Synergistic Enhancement of Methanol Formation by Edge‐Site FeO _x Clusters and In‐Plane Vacancies on MoS₂ for CO₂ Hydrogenation

ABSTRACT

MoS₂ is a promising catalyst for methanol synthesis from CO₂ hydrogenation. It is widely accepted that the in‐plane sulfur vacancies are the active sites for methanol formation, while the edge sulfur vacancies catalyze methane formation, which is typically undesirable. Rather than blocking the edge sites with heteroatom dopants, we demonstrate that decorating the edges of MoS₂ with functional FeO _x clusters effectively boost methanol formation via synergy between the RWGS activity of the anchored FeO _x clusters and the CO hydrogenation activity of the in‐plane sulfur vacancies. Synthetically, this was achieved by a vacuum impregnation method that chemically anchors FeO _x clusters on the edges of MoS₂, forming stable Fe‐S/Fe‐O interfaces. With these Fe‐S/Fe‐O interfaces, the Fe/MoS₂ catalyst shows markedly enhanced methanol selectivity (to 80%) and a high intrinsic space‐time‐yield (STY) of 0.6 mmol_CH3OH·m⁻²·h⁻¹. In situ spectroscopic and microscopic characterizations combined with theoretical calculations corroborate that the formation of amorphous FeO _x clusters at edge sites of MoS₂ could effectively suppress edge S _v formation, generate additional CO^*, and promote the formation of methanol at the in‐plane S _v sites. Broadly speaking, this work has demonstrated the modulation of MoS₂ edge sites using simple and scalable catalyst preparation methods to enhance methanol formation.

FeO _x clusters anchored onto MoS₂ edges via Fe‐S bonds, stabilizing edge S atoms and inhibiting S vacancy formation. These clusters also catalyze the RWGS reaction to generate CO* intermediates, which migrate to in‐plane S vacancies for further hydrogenation into CH₃OH. This mechanism boosts methanol selectivity and overall reaction efficiency.

1. Introduction

The unrestrained consumption of fossil fuels has led to rapidly accelerating global warming and exacerbating environmental challenges [1]. Mitigating climate change and achieving carbon circularity have thus become paramount challenges in today's energy and environmental sectors. Carbon dioxide, as an abundant and inexpensive carbon feedstock, offers great potential for realizing carbon neutrality through its conversion and utilization. Among various carbon reduction strategies, the catalytic hydrogenation of CO₂ with renewable hydrogen (H₂) to form value‐added chemicals and liquid fuels has been recognized as a promising and sustainable approach, as it simultaneously addresses environmental and energy issues [2, 3, 4]. In particular, the hydrogenation of CO₂ to methanol (CH₃OH) has attracted extensive attention, since methanol possesses a high energy density and can serve not only as a clean fuel but also as a versatile feedstock for the synthesis of olefins, gasoline, and other platform chemicals [5, 6, 7]. Since the concept of the “methanol economy” was proposed by George Olah, CO₂ hydrogenation to methanol has become a research focus in both academia and industry, inspiring the development of highly efficient and durable catalytic systems and offering a promising route toward a low‐carbon energy future [8].

Over the past decades, tremendous research efforts have been devoted to developing highly efficient catalysts for CO₂ hydrogenation to methanol. Representative catalytic systems include Cu‐based catalysts (e.g., Cu/ZnO/Al₂O₃ and Cu/ZrO₂) [9, 10, 11], In₂O₃‐based catalysts (e.g., In₂O₃ and Pt/In₂O₃) [12, 13], as well as solid solutions (e.g., ZnO/ZrO₂ and In₂O₃/ZrO₂) [14, 15]. Among them, the Cu/ZnO/Al₂O₃ catalysts have found applications in industry owing to its high CO₂ conversion and low cost. However, its relatively low methanol selectivity and poor structural stability, mainly caused by the aggregation of active sites, have severely hindered further performance improvement [16]. In contrast, In₂O₃‐based catalysts could achieve nearly 100% methanol selectivity. However, In₂O₃‐based catalysts rely on synergy with noble metals to achieve competitive CO₂ conversion, while their long‐term stability remains challenged by aggregation and over‐reduction of indium species in a reactive atmosphere [17, 18]. Therefore, the design and development of novel catalysts featuring high activity, excellent stability, low cost, and facile synthesis are urgently required to advance CO₂ hydrogenation to methanol toward industrial deployment.

In recent years, two‐dimensional (2D) layered MoS₂ catalysts have attracted considerable attention for their ability to hydrogenate CO₂ to methanol. Hu et al. first reported that few‐layer MoS₂ nanosheets containing abundant in‐plane sulfur vacancies (S_v) exhibited excellent methanol synthesis activity under mild conditions (180°C, 3 MPa) [19]. Both experimental and theoretical studies revealed that, unlike conventional metal oxide catalysts, MoS₂ nanosheets feature tunable layered structures, abundant active sites, and well‐defined structure‐activity relationships [20, 21]. Under reductive environments, two distinct types of S_v can be formed on MoS₂ surfaces, viz. (i) in‐plane sulfur vacancies (in‐plane S_v) which are identified as the key active site for CO₂ hydrogenation to methanol, and (ii) edge sulfur vacancies (edge S_v), which possess stronger hydrogen dissociation capability, and favoring the formation of CH₄ byproduct [22]. Intuitively, increasing the fraction of in‐plane S_v is crucial for improving methanol selectivity.

To achieve this goal, three main strategies have been proposed. The first involves reducing the number of stacking layers of MoS₂ to increase the proportion of in‐plane vacancies relative to edge vacancies. This is usually achieved by intricate exfoliation methods (e.g., mechanical, liquid phase and chemical), deposition methods (e.g., PVD, CVD and ALD), and templated synthesis with surfactants (e.g., CTAB, didodecyldimethylammonium). The second strategy focuses on inducing strains to MoS₂ nanosheets, for example, by host‐guest confinement, such as mesoporous silica‐encapsulation and coating outside of CNTs, thus promoting the formation and exposure of more in‐plane vacancies [23, 24]. However, the above strategies require precise control of the synthesis conditions to achieve the desired morphology, making them challenging to scale up. The third strategy involves decorating transition metals (e.g., Cu [25] and Zn [26]) on the edge sites of MoS₂, thereby suppressing the formation of edge S_v and methane. In turn, the blockage of the edge sites sacrifices the catalyst's capacity to activate CO₂.

