Regulating Iron Carbide Evolution over CNT-Supported Fe Catalysts by Mn Incorporation for Selective CO Hydrogenation to Linear α-Olefins

Abstract

Linear α-olefins (LAOs) from CO/H₂ represent an attractive non-petroleum route, yet their selective formation over Fe catalysts is often limited by CO₂ formation via water–gas shift (WGS) reaction and by secondary hydrogenation that consumes terminal olefins. In this work, we demonstrate that these competing pathways can be regulated on carbon-nanotube (CNT) supported Fe catalysts by controlling the CNT interfacial oxygen environment through NO treatment or high-temperature annealing and by adjusting the Mn incorporation protocol between co-impregnation and stepwise addition. Under identical reaction conditions at 280 °C and 3.0 MPa with an H₂-to-CO ratio of 1, high-temperature treated CNTs improve olefin preservation and LAO retention compared with NO-treated CNTs. Mn promotion further shifts selectivity toward α-olefins and lowers CO₂ selectivity. At the same Fe-to-Mn ratio, the Mn introduction sequence produces distinct reducibility and CO-binding behaviors that lead to different steady-state oxide and carbide phases. XPS, H₂-TPR, and CO-TPD collectively suggest that CNT pretreatment and the Mn protocol modulate near-surface oxygen speciation, reduction kinetics, and CO adsorption strength. Mössbauer spectroscopy confirms a predominantly χ-Fe₅C₂ population and indicates the presence of ε-Fe₂C in selected samples together with residual oxide and superparamagnetic Fe species. These results highlight the importance of controlling the CNT–metal interface and Mn–Fe proximity to enhance LAO retention under high-temperature CO hydrogenation.

1. Introduction

The selective synthesis of linear α-olefins (LAOs) from syngas offers a compelling non-petroleum route to high-value chemical feedstocks. LAOs are indispensable intermediates for detergents, lubricants, plasticizers, and polymer production [1,2], yet their efficient generation from CO/H₂ remains difficult because syngas conversion typically suffers from broad product distributions and carbon-efficiency losses caused by CO₂ formation and secondary hydrogenation reactions [3,4]. Recent advances have demonstrated that tailoring iron carbide phases can substantially improve LAO productivity and suppress undesired pathways under industrially relevant conditions, underscoring the importance of active-phase control for LAO-oriented Fischer–Tropsch (F–T) chemistry [1,3].

Iron-based catalysts are attractive for syngas conversion due to their low cost and their intrinsic capability for CO activation, while simultaneously catalyzing the water-gas shift (WGS) reaction [3,4]. However, this dual functionality makes product control particularly sensitive to the dynamic evolution of iron phases during activation and reaction. In practice, catalyst performance is often governed not only by bulk composition but also by how iron transforms among oxides and carbides and how these transformations couple with surface adsorption, chain growth, and termination steps [4,5]. As a result, apparently modest changes in synthesis protocols or promoter incorporation can redirect carburization pathways, shift the steady-state carbide population, and ultimately alter olefin selectivity and stability [3,4,5].

It is now well established that iron carbides constitute the catalytically relevant phases in FT-type reactions [5,6]. Among them, χ-Fe₅C₂ (Hägg carbide) is frequently associated with efficient chain growth and hydrocarbon formation, and phase-controlled χ-Fe₅C₂ has recently enabled notably improved LAO production with reduced CO₂ selectivity [1,6,7]. Meanwhile, ε-Fe₂C has been reported to exhibit high intrinsic activity and comparatively low CO₂ selectivity, but it is often metastable under high-temperature FT conditions and can transform into other iron carbides or oxides depending on the local chemical potential and the catalyst microenvironment [8,9,10,11]. Therefore, a central challenge is to steer iron into a desired carbide ensemble and maintain it under realistic reaction conditions.

Manganese is a widely used promoter in iron-based F-T catalysts and is known to influence reducibility, CO activation and adsorption, and product selectivity; nevertheless, its mechanistic role remains debated because Mn incorporation can modify the electronic properties of Fe species and the surface reaction environment in multiple ways [10,11,12,13]. In particular, how the Mn introduction sequence affects the reducibility, adsorption behavior, and subsequent carburization of Fe species remains insufficiently resolved. Carbon nanotubes (CNTs), featuring tunable surface chemistry, high conductivity, and controllable defect density, provide a useful platform to examine such promoter effects while minimizing strong metal-oxide interactions. Prior studies have shown that CNT surface structure and defects can substantially alter Fe carburization behavior and olefin selectivity in CO hydrogenation to olefins catalysis [14,15], suggesting that CNTs can serve as an effective scaffold to probe promoter-controlled phase evolution.

In this work, we systematically examine how Mn loading and Mn introduction sequence govern carburization pathways and iron carbide evolution over CNT-supported Fe catalysts. By combining catalytic evaluation with XRD, TEM, XPS, H₂-TPR, CO-TPD, and Mössbauer, we correlate Mn-promoted changes in reducibility and CO adsorption behavior with the steady-state distribution of iron carbide phases and LAO selectivity. These insights provide a practical basis for the rational design of Fe-based catalysts, in which promoter effects on reducibility, adsorption behavior, and iron carbide formation are collectively considered to achieve selective LAO synthesis from syngas.

