Ethylene and propylene, traditionally produced by the petroleum route, are extremely important chemical raw materials. The methanol to olefins (MTO) process is a strategic technology for olefin production from the abundant nonpetroleum sources to balance the shortage of petroleum in China. SAPO-34 is the catalyst of choice for MTO due to its high hydrothermal stability, suitable micropores and moderate acidity ^{[1, 2]}. However, it suffers from the rapid coke formation due to diffusion limitation through the small windows (0.38 nm × 0.38 nm), which would drive the process cost up with frequent catalyst regeneration ^[3]. Constructing SAPO-34 crystal with a hierarchical structure which contains micro-, meso- and macropores is a promising approach to improve mass transport to decrease the regeneration frequency required ^{[4, 5]}.

Commonly, hierarchical pore structures are achieved by either direct synthesis with multiple templates or an indirect post-treatment. Wang et al. ^[4] synthesized SAPO-34 nanosheets with a hierarchical structure using the quaternary ammonium-type organosilane surfactant of [3-(trimethoxysilyl)propyl]octadecyl-dimethylammonium chloride (TPOAC) as the mesopore template and diethylamine (DEA) as the micropore template. Wu et al. ^[6] synthesized SAPO zeolite by adding TPOAC and C_22-4-4Br₂ as the organosilane surfactant for the mesopores, but their methanol conversion and catalyst lifetime were even worse than those of conventional SAPO-34 in the MTO process. Li et al. ^[5] synthesized hierarchical SAPO-34/18 zeolite by a hydrothermal method, but unfortunately only a low methanol conversion below 20% was reported. Hollow SAPO-34 cubes with butterfly-shaped spots on the faces prepared from solid precursor gels containing gelatin by a vapor phase transport method has been reported ^[7]. Wang et al. ^{[4, 8]} synthesized hierarchical SAPO-34 catalysts using TPOAC as the mesopore template by direct hydrothermal crystallization, which exhibited a remarkably prolonged catalyst lifetime (350-500 min). A HF-assisted in situ growth etching route was developed to synthesize hierarchical SAPO-34 by the hydrothermal route ^{[9, 10]}, its single-run lifetime for the MTO reactions was up to 600 min. However, the secondary templates or HF used in the direct synthetic system increase the synthesis cost and environmental pollution. Post-synthetic treatment, which provides an alternative approach to get a hollow-structured micro- or nanosized SAPO-34 has been reported. Qiao et al. ^[11] sy nthesized hierarchical SAPO-34 by a hydrochloric acid and tetraethylammonium hydroxide (TEAOH) treatment, respectively, but did not report MTO performance.

Butterfly-shaped porous patterns on four side faces were formed after a nitric acid or oxalic acid treatment in our work. In our previous work, we also developed a facile TEAOH etching post-treatment method to get SAPO-34 with a hierarchical pore structure and single-run MTO lifetime up to 600 min ^[12]. However, this method was limited by the high price of TEAOH and low recovery rate (25%-42%) of SAPO-34 after the alkali treatment ^[11]. An acid treatment for the controllable preparation of a hierarchical SAPO-34 is a potential route to optimize the diffusion efficiency, suppress coke deposition and prolong catalyst lifetime as well as improve the recovery rate of the modified zeolite in an economic way. However, there have been only limited works on the acid treatment of SAPO-34 for the MTO application reported so far.

In this work, we synthesized hierarchical SAPO-34 with high crystallinity by controlling the acid post-treatment conditions. The effects of the acid types (nitric acid, oxalic acid and butanedioic acid) on the hierarchical SAPO-34 structure and MTO performance were investigated. It was found that butterfly-shaped pore structures were formed after the SAPO-34 was treated by nitric acid and oxalic acid, and a significantly prolonged single-run lifetime for MTO reaction together with improved light olefin selectivity were achieved.

The parent SAPO-34 crystals were prepared by a hydrothermal route. The synthesis gel molar composition was 1Al₂O₃:0.44SiO₂:1.1P₂O₅:2.25triethylamine:35H₂O. SAPO-34 crystal seeds with a mass ratio of 1:500 to the gel were mixed in a closed autoclave. Then the mixture was heated from room temperature to 165 ℃ in 7 h and kept at 165 ℃ for 33 h before cooling. The solid product was filtered, washed and dried, and followed by calcination at 600 ℃ for 5 h to obtain the parent SAPO-34, which was denoted as S-0.

