Tailoring the Composition of BaxBO3 (B = Fe, Mn) Mixed Oxides as CO or Soot Oxidation Catalysts in Simulated GDI Engine Exhaust Conditions

Mixed oxides with perovskite-type structure (ABO3) are promising catalysts for atmospheric pollution control due to their interesting and tunable physicochemical properties. In this work, two series of BaxMnO3 and BaxFeO3 (x = 1 and 0.7) catalysts were synthesized using the sol–gel method adapted to aqueous medium. The samples were characterized by μ-XRF, XRD, FT-IR, XPS, H2-TPR, and O2-TPD. The catalytic activity for CO and GDI soot oxidation was determined by temperature-programmed reaction experiments (CO-TPR and soot-TPR, respectively). The results reveal that a decrease in the Ba content improved the catalytic performance of both catalysts, as B0.7M-E is more active than BM-E for CO oxidation, and B0.7F-E presents higher activity than BF for soot conversion in simulated GDI engine exhaust conditions. Manganese-based perovskites (BM-E and B0.7M-E) achieve better catalytic performance than iron-based perovskite (BF) for CO oxidation reaction due to the higher generation of actives sites.


Introduction
Currently, limitation of climate change is one of the most challenging topics due to its dangerous global effects on biological and human systems [1]. Greenhouse gases (GHGs), such as carbon dioxide, methane, and nitrogen oxides, are among the main species causing this problem. Since the industrial revolution, GHGs have been emitted into the atmosphere as the result of human activities, and their release has continuously increased [2]. The transport sector is the most responsible for GHG emissions, and predictions indicate that vehicles will contribute around 85% of the total GHGs emissions in 2040 if effective mitigation measures are not applied [3,4]. This scenario is pushing to the automotive industrial sector to use alternative systems to internal combustion engines, such as electric and fuelcell-based engines [5][6][7]. However, nowadays, electric and fuel-cell-based vehicles are not as available as internal-combustion-engine-based vehicles due to their high economical costs, the limited number of available charging points, and their long charge times [8,9]. For this reason, internal combustion systems cannot be totally eliminated in the coming years, and consequently, additional measures to reduce the amount of pollutants emitted by these engines must continue to be implemented.
In addition to the GHGs, carbon monoxide (CO) and soot are considered dangerous pollutants, as they also increase morbidity rates. These chemical species, which are highlighted as hazardous emissions generated by gasoline direct injection (GDI) engines, have become more relevant in recent last years [10,11]. As GDI-engine-based vehicles offer a high fuel economy and reduced CO 2 emissions (compared to other gasoline engines), they were the most widely purchased passenger cars in 2017 [12].
Perovskite-based catalysts show a good oxygen storage capacity, since O 2 can be adsorbed and activated on the oxygen vacancies [30]. Therefore, BaMnO 3 solids have been used for GDI soot oxidation under the two different atmospheres indicated above [36], as well as for CO oxidation in simulated GDI exhaust engine conditions [37]. Additionally, we recently realized that the use of under-stoichiometric Ba x MnO 3 perovskites improves the performance of raw BaMnO 3 for CO oxidation, showing that Ba 0.7 MnO 3 formulation achieves the best results [38]. Based on this statement and the background described above for CO and soot oxidation, in this study, two series of BaBO 3 perovskite-type catalysts (with Ba x MnO 3 and Ba x FeO 3 formulation of x = 1 and 0.7, respectively) were synthesized, characterized, and tested for CO and soot oxidation reactions as model reactions relevant for the potential use of these catalysts in the control of pollutants generated by GDI engines.

Chemical, Morphological, and Structural Characterization
X-Ray patterns for BM-E, B0.7M-E, BF, and B0.7F-E catalysts are shown in Figure 1a, and the corresponding BF-E and B0.7F samples are presented in Figure 1b. The corresponding XRD data are included in Table 1.
Molecules 2023, 28, x FOR PEER REVIEW 3 of 19 Perovskite-based catalysts show a good oxygen storage capacity, since O2 can be adsorbed and activated on the oxygen vacancies [30]. Therefore, BaMnO3 solids have been used for GDI soot oxidation under the two different atmospheres indicated above [36], as well as for CO oxidation in simulated GDI exhaust engine conditions [37]. Additionally, we recently realized that the use of under-stoichiometric BaxMnO3 perovskites improves the performance of raw BaMnO3 for CO oxidation, showing that Ba0.7MnO3 formulation achieves the best results [38]. Based on this statement and the background described above for CO and soot oxidation, in this study, two series of BaBO3 perovskite-type catalysts (with BaxMnO3 and BaxFeO3 formulation of x = 1 and 0.7, respectively) were synthesized, characterized, and tested for CO and soot oxidation reactions as model reactions relevant for the potential use of these catalysts in the control of pollutants generated by GDI engines.

