On the optimal Cs/Co ratio responsible for the N₂O decomposition activity of the foam supported cobalt oxide catalysts

2.1 Catalyst preparation

Several sets of catalysts differing in the composition of the active phase and deposition procedure were prepared in order to study effect of:

1) various organic impregnation agents,

2) glycerol concentration in impregnation solution,

3) cesium loading for organic free synthesis in bulk and supported catalysts,

4) cesium loading for glycerol assisted deposition in supported catalysts.

The effects 1) and 2) were elucidated on catalyst non-modified by cesium.

The first series of catalysts was prepared using different organic impregnation agents: acetic acid (Ace), citric acid (Cit), glycerol (Glr), glycine (Gly), and urea (Ure). Solutions were prepared by dissolving Co(NO)₃·6H₂O and an organic agent in water and diluting the solution to the desired concentration (Table S1). Compared to the rest of the series, a lower solution concentration was used to accommodate the preparation conditions for different solubility of each organic agent. The molar ratio of the organic agent to Co²⁺ ion in the solutions was always 1:1. A catalyst prepared by conventional (organic-free) impregnation ( $c.i.$ ) was prepared as well. The catalysts are labeled as the active phase low-content series L using a three-letter abbreviation of the organic agent used, e.g. L(Ace).

The second series of catalysts was prepared using glycerol as an impregnation agent with a 2:1 and 1:1 glycerol to Co²⁺ ion molar ratio. Solutions were prepared by mixing an appropriate amount of glycerol with water to volume and subsequently dissolving Co(NO₃)₂·6H₂O at a higher concentration of Co²⁺ compared to the L series (Table S1). The catalysts are labeled as the high-content series H using the numerator part of the glycerol to Co molar ratio, e.g. H(Glr2). A catalyst prepared by conventional (organic-free) impregnation (c.i.) was also prepared for comparison.

The third series was prepared by conventional impregnation with the molar ratio of Cs⁺ to Co²⁺ ions in the solution 1:150, 2:150, 3:150, 4:150, 5:150, 6:150 nominally resulting in the wt.% of Cs in the active phase being 1, 2, 3, 4, 5, 6, respectively. Solutions were prepared by dissolving weights of Co(NO₃)₂·6H₂O and CsNO₃ in a volume of water (Table S1). The catalysts are labeled as c.i. using Cs and nominal wt.% of Cs in the active phase, e.g. c.i.(Cs2). To see the effect of the amount of cesium for the supported and also unsupported catalyst, the bulk active phase in a powder form was prepared from the parent nitrate solutions by drying at 105 °C and subsequent calcination for 4 h at 500 °C. The samples were labelled as bulk and Cs appended by nominal wt.% of Cs in active phase, e.g. bulk(Cs2).

The fourth series consists of Cs-doped catalysts prepared by glycerol-assisted impregnation with glycerol to Co²⁺ ion molar ratio in the solution 1:1 and Cs⁺ to Co²⁺ ion molar ratio equal to those used for the third series i.e. 1:150, 2:150, 3:150, 4:150, 5:150, 6:150. The solutions were prepared by mixing appropriate amount of glycerol with water to volume and subsequently dissolving weights of Co(NO₃)₂·6H₂O and CsNO₃ (Table S1). The catalysts are labeled as Glr series using Cs and nominal wt.% of Cs in active phase, e.g. Glr(Cs2).

For all catalysts commercially available ceramic foam (LANIK, Vukopor A; manufacturer declared composition: 85 wt% Al₂O₃, 14 wt% SiO₂, 1 wt% MgO, 10 ppi porosity) the support in the form of cylinder (48 mm diameter, 20 mm height) was first cleaned with ethanol and dried in air at 105 °C to constant weight. The foam was then immersed in an impregnation solution heated to 60 °C for 20 min. After the foam was removed from the solution, the excess solution was left to escape freely. The foam was then dried in air at 105 °C for 4 h and calcined in air at 500 °C for 2 h with a heating rate of 4 °C min⁻¹.

