Controlling the Activity and Selectivity of HZSM-5 Catalysts in the Conversion of Biomass-Derived Oxygenates Using Hierarchical Structures: The Effect of Crystalline Size and Intracrystalline Pore Dimensions on Olefins Selectivity and Catalyst Deactivation

The conversion of biomass-derived oxygenates over zeolite catalysts constitutes a challenge for the efficient production of bio-based chemicals and fuels due to difficulty in controlling the selectivity and high coke formation of such reactions. This is partly attributed to the microstructure of zeolite catalyst which affects the conversion and selectivity of products derived from biomass-derived oxygenates. In this study, the conversion and deactivation characteristics of three different model oxygenates found in biomass bio-oil (namely, acetol, furfural and guaiacol) over ZSM-5 zeolites of varying acidity, pore and crystal size prepared with bottom-up and top-down approaches were evaluated using a fixed bed microreactor at atmospheric pressure and a space velocity of 5 h−1 at a temperature range of 450–650 °C. Analysis of the experimental results indicates that the optimum temperature for such conversions is in the vicinity of 600 °C allowing for complete conversion of the compounds and high resistance to coking. The mechanisms of those conversions are discussed based on the obtained results. In general, crystal size and mesoporosity induce easier access to active sites improving mass transfer but also alter the location type, and strength of acid sites allowing for higher yields of primary and intermediate products such as olefins.


Introduction
The need to lessen and eventually reverse the environmental burden has set biomass residues as an attractive feedstock and the most abundant source of renewable carbon.Residual biomass can support the production of renewable chemicals which offer a high economic and societal value while at the same time increasing the security of supplies.
Efficient use of the available renewable carbon through thermochemical conversion of biomass residues (i.e., pyrolysis and gasification) provides a sustainable route to produce high value chemicals and fuels.In particular, fast pyrolysis allows the production of renewable liquid feedstock also known as bio-oil.It is a dark brown liquid with a strong phenolic odor with its chemical characteristics dependent on the type of biomass feedstock used and the operating conditions [1].
Bio-oil contains more than 400 various compounds, including carboxylic acids, aldehydes, ketones, alcohols, phenols, and phenolic derivatives and constitutes an excellent feedstock for the production of numerous chemicals or can serve as an energy carrier [2].Nevertheless, characteristics such as high water and oxygen content give rise to undesired properties (i.e., corrosiveness, chemical and thermal instability, high viscosity) and make it incompatible with the conventional fossil-based fuel infrastructure [3]; hence, upgrading is required [4].
Catalytic upgrading to convert bio-oil (or its vapours) to liquid hydrocarbons focuses on the deoxygenation of molecules.However, in-situ conversion exhibits low carbon efficiencies to liquid hydrocarbon products [5].The main catalysts used for such conversion are zeolites [6][7][8], which intend to crack the bulk biomass fragments and oligomers into smaller, lighter, and deoxygenated products.Yet, the high catalytic activity of deoxygenation catalysts results in formation of light gaseous hydrocarbons, coke, and to a lesser extent, liquid deoxygenated molecules.In lieu of trying to produce liquid hydrocarbons for fuels, the use of gaseous molecules to increase the biomass footprint in the chemicals sector could be beneficial.In fact, a total of 144 million tons of CO 2 per year can be saved in the EU by 2050 if renewable carbon is introduced to the chemicals sector which would also be beneficial for economic growth [9].
Light olefins, namely ethylene, propylene and butylenes, are key building blocks for the chemical industry as they are essential for the production of a vast spectrum of products such as polymers (polyethylene, polypropylene), oxo-alcohols, and other organic products [10].Currently, the production of olefins is carried out primarily via steam cracking and fluid catalytic cracking (FCC) of petroleum products [11].Bio-oil represents an alternative feedstock to produce such important building blocks and adds to other alternative production methods utilizing renewable feedstocks, such as methanol to olefins (MTO), ethanol to olefins (ETO), and Fischer-Tropsch to olefins (FTO) among others [12,13].Nevertheless, the yield of light olefins from bio-oil is mainly affected by the ability of the catalyst to selectively crack these various oxygenates [14].
Owing to its strong Brønsted acidity, high thermal stability, and shape-selective micropores, HZSM-5 zeolites, with a molar Si/Al range of 11-40, are regarded as a suitable solid acid catalyst for the conversion of biomass-derived pyrolysis vapors [15,16] with its activity being mainly dependent on its pore structure, crystal size and acid properties.However, the pore size of the conventional HZSM-5 hinders the diffusion of molecules larger than its pore dimensions which results in inaccessible active sites, thus reducing the effectiveness of the catalyst [17].For example, steric hindrance is of importance when considering aromatic structures in biomass with adjacent functional groups.Moreover, the acidity of the catalyst affects the conversion of the reactants, and a high acidity promotes the formation of coke and hence the deactivation of the catalyst [18].As such, multipore zeolite systems are of particular interest when molecules of substantially different sizes undergo conversion [19].
Reduced crystal sizes of HZSM-5 enhance the selectivity towards lighter products while it prolongs the catalyst lifetime, because of reduced mass transfer limitations [20].The introduction of an ordered secondary (hierarchical) porosity improves the use of the active volume and reduces the internal diffusion limitations in conventional ZSM-5 zeolites [21].The latter can be introduced either by templating or post-processing methods [22].For the catalytic upgrading of the biomass-derived oil, it has been reported that the introduction of mesopores enhances the cracking of bulky molecules that are too large to enter the micropores of the catalyst while it allows an easier diffusion of products and coke precursors [3].
Nevertheless, the catalytic upgrading of bio-oil remains a challenge due to the diversity of the chemical classes of the compounds and selectivity can vary depending on catalyst preparation and/or process conditions applied.A proper understanding of the effect of the catalytic cracking of these compounds on product distribution and catalyst deactivation is therefore crucial.Moreover, finding the optimum operating conditions may be a challenge given the varying properties of the reactants [23].The quest for development of more advanced zeolite catalysts with tailored activity-selectivity for efficient biomass utilization requires deeper understanding of the undergoing conversions and detailed studies of the factors influencing catalytic activity.Thus, investigating the behaviour of model compounds with different chemical functionalities and sizes enable a better comprehension of the conversion steps, catalyst selectivity, and deactivation behaviour.
In this study, three model compounds representative of biomass pyrolysis oil (namely, hydroxyacetone, furfural and 2-methoxyphenol) with different chemical functionalities and molecular size have been used for identifying suitable catalyst design properties and operating conditions for efficient conversion to light hydrocarbons (i.e., light olefins).
Hydroxyacetone (acetol) represents the most abundant light ketone in bio-oil [24].It derives from the ring scission of cellulose, and it incorporates a carbonyl and a hydroxyl group.
Moreover, furfural is the main component of furans, which are major intermediates of cellulose and hemicellulose pyrolysis, and are very important precursors for many value-added chemicals [25,26].
Guaiacol (2-methoxy phenol) on the other hand, represents a typical lignin compound containing a hydroxyl and a methoxy group, which conversion leads to excessive coke deposition [27][28][29].
The catalytic upgrading of such compounds is an extremely complex chemical process, and even though there are studies in the literature focusing on the individual effects of catalyst modification on conversion of lignocellulose derivatives, most of the studies focus on the production of liquid product ( i.e., Heavier molecules), with highly aromatic content.Contrary to other studies that investigate the effect of individual catalyst properties on the liquid yield and composition, the focus of this study relates the catalyst properties to light olefins production which is an important intermediate responsible for aromatics and coke generation but also a product with very high industrial importance.Apart from the mesoporosity introduction to zeolite structure with top-down and bottom-up methods, which can affect the access and diffusion of the reactants from and to the active site of the catalyst, the crystal size, which affects the diffusion path of the molecules, is investigated and correlated to reaction mechanisms associated with the formation of gaseous olefins.

