Introduction

Direct valorization of methane into higher hydrocarbons via methane coupling has been a longstanding research effort with high economic interest.¹ Both non-oxidative (NOCM) and oxidative (OCM) methane coupling processes have been studied. Overoxidation of methane to CO_x species is the major route to byproducts in OCM, whereas NOCM suffers from coking.² While each of these processes has specific challenges, they aim to generate similar products, i.e. C₂ or aromatics, and are typically thought to involve different intermediates and reaction mechanisms.^2-6 While surface sites are thought to be important to activate the C−H bond of methane, radicals and gas phase reactions are typically invoked for both OCM and NOCM.

Catalysts for both reactions are typically metal oxides, usually doped with additional elements. In OCM, the most illustrative example is Li/MgO, where Li addition is beneficial for C₂ selectivity owing to a structural reorganization and loss of porosity compared to the pure MgO catalyst.⁷ The effect of Li has been attributed to the formation of [Li⁺O⁻⋅]_MgO surface sites,⁸ which initiate homolytic bond splitting to form methyl radicals that subsequently couple to C₂ products according to the Lunsford mechanism (Eq. 1.^{9, 10}

(1)

Furthermore, the catalytic function of MgO is presumed to be connected to defect sites rather than the MgO phase itself.⁷ While the involvement of [Li⁺O⁻⋅]_MgO is one prevalent rationale for catalytic activity, the nature of the active sites is still a matter of debate,^{3, 8, 11} considering that materials containing neither Li or Mg are also active and selective in OCM toward C₂ products.³ A possible alternative mechanism involves the heterolytic C−H activation of methane on a [Mg²⁺O²⁻] site, yielding a surface OH group and a Grignard-type Mg-methylate ([HO⁻(Mg−CH₃⁺)]) intermediate. Further reaction with O₂ is proposed to liberate the free methyl radicals, which then couple in the gas phase to generate carbon-carbon bonds.¹¹

In NOCM and related processes, which operate at higher temperatures, metal carbides are formed under reaction conditions and can also be employed as catalytic materials.^12-14 In fact, recent work from our group has shown that bulk carbidic carbons in Mo and W carbides can readily exchange with methane present in the gas phase, due to a fast carbon diffusion at these reaction conditions. This study also suggests that carbon homologation may not necessarily involve gas phase processes and that coupling of hydrocarbyl ligands can also take place at the surface of the metal carbide materials.¹²

The involvement of specific mechanisms for OCM and NOCM is certainly consistent with the respective operating conditions, in particular the presence or absence of molecular oxygen as well as the different operating reactor temperatures (700–1100 °C). While thermodynamic limitations in NOCM require the use of high reaction temperatures up to 1100 °C,² OCM is typically operated at 800 °C. Note however that OCM is so exothermic that increase of temperature in the catalyst bed up to 1100 °C has been reported.¹⁵

Bearing the similarities in product and operation temperature in mind, we decided to examine the performance of Li/MgO and MgO under NOCM conditions. In particular, this work shows that, under NOCM conditions, Li/MgO is particularly efficient to promote carbon-carbon bond formation and generate acetylide intermediates. The observation of these intermediates correlates with the high C₂ selectivity, suggesting a relation between them.

Results and Discussion

First, Li/MgO containing 6 wt % Li was synthesized as previously reported.¹⁶ Next, Li/MgO was contacted with a flow of methane at 1000 °C (40 sccm feed gas, 200 mg cat., 10 % CH₄/Ar). Under these conditions, ethane was observed as major product with initial C₂ selectivities reaching 30 % (Figure S2). The C₂ selectivity declined over the course of 2 h and stabilized at around 10 % C₂ selectivity with a steady-state CH₄ conversion of ca. 5 %, showing that Li/MgO can indeed participate in carbon-carbon bond formation under non-oxidative conditions without the need of molecular oxygen. The residual selectivity is ascribed to coke formation.