2.2 Catalyst characterization

Atomic absorption spectroscopy (AAS) was used to determine the chemical composition of the active phase. X-ray diffraction analysis (XRD) was used to characterize the catalyst crystallographic properties and phase composition. N₂ physisorption (S_BET) was used to characterize the catalyst surface area. Scanning electron microscopy (SEM) was used to investigate the catalyst morphology. Temperature programmed reduction by H₂ (H₂-TPR) was used to characterize the catalyst reducibility. For N₂ physisorption, H₂-TPR and XRD analyses, the samples were prepared by crushing the catalysts and grinding them in Pulverisette 6 (Fritsch) planetary mill using a zirconia grinding mill and balls and sieving to obtain particles of size below 165 μm. Information on ceramic foam support can be found in (Pacultová et al., 2020).

The chemical composition was determined by AAS using Analytik Jena ContrAA 700 spectrometer after dissolving the samples in aqua regia by heating them to 200 °C in a microwave reactor Ethos UP (Milestone, IT).

Nitrogen physisorption at − 196 °C was performed on 3Flex automated volumetric apparatus (Micromeritics Instruments, USA) after degassing the material (fraction < 0.25 mm) at 150 °C for more than 24 h under vacuum below 1 torr. Specific surface area was evaluated by the BET method.

Scanning electron microscope Quanta FEG 450 (FEI) with EDS microprobe analysis APOLLO X (EDAX) was used for the characterization of the catalyst morphology. The images were taken using secondary electron and backscattered electron detectors at 15 kV. Microprobe analysis was performed with the EDAX detector and processed with the EDAX software.

Temperature-programmed reduction by hydrogen (TPR-H₂) was carried out on AutoChem II 2920 equipment (Micromeritics, USA). Prior to the TPR experiments, the samples (fraction < 0.2 mm) were outgassed in the flow of pure helium (50 ml min⁻¹) at 500 °C for 60 min. Then, the TPR runs were performed in the flow of 10 mol. % H₂/Ar (50 ml min⁻¹) in the temperature range 20 – 995 °C or 20–450 °C (for Cs modified samples) with the heating rate of 20 °C min⁻¹ and isothermal step at 450 °C for the next 40 min. Water vapor formed during TPR was captured in a cold trap.

SmartLab X-ray powder diffractometer (Rigaku, Japan) equipped with D/teX Ultra 250 position sensitive detector and Co tube (CoKα, λ₁ = 0.178892 nm, λ₂ = 0.179278 nm) was used for phase composition and microstructural determination of prepared catalysts. Recorded patterns were evaluated with appropriate software and compared to ICDD database (PDF-2). Coherent domain size of spinels l_D was calculated from the broadening of the diffraction peak (3 1 1) using Scherer equation l_D = 0.9∙λ / β∙cos θ, where λ is wavelength of the used radiation, β is diffraction width and θ is diffraction position (in 2θ scale).

