See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/258685985 Fischer–Tropsch Synthesis: Catalysts and Chemistry Chapter · August 2013 DOI: 10.1016/B978-0-08-097774-4.00729-4 CITATIONS READS 157 28,544 10 authors, including: Jan van de Loosdrecht F.G. Botes Council for Scientific and Industrial Research, South Africa Suid Afrikaanse Steenkool en Olie 64 PUBLICATIONS 2,745 CITATIONS 19 PUBLICATIONS 838 CITATIONS SEE PROFILE Ionel Mugurel Ciobîcă 34 PUBLICATIONS 1,948 CITATIONS SEE PROFILE Alta C Ferreira Suid Afrikaanse Steenkool en Olie 10 PUBLICATIONS 285 CITATIONS SEE PROFILE SEE PROFILE Some of the authors of this publication are also working on these related projects: Fischer-Tropsch Catalyst Models View project Fischer-Tropsch Synthesis mechanism on Cobalt View project All content following this page was uploaded by Ionel Mugurel Ciobîcă on 29 September 2016. The user has requested enhancement of the downloaded file. This article was originally published in the Comprehensive Inorganic Chemistry II, published by Elsevier, and the attached copy is provided by Elsevier for the author's benefit and for the benefit of the author's institution, for non-commercial research and educational use including without limitation use in instruction at your institution, sending it to specific colleagues who you know, and providing a copy to your institution’s administrator. All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing copies or access, or posting on open internet sites, your personal or institution’s website or repository, are prohibited. For exceptions, permission may be sought for such use through Elsevier's permissions site at: http://www.elsevier.com/locate/permissionusematerial van de Loosdrecht J., Botes F.G., Ciobica I.M., Ferreira A., Gibson P., Moodley D.J., Saib A.M., Visagie J.L., Weststrate C.J. and Niemantsverdriet J.W. Fischer–Tropsch Synthesis: Catalysts and Chemistry. In: Jan Reedijk and Kenneth Poeppelmeier, editors. Comprehensive Inorganic Chemistry II, Vol 7. Oxford: Elsevier; 2013. p. 525-557. Author's personal copy 7.20 Fischer–Tropsch Synthesis: Catalysts and Chemistry J van de Loosdrecht, Sasol Technology Pty (Ltd), Sasolburg, South Africa; Eindhoven University of Technology, Eindhoven, The Netherlands FG Botes, Sasol Technology Pty (Ltd), Sasolburg, South Africa IM Ciobica, Eindhoven University of Technology, Eindhoven, The Netherlands A Ferreira, P Gibson, DJ Moodley, AM Saib, and JL Visagie, Sasol Technology Pty (Ltd), Sasolburg, South Africa CJ Weststrate and JW (Hans) Niemantsverdriet, Eindhoven University of Technology, Eindhoven, The Netherlands ã 2013 Elsevier Ltd. All rights reserved. 7.20.1 7.20.1.1 7.20.1.2 7.20.1.3 7.20.1.4 7.20.2 7.20.2.1 7.20.2.2 7.20.2.3 7.20.2.3.1 7.20.2.3.2 7.20.2.3.3 7.20.2.3.4 7.20.2.4 7.20.2.5 7.20.2.6 7.20.3 7.20.3.1 7.20.3.2 7.20.3.3 7.20.3.3.1 7.20.3.3.2 7.20.3.3.3 7.20.3.3.4 7.20.3.4 7.20.3.5 7.20.4 7.20.4.1 7.20.4.2 7.20.4.2.1 7.20.4.2.2 7.20.4.2.3 7.20.4.3 7.20.4.3.1 7.20.4.3.2 7.20.4.3.3 7.20.4.4 7.20.4.4.1 7.20.4.4.2 7.20.4.4.3 7.20.4.5 7.20.5 References Introduction: Processes, Catalysts, and Recent History ‘Anything’-to-liquids Technology: Syngas Production, FTS, and Product Workup FTS, the Product Distribution Fischer–Tropsch Catalysts and Modes of Operation Historical Development of the FTS Iron-Based FTS Catalysts Introduction Commercial Applications Iron Fischer–Tropsch Catalyst Preparation Fusion Precipitation Improving iron Fischer–Tropsch catalyst precursors by promoters Activation and reduction procedures Selectivity Manipulation of Iron Catalysts Catalyst Stability During FTS Spent Catalyst Management Cobalt-Based FTS Catalysts Introduction Composition of Cobalt Catalysts Preparation of Cobalt Fischer–Tropsch Catalysts Precipitation Preparation methods involving pre-shaped supports Calcination Reduction Cobalt Catalyst Fischer–Tropsch Performance Deactivation and Regeneration of Cobalt Fischer–Tropsch Catalysts Mechanisms and Kinetics of FTS Over Iron and Cobalt Catalysts Introduction Surface Science Studies and Model Reactions Adsorption of CO and hydrogen on model surfaces C–O bond scission Hydrogenation and the stability of C1Hx species DFT Modeling CO Dissociation C þ H reactions Chain growth Macrokinetic Observations and Models General observations regarding kinetics Simple macrokinetic models Selectivity modeling Mechanistic and Kinetic Implications Conclusion Comprehensive Inorganic Chemistry II http://dx.doi.org/10.1016/B978-0-08-097774-4.00729-4 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 526 526 526 527 529 531 531 531 531 532 533 534 534 535 535 536 537 537 537 539 539 540 541 541 541 544 546 546 547 547 548 549 549 549 550 551 551 551 552 552 553 554 554 525 Author's personal copy 526 Fischer–Tropsch Synthesis: Catalysts and Chemistry 7.20.1 Introduction: Processes, Catalysts, and Recent History The Fischer–Tropsch synthesis (FTS) represents technology from the 1920s1,2 that has continuously been revived to provide synthetic hydrocarbon fuels and chemicals from initially coal, later natural gas, and nowadays also biomass. Virtually any source of (hydro)carbon feedstock can be converted to a mixture of synthesis gas, or syngas (CO and H2), which is in fact a key intermediate on which theoretically the entire chemical industry could be based. FTS stands for the reaction(s) of synthesis gas to predominantly straight-chain hydrocarbons, which can be paraffins from CH4 to waxes (CnH2nþ2 with n from 1 to over 100), olefins from ethylene to much longer molecules (CnH2n, with n 2), and to a lesser extent oxygenated products such as alcohols. It produces as main byproducts water and/or carbon dioxide, that is, due to the water-gas shift (WGS) reaction. Being a highly exothermic reaction, it generates large amounts of heat. The process is represented by the simplified reaction equations FTS : WGS : CO þ 2H2 ! CH2 þH2 O CO þ H2 O⇄H2 þ CO2 165 kJ mol1 [1] 42 kJ mol1 gasification. Alternatively, heat can be supplied externally, in which case the gas is reformed only with steam and/or CO2, but no oxygen is added. Examples of this approach include steam reforming (where mainly steam is added), dry reforming (where mainly CO2 is added), and heat exchange reforming (where process heat is supplied to the reformer tubes). Typical reforming catalysts are based on nickel as the active metal.10 The second step in the XTL process is to catalytically convert the syngas to a range of hydrocarbons via the FT synthesis, which mainly yields linear alkanes and 1-alkenes, and which will be the main subject of this chapter hereafter. The third and last step is usually the workup of the hydrocarbons to final products, which are typically fuels, but optionally also chemicals. A popular application at present is to target the production of long chain waxes in the FT synthesis, followed by hydrocracking to middle distillate range components, such as diesel (C9–C22) and jet fuel (C9–C15). Hydrocracking catalysts are bifunctional in nature, with either a noble metal (e.g., Pt) or sulfided base metals (e.g., Ni/W or Co/Mo) as the hydrogenation function on a catalytically active acidic support, such as a silica–alumina. We refer to the literature for further information on this subject.11,12 [2] Reaction [1] represents in essence a polymerization, implying that the product will be a mixture of hydrocarbons with a distribution in molecular weights. Selectivity and control thereof are therefore of key importance in FTS technology. Fischer–Tropsch technology represents a subject of intensive research both in industry and in academia. Many excellent reviews are available.3–9 In this chapter, we first describe the general aspects of the technology in which the FTS features, then the more chemical aspects of the process in relation to the iron and cobalt catalysts that are used in practical applications, and finally mechanistic insight, on the basis of kinetics, surface science, and computational modeling. 7.20.1.1 ‘Anything’-to-liquids Technology: Syngas Production, FTS, and Product Workup The overall process from original carbon source for the syngas to the FTS product is named after the feedstock employed, hence the terminology ‘coal-to-liquids’ (CTL), ‘gas-to-liquids’ (GTL) and ‘biomass-to-liquids’ (BTL), collectively known as XTL (‘anything’-to-liquids). In all instances, the carbon source is first converted to synthesis gas (or ‘syngas’ for short), which is a mixture of CO and H2. Solid feedstocks such as coal or biomass are gasified, usually noncatalytically, by partial oxidation with oxygen (supplying the heat for the endothermic gasification reactions) and reaction with steam (which acts as a gasification agent, hydrogen source, and coolant). When the starting material is natural gas, it can also be adiabatically reformed in the presence of oxygen and steam. There are different embodiments of this approach, such as autothermal reforming (ATR), noncatalytic partial oxidation (POX), and catalytic partial oxidation (CPOX), but in essence the chemistry of all is the same and very similar to that of coal 7.20.1.2 FTS, the Product Distribution At the chemistry level, the FT synthesis is both a CO hydrogenation reaction and a polymerization reaction. The former is reflected by the fact that the C–O bond must be broken and new C–H bonds formed. Additionally, C–C bonds must be formed in order to effect hydrocarbon chain growth. Since the product carbon number distribution approximately follows a statistical function called the Anderson–Schulz–Flory relationship, it is widely accepted that chain growth occurs one carbon atom at a time via a polymerization mechanism. Proposals for the monomer of chain growth, which is produced in situ, have included adsorbed CO, an enol species and a CHx species,7,13–15 and will be discussed further in the section on mechanism and kinetics. The competition between chain growth (yielding a surface intermediate with one higher carbon number) and chain termination (yielding a desorbed final product) is determined by the probability for growth, called the a-value. A higher a-value will result in longer hydrocarbons and thus a heavier product spectrum (Figure 1). If a is independent of carbon number, the scheme presented in Figure 2 applies and the total amount of carbon contained in products with n carbon atoms (namely Cn) can be formulated on a relative basis: C1 ¼ 1ð1 aÞ C2 ¼ 2ð1 aÞa C3 ¼ 3ð1 aÞa2 Cn ¼ nð1 aÞan1 The total amount of carbon in the product spectrum then forms a convergent infinite sum with an analytical solution: 1 X 1 Cn ¼ 1 X nð1 aÞan1 ¼ 1 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 1 1a Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry HTFT 100 Carbon atom selectivity (%) 90 CH4 80 527 LTFT C5+ C20+ waxes 70 C2–C4 60 C5–C11 gasoline 50 40 C9–C22 dieseldistillates 30 20 10 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Chain-growth probability, a 0.8 0.9 1 Figure 1 Hydrocarbon product spectrum that is produced during Fischer–Tropsch synthesis for varying chain growth probability a. High-temperature Fischer–Tropsch technology (HTFT) corresponds approximately to 0.70 < a < 0.75, and low-temperature Fischer–Tropsch (LTFT) to about 0.85 < a < 0.95. Products C1 1-a C* → C2 1-a a C1* → C3 1-a a C 2* → a C3* → ….. Intermediates Figure 2 Carbon chain growth and termination scheme for the derivation of the Anderson–Schulz–Flory equation, with a the chain growth probability factor, Cn (n ¼ 1, 2, 3, . . .) the final products with n carbon atoms, and Cn* the intermediates with n carbon atoms. This means that the selectivity toward products with n carbon atoms on a carbon atom basis, namely Sn, can be expressed as follows: Cn Sn ¼ X1 1 Cn ¼ nð1 aÞ2 an1 After converting this equation to the logarithmic domain and rearranging, it is found that Sn ð1 aÞ2 [3] ¼ n ln a þ ln ln n a As a result, a plot of ln(Sn/n) versus carbon number (n) gives rise to a straight line with a slope equal to ln(a). However, deviations in the actual FT product spectrum from the ideal Anderson–Schulz–Flory distribution are usually observed.4,14,16,17 These include a higher methane and a lower C2 selectivity than predicted by the equation. There is also an increase in the chain growth probability factor and concomitant decrease in the olefin/paraffin ratio with hydrocarbon chain length. In addition to linear alkanes and 1-alkenes, a variety of other products are also formed, including branched aliphatic compounds, alcohols, aldehydes, ketones, acids, and (at sufficiently high operating temperatures) even aromatics. This alludes to the complexity of the reaction and many unresolved issues remain regarding the reaction mechanism. A further complicating factor is that it is not always clear which of these compounds are primary FT products and which are formed subsequently by secondary reactions. For example, olefins and alcohols can undergo a range of secondary reactions, such as hydrogenation, double bond isomerization, skeletal isomerization, and conversion to heavier compounds.4,16 7.20.1.3 Fischer–Tropsch Catalysts and Modes of Operation Metals known to catalyze the FT reaction mainly include iron, cobalt, ruthenium, and nickel.6 Ruthenium is a scarce and expensive metal, whereas nickel only forms methane at reaction temperatures sufficiently high to suppress nickel carbonyl formation (note that methanation is the reverse reaction of methane reforming, for which nickel-based catalysts are commonly used10). As a result, only iron- and cobalt-based FT catalysts have found commercial application.18–20 Iron generally produces more olefins and oxygenates than cobalt (i.e., a less hydrogenated product spectrum), which may be related to the lower hydrogenating ability of iron. While cobalt is active in the metallic state,19 iron catalysts change under Fischer– Tropsch conditions to a complex mixture of iron carbides and oxides.20,21 Byproducts of the FTS originate from the way oxygen from CO is removed. With cobalt catalysts, essentially all oxygen from CO dissociation (typically around 99%) is discarded as water. Iron catalysts differ in this respect, as a significant portion of the oxygen is also discarded as CO2. The latter is often visualized as a separate, consecutive reaction, namely the WGS. Stoichiometrically, the overall process can be represented by reactions [1] and [2], which we repeat here: FTS : WGS : CO þ 2H2 ! CH2 þH2 O CO þ H2 O⇄H2 þ CO2 165 kJ mol1 [1] 42 kJ mol1 [2] The net rate of hydrogen conversion divided by CO conversion (sometimes referred to as the ‘usage ratio’) is extremely important for the gas loop design around an FT reactor. In the one extreme, where virtually no WGS takes place, the usage Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 528 Fischer–Tropsch Synthesis: Catalysts and Chemistry catalysts, yielding a much lower overall usage ratio that is better suited to syngas feeds with a low H2/CO ratio, such as that generally obtained from coal gasifiers. Table 1 presents the current commercial application of the FTS. There are two important aspects to note from this table. First, the worldwide FTS capacity is expected to reach a total of just over 400 000 barrels per day by 2013 (1 barrel ¼ 159 l), which is very small compared to the total crude oil production of around 80–85 million barrels per day. Second, the FTS has been applied in a variety of forms, which determines the type of reactor employed. As indicated in Figure 3, these reactors ratio is only determined by the FT reaction (with a high selectivity to long hydrocarbons and a low selectivity to methane) and assumes a value of around 2. In the other extreme, where almost all water is shifted to CO2, the usage ratio can approach a value of 0.5. The low propensity of cobalt catalysts for the WGS makes them the preferred catalysts for GTL application, since the H2/CO ratio of syngas derived from natural gas is already close to or above the usage ratio. Any additional WGS will result in an excess of hydrogen that would not be fully consumed by the FT reaction, even if CO is converted to extinction. Conversely, the WGS is more facile over iron Table 1 Fischer–Tropsch synthesis, current commercial plants and plants under construction Company Location Carbon feedstock Catalyst type Reactor type Start-up date Sasol Sasolburg, South Africa Initially coal, currently natural gas Fused Fe/K HTFT circulating fluidized bed LTFT multitubular fixed bed LTFT slurry phase 1955 to 1985 1955 Precipitated Fe/K Mostly coal, now supplemented by natural gas Natural gas Precipitated Fe/K (spray dried) Fused Fe/K Sasol Secunda, South Africa Shell Bintulu, Malaysia PetroSA Mosselbay, South Africa Natural gas Co/SiO2 Co/TiO2 Fused Fe/K Sasol-QP (Oryx) Shell (Pearl) Ras Laffan, Qatar Ras Laffan, Qatar Natural gas Natural gas Co/Al2O3 Co/TiO2 Chevron-Sasol Escravos, Nigeria Natural gas Co/Al2O3 Approximate plant capacity (barrels per day) 5000 1993 HTFT circulating fluidized bed HTFT SAS reactora LTFT multitubular fixed bed HTFT circulating fluidized bed (Sasol technology) LTFT slurry phase LTFT multitubular fixed bed LTFT slurry phase 1980–1999 160 000 1995 1992 14 500 1993 22 000 2007 2011 34 000 140 000 2013 34 000 a SAS: Sasol Advanced Synthol, fixed fluidized bed. Stationary bed reactors Moving bed 200–5000 μm particles 1–200 μm particles Multitubular fixed bed 6000 barrels per day LTFT, 200–250 ∞C • 3-phase system: gas–liquid–solid • a = 0.85–0.95 • Products: wax, diesel, naphta • Catalysts: supported cobalt or precipitated iron Slurry bubble column, filled with wax 24 000 barrels per day Microchannel reactor ~ 200–1000 b/d (for assembly) Microchannel process technology module Boiling heat transfer HTFT, 320–350 ∞C High heat flux • 2-phase:gas–solid • a = 0.70–0.75 • Products: petrol and chemicals • Catalysts: fused iron, K-promoted Circulating fluid bed 7000 barrels per day FT 10 times higher heat flux than conventional reactors Fixed fluid bed 20 000 barrels per day Figure 3 Overview of Fischer–Tropsch technology with reactors (figure microchannel reactor: courtesy of the Oxford Catalysts Group). Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry can broadly be classified in two classes, namely two-phase or three-phase reactors, and moving or stationary catalyst bed reactors.22–24 The high-temperature Fischer–Tropsch (HTFT) synthesis process is characterized by operating temperatures of about 320–350 C and the products are essentially only in the gas phase under reaction conditions, giving rise to a gas–solid system without any bulk liquid phase. Originally, this process was operated in circulating fluidized bed reactors and more recently in fixed fluidized bed reactors. Cobalt catalysts would essentially only produce methane at these temperatures, making alkali-promoted iron catalysts the only option for this application. Due to the mechanical demand that these moving bed reactors place on the catalyst, particle strength is an important consideration; consequently, only fused bulk iron catalysts have been employed commercially. The light product spectrum is best suited to the production of gasoline, but the high selectivity toward linear 1-olefins and (to a lesser extent) oxygenates allows for the extraction of chemicals from the product slate. These include monomers such as ethylene and propylene, co-monomers such as 1-hexene and 1-octene, and solvents (e.g., propanol, butanol, methyl ethyl ketone (MEK), and acetaldehyde).25 The low-temperature Fischer–Tropsch (LTFT) synthesis is operated between 200 and 250 C.26,27 Both cobalt and iron catalysts are suitable for this application, although cobalt catalysts would typically be used toward the lower half of the quoted temperature range. The heavy product spectrum extends well into the domain of waxes, which are liquid under reaction conditions. The presence of a bulk liquid phase gives rise to a three-phase gas–liquid–solid system. Originally, only fixed-bed reactors operating in a trickle bed mode were employed for this synthesis. In order to limit the pressure drop over the stationary catalyst bed, catalyst particle sizes must be in the millimeter range, which brings about significant intra-particle diffusion limitations. This not only limits catalyst utilization, but also adversely affects product selectivities due to the differences in diffusion rates between hydrogen and CO that causes higher H2/CO ratios toward the center of the particles. The highly exothermic nature of the FT reaction causes axial and radial temperature profiles in the catalyst bed. More recently, slurry bubble-column reactors have been developed to overcome some of these drawbacks. Syngas is bubbled through a suspension of fine catalyst particles in the liquid product phase. The catalyst particle sizes are usually less than about 100 mm, which is sufficiently small to prevent intraparticle diffusion limitations, while the well-mixed liquid phase ensures virtual isothermal operation of the reactor. There are, however, certain technical challenges associated with FTS slurry reactors. A prerequisite of a slurry process is the development of an efficient solid–liquid separation step to remove product wax from the reactor. It is extremely important to ensure the mechanical integrity of the catalyst to limit the extent of breakup and attrition in the moving bed environment. Of late there have been some new reactor developments for FTS application, but none of these have been commercially applied yet. Microchannel reactors can support very high heat and mass transfer rates and thereby address the problems of traditional fixed-bed reactors, while the stationary bed circumvents the challenges of slurry reactors. This approach shows 529 promise, especially with respect to the small-scale application of a few hundred or a few thousand barrels per day production capacity, and some relatively new commercial companies are actively pursuing this technology.28 Structured reactors (monolith type reactors) for FT application have also attracted the attention of mainly academia, although there has been some limited interest from commercial companies as well.29 In this approach, the active FT metal (e.g., cobalt) is coated onto a large structure with a specific geometry, which is then inserted into a reactor tube. The LTFT synthesis is ideally suited for the production of high-quality middle distillates (diesel and jet fuel) after hydrocracking of the long chain waxes. In addition, the heavy product spectrum provides chemical opportunities in the form of speciality waxes and base oils. The naphtha from the process is also a high-quality feedstock for naphtha steam crackers that produce mainly ethylene, but also some propylene. 