Enhanced Hydrothermal Stability and Propylene Selectivity of Electron Beam Irradiation-Induced Hierarchical Fluid Catalytic Cracking Additives

Abstract

A cheap, environmentally friendly, easily scalable post-treatment of Na-ZSM-5 (Si/Al molar ratio = 20 or 30) via electron-beam irradiation to produce hierarchical H-ZSM-5 as a propylene-increasing fluid catalytic cracking additive was performed. Higher specific surface areas and highly accessible porous systems were obtained among the irradiated samples. A combination of ²⁷Al, ¹H magic angle spinning nuclear magnetic resonance and NH₃-temperature-programmed desorption methods showed that upon irradiation, some of the framework’s tetrahedral Al atoms were removed as non-framework Al atoms via flexible coordination with Si-OH groups (either framework or non-framework defects), thus increasing the H-ZSM-5 acidity and stability during hydrothermal dealumination. The enhanced selectivity and stability toward propylene production over the irradiated H-ZSM-5 samples were attributed to the integration of the reserved population of medium acid sites into the highly accessible hierarchical network. N₂ adsorption–desorption isotherm data showed that the Si-rich H-ZSM-5 samples possessed an obvious ink-bottle-shaped micro-mesopore network and a greater degree of disordered orientation of the straight pore systems toward the exterior surfaces. Micro-activity test data suggested that with an increasing Si/Al ratio, the H-ZSM-5 additives lost some extent of their cracking activity due to the constricted hierarchical pore network toward the exterior surface but gained more stability and selectivity for propylene due to the reserved medium acid sites.

1. Introduction

In the refining industry, fluid catalytic cracking (FCC) of low-value, heavy distillates such as bottoms of vacuum and atmospheric distillation units (vacuum gas oil and atmospheric residue) is the most important process for transforming high-boiling point hydrocarbon fractions to more valuable, lighter products (light olefins, gasoline and diesel) [1]. Light olefins, especially propylene, are highly demanded building blocks for the manufacture of large amounts of products annually in various chemical industries [2]. To maximize the propylene yield from the FCC process and thus refinery profitability, many efforts have been made to manipulate FCC reaction variables (feed properties, catalyst-to-oil ratios, residence times, and reaction temperatures) [3,4] and to modify ZSM-5-based FCC catalysts [5,6,7,8,9,10]. The key feature of ZSM-5-based catalyst modification for increasing light olefin production in the FCC process is the suppression of competitive acid-catalyzed reactions (e.g., cyclization, hydrogen transfer and coke formation reactions), which consume primary olefin products from the primary cracking reaction via beta scission [7]. To minimize the occurrence of these olefin-consuming secondary reactions on the ZSM-5 catalyst and to improve not only the selectivity of the products but also the catalyst stability, a number of strategies have been used to control the density and strength of the ZSM-5 acid sites and overcome diffusion limitations (via appropriately introducing secondary meso- and/or macropores into the ZSM-5 intrinsic micropore network) [5,6,8,9,10]. Desilication and/or dealumination via alkaline and/or acid post-synthesis treatment are common practices for accessing ZSM-5 catalysts due to their simplicity and efficiency in optimizing acidity and porosity to improve catalyst performance [11,12,13]. However, these practices are not sustainable due to the generation of secondary waste streams.

We recently demonstrated that post-synthesis treatment of Al-rich and Si-rich Na-ZSM-5 zeolites via electron-beam irradiation at room temperature with doses up to 80 kGy (1 kGy = 1 kJ/kg) can ease complete detachment of the organic template from zeolites at milder temperatures and concomitantly rearrange framework T atoms [14]. This process led to an increase in the specific surface area and regulation of the micro-mesopore network without leading to the disintegration of the MFI framework. These results are attributed to the direct radiolysis of water molecules coupled with indirect ionization and dissociation along the chain of extra-framework–framework interactions, leading to the partial detemplation of tetra-propyl-ammonium (TPA) ions. The partial detemplation obtained by applying an appropriate dose of irradiation to the samples in ambient conditions is a necessary condition to facilitate the supplementary detemplation as well as to modify the porous structures of the materials. Moreover, the extent of detemplation by irradiation, the extent of distortion in the frameworks, and the hierarchical properties (e.g., sizes and shapes of the mesopores) can be controlled by the absorbed dose of and the hydrophilicity of the materials. The reported results show that a network of large-mouth, funnel-shaped pores was formed in the Al-rich Na-ZSM-5 zeolite (Si/Al = 20), whereas channels with ink-bottle-shaped pores were formed in the Si-rich Na-ZSM-5 zeolite (Si/Al = 30). Apart from the energy saving practice, the environmentally benign possibility of tuning meso-porosity in ZSM-5 zeolites without using multiple steps of chemical treatment and the feasibility of upscaling action of e-beam irradiation treatment would make this technique an alternative for eliminating the internal diffusion limitation of ZSM-5 catalysts. An important question remains as to whether the post-synthesis treatment of TPA⁺-Na⁺-ZSM-5 zeolites via e-beam irradiation is able to tune the density and strength of acid sites in their H-forms for practical application in catalytic cracking of heavy feedstocks or whether e-beam irradiation-derived defects in hierarchical H-ZSM-5 treatment are unstable and inapplicable to catalytic applications. This study aimed to obtain significant findings to answer this question.

In this work, the effectiveness of utilizing e-beam irradiation-induced hierarchical H-ZSM-5 as an FCC catalyst additive for enhancing propylene selectivity and catalyst stability was studied. We focused on the effects of irradiation-induced changes in texture and acidity on the activity and stability of H-ZSM-5 catalysts during the cracking of heavy oil atmospheric residue (AR) feedstock. Parent TPA⁺-Na⁺-ZSM-5 zeolites with Si/Al ratios of ca. 20 and 30 were synthesized and exposed to an irradiation dose of 80 kGy using an electron beam, followed by activation to their H forms. The physicochemical properties, including crystal structure, morphology, and porous texture, Al coordination, and acidic properties of fresh and steamed H-ZSM-5 catalysts were described and compared to those of the irradiated samples. The catalytic performance of these catalysts was evaluated for use as propylene-increasing FCC additives in a fixed-bed micro-activity test (MAT) unit for AR cracking under FCC conditions.

2. Results

Characterization of the H-ZSM-5 Samples

Table 1. Molar composition of the starting gels and the as-synthesized Na-ZSM-5 zeolites.

Sample	Initial Gel Composition		Product Composition 1
	Molar Composition	Calculated Si/Al	Molar Composition	Practical Si/Al
20-NaZ	1SiO2:0.35TPABr: 0.054Na2O:0.025Al2O3: 40H2O	20.0	((C3H7)4N)3.56Na0.99Al4.55Si91.45O192	20.1
30-NaZ	1SiO2:0.35TPABr: 0.054Na2O:0.016Al2O3: 40H2O	31.3	((C3H7)4N)3.11Na0.14Al3.05Si92.95O192	30.5

¹ Determined by TG/DSC and ICP/MS.

shows the unit cell compositions of the as-synthesized 20-NaZ and 30-NaZ samples with practical Si/Al molar ratios of 20.1 and 30.5, respectively. The numbers of tetra-propylammonium (TPA⁺) cations per unit cell for the 20-NaZ and 30-NaZ samples were estimated to be 3.56 and 3.11, respectively. The total number of TPA⁺ and Na⁺ cations was found to be greater than the number of Al atoms in the 30-NaZ unit cell, indicating that structural defects such as clusters of SiO⁻ must be present in the 30-NaZ sample to maintain system charge neutrality. Since the unit cell of the ZSM-5 zeolite contains four channel intersections with a diameter of ca. 9 Å [15], which is in close proximity to the diameter of the TPA⁺ cation [16], each intersection in the unit cell of the as-synthesized 20-NaZ and 30-NaZ was occluded by one TPA⁺ cation with four propyl substituents around the N atom of the TPA⁺ cation pointing along four intersecting channels. Considering that the TPA⁺ cation cannot compensate for the charges of more than one framework Al atom per MFI intersection, as well as the structure–direction competition between small Na⁺ and bulky TPA⁺ cations to balance the anionic aluminate species, we suppose that the Al atoms in the 20-NaZ and 30-NaZ samples are located in both the large intersection and the narrow sinusoidal/straight channels.