Particle size analysis and morphology of the cesium-doped spinel nanocrystals adhered to the support were studied using a FEI Tecnai Osiris transmission electron microscope equipped with an X-FEG Schottky field emitter operating at the voltage of 200 kV. Information about the spatial distribution of the elements in the samples was obtained by EDX analysis, using a windowless detector (Super-X EDX) and the Bruker Esprit software. Prior to the observations, the samples were placed on a copper grid covered with a holey carbon film. For detailed shape analysis of spinel nanocrystals, scanning transmission electron microscopy (STEM) measurements were carried out. Z-contrast imaging was performed using a high-angle annular dark-field detector (HAADF), and the camera length was kept in the range 330–550 mm to maximize the signal intensity. Cobalt spinel particle size analysis was performed using DigitalMicrograph (Gatan) software (Digital Micrograph, 2007). The nanocrystal shape retrieving was based on gradient analysis of the recorded HAADF STEM images. The procedure is based on the Z-contrast analysis using calibrated HAADF STEM images and involves quantification of the relation between the HAADF STEM image intensity (the grey value) and the thickness of the specimen, followed by image gradient analysis to retrieve the diagnostic features of the shape (the diagnostic edge pattern exhibited by the nanocrystals). The analysis was performed employing the Canny algorithm (Canny, 1986) implemented in FeatureJ (FeatureJ, 2003) software as part of the ImageJ package for image processing (Rasband, 1997). The determined edge pattern provided the local orientation of the observed nanocrystals and was used for assigning the (x; y) coordinates of the identified vertices in the observed two-dimensional images. Completed by default values of the missing z coordinates, they were used as input for the convex hull algorithm to obtain the tentative polyhedral shape of the nanocrystal, considering constraints imposed by the spinel crystal symmetry. A more detail description of this method can be found elsewhere (Gryboś et al., 2021). The imaging conditions were adjusted in such a way that the image contrast originates mainly from the changes in the thickness of the examined nanocrystals. By combining the information on the spinel structure with the information on the retrieved edge pattern and the variation of the sample thickness, it was possible to determine the shape of the examined spinel nanocrystals using the reversed Wulff construction (Wulff, 1901).

2.3 Catalyst testing

The catalyst in the form of cylinder (48 mm in diameter, 20 mm in height) was loaded into a fixed-bed reactor (50 mm inner diameter) filled with inert ceramic rings. Before testing, the catalyst was heated to 450 °C at a rate of 6 °C min⁻¹ and kept at 450 °C for 1 h under N₂ flow. Reaction steady state was evaluated for temperatures in the range of 300–450 °C with temperature step of 30 °C. The N₂O concentration of the outlet mixture was continuously measured using the GMS810 (Sick) infrared N₂O analyzer. The composition of the inlet mixture was set to 0.1 mol.% N₂O in N₂ and determined at the beginning and at the end of each catalytic test. Tests were carried out under atmospheric pressure. For the low spinel content catalysts, $\mathrm{G}\mathrm{H}\mathrm{S}\mathrm{V}$  = 1080 (m³ h⁻¹)_gas(m³)_cat.bed was used; with sample sizes in the range 33.3–33.6 ml, corresponding to the volume flow rates of 599–605 ml min⁻¹. For the high spinel content catalysts, $\mathrm{G}\mathrm{H}\mathrm{S}\mathrm{V}$  = 1500 (m³ h⁻¹)_gas (m³)_cat.bed was used; with sample sizes 33.3–33.7 ml corresponding to the volume flow rates of 833–843 ml min⁻¹. Selected samples were also tested in simulated nitric acid plant tail gas at three different inlet gas compositions (0.1 mol.% N₂O, 0–5 mol.% O₂, 0–2 mol. % H₂O, 0–0.02 mol.% NO, balance N₂) at $\mathrm{G}\mathrm{H}\mathrm{S}\mathrm{V}$  = 1500 (m³ h⁻¹)_gas(m³)_cat.bed. The effect of external diffusion was previously tested for catalysts deposited on ceramic foams with a pore density of 10 ppi; effect of external diffusion was confirmed to be insignificant at volume flow rates exceeding 585 ml min⁻¹.

Testing of the bulk Co₃O₄-Cs (particle size of 0.160–0.315 mm) was performed in the reactor with diameter of 5.5 mm at total flow rate of 100 ml min⁻¹ (21 °C, 101325 Pa) and GHSV = 60 L g⁻¹h⁻¹. The catalysts were activated at 450 °C for 1 h in an inert gas, then the N₂O catalytic decomposition at 450 °C was measured for 10 h. After this period, where stable performance was verified, the temperature dependence of conversion was launched with cooling rate of 5 °C min⁻¹ and the catalysts activity was measured for 2 h at each temperature (450–330 °C). The N₂O conversion was calculated from the data obtained at steady state.

