Synthesis, Characterization and Catalytic Evaluation of Cobalt and Niobium Oxide Solids Modified with Alkaline Earth Metals

Cobalt and niobium oxide solids modified with alkaline earth metals (Mg, Ca, and Sr) were synthesized and characterized to compare the second element effect on the catalyst structural and chemical properties. The materials were prepared by co-precipitation of cobalt nitrate and niobium oxalate solutions using ammonium hydroxide as precipitating agent. The samples were characterized by XRD, SEM, Ra-man, BET, TPR, ammonia TPD and cyclohexane dehydrogenation reaction (CHD). Results showed that Mg, Ca or Sr addition changed cobalt niobate crystalline structure, leading to alkaline earth metal nio-bates with lower crystallinity. It was observed that Mg, Ca or Sr increased the solids acidity and changed cobalt niobate reduction behavior so that the lowest amount of metallic cobalt was obtained when Mg was incorporated. Modifiers also decreased cobalt niobate dehydrogenation activity. This effect was higher with Mg or Sr addition.


INTRODUCTION
The use of catalysts is an efficient way to produce chemicals to meet the growing demand for chemical industry. In this context, the use of catalyst for hydrogen production is highlighted as a function of the hydrogen potential as fuel cells energy source. In order to do that, the primary industrial process used is hydrocarbon reforming [1]. The active catalysts used in this process need to have dehydrogenating property, that is, the ability of breaking the hydrogen bond in compounds. The most used catalyst in industry for hydrocarbon reforming is nickel/alumina, which is active and selective at low costs.
Cobalt is another transition metal used as catalyst active phase in industrial reactions. It presents activity to decomposition, dehydrogenation, hydrogenation, hydrogenolysis and hydrodesulfurization reactions. It is also used in catalysts for cyclohexane oxidation [2], methane decomposition [3,4] , Fischer Tropsch synthesis [5,6], methane or ethanol reforming for hydrogen production [3,4,7,8,9], including fuel cell studies [10], although it shows lower selectivity to hydrogen compared to nickel catalyst [11], pollutant degradation [12] and others. Cobalt catalyst activity and selectivity depends considerably on the support type and preparation method [7,13,14,15]. For example, cobalt based supported catalysts were evaluated in dry methane reforming and showed high dependence on the support nature (MgO, CaO, or SiO 2 ) [16]. Nowadays, many studies are based on cobalt supported over different oxides, besides traditional alumina [7,8,14] .
Niobium is another metal that has received a lot of prominence in recent years. It has peculiar physical and chemical characteristics that make different systems useful, such as solid electrolytic capacitors, photochromic devices, transparent conductive oxides or memristors, catalysists [17]. Niobium oxide became a widely studied catalyst support, which presents essential catalytic properties such as stability, among others [18]. Besides, it can act as a catalyst in various reactions such as hydrogenation, dehydrogenation, dehydration, esterification and oxidation [5]. Brazil has the largest niobium reserve in the world (above 90%), therefore it is an extra motivation for the development of new technologies that use niobium derivates [16,19]. Regarding the use of niobium oxide in methane, ethanol or other carbon compounds reforming, niobium oxide may have a disadvantage related to its high surface acidity, which could favor coke formation, carbon-rich species deposition on the catalyst surface and active sites blockage [20,21]. On the other hand, although much studied, there is still contradictory information regarding niobium oxide properties. This occurs especially when studying its phases, due to the tendency to present crystal polymorphs and different phases, consequently running different catalysts [17].
The addition of modifiers to the support could be a solution for such a problem. Due to their basic character, modifiers such as alkali and alkaline earth metal oxides may decrease the support acidity and consequently the amount of carbonaceous deposits on its surface [20,22]. For Ni/Al 2 O 3 catalysts, a decrease in acidity due to alkaline earth metal oxides addition was directly related to coke deposition decrease on the surface [23]. However, other catalytic properties may change with alkaline metal addition, such as metallic dispersion, active phase species reduction degree on the surface and its interaction with the support, which also affects their activity and selectivity [1,10,24]. Studies show that the catalyst behavior can be significantly altered depending on the nature of the alkaline earth metal used as a dopant [25]. In addition, alkaline earth niobate metals formed from the materials mentioned above may be used in other fields, due to their dielectric properties [17,26]. MgNb 2 O 6 , for example, is widely used as a precursor to synthesize other columbites while CoNb 2 O 6 has interesting magnetic properties and CaNb 2 O 6 has useful optical properties [27,28]. Other authors claim that ANb 2 O 6 (A= Ba 2+ ,Sr 2+ ,Mg 2+ ,Sn 2+ , Zn 2+ , among others) columbite structure exhibits excellent performance in tests of magnetic, dielectric, thermoelectric and piezoelectric properties [29].
