Influence of binder concentration in zeolitic ZSM-5/bentonite 3D-printed monoliths manufactured through robocasting for catalytic applications

In this work, a 3D printing method, robocasting was utilized to manufacture zeolite ZSM-5-based woodpile monolith catalysts of approximately 10-mm diameter, using bentonite clay as binding matrix. The effect of three different binder concentrations, in the 40–60 wt.% range, on the rheological, physicochemical, and mechanical properties was examined. The rheometer measurements showed that the printing pastes have identical shear thinning behavior and demonstrate sufficient storage modulus, irrespective of the binder concentration. The printed monoliths had high BET surface areas and porosity. The results showed that the ZSM-5 crystals retained their porous structure, textural characteristics, and crystalline structure during the additive manufacturing process. Pyridine FTIR measurements demonstrated reduced total acidity and number of Brønsted acid sites in the final specimens due to the dilution with the bentonite powder. However, the acidity reduction was roughly proportional to the binder concentration, signifying that the ZSM-5 crystallites also retain their acidity during the robocasting printing. Finally, the mechanical reliability of the thermally treated monoliths was determined by calculating the Weibull modulus values through linear regression of the Weibull equation. The increase in the binder concentration increased the compression strength by a factor of 4.5 and achieved superior mechanical reliability.


Introduction
Zeolites are highly porous crystalline aluminosilicates formed by SiO 4 and AlO 4 tetrahedra. These materials have a wide range of applications as sorbents, ion exchangers in detergents and catalysts in the oil refining and petrochemical industry [1]. With regard to their catalytic functionalities, the preference over other materials, such as disordered porous aluminosilicates, metal-based supported and unsupported catalysts and multi-component oxides, is mostly determined by their increased surface area, well-defined porous structure, and high Brønsted acidity [2]. Moreover, tailored zeolite frameworks with specific pore width, channel system and architecture can be exploited in shape-selective catalysis by acting as a filter that allows molecules of only certain sizes and shapes to pass through [3]. Other less common industrial applications of zeolites include aerospace molecular decontaminations [4] and cleaning liquid nuclear waste [5].
Out of the more than 190 different natural and synthetic aluminosilicates, ZSM-5 is one of the most frequently used catalysts, especially for the cracking of heavy hydrocarbon molecules to more valuable, lighter fuel fractions in crude oil refineries. ZSM-5 has straight channels (5.6 Å × 5.6 Å) interconnected through sinusoidal channels (5.5 Å × 5.6 Å). The channels intersections generate 8.5 Å cavities [6]. This distinctive structure, along with its unique acidic, shapeselective, and thermal stability properties, has established the ZSM-5 zeolite as an important catalyst and sorbent for liquid gas separations.
Pure zeolite crystals can be used without additives, in the form of powder, in packed bed reactors [7] with the cost of increased pressure drop and poor mass transfer. Therefore, in many industrial applications, zeolites are shaped to achieve tailored monolithic structures with specified dimensions, optimal for minimizing pressure drops, maintaining uniform flow distribution and improving mass transfer phenomena. The last decades extrusion has been the main manufacturing procedure to produce such industrial scaled zeolite catalysts in the shape of pelletized extrudate forms or monoliths [8]. The most important disadvantage of pellets, beads, and extrudates is again the large pressure drop they generate at high flow rates and the presence of mass transfer limitations because of slow diffusion of molecules to the core of the particles. Decreasing the pellet size allows to reduce mass transfer limitations, but leads in turn to larger pressure drops. Monoliths for reactors are defined as single pieces of porous materials that contain various types of interconnected or separated channels [9]. Numerous scientific papers and reviews [10,11] discuss the potential advantages of monolithic structures that can overcome most of the previously mentioned limitations. Nevertheless, these types of structures are far more complex in terms of manufacturability, final dimensions, reliability, and mechanical strength. Molds must be manufactured to vary the structural geometry on a case-by-case basis, driving up the manufacturing cost and restricting the versatility of extruded monolith geometries.
