MeO_x/Al₂O₃ and MeO_x/CeO₂ (Me = Fe, Co, Ni) catalysts for high temperature N₂O decomposition and NH₃ oxidation

Abstract

A series of MeO_x/Al₂O₃ and MeO_x/CeO₂ (Me = Fe, Co, Ni) compounds was prepared by incipient wetness impregnation and characterized by XRD, XPS, UV-vis DRS, differential dissolution phase analysis (DDPA). Their activity towards N₂O decomposition and ammonia oxidation at 750–900 °C was shown to depend on oxygen mobility in the sample as characterized by steady state isotopic transient kinetic analysis (¹⁸O SSITKA) and dispersion of MeO_x (Me = Fe) species. Obvious reverse “activity-FeO_x dispersion” dependence was related to the change of contribution from oxygen transfer from the support to active sites through the Me–ceria or not improbably, the Me–alumina interface. It was shown that the high efficiency of CeO₂ based samples in deN₂O can be additionally increased by CeO₂ supporting onto high surface area alumina.

---

1. Introduction

Secondary abatement processes of N₂O formed as an undesirable by-product during the catalytic oxidation of ammonia into nitric oxide implicates a catalyst set up in the ammonia burner operating at 800–900 °C directly after the Pt–Rh gauzes.¹ Bulk or supported mixed metal oxides were developed for catalytic N₂O decomposition and put into practice in nitric acid plants before 2000 (Co₂AlO₄/CeO₂, CuO/Al₂O₃, La_0.8Ce_0.2CoO₃ and Fe/ZrO₂).¹ Moreover, alternative catalytic formulations were evaluated such as yttrium-doped zirconia,² ceria–zirconia,³ A_1−x A^′_xB_1−yB′_yO₃ perovskites with A = La or Ca, B = Mn, Co or Fe, A^' = Sr, B' = Cu or Ni,^4–6 metal-substituted hexa-aluminates,⁷ Fe₂O₃–MeO_x (Me = Al, Zr, Ce, La, Cu, Cr) obtained by co-precipitation,^8,9 Co₃O₄ (ref. 10) and CoO_x–CeO₂ (ref. 11) obtained by calcinations of mixtures of individual oxides. The pilot plant reactor and real plant studies⁸ confirmed high activity and very good mechanical stability of the Fe₂O₃/Al₂O₃ catalysts prepared by co-precipitation as well as no decomposition of nitric oxide. Several aspects with respect to the reaction mechanisms, the structure–activity correlations, the role of various gas inhibitors as well as the strategies followed to adjust the local surface structure of the abovementioned noble metal-free metal oxides were analyzed in the review of Konsolakis.¹²

Ageing of Pt–Rh gauzes is accompanied by decrease of activity and yield of NO + NO₂ (NO_x), whereas N₂O production increases. Taking account of the tendency in the last decade to reduce platinum metal loading in the reactor (including the number of gauzes and wire thicknesses), at gauzes deactivation noticeable slip of ammonia becomes possible followed by its reaction with NO_x. Therefore, development of a bi-functional catalyst that is capable of efficient N₂O abatement and NH₃ oxidation to NO_x becomes urgent for further optimization and the reduction of harmful emissions from the process of HNO₃ production.

An idea for development of an oxide-based catalyst that is capable of converting slipping ammonia with high selectivity to NO_x was successfully realized using a dual bed system, in which a Fe₂O₃–Al₂O₃-based extruded monolith followed fewer numbers of platinum gauzes.¹³ Later studies showed the efficiency of these compositions in both slipping NH₃ oxidation to NO_x and N₂O abatement.¹⁴ At the same time, Al₂O₃-based honeycomb catalysts with the same geometry and FeO_x-based active components supported by impregnation were substantially more active in both reactions and retained high activity after the run in a dual-bed system for 3 months.¹⁴

Earlier, we have shown that catalytic activity of Sr substituted La manganites in high-temperature N₂O decomposition correlated with the coefficient of lattice oxygen self-diffusion.⁵ In the materials with high bulk oxygen mobility, the rates of oxygen transfer in the two-phase system “O_2gas–surface–catalyst bulk” characterized by ¹⁶O/¹⁸O exchange and N₂O decomposition additionally increased either by their decoration by structures characterized by a higher rate of heteroexchange (composites LaSrFeO₄ (surface)–La_0.4Sr_0.6FeO₃ (bulk)⁶) or by modification of near surface layers by hetero-cations (LaFeO₃–CeO₂ composites with Fe³⁺ ions inserted into a fluorite lattice¹⁵). In the former case,⁶ “activity – oxygen mobility” correlation was observed for reaction of NH₃ oxidation as well. Pérez-Ramírez et al.¹⁶ showed that the reaction of NH₃ oxidation to NO_x over oxide catalysts (Fe₂O₃, CeO₂, Cr₂O₃) follows a Mars–van Krevelen-type scheme with participation of lattice oxygen and reoxidation of the so-formed vacancies by both gas-phase O₂ and bulk lattice oxygen. Due to this, the efficiency of NH₃ oxidation to NO_x, especially in oxygen deficient conditions, depended on the rate of bulk oxygen diffusion to surface vacancies.

In accordance with these findings, the idea to combine high activity towards N₂O decomposition and NH₃ oxidation (at high selectivity to NO_x) in the same sample was checked. In addition to Fe/Al₂O₃(CeO₂) samples, Co- and Ni- supported on Al₂O₃ and CeO₂ were tried as well since a vast number of patent applications have claimed a relatively high NO selectivity (90–95%) using Co and Ni single metal oxides. Numerous attempts were made to commercialize Co₃O₄-based catalysts but substantially lower activity and reversible deactivation of Co-systems, due to reduction of Co₃O₄ by NH₃ to the inactive CoO under reaction conditions in the upper parts of the bed, excluded the replacement of Pt gauzes by oxide catalysts. Nevertheless, these systems could be tried as the second catalyst in the dual bed system, especially since the stabilizing effect of CeO₂ in Co₃O₄–CeO₂ mixed oxides with low Co content is well-known.¹¹ Since copper catalyzes ammonium nitrate decomposition, an extremely important safety problem may arise if copper leaches from the catalyst and accumulates in the fertilizer product.¹ This completely excludes any Cu-based system for use in the conditions of the ammonia burner, although Cu/CeO₂ revealed high efficiency in deN₂O at lower temperatures^17,18 and more probably can be located after the absorption column. In addition, it is well known that under these conditions, Cu-containing systems oxidize NH₃ to N₂ with high selectivity.

In the first stage, we showed that intrinsic catalytic activity of different Me/Al₂O₃ (Me = Fe, Co, Ni) samples obtained by impregnation towards N₂O decomposition and NH₃ oxidation was substantially lower compared to corresponding Me/CeO₂ samples with similar surface Me concentration and correlated with the rate of oxygen exchange in the samples. Variation of Fe content in the Fe/Al₂O₃(CeO₂) samples revealed noticeable reverse dependence of deN₂O and NH₃ oxidation activity due to the size effect. In accordance with this, we supported FeO_x onto specially developed Al₂O₃ ¹⁹ characterized by high specific surface area, and formation of highly dispersed FeO_x at increased Fe content in the sample finally resulted in N₂O and NH₃ conversion increase. For such samples, we focused on Fe oxide as the active component because in the case of Me = Co, Ni low active MeAl₂O₄ or NiO_x phases were formed on the alumina surface. Using of the effect of high oxygen mobility in CeO₂ was restricted by its low surface area at the temperature of reaction, independently on preparation method.^3,11,20 Therefore, an attempt was made to disperse CeO₂ using precipitation onto high surface area Al₂O₃.

2. Experimental

2.1. Sample preparation

CeO₂ and Al₂O₃ samples used as the support were obtained by calcination of the corresponding nitrates of chemical purity (99.5%) at 900 °C for 36 h. Al₂O₃-based supports were prepared by calcination of: 1) Al(NO₃)₃·9H₂O of chemical purity (99.5%) at 1000 °C or 1200 °C for 6 h (denoted below as Al₂O₃-1000 and Al₂O₃-1200, respectively), or 2) the product of centrifugal thermal activation of commercial-grade alumina hydrate (hydrargillite)¹⁹ at 1000 °C for 19 h (Al₂O₃-C). To obtain the Ce/Al₂O₃ support, the product of thermochemical activation of hydrargillite was first calcined at 500 °C (S_BET = 210 m² g⁻¹); then, CeO₂ was precipitated onto this product from 0.75 M Ce(NO₃)₃·6H₂O water solution using 1 M (NH₄)₂CO₃ as a precipitation agent, so that the Ce : Al ratio in the final product was 1 : 1 (mole). The resultant precipitate was washed until the pH of the filtrate was 7, dried at 60 °C overnight, followed by calcination at 900 °C for 5 h.