Prior studies indicate that MoS₂ catalyzes CO₂ hydrogenation to methanol via a CO‐mediated pathway [23, 27]. Therefore, we hypothesize that functionalizing the edges of MoS₂ with reverse water‐gas‐shift (RWGS) activity would simultaneously hinder methane formation and promote methanol formation. Accordingly, we propose to decorate the edges of MoS₂ with iron oxides, which are well‐known catalysts for RWGS [28, 29, 30]. To this end, we prepared a series of Fe‐modified MoS₂ catalysts. Here, the FeO _x clusters are preferentially deposited at edge sites of MoS₂ nanosheets through Fe‐S bonds. Instead of blocking and deactivating the edge sites, the FeO _x clusters forms unique FeO _x /MoS₂ interfaces with (nominally) O‐Fe‐S‐Mo moieties which serve two functions (i) stabilizing the edge sulfur atoms from forming CH₄‐selective S_v vacancies and (ii) acting as a new type of active sites to form CO^* intermediates, which are subsequently converted to CH₃OH. Using multilayered MoS₂ as a template (i.e., to show that complex synthesis protocol is needed), an optimal loading of 1.5 wt% of Fe affords a high methanol selectivity (80%), an intrinsic methanol STY of 0.6 mmol_CH3OH·m_cat ⁻²·h⁻¹, and stable performance over 200 h time‐on‐stream (TOS). Combining comprehensive characterization and DFT calculations, we also offer an atomic‐level rationalization of the active centers and structure–activity relationships for the excellent performance of the FeO _x /MoS₂ catalyst.

2. Results and Discussion

2.1. Structural Characterization of xFe/MoS₂ Catalysts

A series of xFe/MoS₂ catalysts with different Fe loading (where x represents the weight percentage of Fe in sample with values of 0.1%, 0.5%, 1.5%, 3% and 10%) were prepared by vacuum impregnation (for details see Supporting Information and Table S1). Here, the base catalyst, MoS₂ was prepared by hydrothermal synthesis following previous report [31]. This MoS₂ contains multiple layers of MoS₂ sheets (∼15 layers) and was chosen to represent a generic MoS₂ structure without any additional structural modification. The powder X‐ray diffraction (XRD) results (Figure S1) showed that all the catalyst samples exhibit the characteristic structural features of hexagonal 2H‐MoS₂ and were also structurally unchanged after CO₂ hydrogenation reactions. The strong (002) peak intensity is indicative of the multilayer structure of the MoS₂ [32]. The Fe species on MoS₂ appear highly dispersed and, or amorphous, showing no observable diffraction peak that can be associated with any Fe‐containing species on the as‐prepared catalysts, with the exception of 10Fe/MoS₂, which shows a small Fe⁰ peak after reaction.

Raman spectroscopy (Figure S2) confirmed that all samples retained the 2H‐MoS₂ phase after the reaction. The two characteristic peaks located at 380 and 406 cm⁻¹, corresponding to the in‐plane Mo‐S phonon mode (E¹ _2g) and the out‐of‐plane Mo‐S mode (A_1g) [33], respectively, gradually shifted to higher wavenumbers when the Fe loading reached 1.5%, indicating the intensified interfacial interaction between FeO _x and MoS₂. Based on the above, we regard the 2H‐MoS₂ phase as the active phase of MoS₂ throughout the CO₂ hydrogenation reactions.

The morphology of MoS₂‐based catalysts was characterized by transmission electron microscope (TEM). As expected, the MoS₂ exhibited a nanoflower‐like morphology, with each sheet composed of about 13 stacked atomic layers (Figure S3a,e,f). To gain a clearer understanding of the local structure of Fe‐decorated MoS₂ samples, 0.1Fe/MoS₂, 1.5Fe/MoS₂ and 10Fe/MoS₂ catalysts were further examined by high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM). It can be seen that the deposition of Fe species had little influence on the morphology of the MoS₂ (Figure S3b–d). As depicted in Figure 1a–b, the Fe species on 0.1Fe/MoS₂ surface are nearly atomically dispersed along the edges of the nanosheets. Given the Z‐contrast nature of HAADF‐STEM imaging (I ∝ Z^1.5−2), substitution of Mo (Z = 42) by Fe (Z = 26) is expected to result in a reduced image intensity. However, the Fe‐associated features in Figure 1c–d exhibit an enhanced HAADF contrast. This observation rules out lattice substitution of Mo by Fe. Instead, the Fe species should be anchored on the exterior surfaces or the edge sites of MoS₂. This mode of deposition contrasts the previous reports of Ni or Fe modified MoS₂, which featured the incorporation of Ni and Fe atoms into the MoS₂ lattice [34, 35]. At an Fe loading of 1.5%, the Fe species remain atomically dispersed, while they become more abundant along the edges of MoS₂. (Figure 1f–h). In addition, a small fraction of the Fe clusters begins to aggregate into nanoparticles with size of approximately 4–5 nm (Figure S3g–i). The cluster size increased to ∼11 nm at a Fe loading of 10%. From Figure 1i‐k, it can be seen that amorphous domains of FeO _x became abundant across the entire field of view, suggesting that the Fe content had exceeded the capacity that the MoS₂ edge sites are capable of anchoring. Based on experimental evidence so far, we confirm the successful synthesis of the Fe/MoS₂ catalysts where FeO _x clusters selectively decorate the edges sites of MoS₂, as schematically illustrated in Figure 1l.

FIGURE 1.

(a–c) HAADF‐TEM images, (d) liner intensity profile corresponding to (c), and (e) elemental mapping of the 0.1Fe/MoS₂ catalysts. (f–h) HAADF‐STEM images and elemental mapping of the 1.5Fe/MoS₂ catalysts. (i–k) HAADF‐STEM images and elemental mapping of the 10Fe/MoS₂ catalysts. (l) Schematic illustration of xFe/MoS₂ catalysts with different distributions of Fe species.

Brunauer–Emmett–Teller (BET) analysis of the N₂ adsorption‐desorption isotherms was conducted to investigate the influence of Fe species on the textural properties of MoS₂, as shown in Figure S4 and Table S1. All catalysts exhibited typical type IV isotherms with hysteresis loops, indicating that the mesoporous structure of MoS₂ remained intact after Fe loading. With increasing Fe content, the specific surface area and pore volume gradually decreased from 40.6 m²/g and 0.234 cm³/g to 5.9 m²/g and 0.042 cm³/g, respectively, while the average pore diameter increased from 11.3 to 25.6 nm. It is generally accepted that the incorporation of isolated metal atoms into the MoS₂ lattice does not significantly influence the surface area or pore volume [31, 34]. Therefore, the pronounced reduction of the porosimeteric parameters suggest the selective deposition of FeO _x clusters at interfacial sites. The increased average pore diameter could be ascribed to the growth of FeO _x clusters on the surface.

To further validate the structures of the FeO _x /MoS₂ catalysts prepared by vacuum impregnation, we separately prepared a series of xFe@MoS₂ (x = 0.5, 1.5, 3, 10) samples by a one‐step hydrothermal method as a control group for comparison [34]. In this control group, the Fe precursors were added during the synthesis of MoS₂ and would be incorporated into the MoS₂ matrix. XRD of the xFe@MoS₂ samples (Figure S5a) confirmed that the incorporation of Fe by one‐step hydrothermal synthesis did not disrupt the crystalline structure of the MoS₂. However, the main peak corresponding to the (002) diffraction plane gradually shifted toward higher angles with increasing Fe content (Figure S5b), suggesting that the Fe atoms were substitutionally doped into the MoS₂ lattice by occupying the Mo sites, resulting in lattice contraction. When the Fe loading exceeded 3%, distinct diffraction signals of metallic Fe were detected in the spent catalysts, suggesting that the limit of substitutional doping of Fe into MoS₂ has been reached. Meanwhile, the BET surface areas of the xFe@MoS₂ samples were similar to that of pristine MoS₂ (see Figure S4 and Table S1), confirming that the xFe@MoS₂ samples consist of large amounts of substitutionally doped Fe in the MoS₂ lattice, whereas the xFe/MoS₂ samples consist of FeO _x clusters that are deposited along the edges of MoS₂ sheets.