The rapid development of 3D printing techniques offers a facile alternative to conventional extrusion of structured monoliths. The ability to tailor shape and size of fabricated parts is a great advantage that enables modulating the catalyst's macroscopic aspects. A variety of different additive manufacturing methods has recently emerged, enabling the construction of highly sophisticated materials such as shape memory alloys [12] fabricated through wire-based directed energy deposition, biodegradable polymers produced through stereolithography [13], and hierarchically porous structures fabricated through robocasting [14]. Although each method presents its own strengths and weaknesses, they are all based on the same principals, where the three-dimensional object is designed through a computer-aided design software, followed by a slicing procedure that reproduces the part layer by layer. This approach allows the fabrication of monoliths with different pore sizes, cross sections, and wall thicknesses, thus maximizing the macroscopic surface and improving the hydraulic behavior and the mechanical properties of catalysts. Robocasting, also called direct ink writing or solid freeform fabrication, is considered an ideal 3D printing technique for processing highly complex catalytic materials, which are typically multi-component with well-specified morphology, structure, textural and physicochemical properties. It consists of building scaffolds or three-dimensional objects layer-by-layer by extruding a continuous rod through a tip guided by a computer-assisted positioning system. It allows the use of a very large range of materials, since inks or pastes can be produced by combining materials of high importance with dispersants, binders, and plasticizers. Several researchers have applied the direct fabrication technique of robocasting for preparing ceramic and zeolitic monoliths, e.g., Al 2 O 3 , TiO 2 , SiO 2 , ZSM-5, and HY, for use as catalysts or catalyst supports [15]. Li et al. prepared 3D-printed ZSM-5 monoliths via a secondary growth method and evaluated their performance in the methanol-to-olefins (MTO) reaction [16]. They showed that the transformation of the zeolite powder to the monolithic structure increased the mesoporosity and moderated the strong zeolite acidity. This in turn led to significant increase in the selectivity to light olefins and substantially higher stability than the powder counterparts. Another work from the same group also demonstrated the superiority of 3D-printed HZSM-5 structured monoliths in the catalytic conversion of methanol to dimethyl ether (DME) [17]. Moreover, Lefevere et al. showed the significant impact of the employed binders and the monoliths' architecture on the catalytic properties of ZSM-5 for the methanol-to-olefins reaction [18].
An important parameter for the smooth operation of industrial reactors that should not be overlooked is the mechanical strength of shaped catalysts [19]. During the reaction, as well as during the reactor loading and unloading process, monoliths are subjected to various thermal and mechanical stresses which could result to failures. These failures formulate fines or fragments that can undermine the homogenous flow distribution, intensify pressure drops, and even result to environmental problems due to the release of small particulates in the flue gas emissions [20]. Furthermore, the catalyst's mechanical failure is a common issue that might lead to process termination and need for catalyst replacement more frequently than catalyst deactivation [21]. Catalysts consisting of oxide supported metals and oxide mixtures present brittle behavior due to their highly porous structure that often presents defects and discontinuities such as crystal edges, voids, and protuberances. These defects cause micro cracks under concentrated stress, and local tensions are largely higher than the ones distributed in a catalyst bulk body [22]. Concentrated stresses are the primary cause of fracturing, leading to mechanical failure of the catalyst. Variations in shape, size, position, and orientation of these defects generate a wide range of mechanical strength data, which are more closely subjected to the Weibull distribution than to a normal or log-normal distribution [23,24].
Given the importance of monolithic structures in catalytic processes and the possibilities and flexibility that 3D printing offers in varying the shape and material composition, 3D printing can be employed for the development of a new generation of monoliths with distinct advantages and superior behavior. However, the fabrication of such catalysts is still in exploratory stage. Although there are several publications on printed monoliths for catalytic applications, the reliability of the structural integrity is still highly unprovable. According to [25], three main configurations of 3D-printed catalyst structures have been identified: woodpile type, isoreticular foam type, and channels' type; the woodpile configuration is used in 87% of literature studies due its simplicity.
