Boosting the Catalytic Performance of Co/Mg/La Catalyst for Ammonia Synthesis by Selecting a Pre-Treatment Method

: The inﬂuence of the calcination process on the physicochemical properties and catalytic behavior of the Co/Mg/La catalysts for ammonia synthesis has been investigated. The catalysts were prepared using the different thermal pre-treatment methods prior to the activation, i.e., drying and calcination, and the respective activities for ammonia synthesis were assessed. It was found out that changing from air calcination prior to activation to direct activation of the co-precipitated species led to the different catalytic performances. The most favorable catalytic performance was achieved with Co/Mg/La prepared by calcination in air. Detailed characterization methods, employing e.g., XRPD, H 2 -TPD, N 2 -TPD, CO 2 -TPD, SEM, and TEM, showed that the superior catalytic behavior of this catalyst was attributed to its strong basicity and favorable adsorption properties toward hydrogen and nitrogen.


Introduction
Heterogeneous catalysts can be included in the group of ceramic materials defined as inorganic non-metallic solids, and along with many functional materials such as engine components, coatings, cutting tools, or bioceramics can be classified as so-called advanced ceramics [1]. It is estimated that 90% of the industrial chemical processes are based on catalysis. The size of the global catalyst market was valued at USD 34 billion in 2019 and is still growing. Thus, the development of new catalytic systems and the improvement of the existing ones are of great interest nowadays. Heterogeneous catalysis and ceramics overlap in a solid-state inorganic chemistry. Ceramics can be used in catalysis either directly as a catalyst or indirectly as a catalyst support, ensuring favorable textural properties, a dispersion of an active phase, and an appropriate mechanical and thermal resistance [1]. However, the development of ceramics as effective catalysts or catalyst supports requires the selection of appropriate compounds and, more importantly, the modification of physicochemical properties and chemical reactivity of their surfaces to adjust the interaction with adsorbed specific reactants of selected processes.
Since ammonia is one of the starting materials for the synthesis of many value-added chemicals, its synthesis is considered as one of the most important processes of the inorganic chemical industry all over the world. Globally, over 80% of ammonia produced is used for fertilizers, either for a direct application or converted to a variety of solid and liquid fertilizers, which are essential for a global food supply chain. Commonly, ammonia is synthesized via the Haber-Bosch process under high temperatures (>450 • C) and pressures Table S1 shows the activity in NH 3 synthesis of the Co/Mg/La catalysts differing by a thermal pre-treatment method. At all the tested temperatures (400, 430, and 470 • C) the NH 3 production rates and H 2 conversions obtained over the calcined Co/Mg/La catalyst were over twice that of the non-calcined Co/Mg/La catalyst. It is noteworthy to point out that with the increase of reaction temperature, the differences in the catalytic performances of Co/Mg/La catalysts were more pronounced. The results indicate that calcination prior to hydrogen activation can improve the catalytic activity of the Co/Mg/La catalyst for ammonia synthesis. The thermal stability is another important factor to evaluate the catalyst properties. The thermal stability tests of the Co/Mg/La catalysts are shown in Figure 1. During the 72 h overheating at 600 • C, the catalytic activities of Co/Mg/La were stable, which indicates their excellent thermal resistance.
To understand why the NH 3 production rate was higher over calcined Co/Mg/La than that achieved with non-calcined Co/Mg/La, we undertook the detailed studies to identify the phenomena occurring during the calcination of the co-precipitated cobalt, magnesium, and lanthanum species. To understand why the NH3 production rate was higher over calcined Co/Mg/La than that achieved with non-calcined Co/Mg/La, we undertook the detailed studies to identify the phenomena occurring during the calcination of the co-precipitated cobalt, magnesium, and lanthanum species. Figure S1 shows the N2 adsorption-desorption isotherms and pore size distributions of the precursors and reduced catalysts differing by the thermal pre-treatment method. Both the Co/Mg/La precursors were characterized by the type IV of nitrogen adsorptiondesorption isotherm with an H3 hysteresis loop, which is typical for the materials with pores of widths in the range of 2-50 nm [25]. It corresponds to the presence of pores formed from non-parallel planes characteristic for the materials of plate-like structure. The non-calcined Co/Mg/La precursor exhibited a unimodal pore size distribution with a dominant pore size of 6 nm. After air calcination, the shape of N2 adsorption-desorption isotherm of the respective calcined Co/Mg/La precursor changed. This indicates changes in its porous structure, as confirmed by the bimodal pore size distribution with a dominant pore size of 3 and 25 nm. After the hydrogen activation, no differences in the shape of N2 adsorption-desorption isotherms and hysteresis loops for both the Co/Mg/La catalysts were observed ( Figure S1). The bimodal pore size distributions with dominant pore sizes of 3 and 15 nm, and 3 and 30 nm for the non-calcined and calcined Co/Mg/La were reported, respectively.

N2 Physisorption
The texture properties of both the precursors and reduced catalysts are summarized in Table 1. Both the Co/Mg/La precursors had a similar BET surface area, ca. 30 m 2 g −1 , while a ca. 2-fold increase in the total pore volume (Vp) and average pore diameter (Φ) for the calcined precursor was reported. An increase in the Vp and Φ after calcination would be expected due to a loss in mass. Air calcination is a typical thermal pre-treatment method to convert e.g., precipitated hydroxide carbonates to respective oxides and remove CO2 and H2O. These gases act as pore-forming agents and induce the formation of large pores. After the hydrogen activation, the differences in the textural parameters were observed (Table 1). For the non-calcined Co/Mg/La catalyst, the SBET and Φ increased 2fold, while the Vp increased 3-fold in comparison to the values measured for the respective precursor. The increase in the SBET, Vp, and Φ after the hydrogen activation was due to the thermal decomposition of the co-precipitated carbonates/hydroxides, and the reduction of Co species to metallic Co during the reduction process. Thus, the CO2 and H2O molecules released upon the reduction process acted as pore forming agents, leading to the  Figure S1 shows the N 2 adsorption-desorption isotherms and pore size distributions of the precursors and reduced catalysts differing by the thermal pre-treatment method. Both the Co/Mg/La precursors were characterized by the type IV of nitrogen adsorptiondesorption isotherm with an H3 hysteresis loop, which is typical for the materials with pores of widths in the range of 2-50 nm [25]. It corresponds to the presence of pores formed from non-parallel planes characteristic for the materials of plate-like structure. The non-calcined Co/Mg/La precursor exhibited a unimodal pore size distribution with a dominant pore size of 6 nm. After air calcination, the shape of N 2 adsorption-desorption isotherm of the respective calcined Co/Mg/La precursor changed. This indicates changes in its porous structure, as confirmed by the bimodal pore size distribution with a dominant pore size of 3 and 25 nm. After the hydrogen activation, no differences in the shape of N 2 adsorption-desorption isotherms and hysteresis loops for both the Co/Mg/La catalysts were observed ( Figure S1). The bimodal pore size distributions with dominant pore sizes of 3 and 15 nm, and 3 and 30 nm for the non-calcined and calcined Co/Mg/La were reported, respectively.

