Effect of MnO_x modification and template type on the catalytic performance of ceria-zirconia in CO and soot oxidation

CO oxidation tests under pulse regime

The catalysts (100 mg) were tested in CO oxidation in a fixed bed flow reactor using pulse feeding of the reagents (2 vol% CO and 1 vol% O₂ in He). The complete description of the experimental setup, the flow mode, the temperature range and the procedure for calculating CO conversion can be found in our previous work [10].

Soot oxidation tests

Soot (Printex-U, Degussa AG) and a catalyst were mixed in a ratio of 1:20. Though the contact between soot and catalyst particles is usually considered to be loose [15] we decided to use the tight contact mode to better understand the unique properties of the catalysts [16]. To provide a tight contact the catalyst and soot were ground in an agate mortar for 3 min. Then 7–14 mg of the mixture was placed into a corundum crucible. The catalyst performance in soot oxidation was studied by simultaneous thermal gravimetric analysis and differential scanning calorimetry (TGA/DSC) on an STA 449C Jupiter instrument (NETZSCH, Germany). The catalyst–soot mixture was heated from 40 to 800 °C at a rate of 10 °C/min in air flow. The TGA conversion curves were normalized by subtracting the weight loss below 200 °C, since the water desorption and decomposition of oxygen-containing species on the soot surface contributes to this low temperature range [17]. The reaction products were monitored by an online quadrupole mass spectrometer.

Composition, textural parameters, and morphology

The total manganese loadings in Mn/CZ (SD) and Mn/CZ(CTAB) determined by AAS (Table 1) are almost the same and close to the target 8 wt%.

The composition, bulk and surface texture, and morphology of the CZ(SD), CZ(CTAB), Mn/CZ(SD) and Mn/CZ(CTAB) catalysts were investigated by the SEM-EDS.

For each type of template, the SEM images of CZ and Mn-CZ are similar (Fig. 1). The CTAB-templated catalysts exhibit a wide range of particle sizes and shapes. The textural characteristics of the biomorphic systems are significantly different: the catalyst surface reproduces the pore structure of wood sawdust.

Figure 1:

SEM images of catalysts.

According to the SEM-EDS data, Ce, Zr, and O were found in all samples, while Mn was only detected in the modified catalysts.

The atomic ratios of the main elements in the catalysts calculated from the EDS data are presented in Table 1. The surface of all samples, except for CZ(SD), is enriched in cerium, and the surface of Mn-CZ(CTAB) is enriched in Mn, since the target atomic ratio of these elements is 4:1:1.2. The manganese percentage on the surface of Mn-CZ(CTAB) exceeds that on Mn-CZ(SD) three times. The enrichment with Ce can be caused by the phase segregation typical for CZ systems [18]. The high concentration of Mn on the surface of Mn-CZ(CTAB) may indicate the presence of separate manganese oxide phases or small particles. The presence of Mn enriched areas in the EDS maps of the surface of Mn-CZ(CTAB) confirms this assumption and indicates weaker interaction between MnO_x and the CZ support in this sample.

CZ(CTAB) and Mn-CZ(CTAB) also contained a small amount of Br (0.5–1.5 at%) remained after the template decomposition, which can affect the catalytic activity [19]. However, considering low concentration of the residual bromine and high stability of oxide systems compared to supported metal catalysts, the Br effect on the catalyst efficiency appears to be small.

The EDS spectra of biomorphic samples comprise low intensity K, Ca and Na lines (less than 0.5 at%). Apparently, EDS underestimates the real content of light elements which are difficult to detect by this method. The signals of these elements can be affected by systematic errors and background subtraction artifacts. However, the presence of basic oxide components (mainly K₂O, CaO, Na₂O and MgO) in the pine sawdust ash is well established [20].

It should be emphasized that a significant fraction of carbon was found by EDS in all samples (from 16 to 27 at%). This fact can be explained by several reasons [11]: (i) the incomplete template decomposition during calcination step, (ii) the presence of strongly bonded carbonate groups on the surface due to the CO₂ adsorption from the air, and (iii) the use of a conductive double sided carbon tape to attach the sample powder to a SEM holder. Probably, all these reasons contribute to a real experiment.

