The Role of the Transition Metal in M₂P (M = Fe, Co, Ni) Phosphides for Methane Activation and C–C Coupling Selectivity

Abstract

Achieving selective, direct conversion of methane into value-added chemicals requires catalysts that can navigate the intrinsic trade-off between C–H bond activation and over-dehydrogenation. Transition metal phosphides (TMPs) have emerged as promising catalysts that can tune this selectivity. This work utilizes density functional theory (DFT) to systematically assess how the transition metal’s identity (M = Fe, Co, Ni) in isostructural M₂P phosphides governs this balance. The findings reveal that the high reactivity of Fe₂P and Co₂P, which facilitates initial methane activation, also promotes facile deep dehydrogenation pathways to coke precursors like CH*. In stark contrast, Ni₂P exhibits a moderated reactivity that kinetically hinders CH* formation while simultaneously exhibiting the lowest activation barrier for the C–C coupling of CH₂* intermediates to form ethylene. This revealed trade-off between the high reactivity of Fe/Co phosphides and the high selectivity of Ni₂P offers a guiding principle for the rational design of advanced bimetallic phosphides for efficient methane upgrading.

1. Introduction

Methane (CH₄), the main component of natural gas, is both a vast C₁ feedstock and a potent greenhouse gas, creating strong motivation for its conversion into value-added chemicals like olefins and aromatics [1,2,3,4,5,6]. Currently, the industrial-scale conversion of methane into liquid fuels and chemicals proceeds primarily through an indirect route. This process involves the initial, energy-intensive reforming of methane to produce synthesis gas (syngas), a mixture of carbon monoxide (CO) and hydrogen (H₂), which is subsequently converted into longer-chain hydrocarbons via established technologies like Fischer–Tropsch synthesis [7,8]. However, the syngas production step requires extreme operating conditions, with temperatures often exceeding 700 °C, which renders the overall process economically costly and entails a substantial energy penalty [9]. Consequently, the direct catalytic conversion of methane in a single step represents a paramount goal in heterogeneous catalysis, offering the potential for simplified process schemes and reduced energy consumption [10].

Direct methane conversion strategies are broadly categorized into oxidative and non-oxidative pathways. The oxidative coupling of methane (OCM) has been extensively investigated but is fundamentally constrained by a persistent challenge: the desired C₂ products (ethane and ethylene) are significantly more reactive than methane itself under oxidizing conditions [11,12]. This leads to the facile over-oxidation of products to thermodynamically favored but undesirable carbon oxides (CO_x), severely limiting the achievable selectivity and yield.

The non-oxidative coupling of methane (NOCM), represented by the reaction:

2CH₄ → C₂H₆ + H₂

(1)

offers a simpler and more carbon-efficient alternative by circumventing the formation of CO_x byproducts [13,14,15,16]. However, the reaction is highly endothermic due to the thermodynamic stability of the methane molecule and the high dissociation energy of its first C–H bond [17], necessitating extreme temperatures to achieve viable conversion rates. These harsh conditions, in turn, promote rapid catalyst deactivation through the formation of coke. The central challenge in NOCM is therefore kinetic: an ideal catalyst must be active enough to cleave the first C–H bond but selective enough to prevent the subsequent, facile dehydrogenation steps that lead to deeply dehydrogenated coke precursors such as CH* and C*.

In this context, transition metal phosphides (TMPs) have emerged as a highly promising class of catalysts. The incorporation of phosphorus into a transition metal lattice induces significant geometric and electronic modifications [18]. Geometrically, P atoms can act as spacers, breaking up large, highly reactive contiguous metal ensembles that are often responsible for deep dehydrogenation and coking on pure metal surfaces. Electronically, charge transfer between the metal and the more electronegative phosphorus atoms modifies the d-band structure of the metal sites, creating a unique bifunctional surface that can be tuned to control catalytic behavior. TMPs are well-established as superior catalysts for hydroprocessing reactions, where they exhibit exceptional selectivity for cleaving C–O, C–S, and C–N bonds, along with high resistance to coking [19,20,21,22,23,24].

