*2.3. Chemisorption Characteristics of the Active Phase Surface (H2-TPD)*

The profiles of hydrogen desorbing from the surface of the promoted cobalt catalysts with different barium content are presented in Figure 2. The hydrogen desorption curves obtained for the samples containing 0–1.1 mmol Ba gCo −1 consist of one broad peak in the low-temperature range (α), extending from about 50 ◦C to 550 ◦C. Its maximum is observed at a temperature of about 170 ◦C. In the profiles of the samples containing 1.4 mmol Ba gCo <sup>−</sup><sup>1</sup> and more (i.e., in the cases where barium is in molar excess to cerium), apart from the low-temperature peak (α), a high-temperature peak (β) appears with a maximum in the range 520–550 ◦C. The low-temperature peak shifts slightly towards lower temperatures as the barium content in the samples increases. In the case of the high-temperature peak, the maximum changes its position slightly. However, there is no clear trend of this change. The low-temperature signal (α) corresponds to the desorption

of hydrogen weakly bound to the cobalt surface. In contrast, the high-temperature signal (β) corresponds to the desorption of H<sup>2</sup> strongly interacting with cobalt [32]. This means that samples containing barium in the content range of 0–1.1 mmol Ba gCo <sup>−</sup><sup>1</sup> have only weak hydrogen-binding sites on their surface. In contrast, on the surface of cobalt in the samples containing 1.4 mmol Ba gCo <sup>−</sup><sup>1</sup> and more, both weakly and strongly hydrogenbinding sites coexist. As the barium content increases, the intensity of the low-temperature peak decreases. The area ratio (β/α, Table 2) of peaks corresponding to strongly and weakly-binding sites on the surface of catalysts increases with barium content above 1.4 mmol Ba gCo −1 , and reaches the highest value for the CoCeBa(2.2) system. Moreover, for the two systems with the highest barium content, CoCeBa(2.2) and CoCeBa(2.6), the high-temperature (β) peak begins to dominate the low-temperature (α) one in terms of the area. The observed phenomena, i.e., peak sharpening, slight temperature shifts of their location, appearance of new peaks, indicate the restructuration of the cobalt systems surface, occurring with the increase in barium content. Not only does the number of hydrogen-binding sites change, but the homogenization of their energy and formation of new types of sites also becomes visible. Therefore, it may be stated that these results support the conclusion that barium exhibits the role of a structural promoter in the studied cobalt catalysts. *Catalysts* **2022**, *12*, x FOR PEER REVIEW 6 of 15

**Figure 2.** Profiles of hydrogen desorption from the surface of the cobalt catalyst promoted with cerium and with a barium loading in the range: 0–2.6 mmol gCo<sup>−</sup>1. **Figure 2.** Profiles of hydrogen desorption from the surface of the cobalt catalyst promoted with cerium and with a barium loading in the range: 0–2.6 mmol gCo <sup>−</sup>1.

in the active phase surface, compared to that of the catalyst without the barium promoter (sample CoCe). The addition of a larger amount of barium (in the range of 0.5–1.6 mmol Ba gCo−1) does not significantly change the Co surface area—for all these systems, the SCo value is constant and amounts to approx. 10 m2 gCo−1. However, a further increase in the content of the barium promoter, i.e., above 1.6 mmol Ba gCo−1, causes the surface of the active phase to gradually decrease. The largest cobalt particle size (dCo-TPD) was determined, and thus the lowest metallic cobalt surface (SCo) was observed for CoCeBa(2.6),

*2.4. Phase Composition of the Precursors and Catalysts in the Reduced form (XRPD)* 

Based on the hydrogen desorption curves (Figure 2) and calculated H2 uptake (Table 2), the average size of metallic cobalt particles (dCo-TPD) and the active phase surface area

The phase composition of the selected promoted cobalt catalysts was analyzed using XRPD. The materials in the oxidized (catalyst precursors) and reduced (catalysts) forms were investigated. The recorded diffraction patterns are presented in Figure 3. Reflexes from Co3O4 are visible in the diffraction patterns of all oxidized samples (Figure 3a). How-

which has the highest Ba content.


**Table 2.** Chemisorption characteristics of the promoted cobalt catalysts.

<sup>1</sup> SCo—surface area of the active phase (cobalt) estimated based on H2-TPD measurement results. <sup>2</sup> dCo-TPD—average cobalt particle size estimated based on H2-TPD measurement results.

Based on the hydrogen desorption curves (Figure 2) and calculated H<sup>2</sup> uptake (Table 2), the average size of metallic cobalt particles (dCo-TPD) and the active phase surface area (SCo) were determined. The data presented in Table 2 show that even a small addition of the barium promoter (0.2 mmol Ba gCo −1 ) results in a significant, i.e., about 33%, increase in the active phase surface, compared to that of the catalyst without the barium promoter (sample CoCe). The addition of a larger amount of barium (in the range of 0.5–1.6 mmol Ba gCo −1 ) does not significantly change the Co surface area—for all these systems, the SCo value is constant and amounts to approx. 10 m<sup>2</sup> gCo −1 . However, a further increase in the content of the barium promoter, i.e., above 1.6 mmol Ba gCo −1 , causes the surface of the active phase to gradually decrease. The largest cobalt particle size (dCo-TPD) was determined, and thus the lowest metallic cobalt surface (SCo) was observed for CoCeBa(2.6), which has the highest Ba content.

#### *2.4. Phase Composition of the Precursors and Catalysts in the Reduced form (XRPD)*

The phase composition of the selected promoted cobalt catalysts was analyzed using XRPD. The materials in the oxidized (catalyst precursors) and reduced (catalysts) forms were investigated. The recorded diffraction patterns are presented in Figure 3. Reflexes from Co3O<sup>4</sup> are visible in the diffraction patterns of all oxidized samples (Figure 3a). However, there are no reflexes from CeO2. This may indicate the presence of a weakly crystallized or amorphous and/or highly dispersed cerium oxide. The presence of signals from two different Ba-containing phases is observed. Barium nitrate signals are clearly visible for the samples with high barium content. Barium nitrate was used for impregnation and the introduction of a substantial amount of this salt could cause a crystallization of this compound in the form of larger particles, detectable by XRPD. For the samples with low barium content, the amount of salt could be too small to form particles of a size appropriate for XRPD analysis or they were better dispersed within the samples. For the catalysts in the reduced form (Figure 3b), signals derived from Ba(NO3)<sup>2</sup> are not detected. According to the literature reports [33], Ba(NO3)<sup>2</sup> is transformed into amorphous BaO<sup>x</sup> species under ammonia synthesis reaction conditions. The subsequent reaction of BaO<sup>x</sup> species with atmospheric CO<sup>2</sup> could cause the formation of BaCO<sup>3</sup> particles (exsitu XRPD measurements for the catalyst samples removed from the ammonia synthesis reactor). Drying barium nitrate at 120 ◦C should not cause the decomposition of the compound. According to the author of [34], the decomposition of pure barium nitrate occurs above 530 ◦C. Nevertheless, the dispersion of barium nitrate on the surface of another material significantly lowers the decomposition temperature. During drying of the catalyst precursors, the dispersed salt could presumably be partially decomposed into BaO, and due to contact with air (containing CO2), it could transform into BaCO3. Hence, there are visible signals of this phase in the catalyst precursor samples (Figure 3a).

phase in the catalyst precursor samples (Figure 3a).

