Empowering Catalyst Supports: A New Concept for Catalyst Design Demonstrated in the Fischer − Tropsch Synthesis

: The Fischer − Tropsch (FT) synthesis is traditionally associated with fossil fuel consumption, but recently this technology has emerged as a keystone that enables the conversion of captured CO 2 with sustainable hydrogen to energy-dense fuels and chemicals for sectors which are challenging to be electrified. Iron-based FT catalysts are promoted with alkali and transition metals to improve reducibility, activity, and selectivity. Due to their low concentration and the metastable state under reaction conditions, the exact speciation and location of these promoters remain poorly understood. We now show that the selectivity promoters such as potassium and manganese, locked into an oxidic matrix doubling as a catalyst support, surpass conventional promoting effects. La 1 − x K x Al 1 − y Mn y O 3 − δ ( x = 0 or 0.1; y = 0, 0.2, 0.6, or 1) perovskite supports yield a 60% increase in CO conversion comparable to conventional promotion but show reduced CO 2 and overall C 1 selectivity. The presented approach to promotion seems to decouple the enhancement of the FT and the water − gas shift reaction. We introduce a general catalyst design principle that can be extended to other key catalytic processes relying on alkali and transition metal promotion.


■ INTRODUCTION
Potassium and manganese are commonly included in ironbased Fischer−Tropsch (FT) catalysts as promoters via impregnation or precipitation. 1 The effect of the alkali promotion on the catalytic performance is based on a modification of the local electron density of the active iron phase. 2 This results in stronger CO adsorption on the catalyst, weakening of the C−O bond, and decreased dissociative adsorption of H 2 . Conventionally, the FT activity increases up to a maximum K concentration upon which further improvement is hindered due to site blockage by the promoter itself or by enhanced C deposits. 3,4 In the presence of K, the product spectrum of the FT synthesis increases in value with a higher olefin content, an overall increased chain growth probability, and a reduced methane selectivity. 5 Concomitantly, an increased water−gas shift (WGS) activity 6,7 has been reported, aiding in the conversion of hydrogen-lean synthesis gas feeds obtained from the gasification of coal or biomass. Manganese promotion increases the number of basic sites while reducing their strength, further enhancing chain growth probability and olefin selectivity, especially at lower hydrocarbon chain lengths. 8,9 Particles of manganese oxide forming under reaction conditions have further been shown to suppress sintering of the active iron phase. 10 The exact speciation of promoter elements under reaction conditions 11 is still not fully resolved. Low concentrations and promoter mobility 4,12 present the most significant challenges in this regard. Without a clear understanding of the interaction between the promoter and the active phase, catalyst optimization through rational design is not possible. Innovative preparation techniques have shown that when an intimate contact between the active phase and the promoter is achieved, the promoting effects observed surpass effects reported for catalysts prepared with conventional impregnation approaches. 13 We have now developed a doped LaAlO 3 -based perovskite of general formula, ABO 3 , with K (10 at. %) partially replacing La on the A site and Mn (20−100 at. %) replacing Al on the B site which serves as a support for pre-prepared γ-Fe 2 O 3 nanoparticles. These catalysts are employed for low-temperature slurry-based FT synthesis. The samples with partial 100 mL with distilled water and filtered before measurement in the Agilent 4200 MP-AES spectrometer.
Inductively Coupled Plasma Optical Emission Spectroscopy. The sample (100 mg) was dissolved in a mixture of concentrated aqueous HCl (3 mL), concentrated aqueous HF (3 mL), and concentrated aqueous HNO 3 (5−10 drops) and heated to 130°C for 1 h using a microwave reactor. The reaction mixture was cooled to 24°C. After cooling, 60 mL of H 3 BO 3 was added. The analysis is conducted in a VARIAN OES-730 ICP−OES spectrometer.