3.1 Effect of organic impregnation agents

The preparation method is based on a fast-redox reaction between a fuel and an oxidant in the presence of metal cations. Organic impregnation agents act as fuel during catalyst preparation, while metal precursors themselves are oxidants. The role of organic agent is to interact with metal cations by chelating and creating a network to maintain the metal cations homogeneously fixed in their positions during combustion and to establish a good connection between the fuel and the oxidant moieties. During combustion, the fuel is oxidized by the oxidants, while metal cations take oxygen from the oxidants and convert into the corresponding oxides (Deganello and Tyagi, 2018). Combustion conditions generally depend on the chemical nature of the reactive solution formed; therefore, on the type of fuel and the ratio of fuel to oxidizer (Varma et al., 2016).

In our case, we decided to test different fuel types, specifically acetic acid and citric acid representing fuels with carboxylic groups, glycerol as a fuel containing the hydroxyl groups, glycine combining the carboxylic and amino groups, and urea containing the amino functions. These categories should have different complexing abilities, and the intensity of the combustion reaction should differ. Furthermore, we used solutions with a constant molar ratio of cobalt and organic agent (1:1) resulting in different fuel rich (glycerol φ = 1.4, citric acid φ = 1.8) or lean (acetic acid φ = 0.8, glycine φ = 0.9, urea φ = 0.6) conditions, defined based on calculations of the propellant chemistry (Deganello and Tyagi, 2018). The glycerol effect was also tested for the sample with two different Co₃O₄ loadings and for two different molar ratios of cobalt to glycerol in solution being 1:1 and 1:2, respectively. The 1:2 ratio was chosen on the basis of the general recommendation for reasonable fuel amount to be equal to twice the total molar amount of metal cations according to (Deganello and Tyagi, 2018; Deganello, 2017).

3.2 Catalyst preparation and characterization

Table 1 shows the physicochemical properties of the prepared catalysts: weight content of the active phase ( $w_{{\mathrm{C}\mathrm{o}}_3{\mathrm{O}}_4}$ ), crystallite size ( $l_{\mathrm{D}}$ ) and lattice parameter ( $a$ ) of the active phase, BET specific surface of the catalyst (S_BET) and calculated specific surface of the active phase $S_{{\mathrm{C}\mathrm{o}}_3{\mathrm{O}}_4}$ . Properties of the ceramic foam substrate, as well as those of the catalyst prepared by conventional impregnation under the same conditions (but for the catalyst containing Cs-Co₃O₄ active phase) as those used for the H(c.i.) sample here, can be found elsewhere (Pacultová et al., 2020).

The weight content of the active phase was determined as the weight difference between the cleaned and dried foam prior to the impregnation and the catalyst after calcination. The values found for high-content catalysts agree with previous research (Pacultová et al., 2020), where conventional impregnation under the same conditions led to an active phase content of 5.4 wt%. As expected, the amount of active components deposited on the low-content catalysts (prepared from an impregnation solution with a lower concentration of cobalt nitrate concentration) is lower, being approximately 3.2 wt%.

XRD diffractograms of the prepared catalysts are shown in Fig. 1. In all the prepared catalysts, a phase corresponding to cobalt spinel oxide (ICDD PDF-2 00–042-1467) was detected along with corundum (Al₂O₃) (01–075-1862) and mullite (Al_4.54Si_1.46O_9.73) (01–079-1456) phases present in the foam support (for details see Table S2). The unmarked low-intensity lines (38,5°) may be attributed to the K $\mathrm{\beta }$ lines of the used X-ray source. Due to the low content of the Co₃O₄ active phase, the high intensity of support diffraction peaks complicates the evaluation of the structural parameters of the active phase.

Close

Fig. 1

XRD diffractograms of the Co₃O₄ catalyst prepared by organic-assisted impregnation. Ascription of the peaks to individual phases can be found in Supplementary materials (Table S2).