**Figure 3.** XRPD patterns of the cobalt catalysts promoted with cerium (CoCe) or cerium and barium (CoCeBa) in the form of a precursor (**a**) and the reduced form (**b**). **Figure 3.** XRPD patterns of the cobalt catalysts promoted with cerium (CoCe) or cerium and barium (CoCeBa) in the form of a precursor (**a**) and the reduced form (**b**).

ever, there are no reflexes from CeO2. This may indicate the presence of a weakly crystallized or amorphous and/or highly dispersed cerium oxide. The presence of signals from two different Ba-containing phases is observed. Barium nitrate signals are clearly visible for the samples with high barium content. Barium nitrate was used for impregnation and the introduction of a substantial amount of this salt could cause a crystallization of this compound in the form of larger particles, detectable by XRPD. For the samples with low barium content, the amount of salt could be too small to form particles of a size appropriate for XRPD analysis or they were better dispersed within the samples. For the catalysts in the reduced form (Figure 3b), signals derived from Ba(NO3)2 are not detected. According to the literature reports [33], Ba(NO3)2 is transformed into amorphous BaOx species under ammonia synthesis reaction conditions. The subsequent reaction of BaOx species with atmospheric CO2 could cause the formation of BaCO3 particles (ex-situ XRPD measurements for the catalyst samples removed from the ammonia synthesis reactor). Drying barium nitrate at 120 °C should not cause the decomposition of the compound. According to the author of [34], the decomposition of pure barium nitrate occurs above 530 °C. Nevertheless, the dispersion of barium nitrate on the surface of another material significantly lowers the decomposition temperature. During drying of the catalyst precursors, the dispersed salt could presumably be partially decomposed into BaO, and due to contact with air (containing CO2), it could transform into BaCO3. Hence, there are visible signals of this

In the case of reduced samples, metallic cobalt is observed (Figure 3b). There are visible signals typical for hexagonal close-packed cobalt (Co hcp: 41.6° and 47.5°) and facecentered cubic cobalt (Co fcc, 51.5°). The reflexes at the 2θ angles = 44.4°, 75.9°, and 92.4° may come from both phases—Co hcp and Co fcc. It is also worth noting that the form of promoter results from the interaction between barium and cerium compounds in the samples. In CoCeBa(0.2), the cerium promoter is present in the form of two chemical compounds: cerium oxide (CeO2) and barium cerate (BaCeO3), whereas barium is only observed in the form of BaCeO3. As indicated in Table 1, in sample CoCeBa(0.2), cerium is present in molar excess to barium, so the phase composition determined by XRPD is consistent with the chemical composition. For other samples (i.e., CoCeBa(1.4), CoCe(2.0), and CoCeBa(2.2)), where the Ba/Ce molar ratio is greater than unity (Table 1), no cerium oxide In the case of reduced samples, metallic cobalt is observed (Figure 3b). There are visible signals typical for hexagonal close-packed cobalt (Co hcp: 41.6◦ and 47.5◦ ) and face-centered cubic cobalt (Co fcc, 51.5◦ ). The reflexes at the 2θ angles = 44.4◦ , 75.9◦ , and 92.4◦ may come from both phases—Co hcp and Co fcc. It is also worth noting that the form of promoter results from the interaction between barium and cerium compounds in the samples. In CoCeBa(0.2), the cerium promoter is present in the form of two chemical compounds: cerium oxide (CeO2) and barium cerate (BaCeO3), whereas barium is only observed in the form of BaCeO3. As indicated in Table 1, in sample CoCeBa(0.2), cerium is present in molar excess to barium, so the phase composition determined by XRPD is consistent with the chemical composition. For other samples (i.e., CoCeBa(1.4), CoCe(2.0), and CoCeBa(2.2)), where the Ba/Ce molar ratio is greater than unity (Table 1), no cerium oxide phase is observed, as Ce is likely to be bound entirely in the form of barium cerate.

phase is observed, as Ce is likely to be bound entirely in the form of barium cerate. Based on the results obtained from XRPD measurements, the average cobalt oxide crystallite size (dCo3O4-XRD) for the precursor samples and the average metal cobalt crystallite Based on the results obtained from XRPD measurements, the average cobalt oxide crystallite size (dCo3O4-XRD) for the precursor samples and the average metal cobalt crystallite size (dCo-XRD) for the reduced samples (ex-situ measurements) were estimated and are presented in Table 3. The average size of cobalt oxide crystallites (dCo3O4-XRD) in all the samples is similar and equals approx. 11 nm. This result may suggest that barium has no structure-forming effect on cobalt in the case of oxidized materials, i.e., it does not increase or decrease the surface area of the cobalt oxide. The estimated average crystallite size of the metallic cobalt in the reduced samples (dCo-XRD) are also similar (in the range of 21–26 nm) and much lower than the dCo-TPD values calculated based on the chemisorption measurements (Table 2). This may be because the XRPD method can determine small cobalt crystallites, structurally ordered fragments of larger aggregates (agglomerates). However, during chemisorption measurements, only the outer surface of the particles is available for the adsorbate. Consequently, the values of dCo-TPD related to cobalt particles may be greater than the values of dCo-XRD related to cobalt crystallites.

#### *2.5. Morphology and Element Distribution of the Catalysts in the Reduced form (SEM-EDX)*

Figure 4 contains SEM images of the selected catalysts in the reduced form. They show that the morphology of CoCe and CoCeBa(0.2) samples is similar. Both materials consist of nanoparticles formed into larger grains. Although the images show the surface morphology regardless of its composition, they confirm previous observations and conclusions drawn for the active phase of the catalyst from H2-TPD and XRPD analyses (Table 2—dCo-TPD values and Table 3—dCo-XRD values, respectively). It was then found that the differences observed between the cobalt crystallite sizes estimated based on these two methods result from the fact that the crystallites of the active phase with an ordered crystal structure (detectable by the XRPD method) may aggregate into larger particles (agglomerates). Their size resulting from the development of their external surface, accessible to gaseous probe molecules, is estimated based on H2-TPD data. The SEM analysis also confirms that CoCeBa(1.4) is a catalyst with a well-developed surface. In fact, it has the most developed surface among the samples tested with this method, which is in good agreement with the results of textural studies of the reduced form of this sample. It can also be seen (Figure 4) that CoCeBa(2.2) differs in morphology from the other samples due to high barium content. The particles which form the grains of the catalyst containing 2.2 mmol Ba gCo <sup>−</sup><sup>1</sup> are much larger than in the case of the other tested catalysts. Consequently, the sample's surface is smaller, i.e., less developed. These observations are also consistent with the results of the specific surface area after reduction (Table 1—S<sup>R</sup> values) and active phase surface area (Table 2, SCo values) obtained for the studied catalysts.


**Table 3.** The crystallite sizes of Co-containing phases of the promoted cobalt catalysts.

<sup>1</sup> The mean cobalt oxide crystallite size (dCo3O4-XRD) for the precursor samples. <sup>2</sup> The mean metal cobalt crystallite size (dCo-XRD) for the reduced samples (ex-situ measurements).

**Figure 4.** SEM images and corresponding BSE images of selected promoted cobalt catalysts in the reduced form (ex-situ measurement): (**a**) CoCe, (**b**) CoCeBa(0.2), (**c**) CoCeBa(1.4), (**d**) CoCeBa(2.2). The selected grains taken into account in the element distribution analysis (EDX) are marked in red. **Figure 4.** SEM images and corresponding BSE images of selected promoted cobalt catalysts in the reduced form (ex-situ measurement): (**a**) CoCe, (**b**) CoCeBa(0.2), (**c**) CoCeBa(1.4), (**d**) CoCeBa(2.2). The selected grains taken into account in the element distribution analysis (EDX) are marked in red.

The distribution of Co, Ce, and Ba elements on the surface of the tested catalyst samples in their reduced form was determined using EDX analysis. The results of the relative ratios of the elements in three randomly selected points (Figure 4) on the surface of the samples are presented in Table 4. In all the tested catalysts, at each of the selected measuring points, the Co/Ce ratio is similar, which indicates a uniform distribution of cobalt and cerium on the catalyst surface. This is due to the preparation of the catalyst precursor by co-precipitation, which ensures a good distribution of the cerium promoter throughout the sample. However, in the Ba-promoted samples, the distribution of barium with respect to cobalt is not uniform. The Co/Ba ratio discrepancies between selected points may be due to the method of sample preparation, i.e., incipient wetness impregnation method, which does not ensure uniform deposition of the promoting element on the catalyst sur-The distribution of Co, Ce, and Ba elements on the surface of the tested catalyst samples in their reduced form was determined using EDX analysis. The results of the relative ratios of the elements in three randomly selected points (Figure 4) on the surface of the samples are presented in Table 4. In all the tested catalysts, at each of the selected measuring points, the Co/Ce ratio is similar, which indicates a uniform distribution of cobalt and cerium on the catalyst surface. This is due to the preparation of the catalyst precursor by co-precipitation, which ensures a good distribution of the cerium promoter throughout the sample. However, in the Ba-promoted samples, the distribution of barium with respect to cobalt is not uniform. The Co/Ba ratio discrepancies between selected points may be due to the method of sample preparation, i.e., incipient wetness impregnation method, which does not ensure uniform deposition of the promoting element on the catalyst surface.