Transition Electron Microscopy. The crystallite size distribution of the synthesised γ-Fe 2 O 3 (maghemite) nanoparticles was extracted from micrographs collected in a Tecnai F20 transmission electron microscope equipped with a field emission gun, operated at 200 kV. Scanning transition electron microscopy (TEM) micrographs and elemental mapping via electron-dispersive spectrometry (EDS) were acquired using two different instruments. In one instance, a double aberrationcorrected JEOL JEM-ARM 200F was used, operated at 200 kV, and equipped with an Oxford Xmax100 TLE EDS detector. Imaging and analysis of the samples were done in a scanning transmission electron microscopy (STEM) mode using a subangstrom-sized probe with a probe current between 68 and 281 pA. Probe current conditions were selected to optimise the beam current but, at the same time, to minimise the risk of beam damage to the specimen. The convergence semi-angle of the probe used was fixed at 23 mrad with acceptance semiangles of the dark-field detector being 34 to 137 mrad. The bright-field (BF) detector acceptance semi-angle was set at 0 to 12 mrad by using an illumination limiting aperture. In the other instance, STEM BF, high-angle annular dark-field, and EDS elemental mapping were recorded using a ThermoFisher Tecnai Osiris, equipped with a FEG operated at 200 kV and four Bruker windowless SSD EDX detectors in the SuperX configuration. TEM samples were prepared by depositing a few drops of the sample suspended in ethanol onto a Quantifoil R1.2/1.3 carbon film on a copper grid followed by drying at room temperature.
X-ray Photoelectron Spectroscopy. X-ray photoelectron spectroscopy (XPS) data were acquired using a Kratos Axis SUPRA using monochromated Al kα (1486.69 eV) X-rays at 15 mA emission and 12 kV HT (180 W) and a spot size/ analysis area of 700 × 300 μm. The instrument was calibrated to gold metal Au 4f (83.95 eV), and dispersion adjusted to give a BE of 932.6 eV for the Cu 2p 3/2 line of metallic copper. Ag 3d 5/2 line FWHM at 10 eV pass energy was 0.544 eV. Source resolution for monochromatic Al Kα X-rays is ∼0.3 eV. Using the Fermi edge of the valence band for metallic silver, the instrumental resolution was determined to be 0.29 at 10 eV pass energy (resolution with a charge compensation system on <1.33 eV FWHM on PTFE). High-resolution spectra were obtained using a pass energy of 40 eV, a step size of 0.1 eV, and a sweep time of 60 s, resulting in a line width of 0.696 eV for Au 4f 7/2 . Survey spectra were obtained using a pass energy of 160 eV. Charge neutralisation was achieved using an electron flood gun with a filament current of 0.45 A, a charge balance of 1 V, and a filament bias of 5 V. Successful neutralisation was adjudged by analysing the C 1s and Al 2p regions, wherein sharp peaks with no lower BE structure were obtained. Spectra have been charge-corrected to the main line of the carbon 1s spectrum (adventitious carbon) set to 284.8 eV. All data were recorded at a base pressure of below 9 × 10 −9 Torr and a room temperature of 294 K. Data were analysed using CasaXPS Ion Scattering Spectroscopy. Ion scattering spectroscopy (ISS) data were acquired on a Kratos Axis SUPRA using a 1 keV He + ion beam (Minibeam 6 GCIS) rastered over an area of 1 × 1 mm and an electron flood gun charge neutraliser. Data were analysed using CasaXPS v2.3.26rev1.1 K. Asymmetric peak modelling was achieved using the line shape [LF- (1.21,8.45,200,83)], which was developed from the analysis of model reference materials, along with a peak area ratio calibration curve (La 2 O 3 , Sigma-Aldrich, 99% and KCl, Fisher Scientific, 99%).
X-ray Absorption Spectroscopy. X-ray absorption spectra (XAS) of the K and Mn K-edges were collected at beamline B-18 of the Diamond Light Source, UK. Access to the beamline was granted via the Block Allocation Group access managed by the UK Catalysis Hub. The perovskite samples investigated were exposed to a reducing atmosphere of flowing hydrogen (16 mL·min −1 ·g catalyst −1 ), heated to 450°C at 2°C·min −1 , and held for 15 h to simulate the activation process in the FT catalytic process. The samples were deposited on a sticky Kapton film before XAS analysis. Data processing was performed using Athena from the Demeter software package. 15 To maximise the number of paths within the extended X-ray absorption fine structure (EXAFS) that could be fitted with the available potassium K-edge k space data, coordination numbers were fixed at appropriate values for the modelled structures, S o 2 was fixed based on the fitting of KCl, and values 2σ 2 were shared between suitable K−O paths. Both R factor and reduced χ 2 fitting parameters are reported with the latter accounting for the quality for the fit with respect to the number of variables fitted.
