m-Xylene Isomerization over IWW Zeolite with a Three-Modal Pore Structure: The Effect of Crystal Morphology

m-xylene isomerization over IWW zeolite with a three-modal pore structure: The effect of crystal morphology Emad Shamma, Alica Seidlov´ a , Subhajyoti Samanta , Michal Mazur, Maksym Opanasenko, Mariya Shamzhy * Department of Physical and Macromolecular Chemistry, Faculty of Science, Charles University, Hlavova 2030/8, Prague 12843, Czech Republic ARTICLE INFO Keywords: Zeolite Xylene isomerization IWW Germanosilicate Crystal morphology Active site Shape-selectivity ABSTRACT Xylene isomerization is a key zeolite-catalyzed petrochemical process for the production of p-xylene, a highly demanded intermediate in the polymer industry. While MFI-type zeolites are widely used in industry as shapeselective catalysts, xylene isomerization also serves as a benchmark reaction for evaluating the shape-selectivity of new zeolite catalysts in relation to their porosity. In this study, IWW zeolite with a three-modal pore network of isolated 8and 12-ring pores intersected by 10-ring channels was investigated for m-xylene isomerization, with a focus on how both the multidimensional pore system and crystal morphology affect catalytic performance. IWW zeolites were synthesized as germanosilicates with platelet-like and needle-like crystals, functionalized by post-synthetic Ge-to-Al substitution, and tested in gas-phase m-xylene isomerization in comparison with reference zeolite catalysts containing unimodal 8-, 10-, or 12-ring channels. Compared to MFI with a similar concentration of acid sites, Al-IWW catalysts exhibited higher m-xylene conversion across a wide range of WHSV values (4.4 – 40 h −1 ), regardless of crystal morphology. Needle-like Al-IWW crystals achieved para-selectivity comparable to that of MFI, while platelet-like Al-IWW outperformed MFI in p-xylene yield at short contact times (WHSV =40 h −1 ). STEM analysis confirmed that the 10-ring channels are aligned along the length of the needlelike crystals, promoting shape-selective p-xylene formation. In contrast, the 12-ring channels running along the extended dimension of the platelet-like crystals facilitate the diffusion of reactant and product molecules to and from the active sites. All in all, the integration of multi-sized pores and tunable crystal morphology in IWW zeolites may offer a promising strategy for balancing selectivity and activity in p-xylene synthesis. 1. Introduction Xylene isomerization is an essential petrochemical process for the selective production of p-xylene, a key building block in the polymers industry. The highly demanded para-isomer typically constitutes only 25 % of the xylene mixture obtained after hydrotreating C9 aromatics in refineries. To increase the p-xylene content, the mixture undergoes selective isomerization over solid acid catalysts, typically aluminosilicate zeolites, with appropriately sized channel entrances and voids which restrict product distribution. Zeolite catalysts are widely used in industry for xylene isomerization due to molecular diffusion control inside their micropores, abundance of acid sites, and long-term stability. Industrially used shape-selective zeolite catalysts (e.g., MFI, MOR, EUO) favor the preferential formation of p-xylene and play a key role in mxylene transformation, which involves a complex reaction network, including (Fig. 1): (A) monomolecular or bimolecular isomerization, that enables the interconversion of xylene isomers; (B) monomolecular dealkylation, that yields toluene and light hydrocarbons through the removal of a methyl group from m-xylene; and (C) bimolecular disproportionation, that produces a mixture of toluene and trimethylbenzenes (TMBs) through methyl group transfer between two aromatic molecules [1−3]. MFI zeolite with a unimodal 3D system of 10-ring pores is the most widely used catalyst for m-xylene isomerization due to its high paraselectivity. It contains two types of intersecting channels: straight channels (0.53 ×0.56 nm) and sinusoidal channels (0.51 ×0.55 nm), with a channel intersection of 0.86 nm. These structural features are key to both product and transition-state shape-selectivity. The kinetic diameter of p-xylene (~0.58 nm) is smaller than those of mand o-xylene (~0.68 nm), facilitating its preferential diffusion through the MFI pores [4]. Adsorption studies with pure xylene isomers show a diffusivity trend of para >ortho >meta in MFI, with p-xylene exhibiting the highest * Corresponding author. E-mail address: [email protected] (M. Shamzhy). Contents lists available at ScienceDirect Catalysis Today journal homepage: www.elsevier.com/locate/cattod https://doi.org/10.1016/j.cattod.2025.115563 Received 27 June 2025; Received in revised form 20 August 2025; Accepted 8 September 2025 Catalysis Today 462 (2026) 115563 Available online 10 September 2025 0920-5861/© 2025 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). diffusion coefficient. Moreover, the transition state complex for the meta-to-ortho isomerization is bulkier (0.67 nm) than that for meta-- to-para isomerization (0.62 nm), further favoring p-xylene formation due to lower steric constraints. As a result, MFI zeolite typically delivers a para/ortho xylene ratio exceeding 2, and a low disproportionation-to-isomerization (Dis/Iso) ratio of ~0.01, with unimolecular pathways dominating the reaction mechanism [3−6]. As such, xylene isomerization