A Highly Active and Selective Manganese Oxide Promoted Cobalt-on-Silica Fischer–Tropsch Catalyst

Silica support material (Grace-Davison Davicat 1454SI silica gel, BET surface area = 500 m² g⁻¹, pore volume = 1.1 mL g⁻¹ and 6 nm average pore size) was dried for 12 h in air at 120 ºC prior to further use. MnO-promoted cobalt catalysts were prepared via a single pore-volume impregnation using an aqueous solution containing Co(NO₃)₂·6H₂O, Mn(NO₃)₂·6H₂O and Pt(NH₃)₄(NO₃)₂, aiming for a cobalt metal loading of 17 wt%. Various catalysts were prepared with a Mn/Co atomic ratio ranging from 0 to 0.25. Platinum (0.05 wt%) was added in all cases as a reduction promoter via co-impregnation. After impregnation the catalyst was dried for 12 h at 70 ºC, with a heating rate of 1 ºC min⁻¹, in stagnant air. Subsequently, the dried catalyst (100 mg) was calcined for 1 h at 240 ºC in a 100 mL min⁻¹ flow of either 1 vol% NO in He (NC) or air (AC). The catalysts (20 mg) were reduced prior to FT catalysis with a heating rate of 5 °C.min⁻¹ at temperatures ranging from 400 ºC to 550 °C for 2 h, using a flow of 30 vol% H₂/He (60 mL min⁻¹). For the various characterization experiments (vide infra) similar reduction conditions were applied, followed by passivation in air at room temperature (rt).

XRD analyses were conducted to determine the average crystallite size of the calcined catalyst precursors. The diffraction patterns were recorded by a Bruker-AXS D8 Advance X-ray diffractometer using Co-K_α radiation (λ = 1.789 Å) scanning from 10 to 90º 2θ. The Co₃O₄ crystallite sizes were determined using the Scherrer equation for the (311) peak at 2θ = 43.1º.

TEM measurements were performed using an FEI Technai 20F. TEM samples were prepared via an ethanol suspension of the passivated catalysts brought onto a carbon film on a copper grid.

STEM–EELS measurements were performed to investigate the spatial distribution of cobalt, manganese and silica in an AC or an NC Co/MnO/Pt/SiO₂ catalyst (Mn/Co = 0.08 at/at). The Co and Mn L_2,3-edges and the O K-edge were studied using a 100 keV STEM apparatus (VG HB 501) equipped with a field emission source and parallel Gatan EELS spectrometer [26, 27]. The EELS-spectra were taken with a 0.99 eV energy resolution and a 1 nm spatial resolution. The reduced (450 ºC) and passivated samples were sonicated in ethanol and brought onto a carbon film on a regular copper EM grid.

X-ray Absorption Near-Edge Spectroscopy (XANES) at the Co K-edge was applied to study the degree of reduction of the catalysts. The measurements were done at DESY synchrotron (beamline C) in Hamburg, using a Si (111) double crystal monochromator detuned to 60% of the maximum intensity to avoid higher harmonics. The catalysts were reduced in situ in a transmission cell in a 30 vol% H₂ in He flow, with a ramp of 5 ºC.min⁻¹ at 450 ºC, for 2 h. Prior to a XANES measurement the samples were cooled in the H₂/He flow to liquid nitrogen temperature. Spectra of Co₃O₄, CoO and cobalt foil were measured as references. The absorption spectra were analyzed using the XDAP code, as described elsewhere. [28, 29] The degree of reduction was calculated using linear combination analysis of the XANES spectra of the catalysts, CoO and Co foil references.

Infrared spectroscopy was used to study the adsorption of CO on the cobalt catalysts, and investigate the influence of MnO addition. For the IR measurements, passivated catalyst (5 mg) was mixed with silica (5 mg, Davicat silicagel), pressed into a self-supporting wafer (~6 mg cm⁻²) and mounted in an IR transmission cell. Prior to the CO adsorption measurements, the catalysts were re-reduced in an H₂ flow (~50 mL min⁻¹) at 450 ºC for 2 h. Afterwards, the catalysts were cooled to room temperature. Below 100 ºC, the hydrogen flow was stopped, and the cell was evacuated (10⁻⁶ mbar) for 15 min. Subsequently, after the admission of 350 mbar 10 vol% CO/He to the cell, several IR spectra were collected during a period of 30 min at ambient temperatures.

