E ﬀ ect of Operating Temperature, Pressure and Potassium Loading on the Performance of Silica-Supported Cobalt Catalyst in CO 2 Hydrogenation to Hydrocarbon Fuel

: Potassium (1–5 wt.%)-promoted and unpromoted Co / SiO 2 catalysts were prepared by impregnation method and characterized by nitrogen physisorption, temperature-programmed reduction (TPR), CO 2 temperature-programmed desorption (TPD), X-ray di ﬀ raction (XRD) and X-ray photoelectron spectroscopy (XPS) techniques. They were evaluated for CO 2 hydrogenation in a ﬁxed bed reactor from 180 to 300 ◦ C within a pressure range of 1–20 bar. The yield for hydrocarbon products other than methane (C 2 + ) was found to increase with an increase in the operating temperature and went through a maximum of approximately 270 ◦ C. It did not show any signiﬁcant dependency on the operating pressure and decreased at potassium loadings beyond 1 wt.%. Potassium was found to enhance the catalyst ability to adsorb CO 2 , but limited the reduction of cobalt species during the activation process. The improved CO 2 adsorption resulted in a decrease in surface H / C ratio, the latter of which enhanced the formation of C 2 + hydrocarbons. The highest C 2 + yield was obtained on the catalyst promoted with 1 wt.% of potassium and operated at an optimal temperature of 270 ◦ C and a pressure of 1 bar.


Introduction
The promoting capabilities of alkali metals, namely potassium, have been investigated for a variety of catalysts and reactions, including steam reforming of bioethanol [1], water gas shift [2], N 2 O decomposition [3], Fischer-Tropsch synthesis (FTS) [4][5][6] and CO 2 hydrogenation [7][8][9][10][11]. One of the earliest studies on the use of potassium as a promoter for the catalyst used in CO 2 hydrogenation to hydrocarbons is that of Russell and Miller [12]. They investigated several copper-activated cobalt catalysts at atmospheric pressure from 448 to 573 K with H 2 /CO 2 ratio varied from 2 to 3. All the catalysts mainly produced methane and liquid hydrocarbons were observed only after potassium addition to the catalyst in the form of either potassium carbonate or phosphate. Potassium was believed to selectively poison methane forming centres, and therefore, promote methylene radicals polymerization by the repression of the competitive hydrogenation reaction. Similarly, Owen et al. [13] studied the effect of potassium, along with that of lithium and sodium, on the performance of Co/SiO 2 catalysts. The catalytic testing was carried out at 643 K, atmospheric pressure and using an H 2 /CO 2 ratio of 3. They showed that with an alkali loading as low as 1 wt.%, the products distribution shifts towards longer chain hydrocarbons. Furthermore, C 2 and C 3 olefins, which did not form over the unpromoted catalyst, were detected in relatively significant amounts over the promoted catalysts. The authors attributed this behaviour to the ability of potassium to enhance the surface to molecule charge transfer, resulting in increased CO and reduced hydrogen binding strength. These findings were further corroborated by a more recent investigation by Shi et al. [8] on a CoCu/TiO 2 system containing 1.5-3.5 wt.% K. Using CO 2 temperature-programmed desorption, the authors were able to link an improved yield of liquid hydrocarbons (C 5+ ) to the increased CO 2 adsorption capacity of the catalyst, when loaded with potassium.
It appears that potassium has an enormous potential in the conversion of CO 2 to liquid hydrocarbons. To derive most of the benefit from this promoter, the study of its effect on the reaction must be integrated with that of the effect of operating conditions. Most studies have reported the effect of potassium on cobalt-based catalysts under pre-selected operating conditions that were not optimized. Hence, the present study aims at systematically evaluating the promoting effect of potassium on a Co/SiO 2 system used in CO 2 hydrogenation under optimized temperature and pressure conditions.

Surface Area and Porosity
The information on the surface area and porosity of the catalysts investigated is presented in Table 1. The data show that cobalt incorporation into the silica support results in a significant drop in the surface area from 186.6 to 133.1 m 2 /g. This behaviour is generally explained by the growth of cobalt oxide particles within the pores of the support during catalyst calcination, leading to some level of pore obstruction. This agrees well with the pore volume data, which show a decrease from 1.5 to 1.0 cm 3 /g. The introduction of potassium, in amounts above 3% in the catalyst, further amplifies this phenomenon. Table 1. Surface area and porosity data.

