Effect of Pre-Treatment Conditions on the Activity and Selectivity of Cobalt-Based Catalysts for CO Hydrogenation

We investigated the effect of pre-treatment conditions on the activity and selectivity of cobalt catalysts for Fischer–Tropsch synthesis (FTS) by varying both the reduction atmosphere and the reduction temperature. Catalysts supported on SiO2, Al2O3, and TiO2, prepared via incipient wetness impregnation, were evaluated, and activation temperatures in the range 250–350 °C were considered. Activation with syngas led to a better product selectivity (low CH4, high selectivity to liquid hydrocarbons, and low paraffin to olefin ratio (P/O)) than the catalysts reduced in H2 at lower activation temperatures. The CoxC species suppressed the hydrogenation reaction, and it is hypothesised that this resulted in the high selectivity of olefins observed for the syngas pre-treated catalysts. On the basis of the experimental results, we postulated that a synergistic effect between Co0 and CoxC promotes the production of the long chain hydrocarbons and suppresses the formation of CH4. In addition, for systems aimed at producing lower olefins, syngas activation is recommended, and for the FTS plants that focus on maximising the production of higher molecular weight products, H2 activation might be considered. These results provide insights for the future FTS catalyst design and for target product-driven operations.


Introduction
Fischer-Tropsch synthesis (FTS) is a structure-sensitive reaction that converts syngas derived from natural gas, coal, and biomass to valuable chemicals and synthetic fuels over a metal-based catalyst [1]. Cobalt (Co) catalysts have attracted more attention in the recent years due to their high intrinsic hydrogenation activity, selectivity towards liquid hydrocarbons, and lower water gas shift (WGS) activity than iron and lower costs compared to noble metals [2,3]. Currently silica (SiO 2 ), alumina (Al 2 O 3 ), and titania (TiO 2 ) are used for commercial FTS operations [4].
The hydrogenation activity of the cobalt metal (Co), which is recognised as the active phase, is highly dependent on its structure. Co particles that are hexagonally packed (hcp) are found to be more active than the face-centred cubic (fcc) structure [5,6]. Evidence from previous studies suggests that the Co particle size is influenced by the support pore size [7][8][9]. Borg [10] studied the dependency of the Co particle size on the Al 2 O 3 -support pore diameter and found that: (i) large Co particles were formed in the large pores and smaller ones formed in the narrow pores, (ii) the degree of reduction increased with the pore size, and (iii) the C 5+ (the long chain hydrocarbons with carbon numbers equal or higher than 5) selectivity also increased with the pore size.
In some cases, the interaction between these supports and the metal can be too strong, which may leave a fraction of the cobalt chemically inactive after reduction. For example, Jacobs et al. [11] reported a lower degree of reduction for Al 2 O 3 -and TiO 2 -supported catalysts due to high metal-support interactions compared to the SiO 2 -support. Strong metal-metal oxide interactions have been demonstrated to play an important role in the 3 h and calcination at 350 • C for 8 h. For, the detailed preparation procedure, please refer to our earlier publication [27].

Catalyst Characterisation
Brunauer-Emmet-Teller (BET) experiments were conducted on the fresh catalysts prior to reduction or reaction to determine the sample surface area and pore size. BET experiments followed the usual procedure. The sample was firstly subjected to a degassing chamber at 200 • C for 6 h, and treatment was performed at a relative pressure of 0.99 (P a /P 0 = 0.99, where P a is the actual gas pressure and P 0 is the vapor pressure of the adsorbing gas) to obtain the pore volume and −196 • C to obtain the surface area and porosity by nitrogen physisorption. Furthermore, the Barrett-Joyner-Halenda (BJH) method was used to obtain the pore sizes from the desorption branches on the isotherms.
The morphology of the catalysts was characterised by transmission electron microscopy (TEM). The samples for TEM studies were prepared by ultrasonic dispersion of the catalysts in ethanol, and the suspensions were added dropwise onto a copper grid. The TEM investigations were carried out using a JEOL-JEM-100CX II (100 kV) transmission electron microscope equipped with a NARON energy-dispersive spectrometer and a germanium detector.
X-ray diffraction (XRD) studies were performed using a Philips PW 1710 spectrometer with monochromatic Cu-Kα radiation to determine the catalyst particle size and crystalline structures. The measurements were made on calcined catalysts and the average Co 3 O 4 particle size was calculated from the most intense peak, with the use of Scherrer formula [6], for each catalyst.
The reduction behaviour and the interaction between the active phase and the support of each catalyst were examined using the temperature programmed reduction (TPR) technique. The TPR experiments were carried out with a thermal conductivity detector (TCD) to determine the hydrogen consumption. The catalyst (500 mg) was placed in a quartz tubular reactor fitted with a thermocouple for continuous temperature measurements. The reactor was heated with a furnaced designed for the TPR machine, at a ramping rate of 10 • C/min, under a mixture of 5 vol % H 2 in an air flow of 30 cm 3 /min.

