Impact of Coordination Features of Co(II)-Glycine Complex on the Surface Sites of Co/SiO2 for Fischer–Tropsch Synthesis

To investigate the effect of coordination features of Co(II)-glycine complex on the performance of Co/SiO2 for Fischer–Tropsch (FT) synthesis, Co(II)-glycine complex precursors were prepared by the conventional method, i.e., simply adding glycine to the solution of Co nitrate and novel route, i.e., reaction of glycine with cobalt hydroxide. The SiO2-supported Co catalysts were prepared by using the different Co(II)-glycine complexes. It is found that glycine is an effective chelating agent for improving the dispersion of Co and the mass-specific activity in FT synthesis when the molar ratio of glycine/Co2+ = 3, which is independent to the preparation method in this study. Significantly, the surface Co properties were significantly influenced by the coordination features of the Co2+ and the molar ratio of glycine to Co2+ in the Co(II)-glycine complex. Specifically, the Co(3gly)/SiO2 catalyst prepared by the novel route exhibits smaller and homogenous Co nanoparticles, which result in improved stability compared to Co-3gly/SiO2 prepared by the conventional method. Thus, the newly developed method is more controllable and promising for the synthesis of Co-based catalysts for FT synthesis.


Introduction
Fischer-Tropsch (FT) synthesis is a promising process to convert syngas (CO + H 2 ) derived from non-petroleum-based resources such as coal, biomass, and natural gas to super-clean fuels and high-value-added fine chemicals [1,2]. Cobalt-based catalysts have been widely investigated for FT synthesis due to its high activity, high resistance to deactivation, low water-gas shift activity, and reasonable reservoir [3,4]. Commonly, the metallic cobalt (Co 0 ) is the active site for FT synthesis. Thus, to increase the dispersion of cobalt, cobalt precursors are generally deposited on porous oxide supports with a high surface area such as alumina, silica, or clays [5][6][7][8]. Cobalt nitrate precursor is often used owing to its high solubility allows for high metal loading in a single impregnation step. However, poor dispersions and inhomogeneous size distributions of cobalt are frequently obtained, which result in the lower mass-specific activity and higher deactivation rate [9,10]. Thus, to achieve higher mass-specific activity of Co-based catalysts, different methods have been explored to increase the cobalt dispersion, such as changing the precursor [9,11], co-impregnation with chelating agents [12][13][14][15][16][17], or modifying the drying or calcination procedure [18][19][20].
Among the above-mentioned methods, cobalt nitrate co-impregnation with chelating agents has been proved an effective approach to decrease the particle size of Co [13]. The essence of this

Preparation of SiO 2 -Supported Catalysts with Different Co(II)-Glycine Complexes
As shown in Figure 1A, the impregnation solution containing different Co complex precursors exhibits different colors and characters. Actually, H 2 O is regarded as a ligand for Co 2+ in the cobalt nitrate aqueous solution. Thus, the complex in the cobalt nitrate impregnating solution can be denoted as [Co(H 2 O) 6 ](NO 3 ) 2 (Figure 1(Aa)). In contrast, the ligand of H 2 O will be substituted by glycine through adding glycine to the cobalt nitrate aqueous solution (Figure 1(Ab)). However, the coordination features of Co in the solution cannot be confirmed, and thus the Co species in the solution is multiform such as [Co(H 2 O) 6 ](NO 3 ) 2 and [Co(glycine) x (H 2 O) 6−2x ](NO 3 ) 2 with different coordination numbers, which is related to the pH of the solution. In the case of the novel route, the Co(II)-glycine complex was prepared by the reaction of glycine with Co(OH) 2 , which resulted in the formation of a homogenous [Co(glycine) 2 (H 2 O) 2 ] complex ( Figure 1(Ac)) [28,29], which is very different in terms of colors and characters compared to the impregnation solution containing Co nitrate and Co nitrate/glycine. After the Co complex was deposited on the SiO 2 support, the crystal structure of Co species over the uncalcined catalysts was investigated by XRD technique. As shown in Figure 1B, no diffraction peaks assigned to Co species are observed over Co/SiO 2 and Co-3gly/SiO 2 . These observations indicate that the [Co(H 2 O) 6 ](NO 3 ) 2 and [Co(glycine) x (H 2 O) 6−2x ](NO 3 ) 2 is uniformly dispersed over the SiO 2 as a noncrystalline structure. In the case of Co(1gly)/SiO 2 , three weak peaks at 2θ of 19.1 • , 37.8 • , and 51.3 • assigned to the Co(OH) 2 phase (PDF#: 74-1057) can be observed. This indicates that not all of the Co(OH) 2 is coordinated by glycine when the molar ratio of glycine/Co(OH) 2 = 1. In contrast, as the molar ratio of glycine/Co(OH) 2 is increased to 3, there are no diffraction peaks assigned to Co species over the Co(3gly)/SiO 2 , which indicates that the Co(OH) 2 is fully coordinated by the glycine and gives the homogenous Co complex with stable and definite structure [Co(glycine) 2 (H 2 O) 2 ] on the SiO 2 . When further increasing the molar ratio of glycine/Co(OH) 2 to 5, the peaks assigned to glycine (PDF# 32-1702) can be clearly seen over Co(5gly)/SiO 2 , indicating the glycine is remarkably excessive for the formation of Co(II)-glycine complex.