Hierarchical SAPO-34 was prepared by an acid etching treatment. In a typical treatment, 300 mL acid aqueous solution (0.05 mol/L) was added into a three-necked flask and heated to 90 ℃. Then 15 g of parent SAPO-34 was mixed with the acid solution for 6 h under reflux. The treated SAPO-34 was subsequently separated by filtration and carefully washed with deionized water five times. The product was dried and calcined at 600 ℃ for 5 h before use. The samples treated by nitric acid, oxalic acid and butanedioic acid were named as S-1, S-2 and S-3, respectively.

The catalytic performance of the SAPO-34 samples for the MTO reaction was tested in a tubular fixed bed stainless steel reactor with the dimension of 550 mm long and I.D. 12 mm under atmospheric pressure. 3 g catalyst (20-40 mesh) was packed into the center of reactor and a mixture of methanol and water (mass ratio of 1:1) was fed into the reactor by an HPLC pump. The feedstock was first vaporized at 150 ℃ before the reactor. The reaction temperature was 450 ℃ and the weight hourly space velocity (WHSV) of methanol was 1.0 h^-1. The liquid products were collected by a cooling trap. The gaseous products were analyzed by an online gas chromatograph (GC 2014C, Shimadzu) equipped with a flame ionization detector (FID) and a HP-PLOT/Q capillary column (30 m × 0.35 mm). The liquid products were analyzed by an offline GC (GC-950, Haixin) equipped with a thermal conductivity detector (TCD) and a packed column (PORAPAK T, 3 m × 3 mm). The conversion and selectivity were calculated on a carbon balance. The selectivity for the MTO products was expressed as mass percentage of each product in all the detected products except dimethyl ether (DME), which was calculated as unreacted methanol.

The crystal structure of SAPO-34 was measured by an X-ray diffractometer (XRD, Rigku, ultima IV) using Cu K_α (0.154 nm) radiation of 40 mA and 40 kV. The morphology and elemental analysis of SAPO-34 was performed by a scanning electron microscope (SEM, Zeiss SUPRA 55 SAPPHIRE) equipped with an energy-dispersion X-ray spectrometer (EDS, Oxford X-max) with acceleration voltage of 2 and 20 kV, respectively. The textural properties, such as specific surface area (BET, DFT), microporous area (t-plot method), pore volume (BJH, DFT and HK) and pore size distribution (BJH) of the samples were derived from N₂ adsorption measurements carried out at -196 ℃ using an automatic physisorption analyzer (Tristar 3020, Micromeritics) after the sample was degassed at 300 ℃ for at least 10 h under 0.133 Pa pressure prior to each run. The macroporous size was measured by mercury intrusion porosimetry (MIP, Micromeritics Autopore V 9620). Fourier transform infrared spectroscopy (FT-IR, Thermo Scientific, Nicolet 6700) was applied to detect the functional groups of SAPO-34. The composition of samples was analyzed by X-ray fluorescence (XRF, Bruker S4PIONEER). The tablets were prepared by mixing boric acid and SAPO-34 with the weight ratio of 2:1.

The acidity of the samples was determined using temperature programmed desorption of ammonia (NH₃-TPD, Tianjin Xianquan TP-5080) automated with a TCD analysis unit under nitrogen flow. Thermal degradation measurements were performed from room temperature to 1000 ℃ with a heating rate of 10 ℃/min in air using a thermogravimetric analyzer (TGA, NETZSCH, STA 449F3). Solid state MAS NMR experiments were performed at 5 kHz on a Bruker AVANCE 400 spectrometer operating at frequencies of 79.5, 104.22 and 161.9 MHz for ²⁹Si, ²⁷Al and ³¹P, respectively. Chemical shifts were referenced to tetramethylsilane (TMS) for ²⁹Si, 1 mol/L of Al(NO₃)₃ for ²⁷Al and 85% H₃PO₄ for ³¹P. The coke composition was analyzed by an Agilent 7890 gas chromatograph coupled to an Agilent 5975C mass spectrometer (GC-MS). Before GC-MS measurement, 0.3 g deactivated SAPO-34 zeolite was treated with 2 mL of HF (40%) followed by extraction with 4 mL methylene chloride to get the organic composition and remove HF.

