Light olefins are important as base raw materials for the petrochemical industry [13]. Of particular interest are ethene and propene, the demand for which is growing every year. Light olefins are generally produced by thermal cracking (pyrolysis) of hydrocarbons. The reported yields of ethene and propene from this process have never exceeded 25 and 13 wt %, respectively. Therefore, this technique can barely satisfy the growing demand for propene. Moreover, thermal cracking has a number of major technological and environmental challenges, such as high energy consumption, costly equipment, heavy environmental pollution, and a limited adjustment range for light olefin yield ratio [46].

When compared to thermal cracking, catalytic cracking over zeolite-based catalysts is able to achieve a higher propene/ethene ratio at lower temperatures [7]. This is because high-carbon-number alkanes are converted to low-molecular-weight alkenes by a carbenium-ion mechanism, which involves β-abstraction of zeolites on acid sites [8]. Zeolites are crystalline aluminosilicates distinguished by a developed micropore system and a high concentration of acid sites of various strengths. Various zeolite types have been used as components of cracking catalysts: FAU, BEA, MOR, MFI, and MSE [914]. Among these, ZSM-5 (MFI type) has achieved the highest yields of light olefins in catalytic cracking due to the high thermal and hydrothermal stability, high coking resistance, and high acidity of this zeolite [4, 5, 12, 15].

Because zeolite crystals are usually much larger in size than micropores, the diffusion of reagent/product molecules within the micropores often acts as the limiting reaction step. Diffusion limitations in the zeolite porous system increase the contribution of secondary processes and, thus, cause carbonaceous coke deposits both within the porous system and on the external surface of the crystal. This leads to the blockage of the porous system and, hence, to rapid catalyst deactivation, short lifetime, and poor zeolite utilization efficiency. To inhibit the deactivation and improve the performance of zeolite catalysts, some researchers have used nanosized or hierarchical zeolites; this approach enabled them to shorten the molecule diffusion path length and reduce the diffusion limitations inside the crystals [1619]. Recently, increasing attention has been paid to MFI zeolites with submicro- and nanocrystals due to their major advantages in terms of catalytic and adsorption performance compared to microsized zeolites [19, 20]. Konno et al. [21] demonstrated that the MFI deactivation rate in the catalytic cracking of n-hexane can be significantly decreased by reducing the crystal size. Furthermore, the same research team showed, with respect to the catalytic cracking of n-hexane, cyclohexane, and methylcyclohexane over MFI zeolites, that smaller crystals enhance both the conversion of model feedstocks and the yield of light olefins [21, 22]. Importantly, most studies have investigated the catalytic properties of pure zeolites, while no relevant data on the properties of MFI-based additives to cracking catalysts have ever been published.

The purpose of the present study was to investigate the crystal size effects of MFI zeolites used as components of additives to cracking catalysts on their catalytic performance. These catalysts were tested in a model cyclohexane cracking reaction and in conversion of hydrotreated vacuum gas oil (HTVGO) as a real hydrocarbon feedstock.

EXPERIMENTAL

Zeolite materials. Four MFI zeolites with similar chemical compositions supplied by different manufacturers were used as parent zeolites to prepare cracking catalysts. The main properties of these zeolites are presented in Table 1. MFI-Zeolyst and MFI-NS refer to commercial MFI samples, specifically CBV-8014 from Zeolyst International and a zeolite MFI from Nizhegorodskiye Sorbents (Nizhny Novgorod, Russia), respectively. MFI-Z-HTC and MFI-Z-VPC are pilot MFI samples synthesized by techniques developed by Zeolitica, Russia, specifically by conventional hydrothermal crystallization and novel vapor-phase crystallization, respectively. Silica gel, sodium hydroxide, sodium aluminate, a tetrapropylammonium hydroxide solution (TPAOH), and distilled water were used to prepare the reaction mixtures. The initial reaction mixtures basically differed in water content: a large amount of water (about 78 wt %) for MFI-Z-HTC and without free water for MFI-Z-VPC. Vapor-phase crystallization significantly (three- or fourfold) enhanced the crystallizer capacity, reduced the energy consumption for crystallization, eliminated liquid wastes produced from crystallization, and reduced the consumption of a costly template by a factor of three or four.