2. Results

The catalyst set was designed to decouple the effects of CNT pretreatment, Mn addition, and the Mn introduction sequence. Mn-free reference catalysts were prepared on NO-treated CNTs and high-temperature-treated CNTs, denoted as Fe/MWCNT-NO and Fe/MWCNT-HT. Co-impregnated Fe–Mn catalysts with an Fe-to-Mn ratio of 4 were prepared on both supports, denoted as Fe₄Mn/MWCNT-NO and Fe₄Mn/MWCNT-HT. To isolate the effect of the introduction sequence, two stepwise routes were further prepared on high-temperature-treated CNTs at the same Fe-to-Mn ratio of 4, namely Mn-first Fe₄–Mn/MWCNT-HT and Fe-first Mn–Fe₄/MWCNT-HT. In addition, a lower-Mn co-impregnated sample, Fe₈Mn/MWCNT-HT, was included to separate the effect of Mn loading from the sequence effect.

2.1. Structural Properties

All fresh catalysts display similar XRD patterns (Figure 1a). The reflections at 2θ = 25.7° and 43.7° can be assigned to the CNT support, indicating that the CNT framework remains intact during catalyst preparation [15]. In addition, peaks at 2θ = 30.1°, 35.4°, 43.0°, 57.0°, and 62.5° are indexed to magnetite Fe₃O₄ (JCPDS 87-2334), suggesting that Fe₃O₄ is the dominant crystalline iron phase in the fresh catalysts. No discernible Mn oxide reflections are observed, implying that Mn species are highly dispersed and largely X-ray amorphous or present as interfacial species, and thus are not readily identifiable by bulk XRD. Notably, both NO-treated and HT-treated CNT-supported samples show the same CNT reflections and Fe₃O₄ as the dominant crystalline phase after calcination. Thus, the NO-HT pretreatment effect is not readily resolved by XRD and is instead captured by surface-sensitive probes, including O 1s XPS, discussed below.

The XRD patterns of the spent catalysts are shown in Figure 1b. Besides the CNT reflections, Fe₃O₄ remains the major crystalline phase. Sharp reflections at 2θ = 20.9°, 26.6°, 50.2°, and 60.0° are attributed to α-quartz (SiO₂, JCPDS 83-0539), likely originating from the quartz sand used for dilution packing in the reaction. Two sharp low-angle peaks at 2θ = 21.5° and 23.9° are most plausibly assigned to crystalline long-chain hydrocarbons (wax) retained on the spent catalyst, which is commonly encountered for Fischer–Tropsch products [16]. In the 2θ = 40–50° region, additional reflections consistent with ε-Fe₂C (PDF#36-1249) and χ-Fe₅C₂ (Hägg carbide, PDF#51-0997) are observed, indicating the likely formation of iron carbide phases under reaction conditions. Notably, the characteristic reflections of ε-Fe₂C and χ-Fe₅C₂ partially overlap with each other and with Fe₃O₄ in this region, making phase discrimination by XRD alone challenging [5]. Moreover, XRD alone is insufficient to reliably resolve and quantify individual iron carbide species in the spent catalysts, and Mössbauer spectroscopy is required as a more sensitive and quantitative probe of iron carbides, oxidic, and superparamagnetic (spm) iron components [9,10,11].

High-resolution transmission electron microscopy (HRTEM) was employed to directly probe the local structure of the MWCNT support and the size and distribution of Fe-containing nanoparticles. Representative TEM images are shown in Figure 2. For Fe/MWCNT-HT (Figure 2a), the nanoparticles exhibit a broader size dispersion with occasional aggregates, indicating locally non-uniform deposition. In contrast, Fe₄Mn/MWCNT-HT (Figure 2b) shows a more intact nanotube framework and a more homogeneous dispersion of nanoparticles, with most particles distributed along the inner cavities and the tube-surface and interface region. Statistical analysis based on multiple particles with clearly resolved edges (Figure 2a,b) yields mean particle sizes of 7.1 nm for Fe/MWCNT-HT and 6.5 nm for Fe₄Mn/MWCNT-HT. The slightly reduced particle size and improved dispersion suggest that Mn incorporation suppresses particle growth and promotes a more uniform distribution on CNTs, consistent with CNT confinement and support effects that can regulate the redox and structural evolution of Fe species [16]. Because smaller Fe domains generally provide a higher density of accessible interfacial sites, such size and dispersion changes may facilitate CO adsorption and activation and promote subsequent carburization on CNT-supported Fe catalysts [11].

Post-reaction structural details were further examined for Fe₄Mn/MWCNT-HT. As shown in Figure 3a, some larger particles are observed inside the nanotubes and at the CNT interface, indicating limited sintering and coalescence during reaction. Importantly, these particles remain intimately associated with the CNT framework, often appearing embedded or strongly anchored at interfacial sites. Lattice-fringe analysis was performed for particles located at the CNT interface. A representative HRTEM image (Figure 3b) gives an average lattice spacing of 0.219 nm, which can be indexed to the (202) plane of χ-Fe₅C₂ (Hägg carbide). This assignment is consistent with reported HRTEM lattice fringes of 0.219 nm for χ-Fe₅C₂ (202) [17]. The corresponding SAED pattern further supports the presence of χ-Fe₅C₂, matching the diffraction features expected for this carbide phase (Figure 3b). Given that χ-Fe₅C₂ is widely recognized as a key iron carbide phase responsible for CO dissociation and C-C coupling in Fischer–Tropsch-type chemistry, the direct observation of χ-Fe₅C₂ lattice fringes at the CNT interface supports the formation of catalytically relevant carbide domains under CO hydrogenation conditions [6].

The elemental distribution of the spent catalyst was then assessed by SEM-EDS (Figure 4). At the micrometer scale, the measured Fe/Mn molar ratio is 3.7, close to the nominal value of 4, indicating no obvious macroscopic deviation at the probed surface region. In addition, no obvious macroscopic segregation of C, O, Fe, or Mn is observed, consistent with good overall dispersion. Nanoscale _TEM-EDX mapping (Figure 5) further shows that Mn remains broadly co-distributed with Fe-containing particles while maintaining intimate contact with the CNT framework. Although distinct MnO_x crystallites are not evident in XRD, the persistent Fe-Mn co-localization implies that Mn modifies the local chemical environment of Fe, which can influence reducibility, CO adsorption behavior, and ultimately the carburization pathway and carbide evolution [18].