N 2 Physisorption
The texture properties of both the precursors and reduced catalysts are summarized in Table 1. Both the Co/Mg/La precursors had a similar BET surface area, ca. 30 m 2 g −1 , while a ca. 2-fold increase in the total pore volume (V p ) and average pore diameter (Φ) for the calcined precursor was reported. An increase in the V p and Φ after calcination would be expected due to a loss in mass. Air calcination is a typical thermal pre-treatment method to convert e.g., precipitated hydroxide carbonates to respective oxides and remove CO 2 and H 2 O. These gases act as pore-forming agents and induce the formation of large pores. After the hydrogen activation, the differences in the textural parameters were observed (Table 1). For the non-calcined Co/Mg/La catalyst, the S BET and Φ increased 2-fold, while the V p increased 3-fold in comparison to the values measured for the respective precursor. The increase in the S BET , V p , and Φ after the hydrogen activation was due to the thermal decomposition of the co-precipitated carbonates/hydroxides, and the reduction of Co species to metallic Co during the reduction process. Thus, the CO 2 and H 2 O molecules released upon the reduction process acted as pore forming agents, leading to the higher surface area and porosity of the non-calcined catalyst. On the other hand, the BET surface area for the calcined Co/Mg/La was lower (ca. 7%) in comparison to the respective precursor. The observed decrease of S BET suggests the agglomeration and/or aggregation of the catalyst particles during the reduction treatment. It is noteworthy that the noncalcined Co/Mg/La catalyst had the highest BET surface area, twice that of the calcined Co/Mg/La catalyst. This implies that the changing from the air calcination method prior to activation to the direct reduction technique would lead to improvement of the textural properties of the catalyst.  4 Φ, average pore diameter estimated based on the BJH isotherm model. 5 The S BET , V p , and Φ determined after the hydrogen activation. 6 Estimated by using the amount of CO 2 desorbed and BET area of the catalysts after the hydrogen activation. Figure 2a presents the XRPD patterns of the Co/Mg/La precursors differing by the thermal pre-treatment method. The non-calcined Co/Mg/La precursor consisted of two phases, i.e., orthorhombic LaCO 3 OH (PDF# 49-0981) and Mg(OH) 2 (PDF# 44-1482). The absence of reflections from Co-bearing phases implies their presence in the amorphous forms and/or highly dispersed. After air calcination of the dried precursor, the phase composition of the respective calcined precursor was modified. In the diffraction pattern of the calcined Co/Mg/La precursor, the observed reflections were ascribed to La 2 O 2 CO 3 phases crystallizing in two polymorphs: hexagonal (II type, PDF# 37-0804) and monoclinic (Ia type, PDF# 48-1113). The reflections of MgO (PDF# 04-0829) were also detected. The presence of these phases was a result of the thermal decomposition of lanthanum carbonate hydroxide and magnesium hydroxide during the calcination process. The crystallite sizes of the respective phases are shown in Table S2. Interestingly, the average crystallite size of the hexagonal polymorph of La 2 O 2 CO 3 was significantly larger than the monoclinic one. Figure 2b shows the XRPD patterns of the Co/Mg/La after the hydrogen treatment. It was observed that after the hydrogen activation at 550 • C for 18 h, the reflections corresponding to MgO (PDF# 04-0829) and hexagonal lanthanum oxide (PDF# 74-1144) were detected. The presence of La 2 O 3 was a result of the thermal decomposition of LaCO 3 OH and La 2 O 2 CO 3 for non-calcined and calcined Co/Mg/La, respectively. Despite the same phase composition of the Co/Mg/La catalysts, the differences in the average crystallite sizes of MgO and La 2 O 3 were observed. MgO crystallites had nearly the same sizes, ca. 10 nm, while the crystallite sizes of La 2 O 3 differed, i.e., 31 and 40 nm for the non-calcined and calcined Co/Mg/La, respectively (Table S2). The observed differences were due to particles sintering during the calcination. As for the Co/Mg/La precursors, no reflections from the metallic Co and/or Co-bearing phases for the reduced catalysts were detected. Figure 3a depicts the DRIFTS spectra of the Co/Mg/La precursors differing by the thermal pre-treatment method. Both the Co/Mg/La precursors displayed a similar shape of DRIFTS spectra. The bands at 3776 and 3765 cm −1 were ascribed to a vibrational absorption band of bound hydroxyl groups. A broad band observed in the frequency region of 3600-3200 cm −1 corresponded to the OH-stretching vibrations of the adsorbed water molecules, while this at 1579 cm −1 was attributed to their bending mode [26][27][28]. The several bands at 1454, 1394, 876, 820, and 750 cm −1 were attributed to CO 3 2− groups of hydrous and/or anhydrous lanthanum carbonates. According to [29], the bands at 1454, 876, 820, and 750 cm −1 were attributed to the CO 3 2− groups occurring between the (LaO) 2 2+ layers, while a band at 1394 cm −1 was assigned to the polydentate carbonate. Figure 4b shows the DRIFTS spectra of the Co/Mg/La catalysts after the hydrogen activation at  [28,29].  