Isotope labelling experiments were then conducted in batch with the goal to track the state of the spent catalyst (Figure 1a). 200 mg of Li/MgO (6 wt % Li) or MgO were loaded into a quartz reactor which was then filled at room temperature with 200 mbar of ¹³CH₄ (molar ratio 1 ¹³CH₄ : 2 Li : 6 MgO for Li/MgO and 1 ¹³CH₄ : 6 MgO for MgO) and heated to 1000 °C for 2 h with aliquots drawn after 1 and 2 h. The reactor was evacuated and refilled at room temperature after 2 h due to significant methane consumption and was allowed to react at 1000 °C for another 0.5 h.

Under these conditions, a rapid consumption of methane can be observed (full conversion within 2 h), while ethane, ethylene and acetylene are detected in the gas phase (Figure 1b). The highest selectivities are observed for ethylene and acetylene while ethane selectivity is comparatively low. Surprisingly, the spent Li/MgO yielded a white powder without visible evidence for coking. Notably, the quartz glass reactor showed signs of devitrification, which is not observed in the absence of Li indicative of intercalation of “Li₂O” in the silica-glass wall and thereby loosing Li from the catalyst.¹⁷ Considering the thermodynamic equilibrium among C₂ species at 1000 °C (see Figure S1), the observed C₂ distribution is in line with expectations. Besides C₂ species, CO₂ and H₂O are detected, showing that the oxygen of the oxide support is accessible and can participate in the reaction.

The corresponding reaction with MgO was further investigated (Figure 1c), where both CH₄ consumption and C₂ product selectivities are significantly lower. This observation parallels what has been observed in OCM where Li doping is crucial for high C₂ selectivities.^{18, 19} Contrary to Li/MgO, the spent MgO turns black, visibly showing coking while no devitrification of the quartz glass reactor was observed.

To understand the difference in catalytic activity and appearance of these two solids post-reaction, they were analyzed by solid-state ¹³C magic-angle-spinning (MAS) NMR spectroscopy. The ¹³C MAS-echo NMR spectrum of post-reaction MgO (Figure 2a) is dominated by an intense ¹³C signal at 129 ppm, consistent with the formation of graphitic carbonaceous deposits,²⁰ the color change of the material post-reaction and the poor C₂ selectivity. In contrast, the ¹³C MAS echo NMR spectrum of Li/MgO (Figure 2b) shows an intense narrow ¹³C signal at 170 ppm along with additional signals at 14 and 3 ppm. A ¹³C{¹H} cross-polarization (CP)-MAS spectrum (Figure 2c), acquired at low temperature (100 K) in order to improve signal sensitivity and to mitigate the influences of surface dynamics, shows an additional weak signal at 129 ppm as well as a distribution of signals in the 14–34 ppm region. The low relative intensity of the ¹³C signal at 129 ppm from coke or related aromatic species in spent Li/MgO compared to MgO indicates that Li effectively suppresses most coke formation and contributes to the enhanced C₂ selectivity of the Li/MgO material. The ¹³C NMR signal at 170 ppm is attributed to MgC₂ based on literature reports²¹ and the formation of the MgC₂ species is only observed in the catalyst material containing Li (see below). Close inspection of the ¹³C{¹H} CPMAS and single-pulse ¹³C spectra of the Li/MgO acquired under the same conditions (Figure S5) show that there are in fact two signals that overlap in this region, with very similar chemical shifts (170 vs 170.5 ppm) but different linewidths (3 vs 5 ppm) that are tentatively attributed to bulk vs near-surface Mg acetylide species, respectively. The minor ¹³C signals at 3 and 14 ppm could arise from Mg alkyl species, which tend to exhibit shielded ¹³C chemical shifts.²² The signal at 34 ppm is noteworthy and suggests the formation of alkyl chains that are at least three carbons long, which could be associated with alkyl magnesium or alkylated aromatic surface species. The ¹H MAS NMR spectrum of spent Li/MgO (Figure S3) shows a range of relatively narrow ¹H signals from 0.4 to 3.2 ppm, indicating the presence of a distribution of alkyl species, as well as broad signals at 5.2 and 7.4 ppm, consistent with the presence of surface-bound water and aromatic species, respectively.

