Surface Chemical Effects on Fischer–Tropsch Iron Oxide Catalysts Caused by Alkali Ion (Li, Na, K, Cs) Doping

Abstract

Among the alkali metals, potassium is known to significantly shift selectivity toward value-added, heavier alkanes and olefins in iron-based Fischer–Tropsch synthesis catalysts. The aim of the present contribution is to shed light on the mechanism of action of alkaline promoters through a systematic study of the structure–reactivity relationships of a series of Fe oxide FTS catalysts promoted with Group I (Li, Na, K, Cs) alkali elements. Reactivity data are compared to structural data based on in situ, synchrotron-based XRD and XPS, as well as temperature-programmed studies (TPR-H₂, TPC-CO, TPD-CO₂, and TPD-H). It has been observed that the alkali elements induced higher carburization rates, higher basicities, and lower adsorbed hydrogen coverages. Catalyst stability followed the trend Na-Fe > unpromoted > Li-Fe > K-Fe > Cs-Fe, being consistent with the ability of the alkali (Na) to prevent active site loss by catalyst reoxidation. Potassium was the most active in promoting high α hydrocarbon formation. It is active enough to promote CO dissociative adsorption (and the formation of FeCx active phases) and decrease the surface coverage of H-adsorbed species, but it is not so active as to cause premature catalyst deactivation by the formation of a carbon layer resulting in the blocking active sites.

1. Introduction

Fischer–Tropsch synthesis (FTS) is the catalytic hydrogenation of CO toward a mixture of hydrocarbons, according to Equations (1) and (2).

nCO + (2n + 1)H₂ → C_nH_{(2n + 2)} + nH₂O

(1)

nCO + 2nH₂ → C_nH_2n + nH₂O

(2)

The products of FTS may be used as high-value-added sulfur-free transportation fuels. Among the Group 7 and 8 metals (Fe, Ru, Os, Co, Rh, Ni) that exhibit catalytic activity for FTS, Co and, particularly, Fe are typically selected due to their relatively low cost, high activity, and favorable selectivity. Several substances have been added to Fe-based catalysts in order to improve performance. In such an effort, the effect of lanthanides [1], chalcogenides [2,3], transition metals [4,5,6,7], group 2 [8,9,10,11], and mainly, group 1 [12,13,14,15,16,17] elements have been studied. In this sense, while Si and Mn have been considered physical promoters, as they contribute to the formation and stabilization of a nanosized iron oxide/carbide phase in the active catalyst [18,19], Cu and K have been regarded as chemical promoters, as they increase catalyst active site density by facilitating its reduction/carburization during the activation step [14,20]. Besides facilitating carburization (i.e., Fe carbide is proposed to be the active phase), small amounts of K also suppress methanation and increase valuable heavy wax productivity [20,21]. K has also been found to promote Fe carbide activity for other important catalytic reactions, such as ammonia synthesis [22] and alkylbenzene hydrogenation [23,24,25,26,27,28]. In the former case, K has been proposed to exert an electron donor effect facilitating the dissociative adsorption of molecular nitrogen [22]. In iron oxide-based alkylbenzene dehydrogenation catalysts, where the active phase is found to be comprised of a mixture of iron suboxides, the presence of small amounts of K is also found to exert an important promoter effect [23]. This has been correlated to the formation of active sites based either on potassium–iron oxides [24,25,26] or on carbon-based functional groups [27,28,29]. Regarding Fe-based FTS catalysts, the active catalyst is comprised of a mixture of partially reduced, non-stoichiometric iron oxides and a complex mixture of iron carbides, from which the Hägg carbide (χ−Fe₅C₂) has been regarded as one of the most important FTS active sites [30]. The role of potassium is proposed to be due to (i) suppressing H₂ dissociative adsorption while increasing CO dissociative adsorption [12,13,31] and (ii) increasing the fraction of iron carbides, which is believed to be the primary active phase for FTS [32]. Regarding the Fe-K catalytic system for FTS, the alkaline promoter is proposed to control hydrocarbon chain growth and product distribution via an electron donor effect based on the Dewar-Chatt-Duncanson complex, which is assumed to be formed between surface iron atoms and hydrocarbon adsorbates involved in chain growth [33]. Although it is expected to acquire the formal positively charged form even in the active FTS Fe catalyst, the alkaline element might contribute to the electronic structure of the solid [34,35,36]. In this aspect, many theoretical studies have been undertaken [34,35,37,38,39,40]. Based on a systematic study of the effect of alkalis on iron oxides by theoretical methods, Yang et al. [34] have suggested that the presence of a small coverage of alkali metals on the Fe₃O₄ surface causes the tetrahedral Fe d_xy, d_yz, d_xz orbitals to be further split due to electron donation from the alkali metal. This alkali-to-Fe₃O₄ surface electron-donating effect follows the following order: Li < Na < K = Rb = Cs. Experimental studies on this subject are, however, more abundant. In this regard, it has been shown [15,31,41] that the presence of alkaline promoters (Li, Na, K, Rb, Cs) in the H₂ pre-reduced catalyst suppresses CO hydrogenation, while they increase the retention time of carbon species in the following order: (Li < Na < K < Rb < Cs), which may, to some extent, positively impact chain growth probability. Similar observations were also made by other authors [42], who concluded that the alkali promoter inhibits H₂ dissociative adsorption while concurrently increasing the rate of CO dissociative adsorption. In that case, the C-C bond formation is preferred, while CH₄ formation and C-H bond formation are deterred. The presence of K has also been correlated to enhanced surface basicity [43]. Ngantsoue-Hoc et al. [17] tested 100 Fe: 4.6 Si: 1.44 Me (Me = Li, Na, K, Rb, Cs) catalysts and found that the alkali promoter leads to (i) higher activities for Li and K at low conversion levels, where FTS is the dominant reaction, and greater activities for K and Na both at moderate (40%) and high (60%) conversion levels, where FTS tends to be restricted by the limited presence of H₂ in the reactive mixture (i.e., unless H₂ is delivered by the WGS reaction); (ii) increased CO₂ production (e.g., WGS activity) in the case of Na− and K−promoted catalysts; and (iii) lower methane selectivity (exception for Li promotion). Furthermore, Rb and Cs might even be considered poisons, as they decreased both FTS and WGS rates. Interestingly, with Pt/m-ZrO₂ catalysts, despite Rb and Cs accelerating the formate turnover rate by straining the adsorbed molecule towards dehydrogenation, the WGS rate was slowed due to excessive basicity, whereby the catalyst surface tended to strongly bind the second intermediate, carbonate, to the catalyst surface, inhibiting its desorption [44,45]. Later studies on this catalyst series also indicated that growth probability consistently increased from Li to Cs promotion, mainly in the high conversion regime (where FTS and WGS are expected to occur simultaneously) [33]. This effect may be correlated to increased alkali-promoted electronic backdonation from Fe three-dimensional states to the 2π* antibonding orbitals of CO, causing increased dissociative adsorption over the catalyst surface; this, in turn, is proposed to facilitate carburization [14,46,47] as the alkali promoter size (and therefore, its basicity), is increased. Further studies have pointed to possible structural modifications in the active catalyst. For example, Huo et al. [48] have shown that the presence of the K promoter stabilized higher indexed and, therefore, more active facets in iron, such as Fe (3 1 0) and Fe (2 1 1). Even though the active iron phase for FTS is known to be carbidic rather than metallic, this study shows that investigating the effect of the alkali promoter on the surface structure of the active phase merits further attention. In some cases, in addition to focusing on the correlation between the activated catalyst structure and the resulting product selectivity, it is also of interest to follow the nature of the formed species at the surface level, as it might provide further insight into the influence of the catalyst surface chemistry on product selectivity. Besides the metallic/carbidic phase, the active FTS catalyst also has an oxidic Fe phase, and this is suggested to be a complex Fe₃O₄/FeO mixture [30], although the oxidic phase is more likely to be comprised of Fe₃O₄, as FeO tends to decompose into a mixture of Fe⁰/Fe₃O₄ at temperatures below 570 °C and FTS is typically performed in the temperature range of 250–350 °C. The resulting interface between the Fe carbide phase and the defect-laden Fe oxide phase has been suggested to promote WGS [49].

