Probing the dynamics of methanol in copper-loaded zeolites via quasi-elastic and inelastic neutron scattering

Abstract

The dynamics of methanol within prototype methanol synthesis catalysts were studied using quasi-elastic neutron scattering. Three Cu-exchanged zeolites (mordenite, SSZ-13 and ZSM-5) were studied after methanol loading and showed jump diffusion coefficients between 1.04 × 10⁻¹⁰ and 2.59 × 10⁻¹⁰ m²·s^–1. Non-Arrhenius behavior was observed with varying temperature due to methoxy formation at Brønsted acid sites and methanol clustering around copper cations.

Graphical abstract

Keywords

quasielastic neutron scattering / inelastic neutron scattering / methanol / diffusion / zeolites

1 Introduction

The search for efficient and sustainable catalytic processes has steadily intensified over the last few decades, largely driven by the growing demand for environmentally friendly energy sources and chemical transformations, with zeolites among the most widely and intensely studied of inorganic materials [1]. These aluminosilicates, primarily composed of Al, Si and O atoms, offer exceptional catalytic and molecular sieving properties, conferred by their microporous structure with inherently high surface area, shape selectivity and solid acid functionality [2]. From the outset of synthetic zeolite discovery, a number of these materials—zeolites A, X and Y—were ion-exchanged with transition metal ions in an attempt to alter their adsorption properties and introduce new catalytic functionality [3]. Copper-exchanged zeolites, in particular, have been under intense scrutiny due to their auspicious application in remediation of NO_x and N₂O from combustion engines and the mild partial-oxidation of unreactive hydrocarbons, such as the direct methane-to-methanol (MTM) conversion. The principal challenge in this latter application lies in the formation of highly-reactive oxygen species adept of activating the strong C–H bond in methane (104 kcal·mol^–1), while simultaneously preventing the thermodynamically favorable complete oxidation which yields carbon dioxide and water [4]. Unsurprisingly, this challenge has already been resolved by nature to a level scientists can only currently aspire to: metalloenzymes (methane monooxygenase) found in methanotrophic bacteria are able to activate and convert methane to methanol on iron and copper active centers at ambient conditions. Zeolites’ ability to stabilize metal-oxygen complexes within their microporous structure, akin to those found in metalloenzymes, thus show a promising path toward an industrially viable, direct MTM conversion operation [5]. Indeed, zeolites mordenite, SSZ-13 and ZSM-5 (framework types MOR, CHA and MFI, respectively) have been consistently shown to convert methane to methanol with exceptional selectivity; albeit at present with conversions below 1% [6]. From elucidating the exact structure of the active sites effecting this conversion to understanding the mechanisms involved, there are many hurdles to overcome in order to optimise and commercialise the MTM conversion process.

One of the key aspects is understanding the dynamic processes occurring at the interface between the catalyst and the reactant, transient and product species. The rate-limiting step may involve the ingress of reactants toward the active site; adsorption on the active site; conversion at the active site; or the egress of products away from the active site [7]. As such, the dynamics of guest molecules adsorbed in the zeolitic framework are fundamental to the catalytic properties of a given material. Studying such dynamics, however, is often complex due to the superposition of disparate motions such as rotations and translations. Macroscopic methods such as, inter alia, gravimetry [8] and membrane permeation [9] probe the mass transfer in the presence of a concentration gradient; while microscopic methods such as pulse field gradient nuclear magnetic resonance [10] and quasi-elastic neutron scattering (QENS) [11] probe the stochastic motions of molecules, enabling the study of molecular self-diffusivity in the absence of a concentration gradient. Neutron scattering is a particularly useful technique for probing the structural and dynamic properties of materials as the neutron’s wavelength and energy are comparable to interatomic distances and kinetics, respectively. Moreover, the exceptionally large incoherent scattering cross section of hydrogen, compared to that of other elements, renders neutrons an especially useful probe when studying dynamics involving protons [12].

