Carbon Pathways, Sodium‐Sulphur Promotion and Identification of Iron Carbides in Iron‐based Fischer‐Tropsch Synthesis

Fisher‐Tropsch Synthesis (FTS) is industrially used for converting a carbon‐containing feedstock, such as coal, natural gas, biomass, and municipal waste, via the production of synthesis gas (a mixture of CO+H2) into hydrocarbons. This review article focuses on Fe‐based FTS catalysis, thereby focusing on the process conditions available for steering the various carbon pathways from input CO and their associated reactions. We will also discuss the effects of alkali‐sulphur chemical promotion and the identification of the FTS reaction active Fe carbides, which are assigned with precise crystal structures and nomenclature. Each observed Fe carbide crystal structure is further assigned with corresponding Mössbauer Absorption Spectroscopy (MAS) hyperfine fields. The expected formation temperatures and experimental conditions for the identified Fe carbides encountered in FTS research, namely ϵ‐Fe3C, η‐Fe2C, χ‐Fe5C2, θ‐Fe3C and θ‐Fe7C3, are reviewed.


Introduction
The Fischer-Tropsch Synthesis (FTS) reaction can be used to convert a CO + H 2 gas mixture (i. e., synthesis gas or syngas) produced from a carbon-containing feedstock into a mixture of hydrocarbons. The FTS reaction can be idealized as (Reaction 1), where the hydrocarbons are polymerized from À (CH 2 )À units and O atom from CO is removed as H 2 O. [1] In an industrial setting, the application of FTS reaction becomes Fischer-Tropsch Process (FTP) where i) a carboncontaining feedstock is converted into syngas, ii) a FTS reaction is utilized and iii) the FTS reaction produced hydrocarbon mixture (i. e., synthesis crude or syncrude) is refined into hydrocarbon products (Figure 1). [2,3] The complete FTP can be labelled after the feedstock: Biomass-to-Liquids (BTL), Coal-to-Liquids (CTL), Gas-to-Liquids (GTL) and Waste-to-Liquids (WTL). The effective H : C ratio of the syngas is the actual ratio of elemental H : C available for the FTS reaction after all of the heteroatoms (N, O, S) in the feedstock are hydrogenated and subsequently removed during a syngas cleaning step (Table 1). [4,5] The used feedstock dependent CO 2 footprints are in the order: GTL < WTL < CTL < BTL with an inverse relation to the effective H : C ratio (Table 1). That is, the increase in the CO 2 footprint due to input feedstock fundamentally comes from the necessity of adding more H to the system for satisfying the required FTS reaction stoichiometry of 4 : 1 H : C. [4,5] Despite that the syngas production step can have a significant impact, [4,6] it is not the only step affecting the FTP's CO 2 footprint. Each process step, from feedstock to refined hydrocarbon products, should be optimized by the Green Chemistry principles: avoiding waste, maximizing atom economy, increasing energy efficiency and designing less hazardous chem-ical syntheses. [4] In other words, from each process input CO molecule maximum amount elemental C should end up to the hydrocarbon products with minimum energy used, and waste products formed.
The C atom efficiencies for the total FTP from feedstock into hydrocarbon products currently stand at~75 % for GTL, [4] and 28-34 % for CTL. [4,5] For comparison, in crude oil refineries the C efficiency is estimated to be~89 %. [4] The production of purified syngas accounts for~30-50 % of the capital costs for GTL-and~60-70 % for CTL-based processes. [2,3] The expensive syngas should be turned into hydrocarbon products as efficiently as possible also from the viewpoint of economic viability of the FTP.
To maximize the FTP C efficiency and adherence to Green Chemistry Principles, the syngas conversion process to syncrude with the FTS reaction still needs improvements. The main ways to achieve this is to alter catalyst compositions and reaction conditions for the FTS reaction.
With varying the catalyst material compositions (e. g., active d-block metal, support material and structural and/or chemical promoters) the catalysts behaviour can be modified. [1,7,16,17,[8][9][10][11][12][13][14][15] General requirements for an industrially good FTS reaction catalyst material are: i) sufficient activity towards FTS reaction, ii) high selectivity for converting the input CO only to the desired hydrocarbons and iii) mechanical and catalytical stability. [4] Despite most of the d-block transition metals being somewhat active towards the FTS reaction, [1,18,19] catalysts based on Fe and Co are the most suitable ones for FTS reaction applications. Consequently, only Fe-and Co-based catalysts have been applied on a large scale in industry. [1,4] The reaction conditions (e. g., reaction temperature, input feed H 2 : CO ratio and pressure) have a significant effect on which hydrocarbon products are formed. In the currently operational industrial Fischer-Tropsch facilities, either Lowtemperature Fischer-Tropsch (200-250°C, LTFT) or High-temperature Fischer-Tropsch (320-375°C, HTFT) have been used with reaction pressures of 15-60 bar. [4] Fe-based catalyst materials have been used on both LTFT and HTFT, while Co has only been used for LTFT. Fe-HTFT is best suited for producing syncrude with short-chained unsaturated hydrocarbons, while the low temperature Fe-or Co-LTFT favours production of syncrude with long-chained saturated hydrocarbons. [4] In terms of catalyst stabilities, Fe-based FTS catalyst materials can be stable up to~4 months for LTFT and 1.5 months for HTFT, while Co-based LTFT catalysts can be used for up to a couple of years. [4] This article reviews a selection of topical aspects on Febased FTS catalysis. The first major topic of this review is to discuss and define the possible reaction pathways for C atoms from input CO in the FTS reaction. How the reaction conditions and some aspects of catalyst composition affect these "C pathways", catalyst phase changes and deactivation, and thus the overall C efficiency towards hydrocarbons are then further reviewed with a clear focus on Fe-based FTS reaction chemistry. The effects from NaÀ S chemical promotion, [20][21][22][23][24] are discussed as well. As a second major topic of this review, the identification and formation conditions of the various Fe carbides encountered in the Fe-based FTS reaction are discussed. Exact crystal structures and Mössbauer Absorption Spectroscopy (MAS) hyperfine fields are assigned for the identified Fe carbides. This task is performed in order to clarify the often imprecise and unclear nomenclature used in FTS literature for the different Fe carbide phases. The clarified Fe carbide nomenclature will help future, and in interpreting the past, studies to assign more definite catalytic properties and effects for the Fe carbide phases in the FTS reaction.
The topic of Fe-based catalyst synthesis is excluded from this review. Discussion of catalyst synthesis effects on FTS reaction can be found from e. g. review article by Torres Galvis and de Jong. [7] 2. Overview of Fischer-Tropsch Synthesis

Reactions
The exothermic main chemical reaction in the FTS is the conversion of the input H 2 : CO gas mixture into various hydrocarbons and the side-product H 2 O, [25,26] Pasi P. Paalanen obtained his BSc degree on Physical Chemistry from the University of Jyväskylä (Finland) in 2011 After this, he completed his MSc degree on Nanomaterials Science at Utrecht University (the Netherlands) in 2013, where he also continued his PhD work focusing on the Fe-based Fischer-Tropsch Synthesis reaction and its spectroscopic characterization.   where Reaction 2 refers to formation of paraffins (i. e., saturated hydrocarbons) and Reaction 3 refers to formation of olefins (i. e., unsaturated hydrocarbons). The olefins formed as the FTS reaction's primary products may re-adsorb on the catalyst material and be hydrogenated into paraffins via secondary reactions. [26,27] Both CO and H 2 adsorb dissociatively under the FTS reaction conditions giving C, O and H atoms adsorbed on the Fe phases' surface. H and C atoms then recombine to form -CH 2 -units which polymerize to hydrocarbon chains, via the so-called "surface carbide mechanism". [25,26,28] Preferably, the O atoms from the CO dissociation recombine with H atoms and desorb as a H 2 O molecule. If H 2 is unavailable, as can be the case with low H 2 : CO ratio syngas, [5] the adsorbed O atom from the CO dissociation may react with an another CO molecule and be removed as CO 2 . [1,28,29] Several mechanisms for the FTS reaction have been proposed to explain the variety of hydrocarbon products and their derivatives. [1,25,28,30] Within this article for the sake of simplicity, their existence is acknowledged, but only the surface carbide mechanism is considered to be active.
The formation of oxygenate derivatives of hydrocarbons (mainly alcohols) can be taken as a side-reaction, as their contribution to the overall FTS reaction products is usually low. [1,4,31] Saturated alcohols form in the FTS reaction according to reaction equation: [26] n CO þ 2n H 2 Ð C n H 2nþ1 OH þ ðn À 1Þ H 2 O Besides the Reaction 4 for exothermic formation of alcohol oxygenates, also some water soluble carboxylic acids and aldehydes may form in the FTS reaction. [1,4,31] The generic reaction equations for the carboxylic acid and aldehyde formation are omitted here, but can be found e. g. as presented by Dry. [1] The Water-Gas Shift (WGS) reaction is used on an industrial scale in the production of H 2 . [32,33] It follows the reaction equation: The WGS reaction is not usually active for Co-based FTS reaction catalyst materials, but can be active under the FTS reaction conditions for Fe-based catalysts (Reaction 5). [3,25,26,[34][35][36] With increasing reaction temperature, the WGS reaction equilibrium shifts towards the formation of the reactants CO and H 2 O (i. e., it is an exothermic reaction). If the reaction is kinetically limited i. e. CO + H 2 O partial pressures are above the equilibrium expectations, the total forward reaction may benefit from an increased reaction temperature. [33,37] In the presence of sufficient amounts of CO 2 and H 2 , the reverse WGS reaction can also take place forming CO and H 2 O. [29,31,[38][39][40] In the industrial application of the WGS reaction, a Cu/ZnO/Al 2 O 3 catalyst material is used for~210-240°C and a Fe 3 O 4 /Cr 2 O 3 catalyst for~310-450°C reaction temperatures. [32,33,37] For both high and low WGS reaction temperatures, the reaction proceeds with an oxidic active phase unable of CO dissociation. [32,33] This is unlike as in the FTS reaction where CO is dissociatively adsorbed by Fe carbides.
The Boudouard reaction is an unwanted reaction pathway under the FTS reaction conditions causing formation of inactive carbon species ("coke") on the catalyst: [41] 2 CO Ð CþCO 2 (6) The Boudouard reaction (i. e., CO disproportionation) deposits adsorbed carbon species on the catalyst, while releasing CO 2 (Reaction 6). When raising the FTS reaction temperature from LTFT at~225°C to HTFT conditions at~345°C, the rate of carbon deposition becomes significant for Fe-based FTS catalyst materials at 0 300°C. [1,[42][43][44] The Boudouard reaction becomes a major reaction consuming CO, if the CO is exposed to Fe at temperatures � 400°C. [45,46] The generic types of inactive carbon formed are either encapsulating carbon over the Fe phases if only CO is present or filamentous carbon if both CO and H 2 are available. [47][48][49] Small amounts of added H 2 , or H 2 O converted to H 2 via the WGS reaction, can increase the carbon deposition rate from the Boudouard reaction. [1,43,50,51] H 2 removes the adsorbed and catalyst surface sites blocking O atoms from the CO dissociation as H 2 O. The O removal is done with H 2 to H 2 O more efficiently than with CO to CO 2 . The improved O removal probably increases the rate at which CO molecules can adsorb and dissociate on the surfaces of (near-)zerovalent Fe phases (i. e. α-Fe or Fe X C). [29,52] However, the increased carbon deposition only takes place with small amounts of added H 2 to input CO. If higher H 2 : CO ratios (0 5 : 1) with above~15-20 bar total reaction syngas pressure are used, the carbon deposition can be decreased during the FTS reaction in relation to the hydrocarbon formation. [1,53]