According to the EDS elemental maps (see Supplementary materials, Fig. S1), the uniform surface distribution of Ce, Zr, O and Mn is observed in all samples except Mn-CZ(CTAB), which contains manganese enriched regions. The possible overestimation of the local signal intensity of manganese species can be associated with the high surface roughness of this sample.

The textural properties were also analyzed by low-temperature N₂ physisorption. The recorded isotherms are shown in Fig.2. According to the IUPAC classification the adsorption-desorption isotherms for the biomorphic catalysts can be ascribed to Type I (micropores), probably, with a slight contribution from Type II (macropores) [21]. A decrease in the nitrogen volume adsorbed by CZ(SD) starting from p/p₀ = 0.2 is an artifact resulted from the errors associated with the low surface area of the sample and complete filling of all micropores at low p/p₀.

Figure 2:

N₂ physisorption isotherms of catalysts (a–d). Inserts in Figs. a–b show the pore size distributions.

The isotherms of the CTAB-templated samples are typical for macro-/mesoporous materials (combination of Types II and IV) with the broad pore size distribution. They shows a hysteresis loop of Type H3, which may be attributed to the pore network consisting of macropores incompletely filled with pore condensate [21]. Pore size distributions (Fig. 2) confirm the significant contribution from micropores and narrow mesopores in the CZ(SD) and the broad distribution of mesopores in the CZ(CTAB).

The BET specific surface areas (A_s) of the catalysts are presented in Table 1. Despite the different pore structure, the CZ(CTAB) and CZ(SD) samples shows similar and relatively high specific surface areas that can have a positive effect on the catalyst performance, especially in soot oxidation. At the meso-scale, the boundary between deposited soot particles and the catalyst can play the significant role in providing sufficient contacts for effective soot oxidation [22]. Moreover, some authors [23], [24] supposed that bulk oxygen mobility is not essential for catalytic activity, while the stabilization of the external surface area and surface active sites could be a more important factor. Despite some controversy in the literature [25], it seems obvious that even for the ceria-based system with the improved structural properties and high defectiveness, the contact area between the soot and catalyst is of critical importance. Taking into account the similar A_s values for CZ(CTAB) and CZ(SD), the comparable activity of these catalysts in the oxidation of soot can be expected, but the wide distribution of mesopores in the former sample can give him an advantage.

Modification with manganese significantly decreased A_s for the CTAB-templated catalyst. Earlier, the similar decrease in A_s after co-precipitation of the triple Mn-CZ(CTAB) system was explained by the instability of the manganese complexes with citric acid under the synthesis conditions, which inhibited the formation of the CZ pore structure [11]. However, in the case of the impregnation with Mn salt used in this work the decrease in A_s can be caused by the inhomogeneous distribution of large MnO_x aggregates on the surface of CZ(CTAB) detected by the SEM-EDS analysis. These aggregates can block pores reducing A_s. In contrast, manganese addition to CZ(SD) only slightly decreased A_s, possibly due to the stronger interaction between manganese and the CZ support and the uniform distribution of manganese species in the sample. It agrees with the SEM-EDS results.

XRD: phase composition and crystal structure

The XRD diffraction patterns of all the samples (Fig. 3) demonstrate broad reflexes attributed to the mixed ceria-zirconia oxide with the cubic fluorite structure (PDF card No. 01-074-8064). No separate ceria or zirconia phases were found, which confirmed the mixed oxide formation. However, the significant peaks broadening and asymmetry may indicate the phase inhomogeneity and the presence of Ce-enriched phases [18], which is consistent with SEM-EDS results about high fraction of Ce on the surface of all samples except CZ(SD). Though the significant peak broadening complicates calculation of the lattice parameters and may lead to misinterpretation [26], we estimated the lattice parameter a for all the samples (Table 1). It is slightly lower for CZ(SD) than for the CZ(CTAB), which can be caused by the increased defectiveness of the latter. For the modified samples, the lattice parameter is almost the same. The average crystallite size (D) of the CZ phase was calculated with the Scherrer equation (Table 1). It was found that modification with Mn did not affect the CZ crystallite size, while the CTAB-templated catalysts showed slightly larger crystallite size than their biomorphic counterparts.

Figure 3:

XRD patterns of catalysts.

