A combined computational and experimental study of methane activation during oxidative coupling of methane (OCM) by surface metal oxide catalysts

Abstract

The experimentally validated computational models developed herein, for the first time, show that Mn-promotion does not enhance the activity of the surface Na₂WO₄ catalytic active sites for CH₄ heterolytic dissociation during OCM. Contrary to previous understanding, it is demonstrated that Mn-promotion poisons the surface WO₄ catalytic active sites resulting in surface WO₅ sites with retarded kinetics for C–H scission. On the other hand, dimeric Mn₂O₅ surface sites, identified and studied via ab initio molecular dynamics and thermodynamics, were found to be more efficient in activating CH₄ than the poisoned surface WO₅ sites or the original WO₄ sites. However, the surface reaction intermediates formed from CH₄ activation over the Mn₂O₅ surface sites are more stable than those formed over the Na₂WO₄ surface sites. The higher stability of the surface intermediates makes their desorption unfavorable, increasing the likelihood of over-oxidation to CO_x, in agreement with the experimental findings in the literature on Mn-promoted catalysts. Consequently, the Mn-promoter does not appear to have an essential positive role in synergistically tuning the structure of the Na₂WO₄ surface sites towards CH₄ activation but can yield MnO_x surface sites that activate CH₄ faster than Na₂WO₄ surface sites, but unselectively.

---

1. Introduction

Methane (CH₄) the main component of natural gas, is a sustainable energy vector and an emerging building block for synthesis of chemicals and fuels.^1–5 Direct CH₄ conversion to ethylene (C₂H₄) via oxidative coupling of methane (OCM) remains an active research frontier with ongoing studies in both catalysis science and reaction engineering.^6–9 Despite continued interest, the lack of the fundamental understanding of the structure–function relationships in the state-of-the-art OCM catalysts has prohibited industrial-scale application of OCM.^2,10,11 Recent work on selective OCM catalyst development chiefly focuses on the combinatorial screening of multicomponent mixed-phase catalyst systems as well as scrutinizing developed catalyst libraries using data science methods.^7,12–14 The structure–activity relationships of any catalyst, however, are the foundation for rational design and optimization of heterogeneous catalysts.^15–17 Under typical OCM reaction conditions (CH₄ + O₂ and high temperatures of up to 850 °C), the catalytic active sites are not mere static geometric forms obtained on calcined supports but undergo dynamic rearrangement dictated by thermodynamics.^18,19 Therefore, well-defined catalysts under reaction conditions should be investigated in order to understand the fundamental OCM catalyst structure–activity relationships.

Dispersed phase catalysts offer key advantages over catalysts containing a mixture of dispersed and crystalline phases,^16,20–23 since dispersed phase catalysts consist of atomically dispersed sites with well-defined molecular and electronic structure that can be tuned via the addition of promoters. Such control of the nature and the structure of catalytic active sites remains challenging in the presence of a crystalline phase in supported mixed phase catalysts, which exhibit an ensemble distribution of the catalytic active sites distributed in the various phases, hence complicating the structure–activity analysis. This difference in dispersed vs. mixed phase catalysts is illustrated in Fig. 1 using silica (SiO₂) supported metal oxide (MO_x) catalysts, such as those used for OCM, with well-defined active sites found in dispersed phase catalysts allowing systematic analysis for structure–activity relationships, and supported mixed oxide phase catalysts resulting in an variety of various types of sites.

Fig. 1 The nature and type of surface sites on SiO₂ support, dispersed and mixed-phase supported tungsten oxide catalysts used in OCM. The support contains surface hydroxyls (–OH) where the active metal oxides (MO_x such as WO_x, MnO_x) anchor in dispersed phase catalysts. For the supported mixed-oxide phase catalysts, the dispersed phase metal oxide anchors onto the surface –OH sites of the support and the crystalline phase is also present resulting in a wide range of metal oxide sites, along with those present at the interface of the crystal-support.

The structure–activity relationships for the state-of-the-art supported Mn–Na₂WO₄/SiO₂ tri-metal oxide OCM catalyst have remained elusive not only due to the demanding reaction conditions, but also due to the mixed-oxide phase nature of this catalyst.²⁴ While dispersed phase tri-metal oxide catalysts, i.e., containing all three metal oxides (W, Na, Mn), have not been studied to date, the structure–activity relationships of active sites in dispersed phase bi-metal oxide catalysts (Na–WO_x/SiO₂, Mn–WO_x/SiO₂) for OCM have been recently reported.^22,25 It was shown that the distorted surface WO₄ sites are present at elevated temperatures (673 K) on the amorphous SiO₂ surface in unpromoted, supported WO_x/SiO₂ catalysts.²² Upon Na-promotion, the extent of distortion of the WO₄ surface site diminishes.^22,25 The dispersed-phase Na-coordinated WO₄ surface sites (Na–WO₄) with Na/W molar ratio less than 2 were shown to be significantly more selective to C₂ products than the unpromoted, surface WO₄ sites and the corresponding mixed-oxide phase catalysts containing crystalline phase Na₂WO₄ (Na/W ∼ 2).^22,25 Moreover, the promotion with Mn i.e Mn–WO_x/SiO₂, did not have a significant impact on the molecular structure and OCM C₂ selectivity of the WO₄ surface sites in comparison to unpromoted WO₄ surface sites.²⁵ Thus, the role of Mn for the OCM catalytic reaction remains unclear.

Older studies of mixed-phase tri-metal oxide catalysts under/near OCM reaction conditions revealed that Mn³⁺ reduces to Mn²⁺ during CH₄ activation, which can then be re-oxidized to Mn³⁺ during the second half of the OCM reaction redox cycle.²⁶ Based on the proposed Mn³⁺ → Mn²⁺ → Mn³⁺ redox cycle, multiple reaction models suggesting Mn sites as the catalytic active site during OCM have been put forth in the literature.^26–29 Note that the observed redox cycle, however, was purely based on transitions observed in the crystalline phases of manganese oxide (Mn₂O₃ and Mn₃O₄) and the redox behavior of the dispersed phase manganese oxide surface sites (MnO_x) in the catalyst are still not known. More recently, an operando synchrotron μ-XRF/XRD/absorption-computed tomography investigation of the mixed-phase tri-metal oxide catalyst found that during phase transformations occurring in the crystalline manganese oxide phases (Mn₂O₃, Mn₇SiO₁₂ and MnWO₄), product distribution did not change. Hence, it was concluded that such manganese oxide phases did not have a critical role in the catalytic OCM reaction.³⁰ It was further concluded that dispersed phase MnO_x surface sites present on the SiO₂ support in the mixed-phase tri-metal oxide catalyst, which bulk X-ray based techniques such as XRD, XAS, XRF cannot effectively monitor, may be involved in the OCM catalytic cycle.³⁰

Herein, we provide the first in silico framework validated with in situ spectroscopy measurements to study the structure–activity relationship of metal oxide sites in model tri-metal oxide OCM catalysts containing Mn, W, Na metal oxides on a SiO₂ support. In particular, the nature and structure of the tri-metal oxide surface sites (Mn–Na₂WO₄/SiO₂) was elucidated and contrasted with the bi-metal oxide surface sites (Na₂WO₄) via ab initio molecular dynamics (AIMD) and periodic density functional theory (DFT). Na₂WO₄ surface sites are used as the base for this comparison as these surface sites were shown to be the selective active sites for CH₄ activation during OCM.^19,22,25 The DFT predicted molecular and electronic structures of the dispersed phase surface sites in bi- and tri-metal oxide OCM catalyst prior to CH₄ activation were also validated by in situ Raman and in situ UV-Vis diffused reflectance (UV-VisDR) spectroscopies at dehydrated oxidative conditions (400 °C, 10% O₂/inert), respectively. Intrinsic kinetics of CH₄ dissociation over surface sites in dispersed phase bi-metal oxide and tri-metal oxide case were calculated and contrasted. Results herein provide molecular-level insights on the nature of surface sites, and their activity towards heterolytic CH₄ dissociation during OCM, providing structure–function relationships for this OCM catalyst.