Figure 1 shows the XRD patterns of the fresh H-ZSM-5 samples obtained from calcination of the NH₄⁺-exchanged 20-NaZ and 30-NaZ samples with and without e-beam irradiation post-treatment. The XRD patterns of the steamed H-ZSM-5 samples, which were obtained after severe steaming of the corresponding fresh H-ZSM-5 samples in a flow of pure steam at 973 K for 8 h, are also included. As shown in Figure 1, all of the fresh and steamed H-ZSM-5 samples exhibited characteristic reflections of highly pure calcined MFI structures. No impurity peaks could be detected in any of the XRD patterns, even in those of the steamed samples (y-HZ-St, y-HZ-Irra-St, y = 20, 30). The XRD patterns of the H-ZSM-5 samples with the same Si/Al molar ratio indicate no apparent change in crystallinity among these samples after treatment with e-beam irradiation and/or steaming. All of the H-ZSM-5 samples exhibited orthorhombic symmetry with the space group Pnma (No. 62) according to the obvious separation between the (051) and (501) reflections observed in the 2-theta range of 23.0–25.0°.

Table 2. Elemental compositions of the fresh H-ZSM-5 and the steamed H-ZSM-5.

Sample	Treatment	Composition 1 (wt.%)				Total Si, Al Contents (mmol g−1)		Bulk Si/Al
		SiO2	Al2O3	Na2O	H2O	QTSi	QTAl
20-HZ	Activated (813 K-3 h)	89.0	3.8	0.07	6.9	14.81	0.74	20.0
30-HZ		92.5	2.6	0.09	4.7	15.40	0.51	30.2
20-HZ-Irra	Irradiated (10 kGy) + Activated (813 K-3 h)	89.9	3.8	0.08	5.3	14.96	0.74	20.2
30-HZ-Irra		93.9	2.6	0.06	3.3	15.63	0.51	31.3
20-HZ-St	Activated (813 K-3 h) + Steamed 973 K-4 h	94.6	3.8	0.09	0.5	15.74	0.75	21.0
30-HZ-St		96.7	2.6	0.07	0.5	16.09	0.52	30.9
20-HZ-Irra-St	Irradiated (10 kGy) + Activated (813 K-3 h) + Steamed (973 K-4 h)	92.8	3.9	0.09	2.3	15.44	0.77	20.1
30-HZ-Irra-St		95.1	2.6	0.09	1.4	15.84	0.51	31.1

¹ Determined by TG/DSC and ICP/MS.

shows the elemental compositions of the fresh and steamed H-ZSM-5 samples. The variation in the contents of structural water within these samples, as determined by TG/DSC, is also shown in Table 2. The analysis of the elemental composition revealed that a high degree of NH₄⁺ ion exchange (ca. 83 to 88%) occurred during the H⁺ activation step, and the bulk Si/Al molar ratios of all the H-ZSM-5 samples were virtually unchanged compared to those of their parent Na-ZSM-5 samples. The analysis also revealed that the water content decreased in the order of y-HZ > y-HZ-Irra > y-HZ-Irra-St > y-HZ-St (y = 20, 30).

SEM images of the fresh H-ZSM-5 and steamed H-ZSM-5 samples were taken and are shown in Figure 2. All the samples exhibited mostly uniform crystals, which are consistent with the high degree of crystallinity obtained from the XRD patterns. An average particle size of ca. 2 µm and 5 µm was observed in the Al-rich H-ZSM-5 and the Si-rich H-ZSM-5 samples, respectively.

The surface areas and pore volumes of the H-ZSM-5 samples, determined by N₂ physisorption at 77 K, are shown in

Table 3. Textural properties of the fresh H-ZSM-5 and steamed H-ZSM-5 samples.

Sample	Textural Properties
	SBET a (m2/g)	Smicro b (m2/g)	Smeso c (m2/g)	Vtotal d (cm3/g)	Vmicro e (cm3/g)	Vmeso e (cm3/g)
Fresh H-ZSM-5
20-HZ	370.8	219.6 (59.2%)	151.2 (40.8%)	0.163	0.108 (66.3%)	0.055 (33.7%)
20-HZ-Irra	459.6	202.6 (44.1%)	257.0 (55.9%)	0.197	0.098 (49.7%)	0.099 (50.3%)
30-HZ	395.0	314.6 (79.6%)	80.4 (20.4%)	0.171	0.124 (72.5%)	0.047 (27.5%)
30-HZ-Irra	399.8	293.1 (73.1%)	106.7 (26.7%)	0.176	0.115 (65.3%)	0.061 (34.7%)
Steamed H-ZSM-5
20-HZ-St	294.4	114.0 (38.7%)	180.4 (61.3%)	0.210	0.078 (37.1%)	0.132 (62.9%)
20-HZ-Irra-St	348.7	128.0 (36.7%)	220.7 (63.3%)	0.285	0.138 (48.4%)	0.147 (51.6%)
30-HZ-St	281.1	100.8 (35.9%)	180.3 (64.1%)	0.288	0.099 (34.4%)	0.189 (65.6%)
30-HZ-Irra-St	295.9	119.2 (40.3%)	176.7 (59.7%)	0.291	0.116 (39.9%)	0.175 (60.1%)

^a: BET surface area; ^b: t-plot of the micropore area; ^c: t-plot of the external surface area; ^d: Single-point adsorption total pore volume of pores at P/P_o = 0.98; ^e: t-plot of micropore volume

. Because of the impossibility of obtaining reliable information about micropore size using N₂ [17,18,19], the surface areas determined from this method can be considered to be approximate values for comparison purposes only. As shown in Table 3, a network of micropores and mesopores was generated in all the samples, in which the extent of meso-porosity growth at the expense of micro-porosity (i.e., micropore area and volume) was found to be different among the samples with different Al contents and/or different treatment histories. For the fresh H-ZSM-5 samples, the contribution of meso-porosity to the total porosity increased in (i) the Al-rich samples under the same thermal treatment conditions (e.g., 20-HZ vs. 30-HZ, 20-HZ-Irra vs. 30-HZ-Irra) and in (ii) the irradiated samples with the same bulk Al content (e.g., 20-HZ-Irra vs. 20-HZ, 30-HZ-Irra vs. 30-HZ). The contribution of meso-porosity to the total porosity also increased in the H-ZSM-5 samples with the same bulk Al content that were exposed to a greater number of high-temperature treatment steps (i.e., 823 K-3 h detemplation, 813 K-3 h H⁺ activation and 973 K-8 h steaming deactivation). Specifically, the 20-HZ sample, which was obtained via the detemplation of its parent Na-ZSM-5 followed by an H⁺ activation step, exhibited a 33.7 vol.% mesopore–66.3 vol.% micropore network, whereas the 20-HZ-St sample, which was exposed to the detemplation, the H⁺ activation, and the steaming deactivation steps, exhibited a network with up to 62.9 vol.% mesopores. Similarly, the 30-HZ sample exhibited a 27.5 vol.% mesopore network, which is much lower than that of the 30-HZ-St sample (a 65.6 vol.% mesopore network). A similar pattern was also observed among the irradiated samples with the same bulk Al content (e.g., 20-HZ-Irra vs. 20-HZ-Irra-St, 30-HZ-Irra vs. 30-HZ-Irra-St). It is noteworthy that the irradiation pre-treatment seems to greatly mitigate the mesoporous expansion in the Al-rich H-ZSM-5 samples while slightly diminishing the mesoporous expansion in the Si-rich H-ZSM-5 samples when they were subjected to severe steaming deactivation. In addition, the meso-porosity was introduced to the pore network of the irradiated samples without severe deterioration of the intrinsic micro-porosity, especially in the samples exposed to many rounds of high-temperature treatment. Indeed, the 20-HZ-Irra sample, which was obtained via the detemplation of its irradiated Na-ZSM-5 followed by the H⁺ activation step, exhibited a 50.3 vol.% mesopore–49.7 vol.% micropore network, whereas the 20-HZ-Irra-St sample, which was exposed to the detemplation, the H⁺ activation, and the steaming deactivation steps, exhibited a network with 51.6 vol.% mesopores. This also explained the higher specific surface areas of the irradiated samples compared to those of the non-irradiated samples. The decreased micropore volume with the mesopore growth might indicate the presence of extra-framework species that can form, aggregate and partially block micropores during deterioration and subsequent H⁺ activation and deactivation steps.