No manganese oxide phases were detected in the XRD patterns of the modified samples. There may be several reasons for this fact [7, 26]: (i) the formation of solid solution of MnO_x in CZ, (ii) the presence of MnO_x as highly dispersed clusters less than 3–5 nm in size, or (iii) a combination of both phenomena. Indeed, the addition of Mn slightly shifted the CZ reflections in the diffraction patterns of Mn-CZ(CTAB) and Mn-CZ(SD) to higher angles. This corresponds to the decrease in the lattice parameter (Table 1) and can be caused by the incorporation of manganese ions into the CZ crystal lattice. The solubility limit of manganese in cerium oxide based on XRD and Raman spectroscopy data is below 3 at% [27] and close to the Mn concentrations used in our work. The reduction of Mn ions that increases the ion radius should lead to an increase in the degree of incorporation of these ions into the crystal lattice of cerium oxide according to work [27], but no direct evidence was provided.

The partial replacement of Ce⁴⁺, Ce³⁺ or even Zr⁴⁺ ions (ionic radii are 0.097, 0.114, and 0.084 nm for Ce⁴⁺, Ce³⁺, and Zr⁴⁺ ions in the 8-fold coordination, respectively [28]) in the CZ cubic lattice with the smaller Mnⁿ⁺ ones (0.083 and 0.065 nm for Mn²⁺ and Mn³⁺, respectively) should affect the lattice parameter. However, since manganese was supported by wet impregnation, the formation of a mixed oxide phase could take place predominantly in the surface and subsurface layers of the catalyst. Such partial incorporation can provide additional structural defects and create more oxygen vacancies on the surface, which is beneficial for the catalytic oxidation. On the other hand, the possibility of small amorphous MnO_x particles formation in the bulk and on the surface of the catalysts cannot be ruled out considering the significant broadening and slight asymmetry of the XRD peaks in the modified catalysts, and the above discussed SEM-EDS data for Mn-CZ(CTAB). It appears that this CTAB-templated sample has higher phase inhomogeneity than Mn-CZ(SD).

It can be concluded that the modification of the CZ catalysts with Mn leads to only partial incorporation of manganese ions into the surface layers of CZ lattice, while most part of manganese is distributed in tiny MnO_x particles on the catalyst surface. These particles can block pores and contribute to the decrease in the specific surface area observed in the Mn modified samples. This blocking is more pronounced for Mn-CZ(CTAB), which exhibits stronger MnO_x segregation on the surface, therefore, the A_s of this sample is the lowest. It seems that in the case of the biomorphic system, the incorporation of manganese into the surface and subsurface layers of CZ lattice proceeds to a greater extent, increasing the structure defectiveness. Compared to Mn-CZ(CTAB), a smaller proportion of manganese oxides is distributed over the surface of biomorphic sample, and this distribution is more uniform, causing only slight decrease in the specific surface area. Based on these results, at least two factors can play a significant role in the oxidation reactions over Mn-modified CZ catalysts: (i) the presence of a sufficient amount of highly dispersed manganese oxide on the surface, which promotes the activation of substrate molecules/particles, and (ii) the incorporation of Mn ions into the CZ lattice, which increases its defectiveness and the mobility of oxygen. However, these factors can contribute differently to the oxidation of CO and soot particles.

H₂-TPR: redox properties

The reducibility of the catalysts was analyzed by the H₂-TPR method. The reduction profiles are shown in Fig. 4.

Figure 4:

H₂-TPR profiles of catalysts.

There are two groups of hydrogen consumption peaks in the H₂-TPR profiles of the unmodified CZ oxides. The first group comprises high temperature peak at ca. 800 °C which corresponds to the bulk reduction of Ce⁴⁺ [29], [30], while Zr⁴⁺ is considered as an unreducible ion [29]. This peak is almost absent in the profile of CZ(SD), probably because of its higher defectiveness that decreases the reduction temperature of the bulk oxide. It can be explained by the influence of ash impurities remaining in the ceramic after decomposition of the wood template. The second group of TPR peaks at 400–600 °C and the low-temperature shoulder can be associated to the reduction of Ce⁴⁺ in the subsurface and surface layers, respectively. The reduction of the surface layers of the CTAB-templated catalyst proceeds at lower temperatures. In addition, the low-temperature shoulder (at 445 °C) is more pronounced for CZ(CTAB). A possible reason for this fact is the relatively high surface reducibility caused by the high rate of oxygen vacancies formation, improved oxygen mobility into subsurface layers, and increased portion of Ce on the surface, detected by SEM-EDS.