While no lithium acetylide species, expected at around 195 ppm,²³ was observed in the ¹³C MAS NMR spectra of the spent catalyst, probably due to low sensitivity and concentration, solid-state ⁶Li MAS NMR analyses of pristine and spent Li/MgO reveal the presence of Li−C species as indicated by the doublet feature likely due to the coupling of ⁶Li and ¹³C nuclei (Figure 2d). A prominent ⁶Li signal at 0.0 ppm observed both before and after reaction corresponds to Li^I bound in a mixed metal oxide Li_xMg_yO_z.

After reaction, additional ⁶Li NMR signals are detected at 0.8 and 1.4 ppm, which can be attributed to distinct types of Li acetylide or related organolithium species.^{24, 25} The solid-state ⁶Li and ¹³C results overall show that Li and Mg acetylides are formed in situ and are stable post-reaction under inert conditions. This finding indicates a close relation between in situ MgC₂ formation and C₂ selectivity, hence providing a potential rationale for enhanced C₂ selectivity of the Li-doped MgO as the formation of MgC₂ is not observed in the absence of Li.

The formation of Mg acetylides is concomitant with efficient NOCM toward C₂ products indicating that it may be an important intermediate for the catalytic reaction.

(2)

Similarly to calcium, the reaction of Mg acetylides and water will yield acetylene (Eq. 2).²⁶ Under NOCM, this reaction could play a role in product formation to some extent as the formation of water has been observed under our reaction conditions.

To rationalize the effect of Li-doping, the formation free energies of Li₂C₂ and MgC₂ were calculated starting from the corresponding oxide using the standard enthalpy and entropy of formation reported in the literature (see Table S1).²⁷ We considered the following equilibria:

(3)

(4)

Li₂C₂ is known to possess an exceptional stability (Δ_fH°(Li₂C₂)=−14.1 kcal mol⁻¹) compared to other alkali and earth alkali acetylides (Δ_fH°(MgC₂)=+19.0 kcal mol⁻¹) and its formation was reported as early as 1896.²⁸ To be more realistic with regard to the experimental conditions, standard pressure for methane and 10⁻² bar pressure for H₂O and H₂ were considered (see Figure S4 for alternative plots under standard conditions). As expected, the formation of Li₂C₂ is much more favorable than MgC₂ (Figure 3) and becomes favorable around 900 °C while MgC₂ is only expected to be a metastable phase at reaction temperature. Based on catalyst bed temperatures in OCM of up to 1100 °C, it is proposed that the formation of Li₂C₂ induces the formation of thermodynamically disfavored MgC₂. The role of Li is thus ascribed to inducing formation of MgC₂.

Considering that Li and Mg acetylides can be considered as a subclass of metal carbides and that both have been shown to be formed in situ, they might play a similar role in methane coupling reactions. These findings parallels the behavior observed for Mo₂C and WC, where bulk carbon can participate in C₂ product formation and diffusion coefficients of carbidic carbon are proposed descriptors for carbidic carbon exchange dynamics, and can describe the extent of exchange between methane in the gas phase and metal carbide (surface and bulk), resulting in incorporation of carbidic carbon in the products.¹² In the case of alkaline (earth) metal acetylides, the bulk structures feature embedded C₂ units (C−C bond length of 1.2 Å), thus the direct evolution of C₂H_x (x=2,4,6) products by hydrogenation or interaction with adsorbed H is probable. To gain insight into the influence of the presence of acetylides on the catalytic performances of Li/MgO, the diffusion mechanism of carbon in MgC₂ and Li₂C₂ was explored employing metadynamics at the density functional theory (DFT) level (see Supporting Information Section 5 for computational details).^29-35 The diffusion of a carbon atom at 1000 °C through an approximately 10×10×10 Å cell of metal acetylide was considered. A history-dependent bias was applied along two collective variables (CV): i) the distance of a selected carbon atom to the geometric center of the cell (d), and ii) the coordination number of this carbon atom to the other carbon atoms CN(C−C). Applying this set of CVs, carbon diffusion pathways were explored while enabling C−C bond cleavage.