2. Experimental

(a) Computational modelling

The periodic calculations have been performed with the Vienna Ab Initio Simulation package (VASP) using the Perdew and Wang (PW91) generalized gradient approximation exchange–correlation functional. PW91 exchange–correlation functional is used as model used herein are adapted from prior models of supported WO_x catalysts published in the literature.^25,31,32 The valence electrons are treated explicitly and their interactions with the ionic cores are described by the Projector Augmented-Wave method (PAW), which allows using low energy cut off equal to 400 eV for the plane-wave basis. A sampling of the Brillouin zone with 2 × 2 × 1 mesh was applied for all models in this work. The positions of the two top-most layer surface slab atoms, as well as the active (un)coordinated WO_x site, in the supercell, were relaxed until the total energy differences decrease below 10⁻⁶ eV (forces acting on atoms fall below 0.005 eV Å⁻¹). The details of the model of the hydroxylated β-cristobalite slab can be found elsewhere.²⁵

Vibrational modes have been calculated for the selected surface species within the harmonic approximation. Only the tungsten active center and its 1st and 2nd neighbors (O–Si and OH groups) were considered in the Hessian matrix. This matrix was computed by the finite difference method followed by a diagonalization procedure. The eigenvalues of the resulting matrix led to frequency values. The assignment of the vibrational modes was done by inspection of the corresponding eigenvectors. Static (NSW = 0, IBRION = −1) self-consistent calculations were performed to compute DOS (ISMEAR = 0, Sigma = 0.02), and the resulting vasprun.xml files were visualized using P4Vasp^33,34 to extract local and total density of states (LDOS, total DOS).

Ab initio molecular dynamics (AIMD) calculations were performed using Anderson thermostat (1700 K, 2 fs steps) to provide sufficient kinetic energy to the simulated catalyst surface for the atoms to rearrange so the global minima for the total energy can be calculated. The antiferromagnetic configuration in dimeric Mn₂O_x cases was consistently lower in energy than ferromagnetic configurations, so all dimeric Mn₂O_x studied here were treated as antiferromagnetic. Typically, ∼8000 fs AIMD simulations resulted in 3–4 lowest energy structures, which were then extracted and further optimized by periodic DFT before subsequent calculations were conducted for frequencies, Bader³⁵ charges and transition states (from nudged elastic band (NEB)^36,37) on the lowest energy structure amongst the 3–4 selected from AIMD. Ab initio thermodynamics (AITD) were calculated according to the framework described in the recent literature.^13,38–40 Specifically, the relative stability of various possible surface Mn_aO_b clusters were computed in reference to the bulk α-Mn₂O₃ phase (unit cell: 32Mn atoms, 48O atoms),⁴¹ which is typically reported to be stable under OCM relevant conditions (T = 1000 K, P_O2 = 0.33–0.1).⁴² The Mn₂O₃ atom positions and cell parameters were relaxed with high precision and cutoff energy of 520 eV to minimize any Pulay stress. Given the unit cell of α-Mn₂O₃ containing 32Mn and 48O atoms, the general governing stoichiometric equation can be written as (1)

with the overall exergonicity or endergonicity of the reaction to form isolated Mn_aO_b clusters, rather than bulk Mn₂O₃, on Na₂WO₄/SiO₂ described via(2)

In this method, the PV contributions of solids were neglected,^13,38–40 and the Gibbs free energies of solids were approximated as their respective electronic energies computed by DFT with thermal corrections in accordance with the methodology described in the literature,⁴³ as shown in eqn (3)

μ(solid) = E^DFT_0K(solid) + G_vib(T). (3)

The chemical potential of the gas phase O₂ depends on the temperature (T) and the corresponding partial pressure (P). At arbitrary T and P, μ_O2 can then be written as:

here, R is the gas constant, 300 K is the reference T, and 1 atm (101 325 Pa; P/P° = 1) is the reference pressure. Δμ(T, P°) were taken from JANAF tables.⁴⁴ was the DFT calculated energy of −9.81 eV and value is 0.098 eV.⁴⁵μ_O (i.e. ½μ_O2) values calculated as a function of T and P are tabulated and plotted in Table S1 and Fig. S1,† respectively, and agree with those reported in the literature.^46,47 Thus, Gibbs free energy of formation (G_form) for a specific case can be written as eqn (2), where μ(Na₂WO₄/SiO₂), μ(Mn_aO_b + Na₂WO₄/SiO₂), and μ(Mn₂O₃) are chemical potentials of the solid phases calculated according to eqn (3), while μ_O is the oxygen chemical potential calculated according to eqn (4).

For example, using the general eqn (1), the Mn₁O₁–Na₂WO₄/SiO₂ system was calculated as:

with

Using eqn (1), the comprehensive Mn_aO_b chemical potential landscape was computed for various mono-atomic (Mn, MnO, MnO₂, MnO₃) and di-atomic (Mn₂O, Mn₂O₂, Mn₂O₃, Mn₂O₄, Mn₂O₅ compositions to compare their respective thermodynamic stability given OCM relevant conditions of 1000 K and P_O2/P° of 0.1–0.33. A higher degree of oligomerization for surface Mn_aO_b compositions was not considered due to the complexity of the possible structures in tri- and tetra-atomic cases such as straight 2D-chain vs. 3D-clusters.

Lastly, intrinsic rate constant of C–H dissociation in CH₄ was calculated using the DFT-computed E₀ values of transition states identified via NEB method, according to equation:

where k_B is the Boltzmann constant (J K⁻¹), h is the Planck's constant (J s), T is the reaction temperature (K), R is the gas constant (J K⁻¹ mol⁻¹), and E_a is the energy difference with E₀ values of initial and transition state structures in J mol⁻¹.

(b) Non-stoichiometric catalyst synthesis

The SiO₂ support (Cabot CAB-O-SIL® EH5 with a surface area of ∼332 m² g⁻¹) was first treated with water, then allowed to dry overnight at room temperature before final calcination at 500 °C for 4 hours under flowing air. This treatment increases the density and surface hydroxyls of the SiO₂ support. The dried SiO₂ obtained after calcination was then crushed into a fine powder. The resulting pore volume of the SiO₂ powder was determined to be ∼0.8 mL g⁻¹ and was utilized for all catalyst preparation steps via incipient-wetness impregnation (IWI) of the metal oxide aqueous solutions unless mentioned otherwise. A NaOH aqueous solution corresponding to the pore-volume equivalent of ∼0.8 mL g⁻¹ was impregnated onto the SiO₂ support and the sample was initially dried overnight, then at 120 °C in an oven under flowing air for 2 hours, and finally calcined at 700 °C under airflow for 2 hours. The resultant supported Na/SiO₂ sample, with pore volume equivalent of ∼0.7 mL g⁻¹, was subsequently impregnated with the desired aqueous concentration of W in the form of AMT ((NH₄)_xW₁₂O₂₈; Alfa Aesar, #44792) and Mn in the form of Mn(NO₃)₂ (Alfa Aesar, #10806-09). The resultant solid was dried overnight, then at 120 °C for 2 hours in airflow and finally calcined at 500 °C for 4 hours under flowing air. The final catalysts were denoted as 8%W/0.4%Na/SiO₂ (Na/W = 0.4)²² in the bi-metal oxide case and 0.5%Mn–8%W/0.4%Na/SiO₂ (Na/W = 0.4, Mn/W = 0.2) in the tri-metal oxide case. Samples calcined at 500 °C calcination did not undergo sintering typical of higher temperature calcination, enabling easier characterization of the surface metal oxide sites, which becomes challenging in low surface area sintered materials.

(c) In situ Raman spectroscopy

The in situ Raman spectra of the Na coordinated WO_x/SiO₂ supported catalysts were obtained with Horiba–Jobin Yvon LabRam HR instrument equipped with three laser excitations (532, 442, and 325 nm) and liquid N₂-cooled CCD detector (Horiba–Jobin Yvon CCD-3000 V). The 442 nm laser was chosen for spectral accumulation since it minimizes sample fluorescence from the SiO₂ supported catalysts. The wavenumber calibration was checked using a standard silicon wafer with a Raman vibration at 520.7 cm⁻¹. A confocal microscope with a 50× objective (Olympus BX-30-LWD) was utilized for focusing the laser on the catalysts. Typically, the spectra were collected for 60 s per scan for a total of three scans with a 1000 μm hole with a spectral resolution of ∼1 cm⁻¹. Approximately 15–20 mg of each catalyst in powder form (100–150 μm size range) was loaded into an environmental cell (Linkam CCR-1000) with a quartz window with O-ring seals, which was kept cool by circulating cooling water. The in situ Raman spectra of the catalysts were collected at 400 °C after dehydration in 10% O₂/Ar (∼30 cc min⁻¹) for 60 min to elucidate the initial molecular structure of the representative bi- and tri-metal oxide surface sites to validate the DFT-optimized structures before reaction with CH₄.

(d) In situ UV-Vis DR spectroscopy

The in situ UV-Vis spectra of the catalysts were obtained using a Varian Cary 5E UV-Vis-NIR spectrophotometer with a Harrick Praying Mantis accessory. Approximately 15–20 mg of each catalyst in powder form was loaded into an in situ Harrick HVCDR2 environmental cell. The UV-Vis spectra of the catalyst samples were collected at 400 °C in the 200–800 nm wavelength range after dehydration (10% O₂/Ar, ∼30 cc min⁻¹) for 60 min, using a scan rate of 15 nm min⁻¹ and a signal averaging time of 0.6 s. MgO was used as a standard for obtaining the background absorbance and was subtracted from the sample absorbance. Kubelka–Munk function F(R_∞) was calculated from the background-subtracted absorbance data of the UV-Vis spectrum of each sample. The edge energy (E_g), or bandgap, was determined by finding the intercept of the straight line for the low-energy rise of a plot of [F(R_∞)hν]²versus hν, where hν is the incident photon energy. The in situ UV-Vis spectra of the catalysts at 400 °C after dehydration in 10% O₂/Ar were used to elucidate the initial electronic structure of the representative bi- and tri-metal oxide surface sites to validate the DFT-optimized structures before reaction with CH₄.

