Suppressing the aromatic cycle of the dimethyl ether to hydrocarbons reaction on zeolites

The influence of the zeolite framework type on the conversion of dimethyl ether (DME) to hydrocarbons (DTH) was investigated for *MRE, MFI and TON zeolite catalysts. Remarkable differences in the catalytic performance of the materials were observed. In particular, the *MRE zeolite showed an exceptionally high yield of olefins (90%) with a substantial ratio of products in the chain length range C 5 -C 11 . Additionally, the longevity of the *MRE zeolite clearly exceeded previously reported data. The comparison of mechanistic parameters (Hydrogen-Transfer-Index HTI C i and C 3 /C 2 ratio) demonstrated for this zeolite, that the formation of aromatics in the reaction network can be almost completely suppressed under suitable reaction conditions. By varying the reaction parameters temperature, DME partial pressure and weight hourly space velocity ( WHSV ), it was possible to identify the optimal combination of selectivity and deactivation resistance for each material. The olefin-rich DTH product of the *MRE zeolite offers manifold possibilities for further conversion to valuable renewably produced low-emission fuels like gasoline or jet fuel.


Introduction
The use of non-fossil carbon sources for fuel synthesis enables the reduction of CO 2 emissions in the transport sector.The production of hydrocarbons from regenerative carbon sources via biomass to liquid (BTL) or power to liquid (PTL) processes offers promising opportunities on the way to CO 2 neutrality in the transport sector.The conversion of dimethyl ether (DME) to hydrocarbons (DTH) could be one step in such a process chain, complementary to the well-known methanol to hydrocarbons (MTH) reaction [1][2][3].Using DME as feedstock is beneficial because: (I) Single-step DME synthesis enables the utilization of syngas with lower H 2 /CO ratio as well as CO 2 -rich syngas in a thermodynamically more advantageous way compared to methanol synthesis [4][5][6].(II) It offers an advantageous hydrocarbon production route with lower adiabatic temperature rise due to lower heat duty and (III) A higher methylation activity, reactivity, hydrocarbon product selectivity and conversion capacity can be implemented [7][8][9][10][11].
Starting from methanol or DME, light olefins are formed as primary hydrocarbon products, which are subsequently transformed to a mixture of paraffins, olefins and aromatics via secondary reactions in a so-called dual cycle mechanism comprising an olefin and an aromatic cycle [12].
The formation of a hydrocarbon pool (HCP) inside the catalyst bed promotes hydrocarbon production in an autocatalytic manner [13,14].Reactant conversion and product composition of the MTH/DTH reaction highly depend on the catalyst and the process conditions.Typically, zeolites with the MFI framework type are used in MTH/DTH processes [1][2][3].A matter of particular interest is the process modification to minimize the yield of aromatics and maximize the yield of olefins.Olefins with a chain length of 5 or more C-atoms (C 5+ ) offer a wide range of possibilities for further processing.For example, low-emission gasoline could be produced by hydrogenation, or synthetic jet fuel could be generated by dimerization.A promising way to generate an enhanced olefin yield would be the elimination of the aromatic cycle during the product formation by selecting a zeolite catalyst with suitable shape selectivity.Efforts in this direction were made using TON framework type zeolites with unidirectional medium (10-ring) pore systems [15][16][17].However, since the aromatic cycle is not completely suppressed, the aromatic products remain mainly in the narrow zeolite pores due to spatial confinement.Thus, a high olefin content prevails in the reactor effluent, but the lifetime of the catalyst is very short due to rapid deactivation [15][16][17].The preparation of mesoporous nanoscale zeolite crystals was reported to counteract this rapid deactivation and increase the longevity of the catalysts to a certain extent [18][19][20].*MRE zeolites exhibit slightly larger one-dimensional 10-ring tubular pores (Table 1).These high-silica zeolites tend to form a highly complex disordered framework.The degree of disorder is variable and it does not block the pores [21].*MRE zeolites were reported to exhibit a high C 2 -C 4 selectivity, employing a high contact time and a reaction temperature below 673 K [22].Catalyst deactivation occurred vastly faster than by using a MFI zeolite under the same conditions.A noticeable yield of paraffins in conjunction with a low yield of aromatics indicated a high deposition of unsaturated compounds on the catalyst, leading to fast coking [22].Teketel et al. [15,23] reported a similar longevity of TON and *MRE zeolites in MTH conversion, with a significant higher selectivity to aromatics for *MRE.By investigating the MTH reaction on a *MRE zeolite in the temperature range of 723-973 K, Zhang et al. [24] achieved a high selectivity to C 3 -C 6 olefins with an *MRE catalyst at comparatively low reaction temperature (723 K), while complete conversion of reactants was only maintained up to a time-on-stream (TOS) of 2.5 h, followed by a decrease of conversion below 20% in a total of less than 10 h TOS.
Within this work, compared to previous reports, deviating results of the catalytic DTH performance on *MRE zeolites were encountered regarding product selectivity and longevity.This enables usage in a modified process directing towards high yields of C 5+ products with very low amounts of aromatics.The aim of the present work is to elucidate this remarkable behavior and elaborate the mechanistic uniqueness compared to common materials.Therefore, we characterized zeolite catalysts of the framework types *MRE, MFI and TON and compared their performances in the conversion of DME.A MFI catalyst was chosen as technically established material, setting the benchmark in reactivity and longevity.TON zeolites are considered to have advantageous structural properties regarding shape selective suppression of aromatics formation thus setting the benchmark in product selectivity.By comparing specific parameters of product selectivity in the case of MFI, TON and *MRE zeolites, peculiarities of the latter zeolite have been identified.Moreover, the influence of varying reaction conditions on the product formation mechanism and longevity of the catalysts has been investigated, determining the most favorable reaction environment to maximize product selectivity and minimize catalyst deactivation.

