Experimental implementation of a catalytic membrane reactor for the direct synthesis of DME from H2+CO/CO2

1 Experimental implementation of a catalytic membrane reactor for the direct synthesis of DME from H2+CO/CO2 Pablo Rodriguez-Vegaa, Ainara Atekaa*, Izumi Kumakirib, Hector Vicentea, Javier Ereñaa, Andres T. Aguayoa, Javier Bilbaoa a Department of Chemical Engineering, University of the Basque Country UPV/EHU, P.O. Box 644, 48080 Bilbao, Spain b Graduate School of Science and Technology for Innovation, Graduate School Science and Engineering, Yamaguchi University, Ube, 755-8611, Japan *Corresponding author. Tel.: 34-94-6015361. E-mail address: ainara.atek[email protected] This is the accepted manuscript of the article that appeared in final form in Chemical Engineering Science 234 : (2021) // Article ID 116396, which has been published in final form at https://doi.org/10.1016/j.ces.2020.116396. © 2021 Elsevier under CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) 2 Abstract The direct synthesis of dimethyl ether (DME) by the hydrogenation of CO2 and CO2/COx mixtures has been studied in an original packed bed membrane reactor (PBMR). The role of the hydrophilic LTA zeolite membrane is to remove H2O from the reaction medium, reducing therefore the thermodynamic limitations of methanol synthesis and dehydration stages. LTA zeolite has the best permeation properties among the studied zeolites (LTX and SOD). The experiments were carried out using a CuOZnO-ZrO2/SAPO-11 catalyst at 275-325 ºC, 10-40 bar, space time of 10 gcat h (molC)-1 and using in the permeate section a sweeping gas flowrate of the same composition as that fed to the reaction section. The results (DME yield, CO2 conversion and product distribution) of the PBMR are compared with those obtained in PBR without membrane. In the hydrogenation of CO2, a DME yield of 12 % and a CO2 conversion of 20 % are obtained at 275 ºC, 40 bar and space time of 10 gcat h (molC)-1 with a great catalyst stability. Keywords: dimethyl ether, CO2, syngas, membrane reactor, LTA zeolite, deactivation 3 1. Introduction The utilization of a membrane in catalytic process reactors pursues reducing both energy consumption and production costs, as well as minimizing the environmental impact. This strategy is part of the engineering approaches for process intensification and has a growing implantation in fuels and chemical products synthesis processes from sustainable sources (N Diban et al., 2013; Cannilla et al., 2017; Tian et al., 2018). Among these processes, the direct synthesis of dimethyl ether (DME) is considered one of the most attractive routes for the large scale CO2 valorization (Olah et al., 2009; Rafiee et al., 2018; Leonzio, 2018). DME has upward commercial interest as automotive and domestic fuel (has a cetane number of 56) (Arcoumanis et al., 2008; Semelsberger et al., 2006) and as reactant (alternative to methanol) for the production of raw chemicals (olefins and aromatics) (Pérez-Uriarte et al., 2016; Cordero-Lanzac et al., 2018) and the production of H2 in vehicles (through steam reforming) (Shimoda et al., 2011; Oar-Arteta et al., 2016). It is also widely used as refrigerant and spray (Good and Francisco, 2003), and in oil extraction (Javanmard et al., 2019). The reaction system for its direct synthesis from CO and CO2 hydrogenation involves: Methanol synthesis: CO + 2H2 ↔ CH3OH (1) CO2 + 3H2 ↔ CH3OH+ H2O (2) Reverse water gas shift (rWGS): CO2 + H2 ↔ CO + H2O (3) Methanol dehydration to DME: 2CH3OH ↔ CH3OCH3 + H2O (4) Hydrocarbons formation (undesired): CO +3H2 ↔ CH4 + H2O (5) Using bifunctional catalysts, methanol synthesis and WGS reactions are catalyzed by the metallic function, while methanol dehydration by the acid function. Performing 4 methanol dehydration in situ in the same reactor displaces the thermodynamic equilibrium of the methanol synthesis reactions. The thermodynamic advantages over methanol synthesis and the synthesis of DME in two separate reaction stages (methanol synthesis and its dehydration) have been quantified in the literature, and are particularly interesting to favor CO2 conversion when co-fed with syngas (Chen et al., 2016; Ateka et al., 2017). Catalyst preparation has received great attention (Sun et al., 2014; Catizzone et al., 2018; Mondal and Yadav, 2019), in particular, seeking to promote CO2 conversion and improve stability. The most studied metallic function is CuO-ZnO-Al2O3 with different metallic oxides such as MgO, CeO2, MnO or ZrO2 among others and their wellestablished behavior in the synthesis of methanol is reported. The metallic oxides were used to replace Al2O3 partially or totally or as promoters to increase the stability of Cu, attenuating its sintering (Frusteri et al., 2015; Bonura et al., 2016; Zhou et al., 2016). The conventional acid function used for methanol dehydration (γ-Al2O3) has been progressively replaced by less hydrophilic materials, such as some zeolites (HZSM-5 and ferrierite have been widely studied) (García-Trenco and Martínez, 2012; Cai et al., 2016; Frusteri et al., 2017) and