Thus, in this work, solids based on cobalt and niobium oxide modified with alkaline earth metals (Mg, Ca, Sr) were synthesized and characterized to evaluate some properties and catalytic behavior, mainly due to dopants.

Samples Preparation
Solids were prepared by co-precipitation method, using cobalt nitrate and niobium oxalate solutions as metallic precursors, and ammonium hydroxide as precipitating agent. Solutions were mixed slowly under constant stirring at pH=9.0. After precipitation, the reaction mixture was heated to 90 ºC under continuous stirring until solvent excess evaporation. The material was dried at 120 °C for 24 h and calcined at 800 °C for 2 h, producing the CoNb sample (the solid was synthesized to obtain Co:Nb at 2:7 atomic ratio). The same procedure/methodology was followed to prepare CoMgNb, CoCaNb, and CoSrNb samples using magnesium nitrate, calcium nitrate and strontium nitrate as metallic precursors, respectively. The solids were synthesized to obtain Co:M:Nb at 2:1:7 atomic ratio, in which M = Mg, Ca, Sr.
X-ray diffraction (XRD) analyses were performed using Shimadzu XRD600 equipment with nickel filter. The sample was submitted to CuKα radiation, which was generated at 30 kV and 20 mA, and the scan ranged from 10º to 80º. The phases were identified using the Inorganic Crystal Structure Database (ICSD).
Scanning Electron Microscopy (SEM) images were obtained through Tescan Vega-3 LMU equipment at 15 kV and 7000x magnification. Energy Dispersive X-ray Spectroscopy (EDS) analyses were performed in an X-act Oxford Instruments apparatus coupled to the electron microscope.
Samples specific area values were determined by physical nitrogen adsorption isotherms at -196 °C in a Micromeritics TriStar II 3020 equipment. Before the experiment, the samples were pre-treated at 10 μmHg for 30 min and then heated at 200 °C for 60 min under nitrogen flow for surface cleaning. After pretreatment, the dead volume and the saturation pressure, P°, were determined using 99.999% pure helium gas. Before analyses, samples were heated at 200 °C (2 h) under 10 -6 Torr vacuum.
Raman spectra were obtained to determine the species crystalline structure in the solids. Analyses were performed at room temperature throug a Jasco NRS-5100 spectrometer, equipped with a TCD cooled at -69 °C, under the following conditions: 532.13 nm wavelength, 20x objective, 100-2000 nm spectral scanning, 6.0 mW laser power and 14 cm -1 spectrum resolution. The sample was exposed to laser ten times to reduce background noise, 10s each time (accumulations number and exposure time, respectively). The analysis was performed repeatedly to observe different parts of the solids.
Temperature programmed reduction (TPR) analyses were performed in a Micromeritics AutoChem 2920 equipment.
Before the experiments, 0.050 to 0.150 g samples were pretreated by heating at 160 o C for 30 min under nitrogen atmosphere in order to remove volatile compounds. During analyses, samples were heated from 30 to 1000 °C at 10 o C.min -1 under 5% H 2 /N 2 flow (30 mL.min -1 ).
Samples whole acidity was determined by ammonia temperature programmed desorption (NH 3 -TPD). The analyses were performed in a Micromeritics TPD/TPR 2900 equipment, using 0.05 g sample, which was initially heated at 150 °C for 1h under He flow to remove impurities. After pretreatment step, the sample was cooled to room temperature and heated again under H 2 flow (35 mL.min -1 ) at 500 °C for 1 h. Ammonia adsorption was carried out by this gas pulses injection through a calibrate loop kept at 75 °C, until sample saturation. The system was cooled to room temperature and then heated to 775 °C, under He flow, to obtain the ammonia temperature programmed desorption profiles. The acidity was calculated from ammonia adsorption peaks areas, taking the last peak for calibration.
Metallic activity was evaluated by cyclohexane dehydrogenation reaction (CHD). The tests were carried out in a continuous flow micro-reactor, at atmospheric pressure, 400 °C and H 2 /cyclohexane (molar) = 26. Before each run, the samples (0.05 g) were reduced in situ at 500 °C for 2 h, under hydrogen flow. Cyclohexane and reaction products were analyzed by gas chromatography using a Chromosorb column and TCD as detector.