The thermal decomposition behavior of different Co precursor and Co(II)-glycine complexes over the catalysts was investigated through thermal gravity (TG) measurement. From the differential thermal gravity (DTG) curves in Figure 2, the decomposition of cobalt nitrate over Co/SiO 2 occurs at about 188 • C. In contrast, the Co-3gly/SiO 2 prepared via co-impregnation with cobalt nitrate and glycine shows four mass loss stages with maximum mass loss rate at about 168, 223, 256, and 290 • C. This observation indicates that the Co species over Co-3gly/SiO 2 should contain multiple coordination states, that are different Co(II)-glycine coexisting over the Co-3gly/SiO 2 catalyst. In the case of Co(xgly)/SiO 2 , i.e., the Co(II)-glycine complexes prepared by the reaction of Co(OH) 2 with glycine, the total weight loss of the catalysts is in the order of Co(1gly)/SiO 2 < Co(3gly)/SiO 2 < Co(5gly)/SiO 2 , which is consistent with the amount of introduced glycine. The maximum mass loss rate for Co(1gly)/SiO 2 is about at 226 and 254 • C. Significantly, when compared to the Co-3gly/SiO 2 , Co(3gly)/SiO 2 shows two mass loss stages with maximum mass loss rate at about 210 and 318 • C, which are higher than that of Co-3gly/SiO 2 . This further indicates that the structure and coordination features of Co complexes in the Co-3gly/SiO 2 and Co(3gly)/SiO 2 prepared by different methods are significantly different with each other. For the Co(5gly)/SiO 2 , the first weight loss stage with maximum mass loss rate at about 211 • C should be attributed to the decomposition of excess glycine. In addition, because the last weight loss stage is at about 367 • C, the calcination temperature for all of the catalysts is fixed at 400 • C.

Textural and Structural Properties of Catalysts
The porous structure of the catalysts was measured by N2 adsorption measurement ( Figure S1) and the textural properties are summarized in Table 1. As expected, the Brunauer-Emmett-Teller (BET) surface area, pore volume, and average pore size of the Co-supported catalysts are obviously decreased compared to the SiO2 support due to the introduction of Co species into the pores and surface of SiO2. It should be noted that the BET surface area, pore volume, and average pore size of Co-3gly/SiO2 are larger than that of Co/SiO2 prepared in the absence of glycine. It has been proved that the viscosity of the impregnation solution will be significantly increased upon evaporation of the solvent in the presence of chelating agents [13]. Thus, this can be applied to explain the above observations. The increased viscosity of the impregnating solution caused by the addition of glycine will suppress the outward flow of the solution and avoid blocking the pores, which leads to more uniform distribution of Co species inside the pores of the support. Moreover, the smaller Co3O4 particle size over Co-3gly/SiO2 may contribute to the larger surface area. In the case of Co(xgly)/SiO2, however, the BET surface area, pore volume, and average pore size are smaller than that of Co-3gly/SiO2. As previously discussed, the coordination features of the Co(II)-glycine complexes prepared by different methods are very different, which result in the different viscosities and interactions with the SiO2 support. Consequently, the porous structures of the catalysts prepared by different Co complexes are obviously different with each other, which will further influence the surface metallic Co site over the catalysts.

Textural and Structural Properties of Catalysts
The porous structure of the catalysts was measured by N 2 adsorption measurement ( Figure S1) and the textural properties are summarized in Table 1. As expected, the Brunauer-Emmett-Teller (BET) surface area, pore volume, and average pore size of the Co-supported catalysts are obviously decreased compared to the SiO 2 support due to the introduction of Co species into the pores and surface of SiO 2 . It should be noted that the BET surface area, pore volume, and average pore size of Co-3gly/SiO 2 are larger than that of Co/SiO 2 prepared in the absence of glycine. It has been proved that the viscosity of the impregnation solution will be significantly increased upon evaporation of the solvent in the presence of chelating agents [13]. Thus, this can be applied to explain the above observations. The increased viscosity of the impregnating solution caused by the addition of glycine will suppress the outward flow of the solution and avoid blocking the pores, which leads to more uniform distribution of Co species inside the pores of the support. Moreover, the smaller Co 3 O 4 particle size over Co-3gly/SiO 2 may contribute to the larger surface area. In the case of Co(xgly)/SiO 2 , however, the BET surface area, pore volume, and average pore size are smaller than that of Co-3gly/SiO 2 . As previously discussed, the coordination features of the Co(II)-glycine complexes prepared by different methods are very different, which result in the different viscosities and interactions with the SiO 2 support. Consequently, the porous structures of the catalysts prepared by different Co complexes are obviously different with each other, which will further influence the surface metallic Co site over the catalysts. The crystal structure of the Co species over the calcined catalysts was further investigated by XRD technique. As shown in Figure 3, only the Co 3 O 4 crystal phase can be seen for all of the catalysts. However, the full width at half maximum (FWHM) of the (311) diffraction peak is obviously different, indicating the different particle size of Co 3 O 4 over the catalysts. Indeed, the particle size of Co 3 O 4 can be calculated by using the Scherrer formula. However, because of the relatively big error for the small crystal Co 3 O 4 for this method, the particle size of Co 3 O 4 is qualitatively discussed in this section. It can be concluded that the particle size of Co 3 O 4 over the Co-3gly/SiO 2 and Co(xgly)/SiO 2 catalysts prepared in the presence of glycine is remarkably smaller than that of Co/SiO 2 prepared using cobalt nitrate as a precursor. In the case of Co(xgly)/SiO 2 , the particle size of Co 3 O 4 is decreased with increasing the molar ratio of glycine to Co(OH) 2 . The Co particle size for the reduced catalysts was investigated by using TEM and H 2 chemisorption in Section 3.3. In addition, it must be pointed out that the formation of the Co 2 SiO 4 phase cannot be detected by XRD because the peak of the Co 2 SiO 4 is overlapped with that of the Co 3 O 4 [31].
