Role of Al in Na-ZSM-5 zeolite structure on catalyst stability in butene cracking reaction

The Na-ZSM-5 catalysts (SiO2/Al2O3 molar ratio = 20, 35, and 50) were prepared by rapid crystallization method to investigate their performance in butene cracking reaction. The XRD, XRF, NH3-TPD, FT-IR, TPO, UV–Vis, and 1H, 27Al, 29Si MAS NMR techniques were used to identify the physical and chemical properties of Na-ZSM-5 catalysts. The silanol group (Si–OH) was the main acid site of Na-ZSM-5, and it was proposed to be the active site for the butene cracking reaction. The butene conversion and coke formation were associated with the abundance of silanol groups over the Na-ZSM-5 catalyst. The dealumination, resulting in the deformation of tetrahedral framework aluminum species was a key factor for Na-ZSM-5 catalyst deactivation, because of the Si–O–Al bond breaking and formation of Si–O–Si bond. The stability of the Si–O–Al bond was linked to the molar number of sodium since the Na atom interacts with the Si–O–Al bond to form Si–ONa–Al structure, which enhances the stability of the silanol group. Therefore, the Si–ONa–Al in zeolite framework was an essential structure to retain the catalyst stability during the reaction. The Na-ZSM-5 with the lowest SiO2/Al2O3 molar ratio showed the best performance in this study resulting the highest propylene yield and catalyst stability.


Results
Catalyst physical and chemical properties. The XRD pattern of synthesized Na-ZSM-5 catalysts, fresh and spent, are depicted in Supplement Figure S1. The fresh and spent Na-ZSM-5 catalysts presented the typical MFI structure. The sodium content on the bulk of catalyst was measured by XRF analysis and summarized in Supplement Table S1. The sodium content on the bulk of catalysts decreased with increasing the SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5 catalyst.
The acid strength and acidity number on Na-ZSM-5 catalysts with various SiO 2 /Al 2 O 3 molar ratios are displayed in Fig. 1 and summarized in Supplement Table S2. Based on Fig. 1, NH 3 -TPD profiles on Na-ZSM-5 catalysts presented two main desorption peaks, related to weak and medium acid sites at about 120 °C and 270 °C, respectively. The number of total acid sites, weak and medium, increased when decreasing the SiO 2 /Al 2 O 3 molar ratio, as detailed in Supplement Table S2. The OH stretching region of Na-ZSM-5 catalyst is detected by FT-IR as illustrated in Supplement Figure S2. The bands at ca. 3,490, 3,580, 3,685, and 3,745 cm −1 were assigned to silanol nests 22 , OH groups interacting with multivalent cations (negative charge compensation) in the zeolite framework 23 , internal silanol groups of hydroxyl nests 24 , and external silanols 25 , respectively. When increasing the SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5, the band intensities related to silanol groups decreased as portrayed in Supplement Figure S2. 1 H MAS NMR spectra of fresh synthesized Na-ZSM-5 catalysts, displayed in Fig. 2a-c and summarized in Supplement Table S3, indicate the types of zeolite acidic structures. The main resonance signal between 3.2 and 3.4 ppm was attributed to the presence of Si-OH groups 26 . The signal of synthesized Na-ZSM-5 with SiO 2 / Al 2 O 3 = 20, 35, and 50 were located at 3.2, 3.4, and 3.2 ppm, respectively. Based on Fig. 2a-c, no Si-(OH)-Al peak was detected over the synthesized Na-ZSM-5 catalysts. Therefore, Na-ZSM-5 could have only one type of Si-OH structure, which exhibited weaker acid strength than Si-(OH)-Al site (strong Brønsted acid site). The Si-OH 1 H MAS NMR band intensity, as well as its abundance, decreased while increasing SiO 2 /Al 2 O 3 molar ratio of synthesized Na-ZSM-5 as shown in Supplement Table S3. 