silicoaluminophosphates (SAPOs), as SAPO-18 and SAPO-11 (Ateka et al., 2016, 2017). These acidic functions must have a high density of sites, but of moderate acidic strength to minimize the formation of coke. The conventional preparation method of bifunctional catalysts consists of physically mixing both functions (hybrid catalysts), and subsequent pelletizing to achieve an adequate particle size and the required mechanical resistance for its use in the reactor. As to the core-shell configuration of the particles regards, even if it can avoid negative effects resulting from the contact between the metallic and acid functions (García-Trenco et al., 2012; Bonura et al., 2020), the preparation method has greater difficulties than those of 5 the hybrid catalyst (Sánchez-Contador et al., 2018a). Considering the reaction system for the direct synthesis of DME (Eqs. (1)-(5)), the presence of H2O in the reaction medium conditions the thermodynamics of the process, limiting the conversion of CO and CO2 yielding DME. Consequently, the proposal of a hydrophilic membrane reactor that favors the separation of H2O from the medium is a challenge of relevant interest. Iliuta et al. (2010, 2011) have studied by simulation the removal of H2O from the reaction medium using a hydrophilic membrane or by means of adsorption in order to increase methanol and DME yields. Diban et al. (2013) have proposed a mathematical model to simulate an isothermal packed bed membrane reactor system, in order to determine the transport characteristics of the most suitable membrane for the direct synthesis of DME, co-feeding CO2 together with CO and H2. In this model unidirectional flow of the feed and the sweeping gas in counter-current mode are considered. These authors emphasize the need of a stable hydrophilic membrane (ZSM5, MOR, SIL) as those used for Fischer-Tropsch (FT) synthesis. They also pointed out the restrictions of H2O permeation selectivity at high reaction temperatures. Using for the simulation an ideal membrane (impermeable to methanol and DME) these authors (Diban et al., 2014) achieved an upgrade of DME yield of over 30 %, compared to that obtained without a membrane. Farsi et al. (2016) have compared, by simulation of a nonisothermal one-dimensional reactor, the performance for the direct synthesis of DME of a double membrane reactor (hydrogen-water), previously proposed for methanol synthesis (Farsi and Jahanmiri, 2012), with that of other reactors with hydrophilic membranes, hydrogen selective membranes and without membrane. De Falco et al. (2016) have demonstrated by simulation the advantages of two alternative reaction systems, thus, a zeolite membrane reactor and two series units, which consisted of a packed bed reactor and a water separation module. Subsequently, these authors (De 6 Falco et al., 2017a) have studied the effect of the operation conditions (CO2/COx and H2/COx ratios in the feed, temperature, pressure, space time) on CO2 conversion and DME yield, at industrial scale conditions with a non-isothermal model in the simulation of the membrane reactor. Among the simulation results, the achievement of the following results is to be highlighted: a DME yield of 0.75 (0.57 in the conventional reactor), a DME selectivity close to 100 % and COx and CO2 conversions of 0.75 and 0.69, respectively, enhancing in 15.4 % and 30.2 % the results in the conventional reactor. Additionally, the results have been improved by these authors (De Falco et al., 2017b), with an operation strategy called Double Recycling Loop DME (DRL-DME). This design consists of the utilization of a pure CO2 stream as sweeping gas in the permeation zone and recirculation to the reactor. Despite the advances in the preparation of zeolite membranes and in their application to membrane reactors in different catalytic processes (Bedard and Liu, 2018), the application of this type of membranes in methanol and DME synthesis is still at the preliminary stage, consisting of the proposal of H2O perm-selective membranes to be used in severe reaction conditions (Gallucci, 2018). Galluci et al. (2004) have experimentally upgraded the yield of methanol synthesis using a LTA membrane, and Fedosov et al. (2015) have used a LTA membrane (NaA Zeolite) for the dehydration of methanol towards DME. Gorbe et. al. (2018) have obtained results on the capacity of A zeolite to selectively separate water and methanol. The experiments have been carried out measuring the permeation of a H2, CO2 and H2O mixture, within the range of interest for methanol synthesis (160-240 ºC, 10-27 bar). Recently, Lee et al. (2021) have used a polyimide hollow fiber membrane reactor for methanol synthesis. In this work, a laboratory scale packed bed reactor provided with a LTA zeolite membrane (selected from a group of materials according to its permselectivity) has been 7 used in the direct synthesis of DME. The study has been conducted in a wide range of operating conditions, analyzing the effects of reaction temperature, pressure and CO2/COx ratio in the feed on various reaction indices (DME and methanol yield and selectivity, COx conversion and CO2 conversion) and on the stability of the catalyst. The experimental results are compared with those obtained in a reactor without membrane to quantify the shift attained in the yield of DME and the conversion of CO2 over the thermodynamic limitations. The composition of the used catalyst (CuO-ZnOZrO2/SAPO-11) has been optimized in previous works (Sánchez-Contador et al., 2018a, 2018b, 2018c). 