In our previous work, we used density functional theory (DFT) to investigate methane activation on the Ni-rich (001) surface of nickel phosphide (Ni₂P) [25]. We found that while Ni₂P was less reactive for the initial C–H bond scission (ΔH_act = 246 kJ mol⁻¹) than pure Ni, this moderated reactivity was key to its selectivity. The presence of phosphorus kinetically hindered the deep dehydrogenation of methane and favored the accumulation of CH₃* and CH₂* intermediates on the surface. The CH₂* species can then be coupled to form ethylene with a relatively low activation barrier compared to further dehydrogenation. These computational findings were consistent with experimental reports showing high selectivity towards C₂ products on silica-supported nickel phosphide catalysts [26,27]. This work established the principle that P-modification can alter reactivity to induce selectivity, but it left open the critical question of how the identity of the transition metal in the M₂P framework affects this balance.

Here, we extend our previous study by using DFT to systematically investigate methane activation and coupling on the (001) surfaces of iron phosphide (Fe₂P) and cobalt phosphide (Co₂P), directly comparing their performance to that of the Ni₂P catalyst. The results reveal a clear trend in reactivity and selectivity across these isostructural catalysts. The findings indicate that Co₂P is the most reactive catalyst, facilitating the initial C–H bond scission with barriers as low as 85 kJ mol⁻¹; however, this high activity is coupled with kinetically facile pathways toward deep dehydrogenation. Conversely, Ni₂P is the least reactive but it exhibits the highest selectivity for ethylene formation by kinetically favoring C–C coupling over C–H scission of the CH₂* intermediate. Fe₂P presents an intermediate case in both reactivity and selectivity. This work elucidates a fundamental reactivity–selectivity trade-off in metal phosphides and suggests that future catalyst design could focus on creating bimetallic phosphide systems that combine the high activity of one metal with the high selectivity of another to optimize performance for direct methane conversion.

2. Results and Discussion

2.1. Optimized Structures and Binding Energies of C₁ Intermediates

This section examines the optimized adsorption geometries and corresponding binding energies (ΔE_ads) of methane-derived intermediates (CH_x*, x = 0–4) and atomic hydrogen (H*) on the isostructural (001) surfaces of Fe₂P and Co₂P. These results are compared with our previous findings for Ni₂P to elucidate trends across the Group 8–10 transition metal phosphides [25].

The DFT-optimized structures and calculated binding energies on Fe₂P and Co₂P are shown in Figure 1. On all three M₂P surfaces, the methane molecule (CH₄*) is weakly physisorbed with a binding energy near 0 kJ mol⁻¹ (Figure 1a,g), indicating minimal interaction with the catalyst surface prior to activation. This is in contrast to the highly reactive oxide surfaces like IrO₂(110), where methane forms a strong dative bond that pre-activates its C–H bond [2,28,29]. The absence of such pre-activation on phosphide surfaces suggests that the initial C–H bond scission is a surface-mediated event requiring the overcoming of a substantial kinetic barrier, rather than a process facilitated by initial molecular destabilization.

Following the initial C–H scission, a notable difference in binding preference emerges compared to our previous work on Ni₂P. On both Fe₂P and Co₂P, all C₁ intermediates (CH₃*, CH₂*, CH*, and C*) preferentially adsorb at high-coordination metallic hollow sites (M₃ or M₄) rather than interacting with surface phosphorus atoms. On the Ni₂P catalyst, however, the less dehydrogenated species CH₃* and CH₂* favored binding to P-containing sites (P₁ and MP, respectively) [25]. These differences in preferred adsorption sites, despite the isostructural nature of the catalyst surfaces examined here, point to a fundamental difference in their electronic properties. The electronic structure of a transition metal phosphide is governed by the interaction between the metal’s d-band and the electron-withdrawing nature of phosphorus. For the more reactive Fe and Co metals, the M₃ and M₄ hollow sites appear to remain the most energetically favorable binding locations for C₁ intermediates, even after modification by phosphorus. In contrast, on the less intrinsically reactive Ni₂P surface, the electronic passivation by phosphorus appears to deactivate the Ni hollow sites to a degree that makes the P-containing sites (P₁ and MP) competitive for the adsorption of CH₃* and CH₂*.

A systematic comparison of the binding energies, visualized in the heatmap in Figure 2, highlights the key trends in intermediate stability. On all three catalysts, the binding strength of CH_x* intermediates increases monotonically with the degree of dehydrogenation (i.e., as x decreases from 3 to 0). This is an expected result, as species with fewer hydrogen atoms have more available electrons to form stronger covalent bonds with the catalyst surface, a trend widely observed in DFT studies of hydrocarbon chemistry on metal surfaces [30,31,32].