Raw Material-Chemicals
In this study, hydroxyacetone (90 vol.%), furfural (99 vol.%), and guaiacol (99 vol.%) were used as model compounds of bio-oil.They were received from Merck and used without any further purification.The chemical formula, structure, and kinetic diameter of each of the compounds are presented in Table 1.

Catalysts Synthesis
A commercial ZSM-5 catalyst was used as a reference for all tests.The catalyst was provided by Süd Chemie (TZP-302) in NH 4 + form (NH 4 + -ZSM-5).Prior to use, the zeolite was calcined in air at 550 °C for 5 h at a heating rate of 2 °C/min to ensure its protonation to HZSM-5.This catalyst is designated as "Com".To evaluate the effects of post-synthetic (top-down) pore size modification, desilication of the catalyst was carried out.The desilication was performed using 0.2 M NaOH solution at 65 °C for 30 min at a ratio of 100 ml per 3 g of catalyst, followed by washing, filtering, and drying overnight at 100 °C.Next, an ion exchange took place four times with 0.1 M NH 4 NO 3 at 60 °C for 1 h (40 ml of solution per gram of zeolite).The desilicated zeolite, designated as "Ds-Com", was again filtered, washed, dried overnight, and calcined at 550 °C to produce the protonated form [31].
The bottom-up pore modification was evaluated by synthesizing ZSM-5 zeolites with nanosized crystals and hierarchical pore structure.The two additional catalysts were synthesized according to [32] and [33].Calculated amounts of aluminum isopropoxide (AIP), tetrapropylammonium hydroxide (TPAOH), NaOH, and water were mixed and stirred for 30 min to form solution A. Tetraethyl orthosilicate (TEOS) was added dropwise to the solution under vigorous stirring and left to stir for 3 h.The obtained gel was then transferred into an autoclave and heated for 72 h at 180 °C, followed by filtration, washing, and drying overnight at 120 °C.The catalyst was finally calcined for 6 h at 550 °C.As for the hierarchical zeolite, the mesoporous agent dimethyloctadecyl [3-(trimethoxysilyl)propyl] ammonium chloride solution TPOAc was mixed with TEOS before being added to solution A. The produced Na-ZSM-5 was later ionexchanged into NH 4 -ZSM-5 with 1 M NH 4 NO 3 at 80 °C for 1 h (repeated three times).The obtained NH 4 -ZSM-5 was again dried overnight and calcined at 550 °C for 6 h to obtain HZSM-5.The nominal composition of the gel mixture was SiO 2 : 0.04 Al 2 O 3 : 0.3 TPAOH: 0.05 NaOH: 11 H 2 O: 0.05 TPOAc.The synthesized catalysts were designated as "HT-mic-20" and "HT-hie-20" and correspond to nanosized microporous and nanosized hierarchical counterparts, respectively.