The FT reaction was performed at 220 °C at 1 bar in a plug-flow reactor with a H₂/CO ratio of 2 v/v. Typically 20 mg catalyst (90–150 μm), mixed with 200 mg SiC (~200 μm), was loaded in the reactor in order to achieve differential and isothermal plug-flow conditions. The catalysts were reduced in situ in an H₂/Ar (20/40 mL min⁻¹) flow at temperatures ranging from 400 to 550 ºC for 2 h, with a ramp of 5 ºC min⁻¹. Gas chromatographic analysis was performed during the FT reaction to determine the activity or Cobalt-Time-Yield (CTY, 10⁻⁵ mol_CO.g _Co ⁻¹  s⁻¹) and selectivity (wt%) towards C₁- and C₅₊-hydrocarbons. The reported activity and selectivity data are obtained after at least 20 h operation and at 2% CO conversion level, which was achieved by tuning the gas flow.

The SSITKA experiments were performed as described elsewhere [30]. Typically 100 mg catalyst (90–150 μm) was diluted with 200 mg SiC (75–150 μm) and loaded in a plug-flow microreactor. Prior to catalysis, the catalysts were reduced at 450 ºC for 2 h, with a flow of 30 vol% H₂ in Ar (40 mL min⁻¹). The experiments were performed at 210 ºC, 1.85 bar, with an H₂/CO ratio of 10 v/v. During steady-state reaction isotopic switches were performed, e.g., from ¹²CO/Ar/H₂ to ¹³CO/Kr/H₂. The transients of labeled reactants and products (e.g., ¹³CO and ¹³CH₄) were monitored with a Mass Spectrometer (MS). The surface residence times and coverages of CO and CH _x intermediates were calculated from those transients. A gas chromatograph equipped with FID and TCD was used to determine the CO conversion.

In Table 1 the Co₃O₄ crystallite sizes of the calcined Co(/MnO)/Pt/SiO₂ catalyst precursors prepared via NC or AC with different Mn/Co ratios are shown.

As can be observed in Table 1, the air calcination treatment yields a significantly larger Co₃O₄ crystallite size for the unpromoted catalyst as compared to that obtained after calcination in a flow of NO/He. This shows the beneficial impact of NC on the Co₃O₄ crystallite size, which is in line with previous findings [5, 6, 31].

For the MnO-promoted catalysts, a smaller difference in Co₃O₄ crystallite size after either AC or NC was observed. Nevertheless, a decrease in crystallite size is observed for both NC and AC catalysts after the addition of a small amount of MnO (Mn/Co = 0.06 at/at) as compared to their unpromoted counter parts. From Table 1 it can moreover be concluded that the crystallite sizes of the Co/MnO/Pt/SiO₂ NC samples are not influenced by the amount of MnO, whereas the Co₃O₄ crystallites of the AC samples showed continuously a decreasing size with increasing MnO content. It should be noted though that the co-impregnation method is expected to yield mixed CoMn-oxides after calcination. As the Mn²⁺ ions possibly affect the stacking in the cobalt oxide crystals, domain sizes rather than crystallite sizes will be detected with XRD. This suggests that from XRD we will underestimate the crystallite sizes. In all cases no diffraction lines of MnO were detected. Please note it was assumed that the passivated catalyst contains Mn in a 2⁺ oxidation state only, based on e.g., a study by Morales et al. on Co/MnO/TiO₂ catalysts. [17].

TEM analysis from reduced (450 ºC) and passivated catalysts was used to investigate the cobalt particle size and distribution in more detail.

From Fig. 1a, c showing TEM images of unpromoted AC and NC Co/Pt/SiO₂ catalysts, respectively, it was concluded that a significantly higher cobalt dispersion is obtained with NO calcination as compared to air calcination. This confirms the XRD results (cf. Table 1). Moreover, clustering of cobalt particles is observed in the case of the air calcined samples (as indicated by the white circle in Fig. 1a) while the cobalt particles of the NC prepared sample (Fig. 1b) are well separated on the silica surface.

For the AC catalysts a significant improvement in cobalt dispersion is observed upon MnO addition, and a decrease in average Co particle size from 12 nm (Mn/Co = 0 at/at) to 5.6 nm (Mn/Co = 0.25 at/at, Fig. 1b) is found. However, TEM analysis on the latter sample also revealed the presence of large Co particles (>20 nm), indicating a broad particle size distribution. Nevertheless it can be concluded that the addition of manganese nitrate to impregnation solution improved the average cobalt dispersion of AC catalysts.