7.20.1.4 Historical Development of the FTS Historically, the first syngas conversion results were published by Sabatier30 in 1902 where it was shown that a mixture of carbon monoxide and hydrogen could be converted into methane over nickel and cobalt catalysts. In the 1920s, Franz Fischer and Hans Tropsch took this process a step further and showed that syngas could be converted into a mixture of higher hydrocarbons that could be used as petrol or diesel (i.e., FTS).1,2 In their first patent,31 they described the production of higher hydrocarbons using iron- and cobalt-based catalysts operated at atmospheric pressure and at temperatures below 300 C. Further research in Germany led to improved versions of this process. The first commercial plant started in 1936. Several others followed and provided Germany and Japan with synthetic fuel during the Second World War. These plants used mainly cobalt catalysts supported on kieselguhr (i.e., silica-based supports) and promoted by magnesia and thoria, in fixed-bed reactors. Further, China had FTS plants in the 1940 through 1960s, all based on cobalt catalysts.32 After the war, the German FT technology came in the hands of the Allied Forces. Many scientists and engineers who contributed to the German developments were interrogated and the entire Fischer–Tropsch technology was extensively investigated at the US Bureau of Mines, which resulted in new twophase HTFT technology. The classical textbook by Storch, Golumbic, and Anderson originates from this period.3 Small plants were built in the US and operated in the 1950s. Large-scale FTS developments mainly occurred in South Africa.33 Sasol started an FTS plant in 1955 based on HTFT to make petrol and on LTFT to produce wax. Both HTFT and LTFT used iron-based catalysts. The HTFT technology formed the basis for the large expansion of Sasol in the late 1970s/early 1980s when Sasol 2 and 3 were built in Secunda (see Table 1). The main reasons for this expansion were the oil crises in the 1970s, which led to a significant increase in the crude oil price (see Figure 4). These oil crises also initiated renewed interest in FTS from other companies like BP, ExxonMobil, Gulf, Shell, and Statoil, which was mainly based on cobalt FTS catalysts.34,35 In the last 20 years, this has led to new commercial GTL plants by PetroSA (South Africa; 1993), Shell (Malaysia; 1992), Sasol-Qatar Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 100 Number of articles/patents 450 90 400 Patents 350 Articles 300 US crude oil price 80 70 60 250 50 200 40 150 Articles Oil price 100 20 Patents 50 0 1970 30 US crude oil price ($) 530 1975 1980 10 1985 1990 1995 Publication year 2000 2005 0 2010 Figure 4 Patents and articles per year compared with the crude oil price (figure inspired by de Smit and Weckhuysen21). FTS ATR ASU Figure 5 Photo of the Sasol-QP Oryx GTL plant in Qatar, showing the air separation units (ASUs), the auto-thermal reformers (ATRs), and the Fischer–Tropsch synthesis (FTS) slurry reactors. The product work-up section located behind the FTS reactors is not visible (photo courtesy of Sasol). Petroleum (Qatar; 2007; see Figure 5), Shell (Qatar; 2011), and Sasol Chevron (Nigeria; under construction – start-up 2013). An overview of the current commercial operations using FTS technology is shown in Table 1. The investment decision to build the Sasol-Qatar Petroleum Oryx-GTL plant was taken in 2003 when the oil price was $25/ barrel. The facility was built at a cost of $1 billion. Currently, a yearly profit is generated of about $500 million.36 Sasol’s much larger Secunda CTL facility is generating currently about $2 billion profit annually.36 Shell’s Pearl plant (both the FTS production – 140 000 barrels per day – and the upstream natural gas condensates – 120 000 barrels per day – together) was built at a cost of $20 billion,37 and Shell announced to make annually $4 billion cash when Pearl is at full production with the oil price at $70/barrel. It is clear that new GTL/CTL facilities require large capital investments, and are heavily dependent on the prevailing crude oil price. However, over the long term these large-scale GTL/CTL facilities do make economic sense. A new challenge to the GTL/CTL technology is global warming and the emission of CO2. Fuel products from GTL facilities have a similar environmental footprint compared to crude oil-derived fuels. However, products from CTL facilities have a much larger CO2 impact, which is immediately clear from the overall stoichiometric equations [1] and [2]. In the hypothetical limit of using a carbon feedstock which does not contain any hydrogen, one CO2 molecule is formed for every carbon atom that ends up in a hydrocarbon. A large portion of the CO2 produced in CTL plants is removed and concentrated, and is therefore ideally suited for capturing, that is, ‘capture ready’. At the same time new focus on products from biomass can stimulate interest in FTS further, as biomass can be used as a carbon source for syngas generation. Recently, Oxford Catalysts has demonstrated their FTS technology using syngas made from wood chips.28 Other opportunities for GTL applications in the future might be the use of associated natural gas in small-scale plants (<1000 barrels per day, as Oxford Catalysts and CompactGTL are pursuing), as well as the use of shale gas in large-scale facilities (as pursued by Sasol). From an academic point of view, renewed interest in FTS was clearly observed in two main waves (see Figure 4). The first wave occurred in the late 1970s, while the second one started around 1995 and is still gaining momentum. The latter Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry observation is also matched by an increase in patenting activity. Of course, this revived activity is related to the increase in the crude oil price, although the present academic interest is certainly also inspired by the notion that crude oil resources are limited. Typical topics in Fischer–Tropsch research with a high academic interest are catalyst preparation methods, deactivation studies, and mechanistic and kinetic studies, in which sophisticated tools such as in situ catalyst characterization, surface science, molecular modeling, and transient kinetic studies are the common ingredients. The increased interest from both the academic as well as the commercial world has created excellent scientific interactions and discussions, which enabled further progress on this exciting topic of FTS. Many questions are still outstanding, such as the state of the catalytically active surface under reaction conditions, and the reaction mechanism in terms of elementary steps. 7.20.2 7.20.2.1 Iron-Based FTS Catalysts Introduction The iron-catalyzed FTS process is, along with ammonia synthesis, one of the most studied systems in the field of heterogeneous catalysis. The reason for this is possibly the fact that the application of the process is so versatile. Not only can iron FTS produce a light hydrocarbon product stream ideal for the fuel and chemical industry, it can also produce heavier hydrocarbons (C35þ) suited for the waxes market. Iron is also a cheap raw material when compared to its cobalt counterpart (cobalt is on average 250 times more expensive than iron raw materials) and it has been commercially applied since the late 1950s by Sasol38 (Table 1). Iron is believed to be more tolerant of poisons, for example, sulfur in synthesis gas than cobalt. It is also known to be responsive to selectivity manipulation by the addition of promoters and a variation of typical process parameters, for example, temperature, pressure, and H2/CO ratio. The disadvantage, however, is the fact that iron FTS catalysts deactivate rather quickly (activity or selectivity loss) and this will be discussed in more detail later in this section. As already mentioned the iron FTS process can be manipulated to produce a range of carbon number distributions with the final product stream depending mainly on the temperature applied during FTS. At lower temperatures, for example, 220–250 C the chain growth probability (a) of the catalyst is approximately 0.94 indicating that the bulk of the products will consist of hydrocarbons longer than C21. In the case of higher temperatures, for example, 320–350 C, the chain growth probability decreases to 0.7 and even lower with the main products being light hydrocarbons utilized for the production of transportation fuel and chemical feedstocks. Figure 1 shows the carbon chain length as a function of chain growth probability. The influence of promoters on selectivity will be discussed later in this section. Although there are many advantages with regard to ironcatalyzed FTS, the transformations of the iron catalyst during activation and FTS are rather complex and still not fully understood. During catalyst preparation, iron oxides (e.g., hematite (Fe2O3) and magnetite (Fe3O4)) are produced and these are transformed to either a-Fe or iron carbides during activation depending on the conditions. 7.20.2.2 531 Commercial Applications Sasol is a leader in the field when it comes to commercializing iron-catalyzed FTS processes. In the early 1950s, Sasol commercialized Fe-catalyzed FTS based on the Ruhrchemie process to produce a variety of synthetic petroleum products using the Arge Tubular Fixed bed reactors (see Table 1 and Figure 3). Later on, they developed the slurry-bed reactor and this reactor together with the Arge reactors are used to produce highmolecular-weight hydrocarbons for the wax industry.38 In the late 1950s, Sasol also commercialized a circulating fluidized bed reactor at their Sasolburg facilities in which fused iron catalyst is fluidized at high temperatures to produce lighter hydrocarbons ideally suited for producing fuel and chemical feedstock. In the late 1970s/early 1980s Sasol’s Secunda plant was built using this circulating fluidized bed technology, which was replaced in the late 1990s by the improved fixed fluidized bed technology. Subsequent to Sasol’s successful commercialization of iron-catalyzed FTS, South Africa’s national oil company (PetroSA) commercialized a GTL facility using previous-generation high-temperature FTS technology (Sasol licensed technology). This technology is based on a fused iron catalyst operated in a fluidized bed reactor at high temperature (330–350 C). In 2010, it was still recognized as one of the world’s largest GTL refineries, producing about 22 000 barrels per day of high-quality FTS-derived fuels. Rentech, based in Colorado, USA has long been investing in iron-based FTS research. Rentech demonstrated their ironbased FTS technology in their Product Demonstration Unit (PDU) in the middle of 2008. The PDU produces approximately ten barrels per day of ultra-clean diesel, aviation fuels, and naphtha.39 Synfuels China has recently emerged as an important player in the Fischer–Tropsch industry.32 Their development of a socalled high-temperature slurry-phase technology (HTSFTP™) and associated iron-based catalyst is novel to the industry. The integrated technology promises improvements in process thermal efficiency and a highly active catalyst. This technology has been demonstrated in a 4000 barrels per day semi-commercial CTL facility owned by the YiTai Coal Liquefaction Company in Xue JiaWan, Erdos, Inner Mongolia. 7.20.2.3 Iron Fischer–Tropsch Catalyst Preparation There are several preparation methods available in the literature for the synthesis of Fe FTS catalysts, like precipitation and fusion. Iron catalysts prepared commercially are actually iron oxides, hydroxides, or oxy-hydroxides, which undergo an activation step such as reduction or pre-treatment in syngas prior to FTS. General requirements for the catalysts are, among others, selectivity (low for methane; high for the targeted hydrocarbon fraction), activity and stability, and mechanical robustness. The operation conditions, for example, high or low temperature (HTFT and LTFT), and the type of reactor employed put specific demands on the catalyst synthesis procedure. For a tubular fixed-bed-type reactor, minimization of mass transfer limitations is an important consideration, and here catalyst strength is less important than catalyst shape and form. For a fluidized bed reactor, however, catalyst strength as well as particle size and density are very important.40 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 532 Fischer–Tropsch Synthesis: Catalysts and Chemistry Table 2 summarizes the typical preparation methods used for the various applications. Key factors when choosing an iron source are cost and availability. Although most of the iron oxides, hydroxides, and oxy-hydroxides are readily available in nature, precursors for iron catalysts are rather chemical grade raw materials.41 This is done to ensure that impurities that can influence the catalyst are either removed or carefully controlled. It is typical for commercial manufacturers of iron catalysts to produce chemical grade iron(III)nitrate from sufficiently pure scrap iron on site as part of the preparation. Large-scale fusion preparation methods use iron ores or mill scale from steel mills. Complex preparation methodologies that involve novel chemicals and/or intricate transformations are rarely commercially viable when compared to the tried and trusted methods of precipitation and fusion. The gains from such novel preparations must be truly unique to justify the additional expense. 7.20.2.3.1 Fusion Fusion produces oxidic iron particles of low surface area, high density, and high strength, which are ideally suited for application in circulating fluidized bed reactors (Figure 3). During the fusion process, the iron oxide raw material together with the promoters are fed into an arc furnace where it is subjected to temperatures above 1000 C. After fusion, the molten material is cast into flat bars (ingots) and cooled. These ingots are milled to a specified particle-size range to ensure optimum fluidization (Figure 6). The disadvantage of fusion is the fact that the inorganic impurities, for example, silica and alumina oxides present in the raw mill scale starting material, form inclusions during cooling of the ingot. The alkali promoters migrate and bind during cooling to these inclusions, which negates the promotion effect. Figure 7(a) is a scanning electron microscopy (SEM) image of a fused ingot showing clearly the inclusions and with scanning electron microscopy energy-dispersive X-ray Catalyst preparation methods used for high and low temperature iron-based Fischer–Tropsch processes Table 2 Reactor HTFT Circulating or fixed fluidized bed reactors, 320–350 C LTFT Tubular fixed bed reactor, 220–250 C Slurry bed reactors, 220–250 C Important catalyst properties Raw material Synthesis method Low surface area (<10 g m2), high density, high strength Mill scale Fusion followed by crushing and milling High surface area, sufficient strength Fe(NO3)3 and silica source Fe(NO3)3 and silica source Precipitation followed by extrusion/shaping Precipitation followed by spray drying and calcination High surface area, small particles (50–250 mm) Promoters Mill scale Oxidation Cooling step Arc furnace Size reduction (milling) Reduction Figure 6 Catalyst preparation diagram for the HTFT fused iron catalyst. Mill scale means iron metal pieces from the steel industry. Inclusion Fe Si Alkali 1 Alkali 2 80 mm (b) (a) Figure 7 (a) SEM image of a HTFT Fe catalyst cast ingot, showing inclusions of silica and alkali and (b) SEM EDX mapping on one of these inclusions. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry spectroscopy (SEM EDX)(Figure 7(b)) one can see the presence of the alkali in these inclusions. The development of fixed fluidized bed reactors diminished the need for particles of high mechanical strength, and permitted the use of catalysts with lower density and higher surface area. One of the prospective routes to alternative HT FTS catalyst precursors is the precipitation of high-density, lowsurface-area iron oxides, hydroxides, and (oxy)hydroxides from solutions of iron(III) salts followed by calcination. 7.20.2.3.2 Precipitation Precipitation of iron(III)oxides from iron(III)nitrate solutions was one of the first methods reported in the literature for the preparation of iron FTS catalysts.42 In the late 1930s, Ruhrchemie developed a large-scale preparation based on precipitation. In this procedure, the iron(III) salt is reacted with a base to form an iron(III) oxide–(oxy)hydroxide precipitate. By variation in process conditions, for example, pH, precipitation rate, and temperature, catalyst properties such as surface area and crystallite size can be controlled. Figure 8 illustrates the decrease of crystallinity versus surface area for the various iron oxides known as precursors for Fe FTS catalysts. After precipitation, the slurry is filtered and washed to remove all the salts (e.g., NH4NO3) from the filter cake (see Figure 9). The latter is then reslurried and impregnated with structural promoters like Si, Al, etc. The application of chemical promoters in the iron catalyst is discussed in more detail later in this section. Next, the slurry is spray-dried to yield spherical particles that are suited to slurry-bed and fixed fluidized-bed reactors. The final step in the catalyst preparation is calcination. Dissolution Precipitation HNO , (aq) Base (aq) 533 This high-temperature treatment removes volatile impurities such as water and NOx and increases the strength of the catalyst particles. Figure 10 shows spherical particles obtained from spray drying. For fixed-bed reactor applications, the impregnated filter cake is extruded and dried at about 150 C. The methodology described above is commercially applied by Sasol for the synthesis of their slurry-bed reactor (SBR) and tubular fixed-bed reactor catalyst precursors. Rentech uses a similar method to synthesize their Fe FTS catalyst precursor.39 As mentioned above, Sasol developed ‘in-house’ precipitation methodology to synthesize a suitable Fe-HTFTS catalyst that offers many advantages over the fused Fe HTFT catalyst.43 Some of these include improved promoter distribution, increased strength, and spherical particles which improve fluidization of the catalyst. By replacing fusion with precipitation, the negating effect of the alkali promoters could be eliminated owing to the purity of the starting iron(III)salt. Hematite: 12 – 27 m2 g-1 Magnetite: 4 – 100 m2 g-1 Goethite: 8 – 200 m2 g-1 Degree of crystallinity Ferrihydrite: 100 – 700 m2 g-1 Figure 8 Degree of crystallinity is decreasing with an increased surface area for various iron oxides. Filtration/ washing Water Iron Re-slurry and impregnation Salt (aq) OR Spray drying Extrusion Calcination Drying Slurry bed catalyst Fixed bed catalyst Figure 9 Catalyst preparation diagram for the slurry-bed and fixed-bed precipitated iron LTFT catalyst. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 534 Fischer–Tropsch Synthesis: Catalysts and Chemistry The stability and composition of the final activated iron phase determine the performance of the catalyst under Fischer– Tropsch conditions. Because of this, the activation procedure has a strong effect on selectivity and activity of the catalyst. For activation in hydrogen, the extent of reduction of iron is governed by the role of water removed during the activation. The degree of reduction is governed by the equation47: DG ¼ DG þ RT ln Figure 10 Spray-dried iron LTFT slurry-bed catalyst particles. 7.20.2.3.3 Improving iron Fischer–Tropsch catalyst precursors by promoters FTS processes catalyzed by unmodified and unpromoted iron catalysts suffer from poor selectivity, low activity, and sintering, but the addition of structural and chemical promoters addresses most of these issues. Structural promoters, for example, Si, Al, and Mg, may suppress sintering, stabilize the active phase, and improve mechanical strength. The addition of alumina and silica typically increases the stability of hematite under FTS conditions.44 In general, it is observed that in the presence of a structural promoter such as silica the surface area of the iron oxide remains high even after calcination at relatively high temperatures. A potential disadvantage, however, of adding structural promoters is that the activation, for example, reduction of the iron oxide, becomes more difficult due to the formation of iron silicates or aluminates. For this reason, chemical promoters such as Cu or Ag are added during catalyst synthesis to increase the rate of reduction, most likely due to hydrogen spillover from the Cu surface to the iron oxide surface. Apart from increasing the rate of reduction, chemical promoters are known to (i) enhance nucleation of iron intermediates which leads to higher surface areas, (ii) increase the number/type of CO adsorption sites, (iii) stabilize selected phases, and (iv) influence the rate of secondary reactions.45 Addition of alkali metals (e.g., potassium) to LTFT iron oxide catalyst precursors is known to enhance the chain growth probability (increased C5þ selectivity), to diminish methane formation, and inhibit secondary hydrogenation reactions, leading to higher olefin to paraffin ratios.46 In a similar way, but perhaps not as effective, alkali earth metals have been shown to increase alpha values and to suppress methane formation. 7.20.2.3.4 Activation and reduction procedures Iron oxides and (oxy)hydroxides are inactive for FTS and must be activated to render an active catalyst. Depending on the application – HTFT or LTFT – and iron oxide precursor, activation is performed in hydrogen, carbon monoxide, or synthesis gas. The optimum activation conditions are influenced by the type and quantity of chemical and structural promoters. pH2 O pH2 [4] where ΔG is the free energy change for the reduction under the conditions employed, ΔG is the standard free energy change for the reduction reaction, R is the gas constant, T is the temperature, and p is the partial pressure of the gases indicated. The equation implies that the rate at which water is removed from the reactor plays a critical role: the faster the water is removed, the faster the reduction process proceeds, and the higher the degree of reduction is. Rewriting expression [4] to include the equilibrium constant in the form of the ratio (pH2O/pH2)eq at equilibrium, enables one to estimate the degree of reduction that can be obtained: " # pH2 O pH2 O DG ¼ nRT ln pH2 pH2 eq [5] As the equilibrium ratio for reduction from Fe2O3 to FeO is 0.7, and for FeO to a-Fe is 0.1, the theoretical degree of reduction would be 50% at 10% water in the gas phase. Hence, for high degrees of reduction the water content should be well below 1%.47,48 Structural promoters such as silica and alumina increase the resistance against reduction. To fully understand activation, it is necessary to understand the possible phase transformations that can occur. x-Ray diffraction (XRD) and Mössbauer spectroscopy are ideal techniques to distinguish between the different iron phases that can arise, while small particle effects, which so often limit the information content of XRD, are generally absent in the unsupported iron FT catalysts.49–51 Activation with CO present in the gas typically leads to a mixture of metallic iron (a-Fe), iron carbides (general formula FexCy), and magnetite (Fe3O4). The relative quantities are a function of the reducing gas, the gas hourly space velocity, and the temperature. It is generally believed that carbides such as Hägg-carbide (w-Fe5C2) are the active phase for FTS.52 The exact nature of the surface carbidic species is still a subject of debate. The stabilities of different bulk iron carbides have been reported to be in decreasing order: e0 -Fe2.2C> e-Fe2C > w-Fe5C2 > y-Fe3C.53 Depending on the type of catalyst (e.g., fused or precipitated), different iron carbides were found to be characteristic for each type after activation. However, the possibility that a-Fe plays a role during FTS cannot be ignored.21 Typically during activation, precipitated catalyst precursors, for example, hematite (Fe2O3), are converted to magnetite (Fe3O4) irrespective of the activation gas used. However, after this transformation the final iron phase will depend on the activation gas used, for example, a-Fe in the case of hydrogen or Hägg-carbide (w-Fe5C2) in the case of CO or synthesis gas. In a study by Herranz et al., it was found that the activation of hematite using CO resulted in mainly cementite (y-Fe3C) while activation in synthesis gas yielded Hägg carbide (w-Fe5C2) (Table 3).54 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry The fused magnetite HTFT catalyst is reduced with hydrogen at temperatures between 350 and 450 C and high linear flow rates to avoid re-oxidation by water, as explained above. During reduction, oxygen atoms are removed from the lattice leading to an increase in surface area from <1 g m2 to 5–8 g m2 55. The extent of reduction under these conditions was measured at about 80% (a-Fe). Under CO or synthesis gas, virtually no reduction of the nonporous magnetite was observed. In the case of precipitated iron oxide catalyst precursors for LTFT catalysts, the activation is usually done under much milder conditions than for fused catalysts. These catalysts are more amorphous with a high pore volume and surface area and the oxide crystallites can sinter under too harsh activation conditions. It is important to note that the success of activation of precipitated iron catalyst precursors is coupled to FTS activity, stability, and selectivity. This is dependent not only on the type of reduction gas but also on process conditions, for example, temperature and pressure. From the literature, it seems as though activation under CO yields the optimally activated precipitated iron catalyst for FTS synthesis, as these gave the best syngas conversion and lowest methane selectivity when compared to catalysts activated with H2 or synthesis gas.56 However, the final catalyst also had a relatively high WGS activity. 7.20.2.4 Selectivity Manipulation of Iron Catalysts A key advantage of iron-catalyzed FTS is the fact that the selectivity of the process can be manipulated, either by process conditions (less responsive) or by catalyst composition (more responsive). Lowering the temperature shifts the selectivity Table 3 Names of the various iron phases a-Fe2O3 a-FeOOH Fe3O4 FeO w-Fe5C2 y-Fe3C Hematite Goethite Magnetite Wustite Hägg carbide Cementite 535 from lighter to heavier hydrocarbons. Although the Anderson– Schultz–Flory (ASF) distribution curve (see Section 7.20.1 and Figure 2) gives a good indication of expected selectivities, alkali promotion of iron catalysts leads to selectivities that tend to deviate from ASF and are characterized by two alpha values (see Figure 11). In attempts to increase the selectivity toward valuable base chemicals, mixed-metal oxides and/or multicomponent metals are typically incorporated in precipitated iron catalysts. Addition of manganese to a typical Ruhrchemie catalyst increases the selectivity toward alpha-olefins at low-temperature FTS conditions (230 C and 20 bar total pressure).57 Another example is the Fe/Zn/Mn/Cu/K/SiO2 catalyst for direct conversion of synthesis gas to chemicals.58 The development of such a technology is known as ChemFT, and has as primary focus to shift the selectivity toward alcohols. Table 4 compares the selectivities of the various iron FTS technologies (HTFT, LTFT, and ChemFT) and illustrates that the alcohol selectivity in ChemFT is much higher than that of a typical Ruhrchemie catalyst under similar LTFT conditions.58 7.20.2.5 Catalyst Stability During FTS Stability is a key characteristic of a successful catalyst. The ideal FTS catalyst should maintain constant activity and a corresponding stable selectivity during time on stream. Commercial reactors and product workup sections are designed for a very narrow set of optimum process conditions. The catalyst must perform within these design constraints for as long as possible. This determines the useful catalyst life. Unfortunately, iron catalysts show considerable loss of performance over time. During recent years, the focus of research has shifted from improving catalyst activity to increasing the lifetime of the catalyst. Deactivation of iron FTS catalysts is usually attributed to the following factors: (i) ‘free’ carbon formation, leading to catalyst fouling, (ii) activity loss due to transformation of the phase, for example, oxidation, LT FT Ln (X ) HT FT Alpha 1 (LT FT) = 0.80 Alpha 2 (LT FT) = 0.94 Alpha (HT FT) = 0.75 0 5 10 15 20 25 30 Carbon number 35 40 45 50 Figure 11 Anderson–Schulz–Flory distribution of hydrocarbons formed over an LTFT catalyst (typical Ruhrchemie catalyst) and a fused HTFT catalyst. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 536 Fischer–Tropsch Synthesis: Catalysts and Chemistry (iii) mechanical break-up of the catalyst, (iv) deposition of poisons in the synthesis gas on the catalyst’s surface, and (v) sintering. Buildup of ‘free’ carbon is one of the major causes of deactivation in HTFT. It leads to a decrease in density and strength of catalyst particles and results in catalyst bed expansion, particle break-up, and carry-over of fine catalyst material into downstream processes.25 Figure 12 shows an SEM image, along with the distribution of elements of a ‘spent’ catalyst retrieved from a commercial fixed fluidized bed reactor. Catalyst break-up and fines formation are easily recognized. As carbon is dispersed through the bulk of the particle, break-up may lead to exposure of new active surfaces, thus helping to maintain activity. At the same time, the carbon that is lost in the form of fines is rich in alkali and thus removes some of the chemical promoter, which degrades the selectivity. Much has been discussed regarding the origin of the free carbon in the catalyst. A plausible explanation is given by the Table 4 Selectivity comparison between LTFT, HTFT, and ChemFT25,58 Product Fe HTFT Fe LTFT Fe ChemFT CH4 (%) C2–C4 olefins (%) C2–C4 paraffin (%) C5–C6 (%) C7–350 C (middle distillate product) 350 C (wax products) Oxygenates as alcohols (%) Oxygenates as acids þ ketones % breakdown (C5–C12 cut) % total paraffins % total olefins % aromatics % oxygenates 8.0 24.0 6.0 16.0 36.0 3.0 4.0 4.5 7.0 26.5 18.0 21.0 17.7 13.3 20.5 5.0 2.8 2.2 51.0 3.8 0.2 0.0 8.3 0.9 13.0 70.0 5.0 12.0 29.0 64.0 0.0 7.0 49.6 37.8 0.0 12.5 so-called competition model.59 After the adsorption and dissociation of CO and H2, three reactions are possible: (i) C* þ iron ! carbides (ii) C* þ xH* ! CHx* (iii) C* þ yC* ! inactive carbon The first reaction describes the formation of iron carbides from the reduced a-Fe under FTS conditions. The dissociated carbon (C*) can either react with dissociated hydrogen atoms (H*) to yield hydrocarbons or react with another carbon atom (C*) to yield inactive/ so-called ‘free’ carbon.59 This type of deactivation can be suppressed by chemical promoters. In recent years, Sasol developed another propriety catalyst involving the addition of chromium to reduce the amount of ‘free’ carbon formed during FTS.60 The formation of ‘free’ carbon is less pronounced in the case of the precipitated LTFT iron catalysts. The main deactivation mechanisms in this case are sintering and oxidation of the active phase. Interconversion of different carbides may lead to a stoichiometric excess of carbon which in turn leads to weakening of catalyst particles. Figure 13 shows a deactivation curve for a typical Rührchemie catalyst under low-temperature FTS conditions. Samples of this catalyst taken from the reactor usually contain mixtures of highly dispersed magnetite and iron carbide (both containing around 2 nm particles).61 The highly dispersed magnetite particles can either react in synthesis gas to the required iron carbide, or they can agglomerate or sinter into larger inactive particles (about 40 nm). The larger magnetite particles can agglomerate further to yield large globules (around 400 nm). Surprisingly, agglomeration or sintering of highly dispersed iron carbide into less active or inactive iron carbide particles of about 20 nm has also been observed (Figure 14). 7.20.2.6 Spent Catalyst Management Regeneration of ‘spent’ iron FTS catalysts is difficult, due to the sintering of the particles during FTS. Successful regeneration requires redispersion of the sintered phase, and this cannot easily be achieved. Reactivation by re-reduction is possible, but the activity of the reactivated catalyst is lower because the Relative activity ratio 2.0 -Fe -C -Si Figure 12 Scanning electron microscopy (SEM) image of a spent, fused Fe HTFT catalyst; color coding: red, iron; yellow, carbon; and green, silicon. 1.5 1.0 0.5 0.0 0 50 100 150 Time on line (h) 200 250 Figure 13 Relative activity versus time on stream for a precipitated Ruhrchemie-type iron LTFT catalyst. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 537 H2 + CO H2 + CO (Fe3O4)HD -H2O CO (FeO)HD CO (Fe)HD -CO2 H2O Sintering -CO2 Oxidation FTS CnHm (FexCy)HD Sintering Sintering (Fe3O4)LP (Fe3O4)G (Fe2C)LP 50 nm 50 nm 200 nm HD – highly dispersed phase » 2 nm LP – larger particles » 20 nm G – globules » 400 nm Figure 14 Deactivation mechanism for a typical Ruhrchemie iron catalyst under low-temperature Fischer–Tropsch synthesis conditions. original surface area cannot be recovered. Multiple reactivation steps are therefore not viable. Spent HTFT catalysts may in principle be recycled to make new catalysts. However, using the material in the fusion process has a high energy cost associated with it, due in part to its high carbon content. Iron is a cheap material, and there is little economic incentive for recovering it. Therefore, spent catalysts have usually been landfilled. Currently, awareness of the environmental impact of such procedures is growing, and reclamation of metal – even iron – from spent catalysts, for example, by acid dissolution is more and more seen as a social responsibility of the industry to reduce the impact of commercial processes on the environment. 7.20.3 7.20.3.1 Cobalt-Based FTS Catalysts Introduction Cobalt as an FTS catalyst was already claimed by Fischer and Tropsch in their original patent of 1925.31 The commercialization of the FTS by Germany and Japan in the period 1938–45 relied fully on cobalt catalysts. Only after World War II did the focus shift to the use of iron catalysts for FTS applications. Since the oil crises of the 1970s the interest in cobalt-based FTS catalysts reappeared, which has resulted in numerous scientific papers and patents (see Figure 4). Many companies showed interest in cobalt FTS, for example, BP, ConocoPhilips, Gulf, ExxonMobil, IFP, Johnson Matthey, Sasol, Shell, Statoil, and Syntroleum. Almost all focused on wax production, followed by hydrotreating to produce diesel. This is also the application that will receive most attention in this section. Cobalt FTS catalysts are exclusively utilized in lowtemperature synthesis or LTFT, and are applied in fixed-bed, slurry-phase, and micro-channel FTS reactors. Catalyst design needs to be adjusted to the targeted reactor as well as the applied FTS conditions. Important for catalyst design are the composition, method of preparation, activity and selectivity behavior, deactivation and regeneration, and mechanical integrity. Cobalt FTS catalysts are currently commercially applied by Sasol/QP in the Oryx GTL plant, Qatar (Co/Al2O3), in a slurryphase reactor, and by Shell in the SMDS plant in Bintulu, Malaysia, as well as in the Pearl plant, Qatar (both Co/Mn/ TiO2), in a fixed-bed reactor. Figure 15 shows the catalyst that is used in slurry-phase application. Exciting academic and industrial research in the last 20 years has increased the fundamental knowledge of cobalt FTS catalysts substantially on topics like the nature of the active site, impact of crystallite size on activity and selectivity, and deactivation mechanisms, owing to the application of surface science techniques, model catalysts, in situ analyses at relevant industrial conditions, and molecular modeling.8,19,27,62–67 The literature of the last 20 years shows that quite a wide variety of cobalt catalyst compositions prepared by numerous methods can result in academically and industrially relevant cobaltbased FTS catalytic systems. 7.20.3.2 Composition of Cobalt Catalysts Modern cobalt catalysts are similar to the ones prepared by Fischer and Tropsch in the sense that they consist of promoted cobalt on a metal oxide support. An inspection of the literature and patents on this topic reveals the following general characteristics, with almost all companies with FTS catalysts having a similar formulation for them18,35,68 (Table 5): (a) Cobalt as the FTS active metal (typically 10–30 wt%) (b) A second metal (usually noble) as a reduction promoter (0.05–1 wt%) (c) A structural oxidic promoter (e.g., Zr, Si, and La) (1–10 wt%) (d) A refractory oxidic support (most likely modified) Cobalt is expensive and to maximize its use, it needs be well dispersed on the support. Since cobalt metal is considered the active phase, it is imperative that there is a high density of cobalt metal sites available. The number of cobalt surface sites is a function of particle size and morphology, extent of Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 538 Fischer–Tropsch Synthesis: Catalysts and Chemistry Cobalt catalyst ~ 60 m Wax 0.1 m -4 5 ⫻ 10 m Support particles 10 m Cobalt Slurry phase reactor Support -8 1 ⫻ 10 m A cobalt nanoparticle 1 ⫻ 10-7 m Cobalt and support Structural promoter Reduction promoter Cobalt Cobalt Cobalt Modified support Figure 15 Cobalt catalysts for application in a slurry-phase reactor, and schematical composition of a typical cobalt-based Fischer–Tropsch catalyst. Table 5 Company Shell ExxonMobil Syntroleum BP Sasol Examples of catalyst formulations, as patented by several industrial Fischer–Tropsch synthesis companies Composition Co (wt%) 2nd metal Structural promoter Support 20 12 20 10 20 – Re (1 wt%) Ru (0.1 wt%) – Pt (0.05 wt%) MnO (Co/Mn ¼ 12) Al2O3 (6 wt%) La (1 wt%), SiO2 (0.1–10.6 Si/nm2) Al2O3 (0.5 wt% Al) SiO2 (0.8 Si/nm2) TiO2 TiO2 Al2O3 ZnO Al2O3 reduction, and particle stability.68 It is preferred to have a fairly high extent of reduction (>60%), but it should also be noted that the cobalt is further reduced during the FTS reaction. An optimum cobalt particle size of just above 8–10 nm is preferred as particles below those have shown to have a lower turnover frequency (TOF).69 Additionally, very small particles (4–6 nm) could be more prone to sintering and also may prove very difficult to reduce due to an increased metal-support interaction. It is important that there is a minimum amount of cobalt-support compounds as these are reducible at very high temperatures and are inactive for the FTS reaction.70 The two most common phases of metallic cobalt in supported cobalt FTS catalysts are face-centered cubic (fcc) and hexagonally close-packed (hcp), which often coexist.19,71 It has been Reference Preparation route WO 199700231 US 5268344 WO 2005058493 WO 19913400 US 7365040B2 Coprecipitation Impregnation Impregnation Impregnation Impregnation reported that for cobalt particles less than 40 nm, the predominant phase should be fcc.72 The mode of activation, addition of promoters, and support may influence the relative amounts of the phases.19 Some authors have reported that the hcp phase is more active for FTS.73 Nanometer-sized cobalt particles when supported on traditional oxidic carriers like silica, alumina, and titania are difficult to reduce due to strong interactions with the support. Therefore, catalysts are often promoted with a second metal (e.g., Ru, Pt, or Re) which leads to improved reducibility of the cobalt oxide particles; the increase in amount of active sites results in higher activity compared to un-promoted catalysts. The more facile cobalt reduction is attributed