2. Experimental 2.1. Catalyst preparation and characterization CuO-ZnO-ZrO2/SAPO-11 catalyst was prepared by physical mixture and subsequent pelletizing of the metallic and acid functions in a 1/2 mass ratio. The atomic Cu:Zn:Zr ratio of the metallic function (determined by ICP-OES analysis) is 2:0.75:1.21. The suitability of this composition and the preparation conditions for each of the functions have been established in previous works (Sánchez-Contador et al., 2018a, 2018b, 2018c). Likewise, the characterization methodologies for each of the functions comprising the catalyst and the bifunctional catalyst have been described in detail in these works. The most significant physico-chemical properties of the final catalyst are listed in Table 1. 8 Table 1. Physico-chemical properties of the CuO-ZnO-ZrO2/SAPO-11 bifunctional catalyst. Physical properties SBET (m2 g-1) Vmicro (cm3 g-1) Vmeso (cm3 g-1) dp (Å) 122 0.029 0.170 97.9 Metallic properties SCu (m 2 gCu -1 ) S´Cu (m 2 gcat -1 ) Cu dispersion (%) 53.5 6.3 8.2 Acid properties Total acidity (mmolNH3 g-1) Acid strength (kJ molNH3-1) 0.17 85 2.2. Membrane preparation and characterization Three types of microporous zeolite membranes, thus, Linde Type A (LTA), Linde Type X (LTX) and Sodalite (SOD) membranes were synthesized on stainless steel supports purchased from Mott Corporation. Table 2 shows the syntheses temperatures and times for the different membranes. The preparation conditions and materials are detailed in the Appendix (Tables A4-A6). Table 2. Hydrothermal conditions for the synthesis of the different membranes. Zeolite membrane Temperature (ºC) Time (h) Linde type A (LTA) 110 12 Linde type X (LTX) 90 8 Sodalite (SOD) 130 48 The morphology of the membranes was analyzed by scanning electron microscopy (SEM, in a JEOL/JSM-7000f equipment, equipped with a W filament, 3.5 eV resolution, attached to energy dispersive X-ray analyzer EDX, Oxford, 133 eV resolution). 9 Pervaporation and vapor permeation were performed to examine the dehydration performances of the membranes. The pervaporation (PV) of the different membranes was evaluated as described in the Appendix. The separation factors (Eqs. (A1) and (A2)) and fluxes (Eq. (A3)) of the PV experiments were calculated for EtOH/H2O and MeOH/H2O mixtures at 75 ºC and 60 ºC, respectively. In addition, permeation of different gases (He, H2, CO2 N2, CH4 and SF6) was measured at 100-200 ºC. Ideal selectivity was calculated from the permeation ratio of two gases. Thermal stability tests were performed by characterizing the membranes before and after thermally treating the membranes at 300 ºC with PV, VP, XRD and SEM. The temperature of the thermal treatment represents the DME synthesis temperature. The used techniques and analysis conditions have been described in the Appendix (section A3). 2.3. Packed bed membrane reactor (PBMR) The stainless steel packed bed membrane reactor (PBMR), shown in Fig. 1, structurally presents two concentric sections: the reaction section, in which the catalytic bed is located, surrounded by another concentric system that allows the sweep of the permeate flow (permeate section). 16 Table 3. Results of H2O pervaporation from EtOH /H2O and MeOH/H2O mixtures through the different membranes at 75 ºC and 60 ºC, respectively. Membrane Composition Feed (wt%) Permeate (wt%) Q (kg m -2 h -1 ) α LTA EtOH H2O 90 10 0.15 99.8 1.59 9329 LTA MeOH H2O 90 10 0.45 99.5 0.92 1486 LTX EtOH H2O 90 10 55.4 44.6 2.40 7.9 LTX MeOH H2O 90 10 69.4 30.6 2.00 2.9 SOD EtOH H2O 90 10 89.7 10.3 0.80 1.1 3.2. Membrane morphology Physical and morphological properties. The porous texture of the LTA zeolite powder has been characterized by CO2 adsorption at 0 ºC (in ASAP 2020 equipment, Micromeritics) (Fig. (A3)). To carry out the analysis, LTA has been prepared following the same procedure described in the Appendix, without introducing the metallic support in the autoclave. Well-defined CO2 adsorption isotherm has been obtained with a practically saturated shape in the range of low relative pressure. Pore width and micropore surface, calculated using the Dubinin-Radushkevich equation and by the Horvath-Kawazoe method, are 3.98 Å and 458 m2 g-1, respectively. These results suggest that the crystalline zeolite presents narrow micropores (3.98 Å). Structural properties. X ray diffraction (XRD) measurements have been used to study the crystallinity and purity of the LTA membrane. From the XRD pattern of the LTA membrane (Fig. 3) it can be confirmed, on one hand, that the zeolite has been appropriately synthesized (Belviso et al., 2018) and, on the other hand, its correct 17 crystallization on the stainless