Fe Phase Changes
Several possible Fe phase changes take place on a Fe-based catalyst material when moving from the as-synthesized catalyst to a catalyst under FTS reaction conditions and, possibly, to a deactivated catalyst material. The generic Fe phase changes for a Fe-based catalyst follow the following order: [1,25,31] The initial FTS reaction catalyst material is commonly in a form of a Fe oxide, mainly either as α-Fe 2 O 3 (hematite), γ-Fe 2 O 3 (maghemite) or Fe 3 O 4 (magnetite). [25] The Fe oxides are not active for the hydrocarbon-forming FTS reactions (Reaction 2 and Reaction 3), [54][55][56] and are commonly reduced to α-Fe (metallic Fe) prior to starting the FTS reaction. Upon exposure to H 2 : CO in the FTS reaction, the α-Fe then readily reacts to form various Fe carbide phases (Fe x C, x = 2-3). [30,[57][58][59][60][61] Alterna-tively, the Fe oxides can also be exposed directly to the FTS reaction conditions, i. e. to a H 2 : CO atmosphere. In this occasion the FTS active Fe carbides form simultaneously with the Fe oxide reduction to α-Fe. [27,46,[62][63][64][65][66] The reaction equations for reducing Fe oxides to metallic α-Fe with either H 2 or CO are: [25,26,67] Fe oxide reduction with H 2 leads to formation of pure metallic α-Fe (Reaction 7). In general, a H 2 reduction temperature of 0 350°C is required, with adequate reaction time, for reducing Fe x O y to α-Fe. [67] The most stable Fe x C phase (namely θ-Fe3C) decomposes between 500-600°C into α-Fe and carbon species. [68][69][70][71] Thus, only α-Fe is observed as the CO reduction product 0 500°C (Reaction 8). [67,70] With CO (or H 2 : CO) reduction temperatures of 9 500°C, Fe x C phases are stable and can form, i. e. the Fe oxides are reduced to α-Fe and simultaneously carburized (Reaction 9). [57,68,70] Furthermore, the H 2 reduction pathway for the Fe oxides is dependent on the used temperature: [67] With H 2 reduction temperatures above 570°C (Reaction 11), an intermediate Fe x O (wüstite) phase appears, which is absent if the reduction is done 9 570°C (Reaction 10). [67] Experimentally, however, the Fe x O phase can be stabilized also at lower temperatures 9 570°C by a strong interaction between the Fe oxide with other oxides (e. g., SiO 2 ) or by addition of alkali (e. g., K). [67,72] The Fe x O phase is an intermediate also if the reduction is done with CO instead of H 2 , with a reduction temperature 0 570°C. [67] At sufficiently high reduction temperatures, Fe 3 O 4 may go through an undesired solid-state reaction with the support or the structural promoter inert oxide. [3,73] As an example, the formation of FeAl 2 O 4 (i. e., Fe aluminate) with a Fe/γ-Al 2 O 3 catalyst at temperatures at~450-470°C, can take place. [74] Such solid-state reactions are undesirable as Fe in the FeAl 2 O 3 compound stays in the oxidized, inactive ionic state for the FTS reaction. [3,73] The FTS reaction active Fe x C phases form in the presence of CO from reduced α-Fe (Reaction 12): xz Fe þ 2z CO ! z Fe x C þ z CO 2 (12) where x = 2-3 and z = 2x-3. If H 2 is also present during the carburization reaction, then the O atoms from CO dissociation may also be removed as H 2 O instead of CO 2 , as was discussed earlier in this article.

Catalyst Deactivation
The primary pathways for Fe-based catalyst deactivation in the FTS reaction are: [1] i) Fe phase changes -conversion of the FTS reaction active Fe x C into inactive Fe 3 O 4 , ii) Metal carbide sintering -loss of active Fe x C surface area due to increasing Fe particle size, iii) Catalyst fouling -deposition of inactive carbon on the catalyst material, and iv) Chemical poisoning -deposition of chemical compounds of e. g. S, Cl, Br, Pb or Sn, on the Fe x C surface. The conversion of Fe x C phases into Fe 3 O 4 takes place if either of the H 2 O:H 2 or CO 2 :CO ratios are above critical molar ratios. [1,32,33,56,62,[75][76][77] In general, H 2 O is more potent Fe oxidant than CO 2 . [32,62,77] When moving along a fixed bed reactor towards the FTS reactor outlet, [64,78] or towards the inner core of the Febased catalyst material particle, [56] the reducing H 2 : CO mixture is progressively converted into H 2 O and CO 2 . Due to the increasing H 2 O and CO 2 concentrations, the oxidation potential towards forming Fe 3 O 4 from Fe x C increases. [1,56,75,76] Up to around 100-200 μm depth from the particle surface, the Fe-based catalyst particle might remain under reductive environment at FTS reaction conditions. For depths beyond > 200-400 μm from the particle surface, the environment may become oxidative and Fe x C oxidation to Fe 3 O 4 can take over. [56] As Fe 3 O 4 is active for the WGS, but not for the FTS reaction, [1,32,33] with sufficient reaction temperature (> 310°C), the catalyst may become a bifunctional catalyst where the Fe x C catalyst partially re-oxidizes to Fe 3 O 4 , which works as a WGS catalyst and the remaining Fe x C part acts as the FTS reaction catalyst. At lower (< 310°C) temperatures, the Fe 3 O 4 phase is a less efficient WGS reaction catalyst. [32,33] The loss of active Fe surface area via Fe particle growth, i. e. sintering, can cause a loss in the number of available active Fe surface sites towards catalytic reactions. This in turn causes lost activity towards the FTS reaction. Deposition of the inactive carbon, i. e. fouling, takes place in the presence of CO or H 2 : CO, causing gradually lost FTS reaction activity as the active Fe sites become harder to access for the reactants. [1] Excessive inactive carbon deposition via the Boudouard reaction (Reaction 6) and/or carbon filament growth causes fragmentation of the, especially bulk, Fe-based catalyst materials into a fine powder. [1,3,4,76,79] The carbon deposition relates to metal carbide sintering in a two-fold manner: i) initially the availability of the FTS reaction active Fe x C sites may increase due to catalyst fragmentation into smaller Fe particles exposing new active Fe x C surface sites. However, ii) after a longer Time-on-Stream (ToS) the Fe tends to sinter into larger Fe particles, thereby resulting in a loss of the available Fe x C surface area. [25,80] The fragmentation of the Fe-based FTS reaction catalyst materials by carbon deposition and growth is a major issue for their industrial application. [79] The catalyst fragments cause clogging downstream from the FTS reactor and the fragments also need to be separated from the liquid/wax hydrocarbon products. [1,3,4,79] If a Fe-based FTS reaction catalyst material is deactivated by carbon deposition but without fragmentation, it can be regenerated with H 2 reduction at temperatures > 350°C by hydrogenating and removing the carbon deposits. [1] However, if the catalyst material has fragmented, the resulting fine Fe powder still needs to be dealt with to avoid the related issues. [1,3,4,79] In this case, the simple H 2 regeneration is an insufficient method for increasing the Fe catalyst's life-time.
Deposition of chemical compounds on the Fe phases can decrease their reactivity under FTS reaction conditions. Commonly, the addition of electropositive compounds as chemical promoters (e. g., K or Na compounds) can have a beneficial effect on the hydrocarbons-forming FTS reaction. [1,34,36,81] While deposition of electronegative compounds, most commonly Scontaining compounds, have a deactivating effect decreasing CO conversion and thus FTS reaction activity. [1,22,[82][83][84] Work function describes the minimum energy that is required to remove an electron from a chemical compound's highest occupied energy level (the Fermi level). [85] In general, lower the work function, the easier for a chemical phase is to donate electrons towards an adsorbate causing it to react. Within context of current discussion, the electronegative compounds increase the α-Fe/Fe x C surface's work function, thereby lowering its reactivity and ability to adsorb and dissociate H 2 and CO. [86][87][88][89] Deactivation of Fe phases with Scontaining compounds is not only via the mentioned electronic effects, but also via geometric effects; i. e., Fe x C surface site dependent effects. [87,88,90] These electron donation, work function and geometric effects are discussed in more detail later in the text.
Thus, to summarize the discussion so far, a C atom from input CO has five distinct pathways to react under FTS reaction conditions when a Fe-based catalyst material is used ( Figure 2): i) with α-Fe to form Fe x C phases (Reaction 12), ii) with H to form hydrocarbons (Reactions 2 and 3), iii) with C to form inactive carbon (Reaction 6), iv) as CO to form CO 2 via Fe x O y reduction (Reaction 9), v) as CO to form CO 2 via WGS reaction (Reaction 5).
The O from CO dissociation is removed as H 2 O or CO 2 , possibly gradually transforming the FTS reaction atmosphere, from reductive/carburizing H 2 : CO towards oxidative H 2 O-and/ or CO 2 -rich conditions. Under the oxidative atmosphere, FTS reaction active Fe x C may re-oxidize to Fe 3 O 4 , a phase which can catalyze the WGS reaction. The reverse WGS reaction in turn may convert CO 2 back to CO. Ideally, only pathway 2) should be active with all the O atoms preferably removed as H 2 O. This in order to achieve the ideal FTS reaction carbon efficiency from the input H 2 : CO (Reaction 1). [5]

Process Conditions
The main FTS reaction process parameters for controlling the formed hydrocarbon products and the C reaction pathways are i) reaction temperature, ii) input H 2 : CO ratio and iii) the used gas space velocity. The effect of total reaction pressure is less straightforward for Fe-based FTS reaction catalysts, based on the open literature available. It is important to note that the CO conversion level directly by itself does not significantly affect the hydrocarbon selectivities. [34,91] The discussion on FTS reaction process conditions is held with assumption of powdered catalysts, i. e. effects arising from e. g. shaped extrudates are not discussed.