Sample
BET Surface Area (m 2 /g) Pore Volume (cm 3 Figure 1 shows the XRD patterns of unpromoted and promoted catalysts before and after reduction. All the unreduced catalysts showed diffraction peaks at 2θ values of approximately 18 • [14].

X-ray Diffraction
After catalysts reduction, the diffraction peaks for Co 3 O 4 , which were present in the unreduced catalysts disappeared (Figure 1b) Table 2. Although there is no observable trend in the data with respect to Co 3 O 4 and Co, it appears that the average crystallite size for CoO decreases with increasing potassium loading in the catalyst. This suggests that potassium controls the size of CoO in the catalyst. After catalysts reduction, the diffraction peaks for Co3O4, which were present in the unreduced catalysts disappeared (Figure 1b). The only visible peaks are those of the lower oxide of cobalt (CoO) at 42.4° and metallic cobalt at 44.5°.
The Scherrer equation was used to calculate the average crystallite sizes of cobalt species in the catalyst, using two theta values of 36.6°, 42.4° and 44.5° for Co3O4, CoO and Co respectively. The data are reported in Table 2. Although there is no observable trend in the data with respect to Co3O4 and Co, it appears that the average crystallite size for CoO decreases with increasing potassium loading in the catalyst. This suggests that potassium controls the size of CoO in the catalyst.

Temperature-Programmed Reduction (TPR)
The effect of potassium addition on the reducibility of silica-supported cobalt catalysts was investigated using TPR analysis. TPR profiles of various potassium-promoted catalysts, along with that of the unpromoted sample are presented in Figure 2. For the unpromoted catalyst, an early and slow reduction process was observed from ca. 170 °C . It became significant from ca. 290 °C, where a fast reduction peak was observed to start and went through a maximum at 365 °C . Subsequent overlapping reduction peaks, with respective maxima at ca. 395, 425 and 466 °C , were also observed. These peaks can be attributed to a two-step reduction of Co3O4 species in the catalyst to CoO and Co 0 . The presence of more than two peaks observed for this reduction process could indicate that not all the cobalt species in the catalyst underwent reduction at the same time. For example, as N2 adsorption data suggest some level of pore obstruction in the catalyst, it is possible that some cobalt species only got reduced after the reduction of some of those that blocked some pores. Adding potassium to the 15

Temperature-Programmed Reduction (TPR)
The effect of potassium addition on the reducibility of silica-supported cobalt catalysts was investigated using TPR analysis. TPR profiles of various potassium-promoted catalysts, along with that of the unpromoted sample are presented in Figure 2. For the unpromoted catalyst, an early and slow reduction process was observed from ca. 170 • C. It became significant from ca. 290 • C, where a fast reduction peak was observed to start and went through a maximum at 365 • C. Subsequent overlapping reduction peaks, with respective maxima at ca. 395, 425 and 466 • C, were also observed. These peaks can be attributed to a two-step reduction of Co 3 O 4 species in the catalyst to CoO and Co 0 . The presence of more than two peaks observed for this reduction process could indicate that not all the cobalt species in the catalyst underwent reduction at the same time. For example, as N 2 adsorption data suggest some level of pore obstruction in the catalyst, it is possible that some cobalt species only got reduced after the reduction of some of those that blocked some pores. Adding potassium to the catalyst reduced the reducibility of cobalt species as per the following observations: (i) The reduction temperatures for the catalysts shifted to higher values. For example, the start of the reduction process moved from 170 • C for the unpromoted catalyst to 210 and 255 • C for catalysts containing 1% and 3-5% K respectively; (ii) the area under the TPR profile below 500 • C decreased, indicating lower degree of catalyst reduction as the amount of potassium increased in the catalyst and (iii) the formation of cobalt species in strong interaction with the support, as shown by a broad reduction peak, with a maximum at ca. 512 • C, observed in the catalyst containing 5% K. The negative effect of potassium on the reduction of cobalt catalyst was also reported by Jacobs et al. [6] who found that (0.5-5%) K shifted Catalysts 2019, 9, 807 4 of 18 the reduction peak temperatures to higher values and lowered the extent of catalyst reduction. This suggests that potassium interacts with the cobalt species and possibly the silica support [15]. moved from 170 °C for the unpromoted catalyst to 210 and 255 °C for catalysts containing 1% and 3-5% K respectively; (ii) the area under the TPR profile below 500 °C decreased, indicating lower degree of catalyst reduction as the amount of potassium increased in the catalyst and (iii) the formation of cobalt species in strong interaction with the support, as shown by a broad reduction peak, with a maximum at ca. 512 °C, observed in the catalyst containing 5% K. The negative effect of potassium on the reduction of cobalt catalyst was also reported by Jacobs et al. [6] who found that (0.5-5%) K shifted the reduction peak temperatures to higher values and lowered the extent of catalyst reduction. This suggests that potassium interacts with the cobalt species and possibly the silica support [15].

