Spectro-kinetics of the methanol to hydrocarbons reaction combining online product analysis with UV–vis and FTIR spectroscopies throughout the space time evolution

Spectro-kinetics of the methanol to hydrocarbons reaction combining online product analysis with UV–vis and FTIR spectroscopies throughout the space time evolution José Valecillos a, ⇑ , Hector Vicente a , Ana G. Gayubo a , Andrés T. Aguayo a , Pedro Castaño a,b, ⇑ a Department of Chemical Engineering, University of the Basque Country (UPV/EHU), P.O. Box 644, Bilbao 48080, Spain b Multiscale Reaction Engineering KAUST Catalysis Center (KCC), King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia article info Article history: Received 22 November 2021 Revised 3 February 2022 Accepted 23 February 2022 Available online 28 February 2022 Keywords: Methanol to hydrocarbons (MTH) ZSM-5 (MFI) zeolite In situ spectroscopy Reaction kinetics Coke deactivation abstract The well-studied methanol to hydrocarbons reaction over a ZSM-5 zeolite catalyst has been used to develop a spectro-kinetic approach to obtain an overall reaction mechanism involving both retained species and gas-phase products. We combined two in situ spectroscopic techniques (ultraviolet–visible and Fourier-transform infrared spectroscopies) with online product analysis to obtain the timeand space time-resolved evolution of the entire reaction media. A ZSM-5 zeolite catalyst was tested in two commercial spectroscopic cells at 400 °C using different space times (different inlet flow rates). Specifically, our work focusses on the effect of the space time (key parameter in any kinetic study) and how to tune other parameters such as partial pressure of methanol to resolve, from the spectroscopic and gas-phase points of view, the mechanisms of reaction and deactivation. Our approach reinforces the previous interpretation of these two combined networks in the selected reaction, thus, proving that the spectro-kinetic approach is a robust methodology to simultaneously build overall reaction and deactivation mechanisms. Ó2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction The catalytic methanol to hydrocarbons (MTH) reaction is the basis for the development of a robust alternative process for the production of gasoline, light olefins (ethylene, propylene, and butylenes), and aromatics [1–6]. The most-used catalysts are based on medium-pore ZSM-5 zeolites and small-pore SAPO-34 zeotypes [3,6–10] for the development of the methanol-to-propylene (MTP) and methanol-to-olefin (MTO) process technologies, respectively. The ZSM-5 catalysts are more stable than SAPO-34 or similar small-pore catalysts (such as SAPO-18) because the ZSM-5 channels facilitate the diffusion of species that are both reaction intermediates and coke precursors, significantly reducing the coke formation and deactivation rates [11,12]. In addition, SAPO-34 and similar small-pore catalysts are remarkably selective to light olefins because of their shape selectivity [12,13], albeit they undergo faster deactivation that can be controlled by increasing the water concentration in the reaction medium [14]. However, the proper balance of acid properties of ZSM-5 catalysts increases the selectivity of light olefins, particularly propylene, with the surplus advantage of their stability provided by their channel structures and balanced acidity [10,15]. The uncertainty of the reaction mechanism and the rapid catalyst deactivation have been key queries for generating hundreds of research studies, with interesting results clarifying these issues, summarized in various reviews [1,6,16–18]. In the most general sense, the current broader consensus is that the MTH reaction proceeds through a mechanism comprising initiation, autocatalysis, and deactivation periods. In the initiation period, methanol and dimethyl ether are adsorbed on the acid sites forming surface methoxy species that react to yield oxygenated intermediates. These intermediates are decomposed into olefins that also react, providing aromatics through oligomerization, cyclization, and hydrogen transfer reactions. Once there is an incipient formation of olefins and aromatics, the autocatalytic period starts with the https://doi.org/10.1016/j.jcat.2022.02.021 0021-9517/Ó2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Abbreviations: FTIR, Fourier-transform infrared; GC, gas chromatograph; MTH, methanol to hydrocarbons; MTO, methanol to olefins; MTP, methanol to propylene; MS, mass spectrometer; UV–vis, ultraviolet-visible Variables; F M0 , weight or carbon-based molar flow rate of methanol in the feed; F W0 , weight flow rate of water in the feed; P M0 , methanol partial pressure in the feed; S i , carbon-based product iselectivity; t, time on stream; T, temperature; W, catalyst weight; X, carbon-based fractional conversion of oxygenates; Y i , carbon-based product iyield. ⇑ Corresponding authors at: Department of Chemical Engineering, University of the Basque Country (UPV/EHU), P.O. Box 644, Bilbao 48080, Spain. E-mail addresses: [email protected] (J. Valecillos), pedro.castano@kaust. edu.sa (P. Castaño). Journal of Catalysis 408 (2022) 115–127 Contents lists available at ScienceDirect Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat buildup of cationic olefins and aromatics, constituting a hydrocarbon pool that acts as a cocatalyst with the acid sites. The hydrocarbon pool species undergo methylation, oligomerization, alkylation, cracking, dealkylation, cyclization, and hydrogen transfer reactions resulting in more olefins and aromatics. The deactivation period starts with a profound degradation of hydrocarbon pool species, particularly aromatics, resulting in coke deposits that block the diffusion of reactants and products. Most of this knowledge acquisition has been possible with the analysis of retained species in the catalyst, following the proposal of a ‘‘carbon pool” that serves as an intermediate in forming