Reduction and Reaction Procedures and Data Analysis
Three fixed bed reactors with the same size (ID = 8 mm) were used in this study. One gram of Co/SiO 2 , Co/Al 2 O 3 , and Co/TiO 2 catalysts was loaded into the three respective reactors. The three catalysts followed the same activation procedures: to reduce in a flow of either H 2 or syngas (H 2 /CO ratio of 2) at two different temperatures, namely, 250 • C and 350 • C, and atmosphere pressure. After catalyst reduction, the catalysts were cooled to 180 • C. Thereafter, the same syngas used for catalyst reduction was introduced for FTS. The catalyst reactivity and product distribution were evaluated at 20 bar, 210 • C, and 60 mL/min with syngas.
The tail gases from the three reactors were monitored and analysed by an online GC (Agilent 7890B): the hydrocarbon products were analysed by a flame ionisation detector (FID), whist the other gases (H 2 , CO, N 2 , and CO 2 ) were analysed by two TCDs. Table 1 lists the physical properties of the catalysts. It shows that the Al 2 O 3 -supported catalyst had a larger pore size and a larger particle size than the catalysts supported on TiO 2 and SiO 2 . For further details on the catalyst physical properties, please refer to our previous publications [27].

Characteristics
The crystal morphology of the catalysts is illustrated in Figure 1, determined via TEM. In the cross sections, the visible darker dense areas represent Co 3 O 4 particles, and the lighter areas correspond to the support. The Co 3 O 4 particles seemed to be more highly dispersed on the TiO 2 -support followed by SiO 2 -support then the Al 2 O 3 -support. This Reactions 2021, 2 261 might have been due to the smaller TiO 2 -support particles that are observed in Figure 1. The large Co 3 O 4 particles observed on the alumina support suggest that the support had a wider pore, thus distributing bigger particles than TiO 2 and SiO 2 . The Co 3 O 4 particles on the Al 2 O 3 -and SiO 2 -supports seemed to be spherical shape, whilst on the TiO 2 , the cobalt particles assumed the shape of the TiO 2 -support particles, in this case cubic/rhombus-shaped. The crystal morphology of the catalysts is illustrated in Figure 1, determined via TEM. In the cross sections, the visible darker dense areas represent Co3O4 particles, and the lighter areas correspond to the support. The Co3O4 particles seemed to be more highly dispersed on the TiO2-support followed by SiO2-support then the Al2O3-support. This might have been due to the smaller TiO2-support particles that are observed in Figure 1. The large Co3O4 particles observed on the alumina support suggest that the support had a wider pore, thus distributing bigger particles than TiO2 and SiO2. The Co3O4 particles on the Al2O3-and SiO2-supports seemed to be spherical shape, whilst on the TiO2, the cobalt particles assumed the shape of the TiO2-support particles, in this case cubic/rhombusshaped. The XRD patterns for the model catalysts (Co/TiO2, Co/SiO2, Co/Al2O3) are presented in Figure 2. XRD characteristics of Co3O4 were detected for all the calcined catalysts with Co/TiO2 and Co/Al2O3, showing distinctive Co3O4 crystalline features, marked with a black circle (see Figure 2). The Co/SiO2 diffractogram showed considerably broad features, which suggests that the silica support is likely amorphous, and contains smaller Co3O4 nanoparticles. The average Co3O4 crystallite size (Table 1) was calculated from the Scherrer equation [6]. The Co3O4 crystallite size varied slightly as a function of the pore size, with Co/Al2O3 showing the biggest size (33.0 nm), owing to its large pore size, followed by Co/TiO2 (21.5 nm) and then lastly Co/SiO2 (17.0 nm). The XRD patterns for the model catalysts (Co/TiO 2 , Co/SiO 2 , Co/Al 2 O 3 ) are presented in Figure 2. XRD characteristics of Co 3 O 4 were detected for all the calcined catalysts with Co/TiO 2 and Co/Al 2 O 3 , showing distinctive Co 3 O 4 crystalline features, marked with a black circle (see Figure 2). The Co/SiO 2 diffractogram showed considerably broad features, which suggests that the silica support is likely amorphous, and contains smaller Co 3 O 4 nanoparticles. The average Co 3 O 4 crystallite size (Table 1) was calculated from the Scherrer equation [6]. The Co 3 O 4 crystallite size varied slightly as a function of the pore size, with Co/Al 2 O 3 showing the biggest size (33.0 nm), owing to its large pore size, followed by Co/TiO 2 (21.5 nm) and then lastly Co/SiO 2 (17.0 nm).
