ZSM-5 zeolites with medium-sized pores (0.54 nm × 0.56 nm) and three-dimensional channels were first synthesized by Mobil in 1972. As a type of high-silica zeolite, ZSM-5 is generally prepared from reaction mixtures containing a certain amount of an organic template, which is used as a structure-directing and charge-balancing agent. However, the zeolite morphology can be influenced by many parameters, such as the silica source ^{[1, 2, 3]}, alumina source, template ^{[4, 5, 6]}, and crystallization conditions ^[7]. Among these parameters, the template is the key factor to control the morphology because different sizes and spatial configurations of the templates can lead to different interactions between the organic templates and zeolite frameworks. Liu et al. ^[8] reported that spheroidal ZSM-5 zeolites could be synthesized using n-butylamine as the template, while after the addition of triethanolamine cuboidal crystals are obtained. Sang et al. ^[9] concluded that ZSM-5 zeolites with cuboidal, lath-shaped, and boat-like morphologies could be prepared using ethylamine, ethylenediamine, and ethanolamine organic templates, respectively. MFI nanosheets were synthesized with the assistance of diquaternary ammonium-type surfactants (C₂₂H₄₅-N⁺(CH₃)₂-C₆H₁₂-N⁺(CH₃)₂- C₆H₁₃) by Choi et al ^[10].

ZSM-5 zeolites with different morphologies (e.g., lath- shaped, spheroidal, hexagonal, and ellipsoidal morphologies) and channel lengths always have different residence times of the reaction intermediates, which affect the catalytic performance and the product distribution ^{[11, 12]}. Wang et al. ^[13] reported that the morphology of ZSM-5 zeolite had a great influence on the product distribution and lifetime of the catalyst in methanol to gasoline reaction. Small-sized ZSM-5 exhibited higher selectivity for C₅₊ and aromatic hydrocarbon formation and longer lifetime than large-sized ZSM-5. Zhang et al. ^[14] investigated the influence of the ZSM-5 zeolite crystal size on the methanol to olefin (MTO) reaction, and found that the stability of the small-crystal catalyst was better than that of the large-crystal catalyst. Bhat et al. ^[15] reported that the morphology of zeolites influenced the selectivity of the toluene disproportionation reaction. Small spheroidal crystal particles were favorable for the formation of m-xylene, while p-xylene was mainly formed over large cylindrical crystals.

In the present study, ZSM-5 zeolite with an unusual snowflake-shaped morphology was successfully synthesized for the first time, and compared with ellipsoidal and boat-like-shaped samples. Various characterizations were carried out to investigate the structural framework, pores, and acidity. The MTO reaction was chosen as a probe reaction to investigate the effect of the morphology and crystal structure on the catalytic performance.

ZSM-5 zeolite with an unusual snowflake-shaped morphology was synthesized following a previous procedure ^[16]. Specific amounts of colloidal silica, Al₂(SO₄)₃·18H₂O, 1,6- diaminehexane (structure-directing agent (SDA)), KOH, and deionized water were mixed in order. After stirring at room temperature for at least 1 h, the resulting homogeneous gel with molar composition 1.0SiO₂:0.26K₂O:0.02Al₂O₃:0.3SDA:40H₂O was transferred into a stainless-steel autoclave with a capacity of 2 L and subjected to hydrothermal treatment at 180 °C for 48 h in a furnace with autogenous pressure under stirring. The obtained zeolite is hereafter denoted as A.

The gel for the conventional ZSM-5 zeolite synthesis was prepared by slowly adding a mixture of Al₂(SO₄)₃, H₂SO₄, and the template agent (ethylamine or ethylenediamine, SDA) to an aqueous solution of Na₂SiO₄ (30 wt% silica) under stirring. The molar composition of the gel was 1.0SiO₂:0.02Al₂O₃:0.74SDA: 40H₂O. Finally, the gel was stirred for at least 1 h at room temperature to obtain a homogenous gel, and then transferred to a 2-L stainless-steel autoclave, which was maintained at 180 °C for 48 h. The obtained samples are hereafter denoted as B (ethylamine templating agent) and C (ethylenediamine templating agent).