Chemical, Morphological, and Structural Characterization
X-Ray patterns for BM-E, B0.7M-E, BF, and B0.7F-E catalysts are shown in Figure 1a, and the corresponding BF-E and B0.7F samples are presented in Figure 1b. The corresponding XRD data are included in Table 1.  For manganese-based samples, BM-E and B0.7M-E catalysts exhibit a hexagonal BaMnO3 perovskite structure (a = b ≠ c) as the main crystalline phase, and BaMn3O16 and Ba2Mn3O16 were detected as minority crystalline phases, appearing in very low amounts. The presence of these minority phases seems not to be related to the modification of the barium content, as a clear trend with barium content is not observed. Thus, the presence For manganese-based samples, BM-E and B0.7M-E catalysts exhibit a hexagonal BaMnO 3 perovskite structure (a = b = c) as the main crystalline phase, and BaMn 3 O 16 and Ba 2 Mn 3 O 16 were detected as minority crystalline phases, appearing in very low amounts. The presence of these minority phases seems not to be related to the modification of the barium content, as a clear trend with barium content is not observed. Thus, the presence of these minority phases may be related to [39] (i) the formation of small precipitates during the gelation step, (ii) the use of a low calcination temperature, or (iii) a very short calcination step. Note that the decrease in barium percentage does not generate significant structural modifications, since the cell parameters are not appreciably different (Table 1).
For iron-based samples, the BF diffractogram shows peaks corresponding to a hexagonal BaFeO 2.67 perovskite structure (a = b = c), while a peak associated with a BaFe 2 O 4 orthorhombic spinel phase (a = b = c) was observed for the B0.7F-E catalyst. M. Sun et al. [40] suggested that the BaFe 2 O 4 structure is formed when FeO 6 octahedrons of the perovskite lattice lose oxygen to form the FeO 4 tetrahedra present in the spinel lattice. Additionally, BaFe 0.72 Fe 0.28 O 2.64 mixed oxide (in which Fe(III) and Fe(IV) ions should coexist) was detected as a minority phase in both samples. In Figure 1b, it is observed that BF-E and B0.7F samples present a mixture of the three crystalline phases previously described for iron-based samples, that is, hexagonal BaFeO 2.67 perovskite, BaFe 2 O 4 orthorhombic spinel, and BaFe 0.72 Fe 0.28 O 2.64 ; however, a predominant phase is not detected. Therefore, in order to simplify, only the samples presenting a clear main crystalline phase, that is, BM-E, B0.7M-E, BF, and B0.7F-E, were selected to carry out the activity studies.
For the selected catalysts, the average crystal size of the main crystalline phases was calculated by employing the Williamson-Hall method [41]; according to the results presented in Table 1, crystal growth is favored for stoichiometric perovskites, that is, for BM-E and BF.
In Table 2, the BET surface area (determined by N 2 adsorption) and the chemical bulk composition (both experimental and nominal values calculated using BaMnO 3 , Ba 0.7 MnO 3 , BaFeO 2.67 , and BaFe 2 O 4 formulae) are included for selected samples. Note that (i) all catalysts present a low surface area, as perovskite-type oxides are solids with negligible porosity [16,[42][43][44][45], and (ii) the nominal and experimental bulk chemical compositions are similar, so the samples achieve the required weight percentages of Ba, Mn, Fe, or O. FT-IR spectra of catalysts are shown in Figure 2. For the manganese-based samples, the following signals are observed: (i) stretching vibration of the Mn-O bonds in the MnO 6 octahedrons appears between 500 and 786 cm −1 , and (ii) asymmetric stretching of carbonate ions at 858 and 1443 cm −1 indicates the presence of BaCO 3 [46][47][48]. For iron-based samples, in addition to carbonate ions that, according to the intensity of signals, are present in a higher amount than in manganese-based samples, the following signals are located: (i) peaks at wavenumbers slightly higher than 598 and 665 cm −1 that correspond to the stretching vibration of the Fe-O bonds of FeO 6 octahedrons, indicating that the Fe-O bond is stronger than the Mn-O bond [49]; (ii) a signal placed at 768 cm −1 corresponding to a BaFe 2 O 4 spinel, which is only detected for the B0.7F-E catalyst [49,50]; and (iii) a signal placed at 1060 cm −1 that corresponds to the stretching vibration of the C-OH bond [51][52][53]. The presence of carbon could be due to the remaining EDTA or citric acid, which was not completely removed during the calcination step. Note that as BaCO 3 was not detected by XRD, it should be an amorphous phase.

Surface Properties
XPS spectra were obtained to determine the surface chemical composition of the selected samples. The spectra of the Ba 3d 5/2 transition for both catalysts series shown in Figure 3 reveal that lattice Ba is the main species on the surface and that the amount of BaCO3 (which was previously observed by FT-IR) is more abundant on the surface of ironbased catalysts than on manganese-based samples. XPS spectra corresponding to the Mn 2p 3/2 transition are shown in Figure 4a. The registered signals are deconvoluted into two contributions associated with Mn(III) and Mn(IV), with Mn(III) located at lower binding energies than Mn(IV) [54]. The Mn(III)/Mn(IV) ratio values (presented in Table 3 and calculated by employing the corresponding area under the deconvoluted signals) reveal a higher amount of Mn(III) than of Mn(IV) on the catalyst surface and that the proportion of Mn(IV) increases as Ba content decreases. Thus, it seems that the oxidation of Mn(III) to Mn(IV) is the charge

Surface Properties
XPS spectra were obtained to determine the surface chemical composition of the selected samples. The spectra of the Ba 3d 5/2 transition for both catalysts series shown in Figure 3 reveal that lattice Ba is the main species on the surface and that the amount of BaCO 3 (which was previously observed by FT-IR) is more abundant on the surface of iron-based catalysts than on manganese-based samples.