gaseous products and coke [19]. From an experimental viewpoint, the investigation of retained species comprises ex situ and in situ analyses using various techniques [12]. The ex situ analysis consists of conducting the reaction in a conventional reactor and recovering the spent catalyst to analyze the retained species using thermogravimetric analysis (typically applying temperature-programmed oxidation), soluble species extraction, microscopies, spectroscopies, and other surface characterization techniques. Thermogravimetric analysis is extended to quantify the amount of retained species or coke [17,20], often providing information on their nature or location, supported by other techniques [21–23]. Likewise, the extraction of soluble species (discovered in the Guisnet group [24]) is often a routine analysis technique in the study of the chemistry of retained species in the MTH reaction [17,25]. Although the ex situ approach provides precise kinetic data based on the product analysis in the effluent stream, the retained species is analyzed at different conditions than the reaction conditions. This implies that the nature of the retained species may have been modified; therefore, the results of the ex situ analysis of retained species may be cautiously interpreted. The in situ analysis consists of carrying out the reaction in a vessel that allows the simultaneous characterization of the catalyst surface using normal thermogravimetric analyses, microscopies, or spectroscopies. The most common techniques for the in situ analysis of retained species include several in situ spectroscopic methodologies based on Fourier-transform infrared (FTIR), ultraviolet–visible (UV–vis), and 13 C solid-state nuclear magnetic resonance (NMR) spectroscopies [26–32]. These techniques allow determining the chemical nature of the hydrocarbon retained species in the catalyst by the spectroscopic signature. In recent years, the Weckhuysen group has published several studies covering the MTH reaction with different catalyst topologies or properties [33– 38] or under different reaction conditions [14,39,40], focusing their attention on the analysis of retained species. These experimental approaches solve the problem of analyzing retained species in conditions different from those of the reaction, but obtaining appropriate kinetic data is a challenge. In recent years, this involves the extended use of commercial spectroscopic cells, facilitating the labor of designing (homemade) spectroscopic cells [27,41] whose performance may differ from that of a model reactor. Despite the issue related to the performance of spectroscopic cells as model reactors, in situ spectroscopic methodologies provide valuable spectro-kinetic analyses because spectroscopic data can be obtained through time on stream or time-resolved spectroscopy [41,42]. The reported spectro-kinetic analyses for the MTH reaction usually consist of analyzing the evolution with time on stream of spectroscopic bands related to retained species, often hydrocarbon pool or coke species. Mores et al. [34] analyzed the time on stream evolution of the 415 nm band in the UV–vis spectrum (attributable to ions of polymethylbezenes) at different reaction temperatures. Upon fitting to a first-order kinetic model, they obtained Arrhenius plots that interpret the effect of temperature on the degradation of these species into coke in ZSM-5 catalysts. Goetze et al. [36] analyzed the time on stream evolution of the 1000 nm band in the UV–vis spectrum (attributable to coke) at three positions along the bed of two ZSM-5 catalysts and correlated this with the methanol conversion, finding that secondary coke (formed from olefins) grows more slowly in a less acidic ZSM-5 catalyst. Similarly, Borodina et al. [39,40] analyzed the time on stream evolution of various UV–vis bands related to the retained species in SAPO-34 and SSZ-13 at variable temperatures. They correlated these analyses with the deactivation kinetics to determine the nature of the active and deactivating species. In our previous works [11,12,15], we used in situ FTIR and UV– vis spectroscopies to assess the formation kinetics of retained species in the MTH reaction on various catalysts. The approach for the spectro-kinetic analysis consisted of analyzing the time on stream evolution of the most representative bands in the FTIR and UV–vis spectra with the conversion of oxygenates. The evolution of the 1570 and 1616 cm 1 bands in the FTIR spectrum during the MTH reaction on the ZSM-5 catalyst provides kinetic data related to the formation of active species, whereas that of the 1481 cm 1 band provides kinetic data related to the formation of deactivating species. Likewise, the time on stream evolution of the 400 nm (actives species) and 500 to 800 nm (coke) bands in the UV–vis spectrum provides spectro-kinetic data on forming several retained species. The results were consistent with the fact that the formation of retained species is slower on less acidic ZSM-5 catalysts, making the formation rates dependent on the concentration of Brønsted acid sites. However, the most relevant variable for the kinetic studies of flow reactors is the space time (with multiple definitions in the field of heterogeneous catalysis) [43], and there is a lack of studies related to spectro-kinetic analyses applied to the MTH reaction in the literature in which the space time is the main study variable. This work aims to develop a robust spectro-kinetic methodology analyzing the effect of time and space time in the wellknown MTH reaction on a ZSM-5 catalyst using commercial spectroscopic cells. The use of commercial cells facilitates the experimental work, providing rapid catalytic tests obtaining as much experimental data (gas phase product and surface species analyses) as possible. To this end, we used two commercial spectroscopic cells (designed for in situ FTIR and UV–vis spectrometries) with online product analysis (mass spectrometry). The timeand space time-resolved mass-spectrometry results were corrected to represent the entire gas-phase reaction media, calibrating the results with those for gas chromatography. The spectroscopic cell for FTIR spectroscopy is used in the transmission mode, whereas that for UV–vis spectroscopy is used in the diffuse reflectance mode. The results obtained using different space times, of the reaction media composition (mass spectrometry) are correlated to the chemistry of the surface species (FTIR and UV–vis spectrometry and retained species extraction) in terms of their time evolution. These correlations redevelop and verify the MTH reaction mechanisms on the ZSM-5 zeolite catalyst, and they may be applied to other reactions and catalysts, tuning several parameters, including partial pressure (as demonstrated in this work). 