Catalyst Characterization
All the catalysts were characterized by means of x-ray diffraction, N 2 -physisorption, NH 3 -TPD, FTIR and TGA to assess the crystal textural, acidic properties and thermal stability, respectively.Furthermore, microscope inspection of the surface structure was performed by means of scanning electron microscopy.
X-ray diffraction patterns were obtained in the range of 2θ = 5-60° at a step size of 0.025° by a Siemens D5000 Diffractometer using CuKα radiation (λ = 1.5408Å) under 40 kV and 30 mA and were used to identify the phase and asses the crystallinity of the catalysts.The relative crystallinity was calculated according to the aggregate intensities of the three peaks at 2θ = 23.0-25.0°.The average crystal size was measured using the Scherrer equation D = K λ cos from XRD peaks between 2θ = 7.0 and 10.0°.K = 0.89 (constant), λ = 0.154 (nm) X-ray wavelength, β = the peak width at half of its height, and θ = Scattering angle.
The morphologies and structures of the different zeolites were examined by scanning electron microscopy using highresolution field-emission SEM Zeiss Ultra 55 GEMINI.The image processing and the determination of the particle size distribution were performed using ImageJ software.
N 2 physisorption at 77 K was obtained by a Micromeritics ASAP 2020 physisorption analyser to determine the specific surface area (S BET ), pore volume (V pore ), and average pore radius (R).Additionally, the pore size distribution was determined using the BJH method while the micropore volume was determined using the t-plot method.Here, it must be noted that BJH results should not be interpreted as absolute values, given the method's inability to accurately determine pore size in the micropore region and it will be used for comparison purposes only.Prior to the measurement, the samples were degassed at 250 °C for 6 h.Temperature-programmed desorption of ammonia (NH 3 -TPD) of catalysts was performed using a Micromeritics AutoChem 2910 in order to determine the amount and the strength of the acid sites in the catalysts.Approximately 100 mg of sample were preheated in helium at 500 °C for 2 h and subsequently were cooled down to 120 °C and subjected to ammonia saturation.Prior to desorption, they were flushed with helium for 20 min to remove any physically sorbed ammonia, thereafter the desorption took place in the range up to a temperature of 800 °C at a temperature ramp rate of 5 °C/ min.The TPD curves were deconvoluted using OriginPro.
The Si/Al ratios were determined using a Rigaku Super-mini200 X-Ray Fluorescence analyzer.
The IR spectra were collected using a Nicolet 5700 FTIR spectrometer (resolution 4 cm −1 ) by means of OMNIC software.All the samples were finely ground in a mortar and pressed in self-supporting wafers (~ 15 mg/cm 2 ).Before IR analysis all samples were heated at 450 °C under high vacuum (10 -6 mbar) for 1 h in order to desorb any possible physisorbed species (activation step).All spectra were collected at 150 °C in order to eliminate the possibility of pyridine condensation.The sample was heated to 250 °C for 30 min and then was cooled down to 150 °C in order to collect the spectrum.The same procedure was followed for temperatures 350 °C and 450 °C.The corresponding bands at 1450 cm −1 and 1545 cm −1 are used for the quantification of the Lewis and the Brønsted acid sites respectively.
All the catalysts were characterized by means of thermogravimetric analysis (TGA), using a NETZCH Jupiter F3 TG/DSC analyzer, to assess the thermal stability of the zeolite.The catalysts were heated at rate of 10 °C/min to a final temperature of 750 °C.It has to be noted that in order to fully assess the thermal stability, successive reaction-regeneration cycles are needed, however, this extends beyond the scope of the study (Figure S4).

Experimental Setup and Catalytic Runs
The experimental set-up is illustrated in Fig. 1.The evaluation of catalytic performance was performed using a fully automated PID Eng & Tech Microactivity reactor setup consisting of a fixed bed millireactor (i.d. 9 mm) at atmospheric pressure (overpressure of few kPa) and a temperature range of 450-650 °C using 0.5 g of catalyst on a porous support.Prior to experiments, the catalyst was heated to the corresponding temperature in N 2 flow (300 Nl/min) for 1 h.The weight hourly space velocity (WHSV) for all the tests equalled 5 h −1 .The reactants were fed to the reactor pulse-wise (8 pulses of 15 min duration each) to allow for chromatographic detection of the products.A total time on stream (ToS) of 75 min for acetol and 2 h for the rest was achieved.An electronic scale was used to record the mass feeding rate of the reactants introduced to the reactor via a HPLC pump (Gilson 307).The liquid products were collected in a condenser for further analysis using GC, whereas the gas products were sent for on-line analysis using an Agilent 7890A equipped with 2 FID detectors for oxygenates and hydrocarbons analyses and a TCD detector for permanent gases (H 2 , N 2 , CO, CO 2 and CH 4 ).The GC conditions and method can be found in [34].Pressure was controlled with a pressure control valve.Duplicates of all the experimental runs were conducted with observed relative deviation being lower than 5%.
The evaluation indices used in this work are expressed as follows: (1) The conversion of a reactant is calculated as The yield of a product is defined as The distribution % of olefins is set as mass of specific olefin mass of total olefins × 100 The effective hydrogen-to-carbon ration [35] is defined as:

