Substitution of Co with Ni in Co / Al 2 O 3 Catalysts for Fischer–Tropsch Synthesis

: The e ﬀ ect of cobalt substitution with nickel was investigated for the Fischer–Tropsch synthesis reaction. Catalysts having di ﬀ erent Ni / Co ratios were prepared by aqueous incipient wetness co-impregnation, characterized, and tested using a continuously stirred tank reactor (CSTR) for more than 200 h. The addition of nickel did not signiﬁcantly modify the morphological properties measured. XRD, STEM, and TPR-XANES results showed intimate contact between nickel and cobalt, strongly suggesting the formation of a Co-Ni solid oxide solution in each case. Moreover, TPR-XANES indicated that nickel addition improves the cobalt reducibility. This may be due to H 2 dissociation and spillover, but is more likely the results of a chemical e ﬀ ect of intimate contact between Co and Ni resulting in Co-Ni alloying after activation. FTS testing revealed a lower initial activity when nickel was added. However, CO conversion continuously increased with time on-stream until a steady-state value (34%–37% depending on Ni / Co ratio) was achieved, which was very close to the value observed for undoped Co / Al 2 O 3 . This trend suggests nickel can stabilize cobalt nanoparticles even at a lower weight percentage of Co. Currently, the cobalt price is 2.13 times the price of nickel. Thus, comparing the activity / price, the catalyst with a Ni / Co ratio of 25 / 75 has better performance than the unpromoted catalyst. Finally, nickel-promoted catalysts exhibited slightly higher initial selectivity for light hydrocarbons, but this di ﬀ erence typically diminished with time on-stream; once leveling o ﬀ in conversion was achieved, the C 5 + selectivities were similar ( ≈ 80%) for Ni / Co ratios up to 10 / 90, and only slightly lower ( ≈ 77%) at Ni / Co of 25 / 75. Author Contributions: Conceptualization, catalyst preparation, catalyst characterization, formal analysis, writing, Catalyst preparation, catalyst characterization, formal analysis, S.C.K. Catalyst preparation, catalyst characterization, formal analysis, R.G. Catalyst preparation, catalyst characterization, formal analysis. C.D.W. Reaction testing, characterization, formal analysis, conceptualization, writing, M.M. Project administration, resources, C.L.M. Catalyst preparation, supervision, resources, D.C.C. Catalyst characterization, data curation, resources, supervision, A.J.K. Formal analysis W.D.S. All authors have read and agreed to the published version of the manuscript.