Specific information on the different textures among these samples can be further derived from the N₂ adsorption–desorption isotherms of the fresh and steamed H-ZSM-5 samples illustrated in Figure 3. All the samples exhibited a composite Type I/IV isotherm, which is associated with the adsorption–desorption process in materials containing a network of micro- and meso-pores. The Al-rich H-ZSM-5 samples exhibited a slight steep increase in the amount of adsorbed N₂ to the saturation point, which is indicative of the condensation of N₂ in funnel-shaped voids. The adsorption and desorption curves closely extended to the relative pressure of 0.4 in the N₂ isotherm measured for the 20-HZ sample, and loosely expanded to a P/P_o value of 0.4 (apparent Type H3 hysteresis loop) in the N₂ isotherm measured for the 20-HZ-St sample provide a supportive evidence that complex hierarchical structures existed in these non-irradiated, Al-rich samples and that considerable pore networking occurred in the 20-HZ-St sample. Different features were observed among the irradiated Al-rich samples. Specifically, the micropores (a sharp uptake of N₂ in the low-pressure region) connected to the uniform, narrow mesopores appeared in the P/P_o range of 0.4–0.5 and extended to the funnel-shaped mesopores (the steeply risen, narrow hysteresis loop to the saturation point) in the irradiated Al-rich H-ZSM-5 samples. The N₂ isotherms measured for the Si-rich H-ZSM-5 samples show a greater uptake of N₂ physisorption with a steep closure of the hysteresis loop (type H4) at a P/P_o of approximately 0.4. These observations indicate that the Si-rich H-ZSM-5 samples had obvious ink-bottle-shaped micro- and meso-pore networks and a slightly greater degree of microporosity than that of the Al-rich H-ZSM-5 samples. In other words, the void spaces within both the non-irradiated and irradiated Si-rich H-ZSM-5 samples were restricted, and had access to the external surface through narrow pore mouths.

The coordination of Al heteroatoms in the studied H-ZSM-5 samples was quantified using ²⁷Al MAS NMR, and the results are shown in Figure 4 and

Table 4. Chemical states of the Al heteroatoms in the fresh H-ZSM-5 and steamed H-ZSM-5 samples.

Sample	Total Al Content a (μmol/g)	Concentrations of Al in Hydrated Samples b (μmol/g)			% Al NMR-Visible c	% de-Al d	Concentrations of Al in Hydrated Samples After Exposure to NH3/He Flow at 373 K e (μmol/g)
		Al(IV) ~55 ppm	Distorted Al(IV) ~30 ppm	Al(VI) ~0 ppm			Al(IV) ~55 ppm	Distorted Al(IV) ~30 ppm	Al(VI) ~0 ppm
20-HZ	73.8	68.0 (99.3%)	0 (0%)	0.5 (0.7%)	93	7.8	68.7 (100%)	0 (0%)	0 (0%)
30-HZ	50.6	49.0 (99.7%)	0 (0%)	0.2 (0.3%)	97	3.1	50.2 (100%)	0 (0%)	0 (0%)
20-HZ-Irra	74.3	54.9 (83.9%)	1.6 (2.5%)	8.9 (13.7%)	88	26.2	61.9 (94.6%)	1.7 (2.5%)	1.9 (2.8%)
30-HZ-Irra	50.7	42.0 (91.6%)	0.4 (0.9%)	3.5 (7.5%)	90	17.0	44.9 (97.8%)	0.5 (1.0%)	0.5 (1.2%)
20-HZ-St	74.5	19.6 (57.3%)	6.4 (18.7%)	8.2 (24.0%)	46	73.7	20.0 (58.5%)	6.5 (19.0%)	7.7 (22.5%)
30-HZ-St	51.5	21.1 (79.0%)	1.3 (4.9%)	4.3 (16.1%)	52	60.0	22.0 (81.8%)	1.1 (4.1%)	3.8 (14.1%)
20-HZ-Irra-St	76.5	32.4 (71.7%)	2.6 (5.7%)	10.3 (22.7%)	59	57.6	39.1 (86.4%)	2.5 (5.5%)	3.7 (8.1%)
30-HZ-Irra-St	51.1	26.8 (83.2%)	0.3 (1.0%)	5.1 (15.8%)	63	47.5	30.27 (95.4%)	0.1 (0.4%)	1.3 (4.2%)

^a: Determined by ICP/MS; ^b,e: estimated from the integrated area of deconvoluted peaks relative to the total integrated area from ²⁷Al MAS NMR spectra; the number in parentheses corresponds to the fraction of each Al coordination number relative to the total number of NMR-visible Al; ^c: fraction of NMR-visible Al atoms relative to the total Al content; ^d: determined from the concentration of Al(IV) and total Al content in the hydrated samples

. The spectra of the 20-HZ and 30-HZ samples displayed a significant resonance centered at ca. 55 ppm and a small feature at ca. 0 ppm, which have been interpreted as framework tetrahedral Al atoms (FAl^IV) and extra-framework octahedral Al atoms (EFAl^VI), respectively [20,21,22,23,24,25,26,27,28,29]. From the mass balance point of the bulk Al content determined by the elemental analysis, a portion of the total Al content was found missing in the ²⁷Al MAS NMR spectra of the y-HZ-St, y-HZ-Irra, and y-HZ-Irra-St samples (y = 20, 30), even though all the measured samples were well hydrated. For the 20-HZ and 30-HZ samples, the concentration of framework tetrahedral Al was close to the bulk Al content determined by ICP/MS, suggesting that almost all the Al atoms in these two samples were incorporated into framework tetrahedral positions; thus, the degrees of dealumination were quite small (ca. 8% for 20-HZ and 3% for 30-HZ). The fraction of octahedral Al was greater in the Al-rich H-ZSM-5 samples. The e-beam irradiation post-synthesis treatment resulted in a decrease in the intensity of the 55 ppm signal accompanied by an increase in the intensity of the 0 ppm signal, as evidenced in the ²⁷Al MAS NMR spectra of the 20-HZ-Irra and 30-HZ-Irra samples. Moreover, a weak signal appeared at ca. 30–36 ppm in these spectra, which has typically been assigned to tetrahedral Al in a distorted environment (EFAl^IV) [27,28]. This reflects greater extents of dealumination in the 20-HZ-Irra and 30-HZ-Irra samples, which were ca. 26% and 17%, respectively. The highly distorted environments of Al species with high quadrupolar coupling constants caused broadening of the resonance of the Al species outside the detection limit of conventional 1D MAS NMR spectroscopy, thus explaining the high fraction of NMR-invisible Al in the ²⁷Al MAS spectra of the 20-HZ-St and 30-HZ-St samples. The same deactivation treatment of the irradiated 20-HZ-Irra and 30-HZ-Irra samples gave rise to less severe extents of dealumination, which were ca. 58% and 48%, respectively, in the 20-HZ-Irra-St and 30-HZ-Irra-St samples. It can, therefore, be inferred that the frameworks of 20-HZ-Irra and 30-HZ-Irra must be more resistant to the hydrolysis of framework Al-O bonds within these irradiated samples during the deactivation step. As shown in Figure 4 and Table 4, the ²⁷Al MAS NMR spectra obtained for the H-ZSM-5 samples upon NH₃ adsorption at 373 K confirmed that parts of the visible EFAl^VI and EFAl^IV were converted into tetrahedral symmetry. The presence of Al species that reversibly convert their coordination upon the adsorption of basic molecules such as NH₃ or pyridine at a sufficiently high temperature has been observed as a general feature of protonic zeolites and amorphous silica alumina (except amorphous Al-rich phases) by many research groups [27,28,29]. These authors attributed this phenomenon to the coordination of water molecules to the Al atoms in the vicinity of framework or non-framework Si-OH groups under partial hydrolysis conditions. As evidenced by the ²⁷Al MAS NMR spectra obtained for the H-ZSM-5 samples upon NH₃ adsorption at 373 K (Figure 4), the NH₃ adsorption at 373 K recovered the framework tetrahedral Al to a great extent at the expense of the extra-framework EFAl^IV and EFAL^VI in both the fresh and steamed forms of the irradiated samples (y-HZ-Irra, y-HZ-Irra-St, y = 20, 30). However, NH₃ adsorption at 373 K insignificantly affects the Al coordination states in the steamed, non-irradiated samples (y-HZ-St, y = 20, 30). This result suggested that a greater number of flexibly coordinated Al species existed in the vicinity of the framework/non-framework Si-OH sites in the irradiated H-ZSM-5 samples and readily changed their coordination to heal the defect sites and stabilize the framework/non-framework sites after dealumination.