to faster hydrogen activation in the presence of promoter metals and Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry subsequent spillover of hydrogen to cobalt oxides and reduction of cobalt species.19,68 In many cases, the promotion with noble metals leads to a smaller average size of either cobalt oxide or cobalt metal particles. Promotion with noble metal may also play a role during the decomposition of the cobalt precursors and can lead to crystallization of smaller cobalt oxide particles. This increased dispersion is most likely due to a higher rate of nucleation, enabled by the promoter.68 Promoter metals such as Ru have also been claimed to lead to the formation of bimetallic particles and alloys. This influences catalyst activity and selectivity, may inhibit deactivation by keeping the surface clean, and allows easier regeneration of the cobalt surface.74 The metal promoter is usually present at levels of 0.1–0.5 wt%. At these low concentrations reduction is efficiently promoted, and the hydrocarbon selectivity is hardly negatively affected. Structural promoters affect the formation and stability of the active phase of a catalyst material. For Co/silica catalysts, it has been shown that promotion with Zr results in a decreased cobalt– silica interaction, which in turn leads to a higher degree of cobalt reduction and increase in the metallic atoms on the surface.19,75 Zr promotion of cobalt/alumina catalysts has been claimed to prevent formation of cobalt aluminate.76 Incorporation of elements such as B77 and Ni78 increases the stability of cobalt catalysts by suppressing carbon formation. Irreducible oxides such as MnO and CeO2 may also slow down cobalt sintering.63 A wide range of promoters has been studied; the reader is referred to a detailed review by Morales and Weckhuysen.63 The support provides mechanical strength and thermal stability to the cobalt crystallites, while facilitating high cobalt dispersion. The properties of the support are an important factor. For alumina, high purity, low acidity, and relatively high surface area (150–250 m2 g1) are required, according to patents from the 1980s.79–81 More recently, however, aluminabased supports of relatively low surface area (50 m2 g1), such as Ni-promoted a-Al2O3, have been reported to have a positive effect on both mechanical strength and C5þ selectivity.82 The pore size of the support can also influence the size of the cobalt crystallites, as shown by Saib et al. for SiO2-supported catalysts.83 Van Steen and Claeys reported that the desired pore size of the support for the optimum cobalt crystallite size should be around 12–16 nm.61 The support needs to be robust under FTS conditions, implying that it should be able to cope with the presence of several bars of steam that occur at high conversion levels. Van Berge et al. found that an unprotected alumina-supported cobalt FTS catalyst was susceptible to hydrothermal attack during realistic FTS conditions, which resulted in contamination of product wax with ultra-fine, cobalt-rich particulates.23,84,85 This problem was solved by pre-coating the support with silica as structural promoter. TiO2 seems to be the support of choice for both Exxon and Shell based on the most recent patents (Table 5). An advantage of TiO2 is that it has a high hydrothermal stability and can withstand high water partial pressures. The rutile/anatase ratio can be tailored, which influences the surface area and mechanical properties. Supported cobalt catalysts should also be resistant to attrition especially if applied in a slurry bubble-column environment. Wei et al.86 noted that the attrition resistance of supported cobalt catalysts follows the sequence: Co/Al2O3 > Co/SiO2 > Co/TiO2. 539 There has also been work conducted on less conventional supports such as MCM-41, SBA-16, and carbon nanofibers, nanotubes, and spheres.19,69,87 These studies are mainly academic in nature but further fundamental understanding of cobalt FTS catalysts considerably. Carbon supports interact weakly with cobalt and allow for a high degree of cobalt reduction, thus enabling the study of cobalt particle-size effects.69,87 7.20.3.3 Preparation of Cobalt Fischer–Tropsch Catalysts The preparation of cobalt FTS catalysts aims to achieve the optimal crystallite size distribution in a particle that is optimal for its application in a specific fixed-bed, bubble-column, or microchannel reactor. As the optimum size range of catalyst particles for the different reactor types varies (see Figure 3), preparation methods and equipment depend on the targeted reactor application. Important considerations for choosing a particular method of preparation and starting components are to minimize poisons (e.g., Na, S, Cl) in the catalyst and the type of waste streams resulting from the chosen method. A number of procedures for preparing cobalt FT-catalyst precursor exist: • • • coprecipitation of cobalt, promoters, and support, followed by catalyst particle shaping. In a variation of this method, the support is added just before particle shaping, precipitation or impregnation of cobalt and promoters onto pre-shaped support particles, and impregnation of cobalt (oxide or metal) particles onto preshaped supports. 7.20.3.3.1 Precipitation Most of the initial FTS-catalysts (e.g., Co/ThO2/kieselguhr) were made by coprecipitation.88 This method has been applied for some of the modern cobalt catalysts as well, for example, for Co/Mn catalysts,89 Co/Mg/SiO2 and Co/ZnO2.90 Catalyst preparation based on coprecipitation usually consists of three steps: precipitation, washing and drying, and shaping. Selection of chemicals is of course an important consideration in view of the associated waste streams. Chemical precipitation of the cobalt, promoter, and support by a precipitation agent can be done batchwise or continuously at constant pH. The cobalt precipitates as a hydroxide, which can exist as green a-Co(OH)2 or pink b-Co(OH)2 polymorphs. The former is metastable and readily transforms into the stable b-phase. Crystallite size and composition of the precipitate are controlled by temperature, precipitation agent, precursor salts, structure directing or organic hydrolysis reagents, aging time, and reaction atmosphere (air or N2). Using Na2CO3 or KOH as precipitation agents in the preparation of Co/SiO2 catalysts would lead to cobalt silicate formation.91 To prevent formation of the inactive cobalt silicate, the silica is added after the precipitation. Filtration and washing of the precipitate is required to remove excess chemicals. Even low levels of alkali metals and halogens left in the washed precipitate can severely degrade the catalyst’s performance. Shaping of the catalyst precursor depends on the reactor application. For bubble beds, the precipitate is usually reslurried and spray-dried to obtain the required particle-size distribution. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 540 Fischer–Tropsch Synthesis: Catalysts and Chemistry on particle geometry (diffusion path length and pore size), viscosity of the suspension, interaction between solution and support surface (contact angle and surface tensions), and diffusion coefficients.93–95 Figure 16 shows the effect of diffusion on the different cobalt distributions observed on small alumina particles for incipient wetness impregnation compared to slurry-phase impregnation, as obtained from SEM/ EDX line scans.96,97 For fixed-bed catalysts, eggshell-type cobalt distributions are sometimes preferred to overcome pore diffusion limitations on performance and selectivity. Concentrating the cobalt in the outer layers of the support is, among others, achieved by adding viscosity enhancers or using cobalt salt melts for impregnation.98 For deposition precipitation onto pre-shaped supports, the same parameters that determine the time required during impregnation to get a homogenous distribution (e.g., particle geometry, diffusion path length, pore size, viscosity of the suspension, interaction between solution and support surface, contact angle and surface tensions, and diffusion coefficients) are important.99 For fixed-bed reactors, the precipitate is extruded or pelletized. Addition of acids to the washed and dried precipitate is done to improve the final catalyst’s particle strength.92 7.20.3.3.2 Preparation methods involving pre-shaped supports Support morphology and characteristics play an important role in optimizing the preparation of cobalt catalysts on pre-shaped support particles. As the aim is to get a desired amount of cobalt crystallites onto the support and maintain a crystallite size of around 8–10 nm, the following support characteristics need consideration: • • • The support pore volume dictates how much cobalt precursor can be added per impregnation. As shown in Table 6, 30 g of metallic cobalt per 100 g of support occupies only 0.03 ml g1 of support material, but when using Co(NO3)26H2O as the precursor, a pore volume of 0.79 ml g1 is required. The time required to get a homogeneous cobalt distribution throughout a support particle during impregnation depends • Pore volume requirements for different cobalt components Table 6 Cobalt compound Molar mass (g mol1) Cobalt mass fraction (%) Density (g cm3) Pore volume required for a loading of 30 g of Co per 100 g of support (ml) Co CoO Co3O4 Co2O3 CoOOH Co(OH)2 Co(NO3)2 Co(NO3)2 6H2O CoCl2 6H2O 59 75 241 166 92 93 183 291 237 100 0.79 0.73 0.71 0.64 0.64 0.32 0.20 0.25 8.9 6.4 6.1 5.2 5.0 3.6 2.5 1.9 1.9 3.4 5.9 6.7 8.2 9.4 13.2 37.7 79.0 62.5 60 50 50 40 40 Mass% Co3O4 Mass% Co3O4 60 30 20 10 -10 30 20 10 0 0 10 20 30 40 50 60 70 Distance from edge of catalyst particle (mm) 80 -10 0 0 10 20 30 40 50 60 Distance from edge of the catalyst particle (mm) 70 Figure 16 Macroscopic cobalt crystallite distribution, as measured by scanning electron microscopy (SEM) line scans for: (a) incipient wetness impregnation followed by immediate fast drying and (b) slurry-phase impregnation allowing 3 h for the cobalt to slowly disperse throughout the alumina particle, followed by fast drying. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry • • In addition, diffusion differences between the precipitation agent and cobalt will impact the final cobalt distribution during precipitation. When depositing bulky cobalt hydroxide crystallites or cobalt metal particles100 onto preshaped support particles, the important parameters are pore size, particle diameter, and bulkiness of the precipitate. During the drying of the catalyst precursor, the same parameters as highlighted above for impregnation and deposition should be taken into account to prevent the cobalt precursor (if not chemically fixed to the support) from migrating out of the particle again. In addition, heat transfer coefficients, evaporation enthalpies, and particle outer surface area need consideration for optimizing the drying phase of the cobalt catalyst preparation. Each preparation method needs to be optimized carefully, and one cannot assume that the optimum procedures for one type of support and support particle shape will be the same for all other supports and support particle shapes. 7.20.3.3.3 Calcination Usually, drying is not fully accomplished and therefore the first stages of calcination actually complete the drying phase. To maintain the cobalt distribution achieved by impregnation or precipitation during drying and calcination, the cobalt component mobility must be hindered. One way of achieving this is to ensure that the cobalt component stays in a viscous or solid form. For catalysts obtained by impregnation from cobalt nitrate solutions, this implies that during calcination the combination of heating rate and air flow must be such that water and NOx are immediately removed.101 As the mobility of the cobalt phase can be minimized by fast calcination, heat flow into the system is also important as both the drying and nitrate decomposition are endothermic. Performing calcination under different atmospheres provides a way to affect the dispersion of the cobalt phase. NO addition during calcination leads to the formation of a less mobile cobalt hydroxyl nitrate.102 Using H2 or CO as decomposition medium at temperatures below those where reduction takes place also gives catalysts with good cobalt dispersions.103 Adding organic additives during impregnation is another method to influence cobalt nitrate decomposition. Oxidation of the additive is an exothermic process, which provides heat for the endothermic nitrate decomposition, and thus accelerates its decomposition. The transmission electron microscopy (TEM) images in Figure 17 illustrate how cobalt distributions change when different calcination conditions are applied. 7.20.3.3.4 Reduction Cobalt catalysts are usually reduced in hydrogen or a diluted hydrogen atmosphere. Examples of CO reductions are also found, but carbon formation on the cobalt crystallites should be avoided. Reduction of cobalt oxide to cobalt metal occurs in two exothermic steps: Co3 O4 þ H2 ! 3CoO þ H2 O [6] CoO þ H2 ! Co þ H2 O [7] 541 For optimal reduction, care must be taken to optimize heat transfer, minimize hydrogen diffusion and mass transfer limitations, and to remove water effectively. The latter benefits from high hydrogen space velocities and application of low heating rates.104 Figure 18 shows the impact of the water partial pressure during reduction of a 30 g Co/0.075 g Pt/100 g alumina catalyst on the starting FTS performance, indicating that low water content should be targeted for maximum activity, in agreement with the thermodynamics of reduction as expressed in eqn [5]. Hence, the hydrogen stream used for reduction must be as dry as possible. For small catalyst particles as used in slurry-bed reactors, fluidized bed reduction reactors are preferred and the above requirements are easily met. For reductions in fixed-bed reactors, more care must be taken to overcome the limitations especially toward the reduction reactor outlet. The maximum temperature required for reduction of cobalt catalysts depends on the level of reduction promoter present (Pt, Ru, Pd, etc.), the presence of other promoters (e.g., alkali metals make reduction more difficult), the support, the support modifiers, and the catalyst precursor used in the preparation. Reduction temperatures that are too high can cause sintering and loss of cobalt metal surface area. In the case of Co/SiO2 catalysts, cobalt silicate formation has been reported for temperatures higher than 350 C. Catalyst performance depends critically on the reduction procedure.105 Application of reduction–oxidation–reduction (ROR) cycles has been reported to improve the FTS performance of cobalt catalysts by up to 30%.106 Some of the reasons given in the literature for this improved performance from ROR treatment are: (1) rougher (more steps on the surface) cobalt crystallites, (2) higher degree of reduction, and (3) redispersion of the cobalt on the support surface. 7.20.3.4 Cobalt Catalyst Fischer–Tropsch Performance Both activity and selectivity are of course important parameters for cobalt-based FTS catalysts. High activity is important for slurry-phase catalysts, while for fixed-bed catalyst the heat removal capacity needs to be balanced with the activity of the catalyst. From a selectivity point of view, a low methane selectivity is normally desired, combined with a high C5þ selectivity or a high chain growth probability (a). Determining the intrinsic catalytic performance of cobalt FTS catalysts is not a straightforward exercise, as it is influenced by the choice of reactor and conditions. Khodakov et al.19 summarize a number of issues and choices related to the testing of FTS catalysts, for example: (i) reactor choice: fixed-bed, slurry-bed (or continuous stirred-tank reactor, CSTR), or high-throughput reactors, (ii) hydrodynamics, (iii) heat transfer and hot spots, (iv) intra particle and external mass-transfer limitations, and (v) atmospheric or elevated pressure. As testing catalysts under different FTS conditions (H2/CO ratio, T, and P) will result in different catalyst performances35,107 and therefore possibly selection of different catalysts, it is important in the early stages of research to understand the long-term scaling-up view, with respect to reactor choice and FTS conditions. Taking the same cobalt catalyst and testing it in different manners can result in very different catalytic activity behavior. Figure 19(a) clearly shows that the FTS activity is influenced by the water partial Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 542 Fischer–Tropsch Synthesis: Catalysts and Chemistry 0.5 μm (a) 0.5 μm (b) 0.5 μm (c) 0.5 μm (d) Figure 17 Cobalt crystallite distributions as measured with STEM for cobalt alumina catalysts calcined in different manners. (a) Cobalt oxide microglobical formation of a 30 g Co/100 g alumina catalysts using a heating rate of 1 C min1 and an air space velocity of 1 m3n per kg Co(NO3)2.6H2O per hour. (b) Cobalt oxide distribution of a 30 g Co/100 g alumina catalysts using optimized heating rate and air space velocity to ensure optimum calcination. (c) Cobalt oxide distribution on a 30 g Co/100 g alumina catalysts using carbon coated alumina , using the same heating rate and air flow rate as in (a). (d) Cobalt oxide distribution on 30 g Co/100 g alumina catalysts using the same heating rate and flow rate as (a) but with 1% NO in He as calcination atmosphere. pressure applied during the test, which possibly impacts factors such as sintering and carbon deposition, as well as surface and active site reconstruction.65 The selectivity behavior of cobalt catalysts is also strongly influenced by parameters such as temperature, hydrogen and carbon monoxide partial pressures, and conversion. Comparing catalysts tested at different conditions should thus be done with care. Figure 19(a) clearly shows that cobalt catalysts are more active in fixed-bed than in slurry-bed reactors. However, cobalt catalysts in slurry-phase reactors are normally applied at temperatures around 230 C, while in fixed-bed reactors they are normally used at temperatures around 210 C. The productivity per gram of catalyst is therefore higher in slurry-phase reactors (Figure 19(b)). For heterogeneous catalysts the activity often increases with smaller particle size, as the metal surface area increases. This was confirmed by Iglesia108 who showed that the activity of cobalt catalysts is directly proportional to the amount of cobalt metal surface in the catalyst. The TOF or the reaction rate per unit of cobalt surface area was stable over the range of cobalt particles that was investigated (9–200 nm). FTS over cobalt catalysts was therefore regarded to be structure insensitive. Thereafter, a number of authors have investigated the effect of cobalt metal particle size on the intrinsic activity of supported cobalt catalysts for smaller cobalt particles, that is, well below 10 nm.69,87,109–112 Bian et al.110 using Co/SiO2 confirmed Iglesia’s results for samples with cobalt particles between 11 and 29 nm, but Barbier et al.,109 Bezemer et al.,69 Martinez and Prieto,111 Coville and coworkers87 all showed that the TOF was stable for catalysts with cobalt particles above 8–10 nm, while it decreased sharply for catalyst with smaller particles. Only Borg et al.113 reported no sensitivity for the Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry As mentioned above, the C5þ selectivity depends strongly on the FTS process conditions. For a constant set of conditions, Iglesia114 found no particle-size effect on selectivity for particles between 9 and 100 nm. A few years later, however, Barbier et al.109 reported a strong dependency of the chain growth probability, a, on particle size. Increasing the cobalt diameter from 4.5 to 9.5 nm caused the a-value to increase from 0.74 to 0.87 (at 170 C, 1 bar). Bian et al.110 showed a similar, though less pronounced, trend with the a-value increasing from 0.85 to 0.89 when the particle size increased from 11 to 29 nm (at 200 C, 10 bar). Bezemer et al.69 reported a very clear particlesize effect on selectivity at atmospheric pressure, with a methane selectivity that was stable for particles larger than 6 nm, but increased sharply for smaller particles (220 C, 1 bar). However, the reported data at high pressure (35 bar and 210 C) clearly show that the C5þ selectivity still increases with increasing particle size up to 15 nm. Xiong et al.,87 Prieto et al.,112 and Borg et al.113 all confirmed the general trend on an increasing C5þ selectivity with increasing particle size extending beyond 10 nm, and up to 20 nm. Little fundamental understanding has been offered to explain this particle-size effect on both activity and selectivity. Interestingly, the effect of size on activity is very pronounced for particles smaller than 10 nm, while the impact on selectivity seems to be more gradual and does not level off above 10 nm. The particle-size effects cannot, as previously suggested, be explained by the oxidation of the smallest particles. Bezemer et al.69 showed with x-ray absorption near-edge structure (XANES) measurements that oxidation did not occur. This is in line with extensively reported research that cobalt oxidation during FTS does not occur for cobalt particles larger than activity of cobalt particles with sizes down to 3 nm. The relationship between TOF (or reaction rate per unit cobalt surface area) and cobalt particle size for above-mentioned publications is summarized in Figure 20. As the TOF numbers were obtained under different FTS conditions (i.e., temperature and partial pressures), they were normalized to enable comparison of trends in the different papers and therefore expressed in arbitrary units. It is clear that for catalysts with cobalt crystals above 10 nm the TOF is structure insensitive, while there is a sharp decrease in activity for particles smaller than 8–10 nm. Relative FT activity 1.0 0.9 0.8 0.7 10 30 50 70 90 110 PH 0 (mbar) 130 150 543 170 2 Figure 18 Impact of water partial pressure during reduction of a 30 g Co/0.075 g Pt/100 g alumina catalyst on its initial Fischer–Tropsch synthesis performance. Productivity (mole CO converted/g cat/s) 1.6 1.4 Activity (au) 1.2 1 0.8 0.6 0.4 0.2 0 0 200 400 600 Time (h) 800 1000 0 200 400 600 800 1000 Time (h) Figure 19 (a) Three Fischer–Tropsch synthesis runs with same Co/Al2O3 catalyst, as tested at 20 bar, 230 C, and H2/CO ¼ 2. Red solid: fixed bed, PH2 O inlet ¼ 0; PH2 O outlet ¼ 3.0 bar. Red open: fixed bed with water co-feeding; PH2 O inlet ¼ 2.6, PH2 O outlet ¼ 4.0 bar. Blue solid: slurry bed; PH2 O inlet ¼ 0; PH2 O outlet ¼ 4.5 bar; all the catalyst is exposed to outlet water partial pressure in a CSTR laboratory slurry reactor. (b) Two Fischer– Tropsch synthesis runs with same Co/Al2O3 catalyst, as tested at 20 bar, 60% conversion, with the fixed bed run at 210 C (red circles) and the slurry phase run at 230 C (blue triangles). Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 544 Fischer–Tropsch Synthesis: Catalysts and Chemistry of Codþ–SiO2 sites in small particles, caused by flattening of the crystals during FTS, and possibly due to carbon-induced surface reconstruction. Another study showed that carbon and oxygen atoms, originating from dissociated CO, were very strongly bonded on small particles, possibly blocking active sites for further CO dissociation.117 As the impact of the particle size is different for activity from for selectivity, the fundamental explanation of these effects seems to involve more than one kinetically relevant mechanistic step in the FTS process. Understanding this requires more fundamental research. 7.20.3.5 Deactivation and Regeneration of Cobalt Fischer–Tropsch Catalysts Catalyst stability is crucial for the economics of cobalt FTS, in addition to other important factors such as high activity, selectivity, and mechanical strength. Understanding catalyst deactivation is essential for improving catalyst stability and for developing effective regeneration procedures. Figure 21 shows a typical deactivation profile for Co FTS catalysts under TOF (au) 2 nm.65 The particle-size effect can also not be explained by sintering as this was not observed by Bezemer et al.69 and by the Coville group.87 It is clear that the Co particle-size effect in FTS extends beyond the classical impact of size, which derives from the fraction and type of surface atoms as a function of crystallite size and normally does not extend beyond 4 nm particles. It was suggested69 that the optimum combination of active sites for the different elemental reactions of FTS (i.e., CO dissociation, hydrogenation, and insertion) requires relatively large cobalt particles, possibly combined with a COinduced surface reconstruction. This might be related to the presence or absence of the so-called B5 site,65 which has been speculated to be the most active site for CO dissociation, and needs a certain particle size to be present in high abundance. Another speculation is that the particle-size effect might be related to specific bonding modes of CO, such as the bridgebonded CO coordination; this mode is believed to be favored on large particles and held responsible for an increased CO dissociation rate, which would lead to an increased reaction rate.87,109 Based on in situ Fourier transform infrared spectrometry results, Prieto et al.112 proposed an enhancement Coville (2011) Fischer (2010) Pietro (2009) Borg (2008) Martinez (2007) De Jong 1 bar (2006) De Jong 35 bar (2006) Bian (2003) Barbier (2001) Iglesia (1997) 0 5 10 15 20 25 30 Co particle size (nm) Figure 20 The TOF or Fischer–Tropsch synthesis rate per unit surface area, as a function of the cobalt metal particle size.69,109,110,112–116 The TOF had to be scaled due to variations in process conditions, and is therefore reported in arbitrary units. 1.0 RIAF 0.8 0.6 0.4 0.2 0.0 0 10 20 30 40 50 60 Time on line (days) Figure 21 Normalized activity stability for a Co/Pt/Al2O3 catalyst during realistic Fischer–Tropsch synthesis in a laboratory scale micro-slurry reactor at fixed CO conversion (230 C, 20 bar, H2 þ CO conversion of 50–70%, feed gas composition of 50–60 vol.% H2 and 30–40 vol.% CO). Adapted from van de Loosdrecht, J.; Bazhinimaev, B.; Dalmon, J. A.; Niemantsverdriet, J. W.; Tsybulya, S. V.; Saib, A. M.; van Berge, P. J.; Visagie, J. L. Catal. Today 2007, 123, 293–302. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 1.8 Hydrogen resistant carbon (wt%) commercially relevant FTS conditions.118 Deactivation is initially stronger, after which it starts leveling off. Similar deactivation profiles have been reported by other laboratories.18,62,119,120 Fundamental studies on catalyst deactivation essentially involve understanding differences in the fresh and spent catalysts with respect to the active site. The B5 sites on metallic cobalt are currently considered as active sites for the FTS.65 Hence, changes to the number or nature of these sites will contribute to deactivation. However, due to the complexity of the FTS process and the lack of suitable techniques to characterize any particular site on a surface, most studies on deactivation are necessarily limited to ‘observables’ such as changes in metallic cobalt surface area. This in itself is no trivial matter, as cobalt catalysts are sensitive to their environment, and spent FT catalysts are embedded in wax (which actually protects them from being exposed to the air). As is so often the case in fundamental studies of catalysts, a combined approach using real catalyst and model systems, advanced in situ and ex situ characterization, combined with molecular modeling has given detailed insight into FTS catalyst deactivation.65 The main deactivation mechanisms of cobalt FTS catalysts and its active sites, as proposed in the literature, are: (1) oxidation, (2) mixed metal-support interaction, (3) carbon deposition and carburization, (4) sintering, (5) poisoning, and (6) surface reconstruction.19,62,64,65,70,71,108,114,118,121–133 Over the last 15 years, oxidation of cobalt by the product water was seen as the major deactivation mechanism in the open literature.65 However, many recent publications have disproved this.62,64,65,71,118,134Following an in-depth study on oxidation using model systems, molecular modeling, surface thermodynamic calculations, and an industrial catalyst tested under commercially relevant conditions, the key finding is that oxidation is crystallite size and condition dependent, that is, under realistic FTS conditions (H2O/H2 ¼ 0.5–3). Co crystallites with diameters larger than about 2 nm will not undergo oxidation. In fact, from XANES analyses of spent Co catalysts from an extended FTS run, it was found that a further reduction took place during FTS118 and has been confirmed by others.135–137 Further, the formation of metal-support compounds such as cobalt aluminate have been considered as a deactivation mechanism.129,138 Although thermodynamically favorable, this reaction needs CoO formation as an intermediate, which does not take place under realistic FTS conditions. Indeed, a recent study showed that the minor cobalt aluminate formed during FTS originates from unreduced CoO present in the fresh catalyst and not from Co metal.65,122 This leaves sintering and carbon deposition as the major contributors to Co FTS catalyst deactivation. Carbon deposition on an FTS catalyst that is covered by growing hydrocarbons and is entirely embedded in product wax represents a real challenge. Nevertheless, deleterious carbon arising from CO or FT products can have a wide range of negative effects on Co FTS catalysts. We mention pore blockage, resulting in mass transfer limitations, formation of bulk or surface carbides, and blockage or alteration of active sites.121 Pore blockage by long hydrocarbon products resulting in diffusion limitations of the reactants CO and H2 has been mentioned as a deactivation mechanism from the onset of the discovery of the FTS.139 A hydrogen treatment or solvent wash of the spent catalyst resulted in a partial recuperation of 545 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 20 40 60 80 Time on stream (days) 100 Figure 22 Build up of polymeric carbon on a cobalt slurry-phase Fischer–Tropsch synthesis catalyst as a function of time on stream.121 FT activity. Although detailed studies are unknown to us, pore blocking is generally accepted as a deactivation mechanism. Recently, strong evidence has been found for the accumulation of a rather inactive polymeric carbon on the metallic surface of cobalt resulting in catalyst deactivation (Figure 22).121 Techniques involved in this work were temperature-programmed hydrogenation, low-energy ion scattering, and energy-filtered TEM. Molecular modeling suggested that the polymeric carbon might be a form of graphene. In general, there is good agreement that bulk cobalt carbide is metastable and will not be present in substantial quantities during FTS for cobalt, although Karaca et al.71 observed small quantities of Co2C in their in situ XRD experiments (which, interestingly, was entirely absent for the first 9 h on stream, but appeared in the measurements after 10 h; exposure to pure CO made the signal grow further). Although characterization of spent cobalt FT catalysts run for several months in a slurry bubble column did not show bulk cobalt carbide formation,118,121 a possible role of subsurface carbon cannot be ruled out. To date, our knowledge on subsurface carbon comes from molecular modeling. Calculations on carbon clusters by Zonneville et al.140 indicate that carbon in subsurface positions affects the CO dissociation rate and may therefore affect the FTS activity as well. More work is needed to ascertain the impact of subsurface carbon. Carbon deposition can be decreased by adding promoters. Examples from the literature include ruthenium74 and boron.77,141,142 Sintering is a thermodynamically driven process whereby smaller, more unstable particles grow to form larger, more stable particles that are lower in surface energy. Sintering as a deactivation mechanism is easier to investigate than carbon deposition, with TEM and XRD being the most common techniques used. Due to this in general there is good agreement in the literature on the importance of sintering as a deactivation mechanism of cobalt FTS catalysts.62,65,70,71,121,134,137,143–145 Key factors that affect the rate of sintering are the reaction temperature and the partial pressure of water: an increase in either of these results in enhanced sintering. The choice of support also plays a key role. Alumina is considered to provide more stability against sintering than silica does, due to the improved metal support interaction in the former. Both Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry particle migration and Ostwald ripening are expected to be important, but further studies are desirable to gain full understanding on the contribution of each. Poisoning of cobalt catalysts by S, NH3, HCN, Hg, and Cl is well known and is an issue for cobalt-based FTS62,146–149 especially for coal-to-liquid applications.65 Sulfur is a strong, irreversible poison with a large adsorption energy. Due to its size and electronic effect, sulfur will also poison adjacent cobalt sites. Once adsorbed, sulfur is difficult to remove and will accumulate with time.146 Sulfur poisoning can however relatively easily be prevented by cleaning the synthesis gas feed properly, for example, by using zinc-oxide or lead-oxide guard beds. Poisoning of cobalt-based FTS catalysts by means of nitrogen-containing compounds such as NH3 and HCN is a known effect and postulated to arise from competitive adsorption.150 The impact of N-compounds is less severe than that of S-compounds and can be reversed by mild hydrogen treatment. Nevertheless, reducing their level to parts per billion is recommended.150 Surface reconstruction is a thermodynamically driven process which results in a lowering of the surface energy and therefore can contribute to catalyst deactivation. Using molecular modeling, Ciobica et al.124 showed that atomic carbon from dissociated CO can cause a reconstruction of the Co fcc (111) surface to a Co (100)-like structure, followed by a clock reconstruction. This surface is less active and could therefore contribute to deactivation. As the reconstruction is accompanied by a change in surface density, it could, ironically, also assist in the formation of more reactive sites, as proposed by Wilson and de Groot.151 This is a complex phenomenon that needs further investigation. Methods to reverse deactivation and regenerate deactivated Co FTS catalysts have been around since the early days of Fischer and Trospch.3 The most common methods reported in the open literature are treatment of the deactivated catalyst in hydrogen or in steam, applying oxidation–reduction cycles, and combinations of these.65,152 With carbon and sintering being the major deactivation mechanisms during FTS, ‘oxidation–reduction’ is considered to be the most robust and preferred method to regenerate spent Co FT catalysts.65 By careful control of the oxidation step, deleterious carbon is removed at temperatures above 250 C. The oxidation step is also key for the redispersion of cobalt (see Figure 23). The 20 nm mechanism of redispersion of cobalt has been proposed to be a two-step process, that is, (1) oxidation to form hollow spheres by the Kirkendall effect and (2) multinucleation of Co3O4 during reduction to produce smaller crystallites.65,153 Poisons such as sulfur are removed by oxidation with steam and air to sulfates, followed by washing them out. However, phases originating from strong metal-support interaction are very difficult to reverse and their formation should be prevented.154 Substantial progress has been made toward understanding deactivation and regeneration of Co FTS catalysts, but more specific knowledge, for example, on deactivation by carbonaceous species and sintering mechanisms, is definitely necessary. With currently available in situ characterization, synchrotron-based techniques, and molecular modeling, we expect major advancements in the coming years. 7.20.4 Mechanisms and Kinetics of FTS Over Iron and Cobalt Catalysts 7.20.4.1 Introduction The complicated nature of the FTS is among others reflected by its complex product spectrum, consisting of methane, C2þ olefins and paraffins (linear and branched), oxygenates (mainly alcohols, but also aldehydes and ketones), and even aromatics (at sufficiently high operating temperatures). Three classes of mechanisms have been proposed, each assuming a different monomer for chain propagation. The carbide mechanism (Table 7) was first formulated by Fischer and Tropsch in 1926, which proposes that CO dissociates before the carbon atom is partially hydrogenated to a CHx species.2 These CHx species combine by the addition of one monomer at a time to effect hydrocarbon chain growth. The growing intermediate can then terminate in different ways before leaving the catalyst surface, giving rise to an alkane, an alkene, or an oxygenate. A number of variations have been considered within the basic carbide mechanism, arising from, for instance, whether CO dissociation occurs unassisted or via interaction with hydrogen, and the number of hydrogen atoms in the monomer (CH or CH2). The enol mechanism (Table 7), proposed by Storch in the 1950s, assumes that CO is partially hydrogenated to a CHOH species (oxymethylene) which then 20 nm 546 20 nm Figure 23 Co particles supported on a flat SiO2 support during different stages of an oxidation–reduction cycle. Cobalt particles in the metallic state before oxidation (left). Hollow oxide particles, formed upon oxidation (middle). Reduction of the hollow particles, which break up into several smaller metallic particles (right).65,153 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry Table 7 Initiation Propagation Termination 547 Fischer–Tropsch Reaction Mechanisms Carbide mechanism Enol mechanism CO insertion mechanism CO ! C þ O O þ 2H ! H2O C þ xH ! CHx R þ CHx ! R–CHx R–CHx þ (2 x)H ! R–CH2 R–CH2 þ H ! R–CH3 R–CH2–CH2–H ! R–CH ¼ CH2 R þ CO ! oxygenates CO þ 2H ! CHOH CO ! C þ O O þ 2H ! H2OH R–C–OH þ CHOH ! R–C–COH þ H2O R–C–COH þ H ! R–CH2–COH R–CH2–COH þ 4H ! R–CH2–CH3 þ H2O R–CH þ CO ! R–CH–CO R–CH–CO þ H ! R–CH2–CH þ H2O R–CH þ 2H ! R–CH3 R–CH2–COH þ nH ! oxygenates R–CH–CO þ nH ! oxygenates NB : R ¼ H, alkyl(CH3, CH3CH2, CH3CH2CH2, . . .). acts as the monomer.3 Chain growth occurs via a condensation reaction with water elimination. Again, different termination routes determine what final product molecule is formed. In the 1970s, Pichler and Schultz introduced the CO insertion mechanism (Table 7), which assumes a similar initiation step to the carbide mechanism.155 The difference resides in the way that chain growth is proposed to occur, namely by direct CO insertion into the growing intermediate followed by hydrogenation to remove the oxygen atom. Even though the FTS has been known since the 1920s, it is evident from the foregoing that its mechanism is still a matter of debate. In order to progress the understanding of the mechanism, a multidisciplinary approach is required. Subsequently, three such disciplines (model surface science experiments, density functional theory (DFT) calculations, and macrokinetic studies) will be briefly discussed with particular emphasis on their implications for mechanistic and kinetic understanding. 7.20.4.2 Surface Science Studies and Model Reactions A surface science approach is very powerful to study elementary reaction steps in isolation. Conceptually, it is very close to the approach taken by DFT calculations: take a well-defined surface, that is, a single-crystal surface of the material you want to study and use an ultrahigh vacuum so that the adsorbate of interest can be introduced with high purity and with a high accuracy, down to a sub-monolayer coverage. The model system can then be studied with many different sophisticated analysis techniques that give information on the atomic level. The number of studies with cobalt and iron single-crystal surfaces related to the FTS is relatively small. Iron carbide single-crystal work is not available, as far as we know. Studies on nickel and rhodium crystals are more numerous, as they are easier to use and because of the fact that both metals are used as a catalyst for a number of reactions. In this section, we discuss the most relevant surface science findings on cobalt, with a focus on studies where a single elementary step was studied in isolation. 