In this study, ZSM-5-based zeolite monoliths of woodpile configuration are prepared via direct 3D printing. The influence of the bentonite binder concertation on the key physicochemical characteristics that affect the catalytic performance is thoroughly investigated. In addition, compressive mechanical strength is examined and statistical parameters describing the reliability of the structure are derived. Rheological characteristics and shrinkage behavior are also evaluated, since achieving the designed dimensions is an important issue allowing for the optimum usage of material, available volume and aiding the practice of accurate simulations.

Catalyst preparation
A commercial equilibrium ZSM-5 catalyst powder was used as starting material. The ZSM-5 powder is an industrial ZSM-5 based formulation that consists of 30 wt.% crystalline ZSM-5 zeolite diluted in a silica-alumina matrix. Bentonite clay (Sigma Aldrich, Germany) was used as the main binder as it allows to manufacture well-structured zeolite-based monoliths with high catalytic efficiency [26,27]. Three different pastes with varying compositions were prepared by mixing different ratios of ZSM-5 powder (40-50-60 wt.%) and bentonite (60-50-40 wt.%). Methylcellulose (Sigma Aldrich, Germany) was used as a secondary organic binder (1 wt.% of the zeolite/bentonite mixture) to promote gelation and shear thinning behavior. The solid powders were mixed in a mortar and distilled water was added gradually until saturation. The as-prepared pastes were used instantly to avoid drying that would affect the printability of the paste. The respective samples are denoted as XXZEO, where XX represents the concentration of the zeolite powder.
Robocasting 3D printing was used as manufacturing technique to produce cylinder monoliths. The specimens were designed with a nominal diameter of 12 mm and a height of 25 mm. However, due to excessive shrinkage during the successive drying and thermal treatment, different dimensional features were accomplished. An Engine HR 3D printer (Hyrel 3D, USA), equipped with an EMO printhead capable of printing pastes with viscosity up to 300,000 mPa s, was used. Syringes filled with the different pastes were loaded into the print head, which was equipped with Luer-Lock tips to define the filament diameter width. Different tip sizes were investigated and although pastes could be extruded through nozzles down to 0.4 mm, a 0.6mm nozzle was used to minimize clogging phenomena. After preliminary tests, a printing speed of 8 mm/s and a layer height of 0.5 mm were chosen, while the volumetric flow rate was adjusted accordingly. A 50% rectilinear infill was used during slicing, which resulted in nominal pore sizes of 0.66 mm. The printed monoliths were dried at ambient conditions for 24 h and were then calcined at 500 °C for 5 h in air flow to ensure the complete removal of the organic binder and achieve good thermal stability for the as prepared materials. A total of 10 specimens were printed for each paste composition to provide sufficient statistical estimates. The equilibrium ZSM-5 catalyst powder was also calcined at 500 °C for 5 h in air flow and was used as benchmark.

Rheological properties
Rheological measurements of the inks were carried out on an Anton Phar ACR 92 rheometer with a 20-mm parallel plate geometry and a 1-mm gap at room temperature. Flow ramps were conducted at strain rates of 0.01-500 s −1 and dynamic mechanical analysis (DMA) was carried out at 1 Hz, varying the oscillation strain from 0 to 100%. Yield stresses were calculated by observing the stress at which the storage modulus (G′) is equal to the loss modulus (G″) during oscillation tests. The maximum shear stress applied to the paste during printing was estimated based on equation: where R is the nozzle radius and Q is the volumetric flow rate that can be calculated from equation: where h is the layer height, w is the layer width, and V is the printing speed.

Morphology and physicochemical properties
The dimensions of the printed monoliths were measured using a digital optical microscope (Dino-Lite AB7013MZE, AnMo Electronics, Hsinchu, Taiwan) to evaluate the shrinkage behavior after drying and calcination. Scanning electron microscopy (SEM) was performed on a Phenom ProX (ThermoFisher Scientific, Waltham, MA, USA) to evaluate the monoliths homogeneity and structure adhesion. The samples were gold-coated using an ion sputtering device (Quorum SC7620, East Sussex, UK) and were mounted onto double adhesive conductive carbon tabs (TED Pella, Redding, CA, USA) on an aluminum stub to be scanned at an accelerating voltage of 10 kV.