2. Experiment 2.1. Catalyst preparation and characterization We prepared the catalyst by mixing 50 wt% of ZSM-5 zeolite with 30 wt% of pseudo-boehmite and 20 wt% of a -alumina, followed by drying at room temperature for 24 h and 110 °C for 24 h, crushing and sieving at 0.125 to 0.300 mm, and calcining at 550 °C for 3 h. The ZSM-5 zeolite is a commercial NH 4 ZSM-5 zeolite (Zeolyst International, CBV8014, SiO 2 /Al 2 O 3 molar ratio = 80), which is calcined at 575 °C, as described in a previous work [15] to obtain the acid (protonic) form: HZSM-5. We characterized the José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 116 catalyst using conventional techniques [15], including x-ray photoelectron spectroscopy, x-ray diffraction, N 2 physisorption, NH 3 adsorption and temperature-programmed desorption (NH 3 -TPD), FTIR spectroscopy, and pyridine adsorption monitored with FTIR spectroscopy. Table 1 summarizes the main catalyst properties. The Brunauer, Emmett, and Teller (BET) specific surface area (S BET ) is within the expected value for a ZSM-5 zeolite (400 m 2 g 1 ), whereas it decreases for the catalyst due to the presence of alumina phases with a less porous specific surface area. Likewise, the zeolite has a more microporous specific surface area (S micro ) than the catalyst due to the absence of micropores in the alumina phases. The total acidity determined with NH 3 adsorption indicates that the zeolite is more acidic than the catalyst because of the dilution of the zeolite with alumina phases that have a low concentration of acid sites. Additionally, the strength of the acid sites determined with NH 3 -TPD indicates that the concentrations of strong acid sites (C SAS ) and weak acid sites (C WAS ) decrease for the catalyst compared with the zeolite, which is related to the dilution effect. However, the concentration of medium-strength acid sites (C MAS ) is slightly higher for the catalyst, indicating that the alumina phases (particularly those of pseudo-boehmite) provide this acidity feature. The concentration of Brønsted acid sites (C BAS ) and Lewis acid sites (C LAS ) is comparable between the zeolite and catalyst, considering that the values are lower for the catalyst because of the dilution of the zeolite with alumina phases. 2.2. Experiments in spectroscopic cells We conducted the MTH reaction in spectroscopic cells at constant conditions and variable space times to monitor the changes on the catalyst surface using FTIR or UV–vis spectroscopy. The conditions of the MTH reaction were temperature (T) = 400 °C, total pressure (P) = 1 bar, methanol partial pressure (P M0 ) = 0.04 or 0.16 bar, methanol flow rate (F M0 ) = 0.94 to 60 mmol h 1 , catalyst weight (W) = 0.012 or 0.048 g, and space time (W/F M0 ) = 0.2, 0.4, 0.8, 1.6, 3.2, 6.4, or 12.8 g h mol 1 . We used two commercial spectroscopic cells: (i) a Specac high-pressure, high-temperature chamber coupled with an FTIR spectrometer (Thermo Scientific, Nicolet 6700), and (ii) a Linkam stage (THMS600) coupled with a UV–vis spectrometer (Jasco, V-780) with a specially adapted compartment (Jasco, ARN-915i) for the cell. Fig. 1 presents a schematic representation of the experimental setup for using commercial spectroscopic cells as reaction systems. The FTIR spectroscopic cell works in the transmission mode, requiring thin samples, whereas the UV–vis spectroscopic cell works in the diffuse reflectance mode using an integrating sphere placed on the cell window. The feed consisted of N 2 with methanol vapor obtained by flowing N 2 through a saturator vessel containing liquid methanol at room temperature or immersed in an ice bath. We calculated the methanol concentration using the thermodynamic equilibrium data at room temperature or 1 °C (the measured temperature in the liquid methanol when the saturator is immersed in an ice bath). To analyze the gaseous effluent, we used a mass spectrometer (MS) (Pfeiffer Vacuum, OmniStar GSD 320O Series) continuously measuring the m/zsignals of 16, 18, 27, 29, 31, 41, 43, 45, 55, 56, 57, 78, and 91. Additionally, we analyzed the gaseous effluent of two experiments by sampling every 20 min and measuring the gas composition in a micro-gas chromatograph (GC) (Varian, CP4900) described in previous work [15] because the GC analysis provides a better identification and quantification of components. We correlated both analyses of the gaseous effluent and used the NIST database [44] to obtain the standard mass spectrum of the expected MTH reaction products, finding the following: The maximum abundance of light olefins (propene and butenes) is at m/z= 41. The analysis of ethene was excluded because it has a maximum abundance at m/z= 28, coinciding with nitrogen used as an inert gas. The maximum abundance of other aliphatic compounds with four or more carbon atoms is at m/z= 43 or 57 for paraffins and 55 for olefins. The analysis of ethane and propane was excluded because they have a maximum abundance at m/ z= 28 or 29, coinciding with nitrogen used as an inert gas. The maximum abundance of aromatic compounds is at m/z=78 