CO2 Temperature-Programmed Desorption (CO2-TPD)
CO2-TPD profiles for 15%Co/SiO2 and 15%Co-1%K/SiO2 are presented in Figure 3. Both catalysts showed two desorption peaks, with the first one centred at 65 °C with near-identical areas. This lowtemperature peak can be attributed to the desorption of physically adsorbed CO2. A second peak, observed for each catalyst, was attributed to the desorption of chemisorbed CO2 and was used as an indication of the strength and amounts of basic sites in the catalyst. As expected, the data show that the addition potassium to the catalyst increases the strength and amounts of basic sites in the catalyst. This is indicated by the large and extended CO2 desorption peak, which goes through its maximum at ca. 187 °C , compared to a corresponding small peak, with a maximum at ca. 134 °C , for the unpromoted catalyst. These data agree with earlier studies [8,16] that also reported an improvement in CO2 adsorption in cobalt-based catalyst upon potassium addition.

CO 2 Temperature-Programmed Desorption (CO 2 -TPD)
CO 2 -TPD profiles for 15%Co/SiO 2 and 15%Co-1%K/SiO 2 are presented in Figure 3. Both catalysts showed two desorption peaks, with the first one centred at 65 • C with near-identical areas. This low-temperature peak can be attributed to the desorption of physically adsorbed CO 2 . A second peak, observed for each catalyst, was attributed to the desorption of chemisorbed CO 2 and was used as an indication of the strength and amounts of basic sites in the catalyst. As expected, the data show that the addition potassium to the catalyst increases the strength and amounts of basic sites in the catalyst. This is indicated by the large and extended CO 2 desorption peak, which goes through its maximum at ca. 187 • C, compared to a corresponding small peak, with a maximum at ca. 134 • C, for the unpromoted catalyst. These data agree with earlier studies [8,16] that also reported an improvement in CO 2 adsorption in cobalt-based catalyst upon potassium addition.

X-ray Photoelectron Spectroscopy (XPS)
XPS spectra, in the Co 2p region, for calcined and activated catalysts are shown in Figure 4.

X-ray Photoelectron Spectroscopy (XPS)
XPS spectra, in the Co 2p region, for calcined and activated catalysts are shown in Figure 4.

X-ray Photoelectron Spectroscopy (XPS)
XPS spectra, in the Co 2p region, for calcined and activated catalysts are shown in Figure 4.  The Co 2p 1/2 and 2p 3/2 peaks for the calcined and unpromoted catalyst were respectively observed at ca. 795.2 and 779.6 eV and are characteristic of Co 3 O 4 [14,17], in agreement with XRD data, discussed in Section 2.2. A shift to lower binding energies can be observed for Co 2p 1/2 (to 793.5 eV) and 2p 3/2 (to 778 eV) following catalyst promotion with potassium. This suggests an electronic donation by potassium as also observed by other studies, where potassium was added to Co/Al 2 O 3 [18] and Pd/Co 3 O 4 and Co 3 O 4 [19] catalysts.
Spectra of reduced catalysts (Figure 4c,d) display features of CoO with broader Co 2p 1/2 and 2p 3/2 peaks and increased intensities of the shake-up satellite features [17,20]. They look similar for both the unpromoted and the potassium-promoted catalysts. These findings indicate that the electronic properties of cobalt in the catalyst were modified by potassium during the calcination process, not during catalyst reduction, causing a different reduction behaviour for the promoted catalyst.