Table 1. Properties of parent zeolites

After crystallization, the isolated products were washed by centrifugation (until pH 9 was reached in wash waters), dried at 60°C, and calcined at 550°C for 12 h at a heating rate of 3°C/min. Then a threefold ion exchange was performed in a 0.1 M ammonium nitrate solution at 80°C for 3 h to prepare the NH4 form of the zeolite. The H-form zeolites were obtained by repeated calcination at 550°C for 6 h.

Zeolite modification and preparation of catalysts. All the parent zeolites were modified with 4 wt % of phosphorus by impregnation from a (NH4)2HPO4 solution. Based on these modified samples, we prepared additives that consisted of the following components: montmorillonite (0.15 wt % Na2O, 2.44 wt % Fe2O3, 24.4 wt % Al2O3), aluminum oxide (0.05 wt % Na2O), and P-containing zeolite, in a weight ratio of 1 : 1 : 2, respectively. The additives were dried at 100°C, then calcined at 600°C for 5 h, and steam-treated according to ASTM D4463 at 788°C in 100% steam for 5 h. The additive samples prepared in this manner were mixed with 5 wt % of a monozeolite cracking catalyst based on Y zeolite (0.38 wt % Na2O, 0.8 wt % REE2O3).

Physicochemical characterization. The chemical compositions were identified by atomic absorption spectrometry on a Shimadzu AA-6300 instrument, with the zeolites pre-decomposed in solutions of mineral acids (specifically, sulfuric, hydrochloric, and hydrofluoric acids). In parallel, the compositions were also determined using a Thermo Scientific ARL PERFORM'X X-ray fluorescence (XRF) spectrometer equipped with a 3.5 kW rhodium tube. Prior to XRF, the zeolites were pelletized with powdered boric acid. The data obtained by both techniques showed good concurrence.

The MFI phase compositions were characterized by XRD on a TongDa TD-3700 diffractometer equipped with a CuKα copper tube and a Mythen2 1D linear semiconductor detector in reflection mode in the 2θ range of 4°–50° with a step of 0.02°.

The porous structure of the samples was examined by low-temperature (–196°C) nitrogen adsorption on a Micromeritics ASAP-2010 porosimeter. Prior to testing, all the samples were evacuated at 350°C. The isotherms were recorded as functions Vads.gas(cm3/g) = f(p/p0). The total adsorption pore volume (Vpore) was measured by nitrogen adsorption at a relative pressure p/p0 = 0.99. The pore sizes were evaluated by the BJH (Barrett–Joyner–Halenda) method.

The crystal morphology and sizes were examined by scanning electron microscopy (SEM) on a Hitachi TM 3030 microscope. Before imaging, the surface was coated with a gold layer by vacuum sputtering.

The acidic properties of the samples was measured by ammonia temperature-programmed desorption (NH3-TPD) using a USGA-101 chemisorption analyzer (UNISIT, Russia). Approximately 0.15–0.2 g of the sample (0.25–1 mm grain size) was placed in a quartz reactor, heated in helium to 250°C at a rate of 10°C/min, and saturated with ammonia in an NH3/N2 mixture (10% ammonia) at 60°C for 30 minutes. Excess ammonia was purged with a 30 mL/min helium flow at 100°C for 60 minutes. The sample was then analyzed in a helium flow at temperatures elevated from 100 to 800°C at a heating rate of 8°C/min. Desorbed ammonia was detected by a thermal conductivity detector.

Catalytic tests. The additive samples were tested using both model and real feedstocks. Cyclohexane (Alfa Aesar; 99+ wt %) was used as a model feedstock, and the real feedstock was represented by HTVGO (EBP 559°C, 300–350 ppm sulfur). The tests were carried out according to ASTM D5154, specifically using a fixed-bed (MAK-10) flow-type experimental setup at 527°C. The feed was delivered over the course of 30 s at a weight hourly space velocity (WHSV) of 30 h–1 regardless of the feed type (in compliance with ASTM D5154). The gaseous products were analyzed on a GC-1000 gas chromatograph equipped with a capillary column (SiO2, 30 m × 0.32 mm) and a flame ionization detector. Liquid cracking products were quantified according to ASTM D2887 (simulated distillation analysis) on a Shimadzu GC2010 gas chromatograph equipped with an Rtx-2887 capillary column (10 m × 0.53 mm × 2.65 µm) and a flame ionization detector. All liquid hydrocarbons boiling below 216°C were considered gasoline fuels. The liquid product components were identified using a Shimadzu GCMS-QP2010 chromatograph/mass spectrometer equipped with an HP-1ms column (60 m × 0.25 μm) and an additional flame ionization detector.

