Carbon-based solid catalysts hold significant research importance in methane catalytic cracking due to their abundant precursor resources, cost-effectiveness, and exceptional thermal stability^[18−20]. Importantly, these catalysts inherently resist sulfur components like H₂S in methane, which typically cause irreversible deactivation in metal catalysts. This resistance eliminates the need for pre-desulfurization in traditional metal catalytic systems, thereby enhancing their suitability for processing sulfur-containing feed gases. Moreover, in carbon-based catalytic systems, carbonaceous deposits formed during reactions can transform into new active sites through in situ activation, demonstrating distinctive autocatalytic properties^[12].

The activity and stability of solid carbon materials as catalysts for methane pyrolysis are primarily influenced by two factors: the atomic-scale electronic/geometric structure and the macroscopic pore structure. At the atomic level, catalytic activity arises from unsaturated carbon sites at edges or defects, known as high-energy sites^{[12−14,21−22]}. Muradov et al.^[21−22] conducted a systematic comparative study on various carbon materials’ catalytic activities. They discovered that disordered (amorphous) carbon materials, such as activated carbon (AC) and carbon black (CB), exhibit significantly higher catalytic activities than highly ordered carbon materials like graphite, diamond and carbon nanotubes. This increased activity is attributed to the abundance of high-energy sites in disordered carbon materials, including crystal edges, defects, and dangling bonds. Muradov et al.^[21] also observed that upon contact with hydrocarbons, CB transitions from a randomly oriented structure to a more ordered turbostratic structure, with an interplanar spacing (d-spacing) of 3.4948 Å between amorphous carbon and pure graphite. This ordering is linked to a reduction in catalytic activity, suggesting that defect sites are the active centers for catalytic reactions. Thus, catalytic activity is inversely related to the crystallinity of carbon, with amorphous carbon being more active than turbostratic carbon, which in turn is more active than graphite. Wang et al.^[13] utilized the full width at half-maximum of the C 1s peak in X-ray photoelectron spectroscopy to assess carbon defect concentrations. This concentration is positively correlated with the initial reaction rate, reaffirming that the defect sites serve as the active centers for catalytic reactions.

Although high initial activity is impressive, it does not guarantee long-term stability. The sustained activity of catalyst largely relies on its physical structural properties, such as specific surface area and pore structure. Muradov et al.^[21] observed that although AC exhibits a high initial activity (e.g., coacervate AC has an initial rate of up to 1.76−1.90 mmol/(min·g)), its numerous micropores with an effective pore size of 25−32 Å are prone to clogging by large carbon crystals (>100 Å) formed during methane decomposition. This clogging leads to a rapid decline in surface area and, consequently, quick deactivation. In contrast, most of the CB surface is “open” and easily accessible, resulting in a slower decay of surface area and more stable catalytic performance^[21]. The properties of various carbon materials such as disordered microporous carbon, mesoporous carbon, ordered mesoporous carbon, and nonporous carbon nanotubes were also compared. Although mesoporous carbon initially shows the highest activity, its pore channels quickly become deactivated due to the rapid clogging by deposited carbon. In contrast, ordered mesoporous carbon, with uniformly large pore size of ~3.5 nm, offers moderate initial activity but provides ample space for carbon deposits within its regular pore system. This structure allows ordered mesoporous carbon to maintain stable catalytic activity over a longer duration, ultimately achieving the highest cumulative hydrogen production of 143 mmol/g^[19]. The study highlights that an ideal carbon catalyst should combine a high defect concentration for abundant active sites with a well-developed, regular mesoporous system to resist deactivation from carbon deposition.

The synergy between carbon materials in composites has shown a significant catalytic effect. Yang et al.^[18] examined the enhanced performance of methane decomposition using both AC and CB. While AC exhibits high initial activity, it deactivates quickly, on the other hand, CB offers stability and autocatalytic properties despite its lower initial activity. When mixed in various ratios, the AC_0.5CB_0.5 composite demonstrated optimal catalytic stability at 900 °C. The initial high activity was attributed to AC, whereas CB contributed to long-term stability. Characterization revealed that on AC, the carbon was mostly amorphous and prone to pore clogging, leading to deactivation. In contrast, on the hybrid AC/CB catalyst, CB altered the morphology and structure of the deposited carbon, where the carbon was more graphitized with curved graphene layers and structural defects (e.g., edges, dangling bonds), creating new active sites for methane decomposition and delaying deactivation.