steel support (fundamental step within the membrane reactor configuration). Neat and well-defined peaks can be observed, indicating an absence of amorphous phase in the material. According to the database, the most intense peaks on the diffractogram clearly identified in Fig. 3 suggest that a crystalline phase with high purity supported on stainless steel can be achieved through this preparation method. The XRD pattern of steel can also be observed in Fig. 3, consisting of three peaks at higher values of 2θ angles (67.1, 79.5 and 129.1 degrees) attributed to the metallic support of the membrane (obtained in a PAN analytical Xpert Pro device). Fig. 3. XRD pattern of the synthesized LTA zeolite crystallized on stainless-steel support. In Fig. 4 SEM micrographs of the synthesized LTA zeolite are depicted. At the front view of the membrane cross-section (Fig. 4a) both phases (zeolite and support) are clearly differentiated. A zeolite layer of regular thickness has been deposited on the surface of the non-polished stainless steel support. The homogeneous texture of the 18 LTA membrane (Fig. 4b) suggests a high effectiveness of the preparation method for the purpose of achieving a complete coverage of the support with zeolite crystals, in accordance with that reported by Belviso et al. (2018). As it is observed in Fig. 4c LTA zeolite morphology consists of cubic aggregates. SEM characterization has demonstrated the correct synthesis method of LTA membranes, highlighting the formation of uniform thickness zeolite layer and its homogeneous distribution on the support surface. 19 Fig. 4. SEM images of the LTA zeolite membrane. Cross-section (a) and top view at different scales (b and c). 3.3. Stability of LTA membrane Pervaporation. The PV properties of the LTA membrane after its thermal treatment are shown in Table 4. Comparing with the results prior to the thermal treatment (Table 3) the separation factors (α), both for EtOH/H2O and MeOH/H2O mixtures, decrease after the treatment (2905 and 1011 respectively). Nevertheless, it maintains its hydrophilicity a) b) c) 20 towards both mixtures. Concerning the fluxes (Q) of the two feed compositions through the membrane, they are slightly raised from 1.59 to 1.70 kg m-2 h-1 for EtOH/H2O (≈ 6 %) and from 0.92 to 1.05 kg m-2 h-1 for MeOH/H2O (≈ 14 %). Table 4. EtOH/H2O and MeOH/H2O mixtures pervaporation results through the thermally treated LTA membrane at 75 ºC and 60 ºC, respectively. Composition Feed (wt%) Permeate (wt%) Q (kg m -2 h -1 ) α EtOH H 2 O 90 10 0.47 99.5 1.70 2905 MeOH H2O 90 10 0.75 99.2 1.05 1011 Vapor permeation. VP performance of the LTA membrane after its thermal treatment is summarized in Table 5. LTA membrane H2O permselectivity remains after its treatment (higher than 3000) and the vapor flux through it rises up to 1.32 kg m-2 h-1. Table 5. EtOH/H2O mixture vapor permeation results through the thermally treated LTA membrane at 125 ºC. Composition Feed (wt%) Permeate (wt%) Q (kg m -2 h -1 ) α EtOH H2O 90 10 0.19 99.8 1.32 3232 Single gas permeation. The permeances for different gases at different temperatures (100, 150 and 200 ºC) through the thermally treated LTA membrane are shown in Table 6. Moreover, in Table 7 the ideal selectivities of different gases at 100, 150 and 200 ºC through the thermally treated LTA membrane have been gathered. Increasing temperature up to 200 ºC, a slight increment in the gas permeances through the 21 thermally treated LTA membrane is observed. In spite of this permeance increase, the analysis demonstrates that even at high temperatures, the LTA membrane has low gas permeability and, therefore, is able to operate at temperatures near to those required in the synthesis of DME. Table 6. Gas permeances through the thermally treated LTA membrane at 100, 150 and 200 ºC (in 10-9 mol m-2 s-1 Pa-1). Temperature (ºC) He H2 CO2 N2 CH4 SF6 100 2.39 2.90 0.56 0.78 0.93 0.31 150 3.47 4.60 0.94 1.24 1.61 0.55 200 9.24 13.2 3.18 3.61 4.34 1.66 Table 7. Ideal selectivities for different gas mixtures through thermally treated LTA membrane at 100, 150 and 200 ºC. Temperature (ºC) H 2 /N 2 H 2 /CH 4 CO 2 /N 2 CO 2 /CH 4 He/SF6 H 2 /SF 6 100 3.7 3.1 0.3 0.2 10 11 150 3.7 2.8 0.8 0.6 6 8 200 3.7 3.0 0.9 0.7 6 8 Inert character. The null activity of the membrane in the reactions involved in the synthesis of DME (synthesis of methanol and its dehydration to DME) has been ascertained by means of experiments feeding syngas and co-feeding methanol with syngas, respectively. In both cases the conversion and the formation of paraffins and coke is null. 3.4. PBMR, Effect of reaction temperature Fig. 5 shows an example of the evolution with time on stream (TOS) of the molar fractions of the reaction products (DME, MeOH, H2O) at the reactor outlet in the 22 reaction section (RS) and in the permeate section (PS). These results, shown as an example, correspond to CO2 hydrogenation at 275 ºC and 40 bar. Under these conditions, the advance of the reverse WGS reaction is facilitated, what leads to a high formation of H2O. It should be noted that this is a key