or 91. The maximum abundance for methanol is at m/z= 31 and for dimethyl ether is at m/z= 45, being both reactive oxygenates easily distinguishable from hydrocarbons. Although the MS analysis does not provide an accurate product analysis, we approached some product groups by grouping m/zsignals: oxygenates (O) summing the intensity of m/z= 31 and 45 (I O =I 31 +2I 45 ), light olefins (LO) taking the intensity of m/z=41 (I LO =3I 41 ), heavy aliphatics (HA) summing the intensity of m/ z= 43, 55, and 57 (I HA =4I 43 +4I 55 +4I 57 ), and aromatics (BTX) summing the intensity of m/z= 78 and 91 (I BTX =6I 78 +7I 91 ). Then, considering that the reacted moles of oxygenates are equal to the formed moles of products in the gaseous effluent, we calculated the conversion of oxygenates based on the MS analysis (X MS )as follows: X MS ¼I LO þI HA þI BTX I O þI LO þI HA þI BTX ð1Þ and the yield of a product group ibased on the MS analysis (Y i MS ) is as follows: Y MS i ¼I i I O þI LO þI HA þI BTX ð2Þ These amounts (X MS and Y i MS ) were compared with those obtained using the GC data (Xand Y i ) for the two experiments, and using Origin 8.5, we found that the best correlation is as follows: X¼ aX MS  b c b þX MS  b ð3Þ Y i ¼ a i Y MS i  b i c i ðÞ b i þY MS i  b i ð4Þ where a,b,c,a i ,b i , and c i are the correlation parameters calculated for each product group (Table S1 in the Supporting Information provides these calculated parameters). Thus, these correlations allow estimating the conversion (X) and yield of product groups (Y i ) from the MS data. A typical experiment consisted of preparing a catalyst sample by pressing catalyst powder into a thin disk by applying 10 tof Table 1 Main catalyst properties. Property Zeolite Catalyst S BET (m 2 g 1 ) 434 290 S micro (m 2 g 1 ) 366 130 Total acidity (mmol g 1 ) 0.33 0.21 C WAS (mmol g 1 ) 0.023 0.012 C MAS (mmol g 1 ) 0.061 0.074 C SAS (mmol g 1 ) 0.26 0.13 C BAS (mmol g 1 ) 0.32 0.13 C LAS (mmol g 1 ) 0.060 0.027 José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 117 pressure in a Specac manual hydraulic press. The prepared catalyst sample was placed in the spectroscopic cell and subjected to thermal treatment at 550 °CinN 2 flow for 1 h to remove contaminants. Afterward, the catalyst was cooled to the reaction temperature, and a reference spectrum was collected (FTIR or UV–vis spectrum). The reaction test started by continuously feeding methanol while collecting differential FTIR or UV–vis spectra (by subtracting the corresponding reference spectrum). Alternatively, we also conducted experiments in the FTIR spectroscopic cell by discontinuously feeding methanol with alternated flushing periods (pulses of methanol feed). In this case, the methanol was input into the cell for a determined time, and then the feed was switched to an N 2 flow to sweep weakly adsorbed or gaseous species from the catalyst surface and collect a spectrum. We collected FTIR spectra every 2 min in the continuous mode, with a measurement range of 1300 to 4000 cm 1 , a resolution of 4 cm 1 , and 100 scans. We collected the UV–vis spectra every 35 s in the continuous mode, with a measurement range of 250 to 850 nm at a scan speed of 4000 nm min 1 , a UV–vis response of 0.24 s, and a data interval of 2 nm. Additionally, we analyzed the retained species after each experiment using extraction with dichloromethane (the procedure is described in previous publications [11,12,15]). Briefly, the procedure consists of (1) dissolving the spent catalyst disk with HF (Merck, 40%) in a Teflon container for 1 h, (2) neutralizing with NaOH, (3) extracting species with 3 cm 3 of dichloromethane (Sigma-Aldrich, 99.9%), and (4) allowing the organic and aqueous phases to separate. The organic phase was analyzed using a GC with an MS (Shimadzu, GCMS-QP2010S), and the components were identified and semi-quantified (peak integration). 3. Results 3.1. Kinetic performance The kinetic performance of the cells was assessed using a gaseous effluent analysis with an MS verified with a GC. For this, we calculated the conversion and yield of product groups using the MS data as described in the experiment section (Section 2). Figs. 2a-b illustrate the time on stream evolution of the conversion at variable space times for the MTH reaction in the spectroscopic cells. In general, the conversion levels increase with the increase in the space time for both reaction systems, which is expected for the MTH reaction [15,45–48]. Additionally, the experimental data reveal that the conversion progressively decreases with increasing times on stream, indicating catalyst deactivation. Particularly, experiments in the FTIR spectroscopic cell demonstrate complete catalyst deactivation at low space times (0.2 and 0.4 g h mol 1 ), whereas those in the UV–vis cell do not exhibit such levels of catalyst deactivation in 180 min on stream. Thus, using high space times improves the catalyst stability in agreement with Fig. 1. Schematic representation of the experimental setup for the reaction system using commercial spectroscopic cells. José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 118 providing slower velocities for the progressive evolution of active and deactivating species through the catalyst bed in the MTH reaction. The reason for the slower deactivation in the UV–vis cell is that this cell seems to exhibit a behavior in which the methanol concentration on the catalyst surface rises more slowly than in the FTIR cell. Thus, the behavior of the former system is closer to a mixed flow reactor, whereas that of the latter is closer to a plug flow reactor. Figs. 2c-d present the evolution with the conversion of the yield of the main product groups at variable space times for the MTH reaction in the spectroscopic cells. The identification of products is approximate, as explained in the experiment section (Section 2), because the MS can identify fragments of compounds and these fragments