Allowing for the study of the electronic states located down to a few atomic monolayers beneath the surface, X-ray photoelectron spectroscopy (XPS) may deliver information about the composition, average oxidation states, and the very chemical environment of the elements located in the structure. Due to the high signal-to-noise ratio, synchrotron X-ray diffraction allows for the study of phase changes that may take place in a catalyst under working conditions. In this regard, the present contribution aims to investigate the effect of the alkali on the surface chemistry of a series of alkali-promoted (Li, Na, K, Cs) iron catalysts toward FTS by in-situ synchrotron X-ray photoelectron spectroscopy (SXPS) and X-ray diffraction (SXRD). The results are correlated with surface reactivity data (TPR-H₂, TPC-CO, TPD-CO₂, and TPD-H) and catalytic testing.

2. Results and Discussion

2.1. SXRD Studies

As a consequence of the Fe-based FTS catalyst being multiphasic in its nature and due to the fact that product selectivity should be a function of phase distribution, catalyst modification during heat treatment under flowing CO(He) and CO/H₂ (He) was followed by SXRD. According to the SXRD profiles displayed in Figure 1, Fe₃O₄ was the main phase at 270 °C under CO(He). After 1 h under these conditions and especially after 1 h under CO + H₂ (He) flow, iron carbide phases emerged, with χ-Fe₅C₂ and θ-Fe₃C being the major contributions. Interestingly, a somewhat distorted FeO phase was also formed in the Li-promoted catalyst.

Table 1. Phase models considered in the Rietveld refinement are considered in this work.

Phase	Code	Space Group	Unit Cell Parameters
			a (Å)	b (Å)	c (Å)	β (deg.)
Fe3O4	ICSD 26410	F d −3 m	8.3941
χ-Fe5C2	COD 1521831	C 2/c	11.588	4.579	5.059	97.75
Θ-Fe3C	amcsd 0013523	P n m a	5.092	6.741	4.527
α-Fe2O3	amcsd 0000143	R −3 c	5.038	5.038	13.772
FeO	amcsd 0013893	F m −3 m	4.326

presents the phases found in our catalysts under in situ conditions based on analysis of the XRD patterns.

Rietveld refinement of the SXRD patterns obtained in this experiment revealed the evolution of the phase composition during heat treatment under CO (He) and CO + H₂ (He) streams, which are depicted in Figure 2. In all alkali-promoted catalyst samples, it is possible to see the stepwise reduction of Fe₂O₃ to Fe₃O₄, with further reduction/carburization to a mixture of θ- and χ-carbides. Note that the intensity of the last carburization step, represented by the fractions of carbides (Hägg and cementite), varied greatly among the samples. No noticeable carburization occurred in the unpromoted sample under our experimental conditions (P = 1 atm); magnetite was the major phase after heat treatment. The Li-promoted sample also displayed limited carburization, and interestingly, cubic FeO and bcc Fe formed along with FeC_X in the last step. More significant carburization was observed with Na and, to a further extent, with K and Cs. The Hägg carbide (χ-Fe₅C₂) was formed to a higher extent than cementite (θ-Fe₃C). Moreover, the onset of Fe₃O₄ formation was displaced from a temperature range of 197–198 °C in the unpromoted and Li-promoted samples to temperatures of 216 °C (Na), 211 °C (K) and 206 °C (Cs). The light alkali promoters (Li and Na) might tend to stabilize the Fe₂O₃ phase, but such a stabilization effect might be more than compensated by the effect of surface activation of CO, especially as observed with the heavier promoters (K, Cs). Such behavior has also been observed by others [16]. It is interesting to note, however, that an inverse effect is observed upon H₂ reduction of Fe₂O₃ to Fe₃O₄, being that the alkalis tend to accelerate the reduction process [50].