27 Al MAS NMR technique was used to identify the aluminum framework and extra-framework species in zeolite catalysts 20,27,28 . Figure 2d-i and Supplement Table S3 present the profile and density of the aluminum species in zeolite structure, respectively. In Fig. 2d-i, the main peak observed at ca. 55 ppm belongs to tetrahedral framework aluminum species (Al IV ) in zeolite framework 20 . The small peak at ca. −20 ppm was assigned to spinning sideband 29 . The fresh Na-ZSM-5 revealed that the Al IV framework concentration increased with increasing Al 2 O 3 content during the catalyst synthesis (decreasing the SiO 2 /Al 2 O 3 molar ratio). No peaks of penta-coordinated aluminum species (Al V ) (ca. 30 ppm), or octahedral aluminum species (Al VI ) (ca. 0 ppm) 20  www.nature.com/scientificreports/ and Al VI species, were produced, however, the intensity of the Al IV species in zeolite framework peak decreased compared to the fresh samples, and the concentration change was proportional to the SiO 2 /Al 2 O 3 molar ratio of catalyst, this confirms that dealumination occurred during the butene cracking reaction on these catalysts. 29 Si MAS NMR spectra present peaks at ca. −112, −102, and −93 ppm which were attributed to the Si(4Si,0Al) site, Si atoms without neighboring Al atom, Si(3Si,1Al) site, Si atoms with one neighboring Al atom, and Si(2Si,2Al) site, Si atoms with two neighboring Al atoms, respectively 30,31 . The spectra of synthesized Na-ZSM-5 zeolites in this research showed resonance signals corresponding to previous studies 30,31 as shown in Fig. 2j-o, the intensities of each signal for all samples are presented in Supplement Table S3. Based on Fig. 2j-o and Supplement Table S3, the fresh synthesized Na-ZSM-5 zeolites displayed one main resonance signal at ca. − 102 ppm corresponding to the Si(3Si,1Al) site, and a low intensity peak at ca. − 113 ppm, attributed to the Si atoms without neighboring Al atom structure. The fresh sample was also presented a small shoulder band, typical of two Si atoms with two neighboring Al atoms at ca. − 94 ppm. Nevertheless, the reaction time (300 min), the signal patterns of spent catalysts were notoriously changed. The band intensity at ca. − 102 ppm reduced, whereas the Si(4Si,0Al) site signal (ca. − 113 ppm) abruptly increased compared to the fresh zeolite catalyst. No peaks for Si(1Si,3Al) or Si(0Si,4Al) sites were observed on fresh and spent Na-ZSM-5 catalysts.
Catalyst performance test. Figure 3 presents the performance of Na-ZSM-5 catalysts for butene cracking reaction as a function of time on stream (TOS). From Fig. 3a, the best stability for the butene cracking reaction was reached over the Na-ZSM-5 (SiO 2 /Al 2 O 3 = 20), and the deactivation rate followed the trend of 7.03%, 8.31%, and 14.32% for SiO 2 /Al 2 O 3 = 20, 35, and 50, respectively. This increase in deactivation led to a reduction in initial butene conversion rate. The propylene selectivity as a function of time enhanced with increasing the SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5 catalysts as shown in Fig. 3b. Remarkably, the Na-ZSM-5 (SiO 2 /Al 2 O 3 = 20) presented excellent propylene production with a final propylene yield of 27.48%, as displayed in Fig. 3c. The initial ethylene and light alkanes (C 1 -C 4 alkanes) selectivities dropped when raising SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5 catalysts, as illustrated in Fig. 3d,e, respectively.
Coke characterization. UV-Vis spectrometry and TPO were used to analyze the carbon species and weight percentage of carbon deposit over the Na-ZSM-5 surface after the butene cracking reaction, respectively.

Discussion
It is widely accepted that the strong Brønsted acid site (Si-(OH)-Al) over acid-solid zeolite, especially for H-ZSM-5 catalysts, dominates the butene cracking reaction 6,38 . However, in the case that the catalyst lacks Si-(OH)-Al sites like Na-ZSM-5, the silanol group can act as an active site in this reaction and plays the same role as strong Brønsted acid site. The concentration of silanol group, determined by 1 H MAS NMR technique as summarized in Supplement Table S3, correlated with the butene conversion over time, as depicted in Fig. 3a. Therefore, the silanol group on Na-ZSM-5 was the active site for butene cracking reaction, which confirmed previous findings where silanol groups could play the same role as the strong Brønsted acid site and these were active sites for different reactions such as 1-butene cracking reaction (silanol nest over silicalite-1) 39 , toluene methylation (silanol nest over silicalite-1) 22 , and methanol conversion (hydrogen-bonded silanol groups over methanol conversion) 40 .