X-ray Diffraction. Crystallite phase composition, structure, and size were probed using X-ray diffraction (XRD). All samples and precursors were analysed in a Bruker D8-ADVANCE diffractometer equipped with a Co source and a position sensitive detector (LYNXSEYE) operating in Bragg− Brentano geometry. For spent samples, the catalyst was first extracted from the FT wax via Soxhlet extraction with xylene. Selected samples were studied in an in-house developed in situ capillary stage 16 fitted to a Bruker D8-ADVANCE diffractometer equipped with a Mo source and a position sensitive detector (Vantec) operating in a parallel beam mode. The samples were heated in a continuous flow of H 2 at atmospheric pressure (40 mL·min −1 ·g catalyst −1 ) from 50 to 450°C at 2°C· min −1 and held for 15 h. Subsequently, the temperature was reduced to 240°C, the pressure increased to 13.5 bar, and the catalyst exposed for 250 min to a stream of H 2 and CO in a 2:1 ratio at a space velocity of 383 mL·min −1 ·g catalyst −1 . To achieve high temporal resolution in the in situ experiments, the scan range was limited to 1/d = 0.2948 to 0.6583 Å −1 , resulting in a scan time of below 5 min. All patterns were evaluated against the International Centre of Diffraction PDF-4+ database (release 2020), and quantitative analysis was performed using Rietveld refinement methodologies 17 incorporated in the TOPAS 5 (BRUKER) software package. Catalytic Performance Evaluation. The activity and selectivity of all catalyst were evaluated in a 600 mL slurrybased continuous stirred tank reactor (CSTR). The catalyst (5 g) was reduced in hydrogen (16 mL·min −1 ·g catalyst −1 ) in a glass reactor positioned in a vertical tubular furnace in a hydrogen flow. The reactor was heated to 450°C at 2°C·min −1 and held for 15 h. The sample was then cooled to room temperature in a flow of Ar and quickly transferred into 50 g of molten wax (Sasol H), which was also protected from air exposure with a blanket of Ar. The wax tablet containing the catalyst was added to 300 g of wax in the CSTR, already molten at 140°C. The reactor was subsequently sealed, pressurised with argon to 15 bar at a stirring speed of 300 rpm, and heated to the reaction temperature of 240°C. A synthesis gas mixture (H 2 /CO = 2) with 10 vol % N 2 as an internal standard was fed to the reactor at 8 or 30 mL·min −1 ·g catalyst −1 to start the FT synthesis. All permanent gases were continuously analysed using a gas chromatograph equipped with a thermal conductivity detector (Agilent 7820A). The hydrocarbon products were collected in pre-evacuated ampoules 18 and analysed using a gas chromatograph equipped with a flame ionization detector (Varian CP-3800). The methane signal was used to link the two chromatography data sets.

