Pure and Fe-doped CeO2 nanoparticles obtained by microwave assisted combustion synthesis: physico-chemical properties ruling their catalytic activity towards CO oxidation and soot combustion

Pure and Fe-doped CeO2 nanoparticles obtained by microwave assisted combustion synthesis: Physico-chemical properties ruling their catalytic activity towards CO oxidation and soot combustion / SAHOO, TAPAS RANJAN; ARMANDI, MARCO; Arletti, Rossella; PIUMETTI, MARCO; BENSAID, SAMIR; Manzoli, Maela; Panda, Sirish R.; BONELLI, BARBARA. In: APPLIED CATALYSIS. B, ENVIRONMENTAL. ISSN 0926-3373. STAMPA. 211(2017), pp. 31-45. [10.1016/j.apcatb.2017.04.032] Original Pure and Fe-doped CeO2 nanoparticles obtained by microwave assisted combustion synthesis: Physico-chemical properties ruling their catalytic activity towards CO oxidation

5 more frequently and at higher energy, facilitating bonds breakage and new bonds formation: the heat needed for the reaction is generated within the mixture, without the necessity of an external source.
MACS allows rapidly obtaining considerable amounts of phases that are stable at high temperature (oxides, nitrides, carbides) and characterized by nanometer size, high specific surface area and high defectivity, all parameters that are crucial for catalytic applications, as well.
To the best of our knowledge, no paper on Fe-doped CeO2 powders obtained by MACS for catalytic applications has been published, so far: in this work, MACS was employed to synthesize both pure and Fe-doped CeO2 powders (with composition Ce1-xFexO2 where x = 0.03 or 0.06).
The choice of Fe as a dopant stems from its higher abundance with respect to Lanthanides, and the possibility to increase the amount of oxygen vacancies, which can lead to a higher number of labile oxygen species at NPs surface, with an ultimate positive impact on the catalytic activity.
As a catalyst for energy and environmental applications, CeO2 has been recently proposed for the splitting of water [19,20], although CeO2-based nanocatalysts gave good performance (in terms of both activity and selectivity) also in the production of H2 from ethanol reforming [21]. Similarly, CeO2-supported Rh, Pd, Pt and Au NPs showed enhanced catalytic properties in two important reactions for H2 purification, i.e. the preferential oxidation of CO (COPROX) in the presence of H2 and the water gas shift (WGS) reaction [22][23][24][25][26][27][28][29][30][31][32][33]. In this work, yet other two reactions of environmental relevance were studied, i.e. oxidation of CO and soot combustion, for which CeO2based nanocatalysts are investigated in the literature [34][35][36][37][38][39][40][41][42][43][44][45]. In the former reaction, the presence of surface labile oxygen species (O) is a crucial aspect, whereas in the latter reaction, the textural and structural properties of the catalyst as well as the contact points between soot and catalyst particles play a crucial role. 6

Materials synthesis
Stoichiometric amounts of analytical grade reagents were used, i.e. 1 g ceric ammonium nitrate ((NH4)2[Ce(NO3)6]), and either 0.023 g or 0.047 g ferric nitrate (Fe(NO3)3) in order to obtain Ce0.98Fe0.03O2 and Ce0.94Fe0.06O2 compositions, respectively. Correspondingly, either 0.5277 g or 0.5395g glycine (fuel) was added as dictated by the oxidizer/fuel ratio [46]. The resulting mixture was exposed to MW irradiation in a microwave oven, heated at high power of 800 W and frequency of 2450 MHz, for a maximum of 3 min. During this period, the redox mixture boiled, underwent dehydration and then was ignited (due to internal heating), resulting in an exothermic reaction with a flame persistent for 2-4 secs. This process is instantaneous, and the resulting residue was powdery and crystalline (see below). The three samples will be referred to as CeO2 (0 at. % Fe), Ce_Fe3 and Ce_Fe6 (3, 6 at. % Fe, respectively).