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 14 The crystal structure of the Co species over the calcined catalysts was further investigated by XRD technique. As shown in Figure 3, only the Co3O4 crystal phase can be seen for all of the catalysts. However, the full width at half maximum (FWHM) of the (311) diffraction peak is obviously different, indicating the different particle size of Co3O4 over the catalysts. Indeed, the particle size of Co3O4 can be calculated by using the Scherrer formula. However, because of the relatively big error for the small crystal Co3O4 for this method, the particle size of Co3O4 is qualitatively discussed in this section. It can be concluded that the particle size of Co3O4 over the Co-3gly/SiO2 and Co(xgly)/SiO2 catalysts prepared in the presence of glycine is remarkably smaller than that of Co/SiO2 prepared using cobalt nitrate as a precursor. In the case of Co(xgly)/SiO2, the particle size of Co3O4 is decreased with increasing the molar ratio of glycine to Co(OH)2. The Co particle size for the reduced catalysts was investigated by using TEM and H2 chemisorption in Section 3.3. In addition, it must be pointed out that the formation of the Co2SiO4 phase cannot be detected by XRD because the peak of the Co2SiO4 is overlapped with that of the Co3O4 [31]. The surface Co species over the calcined catalysts was further identified by XPS measurements based on the chemical shift of the binding energy and the results are shown in Figure 4. For the Co/SiO2 catalyst, the Co 2p3/2 peak is observed at 779.8 eV, which is slightly higher than the binding energy of the pure Co3O4 phase (778.5 eV) reported in the literature [30] owing to the interaction between the smaller Co3O4 particles and the SiO2 support. In contrast, the Co 2p3/2 peak over Co-3gly/SiO2 is almost invisible (all the samples were tested two times to verify the validity of the result). However, Co-3gly/SiO2 exhibits clear peaks of Co3O4 from the XRD result. These observations indicate that the distribution of Co3O4 particles throughout the SiO2 particles was not homogeneous over Co-3gly/SiO2. As previously discussed, the addition of glycine will increase the viscosity of the impregnating solution, which can suppress the outward flow of the solution and leads to the small amount of Co species located at the external surface of the support. In the case of Co(xgly)/SiO2, the The surface Co species over the calcined catalysts was further identified by XPS measurements based on the chemical shift of the binding energy and the results are shown in Figure 4. For the Co/SiO 2 catalyst, the Co 2p 3/2 peak is observed at 779.8 eV, which is slightly higher than the binding energy of the pure Co 3 O 4 phase (778.5 eV) reported in the literature [30] owing to the interaction between the smaller Co 3 O 4 particles and the SiO 2 support. In contrast, the Co 2p 3/2 peak over Co-3gly/SiO 2 is almost invisible (all the samples were tested two times to verify the validity of the result). However, Co-3gly/SiO 2 exhibits clear peaks of Co 3 O 4 from the XRD result. These observations indicate that the distribution of Co 3 O 4 particles throughout the SiO 2 particles was not homogeneous over Co-3gly/SiO 2 . As previously discussed, the addition of glycine will increase the viscosity of the impregnating solution, which can suppress the outward flow of the solution and leads to the small amount of Co species located at the external surface of the support. In the case of Co(xgly)/SiO 2 , the obvious characteristic is that the binding energy of the Co species is higher than that of the Co/SiO 2 catalyst. Based on the literature, the binding energy of the Co 2p 3/2 peak of Co 2 SiO 4 is about 781.5 eV [29]. Thus, the higher binding energy of the Co 2p 3/2 peak over Co(3gly)/SiO 2 (781.6 eV) indicates the formation of Co 2 SiO 4 -like species over these catalysts due to the smaller Co oxide particle size. In addition, it is worth noting that the binding energy of the Co 2p 3/2 peak of Co(1gly)/SiO 2 (782.6 eV) is clearly higher than that of Co(3gly)/SiO 2 (781.6 eV). As revealed from the XRD result, the Co(OH) 2 exists over the Co(1gly)/SiO 2 catalyst. Thus, the higher binding energy of the Co 2p 3/2 peak of Co(1gly)/SiO 2 is attributed to the formation of Co 2 SiO 4 because Co 2 SiO 4 can be easily formed via the reaction of Co(OH) 2 with SiO 2 support [31]. In addition, similar with the Co-3gly/SiO 2 , the Co 2p 3/2 peak over Co(5gly)/SiO 2 is very weak, indicating decreased Co/Si surface ratios. We speculate that the uncombined glycine over Co-3gly/SiO 2 and Co(5gly)/SiO 2 may induce the silica migration, and coating or encapsulation of the CoO x species in the process of drying and calcination [32].