Catalyst materials and preparation
Medium pore zeolites with three different framework types (MFI, *MRE and TON) were employed as catalysts and the main characteristics of the pore channels for the three framework types are listed in Table 1.Zeolite powders in NH 4 -form of each framework type were purchased from an industrial supplier.To remove possible impurities, the samples were ion-exchanged three times in 1 M NH 4 NO 3 solutions for 2 h at 348 K.After drying for 12 h at 353 K and 1 kPa in a vacuum drying oven, the samples were calcined for 6 h at 823 K (heating rate of 2 K min -1 ) under static air conditions to obtain the protonated form.

Catalyst characterization
Structural properties were analyzed by powder X-ray diffraction (XRD) on a PANalytical X'Pert PRO diffractometer using Cu Kα radiation (λ = 0.1541 nm).The measurements were carried out over 2 h in the 2θrange of 5-80 • .
Textural properties were determined by Ar-physisorption measurements at 87 K on a Quantachrome Autosorb IQ-MP2 in a procedure according to ISO 15901-3:2007 [28].The samples were outgassed for 12 h at 623 K under vacuum.Isotherms (in the range of relative pressure 0-0.99 p p 0 -1 ) were used to calculate the specific surface with the BET model (accounting for Rouquerol criteria) [29,30], internal and external surface with the t-plot method and micropore volume by a non-localized density functional theory (NLDFT) kernel.
Particle morphology was investigated by scanning electron microscopy (SEM).SEM images were generated on a Zeiss Gemini SEM with thermal Schottky field emitter cathodes.
Chemical composition was studied by the following characterization techniques: Elemental analysis was determined by using inductively coupled plasma optical emission spectroscopy (ICP-OES) on an Agilent 725 ICP-OES spectrometer.A mixture of 6 mL HNO 3 (65%), 2 mL HCl (30%) and 1 mL HF (40%) (Suprapur®, Merck) was used for microwave digestion of the powder samples. 29Si and 27 Al MAS NMR measurements were conducted on a JEOL JNM-ECZ400R spectrometer, equipped with a 9.4 T Oxford cryo magnet, using a 3.2 mm Automas Solid State Probehead. 29Si MAS MNR spectra were recorded at a resonance frequency of 79.45 MHz with a sample spinning frequency of 15 kHz.4096 scans were accumulated upon π/2 single pulse excitation with 0.9 s pulse width and 10 s recycle delay. 29Al MAS NMR spectra were recorded at a resonance frequency of 104.20 MHz with a spinning frequency of kHz.6000 scans were accumulated upon a π/2 single pulse excitation with 2 s pulse width and 2 s recycle delay.The 29 Si chemical shifts were calibrated relative to TMS (tetramethylsilane) and Q 8 M 8 (Octakis(trimethylsiloxy)silsesquioxane) as secondary standard (δ = 12.5 ppm, lowfield signal).The 27 Al chemical shifts were calibrated relative to potassium aluminum sulfate KAl(SO 4 ) 2 •12 H 2 O (δ = 0 ppm).
The acidic character of the catalysts was investigated by FTIR of adsorbed pyridine in a setup, as described in previous reports [31].The measurements were performed in transmission mode with a Tensor FTIR spectrometer (Bruker) in an in-house made heatable and evacuable reaction cell with water-cooled CaF 2 windows.Prior to the measurements, unsupported wafers of compressed catalyst (r = 1 cm, m = 50 mg) were outgassed for 1 h at 673 K (heating rate of 10 K min -1 ) under vacuum, followed by cooling to room temperature.At 473 K, a spectrum of the unloaded sample was recorded to eliminate background signals later.Dried pyridine with Ar as carrier gas (60 mL min -1 ) was then introduced into the reaction cell until the surface of the sample was saturated.After evacuation of excess gas phase, the reaction cell was heated to 673 K.Then, FTIR spectra were recorded in steps of 50 K. Background signals were subtracted from the IR spectra of pyridine-adsorbed samples at 523 K.The peaks at 1545 cm -1 and cm -1 were used to calculate the concentration of Brønsted and Lewis acid sites, respectively, according to the equation of Emeis [32].

Catalytic tests
The catalytic experiments were performed in a continuously operating laboratory plant with a stainless steel tubular fixed-bed reactor (i.d. = 12 mm).Gas flows were dosed into the reactor by calibrated thermal mass flow controllers from Wagner-Bronkhorst.Before filling the reactor, the protonated zeolite samples were pressed, crushed and sieved to 224-300 µm fractions.To ensure minimal temperature gradients, prevent channeling and avoid problems with axial dispersion throughout the catalytic-bed, the catalyst was diluted with inert silicon carbide (SiC, 100-180 µm sieve fraction) in a catalyst/SiC mass ratio of 1:10.A coaxially movable thermocouple inside the reactor was used to

Table 1
Main characteristics of the pore systems for the employed zeolites (data from [25][26][27]).check isothermal conditions (ΔT < 2 K).Catalyst mass depends on the desired weight hourly space velocity (WHSV, defined as mass flow rate of DME divided by the mass of catalyst in the reactor), because a constant mass flow rate of 0.5 g DME h -1 was set for all experiments.Before each experiment, the reactor was heated to reaction temperature in a flow of 80 mL min -1 N 2 .Then, under the same gas flow, the system pressure is adjusted by a manually controlled pressure regulating valve.To start the experiment, a gas mixture of DME and N 2 (x DME = 5%) was fed into the reactor.