Thus, we present herein an investigation of the dynamics of methanol—the product of MTM conversion—adsorbed onto Cu-exchanged mordenite, SSZ-13 and ZSM-5 utilizing a combination of QENS and inelastic neutron scattering (INS) measurements. Methanol dynamics have been previously reported within H-ZSM-5 [13] after correlating the findings to previous studies, the authors extrapolated that an increase in Si/Al ratio results in an increase in the fraction of mobile atoms. The aims of this study were to provide an insight into the behavior of methanol within Cu-zeolites in relation to the process of methanol desorption after partial oxidation of methane; and to deduce the impact of copper, present within the framework, on the dynamics of methanol within these materials by drawing comparisons to other studies of similar systems.

2 Experimental

2.1 Zeolite preparation

Commercial grade, ammonium forms of mordenite and ZSM-5 (Si/Al = 10 and 15, respectively) were purchased from Alfa Aesar; protonic SSZ-13 (Si/Al = 12) was generously supplied by Johnson Matthey. The ammonium precursors of mordenite and ZSM-5 were calcined in a muffle furnace, under air atmosphere, by heating to 823 K (1 K·min^–1), holding for 8 h and cooling to room temperature to yield H-MOR and H-ZSM-5. Copper ions were subsequently introduced into the framework via liquid ion-exchange by placing each zeolite into 0.1 mol·L^–1 Cu(NO₃)₂ solution and stirring for 24 h at 298 K. The zeolites were then washed with deionised water and resuspended into a fresh Cu(NO₃)₂ solution. The process was repeated three times to maximise copper uptake. As a result, SSZ-13, MOR and ZSM-5 contained 3.53%, 2.37% and 1.38% copper (w/w), respectively, determined via X-ray fluorescence analysis (Malvern Analytical Epsilon 3), equating to ~1 Cu per unit cell (PUC) in each zeolite.

2.2 Methanol dosing

The zeolites were dried and activated by heating in a furnace, under pure oxygen flow, to 723 K (10 K·min^–1); held for 1 h; then cooled to room temperature. Methanol dosing was carried out by flowing pure argon through a bubbler containing HPLC-grade methanol for 30 min. The QENS samples were dosed at 298 K and the INS samples at 350 K to match an intermediate temperature studied with QENS. The pretreatments were carried out in sealable, flow-through cans [14] and all flow rates were maintained at 200 sccm. The samples were then transferred to an argon glovebox and were subsequently sealed in aluminum cans using indium wire gaskets. Methanol uptake was determined via thermogravimetric analysis and the loadings were approximately 9, 7 and 13 molecules per unit cell (MPUC) in Cu-Mordenite, Cu-SSZ13 and Cu-ZSM-5, respectively.

2.3 Neutron scattering

2.3.1 QENS

The QENS measurements were carried out on the cold-neutron backscattering spectrometer, Emu [15], at the Australian Nuclear Science and Technology Organisation. The (111) reflection from the Si analyzer was used, providing a resolution of 1.0 μeV and covering a Q-range of 0.1–2.0 Å^–1. Data obtained were re-binned in Q to provide a satisfactory signal to noise ratio for fitting. This decreased the limits accessible from the theoretical maximum range to 0.2–1.8 Å^–1. The instrumental resolution function was obtained from a vanadium standard measurement and the methanol-dosed zeolites were measured at 300, 325, 350 and 375 K. The experimental spectra obtained were fitted using the instrumental resolution convoluted with combinations of a delta function to represent elastic scattering from immobile atoms and Lorentzian functions to represent the Doppler broadening of the elastic line which occurs during scattering from moving particles. Data reduction, including can subtraction and analysis were carried out using Mantid [16] and DAVE [17].

2.3.2 INS

INS measurements were carried out on the indirect geometry spectrometer TOSCA [18] at the ISIS Pulsed Neutron and Muon Source, UK. The samples were cooled below 20 K prior to spectra collection in order to reduce thermal motions and background signals.