The electronic states and chemical species of interfacial Fe species were investigated by X‐ray photoelectron spectroscopy (XPS), as shown in Figure 2a and Table S2. Here, we focus on characterizing the spent catalysts as they are expected to be more representative of the working catalysts. In the spent xFe/MoS₂ catalysts, binding energy peaks centered around 707.6 eV, 709.5 eV, and 711.1/714.5 eV were assigned to Fe(II)‐S, Fe(II)‐O, and Fe(III)‐O species, respectively [36, 37, 38]. No detectable Fe signal was observed for 0.1%Fe/MoS₂ because its Fe content is below the detection limit. From Figure 2b, it can be seen that, with increasing Fe loading, the surface fraction of Fe‐S species decreased from 18% to 3% on all Fe‐containing surface species, while the proportion of Fe‐O species gradually increased from 82% to 97%. This trend suggests that the extent of formation of Fe‐S is probably limited by the amount of edge MoS₂ sites available. Further increasing Fe loading would lead to the formation of more FeO _x species. In turn, the decline in Fe‐S intensity corresponds to, on average, weakened interaction between Fe and MoS₂. For the control group, that is, the xFe@MoS₂ catalysts, the 1.5Fe@MoS₂ catalyst exhibited as much as 65% Fe‐S species on its surface, suggesting extensive substitution of Fe atoms into the MoS₂ matrix with high Fe‐S coordination numbers.

FIGURE 2.

(a) Fe 2p spectra of used xFe/MoS₂ catalysts and (b) the corresponding percentages of different Fe species. (c) Room‐temperature ⁵⁷Fe Mössbauer spectra of fresh (bottom) and reduced (top) 1.5Fe/MoS₂ catalysts. (d) Ex situ XANES spectra and (e) EXAFS spectra of different xFe/MoS₂ after reaction, alongside the Fe‐based reference standards. (f) In situ O₂‐DRIFTS spectra of MoS₂, 1.5Fe/MoS₂ and 10Fe/MoS₂ catalysts. (g) Schematic illustration of different MoS₂‐based catalysts.

To further elucidate the chemical state of Fe in 1.5%Fe/MoS₂, ⁵⁷Fe Mössbauer spectroscopy was conducted. As shown in Figure 2c and Table S3, the Mössbauer spectrum of the fresh sample (calcined in N₂ at 400°C) was dominated by a doublet assigned to Fe³⁺ (IS = 0.42 mm/s, 66%), while two additional doublets (IS of ≈ 1.0 mm/s) revealed the presence of Fe²⁺ (34% of all Fe species) in different oxygen coordination [39, 40]. After reduction in H₂, the fraction of Fe³⁺ decreased to 47%, indicating the cleavage of Fe‐O bond. Moreover, a new doublet with an IS value of 0.80 mm/s emerged (21% of all Fe species); this is attributed to Fe²⁺ coordinated with sulfur [41, 42]. Compared with Fe²⁺‐O _x species, the significantly lower IS value in Fe²⁺‐S _x reflected an increased electronic density surrounding the Fe²⁺ cations, in good agreement with the XPS analysis.

X‐ray absorption spectroscopy (XAS) was employed to probe the electronic states and local coordination of the Fe species. According to the Fe K‐edge XANES spectra of the reference samples (Figure 2d), the absorption edge at ∼7120 eV and the white line at ∼7130 eV was attributed to the 1s→4p electron transition of Fe‐S_x and Fe(III)‐O _x species, respectively. For 1.5Fe@MoS₂, the Fe‐S feature was the most pronounced, indicating predominant Fe‐S coordination. In contrast, for the xFe/MoS₂ catalysts, the Fe‐S feature at 7120 eV progressively diminished with increasing Fe loading and nearly disappeared at high loadings, while the Fe‐O feature at 7130 eV continuously intensified, eventually resembling the spectrum of Fe₃O₄. This trend suggests that high Fe loading leads to the formation of FeO _x clusters, which form limited interfaces with the MoS₂. EXAFS analysis of the Fe K‐edge (Figure 2e) further confirmed the above structural interpretation. For 1.5Fe@MoS₂, the dominant peak at 1.8 Å (without phase correction) corresponded to the first‐shell of Fe‐S. With increasing Fe content, this peak gradually shifted toward ∼1.5 Å, corresponding to Fe‐O coordination, with an additional weak Fe–Fe scattering path (at ∼2.2 Å) [43] emerged at Fe loading ≥3% and the most abundant Fe coordination changing from FeS_x to FeO _x . Notably, only a single Fe‐O coordination was detected for 1.5Fe/MoS₂. This result could primarily be attributed to the fact that FeO _x clusters are small and dispersed enough to lack long‐range order beyond one Fe‐O shell. Subsequently, the EXAFS fitting was performed to examine the local coordination structure of Fe in 1.5Fe@MoS₂. As described in Figure S6 and Table S4, each isolated Fe atom bonds with about four S atoms with an average bond length of 2.25 Å.

Sulfur vacancies serve as active sites on MoS₂‐based catalysts. Therefore, it is critical to verify the impact of deposition of FeO _x on the concentration and distribution of S_v on MoS₂. This is characterized by in situ diffuse reflectance infrared Fourier‐transform spectroscopy (DRIFTS) of the catalysts after O₂ oxidation (Figure 2f). After oxidizing pristine MoS₂, Mo = O signal could be detected at wavenumbers of 880‐1000 cm⁻¹. The formation of Mo = O is associated with the oxidation of edge sulfur vacancies (S_v), whereas the Mo‐O‐Mo feature (690‐850 cm⁻¹) is associated with the oxidation of the in‐plane S_v. Here, the Mo = O feature is much more intense than the Mo‐O‐Mo feature, indicating the extensive presence of edge S_v (I _{Mo = O}/I _Mo‐O‐Mo≈3) [19, 24]. In contrast, xFe/MoS₂ catalysts after oxidation exhibited markedly different DRIFTS: with Fe loading as low as 0.1 wt.%, the relative intensity of in‐plane S_v has already decreased substantially (with I _{Mo = O}/I _Mo‐O‐Mo ≈ 0.5). The I _{Mo = O}/I _Mo‐O‐Mo ratio further decreased to 0.2 for 1.5Fe/MoS₂, indicating that edge S_v were almost completely eliminated. For higher Fe loadings (3% and 10%), the characteristic Mo = O signal nearly vanished (Figure S7a,b). This observation corroborates that the FeO _x species, which are deposited along the edges of MoS₂ nanosheets could effectively block the formation of edge S_v. In contrast, if Fe were substitutionally doped into the MoS₂ lattice, for example, in the case of 1.5Fe@MoS₂, the edge S_v remained abundant, with an I _{Mo = O}/I _Mo‐O‐Mo ratio of about 4, even higher than that of pristine MoS₂ (Figure S7c). In line with the previous studies on Pt₁/MoS₂ [31], Rh₁/MoS₂ [44], and Ni₁/MoS₂ [34], we conclude that the substitutional doping of Fe into MoS₂ (i.e. to form xFe@MoS₂) introduced more lattice defects and edge S_v. Through this comparison, we confirm that the preferential deposition of FeO _x nanoclusters (i.e., in xFe/MoS₂) via the formation of Fe‐S bonds along the edges of the MoS₂ nanosheets could effectively mitigate the formation of edge S_v. The distinction between pristine MoS₂ and two modes of Fe doping is schematically illustrated in Figure 2g.