All Me/support samples (Me = Fe, Co, Ni) were prepared by incipient wetness impregnation of the corresponding support by water solution of Me nitrate with added citric acid in 10 wt% excess to the stoichiometric amount to form the corresponding citrates and ethyleneglycol, dried in air at 150 °C for 3 h and then calcined at 900 °C for 4 h. In the commonly-used abbreviation nMe/support, “n” corresponded to Me weight content in the sample in terms of the corresponding metal, although oxide compounds obviously formed. For CeO₂ and Al₂O₃-1200, the samples with Me concentration 6.7 × 10¹⁹ at m⁻² have been prepared, which corresponded to Me content of around 2.5–2.7 wt% (Al₂O₃-1200) and 0.86–0.91 wt% (CeO₂). For all supports, we also varied Fe content in the samples from 0.86 to 19.8 wt%. Bulk MeO_x were obtained by calcination of the corresponding chemical purity nitrates at 900 °C for 6 h.

2.2. Sample characterization

X-ray powder diffraction (XRD) patterns were recorded using a Bruker D8 diffractometer with Cu K_α monochromatic radiation. Each sample was scanned in the range of 2θ from 10° to 70° with a step size of 0.05° in 2θ. Surface composition of the samples was investigated by X-ray photoelectron spectroscopy (XPS) using a SPECS spectrometer with Al-K_α irradiation (hν = 1486.6 eV). The binding energy (BE) scale was preliminarily calibrated by the position of the peaks of Au 4f_7/2 (84.0 eV) and Cu 2p_3/2 (932.67 eV) core levels. UV-vis diffuse reflectance spectra (UV-vis DRS) were obtained using a UV-2501 PC Shimadzu spectrometer with an IRS-250A diffusion reflection attachment in the 11 000–54 000 cm⁻¹ range using BaSO4 as a reference. The edge energy (or band gap, E_g) for allowed transitions was determined by finding the intercept of the straight line for the low-energy rise of a plot of [F(R_∞) hν]²versus hν,²¹ where hν is the incident photon energy and F(R_∞) is the Kubelka–Munk function.

The method of differential dissolution phase analysis (DDPA) was used to reveal the composition of the phases (including poorly crystallized, but without information about the oxygen content therein) formed in the supported samples, including possible decoration of one phase by another.²² For this, about 10 mg of the sample was loaded in a microreactor and dissolved in the water based solvent with the composition changing from the lower towards higher acidity in the following order: 1) HCl (pH = 2), 2) (1 ÷ 3) M HCl, while the temperature increased continuously from 20 °C to 90 °C, 3) (3.6 M) HF and running the reactor with a flow rate of 3.6 ml min⁻¹. Change of the outlet mixture composition with time was analyzed by atomic emission spectroscopy (BAIRD spectrometer) using the following spectral lines (nm) of the elements: Fe–238.2, Al–308.2, Ce–413.8, Co–238.8 with 5% accuracy of measurements at sensitivity level 10⁻³ μg ml⁻¹.

Kinetics of oxygen exchange was characterized by steady state isotopic transient kinetic analysis (SSITKA). For SSITKA experiments, the sample (g = 0.025 g) was loaded into a reactor (quartz tube, i.d. = 3 mm) and heated to 800 °C in an 0.5% vol. ¹⁶O₂ + He flow and kept at this temperature for 30 minutes, whereupon the gas mixture was replaced stepwise by the same one containing ¹⁸O₂ and Ar (1 vol%) as an inert tracer. The gas flow rate for all mixtures amounted to 16.7 cm³ s⁻¹. All responses were analyzed as time variation of the ¹⁸O atomic fraction in the gas phase α_g(t) = (¹⁶O¹⁸O + 2¹⁸O₂)/(2ΣOⁱO^j), where ΣOⁱO^j = ¹⁶O₂ + ¹⁶O¹⁸O + ¹⁸O₂.²³

The catalytic activity for samples with particles of 250–500 μm in size was measured in a fixed-bed U-shaped reactor (3 mm i.d. quartz tube) at ambient pressure in the temperature range 750–900 °C and contact time ~10⁻³ s. For NH₃ oxidation, a mixture of 1% NH₃ in air or 1% NH₃ + 2% O₂ in N₂ was fed to the reactor charged by 0.015 or 0.043 g of the sample (see details in the text) with a flow rate 25 l h⁻¹. Outlet mixture composition was measured by IR spectroscopy. For N₂O decomposition, a gas mixture consisting of 0.15 vol% N₂O (+3% O₂ + 3% H₂O) in He flowed through the reactor with a flow rate of 60 l h⁻¹. For samples characterized by high specific surface area (Al₂O₃-C, Ce/Al₂O₃ based), we failed to reach reasonably low values of N₂O conversion at 800 °C. Therefore, the data obtained at the same sample loading (0.038 g) and flow rate (60 l h⁻¹) were considered to analyze the effect of Fe concentration. Outlet mixture composition was analyzed by an on-line gas chromatograph with Porapack T (i.d. = 3 mm, l = 3 m) and NaX (i.d. = 3 mm, l = 2 m) columns.

3. Results and discussion

3.1. Phase composition

3.1.1. XRD analysis. Al₂O₃-1200 (Fig. 1a), Al₂O₃-1000 and CeO₂ used as the supports were single phase samples (Table 1), with values of coherent scattering area d_XRD calculated by the Scherrer equation much less than the particle size estimated from the BET surface area (d_BET). For example, for Al₂O₃-1200 d_XRD = 360 Å, d_BET = 3800 Å, while for CeO₂-680 Å and 6400 Å. TEM images of the CeO₂ sample revealed a mixture of agglomerates composed of concreted crystallites of either 100 nm or 10 nm in size.¹⁵ The same can be reasonably supposed for Al₂O₃-1200 and Al₂O₃-1000 samples as well to explain the difference between d_BET and d_XRD values. Al₂O₃-C represented a mixture of α-Al₂O₃ (Fig. 1B, Table 1) with more dispersed θ-Al₂O₃, γ-Al₂O₃ and traces of κ-Al₂O₃. In the Ce/Al₂O₃ sample, CeO₂ phase with the same lattice parameters but expected more dispersed, than in CeO₂ sample, species, θ-Al₂O₃ and traces of γ-Al₂O₃ were detected (Fig. 1C, Table 1). Dispersion of the θ-Al₂O₃ species was about the same as in Al₂O₃-C, as follows from similar widening of peaks at 2θ of ~67° (γ-Al₂O₃ + θ-Al₂O₃), but cannot be calculated more precisely because of overlapping of its more intense peak at 2θ ~32.9° with that of CeO₂.

Fig. 1 XRD patterns of: (A) Al₂O₃-1200 (1) and (2.5–2.7) Me/Al₂O3-1200 (Me = Co(4), Fe (2), Ni(3)) samples, (B) Fe/Al₂O₃-C samples with different Fe content, and (C) Ce/Al₂O₃ (1) and 9.9Fe/Ce/Al₂O₃ (2) samples.