The addition of MnO_x strongly modified the H₂-TPR profiles of the CZ systems. It is impossible to unambiguously assign the low-temperature peaks in their profiles to particular MnO_x species or specific reduction steps because the peak positions depend not only on the oxidation states but also on the particle sizes and crystallinity of manganese phases [31], [32]. These difficulties are typical in interpreting TPR profiles of multicomponent oxide systems. Nevertheless, we can assume that the low-temperature peaks centered at 240–250 °C in the H₂-TPR profiles of the modified systems can be attributed to the reduction of (i) surface MnO₂/Mn₂O₃ to Mn₃O₄; and (ii) Mn in surface MnCeZrO_x species. As a Mn²⁺ salt was used for the CZ impregnation, it is reasonable to expect the presence of manganese in the 2+ and 3+ oxidation states rather than in 4+ one. In the middle-temperature range the H₂-TPR peaks centered at 385 and 410 °C are probably related to the bulk reduction of larger Mn₂O₃ crystallites to Mn₃O₄, while the peaks at about 505–512 °C can be ascribed to the reduction of Mn₃O₄ to MnO or to transformation of easily reducible Ce⁴⁺ ions in the solid solutions to Ce³⁺ ones [31, 33]. The peak at 512 °C is more intense for Mn-CZ(SD) than for Mn-CZ(CTAB), which additionally proves easier reduction of lattice Ce⁴⁺ ions and the increase in the defectiveness due to stronger interaction of manganese with CZ crystal lattice in the former sample. High temperature peaks of the bulk reduction of Ce⁴⁺ in large crystallites are observed at lower temperatures (at 755–765 °C) than for the unmodified CZ systems. Thus, the H₂-TPR results correlate well with the SEM-EDS and XRD data. However, the reliable detection of the electronic state of manganese in the modified samples requires the use of additional methods, such as XPS.

XPS

The elemental composition and oxidation states of surface species in the catalysts were analyzed by XPS. For this purpose, the survey and high-resolution Ce3d, O1s, Mn2p and Mn3s XPS spectra were recorded. O1s and Mn3s XPS spectra are shown in Fig. 5 (a, b). The survey XPS spectra (see Supplementary materials, Fig. S2) show the contribution from O, Ce, Zr, Mn, C and Br (only in the CTAB-templated samples), which agrees with the SEM-EDS results. The presence of other elements (e.g. ash impurities) was not detected, probably due to their low concentration in the samples.

Figure 5:

High resolution O1s (a) and Mn3s (b) XPS spectra of catalysts.

The atomic ratios of the metals were calculated from the XPS spectra (Table 1). The Ce:Zr ratios are similar for the CZ systems prepared using both templates; it is slightly higher for CZ(CTAB), which agrees with the SEM-EDS data. In the case of the Mn modified catalysts, the larger Mn fraction on the surface was observed in Mn-CZ(CTAB), confirming our assumption about the weaker interaction between MnO_x and the CZ support and higher concentration of MnO_x on the surface of this sample. The similar CTAB templated catalyst with the non-uniform and predominantly surface distribution of manganese showed the excellent catalytic performance in the CO oxidation because of the introduction of additional adsorption sites and second redox pair (Mn³⁺/Mn²⁺), even though MnO_x particles blocked pores on the catalysts surface [11].

The high-resolution Ce3d spectra (see Supplementary materials, Fig. S3) are similar for all catalysts. The satellite structure of the spectra is typical for Ce⁴⁺ species with the possible presence of only minor amounts of Се³⁺ [10], [11]. The calculated Ce³⁺/Ce⁴⁺ ratios for all samples are presented in Table 1. Modification of CZ with manganese decreased the Ce³⁺/Ce⁴⁺ ratio, which can be interpreted in terms of the electron transfer between Mn³⁺ and Ce³⁺ ions [11, 34]:

Even though the calculated Ce³⁺/Ce⁴⁺ ratios are low they could be overestimated because of the reduction of Ce⁴⁺ ions to Ce³⁺ during XPS analysis [35]. In addition, it is believed that Ce³⁺ ions detected by ex-situ and in-situ XPS are not always active species but rather irrelevant spectators, while the active species are short-lived and are hardly detectable by this technique [36]. Moreover, the authors of work [30] suggested that the catalytic activity of ceria-based systems in CO oxidation correlates with the rate of Ce³⁺ formation rather than with the overall oxygen-storage capacity.