The catalyst coke content was measured by the weight loss as the catalyst sample was calcined in air at 550°C.

The contribution of intermolecular hydrogen transfer reactions was quantified using a hydrogen transfer coefficient (HTC), which reflects the butanes to butenes selectivity ratio [23]. The HTC was calculated by the formula:

$${\rm{HTC}} = \;{{{S_{{\rm{butanes}}}}} \over {{S_{{\rm{butenes}}}}}},$$

where HTC is the hydrogen transfer coefficient; and Sbutanes and Sbutenes are the selectivity towards butanes and butenes, respectively.

RESULTS AND DISCUSSION

Physicochemical properties of parent zeolites. The physicochemical properties of the parent zeolites are presented in Table 1 and Figs. 14. The XRD examination showed all the samples to be pure-phase high-crystallinity MFI zeolites (Fig. 1). The chemical compositions indicate similar SiO2/Al2O3 ratios and equal residual sodium concentrations in all the zeolites (Table 1). However, despite their identical phase compositions and similar chemical compositions, the SEM images differ significantly (Fig. 2).

Fig. 1.
figure 1

XRD patterns: (a) MFI-Z-VPC; (b) MFI-Zeolyst; (c) MFI-Z-HTC; and (d) MFI-NS.

Fig. 2.
figure 2

SEM micrographs: (a) MFI-Z-VPC; (b) MFI-Zeolyst; (c) MFI-Z-HTC; and (d) MFI-NS.

The SEM images of MFI-Z-VPC and MFI-Z-HTC show spherical crystals (Figs. 2a, 2c) that, at a higher magnification, can be seen as aggregates of nanocrystallites. The crystal size strongly depends on the synthesis method. The MFI-Z-VPC crystals are 0.15 to 0.45 μm in size, whereas in MFI-Z-HTC we see markedly larger crystals (0.8–2.5 μm). The MFI-Zeolyst crystals are irregularly shaped, with a broad grain size distribution (0.3–0.8 μm, Fig. 2b). The micrograph of MFI-NS (Fig. 2d) displays large (3–6 μm) prismatic crystals. The comparative assessment of the SEM images showed the following ascending order of crystal sizes: MFI-Z-VPC < MFI-Zeolyst < MFI-Z-HTC < MFI-NS.

The textural properties of the zeolites were examined by low-temperature nitrogen adsorption (Fig. 3). The adsorption–desorption isotherm of MFI-NS is typical of large-crystal zeolites (and this corroborates well with the SEM data, Fig. 2d). In contrast, the isotherm of MFI-Z-VPC is typical of zeolites with small crystals (Fig. 2a). The abrupt rise of adsorption volume at high relative pressures (p/p0 > 0.9) is associated with capillary condensation in the mesopores and macropores generated between the zeolite crystals. Finally, MFI-Zeolyst and MFI-Z-HTC have isotherms typical of micro–mesoporous materials with a large contribution of intracrystalline mesopores (Figs. 2b, 2c). The H4 hysteresis loop at p/p0 of about 0.45 serves as evidence of intracrystalline cavitation mesopores with hindered access (in fact, they are accessible only through micropores). These mesopores are 2–5 nm in size. The data in Table 1 indicate that, despite the similar micropore volumes for all the samples, the total pore volume increases with a decrease in crystal size.

Fig. 3.
figure 3

Low-temperature nitrogen adsorption/desorption isotherms.

The acidity of the zeolites was measured by NH3-TPD. The NH3-TPD curves (Fig. 4) feature two peaks, at 220 and 420°C, attributable to ammonia desorption from weak and strong acid sites, respectively. The similar positions of these peaks clearly indicate the approximately identical strength of acid sites in all the samples. Besides, the similar acid site concentrations determined by NH3-TPD correlate well with the SiO2/Al2O3 ratios specified in Table 1.