More importantly, the results show that the Fe₂P and Co₂P surfaces are significantly more reactive towards C₁ intermediates than the Ni₂P surface. This is evidenced by the consistently stronger (more negative) binding energies on Fe₂P and Co₂P across all CH_x* species. For the key CH₂* intermediate, for instance, the binding energies on Fe₂P (−361 kJ mol⁻¹) and Co₂P (−377 kJ mol⁻¹) are more exothermic than on Ni₂P (−329 kJ mol⁻¹). This reactivity difference becomes even more pronounced for the deeply dehydrogenated C* species, a common coke precursor. The C* atom binds very strongly to the Fe₂P (−644 kJ mol⁻¹) and Co₂P (−667 kJ mol⁻¹) surfaces, whereas its interaction with Ni₂P (−561 kJ mol⁻¹) is substantially weaker. This indicates that the Ni₂P surface is electronically distinct and less reactive, a property that suggests it may be more resistant to coking than Fe₂P and Co₂P.

To better understand the electronic origins of these differences in adsorption behavior, a Bader charge analysis was performed on the clean M₂P(001) surfaces [33,34,35]. The results show that in all cases, the more electronegative phosphorus atoms withdraw electron density from the metal atoms, creating partially positive, Lewis acidic metal sites. The calculated average charge on the surface metal atoms follows the order Fe₂P (+0.53 e⁻) > Co₂P (+0.33 e⁻) > Ni₂P (+0.27 e⁻). This analysis quantifies the electronic modification across the catalysts. The significantly lower positive charge on the Ni atoms indicates a more moderated electronic perturbation by phosphorus compared to Fe and Co. This distinct electronic state likely contributes to the weaker binding of intermediates, the unique preference of CH₃* and CH₂* for P-containing sites on Ni₂P, and ultimately, its superior selectivity for C–C coupling over deep dehydrogenation.

2.2. Methane Dehydrogenation Pathways

Following the analysis of the surface thermodynamics, this section investigates the kinetics of the four successive C–H bond activation steps of methane on the Fe₂P, Co₂P, and Ni₂P surfaces. The activation and reaction enthalpies reported in

Table 1. DFT-predicted forward activation enthalpy (ΔH_act = H_TS − H_reactants) and enthalpy of reaction (ΔH_rxn = H_products − H_reactants) for methane dehydrogenation reactions at 1123 K (kJ mol⁻¹).

No.	Reaction	Fe2P(001)		Co2P(001)		Ni2P(001)
		ΔHact	ΔHrxn	ΔHact	ΔHrxn	ΔHact	ΔHrxn
1	CH4* → CH3* + ½H2(g)	101	60	85	47	246	99
2	CH3* → CH2* + ½H2(g)	52	45	49	42	65	44
3	CH2* → CH* + ½H2(g)	14	9	10	0	84	74
4	CH* → C* + ½H2(g)	99	60	76	61	102	41

are referenced to the most stable, pre-adsorbed configurations of the reactant and product species on the surface. For each C–H scission event, the resulting hydrogen atom is assumed to recombine and desorb from the surface as ½H₂(g). The reactant and product structures illustrated in figures such as Figure 3, however, represent the immediate local minima connected by the transition state, excluding any subsequent diffusion to or from more favorable binding sites.

The initial C–H bond scission in methane is widely recognized as the rate-limiting step for direct methane conversion on most catalyst surfaces, because of the high C–H bond dissociation energy of methane [17,30,31,32]. Our calculations identified highly active pathways on the Fe₂P and Co₂P surfaces that occur exclusively on multi-metal hollow sites for the first C–H scission. The activation barriers on these sites, presented in Table 1, are remarkably low: 85 kJ mol⁻¹ on Co₂P and 101 kJ mol⁻¹ on Fe₂P. These values are not only significantly lower than the barrier on Ni₂P (246 kJ mol⁻¹) but are also comparable to, or even lower than, those on reactive pure metal surfaces [30,31,32]. Consequently, for Fe₂P and Co₂P, the initial C–H activation is no longer the most energy-intensive or rate-limiting step. Instead, the phosphorus atoms appear to moderate the subsequent reaction steps rather than simply passivating the initial activity. The structures for this activation step on Fe₂P and Co₂P, shown in Figure 3, reveal a three-centered transition state involving one metal atom and the breaking C–H bond, a common mechanism observed on various metal surfaces.