While composite strategies can enhance the performance of carbon-based catalysts, their low intrinsic activity remains a significant barrier to widespread application^{[12,19,21−22]}. Experimental data indicate that the methane decomposition rates (0.22−2.04 mmol/min·g) for various carbon materials (e.g., AC, CB) generally fall short compared to those of metal catalysts, such as loaded Fe catalysts，achieving up to 1.7−2.9 mmol/(min·g) Fe under identical conditions at 850 °C. This inherent activity disadvantage typically necessitates higher reaction temperatures for carbon-based catalysts to match the methane conversion rates, inevitably increasing both energy consumption and process costs.

Solid metal catalysts are crucial in methane catalytic decomposition (CDM) for hydrogen production, as they lower the activation energy required to break C−H bonds, allowing reactions to occur efficiently at relatively low temperature. The intrinsic activity of methane catalytic decomposition follows a specific order: Ni, Co, Ru, Rh > Gt, Pt, Re, Ir > Gt，Pd, Cu, W, Fe, Mo^[14]. Among these, transition metals like Ni, Co and Fe are particularly promising due to their low cost and high activity caused by the unfilled 3d electronic orbitals^[23−24].

The primary challenge for unloaded catalysts is their rapid deactivation caused by metal particle sintering and carbon encapsulation^[25]. Alloying strategies have proven effective in tackling this problem. For instance, Cunha et al.^[26] incorporated Cu into a nickel catalyst to create a nickel-copper alloy. This modification changed the carbon deposition pathway, favoring the formation of carbon nanofibers or multi-walled carbon nanotubes, which effectively prevents the development of carbon inclusions responsible for deactivation. The study by Lua et al.^[27] expanded alloying into the ternary system, creating carrier-free ternary alloy catalysts by thermally decomposing a Ni-Cu-Co oxalate precursor. They discovered that adding Co significantly enhanced the structural stability at high temperature, such as 750 °C, while suppressing the quasi-liquid phenomenon common in Ni-Cu alloys. However, excessive Co content led to phase separation, negatively affecting catalytic activity. Wang et al.^[28] confirmed the benefits of alloying in loaded systems, demonstrating that the Ni-Fe/Al₂O₃ catalyst maintained its activity at 923 K for 210 h of continuous reaction at an optimal Fe doping ratio (Ni/Fe/Al = 2∶1∶1). The carbon yield reached 562 g/g, attributed to the great ability of Ni-Fe alloy phase to balance the diffusion rate of carbon atoms within the metal body phase and the generation rate of carbon atoms at the surface.