In the case of C 4 olefins cracking reaction, the mechanism is more complicated. The main reaction starts when the acid site of the catalyst is protonated, this proton reacts with C 4 = species to form the carbenium ion (C 4 + ). Then, the oligomerization will occur and crack to light alkenes in various pathways depending on acid strength of the catalyst 41 . Finally, the light alkanes and aromatics are generated via hydrogen transfer reaction 42 and dehydrogenation-aromatization 43 reactions, respectively. Both of these reactions require a strong acid site on catalyst 39 . For Na-ZSM-5 catalyst, the silanol groups worked as the active site in butene cracking reaction. The cracking reaction is composed of three main parallel reactions: Stronger acid site favors the pathway (2) over pathway (3) 6,44,45 . The ratio of propylene to ethylene decreased with a lower SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5 as displayed in Fig. 3f. Therefore, the lower SiO 2 /Al 2 O 3 molar ratio of Na-ZSM-5 could generate stronger acid site, leading to higher selectivity of pathway (2) over pathway (3). Besides, when acid strength over the Na-ZSM-5 was stronger as presented in Fig. 1, the hydrogen transfer reaction to propylene to produce propane was improved, thus, the propylene selectivity of Na-ZSM-5 at low SiO 2 /Al 2 O 3 molar ratio was lower than the Na-ZSM-5 with high SiO 2 /Al 2 O 3 molar ratio. This result was confirmed by earlier research reported by Lin et al. 41 . They found that the acid strength distribution could control the butene conversion and the propylene/ethylene mole ratio (P/E ratio) and the P/E ratio increased with decreasing the density of strong acid site over the catalyst. The sodium-containing ZSM-5 catalyst retards the hydrogen transfer reaction, hindering the light alkanes (C 1 -C 4 alkanes) production since the Na-ZSM-5 only had silanol groups in the zeolite framework and these acid sites were not strong enough to activate this side reaction. However, the hydrogen transfer reaction slightly occurred over silanol groups. The light alkanes production was associated with the acid strength of the Na-ZSM-5 catalysts and increased correspondingly as illustrated in Figs. 1, 3e, and reported in Supplement Table S2. Catalytic performance for butene cracking  Table 1.  Figure S3e). Therefore, the hydrogen transfer reaction to produce light alkanes was promoted over H-ZSM-5 and hindered on Na-ZSM-5, leading to a lower propylene yield for H-ZSM-5 (SiO 2 /Al 2 O 3 = 20) (Supplement Figure S3c). The propylene yield for Na-ZSM-5 (SiO 2 /Al 2 O 3 = ∞) was low (Supplement Figure S4c) because of the lower butene conversion over this catalyst (Supplement Figure S4a In order to investigate the effect of WHSV in butene cracking reaction, Na-ZSM-5 (SiO 2 /Al 2 O 3 = 20) was tested at different WHSVs 1.5 h −1 , 3.0 h −1 , and 6.0 h −1 (Supplement Figure S5). When increasing the WHSV, the butene conversion and ethylene selectivity decreased, and the propylene to ethylene ratio raised. These results corresponded with previous studies 41 . Decreasing the WHSV, lowered the propylene selectivity, and enhanced the light alkanes production. Consequently, hydrogen transfer reaction was inversely proportional to the WHSV. The WHSV = 3.0 h −1 was found to be the best condition providing the highest propylene yield at the initial TOS and the lowest deactivation rate after 300 min of TOS in butene cracking reaction.
The role of Al atom in zeolite structure was related to the Na-ZSM-5 catalysts deactivation as presented in the 27 Al and 29 Si MAS NMR results. The dealumination of tetrahedral framework aluminum species (Al IV ) could occur during the butene reaction, and the loss of Al IV species was followed with the 29 Si MAS NMR spectra. As depicted in Fig. 6a, the Na-ZSM-5 catalyst deactivation rate, and the SiO 2 /Al 2 O 3 molar ratio are directly related. www.nature.com/scientificreports/ The lower Al IV species band intensity of spent Na-ZSM-5 catalyst, compared with fresh samples, confirmed the loss of Al IV species in zeolite framework structure during the reaction, and the dealumination was remarkably escalated with lower aluminum content in zeolite structure, as displayed in Fig. 6b. These results were confirmed by Al-Khattaf et al. 51 previous research, in which they found that the dealumination of H-ZSM-5 (SiO 2 /Al 2 O 3 = 80) occurred because the ratio of Al VI to Al IV species increased after steaming from 0.009 (fresh) to 0.015 (steamed at 350 °C). The 29 Si MAS NMR spectra of silicon environments are presented