also serves as a benchmark reaction for assessing the shape-selectivity of new zeolite catalysts in relation to their porosity [1]. In this way, p-xylene selectivity over MFI zeolite has been further enhanced by tailoring crystal morphology. For instance, chain-like crystals with an extended b-axis showed a 10 % increase in p-xylene selectivity compared to conventional coffin-shaped crystals. However, the narrow pores of MFI limit molecular diffusion and accessibility to active sites, which can promote undesired dealkylation reactions and reduce overall p-xylene yield [4,7,8]. These limitations have prompted the development of alternative zeolite catalysts. One such example is EUO zeolite, designed for industrial m-xylene isomerization. It features 1D 10-ring channels and special 12-ring side pockets, achieving higher para/ortho ratio than MFI (2 vs. 1.35) under similar conditions [9]. In combination with other components, EUO forms the basis of the commercial Oparis® catalyst family, which is widely used for the isomerization of ethylbenzene and m-xylene [8]. Hong and Jones [10,11] compared the performance of zeolites with unimodal 1D 10-ring and 12-ring pore systems to that of MFI. The 1D 10-ring MTT zeolite exhibited a higher para/ortho ratio (3.4) than the 3D 10-ring MFI (2.5) at 20 % conversion. In contrast, increasing pore size, as in 1D 12-ring AFI and 3D 12-ring IFR, resulted in a para/ortho ratio nearly two times lower than that of MFI (approaching ~1). Moreover, the Dis/Iso ratio was lower for MTT compared to AFI and IFR, due to the stronger restriction of bimolecular reactions in the narrower 10-ring pores. MOR zeolite with bimodal 2D system of 8and 12-ring channels shows higher m-xylene conversion than 3D 10-ring MFI. Its larger 12ring pores enable easier diffusion but also promote bimolecular disproportionation and transalkylation reactions, increasing the formation of toluene and trimethylbenzenes (TMBs), and thereby reducing pxylene selectivity. In contrast, MFS zeolite, with a 2D network of 8and 10-ring pores, effectively suppresses bulky byproduct formation and shows a Dis/Iso ratio roughly half that of MFI (~0.005 vs. 0.01) [10,12]. More recently, m-xylene isomerization has been employed to evaluate the shape-selectivity of new UTL-derived isoreticular zeolite catalysts with tailored multimodal porosity, in comparison to MFI [13]. Zeolite PCR with 8and 10-ring pores achieved higher para/ortho ratios (4.5) than MFI (1.4). Overall, the literature survey suggests that designing zeolite catalysts with a combination of different pore sizes represents a promising strategy for optimizing both shape-selectivity and diffusion control in xylene isomerization. Beyond pore size, crystal morphology-induced shape-selectivity, arising from variations in crystal size and shape, has been demonstrated to impact reaction pathways in both xylene isomerization and related acid-catalyzed transformations. The IWW zeolite, a three-modal germanosilicate with a system of 8-, 10-, and 12-ring pores and tunable crystal morphologies, offers a unique opportunity to integrate pore architecture and crystal morphology effects [14,15]. Its 10-ring channels are associated with promoting monomolecular reactions and high p-xylene selectivity, while the 12-ring pores can facilitate the diffusion of reacting molecules. Our study investigates the under-explored zeolite IWW as a promising catalyst for m-xylene isomerization, while proposing strategies to balance product selectivity and catalytic activity. The performance of IWW was systematically compared with zeolites containing unimodal 8-, 10-, or 12-ring pore systems, allowing us to isolate the effects of pore connectivity and size, with MFI serving as the industrial benchmark. This comparison underscores the role of multi-sized micropore network in directing activity and selectivity of IWW catalysts. Furthermore, we demonstrate that tuning the crystal morphology of IWW offers an effective approach to optimize the trade-off between p-xylene selectivity and yield. Collectively, our findings suggest that integrating multi-sized pores with tunable morphology in IWW zeolites represents a promising strategy to balance selectivity and activity in p-xylene synthesis. 2. Experimental part The zeolites investigated in this study are listed in Table 1. They include IWW zeolites featuring a three-modal pore system with 8-, 10-, and 12-ring channels and either needle- (Al-IWW-needles) or plateletlike (Al-IWW-platelets) crystal morphologies, as well as reference zeolites with unimodal pore systems (i.e., having uniform sizes of the pores), either non-intersecting (1D) or intersecting (3D). *BEA (CP811E) and MFI (CBV8014) zeolites were provided by Zeolyst in ammonium form. Al-IWW-needles and Al-IWW-platelets catalysts were synthesized Fig. 1. Scheme for the mechanisms of m-xylene transformation: (A) monomolecular isomerization, (B) monomolecular dealkylation, and (C) bimolecular disproportionation. E. Shamma et al. Catalysis Today 462 (2026) 115563 2 Table 1 Zeolites studied in this work. E. Shamma et al. Catalysis Today 462 (2026) 115563 3 by the post-synthesis Al-for-Ge substitution in hydrothermally synthesized germanosilicate zeolites, IWW-needles and IWW-platelets, respectively. Other zeolites under study were prepared in aluminosilicate form by hydrothermal crystallization. 