Following the initial C–H activation, the kinetics of the subsequent dehydrogenation steps reveal critical differences among these catalysts that govern their selectivity (Table 1). For the second dehydrogenation step to form CH₂*, Co₂P and Fe₂P are again the most active, with barriers of 49 and 52 kJ mol⁻¹, respectively, which are lower than the barrier on Ni₂P (65 kJ mol⁻¹). This reactivity trend remains the same for the dehydrogenation of the CH₂* species. The formation of CH* is kinetically facile on Fe₂P and Co₂P (14 and 10 kJ mol⁻¹, respectively), while the same reaction on Ni₂P requires a high activation barrier of 84 kJ mol⁻¹. The final dehydrogenation step to atomic carbon (C*) exhibits a high activation barrier on all three surfaces, a finding that is consistent with previous studies on pure metal surfaces [30,31,32].

Figure 4 shows the effective enthalpy reaction coordinate diagram used to assess the methane dehydrogenation pathways on all three catalysts. As discussed previously based on Table 1, the initial C–H activation involves a large energetic barrier on Ni₂P compared to Co₂P and Fe₂P. Following this initial step, a clear reactivity trend emerges for the subsequent dehydrogenation reactions, where the overall reactivity follows the order Co₂P > Fe₂P > Ni₂P.

The formation of the CH₂* intermediate from CH₃* proceeds with effective barriers of 96 kJ mol⁻¹ on Co₂P, 112 kJ mol⁻¹ on Fe₂P, and 164 kJ mol⁻¹ on Ni₂P. Once CH₂* is formed, its subsequent dehydrogenation is highly catalyst-dependent. On both Fe₂P and Co₂P, it can readily form CH*, and given that the reverse barriers are also facile while the forward barrier to C* is high, both CH₂* and CH* are likely to be the most abundant surface intermediates. In contrast, the profile for Ni₂P shows a high effective barrier of 227 kJ mol⁻¹ to form CH*, while the reverse barrier to reform CH₂* is only 10 kJ mol⁻¹. This suggests that CH₂* and CH₃* are the most probable surface intermediates on Ni₂P. Furthermore, the effective barrier for C* formation is significantly higher on Ni₂P (319 kJ mol⁻¹) compared to Co₂P (213 kJ mol⁻¹) and Fe₂P (165 kJ mol⁻¹), indicating that it is more resistant to deactivation via coking.

Overall, the reaction coordinate diagram suggests that Ni₂P has a unique potential for controlling selectivity towards ethane or ethylene, while the more active Fe₂P and Co₂P surfaces can readily form CH* species. The possibility of C₂ coupling reactions on these surfaces will be discussed next to better assess the selectivity.

2.3. C–C Coupling Reactions and Selectivity Analysis

To assess the potential of each catalyst for producing higher hydrocarbons, the kinetics of C–C coupling reactions were investigated. The forward activation barriers and reaction energies for the formation of ethane, ethylene, and acetylene are listed in

Table 2. DFT-predicted forward activation enthalpy (ΔH_act = H_TS − H_reactants) and enthalpy of reaction (ΔH_rxn = H_products − H_reactants) for the C–C coupling reactions at 1123 K (kJ mol⁻¹).

No.	Reaction	Fe2P(001)		Co2P(001)		Ni2P(001)
		ΔHact	ΔHrxn	ΔHact	ΔHrxn	ΔHact	ΔHrxn
1	CH3* + CH3* → *CH3CH3*	173	−18	198	0	98	−96
2	CH2* + CH2* → *CH2CH2*	67	−20	105	−18	56	−79
3	CH* + CH* → *CHCH*	107	−2	132	30	56	−133
4	CH3* + CH2* → *CH3CH2*	261	2	270	36	191	−31

. The results indicate that Ni₂P is the most active catalyst for C₂ product formation, exhibiting the lowest activation barriers and highest exothermicities, followed by Fe₂P and then Co₂P. For ethane (*CH₃CH₃*) formation, the activation barrier on Ni₂P is only 98 kJ mol⁻¹, which is significantly lower than on Fe₂P (173 kJ mol⁻¹) and Co₂P (198 kJ mol⁻¹). A similar reactivity trend is observed for ethylene (*CH₂CH₂*) formation, though the barriers are considerably lower than for ethane formation on all catalysts. The reaction pathways for ethylene formation on Fe₂P and Co₂P are shown in Figure 5a,b, where the product is chemisorbed on the surface in a π-bonded configuration. The subsequent desorption of ethylene requires 26 kJ mol⁻¹ on Fe₂P and 52 kJ mol⁻¹ on Co₂P, while it requires only 7 kJ mol⁻¹ on Ni₂P.