2.2. Catalytic Performance for Methanol Synthesis From CO₂

The CO₂ hydrogenation performances of the xFe/MoS₂ catalysts are shown in Figure 3. Under typical reaction conditions (260°C, 4 MPa, CO₂:H₂:N₂ = 23:69:8 and 30 mL/min), the CO₂ conversion over the xFe/MoS₂ catalysts (4.8%) initially remained independent of Fe loading up until 1.5 wt% Fe, beyond which it declined with Fe loading (Figure 3a and Table S5), reaching 3.3% for 10% Fe/MoS₂. In terms of selectivity, pristine MoS₂ exhibited a poor methanol selectivity (46%), with CH₄ being the dominant product. Such poor selectivity is due to the excessively strong hydrogenation activity of the abundant edge sulfur vacancies [19, 24]. Upon depositing FeO _x clusters, the methanol selectivity increased markedly, reaching a maximum of ∼80% at 1.5 wt% Fe and slightly decreased at higher Fe loading, signifying that the suppression of the edge S_v by the FeO _x species could effectively suppress the formation of CH₄ during CO₂ hydrogenation. We also anticipate the introduction of FeO _x to render additional catalytic activity toward CO* (vide infra), which could be subsequently hydrogenated to methanol. Overall, the methanol yield displayed a volcano‐shaped dependence on Fe loading, peaking at 1.5 wt% Fe with a maximum yield of 3.7% in the present set of reaction conditions. Here, the additional methanol yield (relative to pristine MoS₂ with a methanol yield of 2.2%) is associated with the unique FeO _x /MoS₂ interfaces at the edges (as discussed in detail in the section above). When the Fe loading exceeds 1.5 wt%, the edges sites become saturated, such that the excessive FeO _x nanoclusters start to cover in‐plane active sites. As a result, the overall CO₂ conversion and methanol yield decreased, for example, in the case of 3Fe/MoS₂ and 10Fe/MoS₂ (Figure 3a). Therefore, the simultaneous presence of both FeO _x ‐decorated MoS₂ edges and pristine MoS₂ planes is essential to maximizing methanol formation. In contrast, the substitutional doping of Fe to the MoS₂ lattice, (e.g., the 1.5Fe@MoS₂ catalyst) exhibited deleterious CO₂ conversion, high CH₄ selectivity and low methanol yield (Figure S8), suggesting that the confinement of the FeO _x clusters on the edge sites is the key to improved methanol formation and suppressed methane formation. To further confirm the critical role of the FeO _x /MoS₂ interface, control experiments were performed over pristine 1.5Fe/SiO₂, Fe₃O₄ and Fe₃O₄/MoS₂ catalysts, all of which exhibited significantly lower methanol selectivity and yield compared to 1.5Fe/MoS₂ (Figure S9), unambiguously demonstrating that an intimate Fe‐MoS₂ interface is essential for the enhanced catalytic performance.

FIGURE 3.

(a) Catalytic performance of CO₂ hydrogenation over MoS₂‐based catalysts. Reaction conditions: 260°C, 4 MPa, CO₂:H₂ = 1:3 and WHSV = 9 L·h⁻¹·g⁻¹. (b) CO₂ conversion and product selectivity of the 1.5Fe/MoS₂ catalyst as a function of temperature (200‐260°C). (c) Long‐term stability of the 1.5Fe/MoS₂ catalysts. Reaction conditions: 260°C, 4 MPa, CO₂:H₂ = 1:3 and WHSV = 9 L·h⁻¹·g⁻¹. (d) Comparison of intrinsic methanol STY versus methanol selectivity for 1.5Fe/MoS₂ and representative MoS₂‐based catalysts reported in the literature, as summarized in Table S6. Pink area: different weight hourly space velocities (WHSV = 9‐45 L·h⁻¹·g_cat ⁻¹); Green area: different temperature at WHSV of 45 L·h⁻¹·g_cat ⁻¹.

We further examined the effects of reaction temperature, gas flow rate, and pressure on the performance of the 1.5Fe/MoS₂ catalyst for CO₂ hydrogenation. Since CO₂ hydrogenation to methanol is an exothermic reaction, lower temperatures favor methanol formation at the expense of reduced CO₂ conversion (Figure 3b). As the reaction temperature increased, the CO₂ conversion rose markedly, while the methanol selectivity gradually decreased due to the promotion of side reactions, such as the reverse water‐gas shift reaction (RWGS). The gas flow rate also had a pronounced effect on catalytic performance. When the flow rate increased from 10 to 150 mL·min⁻¹, the CO₂ conversion halved (from 5.8% to 2.6%), whereas the methanol selectivity increased from 74% to 89% (Figure S10a). This improvement can be attributed to the shortened residence time of the reaction intermediates on the catalyst surface, which likely suppressed over‐hydrogenation. In addition, as methanol synthesis involves a decrease in gas‐phase volume, increasing the reaction pressure thermodynamically favors methanol formation. Consequently, raising the pressure from 1 to 40 bar led to a substantial increase in CO₂ conversion, methanol selectivity, and methanol yield (Figure S10b).

After rigorously excluding external and intraparticle diffusion limitations (Figure S11a,b), intrinsic kinetics were measured under differential conversion conditions. Arrhenius analysis (Figure S11c,d) reveals a decrease in apparent activation energy from 54 kJ·mol⁻¹ for MoS₂ to 43 kJ·mol⁻¹ for 1.5Fe/MoS₂. This reduction may partly reflect a Fe‐induced pathway shift, where the FeO _x /MoS₂ interface suppresses deep hydrogenation toward CH₄ and favors methanol formation, consistent with the enhanced methanol selectivity observed under steady‐state conditions.

Stability is also a crucial aspect of catalyst performance. As shown in Figure 3c, the 1.5Fe/MoS₂ catalyst exhibited nearly no deactivation over 200 h time‐on‐stream (TOS) at 260°C, 4 MPa, with a feed of CO₂:H₂ = 1:3 at WHSV of 9 L·h⁻¹·g⁻¹, demonstrating excellent long‐term stability. The superior performance of the 1.5Fe/MoS₂ catalyst was further confirmed by comparing it with previously reported MoS₂‐based catalysts. Given that fewer numbers of MoS₂ layers correspond to higher surface areas and more active sites, we compared the intrinsic activity of the MoS₂ catalysts by normalizing the methanol space‐time yield (STY, in mmol_CH3OH g⁻¹ h⁻¹) against their specific surface area (in m₂ g⁻¹), as shown in Figure 3d. Under various reaction temperatures and gas hourly space velocities, 1.5Fe/MoS₂ consistently delivered high surface area‐normalized STY, reaching a maximum surface area‐normalized methanol STY of 0.6 mmol_CH3OH·m⁻² ·h⁻¹ at 260°C, 4 MPa, CO₂:H₂ = 1:3, and a WHSV of 45 L·h⁻¹·g⁻¹.