The TPR reduction profiles presented in Figure 3 for the three kinds of catalysts showed two reduction peaks, which were similar to those observed for bulk Co 3 O 4 oxide. These profiles point to a two-step reduction process: the first one of low intensity started at approximately 200 • C and overlapped with the more intense second peak whose maximum occurred at about 300 • C for Co/TiO 2 and Co/SiO 2 catalysts, and for Co/Al 2 O 3 , the peak started around 300 • C and the second peak emerged at around 450 • C. Other than the fact that the second reduction peak for the Al 2 O 3 -supported catalyst emerged at a higher temperature than the second peak on the TiO 2 -or SiO 2 -supported catalysts, it also extended its shoulder to a higher magnitude, in this case 700 • C. Therefore, the reduction process of Co 3 O 4 can be described by the reduction of Co 3+ ions present in the spinel structure of a fresh catalyst into Co 2+ with subsequent structural change to CoO, followed by the reduction of Co 2+ ions to Co 0 metal. The results observed over the Al 2 O 3 -supported sample suggests that the catalyst supported on Al 2 O 3 is harder to reduce than the one supported by TiO 2 or SiO 2 , which may have been due to strong metal-support interactions, which is in line with the literature [11,12].
The TPR reduction profiles presented in Figure 3 for the three kinds of catalysts showed two reduction peaks, which were similar to those observed for bulk Co3O4 oxide. These profiles point to a two-step reduction process: the first one of low intensity started at approximately 200 °C and overlapped with the more intense second peak whose maximum occurred at about 300 °C for Co/TiO2 and Co/SiO2 catalysts, and for Co/Al2O3, the peak started around 300 °C and the second peak emerged at around 450 °C. Other than the fact that the second reduction peak for the Al2O3-supported catalyst emerged at a higher temperature than the second peak on the TiO2-or SiO2-supported catalysts, it also extended its shoulder to a higher magnitude, in this case 700 °C. Therefore, the reduction process of Co3O4 can be described by the reduction of Co 3+ ions present in the spinel structure of a fresh catalyst into Co 2+ with subsequent structural change to CoO, followed by the reduction of Co 2+ ions to Co 0 metal. The results observed over the Al2O3-supported sample suggests that the catalyst supported on Al2O3 is harder to reduce than the one supported by TiO2 or SiO2, which may have been due to strong metal-support interactions, which is in line with the literature [11,12].
Furthermore, a higher reduction temperature was required for the reduction of the Al2O3-supported catalyst compared to TiO2-and SiO2-supported catalysts, as observed from of the reduction profiles ( Figure 3), which might have been due to the fact that the cobalt particles diffused into the Al2O3 lattice and formed the irreducible compounds, such as aluminates.

Reaction Rate
The Fischer-Tropsch (FT) activity and selectivity of the supported cobalt catalysts is illustrated in Figures 4-10. The catalysts pre-treated in H2 exhibited higher CO reaction rates compared to the samples reduced in syngas, at all reduction temperatures (see Figure 4). For the syngas pre-treatment, the CO reaction rates were found to be higher at a higher reduction temperature (350 °C) for all the samples. Similar results to the syngas pre-treatment were observed, where higher CO reaction rates were achieved at a higher Furthermore, a higher reduction temperature was required for the reduction of the Al 2 O 3 -supported catalyst compared to TiO 2 -and SiO 2 -supported catalysts, as observed from of the reduction profiles (Figure 3), which might have been due to the fact that the cobalt particles diffused into the Al 2 O 3 lattice and formed the irreducible compounds, such as aluminates.