Catalysts Characterization
The XRD patterns of the commercial and synthesized catalysts are shown in Fig. 2. All catalysts show a typical MFI structure (PDF 00-044-0003), with high crystallinity while no impurity phases are observed.
The catalysts exhibit two main characteristic peaks at 2θ = 7-9° and three peaks at 2θ = 23-25°.The hydrothermally synthesized catalysts (HT-mic-20 and HT-hie-20) exhibit intense characteristic peaks at 7-9° which can be attributed to the use of TPAOH as a microporous template [36].The intensity of the characteristic peaks decreases with the introduction of mesoporosity which suggests a decrease in the crystallinity of the mesoporous catalysts [37].Indeed, the crystallinity decreased from 89.5% and 90.5% for Com and HT-mic-20 to 87.2% and 81.6% for Ds-Com and HThie-20, respectively.The reduction in crystallinity due to the addition of TPOAc as a mesoporous agent indicates a hindrance in the crystallization process, hence affecting the formation of zeolite structures [38].The HZSM-5 crystal size determined by the Scherrer equation, for the peak at 2θ = 7.7-7.9° is 38.6 nm for Com, 37.7 nm for Ds-Com, 14.5 nm for HT-mic-20, and 14.3 nm for HT-hie-20. (4) As depicted in Fig. 3, the commercial catalysts Com and Ds-Com have rod-shaped structures, whereas the synthesized catalysts have a spherical-shaped structure.The coffinshaped crystals are aligned so that their lengths are nearly perpendicular to the surface and the 00l and the h0l peaks exhibit exaggerated intensity [39].A close examination of the spherical zeolite particles reveals that they are an aggregate of abundant small crystallites.The average particle size of the aggregates as determined by the SEM image analysis for the commercial catalysts are 1257, 698, 133 and 105 nm for Com, Ds-Com, HT-mic-20 and HT-hie-20 respectively.The corresponding standard deviations of the measurements which averaged 100 particles are determined equal to: 529, 263, 32 and 45 nm.The lumping of these nano-crystallites contributes to the formation of extensive intercrystal mesopores, which effectively increases the coking capacity of such zeolites [40].This can be further confirmed by N 2 physisorption (Fig. 4 and Table 2).
The isotherms of Com and HT-mic-20 exhibit a type I isotherm showing almost no hysteresis loop, which verifies the structure of conventional microporous zeolites [41].
Conversely, the isotherms of Ds-Com and HT-hie-20 exhibit a hysteresis loop typical of type H4 [42], suggesting the existence of mesopores [43], and confirming the observation using SEM, the mesopores of HT-hie-20 zeolite are formed by the accumulation of nanocrystals.The difference in the size of the hysteresis of Ds-Com and HT-hie-20 is proportional to the number of mesopores present in each of the catalysts.Table 2 provides an overview of the textural properties of the zeolite catalysts.
The microporous (Com and HT-mic-20) zeolites have a lower mesoporous area and volume than their mesoporous counterparts.Introducing mesoporosity lead to a decrease of 12 and 26% in the microporous volume of Ds-Com and HT-hie-20, respectively.This is related to the reduction of the crystal size leading to the exposure of the external surface area [44].The reduced micropore volume correlates with the lower intensity observed in the XRD results [45].Furthermore, the desilication of the commercial zeolite caused the formation of mesopores with a minimal effect on the micropores already present in the framework; however, the addition of a mesoporous agent during the synthesis of the hierarchical catalyst resulted in a higher mesoporous over microporous volume [46].The hierarchical factor, determined as (V micro /V pore ) x (S meso /S BET ) [47], was calculated and is equal to 0.054 for Com, 0.081 for Ds-com, 0.076 for HT-mic-20, and 0.082 for HT-hie-20.
The overall acidic properties of the catalysts are determined by NH 3 -TPD.All catalysts exhibit both low and high-temperature peaks.The low-temperature peak is attributed to the adsorbed ammonia on weak total acid sites (Brønsted and Lewis), whereas the high-temperature peak is assigned to the desorption of NH 3 adsorbed on the strong Brønsted and Lewis sites [40].The amount of the weak and strong acidity as derived from peak deconvolution, are listed in Table 2.Although larger crystal catalysts (Com and Ds-Com) have higher total acidity than the nanocrystal zeolites, the creation of mesopores resulted in a higher ratio of strong acid sites as observed for Ds-Com (1.06) and HT-hie-20 (0.8) when compared to Com (0.98) and HT-mic-20 (0.56), respectively.