To verify the role of vacuum impregnation in promoting selective edge anchoring of Fe species, a control sample prepared under ambient atmosphere (1.5Fe/MoS₂‐Air) was evaluated under identical reaction conditions (Figure S12). All three catalysts exhibited comparable CO₂ conversion, while the methanol selectivity of 1.5Fe/MoS₂‐Air (64%) fell between that of bare MoS₂ (46%) and the vacuum‐impregnated sample (80%). The already elevated selectivity of 1.5Fe/MoS₂‐Air relative to MoS₂ indicates that preferential migration of Fe species toward MoS₂ edge sites is an intrinsic characteristic, driven by the locally negative electrostatic potential at edge sites that electrostatically attracts positively charged Fe³⁺ precursors. Nevertheless, the notably higher methanol selectivity of the vacuum‐impregnated sample suggests that vacuum treatment plays an additional and critical role by removing residual gases (e.g., CO₂, O₂) adsorbed on the support surface and within pore channels, thereby exposing a greater number of accessible edge anchoring sites and enhancing the contact efficiency between Fe precursors and edge sites.

2.3. Operando Investigation of FeO _x ‐MoS₂ Interface

Next, we investigate the structure of the Fe/MoS₂ catalysts in reactive conditions through quasi in situ XPS and in situ XAS measurements. Regardless of Fe loading, the surface Fe species (measured by quasi in situ Fe 2p XPS) evolved similarly with varying gaseous environments (Figure 4a–c and Table S2). On the fresh samples, only Fe(III, II)‐O species were detected. After H₂ reduction at 300°C for 1 h, Fe‐S species emerged and continued to be present when the reduced catalyst was exposed to a gas mixture of CO₂ and H₂ at 260°C, 2 bar [36, 37, 38]. The relative intensity of Fe‐S decreased with increasing Fe content, consistent with the ex situ XPS results. Combining these findings with the spatial distribution of Fe species, we propose that the FeO _x ‐MoS₂ interface has dynamically evolved during preparation: amorphous FeO _x nanoclusters initially anchored on the MoS₂ edges through electrostatic interaction. H₂ reduction cleaved some of the Fe‐O bonds, driving the undercoordinated Fe atoms to subsequently bond with neighboring S atoms form Fe‐S bonds, which stabilized the deposition of FeO _x on the edges of MoS₂, as schematically illustrated in Figure 4f.

FIGURE 4.

Quasi in situ XPS in the Fe 2p region of (a) 0.1Fe/MoS₂, (b) 1.5Fe/MoS₂ and (c) 10Fe/MoS₂ catalysts during H₂ reduction and CO₂ hydrogenation. (d) Normalized XANES spectra and (e) EXAFS spectra of 1.5Fe/MoS₂ after sequential exposure to pure H₂ reduction, CO₂/H₂ mixed gas (CO₂:H₂ = 1:3), and pure CO₂ (260°C). (f) Schematic illustration of Fe‐S/Fe‐O interface evolution in 1.5Fe/MoS₂ catalyst under different reaction conditions. k ²‐weighted EXAFS fitting of 1.5Fe/MoS₂ under different conditions: (g) fresh, (h) H₂ activated at 300°C, and (i) under CO₂/H₂ at 260°C, shown in k and R space.

The structural interpretation above is further corroborated by in situ XAS (Figure S13). As shown in Figure S14a and 4d, the Fe K‐edge XANES of the fresh 1.5Fe/MoS₂ exhibits pre‐edge (7113.2 eV) and white‐line (7132.2 eV) features similar to that of Fe₃O₄, indicating mixed Fe(II)/Fe(III) oxidation states. In other words, the freshly prepared catalyst consists of FeO _x clusters electrostatically adhered on the edges of MoS₂. Upon reduction in hydrogen, with increasing reduction temperature and duration (up to 200 min), the white‐line intensity decreased markedly, and two new absorption features at 7121.2 eV and 7125.3 eV appeared, characteristic of the formation of FeS_x species. Correspondingly, in the R‐space EXAFS, the Fe‐O peak at 1.5 Å (without phase correction) gradually weakened, while the Fe‐S peak at 1.7 Å (without phase correction) intensified and stabilized after ∼1 h of reduction in H₂ (Figure S14b and 4e). The asymmetric peak shape between 1 and 2 Å also confirms the coexistence of Fe‐O and Fe‐S bonds, indicating the formation of an atomically seamless Mo‐S‐Fe‐O interface.

No significant changes were observed in the Fe K‐edge spectra when the gas atmosphere was switched from H₂ to a mixture of CO₂ and H₂ (Figures S15 and 4d), suggesting that the Mo‐S‐Fe‐O interface persists in the working catalyst. When exposed to pure CO₂, the absorption edge and white‐line position shifted toward higher oxidation states, accompanied by intensified Fe‐O (1.5 Å, without phase correction) and Fe‐(O)‐Fe (2.6 Å, without phase correction) peaks, suggesting the partial oxidation of Fe‐S species (Figures S16 and 4d,e). To probe the FeO _x /MoS₂ interfacial structure, EXAFS fitting was performed on 1.5Fe/MoS₂ at fresh, H₂‐activated, and in situ reaction states (Figure 4g–i). In the fresh sample, Fe is solely coordinated with O (CN = 5.6, R = 2.00 Å), resembling Fe₂O₃ [45], confirming FeO _x speciation. After H₂ activation (300°C, 3 h), Fe‐S coordination appears alongside residual Fe‐O, yielding CN(O) = 2.6, CN(S) = 1.3, R(Fe‐O) = 2.07 Å, and R(Fe‐S) = 2.61 Å, consistent with a Fe‐O₂S₂ interfacial coordination structure, which will be further corroborated by DFT calculations. Under CO₂ hydrogenation conditions (260°C, 4 MPa), no appreciable change in coordination environment is observed, demonstrating the robust stability of the Fe‐O₂S₂ structure. Overall, the FeO _x ‐MoS₂ interface formed upon reduction dynamically and reversibly responds to the reaction environments: FeO _x clusters are partially sulfide under reducing conditions and could be reoxidized in CO₂ atmosphere, indicating the presence of a redox‐active FeO _x ‐FeS _x interfacial structure along the edges of MoS₂ nanosheets (Figure 4f).