Regarding the active catalyst under FT conditions, the surface composition is important for catalytic activity. In order to assess surface chemistry, we studied our samples using synchrotron XPS, which is discussed in the next section.

2.2. SXPS Studies

The alkali-doped Fe catalyst samples were pre-treated for 30 min under a He-diluted 5% CO:H₂ (0.7 H₂ to CO molar ratio) mixture at 320 °C prior to the SXPS experiment. The XPS spectra are depicted in Figure 3. The effect of the alkaline metal on the surface chemistry of the catalysts was investigated. Regarding the C1s region (Figure 3A), there are two lumped peak regions. The first one, located within the 283–287 eV region, includes contributions from sp²/sp³ characteristics of chemical species such as graphite or hydrocarbon chains [51] or even iron carbides [52,53]. The second region, located in the 285–289 eV region, has been correlated to C-O present in oxygenated adsorbates [54,55,56]. There is still another C1s region located around 290 eV observed in the unpromoted, Li, and Na-promoted samples that have been correlated to adsorbed carbonates [57,58]. Both peak regions (the sp²/sp³ carbon and the oxygenate region, previously indicated) were shown to be of comparable intensities in the unpromoted, Li, K, and Cs-promoted catalyst samples. On the other hand, the tendency of an increased contribution of the oxygenated peak region is clearly observed in the Na-promoted catalyst, indicating that Na induced further formation of adsorbed oxygenated intermediates. This might be linked to the increased selectivity toward olefins found in Na-promoted Fe catalysts [59], considering that olefins may be formed by the deoxygenation of such intermediates [60]. The O1s region peak, displayed in Figure 3B, may be grouped into two contributions. The 529–532 eV peak region has been correlated to transitions from O1s to O2p-Fe3d mixed states in the context of lattice oxygen in iron oxides. The second peak region located around 532 eV has been ascribed to transitions between lattice O1s and Fe4s4p states [61], but also to oxygenated surface species [55]. Surface Si-O-Si and Si-OH is included in the peak located in the 533–535 eV region [59,62]. As Si 2p features have been observed in the survey spectra of the samples, albeit at low intensity, the contribution of surface Si-O from the Si substrate should not be disregarded. Regarding the Fe2p region (Figure 3C), peaks located at B.E. lower than 709 eV have been correlated to iron carbide phases. At higher B.E.s, Fe³⁺ (711 eV) and Fe²⁺ (709 eV) related phases have also been observed [61,63]. Under FTS conditions, the active catalyst is expected to be comprised of a mixture of iron oxides (i.e., mainly magnetite) and iron carbides. A traditional view regarding the role of these Fe-containing phases is that the Fe₃O₄ phase is related to the production of oxygenates and the WGS activity [17], while the iron carbide-related phases are shown to be the active phase for long-chain hydrocarbon synthesis. However, other workers have shown that Fe carbide contributes to WGS activity, and this may be at the interface between Fe carbide and defect-laden Fe₃O₄ [64].

2.3. Temperature Programmed Reaction Studies

Figure 4 displays temperature-programmed reduction profiles of the samples under Ar-diluted H₂ flow (TPR-H₂). Two lumped peaks are observed in all profiles, one in the range of 390–400 °C and the other, a wider peak, located at temperatures above 400 °C. The first, smaller peak has been correlated to the reduction of Fe₂O₃ to Fe₃O₄ [65,66]. In some samples, the peaks in that low-temperature region may be deconvoluted, indicating that the distribution of particle sizes may be bimodal, where it is expected that Fe₂O₃ reduction tends to occur at slightly higher temperatures in larger particles [67]. Regarding the alkali-promoted samples, it may be clearly seen that the presence of Li and, to a higher extent, Na, K, Rb, and Cs tend to shift the hematite–magnetite conversion to higher temperatures. This indicates that the presence of alkali tends either to intrinsically stabilize the hematite phase or to inhibit surface reactivity toward the dissociative adsorption of H₂, which is key to the Fe₂O₃-to-Fe₃O₄ reduction process. On the other hand, the Fe₃O₄ reduction step under H₂ flow does not seem to be overly dependent on the nature of the alkali promoter, as observed by the invariability of the peak temperatures above 450 °C along the promoter series. However, in all catalysts, peaks > 450 °C may be deconvoluted into two different regions representing the reduction of Fe₃O₄ to FeO (which has been ascribed to occur at ca. 590 °C) and finally, FeO to Fe⁰ reduction, at temperatures around 640 °C. The peak at ca. 594 °C is not fully resolved, which is consistent with the fact that FeO is metastable at that temperature and, therefore, forms only transiently.

Figure 5 displays the H-desorption profiles of the differently promoted samples. Many H₂ desorption peaks over the entire 50 to 800 °C temperature range may be observed in all samples, with the multiplicity of H-adsorption site strengths that is expected from a multiphasic catalyst, such as the alkali-promoted Fe-O system. H₂ desorption peaks located within the 50–200 °C range have been related to H₂ desorption from metallic iron surfaces. Hydrogen desorbed from iron surfaces at close to room temperature (Hα) comes from shallow, low coordination surface sites, whereas H₂ desorbed at ca. 180 °C (Hβ) is from hollow fourfold sites [41,67], or even from Fe⁰-FeO surfaces [66,67]. Desorption peaks located at 300 °C and above may be correlated with hydrogen evolution from spillover from the surface OH groups [68,69]. Nevertheless, it is interesting to notice that the presence of the alkali strongly suppresses the Hα/Hβ region, as can be seen from the integration peak areas below 300 °C, as displayed in

Table 2. Integration areas of the deconvoluted temperature-programmed H₂ desorption peaks.