Catalysts 2020, 10, x FOR PEER REVIEW 6 of 14 obvious characteristic is that the binding energy of the Co species is higher than that of the Co/SiO2 catalyst. Based on the literature, the binding energy of the Co 2p3/2 peak of Co2SiO4 is about 781.5 eV [29]. Thus, the higher binding energy of the Co 2p3/2 peak over Co(3gly)/SiO2 (781.6 eV) indicates the formation of Co2SiO4-like species over these catalysts due to the smaller Co oxide particle size. In addition, it is worth noting that the binding energy of the Co 2p3/2 peak of Co(1gly)/SiO2 (782.6 eV) is clearly higher than that of Co(3gly)/SiO2 (781.6 eV). As revealed from the XRD result, the Co(OH)2 exists over the Co(1gly)/SiO2 catalyst. Thus, the higher binding energy of the Co 2p3/2 peak of Co(1gly)/SiO2 is attributed to the formation of Co2SiO4 because Co2SiO4 can be easily formed via the reaction of Co(OH)2 with SiO2 support [31]. In addition, similar with the Co-3gly/SiO2, the Co 2p3/2 peak over Co(5gly)/SiO2 is very weak, indicating decreased Co/Si surface ratios. We speculate that the uncombined glycine over Co-3gly/SiO2 and Co(5gly)/SiO2 may induce the silica migration, and coating or encapsulation of the CoOx species in the process of drying and calcination [32].

Co Particle Size and Reduction Behavior of the Catalysts
H2-TPR measurement was used to evaluate the reduction behavior of the catalysts ( Figure 5). The Co/SiO2 show two discrete peaks at about 296 and 350 °C, which are assigned to the two-step reduction of Co3O4 to CoO and CoO to metallic Co, respectively. In addition, the Co/SiO2 shows a weakly broad peak from 500 to 800 °C, which can be assigned to the reduction of the Co species having a stronger interaction with SiO2. In contrast, the H2-TPR profile of the SiO2-supported Co catalysts prepared with the glycine-assisted method show similar and clear high temperature reduction peaks at a temperature range of 650 to 900 °C, which are attributed to the reduction of the Co2SiO4-like species. However, the reduction behavior of the Co species are significantly influenced by the preparation method of Co(II)-glycine complexes and the molar ratio of glycine to Co(OH)2. In order to quantitatively evaluate the reduction degree of cobalt over the catalysts, the O2 pulse titration is carried out and the results are listed in Table 2. The reduction degree of the catalyst prepared in the presence of glycine is obviously lower than that of Co/SiO2 (52.9%). Moreover, the reduction degree of Co-3gly/SiO2 is higher than that of Co(xgly)/SiO2. For Co(xgly)/SiO2, as the molar ratio of glycine to Co(OH)2 is increased, the reduction degree of the catalysts decrease correspondingly.

Co Particle Size and Reduction Behavior of the Catalysts
H 2 -TPR measurement was used to evaluate the reduction behavior of the catalysts ( Figure 5). The Co/SiO 2 show two discrete peaks at about 296 and 350 • C, which are assigned to the two-step reduction of Co 3 O 4 to CoO and CoO to metallic Co, respectively. In addition, the Co/SiO 2 shows a weakly broad peak from 500 to 800 • C, which can be assigned to the reduction of the Co species having a stronger interaction with SiO 2 . In contrast, the H 2 -TPR profile of the SiO 2 -supported Co catalysts prepared with the glycine-assisted method show similar and clear high temperature reduction peaks at a temperature range of 650 to 900 • C, which are attributed to the reduction of the Co 2 SiO 4 -like species. However, the reduction behavior of the Co species are significantly influenced by the preparation method of Co(II)-glycine complexes and the molar ratio of glycine to Co(OH) 2 . In order to quantitatively evaluate the reduction degree of cobalt over the catalysts, the O 2 pulse titration is carried out and the results are listed in Table 2. The reduction degree of the catalyst prepared in the presence of glycine is obviously lower than that of Co/SiO 2 (52.9%). Moreover, the reduction degree of Co-3gly/SiO 2 is higher than that of Co(xgly)/SiO 2 . For Co(xgly)/SiO 2 , as the molar ratio of glycine to Co(OH) 2 is increased, the reduction degree of the catalysts decrease correspondingly.  