Introduction
Fischer-Tropsch synthesis (FTS) is a catalytic reaction which converts syngas, a mixture of CO and H 2 derived from natural gas, coal, and/or biomass, to high quality fuels. The active metals for FTS are iron, cobalt, nickel, and ruthenium. Among these metals, ruthenium is the most active. However, its application for large-scale FTS plants is impractical because of low abundance and very high XRD patterns for Al 2 O 3 and the oxide catalysts are plotted in Figure 1. All the catalysts show the characteristic reflection peaks associated with Co 3 O 4 (i.e., 2Θ = 36.8 • ). No diffraction peaks correlated with nickel compounds are detected for the nickel-promoted catalysts. This suggests nickel is well dispersed, as well as the formation of a Co-Ni solid oxide solution.
Catalysts 2019, 9, x FOR PEER REVIEW 3 of 31 loading is slightly higher (≈ 30%) than the theoretical value (25%), whereas all Ni/Co ratios are consistent with the nominal values. XRD patterns for Al2O3 and the oxide catalysts are plotted in Figure 1. All the catalysts show the characteristic reflection peaks associated with Co3O4 (i.e., 2Θ = 36.8°). No diffraction peaks correlated with nickel compounds are detected for the nickel-promoted catalysts. This suggests nickel is well dispersed, as well as the formation of a Co-Ni solid oxide solution.  shows TEM and STEM of reduced 25%M (10%Ni-90%Co)/Al2O3. The particle sizes are distributed between 18 and 23 nm, whereas the presence of nickel and cobalt was confirmed by EDS analysis. These two metals are uniformly distributed, confirming the formation of a Co-Ni solid oxide solution, as suggested from XRD results. The Co/Ni weight ratio is close to 10, similar to the theoretical value. The particle sizes for the 25%Co/Al2O3 are between 15 and 20 nm (Figure 3). EDS analysis shows areas richer in cobalt (until 55 wt. %), and others poorer in cobalt (10 wt. %); however, the average cobalt loading is close to 28 wt. % over an extended area.  Figure 2 shows TEM and STEM of reduced 25%M (10%Ni-90%Co)/Al 2 O 3 . The particle sizes are distributed between 18 and 23 nm, whereas the presence of nickel and cobalt was confirmed by EDS analysis. These two metals are uniformly distributed, confirming the formation of a Co-Ni solid oxide solution, as suggested from XRD results. The Co/Ni weight ratio is close to 10, similar to the theoretical Catalysts 2020, 10, 334 4 of 28 value. The particle sizes for the 25%Co/Al 2 O 3 are between 15 and 20 nm (Figure 3). EDS analysis shows areas richer in cobalt (until 55 wt. %), and others poorer in cobalt (10 wt. %); however, the average cobalt loading is close to 28 wt. % over an extended area.  Hydrogen chemisorption with pulse reoxidation results are presented in Table 2. The degree-ofreduction results are similar to those of the TPR-XANES/EXAFS data, as will be shown in the next sections. With increasing Ni/Co ratio, the mixed metal oxides exhibit more facile reduction. There also appears to be a slight increase in average metal diameter with increases in Ni/Co ratio as well. If the traditional approach, designated Method #1, of assuming complete oxidation of reduced metals to their respective oxides is used, then the increase is only slight (i.e., from 11.2 to 15.2 nm). However, if Method #2 is used, the difference is wider (i.e., from 7.0 to 13.5 nm). Method #2 assumes that, during reduction, all Co3O4 reduces to CoO, while only a portion of CoO reduces to Co 0 (and a fraction of NiO reduces to Ni 0 ). Thus, during the reoxidation with O2 pulses, the Ni 0 and Co 0 oxidize to NiO and CoO, and all CoO (including CoO obtained from Co 0 oxidation as well as the CoO previously resulting from merely partial reduction of Co3O4) oxidizes to Co3O4.  Hydrogen chemisorption with pulse reoxidation results are presented in Table 2. The degree-ofreduction results are similar to those of the TPR-XANES/EXAFS data, as will be shown in the next sections. With increasing Ni/Co ratio, the mixed metal oxides exhibit more facile reduction. There also appears to be a slight increase in average metal diameter with increases in Ni/Co ratio as well. If the traditional approach, designated Method #1, of assuming complete oxidation of reduced metals to their respective oxides is used, then the increase is only slight (i.e., from 11.2 to 15.2 nm). However, if Method #2 is used, the difference is wider (i.e., from 7.0 to 13.5 nm). Method #2 assumes that, during reduction, all Co3O4 reduces to CoO, while only a portion of CoO reduces to Co 0 (and a fraction of NiO reduces to Ni 0 ). Thus, during the reoxidation with O2 pulses, the Ni 0 and Co 0 oxidize to NiO and CoO, and all CoO (including CoO obtained from Co 0 oxidation as well as the CoO previously resulting from merely partial reduction of Co3O4) oxidizes to Co3O4. Hydrogen chemisorption with pulse reoxidation results are presented in Table 2. The degree-of-reduction results are similar to those of the TPR-XANES/EXAFS data, as will be shown in the next sections. With increasing Ni/Co ratio, the mixed metal oxides exhibit more facile reduction. There also appears to be a slight increase in average metal diameter with increases in Ni/Co ratio as well. If the traditional approach, designated Method #1, of assuming complete oxidation of reduced metals to their respective oxides is used, then the increase is only slight (i.e., from 11.2 to 15.2 nm). However, if Method #2 is used, the difference is wider (i.e., from 7.0 to 13.5 nm). Method #2 assumes that, during reduction, all Co 3 O 4 reduces to CoO, while only a portion of CoO reduces to Co 0 (and a fraction of NiO reduces to Ni 0 ). Thus, during the reoxidation with O 2 pulses, the Ni 0 and Co 0   Figure 4 is versus temperature, whereas the perspective of Figure 5 is that of photon energy. XANES snapshots at the point of 100% Co 3 O 4 , 100% CoO, as well as the final spectrum at the point of maximum reduction to Co 0 , are shown in Figure 6. Cobalt oxides reduce to metallic compounds in two steps: (I) Co 3 Figure 6 shows that with increasing Ni/Co ratio, the reduction of cobalt oxides systematically moves to lower temperature.
The point of 50%CoO/50%Co 0 was attained at 553 • C (Ni/Co = 0/100), 473 • C (Ni/Co = 5/95), 473 • C (Ni/Co = 10/90), 437 • C (Ni/Co = 25/75), and 386 • C (Ni/Co = 50/50). Thus, up to a ∆ 167 • C decrease in reduction temperature was achieved by doping Co with Ni. It is evident that, unlike the Group 10 metal Pt, where substantial shifts in the reduction of Co oxides were observed with minute amounts of Pt, significantly higher quantities of Ni are required to achieve the same level of reduction. For example, just 0.5%Pt was able to facilitate a decrease of 194 • C for the Co 3 O 4 to CoO transition, and a decrease of 120 • C from CoO to Co 0 [23], which is < 1/40th of the atomic amount. In comparison with the TPR profile of CoO during reduction of undoped 25%Co/Al 2 O 3 , this is likely in part due to the strong interactions between Ni oxides and alumina support.