Magnesium ions incorporation into cobalt niobate crystalline structure is facilitated because Mg 2+ has low ionic radius value (0.72 Å). On the other hand, this did not happen with Ca 2+ and Sr 2+ ions, because they have higher ion radius values, 1.00 and 1.18 Å, respectively [32]. Regarding the variety of phases found, other authors also observed cobalt niobate different phases formation [8,13,31,33] because, just like niobium, cobalt can assume different oxidation states and shows varied phases, depending on the support and the preparation method [7,8,15,29]. Nb 2 O 5 phase is considered to be very stable and, therefore, desirable for catalytic systems operating at high temperatures [34]. However, this phase was not detected in CoCaNb or CoSrNb DRX samples profiles, which may indicate that there was no phase segregation and / or small and dispersed particles formation. Comparing DRX profiles, it can be seen that lower intensity peaks were obtained when alkaline earth metals were incorporated into cobalt niobate. It indicates that dopant addition impaired the solids crystallinity, because the niobate structure has been distorted due to the cation radius [34]. Among the formed niobate phases, MgNb 2 O 6 is the most studied [26,27,30]. It can be difficult to be formed in the single phase, and it may be accompanied by Mg 4 Nb 2 O 9 phase [27,31], which was not observed in this work. SEM images (Fig. 2) suggest the all solids had particles of irregular shapes, highly agglomerated and with different sizes on the surface. This behavior is common for niobates obtained by traditional synthesis routes, as noted by other authors [26,30]. It was also possible to observe the presence of lighter color smaller particles agglomerates over darker color larger particles. It is in accordance with XRD results, which indicate different metal oxides formation in each of the studied solids. Probably, the darker color particles are cobalt niobates, which are present in a higher amount. Materials surface atomic distribution, evaluated from elemental analysis by EDS, showed that Co/Nb and Co/M atomic ratios found for CoNb and CoMgNb samples, respectively, were close to nominal values. However, these atomic ratio values were higher for CoCaNb and lower for CoSrNb, as observed from Table 1. This can indicate that there is a higher concentration of cobalt oxides on CoCaNb sample surface compared to the others.
Raman spectra can be seen in Fig. 3. Scan was performed in order to collect data from different points on the sample surface. It can be observed that for the sample without dopant and the sample with magnesium, two different spectra profiles were obtained, while only one was obtained with the other samples (CoCaNb and CoSrNb). In all samples, the absorption bands found at 870, 530, 702, 283, and 253 cm -1 can be attributed to both niobium and cobalt species [28,33]. The different lines found are in agreement with XRD results which indicated that the cobalt and niobium species can be distributed in different structures: CoNb 2 O 6 (orthorhombic), Co 3 O 4 (spinel), Co 4 Nb 2 O 9 (trigonal) [28]. The highest peak at 870 cm -1 is attributed to niobium species, probably cobalt niobate, with orthorhombic structure [8]. A shift from this peak for higher frequency regions regarding CoMgNb sample may indicate a network distortion as a function of magnesium insert. On the other hand, the shoulder at 780 cm -1 for CoNb and CoMgNb samples spectra can be attributed to niobium oxide phase (Nb 2 O 5 ), as identified by XRD.  The bands shown at 732 and 687 cm -1 , observed almost exclusively for CoSrNb sample, can be attributed to strontium presence. Peaks from 700 to 190 cm -1 can be related to cobalt oxide, Co 3 O 4 [8,34]. Finally, the spectral region from 145 to 186 cm -1 can be associated to niobium oxide, NbO 2 [17]. Thus, Raman spectra also corroborate XRD results, indicating strong interaction between dopant and support, possibly due to magnesium ions incorporation to cobalt niobate structure. CoSrNb sample stands out for presenting more peaks with more intensity than the other ones. This may indicate, in addition to more phases (not detected by XRD), more covalent character metal-O bonds [33], once strontium decreases the solid hygroscopic character, compared to other alkaline earth metals. TPR profiles indicate that cobalt niobate reduction behavior (Fig. 4) was modified with Mg, Ca or Sr addition. In all cases, cobalt species reduction occurred between 250 and 450 °C, in agreement with other works, while niobium species reduction occurred above 700 °C [31]. Also, according to literature [16,27,2], Co 3 O 4 reduction, which is a phase present in all DRX profiles, occurs in two stages. However, in this work this phenomenon was only observed for CoNb and CoMgNb samples, in which the shoulder at 336 °C can be attributed to Co 3+ reduction to Co 2+ while the higher peak, between 349 and 360 °C, is due to CoO reduction to metallic cobalt [2]. The reduction peak at 952 o C is due to the reduction of Nb 2 O 5 phase [35] which is mainly present in the CoNb and CoMgNb samples, as seen by DRX, while the shoulders near to 778 °C could be attributed to the reduction of cobalt niobate phases (CoNb 2 O 6 and Co 4 Nb 2 O 9 ). The cobalt oxide reduction peak at 349 °C as well as the niobium oxide reduction zone at 778 and 952 °C for CoNb sample was slightly shifted to higher temperatures for samples profiles containing modifiers. It shows that the presence of modifiers increases the cobalt oxide interaction with niobium oxide. Other authors [21] also observed this interaction. On the other hand, it was noted that Ca or Sr addition increased cobalt and niobium oxides reducibility, once a greater area under the reduction peaks was observed, according to CoCaNb profiles and CoSrNb samples. The opposite occurred with CoMgNb sample, which could mean that, in this case, a cobalt and magnesium niobate species which is hard to be reduced was probably formed, as seen by XRD.