In MgC₂, carbon diffusion takes place in the plane of the C₂ units (Figure 4a, initial). The biased carbon (marked in red) diffuses as a C₂ unit in the bulk structure, which is energetically preferred over C−C bond cleavage. A transient C₄ chain is formed between the diffusing C₂ and the neighboring C₂ unit (Figure 4a, C₄). The average C−C bond lengths range from 1.26 Å to 1.33 Å upon C₄ chain formation. This transient C₄ chain then breaks and a substitutional diffusion³⁶ of the C₂ fragment takes place to reach a final configuration identical to the initial one (Figure 4a, final). The overall free energy barrier for this diffusion process in MgC₂ is 54.0 kcal mol⁻¹, which is associated to a carbon diffusion coefficient =1.7×10⁻¹⁵ m² s⁻¹, evidencing that carbon diffusion in MgC₂ is accessible at 1000 °C (Figure 4b).

At the same temperature, the Li₂C₂ is in molten phase as expected from the bulk melting temperature (T_mp=452 °C),³⁷ which leads to a facile diffusion of the C₂ units (Figure 4c) featuring an average length of 1.27 Å of the C−C bond. The resulting free energy barrier for C₂ diffusion in Li₂C₂ is significantly lower at 12.6 kcal mol⁻¹, and the associated diffusion coefficient ( =1.1×10⁻⁸ m² s⁻¹) reflects the fast diffusion of carbon in Li₂C₂ at 1000 °C (Figure 4d).

In both materials, C₂ units diffuse without undergoing C−C bond cleavage. The ease of C₂ diffusion suggests facilitated exchange of carbidic carbon with the gas phase for both acetylides, though with much faster diffusion for Li₂C₂ compared to MgC₂. The persistence of the C₂ moiety in the bulk is reflected on the reactivity of the acetylide material leading to increased C₂ selectivity.

Conclusion

This study shows that the archetypical OCM catalyst Li/MgO can efficiently promote carbon-carbon bond formation under NOCM conditions yielding C₂ products, without significant coke formation. Analysis of Li/MgO post-reaction clearly evidences the formation of MgC₂ in the presence of Li, while pure MgO favors coke formation (as do many other oxides). It is proposed that the presence of Li promotes the formation of MgC₂, probably through the formation of Li₂C₂, which explains the origin of the enhanced C₂ selectivity of Li/MgO as opposed to pure MgO.^{38, 39} This increase in selectivity can be partially rationalized by the increased carbon mobility where the carbon of the acetylide diffuses as C₂ unit as shown by metadynamics simulations, though the mechanism of carbon-carbon bond formation remains to be explored. While MgC₂ has thus far not been reported as a catalyst under OCM conditions, one may propose that MgC₂ species are formed as transient intermediates contributing to the higher observed C₂ selectivity in the presence of Li-doping. The formation of organo-lithium/magnesium species on Li/MgO has been previously proposed¹¹ and could be an important step toward carbon-carbon bond formation in the bulk, leading to C₂ (acetylide) unit formation. Notably, this process is reminiscent of the acetylene production based on calcium carbide, which has been a multi-ton scale process since the 70s and remains the most important production route to the present day.^{40, 41}

We think that the discussed parallels between OCM and NOCM open alternative perspectives for the development of catalytic materials. The involvement of metal carbides or acetylides seems to be a recurring theme in methane coupling where carbidic carbon exchange dynamics has to be considered. We are currently pursuing these lines of research in our pursuit to develop alternative catalytic systems.

Conflict of interest

The authors declare no conflict of interest.