The XRD patterns of SAPO-34 before and after acid etching are shown in Fig. 1. All the samples showed the typical diffraction peaks of the CHA structure at 9.6°, 20.7°, 26° and 31°. No impurity crystal or amorphous phase was found, which indicated that the pure SAPO-34 crystals were well preserved after the acid etching. After the acid treatment, the intensity of the diffraction peaks was reduced, suggesting poorer relative crystallinity. Compared with S-0 (taken as 100%), the relative crystallinity of the three samples after acid etching was 25%, 27% and 22% for S-1, S-2 and S-3, respectively. This was reasonable since the acid treatment dissolved part of the crystal and generated defects in the zeolite framework. The recovery rate after each etching was 70%, 72% and 66% after nitric acid, oxalic acid and butanedioic acid etching, respectively, which was about 30% higher compared with those from the alkali treatment ^[12].

Fig. 2 presents the SEM images of SAPO-34 before or after the acid treatment. Typical cubic crystals with particle sizes from 1 to 2 μm can be observed. In contrast to the parent SAPO-34 crystals with smooth faces, the faces of nitric acid and oxalic acid treated SAPO-34 presented butterfly-shaped porous patterns on the four side faces. However, no pore pattern was observed on the butanedioic acid etched sample. Moreover, the etching of nitric acid led to more obvious pores than that by oxalic acid, indicating that the generation of the pore structure was related to the kind of acid used. Although the exact formation mechanism of SAPO-34 is not known, a reasonable conjecture is that the crystal skeleton is first formed by eight pyramidal sub-crystals followed by the subsequent growth to a cubic crystal through the filling of the voids around the center of the crystal. Therefore, the subsequently formed parts of SAPO-34 were preferentially etched to give the hierarchical structure which was probably due to their poorer stability (Fig. 2, S-1, S-2) ^{[6, 12]}. The composition of the SAPO-34 obtained was determined by XRF and shown in Table 1. After the acid treatment, no obvious change in the composition was found, which is a little different from the report that selective dealumination from the framework occurred ^[11].

The pore structure of the SAPO-34 samples was determined by N₂ adsorption (Fig. 3). All samples showed Type I isotherms, which was due to the filling of micropores ^[13]. Interestingly, S-2, especially S-1 exhibited a slight increase in the region 0.99 < p/p₀ < 1.0 with an obvious hysteresis loop, indicating the formation of mesopores and macropores after the etching by nitric acid or oxalic acid, which is consistent with the SEM results ^{[14, 15]}. Both the Horvath-Kawazoe (HK) distributions of S-1 and S-2 showed one sharp peak near 0.5 nm similar to that of S-0 (not shown here), confirming that no obvious change of micropores occurred after the treatment. Meanwhile, the BJH results further revealed that mesopores with the average pore size of 12.5 and 46 nm were formed in S-1 and S-2, respectively. S-1 showed a slight higher micropore surface area and micropore volume (671 m²/g and 0.26 cm³/g) than S-0 (665 m²/g and 0.27 cm³/g) due to the increase of mesopore or macropore surface area and pore volume after the treatment with nitric acid. However, the BET specific surface area and microporous volume of the oxalic acid treated SAPO-34 increased sharply to 876 m²/g and 0.32 cm³/g. In contrast, no obvious mesopores and macropores were formed in the butanedioic acid treated sample, and only a Type I isotherm was observed. Meanwhile, the lower adsorption amount of S-3 also indicated a lower surface area and pore volume, as shown in Table 2. The mercury intrusion porosimetry (MIP) curves in Fig. 4 showed one wide peak at 400 nm, which was attributed to the void spaces between the cubic particles ^[12]. Compared with the parent SAPO-34, additional peaks at 62 and 434 nm appeared on S-1 and S-2, showing that macropores were formed in the acid treated SAPO-34, which was further confirmed from its shift to higher values when the etching time was prolonged (not shown here). Therefore, in combination with the SEM results, the pore size distribution analysis proved clearly that a hierarchical pore structure consisting of micro-, meso- and macropores were successfully formed by the nitric acid and oxalic acid etching of parent SAPO-34.