The high-resolution O1s XPS spectra (Fig. 5a) can be deconvoluted with four components attributed to different surface oxygen species. There are many approaches in the literature to the interpretation of O1s XPS spectra of oxide systems [7, 12, 37, 38]. We believe that the most reasonable approach attributes the intense O_α and O_β components in the range of 529–530 eV to the strongly bound lattice oxygen [13], O_γ component at about 532 eV – to the weakly bound catalytically active adsorbed oxygen molecules, peroxides and superoxide ions or other low coordinated highly polarized oxide species [6, 39]. The forth component O_δ at 532–533 eV is usually attributed to oxygen-containing groups adsorbed on the surface [7, 10].

The calculated O_γ percentages (%) in O1s XPS spectra are presented in Table 1. Among unmodified systems the highest fraction of O_γ species is observed for CZ(CTAB). This fact confirms easier surface reducibility of this sample determined by H₂-TPR. A large number of surface centers for oxygen activation, along with the high specific surface area, can promote more efficient contact between the catalyst and the substrate, which is especially important for the soot oxidation reaction [37]. For Mn-modified systems, the increase in the portion of this oxygen component can be also associated with the strength of interaction between MnO_x and ceria-based support. For example, the strong MnO_x–CeO₂ interaction significantly promoted the reactive surface oxygen availability and catalytic properties in CO oxidation in the relatively low temperature range of 100–200 °C [38]. This fact was confirmed by the spin-polarized energy calculations and charge analysis that elucidated the effect of Mn modification in facilitating the formation of surface oxygen vacancies [40]. Thus, the Mn-modified biomorphic catalyst with the relatively high specific surface area and O_γ content is expected to be active in both soot and CO oxidations.

The Mn2p (see Supplementary materials, Fig. S4) and Mn3s (Fig. 5b) XPS spectra are slightly different for Mn-containing catalysts. In contrast to H₂-TPR data that reflects the bulk state of Mn, the surface sensitive XPS spectra do not provide direct evidence for the presence of Mn⁴⁺ component at about 642.4 eV. The multiplet splitting in Mn3s spectrum of Mn-CZ(SD) is 5.35 eV, which nearly coincides with the value of 5.4 eV for Mn₂O₃ [41]. However, for Mn-CZ(CTAB) the Mn3s splitting (5.25 eV) slightly shifts to lower values typical for MnO₂ (4.4 eV) [42]. The binding energies and satellite structures of the Mn2p_3/2 and Mn3s spectra indicate the predominant Mn³⁺ oxidation state [43], but due to small difference in the binding energies of Mn²⁺ and Mn³⁺ species the presence of Mn²⁺ on the catalyst surface cannot be excluded [44], [45]. In summary, XPS analysis did not provide the precise ratios of different manganese species, but it indicates the predominant presence of Mn³⁺ and possible small contribution from other oxidation states, which agrees with the H₂-TPR data. XPS also confirmed the enrichment of the surface of Mn-CZ(CTAB) with manganese and the absence of this enrichment in the case of Mn-CZ(SD).

Raman spectroscopy

Raman spectroscopy is a convenient tool for analyzing the structural properties of oxide materials due to its ability to detect low-frequency modes of metal-oxygen bond vibrations from different crystalline phases which are not observed with XRD [7].

The triply degenerate Raman active optical phonon F_2g mode located at about 472 cm⁻¹ in the Raman spectra of the catalysts (Fig. 6a) is attributed to the symmetric stretching vibration of the Ce-O₈ crystal unit of the cubic fluorite lattice [46]. Fig. 6b enlarges this region to clarify the differences in the F_2g line position between the samples. Interestingly, F_2g signal in the spectrum of CZ(SD) is blue-shifted and broadened compared to the CTAB-templated sample, which can be explained by the change in the particles size and/or the increase of the number of oxygen vacancies and contraction of the lattice [7, 47, 48]. These effects can also slightly shift the XRD reflections of the biomorphic sample compared to CZ(CTAB) leading to the observed decrease in the lattice parameter.