3 Results and discussion

The spectra collected at all four temperatures (300, 325, 350, and 375 K) contained both elastic and quasielastic components. All of the spectra were adequately fitted using the instrumental resolution convoluted with a delta and a single Lorentzian function to represent the elastic and quasielastic components, respectively. A subset of the data are shown in Fig.1.

Fig.1 Fits applied to the data from (a, d, g) CuMOR, (b, e, h) CuSSZ13, and (c, f, i) CuZSM5 at 375 K for Q values of (a,b,c) 0.4, (d, e, f) 1.0, and (g, h, i) 1.8 Å–1.

Full size|PPT slide

A strong elastic component in all spectra originates from the relatively static atoms in the zeolite framework. These scatter coherently and produce a low-resolution background that contains strong Bragg scattering. Unfortunately it was not possible to obtain a data set of the empty zeolites for subtraction due to time and experimental constraints during the Covid 19 pandemic. Separation of the incoherent and coherent elastic scattering could not be carried out, so analysis of the elastic incoherent structure factor is precluded.

Incoherent inelastic scattering will be dominated by the hydrogen atoms due to its high incoherent scattering cross section. Motions of the methanol were represented within the experimental time window of the instrument as displayed by the presence of a Lorentzian. The delta function represents elastic scattering of both coherent (zeolite) and incoherent (immobile H) origin. The linewidths (Γ) of the Lorentzian component were modeled as a function of Q² in order to deduce the type of motion under observation. These showed nonlinear Q-dependence at all temperatures, corresponding to translational jump—rather than bulk-diffusion. A number of models were applied, with the Singwi & Sjölander (SS) jump diffusion model [19] providing the best fit for the data obtained at all temperatures (Fig.2). This model postulates that a molecule oscillates for a mean time, ${\tau }_0$, at a local site before undergoing an instantaneous jump to an adjacent site. The SS model has a form:

Fig.2 Lorentzian linewidths of the quasielastic component from the QENS spectra and the SS jump diffusion model applied. (a), (b), (c) and (d) correspond to data obtained at 300, 325, 350, and 375 K respectively.

Full size|PPT slide

${\Gamma }_{\mathrm{Q}}=\frac{1}{{\tau }_0}\left[1-\left(\frac{\exp -\left(2W\right)}{1+D{Q}^2{\tau }_0}\right)\right],$

$2W=D{Q}^2{\tau }_0\frac{⟨R^2⟩}{⟨l^2⟩}.$

where $⟨R^2⟩$ is the mean square radius of the thermal cloud developed during the oscillatory motions and $⟨l^2⟩$ is the mean square jump length. D represents the self-diffusion coefficient, i.e., the diffusion occurring without a concentration gradient. It is also commonly denoted as D_s.

The derived diffusion coefficients (D_s), jump lengths and residence times are presented in Tab.1. The calculated jump lengths are smaller than the pores of each framework which suggests that the diffusion observed is intra- rather than inter-crystalline. The D_s obtained from Cu-SSZ-13 are considerably higher ((2.25–2.59) × 10⁻¹⁰ m²·s^–1), across the whole temperature range, than those found in Cu-mordenite ((1.04–1.84) × 10⁻¹⁰ m²·s^–1) or Cu-ZSM-5 ((1.22–1.87) × 10⁻¹⁰ m²·s^–1). This is likely due to the presence of the fewest adsorption sites (Brønsted acid sites (BAS) and Cu) at ~3 PUC compared to ~4 and 6 PUC in mordenite and ZSM-5, (as determind by Si/Al ratio) respectively. Indeed, at 300 K the magnitude of the D_s seems to correspond well to the adsorption site density; i.e., the higher the number of acid sites, the lower the diffusivity due to increased interaction between the methanol and the BAS (H-bonding), as well as Cu within the pores. At 325 K, the D_s of methanol expectedly increases further within mordenite and ZSM-5 while, counterintuitively, decreasing within zeolite SSZ-13. Further increasing the temperature to 350 K appears to have little effect on methanol mobility within Cu-mordenite, a slight reduction in the residence time between jumps within Cu-SSZ13 (80.83 to 72.83 ps) and an appreciable increase in the jump length is observed within Cu-ZSM-5 (2.71 to 3.07 Å).