**Co Ce Ba Co/Ce Co/Ba** 

7.7 7.6

7.6 7.5

8.8 8.6

8.0 8.6 - -

28.6 51.1

> 6.8 6.0

> 7.1 4.7


3.0 1.7

11.7 13.0

11.1 15.9

**Catalyst Point Element Share (%) Elements Ratio** 

3. 88.0 12.0 - 7.3 -

3. 85.6 11.3 3.1 7.6 27.6

3. 69.2 7.8 23.0 8.9 3.0

3. 51.5 6.3 2.2 8.2 23.4

11.5 11.6

11.3 11.5

> 9.0 9.1

> 9.9 8.8

points) on the surface of the promoted cobalt catalysts.

88.5 88.4

85.7 86.8

79.4 77.9

79.1 75.3

face.

2.

2.

2.

2.

CoCe 1.

CoCeBa(0.2) 1.

CoCeBa(1.4) 1.

CoCeBa(2.2) 1.


**Table 4.** Distribution of Co, Ce, and Ba (relative ratio of the elements in three randomly selected points) on the surface of the promoted cobalt catalysts.

#### *2.6. Activity in NH<sup>3</sup> Synthesis (Catalytic Activity Measurements)* Measurements of the catalyst activity were carried out, and the average reaction rate

*2.6. Activity in NH3 Synthesis (Catalytic Activity Measurements)* 

Measurements of the catalyst activity were carried out, and the average reaction rate (rav) of ammonia synthesis was determined. Based on rav values and H<sup>2</sup> uptake values (from H2-TPD measurements), the surface activity of the catalyst, expressed as the turnover frequency (TOF), was determined. The results are shown in Figure 5. (rav) of ammonia synthesis was determined. Based on rav values and H2 uptake values (from H2-TPD measurements), the surface activity of the catalyst, expressed as the turnover frequency (TOF), was determined. The results are shown in Figure 5.

**Figure 5.** Dependence of activity of the promoted cobalt catalysts on the barium promoter content. Activity expressed as an average NH3 synthesis reaction rate (rav, ●) and TOF (■); measurement conditions: T = 400 °C, p = 6.3 MPa, H2/N2 = 3; TOF was determined based on rav values and the hydrogen chemisorption data. **Figure 5.** Dependence of activity of the promoted cobalt catalysts on the barium promoter content. Activity expressed as an average NH<sup>3</sup> synthesis reaction rate (rav, •) and TOF (); measurement conditions: T = 400 ◦C, p = 6.3 MPa, H2/N<sup>2</sup> = 3; TOF was determined based on rav values and the hydrogen chemisorption data.

The addition of a small amount of barium (0.2 mmol Ba gCo−1) results in an almost five-fold increase in average reaction rate and a 3.5-fold increase in the TOF value, compared to the catalysts without barium (CoCe). With a further increase in barium content, i.e., in the range of 0.2–1.4 mmol Ba gCo−1, the rav value increases, which may be a direct result of the development of the cobalt surface (SCo, Table 2) in the catalysts. The average reaction rate (rav) reaches a maximum value for CoCeBa(1.4), and decreases with a further increasing of the barium content. The gradual increase in the activity with the addition of the barium promoter is a result of an electronic effect of barium. Its presence causes a donation of electrons to the cobalt surface, which then facilitates the cleavage of adsorbed dinitrogen. This function of the alkaline dopant was also indicated in our previous studies of the discussed cobalt systems [15,27,35], other cobalt catalysts [17,36,37], and ruthenium The addition of a small amount of barium (0.2 mmol Ba gCo −1 ) results in an almost fivefold increase in average reaction rate and a 3.5-fold increase in the TOF value, compared to the catalysts without barium (CoCe). With a further increase in barium content, i.e., in the range of 0.2–1.4 mmol Ba gCo −1 , the rav value increases, which may be a direct result of the development of the cobalt surface (SCo, Table 2) in the catalysts. The average reaction rate (rav) reaches a maximum value for CoCeBa(1.4), and decreases with a further increasing of the barium content. The gradual increase in the activity with the addition of the barium promoter is a result of an electronic effect of barium. Its presence causes a donation of electrons to the cobalt surface, which then facilitates the cleavage of adsorbed dinitrogen. This function of the alkaline dopant was also indicated in our previous studies of the discussed cobalt systems [15,27,35], other cobalt catalysts [17,36,37], and ruthenium

catalysts for ammonia synthesis [2,11,12,19–24]. Moreover, it is worth nothing that no sign

in industrial processes. When analyzing the surface activity of the catalysts, it should be noted that the TOF value initially increases with increasing barium content in the samples. Then, for samples containing barium in the range of 1.1–2.6 mmol Ba gCo−1, it reaches a constant value of approx. 0.2 s−1. Thus, the decrease in the average reaction rate observed for the samples with the highest barium content may be related to the substantial decrease

The superior performance of the CoCeBa(1.4) catalyst for ammonia synthesis is revealed by comparison to the literature results with similar reaction conditions (Table 5). It can be seen that the ruthenium catalysts display higher NH3 synthesis rates, compared to the iron and cobalt catalysts. However, the activity of the CoCeBa(1.4) catalyst is much higher than that of the commercial fused iron catalyst (about three-fold). Thus, it might

be considered as a valuable alternative to the iron catalyst for ammonia synthesis.

of the active phase surface (SCo, Table 2).

catalysts for ammonia synthesis [2,11,12,19–24]. Moreover, it is worth nothing that no sign of deactivation of the studied catalysts was observed after overheating (600 ◦C, 72 h), indicating that all the catalysts display stable performance—a critical parameter, especially in industrial processes. When analyzing the surface activity of the catalysts, it should be noted that the TOF value initially increases with increasing barium content in the samples. Then, for samples containing barium in the range of 1.1–2.6 mmol Ba gCo −1 , it reaches a constant value of approx. 0.2 s−<sup>1</sup> . Thus, the decrease in the average reaction rate observed for the samples with the highest barium content may be related to the substantial decrease of the active phase surface (SCo, Table 2).

The superior performance of the CoCeBa(1.4) catalyst for ammonia synthesis is revealed by comparison to the literature results with similar reaction conditions (Table 5). It can be seen that the ruthenium catalysts display higher NH<sup>3</sup> synthesis rates, compared to the iron and cobalt catalysts. However, the activity of the CoCeBa(1.4) catalyst is much higher than that of the commercial fused iron catalyst (about three-fold). Thus, it might be considered as a valuable alternative to the iron catalyst for ammonia synthesis.

**Table 5.** The comparison of NH<sup>3</sup> synthesis rate (rNH3) over cobalt, iron, and ruthenium catalysts under the pressure of about 6 MPa and 400 ◦C.