Figure 6:

Raman spectra of catalysts (a) and enlarged F_2g regions of spectra (b).

The modification of both CZ samples with Mn significantly broadened this line and decreased its intensity. In addition, a noticeable redshift of this band was observed upon modification of the biomorphic sample, while for the CTAB-templated one the position of the F_2g line remained practically unchanged. The redshift of the main Raman band indicates the increase in the number of oxygen vacancies or dilation/contraction of the crystal lattice [47]. This effect can arise due to the partial incorporation of manganese into the CZ lattice [47]. The observed changes in Raman spectra confirm the strong interaction between the manganese oxides and CZ support in the Mn-CZ(SD) sample associated with the partial incorporation of manganese into the oxide lattice, which is consistent with the SEM-EDS and H₂-TPR results. The significant decrease in the intensity of all bands in the spectra of biomorphic samples compared to the CTAB-templated ones probably resulted from the large contribution of optical adsorption by ash impurities contained in the biomorphic samples.

The Raman spectra of CZ(SD) and CZ(CTAB) also show a defect-induced O_v band located at ca. 610 cm⁻¹. This band is usually ascribed to oxygen vacancies associated with Frenkel-type defects and to the presence of cations of reduced cerium (Ce³⁺). The intensity of this line is higher for CZ(SD), which confirms the higher defectiveness of this catalyst, most likely induced by promoting ash impurities.

In the Raman spectra of the Mn-modified catalysts, the disorder-associated O_v band is overlapped with a relatively intense line at 660 cm⁻¹ that can be attributed to the A_1g symmetric stretching vibrations of Mn−O bonds in the octahedral MnO₆ units of the Mn₃O₄ spinel hausmannite structure [11, 49]. This fact suggests the simultaneous existence of Mn²⁺ and Mn³⁺ species and confirms the H₂-TPR and XPS results.

The weak band at about 1140–1180 cm⁻¹ in the all Raman spectra can be assigned to 2LO overtone band of the longitudinal optical (LO) mode at about at 595 cm⁻¹ (Raman inactive), which is a characteristic for the multi-phonon relaxation by the resonance Raman effect [50].

No lines characteristic of carbon materials were found in the Raman spectra. Consequently, carbon found in the samples by XPS method is in the composition of functional groups on the surface; in the case of SEM-EDS, it is present also in the material for fixing the samples.

The Raman spectroscopy data mostly agree with the results of other physical-chemical characterization techniques. However, further experimental studies are required to confirm the incorporation of manganese ions into the CZ lattice.

EPR

The EPR spectroscopy was applied to identify the coordination environment of Mn species incorporated into mixed oxides systems (Fig. 7).

Figure 7:

EPR spectra of catalysts.

The EPR spectrum of Mn-CZ(CTAB) exhibits intense broad resonance line at about g = 2.63 without hyperfine structure. This signal arises from the convolution of the spectra with different resonant fields due to the distribution of local effective field and can be associated with the presence of strongly self-interacting Mn²⁺ ions [51]. This observation confirms that a large portion of manganese ions is not surrounded by cerium ions and separate phases of the manganese oxides exist in the sample. The broad low intense EPR spectrum of CZ (CTAB) indicates the presence of some paramagnetic impurities [11].

The EPR spectrum of Mn-CZ(SD) shows a superposition of three spectra: broad line at g = 2.63 and two spectra having hyperfine structure with six lines centered at g = 2.00 and g = 1.97. Hyperfine structure is caused by interaction between electron spin and nuclear spin of manganese ion. First spectrum with g = 2.00 can be associated with the presence of isolated Mn²⁺ ions located somewhere in the solid solutions, most likely in the defect sites with a cubic symmetry in the framework of Ce⁴⁺ ions [52]. Second spectrum centered at g = 1.97 might be interpreted as the spectrum of Mn⁴⁺ [7, 53]. The intensity of the Mn⁴⁺ spectrum is not significant in comparison with the spectrum at g = 2.00 and especially with the broad line at g = 2.63. In this case, an EPR technique which has high sensitivity provides additional information on the catalyst structure which is not obtained by XPS. The spectrum of the CZ (SD) also shows a weak EPR signal because of the small admixture of manganese inherited from the sawdust template. Due to the high sensitivity, the EPR method allows detecting extremely low manganese concentrations in sawdust, which was not possible by other methods. On the other hand, Mn³⁺ ions could not be observed in our EPR experiment because these ions contribute to the spectra at high frequencies and only at very low temperatures [54].