Reaction Temperature
Reaction temperature has the effect of decreasing the α-value (i. e., the hydrocarbon chain length). [1,8,27,91,92] The reaction temperature's effect is more complex and unclear with respect Figure 2. A simplified summary of reaction pathways for C atoms from input CO in a Fe-based Fischer-Tropsch Synthesis (FTS) reaction catalyst particle. The input CO + H 2 gives originally a reductive environment (Red). For C from CO dissociation the pathways to react are: (1) with α-Fe to form Fe x C, (2) with H to form hydrocarbons, (3) with C to form inactive carbon. Without CO dissociation: C pathway (4) may reduce Fe x O y to α-Fe while releasing CO 2 . Removal of O atoms from CO as H 2 O or CO 2 gradually leads to an oxidative environment (Ox) towards the core of the catalyst particle, and the outlet of a fixed bed reactor. If the FTS reaction active Fe x C is oxidized to Fe 3 O 4 , pathway (5) may convert H 2 O + CO to CO 2 + H 2 via the Water-gas Shift (WGS) reaction.
to the olefin formation. With increasing reaction temperature, the olefin content in total may not change or decrease, [1,27,91,93] or might even increase for K-promoted Fe-based catalysts. [27] Conversion of α-olefins into internal olefins increases with increasing reaction temperature. [27,91] On Fe-based FTS catalysts, also the conversion of CO via the WGS reaction, [32,33,37] and via the Boudouard reaction increases with increasing reaction temperature. [1,[42][43][44] This increases the total CO conversion, but away from the FTS reaction hydrocarbon products. The inactive carbon deposition from the Boudouard reaction increases roughly 50 % for every 10°C increase in the reactor bed temperature. [1] How the FTS reaction hydrocarbon production C pathways change in relative terms in comparison to the competing WGS and inactive carbon formation C pathways, against increasing temperature, is not clear for the time being.

Input H 2 : CO Ratio
Increasing the FTS reaction input feed's H 2 : CO ratio increases the hydrogenation ability of the catalyst material. This decreases the α-value via an increased probability for hydrocarbon chain termination and also, the olefin content decreases as any unsaturated C=C bonds are more likely to be hydrogenated. [1,27,91] Short-chained α-olefins are affected more than longer internal olefins, as the short α-olefins are more susceptible towards secondary hydrogenation reactions. [27,91] Also, an increasing input feed H 2 : CO ratio, in general, decreases the inactive carbon deposition rate. [1,3,53] This, while keeping in mind the earlier mentioned exception of a small amount ( � 1 %) of H 2 in the CO feed increasing the inactive carbon deposition in comparison to a pure CO input reactant. [1,43,50,51] Despite the input feed H 2 : CO ratio effect being rather clear -more input H 2 in relation to CO results in more hydrogenation activity -noteworthy emphasis is that the actual H 2 : CO ratio over the different catalyst sites and bed positions may vary. Indeed, the conversion of H 2 : CO in a ratio of~2 : 1 (i. e., ideal FTS), [3,91] WGS activity in forward or reverse directions, [3,29,31,[38][39][40]91] and the possibility of incorporating H and/or C to inactive carbon species, [42,43,[47][48][49] can all modify the local ratio of H 2 : CO along the catalyst particle as well as the reactor bed. The locally altered H 2 : CO ratios can therefore affect the hydrocarbon selectivities during the FTS process. [1,3,56]

Gas Space Velocity
Increasing the reactor gas space velocity has the effect of increasing FTS reaction's hydrocarbon products' olefin content, especially in the short-chained α-olefins fraction. As the space velocity increases, the probability for olefin re-adsorption decreases, thereby decreasing secondary hydrogenation reactions. [1,27,91] The effects of catalyst particle size or catalyst extrudate size/shape on olefin re-adsorption are currently uncertain. The α-value remains unchanged with the changing space velocity. [1,91] The CO conversion decreases, while the FTS and WGS reaction rates, especially with Na and K promoted catalysts, increase with increasing space velocity. [34,35] As the FTS reaction hydrocarbon production rate increases faster than the WGS reaction rate, [34,35] relatively more CO is converted towards the desired hydrocarbon products at higher space velocities than is consumed by the WGS reaction.

Reaction Pressure
The effect of the total FTS reaction syngas H 2 : CO pressure on the hydrocarbon product selectivity is more complex than the effects of the other FTS reaction process parameters. While for a Co-based FTS catalyst the CH 4 selectivity decreases with increasing H 2 : CO pressure within the range of~5-25 bar, no significant change in the CH 4 selectivity with changing reaction pressure is observed for a Fe-based FTS catalyst. [94] This holds when the reactor outlet H 2 : CO ratio (i. e.,~reactor H 2 : CO ratio) is kept constant. [94] With increasing H 2 : CO total pressure within the range~15-25 bar, Todic et al. observed a somewhat increased α-value and decreased olefin content, but also the outlet H 2 : CO ratio dropped, in this way affecting the hydrocarbon selectivities. [91] In experiments at Sasol with Fe-based FTS catalysts, both an increase in the α-value and an increase in the olefin content, as well as no change in either of the hydrocarbon selectivity variables with increasing H 2 : CO reaction pressure, have been observed. [1] An increase in the total H 2 : CO pressure increases the activity of an Fe-based FTS catalyst. [94] This in turn can affect the actual reactor H 2 : CO ratio as well as the partial pressures of H 2 O and CO 2 via an increased CO conversion, thereby altering the hydrocarbon selectivities, as discussed above. In general, the effect of total reaction pressure on the Fe-based FTS hydrocarbon selectivities depend on the individual partial pressures of H 2 , CO, H 2 O and CO 2 present in the FTS reactor, [1,3,91,95] and their relation further depends on the used reaction temperature. [3,95] With regard to the inactive carbon deposition, as briefly mentioned above, with the total reaction syngas pressure of 0 15 bar with a H 2 : CO ratio 0 5, the inactive carbon formation can be suppressed. [1,53] This holds while improving the C pathway selectivity towards the FTS reaction hydrocarbon products. [1,53] With a lower input H 2 : CO ratios of 9 1, the carbon deposition increases with increasing total reaction pressure. [53] According to Todic et al., the WGS reaction extent decreases with an increasing total FTS reaction pressure. [91]

General H 2 and CO Adsorption Principles
The simplified effects of chemical promoters added as electropositive (e. g., Na or K) or electronegative (e. g., S) adsorbates on a transition metal (e. g., Fe) are ( Figure 3): i) Electropositive adsorbates donate electrons towards the transition metal surface, thereby decreasing the measurable surface work function; while the electronegative adsorbates do the opposite. [85,87,100] ii) To modify the electrostatic potential near the transition metal surface, due to dipole moment formed between thepartly or fully ionized -chemical promoter adsorbate and the transition metal surface. [85,100] iii) To possibly offer direct adsorption or reaction sites for reactants (e. g., H 2 or CO). [88,[100][101][102] Under FTS reaction relevant conditions on Fe x C surface, both H 2 and CO adsorb dissociatively due to the electron backdonation from the (near-) zerovalent Fe d-orbitals to the antibonding CÀ O π* and HÀ H σ* molecular orbitals. [86][87][88][89][100][101][102] The dissociation of the adsorbed H 2 and CO on Fe x C surface is affected by i) the availability of electrons in the Fe x C d-orbitals for the back-donation and ii) the stability of the Fe x C surface adsorbed, intermediate states of the H 2 and CO molecules. [86,89,100,101] The addition of alkali elements in moderation increases the Fe d-orbital electron density and availability, i. e., it decreases the surface work function; while the addition of sulphur does the opposite. [86][87][88][89][100][101][102] Via the electrostatic interaction between the formed promoter-surface dipole moment and the reactant adsorbates, the electropositive alkali stabilizes the H 2 À and CO À intermediate states on the Fe x C surface, while the electronegative S destabilizes them. [86,89,100,101] Generally speaking, alkali as an individual chemical promoter in a Fe-based FTS catalyst increases the dissociative adsorption of H 2 and CO. While sulphur as promoter, individually, works vice versa decreasing the dissociative adsorption of H 2 and CO. The dissociative H 2 and CO adsorption promoting effects by K and the inhibiting effects by C, O and S over a surface of metallic state Fe have been experimentally observed by e. g. Benziger and Madix. [88] Furthermore, with the added alkali to the Fe-based FTS catalyst, carbonate or formate species are commonly observed upon CO adsorption, species which are absent without alkali promotion. [100,[103][104][105][106][107] Also, the direct interaction between the alkali and the metallic state Fe surface adsorbed H 2 and CO has been suspected, possibly aiding in the molecules' dissociation. [88,108,109] Similar alkali/CO interaction has also been noted for K promoted Al. [108] Upon addition of alkali (as K), the desorption temperatures for adsorption products of H 2 and CO are increased at the expense of a lower temperature desorption. [88,101,102] That is, the added alkali increases the overall surface binding strength of H 2 and CO adsorption/desorption products. Within the FTS reaction relevant temperatures (9 340°C), the H 2 adsorption products are more strongly bound than the CO products with the addition of alkali K. [88] The addition of electronegative adsorbates (e. g., S or O) has the opposite effect of lowering the desorption temperatures for the adsorbed products of CO and H 2 . [88,101] The measured adsorption energy change for an alkali K promoted Fe metal surface is higher for CO (� + 3.0 kcal/mol) than for H 2 (� + 1.5 kcal/mol). [102] Worth noting is that in an   3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57 often cited study by Dry et al., [110] the heat of H 2 chemisorption was found to decrease due to the addition of K promotion. That is, added K would inhibit H 2 adsorption, in contradiction to the present discussion. However, as noted by Benziger and Madix, [88] the results by Dry et al. were likely obtained on an oxide rather than on a Fe metal (or Fe carbide) surface and the observed H 2 inhibition effect was more likely due to the presence of O atoms rather than the presence of K atoms. [88] Besides the arguments on the electropositivity and electronegativity of the added chemical promoter on H 2 and CO adsorption, also the alkali or sulphur adsorbate binding geometries on Fe surfaces can play a role. [88] Despite the fact that the O atom adsorbates are being more electronegative and the C atom adsorbates having a similar electronegativity as the S atom, the S atom adsorbates on metallic Fe surface showed a more pronounced inhibition effect on CO and H 2 adsorption than either O or C did. This effect was thought to be due to different binding geometries, which the O and C take versus the S adsorbates on the metallic Fe surface. [88] The different binding of O and S elements might be a very relevant factor when a combined NaÀ S promotion is applied in the FTS reaction.