However, the classification of acid sites into weak and strong sites is not enough to have a deeper insight to the catalytic activity of the catalysts.The distinction between Brønsted and Lewis sites is thus necessary.The quantification of Brønsted and Lewis sites was performed by pyridine (Py) adsorption on the acid sites of the different catalysts.The acidities are listed in Table 3.The total amount of acid sites as expressed by μmol of pyridine per gram of catalyst, corroborates the results obtained by the NH 3 -TPD (Table 2).
The amount of Brønsted acid decreased with either mesoporosity introduction method (i.e., top-down or bottomup), whereas the amount of Lewis acid increased.This suggests that the alkaline treatment of commercial microporous ZSM-5 resulted in the removal of internal Si-OH bonds post desilication, whereas the addition of a mesoporous template lead to the reprecipitation of Al on the external surface [48,49].Moreover, the smaller crystal catalysts (HT-mic-20 and HT-hie-20) exhibit lower total acidity compared with their larger counterparts.The latter can be ascribed to more defects that are found in the smaller the crystal size zeolite [50].Nevertheless, the C B /C L ratio of the microporous catalyst was not affected.
The low total acidity of the HT-hie-20 can be attributed to the mesoporous agent TPOAc used.TPOAc is a surfactant agent, which interacts with the silicon atoms, thus providing more silicon atoms that are introduced to the framework hence weakening the total acidity [51] which is supported by the higher Si/Al ratio obtained for this catalyst (Table 2).This as a result affects highly the creation of Brønsted sites [52].Hence, the total acidity as well as the types of acid sites are greatly dependent on the mesopores creation method.
The absorption bands at 440, 540, 795, 1065, and 1220 cm −1 are characteristic of the ZSM-5 framework [53,54].The peaks illustrated in Fig. 5 indicate that all four catalysts have ZSM-5 framework structure as already observed by XRD.The intensity ratio of the peak at 540 cm −1 over that at 440 cm −1 is used as a further assessment of the crystallinity of the zeolite [55,56].The peak at 440 cm −1 is associated with the bending vibration of T-O-T (T being either Si or Al), while the absorption band at 540 cm −1 Fig. 4 Nitrogen adsorption-desorption isotherms and pore size distribution of the zeolite catalysts is related to the asymmetric stretching of the characteristic five-membered ring of the MFI zeolite.As shown, the microporous counterparts (Com, HT-mic-20) exhibit a higher degree of crystallinity (0.363 and 0.251) compared to their mesoporous counterparts (0.336 and 0.172 for Ds-Com and HT-Hie-20 respectively) which supports the XRD findings.The bands at 795 cm −1 and 1220 cm −1 correspond to the symmetric and asymmetric stretching of the external T-O-T bonding respectively.The band at 1065 cm −1 results from the asymmetric stretching of the internal T-O-T [57].
The presence of Brønsted acid sites can be further identified by the peak at 3607 cm −1 .The band at 3740 cm −1 is associated with terminal silanol groups; it is more intense for the hierarchical zeolites and denotes an increase of the ratio between the external/mesopore surface and the surface of micropores that lacks silanol groups [31].The peaks at 3660 and 3680 cm −1 correspond to deposited amorphous extra framework Al indicating that the mesoporosity introduction through desilication results in removal of the Al species out of the framework of the zeolite.Extra framework Al is also observed for the microporous nanosized ZSM-5 catalyst.This is corroborated by the band in the region of 3720 cm −1 which relates to silanol groups undergoing interactions with the extra-framework Al given the high calcination temperature [58][59][60].The observed shift for the hydrothermally prepared nanosized zeolite can be attributed to surface hydroxyl groups that interact with neighbouring species and to internal silanol groups located at framework defect sites [61,62].The band at 3780 cm −1 is related to non-tetrahedral Al and is only observed for the desilicated zeolite supporting the postulation that an alkaline environment partially affects the Al species [63,64].The broad band observed at 3500 cm −1 in HT-mic-20 and HT-hie-20 indicates the presence of internal silanol sites at framework defects (nests) [65,66] and a distorted bridging hydroxyl species [61,[67][68][69].