■ RESULTS AND DISCUSSION
A series of La 1−x K x Al 1−y Mn y O 3−δ with x = 0 or 0.1 and y = 0, 0.2, 0.6, or 1 were prepared via the citrate method. The content of substituents was somewhat lower than that targeted (Supporting Information Table S1), with the perovskite forming a single rhombohedral (R3̅ c) crystalline phase characteristic for lanthanum aluminate (Supporting Information Figure S1a). Good evidence to support the proposed substitution is given by increases in the a and c lattice parameters from 5.37 to 5.50 and 13.11 to 13.31 Å, respectively, as the amount of manganese increases due to the larger atomic radius of Mn 3+ (0.580 Å) compared with Al 3 (0.535 Å) (Supporting Information Figure S1b). The BET surface area of the promoter-free perovskite is 5 m 2 g −1 . The incorporation of potassium increases the surface area to 29 m 2 g −1 , with the further addition of manganese reducing it gradually to 11 m 2 g −1 . No apparent surface enrichment of K or Mn is observed in energy-dispersive X-ray spectroscopy scanning electron microscopy (EDS−STEM) (Figure 1a, Supporting Information Figure S2 for higher magnification). and La 0.9 K 0.1 AlO 3−δ as-prepared and after reductive pre-treatment and of the K 2 CO 3 reference (top) and wavelet transformation of k 1 weighted data from the potassium K-edge extended X-ray absorption fine structure spectra of K 2 CO 3 (left bottom) and the pre-reduced La 0.9 K 0.1 AlO 3−δ (right bottom). (d) Particle size distribution of synthesised γ-Fe 2 O 3 nanoparticles. (e) K/La atomic ratios from X-ray photoelectron spectroscopy analysis with (inset) K 2p + C 1s high-resolution spectra of as-prepared and pre-reduced La 0.9 K 0.1 MnO 3−δ (carbons: red = CC/CH, pink = C−O, green = C�O, yellow = carbonate/adsorbed CO 2 ). (f) K/La atomic ratios from ion scattering spectroscopy (ISS) (derived from calibration curve) with (inset) ISS spectra of as-prepared and pre-reduced La 0.9 K 0.1 MnO 3−δ .
When potassium nitrate is impregnated onto the LaAlO 3−δ perovskite, needle-like structures of potassium and aluminiumenriched areas were observed (Supporting Information Figure  S3b). Potentially, the colocation of potassium and aluminium could indicate the formation of KAlO 2 . However, as the phase is unstable in the presence of CO 2 or moisture, it would phaseseparate into an amorphous aluminium oxide and potassium carbonate in air and/or under FT synthesis. 19 The only sample showing similar structures, albeit much smaller in size, is La 0.9 K 0.1 AlO 3−δ before reduction (Supporting Information Figure S3b). The K and Mn K-edge X-ray absorption near edge structure (XANES) of the samples as-synthesised and after exposure to a stream of hydrogen at 450°C were analysed, simulating the activation procedure of the FT catalyst ( Figure 1b,c, Supporting Information Figures S4 and S5). Mn K-edge spectra prior and post reduction correspond to LaMnO 3 . 20,21 Based on the E 0 position, the oxidation state of Mn in the K-doped perovskites prior to reduction was higher than that in LaMnO 3 , to compensate for the reduced average A site charge. Reduction resulted in a change in Mnoxidation state from 3.5 to 3.2 while retaining the features of a perovskite. In concurrence with XRD and XANES, EXAFS analysis demonstrated that the Mn environment in all samples, prior and post reduction, corresponds to that expected for a perovskite (Table 1 and Supporting Information Figure S4).
