BaFe1−xNixO3 Catalysts for NOx-Assisted Diesel Soot Oxidation

In this work, it is analyzed the effect of the partial substitution of Fe by Ni in a BaFeO3 perovskite to be used as the catalyst for NOx-assisted diesel soot oxidation. A series of BaFe1−xNixO3 (x = 0, 0.2, 0.4 and 0.8) catalysts have been synthesized by using the sol–gel method. The catalysts have been characterized by ICP-OES, XRD, XPS, O2-TPD, H2-TPR- and TEM. The catalytic activity for NO to NO2 oxidation and NOx-assisted diesel soot oxidation have been determined by Temperature Programmed Reaction experiments (NOx -TPR and Soot-NOx-TPR, respectively) and by isothermal reaction at 450 °C. Ni seems not to be inserted in the BaFeO3 perovskite and, instead of that, BaNiO3 perovskite and NiO are detected on the surface of the perovskite BaFeO3. XPS data reveal the coexistence of Fe(III) and Fe(IV) on the catalyst’s surface (being Fe(III) the main oxidation state) and the presence of oxygen vacancies. All catalysts are active for NO oxidation to NO2, showing BaFeO3 and BaFe0.6Ni0.4O3 the best catalytic performance. BaFe0.6Ni0.4O3 shows the highest proportion of nickel on surface and it combines the highest activity and stability for NOx-assisted diesel soot oxidation. Also, this catalyst presents the highest initial soot oxidation rate which minimizes the accumulation of unreacted soot during reaction.


Introduction
Internal combustion engines (ICEs), such as diesel and gasoline engines, are used for primary power generation in the industry, heavy transport, marine, and aviation sectors [1,2]. Diesel engines are preferred for these applications because their efficiency and durability are higher than that of gasoline engines [3,4]. However, diesel engines generate pollutants such as particulate matter (also called soot particles) and nitrogen oxides (NO x ), being both highly dangerous for human health and the environment [5][6][7]. Consequently, vehicle emission regulations are becoming more demanding, and the use of after-treatments systems is mandatory. Thus, a combination of diesel oxidation catalyst (DOC) and diesel particulate filter (DPF) is employed as diesel after-treatment system to meet emission regulations [8,9]. However, DPF is made up of a ceramic honeycomb monolith coated with noble metals (Pt, Pt/Pd or Pd/Rh) [8,9], being these PGMs scarce and expensive [10,11].
Mixed oxides with perovskite-like structure (ABO 3 ) present interesting properties to be used as catalysts for different reactions related to the control of pollutants generated by ICEs [12], and they are considered, among others, as a potential alternative to the use of noble metals for soot removal from diesel engines [13][14][15][16][17]. In a previous study, a series of BaFe 1−x Cu x O 3 catalysts (x = 0, 0.1, 0.3 and 0.4) were synthesized, characterized, and tested for soot oxidation under gasoline direct injection (GDI) exhaust conditions [16,18] and diesel exhaust conditions [16]. The BaFe 1−x Cu x O 3 perovskites catalyze both, the NO 2 -assisted diesel soot oxidation and, to a lesser extent, the soot oxidation during the "fuel cut" stage of GDI exhaust conditions (1% O 2 ). BFC0 is the most active catalyst as the activity seems to be mainly related with the amount of O 2 evolved during an O 2 -TPD (related to lattice oxygen mobility) and decreases with the copper content. On the other hand, it has been published that the addition of Ni to Co 3 O 4 induces a structural distortion, improving the density of oxygen vacancies and enhancing the lattice oxygen mobility [19]. Thus, considering this background, in this paper, the effect of the partial substitution of Fe by Ni in a BaFeO 3 perovskite (BaFe 1−x Ni x O 3 catalysts, with x = 0, 0.1, 0.3 and 0.4) for the catalysis of NO x -assisted diesel soot oxidation reaction has been explored.