Summarizing the presented results, it should be stated that the catalytic properties of the cobalt systems doubly promoted with cerium and barium strictly depend on the content of barium. When the cerium promoter is present in molar excess to barium (Ba/Ce < 1, Table 1), barium acts as a typical structural promoter. It prevents the sintering of cobalt particles during reduction, causing the development of the active phase surface and thus an increase in the activity of the catalysts in ammonia synthesis. However, in cases where the barium to cerium ratio is greater than unity (Ba/Ce > 1, Table 1), the modifying (electronic) character of the barium promoter is also revealed. It was observed that despite the decrease in cobalt surface, the surface activity (TOF values) of the catalysts containing more than 1.4 mmol Ba gCo −1 remained at a high and stable level. However, considering our previous investigation of the synergistic effect of the cerium and barium promoters in the cobalt catalyst [15,28], the properties of the cobalt–cerium–barium systems should also be related to the presence of the BaCeO<sup>3</sup> phase. For barium-rich catalysts (Ba/Ce > 1), the binding of the entire cerium promoter in the form of BaCeO<sup>3</sup> ensures that in all these systems, the amount of this third promoter, which exhibits a strong electron- donating effect on the cobalt surface, is similar. This is reflected by the nearly constant TOF value, indicating a similar surface activity of the active phase of these materials. The lack of a free cerium promoter in the form of CeO<sup>2</sup> (i.e., not bound in BaCeO3) causes the decay of the structural influence of cerium, which explains the decrease in the active phase surface area (Table 2). The effect of the decrease in the cobalt surface for the barium-rich catalysts may also be associated with the phenomenon of surface enrichment with barium, in which the barium promoter introduced in excess in relation to cerium may accumulate on the cobalt particles, blocking the access of the reagents to the active sites of the catalyst. This phenomenon was previously observed in our studies of cobalt systems promoted with barium [36]. Based on the results of the conducted experiments, the optimal barium promoter content in the CoCeBa catalyst was established. The most favorable properties were obtained for the catalytic systems containing 1.1–1.6 mmol Ba gCo −1 .

#### **3. Materials and Methods**

#### *3.1. Preparation of the Catalysts*

In the first step, a mixture of cobalt and cerium oxides (Co3O<sup>4</sup> + CeO2) was prepared using the co-precipitation method and subsequent calcination. Appropriate amounts of cobalt(II) nitrate hexahydrate and cerium(III) nitrate hexahydrate were dissolved in distilled water. Excess potassium carbonate aqueous solution was slowly added under continuous stirring to the nitrates solution until the pH was 9. Both of the solutions were first heated to 90 ◦C. The obtained precipitate was filtered and washed with cold distilled water until the pH was neutral. It was then dried at 120 ◦C in air for 18 h and calcined at 500 ◦C in air for 18 h. Afterwards, the material was impregnated with various amounts of barium using an aqueous solution of barium(II) nitrate (incipient wetness impregnation) and dried in air at 120 ◦C. Finally, the samples were crushed and sieved to obtain grain size in the range of 0.20–0.63 mm. The last step of catalyst preparation was the reduction of precursors carried out directly before measurements, which required a reduced form of the materials and before the catalytic activity studies (details can be found below in a characterization methods description, Section 3.2.). As a result, a series of doubly promoted cobalt catalysts were obtained of cerium content equal to 1.1 mmol gCo −1 , while the barium content varied in the range of 0–2.6 mmol gCo −1 . Cerium content was determined using thermal analysis coupled with mass spectrometry according to the methodology described in [40] for the precursor containing only Co3O<sup>4</sup> and CeO<sup>2</sup> (i.e., the precursor before impregnation with barium). The basis of the discussed method is the fact that under the measurement conditions (heating in argon), cerium oxide is stable, whereas cobalt (II,III) oxide decomposes to cobalt (II) oxide at a temperature of about 750 ◦C. The recorded mass loss allows the determination of the Co3O<sup>4</sup> content in a mixed oxide system. The rest of the sample consists of CeO2. The content of barium promoter in the final precursor samples (before reduction) was calculated based on the mass balance before and after impregnation with barium salt of the precursor samples containing Co3O<sup>4</sup> and CeO2. Barium content and a molar ratio of barium to cerium are listed in Table 1. Materials are denoted as CoCeBa(n), where n is the amount of barium in relation to cobalt, as indicated in Table 1. The sample without barium, donated as CoCe, was a reference material.

#### *3.2. Catalyst Characterisation*

The specific surface area of the precursors (i.e., materials in the unreduced form), total pore volume, and specific surface area of the reduced form of the selected materials were determined by nitrogen physisorption with an ASAP2020 instrument (Micromeritics Instrument Co., Norcross, GA, USA). Before the measurements, each sample of the precursors was degassed in vacuum in two stages: at 90 ◦C for 1 h and then at 200 ◦C for 4 h. Before the measurement for the selected materials in their reduced form, the precursors were reduced in-situ at 550 ◦C for 10 h in hydrogen flow and then subjected to degassing at 150 ◦C for 2 h. The reduction and degassing were conducted in the apparatus directly before N<sup>2</sup> physisorption measurements.

The morphology and element distribution for the selected catalytic materials in the reduced form was studied using scanning electron microscopy (SEM) coupled with energydispersive X-ray spectroscopy EDX (FEI NovaNanoSEM 230, FEI Company, Hillsboro, OR, USA).

The phase composition of the selected precursors and the catalysts in the reduced form were determined using X-ray powder diffraction (XRPD). Data were collected with a Rigaku-Denki Geigerflex (Rigaku Denki Co., Ltd., Tokyo, Japan) diffractometer in Brag– Brentano configuration using CuKα radiation. The samples were scanned in a 2θ range of 15–100◦ with a step of 0.02◦ and counting time 5 s. The average size of Co3O<sup>4</sup> crystallites (in the precursors) and metallic cobalt crystallites (in the reduced catalysts) was estimated based on the Scherrer equation using the integral width of the reflex filled to the analytical Pearson VII function.

A reducibility of the catalyst precursors was studied using Temperature-Programmed Reduction with hydrogen (H2-TPR) at AutoChem2920 (Micromeritics Instrument Co.). Samples of the precursors containing about 0.03 g of Co3O<sup>4</sup> were heated from room temperature to 700 ◦C at a constant rate of 10 ◦C min−<sup>1</sup> in the flow of 10 vol.% H2/Ar (40 mL min−<sup>1</sup> ). The hydrogen consumption was measured by a Thermal Conductivity Detector (TCD).

The catalysts' active phase surface was characterized using temperature-programmed hydrogen desorption (H2-TPD) using a PEAK-4 instrument. Measurements were conducted in a flow set-up supplied with high purity (99.99995 vol.%) gases (total gas flow rate 40 mL min−<sup>1</sup> ) in a quartz U-tube reactor. Samples of the catalyst precursors containing 0.5 g Co3O<sup>4</sup> + CeO<sup>2</sup> were reduced in a flow of H2/Ar = 4:1 mixture (40 mL min−<sup>1</sup> ) at 550 ◦C for 18 h. The system was then flushed with flowing argon at 570 ◦C for 1 h and cooled to 150 ◦C. The H<sup>2</sup> adsorption was carried out at 150 ◦C for 15 min, then continued during cooling of the sample to 0 ◦C and for 15 min at 0 ◦C. After flushing with Ar to remove weakly bound molecules of H2, the temperature was increased to 550 ◦C at a constant rate (10 ◦C min−<sup>1</sup> ) and then kept for 10 min at 550 ◦C while monitoring the concentration of hydrogen desorbing from the surface of the catalyst. The surface area of the active phase (SCo) and average cobalt particle size (dCo) were calculated assuming H/Co = 1 stoichiometry of hydrogen adsorption [41].

### *3.3. Catalytic Tests*

The activity of the catalysts in ammonia synthesis was tested in a tubular flow reactor under steady-state conditions (6.3 MPa, 400 ◦C, H2/N<sup>2</sup> = 3, gas flow rate 70 dm<sup>3</sup> h −1 ). Before the activity measurements, samples of the catalyst precursors (grain size 0.2–0.63 mm) of about 0.5 g were activated in a high purity H2/N<sup>2</sup> = 3 mixture (99.99995 vol%., gas flow rate 30 dm<sup>3</sup> h −1 ) under atmospheric pressure in accordance with the temperature program: 470 ◦C for 72 h, then 520 ◦C for 24 h and finally 550 ◦C for 48 h. The product concentration in the outlet gas was measured interferometrically. The catalytic activity was determined and expressed as an average NH<sup>3</sup> synthesis reaction rate (rav). A detailed description of the set-up and the method for calculating the reaction rate was described in [42]. Moreover, the activity of the catalyst surface expressed as TOF was estimated. The calculation was based on the values of the average reaction rates (rav) and the number of active sites on the cobalt surface determined during chemisorption measurements (H2-TPD).