H 2 TPR-EXAFS
TPR-EXAFS spectra at the Co K-edge are shown in Figures 8 and 9, including a plot that highlights the temperature (Figure 8), and a plot that emphasizes differences as a function of distance from the absorber ( Figure 9). The initial cyan spectrum of each TPR-EXAFS profile represents Co 3 O 4 . It can be differentiated from CoO (i.e., the first green spectrum) by the fact that the Co-O peak is significantly more intense, while the Co-Co coordination peak is broadened due to the presence of additional oxygen atoms (Figure 9). Following the transition to CoO, the CoO slowly converts to Co 0 , resulting in a final well-resolved peak for Co-Co metal coordination. In a manner similar to Pt and Ru promoters, Ni facilitates both steps of reduction; however, as mentioned previously, on an atomically equivalent basis, Ni is far less effective than either Pt or Ru [23]. It differs from Re, which only catalyzes the second step, CoO reduction to Co 0 . Unlike Pt and Ru oxides, which reduce at low temperatures, Re oxide was observed to reduce at a similar range as Co 3 O 4 reduction to CoO (i.e., 300-350 • C), and it was supposed that a reduced form was necessary in order to facilitate CoO reduction through a H 2 dissociation and spillover mechanism [23]. Thus, if a hydrogen dissociation and spillover mechanism operated for the case of Ni, one would expect that NiO should reduce to Ni 0 prior to the reduction of Co oxides. Based on the Ni X-ray absorption spectroscopy results, this does not seem to be the case.

H2 TPR-XANES
TPR-XANES spectra in Figures 11 and 12 at the Ni K-edge reveal that, initially, Ni oxide is associated with Co3O4 (see spectra in cyan color), and this Ni oxide subsequently undergoes a change in electronic structure to a form of Ni oxide associated with CoO (initial green spectrum). This change is best observed by examining the head-on spectra of Figure 13, as well as the XANES snapshots shown in Figure 14. Let us refer to this transition as step one.
Step two is further reduction of Ni 2+ to Ni 0 (dark green spectra in Figure 12).

H 2 TPR-XANES
TPR-XANES spectra in Figures 11 and 12 at the Ni K-edge reveal that, initially, Ni oxide is associated with Co 3 O 4 (see spectra in cyan color), and this Ni oxide subsequently undergoes a change in electronic structure to a form of Ni oxide associated with CoO (initial green spectrum). This change is best observed by examining the head-on spectra of Figure 13, as well as the XANES snapshots shown in Figure 14. Let us refer to this transition as step one.
Step two is further reduction of Ni 2+ to Ni 0 (dark green spectra in Figure 12). Figure 14 provides quantitative information regarding the Ni species present along the TPR trajectory. In step one, the point of 50%Ni 2+ -Co 3 O 4 / 50%Ni 2+ -CoO was reached at 333 • C (Ni/Co = 5/95), 286 • C (Ni/Co = 10/90), 261 • C (Ni/Co = 25/75), and 234 • C (Ni/Co = 50/50). Thus, up to a ∆ 99 • C decrease in reduction temperature was achieved (compared to a ∆ of 60 • C for the cobalt system over the same range of loading), and the temperatures match well with those of the Co 3 O 4 to CoO transitions described earlier at the 50% point of conversion, especially at higher Ni/Co ratios. The point of 100%Ni 2+ -CoO was obtained at 362 • C (Ni/Co = 5/95), 332 • C (Ni/Co = 10/90), 305 • C (Ni/Co = 25/75), and 278 • C (Ni/Co = 50/50). Thus, up to a ∆ 84 • C decrease in reduction temperature was obtained in converting 100%Ni 2+ -Co 3 O 4 to 100%Ni 2+ -CoO (compared to a ∆ of 52 • C for the cobalt system over the same range of loading). Once again, the temperatures match quite well with those of the Co 3 O 4 to CoO transitions described earlier for Co K-edge data, and match nearly perfectly at Ni/Co loadings of 10%, 25%, and 50%.
The similarities in temperature ranges between Co 3 O 4 to CoO and the Ni 2+ -Co 3 O 4 to Ni 2+ -CoO transitions, as well those between CoO to Co 0 and Ni 2+ -CoO to Ni 0 -Co 0 transitions, suggest that the effect may not be simply a H 2 dissociation and spillover mechanism, but rather a chemical effect due to intimate contact between Ni and Co in both oxide (e.g., solid solution) and metallic (e.g., alloy) phases, throughout the TPR trajectory.