After crystallization, the products were washed with deionized water to neutral and dried overnight at 120 °C for 12 h. The organic template was removed by calcination at 550 °C for 10 h. The zeolites were then converted to the protonic form by ion exchange with NH₄NO₃ solution (0.5 mol/L) at 80 °C for 8 h, followed by calcination at 550 °C in air for 8 h. This process was performed twice. The obtained products are denoted as HA, HB, and HC, for samples A, B, and C, respectively.

Dilute nitric acid was added to a mixture of the molecular sieve and Al₂O₃ (mass ratio 4:1), which was extruded to the carriers. The catalysts with 2.0 wt% Ca-loaded zeolite carriers were prepared by the impregnation method with Ca(NO₃)₂ solutions for the metal precursor. The catalysts were dried at 120 °C and then calcined at 550 °C for 4 h. The obtained catalysts are hereafter denoted as Ca-HA, Ca-HB, and Ca-HC.

The morphology and crystal size of ZSM-5 were examined with a Quanta 400F scanning electron microscope (FEI, Eindhoven, Netherlands). X-ray diffraction (XRD) patterns were obtained using a Bruker D8 Advance diffractometer (Germany) with Cu K_α (λ = 1.5418 Å, 40 kV, 40 mA) over a 2θ range of 5°-50°. X-ray fluorescence (XRF) spectroscopy experiments were performed to analyze the SiO₂/Al₂O₃ ratio with a Rigaku ZSX Primus ІІ (Japan). The Brunauer-Emmett-Teller (BET) surface area (A_BET) and pore volume of the ZSM-5 samples were determined by N₂ adsorption-desorption measurements using the BET method and t-plot method on a Micromeritics ASAP 2020 (USA). Prior to the measurements, all of the samples were degassed at 350 °C for 8 h. ²⁹Si and ²⁷Al magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance ІІІ 600 spectrometer (Germany) with frequencies of 119.2 and 156.4 MHz, respectively. The ²⁹Si MAS NMR spectra were acquired using a 7-mm probe with a rotor spin rate of 5 kHz. The spectral data were fitted by Dmfit software (http://nmr.cemhti.cnrs-orleans.fr/dmfit). The ²⁷Al MAS NMR spectra were acquired using a 4-mm probe with a rotor spin rate of 13 kHz. The acidity was determined by temperature-programmed desorption of ammonia (NH₃-TPD) and Fourier transform infrared (FT-IR) spectroscopy of pyridine adsorption (Py-IR). The NH₃-TPD experiments were carried out on a Micromeritics ASAP 2920 (USA) with mass spectra detection. The catalysts (about 200 mg) were pretreated at 550 °C under He flow (50 mL/min) for 1 h, cooled to 100 °C to adsorb NH₃, and then purged with He for 1 h. Finally, the adsorbed samples were heated to 600 °C at 10 °C/min under He flow (50 mL/min). The Py-IR spectra were recorded with a Bruker Vertex 70 FT-IR spectrometer (Germany). The catalyst samples were ground into a fine powder and pressed into self- supported discs. The sample discs were heated to 400 °C at 10 °C/min under vacuum. Then, He flow with pyridine was introduced into the IR cell at room temperature until adsorption saturation. Spectra were recorded at 200 and 350 °C.