#### **4. Conclusions**

In summary, the influence of barium content on the physicochemical properties and catalytic activity of the cobalt catalyst doubly promoted with cerium and barium was investigated. A series of catalysts of various barium promoter content in the range of 0–2.6 mmol gCo <sup>−</sup><sup>1</sup> was prepared, characterized, and tested in ammonia synthesis. The dual nature of the role of the barium promoter(structural and modifying) was revealed, but it strictly depends on the barium-to-cerium molar ratio. For systems of the Ba/Ce molar ratio lower than unity (Ba/Ce < 1), the structural character of barium was observed. It manifested itself mainly in preventing sintering of the active phase during reduction. For the best catalytic performance of the CoCeBa system, the Ba/Ce molar ratio should be greater than unity (Ba/Ce > 1), which results in not only a structural promotion of barium, but also a modifying action associated with the in-situ formation of the BaCeO<sup>3</sup> phase. It was primarily reflected in the differentiation of weakly and strongly binding sites on the catalyst surface and changes of the cobalt surface activity (TOF). The optimal barium content in the range of 1.1–1.6 mmol gCo −1 leads to obtaining a catalyst with the most favorable properties. Its excellent catalytic performance is ascribed to the appropriate Ba/Ce molar ratio. It is also related to the presence of the BaCeO<sup>3</sup> phase, which plays the role of a third promoter of a high electron-donating character.

**Author Contributions:** Conceptualization, A.T., M.Z. and W.R.-P.; methodology, A.T., M.Z. and W.R.-P.; investigation, A.T., M.Z., H.R., W.P., B.M., L.K. and W.R.-P.; writing—original draft preparation, A.T. and M.Z.; writing—review and editing, M.Z., H.R. and W.R.-P.; visualization, A.T., and H.R.; supervision, M.Z.; funding acquisition, W.R.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research has been funded by The National Centre for Research and Development within The Applied Research Programme, grant No. PBS2/A1/13/2014.

**Data Availability Statement:** All data is available within the paper.

**Acknowledgments:** The authors thank Ewa Iwanek from the Faculty of Chemistry, Warsaw University of Technology, for additional proofreading and language corrections.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Paweł Adamski \* , Wojciech Czerwonko and Dariusz Moszy ´nski**

Department of Inorganic Chemical Technology and Environment Engineering, Faculty of Chemical Technology and Engineering, West Pomeranian University of Technology in Szczecin, Piastów Ave. 42, 71-065 Szczecin, Poland; wojciech.czerwonko@zut.edu.pl (W.C.); dmoszynski@zut.edu.pl or dariusz.moszynski@zut.edu.pl (D.M.)

**\*** Correspondence: adamski\_pawel@zut.edu.pl; Tel.: +48-91-449-4024

**Abstract:** The application of cobalt molybdenum nitrides as ammonia synthesis catalysts requires further development of the optimal promoter system, which enhances not only the activity but also the stability of the catalysts. To do so, elucidating the influence of the addition of alkali metals on the structural properties of the catalysts is essential. In this study, potassium-promoted cobalt molybdenum nitrides were synthesized by impregnation of the precursor CoMoO<sup>4</sup> ·3/4H2O with aqueous KNO<sup>3</sup> solution followed by ammonolysis. The catalysts were characterized with the use of XRD and BET methods, under two conditions: as obtained and after the thermal stability test. The catalytic activity in the synthesis of ammonia was examined at 450 ◦C, under 10 MPa. The thermal stability test was carried out by heating at 650 ◦C in the same apparatus. As a result of ammonolysis, mixtures of two phases: Co3Mo3N and Co2Mo3N were obtained. The phase concentrations were affected by potassium admixture. The catalytical activity increased for the most active catalyst by approximately 50% compared to non-promoted cobalt molybdenum nitrides. The thermal stability test resulted in a loss of activity, on average, of 30%. Deactivation was caused by the collapse of the porous structure, which is attributed to the conversion of the Co2Mo3N phase to the Co3Mo3N phase.

**Keywords:** cobalt molybdenum nitrides; ammonia synthesis; phase composition; specific surface area

#### **1. Introduction**

The Haber–Bosch process developed in the early years of the twentieth century had a great influence on the production of ammonia. This process uses an iron catalyst that allows direct bonding of H<sup>2</sup> and N<sup>2</sup> and can be considered efficient; however, due to the huge worldwide production of ammonia, the continuous improvement of catalysts is required for both financial and environmental reasons [1]. An important approach to solving this problem was the application of ruthenium-based catalysts. They exhibit high activity in ammonia synthesis but are burdened by several technological flaws, as well as high cost [2,3]. These led researchers to focus on less expensive and more available materials instead of noble metal-based catalysts [4]. Transition metal nitrides proved to be very effective catalysts that can be obtained as binary [5], ternary [1] and very recently quaternary [6] systems that—with further improvements—can serve as excellent catalysts for ammonia synthesis.

Ternary transition metal nitrides are well-known catalysts for hydrogenation reactions such as hydrosulfurization of thioorganic compounds [7–9], hydroprocessing of organic compounds [10] hydrazine decomposition [11] and NO reduction [12]. For ammonia synthesis, theoretical studies indicate that cobalt molybdenum nitride is the most active among the other chemical substances [13,14].

The studies of cobalt molybdenum nitride promoted by alkali metals were carried out by Kojima and Aika [15]. During these studies, a substantial increase in the catalytic activity, caused by the presence of cesium or potassium in the catalyst, was observed. A specific

**Citation:** Adamski, P.; Czerwonko, W.; Moszy ´nski, D. Thermal Stability of Potassium-Promoted Cobalt Molybdenum Nitride Catalysts for Ammonia Synthesis. *Catalysts* **2022**, *12*, 100. https://doi.org/ 10.3390/catal12010100

Academic Editor: Marco Martino

Received: 22 December 2021 Accepted: 14 January 2022 Published: 16 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

concentration of each of the promoters was required to obtain the optimal catalytic activity. The cesium promoter was found to be more effective compared to potassium [16]. The proposed beneficiary effect of alkali metals on catalyst activity was described in the terms of changes in the electronic properties of the active sites through the electron-donation mechanism [17]. However, studies of other catalytical systems suggest that the effect of the addition of alkali metals is more complicated.

A structural effect of promoter addition was observed in cobalt molybdenum nitride catalysts. In our previous studies [18,19] we reported the complexity of the phase composition of the cobalt molybdenum nitride catalysts, especially the occurrence of two active phases: Co2Mo3N and Co3Mo3N. Their concentration ratio varies with the type and concentration of the applied promoters. Cobalt molybdenum nitride catalysts promoted with chromium [19,20] as well as potassium and chromium [21] were compared. The chromium admixture was found to hinder, and potassium to facilitate the formation of the Co2Mo3N phase. The mechanism that leads to the coexistence of the Co2Mo3N and Co3Mo3N phases was studied in detail elsewhere [22]. The Co2Mo3N phase was shown to be an intermediate phase in the formation of the Co3Mo3N phase. Catalysts with higher concentrations of the Co2Mo3N phase are catalytically more active [18,19].

In addition to the catalytical activity, an important factor for the practical application of catalysts is their thermal stability. Each catalyst changes in the course of the reaction, and its deactivation occurs. For an industrial application, the interval between loads of fresh catalyst should be as long as possible, for example, for the iron catalyst, it is about ten years [23]. In the case of cobalt molybdenum nitrides, the available information about stability during prolonged test runs is limited. The stability of the catalytic properties was examined in detail for the non-promoted and cesium-promoted catalysts by Kojima and Aika [24]. It was shown that during the 24 h process under the atmosphere of the reaction gas N<sup>2</sup> + 3H<sup>2</sup> at 600 ◦C, both the unpromoted catalyst and the Cs-promoted catalysts reached the maximum of their activity after 12 h. Subsequently, their activity gradually decreased. At the same time, their BET surface area decreased by approximately half. The increase in activity was associated with the transformation of the intermediate Co and Mo2N phases into the Co3Mo3N phase.