Synthesis and Characterization of Catalysts
The BaFe 1−x Ni x O 3 (x = 0, 0.2, 0.4 and 0.8) catalysts were synthesized by sol-gel method using citric acid and EDTA as a chelating agent to facilitate the solution of metal precursors in the synthesis conditions [20,21]. Briefly, a 40 ml solution of EDTA (in a 1:2 molar Fe: EDTA ratio) at 60 °C was prepared. Ba(NO 3 ) 2 , Fe(NO 3 ) 3 ·9·H 2 O and Ni(NO 3 ) 2 ·6·H 2 O were added as metal precursors and, finally, citric acid was incorporated using the same molar ratio (1:2, Fe: citric acid). The pH of the solution is adjusted to 9 with an ammonia solution (30% wt). Subsequently, the temperature was increased to 80 °C to ensure the formation of a gel. The obtained gel was dried at 150 °C for 12 h and, as a final step, the solid was calcinated at 850 °C for 6 h.
For sample characterization, the following techniques were employed.
The nickel content was measured by ICP-OES on a Perkin-Elmer device model Optimal 4300 DV. For this analysis, the elements were extracted by the mineralization of the samples using a diluted aqua regia solution (HNO 3 : HCl, 1:3) and stirring at 60 °C for 1 h.
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 20 and 80° 2θ angles with a step rate of 0.4°/min and using Cu K α (0,15,418 nm) radiation in a Bruker D8-Advance device.
The surface chemistry was assessed 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 at 284.6 eV, and the BE and KE values were then determined with the peak-fit software of the spectrometer. The O L / (Ba + Fe + Ni), Ni/(Ba + Fe + Ni), 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 , Fe2p 3/2 , Ni 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) fitted with a Thermal Conductivity Detector (TCD) and using 30 mg of the sample which was heated at 10 °C/min from 25 to 1000 °C in 5% H 2 /Ar atmosphere (40 ml/min). The quantification of the H 2 consumption was carried out using a CuO reference sample.
Temperature Programmed Oxygen Desorption (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 the sample heated at 10 °C/min from room temperature to 950 °C under a 100 ml/min of helium atmosphere. The 18, 28, 32 and 44 m/z signals were followed for H 2 O, CO, O 2 and CO 2 (respectively) evolved during the experiments. The amount of evolved oxygen was estimated using a CuO reference sample.
TEM images were obtained in a 120 kv JEM-1400 PLUS transmission electron microscope from JEOL, available at the technical services of the University of Alicante, equipped with an ORIUS GATAN image acquisition camera. The equipment reaches a resolution of 0.38 nm between two points and 0.2 nm between lines.

Activity Tests
The activity for NO to NO 2 and NO x -assisted diesel soot oxidation was determined by Temperature Programmed Reaction in a quartz fixed-bed reactor, heated up from 25 to 800 °C (10 °C/min), under a gas flow mixture (500 ml/ min) containing 500 ppm NO x , 5% O 2 , balanced with N 2 . For NO to NO 2 oxidation experiments (NO x -TPR), 80 mg of catalyst was diluted with 320 mg SiC. Soot oxidation tests (Soot-NO x -TPR) were performed by mixing 80 mg of catalyst and 20 mg of Printex-U (the carbon black used as model soot) with a spatula to ensure loose contact, and the mixture was diluted with 300 mg of SiC. The most active catalysts were also tested under isothermal soot oxidation conditions at 450 °C for 180 min. The gas composition was monitored by specific NDIR-UV gas analysers for NO, NO 2 , CO, CO 2 and O 2 (Rosemount Analytical Model BINOS 1001, 1004 and 100, Emerson Electric Co., St. Louis, MO, USA). The NO x conversion and the NO 2 generation percentages were calculated using the following equations: (2) NO 2 generation(% ) = NO 2,out NO x,out × 100 where NO 2, out and NO x,out are the NO 2 and NO x (NO + NO 2 ) concentrations measured at the reactor exit, being NO x,in the NO + NO 2 concentrations measured at the reactor inlet. The soot conversion and CO 2 selectivity were determined as: where ∑ t 0 (CO 2 + CO) is the amount of CO 2 and CO evolved at a time t, while ∑ final 0 CO 2 and ∑ final 0 (CO 2 + CO) are the total amount of CO 2 and CO + CO 2 evolved during the test. Table 1 presents the nomenclature used for the catalysts, the actual nickel content of the catalysts, determined by ICP-OES, as well as the nominal content calculated from the stoichiometric formula.