Reaction Rate
The Fischer-Tropsch (FT) activity and selectivity of the supported cobalt catalysts is illustrated in Figures 4-10. The catalysts pre-treated in H 2 exhibited higher CO reaction rates compared to the samples reduced in syngas, at all reduction temperatures (see Figure 4). For the syngas pre-treatment, the CO reaction rates were found to be higher at a higher reduction temperature (350 • C) for all the samples. Similar results to the syngas pre-treatment were observed, where higher CO reaction rates were achieved at a higher reduction temperature for all the three catalysts reduced in H 2 (except for the SiO 2 catalyst reduced in H 2 at 250 • C).
Reactions 2021, 2, FOR PEER REVIEW 7 The CO reaction rates were found to be higher at a higher reduction temperature (350 °C) for all the samples, except for the syngas pre-treated TiO2-supported sample and H2pre-treated SiO2-supported sample. The SiO2-supported catalyst showed higher reaction rates at 250 °C compared to 350 °C (see Figure 4B). Our previous work over the SiO2 sample demonstrated the effect of Co-CoO bonding promoting the FT reaction, thus leading to a higher FT reaction rate at a lower reduction temperature (250 °C), when the CoO density is higher than the density observed at 350 °C [27].

Product Formation Rate
The overall product formation rate as a function of temperature is illustrated in Figures 5 and 6. Changing the reduction medium from H2 to syngas led to a complete change rate followed by Co/SiO2 and then Co/Al2O3, at all reduction temperatur Another observation is that the CH4 formation rate was higher for the syn ples at 350 °C, whereas for the H2-treated samples, higher CH4 formatio served at 250 °C, except for the catalyst supported on SiO2. For long chain hydrocarbons (C5+), higher formation rates were obse treated samples compared to the catalysts reduced in syngas (see Figur CO reaction rates. Figure 6 shows that: (1) for H2 reduction, higher C5 were achieved at a higher reduction temperature for both TiO2-and Al2O alysts, while the higher C5+ formation rates were obtained at a lower red were observed at a higher reduction temperature for both Al2O3-and alysts, while there was only a slight difference between the C5+ forma duction temperatures of 250 and 350 °C for the catalyst supported by The effect of syngas or H2 pre-treatment on the selectivity of the lysts as a function of temperatures is shown in Figures 7 and 8. All syn showed better selectivity (low CH4, high C5+) compared to H2-reduce    The Co/SiO 2 catalyst with the highest surface area and lowest particle size (Table 1) was the most active catalyst when it reduced in H 2 at 250 • C (see Figure 4B). The high surface area of SiO 2 -support and the lower metal-support interaction enhanced the reducibility and the dispersion of the metal. Our previous research [27] reported that the oxidised Co/SiO 2 catalyst, in H 2 at 250 • C, formed a multiphase of CoO-Co/SiO 2 , and this CoO-Co interface promoted the CO dissociation and secondary olefin hydrogenation reactions, thus leading to a higher FT reaction rate [27].
Compared with SiO 2 -and Al 2 O 3 -supported catalysts, the TiO 2 -supported catalyst presented the highest CO reaction rates, when reduced either in H 2 or syngas at all temperatures except for the H 2 -reduction at 250 • C. The lower surface area, observed via BET (Table 1) Table 1) must be related to the strong metalsupport interaction (see Figure 3) and low metal dispersion due to the large cobalt particles (Table 1). For further discussion, please refer to Section 4.

Discussion and Implications
With the aim to understand the reaction pathways observed with different pre-treatment agents at different reduction temperatures, we replotted some of the data reported in Figures 4-9 in order to highlight the important findings of this work-the results are shown in Figure 10. The syngas-treated catalysts afforded a lower CO reaction rate com- The CO reaction rates were found to be higher at a higher reduction temperature (350 • C) for all the samples, except for the syngas pre-treated TiO 2 -supported sample and H 2 -pre-treated SiO 2 -supported sample. The SiO 2 -supported catalyst showed higher reaction rates at 250 • C compared to 350 • C (see Figure 4B). Our previous work over the SiO 2 sample demonstrated the effect of Co-CoO bonding promoting the FT reaction, thus leading to a higher FT reaction rate at a lower reduction temperature (250 • C), when the CoO density is higher than the density observed at 350 • C [27].

