Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Elucidation of radical- and oxygenate-driven paths in zeolite-catalysed conversion of methanol and methyl chloride to hydrocarbons

Elucidation of radical- and oxygenate-driven paths in zeolite-catalysed conversion of methanol... Articles https://doi.org/10.1038/s41929-022-00808-0 Elucidation of radical- and oxygenate-driven paths in zeolite-catalysed conversion of methanol and methyl chloride to hydrocarbons 1,4 1,4 1,4 2 2  ✉ Alessia Cesarini , Sharon Mitchell , Guido Zichittella    , Mikhail Agrachev , Stefan P. Schmid    , 2 3 3 3 1  ✉  ✉ Gunnar Jeschke    , Zeyou Pan , Andras Bodi    , Patrick Hemberger    and Javier Pérez-Ramírez    Understanding hydrocarbon generation in the zeolite-catalysed conversions of methanol and methyl chloride requires advanced spectroscopic approaches to distinguish the complex mechanisms governing C–C bond formation, chain growth and the deposition of carbonaceous species. Here operando photoelectron photoion coincidence (PEPICO) spectroscopy enables the isomer-selective identification of pathways to hydrocarbons of up to C in size, providing direct experimental evidence of methyl radicals in both reactions and ketene in the methanol-to-hydrocarbons reaction. Both routes converge to C mol- ecules that transform into aromatics. Operando PEPICO highlights distinctions in the prevalence of coke precursors, which is supported by electron paramagnetic resonance measurements, providing evidence of differences in the representative molecular structure, density and distribution of accumulated carbonaceous species. Radical-driven pathways in the methyl chloride-to-hydrocarbons reaction(s) accelerate the formation of extended aromatic systems, leading to fast deactivation. By contrast, the generation of alkylated species through oxygenate-driven pathways in the methanol-to-hydrocarbons reaction extends the catalyst lifetime. The findings demonstrate the potential of the presented methods to provide valuable mechanistic insights into complex reaction networks. onosubstituted methanes, including methanol and Comparatively, mechanistic understanding of the methyl methyl chloride (CH X, X = OH and Cl, respectively), chloride-to-hydrocarbons (MCTH) reaction is less developed. Mare attractive building blocks for sustainable fuels and Previous studies have focused on kinetic aspects, showing broadly 1–7 chemicals . These C molecules are the main products of syngas similar product distributions to the MTH process, attributed to conversion, carbon dioxide hydrogenation and halogen-mediated similar confinement effects within the zeolite pore networks on 11,12,32–34 alkane functionalization, and can be readily converted into valuable HCP development . Nevertheless, the absence of oxygenates 8–13 hydrocarbons over zeolite catalysts . However, uncontrolled chain and the faster catalyst deactivation reported in the MCTH process growth typically results in broad product distributions and the suggest that important mechanistic distinctions exist. A system- deposition of heavy carbonaceous species, leading to suboptimal atic comparison of the pathways for the activation and subsequent catalyst selectivity and stability. Towards improving productivity, transformation of CH Cl and CH OH could help decouple direct 3 3 extensive experimental and theoretical efforts have been devoted C–C bond formation routes from oxygenate-driven chain growth. to understanding the mechanism of C–C bond formation and its However, the direct observation of short-lived active intermediates, propagation to higher hydrocarbons in the commercially applied such as ketene and methyl radicals, under operando conditions 1,14–18 methanol-to-hydrocarbons (MTH) reaction . Detailed kinetic remains highly challenging because of the difficulty in discriminat- studies, often combined with isotopic labelling and spectroscopic ing them from stable spectator species due to the limited sensitiv- experiments, evidenced the autocatalytic nature of the process, ity of established methods. Advanced time-resolved spectroscopic provided the basis for the widely accepted dual aromatic–olefin techniques that can distinguish compounds present in low concen- cycles and identified several intermediates expected to play crucial trations and distinct isomeric forms could provide new insights to roles in the early stages and steady-state operation of the reaction help resolve some of the long-standing debates. 19–31 (Supplementary Note 1) . Despite these advances, several ques- Herein, we analyse hydrocarbons desorbed from a representative tions about the mechanism remain strongly debated, including the H-ZSM-5 zeolite catalyst during MTH and MCTH processes using existence and role of radical species, the mechanism of the first operando PEPICO spectroscopy under relevant reaction conditions. 35–37 C–C bond formation and subsequent paths to the early aromatic This technique has shed light on complex reaction networks hydrocarbon pool (HCP) species. Furthermore, archetypical dis- and enables the quantitative, isomer-selective identification of all solution–extraction experiments and spectroscopic measurements short- and long-lived species up to approximately C in size. By have provided limited insight into the specific structures of the coke performing temperature-dependent studies, operando PEPICO precursors and their evolution into carbonaceous species deposited enables the discrimination of primary and secondary intermediates within the zeolite catalyst (Supplementary Note 1). and tertiary products. A complementary study of the used catalysts 1 2 Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland. Laboratory of Physical Chemistry, Department of Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland. Laboratory of Synchrotron Radiation and Femtochemistry, Paul Scherrer Institute, Villigen, Switzerland. These authors contributed equally: Alessia Cesarini, Sharon Mitchell, Guido Zichittella. e-mail: [email protected]; [email protected]; [email protected] NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 605 ms-TPE signal (a.u.) NaTURE CaTalysIs Articles Isomer identification Detection of intermediates Operando PEPICO Ion CH Ion 8 9 10 14 15 16 hν (eV) m/z To GC-FID CH X analysis H-ZSM-5 Kinetics EPR Product analysis CH Coke Coke amount nature Olefins Alkanes BTX 0 MTH MCTH 200 300 400 0 10 20 30 Other B (mT) ω (MHz) Fig. 1 | Approach to mapping the catalysed generation and evolution of hydrocarbons. Schematic of the multi-technique strategy used in this study to unravel the growth of the carbon chain, from the formation of the first C–C bond to the generation of coke in MTH and MCTH over H-ZSM-5. The comparative study of these two C platform molecules enables the main pathways of oxygenate- and hydrocarbon-driven mechanisms to be decoupled and elucidated. Accordingly, operando PEPICO enables the isomer-selective identification of reaction intermediates and coke precursors. This is complemented with EPR measurements that provide insights into the representative molecular structure, density and distribution of deposited carbonaceous species, and kinetic analysis for assessment of the catalytic activity, selectivity and stability. GC-FID, gas chromatography with flame ionization detection. 38,39 via electron paramagnetic resonance (EPR) revealed distinctions attributed to the rapid rupture of Si–O–Al bonds due to interac- in the representative molecular structure, density and distribution tion with HCl. The similar conversions shown in our study suggest 11,33 of the carbonaceous species deposited in the zeolite pore network. that the MFI-type framework of H-ZSM-5 offers higher stability In combination with kinetic analysis, the adopted approach (Fig. 1) and that the leaving group does not severely influence the intrin- provides high molecular resolution of the species formed, enabling sic kinetics of MTH and MCTH. The products are reported in five experimental verification of several open questions about the reac- main groups: CH , C –C olefins, C –C alkanes, benzene–toluene– 4 2 4 2 4 tion mechanisms. xylene (BTX) and other species, including C hydrocarbons. At 5+ 673 K, while the generation of CH , olefins and alkanes was simi- Results lar, the formation of BTX was found more favoured in the MCTH Evaluation of the reaction kinetics. Preliminary insights into the process (52% selectivity) compared with MTH (38% selectivity). An similarities and differences between MTH and MCTH processes opposite trend was observed for C hydrocarbons, with selectivities 5+ were gained by studying the steady-state kinetics of the reactions of 4% and 12% for MCTH and MTH, respectively. Contrariwise, a over an H-ZSM-5 zeolite catalyst with a nominal Si/Al ratio of 40 rise in the reaction temperature (T) led to an increase in the for- under equivalent conditions (Fig. 2; Supplementary Figs. 1 and 2). mation of olefins at the expense of BTX in the MCTH process and The reactivity was assessed by calculating the conversion (X(CH X)) especially in the MTH reaction (Supplementary Fig. 1a), which is 12,32 based on the concentration of CH X at the reactor inlet and outlet. consistent with other studies . A stronger rise in CH genera- 3 4 According to common practice, the reactivity in the MTH process tion was observed in MCTH compared with the MTH process, was also evaluated considering both CH OH and dimethyl ether which suggests that a lower rate of trans-hydrogenation reactions ((CH ) O) as reactants since an equilibrium is typically established in MCTH than in MTH is favoured by the presence of (CH ) O 3 2 3 2 1 40 32 between the two . Since no (CH ) O was observed at 100% conver- (ref. ). In line with the literature , (CH ) O, formed by the revers- 3 2 3 2 sion, recalculation of the conversion and product distribution does ible dehydration of CH OH, was not observed in the MTH reac- not alter the conclusions below (Fig. 2; Supplementary Figs. 1 and tion at 100% conversion. To extract additional kinetic information, 2). Both reactions showed a strong temperature dependence, and the product distribution was also compared at approximately 65% their light-off curves were found to be comparable, particularly CH X conversion, which was achieved by adjusting the space veloc- when considering (CH ) O as a reactant (Fig. 2a; Supplementary ity (Supplementary Fig. 1b). Under these conditions, a considerable 3 2 Fig. 2a). This points to similar kinetic behaviour of the MTH and amount of (CH ) O was observed in the MTH process (27% selec- 3 2 MCTH reactions over H-ZSM-5, which contrasts with previous tivity). In addition, the formation of olefins and C hydrocarbons 5+ findings obtained over H-SAPO-34 zeolite catalysts . In this study was favoured at 65% conversion compared with full CH X conver- by Olsbye et al., lower conversions for the MCTH reaction were sion at 673 K (olefin selectivity: 45% versus 26% for MTH, 40% NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Intensity (a.u.) Intensity (a.u.) ω (MHz) 2 NaTURE CaTalysIs Articles stable during the first 60 h, before BTX production decreased, ulti- mately reaching negligible production at the end of the run, while that of olefins increased. The used catalysts were characterized 80 after selected times on stream in the MTH and MCTH processes to quantify the amount and location of carbonaceous coke species deposited and to confirm the effects of the reaction on the crystal- linity and acidic properties of the H-ZSM-5 zeolite. Consistent with previous reports that have shown coking as a deactivation pathway 34,41,42 in both reactions , thermogravimetric analysis (TGA) provided evidence of the accumulation of substantial carbonaceous deposits 0 (up to 15 wt%) (Fig. 2c). Notably, coke accumulation was faster in MTH 573 673 773 the early stages of the MCTH process compared with MTH, which MCTH T (K) is in line with the higher tendency of this C platform molecule to generate BTX. However, the rate of coke deposition subsequently 0 20 40 60 80 100 120 slowed and, after 120 h on stream, the coke content was slightly Time on stream (h) lower than in the catalyst used in the MTH reaction. This reduction in coking rate is consistent with the strong decrease in the produc- b MTH tion of BTX and C hydrocarbons observed during the long-term 5+ 37 34 32 24 28 test for the MCTH reaction (Fig. 2a,c), which have been identified 41,42 26 as important classes of coke precursors . A linear inverse correla- 15 21 15 10 13 16 tion between the micropore volume and coke content was found 3 2 3 (Fig. 2d). This indicates that carbonaceous species evolve and accu- MCTH mulate similarly within the pore network, despite the different rate 24 25 34 76 of deposition, which is in line with the results from EPR (vide infra). 44 44 44 19 19 18 X-ray diffraction patterns of the used zeolite catalysts after the oxi- 12 10 dative removal of coke species showed no alteration of the crystal 1 2 2 2 5 5 structure after either reaction (Supplementary Fig. 3). By contrast, 10 h 25 h 60 h 120 h the infrared spectroscopy study of adsorbed pyridine highlighted substantial changes in the acidic properties after the MCTH reac- CH Olefins Alkanes BTX Other tion, indicating decreased concentrations of Brønsted acid sites c and increased concentrations of Lewis acid sites (Supplementary d 0.2 Table 1). These observations agree with previous reports on the MTH MTH impact of treatment with gaseous HCl-containing streams, a MCTH MCTH by-product formed in equimolar amounts in the MCTH process, 0.1 33,34 on the zeolite properties . Paths for hydrocarbon formation. Operando PEPICO analysis 0 0 enables us to obtain detailed insight into the intermediates and 0 60 120 0 10 20 products formed in the reactions over H-ZSM-5 at relevant con- Time on stream (h) Coke (wt%) ditions (T ≤ 773 K, P ≤ 0.5 bar), showing a wide range of hydrocar- bons (C –C ) in the MTH and MCTH processes and of oxygenates 1 14 Fig. 2 | Comparative performance of the zeolite catalyst. a, Reactant (C –C ) exclusively in the MTH reaction (Figs. 3–5; Supplementary 1 4 conversion in the MTH and MCTH processes as a function of the time on Figs. 4–23). As detailed in the Supplementary Methods, the reac- stream and temperature (inset) over H-ZSM-5. b, Product distribution as tants and intermediates or products desorbed from the catalyst are a function of time on stream averaged over a period of 1 h. c, Amount of detected via photoionization with monochromatic vacuum ultravi- coke species deposited measured via TGA as a function of time on stream. olet light. The photoions and photoelectrons generated are detected d, Variation in the micropore volume (V ) as a function of the amount micro in coincidence, revealing their mass-to-charge ratio (m/z) and their 3 −1 of coke. Conditions: CH X:He = 1:1, F  = 20 cm  STP min , W  = 0.6 g, 3 T cat isomeric identity based on the photoion mass-selected threshold T = 673 K, P = 1 bar. photoelectron (ms-TPE) spectrum. Two types of experiment were conducted for both reactions . The first type used a packed bed of zeolite catalyst that permitted operation under near-ambient versus 24% for MCTH; other selectivity: 14% versus 12% for MTH, conditions (P = 0.5 bar) (Figs. 3 and 4; Supplementary Figs. 4–7, 11% versus 4% for MCTH) at the expense of BTX (5% versus 38% 14 and 17), whereas the second experiment used a zeolite-coated for MTH, 35% versus 52% for MCTH). These common trends agree microreactor for low-pressure measurements (P = 0.05 bar) with the expected comparable mechanism of chain growth in the (Supplementary Figs. 8–11 and 15–17). Conversion estimates were MTH and MCTH processes, in which olefins and C hydrocarbons obtained by normalizing the feed signals to a xenon (Xe) inter- 5+ act as precursors for the formation of aromatic compounds. nal standard and comparing them with those obtained in blank Comparison of the time-dependent reactivity showed that full experiments at the same temperature (90 and 65% for MTH or CH X conversion could be maintained for 120 h in the MTH process 50 and 35% for MCTH at near-ambient or low pressure, respec- (Fig. 2a,b). The product distribution remained virtually unaffected tively). They confirmed the similar conditions in the operando for the first 60 h on stream, although olefins increased from approxi- PEPICO experiments to the kinetic tests, closing the gap between mately 26% to 40% after 120 h in the MTH reaction at the expense of standard testing and operando measurements. Consistently, no BTX (32% versus 28%) and C hydrocarbons (30% versus 16%). By significant deactivation was observed during the measurements. 