Tab.1 Jump lengths, residence times and diffusion coefficients derived from QENS

Sample	T/K	$\sqrt{⟨l^2⟩}$/Å	${\tau }_0$/ps	Ds/( × 10−10 m2·s–1)
CuMOR	300	2.0 ± 0.3	47.8 ± 11.0	1.3 ± 0.5
CuMOR	325	2.9 ± 0.4	77.2 ± 13.2	1.8 ± 0.6
CuMOR	350	2.9 ± 0.4	76.4 ± 12.2	1.8 ± 0.6
CuMOR	375	1.8 ± 0.2	51.3 ± 10.3	1.0 ± 0.4
CuSSZ13	300	3.7 ± 0.3	90.2 ± 8.9	2.6 ± 0.5
CuSSZ13	325	3.3 ± 0.3	80.8 ± 7.2	2.2 ± 0.4
CuSSZ13	350	3.3 ± 0.2	72.8 ± 6.2	2.5 ± 0.4
CuSSZ13	375	2.7 ± 0.2	53.2 ± 5.0	2.3 ± 0.4
CuZSM5	300	2.4 ± 0.2	77.5 ± 10.3	1.2 ± 0.3
CuZSM5	325	2.7 ± 0.2	81.5 ± 9.1	1.5 ± 0.3
CuZSM5	350	3.1 ± 0.3	83.8 ± 8.4	1.9 ± 0.4
CuZSM5	375	2.4 ± 0.2	59.6 ± 6.3	1.6 ± 0.4

We note the largest changes in methanol diffusivity, in terms of jump lengths, residence times and the D_s, as we reach the highest temperature of 375 K. All three samples displayed a reduction in residence time of ~20 ps and a decrease in jump length of up to 1.11 Å (in CuMOR). Important to note are the inconsistencies between our observations and the various reports on diffusion in zeolites, as we discern a marked variation in calculated jump lengths across the temperature range studied and the overall methanol diffusivity does not appear to scale with temperature. Indeed, neither the D_s nor the residence times display Arrhenius behavior when plotted as a function of 1/T. As the data was well-fitted using a single Lorentzian function convoluted with a delta and the instrumental resolution, we postulate that these discrepancies are caused by at least two dynamic processes occurring within the experimental time-window of the instrument, on very similar time and length scales.

We believe that these two motions differentiate the ‘free’ and the ‘bound’ or ‘activated’ methanol formed at higher temperatures. One possibility is the dissociation of methanol adsorbed on the BAS via H-bonding and the subsequent formation of surface methoxy groups which has been documented within H-ZSM-5 at room temperature [20], although this assignment has been rejected by Zachariou et al., requiring heating to 423 K [21]. Hence, we subjected the three methanol-dosed zeolites to INS, presented in Fig.3. The weak, broad band at ~800 cm^–1 is assigned to H-bonded hydroxyl bending vibrations of methanol. Due to its breadth however, hydrogen bonded water formed in the methoxylation could also be contributing. The sharp bands at 1155 and 1450 cm^–1 are attributed to the CH₃ rocking and bending modes, respectively; there is a broad band that overlaps the latter with a maximum at 1580 cm^–1. This is assigned to methyl vibration modes and low-frequency external modes of methanol molecules; and the broad band at 3200 cm^–1 is a combination of CH and OH stretches in methanol, respectively [20,22]. This is poorly resolved due to instrumental constraints. The presence of the bands in the methyl rocking region and the hydroxyl and methyl deformation region thus confirm the presence of methanol, but not methoxylation; i.e., it is not possible to conclude from the INS alone that methanol has dissociated. Although the sorption temperature of 375 K is lower than the 425 K reported by Zachariou et al. as needed for reaction, the presence of copper in our materials may act as a catalyst to depress this temperature. Since the quasielastic spectra suggest a combination of motions that vary with temperature, this indicates that methoxy formation on the BAS is likely. Our observations, therefore, likely correspond to diffusion of methanol clustered at the cationic sites and the dynamics of the methoxy groups formed, both of which are occurring at similar rates, making their individual delineation challenging.