The ability of the irradiated samples to heal the neighboring internal SiOH groups was verified via ¹H MAS NMR spectroscopy, in which changes in the proton concentration of various types of hydroxyl groups in the studied H-ZSM-5 samples were measured, which are displayed in

Table 5. Acidic properties of the fresh H-ZSM-5 and steamed H-ZSM-5 samples.

Sample	Framework Al Content a (μmol/g)	Concentration of OH Groups b (μmol/g)				Number of Acid Sites (Desorbed NH3) c (μmol/g)				Strong/ (Weak + Medium) Molar Ratio
		Bridging Si-OH-Al (4.4 ppm)	Framework-Connected Al-OH (3.0 ppm)	Silanol Si-OH (2.5 ppm)	Terminal OH (2.0 ppm)	Total Acidity	Weak Acid	Medium Acid	Strong Acid
20-HZ	68.0	58.8 (46.2)	10.3 (8.2%)	22.3 (17.8%)	33.9 (27.1%)	131.5	58.3 (44.4%)	13.3 (10.1%)	59.9 (45.5%)	0.84
30-HZ	49.0	39.0 (39.5%)	9.6 (9.7%)	30.2 (30.5%)	20.0 (20.3%)	98.6	48.5 (49.2%)	10.2 (10.3%)	39.9 (40.5%)	0.68
20-HZ-Irra	54.9	18.6 (17.6%)	35.5 (33.6%)	21.1 (20.0%)	30.4 (28.8%)	108.6	50.2 (46.2%)	32.8 (30.2%)	25.6 (23.6%)	0.31
30-HZ-Irra	42.0	13.1 (14.6%)	27.8 (31.1%)	29.2 (32.6%)	19.5 (21.7%)	85.3	40.0 (46.9%)	26.6 (31.1%)	18.8 (22%)	0.28
20-HZ-St	19.6	10.7 (15.7%)	9.4 (13.8%)	27.3 (40.1%)	20.8 (30.5%)	60.7	33.7 (55.5%)	11.5 (18.9%)	15.6 (25.6%)	0.34
30-HZ-St	21.1	7.8 (13.2%)	10.1 (17.2%)	26.5 (41.7%)	16.36 (27.9%)	48.7	28.2 (58.0%)	9.6 (19.6%)	10.9 (22.4%)	0.29
20-HZ-Irra-St	32.4	14.9 (18.9%)	16.8 (21.0%)	23.6 (29.6%)	24.5 (30.7%)	77.0	43.3 (56.2%)	16.9 (21.9%)	16.8 (21.8%)	0.28
30-HZ-Irra-St	26.8	9.1 (14.5%)	16.9 (26.9%)	20.3 (32.4%)	16.4 (26.1%)	54.5	31.4 (57.5%)	12.1 (22.2%)	12.1 (20.3%)	0.25

^a: The content of Al(IV) associated with the resolved ²⁷Al MAS NMR peak at ~55 ppm (Table 4). ^b: Estimations from the integrated area of deconvoluted ¹H MAS NMR peaks (Figure 5); the number in parentheses corresponds to the fraction of each OH group relative to the total OH concentration. ^c: Estimation from the integrated area of deconvoluted NH₃-TPD peaks (Figure 6); the number in parenthesis corresponds to the fraction of each acid site relative to the total number of acid sites. Weak acid (<493 K), medium acid (493–623 K), and strong acid (>623 K)

. Figure 5 shows the ¹H MAS NMR spectra (solid lines) of the studied H-ZSM-5 samples as well as the deconvolution profiles (dashed lines) of these spectra into peaks with their maxima at chemical shifts of 2.0 ± 0.1 ppm, 2.5 ± 0.4 ppm, 3.0 ± 0.1 ppm, and 4.4 ± 0.4 ppm. The first resolved peak at ca. 2.0 ppm is typically assigned to terminal OH groups, to which the nonbonded Al-rich (hydr)oxide clusters or completely dislodged Al cations (e.g., ⁺Al=O, AlOH²⁺) are coordinatively bonded in various coordination environments [30,31,32,33,34]. The second resolved peak at ca. 2.5 ppm is assigned to the internal Si-OH groups in silanol nests, which are formed during detemplation as defect sites or formed during the removal of framework Al [30,31]. The resolved peak centered at ca. 3.1 ppm is assigned to framework-connected extra-framework Al-OH species (i.e., the Al-OH groups on non-framework Al species pointing toward framework oxygen atoms or the abovementioned ‘flexible-coordination Al species’) [32,33]. The resolved peak centered at ca. 4.4 ppm is assigned to the bridging Si-OH-Al groups [34,35]. The ¹H MAS NMR results confirmed that the concentration of the OH groups from the strong acid sites (Si-OH-Al) decreased with the increasing extent of dealumination in the H-ZSM-5 samples. We noted a smaller concentration of OH groups from the strong acid sites (Si-OH-Al) than from the framework Al atoms in all the studied samples, suggesting that the remaining framework Al atoms were incorporated into their proton-compensating coordinated forms with the non-framework species. The content of framework Al coordinated with the non-framework species was found to be in line with the content of framework-connected Al-OH groups, as shown in Table 5. The data also reveal that a greater content of coordinated framework Al was found in the irradiated samples (y-HZ-Irra, y-HZ-Irra-St, y = 20, 30) than in the corresponding nonirradiated samples (y-HZ, y-HZ-St, y = 20, 30). As evidenced by the ²⁷Al and ¹H MAS NMR spectra, the extent of dealumination increased significantly in the order of y-HZ < y-HZ-Irra < y-HZ-Irra-St < y-HZ-St (y = 20, 30). Considering that the removal of one framework Al atom should create four internal SiOH groups, an accelerated increase in the concentration of the internal SiOH groups should be found in the order of y-HZ < y-HZ-Irra < y-HZ-Irra-St < y-HZ-St samples (y = 20, 30). However, the results showed that the contribution of the internal SiOH groups increased slightly in the irradiated samples (y-HZ-Irra, y-HZ-Irra-St) but notably increased in the non-irradiated samples (y-HZ-St). This result reflects the high healing rate of the silanol nests among the irradiated samples, in addition to the healing power of the strong acid sites (Si-OH-Al), as discussed for the ²⁷Al MAS NMR spectra.