Figure 3a depicts the DRIFTS spectra of the Co/Mg/La precursors differing by the thermal pre-treatment method. Both the Co/Mg/La precursors displayed a similar shape of DRIFTS spectra. The bands at 3776 and 3765 cm −1 were ascribed to a vibrational absorption band of bound hydroxyl groups. A broad band observed in the frequency region of 3600-3200 cm −1 corresponded to the OH-stretching vibrations of the adsorbed water molecules, while this at 1579 cm −1 was attributed to their bending mode [26][27][28]. The several bands at 1454, 1394, 876, 820, and 750 cm −1 were attributed to CO3 2-groups of hydrous and/or anhydrous lanthanum carbonates. According to [29], the bands at 1454, 876, 820, and 750 cm −1 were attributed to the CO3 2-groups occurring between the (LaO)2 2+ layers, while a band at 1394 cm −1 was assigned to the polydentate carbonate. Figure 4b shows the DRIFTS spectra of the Co/Mg/La catalysts after the hydrogen activation at 550 °C for 3 h. Both the Co/Mg/La catalysts displayed a similar shape of DRIFTS spectra. The weak band at 3753 cm −1 and the broad band in the range 3600-3300 cm −1 were assigned to the OHstretching vibrations of the adsorbed water molecules and the additional bands at 1770-750 cm −1 corresponding to the La2O2CO3 species were detected. The bands that appeared at 1768 and 1728 cm −1 were associated with a >C=O group of the La2O2CO3 species. The bands at 1478 and 1136 cm −1 were assigned to type Ia (monoclinic) La2O2CO3, while the bands at 864 and 762 cm −1 corresponded to type II (hexagonal) La2O2CO3. This implies that the thermal decomposition of carbonaceous species requires long-term heating, and even after heating for 3 h, the decomposition was not completed. This is because La2O2CO3 is a stable intermediate in the thermal decomposition of lanthanum carbonate, La2(CO3)3. The band observable at 1394 cm −1 was assigned to the formation of polydentate carbonate. The two bands at 2501 and 2405 cm −1 were associated with physically adsorbed CO2 [28,29].   Figure 4 shows the TG/DTA-MS profiles of the Co/Mg/La precursors differing by the thermal pre-treatment method. For the non-calcined Co/Mg/La precursor (Figure 4a), a three-stage reduction process was distinguished in the temperature ranges:

H2-TPR
(1) 30-150 °C , a slight mass loss (ca. 5%) with a maximum rate of mass loss at 98 °C was observed. This was associated with the desorption of physisorbed water and carbon dioxide, as confirmed by the signals at m/e = 18 and m/e = 44 in the mass spectra. (2) 150-500 °C , a significant mass loss (ca. 25%) with the maximum rates of mass loss at 362 and 434 °C were recorded and associated with an intense water and carbon dioxide release (the signals at m/e = 18 and m/e = 44 in the mass spectrum). This was attributed to the thermal decomposition of the co-precipitated cobalt hydroxy carbonate to cobalt(II,III) oxide and lanthanum hydroxy carbonate to lanthanum oxide   Figure 4 shows the TG/DTA-MS profiles of the Co/Mg/La precursors differing by the thermal pre-treatment method. For the non-calcined Co/Mg/La precursor (Figure 4a), a three-stage reduction process was distinguished in the temperature ranges:

H 2 -TPR
(1) 30-150 • C, a slight mass loss (ca. 5%) with a maximum rate of mass loss at 98 • C was observed. This was associated with the desorption of physisorbed water and carbon dioxide, as confirmed by the signals at m/e = 18 and m/e = 44 in the mass spectra. (2) 150-500 • C, a significant mass loss (ca. 25%) with the maximum rates of mass loss at 362 and 434 • C were recorded and associated with an intense water and carbon dioxide release (the signals at m/e = 18 and m/e = 44 in the mass spectrum). This was attributed to the thermal decomposition of the co-precipitated cobalt hydroxy carbonate to cobalt(II,III) oxide and lanthanum hydroxy carbonate to lanthanum oxide carbonate. The signal at m/e = 18 in the mass spectrum could also be attributable to the dehydroxylation of Mg(OH) 2 occurring at ca. 350 • C [30] and the reduction of Co 3 O 4 to metallic Co, occurring typically in the temperature range of 250-450 • C [31]. Moreover, the hydrogenation of CO 2 (released upon the decarboxylation of lanthanum oxide carbonate) to methane occurred. The confirmation was the appearance of the signal at m/e = 15 in the mass spectrum with the maximum at 425 • C. (3) 500-600 • C, a slight mass loss (ca. 3%) related to the thermal decomposition of La 2 O 2 CO 3 with subsequent hydrogenation of CO 2 to CH 4 was detected. The confirmation was the signal at m/e = 15 in the mass spectrum with the maximum at 551 • C.   Figure 5 shows the SEM and TEM micrographs of Co/Mg/La differenced by the thermal pre-treatment method. The distinct differences in the morphology of the Co/Mg/La precursors were observed both in the SEM and TEM images (Figure 5a). For the non-calcined precursor, the large particles of lanthanum carbonate hydroxide (up to 200 nm) and small flake-like particles of magnesium hydroxide were observed. According to the XRPD, the average crystallite sizes of LaCO3OH and Mg(OH)2 were 20 and 8 nm, respectively (Table S2). Thus, it is observable that the average crystallite size of lanthanum carbonate hydroxide was much smaller than its particle size observed in the SEM and TEM images. This suggests that the observed large particles were aggregates composed of the smaller crystallites. Note that the BET surface area of both the precursors was nearly the same, though there was a noticeable difference in the pore volume ( Table 1). The reason is probably more "dense" packing of LaCO3OH crystallites in the large particles. During air calcination, these large particles of lanthanum carbonate hydroxy transformed into two polymorphs of lanthanum oxide carbonate, hexagonal, and monoclinic. The flake-like Mg(OH)2 transformed into irregular in shape MgO particles with nearly unchanged aver-  After air calcination of the dried precursor, the phase composition of the respective calcined precursor was modified (Figure 2a), and thus, the course of its reduction process differed (Figure 4b). The reduction process of the calcined Co/Mg/La precursor consisted of three stages in the temperature ranges:

SEM and TEM
(1) 30-150 • C, a slight mass loss (ca. 2%) with a maximum rate of mass loss at 83 • C associated with the desorption of physisorbed water and carbon dioxide was observed.