7.20.4.2.1 Adsorption of CO and hydrogen on model surfaces One of the simplest experiments one can do is to study the interaction of a surface with CO and H2, the reactants in the FT reaction. In a typical experiment, a clean close-packed Co surface is exposed to increasing amounts of CO at a sample temperature of 180 K.156 Such a low-temperature experiment in ultrahigh vacuum (UHV) conditions is equivalent to increasing the CO pressure in a room temperature experiment.157 Initially CO adsorbs, with a sticking coefficient of 0.7, on top sites up to a coverage of 0.33 ML (monolayer), with an adsorption energy of 115 kJ mol1.156 This translates into a desorption temperature around 400 K. Upon further dosing, the CO coverage increases, until a saturation coverage of 0.65 ML is reached. Increasing the CO coverage beyond 0.33 ML leads to complex overlayers where CO occupies bridge and threefold sites as well as top sites. The downward shift of the CO desorption temperature for coverages beyond 0.33 ML is mainly caused by repulsive interactions between CO molecules rather than by the difference in adsorption site. This has important implications for the interpretation of vibrational spectra on supported catalyst particles, where occupation of both top and bridge/threefold sites is typically detected. Occupation of bridge and threefold sites can simply be caused by a high CO coverage on the facets of the particle rather than by the presence of special sites. Hydrogen/deuterium adsorption on a close-packed Co surface was studied in a similar fashion: hydrogen adsorbs at 180 K with a low sticking coefficient, up to a coverage of 0.5 ML hydrogen atoms, with an adsorption energy of 33 kJ mol1 (per H atom). Recombinative desorption occurs between 300 and 400 K.158 Hydrogen desorbs at lower temperature from more open surfaces, around 300 K, indicating a weaker adsorption onto those surfaces. The sticking coefficient on the other hand is much higher than on close-packed surfaces: 0.76 for an open surface compared to 0.05 on a close-packed surface.159,160 This enhanced hydrogen sticking is commonly observed on the more open crystal planes of different metal surfaces. When CO is dosed at 180 K on hydrogen-covered surfaces, CO partially replaces the hydrogen, and only 50% of the initial coverage remains on the surface. The remaining hydrogen is less strongly bound due to the CO, and as a result the desorption peak shifts downward by 100 K. CO, on the other hand, is only mildly influenced by the presence of hydrogen, and any influence of hydrogen is only seen at low temperatures.158 These experiments show that repulsive interactions exist between hydrogen and CO, which adds to the barrier to form hydrogenated HxCO species. Other adsorbates such as sulfur, oxygen, and carbon give rise to a similar downward shift of the hydrogen desorption temperature, and those species also (partly) block the surface for hydrogen adsorption.158 Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 548 Fischer–Tropsch Synthesis: Catalysts and Chemistry 7.20.4.2.2 C–O bond scission O1s hv = 650 eV O1s, TP-XPS hv = 650 eV rapidly and are not observed in significant concentrations on the surface. This is in line with the finding that formaldehyde (H2CO) decomposes with 100% selectivity to CO and H2 between 100 and 200 K.160 The experiments show that partially hydrogenated species such as HCO and H2CO are very unstable, and the barrier to decompose via dehydrogenation is obviously much smaller than that of (Hx)C–O bond cleavage. Similar experiments have been reported on 2-nm cobalt particles supported on alumina. In those experiments, 60% of the methanol that was present dehydrogenated, while 40% underwent C–O bond cleavage.164 In this experiment, it was not clear whether the C–O bond scission was assisted by the presence of hydrogen, as CO (produced by methanol dehydrogenation in the methanol experiment) also dissociates on those particles in the absence of hydrogen. In the CO insertion mechanism (Table 7), a CO molecule is inserted into a CxHy, after which the C–O bond has to be cleaved to generate a Cxþ1Hz intermediate that can insert another CO molecule. In other words, the CO molecule is chemically modified by insertion of an alkyl group on the C-end of the molecule. Experiments using ethanol on a closepacked Co surface gave information about the effect of alkyl modification on the C–O bond cleavage.166 Figure 24 shows the result of such an experiment: ethanol adsorbs as an ethoxy species at 160 K. This ethoxy species decomposes around 350 K, via an acetaldehyde intermediate, with an activation barrier of 70 kJ mol1. The products of this dissociation step are atomic O and a C2Hx species, demonstrating C–O bond cleavage. This means that alkyl insertion in the CO molecule facilitates C–O bond scission. For the CO insertion mechanism, it implies that after the CO is inserted the C–O bond can be readily broken and a Cxþ1Hy species is formed that can undergo further chain growth. C1s, TP-XPS hv = 380 eV C1s (high res.) hv = 321 eV 300 K O–CH2–CH3 350 K 250 K 250 K C2H2 Photo-emission intensity (a.u.) Photo-emission intensity (a.u.) O–CH2–CH3 250 K A key step in any FT mechanism is the cleavage of the C–O bond. The close-packed surface of cobalt, Co (0001), is unable to cleave the CO bond: CO desorbs as a molecule with the CO 156 Some open surfaces of cobalt, such as Co bond intact. 161,162 1012 and Co 1120 ,163 and cobalt foils are capable of cleaving the CO molecule directly: after a CO thermal desorption experiment161,162 or a prolonged exposure to CO at elevated temperature,163 carbon and oxygen are found to be left on the surface. CO dissociation stops when enough carbon and oxygen has built up to block all the active sites for dissociation. UHV studies on 2-nm Co particles on alumina show that CO dissociates during a CO thermal desorption experiment, which can be explained by the high defect density on such small nanoparticles.164 These studies did not report exact temperatures or activation barriers for CO dissociation, but in all cases a typical reaction temperature in the order of 400 K can be deduced. In short, surface science shows that direct CO dissociation is possible, but not on the (most abundant) closepacked surface. As direct CO dissociation is not possible on the close-packed surfaces of cobalt, one might consider if the CO molecule is (partly) hydrogenated before the C–O bond breaks. When CO and hydrogen are co-adsorbed onto a close-packed surface at low temperature, the molecules just desorb upon heating, without reaction.158,160 An alternative experimental approach is to study the decomposition of (partly) hydrogenated CO molecules such as methanol and formaldehyde. Methanol adsorbs as methoxy (H3CO) when dosed at 165 K. During heating this methoxy species is stable up to 300 K, after which it decomposes to CO and H2.165,166 Experimentally, this is seen as a single step, indicating that the first dehydrogenation is ratelimiting. The intermediate species, H2CO and HCO, decompose Oad 370 K 370 K 530 528 534 532 Binding energy (eV) 532 530 528 286 285 284 283 Binding energy 287 286 285 284 283 282 Binding energy (eV) Figure 24 Ethanol decomposition on a close-packed cobalt surface, followed with temperature programmed (TP) synchrotron x-ray photoelectron spectroscopy (XPS). At low temperature, the ethoxy intermediate is found. During heating ethoxy decomposes around 350 K, yielding atomic oxygen and C2Hx (acetylene) on the surface, showing clear evidence for C–O bond cleavage. Adapted from Weststrate, C. J.; Gericke, H. J.; Verhoeven, M.; Ciobica, I. M.; Saib, A. M.; Niemantsverdriet, J. W. J. Phys. Chem. Lett. 2010, 1, 1767–1770. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 7.20.4.2.3 Hydrogenation and the stability of C1Hx species C1Hx species are important ingredients in a carbide mechanism as they are the monomeric species responsible for chain initiation and growth. Surface science experiments can give information about the stability of the different C1Hx species. Regarding cobalt there is only one article that addresses this question directly, using CHxCly intermediates on a cobalt foil.167 The authors mention that the behavior on nickel foil was essentially the same. In an experiment on a close-packed nickel surface, a molecular beam was used to dissociate methane at low surface temperature, which generates CH3 species (þH) on the surface.168 Heating of this CH3 layer showed transformation of CH3 to CH around 200 K, and no sign of CH2 was found. On close-packed Pt reported in the same study, a very similar trend was seen: CH3 decomposes around 250 K, yielding solely CH, which decomposes around 500 K.168 These surface science results demonstrate that the CH2 species, which is typically seen as the monomer for chain growth, is particularly unstable, which means that its concentration under equilibrium concentrations will be much lower than that of CH3 and CH. Generally speaking, surface science studies on cobalt and iron are scarce in comparison to those on nickel and the more noble metals. Studies on iron foils and single crystals,169–176 although very interesting from the point of view of surface chemistry, are even further removed from the reality of Fischer–Tropsch reactions than cobalt, as iron FTS catalysts are essentially carbides. Surface science studies on iron carbides are, to the best of our knowledge, not available. 7.20.4.3 DFT Modeling Molecular modeling by DFT offers a relatively new way to understand reaction mechanisms at the molecular level.177,178 Adsorption configurations along with their energetics as well as transition states can be modeled, and thus adsorption energies, activation energies, and heats of reactions can be obtained. In addition, entropy changes over elementary reactions enable one 549 to estimate pre-exponential factors, although these are generally less often calculated than enthalpies. Validation of calculations has to be sought by comparing calculated adsorption energies and activation barriers or vibrational frequencies from stable adsorption states with experimental values, obtained from surface science experiments with single crystals. However, relatively little experimental data are available, partly also because surface science studies are by necessity often performed in vacuum, whereas FTS reaction steps occur at higher pressures. As this is an area of research that is emerging rapidly, we intend to present some examples of how DFT modeling is used in mechanistic studies. Although DFT results refer to temperatures of zero K and pressure, the results give valuable insight into the energetic of the underlying surface chemistry. It is not our intention to give a full review here, as it is still too early for conclusive statements on the FTS reaction mechanism. 7.20.4.3.1 CO Dissociation Conversion of CO and H2 into CxHy þ H2O necessarily implies that the C–O bond has to broken. The question is now if this happens before or after reaction with H-atoms. As an example, we show a computational study of direct CO dissociation on the square (100) surface of bcc-iron in Figure 25.179,180 On this surface, the CO molecule is known to adsorb in a tilted geometry.171 In the transition state for dissociation, the C–O bond elongates, and the energy rises by 1.14 eV ( 109 kJ mol1), which is the activation energy for dissociation. This value compares well with the experimentally measured activation energy of 110 kJ mol1 reported by Bernasek and coworkers.171 When the bond breaks, the C and O atoms each end up in a fourfold hollow site of the Fe(100) surface. However, the two atoms significantly repel each other, implying that the total energy decreases substantially when the two atoms move apart, as shown in the last structure of Figure 25. This series of calculations demonstrates that the direct dissociation of a CO molecule on this (100) surface is very well possible under reaction conditions (a barrier of 110 kJ mol1 corresponds roughly to a reaction temperature of Transition state Adsorbed CO Dissociated CO 1.14 eV (exp 110 kJ mol-1) Dissociated CO repulsion relieved 2.30 eV -0.34 eV 0.82 eV -1.16 eV Extent of reaction Figure 25 Energy diagram for the dissociation of carbon monoxide on the (100) surface of iron, showing the exothermicity of the dissociation, and the effect of repulsion between carbon and oxygen atoms that are adsorbed onto adjacent sites. Energies are given in electron volt (eV); 1 eV 96.5 kJ mol1). Adapted from Bromfield, T. C.; Ferre, D. C.; Niemantsverdriet, J. W. ChemPhysChem 2005, 6, 254–260. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 550 Fischer–Tropsch Synthesis: Catalysts and Chemistry 400–450 K, whereas the FTS reaction temperature is at least 475 K), provided the space is available for the dissociation products to move apart. Hence, this elementary step needs a socalled ‘ensemble’ of iron atoms, which is determined by the condition that the C and the O atom can end up on next nearest neighbor sites, that is, at distances at least equal to √2 times the lattice constant. Table 8 compares the activation energy for dissociation on a number of iron surfaces with increasing reactivity.181,182 The trend is that dissociation becomes easier when the surface becomes more reactive. Note that the (110) surface is unlikely to play a role in CO activation, as the barrier is forbiddingly high.181–183 Several authors have proposed that CO dissociation becomes easier when the CO first reacts with hydrogen.183,185,186 The reasoning is evident, as the C–O bond in an HC–O or C–OH fragment is expected to be weaker than in the CO molecule. However, forming the HCO or COH fragment also costs energy.186 It appears that on surfaces of low reactivity, such as Co(0001), Fe(110), or the (100) surface of the Hägg carbide, Hassisted dissociation indeed leads to a lower activation barrier than direct CO dissociation does.183,185 On more reactive surfaces, and notably on surfaces containing steps, the direct dissociation is favored, and often with a lower barrier than for the reaction H þ CO.186 Nevertheless, it is good to realize that on certain surfaces the H-assisted pathway may be an alternative that gives less reactive facets the chance to play a role in FTS. 7.20.4.3.2 C þ H reactions Many studies have addressed the formation of CHx fragments all the way to methane on several surfaces. Particularly interesting is the comparison made by the Nørskov group.187 Their calculations are based on a full set of DFT calculations on the fcc (211) step of ruthenium surfaces, from which they estimated the adsorbate energies on several other transition and group IB surfaces. According to the Sabatier Principle,188 the optimum pathway for a reaction is that in which the intermediates adsorb at the catalyst surface in a moderate way, that is, not too strongly and not too weakly. The set of profiles in Figure 26 illustrates that metals such as Ni, Rh, and Co are close to ideal methanation catalysts, but that (the stepped surfaces of) metals such as Ru, Fe, and W bind the intermediate species too strongly. On the other extreme are Au and Ag, where formation of intermediates is strongly endothermic and therefore unfeasible. Metals such as Cu, Pt, and Pd are Table 8 Activation energies for direct CO dissociation Iron surface Characteristics CO dissociation activation energy Reference (110) (100) Flat, close packed, least reactive surface Flat, somewhat more open, and thus more reactive than (110) Stepped surface with narrow terraces Stepped surface with broader terraces 149 kJ mol1 103–110 kJ mol1 Sorescu181 Bromfield et al.,179 Scheijen et al.,184 and Sorescu181 70–87 kJ mol1 64–86 kJ mol1 Sorescu181 Sorescu181 (310) (710) Ag 5 Au Energy (eV) Cu 0 CO + 3H2 Pd, Pt CH4 + H2O Ni, Rh Co -5 Ru Fe 4 (g) CH 4 (g) g) + 2 O( H OH *+ H* +C CH H 4 (g) 3* H* + +2 O* +3 H* +C +C O* +4 H* *+ O* 5H H H CH 2* * C* O* + 6H *+ O* *+ +C 2 (g) 3H O* + +C 2 (g) 3H 3H 2 (g) +C O(g ) O* W Figure 26 Density functional calculation for the energy profile of the methanation reaction on the group VIII and I-B metals, showing that nickel, rhodium, and cobalt fall close to the optimum profile for the reaction, while iron and tungsten form too strong bonds with the carbon, oxygen, and the CHx intermediates. Adapted from Jones, G.; Bligaard, T.; Abild-Pedersen, F.; Norskov, J. K. J. Phys. Condens. Matter 2008, 20. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 551 Fischer–Tropsch Synthesis: Catalysts and Chemistry incapable of breaking the CO bond, but can carry out CO hydrogenation to methanol. Not shown in the figure are the barriers for the individual steps. As shown by several authors, activation energies for hydrogenation of adsorbed CHx (x ¼ 0–3) species on uncorrugated (atomically flat) surfaces are in the range of 50– 100 kJ mol1 and therefore not expected to be problematic for the reaction mechanism.189–194 7.20.4.3.3 Chain growth Next comes the question how the hydrocarbon chains grow. This has been and still is a matter of scientific debate. The conventional view is that chains grow by a reaction between an alkyl and a CH2 species, for example, CH3 þ CH2 to C2H5, and C2H5 þ CH2 to C3H7. Termination to an alkane would then occur by hydrogenation of the alkyl, and olefins would form by b-H abstraction from the alkyl fragment. This view originates from the work of Biloen et al., who, however, proposed this mechanism with some reservation, and with a judicious discussion of the assumptions involved.195,196 Ciobica et al. found that reactions between CxHy fragments that contain less hydrogen are energetically more favorable, making chain growth via reactions of the type CH2 þ CH and CH¼CH2 þ CH more likely.197 Nevertheless, the class of mechanisms, based on CO dissociation, formation of CHx species, and incorporation of CHx in a growing chain, albeit with a rich variety in the details, comes close to the original proposal of Fischer and Tropsch, entitled the carbide mechanism (see Table 7). Another class of mechanisms considers CO insertion as the step leading to chain growth (see Table 7). This line of thought also has a long history, the archetype being the Pichler–Schulz mechanism.155 The most detailed mechanism has been presented by Saeys and coworkers,198 who proposed that CO inserts in an adsorbed CH2, which then further hydrogenates to a fragment in which the C–O bond breaks, leading to a C2 intermediate in which the next CO inserts. As discussed in the section on surface science, experimental proof exists that C–O bond breaking in adsorbed species derived from ethanol, such as ethoxy, is a facile step, even on the least reactive surface of cobalt, that is, the close-packed (0001) surface.166 At the time of writing, medio 2011, the authors believed that DFT is a highly valuable tool for getting insight into reaction mechanisms at the level of elementary steps. It is, however, much too early to draw definitive conclusions on which mechanism is prevalent in the FTS. Further, it should be acknowledged that not all catalysts and conditions can be captured under one dominating mechanism, and it is even not at all certain that this would be the case on one catalyst. For example, the initiation by CO dissociation might simultaneously occur via direct dissociation on highly reactive parts of a cobalt particle, and by H-assistance on facets of moderate reactivity. We also point to the differences between cobalt and iron. Whereas the former is believed to operate as a metal, the latter is active as a carbide, in which the intrinsically high reactivity of the iron atoms is considerably decreased by carbon neighbors. The DFT literature has suggested mechanisms varying from Mars-van Krevelen-type reactions199 (in which carbon atoms from the lattice become the CHx species for initiation and chain growth) to H-assisted CO dissociation and CO insertion as the major ingredients for hydrocarbon formation.200 It is obvious that mechanistic understanding would greatly benefit from more surface science work, but unfortunately the possibilities for this branch of physical chemistry research are limited for reactions that by necessity require high pressure conditions. 7.20.4.4 Macrokinetic Observations and Models In order to derive a macrokinetic model, a scheme of elementary reaction steps is usually assumed to represent the mechanistic pathway of the reaction. Normally, a number of simplifying assumptions are made regarding the catalyst surface and the adsorption of species on it, for example, that only one type of adsorption site is considered, and that these sites are homogeneously distributed over the catalyst surface, that species only adsorb in a monolayer, and that adsorbed species do not interact apart from the involved chemical reactions. It is usually further assumed that most reaction steps are sufficiently fast to reach equilibrium, but that one or more ratedetermining steps exist that are relatively slow and control the overall rate of reaction.188 These assumptions allow for the derivation of simple, manageable kinetic expressions, such as those presented in Table 9. 7.20.4.4.1 General observations regarding kinetics Macrokinetic models are mainly developed for use in process modeling, yet their mechanistic importance stems from the fact that they capture the overall behavior of the FT synthesis. Considering eqn [3] from Table 9 as an example, it is seen that the reaction rate is predicted to increase with the square root of the hydrogen partial pressure (everything else being constant). Therefore, it is said that the reaction order of hydrogen is 0.5 in this kinetic model. The variation in reaction rate with CO partial pressure is more complex, since the overall reaction order is not constant. At very low CO partial pressures or at high reaction temperature, KCOPCO is much smaller than 1 and the denominator term can be ignored, yielding an overall CO reaction order of 1. In the other extreme, at very high CO partial pressures or at low temperatures where KCOPCO is much larger than 1, the constant term in the denominator can be ignored and the overall CO reaction order strives to 1. Table 9 Kinetic expressions for the rate of the Fischer–Tropsch synthesis, rFT Equation number Rate expression 1 rFT ¼ A PCO þK2H 2 rFT ¼ A PCO þK2H PH PCO P 2 O H2 O PH0:5 PCO P 2 O H2 O PH0:5 PCO 3 rFT ¼ A 4 rFT ¼ A 5 rFT ¼ A 2 ð1þKCO PCO Þ2 PH2 PCO ð1þKCO PCO Þ2 0:5 PH0:75 PCO 2 2 0:5 P0:25 þK P0:5 P0:25 1þKC=OH PCO O CO H H 2 6 rFT ¼ A 2 0:5 PH0:75 PCO 2 0:5 ð1þKCO PCO Þ 2 K stands for equilibrium constant, P for partial pressure, and A is an effective rate constant. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 552 Fischer–Tropsch Synthesis: Catalysts and Chemistry Similarly, it can be said for eqn [1] that the reaction order in hydrogen is constant at 1, while the overall reaction order in CO varies from 0 to 1 as the conversion increases from low to high values. There are distinct differences between the chemical reaction kinetics over iron and cobalt FTS catalysts. The reaction rate increases roughly linearly with pressure over iron catalysts at constant temperature and H2/CO ratio,201 but only to an order of around 0.5 for cobalt catalysts.202 Furthermore, CO has a strong inhibiting influence over cobalt catalysts to the extent that it has a significantly negative reaction order under commercially relevant reaction conditions. To the contrary, CO has a positive effect on the rate over iron catalysts in the normal operating regime and it has been estimated that CO will only start to negatively influence the kinetics above a partial pressure of around 11 bar.203 The influence of water on the reaction kinetics has proven to be a highly controversial topic. Historically, it was firmly believed that water (and possibly CO2) inhibited the reaction rate over iron catalysts via competitive adsorption, but more recently it has been shown that there is no convincing evidence for this notion.203 The results of water co-feeding studies over cobalt catalysts have been inconsistent, since some have reported a positive and some a negative influence of water on the reaction rate, while others have found no effect at all. At least for alumina-supported cobalt catalysts, it appears as though water in the range of 1–6 bar has no significant influence on the overall rate of CO conversion, but that it does decrease the methane selectivity.204 7.20.4.4.2 Simple macrokinetic models For the derivation of macrokinetic models, a key question is whether or not there is a rate-determining step involved in the formation of the monomer of chain growth. If monomer formation is relatively facile, then the rate of CO conversion and the product distribution obtained are intimately linked and must be modeled together. This typically yields a complex, implicit type model that requires an advanced numerical routine to solve. However, in order to keep these models manageable, questionable simplifying assumptions are often made, for example, the assumption by Yang et al.205 that the monomer is in thermodynamic equilibrium with the gas phase concentrations of CO, H2, and water. Furthermore, such models require a large number of parameters that are inevitably highly crosscorrelated, implying that it is virtually impossible to accurately estimate their values. If, however, there is a rate-determining step in the formation of the monomer, the overall reaction rate can be decoupled from the product distribution. In such a case, the overall rate of CO conversion is determined by the CO hydrogenation reaction (monomer formation), while the product distribution is determined by the polymerization part of the FT reaction (i.e., the competition between chain growth and desorption). The steady state isotopic transient kinetic analysis study of van Dijk206 has indeed provided microkinetic support for the notion that there is a ratedetermining step in the formation of the monomer. The approach followed during FTS kinetic studies by Botes et al.202,207 has been to consider different reaction schemes and rate-determining steps in the formation of the monomer to obtain a variety of explicit rate expressions. A systematic experimental approach was then followed to eliminate nonapplicable models until ultimately a robust kinetic expression remained as the preferred rate equation. A further notable feature of the FTS kinetic studies was to operate the reaction at a baseline condition, where the catalyst is known to be quite stable, for most of the run. Changes to other conditions were only made for short intervals sufficient to allow for hydrodynamic steady state inside the reactor, but insufficient to effect changes in the intrinsic catalyst behavior. The importance of this approach is related to the fact that FT catalysts, especially those based on iron, readily respond to changes in operating conditions. It is imperative to avoid reversible and irreversible changes in the catalyst when chemical reaction kinetics is studied. Originally within Sasol, the rate equation [1] in Table 9, by Anderson201 was used to describe the FTS over iron. Following a systematic in-house study focusing on the reaction order of hydrogen, its exponent was later reduced to a value of 0.5, yielding eqn [2] (Table 9). The most recent study on iron-FT kinetics has considered the implications of the historic rate equations, specifically the single order denominator which implies that hydrogen reacts directly from the gas phase. By applying a second order denominator to obtain a more appropriate Langmuir–Hinshelwood–Hougen–Watson-type equation where both CO and H2 first absorb onto the surface before reaction, and also including a constant term in the denominator to provide for the possibility of vacant sites, it could in fact be shown that there is no statistical justification for including a water term in the kinetic model.207 Experiments were designed to conclusively show that eqn [3] (Table 9) is more accurate than the foregoing expressions. This study highlighted the fact that the historic perception regarding the effect of water on iron-FT kinetics was self-specified by the old models, but never tested. Furthermore, the CO order of unity is consistent with a mechanism where CO interacts with a hydrogen atom before being dissociated. The most recent cobalt kinetic study involved the derivation of several rate equations to cover various reaction schemes of CO hydrogenation.202 Models assuming hydrogen-assisted CO dissociation, such as eqn [4] that was originally proposed by Yates and Satterfield,208 generally described the measured data poorly and could be eliminated early on in the study. Figure 27 illustrates that eqn [4] underestimates the reaction rate at low CO partial pressure, but overestimates it at higher CO partial pressures. Ultimately, after further work to distinguish between those models where CO first dissociates before it is hydrogenated, eqn [5] (Table 9) was the only rate expression that could not be eliminated. Therefore, it was selected as the most appropriate kinetic model. It should be noted that this expression can be very closely approximated by eqn [6], which contains one model parameter less and is thus preferably used from a practical perspective. Figure 27 shows that the preferred equation [6] does not suffer from the same systematic errors as eqn [4], since it is reasonably accurate across a range of CO partial pressures. 7.20.4.4.3 Selectivity modeling The vast number of components in the FT product slate does not allow for the prediction of individual product selectivities; consequently, the product spectrum is rather represented by a product characterization model with a limited number of Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 553 1.5 Equation 4 rFT (measured) / rFT (predicted) rFT = A PH2PCO (1 + KCOPCO)2 1 Equation 6 rFT = A 0.5 0 1 2 3 4 5 CO partial pressure (bar) 6 P 0H.75P 0.5 CO 2 (1 + KCOP 0.5 )2 CO 7 8 Figure 27 Performance of rival kinetic expressions as a function of CO partial pressure for a data series where the CO flow rate into a slurry reactor was varied while that of H2 was maintained constant. parameters. After correlating these parameters with process conditions, an explicit selectivity model can be obtained. The simplest model for describing selectivity is the Anderson– Schulz–Flory distribution, with the chain growth probability (a-value) as the only parameter. Approaches have been proposed to account for deviations from the ideal distribution. The double-a model by Donnelly et al.209 assumes that two types of catalytic sites or two types of mechanisms simultaneously form the observed product spectrum. However, the three model parameters have a high degree of covariance when estimated from experimental data. Furthermore, neither the C1 and C2 selectivities, nor the olefin content of the product spectrum can be predicted. Some of these limitations have been addressed by considering a chain length-dependent desorption model, which assumes that termination by desorption becomes increasingly more difficult as the chain length increases.17 It has also been reported that the chain length effects in the FT product spectrum, in particular the positive deviation of methane and the negative deviation of ethylene, can be explained by symmetry effects as accounted for by the single event kinetic theory.210 Some have ascribed the chain length-dependent deviations to secondary olefin reactions, but many concerns remain over this approach.203 These include the observation that secondary olefin reactions are much less facile (almost negligible) over iron catalysts compared to cobalt catalysts, yet the bend in the Anderson–Schulz–Flory graph is much more pronounced with iron catalysts. 7.20.4.5 Mechanistic and Kinetic Implications Despite the large number of kinetic and mechanistic studies on FTS, there is still substantial uncertainty regarding the most relevant steps in the reaction pathway(s). A variety of elementary reaction steps have been proven to be realistically possible under typical FTS conditions, while very few steps could be eliminated with certainty. For example, and as described before, DFT calculations have shown that the coupling reactions of CHx fractions are reasonably facile, which lends support to the steps of chain growth as proposed by the carbide mechanism. However, DFT calculations have also shown that each of the steps required for the CO insertion mechanism is energetically feasible, while the surface science approach has demonstrated experimentally that the scission of the C–O bond of a CO molecule into which an alkyl group has been inserted is a facile reaction, even on a low reactivity cobalt surface.166 The current inability to clearly discriminate between rival mechanisms partly stems from the overlap between the proposed reaction pathways. For instance, the two most popular mechanisms are not mutually exclusive with respect to each other. The carbide mechanism has to assume direct CO insertion as a termination step at least in order to explain the observed formation of oxygenates in the FTS. On the other hand, the initiation step in the CO insertion mechanism is similar as for the carbide mechanism. The most plausible conclusion currently is therefore that a variety of reaction pathways simultaneously contribute to the overall synthesis. The question still remains though whether one pathway is dominant over the rest and individually determines the bulk of the observed behavior of the system, or whether two or more parallel pathways have similar contributions to the overall kinetics. The answer to this question may not even be absolute, as it may depend on what aspect of the reaction is of interest. Hypothetically speaking, if the carbide mechanism predominates in the formation of olefins and paraffins (the main products of the synthesis), it may well be accurate to describe the overall rate of syngas conversion by only considering this mechanism. However, even in such an event, CO insertion cannot be ignored if the object is to model oxygenate formation. Despite all the foregoing uncertainty, some consistencies have also emerged from the studies performed in the various Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 554 Fischer–Tropsch Synthesis: Catalysts and Chemistry disciplines. For example, the results of DFT calculations suggest that unassisted CO dissociation readily occurs on the more open (high reactivity) metal surfaces, while hydrogen-assisted CO dissociation would be required on the close packed (low reactivity) surfaces.179,181,183,185,186 As described in the section on surface science above, unassisted CO dissociation is not easy on close packed surfaces, while it is facile over open surfaces. It has further been found that HxCO species are more inclined to dehydrogenate than to undergo C–O bond cleavage on cobalt surfaces. During a macrokinetic study on an actual cobalt-FT catalyst, models assuming hydrogen-assisted CO dissociation failed comprehensively, while the preferred model based on unassisted CO dissociation could describe the experimental data over a range of commercially relevant conditions. Together all these findings suggest that, in the case of the cobalt-based FT synthesis, the main pathway in the conversion of CO to a CHx species proceeds via unassisted cleavage of the C–O bond. To the contrary, it is known that the carbiding of iron catalysts substantially decreases the reactivity of iron surfaces. Therefore, one may well expect hydrogen-assisted CO dissociation to predominate over iron-FT catalysts,199 which are in the carbided state under actual synthesis conditions. The most preferred macrokinetic model for iron is indeed consistent with a CO dissociation step that occurs via interaction with hydrogen. A further consistency that is steadily emerging relates to the most likely nature of the species responsible for chain propagation in terms of the carbide mechanism. Originally, it was believed (not necessarily based on strong evidence) that these species are quite saturated with hydrogen, that is, that the growing intermediate is a CH3–CH2 CH2 species, while the monomer being added is a CH2 species.195 DFT calculations have shown that reactions between intermediates that are leaner in hydrogen (e.g., CH¼CH2 and CH) are energetically more favorable than reactions between more hydrogen-saturated species.197 In line with this, it has been found during surface science experiments the coupling of two CH species to from acetylene is facile over nickel catalysts.168,211Further support is provided by the steady state isotopic transient kinetic analysis (SSITKA) study of Govender,212 who concluded from H–D switching experiments that the C2H species is the only abundant C2 intermediate on the fully carbided, working iron-FT catalyst surface. Therefore, even though the carbide mechanism as a whole cannot be discarded as a prominent reaction pathway for the FTS, it seems unlikely that it proceeds in the form that was originally proposed. This has particular significance for the termination toward olefins, since it implies that a hydrogen abstraction is not required (possibly even a hydrogen addition). 7.20.5 Conclusion The FTS represents proven technology, which has secured its position in modern energy technology. Originally used to convert coal into liquid fuels, nowadays the emphasis is on monetizing natural gas, by converting it to diesel fuel, waxes, and naphtha. It is expected that some 500 000 barrels of fuel per day will be produced using Fischer–Tropsch technology by 2013. Although small in comparison to the 85 million barrels of crude oil that are produced daily, the 0.5 million daily barrels of synfuels is undoubtedly significant, particularly locally where the production takes place. The technology has much potential for wider use, for example, in emerging economies, or at a smaller scale in the utilization of biomass. Interesting applications of FTS have been proposed for conversion of remote natural gas at off-shore oil production locations. Both GTL and BTL can be important tools in strategies aimed at reduction of CO2 emissions. CTL technology is clearly disadvantaged here, and will in the future have to be combined with CO2 sequestration technology. The rapid increase in discoveries of shale gas (in, for example, USA and Canada) can also provide the GTL industry with a significant boost. Although proven technology, FTS continues to pose challenges from an industrial perspective. Stability improvement of the catalysts is an important aspect, but also selectivity improvement would be very advantageous. Economically one would like to have the highest possible Cþ 5 and the lowest possible CH4 selectivity, because recycling of CH4 means that the carbon atoms involved have to go through the expensive syngas generation more than once, with the associated efficiency losses. Syngas production is the most expensive part of a GTL plant; it accounts for 40–60% of the capital investments. Increased research efforts on reducing the costs of syngas production will make XTL projects even more viable. Although new XTL facilities require large capital investments and are dependent on the price ratio of crude oil to natural gas, in the long term they are expected to be economically successful. From a more academic perspective, understanding the mechanism of the FTS has been and will be a challenge. It is more and more realized that mechanisms may differ with conditions and catalysts. It is highly unlikely that one unique mechanism can account for all different forms of FTS. Molecular modeling represents a very important tool for getting mechanistic insight, but the problem is that experimental validation of its predictions at the level of elementary steps is very difficult to achieve, as the opportunities for relevant surface science experiments are limited. Mechanistic studies aimed at describing FTS selectivity from first principles are in their infancy and have a long way to go before accurate predictions can be expected. Describing the physical/chemical state of the catalysts under reaction conditions is another field where significant progress has been booked, but major advances would still be very welcome. The advent of in situ imaging tools in combination with realistic catalysts,213 as well as the use of planar model catalysts in simulated environments,153 has proven promising and will almost certainly lead to improved insight in the relation between catalyst properties on the nanoscale and performance in the reaction. The FTS is therefore expected to remain an inspiring source of industrial and academic research for many years to come. For a related chapter in this Comprehensive, we refer to Chapter 7.01 References 1. Fischer, F.; Tropsch, H. Brennst. Chem. 1923, 4, 276–285. 2. Fischer, F.; Tropsch, H. Brennst. Chem. 1926, 7, 97–104. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 3. Storch, H.; Golumbic, N.; Anderson, R. B. The Fischer–Tropsch and Related Syntheses. Wiley: New York, 1951. 4. van der Laan, G. P.; Beenackers, A. A. C. M. Catal. Rev. 1999, 41, 255–318. 5. Kolbel, H.; Ralek, M. Catal. Rev. 1980, 21, 225–274. 6. Dry, M. E.; Hoogendoorn, J. C. Catal. Rev. 1981, 23, 265–278. 7. Roferdepoorter, C. K. Chem. Rev. 1981, 81, 447–474. 8. Iglesia, E., Reyes, S. C., Madon, R. J., Soled, S. L., Eds.; Advances in Catalysis; 1993; Vol. 39, pp 221–302. 9. Steynberg, A. P.; Dry, M. E. Fischer–Tropsch Technology; Elsevier: Amsterdam, 2004; Vol. 152. 10. Rostrup-Nielsen, J. R., Sehested, J., Norskov, J. K., Eds.; Advances in Catalysis; 2002; Vol. 47, pp 65–139. 11. Dancuart, L. P.; de Haan, R.; de Klerk, A. In Studies in Surface Science and Catalysis; Steynberg, A. P., Dry, M. E., Eds.; 152, Elsevier, 2004; pp 482–532. 12. De Klerk, A. Fischer–Tropsch Refining. Wiley-VCH: Weinheim, 2011. 13. Claeys, M.; van Steen, E. Fischer–Tropsch Technology; Elsevier: Amsterdam, 2004; Vol. 152. 14. Schulz, H.; Vansteen, E.; Claeys, M. In Natural Gas Conversion II 1994; Vol. 81, pp 455–460. 15. Claeys, M.; Van Steen, E. In Fischer–Tropsch Technology; Studies in Surface Science: An Catalysis; Steynberg, A. P., Dry, M. E., Eds.; Elsevier: Amsterdam, 2004; Vol. 152, Chapter 8. 16. Botes, F. G.; Govender, N. S. Energy Fuel 2007, 21, 3095–3101. 17. Botes, F. G. Energy Fuel 2007, 21, 1379–1389. 18. Davis, B. H. Ind. Eng. Chem. Res. 2007, 46, 8938–8945. 19. Khodakov, A. Y.; Chu, W.; Fongarland, P. Chem. Rev. 2007, 107, 1692–1744. 20. Dry, M. E. Catal. Lett. 1991, 7, 241–251. 21. de Smit, E.; Weckhuysen, B. M. Chem. Soc. Rev. 2008, 37, 2758–2781. 22. Steynberg, A. P.; Dry, M. E.; Davis, M. E.; Davis, B. H.; Breman, B. B. Stud. Surf. Sci. Catal. 2004, 152, 64–195. 23. Van Berge, P. J.; Van De Loosdrecht, J.; Caricato, E. A.; Barradas, S. Process for Producing Hydrocarbons from a Synthesis Gas, and Catalysts Therefore. Patent WO 9942214A1, 1999. 24. Davis, B. H. Catal. Today 2002, 71, 249–300. 