Surface Properties
XPS spectra were obtained to determine the surface chemical composition of the selected samples. The spectra of the Ba 3d 5/2 transition for both catalysts series shown in Figure 3 reveal that lattice Ba is the main species on the surface and that the amount of BaCO3 (which was previously observed by FT-IR) is more abundant on the surface of ironbased catalysts than on manganese-based samples. XPS spectra corresponding to the Mn 2p 3/2 transition are shown in Figure 4a. The registered signals are deconvoluted into two contributions associated with Mn(III) and Mn(IV), with Mn(III) located at lower binding energies than Mn(IV) [54]. The Mn(III)/Mn(IV) ratio values (presented in Table 3 and calculated by employing the corresponding area under the deconvoluted signals) reveal a higher amount of Mn(III) than of Mn(IV) on the catalyst surface and that the proportion of Mn(IV) increases as Ba content decreases. Thus, it seems that the oxidation of Mn(III) to Mn(IV) is the charge  XPS spectra corresponding to the Mn 2p 3/2 transition are shown in Figure 4a. The registered signals are deconvoluted into two contributions associated with Mn(III) and Mn(IV), with Mn(III) located at lower binding energies than Mn(IV) [54]. The Mn(III)/Mn(IV) ratio values (presented in Table 3 and calculated by employing the corresponding area under the deconvoluted signals) reveal a higher amount of Mn(III) than of Mn(IV) on the catalyst surface and that the proportion of Mn(IV) increases as Ba content decreases. Thus, it seems that the oxidation of Mn(III) to Mn(IV) is the charge compensation mechanism working in this sample in order to counteract the decrease in the positive charge in barium-deficient perovskites [55]. compensation mechanism working in this sample in order to counteract the decrease in the positive charge in barium-deficient perovskites [55].
In the Fe 2p 3/2 transition spectra shown in Figure 4b, the following contributions can be identified: (i) an Fe(III) peak appearing at binding energies lower than those of Fe(IV) and (ii) Fe(IV) and Fe(III) satellite peaks located at approximately 713 eV and 718 eV, respectively [54,56,57]. The Fe(III)/Fe(IV) ratio values presented in Table 3 indicate that both Fe oxidation states coexist on the surface of the BF and B0.7F-E catalysts, with Fe(IV) as the main oxidation state for the two catalysts. However, as the Fe(III)/Fe(IV) ratio increases for B0.7F-E, the Fe(III) oxidation to Fe(IV) seems not to take place to compensate the positive charge deficiency as occurs for the B0.7M-E sample.
Finally, the deconvolution of the O 1 s signal ( Figure 5) shows three contributions: (i) a band located at binding energies around 529 eV, which is associated with the presence of lattice oxygen (OL) [54,58]; (ii) a band located around 531 eV that is related to the existence of defects with low oxygen coordination, that is, the oxygen vacancies formed on the surface (Odef) [59]; (iii) a band detected at 532 eV that indicates the presence of adsorbed oxygen (Oads), hydroxyl (-OH), and carbonate (CO3 2− ) groups on the catalyst surface [54,58,60]; and, finally, (iv) a band at approximately 533 eV corresponding to chemisorbed water (H2Ochem) [61,62]. For iron-based samples (Figure 4b), lattice oxygen peak shows two contributions corresponding to the lattice oxygen of the two main crystalline phases detected by XRD (BaFeO2.67 and BaFe2O4) at binding energies close to 531 eV and the oxygen of the BaFe0.72Fe0.28O2.64 minority phase (observed in XRD) at binding energies of approximately 529 eV. Table 3
In the Fe 2p 3/2 transition spectra shown in Figure 4b, the following contributions can be identified: (i) an Fe(III) peak appearing at binding energies lower than those of Fe(IV) and (ii) Fe(IV) and Fe(III) satellite peaks located at approximately 713 eV and 718 eV, respectively [54,56,57]. The Fe(III)/Fe(IV) ratio values presented in Table 3 indicate that both Fe oxidation states coexist on the surface of the BF and B0.7F-E catalysts, with Fe(IV) as the main oxidation state for the two catalysts. However, as the Fe(III)/Fe(IV) ratio increases for B0.7F-E, the Fe(III) oxidation to Fe(IV) seems not to take place to compensate the positive charge deficiency as occurs for the B0.7M-E sample.
Finally, the deconvolution of the O 1 s signal ( Figure 5) shows three contributions: (i) a band located at binding energies around 529 eV, which is associated with the presence of lattice oxygen (O L ) [54,58]; (ii) a band located around 531 eV that is related to the existence of defects with low oxygen coordination, that is, the oxygen vacancies formed on the surface (O def ) [59]; (iii) a band detected at 532 eV that indicates the presence of adsorbed oxygen (O ads ), hydroxyl (-OH), and carbonate (CO 3 2− ) groups on the catalyst surface [54,58,60]; and, finally, (iv) a band at approximately 533 eV corresponding to chemisorbed water (H 2 O chem ) [61,62]. For iron-based samples (Figure 4b), lattice oxygen peak shows two contributions corresponding to the lattice oxygen of the two main crystalline phases detected by XRD (BaFeO 2.67 and BaFe 2 O 4 ) at binding energies close to 531 eV and the oxygen of the BaFe 0.72 Fe 0.28 O 2.64 minority phase (observed in XRD) at binding energies of approximately 529 eV. Table 3 shows the experimental values of the O L /(Ba + Mn/Fe) ratio (estimated using the areas under the corresponding deconvolutions) and the nominal values (calculated by employing the catalysts formulations of ABO 3 , A 0.7 BO 3 , or AB 2 O 4 ). Thus, as the XPS ratios of BM-E, B0.7M-E, and BF are lower than the nominal ratios, the presence of oxygen vacancies on the surface of catalysts is featured. For manganese-based samples, the amount of oxygen vacancies increases as Ba content decreases, confirming the presence of a band corresponding to the oxygen defects (O def , located at around 531 eV) in B0.7M-E. Thus, it seems that the formation of these oxygen defects also enables counteracting the positive charge defect caused by the decrease in Ba content [63]. On the contrary, the formation of oxygen defects in the B0.7F-E sample is not promoted. Consequently, the change of structure from perovskite (in BF) to spinel (in B0.7F-E) seems to be the unique effect of Ba deficiency in iron-based catalysts.
the XPS ratios of BM-E, B0.7M-E, and BF are lower than the nominal ratios, the presence of oxygen vacancies on the surface of catalysts is featured. For manganese-based samples, the amount of oxygen vacancies increases as Ba content decreases, confirming the presence of a band corresponding to the oxygen defects (Odef, located at around 531 eV) in B0.7M-E. Thus, it seems that the formation of these oxygen defects also enables counteracting the positive charge defect caused by the decrease in Ba content [63]. On the contrary, the formation of oxygen defects in the B0.7F-E sample is not promoted. Consequently, the change of structure from perovskite (in BF) to spinel (in B0.7F-E) seems to be the unique effect of Ba deficiency in iron-based catalysts.