Fig.3 INS spectra of methanol adsorbed on the three zeolites at 350 K.

Full size|PPT slide

4 Conclusions

Methanol dynamics within three Cu-loaded zeolites have been studied using QENS, providing diffusion constants consistent with literature [23−25], but exhibiting non-Arrhenius behavior and variability in calculated jump-lengths, which are indicative of multiple concurrent processes. We confirmed, via INS, the presence of undissociated methanol and these spectra are compatible with some degree of methoxy group formation, which is likely one of the processes influencing our observations. The second process is attributed to the clustering of undissociated methanol at the Cu adsorption sites. These findings have been enabled by the exceptional resolution of the Emu [15] spectrometer which permits the study of much slower processes occurring on the molecular time and length scales. The identified complexities underscore the intricate nature of methanol diffusion within Cu-zeolites, necessitating continued experimentation and further refinement of models to enable a comprehensive understanding of diffusion in zeolitic materials.

References

[1] Van Speybroeck V , Hemelsoet K , Joos L , Waroquier M , Bell R G , Catlow C R A . Advances in theory and their application within the field of zeolite chemistry. Chemical Society Reviews, 2015, 44(20): 7044–7111

[2] Mitra S , Sharma V K , Mukhopadhyay R . Diffusion of confined fluids in microporous zeolites and clay materials. Reports on Progress in Physics, 2021, 84(6): 066501

[3] Vanelderen P , Vancauwenbergh J , Sels B F , Schoonheydt R A . Coordination chemistry and reactivity of copper in zeolites. Coordination Chemistry Reviews, 2013, 257(2): 483–494

[4] Snyder B E R , Bols M L , Schoonheydt R A , Sels B F , Solomon E I . Iron and copper active sites in zeolites and their correlation to metalloenzymes. Chemical Reviews, 2018, 118(5): 2718–2768

[5] Colby J , Stirling D I , Dalton H . The soluble methane mono oxygenase of Methylococcus capsulatus (Bath). Its ability to oxygenate n alkanes, n alkenes, ethers, and alicyclic, aromatic and heterocyclic compounds. Biochemical Journal, 1977, 165(2): 395–402

[6] Park M B , Park E D , Ahn W S . Recent progress in direct conversion of methane to methanol over copper-exchanged zeolites. Frontiers in Chemistry, 2019, 7: 514

[7] O’Malley A J , Catlow C R A . Sorbate dynamics in zeolite catalysts. Experimental Methods in the Physical Sciences, 2017, 49: 349–401

[8] Lin D H , Ducarme V , Coudurier G , Vedrine J C . Adsorption and diffusion of different hydrocarbons in MFI zeolite of varying crystallite sizes. Studies in Surface Science and Catalysis, 1989, 46: 615–623

[9] Millot B , Méthivier A , Jobic H , Moueddeb H , Dalmon J A . Permeation of linear and branched alkanes in ZSM-5 supported membranes. Microporous and Mesoporous Materials, 2000, 38(1): 85–95

[10] Heink W , Kärger J , Pfeifer H , Datema K P , Nowak A K . High-temperature pulsed field gradient nuclear magnetic resonance self-diffusion measurements of n-alkanes in MFI-type zeolites. Journal of the Chemical Society, Faraday Transactions, 1992, 88(23): 3505–3509

[11] BéeM. Quasielastic Neutron Scattering: Principles and Applications in Solid State Chemistry, Biology and Materials Science. Bristol: Adam Hilger, 1988

[12] Mukhopadhyay R , Mitra S . Molecular diffusion and confinement effect: neutron scattering study. Indian Journal of Pure and Applied Physics, 2006, 44: 732–740

[13] Matam S K , Catlow C R A , Silverwood I P , O’Malley A J . Methanol dynamics in H-ZSM-5 with Si/Al ratio of 25: a quasi-elastic neutron scattering (QENS) study. Topics in Catalysis, 2021, 64(9-12): 699–706