2.2. Surface Properties

X-ray photoelectron spectroscopy (XPS) was employed to examine the surface chemical states of Fe and Mn in the calcined catalysts and to assess how CNT pretreatment and the Mn incorporation route modify the near-surface oxygen environment prior to reduction and reaction. Figure 6 shows the Fe 2p and Mn 2p regions, while Figure 7 presents the O 1s spectra and the relative fractions of the deconvoluted oxygen species. The Fe 2p_3/2 binding energies and the Fe³⁺/Fe²⁺ ratios derived from peak deconvolution are summarized in

Table 1. Atomic percentages and binding energies of Fe 2p_3/2 of the fresh catalysts.

Catalysts	Fe2+		Fe3+		Fe3+/Fe2+
	Peak (eV)	Area	Peak (eV)	Area
Fe/MWCNT-NO	711.3	3665.5	713.8	3096.5	0.84
Fe4Mn/MWCNT-NO	710.9	6971.1	713.4	5917.5	0.85
Fe4-Mn/MWCNT-HT	711.0	2705.6	713.5	2131.5	0.79
Mn-Fe4/MWCNT-HT	711.1	2495.4	713.5	2123.2	0.85
Fe4Mn/MWCNT-HT	711.0	3340.2	713.5	2896.7	0.87

.

As shown in Figure 6a, all catalysts exhibit Fe 2p features characteristic of mixed-valence iron oxide. The Fe 2p_3/2 envelope is centered at 711 eV, and the Fe 2p_1/2 feature appears at 724–725 eV, accompanied by weak shake-up satellites, consistent with an Fe₃O₄-like surface environment in the calcined catalysts [19]. This assignment agrees with the XRD results, identifying Fe₃O₄ as the major crystalline phase after calcination [20,21]. Within this common spectral envelope, only minor variations are observed. The Fe 2p_3/2 position spans 710.9–711.3 eV (Table 1), indicating that CNT pretreatment and Mn incorporation route slightly modulate the local electronic environment of surface Fe without altering the dominant oxide identity. Peak deconvolution further suggests that surface Fe remains largely mixed-valent across the series (Fe³⁺/Fe²⁺ = 0.79–0.87). Notably, Fe₄-Mn/MWCNT-HT shows the lowest Fe³⁺/Fe²⁺ value (0.79), whereas Fe₄Mn/MWCNT-HT gives the highest value (0.87), implying that the Mn introduction sequence can bias the relative population of Fe²⁺ rich versus Fe³⁺ rich surface sites. Such differences are often linked to variations in reduction kinetics and carburization behavior in Fe-based CO hydrogenation catalysts [22,23].

The Mn 2p spectra (Figure 6b) display Mn 2p_3/2 at 641.4–641.7 eV and Mn 2p_1/2 at 653.2–653.4 eV for all Mn-containing catalysts, with only marginal differences among Fe₄Mn/MWCNT-HT, Mn-Fe₄/MWCNT-HT, Fe₄-Mn/MWCNT-HT, and Fe₄Mn/MWCNT-NO. Given the limited diagnosticity of Mn 2p alone and the absence of Mn 3s in this study, Mn is discussed conservatively as highly dispersed oxidic MnO_x in comparable near-surface environments [21,24]. Together with the lack of distinct Mn oxide reflections in XRD, these spectra indicate that Mn is present predominantly as highly dispersed MnO_x species associated with Fe oxides, likely forming Fe–Mn–O interfacial motifs, rather than as a separate, well-crystallized MnO_x phase, which is consistent with proximity-sensitive promoter effects reported for Fe-Mn catalysts [25,26].

O 1s analysis reveals a clear CNT pretreatment effect (Figure 7). The O 1s spectra can be deconvoluted into lattice oxygen (O_L, 530.0–530.6 eV), defective or adsorbed oxygen-related species (O_A, 531.4–532.0 eV), and hydroxyl-type oxygen (O_OH, 533.3–533.9 eV). Compared with the NO-treated catalysts, HT pretreatment increases the OA fraction from 40–43% to 48–49% while decreasing O_L from 44–48% to 39–41%, with O_OH remaining nearly constant (11–13%). The 531–532 eV contribution is denoted as O_A because it cannot be uniquely assigned to oxygen vacancies based on ex situ O 1s fitting alone [27,28,29]. This redistribution indicates that HT pretreatment primarily modifies the interfacial oxygen environment by enriching defective or adsorbed oxygen-related species rather than increasing hydroxyl coverage. Such changes may influence the heterogeneity of Fe–O bonding; therefore, the reduction behavior and CO adsorption characteristics discussed in the next section [30,31]. XPS confirms that the calcined catalysts expose predominantly Fe₃O₄-like mixed-valence Fe species together with highly dispersed MnO_x species on the surface. CNT pretreatment and the Mn incorporation route slightly tune the Fe electronic environment and more distinctly regulate the balance between O_L and O_A.

2.3. Reducibility and CO Adsorption

The reduction in Fe species and the subsequent CO adsorption and activation on the reduced surface are prerequisite steps for CO hydrogenation, as they determine the accessible Fe sites and influence carburization. Figure 8 compares the H₂-TPR and CO-TPD profiles to examine the effects of CNT pretreatment, Mn promotion, and Mn incorporation sequence.