Methods
N2 sorption isotherms at −196 °C were measured on the powders previously outgassed at 150 °C in order to remove water and other atmospheric contaminants (Quantachrome Autosorb 1C); samples specific surface area (SSA) was calculated according to the Brunauer-Emmett-Teller (BET) method, the corresponding values being reported in Table 1.
Field Emission Scanning Electron Microscopy (FESEM) micrographs (Fig. SM1) were collected on a Zeiss Merlin microscope equipped with a Gemini II column.
Electron micrographs were obtained on a Jeol 3010-UHR high-resolution transmission electron microscope (HR-TEM) operating at 300 kV and equipped with a LaB6 filament and an Oxford Inca Energy TEM 300 EDX analyser by Oxford Link. Digital micrographs were acquired on an 7 Ultrascan 1000 camera and processed by Gatan digital micrograph. Before experiments, the powder samples were milled in an agate mortar and deposited on a copper grid covered with Lacey carbon film.
To measure X-ray Fluorescence (XRF), samples were pressed in order to obtain pellets of ca. 1 cm 2 area, and analyzed by means of a Rigaku ZSX 100s instrument, on 0.3 cm 2 spots. X-Ray Powder Diffraction (XRPD) experiments were performed at the ID22 high-resolution beamline at ESRF (European Synchrotron Radiation Facility) in Grenoble (France) with a fixed wavelength of 0.4106 Å on a multi-channel detector system. The powder samples were loaded and packed in a 0.3 mm boron capillary, mounted on a standard goniometric head, and spinned during data collection. Structural and unit cell refinements were performed by full profile Rietveld analysis using the GSAS package [47] with EXPGUI interface [48]. The refinements were performed in the space group Fm-3m. The Bragg peak profile was modeled by using a pseudo-Voigt function [49] with 0.001% cutoff peak intensity. The background curve was fitted using a Chebyschev polynomial with 9 variable coefficients. The 2θ−zero shift was accurately refined in separate cycles to avoid correlation with unit cell parameter. The scale factor, unit-cell parameters and displacement thermal parameter were allowed to vary. The high resolution and high flux available on the beamline allowed identifying the presence of CeFeO3 in both Ce_Fe3 and Ce_Fe6 samples as secondary phase (vide infra). The refinements of the collected patterns were treated as a two phases analysis and quantitative phase analyses (QPA) were performed by allowing the variation of the two scale factors.
From XRPD analyses, information on the full width at half maximum (FWHM) was extracted and used to estimate the average micro-strain and crystallite size calculated form Williamson -Hall plot, on the basis of the eq. (1): In eq. (1), β is FWHM,  is the wavelength, t is the crystallite size,  is the Bragg angle and  is the micro-strain. The results of the refinements and the XRPD are reported in Table 2 and Figure   2, respectively. Williamson-Hall plots for the three samples are reported in Figure 3.
XPS (X-ray Photoelectron Spectroscopy) analysis was carried out on a XPS PHI 5000 Versa probe apparatus, using a band-pass energy of 187.85 eV, a 45° take off angle and a 100.0 m diameter X-ray spot size for survey spectra. High-resolution XP spectra were recorded in the following conditions: pass energy of 20 eV, resolution of 0. Diffuse Reflectance (DR) UV-Vis spectra were measured on a Cary 5000 UV-Vis-NIR spectrophotometer (Varian instruments).

Catalytic tests
Catalytic activity of the samples was tested in a temperature-programmed oxidation (TPO) setup, comprising a quartz U-tube micro-reactor; a PID-regulated (Proportional-Integral-Derivative) furnace; a K-type thermocouple (placed at the reactor inlet in such a way that its tip is as close as possible to the reactor bed) and a non-dispersive infrared (NDIR) analyser (ABB Uras 14) as gas detector. In order to get insight into the phenomena related to the behaviour of catalytically active oxygen species in the presence of soot and in the absence of O2(g), soot-TPR measurements ( Fig. SM6) were carried out as follows: a bed mixture like that used for soot combustion tests was prepared in "tight contact" conditions and put inside the fixed-bed micro-reactor where pure N2(g) was fed at 100 ml min -1 . The furnace temperature was increased from ambient to 650 °C (heating rate = 5 °C min -1 ) and CO/CO2 concentrations were measured at the reactor outlet (NDRI analysers).

CO oxidation tests
Additional repeatability tests were carried out: first, fresh powders were tested under the same conditions adopted during previous soot combustion tests, and then each (newly tested) powder was recovered at the end of the first cycle and subjected to a second oxidation cycle (by re-addition of the same amount of soot), to check stability after exposure to high temperatures.  To gain more insights on the samples morphology, selected TEM images of the CeO2, Ce_Fe3 and Ce_Fe6 samples are reported in Fig. 1a, 1c  The XRPD pattern of the three samples are reported in Fig. 2: that of pure CeO2 corresponds to the well-known fluorite-type structure with space group Fm-3m. As a whole, no appreciable shift of the peak positions was observed with Fe-doped CeO2 samples, indicating tiny variations of the unit cell among the three samples. On the contrary, a decrease in the peak broadening from CeO2

Results and Discussion
to Fe-doped CeO2 was clearly recognizable, indicating an increase of grains size, in agreement with morphological analysis (vide supra).
Fe-doped samples maintain the same structure of CeO2, though additional and very low intensity reflections are observed, corresponding to the CeFeO3 crystalline phase (asterisks in Fig. 2). Based on QPA performed by the Rietveld method, the estimated amount of the secondary phase was 1.6 wt. % and 5 wt. % CeFeO3 with Ce_Fe3 and Ce_Fe6, respectively.