Redox Properties
Reducibility and redox properties of the catalysts were analyzed using temperatureprogrammed reduction with H2 (H2-TPR), which represents the consumption profiles shown in Figure 6a.

Redox Properties
Reducibility and redox properties of the catalysts were analyzed using temperatureprogrammed reduction with H 2 (H 2 -TPR), which represents the consumption profiles shown in Figure 6a. the XPS ratios of BM-E, B0.7M-E, and BF are lower than the nominal ratios, the presence of oxygen vacancies on the surface of catalysts is featured. For manganese-based samples, the amount of oxygen vacancies increases as Ba content decreases, confirming the presence of a band corresponding to the oxygen defects (Odef, located at around 531 eV) in B0.7M-E. Thus, it seems that the formation of these oxygen defects also enables counteracting the positive charge defect caused by the decrease in Ba content [63]. On the contrary, the formation of oxygen defects in the B0.7F-E sample is not promoted. Consequently, the change of structure from perovskite (in BF) to spinel (in B0.7F-E) seems to be the unique effect of Ba deficiency in iron-based catalysts.

Redox Properties
Reducibility and redox properties of the catalysts were analyzed using temperatureprogrammed reduction with H2 (H2-TPR), which represents the consumption profiles shown in Figure 6a. For barium manganese-based samples, three different peaks are identified [36][37][38][64][65][66][67][68]: (i) a low-temperature peak, which appears between 400 • C and 500 • C, assigned to the reduction of Mn(IV) and Mn(III) to Mn(II); (ii) a low-intensity signal at temperatures between 700 • C and 800 • C, which corresponds to the reduction of oxygen species on the catalysts surface; and (iii) another low-intensity band at around 900 • C due to the reduction of bulk Mn(III) to Mn(II). In the B0.7M-E sample, a small amount of hydrogen consumption is observed below 400 • C, which confirms the presence of Mn(IV) in the catalyst [69]. It is observed that the reduction of Mn(IV) and Mn(III) takes place at lower temperatures as the barium content decreases, probably due to a higher ionic mobility that improves the reduction process. This improved mobility has to be related to the presence of barium defects and oxygen vacancies in the structure [70].
On the other hand, for iron-based catalysts, the following peaks can be identified: (i) a peak at approximately 320 • C that corresponds to the reduction of Fe(IV) to Fe(III), (ii) a peak between 420 • C and 460 • C due to the partial reduction of Fe(III) to Fe(II), (iii) a third peak assigned to the reduction of the remaining Fe(III) between 650 • C and 740 • C, and finally, (iv) a peak at approximately 815 • C associated with the reduction of Fe(II) to Fe(0) [71]. It is observed that redox processes involving Fe cations take place at slightly lower temperatures in the BaFeO 2.67 perovskite-type structure than in the BaFe 2 O 4 spinel, which seems to be related to the higher amount of vacancies present in the oxygen-deficient perovskite than in the spinel structure, allowing a greater mobility and increased redox properties for the former.
In Figure 6b, the nominal and experimental consumptions of H 2 are compared for both series of catalysts. Nominal consumptions were calculated considering that manganese (either as Mn(III) or Mn(IV)) and iron (either as Fe(III) or Fe(IV)) are reduced to Mn(II) and Fe(0), respectively. Experimental values were calculated from the area under the H 2 -TPR peaks identified in the H 2 -TPR consumption profiles. For manganese-based catalysts, as the experimental consumption is closer to the nominal value corresponding to Mn(IV), it seems that Mn(III) and Mn(IV) species coexist in the bulk of catalysts, with Mn(IV) as the main oxidation state for the B0.7M-E catalyst. This is in agreement with the presence of a higher amount of Mn(IV) on the surface of this catalyst, as revealed by the Mn(III)/Mn(IV) XPS ratio ( Table 3). Note that the higher H 2 consumption determined for the B0.7M-E catalyst with respect to the BM-E catalyst has to be also related to the higher amount of Mn present in this sample (see the chemical bulk composition in Table 2). However, as the normalized value of H 2 consumption (i.e., the percentage of Mn shown in Table 2) is still higher for B0.7M-E (4.2) than for BM-E (2.9), it seems to be confirmed that this sample presents a higher reducibility due to the presence of a higher amount of more reducible Mn(IV). On the other hand, for the iron-based samples, as the experimental values are closer to nominal Fe(III) values, the amount of Fe(III) in the catalysts bulk is higher than that of Fe(IV), as was also deduced by XPS (see the Fe(III)/Fe(IV) ratio in Table 3) for surface composition.