Effect of Temperature
In order to study the effect of temperature, CO 2 hydrogenation was carried out over a 15%Co-3%K/SiO 2 catalyst at atmospheric pressure from 180 to 300 • C. The temperature dependency of CO 2 conversion and its corresponding Arrhenius plot are reported in Figure 5. As expected, the CO 2 conversion continuously increased from 0.6 to 18.4% as the temperature was raised from 180 to 300 • C, in agreement with other earlier studies [21][22][23]. The Co 2p1/2 and 2p3/2 peaks for the calcined and unpromoted catalyst were respectively observed at ca. 795.2 and 779.6 eV and are characteristic of Co3O4 [14,17], in agreement with XRD data, discussed in Section 2.2. A shift to lower binding energies can be observed for Co 2p1/2 (to 793.5 eV) and 2p3/2 (to 778 eV) following catalyst promotion with potassium. This suggests an electronic donation by potassium as also observed by other studies, where potassium was added to Co/Al2O3 [18] and Pd/Co3O4 and Co3O4 [19] catalysts.
Spectra of reduced catalysts (Figure 4c,d) display features of CoO with broader Co 2p1/2 and 2p3/2 peaks and increased intensities of the shake-up satellite features [17,20]. They look similar for both the unpromoted and the potassium-promoted catalysts. These findings indicate that the electronic properties of cobalt in the catalyst were modified by potassium during the calcination process, not during catalyst reduction, causing a different reduction behaviour for the promoted catalyst.

Effect of Temperature
In order to study the effect of temperature, CO2 hydrogenation was carried out over a 15%Co-3%K/SiO2 catalyst at atmospheric pressure from 180 to 300 °C. The temperature dependency of CO2 conversion and its corresponding Arrhenius plot are reported in Figure 5. As expected, the CO2 conversion continuously increased from 0.6 to 18.4% as the temperature was raised from 180 to 300 °C, in agreement with other earlier studies [21][22][23]. From the Arrhenius plot, activation energy of 78 kJ/mol was obtained in a temperature range of 180 to 240 °C. A marked curvature in the Arrhenius plot can be observed at temperatures above 240 °C , suggesting that the catalyst surface underwent some changes, possibly including deactivation by carbon [23]. For comparison, activation energies reported by earlier studies involving cobalt catalysts for CO2 hydrogenation are summarised in Table 3. From the Arrhenius plot, activation energy of 78 kJ/mol was obtained in a temperature range of 180 to 240 • C. A marked curvature in the Arrhenius plot can be observed at temperatures above 240 • C, suggesting that the catalyst surface underwent some changes, possibly including deactivation by carbon [23]. For comparison, activation energies reported by earlier studies involving cobalt catalysts for CO 2 hydrogenation are summarised in Table 3.
The value of the activation energy obtained in this study is similar to most of those reported in earlier studies. Exceptions can be noticed for the data reported by Weatherbee and Bartholomew [23], who reported activation energies of 93 and 171 kJ/mol over 15%Co/SiO2 at 1 and 10 bar respectively. The effect of the operating temperature on products selectivity and yields is summarized in Figure 6. The methane selectivity showed relatively little dependency on temperature from 180 to 225 • C, but continuously increased from 240 to 300 • C, while the selectivity to CO decreased almost linearly with increasing temperature (Figure 6a). Both C 2 -C 4 and C 5+ selectivities increased as the temperature was raised and went through a maximum at 240 • C before decreasing. Figure 6c shows that up to 270 • C, the yields for C 2+ , CH 4 and CO all increased with the rise in temperature, with the yield of CO remaining the highest of the three. The yield of methane and C 2+ hydrocarbons were similar up to 240 • C, above which the yield of methane quickly surpassed that of C 2+ in an exponential manner. * S1 and S3 are carbon supports obtained from saran copolymer. The difference between the two is in the burn-off percentage, i.e., 0 and 20% for S1 and S3 respectively [26].  * S1 and S3 are carbon supports obtained from saran copolymer. The difference between the two is in the burn-off percentage, i.e., 0 and 20% for S1 and S3 respectively [26].
The value of the activation energy obtained in this study is similar to most of those reported in earlier studies. Exceptions can be noticed for the data reported by Weatherbee and Bartholomew [23], who reported activation energies of 93 and 171 kJ/mol over 15%Co/SiO2 at 1 and 10 bar respectively.
The effect of the operating temperature on products selectivity and yields is summarized in Figure 6. The methane selectivity showed relatively little dependency on temperature from 180 to 225 °C , but continuously increased from 240 to 300 °C, while the selectivity to CO decreased almost linearly with increasing temperature (Figure 6a). Both C2-C4 and C5+ selectivities increased as the temperature was raised and went through a maximum at 240 °C before decreasing. Figure 6c shows that up to 270 °C, the yields for C2+, CH4 and CO all increased with the rise in temperature, with the yield of CO remaining the highest of the three. The yield of methane and C2+ hydrocarbons were similar up to 240 °C, above which the yield of methane quickly surpassed that of C2+ in an exponential manner.   The rise in CO yield flattened off around 285 • C and was surpassed by the fast-rising yield of methane around 290 • C. The C 2+ yield went through a maximum at 270 • C, indicating that, above this temperature, the reaction is turning into a preferential methanation process.
The mechanism of CO 2 hydrogenation to hydrocarbons is still subject of some controversies. However, since CO formed during CO 2 hydrogenation, it is most likely that hydrocarbons formed via a typical Fischer-Tropsch mechanism. Indeed, this is a plausible explanation, since some studies [27][28][29] have shown that, in the presence of CO, on cobalt-based catalysts, CO 2 behaves like an inert gas and only reacts when CO is depleted. Also, the rapid increase in methane yields with the temperature at values above 240 • C is typical to FT reaction [22,23]. An operating temperature of 270 • C was selected as optimal for the rest of the study.