Carrier engineering and metal-carrier interactions play a crucial role in catalyst performance, as demonstrated by Lim et al.^[29]. They compared the performance of nickel loaded on dendritic fiber nanosized SiO₂ (KCC-1) with that on conventional Stöber SiO₂. The study revealed that the special morphology of KCC-1 facilitates the formation of more thermodynamically stable 2∶1 Ni-paged silicates with nickel species, thereby causing highly dispersed Ni particles (2.5−4.5 nm) with improved sintering resistance by strong metal-carrier interactions. This strong interaction enhances intrinsic activity, with an initial hydrogen yield of 0.3 mol/(g·h) at 500 °C for Ni/KCC-1, it also intensifies carbon encapsulation during the reaction. This encapsulation alters the growth mode of carbon nanotubes from the typical tip-growth observed on Ni/SiO₂ to base-growth (illustrated in Figure 1(b))^[30]. Similarly, as illustrated in Figure 1(c), Xue et al.^[30] precisely controlled metal-support interaction(MSI) by constructing a core-shell structure. They synthesized SiO₂-loaded Ni nanoparticles and enveloped them with a nanometer-thick TiO₂ shell, forming the embedded core-shell structure as depicted in Figure 1(c)^[31]. This design not only stabilizes the Ni nanoparticles by geometrically confining the domains and preventing their deactivation by carbon nanotubes (CNTs) “pulling out” through the tip growth mechanism, but it also facilitates reactant mass transfer and provides space for CNT growth through the porous TiO₂ shell. In addition, research indicates that carriers with varying properties also can direct the carbon growth path via distinct mechanisms. For instance, when MgO serves as the support, its basic characteristics and lattice structure facilitate the formation of robust interaction phases, such as NiO-MgO solid solutions or magnesium ferrite (Fe₂MgO₄), with active metal precursors (Ni or Fe) at elevated temperatures^[32]. This bonding at the “bulk” scale not only stabilizes metal nanoparticles and inhibits their sintering at high temperatures but also fundamentally predefines the assembly path of carbon atoms. Specifically, multi-walled carbon nanotubes predominantly form on Ni/MgO, while few-layer graphene sheets uniquely develop on Fe/MgO. This observation underscores the capacity of the support to directly dictate the dimensions and macroscopic morphology of carbon products through bulk chemical bonding. In contrast, when Al₂O₃ is employed as the support, its mechanism of action emphasizes “interface engineering”. A systematic study by Yang et al.^[33] demonstrated that in the Ni/Al₂O₃ catalyst, the yield of carbon nanotubes is positively correlated with the presence of a specific β₂-type NiO species, which serves as a strong interaction site associated with the Al-rich mixed oxide phase.

Kutteri et al.^[34] systematically investigated SiO₂-loaded bimetallic catalysts to explore the relationship between alloying and carbon product morphology. They discovered that bimetallic catalysts like 9Ni-1Fe/SiO₂ and 9Ni-1Co/SiO₂ not only reduced metal grain size through the formation of alloying phases such as Ni-Fe and Ni-Co, but also enhanced catalytic performance via electronic effects. This resulted in methane conversion rates of up to 60% and carbon production of up to 2.5 g/g at 650 °C. Significantly, they demonstrated that the Ni-Fe/SiO₂ catalyst predominantly produces fishbone carbon nanotubes with a tip growth mode. In contrast, the Fe/SiO₂ catalyst selectively synthesizes substrate-growing carbon nanotubes at 700 °C, maintaining this growth characteristic even after regeneration. This finding offers a novel approach for catalyst recycling. Additionally, hydrotalcite derived catalysts hold significant promise in CDM research due to the distinctive layered structure and “memory effect”. Sikander et al.^[35] conducted a systematic study on the Mg-Ni-Al hydrotalcite system, revealing that nickel content critically influences performance. At a Ni content of 40% (exLDH-Ni40), the catalyst achieved optimal performance, maintaining approximately 80% methane conversion for up to 6 h at 650 °C. This impressive performance is attributed to the formation of spinel-like structures, such as (Ni, Mg)Al₂O₄, upon calcination at 850 °C. XRD and H₂-TPR characterization confirmed that this structure effectively anchors nickel species, preventing migration and sintering. It also facilitates the uniform growth of carbon nanofibers (CNFs), ensuring that active Ni³⁺ sites remain exposed at the fiber tips to continuously catalyze the reaction. Similarly, the complex role of cobalt in Co_3−xAl_xO₄ (x=0−2) catalysts was explored by Calgaro & Perez-Lopez^[36]. When the Co/Al molar ratio ranges from 1 to 2.3, the catalysts exhibit hydrotalcite-like properties, characterized by a high specific surface area and excellent cobalt dispersion. TPR analysis identified up to six reducible phases, including the Co₃O₄ bulk phase and various Co_3−xAl_xO₄ mixed oxides. The reduction temperatures of these phases increased with higher Al content, indicating stronger interactions between the oxides with the carrier. Although these catalysts show optimal activity and stability at 500 °C, the deactivation pathway is highly dependent on the activation method, where high-temperature reduction in H₂ at 700 °C often leads to sintering, while in situ activation in CH₄ primarily causes deactivation due to the deposition of amorphous carbon.