Fig. 5 gives the NH₃-TPD curves of SAPO-34 before and after the acid treatment. All samples showed two peaks at 190 and 395 ℃, ascribed to NH₃ desorption from weak and strong acid sites, respectively ^{[16, 17]}. Compared with S-0, the acid treated SAPO-34 exhibited peaks at almost the same positions for both the weak and strong acid sites, suggesting that the acid strength was kept almost unchanged after the acid treatment. The peak area for the weak acid sites increased slightly after the acid etching, which may improve the catalytic performance of SAPO-34 for the MTO reactions ^{[18, 19]}. The peak area of the strong acid sites for S-1 and S-2 was somewhat decreased, but S-3 exhibited a little larger amount of strong acid sites. The former may increase while the later may shorten the single-run MTO lifetime ^[18]. The FT-IR spectra in Fig. 6 indicated similar chemical groups for all samples. In detail, the bands at 1082 and 724 cm^-1 belong to the SAPO framework stretching vibration. The peaks at 640 and 490 cm^-1 can be assigned to the vibration of double 6 rings and the bending of T-O. The band at 1220 cm^-1 stems from the asymmetric stretch of T-O tetrahedral groups ^[7]. The FT-IR results allowed us to infer that the acid treatment has no obvious effect on the chemical bonds of the crystals.

The chemical environment in SAPO-34 before and after the acid etching was investigated by solid state ²⁹Si, ²⁷Al and ³¹P MAS NMR (Fig. 7). In the ²⁹Si MAS NMR spectra, all the samples showed a narrow peak centered at -89 assigned to the Si(4Al) species ^{[20, 21]}. The peak area of hierarchical SAPO-34 (S-2) was somewhat larger than that of the parent SAPO-34, indicating a slight increase of the amount of weak acid sites ^{[10, 11]}, which agreed well with the NH₃-TPD results. Three other peaks at -82, -100 and -105 were also observed with the hierarchical SAPO-34. The signal at -82 was assigned to Si(OAl)_n(OH)_4-n (n = 1 or 2) species from the breaking of Si-O-Al bonds ^{[22, 23]}. The peaks at -100 and -105 belong to Si(3Al) and Si(2Al), respectively. The relatively higher amount of these two species suggested a somewhat increased amount of Si after the acid treatment, which agreed well with the XRF results. In the ²⁷Al MAS NMR spectra, signals at 42, 15 and -12 for all samples were detected, which were assigned to Al atoms in a tetrahedral environment ^{[11, 24]}, penta-coordinated Al atoms and octahedrally coordinated Al atoms formed by another two water molecules, respectively ^{[9, 12, 22]}. However, no obvious change in the Al chemical environment was detected before and after the acid treatment. The signals between -26 to -30 in the ³¹P MAS NMR spectra were attributed to P(OAl)₄ species and the broad shoulder peak at -15 stemmed from P atoms coordinated with water (P(OAl)_x(H₂O)_y, x + y = 4) ^{[12, 22, 25]}. The decrease of the ³¹P MAS NMR intensity of S-2 may come from the dephosphorization effect after acid etching ^[26].

The MTO performance of the SAPO-34 before and after the acid treatment was tested at 450 ℃ under atmospheric pressure. Methanol conversion for all samples are shown in Fig. 8. The single-run lifetime (defined as giving methanol conversion >95%) over the parent SAPO-34 was only 210 min, while it was significantly prolonged to 360 and 390 min after the nitric acid and oxalic acid etching. It was reported that the micropores of the parent SAPO-34 were easily blocked by coke species, which will obstruct the access of reagents into the catalyst ^{[27, 28]}. In contrast, the abundant mesopores and macropores in hierarchical SAPO-34 (S-1 and S-2) can greatly promote the diffusion of reagents/products into and out of the crystal, which suppressed coke formation and prolong the lifetime. Moreover, the increase of weak acid sites together with the decrease of strong acid sites in hierarchical SAPO-34 probably also contributed to the longer lifetime. However, only a short lifetime of 100 min was observed on SAPO-34 after it was treated with butanedioic acid. The destruction of some micropores and the absence of mesopores in S-3 would be the main reason for its poor MTO performance. This was supported by the broken crystals in Fig. 2 and its decreased BET surface area. At the same time, the increased amount of strong acid sites accelerated the formation of coke species, which would contribute to rapid deactivation.