25. Steynberg, A. P.; Espinoza, R. L.; Jager, B.; Vosloo, A. C. Appl. Catal. A: Gen. 1999, 186, 41–54. 26. Espinoza, R. L.; Steynberg, A. P.; Jager, B.; Vosloo, A. C. Appl. Catal. A: Gen. 1999, 186, 13–26. 27. Geerlings, J. J. C.; Wilson, J. H.; Kramer, G. J.; Kuipers, H.; Hoek, A.; Huisman, H. M. Appl. Catal. A: Gen. 1999, 186, 27–40. 28. Deshmukh, S. R.; Tonkovich, A. L. Y.; McDaniel, J. S.; Schrader, L. D.; Burton, C. D.; Jarosch, K. T.; Simpson, A. M.; Kilanowski, D. R.; LeViness, S. Biofuels 2011, 2, 315–324. 29. Visconti, C. G.; Tronconi, E.; Lietti, L.; Groppi, G.; Forzatti, P.; Cristiani, C.; Zennaro, R.; Rossini, S. Appl. Catal. A: Gen. 2009, 370, 93–101. 30. Sabatier, P.; Senderens, J. B. Hebd. Seances Acad. Sci. 1902, 134, 514. 31. Fischer, F.; Tropsch, H. Patent DE 484 337, 1925. 32. Liu, Z.; Shi, S.; Li, Y. Chem. Eng. Sci. 2010, 65, 12–17. 33. Dry, M. E. Catal. Today 2002, 71, 227–241. 34. Sie, S. T. Rev. Chem. Eng. 1998, 14, 109–157. 35. Oukaci, R.; Singleton, A. H.; Goodwin, J. G. Appl. Catal. A: Gen. 1999, 186, 129–144. 36. Sasol Financial Report July–December 2010. http://www.sasol.com. 37. Brown, A. Pearl GTL Presentation, XTL Summit, June 2011. 38. Steynberg, A. P. In Studies in Surface Science and Catalysis; Steynberg, A. P., Dry, M. E., Eds.; 152, Elsevier: Amsterdam, 2004; pp 1–63. 39. Penning, R. In New Developments in Synthetic Fuels, CTL/GTL Conference 2010, Brisbane, Australia. 40. Davis, B. H. Catal. Today 2003, 84, 83–98. 41. Schwertmann, U.; Cornell, R. M. Iron Oxides in the Laboratory, Preparation and Characterization. Wiley-VCH: Weinheim, 2000. 42. Smith, D. F.; Hawk, C. O.; Golden, P. L. J. Am. Chem. Soc. 1930, 52, 3221–3232. 43. Bromfield, T. C.; Botes, F. G.; Visagie, R.; Espinoza, R.; Gibson, P.; Van Lawson, K. H. Hydrocarbon Synthesis Catalyst and Process. US Patent 6844370, 2002. 44. Hayakawa, H.; Tanaka, H.; Fujimoto, K. Appl. Catal. A: Gen. 2006, 310, 24–30. 45. O’Brien, R. J.; Xu, L. G.; Spicer, R. L.; Bao, S. Q.; Milburn, D. R.; Davis, B. H. Catal. Today 1997, 36, 325–334. 46. Luo, M. S.; O’Brien, R. J.; Bao, S. Q.; Davis, B. H. Appl. Catal. A: Gen. 2003, 239, 111–120. 47. Niemantsverdriet, J. W. Spectroscopy in Catalysis; An Introduction, 3rd ed.; Wiley-VCH: Weinheim, 2007. 555 48. Kock, A. J. H. M.; Geus, J. W. Prog. Surf. Sci. 1985, 20, 165–272. 49. de Smit, E.; Cinquini, F.; Beale, A. M.; Safonova, O. V.; van Beek, W.; Sautet, P.; Weckhuysen, B. M. J. Am. Chem. Soc. 2010, 132, 14928–14941. 50. Raupp, G. B.; Delgass, W. N. J. Catal. 1979, 58, 348–360. 51. Niemantsverdriet, J. W.; Van der Kraan, A. M.; Van Dijk, W. L.; Van der Baan, H. S. J. Phys. Chem. 1980, 84, 3363–3370. 52. Dry, M. E. In Studies in Surface Science and Catalysis; Steynberg, A. P., Dry, M. E., Eds.; 152, Elsevier: Amsterdam, 2004; pp 533–600. 53. Luo, M. S.; Hamdeh, H.; Davis, B. H. Catal. Today 2009, 140, 127–134. 54. Herranz, T.; Rojas, S.; Perez-Alonso, F. J.; Ojeda, M.; Terreros, P.; Fierro, J. L. G. J. Catal. 2006, 243, 199–211. 55. Dry, M. E. In Catalysis, Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: New York, 1981; Vol. 1, pp 159–255. 56. Luo, M. S.; Davis, B. H. Fuel Process. Technol. 2003, 83, 49–65. 57. Fiato, R. A.; Soled, S. L. Fischer–Tropsch Hydrocarbon Synthesis with High Surface Area Copper and Potassium Promoted Reduced-Carbided Iron/ Manganese Spinels. US Patent 4621102A, 1986. 58. Du Toit, E. Ph.D. Thesis, University of the North West, 2002. 59. Niemantsverdriet, J. W.; van der Kraan, A. M. J. Catal. 1981, 72, 385–388. 60. Bromfield, T. C.; Visagie, R. Chromium Oxide Incorporation into Precipitated IronBased Fischer–Tropsch Catalysts for Increased Production of Oxygenates and Branched Hydrocarbons. Patent WO 2005049765A1, 2005. 61. van Steen, E.; Claeys, M. Chem. Eng. Technol. 2008, 31, 655–666. 62. Tsakoumis, N. E.; Ronning, M.; Borg, O.; Rytter, E.; Holmen, A. Catal. Today 2010, 154, 162–182. 63. Morales, F.; Weckhuysen, B. M. Catalysis 2006, 19, 1–40. 64. van de Loosdrecht, J.; Bazhinimaev, B.; Dalmon, J. A.; Niemantsverdriet, J. W.; Tsybulya, S. V.; Saib, A. M.; van Berge, P. J.; Visagie, J. L. Catal. Today 2007, 123, 293–302. 65. Saib, A. M.; Moodley, D. J.; Ciobica, I. M.; Hauman, M. M.; Sigwebela, B. H.; Weststrate, C. J.; Niemantsverdriet, J. W.; van de Loosdrecht, J. Catal. Today 2010, 154, 271–282. 66. Claeys, M.; van Steen, E. In Studies in Surface Science and Catalysis; Steynberg, A. P., Dry, M. E., Eds.; 152, Elsevier: Amsterdam, 2004; pp 601–680. 67. Iglesia, E.; Reyes, S. C.; Madon, R. J.; Soled, S. L. Adv. Catal. 1993, 39, 221–302. 68. Diehl, F.; Khodakov, A. Y. Oil Gas Sci. Technol. 2009, 64, 11–24. 69. Bezemer, G. L.; Bitter, J. H.; Kuipers, H.; Oosterbeek, H.; Holewijn, J. E.; Xu, X. D.; Kapteijn, F.; van Dillen, A. J.; de Jong, K. P. J. Am. Chem. Soc. 2006, 128, 3956–3964. 70. Jacobs, G.; Patterson, P. M.; Zhang, Y. Q.; Das, T.; Li, J. L.; Davis, B. H. Appl. Catal. A: Gen. 2002, 233, 215–226. 71. Karaca, H.; Hong, J. P.; Fongarland, P.; Roussel, P.; Griboval-Constant, A.; Lacroix, M.; Hortmann, K.; Safonova, O. V.; Khodakov, A. Y. Chem. Commun. 2010, 46, 788–790. 72. Kitakami, O.; Sato, H.; Shimada, Y.; Sato, F.; Tanaka, M. Phys. Rev. B 1997, 56, 13849–13854. 73. Enache, D. I.; Rebours, B.; Roy-Auberger, M.; Revel, R. J. Catal. 2002, 205, 346–353. 74. Iglesia, E.; Soled, S. L.; Fiato, R. A.; Via, G. H. J. Catal. 1993, 143, 345–368. 75. Feller, A.; Claeys, M.; van Steen, E. J. Catal. 1999, 185, 120–130. 76. Moradi, G. R.; Basir, M. M.; Taeb, A.; Kiennemann, A. Catal. Commun. 2003, 4, 27–32. 77. Tan, K. F.; Chang, J.; Borgna, A.; Saeys, M. J. Catal. 2011, 280, 50–59. 78. Rytter, E.; Skagseth, T. H.; Eri, S.; Sjastad, A. O. Ind. Eng. Chem. Res. 2010, 49, 4140–4148. 79. Beuther, H.; Kobylinski, T. P.; Kibby, C. L.; Pannell, R. B. Synthesis Gas Conversion Using Ruthenium-Promoted Cobalt Catalyst Prepared by Nonaqueous Impregnation. US Patent 4585798A, 1986. 80. Beuther, H.; Kibby, C. L.; Kobylinski, T. P.; Pannell, R. B. Fluid Bed Catalyst for Synthesis Gas Conversion and Its Utilization for Preparation of Diesel Fuel. US Patent 4413064A, 1983. 81. Beuther, H.; Kibby, C. L.; Kobylinski, T. P.; Pannell, R. B. Conversion of Synthesis Gas to Diesel Fuel and Gasoline. US Patent 4605680A, 1986. 82. Eri, S.; Kinnari, K. J.; Schanke, D.; Hilmen, A.-M. Preparation and Use of Promoted Cobalt Catalyst with Low Surface Area Alumina for Fischer–Tropsch Reaction with High Olefin Selectivity. Patent WO 2002047816A1, 2002. 83. Saib, A. M.; Claeys, M.; van Steen, E. Catal. Today 2002, 71, 395–402. 84. Van Berge, P. J.; Van De Loosdrecht, J.; Barradas, S. Method of Treating an Untreated Catalyst Support, and Forming a Catalyst Precursor and Catalyst from the Treated Support. Patent EP 1 303 350 B1, 2000. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy 556 Fischer–Tropsch Synthesis: Catalysts and Chemistry 85. Van Berge, P. J.; Van De Loosdrecht, J.; Barradas, S. Production of Fischer– Tropsch Synthesis Produced Wax. Patent EP 1 432 778 B1, 2001. 86. Wei, D. G.; Goodwin, J. G.; Oukaci, R.; Singleton, A. H. Appl. Catal. A: Gen. 2001, 210, 137–150. 87. Xiong, H. F.; Motchelaho, M. A. M.; Moyo, M.; Jewell, L. L.; Coville, N. J. J. Catal. 2011, 278, 26–40. 88. Stranges, A. N. Germany’s Synthetic Fuel Industry 1927–1945. Presented at the AIChE 2003, New Orleans, 2003. http://www.fischertropsch.org. 89. Keyser, M. J.; Everson, R. C.; Espinoza, R. L. Appl. Catal. A: Gen. 1998, 171, 99–107. 90. Co-Precipitated Cobalt–Zinc Catalysts for Fischer–Tropsch Reaction or Functional Group Hydrogenation. Patent EP1358934A1, 2003. 91. Puskas, I.; Fleisch, T. H.; Full, P. R.; Kaduk, J. A.; Marshall, C. L.; Meyers, B. L. Appl. Catal. A: Gen. 2006, 311, 146–154. 92. Joustra, A. H.; Scheffer, B. Process for the Preparation of Alumina-Based Extrudates. Patent EP 455307A1, 1991. 93. Neimark, A. V.; Kheifets, L. I.; Fenelonov, V. B. Ind. Eng. Chem. Prod. Res. Dev. 1981, 20, 439–450. 94. Kheifets, L. I.; Neimark, A. V.; Fenelonov, V. B. Kinet. Catal. 1979, 20, 626–632. 95. Neimark, A. V.; Fenelonov, V. B.; Heifets, L. I. React. Kinet. Catal. Lett. 1976, 5, 67–72. 96. Van Berge, P. J. In Scaling Up of an Alumina Supported Cobalt Slurry Phase Fischer–Tropsch Catalyst Preparation, CatCon – World Wide Catalyst Industry Conference, Houston, TX, USA, June 12–13, 2000. 97. Van Berge, P. J.; Van De Loosdrecht, J.; Caricato, E. A.; Barradas, S.; Sigwebela, B. H. Impregnation Process for Catalysts. Patent WO 2000020116A1, 2000. 98. Soled, S. L.; Baumgartner, J. E.; Reyes, S. C.; Iglesia, E. In Preparation of Catalysts VI: Scientific Bases for the Preparation of Heterogeneous Catalysts, Studies in Surface Science and Catalysis, 1995; Vol. 91, pp 989–997. 99. de Jong, K. P. Deposition Precipitation onto Pre-shaped Carrier Bodies, Possibilities and Limitations; Elsevier: Amsterdam, 1991; Vol. 63. 100. Boutonnet, M.; Jaras, S.; Logdberg, S., Method for Depositing Metal Particles on a Support. Patent EP 1 985 361, 2008. 101. van de Loosdrecht, J.; Barradas, S.; Caricato, E. A.; Ngwenya, N. G.; Nkwanyana, P. S.; Rawat, M. A. S.; Sigwebela, B. H.; van Berge, P. J.; Visagie, J. L. Top. Catal. 2003, 26, 121–127. 102. Wolters, M.; Munnik, P.; Bitter, J. H.; De Jongh, P. E.; De Jong, K. P. Method for Producing a Supported Metal Nitrate. Patent WO 2010109216A1, 2010. 103. Soled, S.L.; Baumgartner, J.E.; Reyes, S.C.; Iglesia, E.; Poncelet, G., J. M. B. D. P. A. J. a. P. G. In Studies in Surface Science and Catalysis; Elsevier, 1995; Vol. 91; pp 989–997. 104. Hoek, A.; Moors, J. H. Catalyst Activation and Rejuvenation Process. Patent WO 9717137A1, 1997. 105. Behrmann, W. C.; Davis, S. M.; Mauldin, C. H. Method for Preparing CobaltContaining Hydrocarbon Synthesis Catalyst. Patent WO 9206784A1, 1992. 106. Oosterbeek, H. Phys. Chem. Chem. Phys. 2007, 9, 3570–3576. 107. van Berge, P. J.; Barradas, S.; van de Loosdrecht, J.; Visagie, J. L. Erdol Erdgas Kohle 2001, 117, 138–142. 108. Iglesia, E. Appl. Catal. A 1997, 161, 59–78. 109. Barbier, A.; Tuel, A.; Arcon, I.; Kodre, A.; Martin, G. A. J. Catal. 2001, 200, 106–116. 110. Bian, G. Z.; Fujishita, N.; Mochizuki, T.; Ning, W. S.; Yamada, M. Appl. Catal. A: Gen. 2003, 252, 251–260. 111. Martinez, A.; Prieto, G. J. Catal. 2007, 245, 470–476. 112. Prieto, G.; Martinez, A.; Concepcion, P.; Moreno-Tost, R. J. Catal. 2009, 266, 129–144. 113. Borg, O.; Dietzel, P. D. C.; Spjelkavik, A. I.; Tveten, E. Z.; Walmsley, J. C.; Diplas, S.; Eri, S.; Holmen, A.; Ryttera, E. J. Catal. 2008, 259, 161–164. 114. Iglesia, E. Appl. Catal. A: Gen. 1997, 161, 59–78. 115. Martinez, A.; Rollan, J.; Arribas, M. A.; Cerqueira, H. S.; Costa, A. F.; SAguiar, E. F. J. Catal. 2007, 249, 162–173. 116. Fischer, N.; van Steen, E.; Claeys, M. Catal. Today 2011, 171, 174–179. 117. Yang, J.; Tveten, E. Z.; Chen, D.; Holmen, A. Langmuir 2010, 26, 16558–16567. 118. Saib, A. M.; Borgna, A.; de Loosdrecht, J. V.; van Berge, P. J.; Niemantsverdriet, J. W. Appl. Catal. A: Gen. 2006, 312, 12–19. 119. Reynhout, M. J. Cobalt-Based Hydrocarbon Synthesis Catalysts Prepared from Solid Solution Mixture of Metal Compound Precursors. Patent WO 2008061970A2, 2008. 120. White Paper – Fischer–Tropsch Catalyst Test on Coal Derived Synthesis Gas, Syntroleum Corporation. http://www.syntroleum.com. 121. Moodley, D. J.; van de Loosdrecht, J.; Saib, A. M.; Overett, M. J.; Datye, A. K.; Niemantsverdriet, J. W. Appl. Catal. A: Gen. 2009, 354, 102–110. 122. Moodley, D. J.; Saib, A. M.; van de Loosdrecht, J.; Welker-Nieuwoudt, C. A.; Sigwebela, B. H.; Niemantsverdriet, J. W. Catal. Today 2011, 171, 192–200. 123. Moodley, D. J.; van de Loosdrecht, J.; Saib, A. M.; Niemantsverdriet, J. W. Chem. Ind. 2010, 128, 49–81. 124. Ciobica, I. M.; van Santen, R. A.; van Berge, P. J.; van de Loosdrecht, J. Surf. Sci. 2008, 602, 17–27. 125. Soled, S. L.; Iglesia, E.; Fiato, R. A.; Baumgartner, J. E.; Vroman, H.; Miseo, S. Top. Catal. 2003, 26, 101–109. 126. Hilmen, A. M.; Schanke, D.; Holmen, A. In Natural Gas Conversion IV; dePontes, M., Espinoza, R. L., Nicolaides, C. P., Scholtz, J. H., Scurrell, M. S., Eds.; Elsevier: Amsterdam, 1997; Vol. 107, pp 237–242. 127. Rothaemel, M.; Hanssen, K. F.; Blekkan, E. A.; Schanke, D.; Holmen, A. Catal. Today 1997, 38, 79–84. 128. Craje, M. W. J.; van der Kraan, A. M.; van de Loosdrecht, J.; van Berge, P. J. Catal. Today 2002, 71, 369–379. 129. Kiss, G.; Kliewer, C. E.; DeMartin, G. J.; Culross, C. C.; Baumgartner, J. E. J. Catal. 2003, 217, 127–140. 130. Das, T. K.; Jacobs, G.; Patterson, P. M.; Conner, W. A.; Li, J. L.; Davis, B. H. Fuel 2003, 82, 805–815. 131. Li, J. L.; Jacobs, G.; Zhang, Y. Q.; Das, T.; Davis, B. H. Appl. Catal. A: Gen. 2002, 223, 195–203. 132. Storsaeter, S.; Borg, O.; Blekkan, E. A.; Holmen, A. J. Catal. 2005, 231, 405–419. 133. Huffman, G. P.; Shah, N.; Zhao, J. M.; Huggins, F. E.; Hoost, T. E.; Halvorsen, S.; Goodwin, J. G. J. Catal. 1995, 151, 17–25. 134. Bezemer, G. L.; Remans, T. J.; van Bavel, A. P.; Dugulan, A. I. J. Am. Chem. Soc. 2010, 132, 8540–8541. 135. Yan, Z.; Wang, Z. J.; Bukur, D. B.; Goodman, D. W. J. Catal. 2009, 268, 196–200. 136. den Breejen, J. P.; Sietsma, J. R. A.; Friedrich, H.; Bitter, J. H.; de Jong, K. P. J. Catal. 2010, 270, 146–152. 137. Ronning, M.; Tsakoumis, N. E.; Voronov, A.; Johnsen, R. E.; Norby, P.; van Beek, W.; Borg, O.; Rytter, E.; Holmen, A. Catal. Today 2010, 155, 289–295. 138. Li, J. L.; Zhan, X. D.; Zhang, Y. Q.; Jacobs, G.; Das, T.; Davis, B. H. Appl. Catal. A: Gen. 2002, 228, 203–212. 139. British Intelligence Objectives Sub-Committee, Interrogation of Dr Otto Roelen of Ruhrchemie A.G., B.I.O.S. Final Report No. 447; Item no 30 (1945). http://www. fischer-tropsch.org. 140. Zonnevylle, M. C.; Geerlings, J. J. C.; van Santen, R. A. Surf. Sci. 1990, 240, 253–262. 141. Tan, K. F.; Xu, J.; Chang, J.; Borgna, A.; Saeys, M. J. Catal. 2010, 274, 121–129. 142. Saeys, M.; Tan, K. F.; Chang, J.; Borgna, A. Ind. Eng. Chem. Res. 2010, 49, 11098–11100. 143. Jacobs, G.; Sarkar, A.; Ji, Y.; Luo, M.; Dozier, A.; Davis, B. H. Ind. Eng. Chem. Res. 2008, 47, 672–680. 144. Tavasoli, A.; Abbaslou, R. M. M.; Dalai, A. K. Appl. Catal. A: Gen. 2008, 346, 58–64. 145. Zhou, W.; Chen, J. G.; Fang, K. G.; Sun, Y. H. Fuel Process. Technol. 2006, 87, 609–616. 146. Bartholomew, C. H. Appl. Catal. A: Gen. 2001, 212, 17–60. 147. Madon, R. J.; Shaw, H. Catal. Rev. 1977, 15, 69–106. 148. Liu, Z. T.; Zhou, J. L.; Zhang, B. J. J. Mol. Catal. 1994, 94, 255–261. 149. Bartholomew, C. H.; Bowman, R. M. Appl. Catal. 1985, 15, 59–67. 150. Leviness, S. C.; Mart, C. J.; Behrmann, W. C.; Hsia, S. J.; Neskora, D. R. Slurry Hydrocarbon Synthesis Process with Increased Catalyst Life. Patent WO 9850487A1, 1998. 151. Wilson, J.; De Groot, C. J. Phys. Chem. 1995, 99, 7860–7866. 152. Luo, M. S.; Davis, B. H. In Catalyst Deactivation 2001, Proceedings, Studies in Surface Science and Catalysis, 2001; Vol. 139, pp 133–140. 153. Thune, P. C.; Weststrate, C. J.; Moodley, P.; Saib, A. M.; van de Loosdrecht, J.; Miller, J. T.; Niemantsverdriet, J. W. Catal. Sci. Technol. 2011, 1, 689–697. 154. Reynhout, M. J. Process for Regenerating a Cobalt Catalyst. Patent EP 1920836A1, 2008. 155. Pichler, H.; Schulz, H. Chemie Ingenieur Technik 1970, 42, 1162–1174. 156. Lahtinen, J.; Vaari, J.; Kauraala, K. Surf. Sci. 1998, 418, 502–510. 157. Beitel, G. A.; Laskov, A.; Oosterbeek, H.; Kuipers, E. W. J. Phys. Chem. 1996, 100, 12494–12502. 158. Habermehl-Cwirzen, K. M. E.; Kauraala, K.; Lahtinen, J. Phys. Scripta 2004, T108, 28–32. 159. Ernst, K. H.; Schwarz, E.; Christmann, K. J. Chem. Phys. 1994, 101, 5388–5401. 160. Bridge, M. E.; Comrie, C. M.; Lambert, R. M. J. Catal. 1979, 58, 28–33. 161. Prior, K. A.; Schwaha, K.; Lambert, R. M. Surf. Sci. 1978, 77, 193–208. 162. Geerlings, J. J. C.; Zonnevylle, M. C.; Degroot, C. P. M. Surf. Sci. 1991, 241, 315–324. 163. Papp, H. Surf. Sci. 1985, 149, 460–470. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 Author's personal copy Fischer–Tropsch Synthesis: Catalysts and Chemistry 164. Nowitzki, T.; Borchert, H.; Jurgens, B.; Risse, T.; Zielasek, V.; Baumer, M. ChemPhysChem 2008, 9, 729–739. 165. Habermehl-Cwirzen, K.; Lahtinen, J.; Hautojarvi, P. Surf. Sci. 2005, 598, 128–135. 166. Weststrate, C. J.; Gericke, H. J.; Verhoeven, M.; Ciobica, I. M.; Saib, A. M.; Niemantsverdriet, J. W. J. Phys. Chem. Lett. 2010, 1, 1767–1770. 167. Steinbach, F.; Kiss, J.; Krall, R. Surf. Sci. 1985, 157, 401–412. 168. Denecke, R. Appl. Phys. Mater. Sci. Process. 2005, 80, 977–986. 169. Krebs, H. J.; Bonzel, H. P.; Gafner, G. Surf. Sci. 1979, 88, 269–283. 170. Dwyer, D. J.; Gland, J.; Albert, M.; Bernasek, S., Intermediates to the Dissociative Chemisorption of Co and Ch3oh on Fe(100). Abstracts of Papers of the American Chemical Society 1987, 193, 30. 171. Moon, D. W.; Cameron, S.; Zaera, F.; Eberhardt, W.; Carr, R.; Bernasek, S. L.; Gland, J. L.; Dwyer, D. J. Surf. Sci. 1987, 180, L123–L128. 172. Dwyer, D. J.; Somorjai, G. A. J. Catal. 1978, 52, 291–301. 173. Dwyer, D. J.; Hardenbergh, J. H. J. Catal. 1984, 87, 66–76. 174. Wedler, G.; Colb, K. G.; McElhiney, G.; Heinrich, W. Appl. Surf. Sci. 1978, 2, 30–42. 175. Wedler, G.; Colb, K. G.; Heinrich, W.; McElhiney, G. Appl. Surf. Sci. 1978, 2, 85–101. 176. Vink, T. J.; Gijzeman, O. L. J.; Geus, J. W. Surf. Sci. 1985, 150, 14–23. 177. van Santen, R. A.; Neurock, M.; Shetty, S. G. Chem. Rev. 2010, 110, 2005–2048. 178. Hammer, B.; Norskov, J. K. In Advances in Catalysis, Impact of Surface Science on Catalysis; Academic Press: San Diego, 2000; Vol. 45, pp 71–129. 179. Bromfield, T. C.; Ferre, D. C.; Niemantsverdriet, J. W. ChemPhysChem 2005, 6, 254–260. 180. Curulla-Ferre, D.; Govender, A.; Bromfield, T. C.; Niemantsverdriet, J. W. J. Phys. Chem. B 2006, 110, 13897–13904. 181. Sorescu, D. C. J. Phys. Chem. C 2008, 112, 10472–10489. 182. Sorescu, D. C.; Thompson, D. L.; Hurley, M. M.; Chabalowski, C. F. Phys. Rev. B 2002, 66, 035416. 183. Ojeda, M.; Nabar, R.; Nilekar, A. U.; Ishikawa, A.; Mavrikakis, M.; Iglesia, E. J. Catal. 2010, 272, 287–297. 184. Scheijen, F. J. E.; Ferre, D. C.; Niemantsverdriet, J. W. J. Phys. Chem. C 2009, 113, 11041–11049. 185. Inderwildi, O. R.; Jenkins, S. J.; King, D. A. J. Phys. Chem. C 2008, 112, 1305–1307. 186. Shetty, S.; van Santen, R. A. Phys. Chem. Chem. Phys. 2010, 12, 6330–6332. 187. Jones, G.; Bligaard, T.; Abild-Pedersen, F.; Norskov, J. K. J. Phys. Condens. Matter 2008, 20, 064239. 188. Chorkendorff, I.; Niemantsverdriet, J. W. Concepts of Modern Catalysis and Kinetics. Wiley-VCH: Weinheim, 2003. 189. Ciobica, I. M.; van Santen, R. A. J. Phys. Chem. B 2002, 106, 6200–6205. 190. Sorescu, D. C. Phys. Rev. B 2006, 73. 191. Lo, J. M. H.; Ziegler, T. J. Phys. Chem. C 2007, 111, 11012–11025. 192. Govender, A. Towards a Mechanism for the Fischer–Tropsch Synthesis on Fe(100) Using Density Functional Theory. Ph.D. Thesis, Eindhoven University of Technology, Eindhoven, The Netherlands, 2010. 193. Cheng, J.; Hu, P.; Ellis, P.; French, S.; Kelly, G.; Lok, C. M. J. Phys. Chem. C 2010, 114, 1085–1093. 194. Cheng, J.; Gong, X. Q.; Hu, P.; Lok, C. M.; Ellis, P.; French, S. J. Catal. 2008, 254, 285–295. 195. Biloen, P.; Helle, J. N.; Sachtler, W. M. H. J. Catal. 1979, 58, 95–107. 196. Biloen, P.; Sachtler, W. M. H. Adv. Catal. 1981, 30, 165–216. 197. Ciobica, I. M.; Kramer, G. J.; Ge, Q.; Neurock, M.; van Santen, R. A. J. Catal. 2002, 212, 136–144. 198. Zhuo, M. K.; Tan, K. F.; Borgna, A.; Saeys, M. J. Phys. Chem. C 2009, 113, 8357–8365. 199. Gracia, J. M.; Prinsloo, F. F.; Niemantsverdriet, J. W. Catal. Lett. 2009, 133, 257–261. 200. Deng, L. J.; Huo, C. F.; Liu, X. W.; Zhao, X. H.; Li, Y. W.; Wang, J. G.; Jiao, H. J. J. Phys. Chem. C 2010, 114, 21585–21592. 201. Anderson, R. B. In Catalysis; Emmett, P. H., Ed.; Reinhold Publishing Company: New York, 1956; Vol. IV. 202. Botes, F. G.; van Dyk, B.; McGregor, C. Ind. Eng. Chem. Res. 2009, 48, 10439–10447. 203. Botes, F. G. Catal. Rev.: Sci. Eng. 2008, 50, 471–491. 204. Botes, F. G. Ind. Eng. Chem. Res. 2009, 48, 1859–1865. 205. Yang, J.; Liu, Y.; Chang, J.; Wang, Y. N.; Bai, L.; Xu, Y. Y.; Xiang, H. W.; Li, Y. W.; Zhong, B. Ind. Eng. Chem. Res. 2003, 42, 5066–5090. 206. van Dijk, H. A. J. Ph.D. Thesis, Eindhoven University of Technology, 2001 207. Botes, F. G.; Breman, B. B. Ind. Eng. Chem. Res. 2006, 45, 7415–7426. 208. Yates, I. C.; Satterfield, C. N. Energy Fuel 1991, 5, 168–173. 209. Donnelly, T. J.; Yates, I. C.; Satterfield, C. N. Energy Fuel 1988, 2, 734–739. 210. Lozano-Blanco, G.; Thybaut, J. W.; Surla, K.; Galtier, P.; Marin, G. B. Ind. Eng. Chem. Res. 2008, 47, 5879–5891. 211. Yang, Q. Y.; Maynard, K. J.; Johnson, A. D.; Ceyer, S. T. J. Chem. Phys. 1995, 102, 7734–7749. 212. Govender, N. S.; Botes, F. G.; de Croon, M.; Schouten, J. C. J. Catal. 2008, 260, 254–261. 213. Weckhuysen, B. M. Angew. Chem. Int. Ed. 2009, 48, 4910–4943. Comprehensive Inorganic Chemistry II: From Elements to Applications, (2013), vol. 7, pp. 525-557 View publication stats 557