Thermogravimetric analysis (TGA) was performed on an STA 449 F5 Jupiter thermogravimetric analyzer. Typically, 20-30 mg of the sample was loaded in an alumina crucible, and the temperature was raised from room temperature to 800 °C at a heating rate of 10 °C/min in synthetic air flow (ASTM E1131). The system was maintained isothermally at 800 °C for 30 min.
Textural characteristics were determined by N 2 adsorption/desorption isotherms at − 196 °C (Nova 2200e Quantachrome flow apparatus). Specific surface area was obtained according to the Brunauer-Emmett-Teller (BET) method at relative pressures in the 0.05-0.30 range, according to the ASTM D3663-03 standard. The specific pore volume was calculated based on the highest relative pressure. Prior to measurements, the samples were degassed at 250 °C under vacuum.
The crystalline structure of the catalysts was determined by powder X-ray diffraction (XRD) on a Siemens D 500 diffractometer operated at 40 kV and 30 mA with Cu Kα radiation (= 0.154 nm). Diffractograms were recorded in the 10-80° (2θ) range and at a scanning rate of 0.01°/s. FT-IR spectra after pyridine adsorption were collected using a Nicolet 5700 FT-IR spectrometer (resolution 4 cm −1 ) equipped with a homemade stainless-steel IR cell with CaF 2 windows by means of OMNIC software. The infrared cell, loaded with self-supporting wafers (~ 15 mg/ cm 2 ), is connected to a high vacuum line consisting of turbomolecular and diaphragm pumps; both sample holder and vacuum line are heated to avoid pyridine condensation. Before IR spectra acquisition, samples were heated at 450 °C under vacuum (10 −6 mbar) for 1 h to desorb any physisorbed species. All spectra were collected at 150 °C in order to eliminate the possibility of pyridine condensation. Adsorption of pyridine was realized at 1 mbar by equilibrating the catalyst wafer with the probe vapor, added in pulses for 1 h. The desorption procedure was monitored by stepwise evacuation of the sample for 30 min at 150, 250, 350, and 450 °C and subsequent cooling to 150 °C, recording the corresponding spectrum after each step. These temperatures represent the very weak, weak, medium, and strong acid sites, respectively. The concentration of acid sites was calculated using the integrated Lambert-Beer's law, corrected for the weight of the catalyst wafer. Pyridine adsorption shows two bands: one at 1450 cm −1 attributed to pyridine coordinated to Lewis sites and a second one at 1545 cm −1 assigned to protonated pyridine on Brønsted acid sites. The acid sites were quantified using the integrated molar extinction coefficients reported by Emeis [28], which are suitable for Si/Al-based catalysts.

Mechanical properties
The compressive strength of the monoliths was determined by performing uniaxial tests on the 3D-printed and thermally treated circular specimens. The experiments were carried out at room temperature on a universal testing machine (M500-50AT, Testometric, Rochdale, UK) with a crosshead moving at a constant rate of 1 mm/min. The applied loads were registered during the experiments through a 500 N loadcell with a resolution of 0.001 N. The compressive force was converted to stress by dividing with the measured cross section area of the sample. The compressive strength was associated with the maximum reached stress on the obtained force-displacement curves. A minimum of 10 samples were tested for each monolith composition to get statically reliable results.