For the NC catalysts a small decrease in cobalt particle size is observed after MnO addition. For the MnO-promoted samples (Fig. 1c, d), small Co particles (3–4 nm, cf. Table 1) are obtained for both low (0.06, C) and high (0.25, D) Mn/Co atomic ratios. From the similar cobalt sizes of the promoted NC catalysts (Fig. 1c, d) it was concluded that the effect of the Mn/Co ratio on the cobalt dispersion is negligible. It should be mentioned though that due to the small difference in Co and Mn mass, a distinction between these elements in TEM could not be made, which therefore complicated an accurate cobalt particle size determination.

To investigate the spatial distribution of cobalt and manganese oxide on the silica surface after reduction and passivation, Scanning Transmission Electron Microscopy measurements combined with electron energy loss spectroscopy (STEM–EELS) were conducted. In this study the catalyst with the optimum FT performance with a Mn/Co ratio of 0.08 at/at (vide infra) prepared via NO calcination was investigated along with the corresponding air calcined sample.

In Fig. 2 composite maps of energy selected STEM images of a NO and AC sample, each measured at two spots on the sample, are shown with the spatially resolved integrated EELS intensities for oxygen, cobalt and manganese. STEM images with integrated EELS intensities for the individual elements are provided in the supplementary information.

From these images it can be concluded that a higher cobalt dispersion is obtained for the NC catalyst than for the AC catalysts, which confirms the XRD and TEM results (vide supra). Moreover, the dispersion of MnO has increased significantly upon NO calcination. Whereas relatively large particles of MnO (up to 4 nm) are formed after air calcination, often present close to the Co particles, the NO calcination seems to yield a homogeneous distribution of MnO over both the cobalt and the silica surface.

X-ray absorption spectroscopy was applied to investigate the degree of reduction. These experiments were conducted at liquid nitrogen temperature, after an in situ reduction treatment. The XANES part of these measurements on various Co(/MnO)/Pt/SiO₂ catalysts, together with Co foil and a CoO reference sample, is shown in Fig. 3.

Using linear combination analysis of CoO and Co foil reference samples, the degrees of reduction of the in situ reduced AC and NC catalysts were determined, and are listed in Table 2.

From this table it is concluded that high degrees of reduction are obtained at 450 ºC for both the AC and NC sample without MnO. Complete reduction is also achieved for the AC catalyst containing MnO (Mn/Co = 0.08 at/at). The small Co particles prepared via NO calcination, however, show a 62% degree of reduction only. Yet, this value increased to 82% after 2 h of FT synthesis (data not shown). The lower degree of reduction might be attributed to the retarding effect of MnO on the extent of reduction, as has been shown earlier e.g., for Co/MnO/TiO₂ catalysts [16].

In a first series of catalysis experiments, the effect of Mn/Co ratio on the FT performance was studied. The catalysts were reduced at 550 ºC to reach a degree of reduction close to 100% (vide supra).

For the CoPt/SiO₂ catalysts (no MnO present) the activity difference between NC and AC treatments is limited, and smaller than reported before [6]. Clearly, the reduction temperature applied (550 ºC) is not optimal for the NC catalyst and reduction at 450 °C has been used before [6].

For the NC catalyst reduced at 550 ºC it can be concluded that the addition of MnO up to Mn/Co = 0.08 brings about an increase of both activity (Fig. 4a) and selectivity (Fig. 4a). For larger amounts of MnO (Mn/Co > 0.13) slightly higher C₅₊-selectivity values were found, however at a significant expense of the activity. This might be due to blockage of the cobalt surface [21] and/or a lower degree of reduction due to the reduction-retarding effect of MnO [16].

For the AC catalyst it was found that the addition of a small amount of MnO caused a decrease in activity, which value even further decreased for higher Mn/Co ratios. This might be due to blocking of the cobalt surface by MnO [21]. Nevertheless, the beneficial effect of MnO is reflected in the increase in C₅₊-selectivity, and an increase in Mn/Co atomic ratio from 0 to 0.25 was accompanied with an increase in C₅₊-selectivity from 56 to 70 wt%.

For the NC catalyst with the optimum amount of MnO (Mn/Co = 0.08 at/at), the effect of reduction temperature was investigated (Fig. 5a). For comparison, the effect of reduction temperature on the performance of the NC Co/Pt/SiO₂ catalyst without MnO was included (Fig. 5b).