Commonly, the surface Co 0 density (Co dispersion) of the catalyst is determined by the particle size and the reduction degree of Co. H2-chemisorption is an effective technique to obtain the actual surface Co 0 density and the dispersion (normalized by the total moles of Co over the catalyst). As shown in Table 2, the Co dispersion of Co-3gly/SiO2 (10.9%) is two-times higher than that of Co/SiO2 (5.4%). This is higher than the result reported by Koizumi et al. [30], in which the dispersion of Co over the catalyst (5 wt.% Co loading) prepared through the co-impregnation of Co nitrate and glycine is almost similar (about 4.9%) with that of Co/SiO2 prepared in the absence of glycine. Through comparing the preparation process of the catalyst in the literature and our work, we can found that aqueous NH3 solution was added to the impregnation solution to maintain the pH at 9-10 in their work, while the pH of the impregnation solution was not regulated in our work. Thus, the distinct difference of the Co dispersion between the two works may be caused by the different pH of the impregnation solution, which may result in the different coordination features of Co(II)-glycine. These observations also proved that the coordination features and the microenvironment of the Co species have prominently impacted on the Co oxide properties over the catalyst after calcination. Moreover, the Co(3gly)/SiO2 shows the highest the Co dispersion in the three catalysts prepared by the novel method. The Co 0 particle size of the reduced catalysts was calculated from the dispersion and the reduction degree of Co. As expected, the Co/SiO2 shows the largest Co 0 particle size (9.4 nm). In contrast, the Co 0 particle size over the Co-3gly/SiO2 is obviously decreased owing to the  Commonly, the surface Co 0 density (Co dispersion) of the catalyst is determined by the particle size and the reduction degree of Co. H 2 -chemisorption is an effective technique to obtain the actual surface Co 0 density and the dispersion (normalized by the total moles of Co over the catalyst). As shown in Table 2, the Co dispersion of Co-3gly/SiO 2 (10.9%) is two-times higher than that of Co/SiO 2 (5.4%). This is higher than the result reported by Koizumi et al. [30], in which the dispersion of Co over the catalyst (5 wt.% Co loading) prepared through the co-impregnation of Co nitrate and glycine is almost similar (about 4.9%) with that of Co/SiO 2 prepared in the absence of glycine. Through comparing the preparation process of the catalyst in the literature and our work, we can found that aqueous NH 3 solution was added to the impregnation solution to maintain the pH at 9-10 in their work, while the pH of the impregnation solution was not regulated in our work. Thus, the distinct difference of the Co dispersion between the two works may be caused by the different pH of the impregnation solution, which may result in the different coordination features of Co(II)-glycine. These observations also proved that the coordination features and the microenvironment of the Co species have prominently impacted on the Co oxide properties over the catalyst after calcination. Moreover, the Co(3gly)/SiO 2 shows the highest the Co dispersion in the three catalysts prepared by the novel method. The Co 0 particle size of the reduced catalysts was calculated from the dispersion and the reduction degree of Co. As expected, the Co/SiO 2 shows the largest Co 0 particle size (9.4 nm). In contrast, the Co 0 particle size over the Co-3gly/SiO 2 is obviously decreased owing to the introduction of glycine. For the Co(xgly)/SiO 2 , the Co 0 particle size is continuously decreased with increasing of the molar ratio of glycine to Co(OH) 2 . However, the Co 0 particle size over Co(1gly)/SiO 2 (7.9 nm) is much larger than Co(3gly)/SiO 2 and Co(5gly)/SiO 2 . As revealed from the XRD result, the Co(OH) 2 phase was detected over the uncalcined Co(1gly)/SiO 2 . Moreover, the lower ratio of glycine to Co will decrease the capsulate effect of glycine. Thus, the bigger Co 0 particle size over Co(1gly)/SiO 2 can be attributed to the two above-mentioned factors. Correspondingly, the smaller Co 0 particle size over Co(3gly)/SiO 2 and Co(5gly)/SiO 2 can be attributed to the stronger capsulate effect of the coordinate glycine around the Co 2+ [33][34][35].