Product Formation Rate
The overall product formation rate as a function of temperature is illustrated in Figures 5 and 6. Changing the reduction medium from H 2 to syngas led to a complete change in the formation of products. A noticeable effect was the lower CH 4 formation rate observed for all the samples treated with syngas compared to H 2 , excluding the SiO 2 -supported catalyst treated with syngas at 350 • C. Co/TiO 2 showed the lowest CH 4 formation rate followed by Co/SiO 2 and then Co/Al 2 O 3 , at all reduction temperatures (see Figure 5). Another observation is that the CH 4 formation rate was higher for the syngastreated samples at 350 • C, whereas for the H 2 -treated samples, higher CH 4 formation rates were observed at 250 • C, except for the catalyst supported on SiO 2 .
For long chain hydrocarbons (C 5+ ), higher formation rates were observed over the H 2treated samples compared to the catalysts reduced in syngas (see Figure 6) due to lower CO reaction rates. Figure 6 shows that: (1) for H 2 reduction, higher C 5+ formation rates were achieved at a higher reduction temperature for both TiO 2 -and Al 2 O 3 -supported catalysts, while the higher C 5+ formation rates were obtained at a lower reduction temperature for the SiO 2 -supported catalysts; (2) for syngas reduction, higher C 5+ formation rates were observed at a higher reduction temperature for both Al 2 O 3 -and SiO 2 -supported catalysts, while there was only a slight difference between the C 5+ formation rates at the reduction temperatures of 250 and 350 • C for the catalyst supported by TiO 2 .

Product Selectivity
The effect of syngas or H 2 pre-treatment on the selectivity of the model cobalt catalysts as a function of temperatures is shown in Figures 7 and 8. All syngas-treated samples showed better selectivity (low CH 4 , high C 5+ ) compared to H 2 -reduced samples when reduced at a lower temperature, 250 • C. Increasing the syngas reduction temperature increased the CH 4 selectivity and decreased C 5+ selectivity, whereas for the H 2 -treated samples, an increase in the reduction temperature decreased the CH 4 selectivity and increased C 5+ selectivity for all the three catalysts ( Figure 8B). An increase in the reduction temperatures caused a slight increase in the selectivity of long chain hydrocarbons.

Paraffin to Olefin (P/O) Ratio
P/O ratio is a very important factor that reflects the selectivity of the paraffin (P) and olefin (O) products-a higher P/O ratio represents the products that are more paraffinic and a lower value indicates a higher selectivity to olefinic products. P n /O n represents the parafin to olefin ratio with carbon number n. In the current work, the ratios of P 2 /O 2 (ethane/ethylene) and P 4 /O 4 (butane/butene) are reported in Figure 9 for the catalysts either reduced by syngas or H 2 at different reduction temperatures. For the Co/TiO 2 catalyst, pre-treatment with syngas (at both 250 and 350 • C) produced more olefins than paraffins compared to the H 2 pre-treatment (low P/O ratios). For Co/SiO 2 , similar results as for the Co/TiO 2 catalyst-lower P/O ratios were obtained for syngas-reduced catalysts compared to H 2 reduction at 250 • C. However, syngas pre-treatment at 350 • C produced more paraffin products over the Co/Al 2 O 3 catalyst compared to H 2 pre-treatment at a similar reduction temperature.