The confirmation was the signals at m/e = 18 and m/e = 44 in the mass spectra. (2) 150-400 • C, a mass loss of ca. 4% related to the reduction of Co 3 O 4 to metallic Co and the thermal decomposition of La 2 O 2 CO 3 to La 2 O 3 was detected. This was confirmed by the signals at m/e = 18 and m/e = 44 in the mass spectrum, respectively. The appearance of the signal at m/e = 15 in the mass spectra was due to the hydrogenation of carbon dioxide (released upon the decarboxylation of La 2 O 2 CO 3 ) to methane.
(3) 400-600 • C, a mass loss of ca. 6% associated with the continuous hydrogenation of carbon dioxide to methane was observable, as confirmed by the signal at m/e = 15 in the mass spectrum with the maximum at 452 • C.
2.6. SEM and TEM Figure 5 shows the SEM and TEM micrographs of Co/Mg/La differenced by the thermal pre-treatment method. The distinct differences in the morphology of the Co/Mg/La precursors were observed both in the SEM and TEM images (Figure 5a). For the noncalcined precursor, the large particles of lanthanum carbonate hydroxide (up to 200 nm) and small flake-like particles of magnesium hydroxide were observed. According to the XRPD, the average crystallite sizes of LaCO 3 OH and Mg(OH) 2 were 20 and 8 nm, respectively (Table S2). Thus, it is observable that the average crystallite size of lanthanum carbonate hydroxide was much smaller than its particle size observed in the SEM and TEM images. This suggests that the observed large particles were aggregates composed of the smaller crystallites. Note that the BET surface area of both the precursors was nearly the same, though there was a noticeable difference in the pore volume ( Table 1). The reason is probably more "dense" packing of LaCO 3 OH crystallites in the large particles. During air calcination, these large particles of lanthanum carbonate hydroxy transformed into two polymorphs of lanthanum oxide carbonate, hexagonal, and monoclinic. The flakelike Mg(OH) 2 transformed into irregular in shape MgO particles with nearly unchanged average crystallite sizes, ca. 10 nm (Table S2). Co/Mg/La). This could be due to the different mechanisms of the phase evolution of both the precursors during the reduction process. An important issue concerning the reduced catalysts is the lack of evidence of metallic Co and/or Co-bearing phases. There were no the Co reflections in the XRPD patterns and no well-defined Co particles in the TEM images. Probably, the sizes of cobalt particles were small, and, what is more, they were at least partly oxidized after exposure to air after the reduction.  Figure 6 displays the EDS elemental mapping of Co/Mg/La in both the precursor and reduced forms differenced by the thermal pre-treatment method. The EDS elemental mapping confirmed that oxygen (O), magnesium (Mg), and cobalt (Co) were rather uniformly distributed on the micrometer scale. On the contrary, lanthanum (La) showed evident inhomogeneities in both the precursor and reduced catalysts. This effect indicates that Lacontaining compounds formed much larger particles than those containing Mg and Co. A more detailed analysis of the element's distribution in the catalysts after the hydrogen activation is presented in a Supplementary Material ( Figure S2 and Table S3). It confirmed the presence of large (up to 1 µ m) La-containing particles. Interestingly, it appears that the concentration of cobalt is noticeably lower in the places where lanthanum prevails (especially for the non-calcined Co/Mg/La). Thus, it can be concluded that Co preferentially deposits on magnesium oxide, rather than lanthanum oxide.  Figure 5b presents the SEM and TEM images of the Co/Mg/La catalysts after the hydrogen activation. Both the catalysts had similar morphology. Two fractions of the particles were observed. The larger ones, showing a higher contrast in TEM images, were lanthanum oxide, while the smaller ones, with a lower TEM contrast, were magnesium oxide. This agrees with the XRPD analysis showing the presence of La 2 O 3 and MgO phases in both the catalysts. Note that there was a significant difference in the BET surface area between the Co/Mg/La catalysts, i.e., the S BET of the non-calcined catalyst was almost twice that of the calcined catalyst (Table 1). This result is unexpected given similar X-ray diffraction and microscopy data. A possible explanation could be a different aggregation of crystallites of phases identified in both the catalysts (more extensive in the calcined Co/Mg/La). This could be due to the different mechanisms of the phase evolution of both the precursors during the reduction process.
An important issue concerning the reduced catalysts is the lack of evidence of metallic Co and/or Co-bearing phases. There were no the Co reflections in the XRPD patterns and no well-defined Co particles in the TEM images. Probably, the sizes of cobalt particles were small, and, what is more, they were at least partly oxidized after exposure to air after the reduction. Figure 6 displays the EDS elemental mapping of Co/Mg/La in both the precursor and reduced forms differenced by the thermal pre-treatment method. The EDS elemental mapping confirmed that oxygen (O), magnesium (Mg), and cobalt (Co) were rather uniformly distributed on the micrometer scale. On the contrary, lanthanum (La) showed evident inhomogeneities in both the precursor and reduced catalysts. This effect indicates that La-containing compounds formed much larger particles than those containing Mg and Co. A more detailed analysis of the element's distribution in the catalysts after the hydrogen activation is presented in a Supplementary Material ( Figure S2 and Table S3). It confirmed the presence of large (up to 1 µm) La-containing particles. Interestingly, it appears that the concentration of cobalt is noticeably lower in the places where lanthanum prevails (especially for the non-calcined Co/Mg/La). Thus, it can be concluded that Co preferentially deposits on magnesium oxide, rather than lanthanum oxide. Figure 7 presents the CO 2 -TPD profiles of the Co/Mg/La catalysts differenced by the thermal pre-treatment method. The profiles could be divided into three kinds of the basic sites at low-, medium-, and high-temperatures relating to the strength of the basic sites. CO 2 desorption observed in the low-temperature region (<250 • C) was ascribed to the hydroxyl groups (OH − ) which are capable of accepting protons serving as the Brønsted basic sites. The CO 2 desorption peaks at medium-(250-550 • C) and high-(>550 • C) temperature regions were attributed to the various oxygen species (e.g., O 2 2− , O − ) which are able to donate electrons serving as the Lewis basic sites. It has been reported [11,13,15,32,33] that medium and strong surface basicity results in an effective electron donation to Co atoms and subsequent electron transfer from Co d-orbitals to the antibonding π* orbitals of N 2 , molecule facilitating the N 2 dissociation by weakening the N≡N bond.