The strength of brittle ceramics is typically not a single value, but is rather represented as a range of values due to the difference in the quantity of defects within each specimen. A parameter called the Weibull modulus was determined for the compressive strength of the measured samples. The Weibull modulus m is an empirical constant that indicates strength inconsistency or sample reliability. Conclusively, it gives information on the spread of the distribution of strength and, therefore, on the number of faults present in the sample. The higher the value of m, the more homogeneous the sample. The Weibull modulus m of the monolith was determined according to ASTM C1239-07 by fitting the compressive strength data to the following equation: where σ 0 is the Weibull scale parameter and P is the probability of failure at stress σ. This equation can be re-written in a simpler form: Hence, a linear interpolation of ln(ln 1 1−P ) vs ln( ) produces a straight line of slope m. The scale parameter σ 0 can be calculated as well. The probability of failure was evaluated with the equation: where n is the number of samples and i is the specimen rank in ascending order of failure stresses.

Printing paste rheology
The printing paste that was used for the 3D printing of the ZSM-5 monoliths of woodpile configuration via robocasting consisted of ZSM-5 powder, bentonite, and methylcellulose. The effect of the bentonite binder was investigated by preparing three different pastes with varying bentonite concentration of 40, 50, and 60 wt.%. According to Peng et al. [29], a paste to be successfully printed should be (5) P = i − 0.5 n free of particle agglomerates to avoid clogging, exhibit a shear thinning behavior, and possess a relatively high storage modulus (G′) and a yield stress > 200 Pa to allow structural self-support and high aspect ratio structures. To thoroughly examine these conditions, static viscosity and dynamic oscillations tests were performed for all three pastes.
The static results (Fig. 1) indicate that the increase of the ZSM-5 powder content has no major effect on the extrusion pastes' viscosity, as all pastes exhibit identical shear thinning behavior which is independent of the paste's composition. The maximum shear rate that the extrusion pastes are subjected to during printing is located at the nozzle's wall and is calculated equal to 124.6 s −1 according to Eq. 1. As presented in the inset of Fig. 1, at this shear rate the viscosity is equal to 39 Pa.s, 51 Pa.s, and 62 Pa.s for 40ZEO, 50ZEO, and 60ZEO samples, respectively.  The storage modulus (G′) and the loss modulus (G″) also provide useful insight on the printability of the pastes. Figure 2 presents the results of the oscillation measurements for the three different extrusion pastes. At low strains, when G′ is larger than G″, the pastes act as viscoelastic solids. Upon increased strain, and subsequently shear stress, the storage modulus starts to decrease, while the loss modulus exhibits a minor increase. All pastes exhibit storage modulus values larger than 500 kPa. This value is considered sufficient to retain structural support after deposition. The yield stress, which is the point where the paste starts to flow, can also be determined as the stress where the condition G″ = G′ is met. It is calculated equal to 90.39 kPa, 85.85 kPa, and 65.55 kPa for 60ZEO, 50ZEO, and 40ZEO respectively. The relatively high values of storage modulus and yield stress are due to the bentonite clay matrix, which provides high cohesion similar to hydrogel-based pastes [30].

Morphology of printed monoliths
As aforementioned, the ZSM-5-based cylinder structures were designed with a nominal diameter of 12 mm and height of 25 mm, as shown in Fig. 3. After the successful 3D printing of the specimens with the use of the different binder concentration pastes, the monoliths' shrinkage due to drying and calcination was investigated with optical microscopy. Three different geometrical measurements were conducted through a digital microscope: the specimens' outer diameter, total length, and rods' diameter. These measurements were performed for the monoliths printed with the different paste concentrations as printed, after drying, and after calcination and are presented in Fig. 3b. After drying, a sharp decrease in all three dimensions can be observed. Further modification after calcination is negligible, suggesting that shrinking occurs primarily during the removal of moisture in the drying process. This was also confirmed by TGA measurements performed with the dried monoliths in the presence of air flow. The temperature was raised from room temperature to 800 °C at a heating rate of 10 °C/min, and the recorded mass loss, as a function of temperature, is shown in Fig. 4, together with that of the reference ZSM-5 powder. All materials demonstrate a total mass decrease of about 14-16%, with the main loss occurring up to 100 °C, attributed to the removal of residual moisture. Furthermore, the TGA results confirm the structural integrity of the 3D-printed monoliths upon calcination and confirm that no crystalline phase change takes place at the employed temperatures.