The MTO catalytic performance of the different catalysts was tested in a continuous flow fixed-bed reactor at atmospheric pressure. For each test, the zeolite catalyst (5 mL) was loaded in the constant temperature zone of the reactor. The reaction conditions were as follows: methanol with liquid hourly space velocity (LHSV) = 2.0 h^-1, n(N₂)/n(CH₃OH) = 2:1, and T = 430-470 °C. For the reaction test of each catalyst sample, five temperature points were sampled with an interval of 3 h between each sampling period, during which the temperature was kept at a constant value. During the sampling period (3 h), the reactor outlet stream was cooled to 2 °C and then the gas and liquid products were separated. The gas products were analyzed by two on-line gas chromatographs (Agilent 6890, USA). One was equipped with an HP-MoleSieve5A (102 m × 0.50 mm) column and a thermal conductivity detector (TCD) for hydrogen and methane, while the other was equipped with an HP-plot Al₂O₃ (50 m × 0.53 mm) column and a flame ionization detector (FID) for alkanes and light olefins. The results were calibrated by the methane concentration. The aqueous and organic phases in the liquid products of the reactor were separated by a decanter. A small portion of the aqueous phase was sent to a gas chromatograph (Agilent 7890A) equipped with an FID and an AB-inowax column for separating water, methanol, and oxygenates. The organic phase was also analyzed by a gas chromatograph (Agilent 7890A) equipped with an FID and an HP-PONA column (50 m × 0.20 mm) for separating hydrocarbons. After the sampling period, the temperature was increased to a higher value for the next test until the maximum conversion was reached, which led to the reaction tests lasting for about 20 h for each catalyst sample.

The reaction performance was assessed by methanol conversion and olefin product selectivity according to the following formulas:

ZSM-5 zeolites with different morphologies were synthesized using various templates. The molar compositions of the gels and the SiO₂/Al₂O₃ ratios of the products are shown in Table 1. The SiO₂/Al₂O₃ ratios of the three samples (34-39) are similar.

Table 1 Molar compositions of the gels and the SiO2/Al2O3 ratios of the different ZSM-5 zeolites a. Sample n(SiO2):n(Al2O3):n(SDA):n(H2O):n(M2O) b n(SiO2)/n(Al2O3) HA 1.0:0.02:0.3:40:0.26 34 HB 1.0:0.02:0.74:40:0.3 37 HC 1.0:0.02:0.74:40:0.3 39 a Crystallization at 180 °C for 48 h. b n(M2O) represents n(K2O) for the HA sample and n(Na2O) for the HB and HC samples.

Table 1 Molar compositions of the gels and the SiO2/Al2O3 ratios of the different ZSM-5 zeolites a.

SEM images of the HA, HB, and HC samples are shown in Fig. 1. The morphologies and particle sizes of the three samples are clearly different. The HA sample has an unusual snowflake-shaped morphology with a crystal size of about 7 μm. The HB and HC samples have conventional ellipsoidal and boat-like morphologies with crystal sizes of 5 and 6 μm, respectively.

Fig. 1. SEM images of the ZSM-5 samples with different morphologies.

The micropore size distributions are shown in Fig. 2. The micropore sizes of the different samples have a maximum at about 5.0 Å, which is typical for 10-membered ring pores, demonstrating the presence of the open microporous structure in the samples. Clear differences between the micropore sizes of the samples cannot be observed because of the limitation of this characterization. The BET surface areas and micropore volumes of the different ZSM-5 zeolites are listed in Table 2. The BET surface areas, micropore areas, and micropore volumes of the three HZSM-5 zeolites are similar. However, the BET surface area and micropore area clearly decrease after Ca modification, suggesting that Ca diffuses into the channels and blocks the micropores of the zeolites. Furthermore, because of the different crystal structures and orientations of the crystal faces, the micropore areas of the Ca-HA and Ca-HB catalysts exhibit much larger decreases than the Ca-HC catalyst after Ca loading.

Fig. 2. Micropore size distributions of the different ZSM-5 zeolites and Ca-loaded ZSM-5 zeolites.

Table 2 BET specific surface areas and micropore volumes of the different ZSM-5 zeolites and Ca-loaded ZSM-5 zeolites. Sample BET surface area (m2/g) Micropore area (m2/g) External surface area (m2/g) Micropore volume (cm3/g) HA 365.91 246.21 119.71 0.11 HB 389.40 240.05 129.35 0.11 HC 372.19 209.32 162.87 0.10 Ca-HA 297.40 211.42 85.98 0.10 Ca-HB 303.35 210.35 93.00 0.10 Ca-HC 286.70 195.30 91.40 0.09

Table 2 BET specific surface areas and micropore volumes of the different ZSM-5 zeolites and Ca-loaded ZSM-5 zeolites.