Fig. 4.
figure 4

NH3-TPD profiles.

Thus, the physicochemical properties of the parent zeolites demonstrate that, despite their identical phase compositions and similar chemical compositions, the samples differed significantly in their crystal size and, hence, textural properties. In particular, the SEM images clearly show submicrocrystals for MFI-Z-VPC and MFI-Zeolyst, and microcrystals for MFI-Z-HTC and MFI-NS.

Effects of MFI crystal size on catalytic performance in cyclohexane cracking. Figure 5 illustrates the results of the catalytic tests of the MFI zeolites with the model cyclohexane feedstock. The GC analysis revealed C1–C4 n-paraffins and C2–C4 olefins in the gaseous products of cyclohexane cracking. Coke deposits did not exceed 1 wt %, which is typical for the cracking of model hydrocarbons.

Fig. 5.
figure 5

Conversion (a) and yield of C2–C4 olefins (b) as functions of MFI crystal size during cyclohexane cracking (MAK-10; 527°C; catalyst/feed ratio 4 g/g).

A comparative assessment of the MFI-based additives demonstrated that the submicrocrystalline MFI-Z-VPC and MFI-Zeolyst zeolites were substantially superior to the microcrystalline MFI-Z-HTC and MFI-NS in terms of cyclohexane conversion and olefin yield. This can likely be explained by the higher rate of reagent diffusion into zeolite pores, higher accessibility of acid sites, and higher efficiency of crystal utilization in zeolites with smaller crystals. This explanation is in concurrence with the data reported in [17]. A slight activity drop for MFI-Z-VPC compared to MFI-Zeolyst may be ascribed to partial amorphization of the sample with the smallest crystal size under steam heating at 788°C. Thus, the highest yield of C2–C4 olefins (20.3 wt %) was achieved when the additive to the cracking catalyst was prepared from MFI-Zeolyst (0.3–0.8 μm crystals).

Effects of MFI crystal size on catalytic activity of catalystadditive systems in HTVGO cracking. Table 2 presents the test data on the catalytic performance of the Y-zeolite-based cracking catalyst with MFI additives in the conversion of HTVGO. The table clearly shows that, when the real feedstock was used, the MFI crystal size had a weaker effect on the cracking performance than in the case of model feedstock. Despite the different crystal sizes of the MFI samples contained in the tested catalyst–additive systems, these systems exhibited similar conversion values and similar yields of the main reaction products. This lack of pronounced effect of crystal size on product yields may be caused by successive conversion of feed components on the base catalyst and on the olefin-boosting additive. In this case, highly reactive intermediates (specifically, C5–C7 olefins) were the compounds that predominantly transformed on the additive.

Table 2. Catalytic performance of catalysts with addition of phosphorus-modified MFI zeolites with different crystal sizes (ASTM D5154 test; 527°C; HTVGO feed; catalyst/feed ratio 4 g/g; feeding time 30 s)

The table also indicates that, unlike conversion and product yield, other process parameters such as coke content and HTC definitely depend on crystal size. The zeolites with smaller crystals exhibited a lower contribution of hydrogen transfer reactions and, hence, a lower coke content. Apparently, decreasing the crystal size shortens the diffusion path length of the feedstock molecules, thus reducing the contribution of secondary hydrogen transfer processes responsible for coking and catalyst deactivation. This finding demonstrates that reducing the crystal size may inhibit the catalyst deactivation.

CONCLUSIONS

A number of MFI zeolites with crystal sizes ranging from 0.15 to 6 μm were investigated as components of additives to a cracking catalyst. The crystal size was shown to have a major effect on the conversion of a model cyclohexane feedstock; in contrast, the crystal size did not influence the conversion of a real HTVGO feed. In the case of model feedstock, the cyclohexane conversion increased from 30–40 to 60–70% as the MFI crystal size was decreased from the micro- to submicroscale, with the yield of light olefins being enhanced from 11–12 to 18–20 wt %. When the real feedstock was used, the downsizing of the MFI crystals added to the cracking catalyst decreased the contribution of hydrogen transfer reactions, thus reducing coke deposits.