CO 2 -TPD
The CO 2 -TPD profiles recorded for the Co/Mg/La catalysts (Figure 7) revealed that the CO 2 desorption peaks had similar shapes and consisted of four overlapping peaks. The desorption peaks at 116-132 • C were ascribed to the adsorbed CO 2 on weak basic sites (OH − groups) and the peaks at 254-265 • C corresponded to the interaction of CO 2 with medium basic sites of magnesium oxide (Mg 2+ -O 2− pairs) [34]. The desorption peaks of carbon dioxide at 540-559 • C were associated with the interaction of CO 2 with the hexagonal La 2 O 3 , while the peaks at 723-774 • C corresponded to the interaction of CO 2 with the cubic La 2 O 3 [26,28]. Note that the strong basicity of La 2 O 3 led to the formation of respective carbonates upon the CO 2 exposure. Thus, CO 2 desorbed at >500 • C originated from the thermal decomposition of these carbonate species.
Based on the total desorption of CO 2 , the basicity of the catalysts was evaluated. No significant differences in the number of basic sites on the catalysts were observed. The differences in the density of the basic sites were more pronounced, i.e., the basic sites density on the calcined Co/Mg/La reached 12.1 µmol m −2 , which was two times that of the non-calcined Co/Mg/La (5.7 µmol m −2 ). These differences were associated with the thermal pre-treatment method. Applying the typical air calcination method resulted in the crystallite growth (Table S2) and the decrease of the surface area after the hydrogen activation ( Table 1). As a result, the calcined Co/Mg/La catalyst possessed such densely distributed basic sites of high electron-donation ability and made the superior performance of this catalyst for ammonia synthesis possible (Table S1 and Figure 1).   Figure 7 presents the CO2-TPD profiles of the Co/Mg/La catalysts differenced by the thermal pre-treatment method. The profiles could be divided into three kinds of the basic sites at low-, medium-, and high-temperatures relating to the strength of the basic sites. CO2 desorption observed in the low-temperature region (<250 °C ) was ascribed to the hydroxyl groups (OH − ) which are capable of accepting protons serving as the Brønsted basic  medium basic sites of magnesium oxide (Mg 2+ -O 2− pairs) [34]. The desorption peaks of carbon dioxide at 540-559 °C were associated with the interaction of CO2 with the hexagonal La2O3, while the peaks at 723-774 °C corresponded to the interaction of CO2 with the cubic La2O3 [26,28]. Note that the strong basicity of La2O3 led to the formation of respective carbonates upon the CO2 exposure. Thus, CO2 desorbed at >500 °C originated from the thermal decomposition of these carbonate species. Based on the total desorption of CO2, the basicity of the catalysts was evaluated. No significant differences in the number of basic sites on the catalysts were observed. The differences in the density of the basic sites were more pronounced, i.e., the basic sites density on the calcined Co/Mg/La reached 12.1 μmol m −2 , which was two times that of the non-calcined Co/Mg/La (5.7 μmol m −2 ). These differences were associated with the thermal pre-treatment method. Applying the typical air calcination method resulted in the crystallite growth (Table S2) and the decrease of the surface area after the hydrogen activation ( Table 1). As a result, the calcined Co/Mg/La catalyst possessed such densely distributed basic sites of high electron-donation ability and made the superior performance of this catalyst for ammonia synthesis possible (Table S1 and Figure 1). Figure 8 shows the H2-TPD and N2-TPD profiles of the Co/Mg/La catalysts differing by the thermal pre-treatment method. In the temperature range of 50-900 °C , the desorbed H2 and N2 molecules were assigned to three types of sites on the cobalt surface, which were related to the strength of the interaction between these molecules and the Co surface: desorbing below 250 °C (LT, i.e., low-temperature), desorbing between 250 and 550 °C (MT, i.e., medium-temperature) and desorbing above 550 °C (HT, i.e., high-temperature). These were due to the desorption of hydrogen and nitrogen molecules that were weakly, intermediately, and strongly bound to the Co surface, respectively. The H2-TPD profile of non-calcined Co/Mg/La was composed of two rather symmetric hydrogen desorption peaks at 127 and 684 °C , respectively. The desorption peak of H2 at 684 °C showed the  Figure 8 shows the H 2 -TPD and N 2 -TPD profiles of the Co/Mg/La catalysts differing by the thermal pre-treatment method. In the temperature range of 50-900 • C, the desorbed H 2 and N 2 molecules were assigned to three types of sites on the cobalt surface, which were related to the strength of the interaction between these molecules and the Co surface: desorbing below 250 • C (LT, i.e., low-temperature), desorbing between 250 and 550 • C (MT, i.e., medium-temperature) and desorbing above 550 • C (HT, i.e., high-temperature). These were due to the desorption of hydrogen and nitrogen molecules that were weakly, intermediately, and strongly bound to the Co surface, respectively. The H 2 -TPD profile of non-calcined Co/Mg/La was composed of two rather symmetric hydrogen desorption peaks at 127 and 684 • C, respectively. The desorption peak of H 2 at 684 • C showed the largest intensity and was associated with a strong chemisorption state of H 2 on the cobalt surface [5]. In addition to the hydrogen desorption peaks at 123 • C and 429 • C, the peak located at 714 • C with a shoulder at 598 • C appeared in the H 2 -TPD profile of the calcined Co/Mg/La. It can be concluded that the calcination prior to the activation led to the decrease of the