H 2 TPR-EXAFS
Comparing the TPR-EXAFS spectra of Figure 15 with Figure 8, Figure 16 with Figure 9, and Figure 17 with Figure 10, it is remarkable how closely the behavior of Ni resembles that of Co. The initial cyan spectra of Figures 15 and 16 reveal that the first Ni-O peak is more intense than that of the green Ni-O peak, suggesting greater coordination to oxygen, while the second peak for Ni-Ni coordination in the initial cyan spectra is more broadened compared to the initial green spectra, consistent with greater oxygen content forcing Ni atoms apart. Figure 17 highlights the differences between the local atomic structures of different species along the TPR profile (e.g., Ni 2+ associated with Co 3 O 4 ; Ni 2+ associated with CoO; and Ni 0 associated with Co 0 ). Figures 18 and 19, as well as Table 3, show EXAFS fittings for the Co K-edge and Ni K-edge data following TPR-EXAFS and cooling to ambient temperature. A simple model was developed previously [20] and applied here, wherein metal coordination to Ni (whether the core atom was Co or Ni) was given as a fraction, X, of metal-cobalt coordination. Using this approach, excellent fittings with low r-factors were obtained. Co-Ni alloy formation is consistent with EXAFS fitting, but it cannot be considered proven, since Co and Ni are too closely related in size.

Catalytic Activity
CO conversion in the first 200 h is shown in Figure 20. The CO conversion for the unpromoted catalyst has an initial value of 42.5%, and then it slightly decreases in the first few hours until a steady-state value of 39.5% was reached. This trend is typical for cobalt-based catalysts. In contrast, CO conversion progressively increases for all nickel-promoted catalysts. 5%Ni-95%Co and 10%Ni-90% exhibit similar CO conversion trends, as the former's initial value is close to 20%, and it continuously increases, reaching 34%, whereas 25%Ni-75Co has a higher initial CO conversion (29%), and it reaches a steady-state value of 36%.
The evolutions of selectivities (CH 4 , CO 2 , C 2 -C 4 , and C 5+ ) with T.o.S. (time on-stream) are shown in Figure 21. CH 4 selectivity and C 5+ selectivities are stable at 7.6% and 80.7%, respectively, for the unpromoted catalyst. The addition of nickel increases initial CH 4 selectivity, whereas it decreases initial C 5+ selectivity. Fortunately, CH 4 selectivity for the nickel-promoted catalyst slowly decreased with T.o.S. For example, the initial CH 4 selectivity for 25%M-25%Ni-75%Co is 13.4%, but it reaches 9.5% after 200 h. Thus, the difference compared to Co/Al 2 O 3 decreases from 5.8% (absolute) to just 1.9%.
The olefin/paraffin ratio for unpromoted and nickel-promoted catalysts decreases with increasing carbon number, starting with ethylene species and moving upward ( Figure 22). This trend is typically observed for cobalt based catalysts. However, even if the trend with the carbon number is similar among the different catalysts, the olefin content decreases by increasing the nickel content. Indeed, the olefin/paraffin ratio for C 3 species is 2.5 for 25%Co, while it is only 1 for the 25%M-75%Co-25%Ni catalyst. At higher carbon numbers, the differences among the different Ni/Co catalysts are not so pronounced because of the tendency of olefins to readsorb on active sites [24]. The high hydrogenation capability of nickel promoted samples is also observed by a slight decrease in chain growth probability. Indeed α c5-c16 slightly decreases from 0.85 for 25%Co to 0.83 for 25%M-25%Ni-75%Co.       Figures 18 and 19, as well as Table 3, show EXAFS fittings for the Co K-edge and Ni K-edge data following TPR-EXAFS and cooling to ambient temperature. A simple model was developed previously [20] and applied here, wherein metal coordination to Ni (whether the core atom was Co or Ni) was given as a fraction, X, of metal-cobalt coordination. Using this approach, excellent fittings with low r-factors were obtained. Co-Ni alloy formation is consistent with EXAFS fitting, but it cannot be considered proven, since Co and Ni are too closely related in size.