13
Such results allow measuring the amount of Fe present in the secondary phase (CeFeO3) and,  Williamson -Hall plots and the derived data are reported in Fig. 3 and Table 2, respectively. As already noticed form the inspection of XRPD patterns in Fig. 2, the grain size increases from pure to Fe-doped CeO2, whereas no differences are observed in the grain size of the two Fe-doped samples. The only observable difference is found in the grains micro-strain, higher with Ce_Fe6 sample, and that could be due to a higher degree of lattice distortion at higher Fe concentration.
Indeed, the crystallinity of the CeO2, Ce_Fe3 and Ce_Fe6 samples was confirmed by HR-TEM analysis (Fig. 1b, 1d and 1f In agreement with XRDP measurement, after Fe doping, the fringes related to the CeFeO3 phase (JCPDS file number 00-022-0166) are also observed with both Ce_Fe3 and Ce_Fe6 samples.
Moreover, due to the much smaller ionic size of Fe 3+ as compared to Ce 4+ , it can be inferred that Fe 3+ can occupy either network Ce 4+ sites or interstitial sites in the fluorite lattice, giving rise to tiny changes in the lattice parameter, as observed by XRDP cell parameter refinement (Table 2).
Therefore, the amount of Fe added by synthesis can rule the formation of different amounts of substoichiometric phases in Ce_Fe3, whereas in Ce_Fe6 sample a higher Fe content likely promotes the formation of the CeFeO3 phase and possibly inhibits formation of the Ce6O11 defective phase.
This result is in agreement with the assumption that with Ce_Fe6 sample a lower amount of Fe 3+ is actually included in the fluorite with respect to Ce_Fe3 sample. Therefore, the addition of Fe affects the formation of defective ceria phases, as observed by the analysis of HRTEM fringes, and the larger amount of interstitial Fe 3+ found for Ce_Fe3 could induce an increase in the structural defectiveness of ceria.
The presence of a Fe-rich shell can be assessed by means of XPS, a technique that allows studying the chemical composition of the surface. The most relevant results obtained by the curve-fit of XPspectra in the O 1s, Ce 3d and Fe 2p BE ranges are reported in Table 3. The position of the Fe2p 3/2 and Fe 2p1/2 lines agrees with the presence of Fe 3+ species, the corresponding XP spectra being reported in the Supplementary Material (Fig. SM2).
Concerning O 1s lines (Fig. 4), a satisfactory curve-fit is obtained by using two components, related to the presence of surface and sub-surface oxygen species, referred to as O (BE ca. 531 eV) and O (BE ca. 529 eV), respectively. The former are labile and more reactive species with respect to sub-surface O species, and play an important role in metal oxides-catalyzed processes [34,53]. With Fe-doped samples, the position of the O 1s lines does not change significantly, but the atomic percentage of O species varies, as shown by the corresponding values in as thoroughly discussed in the literature [54].
The position of Ce peaks is reported in Table 3, along with surface chemical composition. The total amount of surface Ce 3+ species, as calculated according to eq. (2):   Table 1 as normalized to sample unit mass, agree with such hypothesis, in that the total amount of consumed H2 increases with the Fe content.