The stability of the unpromoted cobalt molybdenum nitride catalyst was examined via in situ XRD studies [25]. The phase concentrations of Co2Mo3N and Co3Mo3N under a reaction gas atmosphere at 700 ◦C were stable. The Co3Mo3N phase decomposed into the Co6Mo6N phase under a hydrogen atmosphere. This transformation was explained in terms of the high reactivity of bulk lattice nitrogen present in η-carbide structured Co3Mo3N by Daisley et al. [26].

The outlook on the behavior under industrial reaction conditions of non-promoted and K, Cs, Cr-promoted catalysts before and after thermostability test was given by Nadziejko et al. [27]. It was shown that the specific surface area and activity of the Cspromoted catalysts after the thermostability test decreased the most among the studied catalysts. A serious deactivation for Cs-promoted catalysts was also observed by us, in a detailed study [28]. It was associated with the sintering of the crystallites of the active phases and decreasing of the surface area of the catalysts. Furthermore, in the study of Boisen et al. [5] the influence between the activity of the Cs-promoted catalyst and its surface area was observed. Despite their initial high activity, Cs-promoted catalysts are regarded as highly prone to sintering and are inefficient in the long run.

There are still ambiguities regarding the influence of alkali metals on ammonia synthesis catalysts. In addition to electron transfer from the alkali to the active center, other factors such as the change of surface structure or change of the crystallite sizes must be considered. To address this problem in the present study, the thermal stability of the K-promoted catalyst was examined in detail. Cobalt molybdenum nitrides promoted by potassium were studied as catalysts in ammonia synthesis at 450 ◦C and subsequently after prolonged heating under a H2-N<sup>2</sup> gas mixture at 650 ◦C. The transformation of the phase

structure, porous structure, and catalytic activity of the examined catalysts were related to the concentration of potassium in the catalytic material.

#### **2. Results**

Cobalt molybdenum nitride catalysts modified with an admixture of potassium (henceforth abbreviated as COMON catalysts), were obtained in a three-step process. First, the precipitation of cobalt molybdates, referred to as precursors, was carried out. The precipitation of precursors was followed by their impregnation with the potassium nitrate. Subsequently, the activation process of the catalyst precursors was performed under the flow of pure ammonia (a process called ammonolysis). The obtained catalysts were studied in two chemical states: after ammonolysis of the precursors, and after the activity measurements followed by the aging of the catalysts under increased temperature. More detailed experimental conditions are described in the Section 4.

The X-ray diffraction pattern of the synthesized precursor is consistent with previously published results [20], where the precursor was identified as CoMoO4·3/4H2O (PDF 04- 011-8282). In Figure 1, the XRD pattern acquired after ammonolysis is shown for the exemplary sample containing 0.2 wt.% of potassium. The diffraction reflections observed for the catalysts after ammonolysis, as well as after thermal stability tests, were ascribed exclusively to Co3Mo3N or Co2Mo3N phases. No oxidic phases, metallic cobalt or Mo2N were detected, although the surface of all samples after each process was passivated in diluted oxygen prior to XRD analysis.

**Figure 1.** X-ray diffraction pattern of the COMON catalyst containing 0.2 wt.% of potassium acquired after ammonolysis, with the indicated Rietveld refinement.

The weight fractions of the Co2Mo3N (PDF 04-010-6426) and Co3Mo3N (PDF 04-008- 1301) phases identified in the catalysts were determined by X-ray diffraction analysis with the use of Rietveld refinement. The weight fraction of the Co2Mo3N phase related to the potassium concentration is shown in Figure 2. The Co3Mo3N phase complements the composition of the catalysts. The non-promoted sample contains about 18 wt.% of the cobalt-lean Co2Mo3N phase. The content of this phase in the catalyst grows to about 47 wt.% with increasing potassium concentration to top at 0.8 wt.%. At even higher potassium concentrations, a decrease in Co2Mo3N concentration is observed. At a potassium concentration of 3.5 wt.%, the Co2Mo3N concentration dropped to 18 wt.% of the catalyst. 45

50

*Catalysts* **2022**, *12*, x FOR PEER REVIEW 4 of 10

after ammonolysis

**Figure 2.** Weight fraction of the Co2Mo3N phase as a function of potassium concentration in COMON catalysts. A line is given for eye guidance. **Figure 2.** Weight fraction of the Co2Mo3N phase as a function of potassium concentration in COMON catalysts. A line is given for eye guidance. The catalysts after the thermal stability tests were also subjected to a diffraction analysis, as presented in Figure 2. In general, the relation between the weight fraction of

The catalysts after the thermal stability tests were also subjected to a diffraction analysis, as presented in Figure 2. In general, the relation between the weight fraction of the Co2Mo3N phase and the potassium concentration is similar to that observed for these materials before aging. For the non-promoted catalyst, the share of the Co2Mo3N phase after the aging process was unchanged. However, for the promoted catalysts, the phase composition of the samples changed slightly after the thermostability test and the weight The catalysts after the thermal stability tests were also subjected to a diffraction analysis, as presented in Figure 2. In general, the relation between the weight fraction of the Co2Mo3N phase and the potassium concentration is similar to that observed for these materials before aging. For the non-promoted catalyst, the share of the Co2Mo3N phase after the aging process was unchanged. However, for the promoted catalysts, the phase composition of the samples changed slightly after the thermostability test and the weight fraction of the Co2Mo3N phase was reduced for all promoted catalysts. the Co2Mo3N phase and the potassium concentration is similar to that observed for these materials before aging. For the non-promoted catalyst, the share of the Co2Mo3N phase after the aging process was unchanged. However, for the promoted catalysts, the phase composition of the samples changed slightly after the thermostability test and the weight fraction of the Co2Mo3N phase was reduced for all promoted catalysts. The specific surface area is an important factor that influences the catalytic properties

fraction of the Co2Mo3N phase was reduced for all promoted catalysts. The specific surface area is an important factor that influences the catalytic properties of the catalysts. During the present study, this parameter was measured twice for all of the catalysts: after ammonolysis and after thermostability test. The results are depicted in Figure 3. The specific surface area of fresh, non-promoted COMON catalyst is 15.5 m2/g, a value typical for many carrier-free metallic catalysts [29,30]. The surface area of the remaining catalysts depends on the concentration of potassium. It amounted to 9.2 m2/g for the fresh catalyst containing 0.2 wt.% of potassium and increased in the range of potassium concentration between 0.4 wt.% and 0.8 wt.%. The maximum was observed at 0.8 wt.% of potassium and amounted to 10.7 m2/g. A further increase of potassium concentration resulted in a decrease in the specific surface area. The most prominent loss The specific surface area is an important factor that influences the catalytic properties of the catalysts. During the present study, this parameter was measured twice for all of the catalysts: after ammonolysis and after thermostability test. The results are depicted in Figure 3. The specific surface area of fresh, non-promoted COMON catalyst is 15.5 m2/g, a value typical for many carrier-free metallic catalysts [29,30]. The surface area of the remaining catalysts depends on the concentration of potassium. It amounted to 9.2 m2/g for the fresh catalyst containing 0.2 wt.% of potassium and increased in the range of potassium concentration between 0.4 wt.% and 0.8 wt.%. The maximum was observed at 0.8 wt.% of potassium and amounted to 10.7 m2/g. A further increase of potassium concentration resulted in a decrease in the specific surface area. The most prominent loss was observed for catalysts containing 2.9 wt.% and 3.5 wt.% of potassium, where the specific surface area measured was approximately 5.5 m2/g. of the catalysts. During the present study, this parameter was measured twice for all of the catalysts: after ammonolysis and after thermostability test. The results are depicted in Figure 3. The specific surface area of fresh, non-promoted COMON catalyst is 15.5 m2/g, a value typical for many carrier-free metallic catalysts [29,30]. The surface area of the remaining catalysts depends on the concentration of potassium. It amounted to 9.2 m2/g for the fresh catalyst containing 0.2 wt.% of potassium and increased in the range of potassium concentration between 0.4 wt.% and 0.8 wt.%. The maximum was observed at 0.8 wt.% of potassium and amounted to 10.7 m2/g. A further increase of potassium concentration resulted in a decrease in the specific surface area. The most prominent loss was observed for catalysts containing 2.9 wt.% and 3.5 wt.% of potassium, where the specific surface area measured was approximately 5.5 m2/g.