are common for several compounds that are typical MTH reaction products. However, the trends in Figs. 2c-d agree with those typically expected for the MTH reaction [15]. The yield of light olefins increases faster with increasing conversions (i.e., by changing space time) describing a convex curve, whereas the yield of heavy aliphatics and aromatics slowly increases with increasing conversions (by changing space time) describing a concave curve. This behavior describes the role of products in the reaction, pointing out that light olefins are reaction intermediates and heavy aliphatics and aromatics are final products, in agreement with the global mechanism of the MTH reaction [18]. Thus, light olefins are primarily formed at a high rate (low conversions). When there is a high concentration of light olefins in the reaction medium, they undergo oligomerization, methylation, cyclization, and hydrogen transfer reactions yielding aliphatics and aromatics at increasing rates with increasing conversions. These results evidence a reasonable kinetic behavior of the spectroscopic cells for the MTH reaction, bringing the opportunity to explore more aspects of this reaction using these reaction systems. Thus, the following sections focus on studying the in situ formation of species on the catalyst surface (also referred as surface species) using FTIR and UV–vis spectroscopies. The conditions used in these experiments result in the catalyst undergoing deactivation at 0.2 and 0.4 g h mol 1 in the FTIR spectroscopic cell, which is an interesting experimental scenario to study the species involved in catalyst deactivation. 3.2. In situ FTIR spectroscopy Fig. 3 depicts the time on stream evolution of the differential FTIR spectra (the FTIR spectrum of the fresh catalyst is subtracted) of the catalyst during the experiments at the space times of 0.2 and 1.6 g h mol 1 carried out in the FTIR spectroscopic cell with a continuous methanol feed. Figure S1 in the Supporting Information presents the FTIR spectra evolution for other space times. Exposing the catalyst to methanol results in several FTIR bands related to hydrocarbons in the 1300 to 3200 cm 1 region, which are typical ‘‘signatures” of the MTH reaction [12,15,49]. Although the composition of the catalyst is 50 wt% of zeolite (the rest corresponds to alumina phases), the spectroscopic signature is similar to that obtained for experiments using pure zeolite catalysts, which implies that the alumina phases do not affect the nature of the species formed on the agglomerated catalyst. Likewise, we tested a blank composed of 30 wt% of boehmite and 70 wt% of a -alumina (zeolite is replaced with a -alumina) in a fixed-bed reactor at the same reaction conditions used in this work, and we observed that only the methanol dehydration took place yielding dimethyl ether and water with no evidence of hydrocarbons formation. The bands in the 1300 to 1460 cm 1 region may be associated with the bending vibration of C-H bonds, those in the 1400 to 1700 cm 1 region with the stretching vibration of C = C bonds in the ring of aromatics Fig. 2. Time on stream evolution of the conversion in a (a) FTIR spectroscopic cell and (b) UV–vis spectroscopic cell, and evolution with conversion of the product yield in (c) an FTIR spectroscopic cell and (d) UV–vis spectroscopic cell. Reaction conditions: T= 400 °C, P M0 = 0.16 bar, W/F M0 = 0.2–1.6 g h mol 1 . José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 119 or in olefins, and those in the 2800 to 3100 cm 1 region with the stretching vibration of C-H bonds [28,50–54]. Furthermore, the 3200–4000 cm 1 region bands provide information on the O-H bonds in various species, such as adsorbed methanol and water (positive bands at about 3530 cm 1 ), and the zeolite acid sites exhibit negative bands in the 3550–3800 cm 1 region. We performed a multivariate analysis for the FTIR band assignation for the bands associated with the extracted species in a previous work [12]. Based on the stability and behavior of the bands, we classified the most relevant bands of the spectra. The bands at 2950 and 2844 cm 1 are assigned to –CH 3 and –CH 2 groups, respectively, in oxygenated or non-oxygenated aliphatic and aromatic species that may be on the catalyst surface or gas phase. However, as seen in Fig. 3, other several bands also appear in the 2800–3100 cm 1 region, evidencing the complexity of the reaction medium and making difficult to identify specific species (surface and gas phase species). The band at 1616 cm 1 is assigned to monocyclic aromatic and olefinic species, probably in a cationic form being the active hydrocarbon pool species [54,55]. The band at 1570 cm 1 is assigned to monocyclic and polycyclic aromatic species, and the band at 1481 cm 1 is assigned to linear polycyclic aromatic species. We verified the presence of monocyclic aromatics (xylenes, trimethylbenzenes (triMBs), tetramethylbenzenes (tetraMBs), pentamethylbenzenes (pentaMBs), and hexamethylbenzenes (hexaMBs)) by extracting soluble species in the spent catalyst samples (Fig. 4), which are initially related to the 1616 and 1570 cm 1 bands (Table S2). The abundance of soluble monocyclic aromatic species increases with increasing space times, and the distribution of components changes from being centered at pentaMB/hexaMB at 0.2 g h mol 1 to being centered at tetraMB at 1.6 g h mol 1 . These observations are based on the activity state of the catalyst, which is severely deactivated at low space times and is active at high space times (Fig. 2a). Thus, an active catalyst has a high concentration of monocyclic aromatic species with a composition centered at the most active species (e.g., tetraMB [17]). In contrast, a deactivated catalyst has a low concentration of monocyclic aromatic species with a composition centered at the less active species (e.g., hexaMB, as an inactive species in ZSM-5 catalysts [17]). Fig. 5 illustrates the time on