3. Results and discussion

(a) Thermodynamic stability of Mn_aO_b–Na₂WO₄/SiO₂ catalysts under OCM relevant conditions

The Gibbs free energies of formation (ΔG(T, P) further referred to as ΔG_form) of Mn_aO_b surface sites in Na₂WO₄/SiO₂ catalysts were computed as a function of oxygen partial pressure according to eqn (2) with respect to the bulk Mn₂O₃ phase. In an oxygen containing reaction mixture, the surface Mn_aO_b sites are expected to be oxidized depending on the partial pressure of the oxygen. Therefore, as a starting point, ΔG_form was computed for four possible mono-atomic Mn₁O_b cases Mn, MnO, MnO₂, and MnO₃, where b is 0, 1, 2 and 3, respectively. As shown in the heat maps in Fig. 2a–d, the redder tones signify positive values of ΔG_form and bluer tones signify negative ΔG_form. Positive ΔG_form imply that formation of that cluster is endergonic with respect to bulk Mn₂O₃, while negative values imply that the formation of that cluster is exergonic with respect to bulk Mn₂O₃. Fig. 2a–d show that the formation of isolated Mn, MnO, MnO₃ surface sites are clearly endergonic in comparison to bulk Mn₂O₃ under OCM conditions (∼973–1173 K, P_O2/P = 0.20–0.33). However, the isolated MnO₂ formation is barely endergonic relative to bulk Mn₂O₃ under OCM conditions and in fact becomes exergonic at high P_O2 and low temperatures. Likewise, five stoichiometries possible considered for the di-atomic Mn₂O_b are Mn₂O, Mn₂O₂, Mn₂O₃, Mn₂O₄, and Mn₂O₅ where b is 1, 2, 3, 4, and 5 respectively. The computed ΔG_form for each case are shown as heat maps in Fig. 2e–i. At OCM conditions, besides Mn₂O₅, the formation of di-atomic sites (Mn₂O, Mn₂O₂, Mn₂O₃, Mn₂O₄) is endergonic in reference to the bulk Mn₂O₃ and thus unfavorable. Therefore, under OCM conditions, formation of Mn₂O₅ sites yielding Mn₂O₅–Na₂WO₄/SiO₂ catalyst the most thermodynamically favorable from an AITD standpoint.

Fig. 2 Individual heat maps of computed Gibbs free energy of formation for (Mn_aO_b–Na₂WO₄)/SiO₂ as a function of T (300–1300 K) and P_O2/P = 0–1 (ref. 10) for (a) Mn, (b) MnO, (c) MnO₂, (d) MnO₃, (e) Mn₂O, (f) Mn₂O₂, (g) Mn₂O₃, (h) Mn₂O₄, (i) Mn₂O₅. The dashed rectangle in (i) signifies the OCM relevant conditions.

In a comparison shown in Fig. 3, the entire catalyst structure space of Mn_aO_b–Na₂WO₄/SiO₂ was compared to ascertain the most thermodynamically stable stoichiometry of surface Mn_aO_b sites in Mn_aO_b–Na₂WO₄/SiO₂ catalysts under OCM relevant P_O2/P° of 0.2–0.33, and temperature of 1000 K. Under these conditions, the formation of Mn₂O₅ dimeric sites is evidently thermodynamically favorable (i.e. most exergonic) with reference to bulk Mn₂O₃, especially towards P_O2/P° values ∼0.30–0.33 that represent the OCM reaction stoichiometry i.e. CH₄ : O₂ = 2 : 1. For instance, as the P_O2/P° values increase from 0.25 to 0.35, the ΔG_form of Mn₂O₅ sites decreases from −0.002 eV (−0.20 kJ mol⁻¹) to −0.031 eV (−3.00 kJ mol⁻¹). Note that since the Gibbs energy difference is so small between MnO₂ and Mn₂O₅, a minor presence of MnO₂ sites cannot be precluded under OCM conditions in the catalyst. However, for simplicity, the Mn₂O₅–Na₂WO₄/SiO₂ catalyst was chosen for further study presented in the next sections. Specifically, the molecular and electronic structure of the surface sites in Mn₂O₅–Na₂WO₄/SiO₂ catalyst are calculated and compared against the base case of unpromoted Na₂WO₄/SiO₂ to ascertain the effect of Mn promotion. The base case i.e. Na₂WO₄/SiO₂ contains Na-coordinated WO₄ surface sites that have been shown to be the selective active sites for OCM in such bi-metal oxide catalysts.^22,25 Moreover, the molecular and electronic structures of DFT-optimized structures are also validated against experimental in situ characterization data to ensure agreement. Lastly, CH₄ activation pathways and intrinsic kinetics are also compared in each case to elucidate the role of surface Mn sites in OCM.

Fig. 3 Global comparison of Gibbs free energy of formation of (Mn_aO_b–Na₂WO₄)/SiO₂ at OCM relevant conditions (P_O2/P° = 0.20–0.33 at 1000 K) identifying Mn₂O₅–Na₂WO₄/SiO₂ as the most thermodynamically stable structure under these conditions.

(b) Molecular structure of surface sites in Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂

The DFT-optimized molecular structure of surface sites in Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂ are shown in Fig. 4a. In the case of Na₂WO₄/SiO₂, bi-grafted, dioxo WO₄ surface site is present in a distorted T_d geometry, in agreement with experimental and computational work previously reported in the literature.^22,25 On the other hand, the molecular structure of Mn₂O₅–Na₂WO₄/SiO₂ contains surface WO_x site that changes dramatically in terms of its molecular structure due to interaction with the surface Mn₂O₅ dimer. In this case, the bi-grafted, di-oxo WO_x is coordinated to the adjacent Mn₂O₅ site via a single W–O–Mn bond, and so can be regarded as WO₅. The Mn₂O₅ is present as a dimer bridged via a single Mn–O–Mn bond with both MnO₄ units present as distorted pseudo tetrahedral (T_d).

Fig. 4 (a) Periodic DFT optimized structures minimum energy structures for Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂ catalysts. Comparison of experimental in situ dehydrated Raman (400 °C, 10% O₂/N₂, 532 nm) and DFT calculated frequencies from 600–1200 cm⁻¹ for (b) Na₂WO₄/SiO₂ and (c) Mn₂O₅–Na₂WO₄/SiO₂ OCM catalysts. The vibrations from SiO₂ support are not shown/labelled herein for clarity. The presence of poorly crystalline Mn–WO₃ and MnWO₄ nanoparticles is inferred as well from the experimental in situ Raman spectrum in (c).

Fundamental frequencies for each case were computed and contrasted with experimental in situ Raman spectra of representative dehydrated catalysts to validate the computational model against experimentally observed structures, as shown in Fig. 4b and c. In each case, the computed fundamental modes are assigned in the color-coded tables in Fig. 4b and c and the solid bars are the DFT-calculated frequencies. In both cases, general agreement is observed between computed frequencies and experimentally measured Raman bands, validating the DFT-optimized structural models. From this comparison, it is inferred that W=O bonds are the shortest, given they vibrate in the 945–1000 cm⁻¹ range. Likewise, support-grafting bonds like W–O–Si and Mn–O–Si are significantly longer, given they vibrate in the 800–900 cm⁻¹ range. Lastly, bridging bonds like Mn–O–Mn and Mn–O–W are the longest (weakest), since they vibrate at 616 and 761 cm⁻¹, respectively. It should be noted that we infer a small presence of poorly crystalline nanoclusters/nanoparticles of MnWO₄ and Mn-contaminated WO₃ in the experimental Raman spectrum of the tri-metal oxide catalyst shown in Fig. 4c, which exhibit Raman bands in the same region as dispersed phase sites. However, since operando/in situ characterization studies in the literature suggest that crystalline MnWO₄ is not critical to OCM,^29,30 and because the nature (size, shape, surface structure, defect density) of these nanoparticles is not known presently, the nanoparticles have been excluded from this computational model and only the dispersed phase surface sites were studied for structure–function relationships towards CH₄ activation. Interested readers are directed to a recent, fully experimental study, where further detailed experimental characterization including UV-VisDRS, Raman at 400 °C, Raman during OCM at 900 °C, TPSR, etc. of several model OCM catalysts including the two catalysts compared in Fig. 4b and c can be found.⁴⁸

(c) Electronic structure of surface sites in Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂

Next, the electronic structure of the DFT-optimized models is studied. The calculated Bader charges as well as the bond lengths are shown in Fig. 5a and b for Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂ cases, respectively. In both cases, the W centers have an identical Bader charge of +2.51, which corresponds to an oxidation state of +6 according to the calibration curves shown in Fig. S2,† generated according to literature reported methodology.⁴⁹ On the other hand, the calculated Bader charges for the two Mn atoms in Mn₂O₅–Na₂WO₄/SiO₂ catalyst were +1.66 and +1.63, respectively, corresponding to an oxidation state of ∼+3 (Fig. S2†), which agrees well with the literature reported Mn oxidation state of +3 under oxidative conditions (i.e. no CH₄ at elevated temperatures).²⁶ Experimental in situ UV-Vis DRS was conducted on two representative samples, namely bi-metal oxide Na–WO_x/SiO₂ with Na/W = 0.4, and tri-metal oxide Mn–Na–WO_x/SiO₂ with Na/W = 0.4, Mn/W = 0.5 as shown in Fig. 5d. In both cases, ligand to metal charge transfer energies (∼240 nm or 5.1 eV) indicate that W centers were indeed fully oxidized, i.e. +6, in agreement with the DFT results.^22,50,51 The edge energy (E_g), indicative of the degree of polymerization of surface sites was ∼4.2 eV for Na–WO_x/SiO₂ catalyst, but decreased to 4.0 eV upon Mn-promotion of Na–WO_x sites, suggesting Mn-promotion increases polymerization in surface sites. The oxidation state of Mn atoms, on the other hand, cannot be determined from UV-vis DRS as UV-Vis DRS due to the low concentration of Mn in the Mn–Na–WO_x/SiO₂ representative sample and the lower extinction coefficient of MnO_xvs. WO_x, the LMCT and d–d bands from Mn centers are masked by the absorbance of W sites.

Fig. 5 Structural visualization of (a) Na₂WO₄/SiO₂ and (b) Mn₂O₅–Na₂WO₄/SiO₂ with calculated oxidation states mentioned next to the transition metal atoms in corresponding colors (Mn: red, W: pink). Oxidation states were calculated using a calibrated-fit for Bader charges, shown in Fig. S2 in the ESI section.† Calculated Bader charges for the surface sites in each case are summarized in table (c). (d) In situ dehydrated UV-Vis DRS of Na–WO_x/SiO₂(red) and Mn–Na–WO_x/SiO₂(green) catalysts, with E_g values and LMCT positions marked.

The E_g value after Mn-promotion of Na–WO_x sites is lower than the E_g value in the Na–WO_x case (Fig. 5d), which can be further understood via density of states (DOS) analysis of DFT-optimized structures. Total DOS plots are shown in Fig. 6a and b for Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂, respectively. For the Na₂WO₄/SiO₂ case, the valence band is primarily composed of occupied O 2p and W 5d states, while the conductance band of empty W 5d states. The energy gap between the top of the valence edge and the bottom of the conduction edge is ∼1.7 eV. However, in the Mn₂O₅–Na₂WO₄/SiO₂ case, the valence and conductance bands are primarily composed of Mn 3d states, with some contribution from occupied O 2p states in the valence band. In this case, the lowest part of the conductance band lowers due to low-lying unoccupied Mn 3d states, reducing the bandgap to ∼0 eV. The DOS analysis shows that Mn-addition fundamentally changes the electronic structure of the surface sites by introducing low-lying vacant Mn 3d states above the Fermi level, which will have strong implications for the catalytic CH₄ activation. Elsewhere, it has been found that low-lying empty d states in a transition metal like Pt can act an acceptor for σ-donation from the C–H bond of CH₄ and lead to facile C–H scission.⁵² Thus, it suffices to conclude at this point that experimental in situ UV-Vis DRS and computational DOS analysis are in qualitative agreement evidencing that Mn-promotion of Na₂WO₄ surface sites will lower the E_g values due to the introduction of low-lying empty Mn 3d states above the Fermi level, that Mn is present in its +3 oxidation state, and that W is present in its fully oxidized +6 state.

Fig. 6 Total electronic density of states (total DOS) and calculated band gap values (eV) for (a) Na₂WO₄/SiO₂ and (b) Mn₂O₅–Na₂WO₄/SiO₂. E_F values for Na₂WO₄/SiO₂ and Mn₂O₅–Na₂WO₄/SiO₂ were −3.08 eV and −2.75 eV, respectively. The positive x-axis values correspond to spin-up, while the negative values correspond to spin-down states.

(d) Reaction kinetics of CH₄ dissociation over Mn₂O₅–Na₂WO₄/SiO₂vs. Na₂WO₄/SiO₂

AITD analysis identified Mn₂O₅–Na₂WO₄ as the most stable species under OCM relevant temperature and oxygen partial pressure. Therefore, the intrinsic activity of Na₂WO₄/SiO₂vs. Mn₂O₅–Na₂WO₄/SiO₂ surface sites for CH₄ dissociation were studied, where the C–H bond dissociation is the most kinetically relevant elementary step in the OCM mechanism.^19,53 To enhance readability and comprehension, the various possible reaction pathways for heterolytic C–H scission of CH₄ are grouped based on the type of surface intermediate formed in the forward step. Specifically,

CH₄ + MO_x → M–O–CH₃ (methoxy) + M–OH (hydroxy) (8)

CH₄ + MO_x → M–O–CH₃ + M–H (hydride) (9)

CH₄ + MO_x → M–CH₃ (methyl) + M–OH (10)

where MO_x represents a reactive metal oxide surface site (Mn₂O₅ or WO₄). In each case, the surface intermediate pair formed from CH₄ dissociation is unique, i.e. methoxy–hydroxy (OCH₃, OH) in the red pathway, methoxy–hydride (OCH₃, H) in the blue pathway and methyl–hydroxy (CH₃, OH) in the green pathway, as shown in Fig. 7 and 9.

Fig. 7 (a) Reaction coordinate vs. ΔE_DFT (eV) and (b) calculated intrinsic rate constants (log-scale) for CH₄ dissociation for the three possible pathways over the Na₂WO₄/SiO₂ catalyst. Arrow indicates the dominant (fastest) pathway. (c) Structural depictions of the initial, TS and final states during CH₄ dissociation over Na₂WO₄/SiO₂ catalysts yielding methyl-hydroxy (CH₃, OH) intermediates. Pictorial summary for other pathways can be found in Table S3.†

CH₄ dissociation over Na₂WO₄ surface sites was studied as a function of the possible pathways listed above, and as a function of various active oxygen atoms. The results are summarized in Fig. 7a–c, highlighting the activation barriers associated with various possible pathways over the Na₂WO₄/SiO₂ catalyst. According to the computed energy barriers summarized in Fig. 7a, the highest energy barrier was for the pathway yielding OCH₃, OH surface intermediates (3.4 eV), indicating that CH₄ dissociation turnover frequency for this pathway will be lower than other pathways with smaller transition state barriers. The two pathways with OCH₃, H intermediates formation differ in terms of the oxygen atom of the WO₄ where C–H scission occurs, and consequently have slightly varying TS barriers (1.6, 1.9 eV). Lastly, the pathway yielding CH₃, OH surface intermediates exhibited the lowest energy barrier of 1.1 eV, suggesting that this is the dominant reaction pathway for CH₄ dissociation over Na-coordinated WO₄ surface sites in the Na₂WO₄/SiO₂ catalysts. The resulting intrinsic rate constants, calculated according to eqn (7) are plotted in Fig. 7b, suggesting that the intrinsic rate constant for the CH_3, OH pathway is 2–3 orders of magnitude higher than other pathways making this the dominant pathway. Lastly, the initial, transition state (TS) and final structures for CH₄ activation over Na₂WO₄ surface sites via the dominant reaction pathway are shown in Fig. 7c, showing the formation of W–CH₃ and W–O–H intermediates. Importantly, these results show that Na₂WO₄ surface sites can effectively dissociate CH₄ in the absence of Mn-promoter in agreement with a recently published report,²² as opposed to Lunsford¹⁹ reaction model where Na–O–Mn sites are assumed to be the critical active sites for CH₄ dissociation.