Export to PPT

The mean size of the crystallites was calculated according to the Scherrer equation from full width at half maximum (FWHM) of the spinel (3 1 1) line, corrected for the instrument resolution (LaB₆) (Table 1). In the low-content catalysts, conventional impregnation leads to crystallite size 60 nm, while organic-assisted impregnation regardless of the organic agent leads to approximately half values around 32 nm with the exception of L(Ace) sample having Lc only slightly lower (51 nm). In high spinel content catalysts, conventional impregnation results in crystallite size of 148 nm, glycerol-assisted impregnation at the 1:1 glycerol to cobalt molar ratio halves this value to 81 nm, and doubling of the glycerol content to 2:1 further reduces the crystallite size to 38 nm. Crystallite growth depends on the amount of Co₃O₄ available and the presence of organic agent. Thus, the evolution of gaseous products of the precursor – organic agent reaction disrupts the integrity of the newly formed spinel nanocrystals. The observed decrease in Lc values after using an organic impregnation agent is in accordance with the results of the literature (Zavyalova et al., 2007).

The value of the Co₃O₄ lattice parameter $a$ is around 0.808 (Table 1) for all catalysts, implying that the shape and size of the Co₃O₄ unit cell are unchanged by organic-assisted impregnation.

The specific surface of the foam substrate was previously reported to be 0.7 m²/g (Pacultová et al., 2020). The specific surface of the prepared catalysts was twice larger, 1.9–2.2 m²/g. The catalyst-specific surfaces were recalculated to find the specific surface of the active Co₃O₄ phase based on its weight fraction in the catalyst. Both the measured and calculated values are presented in Table 1.

For the low-content catalysts, recalculated Co₃O₄ surface areas were found in the range 35–45 m²/g in the order: $L(Cit)>L(Ure)\approx L(c.i.)>L(Ace)\approx L(Glr)\approx L(Gly)$ . Interestingly, the active phase of the catalyst prepared by conventional impregnation shows a similar specific surface, despite having the twice larger crystallite size as calculated from XRD. Although smaller crystallites may generally lead to larger specific surfaces, this may be offset by their agglomeration on the support surface. For the high-content catalysts, the conventionally impregnated catalyst exhibits an active phase specific surface area of 15 m²/g, while for both glycerol-impregnated catalysts, the value around 26 m²/g was found despite the different crystallite sizes. Higher active phase loading, therefore, leads to both larger crystallinity and smaller specific surface area of the active phase.

SEM micrographs of the prepared catalysts taken with a backscattered electron detector are shown in Figs. 2 and 3. Because of the higher atomic number of cobalt compared to the other elements present in the ceramic foam, it appears as bright features. On the low-content catalysts, different active phase morphologies are observed, depending on the organic agent used. It should be stressed that the relative weight of the active components is the same on all of these catalysts. On L(Cit) and L(glr), the active phase forms a continuous layer that covers most of the support surface and is disrupted by craters and cracks. It must be added that for the L(grl) catalyst, the places not covered by the active phase were also observed by SEM (S3). On L(Gly), a thin layer covering only parts of the surface is observed, whereas on the rest of the samples clusters of different sizes are scattered over the support surface without any formation of continuous layer similarly to L(c.i.) sample.

Close

Fig. 2

SEM micrographs of the prepared catalysts of L series.

Export to PPT

Close

Fig. 3

SEM micrographs of the prepared catalysts H(Grl) series.

Export to PPT

On the conventionally impregnated high spinel content catalyst (Fig. 3), scattered clusters of the active phase are present similarly to the low spinel content catalyst; in this case, the clusters are larger. On the high spinel content catalysts prepared by glycerol-assisted impregnation, part of the surface is covered by a layer disrupted with cracks; more compact and smoother layer is formed when a higher amount of glycerol was used for the synthesis. The lower values of specific surface observed in the high spinel content catalysts can be explained by a higher agglomeration of the active phase.