stream evolution of the maximum intensity of the 2950, 1616, 1570, and 1481 cm 1 bands at variable space times. The intensity of the 2950 and 1570 cm 1 was directly taken from each spectrum, whereas that of the 1616 and 1481 cm 1 bands was corrected with a baseline as shown in Figures S2 and S3. The reason for this correction is that these bands appear on the tail of the 1570 and 1456 cm 1 bands, like a shoulder, and therefore their intensity is masked. Additionally, Figure S2 also evidences that the 1481 cm 1 band is prone to appear when the catalyst loses activity whose condition is favored at decreasing space times and prolonged times on stream, making evident this band is the signature for the formation of deactivating species. Thus, the spectro-kinetics indicate that the 2950 cm 1 band rapidly increases at the beginning of the reaction and reaches a maximum value that remains stable during the entire reaction. The maximum value depends on the space time, which is higher as the space time decreases. Because the FTIR intensity can be directly related to the species concentration on the sample through Lambert–Beer’s law, this observation also indicates that the concentration of species whose bonds vibrate at 2950 cm 1 is higher as the space time decreases. In fact, it is expected that methoxy species show vibration features at frequencies in the range of 2800–3000 cm 1 ,as reported by Saepurahman et al. [50] who studied the FTIR features of methanol adsorbed on zeolites. Likewise, hydrocarbons also show important vibration features at these frequencies and it is difficult to make a precise difference between methoxy and hydrocarbon species in the reaction, and it is even more complex when working with a continuous methanol feed because the gas phase species seem to show strong vibration features at these frequencies [12]. Albeit these issues, the 2950 cm 1 band may be also associated with reactive species (methoxides, alkoxides, olefins) because the species concentration is theoretically expected to decrease with increasing space times, leading to high conversions (verified by analyzing the gaseous effluent in Fig. 2a). The 1616 cm 1 band intensity also rises very rapidly at the beginning of the reaction (Fig. 3c), reaching a maximum evidenced in the experiments at low space times. This behavior indicates that these species are intermediates because their concentration reaches a maximum and progressively decreases with catalyst deactivation (Fig. 2a). These species are part of the hydrocarbon Fig. 3. Time on stream evolution of the differential FTIR spectra of the HZSM-5 catalyst in the MTH reaction with continuous methanol feed at space times of (a) 0.20 and (b) 1.6 g h mol 1 . Conditions: T= 400 °C, P M0 = 0.16 bar. Fig. 4. Distribution of the components of soluble species in the spent catalyst samples in the experiments in the FTIR spectroscopic cell after 180 min on stream. José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 120 pool in the MTH reaction, primarily comprising monocyclic aromatics probably in cationic forms [54,55]. This is further supported by the disappearance of soluble species at prolonged times on stream and low space times, coinciding with the behavior of this band. Previous works [12,15] demonstrated that monocyclic aromatic species are converted into coke (more condensed aromatic structures) as the catalyst undergoes deactivation. On this catalyst, the presence of absorbed olefins as intermediate species seems to be negligible compared to other small-pore catalysts [12,56]. However, examining the spectra, bands at 1590, 1390, and 1377 cm 1 are the first to rise at a short time on stream, indicating the formation of adsorbed olefins [57], but their intensity is rapidly attenuated, resulting in bands at 1616 and 1570 cm 1 . Thus, the majority of the hydrocarbon pool species would be aromatic species during the reaction. The 1570 cm 1 band intensity also evolves with the time on stream, reaching a maximum but maintaining a high value at the end of the experiments (Fig. 3b). The partial attenuation of this band intensity may indicate the contribution of intermediate monocyclic aromatic species (also showing vibration features at 1616 cm 1 ) that disappear upon catalyst deactivation, whereas the rest of the species may be associated with polycyclic aromatics. It should be mentioned that the intensity of the 1455 and 1510 cm 1 bands (and actually the intensity of the 1400– 1600 cm 1 region) also decreases at prolonged time on stream and, particularly, at low space time (when the catalyst loses activity), which can be clearly seen in Figures S2 and S3 in the Supporting Information. This confirms that active species has vibrational features in this whole range of the FTIR spectrum, overlapping with the vibrational features of other stable species (e.g. polycyclic aromatics). This is also evident when analyzing the expected vibrational features of different monocyclic aromatic species found (Table S2), exhibiting frequencies in the 1400–1700 cm 1 range in an unpredictable way. Thus, the 1616 cm 1 is more representative of these active species because it is isolated (there are no other overlapping bands) and its appearing and disappearing behavior is clearly seen. Based on Raman spectroscopy (complementary to FTIR spectroscopy), several carbon structures typically exhibit bands between 1300 and 1700 cm 1 [56,58–62]. For instance, an appearing and disappearing band at 1602–1606 cm 1 has been related to cationic species (e.g. benzenium species) [61], which can be analogous to the 1616 cm 1 band observed in this work. Likewise, a band at 1510 cm 1 has been related to cyclopentenyl cations in the FTIR spectrum [55]. These findings reported in the literature strengthens the band assignation in this work, and makes evident that the appearing and disappearing behavior (barely reported in the literature) of the 1616 cm 1 band and some bands in