Local DOS (LDOS) analysis was conducted to elucidate the changes in electronic structure of initial versus the TS structure during CH₄ dissociation over the Na₂WO₄ surface sites. LDOS for W 5d states and O 2p states of the initial Na₂WO₄ structure are shown in Fig. 8a. The valence band (below the Fermi level, E_F) is comprised largely of occupied 2p states from the double-bonded oxygen i.e. W=O, while the conductance band is comprised largely of unoccupied W 5d states with a minor contribution from unoccupied O 2p states of W=O. Upon CH₄ dissociation, the lowest part of the conductance band lowers in energy as the unoccupied O 2p states get filled. Moreover, significant overlap of occupied states is observed for W 5d and C 2p states near the top of the valence edge at ca. −1.7 eV, signifying W–C bond formation as in the W–CH₃ surface intermediate. Lastly, occupied H 1s states and occupied O 2p states also overlap in the lower part of the valence edge, confirming O–H bond formation i.e. W–O–H intermediate. The Bader charge for W barely decreases; it changes from +2.51 in initial structure to +2.48 in the TS structure, indicating minimal reduction of the W center. The partial reduction of W center instead of a full reduction to +5 or to +4 might be the reason that reduced W centers could not be observed during OCM in SQUID-EPR experimentally.²⁶ Interestingly, a similar CH₄ dissociation pathway yielding CH₃, OH surface intermediates was also proposed for La₂O₃-based catalysts, where La–CH₃ and La–O–H intermediates formed upon heterolytic dissociation of CH₄.⁵⁴ Likewise, CH₄ dissociation over Al₂O₃ was also shown to proceed via Al–CH₃ and Al–O–H formation.⁵⁵

Fig. 8 LDOS analysis of (a) initial and (b) TS structures evidencing W–CH₃ and W–O–H formation, suggesting that CH₄ dissociation is mediated by a TS where a methyl is coordinated to the W center, while H is abstracted by W=O to form W–O–H. The inset shows the zoomed-in structure of the Na₂WO₄ surface site with the pertinent atoms labeled. E_F for the initial structure was −3.08 eV and for the TS was −2.90 eV. Colored atoms correspond to the corresponding color in DOS plots.

Next, intrinsic kinetics of CH₄ dissociation over Mn₂O₅–Na₂WO₄/SiO₂ catalyst were studied to ascertain if CH₄ still dissociated over the WO_x like it did in the case of Na₂WO₄/SiO₂ catalyst. Once again, the color-coded pathways signify the unique surface intermediates produced. In this case however, CH₄ dissociation was studied as a function of the moiety since three distinct moieties are present in the surface site i.e. WO₅ moiety, W–O–Mn moiety, and Mn₂O₅ moiety. The motivation behind studying CH₄ dissociation as a function of moieties is and to ascertain the affect of Mn₂O₅ promotion on the reaction pathway and intrinsic kinetics. If, CH₄ dissociates preferentially over the Mn₂O₅ moiety, it would indicate that the effect of Mn-promotion in OCM catalysts is not to tune the WO_x surface site but in fact to generate a second active site altogether. On the other hand, if it still activates over WO₅ moiety, the role of Mn₂O₅ would be inferred as a structural/chemical promoter.

As shown in Fig. 9a, CH₄ dissociation becomes energetically unfavorable over the WO₅ moiety i.e. 2–4 eV TS barriers, in stark contrast to facile CH₄ dissociation observed over WO₄ site in Na₂WO₄/SiO₂ without Mn₂O₅ promotion (1.1 eV). The higher TS barriers in the case of WO₅ moiety in Mn₂O₅–Na₂WO₄/SiO₂ catalyst in comparison to the pseudo-T_d WO₄ moiety in Na₂WO₄/SiO₂ catalyst are likely due to the loss of T_d geometry of the WO_x center upon Mn₂O₅ promotion, where the T_d WO₄ serves as a reaction site. The T_d geometry of the WO₄ is often regarded as a critical requirement of the selective active site in OCM in various experimental studies,^{19,22,25,56–60} and that notion finds support from the DFT results herein.

Fig. 9 Reaction coordinate vs. ΔE_DFT (eV) over (a) WO₅ moiety, (b) W–O–Mn moiety, (c) Mn₂O₅ moiety in Mn₂O₅–Na₂WO₄/SiO₂ catalyst. (d) Calculated intrinsic rate constants (log-scale) for CH₄ dissociation for the possible pathways in Mn₂O₅–Na₂WO₄/SiO₂ catalyst. Arrow indicates the dominant (lowest barrier) pathway. All numbered pathways 1–9 are described in ESI Table S4.† (e) Visualization of the dominant (lowest barrier) pathway for CH₄ dissociation over the Mn₂O₅–Na₂WO₄/SiO₂ catalyst, yielding methoxy–hydroxy (OCH₃, OH) surface intermediates.

On the other hand, CH₄ can also dissociate over the W–O–Mn and Mn₂O₅ moieties via various pathways with TS barriers in the 1–5 eV range, shown in Fig. 9b and c. Overall, however, the lowest barrier pathway was the one yielding methoxy–hydroxy (OCH₃, OH) intermediates, over the Mn₂O₅ moiety, with a TS barrier of 0.72 eV, which is 0.3 eV lower than the lowest TS barrier pathway for CH₄ activation over Na₂WO₄/SiO₂ in the absence of Mn-promoter. Fig. 9d compares the calculated intrinsic rate constants for CH₄ dissociation across all moieties via various pathways studied, showing that the lowest TS barrier pathway indeed exhibits a rate constant orders of magnitude higher than other cases and hence is the dominant pathway. Lastly, initial, TS, and final structures for the dominant reaction pathway in this case are summarized in Fig. 9e, evidencing the involvement of Mn–O and Mn–O–Mn bonds in CH₄ dissociation to yield Mn–O–CH₃ and (Mn)₂–O–H surface intermediates.

To ascertain the electronic structural dynamics of the surface sites during CH₄ dissociation, LDOS analysis (Mn 3d, O 2p, C 2p, H 1s states) was conducted for the initial and TS structures of Mn₂O₅–Na₂WO₄/SiO₂, shown in Fig. 10a and b, respectively. Initially, the valence edge is comprised of a mixture of occupied Mn 3d states and O 2p states from Mn–O–Mn and Mn–O bonds. The conductance edge is primarily made up of unoccupied Mn 3d states with some contribution from empty O 2p states. Upon CH₄ dissociation and TS formation, the LDOS changes drastically. The upper part of the valence band lowers in energy, and occupied, spin-up Mn 3d states appear at −2.0 eV, which were not present in the initial structure. Moreover, overlapping O 2p and C 2p states as well as the O 2p and H 1s states confirm formation of Mn–O–CH₃ and (Mn)₂–O–H intermediates. In this case, the occupied C 2p states overlap with occupied O 2p states of Mn–O towards the bottom of the valence edge (−7.0 eV). Lastly, Mn Bader charge decreases from +1.63 to +1.55 from initial to TS structure, indicating reduction of the Mn centers from ∼+3 to ∼+2 in agreement with experimental literature.^26,61

Fig. 10 LDOS analysis of (a) initial, and (b) TS structures, evidencing Mn–O–CH₃ and (Mn)₂–O–H formation, suggesting that the CH₄ dissociation is mediated by a TS where a methyl is coordinated to the oxygens on the Mn center, while Mn–O–Mn bridging bond abstracts the hydrogen from the CH₄. E_F value for the initial structure was −2.75 eV and TS was −1.77 eV. Colored atoms correspond to the corresponding color in DOS plots.