Alkali in Fischer-Tropsch Synthesis
The most common chemical promoters added to a Fe-based FTS catalyst are K and Na. [1,19,34,81,96,111] Other alkali and alkaline earth elements have also been screened for the FTS reaction promotion, but K and Na remain the most suitable, in terms of both FTS reaction activity and selectivity. [34,36,81,95] The addition of alkali by itself to a Fe-based catalyst has the effect on the FTS hydrocarbon selectivity of increasing i) the αvalue and ii) the hydrocarbon product slate olefin content. [1,27,34,52,93,95,111] The alkali hydrocarbon selectivity effect is commonly attributed to an overall decrease in the catalyst's hydrogenation ability, i. e. it effect on the Fe x C surface H : C ratio. The decreased H : C ratio results in a decrease in the hydrocarbon chain termination and in the olefin hydrogenation via primary or secondary pathways. [1,27,52,95,111] The lowered H : C surface ratio with alkali addition can be explained as i) the adsorption energy is increased more for CO (+ 3.0 kJ/mol) than for H 2 (+ 1.5 kJ/mol). [102] Thus, CO is more preferably adsorbed than H 2 when H 2 and CO compete for the alkali-promoted Fe x C surface adsorption sites. ii) the alkali promotor increases the elemental binding strength of H more than that of C on metallic Fe(/Fe x C) surface. [88] The increased binding strength of H to the Fe x C surface results in a depletion of the weakly bound H available for the FTS reaction. [112] Therefore, both alkali addition effects effectively decrease the Fe x C H : C surface ratio available for hydrocarbon-forming FTS reactions.
Increasing the alkali promotion has the intrinsic tendency to increase the inactive carbon deposition C pathway via the Boudouard reaction, along the overall increased CO conversion and dissociation. [1,42,45] Although, a sufficiently low level of added alkali may nevertheless decrease the carbon deposition rate in comparison to an unpromoted Fe-based catalyst, this effect with still observable alkali promotion affects the FTS hydrocarbon selectivity. [1] Alkali also increases the WGS reaction rate, [52,111,113] consuming CO towards the WGS reaction products away from the FTS reaction. [34,36] Alkali furthermore promotes the WGS reaction even without the presence of a transition metal. [105,106] To sum up, as undesired effects from the alkali addition: C pathway selectivity towards FTS reaction hydrocarbon products decreases due to CO being consumed towards the WGS and/or inactive carbon depositing Boudouard reactions. Nevertheless, in the case of low input H 2 : CO ratio, the WGS reaction may become the rate determining step for FTS as the hydrocarbon production becomes dependent on the H 2 from the WGS reaction. [34,35] In this case, the CO-consuming and alkali-promoted WGS reaction becomes a possibly useful property for a FTS catalyst material. [26]

Sulphur in Fischer-Tropsch Synthesis
The effect of sulphur in heterogeneous catalysis is a rather complex phenomenon. [87,90] Too much of added S compounds often result in a catalyst poisoning, [82,83,87,114] while small amounts of added S compounds may have beneficial FTS catalysis effects. S containing impurities in the CO gas can also gradually enrich on the catalyst, thus altering its behaviour. [45] From a review article by Madon and Shaw, some of the possible effects of the S addition on the FTS catalyst are: [87] 1) Avoiding reactor hotspots via decreased CO conversion; 2) Selective S adsorption on different reaction sites i. e. selectively affecting reaction pathways; 3) S itself may become an active site for reactions; and 4) S functions as an electron acceptor i. e. is an electronegative adsorbate on the transition metals and depletes the metals from electrons. As mentioned earlier, the FTS reaction is an exothermic reaction and any hotspots created by the released heat to the reactor bed may cause the bed temperature to increase locally. This results in a decreased α-value, which most easily can be seen by an increased CH 4 selectivity, and/or the inactive carbon deposition via the Boudouard reaction. If a FTS reactor bed is operated non-isothermally, S poisoning decreases CO conversion eliminating the hotspots, which can result in an increased α-value (and therefore in a decreased production of CH 4 ) and in a decreased carbon deposition. That is, the listed point 1) is relatively easy to understand.
As examples of this effect, Dalla Betta, Piken and Shelef observed decreased catalyst activities with added S, as well as decreased CH 4 and increased C 2 + selectivities and also discussed signs of possible non-isothermal reactor bed operation in their data. [115] Similar effects where S addition likely affects the initial reactor bed operating temperature, can be observed in the studies by van Dijk et al. and Zhou et al. with the Fe-based FTS reaction operated at 350°C. [116,117] In the study by van Dijk et al., the added S prevented an initial spike of high FTS catalysis activity in comparison to the non-S containing catalyst. The added S likely prevented any excessive reactor bed heating at the initiation of the FTS reaction, limiting the fast catalyst deactivation from excessive formation of inactive carbon species. [116] The same comments apply to the study by Zhou et al., who exposed their pre-reduced Fe-based FTS catalyst material to H 2 : CO directly at 350°C. [117] This is an initial reaction temperature too high for ensuring isothermal bed operation without excessive reactor bed heating. [71] Definitely pinpointing the effect of the listed points 2), 3) and 4) and their possible combinations in relation to Fe-based FTS catalysis effects is a complex task. In addition, unfortunately, a variety of S poisoning studies with FTS reaction have been conducted with alkali-containing Fe-based catalysts. Thus, mixing both the FTS reaction promotion effects by alkali-S combination and possible poisoning by excessive S addition. This further complicates the task of discussing the S effects on the FTS reaction.
In several studies, positive S promotion effects in the Febased FTS reaction have been observed. However, in these studies, alkali-containing catalysts promoted, intentionally or inadvertently, with synthesis precursors or precipitation agents containing the alkali promoters, have commonly been used. [82,83,96,[118][119][120][121][122] Washing of the prepared catalysts with deionized water in order to remove the alkali, [82,121] or possible S-ions, [121,123] is insufficient in order to have an alkali-or a S-free FTS catalyst material free of promotion or poisoning effects.
From a S poisoning study by Bromfield and Coville, [121] it can be pointed out that Na 2 S was used as the S precursor and Na 2 CO 3 as the catalyst precipitation agent. Despite the subsequent washing to remove the Na atoms from the precipitated Fe catalyst material, with low levels of added S, the catalytic FTS reaction results resemble of what would be expected for the NaÀ S promotion. [121] Presumably, the work led to a later Sasol patent on NaÀ S promotion with Fe-based FTS catalyst materials. [21] Furthermore, in a study by Shultz et al., the initial S addition on K-containing catalysts resulted in an initial promotion effect by KÀ S, but this was followed by S poisoning effects when too much of S was added. [83] In the study, with a low level of COS added as the S-bearing gas compound, a slight increase in the hydrocarbon products olefinicity could be observed. Despite the increased reaction temperature, there was no increase in the CH 4 selectivity. Both effects can be explained by KÀ S promotion. Further addition of COS increased the CH 4 selectivity (and seemingly decreased the α-value), this despite the simultaneously decreased reaction temperature, which should decrease the CH 4 selectivity. And also simultaneously with the increasing amount of S addition, the olefin selectivity decreased for the catalyst. [83] The S poisoning effects of decreased olefin content and increased CH 4 formation can also be observed when different experimental runs with the same reaction temperature are compared. [83] Similar observations have been made by Botes et al., i. e., excessive S addition on alkali containing Fe-based catalysts increases the CH 4 selectivity in the FTS reaction. [22] In the simplest way expressed, the "promotion effects" of increasing CH 4 selectivity and decreasing olefin content by S poisoning of alkali-containing catalyst material can be explained as S being an electronegative adsorbate. That is, the functioning in an opposite way to the alkali addition ( Figure 3). [87] The S addition thus likely increases the overall H : C surface ratio resulting in an increased hydrocarbon hydrogenation and chain termination reactions. However, it is unclear how the S effects, which are not directly related to the S elements electronegativity, i. e. the points 2) and 3) mentioned above, affect the observed selectivities in the Fe-based FTS reaction. Generally speaking, the adsorbate S might selectively block some geometric reaction sites, such as active steps. This might affect the reaction rates and pathways to different degrees in the FTS reaction. [87,90] Some evidence has been observed on the effect of site-blocking for some model catalytic systems. [90] However, how the data obtained from ultra-high vacuum model experiments compares to a more dynamic Fe-based FTS catalyst material behaviour at higher reaction pressures, is still uncertain. [90] As an excessive addition of S eventually covers the active Fe x C surface decreasing its FTS reaction activity, [83,118,119,121,122] and CO dissociation capability, [45,124] it is unlikely that S itself performs as an active site for the FTS reaction. However, some cautious doubt can be expressed as observations of the S compounds possibly assisting in dehydrogenation reactions with Fe-based catalyst materials have been made. [125] In contrast to the cited alkali-containing Fe-based catalyst material studies on the S effects in the FTS reaction, when catalyst precursors free of alkali have been used, the catalyst is directly poisoned by a low level of S addition. S poisoning effects of decreasing: i) α-value, [84] ii) olefin content versus paraffins, [84,116] and iii) overall CO conversion, [82,84] have been observed. Interestingly in the study by Xu et al. with an alkalifree Fe-based catalyst, the CH 4 selectivity remained virtually unaffected despite the increasing S content and decreasing αvalue. [84] This seems to be in contrast to the Fe-based, alkalicontaining catalyst materials for which the CH 4 selectivity increases with excessive addition of S. [22,83] In terms of the various C pathways operative, the effect of the added S might be simply to decrease the overall CO and H 2 dissociation, without altering the relation between the C pathway selectivities. Although the overall carbon deposition decreases with added S, [45,124,126,127] so does the CO conversion. [84] That is, whether S truly suppresses carbon deposition without decreasing the FTS reaction hydrocarbon production, as is sometimes discussed, needs to be compared in relative terms to the CO converted to FTS reaction hydrocarbon products. Such studies do not seem to exist at the moment in the open literature.
On Fe-based WGS reaction catalysts, S behaves as a poison. [33,37] Thus, unlike alkali, S does not promote the WGS reaction.