The NH₃ desorption profiles of the studied H-ZSM-5 samples are shown in Figure 6. The overlapping peaks observed in the NH₃ desorption profiles of all the H-ZSM-5 samples were deconvoluted into different desorption sites, which were defined as weak (peak maxima < 493 K), medium (peak maxima within 493–623 K), or strong (peak maxima > 623 K) acid sites. Typically, weak acid sites represent sites of weak sorption strength, such as (i) silanol nests, which are associated with defect sites in the as-synthesized samples or formed by removal of Al atom(s) from the framework; and (ii) acid-extractable extra-framework species, including nonbonded Al-rich (hydr)oxide clusters or completely dislodged Al cations (e.g., ⁺Al=O, AlOH²⁺), which are coordinatively bonded to terminal OH groups in various coordination environments [35]. The strong acid sites are attributed to bridging hydroxyl groups in various forms of coordination between the Al atom and the neighboring silanol oxygen atom, which are listed in the order of decreasing acid strength as (i) bridging OH in the bonded form between the framework silanol oxygen atom and framework Al(IV) atom (i.e., bridging Si-OH-Al), (ii) bridging OH in the flexible coordination between the framework silanol oxygen and a neighboring Al atom, and (iii) pseudo-bridging OH between the silanol oxygen and a nearby Al atom [35]. Medium acid sites are typically assigned to any proton-compensating coordinated forms between the non-framework species and the framework atoms, such as the interaction of small non-framework species (e.g., ⁺Al=O cations) with an oxygen atom from the internal SiOH groups [36] or the Si-O-Al bond in the framework with residual Na⁺ to produce the Si-ONa-Al form [36]. These medium acid sites exhibit relatively strong acidic properties because of the substitution of monovalent non-framework species with more electropositive protons, as the charge-compensating cations in the zeolite framework have been found to increase the stability and activity of the silanol groups as well as stabilize the Si-O-Al bridges [31,34]. Based on the deconvoluted NH₃ desorption profiles of the studied H-ZSM-5 samples shown in Figure 6, the resolved peaks were shifted to lower temperatures in the Si-rich H-ZSM-5 samples. For the samples with the same content of Al, the resolved peaks were shifted to lower temperatures in the steamed H-ZSM-5 samples (y-HZ-St, y-HZ-Irra-St, y = 20, 30). A shift in the peak toward a lower temperature indicates lower thermal stability and hence a lower strength of the corresponding acid sites. Table 5 summarizes the total number of acid sites and the fraction of each acid site in the H-ZSM-5 samples studied. The data show that the number of total acid sites and the contribution of strong acid sites were greater in the Al-rich H-ZSM-5 samples than in the Si-rich H-ZSM-5 samples during the same treatment history. For the H-ZSM-5 samples with the same Al content, the number of total acid sites and the number of strong acid sites decreased in the steamed, nonirradiated samples compared to those in the steamed, irradiated samples. Treatment in pure steam at 973 K for 8 h sharply reduced the amount of adsorbed NH₃ for all the samples; however, more acid sites were reserved in the irradiated samples than in the non-irradiated samples. This reflects the fact that severe steaming-induced dealumination from the framework of the irradiated samples must be inhibited to some extent via the e-beam irradiation post-synthesis treatment. This is consistent with our observations in the ²⁷Al-, ¹H-MAS NMR spectra. Table 5 shows the amount of framework Al atoms in the studied H-ZSM-5 samples, which is assumed to be equal to the amount of tetrahedrally coordinated Al (viz. the resolved ²⁷Al MAS NMR peak at ca. 55 ppm), correlated well with the sum of the numbers of intermediate and strong acid sites in all the H-ZSM-5 samples. This is consistent with the typical identification of the medium and strong acid sites mentioned above and the results obtained from ¹H MAS NMR, where the framework Al atoms are supposedly incorporated into not only the strong (acidic) bridging Si-OH-Al sites but also the relatively strong (proton-compensating) coordinated forms with the non-framework species. As shown in Table 5, only a small fraction of the intermediate acid sites contributed to the acid population in the non-irradiated H-ZSM-5 samples (y-HZ, y-HZ-St, y = 20, 30). In contrast, a large fraction of the medium acid sites contributed to the acid population on the irradiated H-ZSM-5 samples (y-HZ-Irra, y-HZ-Irra-St, y = 20, 30). The contribution of weak acid sites was almost the same among the steamed samples with and without irradiation, specifically ca. 55.5–58.0% in the y-HZ-St samples (y = 20, 30) and ca. 56.2–57.5% in the y-HZ-Irra-St samples (y = 20, 30). Therefore, it is reasonable to state that the greater number of strong acid sites on the steamed, the more the irradiated H-ZSM-5 samples (y-HZ-Irra-St, y = 20, 30) restrained the acid strength to a medium level. The NH₃-TPD data showed reasonable agreement between the total number of acid sites determined from the total desorbed NH₃ and the total number of protons in various hydroxyl groups (Table 5).

The product yields resulting from the cracking of the BSR atmospheric residue over the Ecat base (100 wt.%) and over the blended Ecat/H-ZSM-5 samples in fresh and steamed forms were compared at a constant conversion of 70% to determine the catalytic performance of the H-ZSM-5 samples. The MAT results are presented in

Table 6. Products of AR cracking at a constant MAT conversion of 70% over 100 wt.% Ecat base and blends of 90 wt.% Ecat/10 wt.% fresh/steamed H-ZSM-5 samples.

Product Yield (wt.%)	100 wt.% Ecat (Base)	90 wt.% E-cat + 10 wt.% H-ZSM-5
		20-HZ	20-HZ-Irra	20-HZ-St	20-HZ-Irra-St	30-HZ	30-HZ-Irra	30-HZ-St	30-HZ-Irra-St
Dry gas	1.15	2.56	1.70	1.69	1.53	1.75	1.62	1.56	1.28
H2	0.11	0.09	0.09	0.13	0.06	0.14	0.15	0.12	0.15
C1	0.41	0.42	0.34	0.43	0.40	0.46	0.47	0.48	0.36
C2=	0.36	1.70	0.97	0.84	0.75	0.85	0.71	0.57	0.51
C2	0.27	0.35	0.29	0.29	0.32	0.30	0.29	0.39	0.26
LPG	12.26	29.53	29.76	21.26	23.02	22.61	23.38	16.21	18.57
C3=	3.36	9.61	11.87	4.12	7.41	8.33	10.96	4.03	6.51
C3	0.70	3.02	1.54	0.99	1.12	2.14	1.01	0.84	0.93
C4=	4.73	10.20	10.37	10.65	9.47	8.17	7.78	7.85	7.64
n-C4	0.71	2.01	1.73	1.62	1.12	1.43	1.13	0.72	0.72
i-C4	2.76	4.69	4.25	3.88	3.90	2.54	2.50	2.80	2.77
Gasoline	53.37	33.48	34.87	42.55	42.40	40.99	41.37	48.36	47.28
LCO	13.95	13.05	14.25	17.56	15.79	12.68	13.53	13.08	13.95
HCO	15.93	16.91	16.07	12.97	14.02	18.34	16.91	17.27	15.73
Coke	3.34	4.47	3.35	3.97	3.24	3.63	3.19	3.52	3.19
Light olefins a	8.45	21.51	23.21	15.61	17.63	17.35	19.45	12.42	14.66
C3= selectivity b	9.33	5.65	12.24	4.90	9.88	9.80	15.44	7.07	12.76
HTC c	0.73	0.66	0.58	0.51	0.53	0.49	0.47	0.44	0.46
CMR d	0.42	0.55	0.40	0.37	0.39	0.69	0.64	0.55	0.46
C3 olefinicity e	0.83	0.76	0.89	0.81	0.87	0.80	0.92	0.83	0.88
ΔC3=/Δgasoline	base	0.31	0.46	0.07	0.37	0.40	0.63	0.13	0.52

^a: C₂₌ to C₄₌; ^b: C₃₌/C₂₌; ^c: hydride transfer coefficient, (nC₄ + iC₄)/C₄₌; ^d: cracking mechanism ratio of dry-gas/iC₄; ^e: C₃⁼/C₃s