The reactor effluent was transferred to a heated line (453 K), where a partial stream was fed into an online gas chromatograph for analysis of the reaction products.Measurements were carried out every hour.An Agilent 7890B gas chromatograph equipped with a HP-PLOT/Q capillary column (Agilent 19095 P-Q04) and coupled with a flame ionization detector (FID) was used for the quantification of products (hydrocarbons) and reactants (DME and methanol).Additionally, a combination of three HayeSep Q columns (Agilent CP1305) plus one MolSieve 5 Å column (Agilent CP1306) coupled with a thermal conductivity detector (TCD) was used for the quantification of H 2 , N 2 , CO and CO 2 .
The remaining effluent stream was collected in a cold trap cooled to 77 K with liquid N 2 .Sampling started at the beginning of the experiment at 100% conversion until a conversion of 40% was reached.This interval is defined as a "catalyst life cycle" in the following, forming the basis for the calculation of the life cycle yield (LCY).Thereafter, at lower conversion consecutive reactions, such as dehydrogenation of coke components on the catalyst with reactants resulting in increased methane formation, became visible in some cases [33].Afterwards, the products of the cold trap were slowly defrosted to a temperature of 278 K, so that the volatile components evaporated and the hydrocarbon products with five or more C atoms (C 5+ ) plus water remained liquid in the collecting vessel of the cold trap.After phase separation, the organic phase was analyzed in a Reformulyzer M4 (PAC) with a multi-dimensional PIONA gas chromatography method in compliance with ISO 22854:2021 [34].A more detailed description of the product analysis procedure is given in the Supplementary Material.

Catalyst properties
The main characteristics of the three different zeolite framework types are listed in Table 2. Si/Al ratios of 40.2, 73.5 and 50.4 were measured for MFI, *MRE and TON, respectively.
Ar-physisorption isotherms are shown in Fig. S1 of the Supplementary Material.BET surface areas of 459, 195 and 228 m 2 g -1 are slightly above literature data for comparable materials (433 m 2 g -1 for MFI (Si/ Al ≈ 40) [35], 148 m 2 g -1 for *MRE (Si/Al ≈ 72) [24] and 196 m 2 g -1 for TON (Si/Al ≈ 50) [15]).The deviation can be explained by the use of Ar as adsorbate, as advised by IUPAC [36], whereas N 2 was used in the literature reports.The external/internal surface ratio of 0.18 for MFI is lower compared to 0.27 for TON and 0.59 for *MRE.In line with this, the micropore volume of 0.22 cm 3 g -1 for MFI is higher compared to 0.10 cm 3 g -1 for TON and 0.07 cm 3 g -1 for *MRE.These observations show a higher proportion of accessible intracrystalline volume inside the MFI framework structure compared to TON and *MRE.This is likely due to the three-dimensional pore system of MFI with linked channels versus the unidirectional one-dimensional TON and *MRE pores.
FTIR spectra of pyridine adsorbed at 523 K are depicted in Fig. S2.The Brønsted acid site density was calculated as 235 µmol g -1 for MFI, 170 µmol g -1 for TON and 97 µmol g -1 for *MRE from the peaks at 1545 cm -1 .This order is consistent with the Si/Al ratio of the materials.The Lewis/Brønsted acid site ratio of 0.25 for *MRE is high compared to 0.16 for MFI and 0.13 for TON, indicating a high proportion of Lewis acidic Al in non-framework positions (extra framework Al, EFAl).
Representative SEM images of the catalyst samples are shown in Fig. 1.The MFI particles are aggregates in the range of 0.1-1 µm, composed of slaty, irregular crystals in the range of 20-200 nm.Depending on the material synthesis, different morphologies of MFI crystals can be formed, which have been classified by Jacobs and Martens [37].The MFI morphology shown in Fig. 1. corresponds to the class of spherulitic agglomerates.The *MRE sample revealed a morphology of irregular aggregates in the range of 0.5-2 µm, assembled by intergrowth of mainly rod/needle and plate shaped crystals with <100 nm length.Similar observations on *MRE crystal morphology have been published previously [15,38].The SEM images of TON show a morphology of columnar aggregates in the range of 0.1-2 µm, formed by uniformly unidirectional intergrown rod/needle shaped crystals with <100 nm length.This is consistent with previous reports and is considered a typical morphology for zeolite framework types with one-dimensional 10 ring pore systems [15,39].
XRD diffractograms of the three zeolites are depicted in Fig. 2. The patterns expose a high degree of crystallinity for all samples.Furthermore, the patterns match the reported structures of orthorhombic MFI [35], *MRE [15,26] and TON [15,27].No impurities of the MFI and *MRE zeolites could be observed.The small peak at approximately 21.7 • 2θ in the TON diffractogram indicates a minor cristobalite impurity [27,40]. 29Si MAS NMR spectra of highly siliceous zeolites are expected to Similar behavior is widely reported in the literature [45,46].For TON, 29 Si MAS NMR spectra of dealuminated materials were reported to reveal four symmetrical peaks with relative intensities in the ratio of 2:1:1:2, which represent the four symmetrically inequivalent T-sites of the unit cell [47,48].However, the total Q 4 [Al 0 ] signal of the TON sample is not symmetrical, as shown in Fig. 3 2).The high content of EFAl could be due to the disordered *MRE framework, as indicated by the symbol *, coming along with more lattice defects compared to the ordered structures.