(e) C–H bond activation over surface metal oxide sites

The results presented so far point out that heterolytic dissociation of CH₄ can occur over surface sites, either WO₄ or Mn₂O₅, via the Mars–van Krevelen (MvK) mechanism (i.e., involvement of lattice oxygen instead of adsorbed O₂, as inferred from general agreement between TS barriers estimated herein and apparent activation barriers reported in the broader OCM literature in the 100–150 kJ mol⁻¹ range.¹⁹ The dissociation of CH₄ over pseudo-T_d WO₄ sites in Na₂WO₄/SiO₂ catalyst is found to be kinetically slower than that on Mn₂O₅ moiety in Mn₂O₅–Na₂WO₄/SiO₂ catalyst, given that the value for C–H scission in CH₄ over Mn₂O₅ sites is 3 orders of magnitude higher than over the surface Na₂WO₄ sites (Fig. 11). However, the formation of surface intermediates in the case of CH₄ dissociation over Na₂WO₄ surface sites, namely W–CH₃ and W–O–H is endothermic by 0.60 eV. On the other hand, the formation of surface intermediates due to CH₄ dissociation over Mn₂O₅ moiety in Mn₂O₅–Na₂WO₄/SiO₂, namely Mn–O–CH₃ and Mn–O–H, is appreciably exothermic by −0.48 eV. The occupied C 2p states of the CH₃ fragment in Mn–O–CH₃ are more stabilized (i.e., lower in energy) when bound to Mn–O (−7.0 eV) vs. when bound to W as in W–CH₃ (−1.7 eV), as shown in Fig. 8 and 10, which explains the enhanced stability of intermediates formed over Mn₂O₅. The higher stability of intermediates formed over Mn₂O₅ moiety increases their likelihood of staying adsorbed and further oxidizing to the unselective CO_x products instead of desorbing as methyl radicals into the gas-phase to form C₂ products.⁶² It is known that the selectivity towards C₂ products in OCM relies on the thermodynamics of adsorption of the methyl fragment (ion or radical) on the catalyst, which can lead to further oxidation to CO or CO₂.⁶³ Therefore, the energy of the adsorbed CH₃ fragment effectively describes the selectivity to C₂ hydrocarbons during the OCM process. Current results can be extrapolated to hypothesize slower kinetics of CH₄ dissociation, but higher C₂ selectivity for CH₄ dissociation over Na₂WO₄ sites in Na₂WO₄/SiO₂ catalysts in comparison to CH₄ dissociation over Mn₂O₅ sites in Mn₂O₅–Na₂WO₄/SiO₂. In agreement with the present hypothesis, experimental results evidence higher CH₄ conversion, but lower C₂ selectivity for Mn–Na₂WO₄/SiO₂ (CH₄ conversion 36%, C₂ selectivity 54%) vs. Na₂WO₄/SiO₂ (CH₄ conversion 15%, C₂ selectivity 62%) catalysts in OCM.⁶⁴ A recent temporal analysis of products (TAP) reactor study of crystalline phase Mn–Na₂WO₄/SiO₂ catalysts further elucidated that the Na₂WO₄/SiO₂ catalyst (without Mn-oxide) possesses two distinct kinds of oxygen species at 800 °C: (i) a dissolved molecular O₂ type species only released from the molten Na₂WO₄ phase and (ii) an atomic lattice O type species associated with surface Na–WO₄ sites that can be removed by reduction with CH₄.⁶⁵ Both these oxygen species are catalytically active for the OCM reaction. On the other hand, the 1.2Mn/SiO₂ catalyst (without Na₂WO₄) only provides molecular O₂ type species associated with the Mn-oxide phase, and no atomic O type species. The O₂ species associated with the Mn-oxide phase were found to highly unselective toward C₂ product formation from CH₄ activation.⁶⁵ It would be interesting to conduct a similar TAP reactor study on model catalysts in the absence of crystalline oxide phases with only dispersed surface sites present to compare how such dispersed sites function versus the respective crystalline oxide phase.

Fig. 11 Comparison of calculated intrinsic rate constants for dominant (lowest barrier) pathways for CH₄ dissociation over the Na₂WO₄/SiO₂ (green) and Mn₂O₅–Na₂WO₄/SiO₂ (red) catalysts.

The faster intrinsic kinetics of heterolytic CH₄ dissociation over the Mn₂O₅ sites versus the Na₂WO₄ surface sites are not surprising. In fact, DFT calculations have been used to correlate the C–H bond activation energy to the surface reducibility (oxygen vacancy formation energy, work function) in the literature.⁶³ A linear correlation was found between the C–H activation energy and the oxygen vacancy formation energy of pure/doped metal-oxides, making surface reducibility a descriptor for predicting catalyst activity and selectivity for CH₄ dissociation in OCM.⁶³ Given that Mn-oxides are more reducible than W-oxides, the CH₄ dissociation over surface Mn₂O₅ sites is expected to be faster than Na₂WO₄. Moreover, according to this correlation between C–H activation barrier and reducibility of the surface metal oxide site, one of the roles of Na-promotion is to increase the reducibility of surface WO_x sites, as evidenced by lowering of peak temperature values during H₂-TPR with Na-addition, reported in a recent study on Na–WO_x/SiO₂ model catalysts.²²

It is known that MnO_x-based catalysts form excellent partial and total oxidation catalysts owing to their enhanced reducibility. Examples of Mn-oxide based catalysts used for oxidation reactions include: Mn₂O₃ for complete oxidation of methane,⁶¹ Mn-oxide–CeO₂ catalysts for formaldehyde oxidation,⁶⁶ Spinel CoMn₂O₄ for toluene oxidation,⁶⁷ mesoporous Mn-oxide catalysts for water oxidation,⁶⁸ alkane oxidation over Mn-perovskites,⁶⁹ CeO₂-supported Mn-oxide for propane oxidation,⁷⁰ In fact, analogous to the CH₄ dissociation pathway over Mn₂O₅ moiety envisioned in the present study, operando FTIR spectroscopy elsewhere⁷¹ confirmed the formation of Mn–OH surface species by abstraction of hydrogen atoms by nucleophilic oxygen atoms (Mn–O–Mn) from propane during propane oxidation, providing support for the DFT modelling results and chemical insights presented herein.

Recent state-of-the-art understanding regarding the role of Mn- and Na-promoters in Mn₂O₃–Na₂WO₄/SiO₂ OCM catalysts puts chemical insights provided in the present work in perspective. Specifically, it was recently shown via experimental CH₄ + O₂ temperature-programmed-surface-reaction of well-defined single site catalysts that Na₂WO₄ surface sites could effectively and selectively activate CH₄ in absence of any Mn-promoter to form C₂ products, making the role of Mn-promoter unclear.²² Further evidence of the dispersed phase Na–WO₄ surface sites being the critical OCM active sites was provided in recent experimental studies^48,72 where temporal analysis of products (TAP) reactor studies in conjunction with steady-state OCM reaction studies of Na₂WO₄/SiO₂ catalysts (without Mn-oxide) demonstrated that the dispersed surface Na–WO₄ sites were responsible for selectively activating CH₄ to yield C₂ and CO products, while molten Na₂WO₄ phase was found to be mainly responsible for the over-oxidation of CH₄ to CO₂, and oxidative dehydrogenation of C₂H₆ to C₂H₄.⁷² Likewise, Mn-containing phases including MnO_x surface oligomers, MnWO₄ and Mn–WO₃ nanoparticles in Mn–Na–WO_x/SiO₂ tri-metal oxide model OCM catalyst were found to be primarily spectating during OCM under differential reaction conditions and when such Mn phases are present in small populations.⁴⁸ An older model study had previously shown that Na-coordinated WO₄ surface sites were more selective but slightly less active than Mn-coordinated WO_x and uncoordinated WO_x sites during OCM in a series of model bi-metal oxide OCM catalysts.²⁵ Lastly, via operando synchrotron μ-XRF/XRD/absorption-computed tomography, it was shown that crystalline manganese oxide phases including Mn₂O₃, Mn₇SiO₁₂ and MnWO₄ are not required components to yield an active catalyst,⁶ while no conclusion could be reached regarding the dispersed phase Mn_aO_b surface sites that can be present on the SiO₂.³⁰

The current study, utilizing in situ spectroscopy, ab initio molecular dynamics, ab initio thermodynamics, and periodic-DFT modelling bridges the knowledge gap present in the literature and provides the following critical mechanistic insights, which suggest that Mn-promotion in fact is likely not essential for a yielding a selective-active OCM catalyst:

(1) CH₄ heterolytically dissociates effectively over the Na₂WO₄ surface sites in absence of Mn-promoter, with a TS barrier of 1.07 eV forming W–CH₃ and W–O–H surface intermediates. The surface intermediate formation is endothermic by 0.60 eV, indicative of their instability and likelihood of facile desorption.

(2) CH₄ heterolytically dissociates over the WO₅ moiety significantly slower when the Mn₂O₅–Na₂WO₄/SiO₂ catalyst is studied due to structural changes incurred upon Mn₂O₅ coordination to the WO₄ site forming the WO₅. This shows that Mn-promotion can infact poison the otherwise pseudo-T_d WO₄ surface active sites, retarding the kinetics of CH₄ dissociation over them.

(3) CH₄ dissociates over the Mn₂O₅ moiety with a TS barrier of 0.72 eV forming Mn–O–CH₃ and (Mn)₂–O–H surface intermediates. The surface intermediates formation is appreciably exothermic (−0.48 eV), indicative of the high stability of the formed intermediates.

(4) The significantly higher stability of surface intermediates formed when CH₄ dissociates over the Mn₂O₅ moiety in Mn₂O₅–Na₂WO₄/SiO₂ catalyst versus the intermediates formed upon CH₄ dissociation over the Na₂WO₄/SiO₂ signifies the likelihood of the Mn–O–CH₃, (Mn)₂–O–H surface intermediates over-oxidizing to CO_x, in agreement with the higher CH₄ conversion but lower C₂ selectivity reported in the OCM literature for MnO_x/SiO₂ catalysts.⁶⁴ Therefore, the selective-active sites for OCM are the Na-coordinated WO₄ surface sites present in Na₂WO₄/SiO₂ catalysts without Mn-promotion. Mn promotion can create a second set of Mn-based surface sites (like the surface Mn₂O₅ sites) that are more active for CH₄ dissociation than the WO₄-based sites, but less selective to C₂ products due to the unfavorable desorption of the reaction intermediates formed over Mn₂O₅.