More detailed imaging of the catalyst particles can be observed by TEM (S4). The results of the particle size analysis for three selected samples are presented in Fig. 4. Different synthesis routes lead to different sizes of the spinel nanocrystals. For the nanocrystals synthesized without any organic agent (Fig. 4a) the average size is 67 nm with a standard deviation of 27 nm. For the Co₃O₄ nanocrystals synthesized with glycerol and acetic acid the average size is smaller and equals to 44 ± 13 nm and 48 ± 18 nm, respectively. Furthermore, for samples prepared with glycerol, a more uniform size of particles was achieved in accordance with (Gudyka et al., 2017).

Close

Fig. 4

Particle size analysis of the samples a) L(c.i.) b) L(grl) c) L(ace).

Export to PPT

The synthesis route also affects the morphology of the nanocrystals (Fig. 2). For the Co₃O₄ nanocrystals synthesized without addition of the auxiliary agent, the nanocrystals have a rather poorly developed subhedral morphology with many defects (cf. Fig. 5a1, a2). Nevertheless, within the polyhedral approximation the shape analysis revealed the following abundance of spinel facets (1 0 0) – 44 %, (1 1 0) – 31 % and (1 1 1) – 25 % (cf. Fig. 5 a3, a4). For the Co₃O₄ nanocrystals synthesized with glycerol, an euhedral habit is well developed and morphological defects were not observed (cf. Fig. 5 b1, b2, c1, c2). As can be seen in Fig. 5 b3-b4 and c3-c4_, spinel nanocrystals prepared with glycerol do not expose the (1 1 0) facet. The abundance of the planes (1 0 0) and (1 1 1) varies from (1 0 0) – 65 %, (1 1 1) – 35 % for the selected nanocrystal shown in Fig. 2b3 – b4, to (1 0 0) – 40 %, (1 1 1) – 60 % for that shown in Fig. 2 c3 – c4. The cobalt spinel nanocrystals synthesized with acetic acid (Fig. 5, d3, d4, e3, e4) expose preferentially the (1 1 1) plane (73 %) along with the (1 0 0) termination (27 %). The effect of the organic agent on the abundance of specific planes was previously published elsewhere (Gudyka et al., 2017; Deganello, 2017).

Close

Fig. 5

Shape analysis of the samples a) L(c.i.) b-c) L(grl) and d-e) L(ace). The analysis was based on HAADF STEM (row a₁ – e₁) and the following: contrast gradient analysis (row a₂ – e₂) the images, edge patterns superimposed with HAADF STEM images (row a₃ – e₃) and the retrieved Wulff shape of the nanocrystals (row a₄ – e₄).

Export to PPT

For all prepared catalysts, a total cobalt spinel reduction was observed during H₂-TPR with the H₂ consumption close to theoretical values, calculated according to the content of the active Co₃O₄ phase (Table 2). The temperatures of the reduction maxima and the total and theoretical H₂ consumption values are presented in Table 2. During Co₃O₄ reduction, two-step reduction: Co³⁺–Co²⁺, Co²⁺-Co⁰ is expected (Jacobs et al., 2007; Fratalocchi et al., 2019). The separation of the two steps observed in the reduction profiles is related to the effects of grain size and shape effects (Wójcik et al., 2020). In the reduction profiles presented in Fig. 6, the peak separation is manifested as a single peak with low-temperature shoulder. For the catalyst prepared by conventional impregnation (both L and H series), the low-temperature shoulder cannot be well distinguished similarly to the L(Ace) sample.

Close

Fig. 6

TPR-H₂ profiles of the prepared catalysts using different organic agents.

Export to PPT

For the low-content catalysts, the reduction onset is observed around 220 °C for all catalysts. The reduction maximum is around 360 °C for the L(c.i.), L(Ace), L(Glr), L(Gly) samples, while for the L(Cit), L(Ure) ones it is shifted to somewhat higher temperatures (371, 390 °C respectively). The total reduction is observed around 460 °C for the L(c.i.), L(Ace), L(Glr), L(Ure) catalysts, and around 500 °C for L(Cit), L(Gly). No reduction was observed at higher temperatures. However, in general, changes in the reducibilities of the samples are very gentle.