The performance of metal catalysts depends not only on their composition and supports but also on the dynamically evolving carbon-metal interface that forms during reaction. Recent works have shown that the initial size of metal nanoparticles is a critical structural determinant of the interface and the catalyst’s fate. Atashi et al.^[31] systematically examined Ni/SiO₂ and Ni/Al₂O₃ and found that the mass-specific hydrogen production rate follows a volcano-shaped dependence on Ni particle size, with a maximum when the particle size is near 10 nm (Figure 1(c)). This optimal size reflects control over the carbon growth mode: Ni particles below 10 nm are readily encapsulated by an initial thin layer of amorphous carbon and thus deactivate quickly, becoming “non-productive” particles; particles larger than this threshold act as stable nucleation sites, sustain tip-growth of CNFs and remain “productive” , preserving hydrogen production activity. This size threshold reflects the minimum energetic and geometric requirements for graphite carbon fiber nucleation. Crucially, the carbon-metal interface is dynamic during the reaction. Carbon species from methane cracking actively interact with and restructure the metal catalyst. Operando XRD and DFT calculations show that adsorbed species such as C and CH₃, formed in the reaction’s initial stage, markedly lower the energy barrier for Ni atoms to detach from surface defect sites on nanoparticles. This effect drives carbon-induced, reaction-driven migration of metal atoms and Ostwald ripening, producing pronounced reorganization and growth of catalyst metal particles under working conditions. Thus, the observed catalytic activity arises from the combined influence of the initial metal size distribution and the dynamic sintering behavior during the reaction. The dynamics of the carbon-metal interface extend beyond transition-metal catalysts. Xi et al.^[37] used in situ TEM to show that, during catalytic methane pyrolysis by nanoporous gold, the gold surface continuously disintegrates and releases Au single atoms (Figure 1(d)). These atoms disperse into the forming amorphous carbon matrix and co-catalyze the reaction alongside the host NPG. Further observations show that the dispersed Au single atoms reversibly aggregate into nanoparticles within the carbon layer and then dissociate, establishing a co-catalytic system in dynamic equilibrium of “single atoms ↔ nanoparticles”. DFT calculations indicate that adsorption of reaction intermediates (for example, CH₃*) on Au clusters weakens metal-metal bonds and facilitates this dissociation. The study therefore demonstrates that the carbon product layer can act as a dynamic carrier for metal active sites and that the metal state at the interface (single atoms or nanoparticles) depends on the carbon deposition microenvironment and the reaction conditions.

In summary, the performance of solid metal catalysts is governed by the active-site structure, support properties, alloy composition, and the dynamic carbon-metal interface that forms during reaction. Coke deposition is not merely an end-stage deactivant but constitutes an active interfacial phase. This phase actively reshapes the catalyst in real time and redirects reaction pathways. Controlling the dynamic interface—by stabilizing the initial metal’s size and state through alloying and support design, and by using the carbon layer to tune the dispersion and electronic state of metal sites—is therefore critical for achieving high activity, long-term stability, and selective synthesis of high-value carbon materials.

The mechanism of methane pyrolysis in molten catalysts primarily involves the interaction between bubble dynamics and interfacial reactions. In this process, methane gas is introduced into the molten medium (metal or salt) as bubbles. Subsequently, pyrolysis reaction occurs (CH₄ → C(s) + 2H₂) as these bubbles rise. During the methane pyrolysis process, the bubble-melt interface serves as the primary reaction site, while the gas-phase reaction within the bubble plays a minor role. Numerical simulations by Mehrabian et al.^[39] demonstrate that methane pyrolysis is predominant at this interface by solving the energy and mass conservation equations:where $ r $ is the radial coordinate (m), $ {k}_{{\mathrm{g}}} $ is the thermal conductivity of the gas (W/(m·K)), $ {T}_{{\mathrm{B}}} $ is the temperature within the bubble (K), $ {R}_{{\mathrm{rxn}}} $ is the reaction rate (mol/(m³·s)), $ \Delta {H}_{{\mathrm{rxn}}} $ is the enthalpy of reaction (J/mol), $ \rho $ is the density (kg/m³), $ {c}_{{\mathrm{p}}} $ is the specific heat capacity (J/(kg·K)), $ t $ is time (s), $ {D}_{i,{\mathrm{mix}}} $ is the diffusion coefficient of species $ i $ in the mixture (m²/s), and $ {C}_{i} $ is the concentration of species i (mol/m³).