The formation of acetylene (*CHCH*) requires a higher barrier than ethylene formation on both Fe₂P (107 kJ mol⁻¹) and Co₂P (132 kJ mol⁻¹) but remains facile on Ni₂P (56 kJ mol⁻¹). The reaction pathways are shown in Figure 5c,d. The acetylene product adsorbs strongly in a 3-fold hollow site, and its desorption is energetically costly, requiring 175 kJ mol⁻¹ on Fe₂P and 192 kJ mol⁻¹ on Co₂P, compared to 100 kJ mol⁻¹ on Ni₂P. The cross-coupling of CH₃* and CH₂* to form an ethyl intermediate was also examined, but this pathway was found to be kinetically unfavorable, exhibiting very high activation barriers on all three catalysts (Table 2).

To better examine the overall selectivity, the activation energy for C–C coupling is compared against the competing C–H bond scission step in the parity plot shown in Figure 6. In this plot, points to the right of the diagonal line indicate a kinetic preference for C–C coupling over further dehydrogenation. The results clearly show that Ni₂P has the best selectivity for C₂ products, as its data points lie in this favorable region. This computational finding is in agreement with experimental studies on silica-supported nickel phosphide catalysts, which have also reported high selectivity towards C₂ products [26,27]. In contrast, Co₂P and Fe₂P are more reactive for C–H activation but exhibit poor selectivity for C₂ coupling, with their data points falling to the left of the line. Fe₂P comes closest to the selective region, with relatively comparable barriers for acetylene formation (107 kJ mol⁻¹) versus C* formation (99 kJ mol⁻¹). While specific experimental data on Fe₂P and Co₂P for methane activation is scarce, this predicted trend of high reactivity but poor C₂ selectivity is consistent with the well-established behavior of iron- and cobalt-based catalysts in related hydrocarbon reactions. These materials are known to be highly active for C–H bond scission but also have a strong tendency to promote deep dehydrogenation pathways that lead to coking, which limits their selectivity for light olefins in non-oxidative conditions [36,37,38]. These findings suggest that future catalyst design could aim to combine the high reactivity of Fe₂P or Co₂P with the superior selectivity of Ni₂P to enhance overall performance in methane activation and upgrading.

3. Computational Methods

All periodic density functional theory (DFT) calculations were performed using the Vienna ab initio simulation package (VASP) as implemented through the Computational Catalysis Interface (CCI) [39,40,41,42,43]. The interactions between ion cores and valence electrons were described using projector augmented-wave (PAW) potentials [44,45]. Exchange and correlation energies were calculated within the generalized gradient approximation (GGA) using the revised Perdew–Burke–Ernzerhof (RPBE) functional [46,47,48]. A plane-wave basis set with an energy cutoff of 396 eV was used for all calculations, and gas-phase molecules were modeled in a periodic 11 × 11 × 11 Å unit cell.

The initial bulk unit cell for Fe₂P, which shares the same space group with Ni₂P ($P\overline{6}2m$), was obtained from crystallographic data and then fully optimized using DFT [49]; its calculated lattice parameters were a = b = 5.81 Å and c = 3.42 Å. The stable bulk phase of Co₂P has the orthorhombic (Pnma) space group, which differs from the hexagonal ($P\overline{6}2m$) space group of Fe₂P and Ni₂P. To specifically isolate the role of the metal’s electronic properties, a theoretical Co₂P model was constructed using the same ($P\overline{6}2m$) framework as the other catalysts. This approach ensures that any observed trends in catalytic activity and selectivity are a direct consequence of the metal’s identity (Fe, Co, or Ni), thereby decoupling these electronic effects from the geometric effects that would arise from comparing different crystal structures. This methodology has been used in previous studies to deconvolute such effects in transition metal phosphides [18]. Moreover, this isostructural approach is also directly relevant to the study of bimetallic systems, where Co atoms could be incorporated into a ($P\overline{6}2m$) host lattice like Ni₂P. The DFT-optimized lattice parameters for this theoretical Co₂P model were a = b = 5.72 Å and c = 3.40 Å. Unlike the non-magnetic Ni₂P system, spin-polarized calculations were employed for both the Fe₂P and Co₂P systems.