O 2 Release during Temperature-Programmed Desorption in He (O 2 -TPD)
In the O 2 -TPD profiles featured in Figure 7a, three different peaks can be identified [72,73]: (i) between 150 • C and 350 • C, corresponding to the desorption of the oxygen adsorbed on surface vacancies (called α-O 2 ); (ii) from 350 • C to 700 • C, the oxygen evolved comes from the adsorbed on the lattice defects (designated as α -O 2 ); and (iii) above 700 • C, where the release of lattice oxygen (β-O 2 ) is detected. This β-oxygen is related to the reduction of Mn(IV)/Fe(IV) to Mn(III)/Fe(III) and the presence of oxygen vacancies in the bulk, which provides information about the oxygen mobility through perovskite lattice, which is a key factor determining the catalytic performance during oxidation reactions [74]. In Figure 7a, the total amount of desorbed oxygen and the weight loss percentage due to oxygen release for each catalyst are included.
Note that for manganese-based perovskites, the desorbed O 2 corresponds mainly to lattice oxygen (β-O 2 ), although the BM-E catalyst also presents a low α-O 2 emission and B0.7M-E shows a small O 2 desorption peak at around 600 • C. Furthermore, the amount of labile oxygen released is notably higher for the barium-deficient perovskite (B0.7M-E) than for BM-E because B0.7M-E presents a higher amount of Mn(IV) than BM-E. The high mobility of oxygen allows a greater reducibility of the catalysts, as shown by H 2 - TPR results (see Figure 6a) [73]. On the other hand, iron-based samples mainly evolve oxygen below 750 • C in the form of α-O 2 and α -O 2 species [75]. Note that in this case, the sample evolving the highest amount of oxygen is the BF catalyst, as expected for an under-stoichiometric perovskite, which is the main crystalline phase for this sample (see Figure 1a). Note that the structural change from perovskite to spinel that occurs due to the decrease in barium content in B0.7F-E does not increase the amount of oxygen vacancies nor the amount of Fe(IV). Note that for manganese-based perovskites, the desorbed O2 corresponds mainly to lattice oxygen (β-O2), although the BM-E catalyst also presents a low α-O2 emission and B0.7M-E shows a small O2 desorption peak at around 600 °C . Furthermore, the amount of labile oxygen released is notably higher for the barium-deficient perovskite (B0.7M-E) than for BM-E because B0.7M-E presents a higher amount of Mn(IV) than BM-E. The high mobility of oxygen allows a greater reducibility of the catalysts, as shown by H2-TPR results (see Figure 6a) [73]. On the other hand, iron-based samples mainly evolve oxygen below 750 °C in the form of α-O2 and α´-O2 species [75]. Note that in this case, the sample evolving the highest amount of oxygen is the BF catalyst, as expected for an under-stoichiometric perovskite, which is the main crystalline phase for this sample (see Figure 1a). Note that the structural change from perovskite to spinel that occurs due to the decrease in barium content in B0.7F-E does not increase the amount of oxygen vacancies nor the amount of Fe(IV).   [76]. The weight loss of manganese-based samples due to O 2 release (Figure 7a) matches with the corresponding TGA profiles shown in Figure 7b. However, for iron-based catalysts, a higher total weight loss than that expected according to the evolved O 2 (data shown in Figure 7a) is detected due to the decomposition of carbonate groups present in a higher amount in iron-based samples than in manganese-based samples (see XPS and IR results).