1

0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 Potassium concentration, wt.% **Figure 3.** Specific surface area of COMON catalysts as a function of potassium concentration, measured after ammonolysis and after all catalytic tests. Lines are given for eye guidance. **Figure 3.** Specific surface area of COMON catalysts as a function of potassium concentration, measured after ammonolysis and after all catalytic tests. Lines are given for eye guidance.

**Figure 3.** Specific surface area of COMON catalysts as a function of potassium concentration,

After thermostability tests, a drop in surface area was observed for all catalysts. In the case of the non-promoted COMON catalyst the surface area decreased insignificantly to 15.0 m2/g. The loss was much more noticeable for COMON catalysts promoted with potassium compounds. For all of these catalysts, the specific surface area after aging decreased by an average of 1–2 m2/g. For example, at a concentration of 1.5 wt.% of potassium, it decreased from about 9.5 to 8.3 m2/g, and at a concentration of 2.9 wt.% of potassium it decreased from about 5.5 to 3.7 m2/g. The most prominent decrease in the specific surface area, by about half, was observed for a catalyst containing 3.5 wt.% of potassium. to 15.0 m2/g. The loss was much more noticeable for COMON catalysts promoted with potassium compounds. For all of these catalysts, the specific surface area after aging decreased by an average of 1–2 m2/g. For example, at a concentration of 1.5 wt.% of potassium, it decreased from about 9.5 to 8.3 m2/g, and at a concentration of 2.9 wt.% of potassium it decreased from about 5.5 to 3.7 m2/g. The most prominent decrease in the specific surface area, by about half, was observed for a catalyst containing 3.5 wt.% of potassium. Catalytic activity measured at 450 °C during the ammonia synthesis reaction carried

After thermostability tests, a drop in surface area was observed for all catalysts. In the case of the non-promoted COMON catalyst the surface area decreased insignificantly

*Catalysts* **2022**, *12*, x FOR PEER REVIEW 5 of 10

Catalytic activity measured at 450 ◦C during the ammonia synthesis reaction carried out under a pressure of 10 MPa for a series of COMON catalysts is shown in Figure 4. Considering fresh catalysts, the one containing 0.2 wt.% of potassium was less active than the non-promoted COMON catalyst. However, with increasing potassium concentration, the catalytic activity grew and its highest value was observed at 1.3% of potassium. In comparison to the non-promoted COMON catalyst, the activity at maximum was about 50% higher. The potassium concentration greater than 1.3% resulted in a drop in catalytic activity. Catalysts containing 2.9% and 3.5% potassium demonstrated very low activity, only about 25% and 10% compared to the non-promoted COMON catalyst, respectively. out under a pressure of 10 MPa for a series of COMON catalysts is shown in Figure 4. Considering fresh catalysts, the one containing 0.2 wt.% of potassium was less active than the non-promoted COMON catalyst. However, with increasing potassium concentration, the catalytic activity grew and its highest value was observed at 1.3% of potassium. In comparison to the non-promoted COMON catalyst, the activity at maximum was about 50% higher. The potassium concentration greater than 1.3% resulted in a drop in catalytic activity. Catalysts containing 2.9% and 3.5% potassium demonstrated very low activity, only about 25% and 10% compared to the non-promoted COMON catalyst, respectively.

**Figure 4.** Catalytic activity of COMON catalysts as a function of the potassium concentration measured after ammonolysis and after the thermostability test. Catalytic tests were carried out at 450 °C under 10 MPa. Lines are given for eye guidance. **Figure 4.** Catalytic activity of COMON catalysts as a function of the potassium concentration measured after ammonolysis and after the thermostability test. Catalytic tests were carried out at 450 ◦C under 10 MPa. Lines are given for eye guidance.

The second activity test was performed after the thermostability test. The catalytic activity of the non-promoted COMON catalyst increased by about 15%. After the thermostability test, significant deactivation was observed for most of the potassiumpromoted COMON catalysts. The catalysts containing 1.3 wt.% and 1.5 wt.% of potassium were still more active than the reference material. However, a decrease of about 15% of their initial activity occurred. In the case of the catalysts containing 2.9 wt.% and 3.5 wt.% of potassium, which were barely active already in the initial stage, the catalytic activity The second activity test was performed after the thermostability test. The catalytic activity of the non-promoted COMON catalyst increased by about 15%. After the thermostability test, significant deactivation was observed for most of the potassium-promoted COMON catalysts. The catalysts containing 1.3 wt.% and 1.5 wt.% of potassium were still more active than the reference material. However, a decrease of about 15% of their initial activity occurred. In the case of the catalysts containing 2.9 wt.% and 3.5 wt.% of potassium, which were barely active already in the initial stage, the catalytic activity remained virtually unchanged.

#### remained virtually unchanged. **3. Discussion**

**3. Discussion**  The literature considering the early studies of cobalt molybdenum nitride catalysts indicates that the compound described by Co3Mo3N stoichiometry is expected to be the only ternary nitride obtained after ammonolysis of the precursor [15]. Metallic cobalt and molybdenum nitride, Mo2N, were observed as intermediates and as decomposition The literature considering the early studies of cobalt molybdenum nitride catalysts indicates that the compound described by Co3Mo3N stoichiometry is expected to be the only ternary nitride obtained after ammonolysis of the precursor [15]. Metallic cobalt and molybdenum nitride, Mo2N, were observed as intermediates and as decomposition products after prolonged gas and heat treatment [7,15,31]. However, in our previous studies [18–21] the occurrence of two cobalt molybdenum nitrides: Co3Mo3N and Co2Mo3N was observed.

products after prolonged gas and heat treatment [7,15,31]. However, in our previous studies [18–21] the occurrence of two cobalt molybdenum nitrides: Co3Mo3N and Unlike previous reports by Kojima and Aika [24], the decomposition products (metallic cobalt or molybdenum nitride Mo2N) were not observed after the thermostability test. Earlier reports indicate that the formation of the Co2Mo3N phase required special treatment [32]. However, a detailed review of older reports indicated that the Co2Mo3N phase also occurred after ammonolysis, but was not adequately identified [33].

In the study focused on the influence of chromium salts on the formation of similar mixtures of cobalt molybdenum nitrides [20], two possible ways of Co2Mo3N formation were considered. The Co2Mo3N phase was considered as a product of the decomposition of Co3Mo3N or as an intermediate on the path to Co3Mo3N formation. Since this phase disappears with increasing ammonolysis temperature, it is not observed in Co3Mo3N decomposition products either, it is supposedly an intermediate product. This is in agreement with our previous study [22], in which data obtained by the in situ XRD method during the non-promoted precursor activation process at 700 ◦C in the presence of a mixture of inert gas and ammonia was analyzed. Co2Mo3N is the first reaction product, which transforms into stable Co3Mo3N by further reconstruction with cobalt atoms.

In the present study, the preparation parameters: relatively low final ammonolysis temperature (700 ◦C) and high heating rate (10 ◦C/min) appear to promote the formation of the Co2Mo3N compound. Herein, a notable variation of the phase composition of the catalysts was observed. The precursors of the samples are virtually identical, and the change in Co2Mo3N concentration is attributed to the only variable parameter, the potassium concentration. The analogous effect of chromium salt addition was previously reported [20]. The content of the Co2Mo3N phase decreases with increasing concentration of chromium in the material. In the case of the potassium admixture, this dependence is more complex. Starting from small concentrations of potassium, the weight fraction of Co2Mo3N phase grows, with the maximum at about 0.8 wt.% of potassium. The excessive potassium content leads to a decrease in Co2Mo3N concentration. The mechanism that explains the observed influence of potassium on the phase composition of the studied samples remains unknown.