adsorption strength on the cobalt surface for H 2 chemisorption. Consequently, the area ratio of medium-temperature to high-temperature peaks (i.e., MT/HT) differed, and the highest ratio was for the calcined Co/Mg/La catalyst (Table 2), indicating the higher presence of intermediate sites on the Co surface for H 2 adsorption. These results indicate that the Co active sites for H 2 adsorption, on which H 2 desorbed at a mediumtemperature range (250-550 • C), were responsible for the synthesis of NH 3 . The sites at a high-temperature region (>550 • C), on which the activation energy for H 2 desorption was high, were not involved in the synthesis of NH 3 (under the reaction conditions, i.e., at temperatures of 400-470 • C). As a result, the dissociative adsorption of nitrogen on the Co active sites was suppressed to some extent, and consequently, so was the activity of the catalyst [35,36]. As revealed in Table 2, ca. 71% of the active sites on the non-calcined Co/Mg/La catalyst were occupied by hydrogen atoms in the strong chemisorption state in comparison to ca. 57% that of calcined Co/Mg/La. This result clearly revealed that a higher number of Co surface atoms were available for NH 3 synthesis under the reaction conditions over the calcined catalyst.  Note that the H2 and N2 molecules competed for the Co active sites on the calcined Co/Mg/La catalyst, which was observed by the corresponding desorption peaks of these molecules in the medium-temperature range, i.e., 250-550 °C (Figure 8). It has been reported [5] that the presence of the active sites for chemisorption of H2 and N2, in which the activation energy for their desorption is comparable, is superior in the terms of activity in NH3 synthesis. This is due to suppressing the inhibition of NH3 production rate by the hydrogen and nitrogen molecules themselves. These results suggest that the enhancement of the performance of the calcined Co/Mg/La catalyst for ammonia synthesis could also be attributable to the high number Co active sites available for ammonia synthesis under reaction conditions and the competitive adsorption of the H2 and N2 molecules on the Co surface. From the chemisorption data, it can be assumed that the different performances of the Co/Mg/La catalysts were related to the nature of the Co active sites. The different nature of the active sites is most likely attributable to the different crystal structure of cobalt. Two crystalline phases of metallic cobalt, i.e., face-centered cubic (FCC) and hexagonal close packed (HCP) have been described in the literature [43]. Kitakami and Sato et al. The N 2 -TPD profiles recorded for the Co/Mg/La catalysts revealed that the desorption profiles of N 2 had a similar shape and consisted of the medium-temperature peak at 313-377 • C with a shoulder at 460-545 • C and high-temperature peak at 712-760 • C. The broad and asymmetric shapes of the medium-temperature peaks were presumably due to the readsorption of desorbed nitrogen molecules within a catalyst bed [37][38][39]. According to Hinrichsen et al. [38,39], a readsorption of N 2 gives rise to an asymmetric peak broadening to the higher temperatures because of the delayed elution of the desorbed nitrogen molecules. The origin of the high-temperature peaks above 700 • C is not clear. Taking into account the weak binding strength between the nitrogen molecules and cobalt, the equilibrium coverages of N 2 on the Co surfaces, in particular, at high temperatures (>550 • C) will be modest [40]. Therefore, it seems reasonable to suggest that N 2 desorbing at the high temperatures did not desorb from the cobalt surfaces. Possibilities could be that N 2 desorbed from the formed new bulk and surface phases partly incorporated nitrogen atoms into the interstitial positions of the cobalt crystal lattice or led to the internal diffusion of N 2 within a porous material [41,42]. Assuming that the desorption of N 2 from the Co surfaces occurred in the medium-temperature range, the rough estimation of the N 2 molecules desorbed was made by integrating the area under the peak to maximum at 313 and 377 • C, respectively, and multiplying by 2. The results of this procedure were 26.3 and 27.2 µmol g −1 of nitrogen desorbed for the non-calcined and calcined Co/Mg/La catalysts, respectively. This corresponded to the 52.4 and 54.4 µmol g −1 active sites assuming a stoichiometry of 1/1 for N-* (*, active site). Based on the number of cobalt surface atoms determined by H 2 -TPD (Table 2), the number of N-* corresponded to the coverage of Θ N = 0.31 and Θ N = 0.30 for the non-calcined and calcined Co/Mg/La, respectively. This is in good accordance with the saturation coverage of Θ N max = 0.33 observed e.g., on the FCC Co(111) surface [40].

H 2 -TPD and N 2 -TPD
Note that the H 2 and N 2 molecules competed for the Co active sites on the calcined Co/Mg/La catalyst, which was observed by the corresponding desorption peaks of these molecules in the medium-temperature range, i.e., 250-550 • C (Figure 8). It has been reported [5] that the presence of the active sites for chemisorption of H 2 and N 2 , in which the activation energy for their desorption is comparable, is superior in the terms of activity in NH 3 synthesis. This is due to suppressing the inhibition of NH 3 production rate by the hydrogen and nitrogen molecules themselves. These results suggest that the enhancement of the performance of the calcined Co/Mg/La catalyst for ammonia synthesis could also be attributable to the high number Co active sites available for ammonia synthesis under reaction conditions and the competitive adsorption of the H 2 and N 2 molecules on the Co surface.