All catalysts show three broad H₂-consumption regions (Figure 8a). For Fe/MWCNT-NO, the low-temperature feature (~360–420 °C) is attributed to the initial reduction in Fe³⁺ to Fe²⁺ (Fe₃O₄ → FeO), whereas the mid-temperature region (~480–520 °C) corresponds to deeper reduction in FeO_x toward metallic Fe (FeO → Fe⁰) and strongly interacting FeO_x species [32]. The high-temperature region (~630–660 °C) is assigned to the most strongly bound and encapsulated FeO_x species and may overlap with CNT-related processes (such as hydrogenation and gasification of carbon species), which broadens the envelope in CNT-supported catalysts [33]. Mn incorporation shifts the low-temperature region to a lower temperature (typically ~360–380 °C), indicating facilitated initial reduction, while the overall peak shapes remain broad; therefore, we discuss trends using temperature ranges rather than over-precise maxima [34].

The CO-TPD profiles (Figure 8b) are discussed relative to Mn-free baselines (Fe/MWCNT-NO and Fe/MWCNT-HT). All samples show two dominant desorption domains. A lower-temperature peak around 494–501 °C and a higher-temperature peak around 585–598 °C, consistent with CO species bound with different strengths or on different Fe surface ensembles [35]. Mn introduction affects the high-temperature domain primarily. At a fixed Fe/Mn ratio of 4 on HT-treated CNTs, co-impregnation (Fe₄Mn/MWCNT-HT) shifts the high-temperature maximum to a higher temperature than the stepwise routes, which may be consistent with stronger CO binding and activation on Fe–Mn proximity sites [36]. Where available, peak deconvolution and area integration (after baseline correction) can be used to quantify the relative contribution of the 494–501 °C domain versus the 585–598 °C domain and to support a more quantitative discussion of CO binding strength distributions [37].

2.4. Iron Species of the Spent Catalysts

Mössbauer spectroscopy was used to quantify the iron species in the spent catalysts, with particular relevance to iron carbides that can be nanosized or poorly crystalline after CO hydrogenation [38,39]. The spectra and fits are shown in Figure 9a–d, and the phase fractions are summarized in Figure 9e and

Table 2. Mössbauer parameters of the spent catalysts after reaction *.

Catalysts	Phase	Hhf (kOe)	IS (mm/s)	QS (mm/s)	Area (%)
Fe4-Mn/MWCNT-HT	Fe3O4	487	0.32	0.00	19.2
		450	0.51	0.00	24.9
	χ-Fe5C2	221	0.30	0.02	7.3
		181	0.22	0.11	7.6
		100	0.27	0.00	4.3
	ε-Fe2C	170	0.23	0.12	20.0
	Fe3+ (spm)	N/A	0.34	0.99	14.5
	Fe2+ (spm)	N/A	1.15	1.03	2.2
Mn-Fe4/MWCNT-HT	Fe3O4	489	0.29	0	6.3
		457	0.64	0	13.5
	χ-Fe5C2	221	0.3	0.03	6.2
		181	0.18	0.08	6.2
		101	0.17	0.1	3.1
	Fe3+ (spm)	N/A	0.29	0.8	54.3
	Fe2+ (spm)	N/A	1.12	2.4	10.5
Fe4Mn/MWCNT-HT	Fe3O4	489	0.30	0.01	15.2
		460	0.47	0.00	25.0
	χ-Fe5C2	223	0.29	0.13	13.4
		182	0.24	0.13	9.0
		100	0.14	0.14	6.5
	Fe3+ (spm)	N/A	0.35	0.72	10.1
	Fe2+ (spm)	N/A	0.95	0.88	20.8
Fe8Mn/MWCNT-HT	Fe3O4	489	0.31	0.00	15.8
		454	0.58	0.00	19.6
	χ-Fe5C2	225	0.30	0.00	3.5
		180	0.24	0.17	4.3
		100	0.13	0.00	1.8
	ε-Fe2C	168	0.25	0.08	44.8
	Fe3+ (spm)	N/A	0.32	0.90	7.4
	Fe2+ (spm)	N/A	0.65	1.96	2.8

* N/A represents the parameter not available.

.

All spent catalysts contain Fe₃O₄ and carburized iron species dominated by χ-Fe₅C₂, while ε-Fe₂C is additionally observed for selected samples. The persistence of Fe₃O₄ indicates that oxide-to-carbide conversion remains incomplete under the present high-temperature CO hydrogenation environment, which is consistent with concurrent carburization and re-oxidation processes under Fischer–Tropsch-type conditions [40]. The co-presence of Fe₃O₄ and χ-Fe₅C₂ is also reasonable because mixed oxide–carbide motifs, including core–shell-like arrangements, have been reported during carburization and under syngas conversion environments [41,42]. Among the carburized phases, χ-Fe₅C₂ is detected in every sample and remains a key component, consistent with its established role in CO activation and chain growth on iron carbide surfaces [43]. All spent samples were cooled under Ar, purged, and handled using the same transfer procedure. Moreover, Mössbauer spectroscopy is bulk-sensitive, reducing the likelihood that minor near-surface air exposure dominates the quantified phase fractions. Nevertheless, we interpret phase–performance correlations conservatively because very near-surface re-oxidation and wax coverage can still influence accessible active sites. Superparamagnetic (spm) Fe²⁺ and Fe³⁺ doublets in the room-temperature Mössbauer spectra are commonly linked to highly dispersed nanoscale iron species with fast magnetic relaxation [44]. In the absence of operando evidence, the spm contribution is interpreted conservatively and is discussed only in terms of its correlation with the measured phase fractions and catalytic trends.