5+ contrast, CH Cl conversion gradually decreased to approximately Blank measurements conducted using an empty reactor con- 80% over the first 60 h before dropping more sharply to less than firmed the absence of products in both reactions (Supplementary 20% after 120 h on stream. The product distribution also remained Figs. 12 and 13). NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Coke (wt%) X(CH X) (%) X(CH X) (%) 3 –1 V (cm g ) micro Intensity (a.u.) Area (counts) NaTURE CaTalysIs Articles MTH MCTH 400 2 CH CHO CH • CH CH O (CH ) O LP LP NA 3 3 3 2 3 2 Blank CH CH CHO 15 15 m/z m/z 9.6 9.8 H-ZSM-5 1 CH O hν (eV) (CH ) O 3 2 0 0 30 44 46 14 15 C H C H C H C H 4 6 C H C H 4 8 4 6 4 8 4 6 4 8 350 40 0 0 57 54 55 C H C H C H C H C H C H 5 6 5 8 5 6 5 8 5 6 5 8 45 3 0 0 66 67 67 C H 5 10 C H 5 10 C H 5 10 230 32 0 0 71 69 70 C H 6 8 C H 7 10 C H C H C H C H 6 8 7 10 6 8 7 10 100 5 0 0 79 80 94 95 79 80 94 95 600 700 600 700 800 m/z m/z T (K) Fig. 3 | Evolution of reaction intermediates. Peak areas of the main reaction intermediates as a function of temperature in the MTH and MCTH reactions over H-ZSM-5 determined using operando PEPICO. The blank reactor experiments are shown as reference (open symbols). Representative mass spectra of the products detected for MTH and MCTH processes over the zeolite at 698 K. The insets show the mass spectra of the detected CH radicals during low-pressure (LP) or near-ambient (NA) experiments at 773 K, as detailed in Supplementary Figs. 9 and 11, confirming their presence under relevant conditions. The photon energies at which the signals of each chemical species were recorded in all PEPICO experiments are shown in Supplementary 3 −1 Figs. 5 and 7. Conditions: CH X:Xe:Ar = 1.0:0.1:20.9, F  = 22 cm  STP min , W  = 0.05 g, P = 0.5 bar (insets, P = 0.05 bar). 3 T cat The near-ambient-pressure measurements confirmed the temperature window in the near-ambient pressure experiments, expected formation of major product groups, such as ethylene the C H , C H and C H signal intensities reached a maximum 3 6 4 8 5 10 (C H ), propylene (C H ), BTX and other C hydrocarbons, includ- at approximately 673 K, and the evolution of C H , C H , C H and 2 4 3 6 4+ 5 8 5 6 6 8 ing C H (m/z 56), C H (m/z 66), C H (m/z 68), C H (m/z 70), C H peaked at around 723 K in both the MTH and MCTH reac- 4 8 5 6 5 8 5 10 7 10 C H (m/z 80) and C H (m/z 94) (Fig. 3; Supplementary Figs. 4–11). tions (Fig. 3; Supplementary Fig. 14). These observations support 6 8 7 10 Extraction of the ms-TPE spectra showed that each of these com- the presence of a consecutive mechanism in both reactions, where pounds was present as a complex mixture of isomers, with varying C compounds are central for chain growth, in line with previous degrees of branching and cyclization (Fig. 4; Supplementary Figs. 19 hypotheses . Conversely, the (CH ) O signal decreased sharply with 3 2 and 20). Oxygenates, including dimethyl ether ((CH ) O; m/z 46), increasing temperature in the MTH process, while that of CH CHO 3 2 3 formaldehyde (CH O; m/z 30) and acetaldehyde (CH CHO; m/z 44) followed a volcano behaviour (Fig. 3). These results confirm the 2 3 were detected in the MTH process (Figs. 3 and 4; Supplementary generation of (CH ) O in the early stages of the MTH process via 3 2 Figs. 5, 9 and 17). The most notable distinction in low-pressure CH OH dehydration, whereas CH CHO is probably involved con- 3 3 experiments was the detection of a readily identifiable signal asso- secutively in the formation of the first C–C bond. By contrast, the ciated with methyl radicals (CH ; m/z 15) and of their character- formation of CH O rose continuously with temperature. The same 3 2 istic vibrational fingerprints in the corresponding ms-TPE spectra trend was observed in the low-pressure experiments in the MTH (Fig. 3; Supplementary Figs. 9, 11 and 16). The reduced pressure reaction, although the amount of CH O generated was considerably suppresses the quenching of this highly reactive species, enabling higher (Fig. 3; Supplementary Fig. 17). This suggests that CH O is its observation. This experimentally confirms the presence of CH predominantly involved in the first steps of the hydrocarbon evolu- species in both MTH and MCTH reactions. The detection of CH tion and not in the generation of coke precursors, in agreement with 19,20 radicals also at near-ambient pressure demonstrates that these spe- recent studies . cies are reactive intermediates under realistic conditions rather than On the basis of these findings, two main reaction pathways just minor desorption by-products from surface methoxy species or appear to be responsible for the first C–C bond formation and mere spectators (Fig. 3; Supplementary Fig. 18). the subsequent chain growth to C compounds. The first involves Signal integration at suitable photon energies enabled quantifi- the evolution of oxygenates, such as (CH ) O and CH CHO, and 3 2 3 cation of the relative abundance of distinct species as a function of is only relevant in the MTH process (Fig. 4a), whereas the second temperature (Fig. 3; Supplementary Figs. 14–17). The increased gen- encompasses olefins and occurs in both MTH and MCTH processes eration of (CH ) O in the MTH reaction and the higher olefin/BTX (Fig. 4b). In the oxygenate-driven mechanism, CH OH probably 3 2 3 ratio in both reactions with decreasing conversion agrees well with initially undergoes dehydration to generate (CH ) O. Calculations 3 2 the kinetic experiments. While BTX increased in the investigated have shown that both CH OH and (CH ) O can react to CH O, 3 3 2 2 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Intensity (a.u.) Area counts NaTURE CaTalysIs Articles + + + + + + (CH ) O CH O C H O CH CHO C H O C H 3 2 2 2 2 3 4 6 5 8 9.6 10.0 10.4 10.0 10.5 11.0 9.4 9.5 9.6 9.7 10.1 10.3 10.5 9.6 10.0 10.4 8.4 8.8 9.2 9.6 hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) CH OH CH O CH O + CO H C O + CH OH 2CH CHO C H O + CH 3 2 2 2 2 3 3 4 6 3 . . . . – H O – H O – H O – H O – 2H – 2H + 2H + H 2 2 2 (CH ) O CH O CO H C O CH CHO + CH O C H O C H 3 2 2 2 2 3 2 4 6 5 8 – CH + + + + C H b CH C H C H C H 5 10 3 2 4 3 6 4 8 9.6 9.8 10.0 10.4 10.5 10.6 9.6 10.0 10.4 8.8 9.2 9.6 8.4 8.8 9.2 9.6 hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) . . . . CH X 2CH C H + CH C H + CH C H +CH 3 3 2 4 3 3 6 3 4 8 3 . . . . – 2H – H – H –H C H C H C H C H CH + X* 2 4 3 6 4 8 5 10 + CH + CH – H – H – 2H + CH – 2H – H – 2H – 2H + CH – H + + + + + + C H C H C H C H C H C H 5 8 5 6 6 8 6 6 7 10 7 8 8 9 10 8 9 10 8 9 10 8 9 10 8 9 10 8 9 10 hν (eV) Fig. 4 | Reaction pathways of catalytic C coupling. a,b, Proposed reaction network of the oxygenate-driven reaction in the MTH process (a) and of the direct CH radical-addition pathway in the MTH and MCTH reactions (b) to yield C intermediates in the micropores of H-ZSM-5. The insets on top of the 3 5 molecular schemes show the ms-TPE (open squares) and reference spectra (solid lines) of the identified species for MTH in a and MCTH in b. The main identified isomers are represented by the molecular models in a,b, while other structures are shown in Supplementary Figs. 18–22. Colour code: C (dark grey), O (red), H (light grey), X (X = OH or Cl; violet). c, Reaction pathway for the chain-propagation reaction of C H to generate benzene and toluene 5 8 in the MTH process. The ms-TPE (open squares) and reference spectra (solid lines) of the identified isomeric products are also shown. Conditions as reported in Fig. 3. 17,43,44 releasing H and CH , respectively . CH O can be oxidized to via Koch carbonylation to yield the ketene ethenone (C H O; m/z 2 4 2 2 2 carbon monoxide (CO) as demonstrated experimentally and theo- 42) , which is assigned by its ionization transition at 9.6 eV in the 15,17 retically . Consistent with previous literature reports, carbon m/z 42 ms-TPE spectrum (Fig. 4a). Although this intermediate monoxide can react with adsorbed CH OH, (CH ) O or CH O had been previously predicted , our data provide the experimental 3 3 2 2 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal ms-TPE signal (a.u.) NaTURE CaTalysIs Articles a b + CH C H MTH 3 10 8 – 4H C H MCTH 10 10 – 3H – 2H C H 10 12 C H + 13 14 C H 11 10 C H – H + CH 11 12 + 2 C H 12 12 C H 9 8 8.1 eV – H + CH C H 10 12 – H + CH 8.0 8.4 8.8 9.2 120 130 140 150 160 170 180 8.0 8.4 8.8 9.2 9.6 hν (eV) hν (eV) m/z c d 8 1.0 3.0 MTH 3D High 0.8 2.8 6 MCTH density Data 0.6 2.6 Fit Fit 0.4 2.4 Data 337 341 Low 0.2 2.2 0 3,000 density B (mT) 2D t (ns) 0 0 2.0 0 40 80 120 0 40 80 120 0 40 80 120 Time on stream (h) Time on stream (h) Time on stream (h) f g 1.1 MTH MCTH High alkylation 30 30 1.0 MTH MCTH 0.9 0.8 0.7 10 10 0.6 Low alkylation 0 0 0.5 0 10 20 30 0 10 20 30 0 40 80 120 ω (MHz) ω (MHz) Time on stream (h) 1 1 Fig. 5 | Generation and evolution of condensed carbonaceous species. a, Mass spectra of the C species detected at the reactor outlet in the MTH 10+ and MCTH reactions over H-ZSM-5. b, Reaction pathways for the formation of naphthalene and methylated naphthalenes in MTH. The insets show the ms-TPE (open squares) and reference spectra (solid lines) of the identified products. c–e, EPR-active concentration (c), Lorentzian fraction (d) and fractal dimension (e) of carbonaceous deposits in the used catalysts as a function of time on stream in MTH and MCTH as determined using CW EPR and pulsed EPR, respectively. The insets in d,e, illustrate how these parameters were extracted. f,g, Weak interaction quadrant of the 2D Hy SCORE spectra of the zeolite after 2 h on stream for MTH (f) and MCTH (g). The insets depict the representative molecular structures of the carbonaceous species. h, H:C molar ratio extracted from the 2D Hy SCORE spectra as a function of time on stream in MTH and MCTH over the zeolite. Conditions: a,b, Conditions as 3 −1 reported in Fig. 3; c–h, CH X:He = 1:1, F  = 20 cm  STP min , W  = 0.6 g, T = 673 K, P = 1 bar. 3 T cat NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal ω (MHz) –3 Intensity (a.u.) Coke spins (μmol cm ) ω (MHz) Lorentzian fraction ms-TPE signal (a.u.) H:C Fractal dimension ms-TPE signal (a.u.) NaTURE CaTalysIs Articles evidence of its formation and role in forming the first C–C bond the autocatalytic HCP pathway . To gain further insight into this in the MTH reaction. Its detection is highly challenging due to its route, the following experiment was performed: after running the −1 high reactivity and the low activation barrier (≤17 kJ mol ) for its reaction under identical conditions to those reported in Fig. 3, the coordination with Brønsted acid sites, generating readily observ- flow of reactants was stopped, the reactor flushed with argon (Ar) 1,19,23,45 able surface-bound acetate or methyl acetate species . In addi- and the temperature increased while analysing the species evolved. tion, C H O has the same integer mass as C H , which convolutes C H and C H were observed throughout the tests in both the 2 2 3 6 2 4 6 6 its identification solely via m/z analysis. The detection of gas-phase MTH and MCTH processes, while C H was found to be almost 3 6 ketene strongly exemplifies the ability of operando PEPICO to dis- negligible (Supplementary Fig. 23). This is in line with the occur- criminate reactive intermediates, either formed or released in the rence of a second pathway to C H and C H that involves the crack- 2 4 6 6 gas phase, from strongly bound species detected using conventional ing of alkylated benzenes . spectroscopic approaches, which are more likely to be spectators in the reaction mechanism. Reproportionation of C H O with Coke generation and evolution. Further growth of the carbon 2 2 CH OH could yield CH CHO and further CH O, whose ioniza- skeleton leads to the formation of heavy carbonaceous species 3 3 2 tion transitions were detected via photoelectron analysis (Fig. 4a). and ultimately to their deposition as coke in the zeolite pore net- 1,10,29,47 CH CHO may undergo aldol condensation to form crotonalde- work . Understanding the generation and evolution of coke hyde (C H O; m/z 70), which has been discussed in the literature . is essential to design efficient and robust catalytic technologies for 4 6 The unambiguous assignment of the crotonaldehyde vibrational converting C platform molecules into fine chemicals and fuels. To fingerprints was not possible in this case, because it has the same achieve this, we have combined operando PEPICO with EPR analy- integer mass as C H , an abundant and isomer-complex C spe- ses, including continuous wave (CW), pulsed and two-dimensional 5 10 5 cies. Upon methylation and dehydration, this aldehyde may gen- hyperfine sublevel correlation (2D HYSCORE) measurements. erate 1,3-pentadiene (C H ; m/z 68) (Fig. 4a). Notably, CH O can By conducting operando PEPICO, a series of C hydrocarbons 5 8 2 10+ also react with CH to form CH CHO. This represents another were observed in both the MTH and MCTH processes at 0.5 bar 3 3 pathway to C–C bond formation, which is thermodynami- (Fig. 5a). These species remained virtually unobserved in the cally favoured compared with reported methane–formaldehyde low-pressure experiments, indicative of reduced chain-propagation (CH –CH O) mechanisms . and condensation-reaction rates. We identify two pathways to 4 2 In the second pathway, CH X can