Given the concentration of the K dopant, evidence for the inclusion of K within the structure cannot be deduced from the Mn edge data. Therefore, potassium K-edge XANES and   EXAFS analyses were performed to understand the local structure of K. Fingerprinting of the reduced La 0.9 K 0.1 Al 1−y Mn y O 3−δ spectra against a database of K standards (Figure 1c and Supporting Information Figure S5), including KOH, KHCO 3 , and K 2 CO 3 (which are reported for conventionally prepared promoted catalysts 11,22 ), demonstrates that K is not present as a phase-separated structure. KAlO 2 can be discounted given its instability in the presence of moisture or CO 2 . Only the XANES of unreduced La 0.9 K 0.1 AlO 3−δ showed any indication of K 2 CO 3 . In comparison, the LaAlO 3−δ perovskite impregnated with potassium nitrate shows a clear potassium K-edge spectrum of KNO 3 , which converted to K 2 CO 3 after reductive treatment (Supporting Information Figure S5b). XPS and ISS surface analyses were employed to further probe potential surface segregation of potassium species (Figure 1e,f). XPS showed an elevated K/La ratio (Supporting Information Table S2), marginally above that anticipated from elemental analysis, though sample probing depth must be accounted for in these systems given that the particles are in a similar size regime to the information depth of photoemission from Al Kα X-rays and at the mercy of topological influences. Both as-prepared and pre-reduced La 0.9 K 0.1 AlO 3−δ showed elevated surface K content, above that found for the La 0.9 K 0.1 Al 1−y Mn y O 3−δ samples, though reductive treatment reduced this apparent surface enrichment. ISS analysis, however, indicated that the outer surface composition is much closer to that expected for a homogeneous perovskite. The difference to the targeted K/La ratio of 0.1 is likely a result of potassium removal by sputtering. This indicates that K enrichment seen using XPS is indeed primarily due to topological effects. Unreduced La 0.9 K 0.1 AlO 3−δ retained a high K content at the surface, suggesting some phase segregation, which disappeared after reduction. Comparison of the EXAFS wavelet transformations of the perovskite supports and K 2 CO 3 reference shows clear differences in R space above 3 Å (Figure 1c and Supporting Information Figure S6). Fitting of the data (excluding La 0.9 K 0.1 AlO 3−δ , which has K 2 CO 3 phase impurities in the unreduced state) shows that the larger and lower charge K + cation (relative to La 3+ ) is accommodated into the perovskite structure through an asymmetric expansion of the first shell oxygen environment (Figure 2 and Table 2 and Supporting Information Figure S7). The model indicates expansion of the 6 K−O bonds within the plane, but no expansion of those above or below the plane (Figure 3). Furthermore, the anticipated over-bonding of the site is alleviated by a reduction in the first shell coordination number from 12 to 11. Such oxygen nonstoichiometry is essential for K incorporation into LaAlO 3−δ , as neither La nor Al cations can accommodate the charge differential as Mn can. This also explains the absence of a shift in lattice parameters when only K is incorporated into the perovskite lattice. The presence of K 2 CO 3 in the unreduced La 0.9 K 0.1 AlO 3−δ , as seen by XANES and the  with K + , extended X-ray absorption fine structure modelling suggests that one O 2− above or below the a−b plane is removed resulting in a reduction of the coordination number from 12 to 11.

ACS Catalysis pubs.acs.org/acscatalysis
Research Article wavelet-transformed EXAFS data, can be explained by the lack of sufficient oxygen nonstoichiometry, which is then remedied by the reduction at 450°C, an observation further supported by a concomitant reduction in carbon species at 289 eV in the XPS analysis (carbonates/adsorbed CO 2 ) by approximately 25% following reduction (Figure 1e/Supporting Information Table S2) (Figure 4). Further evidence that K is located within the perovskite lattice is provided by the successful fitting of the first shell K− La, K−Mn(Al) and second shell K−O paths (Figure 2, Table 2 and Supporting Information Figure S7). To confirm the validity of the model, paths representative of KOH, K 2 O, and K 2 CO 3 were separately fitted to the data but resulted in unrealistic parameters and poor fits (Supporting Information  Figures S7−S9 and Table S3). This confirms that K has been incorporated into the perovskite lattice of all reduced perovskite supports studied.
Maghemite (γ-Fe 2 O 3 ) nanoparticles were prepared via precipitation from an iron(II) chloride precursor solution 25 and subsequently deposited onto the supports with an ultrasound-induced physical mixing technique 26 targeting a maghemite loading of 20 wt % (Figure 3; Supporting Information Table S1; Figure S10). The supported catalysts are termed Fe-La 1−x K x Al 1−y Mn y O 3−δ . The average oxidic crystallite size is 22 nm (Figure 1d), significantly above the size range for the activated metallic and carbide iron species, associated with crystallite size-dependent changes in activity and selectivity in the FT synthesis. 27,28 The reduction and formation behaviour of the supported γ-Fe 2 O 3 nanoparticles was studied in an in-house developed in situ XRD capillary cell 16 (Figure 3, Supporting Information Figures S11 and S12). On the LaAlO 3−δ carrier, the maghemite phase is rapidly converted to magnetite (Fe 3 O 4 ) in a hydrogen atmosphere at 260°C before fully reducing to the metallic α-Fe from 400°C onward. The addition of K to the perovskite structure lowers the reduction temperature of the maghemite to magnetite by 80°C without affecting the reduction of magnetite to the metallic iron. This is in line with our previous observations, showing that potassium promotion of iron catalysts influences the reduction temperature to Fe 3 O 4 but not to α-Fe. 29 In the presence of an additional substitution of Al with Mn in the B site, the enhancement of the first reduction step is not observed. Based on the XANES spectra of the potassium K-edge discussed earlier, it is likely that the facilitation of the first reduction step is a result of free K 2 CO 3 which is only incorporated into the Mn-free perovskite structure upon the reduction-induced oxygen nonstoichiometry.