Table 1 BET surface area, phase composition, structural parameters, and crystallite size estimated from XRD (d_XRD) of Al₂O₃, CeO₂ and MeO_x/CeO₂(Al₂O₃) samples

Sample	S BET, m^2 g^−1	Phase composition	d XRD, Å
Al2O3-1200	4	α-Al2O3 (a = b = 4.756 Å, c = 12.986 Å)	370
Al2O3-1000	7	α-Al2O3 (a = b = 4.760 Å, c = 12.997 Å)	300
Al2O3-C	71	α-Al2O3 (a = b = 4.771 Å, c = 13.020 Å)	480
		θ-Al2O3, κ-Al2O3, γ-Al2O3	100 (γ-Al2O3), 140 (θ-Al2O3)
2.5Fe/Al2O3-1200	4.0	α-Fe2O3 (a = b = 5.035(0) Å, c = 13.741 Å)	260
		α-Al2O3 (a = b = 4.761 Å, c = 12.997 Å)	500
2.7Co/Al2O3-1200	4.0	CoAl2O4 or Co3O4 (a = b = c = 8.104 Å)	260
		α-Al2O3 (a = b = 4.761 Å, c = 12.997 Å)	520
2.7Ni/Al2O3-1200	4.0	NiO (a = b = c = 4.178 Å)	240
		NiAl2O4	≤120
		α-Al2O3 (a = b = 4.760 Å, c = 12.996 Å)	580
2.5Fe/Al2O3-C	60	α-Al2O3 (a = b = 4.778 Å, c = 13.039 Å)	510
		θ-Al2O3, κ-Al2O3, γ-Al2O3	Highly dispersed
6.6Fe–Al2O3-C	46	α-Al2O3: 31% – a = b = 4.770 Å, c = 13.030 Å	510
		69% - a = b = 4.795 Å, c = 13.057 Å	470
		θ-Al2O3, κ-Al2O3, γ-Al2O3	Highly dispersed
13.2Fe–Al2O3-C	33	α-Al2O3: 18% – a = b = 4.760 Å, c = 13.000 Å, 82% – a = b = 4.795 Å, c = 13.057 Å,	510
			520
		Traces of θ-Al2O3, κ-Al2O3, γ-Al2O3	Highly dispersed
		Fe2O3 (a = b = 5.032 Å, c = 13.702 Å)	250
CeO2	1.4	CeO2 (a = b = c = 5.412 Å)	680
0.91Co/CeO2	1.4	CeO2 (a = b = c = 5.412 Å)	630
		Co3O4 (a = b = c = 8.084 Å)	460
0.86Fe/CeO2	1.4	CeO2 (a = b = c = 5.411 Å)	550
0.91Ni/CeO2	1.4	CeO2 (a = b = c = 5.411 Å)	600
2.5Fe/CeO2	1.4	CeO2 (a = b = c = 5.412 Å)	650
		traces α-Fe2O3	—
CeO2/Al2O3	63	CeO2 (a = b = c = 5.411 Å)	400
		θ-Al2O3 and traces of γ-Al2O3	Highly dispersed
9.9Fe/Ce/Al2O3	35	CeO2 (a = b = c = 5.411 Å)	460
		θ-Al2O3 and traces of γ-Al2O3	Highly dispersed
		α-Fe2O3 (a = b = 5.035 Å, c = 13.741 Å)	320

---

Although Co₃O₄ and Co–Al–O spinels are characterized by the same symmetry group, increased lattice parameters of Co compounds in 2.7Co/Al₂O₃-1200 sample (Fig. 1A) (a = b = c = 8.104 (1) Å) compared with those in the 0.91Co/CeO₂ one or in bulk Co₃O₄ (a = b = c = 8.084 (0) Å) favor formation of CoAl₂O₄ in the former,²⁴ probably in the mixture with Co₃O₄. In the 2.7Ni/Al₂O₃-1200 and 2.5Fe/Al₂O₃-1200 samples, additional NiO with admixture of highly dispersed Ni–Al–O compounds of spinel structure and α-Fe₂O₃, respectively, were detected. The lattice parameters of NiO and Fe₂O₃ in the supported samples were exactly the same as for bulk MeO_x (not presented in Table 1 for clarity). However, slight increase of α-Al₂O₃ lattice parameters in all Me/Al₂O₃ samples points to the probability of non-substantial Meⁿ⁺ penetration into the alumina lattice. Taking account of the slightly larger radii of Me³⁺ ions compared with that of Al³⁺ (69–74 pm and 67.5 pm, respectively), such an assumption looks quite reasonable.

FeO_x supported on Al₂O₃-C resulted in continuous decrease of the peaks corresponding to θ-, κ-, γ-Al₂O₃ with an increase of Fe content from 2.5 to 13.2 wt% (Fig. 1b). Instead of these, a second α-Al₂O₃ phase with an increased lattice parameter (a = b = 4.795 Å, c = 13.057 Å) becomes evident and increases in intensity (Table 1), which more obviously points to efficient incorporation of Fe³⁺ ions into the lattices of low temperature aluminas thus promoting their transition to α-Al₂O₃ at a lower temperature.²⁵ This fact agrees well with decrease of S_BET value from 70 m² g⁻¹ for Al₂O₃-C to 22 m² g⁻¹ after supporting 19.8% of Fe (pattern of the last sample not presented for brevity).

In the patterns of (0.86–0.91)Me/CeO₂ samples, traces of crystalline MeO_x were observed only for the case of Me = Co, and some α-Fe₂O₃ was detected for substantially higher Fe content (sample 2.5Fe/CeO₂). Nevertheless, unlike alumina-based samples, neither sintering of CeO₂ nor change of CeO₂ lattice parameters were noted after Me supporting (Table 1). Considering that the radius of Fe³⁺ ions (0.049–0.078 pm, depending on coordination and spin state) is smaller than that of Ce⁴⁺ ions in the typical cubic fluorite lattice (0.097–0.101 nm), the expected lattice shrinkage took place at substantial Fe content in the mixed Fe_x–Ce_1−x oxides.²⁶ However, at x < 0.05, even slight unit cell expansion was observed and related by authors to partial Ce⁴⁺ reduction to the larger (1.23 Å) Ce³⁺ ion²⁷ during calcination in the presence of Fe³⁺. Therefore, penetration of supporting Me into the CeO₂ lattice cannot be excluded, especially taking account of concreted nanocrystallites present in the initial support.

In the 9.9Fe/Ce/Al₂O₃ sample, α-Fe₂O₃ was detected with lattice parameters and d_XRD value similar to that in 13.2Fe/Al₂O₃-C and 2.5Fe/Al₂O₃-1200 samples. Although Fe supporting onto Ce/Al₂O₃ resulted in noticeable drop of the surface area, like in Al₂O₃-C based samples, it did not facilitate formation of α-Al₂O₃ (Tables 1 and 2, Fig. 1C). For similar CeO₂/Al₂O₃ samples, it was shown that it is Ce in the lower oxidation state that stabilizes alumina toward the formation of low surface area phases up to 1100 °C or 1200 °C under oxidizing and reducing conditions, respectively.²⁸ Since CeAlO₃ crystallites at high temperature and under reducing conditions were observed by XRD, stabilization of Ce³⁺ is more probably due to interaction with the OH groups of alumina. Absence of such alumina-bonded hydroxyl groups can prevent penetration of Fe³⁺ into the bulk alumina and the phase transition observed in the case of the Al₂O₃-C sample.

Table 2 Chemical composition of dissoluble part of Me/Al₂O₃-1200(CeO₂) samples (Me = Co, Fe, Ni)

Sample	Sample dissolved, weight%	Me, weight% of the sample in dissolved portion	Element	Distribution of elements dissolved during different steps but O, mole%
				Total	HCl (pH = 2)	0.33 M HCl	3.6 M HF
2.5Fe/Al2O3-1200	5	2.53	Al	38.1	—	—	38.1
			Fe	61.9	—	18.4	43.5
2.7Co/Al2O3-1200	2	0.43	Al	82.1	—	—	82.1
			Co	17.9	0.4	—	17.5
CeO2	4	—	Ce	100	—	—	100
0.91Co/CeO2	95	0.93	Ce	98.2	—	98.2	—
			Co	1.8	Traces	0.39	1.41
0.86Fe/CeO2	54	1.0	Ce	96.8		96.8	—
			Fe	3.2	0.60	1.86	0.74
0.91Ni/CeO2	11	0.81	Ce	91.3	—	—	91.3
			Ni	8.7	Traces	7.8	0.9

---

3.1.2. DDPA. DDPA of (2.5–2.7)Me/Al₂O₃-1200 and (0.86–0.91) Me/CeO₂ samples has been performed to reveal 1) possible modification of support lattice due to Me insertion, and 2) formation of Co–Al spinels. Al₂O₃-1200 supported samples were quite stable towards dissolution. Therefore, about 5 wt% and 2 wt% of the 2.5Fe/Al₂O₃-1200 and 2.7Co/Al₂O₃-1200 samples, respectively, dissolved during all steps (Table 2). Nevertheless, all Fe supported onto Al₂O₃-1200 passed into solution (Fig. 2A) as individual oxide/hydroxide (about 18.4 mole% of the probe dissolved in HCl), or together with some Al (~43.5 mole% of the probe dissolved in HF). Since some dissolution proceeded even after all Fe was removed from the sample, formation of Fe–Al solid solution is excluded, and the structure with FeO_x species located on the surface or on the intergrain boundaries of concreted alumina micro-crystallites and linked to the last, more probably, due to penetration of some Fe³⁺ ions to the disordered near subsurface layers can be considered. In the case of 2.7Co/Al₂O₃-1200, only 0.43 wt% of Co dissolved in all steps, an overwhelming majority, as the solid solution with Co_0.3Al₁ stoichiometry reasonably relates to the mixed Co–Al–O spinel like compounds located in the near surface layer of the sample. The nature of insoluble Co compounds still remains unclear.