The calculation results indicate that under typical reaction conditions, over 97% of methane conversion occurs at the bubble-melt interface, with gas-phase reactions inside the bubble contributing less than 3% (Figure 2(b)). This finding underscores the crucial role of molten catalysts such as Ni_0.27Bi_0.73 by providing active sites at the interface and significantly reducing the activation energy, with 208 kJ/mol of E_a at the interface compared to 422.9 kJ/mol inside the bubble^[39].

The size, frequency, and increasing rate of bubbles significantly affect both the residence time of methane and the gas-liquid interface area, thus determining the conversion rate. Sun et al.^[40] utilized the differential pressure transmitter (DPT) technique to analyze bubble characteristics in high-temperature molten tin and LiCl-KCl systems. By comparing bubble diameters measured by DPT with values predicted by empirical correlation equations, such as those by Jamalahmadi and Gaddis-Vogelpohl, they found that the deviation was less than 8% (e.g., approximately 5% in molten tin). This result confirms the reliability of the DPT technique for characterizing bubbles in high-temperature opaque systems. The bubble volume can be calculated using the following formula:where $ {V}_{{\mathrm{B}}} $ is the volume of a single bubble (m³), $ {V}_{{\mathrm{G,Z}}} $ is the gas volume flow rate (m³/s) and $ {f}_{{\mathrm{B}}} $ is the bubble frequency (s⁻¹). The study also highlighted the critical role of gas preheating. Without adequate preheating, the gas temperature during bubble formation falls below the melt temperature, resulting in conversion deviations.

Iguchi et al.^[41] investigated bubble characteristics in a 160 °C molten iron bath using an electrical resistivity probe. Their findings showed that the bubble diameters aligned with the empirical equations proposed by Mori et al.^[42]:where $ {d}_{{\mathrm{B}}} $ is the bubble diameter (m), $ {d}_{{\mathrm{no}}} $ is the nozzle outer diameter (m), $ \sigma $ is the surface tension (N/m), $ {\rho }_{{\mathrm{L}}} $ is the melt density (kg/m³), $ g $ is the gravitational acceleration (9.81 m/s²), and $ {Q}_{{\mathrm{G}}} $ is the gas flow rate (m³/s)^[41]. Experiments indicate that raising the temperature from 125 to 160 °C reduces the bubble frequency while increasing the diameter, thereby influencing the renewal rate of the reaction interface.

In this interfacial reaction-dominated mechanism, operating parameters significantly influence the conversion rate by altering the interfacial area and the transfer process. Among these factors, melt temperature is the most sensitive because its increase exponentially boosts the interfacial reaction rate. Bubble size is also crucial as reducing the initial bubble radius from 2.6 to 0.6 mm can enhance the conversion rate by approximately 12% by increasing the surface area-to-volume ratio and extending the residence time^[39]. Conversely, increasing system pressure from 0.1 to 8 MPa strongly inhibits conversion, causing the final conversion to drop sharply from over 80% to ~20%. This is primarily due to the Le Chatelier principle, as high pressure diminishes gas diffusion to the interface. The intrinsic physical properties of the melt such as surface tension and density fundamentally influence the kinetic behavior of bubbles. Computational fluid dynamics simulations conducted by Okajima et al.^[43] demonstrate that, bubbles are prone to deformation and can even transition into a jet state in NaBr-KBr molten salts that have low surface tension, significantly enlarging the gas-liquid interfacial area. In contrast, in Fe and Al melts, bubbles maintain more regular shapes. These findings indicate that selecting an appropriate melt medium to control bubble dynamics is crucial for optimizing reactor performance.

Methane pyrolysis in molten catalysts primarily relies on interfacial reactions, with bubble kinetics and molten catalysts jointly influencing reaction efficiency. By strategically optimizing gas injection rate, system factors, and melt selection, the interfacial area and residence time can be maximized, and then significantly boosting hydrogen yield.