The phase distribution depends strongly on how Mn is introduced, and this dependence parallels the selectivity trends [45]. The co-impregnated Fe₄Mn/MWCNT-HT exhibits a comparatively carbide-enriched oxide–carbide ensemble, featuring the highest χ-Fe₅C₂ fraction (28.9%) together with a substantial but non-dominant Fe₃O₄ fraction (40.2%). This ensemble is consistent with its catalytic behavior, namely higher α-olefin and LAO selectivities and relatively lower CO₂ selectivity within the studied set. By comparison, the sequence-controlled samples redistribute the oxide and carbide fractions in ways that may be less favorable for preserving terminal olefins. Fe₄-Mn/MWCNT-HT contains lower χ-Fe₅C₂ (19.2%) and a higher Fe₃O₄ fraction (44.1%), along with a notable ε-Fe₂C contribution (20.0%), which correlates with its lower LAO selectivity relative to Fe₄Mn/MWCNT-HT. Mn-Fe₄/MWCNT-HT shows an even lower χ-Fe₅C₂ fraction (15.5%) and a larger fraction of non-carburized or highly dispersed Fe species, suggesting that this incorporation sequence could suppress the establishment of well-defined carbide domains under steady-state conditions.

Lowering Mn at constant total metal loading shifts the carbide identity and the steady-state distribution. For Fe₈Mn/MWCNT-HT, ε-Fe₂C becomes dominant at 44.8% while χ-Fe₅C₂ decreases to 9.6% and Fe₃O₄ remains significant at 35.4%. This shift indicates that the carburization pathway is sensitive to Mn level and the local microenvironment around Fe, consistent with reports that ε-type carbides can be highly reactive yet sensitive to promoter and interfacial effects under Fischer–Tropsch conditions [46]. In the present catalyst set, however, enriching ε-Fe₂C does not translate into the best LAO performance, suggesting that selectivity is governed by the overall oxide–carbide balance and the accessibility of the active carbide ensemble rather than by a single carbide fraction alone.

Overall, Mössbauer analysis supports that co-impregnation on HT-treated CNTs is the most effective route to establish and maintain a carbide-rich steady-state ensemble while restraining the oxide fraction, thereby favoring chain growth and olefin termination and limiting pathways that increase CO₂ formation.

2.5. Catalytic Performance

Catalytic CO hydrogenation was evaluated at 280 °C and 3.0 MPa with H₂/CO/Ar = 4.5/4.5/1 and a GHSV of 6000 mL·g_cat⁻¹·h⁻¹. The catalytic data in Figure 10 and

Table 3. Catalytic performance of the catalysts for CO hydrogenation ^a.

Catalysts b	CO Conv. (%)	CO2 Sel. (%)	Distribution of Hydrocarbons (C%)				α-Olefins Sel. (%)	LAO Sel. (%)	LAOs/O c (%)	α d
			CH4	C2–C3	C4+=	C4+0
Fe/MWCNT-NO	40.1	40.3	6.2	7.7	24.7	61.4	13.7	9.6	39.0	0.86
Fe/MWCNT-HT	20.9	36.5	3.4	4.0	50.9	41.8	40.8	38.1	74.8	0.75
Fe4Mn/MWCNT-NO	18.3	25.1	1.5	2.4	40.2	55.8	36.2	34.3	85.3	0.74
Fe4-Mn/MWCNT-HT	23.4	21.7	11.3	20.0	52.7	16.0	57.6	41.8	79.3	0.58
Mn-Fe4/MWCNT-HT	13.7	14.0	10.6	12.7	50.4	26.3	51.4	42.1	83.5	0.62
Fe4Mn/MWCNT-HT	16.0	20.3	7.9	15.5	59.2	17.4	62.4	49.3	83.3	0.61
Fe8Mn/MWCNT-HT	21.3	24.1	10.2	17.7	56.2	15.9	59.0	44.8	79.7	0.59

^a Conditions. 3 g of catalyst, 280 °C, 3.0 MPa, H₂/CO/Ar = 4.5/4.5/1, 6000 mL⸱g_cat⁻¹⸱h⁻¹. ^b The selectivity of hydrocarbons (including α-Olefins and LAOs) was based on all hydrocarbons excluding CO₂. ^c The LAOs/O is the ratio of linear α-olefins to olefins in C_4–30 hydrocarbons. ^d The chain growth probability α was calculated by C₁–C₃₀ hydrocarbons.

were collected at 108–120 h on stream to represent steady-state performance. Within this catalyst set, selectivity is governed by the competition among chain growth on carburized Fe ensembles, CO₂ formation via WGS-related pathways, and secondary hydrogenation and readsorption processes that erode olefin retention [2,3,4,5,47].

Fe/MWCNT-NO shows the highest CO conversion (40.1%), yet its selectivity pattern is clearly unfavorable for linear α-olefins. CO₂ selectivity reaches 40.3%, and the hydrocarbon slate is paraffin-dominated, resulting in very low α-olefin and LAO selectivities (13.7% and 9.6%) and a low LAOs/O ratio (39.0%). Switching the CNT pretreatment from NO to HT markedly improves olefin preservation even without Mn. Although CO conversion decreases for Fe/MWCNT-HT (20.9%), α-olefin and LAO selectivities increase sharply to 40.8% and 38.1%, and LAOs/O rises to 74.8, indicating that the HT-treated CNT interface suppresses secondary hydrogenation to a substantial extent [15,48].

Introducing Mn further drives the product distribution toward α-olefins and LAOs, while also lowering CO₂ formation compared with Fe-only catalysts. On the NO-treated support, Fe₄Mn/MWCNT-NO reduces CO₂ selectivity to 25.1% and improves α-olefin and LAO selectivities to 36.2% and 34.3%, together with the highest LAOs/O ratio (85.3%). The most attractive overall balance, however, is achieved on the HT-treated support with co-impregnated Fe₄Mn/MWCNT-HT, which delivers the highest α-olefin selectivity (62.4%) and LAO selectivity (49.3%) while maintaining relatively low CO₂ selectivity (20.3%) and a high LAOs/O ratio (83.3%). This combination identifies Fe₄Mn/MWCNT-HT as the best catalyst in this series for LAO-oriented CO hydrogenation under the present high-temperature conditions.