reaction for the production of DME since CO is more effective than CO2 under these conditions (Aguayo et al., 2007; Ateka et al., 2018). The low concentrations of DME and methanol in the PS, show that the membrane is perm-selective towards these molecules (kinetic diameter of ca. 4.3 Å and 3.8 Å, respectively, and 2.6 Å for H2O), specially for DME. In Fig. A2, the concentration values of all the gaseous components in the reaction and permeate sections are shown for different reaction temperature and different CO2/COx ratios in the feed. The similar H2O concentration values in the reaction and permeation sections observed in Figs. 5 and A2 reveal a good degree of H2O separation attained with the membrane. Nonetheless, the undesired partial permeation of methanol and DME also takes place. The high pressure and temperature required for this reaction have a great responsibility on limiting the permeation selectivity of H2O with respect to oxygenates. 23 30 60 90 120 150 180 0 0.005 0.010 0.015 0.040 0.045 0.050 30 60 90 120 150 1800 0.005 0.010 0.015 0.040 0.045 0.050 b) PS TOS (min)TOS (min) a) RS H2O DME MeOH Molar Fraction Fig. 5. Evolution of H2O, DME and methanol molar fractions with time on stream. (a) results in the reaction section (RS) and (b) results in the permeate section (PS). Reaction conditions: 275 ºC; 40 bar; 10 gcat h (molC)-1; CO2/COx, 1; H2/COx, 3. Permeate conditions: equal composition and flow rate (60 cm3 min-1) as in the reaction section. The common temperature range for DME synthesis studied in the literature is 250300 ºC, since the temperature is limited by the thermodynamic equilibrium (Ateka et al., 2017) and also to preserve the hydrothermal stability of Cu in the catalyst. Nevertheless, the utilization of the membrane reactor (PBMR) diminishes the thermodynamic limitations, due to the lower H2O concentration in the reaction medium, permitting the displacement of the thermodynamic equilibrium of H2O formation reactions (methanol synthesis, reverse WGS and methanol dehydration). Consequently, it allows operating at higher temperatures achieving a higher conversion. Furthermore, the composition and conditions used for the preparation of the catalyst are adequate to confer an acceptable hydrothermal stability at 325 ºC. The fact that H2O concentration in the reaction 24 medium is lower than in the conventional reactor (PBR) is, a priori, another advantage of the PBMR, in which less Cu sintering is expected. Attending to this advantage the studied temperature range has been 275-325 ºC. Fig. 6 shows the effect of temperature in the different reaction indices at zero TOS (Figs. 6a,b) and on the stability of the catalyst (Fig. 6c) for certain reaction conditions (30 bar; 10 gcat h (molC)-1; CO2/COx, 0.5; H2/COx, 3) and H2+CO/CO2 feedstock. Fig. 6a shows that temperature has a great relevance on COx conversion, improving from 13.3 % at 275 ºC to 19.3 % at 325 ºC. DME yield enhances from 275 ºC until 300 ºC, achieving a maximum of 14.8 % at this temperature. Further increasing reaction temperature, a slight decay is observed due to, predictably, the thermodynamic limitation. However, CO2 conversion is favored upon rising temperature, reaching 17.5 % at 325 ºC. MeOH and paraffins yields continue rising with a maximum of 3.7 % and 1.1 %, respectively, at 325 ºC. As a consequence of the evolution of the individual yields with temperature, DME selectivity (Fig. 6b) drops when increasing temperature from 275 to 325 ºC, remaining quasi-constant at higher temperature. On the other hand, paraffins selectivity should be pointed out since it rises up to 5.6 % at 325 ºC. At this temperature, paraffins formation mechanisms are promoted, either from DME and methanol through the hydrocarbon pool mechanism (activated by the acid function) or through methanation or FischerTropsch mechanisms from CO and CO2 on the metallic function. The formation of hydrocarbons facilitates deactivation by coke deposition, due to their role as intermediates for aromatics formation, which condensate to coke. The effect of temperature on the evolution of DME yield with time on stream (TOS) depicted in Fig. 6c demonstrates how temperature affects the stability of the catalyst, which is lower the 25 higher the temperature is within the range of 275-300 ºC. Nevertheless, at 325 ºC catalyst stability is greater than at 300 ºC, which can be related to catalyst sintering. The lower DME yield at 325 ºC than at 300 ºC, at zero time on stream, has therefore lower deactivation as favorable counterpart. It should also be pointed out that deactivation at 275 ºC is significantly slow. 275 300 325 0 5 10 15 20 Yi (%) Temperature (ºC) DME MeOH HC 0 5 10 15 20 a) X CO x X CO 2 X CO 2 , XCO x (%) 275 300 325 0 20 40 60 80 100 b) S i (%) Temperature (ºC) DME MeOH HC 32 20 30 40 0 20 40 60 80 100 b) DME MeOH HC Si (%) Pressure (bar) 020 40 60 80 100 120 140 160 180 0 5 10 15 20 25 c) TOS (min) YDME (%) 40 bar 30 bar 20 bar Fig. 8. Effect of pressure on DME, MeOH and hydrocarbons yield, and COx and CO2 conversion (a) and on product selectivity (b) at zero time on stream; and the evolution of DME yield with time on stream at different reaction pressures (c). Reaction conditions: 300 º; 10 gcat h (molC)-1; CO2/COx, 0.5; H2/CO, 3. Permeate conditions: equal composition and flow rate (60 cm3 min-1) as in the reaction section. 