Fig. 2 DDPA spectra of: (A) 2.5Fe(2.7Co)/Al₂O₃-1200, (B) CeO₂ and 0.91Co/CeO₂ samples.

CeO2 did not dissolve in HCl, and very slow dissolution (about 4% of the total sample weight in 7 minutes) took place after HF feeding (Fig. 2B, Table 2). Me/CeO₂ samples dissolved in substantially milder conditions, but the character and the rate of their dissolution depended on the nature of Me. Therefore, already 95 weight% of the 0.91Co/CeO₂ sample dissolved in these conditions. Increase of solubility can be related first of all to formation of Co–Ce–O solid solution with Co/Ce = 0.004 and including all soluble Ce (dissolved in (1 ÷ 3) M HCl). This obviously means that Co³⁺ ions are able to insert into more dispersed (~10 nm) concreted microcrystallites and modify the lattice of larger (~100 nm) particles increasing their solubility. The rest of Co dissolved slowly as an oxide in the most harsh conditions (3.6 M HF) and more probably, can be related to the well-crystallized Co₃O₄ detected by XRD (Table 1). Absence of any Co in the same flow during dissolution of the 2.7Co/Al₂O₃-1200 sample points to the formation of well-crystallized Co–Al spinel therein. The quantity of 0.86Fe/CeO₂ sample dissolved was somewhat lower (54 wt% of the total) and all Fe passed into solution as it was found for 2.5Fe/Al₂O₃-1200 sample (Fig. S1†). It was included into Fe oxide/hydroxide whose solubility in HCl (pH = 2) and at the substantially higher acidity (3.6 M HF) can be due to different degree of crystallinity or defect structures formed. The parallel dissolution of the remaining Fe and Ce in the flow of (1 ÷ 3) M HCl with the stoichiometry Fe/Ce = 0.02 points to formation of Fe–Ce–O solid solution, more probably, with participation of CeO₂ composites characterized by smaller sizes of crystallites. Unmodified ceria is stable even in 3.6 M HF and can be related to composites with largest crystallites. We cannot exclude that both unmodified CeO₂ and Ce–Fe–O solid solution are decorated by highly defective Fe oxide/hydroxide species dissolved in weak HCl. In the case of the 0.91Ni/CeO₂ sample (Fig. S1†), most of the supported Ni dissolved individually in (1 ÷ 3) M HCl, while the parallel Ni and Ce dissolution took place in the more rigid conditions (3÷6 M HF) and included only 10 weight% of the sample.

Therefore, in the 2.7Co/Al₂O₃-1200 sample, Al³⁺ ions partially replace Co³⁺ in the spinel structure, forming both highly disordered and well-crystallized Co–Al–O species, while incorporation of Fe³⁺ ions into the boundaries of the micro-composites results in the anchoring of Fe₂O₃ crystallites to the alumina surface. Me–ceria interaction results in formation of Me–Ce–O solid solutions in Me/CeO₂ samples; their content and degree of disorder in the sample change in the following order: 0.91Co/CeO₂ > 0.86Fe/CeO₂ > 0.91Ni/CeO₂ ≥ CeO₂.

3.2. Microstructure

UV-vis DR spectra of different Co- and Fe-based catalysts have been presented in Fig. 3a and b, respectively. CeO₂ was characterized by strong absorbance bands at 36 500 and 29 000 cm⁻¹ attributed to Ce⁴⁺ ← O²⁻ charge transfer and the interband transitions, respectively,²⁹ and a band gap energy (E_g) value of 3.35 eV, which is consistent with the reported value of 3.3 eV.^29–31 The spectra of 0.91Co/CeO₂ (Fig. 3a) and 0.86Fe/CeO₂ (Fig. 3b) samples are in fact a superposition of those for corresponding bulk oxides and CeO₂, excluding the red shift of absorption edge of CeO₂ (E_g values of 3.18 and 3.22 eV for 0.91Co/CeO₂ and 0.86Fe/CeO₂, respectively) and additional absorption band at ~20 000 cm⁻¹ (Co case) and 23 500 cm⁻¹ (Fe case). A minimal red shift of the absorption edge compared to that in the support was found for the 0.91Ni/CeO₂ sample (E_g value of 3.27 eV, spectrum not shown for brevity). We already observed these spectral characteristics for a Fe/CeO₂ sample prepared in the same way¹⁵ and explained their appearance by O₂ adsorption on the structural defect, more obviously, clusters of oxygen vacancies arose after the entry of the dopant ions into the bulk of CeO₂. As proposed in,³⁰ a change in the value of the absorption edge indicates that either (1) Me ions in Me/CeO₂ samples substitute into the CeO₂ lattice; or (2) the isolated Me ions form a mixed Me–Ce oxide and strongly interact with CeO₂ leading to the electronic structure changes. In our case, the value of the red shift correlated well with content of Me–Ce mixed solution in the sample, as determined by DDPA.

Fig. 3 UV-vis DR spectra of: A – (1) 2.9Co/Al₂O₃-1200, (2) CoAl₂O₄ supplied by Aldrich, (3) 0.93Co/CeO₂, (4) CeO₂, (5) Co₃O₄; and B – (1) FeO_x, (2) 0.86Fe/CeO₂, (3) 2.7Fe/Al₂O₃-1200, (4) CeO₂.

In the spectrum of the 2.9Co/Al₂O₃-1200 sample, an intense triplet of bands at about 15 900, 17 000 and 18 200 cm⁻¹ with the shoulder at 13 800 cm⁻¹ was detected and related to the transitions of ⁴A₂ → ⁴T₁(⁴P) of the tetrahedral Co²⁺ ions in Co–Al–O spinel and cobalt oxide, respectively.³² The presence of the same triplet in the reference CoAl₂O₄ sample (Aldrich) supports formation of Co–Al–O spinel with admixture of Co₃O₄.

3.3. Surface composition

As follows from the X-ray photoelectron spectra of Ni 2p, Fe 2p, and Co 2p levels in Me/Al₂O₃-1200 and Me/CeO₂ samples presented in Fig. 4, the nature of Me surface compounds depends on the nature of the support. Usually, no difference between various Co compounds can be reliably drawn from binding energies of either Co 2p_1/2 or Co 2p_3/2 signals¹⁰ without accounting for the position of their satellites arising due to the shake-up process of the Co²⁺ compound in the high spin state. Indeed, BE values characterizing the Co 2p_3/2 peak position in CoO, CoOOH and Co₃O₄ are indistinguishable (~780.0 eV). Higher, but also very close, BE values have been reported for CoAl₂O₄ and Co(OH)₂ (781.9 and 781.2 eV, respectively).^33–38 However, for CoO, Co(OH)₂ or CoAl₂O₄, the satellite – main signal distances fall in the range 5–6.5 eV,^34,35,39,40 while the satellites in the spectrum of pure Co₃O₄ and CoOOH should be located approximately 9–10 eV away from the main signals.^34,36,39 In this regard, the BE value of the Co 2p_3/2 signal at 781.6 eV with satellite at 786.6 eV in the 2.7Co/Al₂O₃-1200 sample (Fig. 4a) excluded CoO, Co₃O₄ and CoOOH from our consideration, and either Co(OH)₂³³ or CoAl₂O₄^34,37,38 could form on the surface. According to DDPA data, Co(OH)₂ could be responsible for about 0.4% of Co compounds (Table 2, corresponds to Co dissolved in HCl at pH = 2); therefore, most of the surface Co included in the Co–Al spinel structures. To ensure the correctness of such assignment, the spectrum of CoAl₂O₄ supplied by Aldrich and additionally calcined at 900 °C for 5 h has been presented as well. In the 0.91Co/CeO₂ sample, the position of the main Co 2p_3/2 peak shifted to lower BE values (779.9 eV), while the shoulder at 781.9 eV was still present resulting in the ratio of their intensities of ~3 : 2. Definite assignment of the peak at 779.9 eV to individual CoO, Co₃O₄ or CoOOH was impossible since the presence of satellites at 786.7 and 790.8 eV agrees with formation of any of these compounds. In addition, in the XPS spectrum of the bulk Co₃O₄ sample characterized by the same lattice parameters as the 0.91Co/CeO₂ sample (Table 1), the intensity of the satellite peak was substantially lower. Hence, different oxide/hydroxide-like Co compounds, including those in Co–Ce–O solid solution, have been formed on the surface of the 0.91Co/CeO₂ sample. In particular, Co(OH)₂ in this case could result from hydration of highly dispersed species.