Fig. 3 shows XRD patterns of the HA, HB, and HC samples. All of the samples exhibit a typical characteristic pattern of the MFI-type structure (2θ = 7.91°, 8.76°, 23.11°, and 23.84°) ^[17], while some differences in the intensity of the strongest and second strongest peaks of the different samples can be observed. The ratios of peak intensities at 7.91° and 8.76° for the HA and HC zeolites are significantly higher than that of the HB zeolite. The diffraction peak at 7.91° represents the (101) face of the MFI-type structure, while the diffraction peak at 8.76° can be assigned to the (020) face of the MFI-type structure and corresponds to the straight channel of ZSM-5 zeolites ^{[18, 19]}. The different intensities of the diffraction peaks suggest that the crystal orientations of the ZSM-5 zeolites are different, further indicating the different directions of the straight channels. For the HB sample, the strongest diffraction peak at 8.76° reveals that more of the (020) face is exposed than the (101) face. However, because of twin crystal intergrowth, part of the (020) face of the HC sample is covered, so the intensity of the 8.76° peak is weaker than that of the other samples, as shown in Fig. 3. The XRD pattern of the HA zeolite is similar to that of the HC zeolite, while the relative crystallinity of the former is lower than that of the latter (Table 3), suggesting that the (101) face of the HA sample is mainly exposed.

Fig. 3. XRD patterns of the HA, HB, and HC zeolites.

Table 3 Lattice parameters and relative crystallinity (RC) of the HA, HB, and HC zeolites. Sample a (Å) b (Å) c (Å) V (Å3) RC (%) AlFW a (%) AlEFW b (%) HA 20.1419 20.1413 13.4532 5457.7487 94.5 85 15 HB 20.1037 19.9589 13.4178 5383.8619 98.8 91 9 HC 20.0920 20.1367 13.4410 5438.0482 100.0 92 8 a The relative concentration of tetra-coordinated framework aluminum, which is determined by integrating the NMR signal between the chemical shifts at 30 and 80 ppm. b The relative concentration of hexa-coordinated extra-framework aluminum, which is determined by integrating the NMR signal between −20 and 20 ppm.

Table 3 Lattice parameters and relative crystallinity (RC) of the HA, HB, and HC zeolites.

According to the XRD patterns, the HB zeolite has more of the (020) face, while the (020) face of the HC sample is covered because of twin crystal interlaced growth, as shown in Fig. 4 ^[20]. Because of the relatively complex crystal orientation of the snowflake-shaped HA sample, it still needs to be further studied and characterized.

Fig. 4. Schematic pore structures of the HA, HB, and HC zeolites.

Using the least squares method, the zeolite unit cell parameters and unit cell volume calculated from the XRD peaks in the range of 7°-30° are listed in Table 3. The unit cell parameters and volume decrease in the order HA > HC > HB. This suggests that the micropore skeleton structure of the snowflake-shaped HA zeolite is different from the other two samples.

MAS NMR is a powerful tool to analyze atom coordination in zeolite frameworks. The ²⁷Al MAS NMR spectra of all of the samples show two signals (Fig. 5). In general, chemical shifts at 54 and 0 ppm are assigned to tetra-coordinated framework aluminum and hexa-coordinated extra-framework aluminum ^[21], respectively, indicating that most of the Al atoms are in the framework of the zeolite structure. The peak intensities of the HA sample are lower than that of HB and HC samples because of chemical vibration of the coordination environment of Al atoms (e.g., weak dipole interactions). This observation indicates the presence of distortion, dislocation, and asymmetry in the cross-crystal boundary of the HA zeolite ^[22]. The relative distributions of these two Al species obtained by deconvolution of the ²⁷Al MAS NMR spectra are given in Table 3. Compared with the HA sample, the relative concentration of tetra-coordinated Al signals in the spectra of the HB and HC samples are more intense (Fig. 5), which is consistent with the results of the relative crystallinity of the three samples in Table 3.