At an identical nominal Fe–Mn composition on HT-treated CNTs, the Mn incorporation sequence also impacts selectivity. Fe₄-Mn/MWCNT-HT exhibits a higher share of light hydrocarbons and the lowest chain-growth probability, whereas Mn-Fe₄/MWCNT-HT achieves the lowest CO₂ selectivity (14.0%) while keeping high α-olefin and LAO selectivities (51.4% and 42.1%) and a high LAOs/O ratio (83.5%). Finally, lowering Mn content at constant total metal loading (Fe₈Mn/MWCNT-HT) does not further improve α-olefin and LAO metrics relative to Fe₄Mn/MWCNT-HT, indicating that sufficient Mn in close proximity to Fe is required to sustain the promotion toward olefin-rich products [12,48].

Benchmarking against representative literature, the performance trends observed here are consistent with the broader understanding that olefin and LAO selectivity in Fe-based CO hydrogenation is highly sensitive to the steady-state carbide ensemble and the local interfacial and promoter environment [49]. For example, phase-pure χ-Fe₅C₂ has enabled efficient syngas-to-LAO conversion with reduced CO₂ formation [1], while Mn tuning of ε-Fe₂C or hydrophobic Fe–Mn interfaces has been shown to suppress C1 by-products and improve overall olefin selectivity [10,49]. In addition, Na-promoted Fe–Zn oxides have been reported to selectively produce linear α-olefins under high-temperature F–T conditions [50], and Zn-promoted precipitated Fe catalysts can enhance LAO yields and stability via modulating the surface carbon chemical potential [51]. Together, these comparisons support that tailoring the CNT–metal interfacial oxygen environment and Mn–Fe proximity offers a complementary route to promote terminal-olefin and LAO retention under high-pressure CO hydrogenation.

3. Discussion

Across this catalyst series, selectivity to α-olefins and LAOs is dictated by how CNT pretreatment and the Mn incorporation protocol steer the reduction–carburization evolution, rather than by nominal composition alone. The key result is that Fe4Mn/MWCNT-HT, prepared by co-impregnation, delivers the highest α-olefin and LAO selectivities while maintaining a comparatively low CO₂ selectivity. This outcome reflects an effective balance between suppressing WGS-related CO₂ formation and limiting secondary hydrogenation or readsorption that erodes terminal olefin retention under high-temperature CO hydrogenation conditions [2,3,4,5,14,15].

CNT pretreatment establishes the support contribution that underpins this balance. Although Fe 2p suggests similar Fe₃O₄-like mixed-valence features for the fresh catalysts, O 1s is more responsive to pretreatment and indicates that HT mainly reshapes the interfacial oxygen environment. This is consistent with modified Fe–O interactions and a changed reduction response, and it coincides with the shift from a paraffin-rich slate on Fe/MWCNT-NO to a markedly more olefin-rich distribution on Fe/MWCNT-HT, even without Mn [14,15].

On the HT-derived interface, the advantage of co-impregnation becomes most evident. Introducing Fe and Mn together favors intimate Fe–Mn contact during oxide formation and calcination, increasing the likelihood that Mn acts through interfacial Fe–Mn–O motifs rather than as spatially separated MnO_x domains. Since Mn promotion is proximity dependent, this structural intimacy is expected to influence reducibility and CO binding and to bias the steady-state competition between carburization and re-oxidation. These effects are consistent with the selectivity pattern of Fe₄Mn/MWCNT-HT, where CO₂ formation remains restrained while α-olefins and LAOs are most strongly enriched [12].

Mössbauer spectroscopy provides a phase-level rationale for the performance differences. All spent catalysts retain Fe₃O₄ together with iron carbides, indicating incomplete oxide-to-carbide conversion under the present CO hydrogenation conditions. Within this common phase landscape, the co-impregnated Fe₄Mn/MWCNT-HT shows the highest χ-Fe₅C₂ fraction among the HT-treated Mn-containing samples, together with a moderated Fe₃O₄ contribution. This oxide–carbide balance is consistent with sustaining carbide domains that support chain growth and olefin termination while restraining oxide-associated pathways that promote CO₂ formation. In contrast, the sequence-controlled catalysts shift the steady-state distribution away from χ-Fe₅C₂, either by retaining more Fe₃O₄ or by allocating a larger fraction of Fe to highly dispersed species, which aligns with their less favorable α-olefin and LAO outcomes [7,43,44,45].

The sequence-controlled samples further highlight that how Mn is introduced can be as important as how much Mn is introduced. At the same nominal Fe/Mn ratio on HT-treated CNTs, the stepwise routes redistribute MnO_x relative to FeO_x, changing Fe–Mn proximity and thereby biasing reducibility, CO adsorption energetics, and the resulting χ-Fe₅C₂ and ε-Fe₂C balance. In parallel, changing Mn loading between such as Fe₈Mn and Fe₄Mn at constant total metal alters the local carburizing environment and can shift carbide identity even for co-impregnated samples. Thus, carbide composition reflects the combined effects of Mn proximity (sequence) and Mn level (loading), rather than sequence alone.