Catalytic Activity
CO conversion in the first 200 h is shown in Figure 20. The CO conversion for the unpromoted catalyst has an initial value of 42.5%, and then it slightly decreases in the first few hours until a steadystate value of 39.5% was reached. This trend is typical for cobalt-based catalysts. In contrast, CO conversion progressively increases for all nickel-promoted catalysts. 5%Ni-95%Co and 10%Ni-90% exhibit similar CO conversion trends, as the former's initial value is close to 20%, and it continuously increases, reaching 34%, whereas 25%Ni-75Co has a higher initial CO conversion (29%), and it reaches a steady-state value of 36%.  Figure 21. CH4 selectivity and C5+ selectivities are stable at 7.6% and 80.7%, respectively, for the unpromoted catalyst. The addition of nickel increases initial CH4 selectivity, whereas it decreases initial C5+ selectivity. Fortunately, CH4 selectivity for the nickel-promoted catalyst slowly decreased with T.o.S. For example, the initial CH4 selectivity for 25%M-25%Ni-75%Co is 13.4%, but it reaches The olefin/paraffin ratio for unpromoted and nickel-promoted catalysts decreases with increasing carbon number, starting with ethylene species and moving upward ( Figure 22). This trend is typically observed for cobalt based catalysts. However, even if the trend with the carbon number is similar among the different catalysts, the olefin content decreases by increasing the nickel content. Indeed, the olefin/paraffin ratio for C3 species is 2.5 for 25%Co, while it is only 1 for the 25%M-75%Co-25%Ni catalyst. At higher carbon numbers, the differences among the different Ni/Co catalysts are not so pronounced because of the tendency of olefins to readsorb on active sites [24]. The high hydrogenation capability of nickel promoted samples is also observed by a slight decrease in chain growth probability. Indeed αc5-c16 slightly decreases from 0.85 for 25%Co to 0.83 for 25%M-25%Ni-75%Co.

Discussion
The promotion of nickel on a cobalt-based catalyst does not significantly influence the morphological properties, as all the samples have similar surface areas (≈ 92 m 2 /g), pore volumes, and pore diameters. The effects of nickel on the structural properties were investigated by XRD and

Discussion
The promotion of nickel on a cobalt-based catalyst does not significantly influence the morphological properties, as all the samples have similar surface areas (≈ 92 m 2 /g), pore volumes, and pore diameters. The effects of nickel on the structural properties were investigated by XRD and STEM. XRD patterns for all the calcined catalysts showed the typical peaks associated with Co 3 O 4 , and no diffraction peaks associated with NiO or Ni 0 were observed, suggesting that a mixed metal oxide was formed. Intimate contact between cobalt and nickel was also confirmed by elemental mapping during STEM analysis, as well as by comparing TPR-XANES profiles at nickel and cobalt K-edges. Initially, nickel is associated with Co 3 O 4 , and then it undergoes a change in the electronic structure to a form of nickel associated with CoO. Further evidence for solid solution formation is that the Ni 2+ associated with Co 3 O 4 (Ni K-edge results) and the Co 3 O 4 reduce over a similar temperature range, such that the temperatures at 50% conversion match very well, especially at higher Ni/Co ratios.
TPR-XANES and hydrogen chemisorption/O 2 titration show that nickel has a beneficial effect on cobalt reducibility. The reduction of cobalt oxide is shifted to a lower temperature with increasing Ni/Co ratio, and the percentage of metal reduction is increased. Voss et al. [19] observed an improvement in the reducibility when nickel was added, as the first reduction step shifted to a significantly lower temperature with increasing the nickel loading. The authors proposed a hydrogen spillover effect from the nickel sites to cobalt oxide sites. In the current work, the similarities in temperatures ranges between Co 3 O 4 to CoO and Ni 2+ -Co 3 O 4 to Ni 2+ -CoO, as well as similarities in the second reduction step from CoO to Co 0 and NiO to Ni 0 during TPR-XANES, suggest that in addition to H 2 spillover, a chemical promoting effect exists in leading to the formation of the Co-Ni alloy.
Cobalt-based catalysts are typically characterized by an initial decline and leveling off that occurs within the first few days. Possible mechanisms proposed for deactivation include: reoxidation of small metallic cobalt cluster to inactive CoO X ; sintering; some carbon deposition or solid-state reaction between cobalt and the support [4,25,26]. Thermodynamic calculations show that when the cobalt crystallites have a diameter lower than 4.4 nm, they may be reoxidized in the steam/hydrogen environments of FTS [7,8], but this may occur rapidly before the first measurement is taken. The initial decline and leveling off period for CO conversion typical of cobalt-based catalysts was not observed for the nickel promoted catalysts. Rytter et al. [18] have also observed an activation period in the first 40 h. They proposed a catalyst reconstruction in the first stage of operation where nickel and cobalt partially segregate, thereby exposing the cobalt clusters to FTS reaction. However, additional investigations are needed to speculate the events occurring during this induction time. Interestingly, the performances of nickel-promoted catalysts are stable, despite having lower cobalt loading than 25%Co/Al 2 O 3 . This suggests that nickel can stabilize cobalt metal nanoparticles. Rytter et al. [18] further proposed that the higher stability for Co-Ni alloys could be due to a suppression in carbon deposition and a suppression of reoxidation phenomena via H 2 spillover.
The addition of nickel changed the product distribution. In particular, the initial methane selectivity increases by increasing the nickel loading, whereas the initial C 5+ selectivity decreases. Higher CH 4 selectivities were also observed in previous works, where different Ni/Co ratios were studied [10,16,18]. Interestingly, the selectivities of Co/Ni catalysts improve with time on-stream to nearly match those of pure Co catalysts. Furthermore, the olefin content decreases, as does the chain growth probability with increasing Ni/Co ratio. These results are not surprising because of the high hydrogenation capability of nickel. Furthermore, Ishihara at al. [12] studied Co-Ni alloys supported on SiO 2 and found that cobalt electronically interacts with nickel in the outer shell orbitals by creating adsorption sites possessing a new electron density. These new sites have the highest H adsorption strengths. Thus, hydrogen competes more effectively for adsorption sites during co-adsorption of CO and H 2 with Co-Ni alloys relative to the pure cobalt based catalyst.
Finally, it is interesting to compare the activity/$ (Table 4), as nickel and cobalt have different market prices. The price of cobalt is reported to be 2.13 times the price of nickel. Thus, the partial substitution of cobalt with nickel would be an advantage in terms of total catalyst cost. At steady state conditions, the activity/$ for the 25%M-25%Ni-75%Co is best for both the nickel-promoted and unpromoted catalyst; additionally, at that point, the selectivities of the CoNi catalysts nearly match those of the pure Co based catalyst.