dissociatively adsorb to form naphthalene (C H ; m/z 128), which is considered to be the main 3 10 8 a CH radical and a coordinated leaving group, X* = Cl* or OH* precursor of heavier carbonaceous species (Fig. 5b). The first one (Fig. 4b). The creation of the first C–C bond then occurs via the involves a Diels–Alder dimerization of cyclopentadiene into dicy- reaction of two CH radicals to yield C H upon dehydrogena- clopentadiene (C H ; m/z 132), which isomerizes to 1,2,3,4- and 3 2 4 10 12 tion, analogous to the proposed direct coupling of surface methoxy 1,4,5,8-tetrahydronaphthalene, as observed in the MTH reaction. species to form this olefin . Chain growth proceeds via the addi- The latter species can undergo hydrogen abstraction to form dihy- tion of CH radicals, resulting in a distribution of C H , C H and dronaphthalene and ultimately naphthalene, as identified in both 3 3 6 4 8 C H isomers (Fig. 4b), explaining the complex ms-TPE spectra MTH and MCTH processes via ms-TPE spectrum analysis (Fig. 5b; 5 10 observed for C H and C H (Supplementary Fig. 19). Interestingly, Supplementary Fig. 22). This mechanism was first revealed in detail 4 8 5 10 the signals of isomerized and non-primary olefins were found to be in the catalytic pyrolysis of benzenediols over H-ZSM-5 (ref. ). In stronger, which is in line with their higher thermodynamic stability addition, dicyclopentadiene can hydrogenate and isomerize to ada- compared with primary olefins. CH is formed via the hydrogena- mantine, which can form diamantane, a known coke precursor . • • tion of CH , whereas the CH radical–radical reaction results in The second pathway encompasses the reaction of benzene with pro- 3 3 C H , both products of the MTH and MCTH processes. pyne (C H ; m/z 40) to yield indene (C H ; m/z 116). This agrees 2 6 3 4 9 8 The different C H isomers, which form either via the ketene- with previous studies on ethanol coupling over H-ZSM-5 that have 5 8 driven route in the MTH process or via hydrogen abstraction from shown the generation of indene-like species using solvent extrac- 39 • C H in the MTH and MCTH reactions, can undergo further cycli- tion . Further reaction with CH radicals generates naphthalene 5 10 3 zation/dehydrogenation and methylation to generate cyclopenta- (Fig. 5b; Supplementary Fig. 22), following a known mechanism . diene (C H ; m/z 66) and methyl cyclopentadiene (C H ; m/z 80), Interestingly, by comparing the ms-TPE spectra of m/z 132 photo- 5 6 6 8 respectively, in both reactions (Fig. 4c; Supplementary Fig. 20). The ions, the relative ratio of dicyclopentadiene and tetrahydronaph- detection of (poly)methylated cyclopentadienes corroborates stud- thalenes was lower in MCTH than in the MTH process (Fig. 5b; ies that have discussed their involvement in the MTH mechanism Supplementary Fig. 22). A possible explanation for this behaviour 1,46 and their potential role as further methylating agents . Presumably, is the observed favoured production of aromatics in the MCTH methyl-radical-driven and regular (that is, based on (poly)methyl- reaction that could promote the indene-driven mechanism. Once cyclopentadienes) methylation paths occur in parallel. Although naphthalene forms, it can undergo consecutive methylation to yield it is not possible to distinguish the relative contributions of the alkylated naphthalenes (Fig. 5a,b; Supplementary Fig. 22), which two routes, due to their high reactivity, it is likely that CH radical has been associated with the generation of heavy polyaromatic com- 26,49 species play an important role in the early stages of the reaction, pounds . This C -addition mechanism agrees with experimental 51,52 and that once HCP species form they react indiscriminately in all observations and kinetic models . When switching off the feed steps. Photoelectron analysis detects the dehydrogenation product of the MTH and MCTH reactions, the evolution of heavy species of methylcyclopentadienes (C H ), that is, fulvene (C H ; m/z 78), was also observed (Supplementary Fig. 23). In particular, photoions 6 8 6 6 which generates benzene, the first aromatic ring compound, in a associated with naphthalene were detected after both MTH and subsequent isomerization step (Fig. 4c; Supplementary Fig. 20). MCTH processes, while indene and anthracene could also be iden- Direct methylation of fulvene as well as dehydrogenation–meth- tified to be released after the MCTH reaction. This is in line with the ylation of C H yields methyl fulvene (C H ; m/z 92), the principal favoured generation of indene during the MCTH reaction as well as 6 8 7 8 precursor of toluene (Fig. 4c; Supplementary Fig. 20). This mecha- the higher rate of coke accumulation during the first 60 h on stream nism continues to generate other alkylated benzenes, such as iso- in the long-term tests. mers of xylene and of trimethylbenzene (Supplementary Fig. 21). The evolution of coke species was monitored by studying depos- Vibrational transitions in the ms-TPE spectra of m/z 106 could also ited paramagnetic carbonaceous species via EPR spectroscopy be assigned to ethylbenzene, which can undergo cracking to yield after selected times on stream in MTH and MCTH (Fig. 5c–h). By C H and C H , representing an additional source of olefins from conducting CW EPR, a signal with an isotropic g factor of 2.003 2 4 6 6 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 611 NaTURE CaTalysIs Articles was observed for all investigated samples (Supplementary Fig. 24). enabled us to quantitatively and isomer-selectively map the gen- By applying a reported method for spin counting , it was pos- eration of intermediates during the early stages of the MTH and sible to quantify the specific amount of radicals formed during MCTH processes. In addition, the strategy of jointly studying the the reaction by comparing the double integrals of the EPR signals activation of CH OH and CH Cl was crucial for decoupling and dis- 3 3 with a known standard. Accordingly, the average concentration of criminating the oxygenate- and hydrocarbon-driven mechanisms. EPR-active carbonaceous species increased as a function of the time These results provide experimental evidence for the generation of on stream, which is consistent with the coke contents determined CH radicals in both reactions and of C H O in the MTH process. 3 2 2 −3 via TGA, reaching approximately 4.5 and 7 μmol cm after 120 h in Accordingly, two main C–C bond formation and propagation path- the MTH and MCTH reactions, respectively (Fig. 5c). As detailed in ways are identified. The first, which is dominant in the MCTH pro- Supplementary Note 2, information on the density of these species cess, involves the reaction of two CH radicals to generate C H 3 2 4 was obtained by extracting the Gaussian and Lorentzian lineshape upon hydrogen transfer, which can undergo further methylation to contributions via least-squares fitting of the normalized CW EPR yield complex isomer mixtures of C –C hydrocarbons. The second 3 5 spectra (Fig. 5d, inset; Supplementary Figs. 24 and 25). Whereas occurs only in the MTH process and encompasses the CH OH- and isolated paramagnetic coke generally results in a Gaussian line- (CH ) O-driven formation of CH O, which, after oxidation into 3 2 2 shape, the magnetic interactions of closely packed spins typically CO, can undergo Koch carbonylation to generate C H O, represent- 2 2 yield a Lorentzian-type linewidth. Accordingly, the MCTH reaction ing the first C–C bond generation in the MTH process. This ketene resulted in the formation of highly dense carbonaceous compounds can reproportionate in the presence of CH OH to yield CH O and 3 2 within the first 5 h of reaction, whereas the density of these spe- CH CHO. The latter can undergo aldol condensation into crotonal- cies grew slowly in the MTH reaction (Fig. 5d), consistent with the dehyde, which, upon methylation and dehydration, generates linear higher tendency of CH Cl to generate BTX. C H . Interestingly, CH O can react with a CH radical to generate 3 5 8 2 3 Pulsed relaxation EPR measurements, that is, recording the sig- CH CHO, which is thermodynamically favoured compared with nal decay after microwave pulsing, were performed to further gain proposed CH –CH O routes and represents an additional subroute 4 2 insight into the spatial distribution of the detected paramagnetic for formation of the first C–C bond in the MTH reaction . Both species (Supplementary Fig. 26). Through least-squares fitting of envisioned pathways converge to generate linear and branched C these decay data with a stretched exponential, the corresponding hydrocarbons, such as C H and C H . Following a series of meth- 5 10 5 8 fractal dimension, N, can be extracted (Fig. 5e, inset). Generation ylation, cyclization and hydrogen-transfer reactions, these com- of three-dimensional (3D) deposits (N ≥ 2.8) was observed after pounds form fulvene and methyl fulvenes that isomerize to yield 5 h in the MCTH reaction, whereas N increased slowly in the MTH benzene and toluene, respectively. process, reaching 2.7 after 120 h on stream (Fig. 5e). These results Operando PEPICO spectroscopy also provided insights into the indicate that highly packed carbonaceous species form rapidly in mechanisms for the generation of coke precursors, such as naph- the MCTH process and equally along the three dimensions within thalene and its methylated analogues, in MTH and MCTH. The the zeolite pore network. By contrast, the MTH reaction gener- first involves Diels–Alder dimerization, isomerization and hydro- ates less-dense carbon deposits that grow more slowly, which is gen abstraction of cyclopentadiene, whereas the second encom- in line with the observations gathered using CW EPR, TGA and passes coupling between propyne and benzene to yield indene, sorption analysis. Finally, 2D HYSCORE analyses provide further which forms naphthalene upon methylation. These results were insights into the nature of these deposits (Fig. 5f,g; Supplementary further complemented with EPR measurements that shed light on Figs. 27–31). As detailed in Supplementary Note 2, comparison with the density, spatial distribution and representative molecular struc- density functional theory-based simulations and literature data ture of the carbonaceous species deposited in the zeolite micro- enables the assignment of representative molecular structures to the pores during the MTH and MCTH reactions. By combining CW recorded hyperfine couplings. Marked differences were observed and pulsed EPR with 2D HYSCORE measurements, it could be after 2 h on stream in the MTH and MCTH processes, which observed that the MCTH process results in the fast generation of showed the formation of ethylnaphthalene- and pentacene-like high-density, three-dimensional and low-alkylated carbon deposits, compounds, respectively (Fig. 5f,g). These results are in line with whereas carbonaceous species grow more slowly and form highly the PEPICO experiments performed by switching off the feed, in alkylated aromatics in the MTH process, particularly after a short which naphthalene in both MTH and MCTH together with indene time on stream. However, the degree of alkylation of these depos- and anthracene in MCTH were evolved from the H-ZSM-5 cata- its gradually decreased with time on stream in the MTH reaction lyst (Supplementary Fig. 23). Notably, ethylnaphthalene was pre- until it converged with that obtained in the MCTH reaction after viously observed in ethanol coupling over H-ZSM-5 , suggesting around 60 h. that the generation of coke precursors can be strongly influenced by Determining the different pathways of C–C bond formation and the zeolite framework. Estimation of the corresponding H:C ratio chain propagation that culminate in the generation of carbona- indicates that species with a higher degree of alkylation are formed ceous species has important practical implications for the design of in the MTH process (Fig. 5h). The presence of the ketene-driven catalytic and reactor systems for CH OH and CH Cl coupling, and 3 3 chain-growth route leads to the formation of linear compounds. should favour oxygenate-driven routes to hinder the formation of This, together with the slower initial rate of carbonaceous spe- dense polyaromatic species for developing efficient and robust cata- cies generation observed via CW EPR and TGA, can explain the lytic technologies. On the basis of the map of reactive intermediates reduced tendency of forming highly aromatic compounds in the obtained via the operando analysis in this work, we have proposed MTH process compared with the MCTH reaction. The nature of the a specific reactivity map for those intermediates; further studies will formed deposits converges with time on stream (Fig. 5h), and the undoubtedly provide additional insight that will help to refine fur- representative molecular structure corresponds to alkylated coro- ther our understanding of the actual reaction network. Going for- nenes and tribenzocoronenes after 60 and 120 h in both reactions ward, these techniques provide a valuable platform, potentially in (Supplementary Figs. 28 and 29). combination with co-feeding experiments, for gaining insights into other long-standing questions in hydrocarbon transformations, for Discussion example, to confirm the effects of promoters or reactant leaving The mechanism of C–C bond formation and chain propagation in groups, distinguishing mere spectators from real reaction interme- the zeolite-catalysed coupling of CH OH and CH Cl was assessed diates, and could be applied to virtually all hydrocarbon function- 3 3 using operando PEPICO spectroscopy, whose unique sensitivity alization processes. NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 612 NaTURE CaTalysIs Articles 12. Lin, R., Amrute, A. P. & Pérez-Ramírez, J. Halogen-mediated conversion of Methods hydrocarbons to commodities. Chem. Rev. 117, 4182–4247 (2017). Catalyst synthesis and basic characterization. A commercial ZSM-5 zeolite 13. Yarulina, I. et al. Structure–performance descriptors and the role of Lewis (CBV 8014 from Zeolyst International, nominal Si/Al = 40, ammonium form) was −1 acidity in the methanol-to-propylene process. Nat. Chem. 10, 804–812 (2018). calcined at 823 K (5 K min ) for 5 h to obtain the protonic form. The crystalline 14. Wu, X. et al. Direct mechanism of the first carbon–carbon bond formation order, porous and acidic properties, and the presence of coke in the fresh and in the methanol-to-hydrocarbons process. Angew. Chem. Int. Ed. 56, used catalysts were characterized using powder X-ray diffraction, gas sorption at 9039–9043 (2017). 