Thus, EPR data reliably demonstrate the presence of Mn²⁺ in both modified samples and small amount of Mn⁴⁺ in the Mn-CZ(SD), and incorporation of these ions into the crystal lattice of the double oxide in this catalyst. EPR also confirms the above statement that Mn-CZ(SD) and Mn-CZ(CTAB) differ in the degree of interaction between MnO_x and CZ and the distribution of manganese oxide species on the surface. Obviously, these factors will affect the catalytic properties of the catalysts.

Summary of catalysts characterization

Figure 8 shows a simplified scheme of the chemical and phase composition and structure of the synthesized templated oxide catalysts.

Figure 8:

Scheme of surface and bulk structure of catalysts.

The double oxides CZ(SD) and CZ(CTAB) have the similar A_s values. However, CZ(SD) demonstrates higher defectiveness of the structure and improved oxygen mobility, provided by the presence of ash admixture and possibly the inhomogeneous nature of the SD template. Lower mobility of oxygen in CZ(CTAB) and the presence of bromine inherited from the template can deteriorate its catalytic efficiency in CO oxidation.

The H₂-TPR, XPS and EPR showed the presence of both Mn³⁺ and Mn²⁺ in the Mn-modified catalysts, which can improve the catalytic properties due to the presence of an additional redox pair. The presence of insignificant amount of Mn⁴⁺ state was detected by EPR; it finds an indirect confirmation in the XPS data.

Different structure of the templated CZ oxides led to a different manganese distribution during their impregnation with manganese acetate. Thus, manganese ions are partially incorporated into the surface layer of CZ crystals of Mn-CZ(SD), as it was demonstrated by EPR, while the other part of Mn is homogeneously distributed on the walls of macro- and mesopores in the CZ oxide. In contrast, MnO_x in Mn-CZ(CTAB) is distributed non-homogeneously, mostly on the external surface of the CZ(CTAB) support blocking the pore openings. Thus, the degree of interaction between MnO_x and CZ is much higher in Mn-CZ(SD), providing the additional improvement in oxygen mobility. In addition, the specific surface area of Mn-CZ(SD) is higher than that of Mn-CZ(CTAB).

Apparently, the competition between redox properties of MnO_x modifier located on the surface, and the CZ support itself, which can be activated by the presence of ash components and Mn ions in the crystal lattice, will determine the catalyst efficiency in CO oxidation. As for the soot oxidation, the most important catalyst characteristics are the specific surface area and surface roughness.

Total CO oxidation

The temperature dependence of the CO conversion to CO₂ as a function of the reaction temperature for all the catalysts is presented in Fig. 9a. All catalysts were active in CO oxidation but in different temperature ranges. For comparison, the temperatures of 50 % CO conversion (T₅₀) for all tested catalysts are shown in Table 1.

Figure 9:

CO conversion vs. reaction temperature plots (a), and normalized conversion curves for uncatalyzed and catalyzed soot oxidation (b).

CZ(SD) demonstrated higher CO conversion (Т₅₀ = 220 °C) than CZ(CTAB) (Т₅₀ = 295 °C) in the whole studied temperature range. The possible reason for this fact is the increased surface imperfection and improved oxygen mobility of the former catalysts confirmed by H₂-TPR, XPS and EPR methods. The increase in the number of structure defects in CZ(SD) can be explained by the presence of ash impurities [10].