Moreover, the Co 0 particle sizes over the reduced Co/SiO 2 , Co-3gly/SiO 2 , and Co(3gly)/SiO 2 were also examined via HAADF-STEM technique to further study the effect of coordination features of Co complex on the size distribution of Co 0 particles. As shown in Figure 6, the Co 0 particles' size over the Co/SiO 2 is distributed in the wide range of 3.5-11.5 nm. In contrast, the particle size of Co 0 over Co-3gly/SiO 2 is clearly decreased, and the size distribution of the Co 0 particles is in the range of 2.3-6.3 nm. Significantly, the size distribution of the Co 0 particles is very homogenous and further narrowed in the range of 2-3 nm. These observations indicate that the prepared method and the coordination features of the Co complex have remarkable influence on the size distribution of the Co 0 particles. introduction of glycine. For the Co(xgly)/SiO2, the Co 0 particle size is continuously decreased with increasing of the molar ratio of glycine to Co(OH)2. However, the Co 0 particle size over Co(1gly)/SiO2 (7.9 nm) is much larger than Co(3gly)/SiO2 and Co(5gly)/SiO2. As revealed from the XRD result, the Co(OH)2 phase was detected over the uncalcined Co(1gly)/SiO2. Moreover, the lower ratio of glycine to Co will decrease the capsulate effect of glycine. Thus, the bigger Co 0 particle size over Co(1gly)/SiO2 can be attributed to the two above-mentioned factors. Correspondingly, the smaller Co 0 particle size over Co(3gly)/SiO2 and Co(5gly)/SiO2 can be attributed to the stronger capsulate effect of the coordinate glycine around the Co 2+ [33][34][35]. Moreover, the Co 0 particle sizes over the reduced Co/SiO2, Co-3gly/SiO2, and Co(3gly)/SiO2 were also examined via HAADF-STEM technique to further study the effect of coordination features of Co complex on the size distribution of Co 0 particles. As shown in Figure 6, the Co 0 particles' size over the Co/SiO2 is distributed in the wide range of 3.5-11.5 nm. In contrast, the particle size of Co 0 over Co-3gly/SiO2 is clearly decreased, and the size distribution of the Co 0 particles is in the range of 2.3-6.3 nm. Significantly, the size distribution of the Co 0 particles is very homogenous and further narrowed in the range of 2-3 nm. These observations indicate that the prepared method and the coordination features of the Co complex have remarkable influence on the size distribution of the Co 0 particles. Figure 6. HAADF-STEM images and particle size distributions of Co over the reduced catalyst of Co/SiO2 (a), Co-3gly/SiO2 (b), and Co(3gly)/SiO2 (c).

FT Performance
The catalysts were evaluated for FT synthesis under the conditions of 235 °C, 1 MPa, H2/CO = 2, and W/F = 5.02 g h mol −1 . As shown in Figure 7, the catalytic activity (CO conversion) and stability Figure 6. HAADF-STEM images and particle size distributions of Co over the reduced catalyst of Co/SiO 2 (a), Co-3gly/SiO 2 (b), and Co(3gly)/SiO 2 (c).

FT Performance
The catalysts were evaluated for FT synthesis under the conditions of 235 • C, 1 MPa, H 2 /CO = 2, and W/F = 5.02 g h mol −1 . As shown in Figure 7, the catalytic activity (CO conversion) and stability are obviously different with each other. The CO conversions at steady state over the catalysts are increased in the order of Co-3gly/SiO 2 > Co(3gly)/SiO 2 > Co(5gly)/SiO 2 ≈ Co/SiO 2 > Co(1gly)/SiO 2 , which is consistent with the changing trend of Co 0 dispersion and surface Co 0 density ( Table 2). It should be noted that the deactivation rate of Co-3gly/SiO 2 is obviously higher than that of the others. In contrast, the CO conversion over Co(3gly)/SiO 2 is almost unchanged within the reaction time of 10 h. As revealed from the TEM result, the Co(3gly)/SiO 2 exhibits homogenous particle size distribution of Co. Based on the theoretical calculation, the Ostwald ripening rate could be significantly suppressed by preparing the homogeneously distributed metal particles with identical size [22]. Thus, the different stability of the two catalysts can be reasonably explained by the size distribution of Co particles over the Co-3gly/SiO 2 and Co(3gly)/SiO 2 catalysts. Moreover, for the catalysts prepared by the second method, with an increasing molar ratio of glycine/Co 2+ , the CO conversion is firstly increased, and then clearly decreased. As a result, the highest CO conversion is achieved over Co(3gly)/SiO 2 . Thus, we can conclude that the activity of the catalyst prepared in the presence of chelating agent is not only influenced by the intrinsic properties of the chelating agent, but also influenced by the ratio of chelating agent to metal ion. Moreover, from the cobalt-time-yield (CTY) in Figure 7B, the mass-specific activity of the Co-3gly/SiO 2 and Co(3gly)/SiO 2 catalysts was significantly improved, although the reduction degrees of Co over these two catalysts are lower than that of Co/SiO 2 . However, based on the previous report [30], unfortunately, using the easily available glycine as a chelating agent is almost ineffective for improving mass-specific activity of the Co catalyst. The different results may be caused by the different pH of the impregnation solution, which leads to the different coordination features of Co.