For the high spinel content catalysts, the reduction onset is also observed around 220 °C for all catalysts. The reduction maximum is at 393 °C for the conventionally impregnated catalyst; for the two glycerol-impregnated catalysts it is shifted towards lower temperatures. The total reduction is observed around 500 °C for the conventional-impregnated catalyst and around 550 °C for the two glycerol-impregnated ones. However, most species are reducible below 450° C. Overall, compared to low spinel content catalysts, while an increase in the amount of reducible species is observed, those are mostly reducible at higher temperatures. This agrees with the higher crystallinity, lower specific surfaces, and higher agglomeration observed by XRD, BET, and SEM, respectively.

3.3 N₂O catalytic decomposition

The prepared catalysts were tested in N₂O decomposition, and the obtained N₂O conversions are presented in Figs. 7 and 8. For the low-spinel content catalysts, the conversions obtained are close to each other, with the L(Grl) sample having the highest conversions, and the L(c.i.) one having the lowest. For the high-spinel content catalysts, both conversions and specific rate constants, evaluated according to (Pacultová et al., 2020), follow the same order: H(Glr-1) > H(Glr-2) > H(c.i.). Therefore, for low- and high-spinel content catalysts alike, the use of glycerol as the impregnation agent in the ratio 1: 1 leads to the most active catalyst.

Close

Fig. 7

N₂O conversion and kinetics constants of the prepared catalysts. Conditions: 0.1 mol. % N₂O in N₂, GHSV (20 °C, 101325 Pa) = 1080 m³h⁻¹m_cat⁻³.

Export to PPT

Close

Fig. 8

N₂O conversion and kinetics constants of the prepared catalysts. Conditions: 0.1 mol. % N₂O in N₂, GHSV (20 °C, 101325 Pa) = 1500 m³h⁻¹m_cat⁻³.

Export to PPT

3.4 Effect of cesium loading

The catalysts prepared by glycerol-assisted impregnation were the most active of the tested catalyst series, although the differences were not very high. Another reason to use glycerol is the fact that the main drawback of the fuel such as urea or glycine is that hazardous compounds such as NH₃ or NO_x are formed during synthesis, and moreover, according to the literature, carboxylic groups are able to complex most metal cations. However, complexes with alkaline earth cations are less favorite, while hydroxyl groups are more convenient for this purpose (Deganello and Tyagi, 2018). Therefore, we chose glycerol as the impregnation agent used for the cesium amount optimization process. For this purpose, two sets of catalysts with different amounts of cesium were prepared. The first was prepared by conventional impregnation, and the second was prepared by using glycerol as the impregnation agent, while maintaining the molar ratio of the cobalt to glycerol 1:1. The H(c.i.) sample prepared by conventional impregnation without cesium serves as the reference sample for the conventionally prepared series (c.i.) and the H(Glr) sample serves as the reference sample without cesium for the series prepared by glycerol-assisted impregnation. Moreover, the parent solution was used for the preparation of a powder catalyst to compare the effect of cesium on the activity in comparison to the bulk powdered samples as well.

4.1 N₂O catalytic decomposition

The prepared catalysts were tested for nitrous oxide decomposition to compare the effect of Cs content on catalytic activity in the bulk cobalt active phase and in the active phase deposited on the ceramic support by conventional and glycerol-assisted impregnation. The aim was to find an optimal Cs loading in the cobalt-based catalyst deposited on the ceramic foam for N₂O decomposition. Different cesium loadings strongly influenced the catalytic activity of all samples; however, the results were different for the each of the examined series. N₂O conversion under inert conditions for bulk and conventionally impregnated samples is shown in Figs. 9 and 10a, respectively. The optimal cesium range for the bulk samples is wider for higher temperatures, yet it is very narrow and lies between 0.5 and 1.4 wt%. This finding agrees with the results for cesium-doped spinels prepared from cobalt nitrate using different precipitating agents (Michalík et. al., 2017), where the optimum of 1 wt% was found. The optimal cesium loading shifts to 2–3 wt% after the deposition of the active phase on the foam support by conventional impregnation (Fig. 10a) and the window is markedly enlarged to 1–3 wt% for samples prepared by glycerol-assisted impregnation (Fig. 10b).