Alkali-Sulphur in Fischer-Tropsch Synthesis
In light of individual alkali and sulphur promotion properties with respect to the FTS reaction hydrocarbon selectivities: the combined alkali-S promotion effect resembles both. The increased olefin selectivity and CO conversion are "alkali-like", while the decreased α-value in comparison to alkali-only addition is "sulphur-like". The unique effect of decreased CH 4 selectivity to a lower value than expected from the ASF-distribution is solely a combined alkali-sulphur effect. [22] Together these effects enable an increased yield towards short-chained C 2 -C 4 olefins. [20][21][22][23] The unique disruption of the lowered CH 4 selectivity has been proposed to be due to a changing binding geometry of the chemical promoter compound when the S is added as a counter-ion for the alkali promoter. [24] Commonly, the alkali is thought to be in an oxidized form under the FTS reaction conditions, although the exact chemical compound is uncertain. [100,110] Benziger and Madix have observed the oxidation of K atoms by O from CO dissociation in their science studies on metallic Fe surface, in support of the thought for K oxidation. [88] The added S to Na promotion likely replaces the O as the alkali's counter-ion resulting in a NaÀ S compound instead of a NaÀ O phase. [24] From a combination of experimental data and theoretical calculations, thereby applying Density Functional Theory (DFT), Xie et al. concluded that the counter-ion change from O to S i) increases the Na electron donation towards Fe atoms as S is less electronegative than O; and ii) changes the NaÀ S promoter compound binding geometry on the Fe x C surface. The combined effect decreases the availability of H for the hydrogenation reactions, resulting in especially a suppressed formation of CH 4 . [24] As mentioned before, Benziger and Madix have discussed the individual difference between O (and C) and S adsorbates on a metallic Fe surface. [88] Their experimental results and discussion are in support of the different binding modes for O and S. The S adsorbates bind on the metallic Fe surface in a manner, which more effectively inhibits the adsorptive dissociation of H 2 and CO than do O adsorbates, despite O being more electronegative and Fe surface electron depleting than S. [88] The experimental surface science results from Benziger and Madix thus make similar suggestions as do the theoretical results by Xie et al..
Combined alkali-sulphur promotion affects the C reaction pathways for the dissociated CO by: increasing the overall CO conversion, [23,82,96,98] and might also increase, at least the initial, carbon deposition rate. [96,98,127] Although, in the cited studies regarding the carbon deposition, the vast differences in the CO conversion levels have not been accounted for between the different catalysts and promotion levels. The source for inactive carbon deposited on the catalysts is CO. For low CO conversion carbon deposition is necessarily low and vice versa. For meaningful comparison of carbon deposition between various alkali-S promoted catalysts: the carbon deposition needs to be normalized to the total (i. e. integral of) CO conversion over the measurement period. Otherwise, the carbon deposition results can be misleading and the conclusions misguided.
Nevertheless, as significant CO 2 selectivity in the FTS reaction has been observed in the presence of alkali-sulphur promotion with Fe-based catalysts: [23,96,98,127] either increased WGS and/or carbon-depositing Boudouard reactions, both releasing CO 2 , can be suspected to be operative. An increased CO 2 selectivity with the presence of alkali-sulphur promotion would be akin to the earlier discussed alkali-only addition of increasing WGS and carbon deposition reactions, therefore, being an "alkali-like" property. Moreover, after experiments at Sasol, Botes et al. suggest that the CO conversion selectivity towards FTS reaction hydrocarbon production decreases with the added alkali-sulphur promotion. However, precise data on this observation were not given. [22] In precisely quantified terms, how the alkali-sulphur promotion affects the WGS reaction or the inactive carbon depositing Boudouard reaction, in relation to the FTS reaction hydrocarbon production, remains unclear for the time being.
Summarizing the aspect of promotion, Figure 4 gives a schematic overview on the effects of alkali (Na or K), S and combined alkali-S and their effect on the FTS hydrocarbon and C pathway selectivities, while Table 2 further summarizes the Table 2. The effects of various Fischer-Tropsch Synthesis (FTS) reaction process parameters and alkali, sulphur and alkali-sulphur chemical promoters on the C reaction pathways and on the FTS hydrocarbon selectivity in the Fe-based FTS reaction.

Parameter
Hydrocarbon Selectivity CO Conversion C Pathway Selectivity CH 4 α-value Olefins FTS WGS Carbon Deposition Alkali: " or # -increasing or decreasing FTS reaction property against increasing parameter. $ -no effect. ∽ -complex or unclear effect. Na or K atoms as added promoters.
[a] The FTS reaction rate increases faster than the Water-gas Shift (WGS) reaction rate with increasing space velocity.

Fe Carbides in Fischer-Tropsch Synthesis
The most common Fe carbide observed in the FTS reaction related research is the χ-Fe 5 C 2 phase, which is also known as the Hägg carbide. The so-called "ɛ-carbides" (or (pseudo-) hexagonal) Fe carbides are also often observed and labelled as ɛ'-Fe 2.2 C, Fe 2 C, ɛ-Fe 2 C or ɛ-Fe 3 C in the literature. Less common Fe carbides encountered under FTS reaction conditions are θ-Fe 3 C (cementite) and Fe 7 C 3 (the Eckström-Adcock carbide). The mentioned Fe carbides, while excluding the Fe 7 C 3 phase, are also commonly formed as unwanted precipitates during tempering (i. e. heat-treating) of martensitic steels. Fe carbide unit cells in this section are visualized with the VESTA software. [128]

"ɛ-carbides" (ɛ-Fe 3 C and η-Fe 2 C)
The so-called "ɛ-carbides" or Fe carbides with (pseudo-) hexagonal Fe lattices have traditionally been the most difficult Fe carbides to identify due to an unfortunate confusion with the labelling and identification of the ɛ'-Fe 2.2 C, (ɛ-)Fe 2 C and ɛ-Fe 3 C phases. It is often unclear to which actual crystal lattice/ structure the used nomenclature, as per author, refers to.
The ɛ-Fe 3 C phase has been identified by Nagakura. [68] In their Electron Diffraction (ED) study on Fe thin-films, the author solved the diffracting Fe carbide structure to have a space group of hexagonal P6 3 22 (182). This Fe lattice is hexagonally ordered and the C atoms occupy a one third of the octahedral interstitial Fe spaces in a disordered manner ( Figure 5). The disordered C packing was observed from the broadening of the superstructural diffraction peaks, which for the Fe carbides originate from the interstitial C atoms. By decomposing the formed ɛ-Fe 3 C phase to α-Fe and graphite by heating it under vacuum, its stoichiometry was estimated to be around Fe 2.8 C. [68] As commented by Nagakura, [68] the studied ɛ-Fe 3 C phase had the same lattice parameters, when ignoring the weak superstructural C diffraction, as a Fe carbide labelled as "Fe 2 C" with hexagonal Fe lattice observed earlier by Hofer, Cohn and Peebles. [61] That is, the earlier "hexagonal Fe 2 C" and Nagakura's ɛ-Fe 3 C were the exact same carbide.  The stoichiometry for Hofer's Fe 2 C was determined by a comparison to -what at the time was thought to beisomorphism with a "Hägg carbide Fe 2 C". Experimental C : Fe ratio for the carbide was further determined by the total sample weight increase upon carburization. [61] Both, in retrospective, are inaccurate assumptions for determining a carbide's stoichiometry. Later on, after more accurate structural solutions were provided for the Hägg carbide, [129] the original assumptions in the identification of the "hexagonal Fe 2 C" and its stoichiometry were acknowledged by the researchers at the time to be inaccurate. [130] No evidence of an Fe carbide as "hexagonal Fe 2 C" exists as after Hofer, Cohn and Peebles. [61] Unfortunately, the acknowledgement of the mistaken original labelling went unnoticed and several later studies cite a "hexagonal Fe 2 C" carbide originating from the study by Hofer, Cohn and Peebles. [66,131,132] This has resulted in some unfortunate confusion in the characterization of the actual Fe carbide phases in the FTS reaction.
Furthermore, unfortunately, in a French written study by Dirand and Afqir, [133] the "Fe 2 C" carbide with the data from the study of Hofer, Cohn and Peebles was solved to have a space group of hexagonal P6 3 /mmc (194). Dirand and Afqir also acknowledged the existence of the ɛ-Fe 3 C phase indexed as P6 3 22 (182) after Nagakura, but unfortunately did not acknowledge that the carbides were the same.
The reason why the same carbide was solvable with different space groups is due to the originally used diffraction techniques to collect the data. The ED pattern by Nagakura contains the diffraction originating from both Fe and C in the crystal lattice. The X-ray Powder Diffractometry (XRPD) data collected by Hofer, Cohn and Peebles is not sensitive to the low atomic number element C and essentially, only the diffraction from Fe was observable. Thus, the P6 3 22 (182) is the full, proper space group with Fe and C diffraction while the P6 3 /mmc (194) is an incomplete space group only for the Fe diffraction, for the same hexagonal ɛ-Fe 3 C phase observed by both Nagakura and Hofer, Cohn and Peebles. [61,68] The International Centre for Diffraction Data Powder Diffraction File-4 + (ICDD-PDF-4 +) database lists the Dirand and Afqir's solution based on Hofer, Cohn and Peebles's unit cell as a Fe carbide with space group P6 3 /mmc (194) with database number 00-036-1249. As discussed above, Nagakura's crystal structure solution with space group P6 3 22 (182) and database number 04-008-9572 should be used instead. Dirand and Afqir's P6 3 /mmc (194) incomplete lattice solution should not be used in order to avoid any unnecessary confusion for the future identification of the ɛ-Fe 3 C phase.
Nagakura's ɛ-Fe 3 C is not the only Fe carbide phase observed with a (pseudo-)hexagonal Fe lattice. In a spent FTS catalyst material, Barton and Gale observed a similar hexagonal Fe carbide to what Hofer, Cohn and Peebles observed earlier. However, the Fe carbide differed from the previous observations by i) having split XRPD peaks, suggesting that a distortion from the hexagonal Fe lattice was present, and by ii) having significantly different unit cell parameters. [134] As XRPD was used, Barton and Gale could not conclusively solve the observed Fe carbide structure or the space group due to the missing superstructural diffraction from the interstitial C atoms. Barton and Gale assumed a stoichiometry of Fe 2 C after the study by Hofer, Cohn and Peebles for their carbide.
Barton and Gale, however, calculated from their XRPD data that instead of having a purely hexagonal Fe lattice with lattice parameter γ = 120°, their observed Fe carbide crystal lattice had a distortion from the Fe lattice hexagonality where γ = 120.92°. This angle distortion accounting for the observed peak splitting in the X-ray diffraction patterns. As possible space group solutions for the crystal lattice, Barton and Gale gave trigonal P 3 À m1 (164), hexagonal P6 3 /mmc (194) and orthorhombic Pbcn (60). [134] From tempering studies on martensitic steel, Hirotsu and Nagakura solved a Fe carbide crystal structure on the basis of its electron diffraction patterns, [135] which resembled the earlier hexagonal ɛ-carbide (i. e., the ɛ-Fe 3 C phase) observed by Hofer, Cohn and Peebles and Nagakura. [61,68] However, the Fe carbide was different to Hofer, Cohn and Peeble's observations as the Fe lattice was not purely hexagonal but rather a pseudohexagonal one with the Fe lattice distorted from hexagonality. Hirotsu and Nagakura labelled their Fe carbide as η-Fe 2 C with an orthorhombic space group Pnnm (58) (Figure 6). [135] The C atoms in η-Fe 2 C structure occupy half of the octahedral interstitial spaces in the Fe lattice, in a regular manner. [135] Hirotsu and Nagakura have proposed orthorhombic unit cell dimensions, after their η-Fe 2 C phase, for the Fe carbide observed by Barton and Gale. [136] The crystal structures of the ɛ-Fe 3 C and η-Fe 2 C phases, with the unit cell by Hirotsu and Nagakura corresponding to the Barton and Gale's Fe carbide, are depicted in Figure 7. The carbides indeed have very similar hexagonal Fe planes with differences of, that for i) η-Fe 2 C γ = 120.47°instead of γ = 120.0°a s for the purely hexagonal lattice of ɛ-Fe 3 C and ii) there are no empty octahedral interstitial Fe lattice spaces in the lattice of η- Figure 6. The unit cell of the orthorhombic η-Fe 2 C Pnnm (58). The Fe lattice is distorted from hexagonality and the C atoms occupy half of the octahedral Fe lattice interstitial spaces in an ordered manner.
Fe 2 C. The crystal structures of ɛ-Fe 3 C and η-Fe 2 C are similar and closely related. That is, the orthorhombic structure of the η-Fe 2 C phase with the orthorhombic unit cell dimensions proposed by Hirotsu and Nakagura is a proper solution for the Fe carbide observed by Barton and Gale. It is the crystal structure for what is occasionally labelled in the literature as a pseudo-hexagonal (ɛ'-Fe 2.2 C or Fe 2 C) Fe carbide phase. [59,137] The crystal structure for the η-Fe 2 C phase by Hirotsu and Nakagura is listed with database number 04-013-9510 in the ICDD-PDF-4 + 2016 database.
To summarize the discussion so far, the labelling of a Fe carbide in the FTS reaction as (ɛ-)Fe 2 C phase is improper and should not be used further as there is no evidence that such a Fe carbide exists. What can be grouped as the "ɛ-carbides" are two different carbides: ɛ-Fe 3 C with hexagonal space group P6 3 22 (182) in which the Fe lattice is purely hexagonal and C atoms occupy a third of the octahedral Fe interstitial spaces in a random, disordered manner; η-Fe 2 C with orthorhombic space group Pnnm (58) where the Fe lattice is distorted from the hexagonal symmetry and C atoms occupy half of the octahedral Fe interstitial spaces in a regular, ordered manner. A "pseudohexagonal Fe carbide" either labelled as a Fe 2 C or ɛ'-Fe 2.2 C, should not be used to describe a Fe carbide with distorted Fe lattice but rather the orthorhombic η-Fe 2 C should be used instead. The crystal lattice of the η-Fe 2 C phase contains the distorted (pseudo-)hexagonal Fe lattice.