77 K, Fourier transform infrared spectroscopy of adsorbed pyridine, and TGA, 15. Liu, Y. et al. Formation mechanism of the first carbon–carbon bond and the respectively. Full details are provided in the Supplementary Methods. first olefin in the methanol conversion into hydrocarbons. Angew. Chem. Int. Ed. 55, 5723–5726 (2016). Catalytic evaluation. Methanol and methyl chloride coupling reactions were 16. Comas-Vives, A., Valla, M., Coperet, C. & Sautet, P. Cooperativity between Al performed at ambient pressure using a continuous-flow fixed-bed reactor setup sites promotes hydrogen transfer and carbon–carbon bond formation upon (Supplementary Fig. 32). In a typical test, the zeolite catalyst (catalyst weight, dimethyl ether activation on alumina. ACS Cent. Sci. 1, 313–319 (2015). W = 0.6 g) was loaded in a quartz microreactor and pre-treated at 673 K for 1 h cat 17. Plessow, P. N. & Studt, F. Unraveling the mechanism of the initiation reaction before switching to the desired reaction temperature. Thereafter, a total gas flow, 3 −1 of the methanol to olefins process using ab initio and DFT calculations. F = 20 cm STP min , containing 50 vol% of CH X (X = OH or Cl) in helium, T 3 ACS Catal. 7, 7987–7994 (2017). was fed into the reactor at the desired reaction temperature (T = 523–723 K). 18. Wang, C.-M., Wang, Y.-D., Du, Y.-J., Yang, G. & Xie, Z.-K. Computational Quantification of the reactant conversion and product selectivity, as well as the insights into the reaction mechanism of methanol-to-olefins conversion in error in the carbon mass balance, was conducted according to the protocols H-ZSM-5: nature of hydrocarbon pool. Catal. Sci. Technol. 6, 3279–3288 (2016). described in the Supplementary Methods. 19. Liu, Y. et al. Critical role of formaldehyde during methanol conversion to hydrocarbons. Nat. Commun. 10, 1462 (2019). Operando photoelectron photoion coincidence spectroscopy. PEPICO 20. Wen, W. et al. Formation and fate of formaldehyde in methanol-to- experiments were performed using the CRF-PEPICO endstation at the vacuum hydrocarbon reaction: in situ synchrotron radiation photoionization mass ultraviolet beamline of the Swiss Light Source of the Paul Scherrer Institute, 55–57 spectrometry study. Angew. Chem. Int. Ed. 59, 4873–4878 (2020). Switzerland . The unique reactor configuration extends the lifetime of 21. Yang, L. et al. Role of acetaldehyde in the roadmap from initial carbon– elusive reaction intermediates, thanks to the high dilution of the feed, the low carbon bonds to hydrocarbons during methanol conversion. ACS Catal. 9, pressure inside the reactor (around 0.05–0.5 bar) and the free jet molecular beam 6491–6501 (2019). expansion, enabling their detection (Supplementary Fig. 33). Full details of the 22. Wang, C. et al. Extra-framework aluminum-assisted initial C–C bond protocol are provided in the Supplementary Methods. formation in methanol-to-olefins conversion on zeolite H-ZSM-5. Angew. Chem. Int. Ed. 57, 10197–10201 (2018). Electron paramagnetic resonance spectroscopy. The evolution of coke with the 23. Chowdhury, A. D. et al. Initial carbon–carbon bond formation during the time on stream was monitored using ex situ EPR spectroscopy. A set of experiments, early stages of the methanol-to-olefin process proven by zeolite-trapped including CW, pulsed and 2D HYSCORE measurements, was carried out to acetate and methyl acetate. Angew. Chem. Int. Ed. 55, 15840–15845 (2016). investigate the distinctive features of the deposited carbonaceous species, such as the 24. Hemelsoet, K., Van der Mynsbrugge, J., De Wispelaere, K., Waroquier, M. & density, spatial distribution and representative molecular structure (Supplementary Van Speybroeck, V. Unraveling the reaction mechanisms governing Fig. 34). Full details of the protocol are provided in the Supplementary Methods. methanol-to-olefins catalysis by theory and experiment. ChemPhysChem 14, 1526–1545 (2013). Data availability 25. Chowdhury, A. D. et al. Bridging the gap between the direct and hydrocarbon Data presented in the main figures of the manuscript are publicly available pool mechanisms of the methanol-to-hydrocarbons process. Angew. Chem. through the Zenodo repository (https://doi.org/10.5281/zenodo.6547649). Further Int. Ed. 57, 8095–8099 (2018). data supporting the findings of this study are available in the Supplementary 26. Lezcano-Gonzalez, I. et al. Insight into the effects of confined hydrocarbon Information. All other relevant source data are available from the corresponding species on the lifetime of methanol conversion catalysts. Nat. Mater. 19, author upon request. Source data are provided with this paper. 1081–1087 (2020). 27. Wang, C. et al. Methylbenzene hydrocarbon pool in methanol-to-olefins Received: 22 September 2021; Accepted: 18 May 2022; conversion over zeolite H-ZSM-5. J. Catal. 332, 127–137 (2015). Published online: 27 June 2022 28. Li, J. et al. A route to form initial hydrocarbon pool species in methanol conversion to olefins over zeolites. J. Catal. 317, 277–283 (2014). 29. Müller, S. et al. Coke formation and deactivation pathways on H-ZSM-5 in References the conversion of methanol to olefins. J. Catal. 325, 48–59 (2015). 1. Yarulina, I., Chowdhury, A. D., Meirer, F., Weckhuysen, B. M. & Gascon, J. 30. Wang, S. et al. Origin and evolution of the initial hydrocarbon pool Recent trends and fundamental insights in the methanol-to-hydrocarbons intermediates in the transition period for the conversion of methanol to process. Nat. Catal. 1, 398–411 (2018). olefins over H-ZSM-5 zeolite. J. Catal. 369, 382–395 (2019). 2. Van Speybroeck, V. et al. First principle chemical kinetics in zeolites: 31. Hu, M. et al. Establishing a link between the dual cycles in the methanol-to-olefin process as a case study. Chem. Soc. Rev. 43, methanol-to-olefins conversion on H-ZSM-5: aromatization of cycloalkenes. 7326–7357 (2014). ACS Catal. 10, 4299–4305 (2020). 3. Tian, P., Wei, Y., Ye, M. & Liu, Z. Methanol to olefins (MTO): from 32. Valecillos, J., Manzano, H., Aguayo, A. T., Bilbao, J. & Castaño, P. Kinetic and fundamentals to commercialization. ACS Catal. 5, 1922–1938 (2015). deactivation differences among methanol, dimethyl ether and chloromethane 4. Olivos-Suarez, A. I. et al. Strategies for the direct catalytic valorization of as stock for hydrocarbons. ChemCatChem 11, 5444–5456 (2019). methane using heterogeneous catalysis: challenges and opportunities. ACS 33. Paunović, V., Mitchell, S., Verel, R., Lee, S. S. & Pérez-Ramírez, J. Aluminum Catal. 6, 2965–2981 (2016). redistribution in ZSM-5 zeolite upon interaction with gaseous halogens 5. Čejka, J., Centi, G., Perez-Pariente, J. & Roth, W. J. Zeolite-based materials for and hydrogen halides and implications in catalysis. J. Phys. Chem. C 124, novel catalytic applications: opportunities, perspectives and open problems. 722–733 (2019). Catal. Today 179, 2–15 (2012). 34. Ibáñez, M. et al. Simultaneous coking and dealumination of zeolite H-ZSM-5 6. Zichittella, G. & Pérez-Ramírez, J. Status and prospects of the decentralised during the transformation of chloromethane into olefins. Catal. Sci. Technol. valorisation of natural gas into energy and energy carriers. Chem. Soc. Rev. 6, 296–306 (2016). 50, 2984–3012 (2021). 35. Zichittella, G., Hemberger, P., Holzmeier, F., Bodi, A. & Perez-Ramirez, J. 7. Zhong, J. et al. State of the art and perspectives in heterogeneous catalysis of Operando photoelectron photoion coincidence spectroscopy unravels CO hydrogenation to methanol. Chem. Soc. Rev. 49, 1385–1413 (2020). mechanistic fingerprints of propane activation by catalytic oxyhalogenation. 8. Ilias, S. & Bhan, A. Mechanism of the catalytic conversion of methanol to J. Phys. Chem. Lett. 11, 856–863 (2020). hydrocarbons. ACS Catal. 3, 18–31 (2012). 36. Hemberger, P., Custodis, V. B. F., Bodi, A., Gerber, T. & van Bokhoven, J. A. 9. Olsbye, U. et al. Conversion of methanol to hydrocarbons: how zeolite cavity Understanding the mechanism of catalytic fast pyrolysis by unveiling reactive and pore size controls product selectivity. Angew. Chem. Int. Ed. 51, intermediates in heterogeneous catalysis. Nat. Commun. 8, 15946 (2017). 5810–5831 (2012). 37. Hemberger, P., van Bokhoven, J. A., Pérez-Ramírez, J. & Bodi, A. New 10. Olsbye, U. et al. The formation and degradation of active species during analytical tools for advanced mechanistic studies in catalysis: photoionization methanol conversion over protonated zeotype catalysts. Chem. Soc. Rev. 44, and photoelectron photoion coincidence spectroscopy. Catal. Sci. Technol. 10, 7155–7176 (2015). 1975–1990 (2020). 11. Olsbye, U. et al. Methane conversion to light olefins—how does the methyl 38. Brückner, A. In situ electron paramagnetic resonance: a unique tool for halide route differ from the methanol-to-olefins (MTO) route? Catal. Today analyzing structure–reactivity relationships in heterogeneous catalysis. Chem. 171, 211–220 (2011). Soc. Rev. 39, 4673–4684 (2010). NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 613 NaTURE CaTalysIs Articles 39. Madeira, F. F. et al. Radical species detection and their nature evolution with 56. Bodi, A., Sztáray, B., Baer, T., Johnson, M. & Gerber, T. Data acquisition catalyst deactivation in the ethanol-to-hydrocarbon reaction over HZSM-5 schemes for continuous two-particle time-of-flight coincidence experiments. zeolite. ACS Catal. 1, 417–424 (2011). Rev. Sci. Instrum. 78, 084102 (2007). 40. Martinez-Espin, J. S. et al. New insights into catalyst deactivation and 57. Bodi, A. et al. Imaging photoelectron photoion coincidence spectroscopy with product distribution of zeolites in the methanol-to-hydrocarbons (MTH) velocity focusing electron optics. Rev. Sci. Instrum. 80, 034101 (2009). reaction with methanol and dimethyl ether feeds. Catal. Sci. Technol. 7, 2700–2716 (2017). Acknowledgements 41. Foley, B. L., Johnson, B. A. & Bhan, A. A method for assessing catalyst This publication was created as part of NCCR Catalysis (grant 180544), a National deactivation: a case study on methanol-to-hydrocarbons conversion. ACS Centre of Competence in Research funded by the Swiss National Science Foundation. Catal. 9, 7065–7072 (2019). The operando PEPICO measurements were performed at the VUV beamline of the Swiss 42. Palumbo, L. et al. Conversion of methanol to hydrocarbons: spectroscopic Light Source (SLS) located at the Paul Scherrer Institute (PSI), Villigen, Switzerland. characterization of carbonaceous species formed over H-ZSM-5. J. Phys. Chem. C 112, 9710–9716 (2008). 43. Zang, Z., Zhang, W., Huang, J., Feng, P. & Ding, J. First molecule with Author contributions carbon–carbon bond in methanol-to-olefins process. Chem. Phys. Lett. 737, J.P.-R. conceived the research. G.Z., S.M. and J.P.-R. coordinated the study. G.Z., 136844 (2019). A.C., S.M. and J.P.-R. wrote the article with input from all other co-authors. A.C., 44. Plessow, P. N., Smith, A., Tischer, S. & Studt, F. Identification of the reaction S.M. and G.Z. developed the methodology to quantitatively compare the coupling sequence of the MTO initiation mechanism using ab initio-based kinetics. reactions, and A.C. conducted basic characterization and testing of the catalysts. P.H. J. Am. Chem. Soc. 141, 5908–5915 (2019). and A.B. developed the operando PEPICO technique, and G.Z. and P.H. conducted 45. Chowdhury, A. D. & Gascon, J. The curious case of ketene in zeolite the measurements and analysed the data. A.C. assisted with the measurements. chemistry and catalysis. Angew. Chem. Int. Ed. 57, 14982–14985 (2018). Z.P. conducted the Franck–Condon simulations. G.J. and M.A. developed the EPR 46. Zhang, W. et al. Methylcyclopentenyl cations linking initial stage and spectroscopy analysis, and M.A. and S.P.S. conducted the experiments. highly efficient stage in methanol-to-hydrocarbon process. ACS Catal. 10, 4510–4516 (2020). Competing interests 47. Bleken, F. L. et al. Catalyst deactivation by coke formation in microporous The authors declare no competing interests. and desilicated zeolite H-ZSM-5 during the conversion of methanol-to-hydrocarbons. J. Catal. 307, 62–73 (2013). 48. Pan, Z., Puente-Urbina, A., Bodi, A., van Bokhoven, J. A. & Hemberger, P. Additional information Isomer-dependent catalytic pyrolysis mechanism of the lignin model Supplementary information The online version contains supplementary material compounds catechol, resorcinol and hydroquinone. Chem. Sci. 12, available at https://doi.org/10.1038/s41929-022-00808-0. 3161–3169 (2021). Correspondence and requests for materials should be addressed to 49. Goetze, J. et al. Insights into the activity and deactivation of the Guido Zichittella, Patrick Hemberger or Javier Pérez-Ramírez. methanol-to-olefins process over different small-pore zeolites as studied with operando UV-vis spectroscopy. ACS Catal. 7, 4033–4046 (2017). Peer review information Nature Catalysis thanks the anonymous reviewers for their 50. Mebel, A. M., Landera, A. & Kaiser, R. I. Formation mechanisms of contribution to the peer review of this work. naphthalene and indene: from the interstellar medium to combustion flames. Reprints and permissions information is available at www.nature.com/reprints. J. Phys. Chem. A 121, 901–926 (2017). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 51. Rojo-Gama, D. et al. Deactivation of zeolite catalyst H-ZSM-5 during published maps and institutional affiliations. conversion of methanol to gasoline: operando time- and space-resolved X-ray diffraction. J. Phys. Chem. Lett. 9, 1324–1328 (2018). Open Access This article is licensed under a Creative Commons 52. Janssens, T. V. W., Svelle, S. & Olsbye, U. Kinetic modeling of deactivation Attribution 4.0 International License, which permits use, sharing, adap- profiles in the methanol-to-hydrocarbons (MTH) reaction: a combined tation, distribution and reproduction in any medium or format, as long autocatalytic–hydrocarbon pool approach. J. Catal. 308, 122–130 (2013). as you give appropriate credit to the original author(s) and the source, provide a link to 53. Zichittella, G., Polyhach, Y., Tschaggelar, R., Jeschke, G. & Pérez-Ramírez, J. the Creative Commons license, and indicate if changes were made. The images or other Quantification of redox sites during catalytic propane oxychlorination by third party material in this article are included in the article’s Creative Commons license, operando EPR spectroscopy. Angew. Chem. Int. Ed. 60, 3596–3602 (2021). unless indicated otherwise in a credit line to the material. If material is not included in 54. Ben Tayeb, K. et al. The radical internal coke structure as a fingerprint of the the article’s Creative Commons license and your intended use is not permitted by statu- zeolite framework. Microporous Mesoporous Mater. 289, 109617 (2019). tory regulation or exceeds the permitted use, you will need to obtain permission directly 55. Sztáray, B. et al. CRF-PEPICO: double velocity map imaging photoelectron from the copyright holder. To view a copy of this license, visit http://creativecommons. photoion coincidence spectroscopy for reaction kinetics studies. J. Chem. org/licenses/by/4.0/. Phys. 147, 013944 (2017). © The Author(s) 2022 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Catalysis Springer Journals

Elucidation of radical- and oxygenate-driven paths in zeolite-catalysed conversion of methanol and methyl chloride to hydrocarbons