2.4. Identification of Active Site and Reaction Mechanism

Sulfur vacancies on the MoS₂ surface are known to play a pivotal role in CO₂ activation and hydrogenation [46]. To quantitatively evaluate the influence of Fe loading on surface S vacancies, NO pulse chemisorption measurements were conducted, as NO molecules selectively adsorb on the defective sites [47]. As shown in Figures 5a and S17, the pristine MoS₂ exhibited the highest NO uptake (19.8 µmol·g⁻¹), suggesting the highest number of vacancies. NO uptake sharply decreased with increasing Fe loading (5.9 umol·g⁻¹ for 1.5Fe/MoS₂ and 2.0 µmol· g⁻¹ for 10Fe/ MoS₂), confirming that the formation of the sulfur vacancies was effectively suppressed by the deposition of the FeO _x clusters.

FIGURE 5.

(a) Comparison of NO adsorption capacities of MoS₂, 1.5Fe/MoS₂, and 10Fe/MoS₂ catalysts. (b) H₂‐TPR and (c) CO₂‐TPD profiles of different MoS₂‐based catalysts. (d) Corresponding H₂ consumption and CO₂ desorption amounts, along with two absorbed models (inset). In situ DRIFTS spectra of (e) MoS₂ and (f) 1.5Fe/MoS₂ catalysts in CO₂ hydrogenation after H₂ pretreatment (300°C‐3 h). Test conditions: 260°C, 3 MPa, CO₂:H₂=1:3 and 30 mL·min⁻¹. (g) Normalized intensities of CH₃OH, CH₄ intermediates, and (h) the CH₃OH/CO^* and CH₄/CO^* ratios as a function of reaction time, based on the data in Figure (e–f). (i) CO conversion and product selectivity over MoS₂ and 1.5Fe/MoS₂ catalysts. Reaction conditions: 260°C, 3 MPa, CO:H₂=1:2 and WHSV=9 L·h⁻¹·g⁻¹.

The reducibility of the Fe/MoS₂ catalysts was examined by H₂‐TPR. The TPR profiles can be divided into two regions (Figure 5b): the low‐temperature region (<300°C) corresponding to the removal of labile surface S species [48, 49], and the high‐temperature region (>300°C) associated with the stepwise reduction of the FeO _x species [50, 51]. With increasing Fe loading, the intensity of the low‐temperature peak decreased markedly, indirectly indicating a reduced tendency to form S vacancies due to interfacing with FeO _x clusters (Figure S18 and Table S7), in line with the NO adsorption results. At the same time, the high‐temperature FeO _x reduction features became progressively stronger with increasing Fe loading, demonstrating the presence of partially reduced FeO _x species. In contrast, for 1.5Fe@MoS₂, the H₂‐TPR of sample exhibited only a weak peak corresponding to S removal with nearly no FeO _x reduction signal above 300°C, this is explained by the fact that in 1.5Fe@MoS₂ Fe species were atomically incorporated into the MoS₂ lattice rather than forming surface oxide species.

The CO₂ adsorption behavior of the catalysts was characterized using CO₂‐TPD. All catalysts exhibited desorption peaks at >300°C (Figure 5c), corresponding to chemisorbed CO₂ on medium or strong basic sites, with negligible physisorbed CO₂ [24, 52]. 1.5Fe@MoS₂ showed a CO₂‐TPD profile that is nearly identical to that of MoS₂, indicating that substitutionally doped Fe had little effect on CO₂ adsorption by MoS₂. For xFe/MoS₂ catalysts, although S_v decreased drastically with increasing Fe loading (see Figure 5a), the total CO₂ uptake increased markedly (117→259→610 µmol·g⁻¹, see Table S8). This difference can be explained by the edge‐bound FeO _x nanoclusters providing additional CO₂ adsorption sites (as illustrated in Figure 5d).

In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was carried out to further investigate the catalytic role of the FeO _x /FeS₂ interface under CO₂ hydrogenation conditions. The complete assignments of the main vibrational peaks to the major surface species are detailed in Table S9. Both MoS₂ and 1.5Fe/MoS₂ exhibit similar characteristic absorbed signals, with peaks at 3016, 2955, 2200∼2100, 2077, 1032, 943 and 793 cm⁻¹ corresponding to the typical vibrations of adsorbed CH₄, CH₃O*, CO, CO*, CH₃OH, Mo = O and Mo‐O‐Mo species, respectively (Figure 5e,f) [19, 23, 24]. The surface species observed in the present study are consistent with the literature, also suggesting that the CO₂ hydrogenation mechanism is similar on both MoS₂ and 1.5Fe/MoS₂ [23]. Figure 5g shows the time‐resolved normalized intensities of CH₃OH and CH₄ signals on MoS₂ and 1.5Fe/MoS₂ catalysts. Upon exposure to a 1:3 mixture of CO₂ and H₂ at 260°C and 3 MPa total pressure, both catalysts showed gradually increasing rates of production of both CH₃OH and CH₄. In the case of CH₄, the rates on both catalysts reached a plateau or maximum at around 50 min. Throughout the 120 min of measurements, 1.5Fe/MoS₂ produced more CH₃OH and significantly less CH₄ than MoS₂. This result suggests that the FeO _x /MoS₂ structure not only suppresses the formation of the undesired CH₄ byproduct but also accelerates the formation of methanol, consistent with the packed bed experiments.

On MoS₂, CO* is a key intermediate for methanol formation. Therefore, we examine the formation of CH₃OH and CH₄ in relation to the concentration of CO* on the catalysts, by plotting the ratios of CH₃OH to CO* (CH₃OH/CO*) and CH₄ to CO* (CH₄/CO*) over time on MoS₂ and 1.5Fe/MoS₂, as shown in Figure 5h. Over 120 min, CH₃OH/CO* and CH₄/CO* exhibit opposite trends for the two catalysts. For instance, 1.5Fe/MoS₂ shows consistently higher ratio of CH₃OH/CO*, which also increases more rapidly over time than that on pristine MoS₂, suggesting that the edge‐engineered FeO _x /MoS₂ structure is more efficient in converting CO* to methanol than MoS₂, even though the latter has both in‐plane and edge S_v sites. Meanwhile, the CO* on MoS₂ are more conducive to forming CH₄ than on FeO _x /MoS₂. On the other hand, we have not observed any evidence that the edge‐engineered FeO _x /MoS₂ structure has altered the catalytic pathway for CO₂ hydrogenation, that is, the reaction still proceeds via the RWGS pathway, with additional RWGS activity afforded by FeO _x clusters.

The promotion effect by the FeO _x species was further supported by the in situ DRIFTS experiments performed on the 1.5Fe/MoS₂ catalyst under varying atmosphere at 260°C and 3 MPa. As depicted in Figure S19, distinct CO* signals were detected when only CO₂ gas was introduced for 60 min, confirming that CO₂ can be directly dissociated into CO* and O* on vacancies even in the absence of H₂. Upon switching to a mixed feed of CO₂/H₂, the signals corresponding to methanol (1032 cm⁻¹), CH₃O* (2955 cm⁻¹) and CH₄ (3016 cm⁻¹) increase markedly, while the intensity of CO* decreased sharply, suggesting the facile hydrogenation of CO* by H₂. Moreover, the intensity of CH₃OH was much higher than that of CH₄, suggesting the former is the favored product. Overall, the DRIFTS results fully corroborate the role of the amorphous FeO _x clusters in promoting the transformation of CO₂ to CH₃O* and ultimately CH₃OH, while suppressing the formation of CH₄.