33 The effect of pressure on the reaction indices feeding H2+CO2 has been studied at 275 ºC (Fig. 9). This temperature has been determined in Section 3.1 to be the most appropriate to achieve maximum DME yield with this feedstock. The increase of the operating pressure favors the linear rise of both COx conversion (from 3.0 to 14.9 %) and CO2 conversion (from 9.4 to 25.0 %) within 10-40 bar range (Fig. 9a). Concerning DME and MeOH yields, the enhancement with pressure is clearly observed whereas paraffins formation remains quasi-negligible, achieving a maximum of 0.01 % under 40 bar. Due the constant rise of the products yield within the studied pressure range, MeOH and DME selectivity (Fig. 9b) are barely affected by the increase of the operating pressure (from 23.9 to 20.0 % and from 76.1 to 80.0 %, respectively). Fig. 9c displays the deactivation of the catalyst with time on stream. It demonstrates that, at this temperature and feeding H2+CO2, the deactivation of the catalyst is practically negligible throughout 3 h of reaction at the studied pressure range. Attending to these results under different pressures, even though temperature is different for the two feeds, the effect of pressure for H2+CO/CO2 and H2+CO2 feeds is qualitatively similar. The drop of DME yield decreasing the pressure (interesting for reducing operating costs) is proportionally similar for both feeds, as well as the decay in COx conversion. Regarding DME selectivity, it is constant with pressure in both cases. Nonetheless, feeding syngas, a pressure decrease implies a rise of paraffins selectivity to the detriment of that of methanol (being favored CH4 synthesis and Fischer-Tropsch reactions with respect to methanol synthesis). Feeding CO2 at 275 ºC, on the contrary, paraffins formation is insignificant even under 10 bar. It is also remarkable that catalyst deactivation is slightly favored by the increase of pressure for syngas feeds at 325 ºC, and very slow for H2 +CO2 feeds at 275 ºC. 34 10 20 30 40 0 5 10 15 20 25 a) XCO 2 , XCOx (%) Yi (%) Pressure (bar) DME MeOH HC 0 5 10 15 20 25 XCO2 XCOx 10 20 30 40 0 20 40 60 80 100 b) DME MeOH HC S i (%) Pressure (bar) 35 020 40 60 80 100 120 140 160 180 0 5 10 15 c) TOS (min) YDME (%) 40 bar 30 bar 20 bar 10 bar Fig. 9. Effect of pressure on DME, MeOH and hydrocarbons yield, and COx and CO2 conversion (a) and on product selectivity (b) at zero time on stream; and the evolution of DME yield with time on stream at different pressures (c). Reaction conditions: 275 º; 10 gcat h (molC)-1; CO2/COx, 1; H2/CO, 3. Permeate conditions: equal composition and flow rate (60 cm3 min-1) as in the reaction section. 3.6. Comparison between PBMR and PBR The elimination of H2O from the reaction medium has two potential effects: i) the alteration of the theoretical thermodynamic equilibrium towards an apparent equilibrium, and ii) the increase of the reaction rate of some stages of the reaction. As aforementioned, according to the thermodynamics (Ateka et al., 2017), when CO2 concentration in the feedstock increases, the thermodynamic limitations of the reaction augment, due to the increment of H2O concentration in the reaction medium (formed by methanol synthesis (Eq. (2)) and rWGS (Eq. (3)) reactions). Besides, the H2O is 36 adsorbed on the acid sites of the catalyst competing with the reactants and reducing its capacity to dehydrate methanol (Jun et al., 2002) on the one hand, and, on the other hand, on the metallic sites limiting the methanol synthesis rate (Dadgar et al., 2016). Fig. 10 shows oxygenates yields (MeOH + DME) obtained for certain operating conditions at different temperatures. These results aim to verify how the removal of H2O from the reaction medium using a PBMR permits overtaking the thermodynamic equilibrium of the process predicted for PBR (without membrane for H2O removal). It can be observed that increasing temperature up to 325 ºC, the yield enhancement using a PBMR compared with the process thermodynamic equilibrium for PBR becomes more significant. 275 300 325 0 5 10 15 YOxyg. (%) Temperature (ºC) STD Thermodynamic Equilibrium Experimental PBMR STD Thermodynamics Overtaking Fig. 10. Comparison of the experimental values of oxygenates yields obtained in PBMR with the thermodynamic equilibrium values predicted without membrane. Reaction conditions: 20 bar; 10 gcat h (molC)-1; CO2/COx, 1; H2/COx, 3. Permeate conditions for the PBMR: equal composition and flow rate (60 cm3 min-1) as in the reaction section. 