Catalyst Promoter	Areas < 250 °C	Areas, 250 °C–400 °C	Areas > 400 °C
Unpromoted	66.0	22.6	11.4
Li	53.9	46.1	0
Na	3.6	55.1	41.2
K	21.2	28.1	50.7
Rb	35.9	46.5	17.6
Cs	31.4	20.7	47.8

. Potassium has been suggested to change the electronic structure of the metal surface via its 4s levels [70]. This electronic effect may contribute to increasing repulsion between H₂ and Fe surface hence suppressing adsorption of H₂ onto the catalyst surface. This is expected to lead to lower hydrogenation rates and, therefore, higher long-chain hydrocarbon selectivities. Our data indicate that alkalis other than K exert this same effect, with Na being the most effective. Nevertheless, no apparent correlation between H-coverage and alkali radius (on which the basicity is expected to be dependent) could be observed. This may indicate that another effect could be responsible, such as the chemical state of the alkali in the catalyst surface (whether it is in the form of surface oxides, hydroxides, or carbonates) or the fraction of the alkali that is effectively on the catalyst surface (i.e., and not inside the crystallites and effectively replacing Fe ions in iron oxide related phases).

Figure 6 displays the TPC-CO of the samples. Compared to the unpromoted sample, the Fe₂O₃ transition occurs at increasing temperatures in Li and Na-promoted samples, indicating that those elements tend to stabilize the Fe₂O₃ phase, strengthening Fe-O bonds [12]. However, the K and Cs-promoted samples showed an inverse trend. According to Li et al. [12], the decreasing trend of the Fe₂O₃/Fe₃O₄ transition temperature in the iron oxides promoted with heavier alkali should be due to their enhanced CO activation capability.

CO₂ is an acidic molecule (i.e., making carbonic acid when dissolved in H₂O) and is used as a probe molecule to characterize catalyst surface basicity. Figure 7 shows the TPD-CO₂ profiles of the catalysts considered in this study. CO₂ adsorption on the unpromoted catalyst was lower compared to its promoted counterparts. The MS signals of the unpromoted and Li-promoted catalysts were multiplied by a factor of 4 to enable observation of the CO₂ desorption peaks. It may be observed that all catalysts displayed CO₂ desorption peak lumps located both in the c.a. 100–400 °C range and above 500 °C, which may be ascribed, respectively, to loosely bound CO₂ as surface carbonates on basic sites located on iron oxide surface sites and on alkali containing surface sites [71]. The TPD profile of the unpromoted catalyst showed the lowest temperature CO₂ evolution peaks, with a maximum at 225 °C and shoulder peaks at 303 °C and 379 °C. In addition, 66.6% of the CO₂ was desorbed below 300 °C, while 21.0% was desorbed between 300 °C and 500 °C. In addition to the weakest bonding of CO₂ to the surface,

Table 3. Percentages of CO₂ desorption peaks during TPD-CO₂ having Gaussian peak maxima within the given ranges. The total area of peaks relative to the area of the Na-promoted Fe₂O₃ catalyst is also provided in the final column.

Catalyst	T < 300 °C	300 °C < T < 500 °C	500 °C < T < 700 °C	>700 °C	Relative Total Area Using Na as a Basis
Unpromoted	66.6	21.0	6.2	6.2	8.7%
Li	43.6	32.0	14.6	9.8	11.1%
Na	5.8	3.4	82.2	8.6	100%
K	8.0	6.0	1.2	84.8	62.6%
Cs	7.7	12.4	5.0	74.9	91.9%

shows that it also evolved the lowest amount of CO₂, indicating that this catalyst had the lowest surface basicity. As shown in Table 3, adding Li to the catalyst increased the amount of CO₂ desorbed. Among the alkali-promoted catalysts tested, the majority of the CO₂ desorbed evolved in the lowest temperature ranges (i.e., below 300 °C and between 300 °C and 500 °C), with peak maxima at 220 °C and 364 °C. Adding Na, while a broad low-temperature peak occurred with a maximum of ~281 C, the majority of CO₂ desorbed within the range of 500 °C and 700 °C, with a peak maximum of 653 °C. The Na catalyst evolved the most CO₂ of all the catalysts tested, suggesting the highest number of total basic sites. The majority of basic sites on Na were stronger than those of Li. Adding K also resulted in a low-temperature peak (maximum at 274 °C), but the majority of the CO₂ desorbed in the highest temperature range above 700 °C (i.e., strongest basic sites), with a peak maximum at 927 °C. The Cs-promoted catalyst evolved CO₂ in the low-temperature range at a higher temperature than the other catalysts (i.e., 342 °C), suggesting even stronger basic sites, and the majority of CO₂, as observed with K, desorbed in the highest temperature range (peak maximum of 916 °C). Thus, the general trend is increasing basicity, moving down Group 1.