Notably, gaseous CO signals were also observed during the DRIFTS experiments for both MoS₂ and 1.5Fe/MoS₂, with the latter exhibiting much higher intensity and a faster growth over time. This observation indicates that RWGS takes place on both MoS₂ and FeO _x . Interestingly, the intensity of the gaseous CO signal did not reach a steady state but instead decreased after 40 min, suggesting that the CO in the gas phase could be further consumed, for example, by further adsorption and hydrogenation, (e.g., CO→CO*→ CH₃OH). To verify this interpretation, CO hydrogenation on MoS₂ and 1.5Fe/MoS₂ was performed at 260°C and 3 MPa using a CO/H₂ (1:2) feed. As shown in Figure 5i, CO can be activated and converted on both catalysts. However, the bare MoS₂ exhibited poor activity, with only 0.5% CO conversion and 25% methanol selectivity (balanced by selectivity to CH₄). In contrast, 1.5Fe/MoS₂ showed markedly enhanced activity, with the CO conversion increased nearly fivefold (2.3%) and the methanol selectivity reaching 85%. Therefore, the CO hydrogenation experiments demonstrate that the edge‐modification by FeO _x clusters could significantly promote both the RWGS reaction and the subsequent formation of methanol. As such, the above experimental results and discussion have fully explained the origin of the enhancement in CO₂ conversion and methanol yield of the 1.5Fe/MoS₂ catalyst, whose edge S_v sites were replaced by catalytically active and structurally stable FeO _x clusters, which could effectively catalyze RWGS to form CO* and synergize with the in‐plane S_v sites to boost the formation of methanol via the hydrogenation of CO*.

3. DFT Calculations

DFT calculations were performed to provide ab initio rationale to the experimental observations of significant differences between the catalytic properties of pristine MoS₂ and 1.5Fe/MoS₂ samples. According to our characterization, the main structural distinction between these two catalysts is the presence of small Fe species on the edges of MoS₂ in the latter material (Figures 1 and 3). Therefore, we started our computational study by comprehensively characterizing the structure and composition of small Fe clusters on MoS₂ edges. Firstly, the structure Fe/MoS₂ catalyst is computationally validated. To determine the most stable cluster structures under reaction conditions, we evaluated the relative stability of MoS₂ edges with various supported Fe_xO_yS_z clusters compared to separated pristine MoS₂ and magnetite Fe₃O₄. Magnetite Fe₃O₄ was chosen as the Fe reference because both Fe²⁺ and Fe³⁺ were observed experimentally (Figure 2). The cluster stability (normalized per Fe atom) is evaluated assuming that supported Fe_xO_yS_z species form via reduction of Fe₃O₄ and partial desulfurization of the MoS₂ edge in H₂, accompanied by the formation of gas‐phase H₂O and H₂S (see SI). As a result, the thermodynamic stability of the clusters depends on the partial pressures and chemical potentials of H₂O and H₂S in the environment.

Figure 1 indicates that Fe is nearly atomically dispersed along the MoS₂ edge. We therefore focused on clusters containing one or two Fe atoms (x = 1‐2) and various numbers of O and S atoms. For each unique Fe_xO_yS_z composition, multiple initial geometries were examined, and the most stable optimized structures are summarized in Figure S20. Using the normalized formation energies calculated for these structures, we constructed a phase diagram that includes the supported clusters as well as the separated MoS₂ + Fe₃O₄ reference under the experimental conditions (T = 533.15 K, p(H₂) = 20 bar; Figure 6a). The phase diagram spans a range of oxygen and sulfur chemical potentials, corresponding to different environmental p(H₂O) and p(H₂S), as described in Supporting Information. The phase diagram predicts that, under conditions relevant to our CO₂ hydrogenation reaction, the Fe₁O₁S₁ and Fe₂O₂S₂ clusters are more stable than the separated pristine MoS₂ and bulk Fe₃O₄. This result indicates that Fe₃O₄ can be reduced to thermodynamically stable, edge‐supported clusters containing one or two Fe atoms coordinated by O and S atoms. The corresponding atomic structures are shown in Figure 6b.

FIGURE 6.

(a) Phase diagram identifying the most stable Fe_xO_yS_z cluster composition as a function of the O and S chemical potentials at T = 533.15 K and p(H₂) = 20 bar. The dashed rectangle indicates the range relevant to CO₂ hydrogenation conditions, corresponding to p(H₂O) = 1 × 10⁴‐1 × 10⁵ Pa and p(H₂S) = 0.01‐1 Pa. (b) Front and top views of the pristine MoS₂ edge and the edge‐supported Fe₁O₁S₁ and Fe₂O₂S₂ clusters (structures within the dashed rectangle in (a)). The Gibbs free energy of S vacancy formation (G_svf) for each edge S atom is shown to the right of the corresponding structure. Mo, Fe, S, and O atoms are shown in cyan, purple, yellow, and red, respectively. Edge S atoms are highlighted with a more vivid color.

To assess how these clusters influence S removal, we computed the S‐vacancy formation free energy for each S site on the Mo edge for pristine MoS₂ and for MoS₂ hosting the two predicted stable clusters. For pristine MoS₂, S‐vacancy formation is favorable at most edge S sites, as indicated by negative vacancy formation free energies (Figure 6b). In contrast, when Fe₁O₁S₁ or Fe₂O₂S₂ is present on the edge, all edge S sites exhibit positive S‐vacancy formation free energies, demonstrating that the clusters stabilize edge S and suppress vacancy formation. This suppression is significant because Mo‐edge S vacancies have been identified as active sites for CO₂‐to‐CH₄ conversion, and the reduction in their abundance would therefore inhibit CH₄ formation [19]. This trend agrees with our experimental observation that S‐vacancy formation is hindered on xFe/MoS₂ catalysts compared to pristine MoS₂ (Figure 5). Besides, Bader charges on Fe atoms in the clusters are 0.7 e ⁻ less positive compared to the Fe₂O₃ reference (Table S10), indicating partial reduction of Fe in the supported clusters. This is consistent with the experimentally observed decrease in the Fe³⁺ fraction after H₂ reduction (Figure 2c and Table S3).

The catalytic activity of the Fe₁O₁S₁ and Fe₂O₂S₂ clusters toward the reverse water–gas shift (RWGS) reaction is validated by calculating the reaction free energy profiles (Figure 7a). On both clusters, H₂ dissociation occurs at a cooperative site comprising a cluster O atom and a neighboring Mo atom, forming an OH group on the cluster and an H atom adsorbed on Mo (Figure 7a). The calculated activation barriers for H₂ dissociation are 0.43 eV on Fe₁O₁S₁ and 0.26 eV on Fe₂O₂S₂ (Figure 7a). Subsequently, the Mo‐bound H reacts with the OH group via TS2 (Figure 7a) to form H₂O, which desorbs and leaves an O vacancy on the cluster. These O vacancy sites bind CO₂ strongly, with Gibbs free energy changes of ‐0.48 eV (Fe₁O₁S₁) and ‐0.52 eV (Fe₂O₂S₂) after CO₂ adsorption. The adsorbed CO₂ then dissociates with a very small barrier on Fe₁O₁S₁ (0.04 eV) and a somewhat higher barrier on Fe₂O₂S₂ (0.48 eV). In turn, the desorption of produced CO is thermodynamically favorable, with Gibbs free energy change of ‐0.97 eV on Fe₁O₁S₁ and ‐0.46 eV on Fe₂O₂S₂, indicating facile CO release. The CO* on the edge produced in the edge sites can subsequently be released in the gas phase and readsorb on in‐plane S vacancy sites on MoS₂ for further selective hydrogenation to CH₃OH, as demonstrated in Figure S21.