Characterization of surface species by means of IR spectroscopy and CO adsorption at nominal -196 °C
The nature of surface species can be studied through adsorption of probe molecules as followed by IR spectroscopy, which is a fundamental technique for the study of surface species [58][59][60], like the different types of OH groups [16,61,62] usually found at the surface of ceria and depicted in Scheme 1. Figure 6 reports the IR spectra of the CeO2 sample pretreated as follows: outgassed at 250 °C (curve a), oxidized at 400 °C (curve b) and then reduced in H2 at 200 °C (curve c). The curves in Fig. 6 are characterized by the typical scattering profile of ceria above 600 cm -1 [61]. No remarkable signals ascribable to electronic effects were observed upon reduction (curve (c) in Fig.   6), at variance with literature reports on ceria reduced at higher temperatures [62]. This is likely due to the low reduction temperature adopted in this work.
The IR spectrum of CeO2 sample outgassed at 250 °C shows a band centered at 3630 cm -1 in the OH stretching region (3800-3500 cm -1 , inset to and 3647 cm -1 to OHIIB species where cerium occurred as Ce 4+ and Ce 3+ species, respectively [61]. This assignment is further confirmed by the presence of a weak band at about 2130 cm -1 related to the forbidden 2 F5/2 2 F7/2 electronic transition of Ce 3+ [62] and indicating that outgassing at 250°C has a reducing effect on the surface. Several bands in the 3000 -2800 cm -1 and 1700 -1250 cm -1 range of curve a are assigned to IR modes of glycine [63], indicating that residual molecules of the fuel are likely adsorbed at the surface of CeO2 sample. Such hypothesis is confirmed by TG -Mass analysis (reported in Fig. SM5): the TG curve shows a limited mass loss (ca. 2.5 wt. %) in the temperature range between 100 and 300 °C, accompanied by a corresponding peak with mass = 30 u in the mass spectrum, ascribable to NH2CH2 + species, i.e. the main fragment of glycine [64].
The amount of residual glycine was however very low, as confirmed by the limited mass loss occurring at low temperature (Fig. SM5), and therefore its influence on the catalytic activity of the samples is reasonably negligible. Ce 3+ species [62,65,66]. The 2152 cm -1 band is due to CO molecules H-bonded to OH groups originally absorbing at 3630 cm -1 , as confirmed by the negative band observed in the OH stretch region (3750-3000 cm -1 , inset to Fig. 7a). According to the literature, CO molecules adsorbed on Ce 4+ sites give rise to IR bands in the 2187 -2165 cm -1 range [65], not observed in the present case: Ce 4+ sites, which are strong Lewis acids, are likely occupied by adsorbed glycine molecules.
Nonetheless, the under vacuum treatment at 250 °C has a partially reducing effect on ceria surface [61,62], favoring the formation of surface Ce 3+ species, which were also present on the assynthesized sample, in agreement with XPS analysis.
In order to remove the organic residues, the CeO2 sample was oxidized at 400 C°, the corresponding IR spectrum being reported as curve b in Fig. 6: as expected, IR bands of glycine disappear after oxidation, minor bands in the 3000 -2800 cm -1 range being instead due to surface contamination by vacuum grease always present during the experiments. Peaks in the 1700-1250 cm -1 range are assigned to carbonate-like surface species [62,66,67], usually observed at ceria surface, the assignment of which is not straightforward [62]. All this notwithstanding, the relatively high thermal stability of such carbonate-like species allows inferring that mainly polydentate species are present [62], since lower coordination carbonates are removed above 300 °C. More interestingly, in the OH stretch region (inset to Fig. 6) several new bands are observed upon oxidation at 400 °C: the 3715 cm -1 band is assigned to mono-coordinated OHI species (Scheme 1), whereas signals at 3687, 3650 and 3630 cm -1 to bridged OHIIA and OHIIB species [61,62]. In this case, the band at about 2130 cm -1 is not observed, confirming the effectiveness of the oxidation treatment. Fig. 7b reports difference IR spectra obtained after adsorption of CO at nominal -196 °C on oxidized CeO2 sample. As a first comment, the intensity of the bands due to adsorbed CO species is one order of magnitude higher than in IR spectra reported in Fig. 7a Correspondingly, a band at 2127 cm -1 (arrow) is seen, due to CO molecules adsorbed on reduced Ce 3+ sites. The poor intensity of the band corresponding to CO interacting with Ce 3+ sites is not surprising, since in the present case reduction was carried out in quite mild conditions, as confirmed by the lack of the band at about 2130 cm -1 in curve c of Fig. 6.
In order to understand the effect of oxidation and/or reduction on the hydroxyls population, a curve-fitting procedure was carried on the hydroxyls IR spectra of the CeO2 sample after oxidation at 400 °C and reduction at 200 °C. Fig. 8

UV-Vis spectroscopy
The DR UV-Vis spectra of the three as-synthesized samples under vacuum at room temperature are shown in Figure 11: Ce_Fe3 and Ce_Fe6 samples, respectively. This effect has been already observed with variously doped CeO2, and corresponds to a decrease of the oxide band-gap as consequence of Fe 3+ doping, though restricted to near subsurface layers [70]. In the present case, the band-gap changes from 3.4 eV with pure CeO2 to 3.2 eV and 3.1 eV for Ce_Fe3 and Ce_Fe6 samples, respectively, in agreement with the 0.25-0.12 eV value found in the IR spectra reported in Fig. 9. Such feature may represent a possibility for Fe-doped CeO2 obtained by MACS to have some photocatalytic applications, although not explored in this paper. Figure 12 shows CO conversion to CO2 in the 150 -500 °C range for both the catalized and uncatalized reaction (dotted curve): as a whole, with Fe-doped samples a remarkable improvement of CO oxidation was observed with respect to CeO2 sample.