2.1. Synthesis of zeolite catalysts CHA 3D 8-ring zeolite was synthesized following the procedure reported in the previously published work [16], using N,N, N-trimethyl-1-adamantammonium hydroxide (TCI, 25 % in H 2 O) as SDA and reaction mixture with the molar composition of 1 SiO 2 : 0.35 Na 2 O: 0.05 Al 2 O 3 : 0.35 SDA: 18 H 2 O. 1.6 g of sodium hydroxide (VWR, 99.2 %) was first dissolved in the SDA hydroxide solution (32.96 g of SDA in 0.77 g of water). After the complete dissolution, 0.22 g of aluminium hydroxide (Acros Organics, extra pure) was added, and the mixture was left to stir for 30 min. Lastly, 16.67 g of colloidal silica HS-40 (Sigma Aldrich, 40 % in H 2 O) was added dropwise to the mixture, and the reaction mixture was left to stir for 90 min at room temperature. 0.10 g of CHA previously synthesized according to Zones [17] was added to the mixture as seeds. The reaction mixture was transferred into a 90 mL Teflon-lined autoclave and left to crystallize for 12 days in an oven under rotation (60 rpm) at 160 ◦C. The solid product was isolated by filtration, washed with distilled water and dried overnight at 60 ◦C. The synthesis yielded 3.0 g of CHA zeolite. The zeolite was calcined at 550 ◦C for 5 h with a heating rate of 1 ◦C/min under air flow. ESV 1D 8-ring zeolite was synthesised following the procedure reported by B. J. Campbell et al. [18], using N,N-dimethylpiperidinium hydroxide as SDA, which was prepared by adding dimethylamine to 1, 5-dibromopentane (1:1) in ethanol under reflux and then the bromide salt obtained was filtered, washed with ethanol, and dried. The bromide aqueous solution was ion exchanged with Ambersep® 900(OH) anion exchange resin (Sigma Aldrich, 0.8 mmol of solid per 1 g of resin) to the hydroxide form of the SDA. The molar composition of the reaction mixture was 1.0 SiO 2 : 0.3 Na 2 O: 0.04 Al 2 O 3 : 0.2 SDA: 40 H 2 O. In a Teflon beaker, 22.48 g of the hydroxide SDA was stirred with 7.04 g of water, and then 1.03 g of aluminium sulfate hexadecahydrate (Lechner, 99 %) was added to the solution. The mixture was left to stir for 30 min to dissolve the aluminium salt. Then, sodium silicate solution (37 % in H 2 O, 18.80 g) was added dropwise, and the reaction mixture was left to stir for 1 h at room temperature. The mixture was then transferred into a Teflon-lined 90 mL autoclave. The mixture was crystallized for 14 days in an oven under rotation (60 rpm) at 170 ◦C. The solid product was isolated by filtration, washed with distilled water and dried at 60 ◦C. The synthesis yielded 2.7 g of ESV zeolite. The zeolite was calcined at 680 ◦C for 5 h with a heating rate of 1 ◦C/min under air flow. TON 1D 10-ring zeolite was synthesised following a modified synthesis process reported by K. Hayasaka et al. [19], using a reaction mixture with the molar composition of 112 SiO 2 : 10 K 2 O: 1 Al 2 O 3 : 24.5 SDA: 3265 H 2 O, where 1,6-diaminohexane (Alfa Aeser, 98 %) was used as the SDA. In a Teflon beaker under stirring, 0.69 g of the SDA was dissolved in 11.50 g of distilled water. After a complete dissolution of the SDA, 0.33 g of potassium hydroxide (Lechner, PA) and 0.15 g of aluminium sulfate hexadecahydrate were added, and the mixture was left to stir to dissolve all components. Subsequently, 4.0 g of colloidal silica AS-40 (Sigma Aldrich, 40 % in H 2 O) was added dropwise. The reaction mixture was left to stir for 90 min at room temperature. Afterwards, the mixture was transferred to a Teflon-lined 25 mL autoclave. The mixture was crystallized in 5 days in an oven under rotation (60 rpm) at 160 ◦C. The solid product was isolated by centrifugation, washed with distilled water and dried at 60 ◦C. The synthesis yielded 3.1 g of TON zeolite. The zeolite was calcined at 550 ◦C for 5 h with a heating rate of 1 ◦C/min under air flow. AFI 1D 12-ring zeolite was synthesized according to Ref. [20] using N, N,N-trimethyl-1-adamantammonium hydroxide as SDA and reaction mixture with the molar composition of 30 SiO 2 : 3 NaOH: 1 Al(OH) 3 : 3 SDA: 954 H 2 O. In a Teflon beaker under stirring, 7.78 g of the SDA was added to 47.00 g of distilled water, and then sodium hydroxide (0.38 g) and subsequently aluminium hydroxide (0.24 g) were added to the solution. The mixture was left to stir for 30 min. Lastly, 5.50 g of CAB-O-SIL M5 fumed silica (Acros Organics) was added under vigorous stirring, and the reaction mixture was left to stir for 90 min at room temperature until it became homogenous. The reaction mixture was then transferred into a 90 mL Teflon-lined autoclave and left to crystallize for 8 days in an oven under rotation (60 rpm) at 160 ◦C. The solid product was isolated by centrifugation, washed with distilled water and dried at 60 ◦C. The synthesis yielded 2.2 g of AFI zeolite. The zeolite was calcined at 550 ◦C with a heating rate of 1 ◦C/min under air flow for 5 h. The catalytically active H-form of CHA 3D 8-ring , ESV 1D 8-ring , TON 1D 10-ring and AFI 1D 12-ring aluminosilicates were obtained by ion-exchange with ammonium nitrate (Sigma Aldrich, ≥99 %) and then calcination. The zeolite powders were mixed with 1 M ammonium nitrate solution (1 g/100 mL). This mixture was left to stir for 8 h at room temperature before filtration. The ion-exchange process was repeated five times. Afterwards, the samples were collected, dried at 60 ◦C, and calcined at 350 ◦C for 3 h with a heating rate of 1 ◦C/min under air flow. IWW-needles and IWW-platelets zeolites were synthesized as germanosilicates using 1,5-bis-methylpyrrolidinium pentane as SDA. The preparation of the SDA is described elsewhere [14]. The molar composition of the reaction mixture was varied as follows: (y) SiO 2 : (1-y) GeO 2 : 0.25 SDA: 15 H 2 O, where y=0.90 resulted in crystallization of IWW-needles and y=0.66 resulted in the formation of IWW-platelets. For the IWW-platelets, 0.78 g of the germanium source (GeO 2 , Sigma Aldrich, 99.99 %) was first dissolved in 4.40 g of the SDA solution, containing 3.37 g of water, and then 3.18 g of TEOS (Thermo Scientific, 98 %) was added to the mixture as the silicon source. To synthesize the IWW-needles, 0.21 g of GeO 2 was dissolved in the aqueous solution of 4.40 g of the SDA, containing 3.37 g of water, before adding 4.10 g of TEOS (Thermo Scientific, 98 %) to the mixture. 