Catalytic tests 3.2.1. Catalytic performance of the zeolites MFI, *MRE and TON
The DME conversion versus the cumulative conversion capacity for zeolites MFI, *MRE and TON is shown in Fig. 4. Experiments were carried out at 673 K and 100 kPa DME partial pressure.Due to a strongly varying DTH activity of the materials, catalyst loading in terms of WHSV was individually adjusted (6.0 h -1 for MFI, 1.5 h -1 for *MRE and 0.3 h -1 for TON).Thus, an initial conversion of 100% could be maintained for at least 24 h TOS in all experiments of this series.The highest cumulative conversion capacity was observed for MFI, which is known for its longevity in methanol or DME conversion to hydrocarbons when compared to other zeolites [22].The one-dimensional zeolites *MRE and TON deactivated more rapidly, while *MRE vastly outperformed TON.This observation deviates from previously published reports,  where a similar resistance to deactivation for *MRE and TON type catalysts was stated [15,23].According to the cumulative conversion capacity at 100% DME conversion, in combination with the respective WHSV, the activity of the materials can be ranked in the order MFI > *MRE > TON.The higher activity of MFI compared to *MRE and TON can be explained by the different dimensionality of the pore systems.The diffusion of reactants into the pores, the movement of the molecules through the pores and the release of reaction products from the pores is facilitated by the three-dimensional structure of the MFI pore channels with intracrystalline junctions and openings across the entire outer surface.In contrast, the non-interpenetrating pore channels of *MRE and TON run longitudinally through the entire crystal, so that diffusion of guest molecules occurs only alongside a single channel.The slightly different pore diameters of *MRE (5.3 × 5.6 Å [26]) and TON (5.5 × 4.5 Å [27]) are probably a major reason for the different activities of these two materials with the same dimensionality.
Stability of the materials can be analogously explained.For instance, Guisnet et al. [52,53] showed that the lifetime of zeolites in catalytic reactions for the conversion of hydrocarbons is linked to the dimensionality of the pore system.Accordingly, zeolite structures with one-dimensional pore systems are considered to be particularly sensitive to deactivation due to the blockage of entire pore channels.A small coke deposit in a pore channel may be sufficient to inhibit the activity of all active sites within the channel.Since coke formation and accumulation occurs preferably near the crystal outer surface [54], such deposits can directly block the entire channel in the case of a one-dimensional structure.In contrast, coke deposits in zeolites with a three-dimensional pore system without "trap-cages" may only block individual acid sites or small sections of the pore systems thus avoiding accumulation.Therefore, only small sections within the pore system initially deactivate and can be bypassed by redirecting intramolecular diffusion.In such zeolites, to whom the framework type MFI is allocated, the blockage of entire channels occurs only in the highly coked state and may be attributed to the formation of large, bulky deposits on the outer surface of the crystal [53,55,56].

Product yields for the zeolites MFI, *MRE and TON
The yield of C 5+ hydrocarbons versus DME conversion for the three catalysts is depicted in Fig. 5.The MFI zeolite revealed a relatively high and constant C 5+ yield between 55% and 60% at full conversion.Below 100% reactant conversion, a linear decrease of C 5+ yield was observed.In the case of the *MRE and TON zeolite, the C 5+ yield increased constantly in the beginning of the experiments at 100% conversion.The maximum C 5+ yield of 60% for *MRE and 61% for TON was reached below complete reactant conversion, at 92% and 95%, respectively.Afterwards, in both cases the C 5+ yield decreased with a slightly convex Fig. 3. 29 Si (top) and 27 Al (bottom) MAS NMR spectra of the zeolites.curvature.The corresponding data for the fractions of methane, ethane, ethene, propane, propene, butanes and butenes are shown in the Supplementary Material (Fig. S3).
Volume group type fractions of the liquid C 5 -C 11 product phase during a catalyst life cycle of the zeolites MFI, *MRE and TON, as defined in Chapter 2.3, are depicted in Fig. 6.In the C 5 and C 6 fractions a high content of paraffins, especially branched iso-paraffins, was detected in the case of the MFI catalyst.Regarding C 7+ components, an increasing content of aromatics was formed.The olefin content was consistently low, with only cyclic olefins being formed from C 8 upwards.In contrast, the *MRE catalyst led to a consistently high olefin content in all fractions and iso-olefins dominated the olefin fraction.Only a small content of paraffins and even less aromatics were detected in the liquid product phase.The TON catalyst gave a fairly balanced ratio of paraffins, olefins and aromatics.For low carbon numbers, the liquid product spectrum was dominated by olefins and paraffins.With increasing carbon numbers their proportion decreased in favor of aromatics.The volume fraction of hydrocarbons with 12 or more C-Atoms (C 12+ ) was below 1% for all three zeolites in the PIONA group type analysis.