Hägg Carbide (χ-Fe 5 C 2 )
The most commonly observed Fe carbide in the FTS reaction is the χ-Fe 5 C 2 phase, which is mostly properly and consistently identified. The first observations of the Hägg carbide dates back to 1934, [138] and the carbide's crystal structure was later solved to be monoclinic, with a stoichiometry of Fe 5 C 2 . [129] Retief has further improved the accuracy of the structural solution for the known χ-Fe 5 C 2 phase with a monoclinic space group C2/c (15) (Figure 8). [139] Du Plessis et al. have argued that the proper space group for the χ-Fe 5 C 2 phase is not a monoclinic C2/c (15) but a triclinic P � 1 (2). [140] However, Leineweber et al. have explained the observations by du Plessis et al. being due to anisotropic microstrain broadening rather than a different unit cell symmetry. [141] The crystal structure solved by Retief is a proper solution for the Hägg carbide χ-Fe 5 C 2 and is listed in the ICDD-PDF-4 + 2016 database with number 00-051-0997. The crystal structure solutions provided by Leineweber et al. are also accurate for the Hägg carbide. [141] Figure 7. The crystal lattices of a) the hexagonal ɛ-Fe 3 C P6 3 22 (182) viewed along the c-axis, b) the orthorhombic η-Fe 2 C Pnnm (58) viewed along the b-axis, so that the (distorted-)hexagonal Fe planes are in plane with the paper. The η-Fe 2 C unit cell is after Hirotsu and Nagakura corresponding the Fe carbide by Barton and Gale. [136] Blue, dashed line -the respective unit cells for each crystal lattice, Black, solid line -In plane Fe atom positions used for measuring the lattice angle parameter γ. For the ɛ-Fe 3 C γ = 120.0°i. e. the Fe lattice is hexagonal, while for η-Fe 2 C γ = 120.47°i. e. the Fe lattice is distorted from hexagonality. In the η-Fe 2 C lattice the interstitial Fe spaces are occupied by C to a higher degree than in the ɛ-Fe 3 C lattice.

Cementite (θ-Fe 3 C)
As being the most thermally stable Fe carbide, [57,142] cementite θ-Fe 3 C is -akin to χ-Fe 5 C 2 -a well-studied and consistently identified Fe carbide phase. Already the early studies have identified the cementite θ-Fe 3 C crystal structure with an orthorhombic space group Pnma (62). [143] Later, Wood et al. have applied Neutron Powder Diffraction (NPD) to further improve the cementite θ-Fe 3 C structure solution. These experiments provide to date the most accurate crystal structural solutions for the θ-Fe 3 C phase, at various temperatures. [69] The crystal structure of θ-Fe 3 C by Wood et al., at room temperature, is available in the Crystallography Open Database (COD) with the database number 9014027 (Figure 9). [144]

Eckström-Adcock Carbide (θ-Fe 7 C 3 )
The Eckström-Adcock carbide is rather rarely observed in the FTS reaction, but has been known to exist since the initial observations of a new carbide by Eckström and Adcock. [145] It may form during the FTS reaction after extended times under industrial high-pressure reaction conditions. [1] The initial identification of the carbide observed by Eckström and Adcock was done as a hexagonal Fe 7 C 3 with a space group P6 3 mc (186) by Herbstein and Snyman. [146] However, in later studies Fruchart et al. proposed the crystal structure of Fe 7 C 3 to be an orthorhombic Pnma (62). [147] In a single crystal ED study by Audier, Bowen and Jones, the orthorhombic space group was agreed over the hexagonal one. [148] Therefore, the preferred crystal structure for the Eckström-Adcock carbide Fe 7 C 3 has an orthorhombic space group Pnma (62) by Fruchart et al. (Figure 10). [147] The orthorhombic crystal structure for the Fe 7 C 3 phase is available in the ICDD-PDF-4 + database with the database number 04-002-8165.
To make the distinction and to avoid any confusion between the Fe 7 C 3 crystal lattices, the orthorhombic Fe 7 C 3 Pnma (62) is arbitrarily labelled as θ-Fe 7 C 3 akin to same space group θ-Fe 3 C. The use of various Greek prefix letters, [149] to name the Fe carbides is rather arbitrary and not systematic.
The crystal lattice details for the discussed Fe carbides observable in the FTS reaction research i. e. ɛ-Fe 3 C, η-Fe 2 C, χ-Fe 5 C 2 , θ-Fe 3 C and θ-Fe 7 C 3 are listed in the Table 3. The crystal structure solutions are available from the listed authors in Table 3 and either the ICDD-PDF-4 + or COD databases.