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/elucidation-of-radical-and-oxygenate-driven-paths-in-zeolite-catalysed-QzrSAvLLSP

References (61)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
eISSN
2520-1158
DOI
10.1038/s41929-022-00808-0
Publisher site
See Article on Publisher Site

Abstract

Articles https://doi.org/10.1038/s41929-022-00808-0 Elucidation of radical- and oxygenate-driven paths in zeolite-catalysed conversion of methanol and methyl chloride to hydrocarbons 1,4 1,4 1,4 2 2  ✉ Alessia Cesarini , Sharon Mitchell , Guido Zichittella    , Mikhail Agrachev , Stefan P. Schmid    , 2 3 3 3 1  ✉  ✉ Gunnar Jeschke    , Zeyou Pan , Andras Bodi    , Patrick Hemberger    and Javier Pérez-Ramírez    Understanding hydrocarbon generation in the zeolite-catalysed conversions of methanol and methyl chloride requires advanced spectroscopic approaches to distinguish the complex mechanisms governing C–C bond formation, chain growth and the deposition of carbonaceous species. Here operando photoelectron photoion coincidence (PEPICO) spectroscopy enables the isomer-selective identification of pathways to hydrocarbons of up to C in size, providing direct experimental evidence of methyl radicals in both reactions and ketene in the methanol-to-hydrocarbons reaction. Both routes converge to C mol- ecules that transform into aromatics. Operando PEPICO highlights distinctions in the prevalence of coke precursors, which is supported by electron paramagnetic resonance measurements, providing evidence of differences in the representative molecular structure, density and distribution of accumulated carbonaceous species. Radical-driven pathways in the methyl chloride-to-hydrocarbons reaction(s) accelerate the formation of extended aromatic systems, leading to fast deactivation. By contrast, the generation of alkylated species through oxygenate-driven pathways in the methanol-to-hydrocarbons reaction extends the catalyst lifetime. The findings demonstrate the potential of the presented methods to provide valuable mechanistic insights into complex reaction networks. onosubstituted methanes, including methanol and Comparatively, mechanistic understanding of the methyl methyl chloride (CH X, X = OH and Cl, respectively), chloride-to-hydrocarbons (MCTH) reaction is less developed. Mare attractive building blocks for sustainable fuels and Previous studies have focused on kinetic aspects, showing broadly 1–7 chemicals . These C molecules are the main products of syngas similar product distributions to the MTH process, attributed to conversion, carbon dioxide hydrogenation and halogen-mediated similar confinement effects within the zeolite pore networks on 11,12,32–34 alkane functionalization, and can be readily converted into valuable HCP development . Nevertheless, the absence of oxygenates 8–13 hydrocarbons over zeolite catalysts . However, uncontrolled chain and the faster catalyst deactivation reported in the MCTH process growth typically results in broad product distributions and the suggest that important mechanistic distinctions exist. A system- deposition of heavy carbonaceous species, leading to suboptimal atic comparison of the pathways for the activation and subsequent catalyst selectivity and stability. Towards improving productivity, transformation of CH Cl and CH OH could help decouple direct 3 3 extensive experimental and theoretical efforts have been devoted C–C bond formation routes from oxygenate-driven chain growth. to understanding the mechanism of C–C bond formation and its However, the direct observation of short-lived active intermediates, propagation to higher hydrocarbons in the commercially applied such as ketene and methyl radicals, under operando conditions 1,14–18 methanol-to-hydrocarbons (MTH) reaction . Detailed kinetic remains highly challenging because of the difficulty in discriminat- studies, often combined with isotopic labelling and spectroscopic ing them from stable spectator species due to the limited sensitiv- experiments, evidenced the autocatalytic nature of the process, ity of established methods. Advanced time-resolved spectroscopic provided the basis for the widely accepted dual aromatic–olefin techniques that can distinguish compounds present in low concen- cycles and identified several intermediates expected to play crucial trations and distinct isomeric forms could provide new insights to roles in the early stages and steady-state operation of the reaction help resolve some of the long-standing debates. 19–31 (Supplementary Note 1) . Despite these advances, several ques- Herein, we analyse hydrocarbons desorbed from a representative tions about the mechanism remain strongly debated, including the H-ZSM-5 zeolite catalyst during MTH and MCTH processes using existence and role of radical species, the mechanism of the first operando PEPICO spectroscopy under relevant reaction conditions. 35–37 C–C bond formation and subsequent paths to the early aromatic This technique has shed light on complex reaction networks hydrocarbon pool (HCP) species. Furthermore, archetypical dis- and enables the quantitative, isomer-selective identification of all solution–extraction experiments and spectroscopic measurements short- and long-lived species up to approximately C in size. By have provided limited insight into the specific structures of the coke performing temperature-dependent studies, operando PEPICO precursors and their evolution into carbonaceous species deposited enables the discrimination of primary and secondary intermediates within the zeolite catalyst (Supplementary Note 1). and tertiary products. A complementary study of the used catalysts 1 2 Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland. Laboratory of Physical Chemistry, Department of Chemistry and Applied Biosciences, ETH Zurich, Zurich, Switzerland. Laboratory of Synchrotron Radiation and Femtochemistry, Paul Scherrer Institute, Villigen, Switzerland. These authors contributed equally: Alessia Cesarini, Sharon Mitchell, Guido Zichittella. e-mail: [email protected]; [email protected]; [email protected] NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 605 ms-TPE signal (a.u.) NaTURE CaTalysIs Articles Isomer identification Detection of intermediates Operando PEPICO Ion CH Ion 8 9 10 14 15 16 hν (eV) m/z To GC-FID CH X analysis H-ZSM-5 Kinetics EPR Product analysis CH Coke Coke amount nature Olefins Alkanes BTX 0 MTH MCTH 200 300 400 0 10 20 30 Other B (mT) ω (MHz) Fig. 1 | Approach to mapping the catalysed generation and evolution of hydrocarbons. Schematic of the multi-technique strategy used in this study to unravel the growth of the carbon chain, from the formation of the first C–C bond to the generation of coke in MTH and MCTH over H-ZSM-5. The comparative study of these two C platform molecules enables the main pathways of oxygenate- and hydrocarbon-driven mechanisms to be decoupled and elucidated. Accordingly, operando PEPICO enables the isomer-selective identification of reaction intermediates and coke precursors. This is complemented with EPR measurements that provide insights into the representative molecular structure, density and distribution of deposited carbonaceous species, and kinetic analysis for assessment of the catalytic activity, selectivity and stability. GC-FID, gas chromatography with flame ionization detection. 38,39 via electron paramagnetic resonance (EPR) revealed distinctions attributed to the rapid rupture of Si–O–Al bonds due to interac- in the representative molecular structure, density and distribution tion with HCl. The similar conversions shown in our study suggest 11,33 of the carbonaceous species deposited in the zeolite pore network. that the MFI-type framework of H-ZSM-5 offers higher stability In combination with kinetic analysis, the adopted approach (Fig. 1) and that the leaving group does not severely influence the intrin- provides high molecular resolution of the species formed, enabling sic kinetics of MTH and MCTH. The products are reported in five experimental verification of several open questions about the reac- main groups: CH , C –C olefins, C –C alkanes, benzene–toluene– 4 2 4 2 4 tion mechanisms. xylene (BTX) and other species, including C hydrocarbons. At 5+ 673 K, while the generation of CH , olefins and alkanes was simi- Results lar, the formation of BTX was found more favoured in the MCTH Evaluation of the reaction kinetics. Preliminary insights into the process (52% selectivity) compared with MTH (38% selectivity). An similarities and differences between MTH and MCTH processes opposite trend was observed for C hydrocarbons, with selectivities 5+ were gained by studying the steady-state kinetics of the reactions of 4% and 12% for MCTH and MTH, respectively. Contrariwise, a over an H-ZSM-5 zeolite catalyst with a nominal Si/Al ratio of 40 rise in the reaction temperature (T) led to an increase in the for- under equivalent conditions (Fig. 2; Supplementary Figs. 1 and 2). mation of olefins at the expense of BTX in the MCTH process and The reactivity was assessed by calculating the conversion (X(CH X)) especially in the MTH reaction (Supplementary Fig. 1a), which is 12,32 based on the concentration of CH X at the reactor inlet and outlet. consistent with other studies . A stronger rise in CH genera- 3 4 According to common practice, the reactivity in the MTH process tion was observed in MCTH compared with the MTH process, was also evaluated considering both CH OH and dimethyl ether which suggests that a lower rate of trans-hydrogenation reactions ((CH ) O) as reactants since an equilibrium is typically established in MCTH than in MTH is favoured by the presence of (CH ) O 3 2 3 2 1 40 32 between the two . Since no (CH ) O was observed at 100% conver- (ref. ). In line with the literature , (CH ) O, formed by the revers- 3 2 3 2 sion, recalculation of the conversion and product distribution does ible dehydration of CH OH, was not observed in the MTH reac- not alter the conclusions below (Fig. 2; Supplementary Figs. 1 and tion at 100% conversion. To extract additional kinetic information, 2). Both reactions showed a strong temperature dependence, and the product distribution was also compared at approximately 65% their light-off curves were found to be comparable, particularly CH X conversion, which was achieved by adjusting the space veloc- when considering (CH ) O as a reactant (Fig. 2a; Supplementary ity (Supplementary Fig. 1b). Under these conditions, a considerable 3 2 Fig. 2a). This points to similar kinetic behaviour of the MTH and amount of (CH ) O was observed in the MTH process (27% selec- 3 2 MCTH reactions over H-ZSM-5, which contrasts with previous tivity). In addition, the formation of olefins and C hydrocarbons 5+ findings obtained over H-SAPO-34 zeolite catalysts . In this study was favoured at 65% conversion compared with full CH X conver- by Olsbye et al., lower conversions for the MCTH reaction were sion at 673 K (olefin selectivity: 45% versus 26% for MTH, 40% NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Intensity (a.u.) Intensity (a.u.) ω (MHz) 2 NaTURE CaTalysIs Articles stable during the first 60 h, before BTX production decreased, ulti- mately reaching negligible production at the end of the run, while that of olefins increased. The used catalysts were characterized 80 after selected times on stream in the MTH and MCTH processes to quantify the amount and location of carbonaceous coke species deposited and to confirm the effects of the reaction on the crystal- linity and acidic properties of the H-ZSM-5 zeolite. Consistent with previous reports that have shown coking as a deactivation pathway 34,41,42 in both reactions , thermogravimetric analysis (TGA) provided evidence of the accumulation of substantial carbonaceous deposits 0 (up to 15 wt%) (Fig. 2c). Notably, coke accumulation was faster in MTH 573 673 773 the early stages of the MCTH process compared with MTH, which MCTH T (K) is in line with the higher tendency of this C platform molecule to generate BTX. However, the rate of coke deposition subsequently 0 20 40 60 80 100 120 slowed and, after 120 h on stream, the coke content was slightly Time on stream (h) lower than in the catalyst used in the MTH reaction. This reduction in coking rate is consistent with the strong decrease in the produc- b MTH tion of BTX and C hydrocarbons observed during the long-term 5+ 37 34 32 24 28 test for the MCTH reaction (Fig. 2a,c), which have been identified 41,42 26 as important classes of coke precursors . A linear inverse correla- 15 21 15 10 13 16 tion between the micropore volume and coke content was found 3 2 3 (Fig. 2d). This indicates that carbonaceous species evolve and accu- MCTH mulate similarly within the pore network, despite the different rate 24 25 34 76 of deposition, which is in line with the results from EPR (vide infra). 44 44 44 19 19 18 X-ray diffraction patterns of the used zeolite catalysts after the oxi- 12 10 dative removal of coke species showed no alteration of the crystal 1 2 2 2 5 5 structure after either reaction (Supplementary Fig. 3). By contrast, 10 h 25 h 60 h 120 h the infrared spectroscopy study of adsorbed pyridine highlighted substantial changes in the acidic properties after the MCTH reac- CH Olefins Alkanes BTX Other tion, indicating decreased concentrations of Brønsted acid sites c and increased concentrations of Lewis acid sites (Supplementary d 0.2 Table 1). These observations agree with previous reports on the MTH MTH impact of treatment with gaseous HCl-containing streams, a MCTH MCTH by-product formed in equimolar amounts in the MCTH process, 0.1 33,34 on the zeolite properties . Paths for hydrocarbon formation. Operando PEPICO analysis 0 0 enables us to obtain detailed insight into the intermediates and 0 60 120 0 10 20 products formed in the reactions over H-ZSM-5 at relevant con- Time on stream (h) Coke (wt%) ditions (T ≤ 773 K, P ≤ 0.5 bar), showing a wide range of hydrocar- bons (C –C ) in the MTH and MCTH processes and of oxygenates 1 14 Fig. 2 | Comparative performance of the zeolite catalyst. a, Reactant (C –C ) exclusively in the MTH reaction (Figs. 3–5; Supplementary 1 4 conversion in the MTH and MCTH processes as a function of the time on Figs. 4–23). As detailed in the Supplementary Methods, the reac- stream and temperature (inset) over H-ZSM-5. b, Product distribution as tants and intermediates or products desorbed from the catalyst are a function of time on stream averaged over a period of 1 h. c, Amount of detected via photoionization with monochromatic vacuum ultravi- coke species deposited measured via TGA as a function of time on stream. olet light. The photoions and photoelectrons generated are detected d, Variation in the micropore volume (V ) as a function of the amount micro in coincidence, revealing their mass-to-charge ratio (m/z) and their 3 −1 of coke. Conditions: CH X:He = 1:1, F  = 20 cm  STP min , W  = 0.6 g, 3 T cat isomeric identity based on the photoion mass-selected threshold T = 673 K, P = 1 bar. photoelectron (ms-TPE) spectrum. Two types of experiment were conducted for both reactions . The first type used a packed bed of zeolite catalyst that permitted operation under near-ambient versus 24% for MCTH; other selectivity: 14% versus 12% for MTH, conditions (P = 0.5 bar) (Figs. 3 and 4; Supplementary Figs. 4–7, 11% versus 4% for MCTH) at the expense of BTX (5% versus 38% 14 and 17), whereas the second experiment used a zeolite-coated for MTH, 35% versus 52% for MCTH). These common trends agree microreactor for low-pressure measurements (P = 0.05 bar) with the expected comparable mechanism of chain growth in the (Supplementary Figs. 8–11 and 15–17). Conversion estimates were MTH and MCTH processes, in which olefins and C hydrocarbons obtained by normalizing the feed signals to a xenon (Xe) inter- 5+ act as precursors for the formation of aromatic compounds. nal standard and comparing them with those obtained in blank Comparison of the time-dependent reactivity showed that full experiments at the same temperature (90 and 65% for MTH or CH X conversion could be maintained for 120 h in the MTH process 50 and 35% for MCTH at near-ambient or low pressure, respec- (Fig. 2a,b). The product distribution remained virtually unaffected tively). They confirmed the similar conditions in the operando for the first 60 h on stream, although olefins increased from approxi- PEPICO experiments to the kinetic tests, closing the gap between mately 26% to 40% after 120 h in the MTH reaction at the expense of standard testing and operando measurements. Consistently, no BTX (32% versus 28%) and C hydrocarbons (30% versus 16%). By significant deactivation was observed during the measurements. 