Fig. 4 (A) Co2p, (B) Ni2p, (C) Fe2p spectra of Me/Al₂O₃-1200(CeO₂) samples (Me = Co, Ni, Fe) with similar (~7 × 1019 at/m²) concentration of Me, and Fe 2p spectrum of 3.8Fe/Ce/Al₂O₃ sample (C).

The Ni 2p_3/2 peak at 856.3 eV in the 2.7Ni/Al₂O₃-1200 sample (Fig. 4b) can be due to both NiAl₂O₄ (BE ranging from 855.2 to 857.2 eV) and Ni(OH)₂ (855.6–856.6 eV),^41–45 the former being detected as well by XRD (Fig. 1, Table 1). NiO state, characterized by lower BE value (854.2 eV),⁴⁵ was detected in the 0.91Ni/CeO₂ sample together with Ni(OH)₂ (NiO : Ni(OH)₂) = 1 : 1) resulting in the widening of both the main Ni 2p peak and the energy shake-up satellite peak about 6 eV above the main one. As in the 0.91Co/CeO₂ sample, Ni(OH)₂ detection could indicate the presence of highly dispersed NiO species both in Al₂O₃ and CeO₂-based samples. In this case, their relative content in the samples is higher compared with the Co case.

While the BE value of the Fe 2p_3/2 spectra in the 0.86Fe/CeO₂ sample at 710.4 eV (Fig. 4c) indicates a preferential Fe³⁺ state in the oxide, its high energy shift (BE = 711.8 eV) detected in both the 2.5Fe/Al₂O₃-1200 specimen and all Al₂O₃-1000, Al₂O₃-1200 and Al₂O₃-C supported samples can be related to formation of Fe³⁺ oxyhydroxide.⁴⁶ In the 3.8Fe/Ce/Al₂O₃ sample, both states of Fe³⁺ have been detected also, but the fraction of Fe³⁺ state in the oxide is somewhat higher than in Fe/Al₂O₃-based samples. Therefore, FeO_x species anchor both to the Al₂O₃ and CeO₂ surfaces.

In the Ce 3d spectra of the 3.8Fe/Ce/Al₂O₃ sample, the normal complex form due to shake-down satellites from an O 1s to Ce 4f electron transfer observed for CeO₂ and 0.86Fe/CeO₂ was supplemented by the features marked as v' and u' due to anion defects and Ce³⁺ (Fig. S2†).⁴⁷ This means that besides smaller crystallized CeO₂ species, X-ray amorphous CeAlO₃ and probably isolated Ce³⁺ species can reasonably form after CeO₂ precipitation onto high surface area alumina,^48,49 thus stabilizing alumina toward the formation of low surface area phases, as proposed in reference.²⁷

Therefore, spinel-like surface structures including Al have been formed on the surface of the 2.7Co/Al₂O₃-1200 sample and probably their mixture with Ni(OH)₂ in the 2.7Ni/Al₂O₃-1200 sample. It is quite reasonable to suppose that above 700 °C, surface Fe oxyhydroxide formed after Fe supporting onto Al₂O₃ and all Ni- and Co-containing hydroxylated species convert to corresponding oxides, and thus similar types of surface compounds participate in the reaction in both Al₂O₃-1200, CeO₂ and CeO₂/Al₂O₃ based samples.

Table 3 lists the data on the surface concentration of Me in the different Al₂O₃-1200 and CeO₂ based samples. Similar values were measured in the (2.5–2.7)Me/Al₂O₃-1200 and 0.91Co(0.86Fe)/CeO₂ samples with the same as related per surface unit (6.7 × 10¹⁹ at m⁻²) content of Me therein that means similar MeO_x (Co–Al–O) species dispersion (D_Me) therein. Its relative value can be estimated as follows: D_Me = Me_s/C_Me, where Me_s is the surface atomic concentration of Me in the sample as measured by XPS (Table 3) [%] and C_Me is total content of Me in the samples [atoms/m²]. Noticeably higher content of Ni in the near surface layers of the CeO₂ based sample can be explained either by lower degree of Ni penetration into CeO₂, resulting in the formation of Ni–Ce–O solid solutions (Table 2) or by higher dispersion of NiO_x species, which agrees well with high content of hydroxylated species.

Table 3 Physicochemical properties of different Me/Al₂O₃, Me/CeO₂ and 3.8Fe/Ce/Al₂O₃ samples

Sample	S BET, m^2 g^−1	Mes, at% (by XPS)	Feexp, rel. units	DMe, rel. units
CeO2 based
0.86Fe/CeO2	1.3	1.0	1.3	1.5
0.91Co/CeO2	1.3	1.2	—	1.8
0.91Ni/CeO2	1.3	2.0	—	2.7
2.5Fe/CeO2	1.3	1.5	2.1	0.72
Al2O3-1200 and Al2O3-1000 based
2.5Fe/Al2O3-1200	4	1.1	4.4	1.6
2.7Co/Al2O3-1200	4	1.0	—	1.5
2.7Ni/Al2O3-1200	4	1.4	—	2.1
0.86Fe/Al2O3-1200	4	1.1	4.4	4.8
2.5Fe/Al2O3-1000	7	1.4	9.8	3.7
Al2O3-C based
2.5Fe/Al2O3-C	61	0.46	28.1	10.5
6.6Fe/Al2O3-C	46	1.3	59.8	8.5
13.2Fe/Al2O3-C	33	2.2	72.6	5.1
19.8Fe/Al2O3-C	22	1.6	35.2	1.7
CeO2/Al2O3 based
3.8Fe/CeO2/Al2O3	50	1.3	65.0	16.0

---

3.4. ¹⁸O SSITKA

Fig. 5 shows the time dependencies of exchanged oxygen for different samples as related to unit of the surface (N_O) calculated using the formulae
where α_g^input-isotope fraction in the inlet mixture (0.95), C_O2 − O₂ concentration (mol mol⁻¹), U–flow rate of the reaction mixture (mole/s), and N_A–Avogadro number. It is seen that for (2.5–2.7)Fe(Co)/Al₂O₃-1200 samples, N_O(t) dependencies (at higher rate of exchange on 2.5Fe/Al₂O₃-1200, as determined from the slope of N_o curves) reached the plateau after about 15 s after isotope admission corresponding to exchanged 4.9 × 10²⁰ and 3.2 × 10²⁰ O atoms per g for 2.5Fe/Al₂O₃-1200 and 2.7Co/Al₂O₃-1200 samples, respectively. Taking account of the quantity of supported Me (2.7 × 10²⁰ atoms per g), one can suppose that within the accuracy of measurement, solely the oxygen of Fe₂O₃ and Co oxidic compounds (Co₃O₄ + Co_3−xAl_xO₄) participated in the exchange. The integral quantity of exchanged oxygen in the 2.5Fe/Al₂O₃-C sample (7.9 × 10²⁰ atoms per g) substantially exceeds that in supported Fe oxide. Provided substantially more dispersed FeO_x species have been formed in this sample compared with those in 2.5Fe/Al₂O₃-1200 one (Table 3), it is more probably support oxygen that can transfer to O₂ molecule through developed Fe-alumina interface.