2.4. Catalytic Testing

CO and H₂ conversions are depicted in Figure 8a and Figure 8b, respectively, while their normalized activities—CO and H₂ conversions normalized by their initial conversions—are shown in Figure 8c and Figure 8d, respectively. The highest and most stable conversions were attained with the unpromoted and Na-promoted (Na-Fe) catalysts, such that Na-Fe was slightly more stable than the unpromoted catalyst over 250 h of FT synthesis. Although K-Fe displayed the highest initial CO conversion among all the catalysts evaluated, its catalytic activity declined rapidly over time. The decreasing trend in H₂ conversion (Figure 8b) follows a similar trend to that of the CO (Figure 8a) for all catalysts. The normalized activity (a = r_t/r₀) obtained on various catalysts is shown in Figure 8c,d. It is derived by means of dividing the reaction rate at time ‘t’ (r_t) by the reaction rate at time zero (first data point, r₀). The linear model fits with the normalized activity of various catalysts. The slope of normalized activity versus normalized time yields the deactivation constant (k_d), which decreased in the following order: CsFe > KFe > LiFe > Fe > NaFe. For instance, the deactivation constant (k_d, h^-1) corresponding to CO for Cs-Fe (0.0036) is ~10 times higher than NaFe (0.0003), indicating that Cs-Fe underwent faster deactivation than Na-Fe under our testing conditions. It seems that smaller-sized alkali metals such as Li and Na are more efficient in keeping the catalyst from deactivating too rapidly as compared to larger alkali metals, as is supported by the earlier findings from the literature [17]. The deactivation pattern observed in Cs-Fe and K-Fe shown in Figure 7c,d was found to be roughly the same (although K-Fe displayed higher activities than Cs-Fe (Figure 7a,b) over the entire testing time), indicating that both promoted catalysts shared similar deactivation mechanisms. Furthermore, this result is consistent with our SXRD results that indicate higher carburization extents for K and Cs promoted catalysts. In the previous study for carbon dioxide hydrogenation with alkali-doped Fe prepared by following an oxalate decomposition method, it was predicted that catalyst deactivation occurs predominantly by deposition of amorphous carbon as moving down on alkali Group I metals, especially from K to Cs that mainly originate from the basic nature of alkali [72]. Bartholomew and Xu have investigated carbonaceous surface species of various forms along with bulk iron carbides using Temperature-Programmed Hydrogenation (TPH) and Mössbauer spectroscopy. The authors suggest that several types of carbon (atomic, lightly polymerized, and graphitic surface carbons) can contribute to catalyst deactivation [73]. The presence of K and Cs in the high concentrations considered in this study (5:100 alkali to Fe molar ratio) contributed to excessive surface reactivity toward CO activation, and therefore, a larger inventory of adsorbed carbon species was formed that contributed to the formation of a greater larger fraction of the less active θ-Fe₃C phase. The formation of plain amorphous carbon in such conditions is a contributor to early catalyst deactivation [30] of KFe and CsFe should not be dismissed, either. It is understood that the presence of multiple bulk phases along with amorphous carbon makes it difficult to identify factors that really contribute to catalyst deactivation.

The selectivity of hydrocarbons, including methane, light gases (C₂-C₄), and other hydrocarbons, is compared for various catalysts at similar CO or syngas conversions. As may be observed in

Table 4. The sensitivity of the conversion and FT product selectivity of Fe and Group 1 alkali metal doped Fe catalysts to the nature of alkali present in the catalyst during iron FT synthesis.

Catalysts	TOS(h)	Conv. (%)		H2/CO Usage	CO2 sel. (mol-%)	HC sel. (wt%)			(C2–C4) = /(C2–C4)o	Ref.
		CO	CO + H2			CH4	C2–C4	C5+
Fe	94	63.2	65.0	0.75	44.6	6.4	32.3	61.3	1.92	this work
LiFe	41	50.3	52.9	0.79	42.6	4.6	25.1	70.3	3.21	this work
NaFe	91	62.6	58.9	0.60	49.2	5.9	22.5	71.6	3.49	this work
KFe	98	61.4	57.1	0.58	47.3	3.3	13.8	82.9	5.26	this work
CsFe	50	55.3	52.8	0.62	49.1	3.9	13.7	82.4	5.48	this work
FeSi	140	53.4	40.7	-	33.7	24.9	49.4	25.7	0.45	[41]
LiFeSi	143	30.1	25.7	-	27.1	16.0	39.7	44.3	0.56	[41]
NaFeSi	142	42.4	27.6	-	29.2	14.9	35.5	49.6	0.63	[41]
KFeSi	141	57.3	33.3	-	46.1	12.9	30.3	56.8	0.61	[41]
CsFeSi	141	57.1	38.5	-	36.5	15.6	31.2	53.2	0.64	[41]
Fe/CNT	-	46.4	-	-	9.4	29.1	13.5	57.4	-	[77]
Fe-3Li/CNT	-	62.2	-	-	10.3	30.2	17.6	52.2	-	[77]
Fe-3Na/CNT	-	73.0	-	-	14.0	12.0	25.1	63.0	-	[77]
Fe-3K/CNT	-	38.4	-	-	8.5	13.6	29.9	56.5	-	[77]
FeCu/SiO2	201	28.8	-	-	30.4	9.9	41.4	48.7	2.38	[78]
FeCuNa/SiO2	202	63.5	-	-	45.1	6.6	29.3	64.1	3.60	[78]
FeCuK/SiO2	199	69.1	-	-	47.1	5.2	19.9	75.0	3.72	[78]
FeSiKCu	-	86.2	82.5	0.60	45.6	-	-	61.4	-	[79]

Reaction conditions: this work T-270 °C, P-175 psig, SV-3 NL/h/g cat., H₂/CO-0.7. [41] T-260 °C, P-217 psig, GHSV-2000 h⁻¹, H₂/CO-2. [77] T-275 °C, P-116 psig, GHSV-2400 h⁻¹, H₂/CO-2, HC selectivity (mol%). [78] T-250 °C, P-217 psig, GHSV-2000 h⁻¹, H₂/CO-0.67. [79] T-270 °C, P-175 psig, SV-3.1 NL/h/g Fe, H₂/CO-0.7.

, methane selectivity decreases from 6.4% for undoped Fe to 4.6 for Li-Fe, 5.9 for Na-Fe, 3.3 for K-Fe, and 3.9% for Cs-Fe. Similarly, light gas selectivity was also found to be lower with the presence of the alkali in the catalyst. This result is consistent with the higher surface reactivity toward CO, leading to higher carburization extents, as shown by the SXRD results. This trend indicates that the alkali does affect the FT reaction pathway probably by altering the electronic interactions so as to increase the interaction energy between adsorbed hydrocarbon chains and the iron carbide surface, which are claimed to be the active phases of Fe FT synthesis [74,75,76].

