), located on the zeolite surface. Carbon deposits that cover the Pt can deactivate the catalyst and so, following the 10-h reaction, both the H2 uptake (Table 2, entry 4) and the activity of the PtSnNa/ZSM-5 were significantly decreased.
Previous studies have indicated that the interactions between the Pt, Sn and support are important factors affecting the state of the Sn component, the dispersion of the Pt and the dehydrogenation performance of the catalyst [34, 35]. In the case of catalysts with high levels of added metal components but low dispersion, it is difficult to obtain useful information about the catalyst active phase structure based on characterizations with XRD and X-ray photoelectron spectroscopy. To further assess the interaction of each component in the PtSnNa catalyst, TPR with H2 was performed. As shown in Fig. 7, a broad but weak reduction peak was observed at 350 ℃ for the Pt/SUZ-4 sample, indicating that the majority of the Pt (Ⅱ) was oxidized to Pt (0) during heating (Fig. 7, curve (1)) [36]. In contrast, the Sn/SUZ-4 generated two reductive peaks. The peak at 300 ℃ is attributed to the reduction of Sn (Ⅳ) to Sn (Ⅱ), while the peak at 520 ℃ clearly indicates the reduction of Sn (Ⅳ) or Sn (Ⅱ) to Sn (0) (Fig. 7, curve (2)). The reduction of the Sn component was evidently much easier following the addition of Pt, as the signals produced by the PtSn/SUZ-4 were moved to lower temperatures (Fig. 7, curves (3) vs. (2)). Compared with the PtSn/SUZ-4, the Sn (Ⅱ) to Sn (0) reductive peak of the PtSnNa/SUZ-4 was increased in intensity and situated at a lower temperature (400 ℃), demonstrating that the addition of Na reduced the interactions between the Sn species and the support, and promoted the reduction of Sn (Fig. 7, curves (4) vs. (3)). The ZSM-5 supported catalysts exhibited similar behavior (Fig. 7, curves (5-8)). However, as shown by the much greater peak area in curve (8), the higher valency Sn on the ZSM-5 was reduced to metallic Sn much more easily, likely because of the weaker acidity of the PtSnNa/ZSM-5, as shown by the NH3-TPD profiles (Fig. 4). The coverage of surface Pt by metallic Sn species in the Pt-Sn alloys may lead to decreased H2 uptake and lower catalytic activity during hydrocarbon dehydrogenation [37, 38].
PtSnNa/SUZ-4 was found to exhibit higher activity and improved stability compared with PtSnNa/ZSM-5 during propane dehydrogenation. Variations in the Pt distribution were likely the main reason for the different performance of the two catalysts. The higher catalytic activity and stability of the PtSnNa/SUZ-4 catalyst can possibly be attributed to a greater quantity of Pt on its external surfaces, and the minimal deactivation of this Pt by carbon deposition that blocked the zeolite pores. In the case of the PtSnNa/ZSM-5, the interactions between Sn oxides and the support were lessened because of the weaker acidity of the ZSM-5 zeolite. The dispersed Sn oxides were therefore easier to reduce to the metallic state, thus decreasing the catalytic activity for hydrocarbon dehydrogenation.
We thank the High Level Talent Support Project of Yangzhou University for financial support. We thank Dr. Lei Fan for suggestions. We thank the analysis centre of Yangzhou University for assistances.
2017,
Vol. 38 Issue (3): 529-536 
PDF
2017,
Vol. 38 Issue (3): 529-536 PDF