Fig. 5 Time dependence of the quantity of exchanged oxygen (N_O) in ¹⁸O SSITKA experiments for different Me/Al₂O₃ and Me/CeO₂ samples.

Substantially more profound and faster oxygen exchange took place on CeO₂ based samples, obviously involving the bulk of the support starting from the very low time after ¹⁸O₂ admission. The rate of exchange changed in the following order: 0.91Co/CeO₂ ≥ 0.86Fe/CeO₂ > 0.91Ni/CeO₂ ≈ CeO₂ and correlated with values of the red shift of the CeO₂ band gap energy (E_g) in turn depending on the content of Me–Ce mixed solution in the sample characterized by increased degree of disorder. Therefore, oxygen transfer from fluorite lattice to the surface by extended vacancies arising after insertion of Meⁿ⁺ ions into the fluorite lattice can be responsible for enforced rate of isotope exchange on Me/CeO₂ (Me = Co, Fe) samples compared with pure CeO₂.

3.5. Catalytic activity

3.5.1. N₂O decomposition.
Effect of the support nature. Among Al₂O₃-1200-based samples with similar surface concentration as supported Me (6.7 × 10¹⁹ at m⁻²), catalytic activity in N₂O decomposition at T = 750–900 °C can be ordered as follows: 2.5Fe/Al₂O₃-1200 > 2.7Co/Al₂O₃-1200 > 2.7Ni/Al₂O₃-1200 > Al₂O₃-1200 (Fig. 6). The (0.86–0.91)Me/CeO₂ samples with quite close, both calculated (6.7 × 10¹⁹ at m⁻²) and real (Table 3), surface concentration of Me as in (2.5–2.7)Me/Al₂O₃-1200 samples revealed substantially higher intrinsic activity. In the case of Co and Ni containing samples such increase of the activity could be partly related to the formation of principally different state of active component (oxides/hydroxides in CeO₂ instead of corresponding Co(Ni)–Al–O spinels located on the surface of Al₂O₃). Nevertheless, a similar effect observed for Fe based samples characterized by similar type of active species on both supports obviously points to the importance of increased oxygen mobility of CeO₂, which in turn was substantially more active than Al₂O₃. The general scheme of N₂O decomposition obeys the Langmuir–Hinshelwood mechanism:

N₂O + S → N₂O-S (1)

N₂O-S → N₂ + O − S (2)

2O-S ↔ O₂ + 2S (3)

where S is surface active site, N₂O-S is adsorbed or chemisorbed N₂O, and O-S is adsorbed or chemisorbed O, the rate of O₂ desorption by step 3) in the case of (0.86–0.91)Me/CeO₂ samples increases due to alternative fast supply of oxygen to reduced site S from the subsurface layers (O_ss) of CeO₂ characterized by increased bulk oxygen mobility through Me–CeO₂ interface containing many oxygen vacancies to obtain additional O-S site:

O-S ↔ O_ss + S (4)

A similar scheme considering two alternative pathways of regeneration of active sites was also proposed for Cu/CeO₂ samples tested in substantially lower temperature range (between 300 and 550 °C) and included recombination step (if two oxidized copper sites are close enough to each other) and the step of Cu²⁺−O-species reduction by Ce³⁺.⁵⁰

Fig. 6 Temperature dependence of the activity of different (2.5–2.7) Me/Al₂O₃-1200 (open symbols) and (0.86–0.91) Me/CeO₂ (solid symbols) samples (Me = Co, Fe, Ni) in the reaction of N₂O decomposition.

Effect of Fe content and FeO_x dispersion. With variation of Fe content in Al₂O₃-1200-based samples from 0.89 to 6.6 wt% (Fe was chosen as the most active among Me) we observed the inverse “activity (as related both to sample surface unit and to one Fe atom) – Fe content” dependence (Fig. 7A). The same Fe_s values for 0.86Fe/Al₂O₃-1200 and 2.5Fe/Al₂O₃-1200 samples (Table 3) indicate decrease of FeO_x species dispersion (D_Fe, Table 3) at higher Fe content. Finally, similar rates of N₂O decomposition were measured for bulk Fe₂O₃ and all Fe/Al₂O₃-1200 samples (Fig. S3†), although in the former case the Fe_s value should be obviously higher. Therefore, activity of surface Fe sites decreases substantially with enlargement of FeO_x species.

Fig. 7 Reaction rate (A) and N₂O conversion as measured at the same sample loading and flow rate (B) at 800 °C depending on Fe content in Al₂O₃-1200, Al₂O₃-1000, Al₂O₃-C, CeO₂ and Ce/Al₂O₃ based samples.

With variation of Fe content in CeO₂-based samples from 0.5 to 2.5 wt% (i.e. at similar interval of supported Fe to that in the case of Al₂O₃-1200-based samples as related per surface unit and thus similar dispersion of FeO_x species) their higher activity due to increased mobility of oxygen is more prominent at lower quantity of supported Fe. However, activity drop was substantially more prominent, although even higher Fe_s values were measured for the 2.5Fe/CeO₂ sample than for the 0.86Fe/CeO₂ one (Table 3). Supposing the rate of oxygen transfer from the bulk to active sites should depend on the length of MeO_x–CeO₂ interface, its decrease at enlargement of FeO_x species (Table 3), revealed as start of oxygen exchange at higher temperature (Fig. S4†) should make an additional negative contribution to the activity.

Effect of the support surface area. In accordance with observed effect of FeO_x dispersion on activity, use of the support characterized by higher specific surface area allows obtaining more dispersed FeO_x species at substantially higher total Fe content (Table 3). This opens up additional possibilities for increase of conversion as measured at the same sample loading. Indeed, at a similar quantity of supported Fe, N₂O conversion increased with S_BET value of Al₂O₃ (Fig. 7B).

Unfortunately, for Fe/Al₂O₃-C samples we failed to perform catalytic tests at reasonably low conversion values for correct calculation of the reaction rate values. This fact and substantial negative effect of the quantity of supported Fe on S_BET value (Tables 2 and 3) complicated analysis of parameters determining the activity. To overcome this problem, we calculated the concentration of exposable FeO_x species (Fe_exp) per weight unit by the following formula: Fe_exp = Fe_s*S_BET, where Fe_s is the surface concentration of Fe as measured by XPS (Table 3), and compared conversion values were measured for the same catalyst loading as dependent on both Fe content in the sample and Fe_exp. It turned out that N₂O conversion decreased with increase of Fe content from 2.5% to 19.8% (Fig. 7B) or FeO_x dispersion drop (Table 3), but this order cardinally differed from that for Fe_exp (13.2Fe/Al₂O₃-C > 6.6Fe/Al₂O₃-C > 19.8Fe/Al₂O₃-C > 2.5Fe/Al₂O₃-C). Therefore, similar to Fe/Al₂O₃-1200 samples, it is the size effect of FeO_x species that seems to be the dominant factor determining descending character of “conversion-Fe content” dependence. Similar reverse dependence was already observed earlier for lower (500–650 °C) temperatures.⁵¹ Taking account of SSITKA data obtained for 2.5Fe/Al₂O₃-1200 and 2.5Fe/Al₂O₃-C samples, higher activity of smaller FeO_x species can be due to growing ability of oxygen supply to reduced active sites S from the alumina regions adjacent to developed MeO_x-alumina interface (step 4). Oxygen (3 vol%) and H₂O (3 vol%) addition into the reaction mixture, in accordance with considered abovementioned scheme of the mechanism, decreased the observed N₂O conversion, but did not change the order of activity (Fig. S5†).

Effect of CeO₂ dispersion. Using the effect of high oxygen mobility of CeO₂ for the increase of the efficiency of N₂O decomposition (Fig. 7B) is restricted by its low surface area under the reaction conditions.^3,20 Therefore, in addition to traditionally used increase of concentration of active sites by supporting more active component, we tried to disperse CeO₂ by supporting onto high surface area Al₂O₃. Only Fe oxide was used in this case to avoid formation of low active Ni (Co)–Al spinels. The 3.8Fe/Ce/Al₂O₃ and 9.9Fe/Ce/Al₂O₃ samples revealed that the highest N₂O conversion (Fig. 7B) was due to the part of FeO_x that was anchored to the surface of higher dispersed ceria particles (Table 1, Fig. 4c). At the same time, the size effect, at least for the sample 3.8Fe/Ce/Al₂O₃, cannot be completely excluded (compare D_Fe values in the Table 3 for the samples 3.8Fe/Ce/Al₂O₃ and 6.6Fe/Al₂O₃-C with similar S_BET).