37 For studying quantitatively the influence of incorporating a membrane on the yield of oxygenates and on CO2 conversion Fig. 11 is presented. The comparison between oxygenates (DME and MeOH) yield evolution with temperature obtained in PBR and PBMR reactors feeding H2+CO2 is depicted in Fig. 11a. In all the cases, within the studied temperature range (275-325 ºC) and at 30 bar, oxygenates yield is enhanced using a PBMR. This improvement is greater when increasing temperature, from 2.80 to 6.57 % at 325 ºC. The conversion of CO2 is also greater in the PBMR within the studied temperature range, observing the highest enhancement (37 %) at 325 ºC (Fig. 11b). The results in Figs. 10 and 11 highlight that in the operation in the PBMR, the separation of H2O has a great effect on increasing the yield of oxygenates and CO2 conversion. As a consequence of the partial separation of H2O from the reaction medium, a pseudoequilibrium state is reached, whose conversion also decreases with increasing temperature. In this pseudo-equilibrium, the decrease of limit conversion with rising temperature is attenuated with respect to the thermodynamics prediction. Fig. 11c shows the effect of CO2/COx ratio in the feed from 0 to 1 on oxygenates yield at 325 ºC. Higher yield is observed in PBMR whatever the CO2 content in the feed. Being the most remarkable, the greatest upgrade obtained with the highest CO2 content in the feed, improving oxygenates yield by 86 % when feeding H2+CO2. 38 275 300 325 0 5 10 a) YOxyg. (%) Temperature (ºC) PBR PBMR 275 300 325 0 5 10 15 20 25 30 b) PBMR PBR X CO2 (%) Temperature (ºC) 39 0.0 0.5 1.0 0 10 20 30 40 50 YOxyg. (%) CO2/COx PBR PBMR c) Fig. 11. Comparison of the PBMR and PBR. Oxygenates yield (a) and CO2 conversion (b) at different temperatures; and oxygenates yield for different CO2/COx ratios in the feed (c). Reaction conditions: 325 ºC; 30 bar; 10 gcat h (molC)-1; H2/COx, 3; CO2/COx, 1. Permeate conditions for the PBMR: equal composition and flow rate (60 cm3 min-1) as in the reaction section. The effect of membrane utilization is more relevant with increasing space time. In Fig. 12, the yield of DME and the conversion of CO2 at zero time on stream can be observed for different space time values. As it can be observed, the better performance of the PBMR over the PBR is greater upon increasing space time (industrially operating conditions). Indeed, for the studied conditions (325 ºC, 30 bar) and a H2+COx feedstock with CO2/COx= 0.5, using a membrane reactor doubles the yield of DME obtained with a PBR for a space time of 100 g h (molC)-1, and boosts CO2 conversion in a major extent. That is, at large-scale production conditions, thus, at high space time values, the 40 separation of H2O from the reaction medium has a greater effect because it shifts the values of DME yield and CO2 conversion with respect to the thermodynamic equilibrium in the PBR. 020 40 60 80 100 0 10 20 30 40 Space time (g h (molC)-1) PBR PRMR YDME XCO2 YDME , XCO2 (%) Fig. 12. Comparison of the evolution with space time of DME yield and CO2 conversion in PBMR (continuous lines) and PBR (dashed lines). Reaction conditions: 325 ºC; 30 bar; H2/COx, 3; CO2/COx, 0.5. Although the removal of H2O from the reaction medium improves the reaction rates, it has an undesired side effect, because it is well established that the presence of H2O in the reaction medium reduces coke formation. Sierra et al. (2011) explain this effect by the attenuation of the formation of methoxy ions from methanol and DME. It is also well established that these ions are the intermediates precursors of hydrocarbons formation in the acid function (Bjørgen et al., 2007). In addition, the greater capacity of DME for the formation of methoxy ions is considered to be the main cause of the higher rate of hydrocarbons and coke formation compared to that of methanol (Cordero-Lanzac 41 et al., 2018; Ibáñez et al., 2017). This effect has been studied by determining the coke content in the used catalysts by means of thermogravimetric analyses of their combustion with air. For the conditions in Fig. 12 and CO2/COx = 0.5, coke content in 3 h is 0.32 wt% in the PBR and 0.61 wt% in the PBMR. This deposition of coke results in a partial deterioration of the properties of the catalyst. The textural, metallic and acid properties of the catalyst after its use under different reaction conditions have been gathered in Table A3. Consequently, and as expected, the removal of H2O from the reaction medium has the unfavorable effect of upturning coke content. However, this effect is not very relevant and has little consequence in the evolution of the reaction indices with TOS. To explain this low incidence of H2O separation on deactivation, it must be taken into account that the greater advance of the reaction in the PBMR also yields higher H2O content. Thus, despite the separation, a sufficient H2O concentration is present to limit the formation of coke. In addition, this problem is less relevant when feeding CO2, since the formation of H2O is favored and therefore, the attenuation of deactivation. The stability of the aforementioned results has been experimentally ascertained by means of repeated runs with the same membrane. It should be noted that the results in this work correspond to per-pass conversions. Indeed, to assess the effect of the reaction variables in the performance of the catalyst, the study has