Another characteristic pattern observed during this study is that the selectivity to olefins, defined by the fraction of C₂–C₄ olefins in the entire C₂–C₄ product [(C₂–C₄) = /(C₂–C₄+)^o], was found to increase linearly with increasing size of the Group 1 ion. Considering that the pathway for paraffin formation should involve the hydrogenation of the olefin desorbed from the catalyst surface, the trend in the olefin to paraffin ratio should be understood in terms of a decreased coverage of adsorbed H species, caused by the presence of the alkali, which is expected to increase electronic repulsive interaction between the catalyst surface and the H₂ molecule (Pauli repulsion) [41,67]. This result is consistent with the results of H-TPD, which shows decreased H desorption areas below 250 °C in the alkali-promoted catalyst, as evidenced in Figure 5 and Table 2, and may be correlated to the alkali’s capacity to inductively increase electronic density in the surface active sites [13].

Iron-based catalysts are known to exhibit a steady water-gas shift activity, which produces carbon dioxide and hydrogen from carbon monoxide and water. It is well-known that the alkali enhances the water-gas shift activity of Fe. In doing so, it allows the catalyst to be operated at a low H₂/CO ratio to form hydrocarbons [70]. The calculated H₂/CO usage ratio on pure Fe was 0.75 and decreased to the 0.58–0.62 range for Na, K, and Cs doped iron catalysts. This reveals that the alkali (Na, K, and Cs) promotes the WGS activity of iron. The higher CO₂ production rate observed on alkali-doped iron catalysts supports the above-mentioned conclusion. Bare Fe produces about 40.1% CO₂ from FT synthesis, which increased with Li-Fe to 42.6% and slightly further increased to 49.1, 46.8, and 49.2% for Na-Fe, K-Fe, and Cs-Fe, respectively. The increased WGS selectivity is consistent with other authors [17]. It can be concluded from Figure 9 that Group 1 alkali metal addition to Fe tends to decrease selectivity to hydrocarbons with increasing size of the alkali and increases olefin formation from syngas. The primary alcohol selectivity does not vary appreciably across various Group 1 alkali metal-promoted iron catalysts as compared to the unpromoted catalyst, except in the case of LiFe, where alcohol selectivity was found to be significantly lower. It indicates that the formation of olefins during iron FT synthesis is sensitive to the nature of the alkali present in the catalyst. The product selectivity observed in the present investigation with alkali-doped Fe is comparable to works found in the literature [41,77,78,79], as shown in Table 4. Irrespective of the presence of structural promoter [78], silica, and reduction promoter (e.g., Cu), both Na- and K-doped Fe exhibit similar CO₂ selectivity (45.1%,47.1%), and hydrocarbon selectivity is also comparable with current data. This indicates that the catalytic activity results presented in this study are indeed relevant from an industrial perspective.

The alpha plots for hydrocarbons and alcohols formed on various catalysts are shown in Figure 10. The chain growth probability, alpha (α), was calculated based on FT products in the carbon range from C₁ to C₃₁ for hydrocarbons and C₁ to C₁₇ for alcohols, which is shown as an inserted table in Figure 10. In general, the addition of alkali did not affect the chain growth probability significantly, except in the case of Cs, whose value for hydrocarbons was found to be significantly lower. KFe exhibits a slightly higher chain growth probability (0.76) for hydrocarbons than undoped Fe (0.73) and even other alkali-promoted catalysts. This indicates that Cs can be detrimental to carbon–carbon chain growth for iron Fischer–Tropsch synthesis. Taking into account the notion that α might increase having the coverage of adsorbed carbon species as a parameter, it might be expected then that α should monotonically increase from Li to Cs. For example, as a function of alkali radius, the ability to inductively increase electron density on the catalyst surface should increase. The unexpected effect of Cs on α might then indicate that although the heavy alkali increases surface reactivity toward CO activation (as observed from the higher carburization rates that were attained), this very activity might be detrimental to the ability to catalyze the growth of long-chain hydrocarbons, likely due to the formation of active site poisoning adsorbed carbon species, such as amorphous carbon. Regarding the selectivity to oxygenates (n-alcohols), it may be observed that unlike Li, K, and Cs, the Na promoter does not seem to negatively impact α, which suggests that Na is not blocking active sites for n-alcohol synthesis at the alkali concentration considered in this work (5:100 alkali to Fe atomic ratio).

Olefins are an integral part of the broad spectrum of FT products based on Fe. Under similar CO conversions, KFe and CsFe (as compared to un-doped and Li and Na-doped iron catalysts) produce a higher fraction of olefin content. Li et al. obtained similar results with Si-promoted Group 1 alkali metal-doped iron catalysts [41]. This is primarily attributed to the lower hydrogenation activity of alkali-doped iron, causing enhanced olefin formation from syngas [32]. It is evident from Figure 11 that the maxima of O/P occur at either C-3 or C-4. The olefin/paraffin ratio decreases with increasing carbon number after C-3 or C-4 indicating that higher olefins were taking part in the secondary hydrogenation reaction. The Fe promoted by heavier alkali metals produced a significant fraction of higher olefin, meaning that secondary hydrogenation was suppressed with alkali addition over Fe. This observation is also aligned with earlier work reported in the literature [41], as well as the suppressive effect of alkali on the catalysts´ hydrogenation function [80]. Bukur et al. [81] demonstrated from an experiment performed under supercritical conditions that α-olefins and, to some extent, n-paraffins are the primary products of FTS. The authors came to this conclusion based on the observation of enhanced α-olefins formation under supercritical conditions compared to conventional FTS conditions due to α-olefins having higher diffusivity and desorption rate than 2-olefins whose selectivity did not seem to change with reaction condition. Hence, it is likely that larger alkalis, particularly in our case, K and Cs, might induce the carbon–carbon chain growth through modified electronic interactions of CO and intermediate species with iron, presumably by excessive carbon formation over the iron surface. However, this ultimately has an impact on catalyst stability as well. Catalyst deactivation with Fe can be attributed to many factors, including sintering, oxidation, and inactive carbon formation, and, in the present investigation, any of these factors could contribute to the deactivation observed with K-Fe and Cs-Fe catalysts.