The catalytic activity of COMON catalysts in the process of ammonia synthesis is high. The highest reaction rate constant observed for the catalyst containing 1.3 wt.% of potassium was 1.5 gNH<sup>3</sup> ·g −1 ·h −1 . It is approximately twice as high as the reaction rate constant observed under identical process conditions for industrial iron catalysts (0.6 ÷ 0.7 gNH<sup>3</sup> ·g −1 ·h −1 ) [34]. The optimal potassium concentration corresponds well to the studies we previously reported on COMON catalysts at 400 ◦C [18,27]. A comparable amount of potassium promoter was also claimed as optimal in the report by Kojima and Aika [15]. In their study, the catalyst containing approximately 1.2 wt.% of potassium (that is, 0.05 mol K per mol Mo) was the most active one. It must be stated that the latter study was carried out under lower pressures (between 0.1 MPa and 3.1 MPa).

The detrimental influence of excessive potassium admixture confirms earlier reports [15,18]. Especially beyond 2.9 wt.% of potassium, the catalytic activity of COMON catalysts was very low. An excess of alkali metal was supposed to prevent the proper development of the catalyst surface [15]. The surface area of cobalt molybdenum nitride mixtures was prominently affected by the potassium content. It decreased by about 30% compared to the potassium-free COMON catalyst. It also varied considerably with potassium concentration. The optimal potassium content in COMON catalysts needed for the development of the highest specific surface area is ambiguous. The values of this parameter observed at 0.8 wt.% and 1.3 wt.% are relatively close to each other, and the optimal potassium concentration supposedly lies between them.

The influence of the potassium content on the activity of the COMON catalysts is analogous to that observed for potassium-promoted iron catalysts, both fused [35] and supported [36]. Low and very high potassium concentrations in these catalysts also result in a relatively low catalytic activity. We suppose that this is a general property resulting from the presence of potassium atoms on the surface of the catalysts. Alkali metals are assumed to affect the electronic properties of the surface active sites. However, the possibility that potassium modifies the structure of catalysts should not be overlooked [37,38].

The loss of activity after the thermostability test correlates well with the decrease of the surface area, which was observed for all catalysts, apart from the unpromoted one. Because the only variable between the studied materials was the potassium concentration, this phenomenon is associated with the difference in the structure of the catalysts, which apparently is the change of Co2Mo3N and Co3Mo3N concentrations.

#### **4. Materials and Methods**

#### *4.1. Precursor Synthesis*

Catalyst precursors were obtained using the method described in the previous work [18]. Briefly, water solutions of cobalt nitrate, Co(NO3)2·6H2O, and ammonium molybdate, (NH4)6Mo7O24·4H2O, were stirred and heated to about 90 ◦C. These solutions were mixed, while the pH of the resulting solution was controlled by addition of a 25% aqueous ammonia solution (NH3·H2O) to remain at pH = 5.5. A purple-blue precipitate was isolated by vacuum filtration, rinsed three times with distilled water and once with ethanol, then dried overnight at 150 ◦C. Potassium-promoted samples were obtained by impregnation of the precipitate in aqueous solutions of potassium nitrate, KNO3, in a vacuum evaporator at 60 ◦C. The concentration of potassium ions in solution was chosen as such to obtain the potassium concentration in the final, nitrided form of catalysts in the range 0.2 – 3.5% by weight.

#### *4.2. Nitriding of Precursor*

The active form of the catalysts was synthesized via the reduction process of the oxidized precursor under the flow of pure ammonia in a horizontal steel reactor placed inside an electric oven and then following the procedure described elsewhere [31,39]. Approximately 6 g of the oxidized precursor powder was placed in a ceramic boat. After flushing the reactor with pure ammonia, the precursor was heated under flowing ammonia gas (NH<sup>3</sup> flow—250 sccm, heating rate—10 ◦C/min., maximum temperature—700 ◦C). The sample was kept under ammonia flow at 700 ◦C for 6 h and then cooled to room temperature. The resulting fine-crystalline substrate was pyrophoric, and therefore each sample was left overnight in the flow of oxygen/nitrogen mixture (1:100) for passivation. The materials were then removed from the reactor and pressed into pellets, which were subsequently crushed and sieved. The 1.0–1.2 mm grain fraction was selected and used for the activity experiments.

#### *4.3. Material Characterization*

The phase composition of the materials was analyzed by powder X-ray diffraction (XRD). The Philips X'pert PRO MPD diffractometer was used in Bragg–Brentano geometry, with a Cu radiation source. To avoid fluorescence effects, a graphite monochromator was used. Phase identification was performed with the use of the ICDD PDF-4+ database [40]. A full-pattern fit based on the Rietveld method, using the formalism described by Hill and Howard [41], was applied to calculate the weight fractions of the crystallographic phases identified in the material. A semi-automatic Rietveld refinement procedure included in the HighScore Plus software [42] by PANalytical B.V. was used. All the data required for initialization of the Rietveld refinement were retrieved from the ICDD database. During the Rietveld refinement, the scale factor, unit cell parameters, full width at half maximum, and peak shape parameters of the phases have been refined. The pseudo-Voigt function was used.

The specific surface area of the nitrided samples was measured by the volumetric method using the N<sup>2</sup> adsorption-desorption isotherm at 77 K. The Quantachrome Quadrasorb SI-Kr/MP apparatus was used. Before measurements, the samples were degassed in vacuo for 6 h at 400 ◦C. The specific surface area was calculated using the Brunauer– Emmett–Teller (BET) equation. It was performed using commercial QuadraWin software.

### *4.4. Catalytic Activity Tests*

Catalytic activity tests were performed in the apparatus described in detail elsewhere [43]. The equipment consists of a 6-channel high-pressure steel reactor with a gas purification stage and enables the synthesis of ammonia under the pressure up to 10 MPa at a temperature reaching 650 ◦C. The 1 g samples of nitrided catalysts were placed inside the reactor in separate channels. Cobalt molybdenum nitride without alkali admixture was used as a reference sample. The samples were activated under a flowing reactant mixture (N<sup>2</sup> + 3H2, 330 sccm, 0.1 MPa) according to the following temperature program: 2 h at 350 ◦C, 3 h at 400 ◦C, 14 h at 450 ◦C and 24 h at 500 ◦C. This procedure is intended to remove the superficial oxide layer formed on the surface of catalysts during the passivation stage. The ammonia synthesis process was carried out with parameters as follows: pressure—10 MPa, gas reactants flow—330 sccm, temperature—450 ◦C. The ammonia concentration was measured in the outlet gas stream using a Siemens ULTRAMAT 6 NDIR (non-dispersive infrared absorbance) gas analyzer. The reaction rate constants of the ammonia synthesis reaction was calculated for each catalyst utilizing the modified Tiemkin–Pyzhev equation described elsewhere [44].

After the first activity test, all catalysts were heated in the reactor under a flowing ammonia–hydrogen mixture at 650 ◦C for 12 h. This procedure was intended to simulate the long-run action of the catalyst and is further referred to as thermostability test. Subsequently, the temperature was lowered to 450 ◦C and the activity test was repeated under identical conditions as described above.

#### **5. Conclusions**

Catalysts based on the mixture of Co3Mo3N and Co2Mo3N phases are highly active in the process of ammonia synthesis. Admixture of potassium compounds promotes the catalytic activity. Additionally, the phase composition of the catalysts is affected by the potassium content. After the thermostability test, the potassium-free catalyst remains virtually unchanged. Potassium-promoted catalysts lose catalytic activity as a result of the decrease of their surface area. The characteristic shape of the relation between potassium concentration and all of the measured parameters was observed for fresh catalysts as well as after the thermostability test. The maximum of the surface area and activity was observed for the catalysts with the greatest concentration of the Co2Mo3N phase, around 0.8–1.3 wt.% of potassium.

**Author Contributions:** Conceptualization; methodology; investigation; writing—original draft preparation; writing—review and editing; visualization; project administration; funding acquisition, P.A. Investigation; writing—original draft preparation; writing—review and editing; visualization, W.C. Conceptualization: methodology; investigation; writing—original draft preparation; writing—review and editing; visualization; supervision, D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** The scientific work was financed by the Polish National Centre for Research and Development, grant "Lider", project No. LIDER/10/0039/L-10/18/NCBR/2019 (Paweł Adamski); and the National Science Centre, Poland, grant "Preludium Bis", project No. 2019/35/O/ST5/02500 (Wojciech Czerwonko).

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