By combination of the PIONA group type analysis data for the product group C 5+ with the average values of the C 1 -C 4 yields, the yield of all products (C 1 -C 11 ) was calculated for the three catalysts (Fig. 7).For MFI, the highest yield was achieved in the case of C 4 products.Paraffins were the dominant product group (52%), followed by almost similar proportions of olefins (26%) and aromatics (22%).For *MRE, C 4 compounds also represented the largest product fraction and olefins were by far the major product group (90%).Small amounts of paraffins (9%) and even smaller amounts of aromatics (<1%) were formed.The main product fraction in the case of TON was the C 3 fraction.Olefins were the largest hydrocarbon fraction (50%), followed by paraffins (37%) and aromatics (13%) with significantly higher proportions compared to *MRE.Conversion capacity, yield of liquid C 5+ phase as well as C 5+ mass fractions for a catalytic life cycle of all catalysts are listed in Table 3.The high olefin formation of the *MRE and TON zeolites seems to cause the rising selectivity to C 5+ products with increasing deactivation, which is shown in Fig. 5. Simonetti et al. [57,58] showed by studying the chain growth of light olefins using DME as methylation reagent, that for products in the chain length range <C 8 , olefin cracking has the highest activation barrier and thus the lowest reaction rate compared to methylation and hydride transfer.Guisnet et al. [53] reported that for zeolites with unequal acid strength within the pore system, the strongest, most active centers will deactivate first, thereby reducing the rate   of reaction steps with high activation barrier, like the cracking of higher olefins in this case.This explains the increased formation of C 5+ olefins during deactivation of *MRE and TON zeolites.Analysis of the yields, either online or offline after collection of the product mixtures, indicate major differences in product formation for the three zeolites.As widely reported in the literature, DTH product formation on zeolites occurs via the dual cycle model [12].The so-called Hydrogen-Transfer-Index (HTI Ci , defined as the ratio of saturated products to total products of carbon number i) is a qualitative indicator of the hydride transfer during the DTH reaction [59].Hydride transfer between olefins is mandatory for the formation of paraffins and aromatics in the reaction network.An indicator for the formation of products via the aromatic cycle of the dual cycle model is the formation of ethene.Ethene is inherently linked to alkylation/dealkylation of aromatics and can be excluded as cracking product of higher olefins [60].In addition, methylation activity of ethene in the olefin cycle of the dual cycle model is very low compared to propene and butenes [60].Since the oligomerization of light olefins is suppressed in the presence of methanol/DME [61,62], the conclusion that ethene leaves the reactor without a subsequent reaction is valid.In contrast, due to steric constraints in medium pore zeolites, propene formation differs, from a mechanistic point of view, from cracking reactions of polymethylated aromatics [63,64].Accordingly, for 10-ring zeolite framework types, propene is considered a key component of the olefin cycle which is formed mainly from olefin methylation and interconversion.Consequently, the C 3 /C 2 product ratio can be considered as an indicator for the intensity of the olefin cycle compared to the aromatics cycle.
Fig. 8 depicts the HTI C3 and the C 3 /C 2 ratio versus DME conversion.It is clearly visible, that at high conversion the HTI C3 of the *MRE catalyst was comparatively low, whereas the C 3 /C 2 ratio was exceptionally high.This means that the hydride transfer in the HCP is low and product formation hardly occurs via alkylation/dealkylation of aromatics.Based on these indicators, the conclusion can be drawn, that the aromatic cycle is suppressed on *MRE under the given conditions, which is also in accordance with the measured yields.These observations deviate from previously published data on *MRE type catalysts, where substantial amounts of aromatics, high HTI values and low C 3 /C 2 product ratios were reported [23].These differences are probably due to different reaction conditions, which will be considered in Chapter 3.2.3.Catalyst deactivation is associated with a reduction of active catalyst mass in reactor.This implies a reduction of the contact time (contact time τ 0 , defined as mass of catalyst in the reactor divided by the total gas volume flow in the reactor) and is expected to cause a decrease of HTI C3 with increasing catalyst deactivation [59].However, this effect is only observed for the MFI zeolite, whereas the *MRE and TON zeolites behave differently with an increasing HTI C3 during deactivation.
Since hydride transfer is a bimolecular reaction step in the HCP reaction network, the reaction rate is expected to decrease with increasing WHSV for a given catalyst.Despite a four times higher WHSV, significantly higher HTI C3 values were measured for the MFI catalyst at high conversion, compared to *MRE.The high hydride transfer rate explains the large fractions of paraffins and aromatics in the product spectra while the very low C 3 /C 2 ratio indicates enhanced product formation via cracking of methylated aromatics.
For TON, the HTI C3 was higher compared to the two other zeolites.Besides the structural properties of the TON framework, the comparatively high τ 0 may have a noticeable impact on the hydride transfer during the reaction.Resulting from the high hydride transfer rate, high amounts of aromatics are formed, whose mobility is severely restricted in the narrow TON pore channels.Polymethylated benzenes reside inside the TON crystal and block the channels, which explains the rapid deactivation and low cumulative conversion capacity of the material [15][16][17].The C 3 /C 2 ratio of TON is in between MFI and *MRE indicating that, despite the potentially high aromatics formation rate, dealkylation of aromatics is limited due to steric constraints of the zeolite framework.A significant part of the products is thus formed via the olefin cycle.Thereby, the performance of the TON catalyst, yielding a product composition between the two extremes *MRE (dominant olefin cycle) and MFI (dominant aromatic cycle), can be explained.

Variation of reaction conditions
The three catalysts were tested with varying temperature, DME partial pressure and WHSV to determine the influence of reaction conditions on the DTH reaction network and to investigate practical aspects like effects on product selectivity or catalyst lifetime.The conditions for experiments presented in Sections 3.2.1 and 3.2.2(T = 673 K, p DME = 100 kPa, WHSV = 6.0 h -1 for MFI, 1.5 h -1 for *MRE and 0.3 h -1 for TON) represent the base case for this test series.Starting from these values, one parameter was varied in each case.The cumulative conversion versus C 5+ product yield (top) and C 3 /C 2 ratio versus HTI C3 (bottom) of the MFI, *MRE and TON catalysts are shown in Fig. 9. C 5+ product yield, C 3 /C 2 ratio and HTI C3 are given as average values for the period from the beginning of the experiment until the point of 40% DME conversion (see above).Further data on this experimental series are shown in the Supplementary Material.