70 mg of a previously prepared IWW-needles sample was added to the synthesis mixture as seeds. The mixture was left to stir for enough time to evaporate the alcohol before charging the final gel into a 25 mL autoclave. The synthesis continued for ten days at 175 ◦C under rotation with a speed of 20 rpm. Later, the synthesis was quenched, and the solid product was collected by centrifugation, washed with water and dried at 60 ◦C. The described synthesis procedure yielded 1.4 g of IWW-platelets and 1.6 g of IWW-needles. The as-synthesized IWW samples were calcined at 550 ◦C for 5 h with a heating rate of 1 ◦C/min under air flow. Al-IWW-needles and Al-IWW-platelets were prepared by postsynthesis Al-for-Ge substitution in IWW-needles and IWW-platelets zeolites according to Ref. [21]. For that, the parent germanosilicate was treated with 1 M aluminium nitrate (Thermo Scientific, 98 %) solution (1 g/100 mL) at 96 ℃ for four days. The solid product was collected by filtration, washed with excess water and 0.1 M HCl, and then dried at 60 ◦C for one day. The dry solid was calcined at 450 ◦C for 4 h. 2.2. Characterization Powder XRD patterns were collected on a Bruker D8 Advance diffractometer equipped with a LYNXEYE XE-T detector and a Cu-K α radiation generator (λ=1.54 Å). Zeolite samples were homogenized using a mortar and pestle and loaded into plastic holders for measurements. Diffractograms were obtained in the 2θ range between 3 ◦and 40 ◦. Physisorption measurements were performed on a Micrometrics 3Flex volumetric Surface Area Analyzer using 0.1 g of the catalyst. Each sample was outgassed at 110 ◦C for 1 h, with a heating rate of 1 ◦C/min, using the Micrometrics Smart Vac Prep system. Subsequently, the sample was activated at 250 ◦C for 8 h with the same heating rate. A total of 40 data points with an equilibration interval of 180 s were collected for each nitrogen adsorption isotherm at –196 ◦C (77 K) in the p/p◦range between 0 and 1. The micropore volume and external surface area were obtained using the t-plot method [22,23]. The total pore volume was E. Shamma et al. Catalysis Today 462 (2026) 115563 4 calculated at p/p◦=0.95. SEM images were obtained using JEOL JSM-IT800 and a Thermo Fisher Scientific Scios 2 DualBeam FIB-SEM microscopes using secondary and backscattered electron detectors. The sample was ground and spread over carbon tape. Images were acquired at an accelerating voltage of 1–3 kV, a beam current of 10–50 pA, and a working distance up to 10 mm. The size of crystals was estimated by measuring and averaging the dimensions of at least seven crystals using the ImageJ software. Elemental composition analysis was conducted by EDS using the FEI Quanta 200 F X-ray detector integrated into the SEM system. Data collection was performed at an accelerating voltage of 15 kV, a beam current of 10 nA, and a working distance of 10 mm. In situ FTIR spectroscopic measurements were performed using a Nicolet iS50 spectrometer equipped with a DTGS detector in Transmittance mode. Spectra were collected in the range 4000–400 cm −1 with a resolution of 4 cm −1 with 64 scans. The sample was homogenized by grinding and formed into a self-supported wafer with a density between 8 and 12 mg/cm 2 . The wafers were activated under high vacuum (≈6×10 −4 Torr) at 450 ◦C for 4 h with a heating rate of 5 ◦C/min. Trideuteroacetonitrile (AN), pyridine (Py) and 2,6-di-tert-butylpyridine (DTBPy) were used as probe molecules and were degassed by freezingpump-thaw cycles prior to adsorption over activated sample wafers. AN was adsorbed at room temperature at a partial pressure of 3.5 Torr for 20 min. Subsequently, it was desorbed at room temperature for 20 min. Py was adsorbed at 150 ◦C at a partial pressure of 3.5 Torr for 20 min. Afterwards, it was desorbed at 150 ◦C, 250 ◦C, 350 ◦C and 450 ◦C for 20 min. DTBPy was adsorbed at 150 ◦C at the equilibrium vapor pressure of the probe molecule for 20 min and then desorbed at the same temperature for 60 min. Spectra were obtained (i) before activation, (ii) after activation, (iii) after adsorption of the probe molecule and (iv) after desorption of the physisorbed probe molecule. Concentrations of Brønsted (BAS) and Lewis (LAS) acid sites were calculated by integrating the intensities of specific IR bands. For AN, bands at 2298 cm −1 , 2310 cm −1 and 2326 cm −1 were analysed. Molar absorption coefficients used for evaluation of acid site concentrations were: ε (2298 cm −1 , BAS) =2.05 cm/ μ mol and ε (2310, 2326 cm −1 , weak and strong LAS) =3.60 cm/ μ mol [24]. For pyridine, bands at 1455 cm −1 and 1546 cm −1 were analysed and the following molar absorption coefficients were used: ε (1455 cm −1 , LAS) =1.71 cm/ μ mol; ε (1546 cm −1 , BAS in 10-ring) =1.09 cm/ μ mol and ε (1546 cm −1 , BAS in 12-ring) =1.12 cm/ μ mol [25]. For DTBPy, the band at 1615 cm −1 was analysed with the corresponding integrated molar absorption coefficient ε (1615 cm −1 , BAS) =5.3 cm 2 / μ mol [26]. 