Fe Carbide Hyperfine Fields
Just as was the case with their X-ray and electron diffraction data based crystal lattice solutions, the literature data with various Fe carbides identified in FTS research with Mössbauer Absorption Spectroscopy (MAS) can be challenging to follow due to the unclear nomenclature, as per author.
As based on the solved crystal structures for the Fe carbides, by the authors listed in Table 3, the total count and the site ratios for the crystallographically non-equivalent Fe positions can be determined. Each non-equivalent Fe position should ideally give a distinct MAS hyperfine field component with fields' spectral contribution ratios following the non-equivalent Fe site ratios. [150] Determining the non-equivalent Fe positions from the crystal structures gives the ideal Fe site ratios as: i) the ɛ-Fe 3 C and η-Fe 2 C phases have one equivalent Fe position, ii) the χ-Fe 5 C 2 phase has three non-equivalent Fe positions in ratio of 2 : 2 : 1, iii) the θ-Fe 3 C phase has two non-equivalent Fe positions in ratio 2 : 1 and iv) the θ-Fe 7 C 3 phase has five nonequivalent Fe positions in ratio 2 : 2 : 1 : 1 : 1.
Le Caër et al. have discussed the MAS hyperfine field assignment of the "ɛ-carbides" on the basis of experimental data gathered from literature. [60] As Le Caër et al. note, the "ɛcarbides" can be grouped around two hyperfine field values of 24.5 � 0.8 T and 17.2 � 0.3 T (at room temperature). Furthermore, increasing the amount of interstitial C atoms in the Fe lattice decreases the measured hyperfine field. [60,[151][152][153] Thus, the field 24.5 � 0.8 T belongs to a "ɛ-carbide" with a low C content and the field 17.2 � 0.3 T for a "ɛ-carbide" with a higher C content. Recalling from the earlier discussion on the crystal  structures of the Fe carbides: for the ɛ-Fe 3 C phase third of the interstitial Fe lattice sites are occupied by C atoms, while for the η-Fe 2 C phase half of the interstitial sites are occupied by C. That is, the η-Fe 2 C phase has the higher C content in its crystal structure. Therefore, the hyperfine field 24.5 � 0.8 T is assigned for the one equivalent Fe site in the ɛ-Fe 3 C crystal structure and 17.2 � 0.3 T for the one Fe site in the η-Fe 2 C phase. Occasionally, also an additional minor intensity hyperfine field at~13.0 T has been assigned for "ɛ-carbides". [60,154,155] However, whether this is correct and from where the low~13.0 T hyperfine field arises from, is uncertain. The two assigned hyperfine fields are the primary ones for the ɛ-Fe 3 C and η-Fe 2 C phases for correctly identifying the carbides.
The experimentally assigned, averaged hyperfine fields from the cited studies, for the Hägg carbide χ-Fe 5 C 2 are the fields 21.4 � 0.3 T, 18.1 � 0.4 T and 10.9 � 0.9 T with respective spectral contribution ratio of~2 : 2 : 1. [59,77,155,156] The spectral contribution ratio is in agreement with the expected non-equivalent Fe site ratios for the χ-Fe 5 C 2 phase. In addition to the commonly assigned hyperfine fields, Barinov, Protasov and Surikov have proposed an additional field of~16.0 T to account for a possible anisotropy in the χ-Fe 5 C 2 crystal structure. [156] The proposal by Barinov, Protasov and Surikov perhaps could account for the earlier discussed anisotropy phenomena, as observed by Leineweber et al. for the χ-Fe 5 C 2 phase. [141] However, whether or not such anisotropy is present, the three assigned hyperfine fields are sufficient for identifying the χ-Fe 5 C 2 phase.
The experimentally assigned, averaged hyperfine fields for cementite θ-Fe 3 C can be obtained from studies by Bi et al. and Kniep et al. and have values of 20.8 � 0.3 T and 20.0 � 0.3 T. [157,158] However, as the values are very close to each other, commonly only one average hyperfine field value of 20.8 � 0.5 T is used for the MAS fitting of θ-Fe 3 C. [59,132,159,160] Neither Bi et al. nor Kniep et al. report the MAS-based relative spectral contribution areas for the θ-Fe 3 C assigned hyperfine fields. Thus, how the crystallographically expected 2 : 1 non-equivalent Fe site ratio is assigned to the hyperfine fields, is unclear for the time being for the θ-Fe 3 C phase.
The assignment of the hyperfine fields from the experimental literature data for the θ-Fe 7 C 3 phase is problematic as the carbide is less often observed than the other carbides, and therefore less studied. For the orthorhombic crystal lattice of the θ-Fe 7 C 3 phase, five non-equivalent Fe sites are expected with corresponding hyperfine fields. However, only three fields have been experimentally assigned so far in the literature. Averaging the hyperfine fields from the studies of Bi et al. and Yamada et al. gives field assignments as 22.9 � 0.1 T, 18.5 � 0.1 T and 16.3 � 0.1 T. [157,160] Bi et al. have also assigned a fourth field for their (θ-)Fe 7 C 3 phase as 21.0 T which might belong to the θ-Fe 3 C impurity from the sample synthesis. [157] In a theoretical MAS study, Liu et al. have predicted the hyperfine fields of 22.9 T, 22.9 T, 19.9 T, 18.7 T and 17.4 T for the orthorhombic θ-Fe 7 C 3 phase. [153] The assignments by Liu et al. are relatively close to the experimental assignments by Bi et al.. [157] That is, if the hyperfine field 21.0 T is assumed belonging to the θ-Fe 7 C 3 phase rather than an impurity θ-Fe 3 C phase and if the same field of 22.9 T is indeed assignable for two non-equivalent Fe positions. Bi et al. considered the Fe 7 C 3 phase to only have three non-equivalent Fe sites, [157] as based on the hexagonal Fe 7 C 3 structure by Herbstein and Snyman. [146] However, as discussed earlier, the hexagonal crystal lattice for the Fe 7 C 3 phase is not a fully satisfactory crystal structure solution as based on the diffraction studies. The orthorhombic θ-Fe 7 C 3 crystal structure should be used instead. [148] None of the cited studies assigns spectral contribution ratios for the [a] Lattice parameters are translated according to Hirotsu and Nagakura, [136] to correspond the distorted Fe carbide observed by Barton and Gale. [134] [b] Crystallography Open Database (COD) number.
[c] The original article written in French by Fruchart et al. is not readily available but some discussion on the crystal structure can be found in publication by Audier, Bowen and Jones. [148] assigned hyperfine fields for the θ-Fe 7 C 3 phase. Therefore, the hyperfine fields corresponding to the specific non-equivalent Fe sites cannot be definitively identified from the literature for the θ-Fe 7 C 3 phase at the moment. To conclude, the MAS hyperfine field assignments for the θ-Fe 7 C 3 phase are uncertain for the time being. However, as based on both experimental and theoretical data, the possible field assignments for the expected five sites for the orthorhombic θ-Fe 7 C 3 phase are: 22.9 � 0.1 T twice, 18.5 � 0.1 T and 16.3 � 0.1 T, with the field 21.0 T being uncertain. The hyperfine field spectral contribution ratios corresponding to the non-equivalent Fe site ratios are unassigned.
The assigned hyperfine fields for MAS fingerprinting of the Fe carbides encountered in the FTS reaction research, as based on the experimental literature data, are summarized in the Table 4. Hyperfine fields based on theoretical calculations by Liu et al., are reproduced in the table for comparison. [153]

Formation of Fe Carbides
Fe carbides form by dissolution of carbon atoms into the interstitial spaces in the α-Fe lattice. The formation of the various Fe carbides is primarily dependent on the temperature where the α-Fe lattice is exposed to an available source of carbon atoms. Fe in Fe carbides has a valency state close to zero, as for example in the χ-Fe 5 C 2 phase. [24] Fe carbides can form from α-Fe by: i) gas phase carburization where e. g. the CO, either from pure CO or H 2 : CO, acts as the C source; [57,59] ii) heat treating the pre-formed Fe carbides to interchange the carbide phases, [57,68,77] and iii) tempering preformed martensite (or austenite) resulting in Fe carbide precipitation. [142,161] The ɛ-Fe 3 C, η-Fe 2 C, χ-Fe 5 C 2 and θ-Fe 3 C phases form relatively easily from an α-Fe/martensite precursor with any of the listed methods, i. e. by varying the heat treatment or gas phase carburization temperature. The formation of the θ-Fe 7 C 3 phase seems to be a more complex phenomenon, as based on the available literature. It will be discussed separately after the formation conditions of the more common Fe carbides.
The observation temperatures for the ɛ-Fe 3 C, η-Fe 2 C, χ-Fe 5 C 2 and θ-Fe 3 C phases from the literature are gathered with the following criteria: i) the reaction takes place from metallic, zero valency α-Fe, in order to decouple the Fe reduction from carburization effects, ii) reaction treatment time is reported, and iii) it is possible to re-evaluate the reported data to correspond to the nomenclature, as defined in Table 3 and Table 4. From criteria iii) follows that only MAS-and ED-based results are included to evaluate the presence of the ɛ-Fe 3 C and η-Fe 2 C phases. XRPD and Thermomagnetic (TM) data are also included for evaluating the χ-Fe 5 C 2 and θ-Fe 3 C phases' formation behaviour. The references to the studies used to collect the data points in Figure 11 are collected in Table 5.
The collected data points in Figure 11 refer to the experimental observations on the Fe carbides after respective Fe carbide forming treatment. Possibly coexisting phases of α-Fe, martensite or Fe oxides are excluded from the figure.
The expected observation temperatures and carbide-tocarbide transition temperatures for the Fe carbides with 5 h treatment time can be roughly estimated as ( Figure 11): e-Fe 3 C ð100 À 265 � CÞ < h-Fe 2 C ð160 À 350 � CÞ < c-Fe 5 C 2 ð240 À 350 � CÞ < q-Fe 3 C ð310 À 600 � CÞ [a] Determined on the basis of the crystal structures from authors in Table 3. The nominal relative ratios in parentheses () could not be conclusively assigned.
[b] Mean hyperfine fields from the experimental literature data with their Standard Deviations (STDs).
[e] Commonly, a single (averaged) hyperfine field is used for the θ-Fe 3 C phase in the experimental literature data.
[f] The assignment of hyperfine field 21.0 T for the θ-Fe 7 C 3 phase is uncertain.
These expectations on the Fe carbide thermal stabilities and the formation temperatures mainly compare well with the expectations presented by other authors. [57,68,133,142,161] This holds bearing in mind the unclear nomenclature in the literature regarding the "ɛ-carbides" i. e. the identification of the ɛ-Fe 3 C and η-Fe 2 C phases.
A good example on the behaviour of the ɛ-Fe 3 C and η-Fe 2 C phases is from a study by Boellaard, van der Kraan and Geus. [107] With the FTS reaction experiments at reaction temperatures between 200-275°C, either increasing the reaction time or the temperature caused Fe carbide formation to shift from the ɛ-Fe 3 C towards η-Fe 2 C phase. Already at~200°C with~1 h reaction time, the η-Fe 2 C phase is the Fe carbide to form in larger quantities over the ɛ-Fe 3 C phase. With further increasing the reaction time or temperature, the η-Fe 2 C phase content increases. [107] In a martensite tempering study by Williamson, Nakazawa and Krauss, [132] at 200°C the ɛ-Fe 3 C phase (component M2 in their study) disappears if the tempering time is increased above 1 h. At temperatures < 150°C, ɛ-Fe 3 C is stable and is able to withstand longer tempering times. [132] In short, the ɛ-Fe 3 C phase is the least stable Fe carbide that easily transforms to the η-Fe 2 C phase, with increasing treatment time or temperature. Minor quantities of the ɛ-Fe 3 C phase may nevertheless be observed with temperatures up to 250-270°C. [58,65,107,155,162] Within the temperature range of~240-350°C, both of the η-Fe 2 C and χ-Fe 5 C 2 phases may co-exist. For these carbides, the reaction exposure time seems to be less influential than in the case of relative formation of the ɛ-Fe 3 C and η-Fe 2 C phases. For at least up to 50 h in the FTS reaction at 300°C, both the η-Fe 2 C and χ-Fe 5 C 2 phases can be observed, although at the higher temperatures the χ-Fe 5 C 2 phase becomes the Fe carbide to form in higher quantities. [93] After a short 3 h FTS reaction, Niemantsverdriet et al. were able to observe low quantities of the η-Fe 2 C phase even at 350°C, while the χ-Fe 5 C 2 and θ-Fe 3 C phases were the major Fe carbide phases. [59] Also, after a 24 h FTS reaction at 340°C, Wezendonk et al. observed a significant presence of the η-Fe 2 C phase. [77] Few catalyst composition properties are known to increase the stability of the ɛ-Fe 3 C or η-Fe 2 C phases (i. e., "ɛ-carbides") in relation to the χ-Fe 5 C 2 phase. Addition of i) alkali, as at least if added as Na or K, [27,55,63,163] or ii) SiO 2 as an inert oxide, [60,72,155] to the Fe catalyst composition can stabilize the ɛ-Fe 3 C and/or η-Fe 2 C phase(s). Also, iii) smaller Fe particle size tends to favour the formation of the ɛ-Fe 3 C and/or η-Fe 2 C phases over the χ-Fe 5 C 2 . [59,155,164,165] Unfortunately, from the literature it is not possible to conclusively determine which of the "ɛ-carbides" i. e.