Overall, HT pretreatment and co-impregnation act cooperatively to guide the activation trajectory toward a phase ensemble that enhances α-olefin and LAO retention while limiting CO₂ formation. This identifies CNT interfacial control and the Mn introduction protocol as practical levers for improving selectivity in high-temperature CO hydrogenation [2,3,4,5,14,15]. The interface-related discussion is kept conservative and is based on catalyst-level evidence and internally consistent trends in activity and selectivity. Direct characterization of the CNT supports before and after treatment would further strengthen the mechanistic interpretation. Future work will include dedicated support characterization and quantitative impurity analysis to validate the role of CNT oxygen environments in governing Fe–Mn proximity and carbide evolution.

4. Materials and Methods

4.1. Materials and Catalyst Preparation

Multi-walled carbon nanotubes (MWCNTs, length 5–15 μm, diameter 10–40 nm) were supplied by Naco New Materials Company (Jiaxing, China). Ammonium ferric citrate (FAC), manganese(II) nitrate tetrahydrate (Mn(NO₃)₂·4H₂O), nitric acid (HNO₃), and other chemicals were purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). All reagents were used as received without further purification. Deionized water was used throughout.

MWCNTs were pretreated prior to metal loading. The raw MWCNTs were purified and oxidized by refluxing in 8 M HNO₃ at 110 °C for 4 h under stirring to remove residual metal species and introduce oxygenated anchoring sites. After cooling, the solids were washed with deionized water until the filtrate reached neutral pH (pH ≈ 7), filtered, and dried at 60 °C overnight. The obtained nitric-acid-treated CNTs were denoted as MWCNT-NO. In this work, the effects of CNT pretreatments are discussed primarily based on catalyst-level surface chemistry (XPS), reducibility (H₂-TPR), CO adsorption (CO-TPD), and catalytic performance.

MWCNT-HT was obtained by further Ar annealing of MWCNT-NO to modify the CNT surface prior to metal loading. In this case, the MWCNTs-NO sample was further heat-treated in a tubular furnace under flowing Ar (100 mL·min⁻¹) to tune defect density and remove unstable oxygenated groups. The temperature was increased to 850 °C at 5 °C·min⁻¹ and maintained for 2 h, then cooled to room temperature under Ar. The resulting sample was labeled MWCNT-HT.

FAC was used as the Fe precursor. CNT-supported Fe catalysts were prepared by incipient wetness impregnation to obtain a nominal total metal loading of 30 wt%. An aqueous FAC solution was added dropwise onto MWCNT-NO or MWCNT-HT, followed by ultrasonic dispersion and aging at room temperature for several hours. The samples were dried at 60 °C overnight. The dried powders were calcined under Ar at 500 °C for 2 h with a heating rate of 2 °C·min⁻¹, yielding Fe/MWCNT-NO and Fe/MWCNT-HT.

For Mn-modified catalysts prepared by co-impregnation, FAC and Mn(NO₃)₂·4H₂O were simultaneously dissolved in deionized water at a Fe: Mn molar ratio of 4:1, while maintaining the total metal loading at 30 wt%. The mixed solution was impregnated onto CNT supports following the same drying and calcination procedures, producing Fe₄Mn/MWCNT-NO and Fe₄Mn/MWCNT-HT.

To investigate the effect of Mn introduction sequence, catalysts with identical Fe: Mn molar ratios (4:1) and total metal loadings (30 wt%) were prepared by stepwise impregnation on MWCNT-HT. In the “Mn-first” route, MWCNT-HT was impregnated with manganese nitrate solution, dried, then impregnated with FAC solution and dried again. the resulting sample was labeled Fe₄–Mn/MWCNT-HT. In the “Fe-first” route, MWCNT-HT was first impregnated with FAC solution, dried, then impregnated with manganese nitrate solution and dried again. The resulting sample was labeled Mn–Fe₄/MWCNT-HT. Each impregnation step was followed by drying prior to the subsequent metal addition. For comparison, the co-impregnated Fe₄Mn/MWCNT-HT catalyst was also prepared. All stepwise and co-impregnated samples were finally calcined together under Ar at 500 °C for 2 h.

4.2. Catalyst Characterization

Powder X-ray diffraction (XRD) patterns of fresh and spent catalysts were collected on a Bruker D8 Advance diffractometer using Cu Kα radiation (40 kV, 40 mA) to identify crystalline phases.

The morphology and microstructure of the catalysts were examined by scanning electron microscopy (SEM), and elemental distributions were obtained by energy-dispersive X-ray spectroscopy (EDS) using a Helios 5 UC microscope (Thermo Fisher Scientific, Hillsboro, OR, USA) equipped with an Ultim Max 65 detector (Oxford Instruments, High Wycombe, UK). Transmission electron microscopy (TEM) observations were performed on a JEM-2100F microscope (JEOL Ltd., Tokyo, Japan) operated at 200 kV to evaluate particle size and dispersion. Particle sizes were measured from HRTEM images using ImageJ software (v1.53k, ≥100 particles per sample). Only particles with clearly resolved boundaries were included, and results are reported as mean ± standard deviation.

X-ray photoelectron spectroscopy (XPS) measurements were carried out on a Thermo Scientific ESCALAB 250Xi spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) using Al Kα radiation to analyze surface chemical states. Binding energies were calibrated using the C 1s peak at 284.8 eV as reference. Fe 2p, Mn 2p, and O 1s spectra were deconvoluted after Shirley background subtraction.

H₂ temperature-programmed reduction (H₂-TPR) and CO temperature-programmed desorption (CO-TPD) experiments were conducted on a Micromeritics AutoChem II 2930 analyzer (Micromeritics Instrument Corporation, Norcross, GA, USA). For H₂-TPR, 50 mg of sample was heated from room temperature to 800 °C at 10 °C·min⁻¹ under 5% H₂/Ar (30 mL·min⁻¹). For CO-TPD, the samples were first reduced in situ under H₂ at 350 °C for 1 h, followed by CO adsorption using 5% CO/He at 50 °C. After purging with He to remove physisorbed CO, desorption profiles were recorded by heating to 800 °C at 10 °C·min⁻¹.