Based on our previous analysis for Ni₂P showing the high thermodynamic stability of the M-rich termination of the (001) facet [22], this surface was used to model both catalysts as shown in Figure 7. To further validate this choice, the surface formation energies for the low-index (001), (100), and (111) facets were calculated. These calculations, detailed in the Supporting Information (Table S1), confirm that the (001) facet is indeed the most thermodynamically stable surface for both Fe₂P and Co₂P catalysts. The surfaces were modeled as 2 × 2 slabs consisting of four atomic layers, with a 10 Å vacuum region added in the z-direction. During geometric optimization, the bottom two layers were fixed in their bulk positions, while the top two layers and any adsorbates were allowed to fully relax. A 3 × 3 × 1 Monkhorst–Pack k-point mesh was used for sampling the Brillouin zone [50,51]. All structures were relaxed until the forces on each atom were below 0.05 eV Å⁻¹, with the electronic energy converged to within 10⁻⁶ eV.

Transition state structures were identified using the nudged elastic band (NEB) and dimer methods [52,53,54]. Vibrational frequency calculations were conducted to determine the thermal enthalpy corrections at a temperature of 1123 K. The binding energy of adsorbates (ΔE_ads) was calculated relative to the clean catalyst surface and the corresponding molecule in the gas phase using Equation (2):

$$∆E_{ads}=E_{species/surf}-E_{species\left(g\right)}-E_{surf}$$

(2)

Therefore, a more negative value signifies a more stable, exothermic adsorption. To evaluate the overall reaction kinetics, the effective enthalpy barrier is defined as the enthalpy of forming the transition state and a stoichiometric amount of gas-phase H₂ from gas-phase methane:

$$∆H҂=H҂+\lambda H_{H2\left(g\right)}-H_{CH4\left(g\right)}-H_{surf}$$

(3)

In this equation, λ corresponds to the number of H₂ molecules produced during the dehydrogenation steps. Additional computational details are available in Section S1 of the Supporting Information.

4. Conclusions

In this work, a systematic DFT investigation was conducted to compare methane activation and C–C coupling pathways on the isostructural (001) surfaces of Fe₂P, Co₂P, and Ni₂P. The study aimed to deconvolute the role of the transition metal identity on the catalytic reactivity and selectivity for the non-oxidative coupling of methane.

The results demonstrate a clear trend in thermodynamic surface reactivity towards C₁ intermediates, which follows the order Co₂P > Fe₂P > Ni₂P and correlates with the binding strength of the adsorbates. A new, highly active pathway was identified for the initial C–H scission on purely metallic sites of Fe₂P and Co₂P, with activation barriers (85–101 kJ mol⁻¹) that are comparable to reactive pure metals. The high activity of Fe₂P and Co₂P for the initial C–H scission also leads to facile pathways for subsequent C–H bond breaking, promoting the formation of a mixture of deeply dehydrogenated intermediates. In contrast, the much higher initial activation barrier on Ni₂P makes it significantly less active.

The selectivity of the catalysts is ultimately dictated by the kinetic fate of the CH₂* intermediate, the primary precursor for ethylene. On the Ni₂P surface, a direct comparison of the competing pathways shows that the activation barrier for C–C coupling to ethylene (56 kJ mol⁻¹) is significantly lower than that for the undesired dehydrogenation to CH* (84 kJ mol⁻¹). This favorable kinetic balance, which is not observed on Fe₂P or Co₂P, promotes the selective conversion of surface CH₂* species to C₂ products. Both Fe₂P and Co₂P heavily favor decomposition, as the barriers for CH₂* dehydrogenation (14 and 10 kJ mol⁻¹, respectively) are substantially lower than those for ethylene coupling (67 and 105 kJ mol⁻¹, respectively). These findings are consistent with experimental reports that have demonstrated the high C₂ selectivity of Ni₂P catalysts.

Overall, this work elucidates a fundamental reactivity–selectivity trade-off in this class of materials. While Fe₂P and Co₂P are more active for C–H activation, their high reactivity leads to poor selectivity. Ni₂P, while significantly less active, is an exceptionally selective catalyst for ethylene formation. These findings provide a guiding principle for the rational design of next-generation catalysts, suggesting that bimetallic phosphide systems could be engineered to combine the activity of one metal with the selectivity of another, offering a promising route towards efficient and selective direct methane conversion.