(5) Computational results herein also suggest that CH₄ activation via the MvK mechanism where the oxygen species involved in the reaction originate in the solid phase catalytic active site is quite possible. The DFT-calculated intrinsic barriers of the MvK type reaction pathways are in general agreement with the experimental apparent activation energy values reported in the literature, providing qualitative support for the possibility of the MvK mechanism prevailing during catalytic OCM. However, we note that only the heterolytic CH₄ activation pathway is studied herein and the homolytic CH₄ activation pathway over these surface sites will be addressed in a future study. Literature studies have shown that both heterolytic and homolytic activation of CH₄ occurs during OCM, where both pathways yield CH₃ radicals in the gas-phase.^73,74 Typically, homolytic pathways exhibit higher energy barriers than the heterolytic pathway⁵⁵ due to the stabilization of the CH₃ fragment from surface interactions in the heterolytic pathway.

4. Conclusions

The study herein utilized in situ spectroscopy, AIMD, AITD, and periodic-DFT to provide unprecedented chemical insights regarding CH₄ activation over Na₂WO₄/SiO₂versus Mn–Na₂WO₄/SiO₂ catalysts. We report, for the first time, that Mn-promotion is not essential to tune the pseudo-T_d WO₄ surface sites known to be active for OCM. Contrary to previous understanding, we show that Mn-promotion poisons the pseudo-T_d Na₂WO₄ surface site, with the resulting surface WO₅ site exhibiting retarded kinetics for CH₄ activation. With Mn-promotion, however, new Mn-based surface sites (Mn₂O₅ in this study) form that can activate CH₄ faster than the WO₄-based surface sites. The surface intermediates formed during CH₄ dissociation over the Mn₂O₅ sites (Mn–O–CH₃, (Mn)₂–OH) are more stable than those formed via CH₄ dissociation over the Na₂WO₄ site (W–CH₃, W–O–H), and prone to over-oxidation to CO_x due to the unfavorable desorption. Considering these insights that agree with recent state-of-the-art in situ and operando characterization studies, it is proposed that the addition of MnO_x does not promote the pseudo-T_d, Na-coordinated WO₄ sites where CH₄ activates selectively (i.e., essential selective active sites for OCM). The Mn-promotion likely creates a second set of active MnO_x surface sites that are more active for CH₄ activation, but less C₂ selective than the Na–WO₄ surface sites. These findings contradict Lunsford's OCM model, where the Na–O–Mn sites are hypothesized to be critically involved in the redox mechanism during OCM.⁷⁵ The results herein also do not completely agree with Li's OCM model, as W⁶⁺ is barely seen to reduce during CH₄ dissociation step, while Li proposed that W reduces from +6 to +5/+4 during the CH₄ dissociation step, and Mn₂O₃ provides electron-coupled oxygen-spillover to re-oxidize the reduced W sites back to +6 oxidation state.⁷⁶ Lastly, note that the findings herein do not account for the crystalline nanoclusters/nanoparticles (<2–3 nm) of MnWO₄ or Mn-contaminated WO₃ which are present in the tri-metal oxide model catalyst studied herein. The exact nature of these nanoclusters/nanoparticles is not understood presently and they can only be modelled accurately to generate structure–function relationships for OCM when their size, shape, surface composition, vacancy/defect density, etc. are well-characterized. It suffices to note here that crystalline MnWO₄, even when present, does not contribute significantly to OCM.^29,30

Future investigations on the role of surface MnO_x sites during CH₄ activation in well-defined catalysts should include CH₄ + O₂ temperature-programmed surface reaction, steady state performance tests, isotope-switch experiments in temporal analysis of products (TAP) reactor, and computational analysis of the homolytic pathway for CH₄ activation over the catalyst models generated in this study.

Data availability

Relevant data is present in the ESI file.†

Author contributions

DK and JB conducted all theoretical work. DK and SS conducted spectroscopy experiments. DK, JB, and IEW wrote the manuscript. All authors have agreed to the publication of this manuscript.

Conflicts of interest

The authors declare no competing interests.

Acknowledgements

This work is supported by NSF CBET 1706581. DK also gratefully acknowledges John C. Chen endowed fellowship for financial support. This work used the Extreme Science and Engineering Discovery Environment (XSEDE) (Comet cluster), which is supported by National Science Foundation grant number ACI-1548562, under allocation ID TG-CTS150038 with XSEDE.⁷⁷