Fig. 5. 27Al MAS NMR spectra of the HA, HB, and HC zeolites.

Fig. 6 shows the ²⁹Si MAS NMR spectra of the HA, HB, and HC samples. In these cases, the resonance at around -118.0 ppm is assigned to Si(4Si) sites, while the resonance at -115.1 ppm originates from Si(4Si) sites with 1H (OH group) as the next-nearest neighbor ^[23]. The bands around -111.0 and -108.0 ppm correspond to the Si(3Si) sites ^[24]. After deconvolution, the signal of HA at -115.1 ppm is significantly wider than the signals of the conventional samples, further confirming the presence of distortion, dislocation, and asymmetry in the cross-crystal boundary of the HA zeolites ^{[25, 26]}. Consequently, the ²⁷Al and ²⁹Si MAS NMR spectra are consistent with the XRD results.

Fig. 6. 29Si MAS NMR spectra of HA, HB, and HC zeolites. The black curve represents the peak of the original data; the red curve represents the peak sum of all the fitted curves.

NH₃-TPD profiles of the HZSM-5 zeolites and Ca/HZSM-5 catalysts are shown in Fig. 7. The resulting peaks are classified as two types of acid sites with different acid strengths, corresponding to the peak maximum at the desorption temperature. The temperature ranges are 400-450 °C for strong acid sites, and 200-230 °C for weak acid sites. The total number of acid sites decreases in the order HA > HC > HB. The larger number of acid sites for HA is probably because of the presence of distortion, dislocation, and asymmetry in the framework of the HA zeolite, as confirmed by XRD and MAS NMR. After modification with Ca, the total acidity of all of the catalysts, especially the strong acidity, significantly decreases, and some differences can be observed. The total number of acid sites of HA markedly decreases with Ca loading. However, the total acidity of HC only slightly decreases with Ca loading. This result may be because the introduced metal not only interacts with acid sites on the surface but also with acid sites located in the micropores. For the snowflake-shaped Ca-HA and ellipsoidal Ca-HB catalysts, the Ca²⁺ ions favorably diffuse into the micropores and interact with acid sites. However, diffusion of Ca²⁺ ions is limited by the zigzag channels of the HC zeolite, leading to Ca mainly interacting with acid sites at the surface. The variation of the surface areas of the different samples before and after Ca loading can give indirect evidence of the Ca spatial distribution. The micropore areas and external surface areas of the three samples after loading Ca exhibit different trends. The micropore areas of HA and HB exhibit a much sharper decrease than that of HC after loading with Ca. However, the external surface areas of HA and HB exhibit a smaller decrease than that of HC because of the larger number of zigzag channels with high diffusion limitation in the HC zeolite than in the HA and HB samples.

Fig. 7. NH3-TPD profiles of the HA, HB, and HC zeolites, and Ca-HA, Ca-HB, and Ca-HC zeolite catalysts.

The acid types of the different zeolites were investigated by Py-IR. The density and distribution of the acid sites were estimated according to the areas of the corresponding IR bands. The results are summarized in Table 4. The amount of B acid sites is in the order HC ≈ HB > HA, which is not consistent with the NH₃-TPD results. This may be because of the effects of the geometric structure of the snowflake-shaped HA zeolite and the distortion and dislocation of the grain boundaries on the pore structure in the framework. This could suppress diffusion of pyridine molecules, which are larger than NH₃ molecules, and thus lead to decreased adsorption of pyridine molecules on the acid sites. After modification with Ca, the total amount of acid sites markedly decreases. The amount of B acid sites decreases, while the amount of L acid sites exhibits the opposite trend, probably owing to the B acid sites of HZSM-5 interacting with Ca²⁺ to form L acid sites (Ca²⁺ + OH^- → Ca(OH)⁺) ^{[27, 28]}.