Mössbauer spectra of the spent catalysts were recorded at room temperature using a conventional constant-acceleration spectrometer with a ⁵⁷Co/Rh source. Spectra were fitted to quantify iron species, including Fe₃O₄, χ-Fe₅C₂, ε-Fe₂C, and spm Fe²⁺/Fe³⁺ components.

4.3. Catalyst Performance Evaluation

CO hydrogenation reactions were carried out in a fixed-bed stainless-steel reactor. Typically, 3.0 g of catalyst (20–40 mesh) was thoroughly mixed with an equal mass of quartz sand and loaded into the isothermal zone of the reactor. Prior to reaction, the catalyst was reduced in situ under a syngas atmosphere (H₂/CO/Ar = 6/3/1) at 280 °C and atmospheric pressure with a space velocity of 2000 mL·g_cat⁻¹·h⁻¹ for 8 h. After reduction, the reactor was purged with Ar and then pressurized to the desired reaction pressure.

All CO hydrogenation tests were conducted at 280 °C and 3.0 MPa with a feed gas composition of H₂/CO/Ar = 4.5/4.5/1 (molar ratio), containing 10 vol% Ar as an internal standard. The total gas hourly space velocity (GHSV) was maintained at 6000 mL·g_cat⁻¹·h⁻¹. The reactor effluent was analyzed online using a gas chromatograph equipped with a thermal conductivity detector (TCD) for permanent gases (H₂, CO, and CO₂) and a flame ionization detector (FID) for hydrocarbons. CO conversion and product selectivities were calculated on a carbon basis.

The olefin-to-paraffin (O/P) ratio was defined as the molar ratio of total C₂₊ olefins to the corresponding paraffins. The selectivity of linear α-olefins (LAOs/O) was defined as the fraction of terminal α-olefins in C₄₊ hydrocarbons relative to total olefins. Each catalytic test was performed until steady-state operation was achieved. After reaction, the catalysts were purged with Ar and cooled to room temperature prior to post-reaction characterization.

CO conversion (${\mathrm{X}}_{\mathrm{CO}}$), CO₂ selectivity (${\mathrm{S}}_{{\mathrm{CO}}_2}$), and hydrocarbon selectivities (${\mathrm{S}}_{{\mathrm{C}}_{\mathrm{i}}{\mathrm{H}}_{\mathrm{m}}}$) were calculated according to Equations (1)–(3).

$${\mathrm{X}}_{\mathrm{CO}} (\%)={\displaystyle \frac{{\mathrm{N}}_{\mathrm{CO}}^{\mathrm{in}}-{\mathrm{N}}_{\mathrm{CO}}^{\mathrm{out}}}{{\mathrm{N}}_{\mathrm{CO}}^{\mathrm{in}}}} \times 100\%,$$

(1)

$${\mathrm{S}}_{{\mathrm{CO}}_2} (\%)={\displaystyle \frac{{\mathrm{N}}_{\mathrm{C}{\mathrm{O}}_2}^{\mathrm{out}}}{{\mathrm{N}}_{\mathrm{CO}}^{\mathrm{in}}-{\mathrm{N}}_{\mathrm{CO}}^{\mathrm{out}}}} \times 100\%,$$

(2)

$${\mathrm{S}}_{{\mathrm{C}}_{\mathrm{i}}{\mathrm{H}}_{\mathrm{m}}} (\%)={\displaystyle \frac{\mathrm{i}{\mathrm{N}}_{{\mathrm{C}}_{\mathrm{i}}{\mathrm{H}}_{\mathrm{m}}}^{\mathrm{out}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\mathrm{i}{\mathrm{N}}_{{\mathrm{C}}_{\mathrm{i}}{\mathrm{H}}_{\mathrm{m}}}^{\mathrm{out}}}} \times 100\%,$$

(3)

where ${\mathrm{N}}_{\mathrm{CO}}^{\mathrm{in}}$, ${\mathrm{N}}_{\mathrm{CO}}^{\mathrm{out}}$, and ${\mathrm{N}}_{\mathrm{C}{\mathrm{O}}_2}^{\mathrm{out}}$ represent the molar flow rates of CO and CO₂ at the reactor inlet and outlet, respectively.

5. Conclusions

CNT pretreatment and the Mn incorporation protocol are key variables for steering olefin and LAO selectivity in high-temperature CO hydrogenation over CNT-supported Fe catalysts. Switching from NO-treated to HT-treated CNTs improved olefin preservation (such as LAOs/O increased from 39.0% on Fe/MWCNT-NO to 74.8% on Fe/MWCNT-HT) by suppressing secondary hydrogenation and readsorption. Beyond promoter loading, the Mn introduction protocol strongly affected CO₂ selectivity and LAO retention at the same nominal Fe/Mn ratio. Among all samples, the co-impregnated Fe₄Mn/MWCNT-HT achieved the most favorable balance at steady state (108–120 h), combining CO conversion of 16.0% with CO₂ selectivity of 20.3%, the highest α-olefin selectivity of 62.4%, and the highest LAO selectivity of 49.3% (LAOs/O = 83.3%). Mössbauer analysis confirms that these trends track changes in the steady-state oxide and carbide/spm ensemble, with co-impregnation on HT-treated CNTs favoring a higher fraction of catalytically relevant χ-Fe₅C₂ while maintaining a moderated residual oxide fraction. Overall, regulating the CNT interfacial environment together with controlled Mn–Fe proximity provides a practical route to suppress CO₂ formation and enhance LAO retention under high-temperature CO hydrogenation.