3.5.2. NH₃ oxidation.
Effect of the support nature. Among (2.5–2.7) Me/Al₂O₃-1200 samples, the 2.5Fe/Al₂O₃-1200 one was the most active (Fig. 8A) and characterized by the highest selectivity to NO + NO₂ (86–88%). As a result, the orders for the activity and the efficiency of NO_x formation evaluated as NO_xyield = Conversion × NO_xselectivity were similar to that for the activity of N₂O decomposition: 2.5Fe/Al₂O₃-1200 ≫ 2.7Co/Al₂O₃-1200 > 2.7Ni/Al₂O₃-1200–Al₂O₃-1200. In addition, the N₂O yield value calculated in a similar manner as for NO_xyield on the 2.5Fe/Al₂O₃-1200 sample in this temperature interval was lower than 2%, which is comparable to that for Pt–Rh gauzes,¹ while continuous increase of N₂O yield with temperature was noted for all other samples. At lower O₂ concentration in the inlet mixture (2.5%), the yield of NO_x decreased to 77–78%, mainly because of lower selectivity. For Co- and Ni-based samples and for pure Al₂O₃-1200, negative changes of conversion and NO_x selectivity were substantially more prominent.

Fig. 8 Temperature dependence of NH₃ conversion and N₂O yield for (2.5–2.7) Me/Al₂O₃-1200 (A) and (0.86–0.91) Me/CeO₂-900 (B) samples. Catalyst loading 0.015 g.

We failed to calculate the rate of ammonia oxidation in all temperature intervals due to non-differential reactor operating conditions. Hence, to compare intrinsic activity of samples characterized by different S_BET values, we tested catalytic properties of (0.86–0.91) Me/CeO₂ samples at catalyst loading of 0.043 g corresponding to the same surface area as for Al₂O₃-1200-based samples. In these conditions, at 750 °C, complete NH₃ conversion was measured on all (0.86–0.91)Me/CeO₂ samples, and 96% for CeO₂ that is obviously due to capability of CeO₂ itself to donate lattice oxygen for ammonia oxidation.¹⁶ Even at the same catalyst loading as for Al₂O₃-1200-based samples (0.015 g), i.e. at about a 3 times lower number of Me sites therein (Table 3), higher values of NH₃ conversion were observed for corresponding CeO₂ based samples (Fig. 8B). This increase, at least in the case of Fe-based samples, is exclusively due to the ability of the support to donate oxygen to FeO_x, since substantially lower conversion values were measured on CeO₂ and bulk FeO_x characterized by similar S_BET values. N₂O yield values were reasonably lower than on corresponding Al₂O₃-1200-based samples. In addition, at 800 °C, the order of NH₃ conversion and NO_x yield (not shown for brevity), 0.91Co/CeO₂ ≥ 0.86Fe/CeO₂ > 0.91Ni/CeO₂ ≈ CeO₂, was exactly the same as for the efficiency towards oxygen exchange therein (Fig. 4).

Strong decrease of both NH₃ oxidation⁵² and N₂O decomposition³² over Co oxide at high temperatures was shown to be due to Co₃O₄ reduction to CoO. We did not observe such deactivation of the 0.91Co/CeO₂ sample with temperature both at N₂O decomposition (Fig. 6), as in Co₃O₄–CeO₂ mixed oxides with low Co content,¹¹ and at ammonia oxidation (Fig. S7B†), although Co₃O₄ reduction under TPD conditions was observed (Fig. S6†). It obviously can be explained by higher resistance of the Co–Ce interface towards reduction due to some oxygen supply from CeO₂ by reaction (4). Nevertheless, at lower temperature, the 0.91Co/CeO₂ sample with Co state closer to Co₃O₄ was already substantially more active than 0.87Fe/CeO₂ both in N₂O decomposition (Fig. S7A†) and NH₃ oxidation (Fig. S7B†).

Effect of Fe content and FeO_x dispersion. Decreasing dependence of NH₃ conversion vs. Fe content in Al₂O₃-1200 and CeO₂-based samples (0.015 g sample loading for all these experiments) was observed (Fig. 9), similarly to the reaction of N₂O decomposition, and thus can be related to size effect. Moreover, for the Fe/CeO₂ case, it was more prominent due to attenuation of oxygen transfer processes through the Fe–CeO_x interface (Fig. S4†). Nevertheless, unlike the reaction of N₂O decomposition, FeO_x dispersion is not a predominant factor for NH₃ conversion any more. Therefore, at Fe content lower than 3 × 10¹⁹ atoms per m² (the case with Fe/Al₂O₃-C and Fe/Ce/Al₂O₃ samples), the NH₃ conversion value depended mainly on the quantity of exposable Fe (Fe_exp).

Fig. 9 Dependence of NH₃ conversion (solid symbols) and NO_x yield (open symbols) on Fe content in different Fe/(Al₂O₃,CeO₂) samples. Catalyst loading 0.015 g.

At the same time, these were larger FeO_x species that converted NH₃ to NO_x more selectively in all Al₂O₃-based samples (Fig. S8†), finally resulting in quite similar (77–81%, in the case of Fe/Al₂O₃-1200 samples) or growing with Fe content (up to 79% for Fe/Al₂O₃-C) NO_x yield values (Fig. 9). In accordance with findings of Pérez-Ramírez and Kondratenko¹⁶ on bulk Fe₂O₃, the desired reaction follows a Mars–van Krevelen-type scheme involving the participation of lattice oxygen in the NH₃ conversion to NO. The degree of reduction of the oxide surface was shown to determine the product distribution. One can reasonably suppose that for small species, the FeO_x surface should be in a more reduced state because of a smaller rate of dissociative activation of O₂ molecules on the Fe–Al interface. In addition, a growing contribution from low selective NH₃ oxidation on Al₂O₃ should be accounted as well.

We also believe that the more oxidized state of FeO_x anchored to CeO₂, resulting from more efficient O transfer through the Fe–Ce interface, causes higher selectivity to NO_x in the 3.8Fe/Ce/Al₂O₃ and 9.9Fe/Ce/Al₂O₃ samples compared to the Fe/Al₂O₃-C ones with similar dispersion. In addition, substantially smaller values of N₂O yield were measured on Fe/Ce/Al₂O₃ samples, which agree with their higher activity towards N₂O decomposition.

Conclusions

We have shown that replacement of Al₂O₃ by CeO₂ as the support for MeO_x (Me = Fe, Co, Ni) characterized by increased lattice oxygen mobility resulted in substantial increase of the rates of ¹⁶O/¹⁸O exchange at 800 °C, N₂O decomposition and NH₃ oxidation on corresponding samples. Oxygen transfer from the fluorite lattice to the surface by extended oxygen vacancies arose after insertion of Meⁿ⁺ ions into the fluorite lattice was shown to be responsible for the enforced rate of isotope exchange.

Formation of low active spinel-like Me–Al–O structures restricted the application of Co and Ni oxides in any Al₂O₃ containing catalysts. For CeO₂- and Al₂O₃-based samples with supported FeO_x, an obvious size effect was observed in both reactions. Therefore, highly dispersed FeO_x was substantially more active due to change of contribution of oxygen supply from the support (both ceria and, probably, alumina) to reduced surface sites through Fe–CeO_x interface thus increasing the rate of O₂ desorption. However, smaller species oxidized NH₃ to NO_x less selectively.

Using high surface area Al₂O₃-C, the samples with high content of dispersed FeO_x in the weight unit were synthesized to increase observable N₂O conversion. To increase the surface area of CeO₂ and thus efficiently use the promoting effect of oxygen mobility, CeO₂ was dispersed onto alumina by precipitation. Fe/Ce/Al₂O₃ samples with optimal content of Fe revealed superior activity in N₂O decomposition and NH₃ oxidation to NO_x compared with Fe/Al₂O₃ with similar S_BET and dispersion of FeO_x species values.

Acknowledgements

The authors thank Mrs. I. Kharina for preparation of the Al₂O₃-C sample.