Catalytic Conversion of Bio-Oil Model Compounds
In all the experiments a mass balance closure greater than 70% has been obtained (Table S1, Table S2 and Table S3).
The main losses are associated with unidentified gaseous species.

Hydroxyacetone
Hydroxyacetone (acetol), exhibits a high initial conversion at all temperatures irrespective of the pore size of the catalysts which correlates to the kinetic diameter of acetol (Table 1).At temperatures higher than 600 °C complete conversion of the molecule is achieved (Fig. 6).Reduced acidity and nanosized crystal of hydrothermally synthesized zeolites allow full conversion with increased resistance to coking, maintaining the activity at temperatures greater than 600 °C.At lower temperatures (450 and 550 °C), coking is evident with activity being diminished due to aromatization and oligomerization reactions over the Brønsted acid sites of the catalysts, an observation that is corroborated by the increased aromatics (Benzene, Toluene, Xylene, Naphthalene -BTXN) yield (Fig. 7)-which are secondary products of olefins (BTXN compounds exhibit the same trend as olefins regardless of temperature and catalyst used) [27]obtained at this temperature range [70,71].Nevertheless, the yield of aromatics formed over nanosized zeolites (HTmic-20 and HT-hie-20) is lower, indicating that the reduced total acidity as well as the crystal size has a significant effect on suppressing the formation of coke precursors and maintaining the activity of the catalyst longer as stated above.The importance of more efficient mass transport of primary deoxygenation products to the reduction of aromatics is also demonstrated by the lower aromatics yield obtained when using the desilicated counterpart (Ds-Com).The yields of benzene, toluene, para-xylene, ortho-xylene, and naphthalene are plotted in Figure S7.Their production decreases with time with toluene being the main product followed by benzene and para-xylene.Additionally, the use of hierarchical zeolites results in higher olefin yields compared to their corresponding microporous catalysts, with a maximum yield reached at 600 °C.An increase of temperature to 650 °C resulted in a decline in the yield of olefins produced on microporous catalysts, accompanied by an increase in H 2 formation as shown in Figure S6 since H 2 is incorporated in the formation of olefins.
The lower zeolite occupancy with primary olefinic products due to faster diffusion from the porous network as well as the high temperature result in a decreased rate of  formation of oligomers and aromatics and their subsequent reaction products [71].At higher temperatures, the aromatization/cracking mechanism involves dealkylation of aromatic products formed resulting in higher olefins and gaseous products formation, supporting the results obtained for all the catalysts tested.This can explain the lower rate of coke formation on the catalysts as expressed by the prolonged (almost unchanged) activity that is observed at temperatures greater than 600 °C.Similar results have been reported elsewhere [23].
HT-hie-20 maintained a stable activity throughout the ToS at any temperature.Com and HT-mic-20 can be correlated due to the similar microporous structure and the same C B /C L ratio.It can then be postulated that differences in the activity of the microporous catalysts are related also to their different crystal size.In that regard, a reduction in the crystal size of the microporous HZSM-5 leads to a higher conversion of acetol and a slower deactivation of the zeolite.This observation holds also when comparing Ds-Com and HT-hie-20.
The effect of temperature on the CO production is apparent as the yield of CO increases substantially with the increase in temperature.This indicates that the CO is the primary product of acetol deoxygenation via direct decarbonylation, as also suggested by other studies [23,27,72].
The fact that CO 2 is not directly affected by temperature suggests that direct decarboxylation of acetol is not one of the main acetol direct deoxygenation pathways and therefore CO 2 is not a product of direct decarboxylation of acetol.This contradicts the postulation made by Chen et al. [27] stating that the primary products of acetol deoxygenation are CO, CO 2 , CH 4 , and CH 3 + which subsequently undergo secondary reactions.
Other products of acetol conversion include acetone and acetic acid [27,73,74] and their yields are plotted in Figure S6 in addition to propanoic acid and H 2 .It is known that the weak acidic silanol groups are efficient in catalysing the ketonization reactions of carboxylic acids forming acetone and CO 2 [75,76] .This mechanism supports also the observed increased acetone yield with temperature.The mechanism is further supported by the spectroscopic characterisation of zeolites indicating an abundance of silanol groups in the nanosized mesoporous catalyst.This catalyst exhibits a higher acetone formation while the nanosized crystal hinders the occurrence of secondary reactions of acetone.
Based on the obtained results, the conversion occurs through cascade reactions of decarbonylation and dehydration, and ketonization rather than direct decarboxylation with the acidic properties and crystal size affecting the extent of the corresponding steps.
In order to investigate the influence of the zeolites' properties on the olefins distribution, their percentage distribution is plotted in Fig. 8.The dominance of ethylene production over propylene and butenes is evident.All the catalysts demonstrate a similar behaviour with a constant percentage of each of the olefins with time.The production trend is ethylene > propylene > butenes, averaging ca.80%, 15% and 5% respectively.The only exception is at 450 °C where the ethylene production increases at the expense of the propylene percentage, ethylene being the dominant olefin for Com and Ds-com.This observation can be associated with the higher amount of strong acid sites present in the microsized catalysts compared to the nanosized zeolites and higher reactivity of higher olefins in oligomerization and aromatization reactions [23].
The reduction of the diffusion path in the nanosized counterparts does in fact result in more propylene, preventing the readsorption of primary olefinic products.
In addition, there is no apparent effect of temperature on the olefin distribution on HT-hie-20 in contrast to the rest of the catalysts with the percentage of ethylene increasing with temperature which is in agreement with the oligomerization/ cracking mechanism.HT-mic-20 displays an exceptional trend where due to moderate surface acidity, higher propylene selectivity is achieved by increasing the temperature from 550 to 650 °C, similar behaviour is observed during ethanol conversion [77].