From the chemisorption data, it can be assumed that the different performances of the Co/Mg/La catalysts were related to the nature of the Co active sites. The different nature of the active sites is most likely attributable to the different crystal structure of cobalt. Two crystalline phases of metallic cobalt, i.e., face-centered cubic (FCC) and hexagonal close packed (HCP) have been described in the literature [43]. Kitakami and Sato et al. [44,45] found that there is a close relationship between the particle size and crystal structure of cobalt. Co fine nanoparticles, synthesized by magnetron sputtering, were reported to consist of pure FCC Co for particles smaller than 20 nm, while the concentration of HCP Co increases significantly with size to ca. 85% in particles larger than 40 nm. However, the heterogenous catalysts are often very complex solid systems, and e.g., the presence of support, promoter, and thermal pre-treatment method influence the actual crystal structure of the Co active species. Such changes in the crystal structure of metallic cobalt play an important role in the performance of the catalysts [46,47]. For example, the effect of the cobalt particle size in Fischer-Tropsch synthesis (FTS) was studied with carbon nanofiber-supported Co catalysts [48]. The results showed that CO hydrogenation activity was independent of the cobalt particle size with sizes larger than 6 nm, while the activity decreased when the cobalt particle size was smaller. This demonstrated that there was a minimal required cobalt particle size for FTS, i.e., 6 nm, and that was explained by classical structure sensitivity. In the case of the reaction of NH 3 synthesis, Rambeau et al. [49] indicated that the rate of N 2 chemisorption and equilibrium constants of H 2 chemisorption and of hydrogenation of N ads were sensitive to the crystal structure of cobalt. It affects the NH 3 synthesis rate and clearly means that ammonia synthesis is a structure-sensitive reaction on metallic cobalt. HCP Co exhibits higher activity, 0.0008 s −1 , which is two times that of FCC Co, 0.0004 s −1 [25]. These results are in agreement with computational studies. Based on a first-principles kinetic study, involving DFT calculations, Zhang et al. [41] showed that HCP Co displayed higher activity than FCC Co for nitrogen molecule dissociation and thus for ammonia synthesis. The origin of the higher activity of HCP Co resulted from the higher intrinsic activity and high density of the favorable active sites on HCP Co that were not available for FCC Co (due to their distinct crystal structures and morphologies) [40,50]. The reports focusing on the hydrogen chemisorption of Co [50][51][52][53][54][55][56][57][58] showed that the adsorption strength of H atoms on the Co surface depends on its crystal structure.
Thus, regarding prior results from the literature, we deduced that the reduction of the Co/Mg/La precursor prepared by air calcination resulted in the larger Co nanoparticles of the crystal structure, which were different than those produced from the reduction of the respective dried precursor. This was because Co oxide agglomerated into the larger particles (aggregates) during the high-temperature calcination due to sintering. The Co nanoparticles produced upon the reduction of these Co oxide species were larger than those produced from the direct reduction of the precipitated Co species (Table 2). Thus, these cobalt particles (>10 nm) could have played an important role in tuning the NH 3 synthesis rate, while the small nanoparticles (<10 nm) were not appropriate. This is because the larger Co nanoparticles are reported to consist of a mixture of FCC and HCP Co phases, while the small Co nanoparticles consist of pure FCC Co phase [45,46]. As it was pointed out, the HCP Co phase gives rise to the various favorable active sites with lower nitrogen and hydrogen dissociation barriers, resulting in the superior activity of the catalyst in NH 3 synthesis. Therefore, it was reasonable to deduce that the calcined Co/Mg/La was composed of metallic Co, involving mainly an HCP structure, coexisting with a minor portion of FCC structure. However, despite the confirmation that the calcined catalyst had the larger Co nanoparticles than that of the non-calcined catalyst (Table 2), the actual crystal structure of these Co nanoparticles was unable to detect ( Figure 2).

Catalysts Synthesis
All chemicals used were p.a. grade from POCH company. The Co/Mg/La catalyst precursors were prepared using a co-precipitation method. An aqueous solution of Co 2+ , Mg 2+ , and La 3+ nitrates was used while maintaining a Mg/La molar ratio of 7 and the Co percentage of ca. 10 wt %. During the co-precipitation, the solution of KOH and K 2 CO 3 was added dropwise to the nitrates solution under vigorous stirring at 30 • C. The co-precipitation was carried out up to pH = 11. The precipitate was kept at 70 • C for 18 h for aging, filtered, and washed until free from NO 3 − , CO 3 2− , OH − , and K + ions. Then, the solid sample was divided into two portions to obtain (1) non-calcined and (2) calcined Co/Mg/La catalyst precursors. The first portion was dried at 120 • C for 18 h, while the second was dried at 120 • C for 18 h and then calcined at 500 • C for 18 h in a muffle furnace. Finally, the catalyst precursors were pressed into the cylindrical pellets, crushed, and sieved to the particle size between 0.2 and 0.63 mm.

Characterization Studies
The specific surface area (S BET ), total pore volume (V p ), and average pore diameter (Φ) of the prepared materials were evaluated by the N 2 physisorption method on the ASAP2020 (Micromeritics) at −196 • C. Before analysis, the catalyst precursors of 0.5 g were degassed under vacuum for 1 h at 50 • C and the next 4 h at 80 • C. The results of the measurements were approximated by the Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) adsorption isotherm models to determine the S BET , V p , and Φ. For determining the S BET , V p , and Φ of the catalysts after treatment in hydrogen atmosphere, the precursors of 0.5 g were subjected to the in situ reduction process in the ASAP 2020 instrument using H 2 flow at temperature of 550 • C for 18 h. Then, the samples were degassed under vacuum for 2 h at 150 • C, and N 2 physisorption measurements were performed. The experimental uncertainty in the S BET , V p , and Φ was ±1%.
H 2 -TPR experiments were performed using a thermogravimetric analyzer (TGA, NETZSCH STA449C) integrated with a quadrupole mass spectrometer NETZSCH QMS Aëolos 403C (QMS; m/e range between 10 and 300). A catalyst precursor sample of 0.03 g was loaded into alumina plates and heated up to 600 • C at a ramping rate of 5 • C min −1 under 50 vol% H 2 /Ar mixture flow of 100 cm 3 min −1 .
Powder X-ray diffraction patterns were recorded at room temperature on a Bruker D8 Advance diffractometer equipped with a LYNXEYE position sensitive detector, using Cu-Kα radiation (λ = 0.15418 nm). The data were collected in the Bragg-Brentano (θ/θ) horizontal geometry (flat reflection mode) between 10 • and 70 • (2θ) in a continuous scan using 0.03 • steps 10 s/step. The diffractometer incident beam path was equipped with a 2.5 • Soller slit and a 1.14 • fixed divergence slit, while the diffracted beam path was equipped with a programmable antiscatter slit (fixed at 2.20 • ), a Ni β-filter, and a 2.5 • Soller slit. Data were collected under standard laboratory conditions (temperature and relative humidity). Using the Scherrer equation (k = 0.89), the average crystallite sizes of respective phases were calculated.