As to the effect of the binder's concentration, the monoliths produced with the highest amount of bentonite (40ZEO) experienced the largest shrinkage. This can be clearly   Fig. 5, which illustrates the rod diameter values at each condition and binder concentration. As the bentonite percentage increases, the variation in the final dimensions of the monoliths is accentuated. It is speculated that this is due to the bentonite's ability to absorb water during the paste formulation stage; this moisture is discarded during the drying phase [31]. This leads to large differences between the designed geometrical characteristics of the monoliths and their real final dimensions, as in this work the rod diameter decreases by 40%, 25%, and 13% for 40ZEO, 50ZEO, and 60ZEO, respectively. The homogeneity of the calcined monoliths was evaluated through SEM images (Fig. 6). The 40ZEO specimen displays good dispersion of the zeolite particles within the matrix, with no surface discontinuities. The decrease of the bentonite matrix concentration in the 50ZEO and 60ZEO samples produces monoliths with surface cracks and larger agglomerates. This is expected to have a major influence on the mechanical strength of the monoliths, as it is well known that ceramics tend to fail due to crack propagations generated from surface flaws [32].

Physicochemical properties of printed monoliths
As the 3D-printed ZSM-5-based zeolite monoliths are intended to be used as catalysts in acid-catalyzed reactions, such as methanol dehydration to dimethyl ether (DME), it is important to assess the properties that govern their catalytic performance, i.e., crystalline structure, surface area and pore size, and type and strength of acidity. Table 1 presents the textural properties of the printed and calcined 3D-printed structures, as well as those of the pure ZSM-5 powder and the bentonite binder both calcined also at 500 °C in air. The corresponding N 2 adsorption/desorption isotherms and the Barrett-Joyner-Halenda (BJH) pore-size distribution are shown in Fig. 7a and b, respectively. The commercial equilibrium ZSM-5 powder presents a type IV isotherm, typical for mesoporous materials with pore width in the 2-50 nm range. The isotherm has an H4 hysteresis loop associated with narrow slit-like pores. Crystalline ZSM-5 is known to be microporous, with pores in the range of 0.3-0.6 nm. The mesoporosity observed here is due to the silica-alumina matrix used to dilute the ZSM-5  crystals in the equilibrium catalyst [33]. Bentonite has large pores and low surface area, which is mainly external, i.e., it is the outer surface area of the discrete particles/agglomerates of the material. The 3D-printed structures exhibit similar type IV isotherms to the zeolite powder, with however preferential adsorption at higher relative pressures and more pronounced hysteresis loops. This increases with the decrease of the ZSM-5 concentration in the monoliths and suggests the enhancement of mesoporosity and the formation of a significant number of large pores [34]. Moreover, the increase of the adsorbed nitrogen volume in the monoliths at high relative pressure indicates an important external surface area contribution. The BJH pore size distribution curves demonstrate in all materials a well-defined pore distribution around 3-4 nm. This feature does not represent real pores and is caused by a transition of the adsorbed phase from a lattice fluid-like phase to a crystalline-like solid phase, typical for ZSM-5 zeolites [35]. A broad pore size distribution centered on 7-9 nm is apparent in both the ZSM-5 powder and the printed monoliths. The latter present in addition some very large macropores with size > 50 nm. The similarity between the samples is also confirmed by the textural properties in Table 1. All materials have comparable pore volume, while the BET surface area of the printed monoliths with different ZSM-5 concentrations varies between that of the ZSM-5 powder and the pure bentonite according to the respective ZSM-5/binder ratio. These promising findings clearly demonstrate that the ZSM-5 crystals retain their porous structure and textural characteristics and as expected not affected by the additive manufacturing process. This is also confirmed by the XRD patterns presented in Fig. 8. The well-defined crystalline structure of the equilibrium catalyst is retained in all monoliths, irrespective of the ZSM-5 concentration, with similar sharpness and intensity of the diffraction peaks. The additional diffractions at 21.9, 27.8, 35.6, 54.1, and 61.9° in the diffractograms of the structured samples are characteristic of the bentonite binder crystalline structure.