The cumulative conversion for the MFI catalyst decreased with increasing temperature, while for *MRE and TON a maximum cumulative conversion was achieved at 673 K under the given conditions.The initial DTH conversion in the case of the *MRE catalyst at 648 K was below 40% (Fig. S4) and thus too low for the criteria of the evaluation shown in Fig. 9.The maximized cumulative conversion in each case should be closely associated with a minimized formation of deactivating species.Formation of such species occurs through manifold pathways driven by reactants [65][66][67], interactions of reactants and HCP products [68,69] or HCP products alone [70,71].This can result in countervailing deactivation mechanisms, especially depending on the temperature.At low temperatures, a decreasing cracking rate of olefins and dealkylation rate of aromatics favors the formation of bulky molecules within the zeolite pores.Additionally, the built-up of the autocatalytic HCP is slowed down, resulting in a prolonged induction period [72].On the other hand, coking reactions, in which the oxygenate reactants act as precursors seem to proceed preferentially at higher temperatures [67].Based on early observations of Hutchings et al. [65], a mechanism of coke formation related to formaldehyde formation by hydride transfer of chemisorbed methanol was proposed lately [66,67,73,74].A direct disproportionation of the methylation agent DME to form formaldehyde is not possible.However, methanol is constantly generated during the catalytic reaction, by DME methylation or hydrolysis of DME [11].Formaldehyde can act as precursor for strongly adsorbed oxygen-containing surface species, which are converted to aromatics with increasing time on stream [67].Aromatics formed by this mechanism will not easily desorb from the active sites of the catalyst, causing deactivation by site blocking [67].A temperature dependency of this process was observed and higher temperatures led to higher amounts of coke deposits [67].For MFI, strongly chemisorbed aromatics can enhance product formation via alkylation/dealkylation according to the aromatic cycle, which explains the decreasing C 3 /C 2 ratio.As more aromatics are retained on the active sites inside the catalyst at higher temperatures, C 5+ product yield and cumulative conversion decrease because of faster deactivation.This assumption is underlined by an increased C 1 yield, known as a by-product of coke formation from oxygenates, at higher temperatures (Table S1).In contrast, the formation of bulky, but mobile, molecules at lower temperatures seems to have only a slight negative effect in the three-dimensional MFI pore system with spacious channel intersections.For *MRE, a decrease of the cumulative conversion and C 5+ product yield were observed together with a drastically reduced C 3 /C 2 ratio and slightly increased HTI C3 by increasing the temperature from 673 K to 698 K.These findings suggest an increased formation of aromatics, as well as an increased contribution of the aromatic cycle to the reaction.Similar observations were published by Zhang et al. [24] for temperatures in the range of Fig. 9. Cumulative conversion versus C 5+ product yield (top) and C 3 /C 2 ratio versus HTI C3 (bottom) of the zeolites MFI, *MRE and TON for varying reaction conditions.The diamond symbols represent the base case conditions for each catalyst (T = 673 K, p DME = 100 kPa, WHSV = 6.0 h -1 for MFI, 1.5 h -1 for *MRE and 0.3 h -1 for TON), other symbols refer to the variation of one parameter.
723-973 K.The enhanced formation of products via the aromatic cycle could be attributed to aromatics produced by the oxygenate coking mechanism.It can be assumed that, due to the steric constraints of the *MRE framework, the formation of transition states for the dealkylation of strongly bound aromatics inside the pore channels are hindered [15].Therefore, alkylation of these aromatics can rapidly lead to deactivation by pore blockage, which could explain the lower cumulative conversion and lower C 5+ product yield at higher temperatures.In addition, an increased olefin cracking rate at higher temperatures may affect the C 5+ yield.Despite the reduced C 3 /C 2 ratio, the olefin cycle appears to remain dominant over the aromatic cycle in comparison to the other zeolites studied within this work.As a result of the high olefin concentration in the HCP inside the *MRE pores, the formaldehyde formed during oxygenate hydride transfer may react with olefins to form dienes.These in turn are assumed to be highly reactive for cyclization with subsequent hydride donation resulting in the formation of aromatics [75].Thus, this mechanism could contribute to both, the reduced C 3 /C 2 ratio and the increased HTI C3 .For TON, the C 3 /C 2 ratio decreased at higher temperatures, whereas the HTI C3 showed a maximum at 673 K.The little cumulative conversion at 648 K can be explained by reduced aromatic dealkylation at lower temperatures (higher C 3 /C 2 ratio), resulting in more bulky molecules (higher C 5+ yield) which block the narrow pore channels of the zeolite.The low HTI C3 can be linked to the fast deactivation, which shortens the average τ 0 by fast reduction of active catalyst mass [59].With increasing temperature, the aromatic dealkylation rate increases, which reduces pore blocking (lower C 3 /C 2 ratio and C 5+ yield).But at the same time, the rate of oxygenate hydride transfer, indicated by increasing C 1 yield (Table S1), may increase as well, facilitating the formation of deactivating species according to this mechanism.At 698 K, this disadvantage seems to outweigh the advantages of the increased dealkylation rate, which lowers the cumulative conversion compared to the experiment at 673 K.The low HTI C3 in turn may result from the short average τ 0 .