in Fig. 2j-o, a difference between fresh and spent samples is evident. The main type of silicon environment on fresh Na-ZSM-5 was Si(3Si,1Al) site, implying the presence of Si-O-Al bond in zeolite structure. While after the reaction, the Si(4Si,0Al) site was the major silicon surrounding type. This result confirms that Si-O-Al bond in the zeolite framework was broken during the cracking reaction, leading to dealumination. Based on Fig. 6c, when the SiO 2 /Al 2 O 3 molar ratio raised, the Si-O-Al bond intensity variation between fresh and spent catalyst was higher, leading to an increase of deactivation rate. This outcome verified that the Si-O-Al bond was lost, and Si-O-Si bond formed during the reaction as illustrated in Fig. 7a. However, the framework of spent samples retained the typical MFI structure of ZSM-5 catalysts (supplement figure S1). To further investigate the effect of sodium bulk content (Supplement Table S2 Another deactivation cause was the carbon deposition on the catalyst surface. It is well established that coke formation takes place on a Si-(OH)-Al structure (strong Brønsted acid site) 11 . In the case of catalysts which only had the silanol groups as an active site, the coke deposits could also occur on these sites. The coke content over the catalysts after the reaction was determined by TPO technique. The TPO results are displayed in Fig. 4. It exhibited direct correlation with the number of silanol groups detected by 1 H MAS NMR (Supplement Table S3). In case of Na-ZSM-5, the acid strength was not high enough to produce a high concentration of polyaromatic coke species. As shown in Fig. 5, the intensity in the range of UV-Vis band representing charged & neutral polyaromatics was lower than the range of UV-Vis bands representing charged poly-alkylated benzenes and charged alkylated naphthalenes. Moreover, the result determined by TPO and UV-Vis techniques demonstrated that the Na-ZSM-5 (SiO 2 /Al 2 O 3 = 20) could generate a greater deal of coke content than the other samples (Na-ZSM-5 at SiO 2 /Al 2 O 3 = 35, and 50) as seen from higher absorption intensity in the whole wavenumber range of coke species bands as depicted in Fig. 5. Besides, the charged & neutral polyaromatic coke species were more abundant over the H-ZSM-5 (SiO 2 /Al 2 O 3 = 20), because of the highest acid strength of this catalyst (Supplement Figure S3e,h). The proposed mechanism for coke generation 41,53,54 related to the acid strength of the catalyst is illustrated in Fig. 7c.
Finally, the results suggested the deactivation of Na-ZSM-5 catalyst in butene cracking reaction was strongly affected by the removal of framework aluminum in zeolite structure, while the coke formation was of minor effect. The loss of tetrahedral framework aluminum species during the reaction was caused by the Si-O-Al bond breaking in Si(3Si,1Al) site to form Si-O-Si in Si(4Si,0Al) site, and Na content of the catalyst correlated with the Si-O-Al structure. Hence, the Si-O-Al bond in zeolite framework with the Na atom to form Si-ONa-Al structure www.nature.com/scientificreports/ was the determining factor to inhibit the dealumination process, and improved the stability of the silanol groups over Na-ZSM-5 catalysts in butene cracking reaction.

Conclusion
The Na-ZSM-5 catalysts (SiO 2 /Al 2 O 3 molar ratio = 20, 35, and 50) were synthesized and tested for butene cracking reaction to investigate the catalyst deactivation and coke formation. Based on the characterization, the fresh Na-ZSM-5 presented the typical MFI structure, and only silanol groups (Si-OH) were detected over Na-ZSM-5 catalyst surface as the active sites. The butene conversion was associated to the acid strength of these sites. The coke content exhibited a similar trend as the butene conversion. Regarding catalyst deactivation, the dealumination during the reaction critically affected the Na-ZSM-5 stability due to the loss of the tetrahedral framework aluminum species. Consequently, the break of Si-O-Al bonds in Si(OSi) 3 OAl framework led to the formation of new Si-O-Si bonds and Si(OSi) 4 structures. Also, the degree of dealumination of Na-ZSM-5 catalyst was a function of the alumina content in zeolite structure (SiO 2 /Al 2 O 3 molar ratio), and the relation between Si-O-Al bond stability and sodium content of catalyst was established. The formation of Si-ONa-Al structure led to enhanced silanol group stability. Finally, the Na-ZSM-5 (SiO 2 /Al 2 O 3 = 20) was the most promising catalyst in this study rendering the highest propylene yield, catalytic activity, and stability.