5+ contrast, CH Cl conversion gradually decreased to approximately Blank measurements conducted using an empty reactor con- 80% over the first 60 h before dropping more sharply to less than firmed the absence of products in both reactions (Supplementary 20% after 120 h on stream. The product distribution also remained Figs. 12 and 13). NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Coke (wt%) X(CH X) (%) X(CH X) (%) 3 –1 V (cm g ) micro Intensity (a.u.) Area (counts) NaTURE CaTalysIs Articles MTH MCTH 400 2 CH CHO CH • CH CH O (CH ) O LP LP NA 3 3 3 2 3 2 Blank CH CH CHO 15 15 m/z m/z 9.6 9.8 H-ZSM-5 1 CH O hν (eV) (CH ) O 3 2 0 0 30 44 46 14 15 C H C H C H C H 4 6 C H C H 4 8 4 6 4 8 4 6 4 8 350 40 0 0 57 54 55 C H C H C H C H C H C H 5 6 5 8 5 6 5 8 5 6 5 8 45 3 0 0 66 67 67 C H 5 10 C H 5 10 C H 5 10 230 32 0 0 71 69 70 C H 6 8 C H 7 10 C H C H C H C H 6 8 7 10 6 8 7 10 100 5 0 0 79 80 94 95 79 80 94 95 600 700 600 700 800 m/z m/z T (K) Fig. 3 | Evolution of reaction intermediates. Peak areas of the main reaction intermediates as a function of temperature in the MTH and MCTH reactions over H-ZSM-5 determined using operando PEPICO. The blank reactor experiments are shown as reference (open symbols). Representative mass spectra of the products detected for MTH and MCTH processes over the zeolite at 698 K. The insets show the mass spectra of the detected CH radicals during low-pressure (LP) or near-ambient (NA) experiments at 773 K, as detailed in Supplementary Figs. 9 and 11, confirming their presence under relevant conditions. The photon energies at which the signals of each chemical species were recorded in all PEPICO experiments are shown in Supplementary 3 −1 Figs. 5 and 7. Conditions: CH X:Xe:Ar = 1.0:0.1:20.9, F  = 22 cm  STP min , W  = 0.05 g, P = 0.5 bar (insets, P = 0.05 bar). 3 T cat The near-ambient-pressure measurements confirmed the temperature window in the near-ambient pressure experiments, expected formation of major product groups, such as ethylene the C H , C H and C H signal intensities reached a maximum 3 6 4 8 5 10 (C H ), propylene (C H ), BTX and other C hydrocarbons, includ- at approximately 673 K, and the evolution of C H , C H , C H and 2 4 3 6 4+ 5 8 5 6 6 8 ing C H (m/z 56), C H (m/z 66), C H (m/z 68), C H (m/z 70), C H peaked at around 723 K in both the MTH and MCTH reac- 4 8 5 6 5 8 5 10 7 10 C H (m/z 80) and C H (m/z 94) (Fig. 3; Supplementary Figs. 4–11). tions (Fig. 3; Supplementary Fig. 14). These observations support 6 8 7 10 Extraction of the ms-TPE spectra showed that each of these com- the presence of a consecutive mechanism in both reactions, where pounds was present as a complex mixture of isomers, with varying C compounds are central for chain growth, in line with previous degrees of branching and cyclization (Fig. 4; Supplementary Figs. 19 hypotheses . Conversely, the (CH ) O signal decreased sharply with 3 2 and 20). Oxygenates, including dimethyl ether ((CH ) O; m/z 46), increasing temperature in the MTH process, while that of CH CHO 3 2 3 formaldehyde (CH O; m/z 30) and acetaldehyde (CH CHO; m/z 44) followed a volcano behaviour (Fig. 3). These results confirm the 2 3 were detected in the MTH process (Figs. 3 and 4; Supplementary generation of (CH ) O in the early stages of the MTH process via 3 2 Figs. 5, 9 and 17). The most notable distinction in low-pressure CH OH dehydration, whereas CH CHO is probably involved con- 3 3 experiments was the detection of a readily identifiable signal asso- secutively in the formation of the first C–C bond. By contrast, the ciated with methyl radicals (CH ; m/z 15) and of their character- formation of CH O rose continuously with temperature. The same 3 2 istic vibrational fingerprints in the corresponding ms-TPE spectra trend was observed in the low-pressure experiments in the MTH (Fig. 3; Supplementary Figs. 9, 11 and 16). The reduced pressure reaction, although the amount of CH O generated was considerably suppresses the quenching of this highly reactive species, enabling higher (Fig. 3; Supplementary Fig. 17). This suggests that CH O is its observation. This experimentally confirms the presence of CH predominantly involved in the first steps of the hydrocarbon evolu- species in both MTH and MCTH reactions. The detection of CH tion and not in the generation of coke precursors, in agreement with 19,20 radicals also at near-ambient pressure demonstrates that these spe- recent studies . cies are reactive intermediates under realistic conditions rather than On the basis of these findings, two main reaction pathways just minor desorption by-products from surface methoxy species or appear to be responsible for the first C–C bond formation and mere spectators (Fig. 3; Supplementary Fig. 18). the subsequent chain growth to C compounds. The first involves Signal integration at suitable photon energies enabled quantifi- the evolution of oxygenates, such as (CH ) O and CH CHO, and 3 2 3 cation of the relative abundance of distinct species as a function of is only relevant in the MTH process (Fig. 4a), whereas the second temperature (Fig. 3; Supplementary Figs. 14–17). The increased gen- encompasses olefins and occurs in both MTH and MCTH processes eration of (CH ) O in the MTH reaction and the higher olefin/BTX (Fig. 4b). In the oxygenate-driven mechanism, CH OH probably 3 2 3 ratio in both reactions with decreasing conversion agrees well with initially undergoes dehydration to generate (CH ) O. Calculations 3 2 the kinetic experiments. While BTX increased in the investigated have shown that both CH OH and (CH ) O can react to CH O, 3 3 2 2 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal Intensity (a.u.) Area counts NaTURE CaTalysIs Articles + + + + + + (CH ) O CH O C H O CH CHO C H O C H 3 2 2 2 2 3 4 6 5 8 9.6 10.0 10.4 10.0 10.5 11.0 9.4 9.5 9.6 9.7 10.1 10.3 10.5 9.6 10.0 10.4 8.4 8.8 9.2 9.6 hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) CH OH CH O CH O + CO H C O + CH OH 2CH CHO C H O + CH 3 2 2 2 2 3 3 4 6 3 . . . . – H O – H O – H O – H O – 2H – 2H + 2H + H 2 2 2 (CH ) O CH O CO H C O CH CHO + CH O C H O C H 3 2 2 2 2 3 2 4 6 5 8 – CH + + + + C H b CH C H C H C H 5 10 3 2 4 3 6 4 8 9.6 9.8 10.0 10.4 10.5 10.6 9.6 10.0 10.4 8.8 9.2 9.6 8.4 8.8 9.2 9.6 hν (eV) hν (eV) hν (eV) hν (eV) hν (eV) . . . . CH X 2CH C H + CH C H + CH C H +CH 3 3 2 4 3 3 6 3 4 8 3 . . . . – 2H – H – H –H C H C H C H C H CH + X* 2 4 3 6 4 8 5 10 + CH + CH – H – H – 2H + CH – 2H – H – 2H – 2H + CH – H + + + + + + C H C H C H C H C H C H 5 8 5 6 6 8 6 6 7 10 7 8 8 9 10 8 9 10 8 9 10 8 9 10 8 9 10 8 9 10 hν (eV) Fig. 4 | Reaction pathways of catalytic C coupling. a,b, Proposed reaction network of the oxygenate-driven reaction in the MTH process (a) and of the direct CH radical-addition pathway in the MTH and MCTH reactions (b) to yield C intermediates in the micropores of H-ZSM-5. The insets on top of the 3 5 molecular schemes show the ms-TPE (open squares) and reference spectra (solid lines) of the identified species for MTH in a and MCTH in b. The main identified isomers are represented by the molecular models in a,b, while other structures are shown in Supplementary Figs. 18–22. Colour code: C (dark grey), O (red), H (light grey), X (X = OH or Cl; violet). c, Reaction pathway for the chain-propagation reaction of C H to generate benzene and toluene 5 8 in the MTH process. The ms-TPE (open squares) and reference spectra (solid lines) of the identified isomeric products are also shown. Conditions as reported in Fig. 3. 17,43,44 releasing H and CH , respectively . CH O can be oxidized to via Koch carbonylation to yield the ketene ethenone (C H O; m/z 2 4 2 2 2 carbon monoxide (CO) as demonstrated experimentally and theo- 42) , which is assigned by its ionization transition at 9.6 eV in the 15,17 retically . Consistent with previous literature reports, carbon m/z 42 ms-TPE spectrum (Fig. 4a). Although this intermediate monoxide can react with adsorbed CH OH, (CH ) O or CH O had been previously predicted , our data provide the experimental 3 3 2 2 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal ms-TPE signal (a.u.) NaTURE CaTalysIs Articles a b + CH C H MTH 3 10 8 – 4H C H MCTH 10 10 – 3H – 2H C H 10 12 C H + 13 14 C H 11 10 C H – H + CH 11 12 + 2 C H 12 12 C H 9 8 8.1 eV – H + CH C H 10 12 – H + CH 8.0 8.4 8.8 9.2 120 130 140 150 160 170 180 8.0 8.4 8.8 9.2 9.6 hν (eV) hν (eV) m/z c d 8 1.0 3.0 MTH 3D High 0.8 2.8 6 MCTH density Data 0.6 2.6 Fit Fit 0.4 2.4 Data 337 341 Low 0.2 2.2 0 3,000 density B (mT) 2D t (ns) 0 0 2.0 0 40 80 120 0 40 80 120 0 40 80 120 Time on stream (h) Time on stream (h) Time on stream (h) f g 1.1 MTH MCTH High alkylation 30 30 1.0 MTH MCTH 0.9 0.8 0.7 10 10 0.6 Low alkylation 0 0 0.5 0 10 20 30 0 10 20 30 0 40 80 120 ω (MHz) ω (MHz) Time on stream (h) 1 1 Fig. 5 | Generation and evolution of condensed carbonaceous species. a, Mass spectra of the C species detected at the reactor outlet in the MTH 10+ and MCTH reactions over H-ZSM-5. b, Reaction pathways for the formation of naphthalene and methylated naphthalenes in MTH. The insets show the ms-TPE (open squares) and reference spectra (solid lines) of the identified products. c–e, EPR-active concentration (c), Lorentzian fraction (d) and fractal dimension (e) of carbonaceous deposits in the used catalysts as a function of time on stream in MTH and MCTH as determined using CW EPR and pulsed EPR, respectively. The insets in d,e, illustrate how these parameters were extracted. f,g, Weak interaction quadrant of the 2D Hy SCORE spectra of the zeolite after 2 h on stream for MTH (f) and MCTH (g). The insets depict the representative molecular structures of the carbonaceous species. h, H:C molar ratio extracted from the 2D Hy SCORE spectra as a function of time on stream in MTH and MCTH over the zeolite. Conditions: a,b, Conditions as 3 −1 reported in Fig. 3; c–h, CH X:He = 1:1, F  = 20 cm  STP min , W  = 0.6 g, T = 673 K, P = 1 bar. 3 T cat NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal ω (MHz) –3 Intensity (a.u.) Coke spins (μmol cm ) ω (MHz) Lorentzian fraction ms-TPE signal (a.u.) H:C Fractal dimension ms-TPE signal (a.u.) NaTURE CaTalysIs Articles evidence of its formation and role in forming the first C–C bond the autocatalytic HCP pathway . To gain further insight into this in the MTH reaction. Its detection is highly challenging due to its route, the following experiment was performed: after running the −1 high reactivity and the low activation barrier (≤17 kJ mol ) for its reaction under identical conditions to those reported in Fig. 3, the coordination with Brønsted acid sites, generating readily observ- flow of reactants was stopped, the reactor flushed with argon (Ar) 1,19,23,45 able surface-bound acetate or methyl acetate species . In addi- and the temperature increased while analysing the species evolved. tion, C H O has the same integer mass as C H , which convolutes C H and C H were observed throughout the tests in both the 2 2 3 6 2 4 6 6 its identification solely via m/z analysis. The detection of gas-phase MTH and MCTH processes, while C H was found to be almost 3 6 ketene strongly exemplifies the ability of operando PEPICO to dis- negligible (Supplementary Fig. 23). This is in line with the occur- criminate reactive intermediates, either formed or released in the rence of a second pathway to C H and C H that involves the crack- 2 4 6 6 gas phase, from strongly bound species detected using conventional ing of alkylated benzenes . spectroscopic approaches, which are more likely to be spectators in the reaction mechanism. Reproportionation of C H O with Coke generation and evolution. Further growth of the carbon 2 2 CH OH could yield CH CHO and further CH O, whose ioniza- skeleton leads to the formation of heavy carbonaceous species 3 3 2 tion transitions were detected via photoelectron analysis (Fig. 4a). and ultimately to their deposition as coke in the zeolite pore net- 1,10,29,47 CH CHO may undergo aldol condensation to form crotonalde- work . Understanding the generation and evolution of coke hyde (C H O; m/z 70), which has been discussed in the literature . is essential to design efficient and robust catalytic technologies for 4 6 The unambiguous assignment of the crotonaldehyde vibrational converting C platform molecules into fine chemicals and fuels. To fingerprints was not possible in this case, because it has the same achieve this, we have combined operando PEPICO with EPR analy- integer mass as C H , an abundant and isomer-complex C spe- ses, including continuous wave (CW), pulsed and two-dimensional 5 10 5 cies. Upon methylation and dehydration, this aldehyde may gen- hyperfine sublevel correlation (2D HYSCORE) measurements. erate 1,3-pentadiene (C H ; m/z 68) (Fig. 4a). Notably, CH O can By conducting operando PEPICO, a series of C hydrocarbons 5 8 2 10+ also react with CH to form CH CHO. This represents another were observed in both the MTH and MCTH processes at 0.5 bar 3 3 pathway to C–C bond formation, which is thermodynami- (Fig. 5a). These species remained virtually unobserved in the cally favoured compared with reported methane–formaldehyde low-pressure experiments, indicative of reduced chain-propagation (CH –CH O) mechanisms . and condensation-reaction rates. We identify two pathways to 4 2 In the