been carried out under kinetic regime, ensuring there is no thermodynamic restrictions conditioning the results; enabling therefore the comparison of PBMR and PBR. The scaling up this process should be carried out recycling the nonconverted reactants (H2, CO and CO2) after the condensation of the oxygenates and H2O. This recirculation strategy is used industrially in the synthesis of methanol (Bozzano and Manenti, 2016) and has also been proposed 48 polyimide hollow fiber membrane reactor. Chem. Eng. J. 403, 126457–126466. https://doi.org/10.1016/j.cej.2020.126457 Leonzio, G., 2018. State of art and perspectives about the production of methanol, dimethyl ether and syngas by carbon dioxide hydrogenation. J. CO2 Util. 27, 326– 354. https://doi.org/10.1016/j.jcou.2018.08.005 Mondal, U., Yadav, G.D., 2019. Perspective of dimethyl ether as fuel: Part I. Catalysis. J. CO2 Util. 32, 299–320. https://doi.org/10.1016/j.jcou.2019.02.003 Oar-Arteta, L., Remiro, A., Epron, F., Bion, N., Aguayo, A.T., Bilbao, J., Gayubo, A.G., 2016. Comparison of Noble Metaland Copper-Based Catalysts for the Step of Methanol Steam Reforming in the Dimethyl Ether Steam Reforming Process. Ind. Eng. Chem. Res. 55, 3546–3555. https://doi.org/10.1021/acs.iecr.6b00126 Olah, G.A., Goeppert, A., Prakash, G.K.S., 2009. Chemical recycling of carbon dioxide to methanol and dimethyl ether: From greenhouse gas to renewable, environmentally carbon neutral fuels and synthetic hydrocarbons. J. Org. Chem. 74, 487–498. https://doi.org/10.1021/jo801260f Osatiashtiani, A., Puértolas, B., Oliveira, C.C.S., Manayil, J.C., Barbero, B., Isaacs, M., Michailof, C., Heracleous, E., Pérez-Ramírez, J., Lee, A.F., Wilson, K., 2017. On the influence of Si:Al ratio and hierarchical porosity of FAU zeolites in solid acid catalysed esterification pretreatment of bio-oil. Biomass Convers. Biorefinery 7, 331–342. https://doi.org/10.1007/s13399-017-0254-x Perez-Carbajo, J., Balestra, S.R.G., Calero, S., Merkling, P.J., 2020. Effect of lattice shrinking on the migration of water within zeolite LTA. Microporous Mesoporous Mater. 293, 109808. https://doi.org/10.1016/j.micromeso.2019.109808 Pérez-Uriarte, P., Ateka, A., Aguayo, A.T., Gayubo, A.G., Bilbao, J., 2016. Kinetic model for the reaction of DME to olefins over a HZSM-5 zeolite catalyst. Chem. Eng. J. 302, 801–810. https://doi.org/10.1016/j.cej.2016.05.096 Rafiee, A., Rajab Khalilpour, K., Milani, D., Panahi, M., 2018. Trends in CO2 conversion and utilization: A review from process systems perspective. J. Environ. Chem. Eng. 6, 5771–5794. https://doi.org/10.1016/j.jece.2018.08.065 Sánchez-Contador, M., Ateka, A., Aguayo, A.T., Bilbao, J., 2018a. Direct synthesis of dimethyl ether from CO and CO2 over a core-shell structured CuO-ZnOZrO2@SAPO-11 catalyst. Fuel Process. Technol. 179, 258–268. https://doi.org/10.1016/j.fuproc.2018.07.009 Sánchez-Contador, M., Ateka, A., Aguayo, A.T., Bilbao, J., 2018b. Behavior of SAPO11 as acid function in the direct synthesis of dimethyl ether from syngas and CO2. J. Ind. Eng. Chem. 63, 245–254. https://doi.org/10.1016/j.jiec.2018.02.022 Sánchez-Contador, M., Ateka, A., Rodriguez-Vega, P., Bilbao, J., Aguayo, A.T., 2018c. Optimization of the Zr Content in the CuO-ZnO-ZrO2/SAPO-11 Catalyst for the Selective Hydrogenation of CO+CO2 Mixtures in the Direct Synthesis of Dimethyl Ether. Ind. Eng. Chem. Res. 57, 1169–1178. https://doi.org/10.1021/acs.iecr.7b04345 Semelsberger, T.A., Borup, R.L., Greene, H.L., 2006. Dimethyl ether (DME) as an alternative fuel. J. Power Sources 156, 497–511. 49 https://doi.org/10.1016/j.jpowsour.2005.05.082 Shimoda, N., Faungnawakij, K., Kikuchi, R., Eguchi, K., 2011. A study of various zeolites and CuFe2O4 spinel composite catalysts in steam reforming and hydrolysis of dimethyl ether. Int. J. Hydrogen Energy 36, 1433–1441. https://doi.org/http://dx.doi.org/10.1016/j.ijhydene.2010.10.088 Sierra, I., Ereña, J., Aguayo, A.T., Arandes, J.M., Olazar, M., Bilbao, J., 2011. Cofeeding water to attenuate deactivation of the catalyst metallic function (CuO– ZnO–Al2O3) by coke in the direct synthesis of dimethyl ether. Appl. Catal. B Environ. 106, 167–173. https://doi.org/http://dx.doi.org/10.1016/j.apcatb.2011.05.021 Sierra, I., Ereña, J., Aguayo, A.T., Ateka, A., Bilbao, J., 2013. Kinetic modelling for the dehydration of methanol to dimethyl ether over γ-Al2O3. Chem. Eng. Trans. 32, 613–618. https://doi.org/10.3303/cet1332103 Sun, J., Yang, G., Yoneyama, Y., Tsubaki, N., 2014. Catalysis chemistry of dimethyl ether synthesis. ACS Catal. 4, 3346–3356. https://doi.org/10.1021/cs500967j Tian, Y., Demirel, S.E., Hasan, M.M.F., Pistikopoulos, E.N., 2018. An overview of process systems engineering approaches for process intensification: State of the art. Chem. Eng. Process. - Process Intensif. 133, 160-210. https://doi.org/10.1016/j.cep.2018.07.014 Zhou, X., Su, T., Jiang, Y., Qin, Z., Ji, H., Guo, Z., 2016. CuO-Fe2O3-CeO2/HZSM-5 bifunctional catalyst hydrogenated CO2 for enhanced dimethyl ether synthesis. Chem. Eng. Sci. 153, 10–20. https://doi.org/https://doi.org/10.1016/j.ces.2016.07.007