ChemCatChem
Reviews doi.org/10.1002/cctc.202000535 ɛ-Fe 3 C and/or η-Fe 2 C is stabilized, in relation to the mentioned catalyst composition properties. Above the temperature of~300-350°C, the χ-Fe 5 C 2 and θ-Fe 3 C phases are the expected main Fe-carbide phases. The highest temperature stability for the χ-Fe 5 C 2 phase included in this review, has been observed to be up to~400°C from a martensite tempering study. [161] Also, the χ-Fe 5 C 2 phase has been observed after heat treating of preformed Fe carbides under vacuum at 400°C for short periods of time (~0.1 h). [68] Above 350°C, the preferred Fe carbide to form is θ-Fe 3 C. [59,68,166,167] From martensite tempering studies, Nagakura et al. have reported the formation of θ-Fe 3 C after tempering martensitic steel at 267°for 1 h, [161] and Williamson, Nakazawa and Krauss after 250°C for 5 h. [132] Both of these temperatures seem very low for θ-Fe 3 C formation when comparing to the other literature data, and are excluded from the Figure 11.
Decomposition of the θ-Fe 3 C phase into α-Fe and C under vacuum heat treatment may start already slowly at a temperature as low as 377°C and becomes rapid at > 527°C as has been observed by Wood et al. [69] Nagakura in comparison observed that the θ-Fe 3 C decomposition to α-Fe only started at 600°C under vacuum heat treatment. [68] The increase in the most thermally stable Fe carbide's (i. e., θ-Fe 3 C) decomposition temperature has been explained by Podgurski et al. to be due to the formed Fe carbide oxidation shell. [71] The shell can retard the decomposition of the θ-Fe 3 C core to α-Fe and C. The Fe carbides readily form a~4-5 nm oxidation layer if any O is available for the oxidation reaction. [168] As mentioned, the main Fe carbide phase determining factor is the temperature. For the gas phase carburization, factors such as H 2 : CO ratio and reaction pressure, do not seem to play a significant role when the Fe carbides are formed from a zero valency α-Fe.
For example, with the FTS reaction at 250°C with H 2 : CO ratios of 1 : 1 (0.5 h), [58] 2 : 1 (1.1 h), [107] and 3 : 1 (0.66 h), [155] only the ɛ-Fe 3 C (minor phase) and the η-Fe 2 C (major phase) phases are observed along with differences in the extent of α-Fe carburization. At 270°C after 0.66 h of FTS reaction with 10 : 1 H 2 : CO, Tau et al. observed the formation of the χ-Fe 5 C 2 phase in addition to the ɛ-Fe 3 C and η-Fe 2 C phases from α-Fe, [65] as would be expected on the basis of the reaction temperature ( Figure 11). In a systematic study, Jung et al. have observed that an increasing H 2 : CO ratio decreases the rate of carburization of α-Fe. [164] Despite the CO-rich atmosphere being favoured for the increased carburization rate, small amounts of (~1 %) H 2 can still aid the carburization more efficiently than CO by removing the surface O from CO dissociation, [29,52] as was also discussed in the previous sections. All in all, despite the changing H 2 : CO ratios, the reaction temperature is still the determining factor for Fe carbide formation.
It is, however, worthy to emphasize again that when carburization with H 2 : CO is done with a Fe oxide as the starting material instead of α-Fe, the Fe carbide formation is dependent on the simultaneous Fe reduction behaviour. [63,66,78,170,171] This can cause perceived deviations from the formation regions in Figure 11.
The effect of the overall reaction pressure under which the α-Fe phase is carburized from the gas phase is somewhat unclear from the literature, as systematic studies are lacking. It seems, however, that the increased pressure does not change the expected order of the Fe carbide formation, again, when the carburization is done from the α-Fe phase. Amelse et al. performed 300°C FTS reaction (H 2 : CO 3 : 1) for 24 h atmospheric pressure, [162] and Jiang et al. 300°C FTS reaction (H 2 : CO 1 : 1) at 10 bar pressure for 50 h. [93] In both studies, only the η-Fe 2 C and χ-Fe 5 C 2 phases were observed as the Fe carbides. After 30 bar FTS reaction (H 2 : CO 2 : 1) for 0.3 h at 170°C, Xu et al. observed only the ɛ-Fe 3 C and η-Fe 2 C phases as the Fe carbides. [154] With 23 bar FTS reaction (H 2 : CO 3 : 2) at 235°C for 12 h, only ɛ-Fe 3 C and η-Fe 2 C phases were present in a study by Wang et al., [165] which is a quantification result comparable to 1.03 bar FTS reaction (3.9 h, 250°C, H 2 : CO 2 : 1) by Boellaard, van der Kraan and Geus. [107] In each of the mentioned studies, the Fe carbide formation follows the expectations from the reaction temperature regardless of the applied reaction pressure. As noted by Xu et al., the elevated pressure only seems to increase the rate of carburization from α-Fe (see Supplementary Figure 2, in their study). [154] The formed Fe carbide phases, however, are not affected.
Finally, it is important to note that the formation of the θ-Fe 7 C 3 phase is a special case, as it does not follow the temperature formation regions, presented in Figure 11. In literature, the θ-Fe 7 C 3 phase is either observed to form at extreme conditions, such as at 60 kbar with > 1100°C from a mixture of graphite and α-Fe, [172] or via laser pyrolysis of Fe(CO) 5 under C 2 H 2/4 . [157,160,173] A more interesting case for the FTS reaction related discussion in the realm of θ-Fe 7 C 3 formation is observed in a study by Tajima and Hirano, where Ba-promoted Fe 3 O 4 was carburized with pure CO at unspecified reaction pressure. [174] Within the temperature range of 300-375°C, the Ba-Fe 3 O 4 carburized into a mixture of the χ-Fe 5 C 2 and θ-Fe 7 C 3 phases. [174] Without Ba, only the χ-Fe 5 C 2 phase was formed within the same temperature range from the unpromoted Fe 3 O 4 . [174,175] In a single crystal ED study of the θ-Fe 7 C 3 phase by Audier, Bowen and Jones, single crystals of the θ-Fe 7 C 3 phase were formed by carburizing the α-Fe phase under 3 : 1 CO : CO 2 at 500°C. [148] As the carburization to, at least some amounts of, the θ-Fe 7 C 3 phase took place without Ba or a Fe oxide, perhaps, the CO 2 might play a role in facilitating the formation of the θ-Fe 7 C 3 phase. The θ-Fe 7 C 3 phase has also been observed in catalysts extracted from industrial FTS reactors after long reaction times at high-pressure. [1] A mechanical synthesis route for the θ-Fe 7 C 3 phase has been attempted by Barinov, Tsurin and Surikov. [176] The route has more likely to have only formed the χ-Fe 5 C 2 phase, [156] rather than any observable quantities of the θ-Fe 7 C 3 phase.
Regarding the θ-Fe 7 C 3 phase thermal stability, a heat treatment of the θ-Fe 7 C 3 phase under He leads to the phase's decomposition, with decomposition temperature depending on the O atom impurity content in the θ-Fe 7 C 3 crystal lattice. [173] High O atom impurity content causes the θ-Fe 7 C 3 phase to decompose to Fe oxides and α-Fe between 300-400°C. O atom free θ-Fe 7 C 3 phase can survive up to~400-500°C before decomposing directly to α-Fe and C atoms. [173] To conclude, the formation of the θ-Fe 7 C 3 phase does not follow the same route as for the ɛ-Fe 3 C, η-Fe 2 C, χ-Fe 5 C 2 or θ-Fe 3 C phases, but might have its own formation route. Nevertheless, θ-Fe 7 C 3 phase formation is possible under relatively mild and FTS reaction relevant conditions, although more systematic research is needed on its formation specifics.

Summary and Outlook
Within this review article, the possible "C pathways" for C from CO and the associated chemical reactions in Fe-based Fischer-Tropsch Synthesis (FTS) were reviewed. Especially, effects from NaÀ S promotion were discussed. Furthermore, Fe carbides, which are considered to be the FTS active Fe phases, often have an unclear identification and their formation conditions were critically assessed based on the existing literature.
The following pathways can be active in Fe-based FTS reaction: C from CO dissociation can react i) with α-Fe to form Fe carbides, ii) with H from H 2 to form hydrocarbons, iii) with another C to form inactive carbon. Without CO dissociation, CO may iv) reduce Fe x O y or v) react via the WGS reaction with both pathways converting CO to CO 2 . As catalyst deactivation pathways: the Fe x C phases may re-oxidize to Fe 3 O 4 under FTS reaction conditions if the local H 2 O:H 2 or CO 2 :CO ratios become too high. The Boudouard reaction as pathway iii) results in hindered reactant access to the catalysts surface and catalyst fragmentation. Modifying the process conditions and/or the catalyst compositions can be used as ways to modify the Febased FTS hydrocarbon and C pathway selectivities. Input H 2 : CO ratio, reaction temperature and reactor space velocity are the main process parameters for modifying the α-value or olefin content of the FTS reaction produced hydrocarbons.
Addition of alkali individually increases the α-value and olefin content in FTS-synthesized hydrocarbons, while the individual addition of S does the opposite. Addition of combined alkali-S in the molar ratio of � 2 : 1 increases the αvalue less than with alkali only addition, and as an unique alkali-S promotion feature, decreases the CH 4 selectivity to a value lower than expected from the Anderson-Schulz-Flory (ASF) distribution. Also, the olefin content increases with the alkali-S addition akin to alkali-only addition. Possibly, S replaces O as the alkali counter-ion changing the chemical promoter compound binding on the Fe x C surface, thereby affecting the promotion effects by alkali-S. On the C pathways, however, added alkali increases the inactive carbon deposition and WGS reactions, thereby decreasing the CO conversion towards hydrocarbon products. Unfortunately, the effect of individual S on the C pathways is still unclear. An increased CO 2 selectivity suggests an increased inactive carbon formation and/or WGS C pathways for alkali-S promoted catalysts -effects akin to alkalionly addition.
For further research, the exact effects of the different FTS active Fe carbides on the C pathways are still unclear. Determination of these effects has been hindered in the past by the imprecise and unclear Fe carbide nomenclature between different studies and characterization techniques. More precise understanding on the Fe carbide properties might offer an additional tool for altering the C pathways and overall C efficiency towards the hydrocarbon synthesis.