From Fig. 5a it can be concluded that for NC Co/MnO/Pt/SiO₂ a slight increase in C₅₊-selectivity is obtained with an increase in reduction temperature. For the activity a shallow optimum temperature of 450 ºC is found. The lower activities obtained at high reduction temperatures (>450 ºC) might be due to sintering.

For the unpromoted NC Co/Pt/SiO₂ catalyst (Fig. 5b), an initial increase in activity with increasing reduction temperature is shown. At high temperatures (>450 ºC) however, a significant drop in activity is observed, which is attributed to sintering concluding from additional XRD results. It is interesting to note that the drop in activity at too high reduction temperatures (<450 ºC) is larger for the Co/Pt/SiO₂ catalyst as compared to the Co/MnO/Pt/SiO₂ catalyst. This might indicate that the presence of MnO promoter inhibits sintering of the Co particles during reduction. For the Co/Pt/SiO₂ catalyst relatively low C₅₊-selectivities were found, although higher values (up to 48 wt%) were obtained at higher (>450 ºC) reduction temperatures possibly related to the presence of larger Co particles [2‐4].

From the comparison of the NC Co/Pt/SiO₂ and NC Co/MnO/Pt/SiO₂ catalysts reduced at their optimum reduction temperature (450 ºC) to obtain maximum activity it can be concluded that a significant increase in C₅₊-selectivity (from 32 wt% to 54 wt%) is achieved by the addition of MnO. Most notably, the highest activity (4.6·10⁻⁵ mol_CO.g _Co ⁻¹  s⁻¹) found for the MnO promoted catalyst is close to the value of 4.9·10⁻⁵ mol_CO.g _Co ⁻¹  s⁻¹ obtained for the unpromoted catalyst, cf Fig. 5b. Moreover, this relatively high activity of the MnO promoted catalyst is obtained for Co particle sizes (~4 nm) smaller than the optimum particle size (~5 nm) [6], and hence a significantly lower activity was expected. This might indicate that the MnO promoted catalysts do not show a cobalt particle size effect identical to that of unpromoted catalyst [6], possibly caused by a different Co surface structure or particle shape. However, it might also indicate that manganese oxide acts both as selectivity and activity promoter [32], thereby boosting the performance of small Co particles.

IR spectroscopy of adsorbed carbon monoxide was applied to investigate the effect of MnO on the nature of the cobalt surface of NC Co(/MnO)/Pt/SiO₂ catalysts. As CO can bind to the cobalt surface in a linear, bridged and multiple-bridged form, which all have a characteristic vibrational frequency in the infrared region, detailed information about the structure and electronic properties of the surface sites can possibly be obtained [22, 33, 34]. In this study, NO-calcined Co/MnO/Pt/SiO₂ catalysts with a Mn/Co atomic ratio of 0, 0.08 and 0.13 were investigated as well as air-calcined Co/Pt/SiO₂. Prior to the IR measurements, the passivated catalyst samples were re-reduced in situ at 450 ºC (see details in experimental). CO was adsorbed at ambient temperatures.

Figure 6 shows the region of the CO vibrations in the IR spectra. Small bands developed at 2,180 and 2,126 cm⁻¹ are assigned to gaseous CO [22]. For the air and NO calcined Co/Pt/SiO₂, bands at 2,057 and 1,892 cm⁻¹ are found, which are indicative for linear and bridged bonded CO on metallic cobalt particles, respectively [35]. From the peak areas it can be concluded that the amount of adsorbed CO on the NC catalyst is almost twice as high as on the AC catalyst, which is due to the enhanced Co dispersion. Moreover, a two times higher linear:bridge ratio of adsorbed CO is found for the NC catalyst, which might be ascribed to a higher fraction of edge and corner sites at the surfaces of the small cobalt particles in the NC sample than present at the larger particles of the AC sample. A similar effect has been observed for adsorption of CO on a defect rich Co (0001) surface, where a lower quantity of bridge-bonded CO was found as compared to the amount present on an annealed surface with a low amount of defects [36].

The presence of MnO in the NC samples induced a significant change in both the CO coverage and bonding mode. This proves the close interaction of the Co particles and MnO promoter. The lower IR absorption signal indicates a lower amount of CO adsorbed on the manganese promoted catalysts, probably due to blocking of part of the Co surface by MnO [21]. Moreover, next to the peak at 2,057 cm⁻¹, a second distinct peak of linearly bonded CO is observed at around 2,012 cm⁻¹ region. Whereas the first peak (2,057 cm⁻¹) is ascribed to CO adsorption on fcc cobalt [37], the latter peak has been attributed to the linear adsorption of CO on low-index surface crystallographic planes or corners and steps sites with coordinatively unsaturated sites [4, 38, 39]. This indicates that MnO-promoted catalysts exhibit a different cobalt surface structure as compared to unpromoted catalysts.