. To monitor the yield of the products and compare the catalytic performance among the samples, the C₃ olefinicity ratio (propylene/total C₃), propylene selectivity ratio (C₃₌/C₂₌), percent increase in propylene yield per unit decrease in gasoline yield (ΔC₃₌/Δgasoline), hydride transfer coefficient (HTC = butanes/butenes) and cracking mechanism ratio (CMR = dry gas/i-C₄) were also included. A yield comparison of the products, including dry gas, LPG, gasoline, and total light olefins, is presented in Figure 7. The specific yields of ethylene, propylene, and butenes along with the selectivity for propylene are shown in Figure 8. With a C/O ratio of ca. 2, a standard MAT conversion of 70.12% was obtained over the Ecat base, producing cracking products containing 1.15 wt.% dry gas, 12.26 wt.% LPG, 53.37 wt.% gasoline, 13.95 wt.% LCO, 15.93 wt.% unconverted HCO, and 3.34 wt.% coke. This product distribution is typical of an FCC catalyst containing a mesoporous matrix and microporous zeolite component. Large, bulky molecules of paraffins and aromatics in the AR feedstock were progressively converted to LCO-range hydrocarbons through the Lewis acid sites in the mesopores of the matrix via the formation of carbenium ions. The produced LCO hydrocarbons access the micropore domain of the zeolite component, where they are further converted to gasoline-bound hydrocarbons over protonic acid sites via competitive reactions, including β-scission, isomerization, cyclization, mono-molecular (protolytic Haag–Dessau) cracking, and bimolecular (hydride transfer) reactions [37]. The C₃ olefinicity and the ratios of HTC to CMR, which illustrate the importance of acid strength and field gradients in zeolites with different topologies [38], were found to be slightly different between the Ecat base and the E-cat/HZSM-5 blends, as shown in Table 6. Because H-ZSM-5 additives with the same MFI topology were tested in this study, the differences in the C₃ olefinicity, HTC, and CMR ratios could be attributed to the accessibility of the acid sites and the modification of acid strength by the e-beam irradiation treatment. As shown in Table 6 and Figure 7, the yield of dry gas (H₂, C₁-C₂) increased from 1.15 wt.% over the Ecat base to 1.28–2.56 wt.% over the Ecat/H-ZSM-5 blends, which was mostly associated with the increased yield of ethylene. For all the Ecat/H-ZSM-5 blends, the yield of the gaseous fraction, mainly LPG, increased significantly at the expense of the gasoline yield compared to that of the Ecat base. It is well established that the cracking of LCO-range (C₅ to C₁₂) carbenium ions by solid catalysts with a high density of strong acid sites can proceed via either the β-scission mechanism to produce C₃-C₄ alkanes or via the protolytic Haag–Dessau mechanism to produce dry gas [37,38,39]. The protolytic Haag–Dessau mechanism has been shown to be predominant when reactants enter small cavities or tortuous channels, extending their residence times long enough to close to strong acid sites so that they can be further cracked into dry gas by secondary cracking reactions [38,39]. Strong acid sites in solid catalysts have also been demonstrated to promote hydride transfer reactions and cyclization, leading to a decrease in the yield of light olefins and an increase in the coke factor [39]. Among all the samples, 20-HZ possesses the largest number of strong acid sites and its complicated interconnection of micro- and meso-pores, as mentioned above. This explains the significantly high cracking activity of the Ecat/20-HZ blend, from which the highest yields of dry gas (2.56 wt.%) and coke (4.47 wt.%) were obtained. In comparison to those of the Ecat base, a smaller yield of LCO and a higher yield of HCO accompanied by significant cracking of gasoline to LPG (29.53 wt.%) and a higher CMR (0.55) were obtained for the Ecat/20-HZ blend. This is certainly a consequence of the limited accessibility of the reactants to a great number of strong acid sites and the limited diffusion of the products from these reactive sites within the 20-HZ sample. In contrast, the Ecat/20-HZ-Irra blend produced lower yields of dry gas (1.70 wt.%) and coke (3.35 wt.%) and a comparably high LPG (29.76 wt.%) at the similar expense of gasoline yield (34.87 wt.%) as did the Ecat/20-HZ blend. Moreover, Table 6 and Figure 8 show that the Ecat/20-HZ-Irra blend rendered the highest yield of propylene (11.87 wt.%) among the Ecat blends. It is noteworthy that, from the acidic properties shown in Table 5, the 20-HZ-Irra sample possesses the highest density of medium acid sites, which are distributed in a comparable number of total acid sites and a well-interconnected pore network. This finding supports the essential contribution of medium acid sites to the enhanced selectivity of propylene. This finding is also in good agreement with the related research of C. Auepattana-aumrung et al. [36], who studied propylene production over medium acid sites of ZSM-5 and claimed that propylene selectivity is directly proportional to the density of medium acid sites.

The studied Al-rich H-ZSM-5 samples were deactivated upon pure steaming at 973 K for 8 h, after which their stability during AR cracking was tested. In comparison to the MAT data of the Ecat/20-HZ-St blend, the MAT data of the Ecat/20-HZ-Irra-St blend show greater stability and selectivity for propylene during the cracking reaction. At a constant MAT conversion of 70%, the Ecat/20-HZ-Irra-St blend yielded more LPG (23.02 wt.%) and light olefins (17.63 wt.%) but less dry gas (1.53 wt.%) and coke (3.24 wt.%) than the Ecat/20-HZ-St blend and the Ecat base. A reasonable explanation for the high yield of light olefins and the low yield of coke obtained over the Ecat/20-HZ-Irra-St blend is its low hydrogen-transfer activity in the highly accessible pore structure. This process may shorten the residence times of bulky intermediates and thus hinder secondary β-scission and hydride transfer reactions. Similar patterns of product yields were obtained for all the Ecat/Si-rich H-ZSM-5 blends, in which the blends containing the irradiated samples exhibited better propylene selectivity and greater stability during the AR cracking than the blends containing the nonirradiated samples. This is reasonable because the 30-HZ-Irra and 30-HZ-Irra-St samples possess more medium acid sites, which is consistent with the improvements in reactant and product diffusion through the porous network compared to those of 30-HZ and 30-HZ-St, respectively.

3. Discussion

As shown in Figure 2, all of the Al-rich and Si-rich sample crystals were inter-penetrative twinned between the (010) and (100) faces. Many published works have demonstrated that 90° intergrowth between the (h00) and (0k0) facets of ZSM-5 zeolite is a fundamental possibility due to the similarity of MFI unit cell parameters (a = 20.072 Å, b = 19.937 Å). An excellent example can be found in the work of M.B.J. Roeffaers et al. [40], in which the authors studied the intergrowth phenomena and pore accessibility in a set of five ZSM-5 samples with different sizes, aspect ratios, surface defects from ramps to intergrowths and degrees of crystal intergrowth. In our work, the larger crystal size of the Si-rich H-ZSM-5 samples contained 30% less Al. And additional elevated terraces were particularly observed on the Si-rich H-ZSM-5 exterior surfaces. These might be due to to a greater number of internal defects existing on the crystals, including but not limited to a greater degree of disordered orientation in the straight pore systems toward the exterior surface in the Si-rich H-ZSM-5 samples. For the Al-rich H-ZSM-5 samples, the smaller crystal sizes obtained from nucleation in the Al-rich synthesis gel contributed more less-stable, high-index phases than the larger crystals obtained from the Si-rich H-ZSM-5 samples.

A possible explanation for the steric hindrance of accessibility to the active sites within the micropore domain of the Si-rich H-ZSM-5 samples may be related to the greater degree of disorder in the straight pore systems toward the exterior surfaces in these samples, as already stated from the SEM images. As illustrated in Figure 2, the basal body of large Si-H-ZSM-5 crystals consists of many subunits that form in terraces, thus restricting parts of the straight pore systems to the open surface. This is consistent with the preferred expansion followed by the contraction of the sinusoidal channels in the 30-NaZ sample upon exposure to e-beam irradiation, as demonstrated in our previous work [14]. The difference in textural properties between the H-ZSM-5 samples treated with and without irradiation in this work could be understood by applying the same concept, which was described in our previous work, to explain the textural differences among the parent Na-ZSM-5 samples treated with or without irradiation [14]. As demonstrated in our previous work, there are two simultaneous impacts of e-beam irradiation at room temperature on the as-synthesized TPA⁺-Na-ZSM-5 samples, including (1) dehydration via water radiolysis and (2) partial removal of template cations via radical-induced oxidation. On the one hand, these impacts create partially open spaces within the confined crystals. Compared to the confined spaces in the as-synthesized Na-ZSM-5 samples, the partially open, sub-micron environments in the irradiated Na-ZSM-5 samples experienced less severe steaming during the subsequent high-temperature treatments. On the other hand, these impacts have driven local charge imbalances that enforce the rearrangement of T (Al, Si) atoms in the material confinement to regain coherence with the crystalline lattice in the distorted structures. Accordingly, these concomitant impacts on the irradiated Na-ZSM-5 samples led to the formation of hierarchical structures during the detemplation of the irradiated Na-ZSM-5 samples at 823 K without serious collapse of the porous structures, as evidenced by the enhanced specific areas and meso-porosities of the 20-NaZ-Irra and 30-NaZ-Irra samples. Considering that the H⁺ activation step was carried out in air for 3 h at 813 K, which is a lower temperature than that used in the detemplation step (823 K), we suppose that the coupling impacts induced by the e-beam irradiation treatment on improving the porous textures of the irradiated Na-ZSM-5 samples after detemplation might be considered a beneficial treatment step for stabilizing porous structures against thermal treatment. As such, structural water—the water molecules hosted within confined spaces of the framework—plays an important role in initiating radiolysis to form highly reactive radical species for the subsequent oxidation of template molecules, thus creating partially open spaces for mild steaming. L.R. Aramburo et al. [33] stated that the meso-porosity generated in H-ZSM-5 zeolite by mild steaming is open to crystal surfaces, while severe steaming tends to create more mesopores distributed within the bulk of crystals. Therefore, the generation of different micro- and meso-pore interconnectivities among the irradiated Al-rich H-ZSM-5 and Si-rich samples in this work can be understood as a result of differences in hydrophilicity and the preferred openings of the partially open voids facing toward the exterior surface.