3. Materials and Methods

3.1. Catalyst Synthesis

Fe-based catalysts were prepared by the precipitation of Fe(NO₃)₃. 9H₂O (Aldrich, St. Louis, MO, USA, 99%) solution with ammonia (Dinamica, (Indaiatuba, São Paulo, Brazil), 28–30%), molar ratio 1:10. After filtration of the precipitated cake and drying at 120 °C for 10 h, it was calcined in a muffle furnace at 400 °C for 10 h. This hematite support was then impregnated to the point of incipient wetness with solutions containing alkali metal carbonates (Vetec, (Rio de Janeiro, Rio de Janeiro, Brazil), 98%). The catalysts were prepared to obtain an atomic ratio of 5 Me: 100 Fe (Me = Li, Na, K, Cs). Each catalyst was then calcined at 400 °C for 4 h.

3.2. Catalyst Characterization by In-Situ SXPS

The effect of Group I alkali metal promotion on the surface chemistry of Fe catalysts was studied by in-situ synchrotron X-ray photoelectron spectroscopy (SXPS) using the SXS beamline located at the Brazilian Synchrotron Light Source (LNLS-Campinas). In order to avoid distortions in the XPS spectra due to eventual charge buildup, the powders were first dispersed on n-type Si (1 1 1) wafers, following an established procedure [34]. According to that procedure, the catalyst powders were suspended in isopropyl alcohol and then droplets were added to the Si (1 1 1) wafers and gently heated on a hot plate. Several drops were added to the wafers to ensure uniform coverage. The Si (1 1 1) wafers were meticulously cleaned with piranha solution prior to impregnation with the catalyst suspension.

The samples (Li, Na, K, Cs-doped, and undoped iron oxide) were then placed in a sample holder fit for holding eight samples and then placed inside the XPS pre-treatment chamber for 90 min of treatment under a He diluted H₂/CO mixture (H₂:CO 1:1 molar ratio) at 1 atm and 350 °C (White Martins (Campinas, São Paulo, Brazil)). After that, the sample holder was then directed into the XPS chamber using mechanical arms and avoid exposure to the atmosphere. The sample was studied using 1840 eV monochromatic light, which was selected by a Si (1 1 1) monochromator. From each sample, a general survey spectrum and more detailed C 1s, O 1s, and Fe 2p spectra were recorded. XPS spectra were calibrated for aliphatic C 1s, known to be located at 284.6 eV. The general survey (1700 to –10 eV) was recorded at SXS beamline with a pass energy of 40 eV, a bias voltage of 90 V, and a dwell time of 0.25 s. Specific spectra were recorded to monitor Fe 2p (745 to 700 eV), O 1s (550 to 505 eV), and C 1s (300 to 250 eV), with a pass energy of 20 eV and a bias voltage of 90 V, and two scans of each region were conducted. A Lorentzian curve fitting function was used for peak deconvolution.

3.3. Characterization by In-Situ SXRD

The evolution of the phases present in the catalysts while exposed to CO and CO + H₂ mixtures (at 1 bar) was studied by synchrotron X-ray diffraction (SXRD) in the XPD beamline located in the Brazilian Synchrotron Light Source (LNLS-Campinas). For the in situ XRD measurements, the samples were placed within a rotating sample holder located inside a furnace (Canário) that allowed gas to flow during the analysis. The samples were then irradiated with 7000 eV (λ = 1.7712 Å) light in a Hübner 4 + 2 circle diffractometer. Heat treatment of the samples was comprised of three steps: (i) temperature ramping (15 °C min⁻¹) up to 30 min soaking at 450 °C, under 100 mL min⁻¹ synthetic air flow (White Martins (Campinas, São Paulo, Brazil); (ii) room temperature cooling under synthetic air flow and then temperature ramping (2 °C min⁻¹) under 100 mL min⁻¹, 5% CO (He) (White Martins (Campinas, São Paulo, Brazil)) up to a 1 h temperature soaking at 270 °C and (iii) 1 h flowing 100 mL min⁻¹ of a mixture containing 2.5% CO and 1.75% H₂ diluted in He (1:0.7 CO to H₂ molar ratio) (White Martins (Campinas, São Paulo, Brazil)), while the temperature was kept at 270 °C.