Method
Material synthesis. The Na-ZSM-5 zeolite catalyst was synthesized by rapid crystallization procedure 55 as reported by the previous study 17 . Sodium silicate solution (Merck), and aluminum sulfate octadecahydrate (UNIVAR) were used as the silica source, and alumina source, respectively. The template was tetrapropylammonium bromide (Sigma-Aldrich), and the alkali source was sodium hydroxide (Merck). The molar composition of the resulting synthesis mixture gel with different SiO 2 /Al 2 O 3 molar ratio was 0.    = 20) were synthesized and the related information on catalytic performance and the characterization were provided in Supplementary information. The Na-ZSM-5 without the alumina source (SiO 2 /Al 2 O 3 = ∞) was synthesized with the same method which was mentioned earlier and the H-ZSM-5 (SiO 2 /Al 2 O 3 = 20) was further synthesized from Na-ZSM-5 (SiO 2 / Al 2 O 3 = 20). The ion-exchange with 1 M ammonium nitrate solution (NH 4 NO 3 ) at 80 °C for 2 h was applied to transfer Na + ion to NH 4 + ion in ZSM-5 structure, and then the catalyst was rinsed with deionized water, dried in the oven, and calcined as same as the method of Na-ZSM-5 which was mentioned earlier, respectively. Finally, the H-ZSM-5 with 20 of the SiO 2 /Al 2 O 3 molar ratio was observed.
Catalyst characterization. The crystallinity and crystal structure of fresh and spent samples were analyzed by X-ray powder diffraction (XRD) technique with a Bruker AXS D8 Advance (Ni-filtered CuK α radiation). The 2 range between 5° and 50° with a step size of 0.01 was applied to record in XRD pattern of the sample. The X-ray fluorescence (XRF) with a Bruker S8 TIGER was used to analyze the bulk composition of catalyst sample. The acidity strength was characterized by temperature-programmed desorption of ammonia (NH 3 -TPD) technique with a Micromeritics Chemisorb 2,750 automated system. The sample weight (0.1 g) was moved into the U-quartz tube reactor under 25 ml/min of He gas stream at 550 °C in a preheating step, then the temperature was dropped to 40 °C to adsorb the ammonia under 25 ml/min of 15% NH 3 /He mixed gas stream for half an hour. Subsequently, the physisorbed ammonia was eliminated with He gas stream. Finally, the desorbed ammonia was recorded by TCD detector with linearly temperature increasing from 40 to 600 °C (10 °C/min). The stretching vibration of OH groups on catalyst sample was determined by the Fourier Transform-Infrared Spectroscopy (FT-IR) with a Bruker VERTEX 70v FT-IR Spectrometer. The 0.2 g of sample was transferred into the in situ IR cell with the KBr windows to preheat at 550 °C (10 °C/min) for 1 h with 25 ml/min of N 2 gas stream. Afterward, the temperature was reduced to 40 °C for the OH group spectra recording, subtracting automatically with the background spectrum. The Fourier Transform Nuclear Magnetic Resonance Spectrometer 400 MHz (Solid) with a Bruker AVANCE III HD (Ascend 400 WB) spectrometer (4 mm diameter rotor, 8 kHz of speed rate, and 400.20 MHz of resonance frequency) was used to characterize the acidity structure, type of aluminum, and Si environment by 1 H, 27 Al, and 29 Si MAS NMR methods over the sample, respectively. The coke content and species were identified by the temperature-programmed oxidation (TPO) and ultraviolet-visible spectroscopy (UV-Vis) techniques, respectively. The TPO profile was analyzed by a Micromeritics Chemisorb 2,750 automated system and the CO 2 formation was detected by a gas chromatography with Rt-Q-BOND-Fused Silica Plot column. The 0.05 g of spent sample was weight and transferred into the U-quartz tube reactor under 25 ml/min of 1% O 2 /He mixed gas stream and the signal was recorded with 5 °C/min of step size in the range between room temperature and 970 °C. The Lambda 650 UV-Vis spectrophotometer was applied to characterize the intensity signal variation between spent and fresh catalyst samples at the range from 12,500 to 50,000 cm −1 .
Catalyst testing. The fixed-bed tubular reactor with a K type of thermocouple was used to test the catalytic cracking reaction of butene. The reactor was made up from stainless steel with 19.05 mm of inner reactor diameter. The steps of reaction testing under atmospheric pressure were following: (1) the catalyst sample (1 g) was preheated at 550 °C with N 2 for 1 h, and (2) the reaction temperature was at 500 °C with the weight hourly space velocity (WHSV) of the reactant gas at 3 h −1 (the molar composition reactant gas between butene and N 2 was 65:35). The product compositions were evaluated by a gas chromatography, an Agilent J&W HP-PLOT Al 2 O 3 S with FID detector. The conversion, product selectivity, product yield, and deactivation rate after 300 min with time on stream were determined by Eqs. (1) -(4), respectively.
Here W 0 , W t , and (W i ) t are defined as the weight percentages of butenes in the feed, of butenes in the product, and of any hydrocarbons in the product, detected by the gas chromatography, respectively.
And X 0 and X f are defined as the butene conversion at 50 min after reaction test and 300 min after reaction test, respectively.