The impact of varying WHSV on product formation of the different catalysts is more distinct than the impact of temperature variation, since no countervailing effects occur.Increasing WHSV values cause a reduction of τ 0 .Lower τ 0 should reduce the probability of bimolecular reactions like olefin hydride transfer (lower HTI C3 ) and most likely favor monomolecular reactions like olefin cracking (lower C 5+ yield) or dealkylation of aromatics (lower C 3 /C 2 ratio and lower C 5+ yield).These effects should have a positive impact on the longevity of the catalysts, as reaction steps that contribute to the formation of deactivating species are reduced.However, to build up the HCP during the initial phase of the reaction, a critical concentration of autocatalytic species must be generated inside the catalyst bed.The critical contact time τ 0,crit is defined as the contact time needed to build up a HCP concentration where the autocatalytic reaction is dominating the overall reaction rate [76].The axially position where the accumulation of HCP species begins, depends on the contact time [77].From there, the HCP may firstly arise in the opposite direction to the reactant flow and the HCP can move later with the flow [77].If the accumulation of the active species begins early enough in the catalyst bed, the entire catalyst volume is used for the autocatalytic reaction during time on stream.In this case, the cumulative conversion capacity of a catalyst does not change, despite different WHSV values, as observed for MFI at WHSV values of 6.0 h -1 and 12.0 h -1 .If the accumulation zone of active species shifts towards the end of the catalyst bed due to shorter τ 0 , the entire catalyst volume is probably not used during the autocatalytic phase, since the spatial expansion of HCP formation against flow direction is limited.The phenomenon of an "unused" coke free inlet zone of a deactivated catalyst bed at short τ 0 has been photographically documented by Bleken et al. [78].In this case, the cumulative conversion capacity of a catalyst decreases with increasing WHSV, as observed for MFI at 18.0 h -1 , or for TON at 0.5 h -1 and 0.7 h -1 .A moderately lower cumulative conversion capacity in the case of MFI and a considerable decrease in the case of TON were observed with increasing WHSV at full initial conversion and these findings are in line with observations of Rojo-Gama et al. [79].The different behavior may be attributed to the influence of τ 0,crit , which was reported to be minimal on MFI and clearly noticeable on TON [76,79].Catalyst deactivation driven by reactants strongly depends on the influence of τ 0,crit , as this is the predominant deactivation mechanism until the HCP is built up and reactant-HCP interactions become dominant.This is underlined by the observation of enhanced methane formation at lower τ 0 /τ 0,crit ratios corresponding to increasing WHSV (Table S2).For very high WHSV, τ 0 drops below τ 0,crit , so that the initial accumulation of autocatalytic species can no longer occur at complete reactant conversion, as seen for *MRE when the WHSV was increased from 1.5 h -1 to 2.0 h -1 (Fig. S5).The cumulative conversion capacity of the catalyst was reduced in this case, since the deactivation of the catalyst caused by oxygenate precursors [66,67] is already underway during the slow build up of the HCP.This interpretation is supported by an increased C 1 yield at a WHSV of 2.0 h -1 (Table S2).When the WHSV was further increased to 2.5 h -1 , the catalytic activity of *MRE was insufficient to build up the HCP.Mainly methane was produced at a very low conversion level (<20%).
Lowering the DME partial pressure at constant WHSV and feed composition can be understood as a reduction of τ 0 , analogous to an increase of WHSV at constant partial pressure.However, as the reduced τ 0 results from an increased flow velocity inside the reactor, DME concentration at the reactor entrance is reduced at lower reactant partial pressure.This leads to a shift in the ratio of primary methylation reactions to secondary HCP interconversions, in favor of the latter.
Furthermore, a shortened τ 0 promotes monomolecular secondary reactions like olefin cracking over bimolecular secondary reactions like olefin hydride transfer.The consequences of these effects are represented in the experimental data by a lower C 5+ product yield, C 3 /C 2 ratio and HTI C3 at reduced DME partial pressures.This is completely in line with literature data for MFI zeolites which describe, that at low reactant partial pressure formation of short-chain olefins is promoted and aromatics formation is reduced [22,80,81].Since the catalytic activity of the MFI zeolite is high enough to compensate for the shortened τ 0 , lower DME partial pressure had a perceptible positive effect on the cumulative conversion capacity of the material by reducing the formation of bulky unsaturated products, which can act as coke precursors.Contrary to a reduction of the WHSV at constant DME partial pressure, the reduction of the DME partial pressure at constant WHSV causes a mitigation of τ 0,crit besides shortened τ 0 , due to the lower DME concentration [76].This relationship is particularly supported by the fact that for *MRE the HCP was build up at complete conversion for each level of DME partial pressure.At both 25 and 50 kPa DME partial pressure, τ 0 falls below the value of operating at 100 kPa DME partial pressure and WHSV of 2.5 h -1 where, as shown above, the catalytic activity of the zeolite was insufficient to build up the HCP.Since bimolecular hydride transfer and subsequent reaction steps of aromatics formation generally appear to be very weak for *MRE zeolites, the combined effects of shortened τ 0 and low DME concentration seem to have only a minor impact on the longevity of the catalyst.Rather, decreasing τ 0 /τ 0,crit ratio appears to accelerate reactant-induced deactivation, mitigating the cumulative conversion at lower DME partial pressure levels.For the TON zeolite, lowering the DME partial pressure has the same effect on the cumulative conversion as for *MRE.