Close

Fig. 9

Dependence of N₂O conversion on the Cs content in Co₃O₄ over bulk active phase. Conditions: 1000 ppm N₂O in N₂, GHSV = 60 L g⁻¹h⁻¹. Cs wt. % are related to Co₃O₄ calculated from AAS results and known weight of deposited active phase.

Export to PPT

Close

Fig. 10

Dependence of N₂O conversion on the Cs content (AAS) in Co₃O₄ over supported catalysts prepared by a) conventional impregnation b) glycerol assisted impregnation. Conditions: 1000 ppm N₂O in N₂, GHSV (20 °C, 101325 Pa) = 1500 m³h⁻¹m_cat⁻³. Cs wt. % are related to Co₃O₄ calculated from AAS results and known weight of deposited active phase.

Export to PPT

Obvious differences in catalytic activity for the two supported samples can be observed in the streams simulating off gases from nitric acid plats, that is containing besides 1000 ppm of N₂O also 5 mol.% O₂, 2 mol.% H₂O and 200 ppm NO (Figs. 11 and 12). The positive influence of glycerol is even stronger, resulting not only in the wide range window of the optimal cesium loading compared to that of the catalyst prepared by conventional impregnation but also in conversion increase especially at low temperatures.

Close

Fig. 11

Dependence of N₂O conversion on the Cs content (AAS) in Co₃O₄ over supported catalysts prepared by a) conventional impregnation b) glycerol assisted impregnation in the presence of oxygen and water vapor. Conditions: 1000 ppm N₂O in N₂ + 5 mol.% O₂ and 2 mol.% H₂O, GHSV (20 °C, 101325 Pa) = 1500 m³h⁻¹m_cat⁻³. Cs wt. % are related to Co₃O₄ calculated from AAS results and known weight of deposited active phase.

Export to PPT

Regarding comparison the activity results with results from scientific literature, it must be noticed, that mostly unsupported catalysts were tested and presented and these cańt be applied at industrial level. The newest results obtained using structured foam catalysts was published by Ho (Ho et al., 2022), who also showed nice comparison for Co containing catalysts reported in the literature, showing his catalyst as very active. Anyway, the results shown in Fig. 12b) are the best among shaped catalysts designed for N₂O decomposition. Direct comparison of conversion is difficult due to different conditions used but keeping in mind the difference in GHSV applied rough comparison can be obtained. We have achieved 75 % conversion at 420 °C and almost 100 % at 450 °C at GHSV = 1500 h⁻¹ in the presence of O₂, NO and H₂O, while Ho achieved almost zero conversion at 420 °C and <30 % at 450 °C at GHSV = 3500 h⁻¹ in the presence of O₂ and NO. The results confirm that Co₃O₄-Cs deposited on an open cell ceramic foam as catalysts for N₂O decomposition are a promising alternative to conventional fixed-bed reactors.

Close

Fig. 12

Dependence of N₂O conversion supported catalysts on the Cs content (AAS) in Co₃O₄ prepared by a) conventional impregnation b) glycerol assisted impregnation in the presence of oxygen, water vapor and NO. Conditions: 1000 ppm N₂O in N₂ + 5 mol.% O₂, 2 mol.% H₂O and 200 ppm NO, GHSV (20 °C, 101325 Pa) = 1500 m³h⁻¹m_cat⁻³. Cs wt. % are related to Co₃O₄ calculated from AAS results and known weight of deposited active phase.

Export to PPT