Effect of Pressure
The effects of pressure on CO 2 conversion, and products selectivities and yields are reported in Figure 7. An increase in operating pressure, from 1 to 15 bar, resulted in an increase in CO 2 conversion. As can be seen from Figure 7a, the CO 2 conversion measured at 1 bar was ca. 12.5%; it increased to ca. 21%, 22%, and 27% when the pressure was increased to 5, 10 and 15 bar, respectively. Further increase in the operating pressure to 20 bar resulted in a slight decrease of CO 2 conversion to ca. 26%. The increase in CO 2 conversion with the operating pressure from 1 to 15 bar was expected because of an increase in reactants partial pressures. However, the decrease in CO 2 conversion observed when the operating pressure was increased from 15 to 20 bar was not expected; it is possible that some CO 2 or reaction intermediate species irreversibly adsorbed on the catalyst surface, blocking some active sites.
An increase in operating pressure from 1 to 5 bar significantly decreased the selectivities to CO and C 2+ hydrocarbons from ca. 48% and 21% to 8% and 11%, respectively (Figure 7b). Further increase in pressure only resulted in slight decreases in CO and C 2+ hydrocarbons selectivities. An opposite behaviour was observed for CH 4 selectivity, which increased from 30 to 81% when the operating pressure was increased from 1 to 5 bar. Further increase in operating pressure resulted in a relatively slight increase in CH 4 selectivity. Similar trends can be observed for CO and CH 4 yields as a function of the operating pressure ( Figure 7c); however, the C 2+ yield was not significantly affected by changes in operating pressure. It remained between 2.1% and 2.7% over the range of pressure used. Under these conditions, operating at 1 bar is optimal.  Figure 8 shows the effect of potassium addition on CO2 conversion, and products selectivity and yields. It is observed that the presence of potassium at a loading of as low as 1% results in a significant drop in CO2 conversion from 39 to 16% when compared to the unpromoted catalyst (Figure 8a).  Figure 8 shows the effect of potassium addition on CO 2 conversion, and products selectivity and yields. It is observed that the presence of potassium at a loading of as low as 1% results in a significant drop in CO 2 conversion from 39 to 16% when compared to the unpromoted catalyst (Figure 8a). Adding more potassium further exacerbates this behaviour, but with an attenuated effect. The following can explain these observations: (i) Coverage of active sites by potassium, although considered to happen at a low extent because of the low potassium loading employed; (ii) increase in CO2 adsorption capacity: As discussed earlier in Section 2.4, CO2-TPD results have shown that the CO2 adsorption capacity for the catalyst improves upon potassium addition. As a consequence, the H/C molar ratio on the catalyst surface is decreased, leading to a drop in CO2 conversion. This is in agreement with numerous experimental [30][31][32][33] and theoretical [34,35] investigations that reported a drop in CO2 conversion with a decrease in H2/CO2 molar ratio. The decrease in CO2 conversion with potassium addition was also reported by Owen et al. [13] and Shi et al. [8] on Co/SiO2 and CoCu/TiO2 catalysts, respectively. Using H2 and CO2 temperature-programmed desorption analyses, Shi et al. [8] were able to show that potassium promotion decreases the H2 adsorption capacity of the catalyst, while that of CO2 is enhanced; (iii) the oxidation state of cobalt in the catalyst: XRD results have shown the presence of CoO and metallic cobalt phases in all the reduced and passivated catalysts. TPR analyses, on the other hand, confirmed that these catalysts have different reducibility properties. Promotion with potassium limits the reducibility of the catalysts, resulting in limited amounts of Adding more potassium further exacerbates this behaviour, but with an attenuated effect. The following can explain these observations: (i) Coverage of active sites by potassium, although considered to happen at a low extent because of the low potassium loading employed; (ii) increase in CO 2 adsorption capacity: As discussed earlier in Section 2.4, CO 2 -TPD results have shown that the CO 2 adsorption capacity for the catalyst improves upon potassium addition. As a consequence, the H/C molar ratio on the catalyst surface is decreased, leading to a drop in CO 2 conversion. This is in agreement with numerous experimental [30][31][32][33] and theoretical [34,35] investigations that reported a drop in CO 2 conversion with a decrease in H 2 /CO 2 molar ratio. The decrease in CO 2 conversion with potassium addition was also reported by Owen et al. [13] and Shi et al. [8] on Co/SiO 2 and CoCu/TiO 2 catalysts, respectively. Using H 2 and CO 2 temperature-programmed desorption analyses, Shi et al. [8] were able to show that potassium promotion decreases the H 2 adsorption capacity of the catalyst, while that of CO 2 is enhanced; (iii) the oxidation state of cobalt in the catalyst: XRD results have shown the presence of CoO and metallic cobalt phases in all the reduced and passivated catalysts. TPR analyses, on the other hand, confirmed that these catalysts have different reducibility properties. Promotion with potassium limits the reducibility of the catalysts, resulting in limited amounts of metallic cobalt sites for CO 2 conversion. A similar relationship between catalyst reducibility and CO 2 activity was reported by Melaet et al. [36] who conducted CO 2 hydrogenation on Co/SiO 2 catalysts activated at 523 K and 723 K. They established, by means of XPS, that CoO and metallic cobalt formed upon activation at 523 K and 723 K respectively. The catalyst reduced at 723 K showed higher activity.