the 1400–1600 cm 1 region corresponds to vibrational features of active hydrocarbon pool species (cationic forms). The 1481 cm 1 band intensity rises when the catalyst undergoes severe deactivation (prolonged time on stream and low space times) and when the intermediate species decrease (Fig. 3a), which can be clearly seen in Figures S2 and S3 in the Supporting Information. Thus, the species with bonds vibrating at 1481 cm 1 are associated with deactivating species. According to Lee et al. [63], these deactivating species may be associated with linear monocyclic aromatics growing in the zeolite channels. Additionally, Rojo-Gama et al. [56] have assigned a Raman band at 1485 cm 1 to fluorene, supporting that linear polycyclic aromatics can show vibrational features at this frequencies. When analyzing the band evolutions over the space time (different activity states of the catalyst), we observed that all retained species evolve faster at low space times. This observation also Fig. 5. Time on stream evolution of the intensity of FTIR bands of (a) 1481, (b) 1570, (c) 1616, and (d) 2950 cm 1 in the MTH reaction with a continuous methanol feed at variable space times. Conditions: T= 400 °C, P M0 = 0.16 bar, W/F M0 = 0.20–1.6 g h mol 1 . José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 121 applies to forming aliphatic species (2950 cm 1 ) and heavier polycyclic aromatic species (1481 and 1570 cm 1 ). We also observe a more rapid conversion of active into deactivating species (1616 cm 1 ) with decreasing space times, which is also evidenced when analyzing the evolution of soluble species over space time: the most active monocyclic aromatic species for the MTH reaction disappear faster at low space times according to the catalyst activity state. In this way, the results imply that aliphatic and methoxy/alkoxy species (2950 cm 1 ) and the most active hydrocarbon pool species (olefins and monocyclic aromatic hydrocarbons, 1616 cm 1 ) are generated very quickly over the surface. The latter species group (active) degrades progressively to polycyclic aromatic species, as observed in the evolution of the 1616, 1570, and 1481 cm 1 bands. As the reaction evolves with time on stream, the intermediate monocyclic aromatic species in the micropores polymerize into heavy polycyclic aromatic species (1481 cm 1 ), causing catalyst deactivation by blocking access to acid sites on the microporous surface. 3.3. In situ UV–vis spectroscopy Figure S4 in the Supporting Information presents the time on stream evolution of the differential UV–vis spectra (the UV–vis spectrum of the fresh catalyst is subtracted) for experiments with different space times (0.2, 0.4, 0.8, and 1.6 g h mol 1 ) conducted in the UV–vis spectroscopic cell with a continuous methanol feed. As observed, several bands occur in the UV (250 to 330 nm) and visible (330 to 850 nm) regions. The first bands appear around 380 to 400 nm, and then other bands appear at about 500, 604, and 704 nm. These bands have been extensively discussed in the literature [11,15,33,34,36,37] and correspond to cations of monocyclic aromatics, monoenes, or polyenes (380 to 400 nm) and cationic or neutral polycyclic aromatics (500 to 800 nm). Figure S5 presents the time on stream evolution of the intensity of selected UV–vis bands. The spectro-kinetic data reveal that the evolution of these bands provides limited kinetic information because the intensity rapidly reaches a stable value through time on stream, which occurs because this spectroscopic technique is limited by the color of the catalyst sample. The fast darkening of the catalyst sample provokes a rapid saturation of the spectra as evidenced by the broad background absorption across the whole visible region. This similar observation was reported by Mores et al. [34] for the methanol conversion on a ZSM-5 catalyst at high temperatures (above 400 °C). To address this problem, we varied other reaction conditions, such as the methanol partial pressure and space time to obtain more incipient reaction conditions. Decreasing these variables slows the reaction rates of all kinetic steps occurring in the MTH reaction without significantly changing the reaction mechanisms, so that a slower formation of retained species and, therefore, an attenuation on the spectra saturation is expected. Fig. 6 compares the kinetics of the MTH reaction at the two methanol partial pressures in the UV–vis spectroscopic cell. The evolution with space time of the conversion of oxygenates (Fig. 6a) is slower as the methanol partial pressure decreases, requiring higher space times to reach high conversions. Likewise, the time on stream evolution of the conversion (Fig. 6b) is highly stable at a low methanol partial pressure, providing insignificant evidence of catalyst deactivation compared with experiments at a high methanol partial pressure (Fig. 2b). The product distribution at a low methanol partial pressure (Fig. 6c) is comparable with that at a high methanol partial pressure (Fig. 2d), indicating that this variable has a poor effect on changing the reaction mechanisms but has a strong influence on the reaction rate. Fig. 7 depicts the corresponding UV–vis spectra for experiments with a low methanol partial pressure and space times of 0.8 and 6.4 g h mol 1 . Additionally, Figure S6 in the Supporting Information presents the spectral evolution for other space times. The spectral evolution indicates that these reaction conditions significantly attenuate the spectral saturation due to slowing the darkening of the catalyst sample, and this attenuation is even more favorable at high space times. The UV–vis bands are the same as those observed for the experiments at a high methanol partial pressure; therefore, the band assignation is analogous. We verified the presence of monocyclic aromatic species (xylenes, triMB, tetraMB, pentaMB, and hexaMB) by extracting soluble species in the spent catalyst samples (Fig. 