Conclusions
The catalytic DTH conversion on zeolites with the framework types MFI, *MRE and TON revealed notable different shape selectivity of the materials.High quantities of paraffins and aromatics were formed on MFI, which decreased to medium concentrations on TON, in favor of olefins.On *MRE, almost exclusively olefins were produced, extinguishing other hydrocarbons from the product spectrum.By examination of HTI C3 and C 3 /C 2 it can be assumed that formation of products via the aromatic cycle is almost completely suppressed inside the *MRE pores.Thus, the exceptional high content of more than 90% olefins within the C 3 -C 11 product fraction is most probably generated via the olefin cycle.Simultaneously, significantly increased *MRE catalyst stability was found compared to previous reports [15,23,24], enabling a cumulative conversion capacity of more than 100 g DME g Catalyst -1 . This can be attributed to the minimized formation of deactivating aromatic species inside the catalyst pores.
Furthermore, a strong dependency of DTH product formation and catalyst deactivation on reaction conditions was observed.It became evident, that the effects of altered reaction environment strongly depend on the nature of the zeolite catalyst, in particular on the dimensionality of the pore system.Thus, it was impossible to derive general trends that are valid for different zeolite framework types.Most beneficial conditions for DTH conversion on zeolites *MRE and TON were medium temperature (673 K), low WHSV (1.5 h -1 and 0.3 h -1 , respectively) and high DME partial pressure (100 kPa).Under these conditions, the disadvantages of lower temperatures (slow build-up of the autocatalytic HCP and low aromatic dealkylation rate), higher temperatures (formation of strongly chemisorbed aromatics), low τ 0 /τ 0,crit ratio (slow HCP build-up) and low DME concentration (increase of secondary reactions) can be minimized.Such conditions counteract the steric constraints of the narrow one-dimensional 10-ring pore channels enabling a high cumulative conversion capacity and a high C 5+ yield.For MFI, a lower reaction temperature, comparably high WHSV and low DME partial pressure are favorable for high cumulative conversion capacity and/or C 5+ yield.As the three-dimensional pore system is accessible for partial substituted methylbenzenes, which facilitate fast HCP build-up and exhibit high reactivity as well as high tolerance for secondary reactions, no inhibition by the corresponding mechanistic effects inside the zeolite channels was observed.This way, a wider range of reaction conditions can be utilized for the material, where coking driven by reactants can be minimized.
The new insights into DME conversion on *MRE demonstrate, that this material was undervalued in the past and offers a highly attractive perspective to new MTH/DTH process routes.The combination of a remarkably high cumulative conversion capacity with the unique product selectivity could become a pathway for the production of synthetic jet fuel, e.g. by recycling of light olefins in the process and dimerization/oligomerization of higher olefins in a subsequent process step.Moreover, a great potential for optimizing the catalytic performance by customized design of the *MRE catalyst is evident.For instance, unfavorable material properties such as the high ratio of Lewis/Brønsted acid sites could be reduced.Additionally, proven concepts for improving zeolite catalysts, such as the introduction of mesopores and the development of hierarchical structures, could be employed.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

reveal Q 4 [
Al 0 ] signals in the range of − 110 to − 117 ppm, crystallographically corresponding to distinct Si sites with varying Si-Si distances and Si-O-Si angles[41].The insertion of Al in the framework broadens the peaks and draws some intensity to the Q 4 [Al 1 ] resonance around − 105 ppm.SiOH type structural defects (Q 3 [Al 0 ]) can be observed at − 102 ppm.Orthorhombic [Si-O]-MFI was reported to reveal 12 resonances in an asymmetric arrangement, assigned to the different T-sites of the unit cell[42,43].Due to the Si/Al ratio of 40, the MFI sample of this work is expected to be in the orthorhombic phase[44].The broad overlain total Q 4 [Al 0 ] signal of MFI presented in Fig.3(top) comprises a main peak at − 112.2 ppm with a shoulder at − 115.3 ppm.Furthermore, the broad signal at lower shift indicates the presence of Al (Q 4 [Al 1 ] signal at − 105.5 ppm) and SiOH (Q 3 [Al 0 ] signal at − 101.5 ppm).

Table 2
[49]).This might be caused by a non-uniform distribution of Al over the inequivalent T-sites[49].The broad and weak signal beginning at − 100 ppm is formed by the combination of Q 4 [Al 1 ] and Q 3 [Al 0 ] peaks.For *MRE, no29Si NMR spectra of highly siliceous samples with clearly defined Q Main characteristics of the employed zeolites.
4 [Al 0 ] signals of different T-sites could be found in the literature.The highly symmetric peak at − 113 ppm, which was observed for the *MRE sample, coincides with reports of Al-containing *MRE zeolites in the literature [50,51].The lower signal intensity at − 105 ppm in relation to the main peak indicates a lower Al content of the *MRE sample compared to MFI and TON.The 27 Al NMR spectra in Fig. 3 (bottom) show the typical signals of a ICP-OES.b t-plot.c NLDFT.d Pyridine FTIR.B. Niethammer et al. tetrahedrally coordinated framework Al at ~55 ppm and octahedrally coordinated extra framework Al (EFAl) at ~3 ppm.The signal intensity of the *MRE sample at ~55 ppm was lower compared to MFI and TON, which can be attributed to the lower Al content of the material.Furthermore, it is clearly noticeable, that the content of EFAl in *MRE exceeds the EFAl contents in MFI and TON.This is in good agreement with the comparatively high content of Lewis acid sites of the *MRE sample, probably caused by EFAl, which was detected by pyridine adsorption measurements (Table

Table 3
Performance parameters for the zeolites MFI, *MRE and TON.