Effect of Potassium Addition
The selectivity towards methane decreased from 96 to 37.6% (Figure 8b) upon adding 1 % of potassium to the catalyst. Meanwhile, the selectivity of both CO and C 2+ hydrocarbons increased. Additional amounts of potassium resulted in a further decrease in methane selectivity and increase in CO selectivity. The selectivity of C 2+ hydrocarbons, on the other hand, decreased with further increase in potassium loading above 1%. The improvement of C 2+ hydrocarbons selectivity at 1% potassium can be attributed to a decreased surface H/C ratio as discussed earlier. This implies that carbon-containing species from CO 2 dissociation can polymerize rather than being hydrogenated as is often the case in a hydrogen-rich environment.
The decreased C 2+ hydrocarbons selectivity at 3% and 5% potassium loading is also explained by an increased CO 2 adsorption capacity of the catalyst, causing a decrease in the surface H/C ratio. Since the CO yield is shown to increase and undergo little variations with further increase in potassium loading, while both CH 4 and the C 2+ yields decrease (Figure 8c), it is possible that there is not enough surface hydrogen to readily react with both CO 2 and CO on the catalyst surface.
The highest C 2+ yield achieved in this study was ca. 5% and was measured on the 15%Co/1%K/SiO 2 catalyst at 1 bar and 270 • C. This condition is compared to results reported in other studies that used cobalt-based catalysts for CO 2 hydrogenation under various conditions, as summarized in Table 4. Of the catalysts that produced C 5+ products at low pressure (<2 bar), our catalyst had the lowest methane selectivity under the optimized operating temperature and pressure. This is particularly important, since it offers opportunities to limit the formation of the undesirable methane without the need for excessive operating pressures that will make the process more energy-intensive.