Molecular changes of the cobalt catalysts' structure were determined by DRIFT spectroscopy using a Thermo FTIR iS10 spectrometer equipped with an MCT detector and DRIFT High-Temperature Reaction Chamber (Harrick Scientific Co.). The DRIFT analyses were performed by averaging 128 scans with 4 cm −1 resolution. The DRIFT spectra of the catalyst precursors were performed under 50 cm 3 min −1 in the Ar flow at ambient conditions. The in situ reduction of the catalyst precursors was performed under 50 cm 3 min −1 in the 5 vol% H 2 /N 2 mixture at 550 • C for 3 h; then, the temperature was subsequently decreased to 150 • C, and the spectra of the reduced catalysts were taken under dehydrated conditions at 150 • C in a 50 cm 3 min −1 in the Ar flow.
The SEM micrographs and EDS maps of the samples were acquired by using the Field Emission Scanning Electron Microscope (FE-SEM) (FEI Nova NanoSEM 230) equipped with an EDX spectrometer (EDAX Genesis XM4). The samples were dispersed in alcohol; then, a drop was placed on the carbon stub. In the SEM, the samples were analyzed by electron beam at a low accelerating voltage of 5.0 kV using a secondary electron detector. The SEM images were recorded in a beam deceleration mode in order to show more detailed morphological features of the studied materials. In SEM-EDS measurements, the samples were scanned at 18.0 kV using EDAX Apollo X Silicon Drift Detector (SDD).
The TEM micrographs of the samples were performed by using the Transmission Electron Microscopy JEOL 2100HT and TVIPS camera. Before screening, the samples were put on copper 300 mesh grids covered with formvar/carbon film. CO 2 -TPD measurements were carried out using an AutoChem II 2920 (Micromeritics) chemisorption analyzer. Prior to the experiment, the catalyst precursor sample of 0.5 g was loaded into a U-shape quartz reactor and reduced at 550 • C at a ramping rate of 10 • C min −1 for 18 h in the H 2 flow (40 cm 3 min −1 ). Next, the gas flow was switched to helium for purging at 570 • C for 1 h. The CO 2 adsorption was carried out at 40 • C for 2 h. Then, the sample was rinsed with helium at 40 • C for 1 h to remove physisorbed carbon dioxide. The CO 2 -TPD was carried out in the helium flow (40 cm 3 min −1 ) at a ramping rate of 5 • C min −1 up to 900 • C. The outlet gas was analyzed by a thermal conductivity detector (TCD). The uncertainty in the carbon dioxide uptake was ±3%.
H 2 -TPD measurements were carried out using the same AutoChem II 2920 chemisorption analyzer. Prior to the experiment, the catalyst precursor sample of 0.5 g was loaded into a U-shape quartz reactor and reduced at 550 • C at a ramping rate of 10 • C min −1 for 18 h in the H 2 flow (40 cm 3 min −1 ). Next, the gas flow was switched to argon for purging at 570 • C for 1 h. The H 2 adsorption was carried out at 150 • C for 15 min, and it was continued when cooling to 0 • C and left for a further 15 min at 0 • C. Then, the sample was rinsed with argon at 0 • C for 1 h to remove physisorbed hydrogen. The H 2 -TPD was carried out in the argon flow (40 cm 3 min −1 ) at a ramping rate of 5 • C min −1 up to 900 • C. The outlet gas was analyzed by a thermal conductivity detector (TCD). Based on the hydrogen uptake, the cobalt dispersion (i.e., fraction exposed, FE) and its average particle size (d H 2 ) were calculated [59,60]. The uncertainty in the hydrogen uptake was ±3%, and the uncertainty in the cobalt dispersion and its average particle size was ±3%. N 2 -TPD measurements were carried out using the same AutoChem II 2920 chemisorption analyzer. Prior to the experiment, the catalyst precursor sample of 0.5 g was loaded into a U-shape quartz reactor and reduced at 550 • C at a ramping rate of 10 • C min −1 for 18 h in the H 2 flow (40 cm 3 min −1 ). Next, the gas flow was switched to argon for purging at 570 • C for 1 h. The N 2 adsorption was carried out at 200 • C for 14 h, and it was continued when cooling to 0 • C. Then, the sample was rinsed with helium at 0 • C for 1 h to remove physisorbed nitrogen. Furthermore, the temperature was increased from 0 to 900 • C at a ramping rate of 5 • C min −1 in the helium flow (40 cm 3 min −1 ). The outlet gas was monitored by a thermal conductivity detector (TCD). The uncertainty in the nitrogen uptake was ±3%.

Conclusions
The influence of the calcination process on the physicochemical properties (e.g., the textural, structural, and basic properties) and catalytic behavior of the Co/Mg/La catalysts for ammonia synthesis have been investigated. The application of the conventional calcination method prior to the activation resulted in the decrease of the BET surface area, pore volume, and the increase of the surface basicity of the catalyst and the particle size of metallic cobalt. This had a substantial effect on the chemisorption properties and nature of the Co active sites, and thus, it affected the activity for ammonia synthesis. The most favorable catalytic performance was achieved with the Co/Mg/La catalyst prepared by the air calcination method. This catalyst was over twice more active than the corresponding non-calcined catalyst at all the reaction conditions tested. The superior performance of this catalyst was attributable to the following: (1) Strong basicity, (2) More availability of the Co active sites under the reaction conditions, and (3) Competitive adsorption between the H 2 and N 2 molecules.
These findings emphasize the essential role of the materials synthesis procedure in the designing of better, stable catalysts with maximum mass-specific reactivity.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/catal11080941/s1, Figure S1: N 2 adsorption-desorption isotherms (left) and pore size distribution curves (right) of the calcined (--) and non-calcined (--) Co/Mg/La catalysts in the precursor and reduced forms, Figure S2: SEM micrographs and EDS maps of the Co, Mg, and La elements of the Co/Mg/La catalysts in the reduced form, Table S1: Catalytic activity of the Co/Mg/La catalysts in NH 3 synthesis, Table S2: Average crystallite sizes of the respective phases presented of the Co/Mg/La catalysts in the precursor and reduced forms, Table S3: EDS analysis of the selected areas (see Figure S2) of the Co/Mg/La catalysts in the reduced form.