Catalysts 2020, 10, x FOR PEER REVIEW 9 of 14 are obviously different with each other. The CO conversions at steady state over the catalysts are increased in the order of Co-3gly/SiO2 > Co(3gly)/SiO2 > Co(5gly)/SiO2 ≈ Co/SiO2 > Co(1gly)/SiO2, which is consistent with the changing trend of Co 0 dispersion and surface Co 0 density ( Table 2). It should be noted that the deactivation rate of Co-3gly/SiO2 is obviously higher than that of the others. In contrast, the CO conversion over Co(3gly)/SiO2 is almost unchanged within the reaction time of 10 h. As revealed from the TEM result, the Co(3gly)/SiO2 exhibits homogenous particle size distribution of Co. Based on the theoretical calculation, the Ostwald ripening rate could be significantly suppressed by preparing the homogeneously distributed metal particles with identical size [22]. Thus, the different stability of the two catalysts can be reasonably explained by the size distribution of Co particles over the Co-3gly/SiO2 and Co(3gly)/SiO2 catalysts. Moreover, for the catalysts prepared by the second method, with an increasing molar ratio of glycine/Co 2+ , the CO conversion is firstly increased, and then clearly decreased. As a result, the highest CO conversion is achieved over Co(3gly)/SiO2. Thus, we can conclude that the activity of the catalyst prepared in the presence of chelating agent is not only influenced by the intrinsic properties of the chelating agent, but also influenced by the ratio of chelating agent to metal ion. Moreover, from the cobalt-time-yield (CTY) in Figure 7B, the mass-specific activity of the Co-3gly/SiO2 and Co(3gly)/SiO2 catalysts was significantly improved, although the reduction degrees of Co over these two catalysts are lower than that of Co/SiO2. However, based on the previous report [30], unfortunately, using the easily available glycine as a chelating agent is almost ineffective for improving mass-specific activity of the Co catalyst. The different results may be caused by the different pH of the impregnation solution, which leads to the different coordination features of Co.  Figure 7B shows the product selectivity of the FT synthesis over these catalysts. The selectivity of CH4 and C2-C4 over Co-3gly/SiO2 is slightly increased and the selectivity of C21 + is slightly decreased compared to that of Co/SiO2. However, in comparison with Co/SiO2, the selectivity of CH4 and C2-C4  Figure 7B shows the product selectivity of the FT synthesis over these catalysts. The selectivity of CH 4 and C 2 -C 4 over Co-3gly/SiO 2 is slightly increased and the selectivity of C 21 + is slightly decreased compared to that of Co/SiO 2 . However, in comparison with Co/SiO 2 , the selectivity of CH 4 and C 2 -C 4 over Co(3gly)/SiO 2 is significantly increased, while the selectivity of C 21 + is significantly decreased.
Based on the previous report, the surface hydrogen coverage is increased with decreasing Co particle size over the catalysts with average Co 0 particles smaller than 6 nm [36,37]. Thus, the above results about the product selectivity over the Co-3gly/SiO 2 and Co(3gly)/SiO 2 can be explained according to the decreased particle size of the Co 0 over the catalysts. In addition, the product distributions over the three catalysts prepared by the second manner are very similar to each other. The highest selectivity of C 21 + over Co(3gly)/SiO 2 should be attributed to the higher CO conversion. From the above discussion about the catalytic performance, we can conclude that glycine is an effective chelating agent to improve the dispersion of Co and the mass-specific activity of Co catalysts in Fischer-Tropsch synthesis independent of the preparation method used in this study, although the reduction degree of Co is very low compared to the Co/SiO 2 prepared in the absence of glycine. However, the size distribution of Co and the stability in the FT synthesis was significantly influenced by the coordination features (preparation method) of the Co(II) complexes. Moreover, it should be pointed that the higher selectivity of light hydrocarbons (C 1 -C 4 ) over the Co(3gly)/SiO 2 is not desirable, which makes the FT synthesis process inefficient. Thus, the preparation of the Co catalyst using glycine as a chelating agent still needs optimized.

Preparation of Catalyst
The conventional Co/SiO 2 catalyst was prepared from the impregnation solution containing only Co nitrate by incipient wetness impregnation method (control experiment). The SiO 2 -supported Co catalysts using Co(II)-glycine as the precursor were prepared by two different methods: (1) the Co(II)-glycine precursor in the first method was prepared by directly adding the glycine to the solution of Co nitrate with a molar ratio of glycine/Co 2+ = 3. Then the SiO 2 -support Co catalyst was prepared from the impregnation solution containing both Co nitrate and glycine, and this catalyst was denoted as Co-3gly/SiO 2 ; (2) the Co(II)-glycine precursors in the second method were prepared by the reaction of glycine with Co(OH) 2 with a different molar ratio of glycine to Co(OH) 2 (glycine/Co(OH) 2 = 1, 2, and 3). Typically, the Co(OH) 2 was prepared by the reaction of Co(NO 3 ) 2 with NaOH using the molar ration of OH -/CO 2+ = 2. After the formation of the Co(OH) 2 precipitate, the Co(OH) 2 was recovered by filtration and thoroughly washing with deionized water, and then the wet Co(OH) 2 were dried at 80 • C for 12 h. For the preparation of the Co(II)-glycine complex, the Co(OH) 2 powder was slowly added to the 0.2 mol/L glycine aqueous solution at 80 • C, in which the molar ratio of glycine to Co(OH) 2 was 1, 3, or 5. After the addition of Co(OH) 2 powder, the solution was continuously stirred for 2 h at 80 • C. Finally, the SiO 2 powder was impregnated with the obtained solution containing Co(II)-glycine complex. These kind of catalysts were denoted as Co-xgly/SiO 2 where x = molar ratio of glycine to Co(OH) 2 . All the catalysts were dried at 100 • C for 12 h, and then calcined at 400 • C for 4 h. The metallic Co loading for all of the catalysts was 7 wt.%.