FIGURE 7.

Reaction Gibbs free energy profiles for (a) the reverse water–gas shift (RWGS) reaction and (b) CH₃OH dissociation followed by hydrogenation to CH₄ on the Fe₁O₁S₁ and Fe₂O₂S₂ clusters determined in Figure 6. Insets show top and front views of selected key intermediates and transition states. Mo, Fe, S, O, C, and H atoms are shown in cyan, purple, yellow, red, gray, and white, respectively. Edge S atoms are highlighted with a more vivid color, and atoms directly involved in the reactions are shown with smaller sizes.

On MoS₂, CH₄ formation has been proposed to proceed via CH₃OH dissociation followed by hydrogenation of the resulting methyl species at both edge and in‐plane S vacancy sites [19]. To assess whether the Fe₁O₁S₁ and Fe₂O₂S₂ clusters can also promote CH₄ formation, we calculated the free energy profiles for CH₃OH dissociation and subsequent hydrogenation on these clusters (Figure 7b). The initial C‐O bond scission of CH₃OH to form *CH₃ and *H is kinetically prohibitive, with activation barriers of 1.48 eV on Fe₁O₁S₁ and 1.73 eV on Fe₂O₂S₂, indicating that these supported clusters are unlikely to catalyze the formation of CH₄.

To compare the overall catalytic activity across different reactions and catalysts, we applied the energetic span model (Table 1). In this framework, the turnover frequency (TOF) for a full catalytic cycle follows an Arrhenius‐like expression in which the effective barrier is replaced by the energetic span (δE). δE is defined as the Gibbs free energy difference between the TOF‐determining transition state (TDTS) and the TOF‐determining intermediate (TDI), with an additional term involving the overall reaction free energy when required [53]. The TOF‐determining states and the resulting energetic spans are summarized in Table S11. For Fe₁O₁S₁, the energetic spans for RWGS (δE = 1.86 eV) and CH₄ formation from CH₃OH (δE = 1.83 eV) are comparable. For Fe₂O₂S₂, RWGS is favored over CH₄ formation, with δE = 1.96 eV for RWGS versus 2.21 eV for CH₄ formation (Tables 1, S11). Comparison to the respective values on pristine MoS₂ (Tables 1, S12) reveals the deposition of Fe species on the MoS₂ edges significantly increases their activation barrier in CO₂ methanation (from just 1.09 eV) with much less significant increase in the RWGS barrier (from 1.66 eV) [19]. At the same time, the FeO _x S _y clusters are found to be much more active in catalyzing RWGS than pairs of adjacent in‐plane sulfur vacancies (with δE = 1.97 eV). Thus, our simulations corroborate that in the prepared catalysts, the FeO _x clusters on MoS₂ edges serve as active sites for CO₂ conversion to CO, whereas in‐plane sulfur vacancies further hydrogenate CO₂ and CO to methanol.

TABLE 1.

Energetic span analysis for CH₃OH hydrogenation to CH₄ and RWGS. The analysis is performed based on our calculated reaction free energy profile for Fe₁O₁S₁ and Fe₂O₂S₂ clusters and on the reported reaction free energy profiles in Ref. [19] for pristine MoS₂.

Reaction	Active site	Energetic span δE (eV)
RWGS	Fe1O1S1	1.86
	Fe2O2S2	1.96
	In‐plane Sv on pristine MoS2	1.97
	Edge Sv on pristine MoS2	1.66
CH3OH hydrogenation to CH4	Fe1O1S1	1.83
	Fe2O2S2	2.21
	In‐plane Sv on pristine MoS2	1.69
	Edge Sv on pristine MoS2	1.09

Overall, the DFT calculation results show that the Fe₁O₁S₁/Fe₂O₂S₂ clusters shifts hinder the formation of sulfur vacancies at MoS₂ edges and shift the selectivity of CO₂ hydrogenation from methanation toward CO formation via RWGS pathway. Finally, we benchmarked the RWGS energetic spans of Fe₁O₁S₁ and Fe₂O₂S₂ against reported values for Fe₂O₃ and Fe₃O₄ surfaces [28]. For a fair comparison, our energetic spans were recalculated under the same reaction conditions used in those literature calculations. This analysis shows that the edge‐supported Fe₁O₁S₁ and Fe₂O₂S₂ clusters are more active for RWGS than pristine Fe₂O₃ or Fe₃O₄ surfaces (Table S13).

4. Conclusion

In summary, this work demonstrates that modifying the edges of MoS₂ with catalytically active FeO _x clusters renders synergy that leads to enhanced methanol formation from CO₂ hydrogenation. While unmodified edge sites are detrimentally selective toward methane formation catalysts, the unique FeO _x /MoS₂ interface at MoS₂ edges could effectively suppress the formation of edge sulfur vacancies and alters the CO₂ hydrogenation pathway on these sites from methanation to RWGS. The additional CO formed at the edge sites subsequently migrates to the in‐plane S_v sites to be converted to methanol, rendering a net increase in the overall rate of methanol production and a decrease in selectivity to methane. This mechanistic insight is corroborated by comprehensive in situ characterization, ex situ characterization, and DFT calculations. More importantly, the preparation of the edge‐modified FeO _x /MoS₂ is simple and scalable, making it a practical solution to address the undesirable formation of methane on MoS₂ catalysts during CO₂ hydrogenation.

Author Contributions

Huibo Zhao: conceptualization, methodology, investigation, validation, formal analysis, writing – original draft, data curation. Wenrui Ma: writing – original draft, visualization, software, formal analysis, validation, methodology. Wenjie Liu: investigation, formal analysis, validation. Jiabin Niu: investigation, methodology, formal analysis. Chuande Huang: investigation, methodology, formal analysis, visualization, resources. Yao Wu: investigation, methodology, formal analysis. Shaohui Xiong: investigation, methodology, formal analysis. Mingwu Tan: investigation, methodology, resources. Longgang Tao: investigation, methodology, resources. Qian He: methodology, resources, investigation, supervision. Takeshi Watanabe: investigation, formal analysis, methodology, resources. Takuma Higo: investigation, methodology, formal analysis, resources. Yasushi Sekine: resources, methodology, investigation, supervision. Li Tan: investigation, methodology, resources. Sergey M. Kozlov: methodology, supervision, writing – review and editing, formal analysis, resources. Wen Liu: conceptualization, supervision, funding acquisition, writing – review and editing, resources, project administration.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72844‐sup‐0001‐SuppMat.doc.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.