Acidity is a key property for the catalytic performance of the 3D-printed ZSM-5-based zeolite monoliths, as their intended application is the use in the acid-catalyzed reaction of methanol dehydration to DME. Chemisorption of pyridine followed by IR was employed to detect the number and strength of surface Lewis and Brønsted acid sites. The total acidity, the number and strength of Brønsted and Lewis acid sites, as well as the B/L ratio, are shown in Fig. 9. As expected, the ZSM-5 equilibrium catalyst has the highest number of Brønsted acid sites (attributed to bridging framework hydroxyls) [36], with the highest percentage being of medium and high strength. There is also a considerable amount of Lewis acidity, originating from the silica-alumina matrix, and an overall B/L acid site ratio of 1.4. The fabricated monoliths exhibit reduced acidity due to dilution with the bentonite powder used as binder. There is also a progressive reduction in the B/L ratio due to the increased contribution of the (low) Lewis acidity of the bentonite binder. Overall, the decrease is roughly proportional to the concentration of the binder, suggesting that the ZSM-5 crystallites retain their acidity during the additive manufacturing process. These results are very promising for the further utilization of the 3D-printed ZSM-5-based zeolite monoliths in catalytic applications.

Mechanical properties
To assess the mechanical properties of the materials, compression tests were carried out on the 3D-printed ZSM-5-based zeolite monolith samples and representative stress-strain curves are presented in Fig. 10. At the start of the test, there is a region exhibiting a small increase in stress with increasing compressive strain. This is common occurrence [37] in non-post processed structures. The specimens were tested in their as-fired condition without any polishing treatment of their top and bottom surfaces to avoid any local damage. Subsequently, their bottom and top surfaces are neither parallel nor flat; thus, there is a displacement pre-required to achieve uniform loading. Specimens fabricated with 40 wt.% zeolite powder present a multipeak profile due to successive failures of non-critical rods. The repetition of this behavior produces a jagged stress-strain curve until a relatively sharp decrease in stress that signifies the critical rod failure. In contrast, specimens fabricated with 50 wt.% and 60 wt.% ZSM-5 powder demonstrate relatively smoother stress-strain behavior, with less sharp stress drops. This alteration can be possibly explained by the additional bentonite that the 40ZEO contains, which acts as a binder and during failure; larger sections of material (Fig. 10a) are detached. The 50ZEO and 60ZEO pastes produce specimens with less binder that break down to very small, crumbled pieces (Fig. 10b) that do not play a major role in stress resistance. Figure 11 summarizes the mean maximum compressive strength achieved for each monolith consistency. It can be concluded that the zeolite powder concentration greatly impacts the material's ability to withstand loads. An increase in the zeolite concentration by 20 wt.% with a respective decrease in the bentonite content results to approximately 4.5 times lower maximum compressive strength, which varies from 1.61 MPa to 0.34 MPa for 40ZEO and 60ZEO, respectively. A further interpretation of this effect can be that the large amount of surface cracks that the 50ZEO and 60ZEO monoliths retain, as shown by the SEM analysis, act as stress concentration points and greatly intensify crack propagation.
The mechanical response of brittle materials is sensitive to microstructural internal or surface flaws, such as pore and micro-cracks. Therefore, the Weibull modulus is used as a quantitative measure of the 3D structure's mechanical reliability by estimating the probability of failure under a given stress. Figure 12 presents plots of the compressive strength data based on Eq. (4), where least squares fitting of a straight line was used through the data for the calculation of the Weibull modulus (m) and scale parameter(σ 0 ). The results for the three different monoliths are presented in Table 2. All data present a satisfactory fitting parameter (R 2 ), which is indicative of the degree of fit of each model. The Weibull modulus exhibits a value of 13.92 for 40ZEO, which can be considered an indication of high reliability for ceramic materials. In contrast, the 50ZEO and 60ZEO specimens exhibit a much smaller Weibull modulus due to the lower binder concentration and the surface microcracks that increase the number of materials flaws. To the best of our knowledge, no published data exist in the literature describing the Weibull modulus of zeolite-bentonite monoliths produced by robocasting. Thereby, a direct comparison with analogous data from literature is not possible; nonetheless, a few results concerning ceramic catalysts produced by different manufacturing methods can be compared. Samimi et al. [38] examined the effect of binder concentration and forming conditions on the Weibull modulus of alumina pellets. Almost all samples exhibited values lower than 10. Ojuva et al. [39] produced zeolite pellets by freeze casting that demonstrated Weibull modulus values close to 3 and 7.