Catalyst Synthesis
The catalyst synthesis consisted essentially of two steps, namely support preparation and metal loading. Fumed silica with an average particle size range of 0.2-0.3 µm, supplied by Sigma-Aldrich South Africa, served as the catalyst supporting material. Given its small particle size, it was pre-treated with deionized water and agglomerated by drying overnight at 120 • C in the air before crushing and sieving to obtain a powder with particles within the size range of 212-500 µm. The powder so obtained was subsequently calcined at 400 • C for 6 h in the air to lock its properties before loading the metals. The addition of cobalt and potassium was done through co-impregnation with solutions of cobalt and potassium nitrates-both purchased from Sigma-Aldrich. After impregnation, the catalysts were dried overnight at 120 • C and calcined at 400 • C in the air for 6 h. All the prepared catalysts contained 15 wt.% cobalt with varying potassium loading (0-5 wt.%). The amount of silica used in catalyst preparation was reduced to account for the addition of potassium. This allowed for the cobalt loading to be kept constant for all the catalysts.

Catalyst Characterization
The surface area and the porosity of the catalysts were measured by nitrogen physisorption at −196 • C using an Accelerated Surface Area and Porosimetry System (ASAP 2460) from Micromeritics. Each analysis was preceded by degassing the sample at 150 • C for 4 h. The multipoint Brunauer-Emmett-Teller (BET) method was used to determine the surface area of the materials analysed.
The reducibility of the catalysts was studied by means of temperature-programmed reduction (TPR). An in-house built instrument, equipped with a thermal conductivity detector (TCD), was used for this purpose. In a typical analysis, 100 mg of catalyst was loaded in a stainless-steel reactor and heated to 300 • C for one hour under 70 NmL/min of helium to remove traces of moisture and other ambient contaminants. This step was referred to as degassing. After allowing the reactor to cool to room temperature, helium was switched with a gas mixture containing 5% H 2 in argon at a flow of 65 NmL/min. In the final step, the temperature was raised from room temperature to 700 • C at a heating rate of 10 • C/min, while recording the signal of the TCD.
Temperature-programmed desorption (TPD) of CO 2 was carried out using the same instrument as described for TPR analysis. Different to TPR analysis, the catalysts used in this analysis were first reduced at 335 • C for 17 h, using the same reactor and conditions as for the reduction of catalyst samples used in the CO 2 hydrogenation testing as will be described in Section 2.3. The reduced catalysts were passivated using 5% O 2 in helium for 2 h at ambient temperature before their transfer from the CO 2 hydrogenation reactor to the TPD apparatus. Two hundred milligrams of catalyst sample was degassed in a similar manner as for TPR analysis. After degassing and cooling to room temperature, the temperature was raised to 335 • C at a heating rate of 10 • C/min and maintained at this value for 30 min under a flow of 5% H 2 in argon. This step was necessary for the removal of the cobalt oxide layer formed during catalyst passivation. Thereafter, the reactor was cooled and maintained at 50 • C for at least 10 min before replacing H 2 (5% in argon) with CO 2 (10% in helium). CO 2 adsorption was performed at 50 • C for 1 h before re-introducing helium, but this time to remove the physically adsorbed CO 2 molecules. TPD was then performed under helium flow, after stabilization of the TCD signal, from 50 to 700 • C at a heating rate of 5 • C/min. X-ray diffraction (XRD) analysis was performed to identify the oxidation state of cobalt species in the unreduced and reduced catalyst samples. The instrument used for this purpose was a Rigaku Ultima IV equipped with a copper target. The voltage and current at which the diffractometer was operated were 40 kV and 30 mA respectively. Spectra were acquired in the range of 2θ from 10 • to 90 • with a step size of 0.01 • at the scanning speed of 1 • /min. X-ray photoelectron spectroscopy (XPS) was used to determine the oxidation states of the elements present on the surface of the catalysts. This analysis was performed on a Specs Phoibos 150 spectrometer with a monochromatic X-ray source Al Kα at 1486.71 eV. A low-energy electron flood gun operated at 2.0-2.5 eV and 20 µA was used to stabilize the sample surface charge. The spectrometer was operated at constant pass energy of 40 eV. The shift in binding energy peaks position, due to the surface charging effect was corrected by setting the C 1s binding energy to 284.8 eV [14].