Catalytic Activity
The effect of decreasing the Ba content on the catalytic performance of BM and BF perovskites for CO and soot oxidation in simulated GDI engine exhaust conditions was analyzed. Figure 8 features the CO conversion profiles for the two tested reaction atmospheres, and Table 4 presents the temperature at which 50% of CO conversion was reached (T 50% values). The CO conversion profiles reveal that all catalysts are active for CO oxidation under the tested conditions, as CO is not oxidized in the absence of a catalyst. However, all catalysts are less active than the 1% Pt/Al 2 O 3 used as reference catalyst. On the other hand, for a stoichiometric gaseous reactant mixture (1% CO, 1% O2), Figure 8b shows that the B0.7M-E sample is more active than the BF sample. Therefore, manganese-based mixed oxides with perovskite structures are more active than ironbased mixed oxides with perovskite structures (BF) for CO oxidation under the two tested conditions. This is because manganese-based perovskite presents a higher amount of active sites for CO oxidation than iron-based perovskite, as it presents not only a high fraction of Mn(IV) but also a higher amount of oxygen vacancies. Therefore, it is confirmed that the nature of the B cation in perovskite determines the catalytic performance in oxidation reactions. In this sense, it is well known that electronic configuration of the B cation in perovskite is a key factor for adsorption/activation of reactants (CO and O2 in this case). In fact, the occupancy of the eg orbital (which presents the ideal geometry for interacting with the molecular orbitals of CO and O2 molecules) establishes the adsorption/desorption For the low CO content gaseous mixture (0.1% CO, 1% O 2 ), Figure 8a indicates that manganese-based catalysts are more active than the iron-based catalysts, as Mn is more efficient than Fe in terms of promoting redox processes due to a greater redox versatility [77]. Note that for manganese-based samples, Ba deficiency improves CO conversion at low temperatures, since the B0.7M-E catalyst presents a higher amount of oxygen vacancies (that are active sites for the reaction), a higher amount of Mn(IV), and a higher reducibility than BM-E perovskite. On the contrary, almost no effect of Ba deficiency is detected for iron-based samples because a structural change from perovskite to spinel occurs in this sample instead of an increase in the amount of oxygen vacancies and/or in the amount of Fe(IV). In fact, B0.7F-E presents a higher T 50% value than BF (see Table 4), suggesting that a spinel structure, which is the main crystalline phase for B0.7F-E, is less active than a perovskite structure for the CO oxidation reaction because of its lower amount of oxygen vacancies and Fe(IV) with respect to BF with a perovskite structure (see XPS and H 2 -TPR data in Table 3 and Figures 5b and 6) [78]. On the other hand, for a stoichiometric gaseous reactant mixture (1% CO, 1% O 2 ), Figure 8b shows that the B0.7M-E sample is more active than the BF sample. Therefore, manganese-based mixed oxides with perovskite structures are more active than ironbased mixed oxides with perovskite structures (BF) for CO oxidation under the two tested conditions. This is because manganese-based perovskite presents a higher amount of active sites for CO oxidation than iron-based perovskite, as it presents not only a high fraction of Mn(IV) but also a higher amount of oxygen vacancies. Therefore, it is confirmed that the nature of the B cation in perovskite determines the catalytic performance in oxidation reactions. In this sense, it is well known that electronic configuration of the B cation in perovskite is a key factor for adsorption/activation of reactants (CO and O 2 in this case). In fact, the occupancy of the e g orbital (which presents the ideal geometry for interacting with the molecular orbitals of CO and O 2 molecules) establishes the adsorption/desorption energy of reactants [79], as the energy of adsorption increases as the number of electrons in the e g orbital decreases. Therefore, as Fe(III) has two e g electrons but Mn(III) has only one (and Mn(IV) presents an empty e g orbital), the CO adsorption energy is higher for manganese perovskites than for iron perovskites. Consequently, manganese allows a more efficient CO activation on the catalyst surface, promoting Langmuir-Hinshelwood (LH) and Eley-Rideal (ER) mechanisms. Additionally, as the amount of oxygen vacancies is higher for manganese perovskites than for iron perovskite, a higher number of active sites is present in the former samples. Note that the prevalence of LH or ER mechanisms mainly depends on the gas mixture composition [59].
Additionally, the data presented in Table 4 shows that the increase in the CO partial pressure has a different effect for manganese than for iron-based samples, i.e., the CO conversion decreases for the former but improves for the latter. The observed decrease in the catalytic activity for manganese-based samples indicates that the higher demand of active sites at higher reactant partial pressure only seems to be achieved at higher temperatures. On the other hand, the increase in the CO conversion at higher CO partial pressures observed for the BF catalyst reveals the participation of α-O 2 and α -O 2 present (detected during O 2 -TPD experiments, see Figure 7a) as active species. In fact, the amount of oxygen evolved by BF and B0.7F-E samples at temperatures below 700 • C (included in Figure 7a) is in agreement with the ∆T 50% values presented in Table 4. Finally, by comparing ∆T 50% for B0.7M-E and B0.7F-E catalysts with that observed for the 1% Pt/Al 2 O 3 reference sample, it seems that mixed oxide catalysts are more stable than Pt-based catalysts against fluctuations in the composition of the reactant mixture, so they are more resistant to changes in the amount of O 2 and CO in the exhaust evolved by GDI engines, which seems to be an advantage for a potential application.
The soot conversion profiles obtained during the soot oxidation experiments carried out under the simulated conditions close to those of GDI engine exhaust, i.e., an inert atmosphere and a 1% O 2 /He gas mixture, are featured in Figure 9a,b. T 25% , T 50% (temperatures required to achieve the 25% and the 50% of soot conversion, respectively) and the selectivity to CO 2 are featured in Table 5. The data confirm that the two types of catalysts are also able to increase the soot conversion in the two tested atmospheres, as soot oxidation takes place at temperatures lower than for an uncatalyzed reaction. However, the improvement in the soot conversion is lower than that observed for the CO oxidation reaction, being the increase in the selectivity to CO 2 more relevant than the decrease in T 25% and T 50% values. In 1% O2/He, the B0.7F-E sample presents the lowest T50% value, so it seems that the spinel structure of B0.7F-E catalysts presents a higher ability to adsorb and dissociate O2 from the gas phase than a perovskite structure [80].
In the absence of oxygen, as expected [43,81,82], all catalysts show lower soot conversions due to the lower amount of oxygen available (which only comes from the samples). Under these conditions, the BF sample shows the lowest T25%, as it evolves the highest amount of low and intermediate oxygen species (α-O2 and α´-O2; see Figure 7a) able to oxidize the soot. However, at temperatures higher than around 800 °C , the conversion in the presence of manganese notably increases, and, consequently, lower T50% values are shown for manganese-based samples than for iron-based samples. This behavior seems to be related to the higher emission of -O2 able to oxidize soot.
On the other hand, the selectivity to CO2 values shown in Table 5 reveal that the  In 1% O 2 /He, the B0.7F-E sample presents the lowest T 50% value, so it seems that the spinel structure of B0.7F-E catalysts presents a higher ability to adsorb and dissociate O 2 from the gas phase than a perovskite structure [80].
In the absence of oxygen, as expected [43,81,82], all catalysts show lower soot conversions due to the lower amount of oxygen available (which only comes from the samples). Under these conditions, the BF sample shows the lowest T 25% , as it evolves the highest amount of low and intermediate oxygen species (α-O 2 and α -O 2 ; see Figure 7a) able to oxidize the soot. However, at temperatures higher than around 800 • C, the conversion in the presence of manganese notably increases, and, consequently, lower T 50% values are shown for manganese-based samples than for iron-based samples. This behavior seems to be related to the higher emission of β-O 2 able to oxidize soot. On the other hand, the selectivity to CO 2 values shown in Table 5 reveal that the generation of CO 2 is highly dependent on the catalyst and on the atmospheric composition. As expected [43,80], the selectivity values in the presence of oxygen in the reaction atmosphere are higher than in inert atmosphere, as in the latter, less oxygen is available for the oxidation of soot. Note that in the presence of O 2, the increase in the selectivity to CO 2 is higher for manganese than for iron-based samples. This is in agreement with the higher ability of Mn with respect to Fe to catalyze the CO oxidation to CO 2 , as previously discussed. Finally, it is remarkable that in an inert atmosphere, the Fe based catalysts present a higher selectivity to CO 2 than Mn-based catalysts because the latter metal promotes soot oxidation at higher temperatures, at which CO is the main product of carbon (soot) oxidation [83].
Thus, it can be concluded that the role of Ba x BO 3 samples as catalysts in oxidation reactions depends on the reaction (CO or soot oxidation) and on the nature of the B cation, as Mn-based are samples more active for CO oxidation, with B0.7F-E showing a higher increase in soot conversion in the presence of oxygen, even though the catalytic effect is more significant for the CO oxidation reaction. Additionally, it has been observed that the decrease in the Ba content improves the catalytic performance of both catalysts, as B0.7M-E is more active than BM-E for the catalysis of CO oxidation, as B0.7F-E is a better catalyst than BF for soot conversion under simulated GDI engine exhaust conditions. This shows that the composition of BaBO 3 mixed oxides can be tailored as a function of the oxidation reaction that has to be catalyzed.
Finally, considering that the four samples studied present a moderate ability as catalysts for CO and soot oxidation compared with other noble-metal-free catalysts [83][84][85], further optimization of perovskite composition is mandatory. In this sense, it is well known that copper plays an important role as an active phase during CO and soot oxidation reactions [84][85][86]. Consequently, the next step will be the analysis of the effect of the decrease in the barium content in copper-manganese perovskites [73].