According to Triantafillidis C.S. et al. [30] and Holzinger J. et al. [31], the mass of structural water in the H-ZSM-5 sample correlates well with the framework Al content, and the H-ZSM-5 zeolites with Si/Al ratios of ca. 20 and 30 contain an average of 4.5 ± 1 water molecules per framework Al site. Therefore, the decrease in the structural water contents in the order of y-HZ > y-HZ-Irra > y-HZ-Irra-St > y-HZ-St (y = 20, 30), as presented in Table 2, suggested a similar trend of variations in the framework Al contents within these samples. It could be inferred from these observations that the different extents of dealumination and micro-, meso-pore interconnectivity among the studied H-ZSM-5 samples might be associated with remarkable variations in the accessibility of acid sites with different natures, densities and strengths across the H-ZSM-5 framework. Indeed, the ²⁷Al MAS NMR spectra confirmed the above-mentioned inference of there being a dealumination trend among the studied samples during the H⁺ activation and steaming deactivation. In our previous work, we demonstrated that upon e-beam irradiation, the initial ionization and excitation of water molecules by primary high-energy electrons followed by a sequence of excitation, dissociation, ionization, and recombination along the chain of extra-framework–framework interactions increase the mobility of framework T-O-T bonds and produce secondary electrons within the samples via water radiolysis and collision with the atomic electrons in the samples [14]. These energetic secondary electrons successively lose energy and localize to the Na-ZSM-5 framework, especially in the polar Al-rich state, leading to the formation of excited states in the framework. In addition, loosely bonded species such as water molecules and extra-framework cations are particularly ionized as separate ions (e.g., H₃O⁺, Na⁺, (C₃H₇)_4-nNH_n⁺ (1 ≤ n ≤ 4)). During significant dehydration and detemplation, a local charge imbalance arises, forcing a large number of highly mobile ion clusters to occupy positions close to the framework oxygen due to a favorable electrostatic potential. This leads to local strains on the framework tetrahedral Al atoms, and these Al atoms eventually migrate to non-framework positions. However, these non-framework Al atoms are still in the vicinity of the framework Si-OH groups (Brönsted acid sites) due to the high electron affinity of the proton, thus giving rise to a locally distorted environment. These non-framework Al atoms may be present in various forms, ranging from charged species such as Al^(3-n)+(OH)_n (0 ≤ n ≤ 2) to basic species such as Al(OH)₃, Al(OH)₃(H₂O) and Al(OH)₄(H₂O)₂. A similar picture can also be envisioned for amorphous silica–alumina species, which are typically formed and immobilized by the zeolite framework inside narrow channels during complete hydrolysis upon steaming. Several concepts can be proposed to explain the synergetic power of flexibly coordinated Al species in stabilizing defect sites and retaining site activity to some extent, including, for example, pseudo-bridging between nearby non-framework Al species and protons at Brønsted acid sites or the hydroxylation–dehydration of neighboring internal SiOH groups into intact Si-O-Si bonds.

The detailed characterizations of the acidic properties and the Al coordination states in the studied H-ZSM-5 samples using the ²⁷Al-, ¹H-MAS NMR and NH₃-TPD methods revealed that the non-framework Al species that formed during the e-beam irradiation treatment of the parent Na-ZSM-5 samples stabilized the framework against dealumination during the following activation and deactivation steps. The coordination flexibility between the non-framework Al species and the framework Si-OH sites increased the strength of the original framework Si-OH-Al sites. Because of the irradiation-induced dealumination and consequential healing effect, a large fraction of the strong acid sites was inclined toward the medium acid sites; hence, the density of acid sites in the irradiated samples (y-HZ-Irra, y = 20, 30) was negligibly lower than that in the nonirradiated samples (y-HZ, y = 20, 30). Compared to the steamed samples without irradiation (y-HZ-St, y = 20, 30), the steamed samples with irradiation retained more acid sites.

The MAT data show that lower yields of dry gas, LPG, and light olefins were obtained from the Si-rich H-ZSM-5 blends than from the Al-rich H-ZSM-5 blends. However, higher a C₃₌ selectivity, C₃ olefinicity, and percent increase in propylene yield per unit decrease in gasoline yield (ΔC₃₌/Δgasoline) were obtained for the Si-rich H-ZSM-5 blends. These features can be attributed to the lower density and strength of the strong acid sites as well as the significantly high density of medium acid sites in the Si-rich H-ZSM-5 samples, as discussed earlier for the NH₃-TPD measurements. Lower coke factors were found in all the Ecat blends with the Si-rich H-ZSM-5 samples in comparison to the Ecat blends with the Al-rich H-ZSM-5. We noted smaller yields of LCO and higher yields of HCO in all the Ecat blends containing the Si-rich H-ZSM-5 samples than in those containing the Al-rich H-ZSM-5. These features indicate that the reactants must be limited to the micropore domain in the Si-rich H-ZSM-5 blends. This is also consistent with the realization that the Si-rich H-ZSM-5 samples have an obvious ink-bottle-shaped micro- or meso-pore network and a greater degree of disordered orientation of the straight pore systems toward the exterior surfaces, as discussed in the N₂ physisorption measurements. These MAT data suggest that with increasing Si/Al ratio, the H-ZSM-5 additives lost some of their cracking activity due to the constricted hierarchical pore network toward the exterior surface but gained more stability and selectivity for propylene due to the reserved medium acid sites.

4. Materials and Methods

Figure S1 shows the experimental procedure for synthesizing fresh H-ZSM-5 samples for FCC application. Na-ZSM-5 zeolites with theoretical Si/Al ratios of ca. 20 and 30 were first synthesized from a hydrothermal gel consisting of tetrapropylammonium bromide (TPABr, Sigma Aldrich, Burlington, MA, USA), boehmite (CATAPAL C1, Sasol Chemicals, Sandton, South Africa), fumed silica (Sigma Aldrich, Burlington, MA, USA), sodium hydroxide (NaOH, Prolabo, Chattanooga, TN, USA), and water. The obtained powders were labeled 20-NaZ and 30-NaZ, respectively.

Hierarchical Na-ZSM-5 zeolites with different levels of mesoporosity were then prepared via a conventional calcination route from dried 20-NaZ and 30-NaZ powders with and without e-beam irradiation treatment. The detailed settings of the e-beam irradiation can be found in reference [14]. With this setting, the sample surface was treated with a total dose of 80 kGy (1 kGy = 1 kJ/kg), which was subsequently combined after eight scans (10 ± 0.5 kGy per scan). Afterwards, the nonirradiated samples and irradiated samples were detemplated completely in air at 823 K for 3 h at a heating rate of 3 K min⁻¹.