Furfural
As shown in Fig. 9, the temperature increase favours furfural conversion with 600 °C being the most efficient temperature for its conversion providing higher conversion and higher resistance to coking (i.e., stable activity for longer ToS).The conversion data indicates that at lower temperatures (450 °C), the diffusion path and acid sites access are important for obtaining the cracking products with the nanosized catalysts and the Ds-com achieving higher conversions than the commercial microporous catalyst.Notably, the activity of the nanosized microporous catalyst is stable for the whole duration of the experimental run at temperatures lower than 650 °C.The conversions achieved with the desilicated counterpart (Ds-Com) show insensitivity to temperature as the same conversion and deactivation trends are observed at low temperatures, indicating the difficulty in converting the primary products at this temperature range.The conversion reaches its maximum at 600 °C.The conversion over HThie-20 increases with temperature and reaches its maximum also at 600 °C before decreasing again at 650 °C.While a stable conversion of furfural can be achieved at the lower temperatures, contrary to acetol, the activity of the catalyst degrades rapidly at high temperatures.This difference can be attributed to the different cracking mechanisms and the formation of benzofurans which are known coke precursors [78].The latter is supported by the postulated mechanism of acetol conversion suggested by Chen involving a series of consecutive reaction until the furan formation [27].In general, the deactivation of the catalysts over time is evident which can be attributed to the excessive formation of coke observed in the catalytic cracking of furfural [27].
The primary products of the conversion are CO and furan (Fig. 10) suggesting that the primary step for its Fig. 9 The conversion of furfural over the different zeolite catalysts as a function of time conversion is direct decarbonylation as proposed also by Che et al. [79].Furan is the most abundant specie at temperatures up to 600 °C indicating that it is difficult to further convert it at this temperature range.The latter is in agreement with other works reporting furan's resistance to cracking compared to furfural [80].Thus, furfural conversion is limited by an initial protonation step which is considered the rate-determining step [81].
Here it must be noted that the preliminary thermal run in the investigated temperature range did not produce any furan (see supporting information for more details) which is in accordance with observations by Vermeire et al. [82] where furan formation is not the preferred pathway during thermal decomposition of furfural at temperatures below 1000 K.
Based on the yields of the products obtained (Fig. 10), the presence of smaller crystal size in microporous catalyst allows more efficient conversion of furfural given that the acid sites density and relative strengths of Brønsted and Lewis sites are similar between the microporous counterparts.The secondary conversion includes the conversion of produced furan which undergoes ring-opening reactions, followed by a series of dehydration, decarbonylation and other reactions to form olefins, paraffins, and aromatics.Nevertheless, hydrogen transfer reactions cannot be excluded from the reaction scheme [80].
To elucidate the conversion mechanism and the effects of the different catalysts, the furan conversion was calculated.The furan conversion is calculated based on the overall conversion and on the assumption that it is the only primary product of furfural conversion together with CO as explained above (Fig. 11).This assumption can hold true for temperatures up to 600 °C as also reported by Chen et al. [83].At temperatures greater than 600 °C, however a relatively constant CO production is attained which allows to expect a minimal contribution from furan conversion and thus introduces an insignificant error.As shown, the hierarchical zeolites exhibit a better performance in converting furan at lower temperatures compared to their corresponding microporous HZSM-5 and a poorer one at 650 °C.At 600 °C, both the effects of the porosity and crystal size are apparent.The hierarchical nanosized zeolite converts almost all furan produced (even with the lowest overall acidity), while the microporous microsized zeolite results in the Fig. 10 The yields of olefins, CO, CO 2 , furan, and aromatics as a function of time for the catalytic conversion of furfural over Com, Ds-Com, HT-mic-20, and HT-hie-20 catalysts lowest conversion.As for Ds-Com, it successfully converts around 60% of the furan produced over time before the conversion of furan starts decreasing gradually.The conversion of furfural for the same time is however stable, showing that while the Ds-Com has the active sites to convert the furfural molecule into furan and CO, it fails to convert the furan produced into other compounds indicating a competitive conversion mechanism with higher selectivity to primary products, and plausibly highlighting the importance of Lewis acid sites in converting furan.This is however not the case for HT-mic-20, over which the conversion of furan molecules decreases dramatically then increases again to reach 80% conversion at the end of the time of experiment.
The formation of olefins, CO 2 and aromatics is considerably low (below 5 wt.%) regardless of the catalyst used, and positively correlates with increasing temperature supporting the need for high temperatures to break the intermediates into light products [84].
The wt.% of individual olefins are plotted in Fig. 12.The main olefins produced are ethylene and propylene.The amount of butenes does not exceed 10% and showed a stable trend except for the increase at 650 °C over Com and Ds-Com indicating the importance of Brønsted acid sites (Com and Ds-Com exhibit a higher amount of Brønsted acid sites compared to the nanosized catalysts) which are catalyzing the conversion of ethylene [85].As a general trend, the increase in temperature results in a decrease in ethylene and an increase in propylene produced.This however is not observed at a temperature of 600 °C.The propylene selectivity is increased when reducing the crystal size of the catalyst, which is even more pronounced on hierarchical zeolites.Introducing mesoporosity alone boosts the formation of propylene while ethylene remains the main olefinic product.However, reducing the crystal size, even in a microporous structure, alters the olefin distribution by a drastic reduction in ethylene accompanied with an increase in propylene.This can be explained by Diels-Alder condensation reaction of furan which results in formation of benzofuran and water [83].A decarbonylation step of benzofuran and furan yields benzene and allene, respectively.Allene then undergoes multiple reactions, and combines into propylene (mainly), benzene, toluene, and ethylene.This mechanism can be seen in Figure S8, as these compounds display a similar trend.