2.3. Catalytic evaluation Gas-phase m-xylene isomerization experiments were carried out on a Micromeritics Microactivity Effi microflow experimental unit equipped with a stainless-steel fixed-bed reactor (internal diameter 11 mm). Prior to the experiment, zeolite powders were pressed, grained, and sieved to obtain particle sizes between 200 and 500 μ m. The granulated zeolite catalysts (0.22 – 2.00 g depending on weight hour space velocity (WHSV) values, Table S1) were diluted with carborundum (particle size 500 μ m) to reach a consistent reactor bed volume across all samples. The reaction was performed under atmospheric pressure and at 350 ◦C using N 2 as the carrier gas. Before the reactions, the catalyst was activated at 450 ◦C for 90 min under an airflow of 200 mL/min. Subsequently, the reactor was flushed with N 2 and cooled down to the reaction temperature. Liquid m-xylene was pumped from a reservoir to an evaporator, with a flow of 0.17 mL/min, using an HPLC pump. Gaseous m-xylene was blended with N 2, and this mixture was fed to the catalyst bed. The flow rates of N 2 and m-xylene were maintained at 8.77 g.h −1 and 170 mL.min −1 , respectively. To prevent condensation of the reactant and products, connecting lines and valves were heated to 145 ◦C. Typically, the duration of the experiment was 215 min of time-onstream (T-O-S). The long-term performance of Al-IWW-platelets and Al-IWW-needles catalysts was studied for 24 h TOS. Samples were collected at 30-minute intervals, with the zeroth sample taken at 5 min T-O-S. Samples of the reaction mixture were analysed using an online-connected Agilent 8890 Gas Chromatograph (GC) equipped with a heated 6-port sampling valve, VF-WAXms column (30 m ×0.25 mm ×1.00 µm) and an FID detector using N 2 as carrier gas and Air/H 2 as flame ignitor. FID signals (area) were used for the catalytic activity evaluation. The values of WHSV were tuned in the range of 4.4–40h −1 by varying the amount of the zeolite catalyst (Table S1) and calculated using Eq. (1): WHSV = ρ m−xylene •Fm−xylene mzeolite (1) where ρ m-xylene is the density of m-xylene, [g/mL]; F m-xylene is the flow rate of m-xylene, [mL/h]; and m zeolite is the mass of the zeolite catalyst in the reactor bed, [g], respectively. Carbon balances were determined from GC-FID analyses. The FID response was normalized to carbon atom number, assuming nearconstant sensitivity per carbon. The molar flow was obtained from the GC peak area, multiplied by its number of carbon atoms to yield the carbon contribution. The overall carbon balance was then calculated as the ratio of total carbon in the effluent to the carbon fed with m-xylene. In all catalytic tests, the balance was 92–99 %, except in the cases (carbon balances of 85–86 %, for *BEA and AFI zeolites, Table S1) where framework-dependent coke formation led to partial carbon loss. Conversion of m-xylene, X m-xylene , was calculated using Eq. (2): Xm−xylene =1−Am−xylene ∑Ai,norm. •100 % (2) where A m-xylene is the integrated area of the GC signal for m-xylene and A i, norm. is the integrated area of the GC signal for every compound in the reaction mixture, normalized to the molar response of m-xylene based on the product carbon number. Yield of a product, Y p , was determined using Eq. (3): Yp=Ap,norm. ∑Ai,norm. •100 % (3) where A p, norm is the integrated area of the GC signal for a product and A i, norm. is the integrated area of the GC signal of every compound in the reaction mixture, normalized to the molar response of m-xylene based on product carbon number. p-Xylene selectivity, S p-xylene , was determined using Eq. (4): Sp−xylene =Yp−xylene Xm−xylene •100 % (4) where Y p-xylene is the yield of p-xylene and X m-xylene is the conversion of m-xylene. The ratio of disproportionation to isomerization was determined using Eq. (5): Dis/Iso =YDis. YIso. =Ytoluene +YTMB Yp−xylene +Yo−xylene (5) where Y toluene is the yield of toluene, Y TMB is the yield of TMB, Y pxylene is the yield of p-xylene, and Y o-xylene is the yield of o-xylene. 3. Results and discussion 3.1. Structure, texture, and acidity of the zeolite catalysts Powder XRD patterns revealed the structural identity and phase purity of the studied zeolites, both reference zeolite catalysts with unimodal system of 8-, 10or 12-ring channels (Fig. S1) and multimodalpore IWW samples. Fig. 2-A presents the diffractograms of the calcined E. Shamma et al. Catalysis Today 462 (2026) 115563 5 IWW samples with peak positions identical to the reference patterns from [27]. The broader diffraction lines in the IWW-needles sample stem from the smaller particle size (vide infra) and suggest a distinct crystal morphology of the sample. The distinct features observed in the XRD patterns of IWW zeolites with different crystal morphologies are presented in Fig. S2 of the Supporting Information and discussed therein. The PXRD patterns of the parent IWW samples after Al incorporation maintained the positions of characteristic reflexes but showed a decrease in their intensity. The result suggests the preservation of the structure ordering of IWW zeolite at a decrease in