second pathway, CH X can dissociatively adsorb to form naphthalene (C H ; m/z 128), which is considered to be the main 3 10 8 a CH radical and a coordinated leaving group, X* = Cl* or OH* precursor of heavier carbonaceous species (Fig. 5b). The first one (Fig. 4b). The creation of the first C–C bond then occurs via the involves a Diels–Alder dimerization of cyclopentadiene into dicy- reaction of two CH radicals to yield C H upon dehydrogena- clopentadiene (C H ; m/z 132), which isomerizes to 1,2,3,4- and 3 2 4 10 12 tion, analogous to the proposed direct coupling of surface methoxy 1,4,5,8-tetrahydronaphthalene, as observed in the MTH reaction. species to form this olefin . Chain growth proceeds via the addi- The latter species can undergo hydrogen abstraction to form dihy- tion of CH radicals, resulting in a distribution of C H , C H and dronaphthalene and ultimately naphthalene, as identified in both 3 3 6 4 8 C H isomers (Fig. 4b), explaining the complex ms-TPE spectra MTH and MCTH processes via ms-TPE spectrum analysis (Fig. 5b; 5 10 observed for C H and C H (Supplementary Fig. 19). Interestingly, Supplementary Fig. 22). This mechanism was first revealed in detail 4 8 5 10 the signals of isomerized and non-primary olefins were found to be in the catalytic pyrolysis of benzenediols over H-ZSM-5 (ref. ). In stronger, which is in line with their higher thermodynamic stability addition, dicyclopentadiene can hydrogenate and isomerize to ada- compared with primary olefins. CH is formed via the hydrogena- mantine, which can form diamantane, a known coke precursor . • • tion of CH , whereas the CH radical–radical reaction results in The second pathway encompasses the reaction of benzene with pro- 3 3 C H , both products of the MTH and MCTH processes. pyne (C H ; m/z 40) to yield indene (C H ; m/z 116). This agrees 2 6 3 4 9 8 The different C H isomers, which form either via the ketene- with previous studies on ethanol coupling over H-ZSM-5 that have 5 8 driven route in the MTH process or via hydrogen abstraction from shown the generation of indene-like species using solvent extrac- 39 • C H in the MTH and MCTH reactions, can undergo further cycli- tion . Further reaction with CH radicals generates naphthalene 5 10 3 zation/dehydrogenation and methylation to generate cyclopenta- (Fig. 5b; Supplementary Fig. 22), following a known mechanism . diene (C H ; m/z 66) and methyl cyclopentadiene (C H ; m/z 80), Interestingly, by comparing the ms-TPE spectra of m/z 132 photo- 5 6 6 8 respectively, in both reactions (Fig. 4c; Supplementary Fig. 20). The ions, the relative ratio of dicyclopentadiene and tetrahydronaph- detection of (poly)methylated cyclopentadienes corroborates stud- thalenes was lower in MCTH than in the MTH process (Fig. 5b; ies that have discussed their involvement in the MTH mechanism Supplementary Fig. 22). A possible explanation for this behaviour 1,46 and their potential role as further methylating agents . Presumably, is the observed favoured production of aromatics in the MCTH methyl-radical-driven and regular (that is, based on (poly)methyl- reaction that could promote the indene-driven mechanism. Once cyclopentadienes) methylation paths occur in parallel. Although naphthalene forms, it can undergo consecutive methylation to yield it is not possible to distinguish the relative contributions of the alkylated naphthalenes (Fig. 5a,b; Supplementary Fig. 22), which two routes, due to their high reactivity, it is likely that CH radical has been associated with the generation of heavy polyaromatic com- 26,49 species play an important role in the early stages of the reaction, pounds . This C -addition mechanism agrees with experimental 51,52 and that once HCP species form they react indiscriminately in all observations and kinetic models . When switching off the feed steps. Photoelectron analysis detects the dehydrogenation product of the MTH and MCTH reactions, the evolution of heavy species of methylcyclopentadienes (C H ), that is, fulvene (C H ; m/z 78), was also observed (Supplementary Fig. 23). In particular, photoions 6 8 6 6 which generates benzene, the first aromatic ring compound, in a associated with naphthalene were detected after both MTH and subsequent isomerization step (Fig. 4c; Supplementary Fig. 20). MCTH processes, while indene and anthracene could also be iden- Direct methylation of fulvene as well as dehydrogenation–meth- tified to be released after the MCTH reaction. This is in line with the ylation of C H yields methyl fulvene (C H ; m/z 92), the principal favoured generation of indene during the MCTH reaction as well as 6 8 7 8 precursor of toluene (Fig. 4c; Supplementary Fig. 20). This mecha- the higher rate of coke accumulation during the first 60 h on stream nism continues to generate other alkylated benzenes, such as iso- in the long-term tests. mers of xylene and of trimethylbenzene (Supplementary Fig. 21). The evolution of coke species was monitored by studying depos- Vibrational transitions in the ms-TPE spectra of m/z 106 could also ited paramagnetic carbonaceous species via EPR spectroscopy be assigned to ethylbenzene, which can undergo cracking to yield after selected times on stream in MTH and MCTH (Fig. 5c–h). By C H and C H , representing an additional source of olefins from conducting CW EPR, a signal with an isotropic g factor of 2.003 2 4 6 6 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 611 NaTURE CaTalysIs Articles was observed for all investigated samples (Supplementary Fig. 24). enabled us to quantitatively and isomer-selectively map the gen- By applying a reported method for spin counting , it was pos- eration of intermediates during the early stages of the MTH and sible to quantify the specific amount of radicals formed during MCTH processes. In addition, the strategy of jointly studying the the reaction by comparing the double integrals of the EPR signals activation of CH OH and CH Cl was crucial for decoupling and dis- 3 3 with a known standard. Accordingly, the average concentration of criminating the oxygenate- and hydrocarbon-driven mechanisms. EPR-active carbonaceous species increased as a function of the time These results provide experimental evidence for the generation of on stream, which is consistent with the coke contents determined CH radicals in both reactions and of C H O in the MTH process. 3 2 2 −3 via TGA, reaching approximately 4.5 and 7 μmol cm after 120 h in Accordingly, two main C–C bond formation and propagation path- the MTH and MCTH reactions, respectively (Fig. 5c). As detailed in ways are identified. The first, which is dominant in the MCTH pro- Supplementary Note 2, information on the density of these species cess, involves the reaction of two CH radicals to generate C H 3 2 4 was obtained by extracting the Gaussian and Lorentzian lineshape upon hydrogen transfer, which can undergo further methylation to contributions via least-squares fitting of the normalized CW EPR yield complex isomer mixtures of C –C hydrocarbons. The second 3 5 spectra (Fig. 5d, inset; Supplementary Figs. 24 and 25). Whereas occurs only in the MTH process and encompasses the CH OH- and isolated paramagnetic coke generally results in a Gaussian line- (CH ) O-driven formation of CH O, which, after oxidation into 3 2 2 shape, the magnetic interactions of closely packed spins typically CO, can undergo Koch carbonylation to generate C H O, represent- 2 2 yield a Lorentzian-type linewidth. Accordingly, the MCTH reaction ing the first C–C bond generation in the MTH process. This ketene resulted in the formation of highly dense carbonaceous compounds can reproportionate in the presence of CH OH to yield CH O and 3 2 within the first 5 h of reaction, whereas the density of these spe- CH CHO. The latter can undergo aldol condensation into crotonal- cies grew slowly in the MTH reaction (Fig. 5d), consistent with the dehyde, which, upon methylation and dehydration, generates linear higher tendency of CH Cl to generate BTX. C H . Interestingly, CH O can react with a CH radical to generate 3 5 8 2 3 Pulsed relaxation EPR measurements, that is, recording the sig- CH CHO, which is thermodynamically favoured compared with nal decay after microwave pulsing, were performed to further gain proposed CH –CH O routes and represents an additional subroute 4 2 insight into the spatial distribution of the detected paramagnetic for formation of the first C–C bond in the MTH reaction . Both species (Supplementary Fig. 26). Through least-squares fitting of envisioned pathways converge to generate linear and branched C these decay data with a stretched exponential, the corresponding hydrocarbons, such as C H and C H . Following a series of meth- 5 10 5 8 fractal dimension, N, can be extracted (Fig. 5e, inset). Generation ylation, cyclization and hydrogen-transfer reactions, these com- of three-dimensional (3D) deposits (N ≥ 2.8) was observed after pounds form fulvene and methyl fulvenes that isomerize to yield 5 h in the MCTH reaction, whereas N increased slowly in the MTH benzene and toluene, respectively. process, reaching 2.7 after 120 h on stream (Fig. 5e). These results Operando PEPICO spectroscopy also provided insights into the indicate that highly packed carbonaceous species form rapidly in mechanisms for the generation of coke precursors, such as naph- the MCTH process and equally along the three dimensions within thalene and its methylated analogues, in MTH and MCTH. The the zeolite pore network. By contrast, the MTH reaction gener- first involves Diels–Alder dimerization, isomerization and hydro- ates less-dense carbon deposits that grow more slowly, which is gen abstraction of cyclopentadiene, whereas the second encom- in line with the observations gathered using CW EPR, TGA and passes coupling between propyne and benzene to yield indene, sorption analysis. Finally, 2D HYSCORE analyses provide further which forms naphthalene upon methylation. These results were insights into the nature of these deposits (Fig. 5f,g; Supplementary further complemented with EPR measurements that shed light on Figs. 27–31). As detailed in Supplementary Note 2, comparison with the density, spatial distribution and representative molecular struc- density functional theory-based simulations and literature data ture of the carbonaceous species deposited in the zeolite micro- enables the assignment of representative molecular structures to the pores during the MTH and MCTH reactions. By combining CW recorded hyperfine couplings. Marked differences were observed and pulsed EPR with 2D HYSCORE measurements, it could be after 2 h on stream in the MTH and MCTH processes, which observed that the MCTH process results in the fast generation of showed the formation of ethylnaphthalene- and pentacene-like high-density, three-dimensional and low-alkylated carbon deposits, compounds, respectively (Fig. 5f,g). These results are in line with whereas carbonaceous species grow more slowly and form highly the PEPICO experiments performed by switching off the feed, in alkylated aromatics in the MTH process, particularly after a short which naphthalene in both MTH and MCTH together with indene time on stream. However, the degree of alkylation of these depos- and anthracene in MCTH were evolved from the H-ZSM-5 cata- its gradually decreased with time on stream in the MTH reaction lyst (Supplementary Fig. 23). Notably, ethylnaphthalene was pre- until it converged with that obtained in the MCTH reaction after viously observed in ethanol coupling over H-ZSM-5 , suggesting around 60 h. that the generation of coke precursors can be strongly influenced by Determining the different pathways of C–C bond formation and the zeolite framework. Estimation of the corresponding H:C ratio chain propagation that culminate in the generation of carbona- indicates that species with a higher degree of alkylation are formed ceous species has important practical implications for the design of in the MTH process (Fig. 5h). The presence of the ketene-driven catalytic and reactor systems for CH OH and CH Cl coupling, and 3 3 chain-growth route leads to the formation of linear compounds. should favour oxygenate-driven routes to hinder the formation of This, together with the slower initial rate of carbonaceous spe- dense polyaromatic species for developing efficient and robust cata- cies generation observed via CW EPR and TGA, can explain the lytic technologies. On the basis of the map of reactive intermediates reduced tendency of forming highly aromatic compounds in the obtained via the operando analysis in this work, we have proposed MTH process compared with the MCTH reaction. The nature of the a specific reactivity map for those intermediates; further studies will formed deposits converges with time on stream (Fig. 5h), and the undoubtedly provide additional insight that will help to refine fur- representative molecular structure corresponds to alkylated coro- ther our understanding of the actual reaction network. Going for- nenes and tribenzocoronenes after 60 and 120 h in both reactions ward, these techniques provide a valuable platform, potentially in (Supplementary Figs. 28 and 29). combination with co-feeding experiments, for gaining insights into other long-standing questions in hydrocarbon transformations, for Discussion example, to confirm the effects of promoters or reactant leaving The mechanism of C–C bond formation and chain propagation in groups, distinguishing mere spectators from real reaction interme- the zeolite-catalysed coupling of CH OH and CH Cl was assessed diates, and could be applied to virtually all hydrocarbon function- 3 3 using operando PEPICO spectroscopy, whose unique sensitivity alization processes. NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 612 NaTURE CaTalysIs Articles 12. Lin, R., Amrute, A. P. & Pérez-Ramírez, J. Halogen-mediated conversion of Methods hydrocarbons to commodities. Chem. Rev. 117, 4182–4247 (2017). Catalyst synthesis and basic characterization. A commercial ZSM-5 zeolite 13. Yarulina, I. et al. Structure–performance descriptors and the role of Lewis (CBV 8014 from Zeolyst International, nominal Si/Al = 40, ammonium form) was −1 acidity in the methanol-to-propylene process. Nat. Chem. 10, 804–812 (2018). calcined at 823 K (5 K min ) for 5 h to obtain the protonic form. The crystalline 14. Wu, X. et al. Direct mechanism of the first carbon–carbon bond formation order, porous and acidic properties, and the presence of coke in the fresh and in the methanol-to-hydrocarbons process. Angew. Chem. Int. Ed. 56, used catalysts were characterized using powder X-ray diffraction, gas sorption at 9039–9043 (2017). 