Catalyst Preparation
The conventional slurry impregnation method was used to prepare the catalyst containing 25% metal by weight, with the following Ni/Co atomic ratios: 0/100, 5/95, 10/90, 25/75, and 50/50. The support was Catalox 150 γ-alumina with a surface area of 150 m 2 /g. Nickel nitrate and cobalt nitrate (Alfa Aesar) served as the precursors to load the nickel and cobalt together (i.e., in a single solution) onto the γ-Al 2 O 3 support. In this method, the ratio of the weight of alumina to the volume of solution used was 1:1, as reported in a Sasol patent [1], such that the loading solution prepared was approximately 2.5 times that of the pore volume. The total metal loading was added by two impregnation steps with 12.5% of metal by weight for each step. Between each step, the catalyst was dried at 60 • C under vacuum in a rotary evaporator; then, the temperature was slowly increased until 100 • C. After the second impregnation, the catalyst was dried and then calcined in flowing air for 4 h at 350 • C.

Characterization
A Micromeritics 3-Flex system (Norcross, GA, USA) using N 2 (UHP N 2 , Airgas, Lexington, KY, USA) for physisorption was used to measure BET surface area and porosity properties. Before testing, the samples were pretreated at 160 • C and 50 mTorr for at least 12 h. The BJH method was employed to calculate the average pore volume and pore diameter.
XRD spectra were collected with a Philips X'Pert diffractometer (Amsterdam, The Netherlands) with monochromatic Cu Kα radiation (λ = 1.54Å). The conditions employed included a scan rate of 0.01 • per step, 2θ range of 10-90, and scan time of 4 s per step.
Prior to STEM characterization, the samples were pretreated in hydrogen (American Welding and Gas, Lexington, KY, USA) at 350 • C for 18 h, cooled down to room temperature, and then passivated with a mixture of 1% O 2 in nitrogen (American Welding and Gas, Lexington, KY, USA). STEM analysis was performed with an FEI Talos F200X instrument (Thermo Scientific, Waltham, MA, USA) equipped with BF, DF2, DF4, and HAADF detectors. The imaging was collected with a field emission gun using an accelerating voltage of 200 kV and a high speed Ceta 16M camera; Velox software (Thermo Scientific, Waltham, MA, USA) was used for data processing. The samples were dispersed in ethanol (Alfa Aesar, Haverhill, MA, USA), sonicated, and then dropped onto a carbon-coated copper grid and dried in air.
Hydrogen chemisorption and the ensuing pulse reoxidation were carried out using an Altamira AMI-300 unit (Altamira Instruments, Pittsburgh, PA, USA). The sample was reduced at 350 • C (ramping rate at 2 • C/min) for 10 h in 10 cm 3 /min of UHP H 2 (Airgas, San Antonio, TX, USA) blended with 20 cm 3 /min of UHP argon (Airgas, San Antonio, TX, USA). Then, the temperature was cooled to 100 • C, and UHP argon (30 cm 3 /min) was flowed through the catalytic bed to avoid the adsorption of weakly bound hydrogen. Next, the temperature was increased to 350 • C at 10 • C/min in flowing argon to desorb the chemisorbed hydrogen. The hydrogen peak obtained during the temperature programmed desorption was integrated and the moles of hydrogen evolved was determined by comparing to calibration pulses. Pulses of UHP O 2 (Airgas, San Antonio, TX, USA) were then passed through the reactor to reoxidize the catalyst until saturation was achieved. The percentage of reduction was estimated with two different methods. In the first approach, nickel and cobalt metal were assumed to oxidize to NiO and Co 3 O 4, respectively. However, in a second approach, we assumed all Co 3 O 4 converted at least to CoO during the reduction, and that a fraction of NiO and CoO converted to Ni 0 and Co 0 . Thus, during the reoxidation step, the Ni 0 and Co 0 are first oxidized to NiO and CoO. Then, all CoO oxidizes to Co 3 O 4 . These two approaches set minimum and maximum limits for the cobalt cluster size when the uncorrected dispersion is modified by considering the percentage of reduction by the metal, as follows: % Dispersion (uncorrected) = (# metal atoms on the surface)/(# metal atoms in the sample) % Disp. (corrected) = (# metal atoms on the surface)/((# metal atoms in the sample)(% reduction))