[14] Silverwood I P , Hamilton N G , McFarlane A , Ormerod R M , Guidi T , Bones J , Dudman M P , Goodway C M , Kibble M , Parker S F . . Experimental arrangements suitable for the acquisition of inelastic neutron scattering spectra of heterogeneous catalysts. Review of Scientific Instruments, 2011, 82(3): 034101

[15] de Souza N R , Klapproth A , Iles G N . EMU: high-resolution backscattering spectrometer at ANSTO. Neutron News, 2016, 27(2): 20–21

[16] Arnold O , Bilheux J C , Borreguero J M , Buts A , Campbell S I , Chapon L , Doucet M , Draper N , Ferraz Leal R , Gigg M A . . Mantid—Data analysis and visualization package for neutron scattering and μSR experiments. Nuclear Instruments & Methods in Physics Research. Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, 2014, 764: 156–166

[17] Azuah R T , Kneller L R , Qiu Y , Tregenna-Piggott P L W , Brown C M , Copley J R D , Dimeo R M . DAVE: a comprehensive software suite for the reduction, visualization, and analysis of low energy neutron spectroscopic data. Journal of Research of the National Institute of Standards and Technology, 2009, 114(6): 341–358

[18] Parker S F , Fernandez-Alonso F , Ramirez-Cuesta A J , Tomkinson J , Rudic S , Pinna R S , Gorini G , Fernández Castañon J . Recent and future developments on TOSCA at ISIS. Journal of Physics: Conference Series, 2014, 554: 012003

[19] Singwi K S , Sjölander A . Resonance absorption of nuclear gamma rays and the dynamics of atomic motions. Physical Review, 1960, 120(4): 1093–1102

[20] O’Malley A J , Parker S F , Chutia A , Farrow M R , Silverwood I P , García-Sakai V , Catlow C R A . Room temperature methoxylation in zeolites: insight into a key step of the methanol-to-hydrocarbons process. Chemical Communications, 2016, 52(14): 2897–2900

[21] Zachariou A , Hawkins A , Parker S F , Lennon D , Howe R F . Neutron spectroscopy studies of methanol to hydrocarbons catalysis over ZSM-5. Catalysis Today, 2021, 368: 20–27

[22] Schenkel R , Jentys A , Parker S F , Lercher J A . Investigation of the adsorption of methanol on alkali metal cation exchanged zeolite X by inelastic neutron scattering. Journal of Physical Chemistry B, 2004, 108(23): 7902–7910

[23] Plant D F , Maurin G , Bell R G . Molecular dynamics simulation of methanol in zeolite NaY. Studies in Surface Science and Catalysis, 2005, 158: 963–970

[24] Woodward C L M , Porter A J , Morton K S C , O’Malley A J . Methanol diffusion in H-ZSM-5 catalysts as a function of loading and Si/Al ratio: a classical molecular dynamics study. Catalysis Communications, 2022, 164: 106415

[25] Botchway C H , Tia R , Adei E , O’Malley A J , Dzade N Y , Hernandez-Tamargo C , De Leeuw N H . Influence of topology and brønsted acid site presence on methanol diffusion in zeolites beta and MFI. Catalysts, 2020, 10(11): 1342

[26] GibsonESkukasukasVSilverwoodI P. INS vibrational spectroscopy of surface species relevant to the partial oxidation of methane. STFC ISIS Neutron and Muon Source, 2022

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

We acknowledge the support of the Australian Centre for Neutron Scattering, ANSTO and the Australian Government through the National Collaborative Research Infrastructure Strategy, in supporting the neutron research infrastructure used in this work via ACNS proposal P13488. The ISIS neutron and Muon source also awarded a grant of beamtime (RB2220457). This is publicly available at the ISIS data archive [26]. We also wish to thank Johnson Matthey for their gift of the zeolite SSZ-13 sample.

RIGHTS & PERMISSIONS

2024 Higher Education Press