77 K, Fourier transform infrared spectroscopy of adsorbed pyridine, and TGA, 15. Liu, Y. et al. Formation mechanism of the first carbon–carbon bond and the respectively. Full details are provided in the Supplementary Methods. first olefin in the methanol conversion into hydrocarbons. Angew. Chem. Int. Ed. 55, 5723–5726 (2016). Catalytic evaluation. Methanol and methyl chloride coupling reactions were 16. Comas-Vives, A., Valla, M., Coperet, C. & Sautet, P. Cooperativity between Al performed at ambient pressure using a continuous-flow fixed-bed reactor setup sites promotes hydrogen transfer and carbon–carbon bond formation upon (Supplementary Fig. 32). In a typical test, the zeolite catalyst (catalyst weight, dimethyl ether activation on alumina. ACS Cent. Sci. 1, 313–319 (2015). W = 0.6 g) was loaded in a quartz microreactor and pre-treated at 673 K for 1 h cat 17. Plessow, P. N. & Studt, F. Unraveling the mechanism of the initiation reaction before switching to the desired reaction temperature. Thereafter, a total gas flow, 3 −1 of the methanol to olefins process using ab initio and DFT calculations. F = 20 cm STP min , containing 50 vol% of CH X (X = OH or Cl) in helium, T 3 ACS Catal. 7, 7987–7994 (2017). was fed into the reactor at the desired reaction temperature (T = 523–723 K). 18. Wang, C.-M., Wang, Y.-D., Du, Y.-J., Yang, G. & Xie, Z.-K. Computational Quantification of the reactant conversion and product selectivity, as well as the insights into the reaction mechanism of methanol-to-olefins conversion in error in the carbon mass balance, was conducted according to the protocols H-ZSM-5: nature of hydrocarbon pool. Catal. Sci. Technol. 6, 3279–3288 (2016). described in the Supplementary Methods. 19. Liu, Y. et al. Critical role of formaldehyde during methanol conversion to hydrocarbons. Nat. Commun. 10, 1462 (2019). Operando photoelectron photoion coincidence spectroscopy. PEPICO 20. Wen, W. et al. Formation and fate of formaldehyde in methanol-to- experiments were performed using the CRF-PEPICO endstation at the vacuum hydrocarbon reaction: in situ synchrotron radiation photoionization mass ultraviolet beamline of the Swiss Light Source of the Paul Scherrer Institute, 55–57 spectrometry study. Angew. Chem. Int. Ed. 59, 4873–4878 (2020). Switzerland . The unique reactor configuration extends the lifetime of 21. Yang, L. et al. Role of acetaldehyde in the roadmap from initial carbon– elusive reaction intermediates, thanks to the high dilution of the feed, the low carbon bonds to hydrocarbons during methanol conversion. ACS Catal. 9, pressure inside the reactor (around 0.05–0.5 bar) and the free jet molecular beam 6491–6501 (2019). expansion, enabling their detection (Supplementary Fig. 33). Full details of the 22. Wang, C. et al. Extra-framework aluminum-assisted initial C–C bond protocol are provided in the Supplementary Methods. formation in methanol-to-olefins conversion on zeolite H-ZSM-5. Angew. Chem. Int. Ed. 57, 10197–10201 (2018). Electron paramagnetic resonance spectroscopy. The evolution of coke with the 23. Chowdhury, A. D. et al. Initial carbon–carbon bond formation during the time on stream was monitored using ex situ EPR spectroscopy. A set of experiments, early stages of the methanol-to-olefin process proven by zeolite-trapped including CW, pulsed and 2D HYSCORE measurements, was carried out to acetate and methyl acetate. Angew. Chem. Int. Ed. 55, 15840–15845 (2016). investigate the distinctive features of the deposited carbonaceous species, such as the 24. Hemelsoet, K., Van der Mynsbrugge, J., De Wispelaere, K., Waroquier, M. & density, spatial distribution and representative molecular structure (Supplementary Van Speybroeck, V. Unraveling the reaction mechanisms governing Fig. 34). Full details of the protocol are provided in the Supplementary Methods. methanol-to-olefins catalysis by theory and experiment. ChemPhysChem 14, 1526–1545 (2013). Data availability 25. Chowdhury, A. D. et al. Bridging the gap between the direct and hydrocarbon Data presented in the main figures of the manuscript are publicly available pool mechanisms of the methanol-to-hydrocarbons process. Angew. Chem. through the Zenodo repository (https://doi.org/10.5281/zenodo.6547649). Further Int. Ed. 57, 8095–8099 (2018). data supporting the findings of this study are available in the Supplementary 26. Lezcano-Gonzalez, I. et al. Insight into the effects of confined hydrocarbon Information. All other relevant source data are available from the corresponding species on the lifetime of methanol conversion catalysts. Nat. Mater. 19, author upon request. Source data are provided with this paper. 1081–1087 (2020). 27. Wang, C. et al. Methylbenzene hydrocarbon pool in methanol-to-olefins Received: 22 September 2021; Accepted: 18 May 2022; conversion over zeolite H-ZSM-5. J. Catal. 332, 127–137 (2015). Published online: 27 June 2022 28. Li, J. et al. A route to form initial hydrocarbon pool species in methanol conversion to olefins over zeolites. J. Catal. 317, 277–283 (2014). 29. Müller, S. et al. Coke formation and deactivation pathways on H-ZSM-5 in References the conversion of methanol to olefins. J. Catal. 325, 48–59 (2015). 1. Yarulina, I., Chowdhury, A. D., Meirer, F., Weckhuysen, B. M. & Gascon, J. 30. Wang, S. et al. Origin and evolution of the initial hydrocarbon pool Recent trends and fundamental insights in the methanol-to-hydrocarbons intermediates in the transition period for the conversion of methanol to process. Nat. Catal. 1, 398–411 (2018). olefins over H-ZSM-5 zeolite. J. Catal. 369, 382–395 (2019). 2. Van Speybroeck, V. et al. First principle chemical kinetics in zeolites: 31. Hu, M. et al. Establishing a link between the dual cycles in the methanol-to-olefin process as a case study. Chem. Soc. Rev. 43, methanol-to-olefins conversion on H-ZSM-5: aromatization of cycloalkenes. 7326–7357 (2014). ACS Catal. 10, 4299–4305 (2020). 3. Tian, P., Wei, Y., Ye, M. & Liu, Z. Methanol to olefins (MTO): from 32. Valecillos, J., Manzano, H., Aguayo, A. T., Bilbao, J. & Castaño, P. Kinetic and fundamentals to commercialization. ACS Catal. 5, 1922–1938 (2015). deactivation differences among methanol, dimethyl ether and chloromethane 4. Olivos-Suarez, A. I. et al. Strategies for the direct catalytic valorization of as stock for hydrocarbons. ChemCatChem 11, 5444–5456 (2019). methane using heterogeneous catalysis: challenges and opportunities. ACS 33. Paunović, V., Mitchell, S., Verel, R., Lee, S. S. & Pérez-Ramírez, J. Aluminum Catal. 6, 2965–2981 (2016). redistribution in ZSM-5 zeolite upon interaction with gaseous halogens 5. Čejka, J., Centi, G., Perez-Pariente, J. & Roth, W. J. Zeolite-based materials for and hydrogen halides and implications in catalysis. J. Phys. Chem. C 124, novel catalytic applications: opportunities, perspectives and open problems. 722–733 (2019). Catal. Today 179, 2–15 (2012). 34. Ibáñez, M. et al. Simultaneous coking and dealumination of zeolite H-ZSM-5 6. Zichittella, G. & Pérez-Ramírez, J. Status and prospects of the decentralised during the transformation of chloromethane into olefins. Catal. Sci. Technol. valorisation of natural gas into energy and energy carriers. Chem. Soc. Rev. 6, 296–306 (2016). 50, 2984–3012 (2021). 35. Zichittella, G., Hemberger, P., Holzmeier, F., Bodi, A. & Perez-Ramirez, J. 7. Zhong, J. et al. State of the art and perspectives in heterogeneous catalysis of Operando photoelectron photoion coincidence spectroscopy unravels CO hydrogenation to methanol. Chem. Soc. Rev. 49, 1385–1413 (2020). mechanistic fingerprints of propane activation by catalytic oxyhalogenation. 8. Ilias, S. & Bhan, A. Mechanism of the catalytic conversion of methanol to J. Phys. Chem. Lett. 11, 856–863 (2020). hydrocarbons. ACS Catal. 3, 18–31 (2012). 36. Hemberger, P., Custodis, V. B. F., Bodi, A., Gerber, T. & van Bokhoven, J. A. 9. Olsbye, U. et al. Conversion of methanol to hydrocarbons: how zeolite cavity Understanding the mechanism of catalytic fast pyrolysis by unveiling reactive and pore size controls product selectivity. Angew. Chem. Int. Ed. 51, intermediates in heterogeneous catalysis. Nat. Commun. 8, 15946 (2017). 5810–5831 (2012). 37. Hemberger, P., van Bokhoven, J. A., Pérez-Ramírez, J. & Bodi, A. New 10. Olsbye, U. et al. The formation and degradation of active species during analytical tools for advanced mechanistic studies in catalysis: photoionization methanol conversion over protonated zeotype catalysts. Chem. Soc. Rev. 44, and photoelectron photoion coincidence spectroscopy. Catal. Sci. Technol. 10, 7155–7176 (2015). 1975–1990 (2020). 11. Olsbye, U. et al. Methane conversion to light olefins—how does the methyl 38. Brückner, A. In situ electron paramagnetic resonance: a unique tool for halide route differ from the methanol-to-olefins (MTO) route? Catal. Today analyzing structure–reactivity relationships in heterogeneous catalysis. Chem. 171, 211–220 (2011). Soc. Rev. 39, 4673–4684 (2010). NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal 613 NaTURE CaTalysIs Articles 39. Madeira, F. F. et al. Radical species detection and their nature evolution with 56. Bodi, A., Sztáray, B., Baer, T., Johnson, M. & Gerber, T. Data acquisition catalyst deactivation in the ethanol-to-hydrocarbon reaction over HZSM-5 schemes for continuous two-particle time-of-flight coincidence experiments. zeolite. ACS Catal. 1, 417–424 (2011). Rev. Sci. Instrum. 78, 084102 (2007). 40. Martinez-Espin, J. S. et al. New insights into catalyst deactivation and 57. Bodi, A. et al. Imaging photoelectron photoion coincidence spectroscopy with product distribution of zeolites in the methanol-to-hydrocarbons (MTH) velocity focusing electron optics. Rev. Sci. Instrum. 80, 034101 (2009). reaction with methanol and dimethyl ether feeds. Catal. Sci. Technol. 7, 2700–2716 (2017). Acknowledgements 41. Foley, B. L., Johnson, B. A. & Bhan, A. A method for assessing catalyst This publication was created as part of NCCR Catalysis (grant 180544), a National deactivation: a case study on methanol-to-hydrocarbons conversion. ACS Centre of Competence in Research funded by the Swiss National Science Foundation. Catal. 9, 7065–7072 (2019). The operando PEPICO measurements were performed at the VUV beamline of the Swiss 42. Palumbo, L. et al. Conversion of methanol to hydrocarbons: spectroscopic Light Source (SLS) located at the Paul Scherrer Institute (PSI), Villigen, Switzerland. characterization of carbonaceous species formed over H-ZSM-5. J. Phys. Chem. C 112, 9710–9716 (2008). 43. Zang, Z., Zhang, W., Huang, J., Feng, P. & Ding, J. First molecule with Author contributions carbon–carbon bond in methanol-to-olefins process. Chem. Phys. Lett. 737, J.P.-R. conceived the research. G.Z., S.M. and J.P.-R. coordinated the study. G.Z., 136844 (2019). A.C., S.M. and J.P.-R. wrote the article with input from all other co-authors. A.C., 44. Plessow, P. N., Smith, A., Tischer, S. & Studt, F. Identification of the reaction S.M. and G.Z. developed the methodology to quantitatively compare the coupling sequence of the MTO initiation mechanism using ab initio-based kinetics. reactions, and A.C. conducted basic characterization and testing of the catalysts. P.H. J. Am. Chem. Soc. 141, 5908–5915 (2019). and A.B. developed the operando PEPICO technique, and G.Z. and P.H. conducted 45. Chowdhury, A. D. & Gascon, J. The curious case of ketene in zeolite the measurements and analysed the data. A.C. assisted with the measurements. chemistry and catalysis. Angew. Chem. Int. Ed. 57, 14982–14985 (2018). Z.P. conducted the Franck–Condon simulations. G.J. and M.A. developed the EPR 46. Zhang, W. et al. Methylcyclopentenyl cations linking initial stage and spectroscopy analysis, and M.A. and S.P.S. conducted the experiments. highly efficient stage in methanol-to-hydrocarbon process. ACS Catal. 10, 4510–4516 (2020). Competing interests 47. Bleken, F. L. et al. Catalyst deactivation by coke formation in microporous The authors declare no competing interests. and desilicated zeolite H-ZSM-5 during the conversion of methanol-to-hydrocarbons. J. Catal. 307, 62–73 (2013). 48. Pan, Z., Puente-Urbina, A., Bodi, A., van Bokhoven, J. A. & Hemberger, P. Additional information Isomer-dependent catalytic pyrolysis mechanism of the lignin model Supplementary information The online version contains supplementary material compounds catechol, resorcinol and hydroquinone. Chem. Sci. 12, available at https://doi.org/10.1038/s41929-022-00808-0. 3161–3169 (2021). Correspondence and requests for materials should be addressed to 49. Goetze, J. et al. Insights into the activity and deactivation of the Guido Zichittella, Patrick Hemberger or Javier Pérez-Ramírez. methanol-to-olefins process over different small-pore zeolites as studied with operando UV-vis spectroscopy. ACS Catal. 7, 4033–4046 (2017). Peer review information Nature Catalysis thanks the anonymous reviewers for their 50. Mebel, A. M., Landera, A. & Kaiser, R. I. Formation mechanisms of contribution to the peer review of this work. naphthalene and indene: from the interstellar medium to combustion flames. Reprints and permissions information is available at www.nature.com/reprints. J. Phys. Chem. A 121, 901–926 (2017). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 51. Rojo-Gama, D. et al. Deactivation of zeolite catalyst H-ZSM-5 during published maps and institutional affiliations. conversion of methanol to gasoline: operando time- and space-resolved X-ray diffraction. J. Phys. Chem. Lett. 9, 1324–1328 (2018). Open Access This article is licensed under a Creative Commons 52. Janssens, T. V. W., Svelle, S. & Olsbye, U. Kinetic modeling of deactivation Attribution 4.0 International License, which permits use, sharing, adap- profiles in the methanol-to-hydrocarbons (MTH) reaction: a combined tation, distribution and reproduction in any medium or format, as long autocatalytic–hydrocarbon pool approach. J. Catal. 308, 122–130 (2013). as you give appropriate credit to the original author(s) and the source, provide a link to 53. Zichittella, G., Polyhach, Y., Tschaggelar, R., Jeschke, G. & Pérez-Ramírez, J. the Creative Commons license, and indicate if changes were made. The images or other Quantification of redox sites during catalytic propane oxychlorination by third party material in this article are included in the article’s Creative Commons license, operando EPR spectroscopy. Angew. Chem. Int. Ed. 60, 3596–3602 (2021). unless indicated otherwise in a credit line to the material. If material is not included in 54. Ben Tayeb, K. et al. The radical internal coke structure as a fingerprint of the the article’s Creative Commons license and your intended use is not permitted by statu- zeolite framework. Microporous Mesoporous Mater. 289, 109617 (2019). tory regulation or exceeds the permitted use, you will need to obtain permission directly 55. Sztáray, B. et al. CRF-PEPICO: double velocity map imaging photoelectron from the copyright holder. To view a copy of this license, visit http://creativecommons. photoion coincidence spectroscopy for reaction kinetics studies. J. Chem. org/licenses/by/4.0/. Phys. 147, 013944 (2017). © The Author(s) 2022 NAtuRE C At Aly SiS | VOL 5 | JuLy 2022 | 605–614 | www.nature.com/natcatal

Journal

Nature Catalysis – Springer Journals

Published: Jul 1, 2022

There are no references for this article.