H 2 -TPR XANES-EXAFS
In-situ H 2 -TPR XAFS experiments were carried out at the Materials Research Collaborative Access Team (MR-CAT) beamline at the Advanced Photon Source, Argonne National Laboratory. A cryogenically cooled Si (1 1 1) monochromator selected the incident energy and a rhodium-coated mirror rejected higher order harmonics of the fundamental beam energy. The experiment setup was analogous to that outlined by Jacoby [27]. The in-situ TPR of 6 catalysts were performed in stainless-steel multi-sample holder (3.0 mm i.d. channels). Approximately 6 mg of each catalyst was loaded as a self-supporting wafer in each channel. The catalyst was diluted with alumina in a weight ratio of approximately 1:1. The holder was located in the center of a quartz tube and equipped with Kapton windows, a thermocouple, and gas ports. The amount of catalyst loaded was optimized for the Co and Ni K edges, considering the absorption by aluminum of the Al 2 O 3 . The quartz tube was positioned in a clamshell furnace mounted on the positioning table. Each sample cell was placed relative to the beam, and the position of the table was adjusted to an accuracy of 20 µm (for repeated scans). Once the catalyst positions were fine-tuned, the reactor was purged with He (100 mL/min) for more than 5 min, and then the reactant gas (a mixture H 2 /He, 3.5%) was flowed through the samples (100 mL/min). The temperature was increased to 700 • C (ramp of ≈ 1.0 • C/min) and then held for 4 h. The Ni and Co K-edge spectra were collected in transmission mode. The Co metallic foil spectrum was also recorded simultaneously for energy calibration. X-ray absorption spectra for each catalyst were collected from 7500 to 9000 eV.
The WinXAS program was used to analyze the spectra collected during H 2 -TPR EXAFS/XANES experiments [28]. Additional details of the EXAFS and XANES analyses for Co K-edge data are reported in our previous work [29]. Ni K-edge data were processed in a similar manner (i.e., same ∆k and ∆R in fittings). For qualitative comparisons of EXAFS and XANES results, the references used for Co 0 , CoO, and Co 3 O 4 were the final spectrum, the point of maximum CoO content, and the initial spectrum of the TPR trajectory of undoped cobalt catalyst (25%Co/Al 2 O 3 ). For Ni 0 and NiO, the references were NiO (Alfa Aesar, Puratronic, 99.998%, Tewksbury, MA, USA) and a Ni 0 foil.
For XANES analyses, linear combination fittings were carried out considering as reference compounds for Co K-edge data, the initial spectrum (a mixture of Co 3+ and Co 2+ similar to Co 3 O 4 ), the point of maximum CoO content, and the final spectrum after H 2 TPR (representing Co 0 ). At the Ni K-edge, the reference compounds were the first spectrum (Ni 2+ associated with Co 3 O 4 ), the spectrum with highest CoO content, where Ni 2+ is associated with the CoO, and the final spectrum after H 2 TPR (representing ≈ 100% Ni 0 ). The data reduction and fitting for EXAFS were performed using the catalysts in their final state following TPR and cooling in flowing H 2 using the WinXAS [28], Atoms [30], FEFFIT [31], and FEFF [31] programs. The k-range chosen for the fittings was 3-10 Å −1 . Fitting was confined to the first metallic coordination shell by applying a Hanning window in the Fourier transform magnitude spectra, and carrying out the back-transform to isolate that shell.