Guaiacol
Guaiacol is considered as a representative lignin molecule with adjacent methoxy and hydroxyl functional groups which induce steric effects during their conversion in the micropores of HZSM5 catalysts [86].The effect of temperature on the conversion of guaiacol is illustrated in Fig. 13.The conversion is positively correlated with temperature increase for all catalysts.HT-hie-20 exhibits a superior performance over HT-mic-20 at all temperatures except 600 °C while Ds-Com has a lower activity than Com except The catalytic cracking of guaiacol over the different catalysts has revealed a large production of CO, followed by Fig. 12 The percentage distribution of C 2 -C 4 olefins formed during the conversion of furfural over the different zeolites Fig. 13 The conversion of guaiacol over the different zeolitic catalysts in function of time CH 4 and almost no CO 2 , indicating that CO and CH 4 are primary products of the conversion suggesting that the prevailing reactions include decarbonylation and demethylation of guaiacol molecule [29,79] while the minimal amount of CO 2 indicate limited decarboxylation extent.As shown in Fig. 14, high temperatures are needed for the complete conversion of guaiacol.The production of olefins, aromatics and single carbon molecules was proportional to the cracking temperature.Demethylation and decarbonylation reactions result in production of phenols which in turn are subject to ring opening reactions leading to the production of ethylene and propylene [79].The rest of the phenols are transformed to a pool of oxygenates and hydrocarbons which will either crack into olefins or aromatize into coke precursors [29].Although this is common regardless of the catalyst, it is important to note that more CO is produced over the hierarchical zeolites compared to the microporous ones, especially at 550 °C.The presence of mesopores allows the occurrence of the ring opening reactions which are accompanied with a release of a CO molecule.In addition to the aforementioned reactions, the lower yield of aromatics produced over Ds-Com and HT-hie-20 compared to their corresponding microporous catalysts is indicative of the cracking reactions of these molecules over the exposed acidic sites.In fact, the microporous channels of Com and HT-mic-20 hinder the cracking on the inner active sites since guaiacol has a kinetic diameter bigger than the pore opening of the channels [79].Moreover, the low yield of olefins in general is also in agreement with the cracking mechanism and can be attributed to the fact that olefins are actually by-products of the cracking of guaiacol and not primary products [2].
The catalyst design affects the olefin distribution as shown in Fig. 15.It is noticeable that the selectivity to butenes increases with temperature.The hierarchical system in Ds-Com and HT-hie-20 enables the formation of butenes, a bigger molecule than ethylene and propylene, reaching almost 50% of the olefins produced.Furthermore, there is an apparent relation between the amount of propylene and butenes produced over the Ds-Com.
The reduction in the crystal size diminishes the percentage of ethylene in favour of propylene and butenes at lower temperatures, whereas butenes production is more pronounced over these two catalysts compared to the larger crystal zeolites.Similar behaviour can also be seen over the These findings support the postulation of olefins being products of ring opening reactions and is in line with the classical acid strength-dimerization relationship, suggesting that dimerization reactions of ethylene cannot be the prevailing mechanism for such behaviour over the nanosized catalysts.Moreover, the propylene percentage seems unaffected by the temperature as seen for HT-mic-20 and HT-hie-20.It can therefore be concluded that enhanced diffusion rate does allow for more efficient mass transfer from the catalyst pores preventing further cracking and/or aromatization of the C 3 and C 4 olefins produced.

Conclusions
The effects of microstructure and inherent acidic properties of ZSM-5 zeolites on the primary products and coking characteristics during conversion of model compounds of biomass-derived oxygenates in a fixed bed reactor at a temperature range of 450-650 °C have been investigated.Detailed characterization of the catalysts prepared with topdown as well as bottom-up approaches indicates that apart from the porous structure, preparation method and templates used can affect the acidic characteristics and the relative Brønsted and Lewis acid sites.For all the model compounds tested at a WHSV of 5 h −1 , optimum conversion temperatures lie in the vicinity of 600 °C providing almost complete conversion with increased resistance to coking for the time on stream (75-120 min).
The hierarchical porous system and other confinement effects from the zeolite pores alter the product selectivity irrespectively of the molecule size.Even though not investigated in this study, the increased resistance to coking could suggest lower refractory coke deposition due to the presence of increased Lewis acidity.
In particular, for the small acetol molecule one should not expect large impact of the porous system.Nevertheless, introduction of mesoporosity and nanosized zeolite crystal with moderate acidity increases the selectivity towards propylene.Similar trend is observed during furfural conversion with propylene selectivity being positively correlated with the reduction in crystal size and introduction of a hierarchical porous system.For the larger molecule of guaiacol, introduction of mesoporosity increases the initial conversion at the low temperature end (450 °C) but induces a faster deactivation at longer time on stream.Reduction in crystal size offers a more coke resistant performance which is also partly attributed to the decreased acidity of the catalysts.Temperatures in the vicinity of 550 and 600 °C offer a compromise between overall conversion and deactivation resistance.Nanosized crystal alters the olefins distribution by promoting the faster diffusion of the produced intermediates giving rise to higher amount of butenes and propylene.Concluding, crystal size and mesoporosity induce easier access to active sites improving mass transfer but also alter the location, type, and strength of acid sites allowing for higher yields of primary and intermediate products, and thus olefins.

Fig. 1
Fig. 1 Schematic diagram of the experimental set-up

Fig. 2
Fig. 2 XRD patterns of the zeolite catalysts Fig. 3 SEM images of a Com, b Ds-Com, c HT-mic-20 and d HThie-20

Fig. 8
Fig. 8 Distribution of C 2 -C 4 olefins formed during the conversion of acetol over the different zeolites

Fig. 11
Fig. 11 Furan conversion as function of time at different temperatures over the zeolite catalysts

Fig. 14
Fig.14 The yields of olefins, single carbon compounds, and aromatics in function of temperature for the catalytic cracking of guaiacol over the tested catalysts

Fig. 15
Fig.15 The percentage distribution of C 2 -C 4 olefins formed during the cracking of guaiacol over the different zeolites

Table 2
Textural properties and acidity of the zeolite catalysts