Additionally, the probability of failure is calculated and is shown in Fig. 13. It is obvious that only 40ZEO specimens are reliable at application stresses above 1 MPa. However, when it comes to lower stresses, 50ZEO and 60ZEO specimens demonstrate similar results. Thus, for stresses close or lower than 0.5 MPa, a more thorough consideration of the monoliths' consistency is required.

Conclusions
In this work, ZSM-5-based zeolite monoliths of woodpile configuration were produced through robocasting. The effect of bentonite clay binder concentration in the range of 40-60 wt.% on the rheological, physicochemical, and mechanical properties was examined. The material homogeneity, shrinkage behavior, and achieved dimensions were also assessed.
The static rheological results indicate that increasing the ZSM-5 content has no major effect on the viscosity, since all pastes exhibit identical shear thinning behavior. All three pastes demonstrate sufficient storage modulus, since structural support was retained after deposition. However, as the bentonite concentration increases, the monoliths' shrinkage due to drying is accentuated, as bentonite absorbs more water during the paste formulation stage and discards it during the drying phase. This effect can lead to considerable differences between the designed geometrical dimensions and the final attained values. In addition, decreasing the bentonite matrix concentration produces monoliths with surface cracks and larger agglomerates.
As the 3D-printed ZSM-5-based zeolite monoliths are intended to be used as catalysts in methanol dehydration to DME, the basic physicochemical properties that define the catalytic performance, i.e., surface area, crystalline structure, and type and strength of acidity, were determined. The BET surface area of the printed monoliths with different ZSM-5 concentration was found to vary between that of the ZSM-5 powder and the pure bentonite, demonstrating that the ZSM-5 crystals retain their porous structure and textural characteristics and are not affected by the additive manufacturing process. XRD also confirmed the existence of crystallites with well-defined MFI structure in the printed monoliths. With regard to acidity, a key property for application in acid-catalyzed reactions, the fabricated monoliths exhibit reduced acidity due to dilution with the bentonite powder used as binder and progressive reduction in the Brønsted/Lewis acid sites ratio. The reduction in acidity was found to be roughly proportional to the concentration of the binder, signifying that the ZSM-5 crystallites also retain their acidity during the additive manufacturing process.
Finally, the mechanical properties of the materials were assessed. Compressions tests of over 10 specimens for each composition were conducted to produce statistical reliable data. By increasing the bentonite concentration from 40 to 60 wt.%, the mean compressive strength increased from 0.34 MPa to 1.60 MPa. This effect can be explained by the large amount of surface cracks that the 50ZEO and 60ZEO monoliths retain that greatly intensify crack propagation. The fabricated monoliths' mechanical reliability was evaluated by calculating the Weibull modulus. The results showed that the monoliths' consistency under applied stress was heavily  Overall, this study demonstrates the successful preparation of 3D-printed ZSM-5 monoliths that retain their key physicochemical properties, in addition to increased mechanical strength. Future work will address the evaluation of the performance of these materials in the reaction of methanol dehydration to dimethyl ether.
Author contribution Eleni Heracleous and Dimitrios Tzetzis contributed to the study conception and design. Material preparation and data collection was performed by Savvas Koltsakidis and Vasiliki Koidi, while analysis of data was performed by all authors. The first draft of the manuscript was written by Savvas Koltsakidis and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.