Catalytic tests concerning activity of the samples towards CO oxidation
In particular, the Ce_Fe3 sample exhibited the best perfomance in terms of both T10%-50%-90% values (277, 298, 327°C, respectively) and specific CO oxidation rate (= 59.13 mmol m -2 h -1 at 285°C) ( Fe content seems to inhinibt the formation of the Ce6O11 defective phase, with consequent formation of a higher amount of both CeO2 and CeFeO3 phases. The latter phase appears less effective towards CO oxidation with respect to highly dispersed Fe-species within the ceria lattice. Finally, the CeO2 sample resulted the least performing CO oxidation catalyst, in terms of both T10%-50%-90% values and specific oxidation rate, despite its higher BET surface area, likely due to 25 the lack of the positive Ce-Fe synergy that, according to the literature, is achieved by combining the redox behavior of cerium (Ce 4+ /Ce 3+ ) and iron (Fe 3+ /Fe 2+ ) species [50]. Interestingly, a fair correlation is observed between CO oxidation activity of the samples and the corresponding amount of surface Oα species (Table 3), as measured XPS (vide supra). The latter finding confirms that Oα species can be directly involved in the catalytic oxidation of CO, at variance with other ceria-based nanostructures (namely, Ce-Zr-O and Ce-Pr-O systems) [36,71,72]. According to previous work [73], this suggest the occurrence of either a Langmuir-Hinshelwood or Eley-Rideal type mechanism, in which the determining steps occur in the gas phase: however, further considerations about reaction mechanisms are out of the scopes of this paper and will be the matter of future work. Figure 13 shows the soot conversion to COx (%) in the 150 -650 °C range during both the catalysed reaction (in "tight contact" conditions) and the non-catalysed reaction (soot only).
The contact points between soot and catalyst particles play a key role in such solid-solid reaction mediated by gas-phase oxygen (O2(g)) and therefore the smaller size of CeO2 NPs improves the contact interactions with soot aggregates, finally leading to a higher activity. If, however, we consider that in "tight contact" conditions the soot is completely surrounded and in contact with the catalyst (being the latter in large excess, as tests were performed with a catalyst/soot mass ratio equal to 9, vide supra), one can assume that soot reactivity directly depends on the specific surface (which is not the case in "loose" contact) [41,42,45]. In other words, soot is in direct contact with the same mass of catalyst in all experiments, and the mechanical force exerted in "tight contact" conditions makes fully accessible the catalyst surface, provided that no inner porosities inaccessible to soot (i.e. well below soot particle sizes) are present. This, in fact, occurs for the investigated samples, for which microporous intra-particle volume is nearly absent and the interparticle volume is mesoporous, according to N2 sorption isotherms at -196 °C (vide supra). Further insight into soot oxidation can be derived by the soot-TPR curves reported in the Supporting Material (Fig. SM6), showing that the CeO2 sample provides the largest amount of CO2: also in the absence of O2(g), CeO2 NPs confirm their higher activity, ascribable to a higher tendency of surface oxygen species to react with soot particles. Above 450 °C, the CO2 emissions of the three catalysts are very similar: since surface oxygen species have been consumed at low T, reaction occurs with oxygen species coming form the bulk. The latter phenomenon requires higher T, at which the three catalysts perfom similarly, being Fe-doping mostly related to the surface.
Concerning soot combustion, a second set of experiments was carried out with the fresh powders, to test both the repeatibility of the catalytic measurements and the quality (and the overall homogeneity) of samples obtainable by MACS. Figure 14 reports the two catalytic cycles obtained with fresh powders: the first cycles nearly cohincide to previous ones, as comparable results are obtained with respect to Fig. 13, given the very limited differences between the values of T50% in Fig. 13 and Fig. 14  The catalysts were then recovered, and the soot-catalyst bed was re-prepared by adding soot and repeating the measurement: the obtained soot conversion curves (hollow symbols in Fig. 14) show that the catalyst degradation was limited, though selectivity to CO2 decreased, especially with CeO2 sample (from 86% to 60% during the second cycle). This effect was also evident with the Fe-doped samples, where CO oxidation activity decreased shifting towards higher T. Future work will concern the improvement of catalysts stability, with the aim of obtaining catalysts able to withstand (technologically relevant) phenomena of ageing.   Table 3 Most relevant results as obtained by means of XPS analysis and related curve-fitting procedure.