CoCaNb CoSrNb
Solids showed low specific area values from 2.0 to 6.0 m²g -1 (Table I), which are typical of niobium oxide [27] and cobalt oxide [4] obtained at high temperatures. Commonly, alkaline earth metals do not contribute to area increase [24], although modifiers addition slightly increased the specific area and decreased cobalt niobate mean pore diameter (CoNb). This effect was more pronounced in samples containing strontium (CoSrNb) which, as can be seen in all characterization results, was the dopant which changed CoNbOx properties more effectively.
Modifiers also increased cobalt niobate whole acidity, as shown in Table 1, but the strontium-containing material (CoSrNb) showed higher total acidity than the other ones. According to NH 3 -TPD curves (Fig. 5), only two defined desorption regions were observed in the sample profile, corresponding to moderate acidic sites (peak with maximum at 350 °C) and strong acidic sites (peak with maximum at 680 °C). For the other samples, it is probable that the adsorbed ammonia quantity was so low that, due to the equipment detection limit, it was not possible to identify the desorption peaks in TPD profiles. In all cases, whole acidity values were lower than the ones obtained by other authors [32] for pure niobium oxide (403 mmol NH 3 .g -1 ). It could show that cobalt interaction decreased niobium oxide acidity. By adding the dopants, Co-Nb interaction weakened and acidic sites concentration in cobalt niobate increased. This effect may be more pronounced for CoSrNb sample because its Sr/Nb ratio is the highest one when compared to M/Nb ratio for the other samples (Table 1).  The effect of modifiers (Mg, Ca, and Sr) on Co-Nb based catalysts metallic function activity was evaluated by cyclohexane dehydrogenation reaction and the results are shown in Table 2. Cyclohexane dehydrogenation is frequently used as a model reaction to evaluate metallic sites activity because it is a non-demanding reaction [36], that is, it does not require specific atomic arrangement to take place. Therefore, the conversion values are a function of the active centers number exposed on the catalyst surface. According to TPR profiles, magnesium and niobium oxides were not reduced in the reduction conditions used in this work. Therefore, they remained mainly as oxides species, without activity for dehydrogenation. Thus, the cyclohexane conversion values were due to cobalt atoms, which are active for dehydrogenation [5,13]. Table 2 shows that cyclohexane conversion with CoNb decreased when the modifier was added. It means that Mg, Ca, or Sr could be electronically interacting with Co, making it rich in electrons. Consequently, its hydrogen chemisorption capacity could have been reduced (electronic effect) or cobalt atoms could have been covered by alkaline earth metal niobates, as seen by the SEM images, and so be less available to dehydrogenate cyclohexane (geometric effect). The inhibitor effect regarding Co activity was lower with Ca addition, probably due to higher cobalt amount which was reduced for CoCaNb catalysts, as it can be seen by the TPR profile first peak underneath area. However, a higher amount of cobalt species was present on CoCaNb sample surface when compared to the other samples, as showed by EDS data. Some authors attribute the cobalt metallic activity decrease to SMSI effect [31,32]. Niobium oxide partial reduction, confirmed by TPR results, can contribute to cover the cobalt particle metal surface. On the other hand, the conversion values were similar to the samples containing Mg or Sr (CoMgNb and CoSrNb, respectively). In this case, using Mg or Sr as dopants in cobalt niobate based catalysts can be advantageous, because they decrease the highly hydrogenated coke precursors production during carbon compounds reforming.

CONCLUSIONS
In this work, cobalt and niobium oxide solids modified with alkaline earth metals (Mg, Ca, and Sr) were synthesized by co--precipitation, characterized and evaluated for cyclohexane dehydrogenation reaction. The results showed that the second element addition: (i) changed cobalt niobate crystalline structure, producing alkaline earth metal niobates with lower crystallinity as well as segregated cobalt oxides; (ii) slightly increased cobalt niobate specific area; (iii) contrary to expectations, increased cobalt niobate acidity, and this effect was more pronounced for solids containing strontium; (iv) modified cobalt niobate reduction behavior so that the amount of reduced cobalt species was higher with Ca-containing material and lower with Mg-containing solid, probably due to a stable magnesium and cobalt niobate phase formation, hard to be reduced.
The solids showed slight activity for cyclohexane dehydrogenation, which decreased with the presence of donates. However, the solid with calcium showed the best result. These effects were more pronounced with Mg or Sr addition. It indicates that their use in cobalt niobate based catalysts could be suitable to decrease cobalt dehydrogenation activity, it may generate less highly dehydrogenated coke precursors on the catalyst surface during carbon compounds reforming. However, one must be careful with the Sr effect on increasing cobalt niobate acidity, as this may favor cracking reactions, which contributes to coke formation.