Synthesis and Characterization of Catalysts
The Ba x MnO 3 and Ba x FeO 3 perovskites were synthesized by employing the sol-gel method adapted to aqueous medium [80,87]. First of all, a 40 mL solution (containing EDTA as a chelating agent in a 1:2 molar Mn/Fe:EDTA ratio) was prepared at 60 • C. Ba(NO 3 ) 2 was added as a Ba precursor; then, Mn(NO 3 ) 2 ·4 H 2 O or Fe(NO 3 ) 3 ·9 H 2 O was included as an Mn or Fe precursor, respectively. After that, citric acid was incorporated using a 1:2, Mn(Fe):citric acid molar ratio. Subsequently, the temperature was increased to 80 • C to assure gel formation. The described procedure was carried out keeping the pH at 9 using an ammonia solution (30% wt). Finally, the obtained gel was dried at 150 • C for 12 h, and as the last step, the solid was calcinated at 850 • C for 6 h. Additionally, in an attempt to decrease the number of crystalline phases present in the samples, Ba x FeO 3 catalysts were synthesized using the same method but avoiding the use of EDTA (named BF and B0.7F).
Thus, even though the synthesis method was the same, some experimental variables were slightly modified with respect to the procedure described above: (i) citric acid was dissolved in a 1:2 molar Mn (Fe):citric acid ratio before adding the metal precursors, (ii) barium acetate was used instead of barium nitrate, (iii) the gelation process was conducted at 65 • C, and (iv) the drying process was carried out at 90 • C for 48 h.
For sample characterization, the following techniques were employed. The chemical bulk composition was analyzed by µ-XRF on an Orbis EDAX analyzer. The textural properties were determined by N 2 adsorption at −196 • C using an Autosorb-6B instrument from Quantachrome (Anton Paar Austria GmbH). The samples were degassed at 250 • C for 4 h before the N 2 adsorption experiments.
The crystalline structure was obtained using X-ray diffraction (XRD). The X-ray patterns were recorded between 20º and 80 • 2θ angles with a step rate of 0.4 • /min and using Cu K α (0.15418 nm) radiation in a Bruker D8-Advance device. FT-IR spectra from 4000 to 500 cm −1 were also obtained using a BRUKER IFS 66/S apparatus with a spectral resolution of 4 cm −1 .
Surface chemistry was studied by X-ray photoelectron spectroscopy (XPS) using a K-Alpha photoelectron spectrometer by Thermo-Scientific with an Al K α (1486.7 eV) radiation source. To obtain XPS spectra, the pressure of the analysis chamber was maintained at 5 × 10 −10 mbar. The binding energy (BE) and kinetic energy (KE) scales were adjusted by setting the C 1 s transition to 284.6 eV, and the BE and KE values were determined with the peak-fit software of the spectrometer. The O L /(Ba + Mn/Fe), Mn(IV)/Mn(III), and Fe(IV)/Fe(III) XPS ratios were calculated using the area under the suggested deconvoluted bands of O 1 s, Mn 2p 3/2 , Fe 2p 3/2 , and Ba 3d 5/2 .
The reducibility of catalysts was determined by temperature-programmed reduction with H 2 (H 2 -TPR) in a Pulse Chemisorb 2705 (from Micromeritics) provided by a thermal conductivity detector (TCD) and using 30 mg of sample heated at 10 • C/min from 25 • C to 1000 • C in a 5% H 2 /Ar atmosphere (40 mL/min). Quantification of the H 2 consumption was carried out using a CuO reference sample. O 2 -TPD experiments were performed in a TG-MS (Q-600-TA and Thermostar from Balzers Instruments (Pfeiffer Vacuum GmbH, Germany), respectively) with 16 mg of sample heated at 10 • C/min from room temperature to 950 • C under a 100 mL/min helium atmosphere. Moreover, all the samples were preheated to 150 • C for 1 h for moisture removal. The 18, 28, 32, and 44 m/z signals were registered for the H 2 O, CO, O 2 , and CO 2 evolved during the experiments, respectively. The amount of evolved oxygen was estimated using a CuO reference sample, and the weight data recorded by the TG system were used to obtain the mass loss profiles.