Table 4 Acid amounts of the different ZSM-5 zeolites and Ca-loaded ZSM-5 zeolites. Sample Acidity (μmol Py/g sample) Weak acid sites Strong acid sites B L B L HA 208 103 163 44 HB 267 116 266 90 HC 284 121 260 72 Ca-HA 23 159 13 60 Ca-HB 33 242 16 156 Ca-HC 37 218 16 60

Table 4 Acid amounts of the different ZSM-5 zeolites and Ca-loaded ZSM-5 zeolites.

According to the above characterizations of the different samples, it can be inferred that these catalysts with similar surface areas and SiO₂/Al₂O₃ ratios will exhibit different catalytic performance because of the presence of distortion, dislocation, and asymmetry in the framework of the HA sample, which has a completely different morphology than the other zeolites. The MTO reaction was chosen as a probe reaction to further investigate the effect of zeolite morphology on catalytic performance.

To rationalize the observed data, a brief review of the mechanistic aspects of the MTO reaction is required. First, methanol is dehydrated to form dimethyl ether (DME) at the L acid sites. Second, methanol and DME react through the hydrocarbon pool mechanism to form the carbenium ions (C_x⁺), which can then be converted to C₂-C₄ olefins. Finally, C₂-C₄ olefins react with each other to form n-alkanes and aromatics at B acid sites ^{[29, 30]}.

The catalytic performance of the Ca/HZSM-5 catalysts for the MTO reaction is shown in Fig. 8. The selectivity of total olefin and propene formation over the different catalysts first increases and then decreases with increasing reaction temperature. The product distribution over the different catalysts at about 100% conversion is shown in Table 5.

Fig. 8. Variation of the selectivity of C2-C4 olefin formation over different Ca/ZSM-5 catalysts. : ethylene; : propene; : butylene.

Table 5 Variation of the selectivity of products over different Ca/ZSM-5 catalysts at about 100% methanol conversion. Catalyst XMeOH/% Selectivity (%) Alkanes Aromatics Total = a / Ca-HA 99.78 10.67 39.22 22.52 22.89 4.70 72.41 3.68 Ca-HB 99.24 8.10 36.49 19.51 32.25 3.66 64.09 4.50 Ca-HC 98.13 9.43 32.96 16.34 35.25 6.02 58.73 3.49 a The sum of ethylene, propene, and butylene.

Table 5 Variation of the selectivity of products over different Ca/ZSM-5 catalysts at about 100% methanol conversion.

The snowflake-shaped zeolite catalyst (Ca-HA) exhibits the highest selectivity for olefin (72%) and propene (39%) among all of the catalysts. According to the MTO reaction mechanism, the presence of B acid sites always induces cyclization and dehydrogenation of light olefins. Thus, the ellipsoidal Ca-HB catalyst with more B acid sites exhibits lower selectivity for light olefin formation than the Ca-HA catalyst. Moreover, compared with the HB zeolite, the channel length of the snowflake-shaped HA zeolite is much shorter, leading to lower diffusion limitation and shorter residence time, which further suppresses hydrogenation and/or aromatization of propene.

Among the three catalysts, the Ca-HC catalyst exhibits the lowest selectivity for light olefin formation because the HC zeolite has the most zigzag channels, which suppress diffusion and increase the residence time of the products, leading to the further conversion of light olefins at B acid sites.

The ZSM-5 zeolite with an unusual snowflake-shaped morphology was hydrothermally synthesized and compared with conventional ellipsoidal and boat-like-shaped ZSM-5 zeolite samples. Because of the presence of distortion, dislocation, and asymmetry in the framework, the snowflake-shaped ZSM-5 zeolite possesses a higher amount of acid sites and lower diffusion limitation than the conventional zeolite samples. Catalysts for the MTO reaction were prepared by loading Ca on the three samples. The snowflake-shaped Ca/ZSM-5 zeolite catalyst exhibits excellent selectivity for light olefin formation (72%) and good selectivity for propene formation (39%) in the MTO reaction. Our results demonstrate how the macroscopic zeolite morphology affects the microscopic structure and catalytic performance.

The authors are grateful to Synfuels China Co. Ltd. for the financial and equipments support.