The modification of both CZ types with MnO_x significantly improved the catalyst efficiency. Mn-modified samples showed the similar efficiency in CO oxidation. In the case of the biomorphic system, the improvement in catalytic performance is primarily caused by the increase in oxygen storage capacity due to incorporation of manganese ions into the surface and subsurface layers of CZ crystal lattice. The introduction of dopant ions into the fluorite structure creates oxygen vacancies and improves oxygen mobility in the ceria-based supports [4, 6, 7, 33], and the arrangement of such vacancies in the surface layer is very favorable for heterogeneous catalysis. This advantage of the doped catalysts is especially evident at high temperatures, when the oxygen mobility is further increased. That is why Mn-CZ(SD) provided slightly lower CO conversion compared to Mn-CZ(CTAB) up to 150 °C, while at high temperatures it showed the superior catalytic efficiency. On the other hand, the excellent catalytic properties of the Mn-CZ(CTAB) can be explained by intrinsic activity of the manganese oxides particles, distributed mostly on the surface, or sealed inside CZ particles. These particles can participate in the reaction through transitions between different oxidation states of manganese [11].

Soot oxidation

The conversions of soot during combustion in uncatalyzed and catalyzed tests are shown in Fig. 9b. Burning of soot particles leads to a weight loss and thermal effects in the DSC curve. All the catalysts shifted the TG curves of soot oxidation to low temperatures. As a result, a complete oxidation was observed in the temperature range from ca. 220 to 800 °C, whereas without catalyst only 50 % soot was converted at 800 °C. The temperatures of 50 and 90 % (T₅₀ and T₉₀) soot conversions were used to compare the catalyst activities (Table 1). At the maximum rate of weight loss, the DSC curves exhibit distinct peaks of heat release associated with the combustion of soot. According to the MS data (not presented), CO₂ was detected as the only carbon containing product for all catalysts in the whole studied temperature range.

Considering the small amount of the soot-catalyst mixture used in the TGA/DSC experiments, the values of characteristic temperatures (T₅₀ and T₉₀) were confirmed in scaled experiments using 7-fold increased amount of the soot – catalyst mixture (an example is presented in Supplementary materials, Fig. S5). For this purpose, the mixture was annealed at T₅₀ for 2 h and weighted after cooling. Then the catalyst was annealed at T₉₀ and weighted again. This experiment confirms the reliability of the data obtained in the non-isothermal conditions.

Catalytic soot combustion is a topochemical heterogeneous process. That is why the catalyst – soot contact interface plays an important role in the oxidation activity of the catalysts [16]. The key factors affecting the catalyst activity in soot oxidation are the presence of surface active sites and the catalysts ability to act as an oxygen buffer, providing oxygen transfer from the bulk to the surface [55], [56]. These properties strongly depend on A_s and surface/bulk composition [55].

The CZ(CTAB) demonstrated the best soot oxidation activity: the maximum oxidation rate in the TG and DSC curves was observed at the lowest temperature (Т₅₀ = 373 °C and Т₉₀ = 427 °C). The CZ(SD) was less efficient (T₅₀ = 411 °C). Both samples have high A_s, but the CZ(CTAB) shows a broad distribution of mesopores and provides therefore the efficient contact between the catalyst surface and the soot particles. In addition, according to the H₂-TPR results, the reduction of this sample starts at lower temperature, and the peak at 445°С is more intense than in the H₂-TPR profile of CZ(SD), confirming the improved redox properties of the former catalysts. That is why CZ(SD) demonstrated the lower efficiency especially at relatively low temperatures in the range of 220–420 °C, at which oxygen mobility in this sample was not so high according to the H₂-TPR data.

In contrast to CO oxidation, the modification of CZ(CTAB) with manganese did not improve the catalyst efficiency in soot oxidation. Probably, this fact may be explained by the drastic decrease in the specific surface area of Mn-CZ(CTAB). On the other hand, modification of CZ(SD) with Mn slightly increased soot conversion especially in the low temperature region: the difference between the T₅₀ temperatures of Mn-CZ(SD) and CZ(SD) was 14 °C. The specific surface area of Mn-CZ(SD) was lower than of CZ(SD), but the difference is not so prominent, as for CTAB-templated counterparts. The improved catalytic performance of Mn-CZ(SD) (Т₅₀ = 385 °C and Т₉₀ = 513 °C) compared to CZ(SD) can be explained by the increased defectiveness of its structure. Interestingly, despite the significantly different A_s values, both modified samples showed the similar soot conversions, the same way as in the case of CO oxidation. The similar redox properties of both Mn-containing catalysts are evident from the TPR results. In addition, the presence of highly dispersed MnO_x particles or even tiny clusters containing manganese in different oxidation states (2+ and 3+) contributes mainly to the efficiency of Mn-CZ(CTAB).