the framework density due to the non-equivalent substitution of Ge for Al, as previously reported in Ref. [21,28,29]. The crystal morphology of the prepared zeolites was examined by scanning electron microscopy (SEM) imaging (Fig. S3 and Fig. 3). Zeolite IWW was obtained in two distinct crystal morphologies, depending on the Si/Ge ratio in the synthesis mixture (Fig. 3). While the IWW-platelets zeolite synthesized from a synthesis gel with Si/Ge =2 showed small platelet-shaped crystals (0.43 ×0.21 ×<0.10 μ m, Table 2) stacked on top of each other, the IWW-needles sample obtained from a synthesis mixture with Si/Ge =10 demonstrated piles of very thin needle-shaped crystals (0.48 ×<0.01 ×<0.10 μ m). SEM images were recorded at various scales to confirm the uniformity of the samples and to show particle dimensions. Fig. 3B,C compares the size and dimensionality of IWW platelets and needles at the same scale: the needles appear as thinner, elongated 2D particles, whereas the platelets exhibit larger 3D growth. At higher magnification (Fig. 3D), individual needles are clearly visible, confirming the absence of a large fraction of dense, bulky aggregates and substantiating the needle-like morphology of the sample. SEM images of the Al-IWW samples after aluminium incorporation indicate that IWW-platelets and IWW-needles preserved crystal morphology, and there was no distortion or fragmentation of the crystals (Fig. 3E, F). Textural properties of the zeolite samples were determined by N 2 physisorption (Fig. S4 and 2-B). IWW germanosilicates display type-I isotherm, which confirms their microporous nature (Fig. 2-B). For IWW-needles and Al-IWW-needles samples, a steep increase in adsorbed volume at p/p◦>0.8 was observed, which likely reflects an intercrystalline adsorption due to the loosely packed, small, needle-like crystals. In contrast, this feature is absent in the isotherm of IWW-platelets samples, which consists of larger, more densely packed crystals that minimize interparticle voids. A minor increase in the micropore volume in IWW samples upon the Ge-for-Al substitution, was observed independently of crystal morphology (Table 2). This increase is consistent with previous reports and is attributed to the non-equivalent replacement of Ge by Al, which slightly increases the framework void volume [29]. Chemical analysis supports this interpretation as the parent IWW germanosilicates have Si/Ge ratios of 4.7 (IWW-platelets) and 12.5 (IWW-needles), while the aluminated samples show lower Ge content with Si/Ge ratios of 21 and 31, and Si/Al ratios of 12 and 14 for Al-IWW-platelets and Al-IWW-needles, respectively (Table 2). Variation in the Si/Ge ratio not only influenced the morphology of the IWW crystals but also directed their crystal growth. Scanning transmission electron microscopy (STEM) revealed the spatial arrangement of perpendicular 12and 10-ring channels within individual crystals (Fig. 4). IWW-platelets sample exhibits 10-ring channels penetrating the largest crystal plane (i.e., 0.43 ×0.21 μ m, Fig. 4-A). When oriented along the (210) plane, these platelets display well-defined crystalline IWW layers with a characteristic d-spacing of 1.2 nm, consistent with reference [30]. Unlike the platelet-like crystals, each crystal of IWW-needles sample grows along the 10-ring channels, while shorter 12-ring channels extend from the top and bottom surfaces (Fig. 4-B). These findings are essential when evaluating the role of pore size and crystal morphology in m-xylene isomerization (Section 3.2). The concentration of acid sites, both BAS and LAS, was assessed by FTIR spectroscopy of adsorbed basic probe molecules of varying kinetic diameters (Fig. S5): AN (~0.38 nm), Py (~0.54 nm), and DTBPy (~0.79 nm). AN was used to quantify the total acid site concentration, as this smaller probe molecule can access the acid centres in all 8-, 10and 12ring pores. Py has a size comparable to that of the targeted product in the xylene isomerization reaction, p-xylene, and was used to probe acid sites relevant to catalysis. DTBPy was employed to assess acid sites located on the external surface of zeolites with unimodal 8and 10-ring pore systems, as its bigger size prevents diffusion into <12-ring micropores. The concentration of acid sites accessible for AN ranges from 0.06 to 0.70 mmol/g across the samples, with BAS (active sites in xylene isomerization [1,31,32]) fractions between 39 % and 88 %, indicating evident Lewis acidity. Notably, the number of acid sites accessible for AN was lower than the total Al content, as reflected by the ΣAN/Al (Table 3), which ranges from 0.04 to 0.80. This discrepancy is commonly observed in literature; for instance, the results of Zholobenko et al. suggest (BAS +LAS)/Al ratios between 0.47 and 0.82 for various zeolites [25]. In our study, this ratio was significantly lower for zeolites with 1D unimodal pores (0.04 for ESV 1D 8-ring , 0.28 for TON 1D 10-ring , 0.16 for AFI 1D 12-ring ) than that for zeolites with 3D unimodal channel systems (0.58 for CHA 3D 8-ring , 0.80 for MFI 3D 10-ring , 0.72 for *BEA 3D 12-ring ) and multimodal-pore IWW catalysts (0.51 for both Al-IWW-needles and Al-IWW-platelets). As we can exclude the inaccessibility of BAS for AN (the Fig. 2. A) PXRD patterns and B) N 2 physisorption isotherms of IWW samples: IWW-needles, Al-IWW-needles, IWW-platelets, and Al-IWW-platelets. E. Shamma et al. Catalysis Today 462 (2026) 115563 6 Fig. 3. SEM images of IWW with two distinct crystal morphologies: A, B) IWW-platelets at low and high magnification, C, D) IWW-needles at low and high magnification, E) Al-IWW-platelets, and F) Al-IWW-needles. Table 2 Crystal size, elemental composition and textural properties of the IWW samples a . Catalyst Si/Al Si/Ge Crystal size, μ mV mic (cm 3 /g) V tot (cm 3 /g) S ext (m 2 /g) IWW-platelets *n.d. 4.7 0.43 ×0.21 ×<0.10 0.17 0.24 81 IWW-needles *n.d. 12.5 0.48 ×<0.10 ×<0.10 0.16 0.39 107 Al-IWW-platelets 12 21 0.43 ×0.21 ×<0.10 0.24 0.32 70 Al-IWW-needles 14 31 0.48 ×<0.10 ×<0.10 0.17 0.41 110 *n.d.