For the AC Co/Pt/SiO₂ and the MnO promoted NC catalysts higher C₅₊-selectivities were found as compared to the C₅₊-selectivity of the NC Co/Pt/SiO₂ catalyst. Since the AC catalyst shows a higher amount of bridged-bonded CO and the MnO promoted catalysts possibly show a different surface structure as compared to NC Co/Pt/SiO₂, it might be concluded that the C₅₊-promoting effect of either a relatively larger Co size (AC sample) or the presence of MnO has a different origin.

Isotopic switches (¹²CO/¹³CO and H₂/D₂) were applied to study the amount and residence times of various FT surface intermediates. First the carbon intermediates were investigated by a switch from ¹²CO/Ar/H₂ to ¹³CO/Kr/H₂ after reaching steady-state conversion at 210 ºC, H₂/CO = 10 v/v and 1.85 bar. This was followed by a back-switch, after reaching an isotopic steady-state. From this back-switch, the residence times of CO and CH _x were calculated. In this case the CH _x intermediates represent the surface species which eventually produce CH₄. A detailed description of the transient analyses has been published elsewhere [30]. For AC Co/Pt/SiO₂ and NC Co(/MnO)/Pt/SiO₂ with Mn/Co atomic ratios of 0, 0.08 and 0.25 reduced at 450 ºC, the amounts (N) of CO and CH _x bonded to the cobalt surface were calculated using the residence times and gas flows. (Table 3) Since part of the surface might be blocked by MnO, as the IR CO adsorption measurements suggest, the determination of the number of Co surface sites via the cobalt dispersion is cumbersome. Hence, the CO and CH _x surface coverage on the promoted catalysts could not be calculated. Based on previous SSITKA studies using Co/CNF catalysts [3, 30] however, it was assumed that the number of cobalt surface sites equals two times the number of reversibly bonded CO. This allowed to calculate the CH _x and CO coverage via θ_{CHx or CO} = N_{CHx or CO}/(2N_CO).

Table 3 shows that the residence time and amount of reversibly adsorbed CO on the NC Co/Pt/SiO₂ catalyst is twice as high as compared to the amount of CO on its air-calcined counterpart. This is in line with the IR results (vide supra) and is ascribed to the two times higher Co surface area per gram of catalyst.

For the CH _x intermediate, both a lower residence time and surface coverage was found for the NC Co/Pt/SiO₂ catalyst as compared to its AC counterpart. However, upon MnO addition, an increase in residence time and coverage of CH _x intermediates was observed with increasing Mn/Co ratio, which is in line with findings by Vada et al. using LaO _x promotion for a Co/Al₂O₃ catalyst [40]. At the same time, both a decrease in the residence time and amount of reversibly adsorbed CO was observed. It is interesting to note that also for ZrO _x promoted Co/Al₂O₃ catalysts an increase in CH _x -coverage and TOF is observed upon addition of the Zr promoter [41, 42]. However, in this case no change in CO coverage is observed.

From the CH _x coverage and residence time, and assuming pseudo first-order kinetics, the TOF was calculated (Table 3) [3]. The obtained values show a similar surface-specific activity for the AC and NC Co/Pt/SiO₂ catalyst. Interestingly, for the MnO promoted catalysts a higher calculated TOF was found, which indicates that MnO also acts as an activity promoter. This is also in line with findings for ZrO _x -promoted catalysts [41‐43]. This might explain the fact that a higher selectivity of the MnO promoted NC samples was found, without showing a significant loss in activity.

In order to provide a qualitative understanding of the higher C₅₊-selectivity of the MnO promoted catalysts and the air-calcined Co/Pt/SiO₂ catalysts, the CH _x coverage (H₂/CO = 10, 210 °C, 1.85 bar) was plotted versus the CH₄ and C₅₊-selectivities both determined via GC analyses (H₂/CO = 2, 220 °C, 1 bar) in Fig. 7.

In this figure a clear trend of increasing C₅₊-selectivity and decreasing C₁-selectivity with increasing CH _x coverage is visible. These trends in selectivity might be rationalized by a higher C–C coupling probability with higher CH _x coverages leading to a higher C₅₊- and lower C₁-selectivity [44].