To prepare fresh H-forms of ZSM-5 samples with different meso-porosities, the nonirradiated samples and the irradiated samples were ion-exchanged with 1 M NH₄NO₃ (98%, Fischer Scientific, Pittsburgh, PA, USA) solution. Proton forms of the samples were obtained by thermally treating the NH₄⁺-exchanged samples at 813 K for 3 h in air using a heating rate of 3 K min⁻¹. The obtained samples were labeled 20-HZ and 30-HZ (non-irradiated) and 20-HZ-Irra and 30-HZ-Irra (80 kGy irradiated).

To determine the catalyst stability in the FCC regenerator, steamed H-forms were prepared from the as-prepared H-ZSM-5 samples. The non-irradiated y-HZ samples and the irradiated y-HZ-Irra samples (y = 20, 30) were subjected to severe steaming (100% steam, 20 mL min⁻¹) in a fixed-bed reactor at 973 K and atmospheric pressure for 8 h, as described in the standard ASTM D4463-96 [41]. The obtained samples were labeled 20-HZ-St and 30-HZ-St (non-irradiated) and 20-HZ-Irra-St and 30-HZ-Irra-St (80 kGy irradiated). The samples that were not steamed were designated “fresh”.

Inductively coupled plasma–mass spectrometry (ICP-MS) and thermogravimetric analysis (TG/DSC) were used to determine the elemental compositions of the samples using a NexION^®2000 (Perkin Elmer, Shelton, CT, USA) and a LABSYS evo TGA/DSC at 1600 °C (Setaram Instrumentation, Caluire, France), respectively.

The crystalline structure of the samples was determined by X-ray powder diffraction (XRD) analysis using a Bruker AXS D8 diffractometer (Karlsruhe, Germany) with Cu Kα radiation, λ = 1.5418 in the 2-theta range of 6–80° at a scan rate of 1.2° min⁻¹.

The textural properties of the samples were characterized by N₂ adsorption/desorption isotherms at 77 K using a TriStar 3020 Micromeritics unit (Norcross, GA, USA). Prior to the isotherm measurements, the samples were evacuated under vacuum (1.0 × 10⁻⁶ mbar) at 323 K for 5 h. The surface areas and volumes of the pores were calculated using the t-plot and the BJH models, as described in the standard ASTM D4365-13 [42].

The morphology of the zeolite samples was observed by scanning electron microscopy (SEM). The images were taken over the Pt-coated samples on an S-4800 FESEM (Hitachi, Chiyoda, Tokyo, Japan) with a cold field-emission gun operating at 10 kV and 10 μA.

Magic angle spinning (MAS) NMR spectra of ¹H and ²⁷Al were obtained using a Bruker Advance AMX-500 NMR (Karlsruhe, Germany) spectrometer at room temperature with a magnetic field of 11.75 T. For ¹H MAS NMR spectra, all samples were activated in vacuum at 673 K for 15 h to remove any adsorbed water from the ambient environment. The ¹H chemical shift was referenced to adamantane (C₁₀H₁₆, Sigma Aldrich, Burlington, MA, USA). An excitation pulse with a power level of 6 dB, length of 1.6 μs and relaxation time of 2 ms was applied for recording all the ¹H spectra. For ²⁷Al MAS NMR measurements, all the samples were hydrated over a saturated MgCl₂ solution. The ²⁷Al chemical shift was referenced to a saturated Al(NO₃)₃ solution. The ²⁷Al MAS NMR spectra were recorded at a resonance frequency of 104.01 MHz (0.6 μs pulse width, 250 ms relaxation time) with ca. 2400 scans. The ²⁷Al MAS NMR spectra were also recorded for the samples that were treated with a flow of He containing 10% NH₃ at 373 K for 1 h.

The density and strength of the acid sites in the H-ZSM-5 samples were characterized via temperature-programmed desorption (NH₃-TPD) on a modified AMI-1 unit (Altamira Instruments, Cumming, GA, USA) equipped with a TCD detector. The sample (100 mg) was loaded into a U-tube reactor with a 1/4-inch diameter. The sample was pretreated in a high-purity He stream (99.999% purity) at 873 K for 1 h at a flow rate of 30 mL min⁻¹ to remove any absorbed volatile impurities. The sample was then cooled to 373 K and subjected to NH₃ adsorption in a 30 mL min⁻¹ flow of 10% NH₃/He gas for 30 min. After exposure to the NH₃ adsorbate, the sample was purged with 30 mL min⁻¹ of flowing He to desorb the weakly physisorbed NH₃. To start the TPD, the sample was heated from 373 K to 923 K at a heating rate of 5 K min⁻¹, and the desorbed NH₃ was continuously monitored. The area under the generated curves was used to calculate the total NH₃ uptake (mmol g⁻¹) based on loop calibration.

The H-ZSM-5 samples were tested as FCC catalyst additives in a short contact-time micro-activity testing unit (SCT-MAT) licensed by Grace Davison (Columbia, MD, USA) with a systematic error of less than 2%. The complete details of the SCT-MAT method are given in reference [43]. Experiments were carried out over a physical mixture of 90 wt.% of an equilibrium catalyst (Ecat, Binh Son Refinery, Dung Quat Economic Zone, Quang Ngai, Vietnam) and 10 wt.% of the H-ZSM-5 additive using an atmospheric residue (AR, Binh Son Refinery, Dung Quat Economic Zone, Quang Ngai, Vietnam) as the feedstock. The properties of the Ecat and the AR feedstock are listed in Table S1 and Table S2, respectively. The catalyst was tested at 793 K under constant nitrogen flow and autogenous pressure with a contact time of 12 s. In this work, different catalyst-to-oil (C/O) mass ratios were used to maintain a constant MAT conversion of ca. 70% over the studied mixtures. To maintain a constant volume of the catalyst bed and to help improve the reaction efficiency, glass beads with particle sizes in the range of 180–250 μm were mechanically mixed with the catalyst mixture. To secure a homogenous blending of the glass beads, the E-cat and the studied catalysts with different particle sizes, a special glass vial was designed by Vinci Technologies, where vertical glass baffles are attached to the vial wall to improve mixing efficiency. Therefore, consistency and uniformity of the materials within each individual batch and across multiple batches have been achieved, and the obtained MAT results was accurate and reproducible. Gaseous products (dry gas and LPG) were analyzed using a 7890A GC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a 27 m long, 0.32 mm diameter RGA capillary column (Agilent) and FID/TCD detectors. The liquid products were cut into three different boiling temperature ranges: gasoline (C5+, 494 K), light cycle oil (LCO, 494–633 K) and heavy cycle oil (HCO, +633 K). The weight percentage of the liquid products was determined by a simulated distillation GC equipped with a 10 m long, 0.53 mm diameter capillary column and an FID detector according to ASTM D2887-14 [44]. The coke content in the catalyst was determined by a CS600 (LECO, St. Joseph, MI, USA) according to the standard guide ASTM E1915-13 [45]. The product yield (Y_i, wt.%) was defined as the mass ratio between the product and the feedstock. The standard MAT conversion was defined as 100% − (Y_HCO + Y_LCO).

5. Conclusions

We have demonstrated that the post-synthesis treatment of Al-rich and Si-rich TPA⁺-Na⁺-ZSM-5 zeolites via e-beam irradiation at room temperature results in the formation of highly crystalline, pure H-ZSM-5 catalysts with different micro- and meso-pore interconnectivities to the exterior surface. The ²⁷Al MAS NMR, ¹H MAS NMR, and NH₃-TPD results revealed that the irradiation-induced impacts of dehydration via water radiolysis and the partial removal of TPA⁺ cations via radical-induced oxidation not only create partially open spaces within the confined crystals but also enforce the migration of framework Al atoms to locally distorted non-framework environments. Upon dehydration and detemplation at 823 K, mild steaming in partially open sub-micron environments results in the formation of non-framework Al species in the vicinity of the framework Si-OH sites, where these non-framework Al species establish flexible coordination with the framework oxygen to regain cohesion with the crystalline lattice. The presence of flexibly coordinated Al species manipulates the strong acidic sites and increases the framework stability against hydrothermal dealumination. By preserving the density of acid sites and restraining strong acid sites within the modestly strong acidity range, the irradiated H-ZSM-5 catalysts retain their cracking activity and improve their stability and selectivity for propylene.