Catalyst Testing
Carbon dioxide hydrogenation was carried out in a system which consisted mainly of a stainless steel fixed-bed reactor (16 mm i.d. × 220 mm length) mounted in an electrical furnace, a mass flow controller (Aalborg), a back-pressure regulator and a product collection pot. The furnace temperature was controlled using a programmable temperature controller connected to a K-type thermocouple and the furnace heating element. Accurate reaction temperatures were measured by means of another K-type thermocouple in direct contact with the catalyst bed held in place by plugs of quartz wool. Any liquid product formed was collected in a cold pot mounted at the bottom of the reactor. There was no need for a hot trap since the products were mainly light hydrocarbons. The reactor outlet was connected to a three-way valve, which made it possible to either send the reaction products to vent or to an online gas chromatograph (GC) for analysis. The Dani Master GC used in this study was equipped with a flame ionization detector (FID) connected to a capillary column (Supel-Q TM PLOT) that separated hydrocarbons and oxygenates, and a thermal conductivity detector (TCD) connected to a packed column (60/80 Carboxen 1000) for the separation of H 2 , N 2 , CO and CO 2 .
Prior to testing, 500 mg of catalysts were reduced in flowing hydrogen (23 NmL/min) at 335 • C and atmospheric pressure for 17 h. Catalyst testing was done at temperatures ranging from 180 to 300 • C with an increment of 15 • C and at pressures within a range of 1-20 bar at a space velocity of 0.92 NL/g cat /h. The feed gas was premixed and contained 21.8% CO 2 , 68.6% H 2 and 9.6% N 2 . After testing, all catalysts were passivated in 5% O 2 in helium (23 NmL/min) at room temperature for 2 h. The nitrogen present in the feed gas was used as an internal standard for mass balance calculations. The CO 2 conversion, the rate of CO 2 conversion, the rate of products formation, selectivity and yield were calculated according to Equations (1)-(5), where F and X indicate the total molar gas flow rate and mole fraction respectively. The subscripts "in" and "out" refer to the gas streams entering or leaving the reactor. CO 2 conversion (%) = X CO 2, in − X N 2, in X N 2, out × X CO 2, out Rate of CO 2 conversion = F in X CO 2, in − X N 2, in X N 2, out × X CO 2, out Catalyst mass , Rate of formation of product i = F out × X i, out Catalyst mass , Selectivity of product i (%) = moles of carbon in product i per unit time Rate of CO 2 conversion × Catalyst mass ×100 , Yield of product i (%) = Selectivity of product i × CO 2 conversion 100 .
After a change in operating conditions or in catalyst sample, the reactor was allowed to reach a steady state and maintained at the new conditions for at least two days. At least two data points were generated per day. To ensure reproducibility, each data point was an average of three independent measurements that were closer to each other within 5% error range.

Conclusions
The aim of this study was to investigate the effects of operating conditions (temperature, pressure) and potassium loading on the performance of silica-supported cobalt catalysts in CO 2 hydrogenation. The highest yield in C 2+ hydrocarbons was measured at 1 bar and 270 • C. Potassium was found to negatively affect the reducibility of the catalyst, while enhancing its CO 2 adsorption capacity. The improved CO 2 adsorption capacity of the catalyst leads to a lower surface H/C ratio, which promotes chain growth reactions. The limited catalyst reducibility resulted in low catalyst activity and is explained by an electric donation of potassium to cobalt species during the calcination process of the catalyst. The optimal operating pressure and temperature determined in this study, combined with catalyst promotion with 1 wt.% of potassium, significantly lowered the undesirable methane selectivity when compared to other cobalt-based catalysts that also produced some C 5+ hydrocarbons at low pressures (<2 bar). This constitutes a significant further step in the development of efficient catalysts for CO 2 hydrogenation to liquid fuels.