Activity Tests
To determine the catalytic activity for CO oxidation, temperature-programmed reaction (CO-TPR) experiments were developed using two reactant mixtures composed of (i) 0.1% CO and 1% O 2 in He (low CO reactant mixture) and (ii) 1% CO and 1% O 2 in He (near-stoichiometric reactant mixture, which simulates the actual working conditions of GDI engines). For the experiments, 50 mg of catalysts and 100 mg of SiC were loaded into a U-shaped quartz reactor, and a 10 • C/min heating rate was used from room temperature to 500 • C. The mixture catalyst-SiC was preheated for 1 h at 600 • C in a 5% O 2 /He gas mixture in order to clean the surface of the catalysts. For quantification of reaction products, a HP6890 gas chromatograph equipped with a thermal conductivity detector and two packed columns (Porapack-Q and MolSieve-13X) was employed.
The CO conversion was determined using the following equation.
CO conversion (%) = ((CO in − CO out )/CO in )·100 The catalytic tests for the soot oxidation reaction under simulated GDI conditions were performed in the TG-MS system used for O 2 -TPD experiments. To develop these tests, 16 mg of a catalyst and soot mixture (soot:catalyst ratio of 1:8, using Printex-U as model soot in loose contact mode) was preheated for 1 h at 150 • C in a 1% O 2 /He gas mixture (100 mL/min); then, the temperature was increased at 10 • C/min until 900 • C (soot-TPR). Two different reactant atmospheres were employed: (i) 1% O 2 /He, which simulates "fuel cuts" GDI exhaust conditions, and (ii) 100% He, which reproduces regular stoichiometric GDI operations [43].

Conclusions
In this study, two series of Ba x MnO 3 and Ba x FeO 3 (x = 1 and 0.7) catalysts were synthetized, characterized, and tested for CO and soot oxidation reactions under simulated GDI exhaust conditions. Based on the results discussed above, the following conclusions can be exposed: (1) Manganese-based samples present a BaMnO 3 hexagonal perovskite structure, as the decrease in the amount of Ba does not significantly modify the crystalline structure. For iron-based samples, the decrease in the Ba content promotes the transition from a BaFeO 2.67 hexagonal perovskite structure to a BaFe 2 O 4 spinel structure. (2) To compensate for the Ba deficiency in the manganese-based samples, the amount of Mn(IV) and the oxygen vacancies increases, allowing a higher reducibility and oxygen mobility. In iron-based samples, the Ba deficiency only causes a change in the structure from perovskite to spinel. (3) Manganese-based perovskites (BM-E and B0.7M-E) show a better catalytic performance than iron-based perovskite (BF) for CO oxidation reactions due to the higher generation of actives sites. (4) The decrease in the Ba content improves the catalytic performance of both catalysts, as B0.7M-E is more active than BM-E for CO oxidation, and B0.7F-E presents a higher activity for soot conversion than BF under simulated GDI engine exhaust conditions. This shows that the composition of BaBO 3 mixed oxides can be tailored as function of the oxidation reaction to be catalyzed.  Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.