8), which are observed at 400 nm in the spectroscopic signature. The concentration of these species increases with decreasing space times, whereas the distribution changes from being centered at hexaMB at low space times to Fig. 6. Kinetic performance of the UV–vis spectroscopic cell at variable methanol partial pressure: (a) maximum conversion curves at a methanol partial pressure of 0.050 and 0.16 bar, (b) time on stream evolution of the conversion at 0.050 bar, (c) evolution with conversion of the product yield at 0.050 bar. Conditions: T= 400 °C, P M0 = 0.050 to 0.16 bar, W/F M0 = 0.2 to 1.6 g h mol 1 . José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 122 being centered at pentaMB at high space times. In these experiments, the catalyst was active at all tested space times (as shown in Fig. 6b with no evidence of deactivation). Fig. 9 reveals the time on stream evolution of the intensity of the selected UV–vis bands for these reaction conditions. The spectro-kinetic data indicate that these reaction conditions significantly slow the intensity evolution of the bands, controlling the rapid spectral saturation. The data reveal that the monocyclic aromatic species (around 400 nm) are the first to grow, followed by other polycyclic aromatic species observed at high wavelengths (604 and 704 nm). In all experiments conducted in the UV–vis spectroscopic cell, we only observed the formation dynamics of species without evidence of degradation indicated by the disappearance of bands compared to the observations in the experiments in the FTIR spectroscopic cell. We infer that we cannot observe the degradation of active species in these experiments because the catalyst did not undergo deactivation (Figs. 2b and 6b). We reported similar observations when comparing different catalysts using a similar experimental setup [11,15]. However, Borodina et al. [39,40] and Goetze et al. [64] reported the disappearance of active species detected using in situ UV–vis spectroscopy in small-pore catalysts during deactivation. The difference between conducting experiments with medium-pore catalysts, such as in this work, relies on forming large polycyclic aromatic species at high temperatures that lead to a rapid darkening of the catalyst sample and the saturation of the spectra. Although this effect can be controlled by operating at less severe reaction conditions, the deactivation period is even more delayed. Thus, the observations in this work using a medium-pore catalyst are limited to the formation of retained species due to reaction conditions that do not allow the observation of severe catalyst deactivation. When analyzing the band evolution over space time, we observe that all retained species evolve faster at decreasing space times (similar to the observations from the FTIR spectra data). Another peculiar kinetic observation is the sigmoidal profile for the UV–vis band evolutions (Fig. 9), which may be associated with the typical kinetic periods of induction and autocatalysis for the MTH reaction, as reported in other work using in situ spectroscopic approaches [14,35]. First, the slow increase of the band intensity indicates the induction period, which is longer as the space time increases. Second, the autocatalytic period starts when the band intensity increases rapidly, reaching a stable value, which seems to be slower as space time increases. The observed data are from the induction and autocatalytic period during a transient state governed by the increase in the reactive species concentration on the catalyst surface from zero to a maximum value. Thus, the use of low space times increases the contact velocity (inverse of space time), resulting in a rapid increase in the reactive species concentration and the rapid formation of retained species on the catalyst surface. In contrast, using high space times decreases the contact velocity, causing a slower increase in the reactive species concentration and the formation of retained species on the catalyst surface. The kinetic analysis of the retained species on the catalyst surface using FTIR and UV–vis spectroscopies is useful to obtain mechanistic information when these species participate in the reaction network, such as in the case of the MTH reaction. Using the kinetic information from both techniques, we sketched a reaction mechanism based on the identified retained species (Fig. 10). Thus, the formation of reactive species from oxygenates (band at 2950 cm 1 , albeit this band is not unambiguously assigned because it may also include the contribution of –CH 3 groups in other aliphatic and aromatic species) on the catalyst surface forms active species of the hydrocarbon pool (1616 cm 1 or 400 nm) retained in the catalyst micropores. These species form small olefins, paraffins and aromatics (m/z= 43, 55, 78...) that diffuse into the gas phase as products and form large polycyclic aromatic species (1570 cm 1 or 500 to 700 nm) deposited on the catalyst mesoporous surface with a poor deactivating effect. Eventually, the active species polymerize into linear polycyclic aromatics (1481 cm 1 ) in the catalyst microporous surface, blocking the micropores and severely depleting the catalytic activity. As discussed, some of these mechanisms have been studied in detail in other works. We intend to construct an overall kinetic reaction network and deactivation from the experiments conducted with the spectro-kinetic approach. Fig. 7. Time on stream evolution of the differential UV–vis spectra of the HZSM-5 catalyst in the MTH reaction with continuous methanol feed at space times of (a) 0.80 and (b) 6.4 g h mol 1 . Conditions: T= 400 °C, P M0 = 0.16 bar. Fig. 8. Distribution of the components of soluble species found in the spent catalyst samples in the experiments in the UV–vis spectroscopic cell at 0.05 bar after 120 min on stream. José Valecillos, H. Vicente, A.G. Gayubo et al. Journal of Catalysis 408 (2022) 115–127 123