The distribution of the products is shown in Fig. 9. The C₂⁼-C₄⁼ selectivity from the parent SAPO-34 was around 90% during the single-run process, and contained 37.4% of ethylene, 38.8% of propylene and 14.0% of butene. Although the distribution of C₂⁼-C₄⁼ over hierarchical S-1 was similar to that of the parent SAPO-34, the total selectivity for light olefins increased to 94%. However, the distribution of the products from S-2 was quite different from those of S-0 and S-1. The selectivity to ethylene increased to 51.5% and the selectivity to propylene and butene decreased to 31.7% and 8.1%, respectively. As a result, the total C₂⁼-C₄⁼ selectivity was 92%. This suggested that the distribution of C₂⁼-C₄⁼ was controlled by the channel structure of the hierarchical SAPO-34. For the S-3 sample, the distribution of light olefins was not changed compared with that of S-0, but the lifetime of the catalyst was decreased dramatically. The MTO results showed that a hierarchical SAPO-34 could not only prolong the lifetime of the catalyst, but also optimize the distribution of the products, which provides the possibility to increase or decrease the ethylene ratio according to the market demand.

TGA measurements of the spent catalysts under flowing air are shown in Fig. 10. These curves exhibited two obvious weight loss steps. The weight loss before 300 ℃ was attributed to adsorbed water. The weight loss between 500 and 700 ℃ was associated with the decomposition of coke. It was found that the weight loss of water was the same for these two samples, but the amount of deposited coke was 15% and 18% for the parent and hierarchical SAPO-34, respectively, suggesting that the coke capacity of SAPO-34 was increased after the oxalic acid etching, which also contributed to its longer lifetime ^[29-31].The average coke formation rate and the amount of coke deposition per methanol unit were calculated and shown in Table 3. This revealed that the coke formation rate for hierarchical SAPO-34 was much lower than that of the parent one (0.046 vs. 0.071 g/min). At the same time, the amount of coke deposition per unit of methanol decreased from 0.059 to 0.038 g/g_MeOH, indicating that coke formation was suppressed over hierarchical SAPO-34 due to the synergistic effect of the micro-, meso- and macroporous channels and suitable acid sites.

In order to further investigate the effect of the hierarchical pores on coke deposition, the composition of the coke in spent S-0 and S-2 was analyzed by GC-MS (Fig. 11). For the spent S-2, the coke was mainly composed of polycyclic aromatics, such as phenanthrene and pyrene, and the weight percent of these was 73%. However, the polycyclic aromatics deposited on the spent parent SAPO-34 was decreased remarkably to 49% while the lighter coke species like benzene and polymethyl benzene were increased. The higher amount of polycyclic aromatics in S-2 may come from its hierarchical structure, due to improved diffusion of the products, thus benzene or polymethal benzene diffuse easily out of the pores while larger molecules such as polycyclic aromatics remained in the pores.

Hierarchical SAPO-34 was prepared by hydrothermal synthesis followed by a nitric acid or oxalic acid post-treatment. Their catalytic performance of hierarchical SAPO-34 was investigated in MTO reaction. The SAPO-34 treated with nitric acid or oxalic acid exhibited butterfly-shaped pores on four side faces and the pores comprised micropores, mesopores (40-50 nm) and macropores (62-500 nm). The number of acid sites was also optimized after the nitric acid and oxalic acid treatment. The hierarchical SAPO-34 treated with nitric acid or oxalic acid show 100% methanol conversion and higher selectivity for light olefins (92%-94%). The synergistic effect of the hierarchical pores and acid sites helped the single-run lifetime of hierarchical SAPO-34 to increase sharply from 210 to 360 and 390 min, and ethylene selectivity changed to be between 37.4% and 51.5%. The hierarchical structure also increased the coke capacity and the ratio of smaller coke molecules, which also contributed to the longer lifetime. However, we did not get hierarchical SAPO-34 by treating SAPO-34 with butanedioic acid, and its lifetime was only 100 min. This work demonstrated a simple and efficient route to prepare hierarchical SAPO-34 with appropriate acid sites, which may be extended to make other hierarchical zeolites.

Thanks to Lu'an Mining Group (Changzhi, China) for supporting our research and technical assistance.