Reaction Testing
Activity tests were performed using a 1 L continuously stirred tank reactor (CSTR) (PPI, Warminster, PA, USA). Additional information on the laboratory-scale rig can be found elsewhere [32]. In a typical test, 9.6 g of catalyst (63 < dp < 125 µm) were loaded into a fixed bed reactor. The catalyst was reduced at 350 • C for 20 h, while being fed a 30 Nl/h H 2 /He mixture (1:3 v/v, American Welding and Gas, Lexington, KY, USA) at atmospheric pressure. The reduced catalyst was transferred by pneumatic transfer under the protection of inert gas to a CSTR containing 310 g of melted Polywax 3000 (Baker Petrolite, Houston, TX, USA). In-situ reduction of the transferred catalyst was performed at atmospheric pressure and 230 • C with overnight feeding of 30 Nl/h pure H 2 (American Welding and Gas, Lexington, KY, USA). In this work, the catalytic testing was carried out at the following process conditions: P = 20.2 bar, T = 220 • C, a stirring speed of 750 rpm, and H 2 /CO = 2 mol/mol. The unconverted reactants and the products leaving the CSTR were sent to a warm trap, in which the temperature was set at 100 • C, and then to a cold trap maintained at 0 • C. The uncondensed stream was reduced to atmospheric pressure, while the flowrate and the composition were measured by a wet test meter and by an online 3000A micro-GC (Agilent, Santa Clara, CA, USA), respectively. The micro-GC is equipped with four different columns (Plot U, Molecular Sieve, OV-1 and, Alumina) and TCD. The reaction products were collected in three traps maintained at different temperatures: a hot trap (200 • C), a warm trap (100 • C), and a cold trap (0 • C). The products were separated into different fractions (wax, oil, and aqueous) for quantification. The oil (C 4 -C 20 ) fraction was analyzed with a 7890 GC (Agilent, Santa Clara, CA, USA) equipped with DB-5 (60 m × 0.32 mm × 0.25µm, Agilent J&W) column and FID, whereas waxes (C 21 -C 60 ) were analyzed with an HP 6890 GC equipped with ZB-1HT column (30 m × 0.25 mm × 0.10 µm, Zebron) and FID.

Conclusions
Bimetallic catalysts with different Ni/Co ratios were prepared by standard aqueous incipient impregnation. N 2 adsorption/desorption results show that the addition of nickel had no effect on the morphological properties, as similar surface areas, pore volumes, and pore diameters were obtained independently from the Ni/Co ratio. XRD patterns of the samples exhibited peaks associated with Co 3 O 4 , whereas no diffraction peaks associated with Ni were observed. Thus, this suggests a Co-Ni solid oxide solution might be formed. STEM and TPR-XANES show the nickel and cobalt are in intimate contact, strongly suggesting the formation of a Co-Ni alloy from a Co oxide -Ni oxide solid solution. Moreover, TPR-XANES results indicate that nickel promotion improves the cobalt reducibility by systematically shifting the reduction profiles to lower temperatures. The similarities in temperature ranges between Co 3 O 4 to CoO and Ni 2+ -Co 3 O 4 to Ni 2+ -CoO, and the second reduction step to the metallic phase (i.e., CoO to Co 0 and NiO to Ni 0 ) during TPR-XANES, suggest that not only is H 2 spillover involved, but a chemical effect is likely. This is due to intimate contact of Co and Ni in the solid solution that leads to the formation of the Co-Ni alloy.
The catalyst's performance during FTS, in terms of conversion, selectivity, and stability, was evaluated using a CSTR. Nickel-promoted catalysts have lower initial CO conversion, which progressively increases with T.o.S. until a steady-state value is achieved. This CO conversion trend is inverted compared to the typical induction period observed for Co/Al 2 O 3 , where the activity progressively declines and levels off because of deactivation phenomena. The stability of the Co-Ni alloy may be due to the stabilization of metallic cobalt nanoparticles by nickel addition, resulting in robust nanoparticles even at lower cobalt content compared to commercial Co loadings. CH 4 selectivity increases by increasing nickel loading because of the higher hydrogenation capability of the Co-Ni alloy. However, the difference in CH 4 selectivity between the Ni-promoted and unpromoted catalyst (in terms of the absolute value) decreases with T.o.S., which is beneficial; in fact, after stabilization, the C 5 + selectivities were quite similar between catalysts prepared with Co-Ni bimetallic mixtures and Co alone. Finally, the steady-state activity/$ for a Ni/Co ratio of 25/75 is slightly higher than that of 25%Co/Al 2 O 3 .