Following the reductive treatment, the catalysts were exposed to FT conditions for approximately 4 h to monitor the carburization process. On the unmodified perovskite support, Hagg carbide (χ-Fe 2 C 5 ) formation is observed after 1.3 h and all metallic iron reflexes disappear after 1.8 h. When K is present in the perovskite structure, this process is significantly enhanced with the first evidence of carbide recorded after just 0.25 h and full conversion of the metallic phase after 0.5 h. The enhancement of the carburization process through K has been reported extensively for classic promotion and is rationalised by a donation of electron density from the alkali metal to the iron phase, enhancing CO adsorption over H 2 adsorption and weakening the carbon− oxygen bond. 2,30 The observed effect on carburization is therefore direct evidence that K incorporated in a perovskite structure does provide promotional activity to a catalytically active phase. The additional presence of Mn does not influence this phase transition on bulk scale. Effects on the detailed composition of the carbide phase cannot be excluded at this stage.
All catalysts were studied for their FT performance in a 600 mL slurry CSTR reactor. The unmodified Fe-LaAlO 3−δ compares well with more classical supports such as SiO 2 , Al 2 O 3 , ZrO 2 , and TiO 2 , both with regard to CO conversion as well as product selectivity (Supporting Information Figure S13 and Table S4). A 10 at. % replacement of La with K increases . Subsequently, the reactor temperature was reduced to 240°C, the pressure increased to 13.5 bar, and the catalyst exposed for 4.17 h to a stream of H 2 and CO in a 2:1 ratio as a pace velocity of 383 mL·min −1 ·g catalyst −1 .
the CO conversion from 45 to 76% which equates a rate increase per mol iron in the reactor of 1.6 (Supporting Information Table S4). This increase is paralleled by a nine percentage points increase in chain growth probability and an almost fourfold increase in CO 2 selectivity. The increased chain growth materialises predominantly in the short chain hydrocarbon fraction with an overall increased olefinicity, that is, an increased ratio of primary olefin to n-paraffin concentration. For a realistic comparison of catalyst selectivity, the K-bearing catalyst was re-evaluated at a 3.75 times higher space velocity, resulting in a CO conversion of 34% ( Figure 5). At this near iso-conversion condition, the CO 2 selectivity is only slightly enhanced (12.7 vs 15.6 C %) at an even further increased chain growth probability (12% points increase compared to the Fe-LaAlO 3−δ catalyst), resulting in a reduction of methane and an increase in higher hydrocarbon concentrations in the product stream. The olefinicity remains elevated. The additional substitution of 20 at. % of Al with Mn retains the enhanced activity and results in a further increased chain growth probability, a reduced methane selectivity, and significantly increased C 5+ selectivity. Furthermore, CO 2 formation is suppressed by the presence of Mn. Higher concentrations of Mn, both in the presence and absence of K, reduce the activity to levels below the unpromoted sample, combined with increased WGS activity and FT chain growth probability (Supporting Information Figure S13c−f).