References

1. X. Li, J. Xie, H. Rao, C. Wang and J. Tang, Angew. Chem., Int. Ed., 2020, 59, 19702–19707.
2. A. M. Arinaga, M. C. Ziegelski and T. J. Marks, Angew. Chem., Int. Ed., 2021, 60, 10502.
3. V. Paunović, G. Zichittella, M. Moser, A. P. Amrute and J. Pérez-Ramírez, Nat. Chem., 2016, 8, 803–809.
4. M. Moser, L. Rodríguez-García, A. P. Amrute and J. Pérez-Ramírez, ChemCatChem, 2013, 5, 3520–3523.
5. A. Caballero and P. J. Perez, Chem. Soc. Rev., 2013, 42, 8809–8820.
6. M.-S. Salehi, M. Askarishahi, H. R. Godini, O. Görke and G. Wozny, Ind. Eng. Chem. Res., 2016, 55, 3287–3299.
7. T. N. Nguyen, T. T. P. Nhat, K. Takimoto, A. Thakur, S. Nishimura, J. Ohyama, I. Miyazato, L. Takahashi, J. Fujima, K. Takahashi and T. Taniike, ACS Catal., 2020, 10, 921–932.
8. Z. Cheng, D. S. Baser, S. G. Nadgouda, L. Qin, J. A. Fan and L.-S. Fan, ACS Energy Lett., 2018, 3, 1730–1736.
9. S. Parishan, P. Littlewood, A. Arinchtein, V. Fleischer and R. Schomäcker, Catal. Today, 2018, 311, 40–47.
10. A. L. Tonkovich, R. W. Carr and R. Aris, Science, 1993, 262, 221.
11. X. Guo, G. Fang, G. Li, H. Ma, H. Fan, L. Yu, C. Ma, X. Wu, D. Deng, M. Wei, D. Tan, R. Si, S. Zhang, J. Li, L. Sun, Z. Tang, X. Pan and X. Bao, Science, 2014, 344, 616–619.
12. K. Takahashi, L. Takahashi, T. N. Nguyen, A. Thakur and T. Taniike, J. Phys. Chem. Lett., 2020, 11, 6819–6826.
13. L. Grajciar, C. J. Heard, A. A. Bondarenko, M. V. Polynski, J. Meeprasert, E. A. Pidko and P. Nachtigall, Chem. Soc. Rev., 2018, 47, 8307–8348.
14. P. Schlexer Lamoureux, K. T. Winther, J. A. Garrido Torres, V. Streibel, M. Zhao, M. Bajdich, F. Abild-Pedersen and T. Bligaard, ChemCatChem, 2019, 11, 3581–3601.
15. J. A. Labinger, Science, 1995, 269, 1833.
16. D. Kiani, S. Sourav, Y. Tang, J. Baltrusaitis and I. E. Wachs, Chem. Soc. Rev., 2021, 50, 1251–1268.
17. D. Kiani, G. Belletti, P. Quaino, F. Tielens and J. Baltrusaitis, J. Phys. Chem. C, 2018, 122, 24190–24201.
18. R. K. Grasselli and A. W. Sleight, in Studies in Surface Science and Catalysis, ed. R. K. Grasselli and A. W. Sleight, Elsevier, 1991, vol. 67, p. ix.
19. D. Kiani, S. Sourav, J. Baltrusaitis and I. E. Wachs, ACS Catal., 2019, 9, 5912–5928.
20. Z. Li, S. Ji, Y. Liu, X. Cao, S. Tian, Y. Chen, Z. Niu and Y. Li, Chem. Rev., 2020, 120, 623–682.
21. L. Zhang, M. Zhou, A. Wang and T. Zhang, Chem. Rev., 2020, 120, 683–733.
22. D. Kiani, S. Sourav, I. E. Wachs and J. Baltrusaitis, Catal. Sci. Technol., 2020, 10, 3334–3345.
23. I. E. Wachs and K. Routray, ACS Catal., 2012, 2, 1235–1246.
24. S. Arndt, T. Otremba, U. Simon, M. Yildiz, H. Schubert and R. Schomäcker, Appl. Catal., A, 2012, 425–426, 53–61.
25. D. Kiani, S. Sourav, W. Taifan, M. Calatayud, F. Tielens, I. E. Wachs, J. Baltrusaitis, I. E. Wachs and J. Baltrusaitis, ACS Catal., 2020, 10, 4580–4592.
26. W. Riedel, L. Thum, J. Möser, V. Fleischer, U. Simon, K. Siemensmeyer, A. Schnegg, R. Schomäcker, T. Risse and K.-P. Dinse, J. Phys. Chem. C, 2018, 122, 22605–22614.
27. W. Pengwei, Z. Guofeng, Y. Wang, L. Yong, P. Wang, G. Zhao, Y. Wang and Y. Lu, Sci. Adv., 2017, 3, 1–10.
28. P. Wang, G. Zhao, Y. Liu and Y. Lu, Appl. Catal., A, 2017, 544, 77–83.
29. M. J. Werny, Y. Wang, F. Girgsdies, R. Schlögl and A. Trunschke, Angew. Chem., Int. Ed., 2020, 59, 14921.
30. A. Vamvakeros, D. Matras, S. D. M. Jacques, M. di Michiel, S. W. T. Price, P. Senecal, M. A. Aran, V. Middelkoop, G. B. G. Stenning, J. F. W. Mosselmans, I. Z. Ismagilov and A. M. Beale, J. Catal., 2020, 386, 39–52.
31. H. Guesmi, R. Grybos, J. Handzlik and F. Tielens, RSC Adv., 2016, 6, 39424–39432.
32. K. Kurleto, F. Tielens and J. Handzlik, J. Phys. Chem. C, 2020, 124(5), 3002–3013.
33. M. Lahti, A. Chaudhuri, K. Pussi, D. Hesp, I. M. McLeod, V. R. Dhanak, M. O. King, M. Kadodwala and D. A. MacLaren, Surf. Sci., 2014, 622, 35–43.
34. http://www.p4vasp.at/, 2021.
35. M. Yu and D. R. Trinkle, J. Chem. Phys., 2011, 134, 064111.
36. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
37. G. Henkelman and H. Jónsson, J. Chem. Phys., 2000, 113, 9978–9985.
38. G. Li, P. Vassilev, M. Sanchez-Sanchez, J. A. Lercher, E. J. M. Hensen and E. A. Pidko, J. Catal., 2016, 338, 305–312.
39. N. Kosinov, A. S. G. Wijpkema, E. Uslamin, R. Rohling, F. J. A. G. Coumans, B. Mezari, A. Parastaev, A. S. Poryvaev, M. V. Fedin, E. A. Pidko and E. J. M. Hensen, Angew. Chem., 2018, 130, 1028–1032.
40. G. Li, I. Vollmer, C. Liu, J. Gascon and E. A. Pidko, ACS Catal., 2019, 9, 8731–8737.
41. The Materials Project, Materials Data on Mn2O3 by Materials Project, United States, 2014,  DOI:10.17188/1272415.
42. Y.-B. Kang and I.-H. Jung, Metall. Mater. Trans. E, 2016, 3, 156–170.
43. W. E. W. E. Taifan, T. Bučko and J. Baltrusaitis, J. Catal., 2017, 346, 78–91.
44. M. W. Chase Jr, NIST-JANAF thermochemical tables, American Chemical Society, American Institute of Physics for the National Institute of Standards and Technology, Washington, DC, New York, 4th edn, 1998.
45. K. K. Irikura, J. Phys. Chem. Ref. Data, 2007, 36, 389–397.
46. K. Reuter and M. Scheffler, Phys. Rev. B, 2001, 65, 035406.
47. R. Jutta and K. Reuter, Max-Planck-Gesellschaft Zur Foerderung Der Wissenschaften ev Berlin (GERMANY FR), Fritz-Haber-INST, 2006.
48. D. Kiani, S. Sourav, J. Baltrusaitis and I. E. Wachs, ACS Catal., 2021, 11, 10131–10137.
49. F. Yang, J. Graciani, J. Evans, P. Liu, J. Hrbek, J. F. Sanz and J. A. Rodriguez, J. Am. Chem. Soc., 2011, 133, 3444–3451.
50. E. I. Ross-medgaarden and I. E. Wachs, J. Phys. Chem. C, 2007, 111, 15089–15099.
51. S. Lwin, Y. Li, A. I. Frenkel and I. E. Wachs, ACS Catal., 2016, 6, 3061–3071.
52. Q. Wan, V. Fung, S. Lin, Z. Wu and D.-e. Jiang, J. Mater. Chem. A, 2020, 8, 4362–4368.
53. K. Takanabe and E. Iglesia, J. Phys. Chem. C, 2009, 113, 10131–10145.
54. Z. Liu, J. P. Ho Li, E. Vovk, Y. Zhu, S. Li, S. Wang, A. P. van Bavel and Y. Yang, ACS Catal., 2018, 8, 11761–11772.
55. M. C. Cholewinski, M. Dixit and G. Mpourmpakis, ACS Omega, 2018, 3, 18242–18250.
56. J. Wu, S. Li, J. Niu and X. Fang, Appl. Catal., A, 1995, 124, 9–18.
57. S.-f. Ji, T.-c. Xiao, S. S.-B. B. Li, C.-z. Xu, R.-l. Hou, K. S. Coleman, M. L. H. Green, J. Wu, S. S.-B. B. Li, Z. C. Jiang, C. J. Yu, X. P. Fang, S. S.-B. B. Li, H. L. Wang, J. Wu, S. S.-B. B. Li, J. Niu, X. P. Fang, D. J. Wang, M. P. Rosynek, J. H. Lunsford, S. S.-B. B. Li, A. Palermo, J. Pedro, H. Vazquez and R. M. Lambert, Appl. Catal., A, 1995, 68, 191–196.
58. J. Wu and S. Li, J. Phys. Chem., 1995, 99, 4566–4568.
59. S. Ji, T. Xiao, S. Li, L. Chou, B. Zhang, C. Xu, R. Hou, A. P. E. E. York and M. L. H. H. Green, J. Catal., 2003, 220, 47–56.
60. S. Li, J. Nat. Gas Chem., 2003, 12, 1–9.
61. Y.-F. Han, K. Ramesh, L. Chen, E. Widjaja, S. Chilukoti and F. Chen, J. Phys. Chem. C, 2007, 111, 2830–2833.
62. M. J. Capitan, J. A. Odriozola, A. Marquez and J. F. Sanz, J. Catal., 1995, 156, 273–278.
63. G. Kumar, S. L. J. Lau, M. D. Krcha and M. J. Janik, ACS Catal., 2016, 6, 1812–1821.
64. S. Hou, Y. Cao, W. Xiong, H. Liu and Y. Kou, Ind. Eng. Chem. Res., 2006, 45, 7077–7083.
65. S. Sourav, Y. Wang, D. Kiani, J. Baltrusaitis, R. R. Fushimi and I. E. Wachs, ACS Catal., 2021, 11, 10288–10293.
66. J. Quiroz, J.-M. Giraudon, A. Gervasini, C. Dujardin, C. Lancelot, M. Trentesaux and J.-F. Lamonier, ACS Catal., 2015, 5, 2260–2269.
67. C. Dong, Z. Qu, Y. Qin, Q. Fu, H. Sun and X. Duan, ACS Catal., 2019, 9, 6698–6710.
68. C.-H. Kuo, I. M. Mosa, A. S. Poyraz, S. Biswas, A. M. El-Sawy, W. Song, Z. Luo, S.-Y. Chen, J. F. Rusling, J. He and S. L. Suib, ACS Catal., 2015, 5, 1693–1699.
69. G. Koch, M. Hävecker, D. Teschner, S. J. Carey, Y. Wang, P. Kube, W. Hetaba, T. Lunkenbein, G. Auffermann, O. Timpe, F. Rosowski, R. Schlögl and A. Trunschke, ACS Catal., 2020, 10, 7007–7020.
70. H. Wang, H. Zhou, S. Li, X. Ge, L. Wang, Z. Jin, C. Wang, J. Ma, X. Chu, X. Meng, W. Zhang and F.-S. Xiao, ACS Catal., 2020, 10, 10559–10569.
71. X. Li, D. Teschner, V. Streibel, T. Lunkenbein, L. Masliuk, T. Fu, Y. Wang, T. Jones, F. Seitz, F. Girgsdies, F. Rosowski, R. Schlögl and A. Trunschke, Chem. Sci., 2019, 10, 2429–2443.
72. S. Sourav, Y. Wang, D. Kiani, J. Baltrusaitis, R. R. Fushimi and I. E. Wachs, Angew. Chem., Int. Ed., 2021, 60, 21502.
73. J. H. Lunsford, Angew. Chem., Int. Ed. Engl., 1995, 34, 970–980.
74. D. Dissanayake, J. H. Lunsford and M. P. Rosynek, J. Catal., 1994, 146, 613–615.
75. D. J. Wang, M. P. Rosynek and J. H. Lunsford, J. Catal., 1995, 155, 390–402.
76. S. Li, J. Nat. Gas Chem., 2003, 12, 1–9.
77. J. Towns, T. Cockerill, M. Dahan, I. Foster, K. Gaither, A. Grimshaw, V. Hazlewood, S. Lathrop, D. Lifka, G. D. Peterson, R. Roskies, J. R. Scott and N. Wilkins-Diehr, Comput. Sci. Eng., 2014, 16, 62–74.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1sc02174e

---