Discussion and Implications
With the aim to understand the reaction pathways observed with different pretreatment agents at different reduction temperatures, we replotted some of the data reported in Figures 4-9 in order to highlight the important findings of this work-the results are shown in Figure 10. The syngas-treated catalysts afforded a lower CO reaction rate compared to H 2 -treated samples ( Figure 10A). This could be attributed to the lower Co site density caused by incomplete reduction of Co 3 O 4 to metallic Co 0 . Metallic Co 0 is known to be the active phase for the conversion of syngas to hydrocarbon products [20]; therefore, the lower the Co 0 density, the lower the CO hydrogenation activity. Our findings are in line with Gnanamani et al. [28] who reported that cobalt catalysts do not reduce completely under syngas treatment.
Catalyst pre-treatment is a way to transform cobalt oxides to active sites. For H 2 reduction, cobalt oxides are reduced to metallic Co 0 ; in the meantime, there is still some cobalt oxides (Co 3 O 4 and/or CoO) left due to partial reduction depending on the reduction temperature or the extent of metal-support interactions. For syngas (a mixture of H 2 /CO) reduction, the presence of CO in the mixture can also react with the cobalt oxides to form cobalt carbides (Co x C), which has been confirmed by Peacock et al. [29] and Claeys et al. [30] using the in situ magnetometer. In addition, the Boudouard reaction (2CO = CO 2 + C) may occur when the operating temperature is high. Figure 11 lists the possible cobalt phases after cobalt catalyst reduction under different atmospheres.
Reactions 2021, 2, FOR PEER REVIEW 14 Catalyst pre-treatment is a way to transform cobalt oxides to active sites. For H2 reduction, cobalt oxides are reduced to metallic Co 0 ; in the meantime, there is still some cobalt oxides (Co3O4 and/or CoO) left due to partial reduction depending on the reduction temperature or the extent of metal-support interactions. For syngas (a mixture of H2/CO) reduction, the presence of CO in the mixture can also react with the cobalt oxides to form cobalt carbides (CoxC), which has been confirmed by Peacock et al. [29] and Claeys et al. [30] using the in situ magnetometer. In addition, the Boudouard reaction (2CO = CO2+ C) may occur when the operating temperature is high. Figure 11 lists the possible cobalt phases after cobalt catalyst reduction under different atmospheres. From our experimental data (see Figure 10B-D), with syngas reduction at 250 °C, all the three catalysts had a much lower CH4 selectivity and lower P/O ratio compared with the catalysts treated with H2 at 250 °C, which indicates that the reactions of CO hydrogenation to paraffins were suppressed during the low temperature syngas reduction. These experimental results may provide evidence that the Co2C phase promotes the formation of olefins by suppressing the olefin hydrogenation reaction. Furthermore, the existence of the metallic Co-hcp phase, obtained from further reduction of cobalt via the CoxC intermediate, could catalyse the FT chain growth reaction by converting syngas to light olefins, which in turn react to form longer chain hydrocarbons for syngas-treated catalysts reduced at 250 °C. This reaction path is not the dominant mechanism for H2reduced catalysts at these reduction temperatures. On the basis of our results, we hypothesised that cobalt in association with the cobalt carbides enhances the production of higher hydrocarbons. Jiao et al. [31] and Gnanamani et al. [28] both supported this hypothesis in that they both suggested that Co2C contributes to the selectivity of light olefin products and alcohol formation via a CO insertion mechanism. In addition, Jalama et al. [18] reported a higher olefin to paraffin ratio for the samples pre-treated in syngas than in H2, which was attributed to the presence of the Co2C phase.
The large shift in product selectivity by changing the activation conditions (as shown in Figures 5-9) led us to believe that the method of activation of cobalt as well as the corresponding temperature play vital roles in the subsequent hydrogenation of CO. A higher reduction temperature was found to increase the selectivity of C1 in detriment to all other hydrocarbons (Figure 10), when syngas was used as a reducing agent. When reducing at 350 °C, the Boudouard reaction may occur, and the carbonaceous deposits on the surface could act as methanation sites and decrease the number of Co active sites available for FTS. Findings over these catalysts suggest that carburisation and carbon deposits are feasible under syngas reduction, and that this may cause an increase in both the C5+ selectivity From our experimental data (see Figure 10B-D), with syngas reduction at 250 • C, all the three catalysts had a much lower CH 4 selectivity and lower P/O ratio compared with the catalysts treated with H 2 at 250 • C, which indicates that the reactions of CO hydrogenation to paraffins were suppressed during the low temperature syngas reduction. These experimental results may provide evidence that the Co 2 C phase promotes the formation of olefins by suppressing the olefin hydrogenation reaction. Furthermore, the existence of the metallic Co-hcp