To contextualize the enhanced performance of the potassium-containing perovskites, LaAlO 3−δ was impregnated with 2, 1, or 0.5 wt % K before the deposition of the maghemite nanoparticles. 2 wt % equates the total potassium concentration in the perovskites with a 10 at. % replacement of La with K. However, the resulting surface concentration is much higher in the impregnated sample. Of the three catalysts, the two lower promoter concentrations yield an elevated CO conversion (76 and 78%; Supporting Information Figure  S13c,d) comparable to the enhancement observed for the potassium-containing perovskite samples. Compared to Fe-La 0.9 K 0.1 AlO 3−δ and Fe-La 0.9 K 0.1 Al 0.8 Mn 0.2 O 3−δ at an elevated space velocity, that is, at conversion levels also comparable with the unpromoted catalyst, the hydrocarbon product slate is slightly heavier, due to a lower C 1 −C 4 fraction (Figure 5.). However, the CO 2 selectivity is up to 60% higher for the impregnated samples due to a parallel promotion of the WGS activity, resulting in an overall higher undesired C 1 product stream.
By anchoring potassium within the perovskite matrix, we have been able to deconvolute its enhancement of the FT and WGS activity. The potassium-containing perovskites yield less CO 2 at equivalently enhanced CO conversion rates. Traditionally, the WGS reaction is beneficial when operating largescale FT plants with a H 2 lean feedstock, for example, as obtained from the gasification of coal. In a future scenario of small-scale power-to-liquid plants, the suppression of the WGS activity provides the opportunity to employ a low-cost ironbased FT catalyst instead of their cobalt-based counterparts, without loss of product quality. The developed promoter presentation further overcomes promoter mobility, especially at the expected harsh power-to-liquid reaction conditions.
After 48 h at FT conditions, the spent catalysts were recovered in solidified wax. Fragments of the wax-covered catalyst were digested for elemental analysis (Supporting Information Table S5). No significant change in composition was detected. To prepare the samples for XRD, wax residues were removed via Soxhlet extraction in xylene at 140°C for 4 days. The obtained diffraction patterns show a stable perovskite structure (Supporting Information Figure S14). In the presence of potassium, the iron phase is predominantly carbide with minor traces of magnetite. However, in the absence of the alkali, about 6 wt % of metallic iron is detected in the catalyst alongside the Hagg carbide and traces of Fe 3 O 4 . The speciation of potassium does not change after 48 h under reaction conditions (Supporting Information Figure S15). For all samples with K present, the space velocity was elevated by a factor of 3.75 to achieve comparable conversion levels. The stacked bar chart in the middle of the figure represents the composition (CH 4 , C 2−4 , and C 5+ ) of the hydrocarbon fraction in the product stream as well as the selectivity toward CO 2 of the same catalysts. In the bottom panel, the ratio of primary olefins to linear paraffins in the C 5 hydrocarbon faction is shown. Error bars are provided for all datasets. All FT experiments were carried out in a 600 mL slurry CSTR reactor at 240°C, 15 bar pressure, a H 2 /CO ratio of 2 with 10 vol % N 2 as internal standard, and a space velocity of 8 mL·min −1 ·g catalyst −1 for Fe-LaAlO 3 and 30 mL·min −1 ·g catalyst −1 for the K-containing catalysts. Data collected after 48 h time on stream.

■ CONCLUSIONS
We have synthesised new support materials by anchoring promoting elements within a perovskite lattice. We demonstrated that these novel materials can act as empowered catalyst supports by providing both the physical functionality of common catalyst support materials as well as the chemical and electronic effects of conventional promoters. Due to the versatility of perovskites, it is expected that improved promoter incorporation can be achieved by changing the composition of the parent perovskite. 31 While the materials in the present study displayed low specific surface areas, a number of synthesis techniques using templating approaches have been reported to overcome this perovskite-specific limitation. 32,33 In the case of the iron-based FT synthesis, this new approach has outperformed conventional techniques by matching conversion enhancements of conventional promotion while promoting the FT reaction more selectively over the WGS reaction. These enhancements were stable over the evaluation period of 48 h. There is no reason to assume that the observed effects represent a singularity regarding the promoter elements and the chosen reaction. We expect that the underlying design principles can be transferred to other catalytic processes breaking the boundaries of known performance enhancement abilities of promoter elements.
■ ASSOCIATED CONTENT * sı Supporting Information