- not detected. a Information on the physicochemical characteristics of the zeolites with unimodal pore systems is provided in Table S2, Supporting Information. E. Shamma et al. Catalysis Today 462 (2026) 115563 7 disappearance of 3610 cm −1 band of bridging ≡Si-(OH)-Al≡groups after AN adsorption is observed for all the samples), the results are most probably related to variations in the spatial distribution and coordination environment of aluminium, which affect the strength and accessibility of LASs, rendering some inactive or undetectable by AN under the applied conditions. Py accessed the same number of acid sites as AN in zeolites AFI 1D 12ring and *BEA 3D 12-ring with the unimodal 12-ring pores, and in the multimodal Al-IWW samples of both morphologies, while no acid sites were accessible to Py in zeolites ESV 1D 8-ring and CHA 3D 8-ring with Fig. 4. STEM images of: A) IWW-platelets and B) IWW-needles. Table 3 Concentration of BASs and LASs in Al-IWW catalysts and reference zeolites with unimodal pore systems, as determined by FTIR spectroscopy of adsorbed probe molecules of variable sizes. Catalyst c(Al) a (mmol/g) c(AN) b (mmol/g) c(Py) b (mmol/g) c(DTBPy) b (mmol/g) BAS LAS Σ Σ AN /Al BAS LAS Σ Σ Py /Al BAS BAS DTBPy / Σ BAS ESV 1D 8-ring 1.54 0.04 0.02 0.06 0.04 *n.d. *n.d. *n.d. *n.d. *n.d. *n.d. CHA 3D 8-ring 1.20 0.47 0.23 0.70 0.58 n.d. n.d. n.d. n.d. 0.03 0.06 TON 1D 10-ring 0.60 0.15 0.02 0.17 0.28 0.29 0.02 0.31 0.52 0.01 0.02 MFI 3D 10-ring 0.46 0.25 0.12 0.37 0.80 0.26 0.11 0.36 0.78 0.03 0.12 AFI 1D 12-ring 0.64 0.05 0.06 0.11 0.17 0.08 0.07 0.15 0.23 0.04 0.67 *BEA 3D 12-ring 0.43 0.14 0.17 0.31 0.72 0.20 0.15 0.35 0.81 0.19 1 Al-IWW-platelets 1.20 0.30 0.31 0.61 0.51 0.29 0.23 0.53 0.44 0.30 1 Al-IWW-needles 1.12 0.22 0.36 0.57 0.51 0.32 0.27 0.59 0.53 0.23 1 a – based on EDS results. b – based on in situ FTIR spectroscopic results. * n.d.- not detected E. Shamma et al. Catalysis Today 462 (2026) 115563 8 unimodal 8-ring channels (Table 3). In turn, based on DTBPy adsorption, all 8and 10-ring zeolites with unimodal pore systems showed negligible concentration of external BASs (0.01 – 0.03 mmol/g), representing <12 % of the total Brønsted acidity determined with AN. On the other hand, all BAS detected by smaller probe molecules were also accessible to DTBPy in the 3D 12-ring *BEA 3D 12-ring zeolite and the multimodal Al-IWW zeolites. A clear difference in acid site accessibility to DTBPy was observed between 12-ring zeolites with 1D and 3D pore systems: only 67 % of BAS were accessible to DTBPy in AFI 1D 12-ring compared to 100 % in *BEA 3D 12-ring . The comprehensive acidity characterization of the studied zeolites reveals two important trends: 1) The total BAS concentrations vary widely across the zeolites (0.04–0.47 mmol/g). Multimodal-pore Al-IWW catalysts exhibit BAS concentrations comparable to the benchmark MFI 3D 10-ring zeolite (0.22–0.30 mmol/g), while 1D zeolites such as ESV 1D 8-ring and AFI 1D 12-ring showed much lower values (0.04–0.05 mmol/g). This variation in active site numbers, along with the differences in pore size and dimensionality, is expected to strongly influence m-xylene conversion in the isomerization reaction. 2) The concentration of external BAS is negligible in 8and 10-ring zeolites (<0.03 mmol/g; <12 % of total BAS), suggesting a minimal role in catalysis. Overall, according to their basic characteristics, the prepared series of zeolites (Table 1) constitutes a representative and well-defined set of catalysts suitable for addressing the role of the multi-size pore architecture and crystal morphology of IWW zeolite catalysts in the acidcatalyzed transformation of m-xylene. 3.2. Catalytic performance in m-xylene isomerization The catalytic performance of the Al-IWW and reference zeolites was analyzed based on the m-xylene conversion, p-xylene selectivity and yield, Dis/Iso, and p-xylene/o-xylene ratios. MFI with unimodal 3D system of 10-ring pores was used as the core reference for comparison because of its industrial significance. A direct comparison between zeolites listed in Table 1 at similar WHSV does not clearly reveal the effect of pore size or channel architecture on key catalytic performance characteristics in m-xylene isomerization. The key parameters of the catalytic performance (i.e., p-xylene selectivity, p-xylene/o-xylene and Dis/Iso ratios) are strongly dependent on the conversion and vary widely among the studied catalysts due to significant differences in BAS concentrations (0.03–0.47 mmol/g, Table 3). Specifically, at comparable WHSV (4.4 – 8.8 h −1 ), m-xylene conversion ranged from 0 (for ESV 1D 8-ring ) to 48 % (for MFI 3D 10-ring , Fig. S6-A). Therefore, to concentrate on the effect of pore structure, the values of WHSV were tuned, yielding a narrower range of conversion between 20 % and 35 %, for better comparison (Fig. S6-B). Based on the analysis of the catalytic performance of zeolites with unimodal pore systems under optimized conditions and taking into account pore size-catalytic performance correlations known from the literature, the influence of pore size of zeolite catalysts on the selectivity in m-xylene isomerization can be summarized as follows: •Although the 8-ring pores of ESV 1D 8-ring and CHA 3D 8-ring are generally considered too small for m-xylene (0.73 nm) to enter [5,33, 34], ESV 1D 8-ring shows minimal activity (1–2 % conversion), whereas CHA 3D 8-ring achieves 58 % p-xylene selectivity at 20 % conversion. Given limited external BAS (Table 3), the performance of CHA 3D 8-ring can be attributed to its dynamic framework flexibility, enabling molecular diffusion and promoting monomolecular isomerization, as supported by a low Dis/Iso (0.04) and high p/o-xylene ratio (1.5), unlike ESV 1D 8-ring , whose rigid pores and scarce accessible BAS result in poor activity (Fig. S6-A; Fig. S7). 8-ring zeolites (e. g., CHA 3D 8-ring ) were unexpectedly active and selective, achieving up to 58 % p-xylene selectivity with negligible side reactions. •Both 1D and 3D 10-ring zeolites, TON 1D 10-ring and MFI 3D 10-ring , showed the highest p-xylene selectivity of 63 % (in the range of 20–35 % conversion) and the lowest Dis/Iso (0.01), confirming monomolecular isomerization predominance, though only MFI 3D 10ring maintains stable activity without deactivation. •Despite lower BAS (0.14 vs. 0.25 mmol/g) and shorter contact time, 3D 12-ring *BEA 3D 12-ring shows similar m-xylene conversion to MFI 3D 10-ring but the lowest p-xylene selectivity (44 %) due to its large pores promoting bimolecular pathways, as evidenced by the highest Dis/Iso (~0.30). In contrast, 1D 12-ring AFI 1D 12-ring , with comparable pore size, surpasses *BEA 3D 12-ring in selectivity (51 % vs. 44 %) after 35 min, rising to 60 % with time as coke formation likely narrows pores and shifts the reaction toward monomolecular isomerization (Dis/Iso drops from 0.18 to 0.11). 12-ring zeolites (e.g., *BEA 3D 12-ring ) showed the lowest p-xylene selectivity due to the prevalence of bimolecular side-reactions, such as disproportionation. When comparing the catalysts of interest (IWW) with the wellestablished standard (MFI), both plateletand needle-like Al-IWW samples consistently showed greater m-xylene conversion than MFI across all the tested contact times (Fig. 5). At long contact time (WHSV 4.4 h −1 ), Al-IWW-platelets and Al-IWWneedles exhibited similar conversions (~65 %), notably exceeding that of MFI zeolite (~47 %). Given their comparable BAS concentrations (0.26–0.32 mmol/g, Table 2), the higher conversion of Al-IWW catalysts can be attributed to the enhanced molecular transport through their 12ring pore system. Decreasing the contact time (WHSV increase 4.4 →20 →40 h −1 ) resulted in a steady decline in m-xylene conversion in MFI (47 → 35 → 32 %) and Al-IWW-needles (65 → 55 → 44 %) (Fig. 5) in relevance with the common explanations: (i) shorter residence time limits the extent of the reaction, (ii) diffusion constraints hinder access to active sites, and (iii) insufficient time to reach equilibrium lowers observed conversion [10,35,36]. Notably, in comparison to Al-IWW-needles and MFI, Al-IWWplatelets maintained m-xylene conversion in the range 65-55 % across the same WHSV range, suggesting more stable performance under reduced contact time (Fig. 5). Further insights into the long-term stability of Al-IWW-platelets and Al-IWW-needles were obtained from catalytic tests extended to 24 h on stream (Fig. S8). The results suggest that Al-IWW-platelets deactivate more slowly than Al-IWW-needles, with deactivation constants of 0.020 and 0.025 h −1 (considering the 1st order deactivation model). This stability may theoretically originate from (i) higher intrinsic activity of BAS in Al-IWW-platelets leading to faster reaction rates; (ii) higher accessibility of BASs leading to their better utilization at high throughput due to more favourable pore arrangement, minimizing internal diffusion limitations. Hypothesis (i) was ruled out by pyridine-FTIR thermodesorption results (Fig. S9), which showed similar acid strength distributions in IWW catalysts of both studied morphologies. In contrast, hypothesis (ii) was in line with FTIR-DTBPy results, showing higher concentrations of accessible BAS in Al-IWW-platelets (Table 3). Furthermore, STEM results reveal the pore arrangement in the Al-IWW samples, which may result in more efficient mass transport in the Al-IWW-platelets. Indeed, the platelet-shaped crystals (0.43 ×0.21 ×<0.10 µm) offer more balanced pore orientation and diffusion paths compared to the needle-shaped Al-IWW crystals, where 10-ring channels run along the long edge of a crystal (~0.48 µm), and 12-ring pores are confined to much shorter dimensions (<0.1 µm). This anisotropy in the needle morphology likely restricts diffusion and contributes to the observed decline in conversion at higher WHSV. While providing benefits for m-xylene conversion at short contact time, the platelet-like morphology of Al-IWW was less efficient for achieving high para-selectivity. Morphology-related differences in pxylene selectivity between the Al-IWW catalysts are apparent in Fig. 6-A. At similar conversion levels (44–55 %), MFI outperforms both Al-IWW E. Shamma et al. Catalysis Today 462 (2026) 115563 9