phase, obtained from further reduction of cobalt via the Co x C intermediate, could catalyse the FT chain growth reaction by converting syngas to light olefins, which in turn react to form longer chain hydrocarbons for syngas-treated catalysts reduced at 250 • C. This reaction path is not the dominant mechanism for H 2 -reduced catalysts at these reduction temperatures. On the basis of our results, we hypothesised that cobalt in association with the cobalt carbides enhances the production of higher hydrocarbons. Jiao et al. [31] and Gnanamani et al. [28] both supported this hypothesis in that they both suggested that Co 2 C contributes to the selectivity of light olefin products and alcohol formation via a CO insertion mechanism. In addition, Jalama et al. [18] reported a higher olefin to paraffin ratio for the samples pre-treated in syngas than in H 2 , which was attributed to the presence of the Co 2 C phase.
The large shift in product selectivity by changing the activation conditions (as shown in Figures 5-9) led us to believe that the method of activation of cobalt as well as the corresponding temperature play vital roles in the subsequent hydrogenation of CO. A higher reduction temperature was found to increase the selectivity of C 1 in detriment to all other hydrocarbons (Figure 10), when syngas was used as a reducing agent. When reducing at 350 • C, the Boudouard reaction may occur, and the carbonaceous deposits on the surface could act as methanation sites and decrease the number of Co active sites available for FTS. Findings over these catalysts suggest that carburisation and carbon deposits are feasible under syngas reduction, and that this may cause an increase in both the C 5+ selectivity or CH 4 selectivity depending on the amount of surface carbon available. Our findings are in line with Lee et al. [32], who reported that surface carbonaceous deposits can exist in two forms, namely, active carbon or graphitic carbon, and that the active carbon can hydrogenate to methane under normal FT conditions.
In the case of H 2 pre-treated samples, an increase in the reduction temperature from 250 to 350 • C led to the production of higher hydrocarbons. These results demonstrate that C 5+ hydrocarbon formation is a function of temperature and that a high reduction temperature is associated with a higher reducibility of Co 3 O 4 to Co 0 ; therefore, it can be deduced that Co 0 is selective to the production of C 5+ hydrocarbons.
This study also reflects on the effect of support properties on the performance of cobalt catalysts under different pre-treatment conditions. The catalyst supported on TiO 2 exhibited the highest selectivity towards liquid products with the lowest CH 4 selectivity when treated in both H 2 and syngas. This can be attributed to the higher Co site density observed via TEM ( Figure 1A) and a higher reducibility, as established by the TPR profile in Figure 3, due to weaker metal interactions. On the other hand, Al 2 O 3 showed the least activity ( Figure 4) when both syngas and H 2 were used as a pre-treatment feed, due to (1) the strong metal-support interaction, resulting in lower reducibility, as observed via TPR, and (2) the lower metal dispersion, observed via TEM, caused by the larger Co 3 O 4 particles that formed in the large alumina pores, observed via TEM, XRD, and BET (see Table 1). In the case of the SiO 2 -support, the TPR reduction profile resembled that of the TiO 2 support, suggesting that the SiO 2 support is also weakly bonded to the Co metal. However, the Co/SiO 2 catalyst showed a different reactivity to that of TiO 2 , which may have been due to a large surface area and a lower metal dispersion than the catalyst supported on TiO 2 (see Figure 1C and Table 1).
The support identity therefore plays a major role in the performance of the catalyst. Two parameters seem to determine the catalytic activity of the Co 3 O 4 nanoparticles: (1) the Co particle size, which is influenced by the structure of the support, and (2) the extent of the metal-support interaction, metal-metal oxide interaction, and the metal-metal carbite interaction, which determines the specific nature of the active sites and their intrinsic catalytic activity.

Conclusions
The purpose of this study is to show the advantages of using syngas as a reducing agent for Co-FTS catalysts. To this end, we demonstrated the effect of the pre-treatment conditions by comparing the activity and product selectivity of the catalysts treated with syngas or H 2 at different temperatures. A lower CH 4 selectivity, higher C 5+ selectivity, and lower P/O ratio were observed for the catalysts treated with syngas at 250 • C compared to the catalysts reduced either with H 2 -at 250 • C or syngas at 350 • C. The formation of the Co x C phase during the reduction in syngas may either: (1) act as an active site for the production of lower olefins or (2) suppress the hydrogenation reaction. On the basis of the experimental results, we hypothesised that there may be synergy between Co 0 and Co x C to convert CO and H 2 to long chain hydrocarbons.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data presented in this article is available from the corresponding author on reasonable request.