Characterizations
XRD measurement was performed on an X-ray diffractometer (D8 Advance) with Cu Kα radiation operated at 40 kV and 40 mA. The speed of scanning was 4 • /min with a step size of 0.02 • .
Low temperature N 2 adsorption was performed at −196 • C using BelSorp-Max (Bel Japan Inc., Osaka, Japan). The samples were outgassed at 300 • C for 12 h. The specific surface area was calculated by the BET method. The pore size distributions were obtained from the BJH method using adsorption branch. Thermogravimetric (TG) measurement was carried out on a Q1000DSC thermogravimetric analyzer. The uncalcined catalysts were heated from room temperature to 800 • C at a heating rate of 10 • C/min in an air atmosphere. X-ray photoelectron spectroscopy (XPS) analyses were carried out on an Axis Ultra spectrometer (Kratos Analytical Ltd., Manchester, UK) using an Al monochromatic X-ray source (Al Ka = 1486.6 eV). The C1s binding energy of adventitious carbon (284.8 eV) was taken as an internal standard for correcting any charge-induced peak shifts. The H 2 -TPR was carried out on a BELCAT II (MicrotracBEL) instrument. The catalysts, 0.05 g, were first purged in a flow of argon at 200 • C for 30 min. After the temperature was decreased to 35 • C, the catalysts were heated to 900 • C at a heating rate of 10 • C/min under 10 vol.% hydrogen-argon mixtures with a flow rate of 30 cm 3 /min. The reduction degree of cobalt was determined by O 2 pulse titration method. Firstly, about 0.1 g of catalyst was reduced in situ for 6 h at 500 • C using pure hydrogen. Afterwards, the temperature of the sample was decreased to 400 • C, and flushed with pure Ar for 1 h. At the same temperature, 3 vol.% O 2 was injected with pulse mode to oxidize the reduced catalyst. The reduction degree (RD, %) of the catalyst was estimated based on the consumption of oxygen assuming that metallic Co was converted to Co 3 O 4 . H 2 -chemisorption measurements were performed on a Micromeretics ASAP 2020C instrument to evaluate the Co dispersion. Before measurement, the sample was reduced on the analysis station in situ in flowing H 2 at 500 • C for 6 h. Afterwards, the temperature was decreased to 100 • C and the H 2 -chemisorption was measured at this temperature. The H 2 uptakes and Co dispersion (D, %) were determined using the method reported in the literature [30], assuming hemispherical geometry of the metallic Co, with surface atomic density of 14.6 atoms/nm 2 . The particle sizes of Co 0 were calculated using d(Co) H = 96 × RD/D formula. TEM micrographs were obtained on an FEI Tecnai G2 F20 S-TWIN at an acceleration voltage of 200 kV. Prior to measurement, the catalysts were reduced at 500 • C for 6 h in hydrogen, then passivated by 1% O 2 in argon for 1 h at room temperature.

Catalytic Reactions
The catalytic performances of the catalysts in FT synthesis were tested in a fixed-bed reactor. Typically, 0.5 g of catalyst (40-60 mesh diluted with quartz sands) was firstly reduced in situ in a flow of pure H 2 (50 cm 3 /min) at 500 • C for 6 h. Then the temperature was decreased to 190 • C and the syngas (H 2 /CO = 2, 4% Ar as an internal standard) was fed into reactor. The reaction conditions were at 235 • C, 1.0 MPa, and W/F = 5.0 g·h·mol −1 . In order to prevent condensation of the products, the pipeline from the outlet of the reactor to the inlet of the gas chromatography (GC) was heated at 180 • C. The hydrocarbons in the effluent were analyzed online by GC with an HP-PONA capillary column (0.20 mm × 50 m, 0.5 µm) and a flame ionization detector (FID) (SP-3420A, Beijing Beifen-Ruili Analytical Instrument (Group) Co., Ltd. Beijing, China). The CO, CH 4 , Ar, and CO 2 in the effluent were analyzed online by GC with a packed activated-carbon column and a TCD detector (SP-3420A). The selectivity for hydrocarbons was calculated on the basis of carbon number.

Conclusions
In summary, SiO 2 -supported Co catalysts were prepared from Co(II)-glycine complex with different coordination features obtained by two different methods. It is found that glycine is an effective chelating agent to improve the dispersion of Co and the mass-specific activity of Co catalysts in Fischer-Tropsch synthesis independent of the preparation method used in this study. The characterization results revealed that the dispersion, particle size distribution, and reduction degree of Co are significantly influenced by the preparation method of Co(II)-glycine complex and the ratio of glycine/Co 2+ . In comparison to the Co-3gly/SiO 2 prepared by the traditional co-impregnation method, the Co(3gly)/SiO 2 prepared by the newly developed method exhibits smaller Co particle size and homogenous size distribution of Co. The mass-specific activity of the Co-based catalyst in the FT synthesis is significantly increased when the glycine/Co 2 + = 3. However, the stability of Co(3gly)/SiO 2 is evidently improved compared to Co-3gly/SiO 2 , which is attributed to the homogeneously distributed Co particles with identical size. Thus, the newly developed method is more controllable and promising for the synthesis of Co-based catalysts for FT synthesis.