3.4. Characterization by TPR-H₂/TPC-CO/TPD-H₂/TPD-CO₂

H₂ temperature programmed reduction (TPR-H₂), temperature-programmed carburization mass spectrometry (TPC-CO), temperature-programmed desorption of hydrogen (TPD-H₂), and temperature-programmed desorption of CO₂ were performed using an Altamira AMI-300R (Pittsburgh, PA, USA) catalyst characterization unit coupled to a Hiden quadrupole mass spectrometer and a TCD. For TPR-H₂, 10% H₂ in Ar (UHP, Airgas, San Antonio, TX, USA) flowed at 30 cm³/min, and the temperature was increased from 30 to 1000 °C at 10 °C/min. The thermocouple was located inside the catalyst bed, with approximately 33 mg of sample being used. TPC-CO experiments were performed with a mixture of 4% CO (He balance) (Airgas, San Antonio, TX, USA) being flowed at 25 cm³/min as the temperature was ramped at 10 °C/min to 570 °C and held for 50 min. The MS signal (m/z of 44) of CO₂ was followed. Regarding TPD-H₂ analysis, catalysts (400 mg) were activated under H₂ flow (30 mL min⁻¹) (Airgas, San Antonio, TX, USA) by ramping (1 °C min⁻¹) at 400 °C. Temperature was then held at 400 °C over 1 h. After cooling the catalyst samples to 50 °C and holding at this temperature for 1 h, both under flowing H₂, the temperature was then ramped to 800 °C at a 10 °C min⁻¹ rate under Argon 30 mL min⁻¹ (flowing pure Argon (Airgas, San Antonio, TX, USA)). For TPD-CO₂, the catalyst (400 mg alkali-promoted, 800 mg unpromoted) was reduced at 400 °C using 10 cm³/min H₂ and 20 cm³/min argon for 1 h, purged in 30 cm³/min flowing Ar for 20 min, and cooled to 30 °C. Then, the sample was saturated with CO₂ using 25 cm³/min of 25% CO₂ (balance helium) for 15 min. at 30 °C, and finally, the temperature was increased to 1000 °C (10 °C/min) in 30 cm³/min flowing Ar while the MS signal (m/z of 44) of CO₂ was followed.

3.5. Fischer-Tropsch Synthesis

The description of a 1 L continuously stirred tank reactor (CSTR, Pressure Products Industries (PPI), Warminster, PA, USA) is conveniently divided into four parts: (1) feed system, (2) reactor, (3) condensed products recovery, and (4) gaseous products analysis. (1) Feed system: The feed system consists of hydrogen (UHP grade supplied by American Welding & Gas, Lexington, KY, USA) and carbon monoxide (UHP grade supplied by American Welding & Gas, Lexington, KY, USA) continuously fed to the reactor from individual pressurized cylinders in which the flow rates were controlled by pre-calibrated Brooks mass-flow controllers (5850 E series, Brooks Instruments, Halfield, PA, USA). All lines were made of stainless-steel ¼” tubing, which connected the feed system, reactor, condensed products recovery unit, and gaseous product analysis equipment. Gas mixing was aided by a 500 cc bomb, which was connected between the feed system and the reactor. (2) Reactor: The autoclave (1 L) is made up of a 316 stainless-steel housed between a ceramic electric heater. The feed stream fed into the reactor via a dip-tube. A K-type thermocouple was placed inside the thermowell and measured the temperature of the slurry medium, with molten Polywax 3000 (Baker Hughes, Houston, TX, USA) serving as the startup solvent. The slurry-containing catalyst was stirred at a constant stirring speed of 750 rpm. (3) Condensed products recovery: The effluent from the reactor was passed through two 500 cc bombs maintained at 100 °C and 5 °C to condense liquid products. The hot trap received heavier hydrocarbons, which otherwise resided inside the reactor due to the high boiling point. (4) Gaseous products analysis: Uncondensed gaseous hydrocarbons and unreacted syngas were passed through a backpressure regulator, and the flow rate was measured using a dry test meter (Model: DTM-200A, Elster American Meter, NE, USA). The composition of the effluent gas stream was analyzed using an online micro-GC system (Inficon, East Syracuse, NY, USA). The organic material condensed in the cold and warm traps was analyzed using an offline gas chromatograph (GC) system (Model: 7890A GC from Agilent Inc., USA) equipped with a DB-5 capillary column and FID detector. For quantification of hydrocarbons, the raw peak area is normalized to equivalent weight percent while the response factor (published in the literature [82]) was applied to calculate primary alcohol weight percent. A similar procedure was followed as described above for the analysis of various organics present in the water phase. The separation of components was realized using an SRI 8610C GC (Torrance, CA, USA) containing a HayeSep D packed column and a TCD detector. Typically, 8 g of Fe oxide catalyst was dropped into the 1 L CSTR containing 310 g of melted Polywax, 3000 (Avg. MW = 3000). The catalyst was then activated in flowing CO (24 slph) at a temperature of 270 °C and CO pressure of ~175 psig for 24 h. After CO activation, the feed gas was switched from CO to the H₂/CO mixture at a space velocity of 3 (NL/h/g cat) with an H₂:CO ratio of 0.7.

The conversion and selectivity parameters are defined in Equations (3) and (4) respectively:

$$\%CO conversion=100\times {\displaystyle \frac{{\mathrm{n}\mathrm{C}\mathrm{O}}_{in}-{nCO}_{out}}{{nCO}_{in}}}$$

(3)

$$Selectivity \left(\%\right)=100\times {\displaystyle \frac{{\mathrm{n}}_{product out}·\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{n}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}}{{nCO}_{in}-{nCO}_{out}}}$$

(4)

where n(CO)_in and n(CO)_out are the number of moles of CO entering and exiting, respectively. Selectivity is defined as the percentage of moles of CO consumed to form a particular C_n product (hydrocarbon, CO₂, or oxygenate), normalized by the amount of CO consumed.

4. Conclusions

The present contribution aimed to study the effect of the alkali metal on the structure–reactivity relationships of iron-based catalysts toward FTS. Catalyst structure was elucidated by in-situ synchrotron-based XRD and XPS, which revealed that the alkali elements induced higher carburization rates, as well as lower adsorbed hydrogen coverages as an indicator of their ability to alter surface reactivity. This was likely via an inductive effect on the surface electronic structure of the catalyst. Catalyst stability followed the trend Na-Fe > unpromoted > Li-Fe > K-Fe > Cs-Fe, being consistent with the ability of the alkali (Na) to prevent active site loss by catalyst reoxidation, as well as the ability to prevent deactivation from the formation of excessive carbon, which is a function of alkali radius/basicity. Potassium was the most active in promoting high α hydrocarbon formation. It is active enough to promote CO dissociative adsorption (and the formation of FeCx active phases) and decrease the surface coverage of H-adsorbed species, but it is not so active as to cause catalyst deactivation by the formation of an active site-blocking carbon layer.