Notes and references

1. J. Pérez-Ramírez, F. Kapteijn, K. Schöffel and J. A. Moulijn, Appl. Catal., B, 2003, 44, 117.
2. P. Granger, P. Esteves, S. Kieger, L. Navascues and G. Leclercq, Appl. Catal., B, 2006, 62, 236.
3. P. Esteves, Y. Wu, C. Dujardin, M. K. Dongare and P. Granger, Catal. Today, 2011, 176, 453.
4. D. V. Ivanov, E. M. Sadovskaya, L. G. Pinaeva and L. A. Isupova, J. Catal., 2009, 267, 5.
5. Y. Wu, X. Ni, A. Beaurain, C. Dujardin and P. Granger, Appl. Catal., B, 2012, 125, 149.
6. D. V. Ivanov, L. G. Pinaeva, L. A. Isupova, E. M. Sadovskaya, I. P. Prosvirin, E. Yu Gerasimov and I. S. Yakovleva, Appl. Catal., A, 2013, 457, 42.
7. J. Pérez-Ramírez and M. Santiago, Chem. Commun., 2007, 619.
8. G. Giecko, T. Borowiecki, W. Gac and J. Kruk, Catal. Today, 2008, 137, 403.
9. J. Kruk, K. Stołecki, K. Michalska, M. Konkol and P. Kowalik, Catal. Today, 2012, 191, 125.
10. E. Wilczkowska, K. Krawczyk, J. Petryk, J. W. Sobczak and Z. Kaszkur, Appl. Catal., A, 2010, 389, 165.
11. E. Iwanek, K. Krawczyk, J. Petryk, J. W. Sobczak and Z. Kaszkur, Appl. Catal., B, 2011, 106, 416.
12. M. Konsolakis, ACS Catal., 2015, 5, 6397.
13. V. A. Sadykov, L. A. Isupova, I. A. Zolotarskii, L. N. Bobrova, A. S. Noskov, V. N. Parmon, E. A. Brushtein, T. V. Telyatnikova, V. I. Chernyshev and V. V. Lunin, Appl. Catal., A, 2000, 204, 59.
14. L. Pinaeva, E. Sutormina, L. Isupova, N. Kulikovskaya and A. Marchuk, RU Pat., 2 430 782 C1, 2010.
15. L. Pinaeva, L. Isupova, I. Prosvirin, E. Sadovskaya, I. Danilova, D. Ivanov and E. Gerasimov, Catal. Lett., 2013, 143, 1294.
16. J. Pérez-Ramírez and E. Kondratenko, J. Catal., 2007, 250, 240.
17. M. Zabilskiy, B. Eravec, P. Djinović and A. Pintar, Chem. Eng. J., 2014, 254, 153.
18. M. Konsolakis, S. A. C. Carabineiro, E. Papista, G. E. Marnellos, P. B. Tavares, J. Agostinho Moreira, Y. Romaguera-Barcelay and J. L. Figueiredo, Catal. Sci. Technol., 2015, 5, 3714.
19. L. Isupova, Yu. Tanashev, I. Kharina, E. Moroz, G. Litvak, N. Boldyreva, E. Paukshtis, E. Burgina, A. Budneva, A. Shmakov, N. Rudina, V. Kruglyakov and V. Parmon, Chem. Eng. J., 2005, 107, 163.
20. I. Danilova, E. Slavinskaya, V. Zaikovskii, A. Ivanova, A. Boronin, R. Gulyaev and Yu. Amosov, Kinet. Catal., 2010, 51, 143.
21. J. Tauc, R. Grigorovici and A. Vancu, Phys. Status Solidi B, 1966, 15, 627.
22. V. Malakhov, N. Boldyreva, A. Vlasov and L. Dovlitova, J. Anal. Chem., 2011, 66, 458.
23. L. G. Pinaeva, E. M. Sadovskaya, A. P. Suknev, V. B. Goncharov and B. S. Bal'zhinimaev, in Mass Spectrometry Handbook, ed. M. S. Lee, Wiley, Hoboken, 2012, pp. 1229–1256.
24. S. Kurajica, J. Popović, E. Tkalčec, B. Gržeta and V. Mandić, Mater. Chem. Phys., 2012, 135, 587.
25. O. Kirichenko, V. Ushakov, E. Moroz and M. Vorob'eva, Kinet. Catal., 1993, 34, 739.
26. Z. Zhang, D. Han, S. Wei and Y. Zhang, J. Catal., 2010, 276, 16.
27. A. Gupta, A. Kumar, U. V. Waghmare and M. S. Hegde, Chem. Mater., 2009, 21, 4880.
28. A. Piras, S. Colussi, A. Trovarelli, V. Sergo, J. Llorca, R. Psaro and L. Sordelli, J. Phys. Chem. B, 2005, 109, 11110.
29. A. Bensalem, J. C. Muller and F. Bozon-Verduraz, J. Chem. Soc., Faraday Trans., 1992, 88, 153.
30. R. Zhang, J. T. Miller and C. D. Baertsch, J. Catal., 2012, 294, 69.
31. R. C. Olegário, E. C. F. Souza, J. F. M. Borges, J. B. M. Cunha, A. V. C. Andrade, S. R. M. Antunes and A. C. Antunes, Dyes Pigm., 2013, 97, 113.
32. C. He, M. Paulus, W. Chu, J. Find, J. A. Nickl and K. Kohler, Catal. Today, 2008, 131, 305.
33. I. G. Casella and M. R. Guascito, J. Electroanal. Chem., 1999, 476, 54.
34. H. Xiong, Y. Zhang, K. Liew and J. Li, J. Mol. Catal. A: Chem., 2005, 231, 145.
35. M. H. Kim and K.-H. Choo, Catal. Commun., 2007, 8, 462.
36. J. Yang, H. Liu, W. N. Martens and R. L. Frost, J. Phys. Chem., 2010, C114, 111.
37. X. Duan, M. Pan, F. Yu and D. Yuan, J. Alloys Compd., 2011, 509, 1079.
38. N. Srisawad, W. Chaitree, O. Mekasuwandumrong, P. Praserthdam and J. Panpranot, J. Nanomater., 2012, 2012, 1.
39. D. Barreca and C. Massignam, Chem. Mater., 2001, 13, 588.
40. S. C. Petitto, E. M. Marsh, G. A. Carson and M. A. Langell, J. Mol. Catal. A: Chem., 2008, 281, 49.
41. D. Nikolova, R. Edreva-Kardjieva, G. Gouliev, T. Grozeva and P. Tzvetkov, Appl. Catal., A, 2006, 297, 135.
42. P. Dufresne, E. Grimblot and J. P. Bonnelle, J. Phys. Chem., 1981, 85, 2344.
43. E. Laurent and B. Delmon, J. Catal., 1994, 146, 281.
44. C. Jiménez-González, Z. Boukha, B. de Rivas, J. J. Delgado, M. A. Cauqui, J. R. González-Velasco, J. I. Gutiérrez-Ortiz and R. López-Fonseca, Appl. Catal., A, 2013, 466, 9.
45. P. Prieto, V. Nistor, K. Nouneh, M. Oyama, M. Abd-Leftil and R. Diaz, Appl. Surf. Sci., 2012, 258, 8807.
46. S. Suzuki, K. Yanagihara and K. Hirokawa, Surf. Interface Anal., 2000, 30, 372.
47. L. Chen, P. Fleming, V. Morris, J. D. Holmes and M. A. Morris, J. Phys. Chem. C, 2010, 114, 12909.
48. J. Z. Shyu, W. H. Weber and H. S. Gandhi, J. Phys. Chem., 1988, 92, 4964.
49. C. Ge, L. Liu, Z. Liu, X. Yao, Y. Cao, C. Tang, F. Gao and L. Dong, Catal. Commun., 2014, 51, 95.
50. M. Zabilskiy, P. Djinović, E. Tchernychova, O. Tkachenko, L. Kustov and A. Pintar, ACS Catal., 2015, 5, 5357.
51. P. Pomonis, D. Vattis, A. Lycourghiotis and C. Kordulis, J. Chem. Soc., Faraday Trans. 1, 1985, 81, 2043.
52. M. C. Goromonzi, S. K. Raman, B. D. Padalia, A. Manohar and P. N. Koul, React. Kinet. Catal. Lett., 1980, 15, 131.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5cy01381j

---