Chemical and Morphological Properties
In general terms, the actual Ni contents are similar to the nominal ones, so, the synthesis method allows the incorporation of the required amount of this metal. On the other hand, BET data reveal that all BaFe x−1 Ni x O 3 catalysts present a very low surface area, as it is expected for mixed oxides with perovskite like structure [22].

Crystalline Structure
XRD diffractograms obtained for BaFe 1−x Ni x O 3 catalysts are shown in Fig. 1. The unique crystalline phase detected for the BaFeO 3 catalyst is the oxygen-deficient hexagonal perovskite-type structure (BaFeO 2.67 ) . Note that the incorporation of nickel significantly modifies the XRD patterns of BF-E. First, the catalysts with nickel present the oxygendeficient hexagonal perovskite structure BaNiO 2.36 so, it seems that nickel is not inserted into the BaFeO 3 perovskite structure and instead of that, the hexagonal BaNiO 3 perovskite is formed as main crystal phase. This founding can be justified based on the tolerance factor value (t) for perovskites series [12] (featured in Table 2) that have been calculated using the ionic radius of the components of perovskites (Ba +2 : 175 ppm, Fe 3 : 64 ppm, Ni 2+ :78 ppm and O 2− :140 ppm), which is closer to 1 for BaNiO 3 than for any of the BaFe 1−x Ni x O 3 perovskites.  However, note that the main peak corresponding to hexagonal BaNiO 2.36 is overlapped with the main peak of BaFeO 3 perovskite and this fact hinders the assignation of XRD peaks. On the other hand, BFN2E, BFN4E, and BFN8E also present NiO and BaCO 3 peaks showing the heterogeneity of samples after nickel addition. It is remarkable that, in the presence of Ni, the structure of perovskite does not change to cubic, as observed for BaFe 1−x Cu x O 3 series [16,18], thus, it seems to support that Ni is not inserted into the BaFeO 3 perovskite. Instead of that, Ni is forming a new BaNiO 3 perovskite because, as commented above, the tolerance factor (t) for nickel perovskite is closer to the ideal value (see Table 2).

Surface Properties
The XPS technique allows the study of the surface composition of the samples (up to 5 nm in depth) and the oxidation state of the elements [23]. The XPS spectra for the transition Ni2p 3/2 (Fig. 2a), Fe2p 3/2 (Fig. 2b) and O1s (Fig. 2c) for the BaFe 1−x Ni x O 3 catalysts are presented in Fig. 2. Figure 2a shows the XPS spectra for the Ni2p 3/2 transition. According to the literature [24], binding energy values between 854 and 856 eV indicate the presence of Ni(II) species on the surface. The presence of Ni(II) is confirmed by the satellite peaks detected at 860 and 864 eV. The main signal is deconvoluted in two components corresponding to: (i) Ni (II) in BaNiO 3 perovskite, at higher binding energy and (ii) Ni(II) in NiO particles on the catalyst surface, at  Table 3 compares the percentage of nickel on the surface calculated by XPS and the percentage of nickel determined by ICP-OES. In the BFN2-E and BFN8-E catalysts, the XPS value is much lower than that of ICP-OES, so the fraction of nickel on the surface is low. However, in the BFN4-E catalyst, the XPS value is more similar to the nickel content determined by ICP-OES than for the other samples, indicating that Ni is more homogeneously distributed. Therefore, Ni species are present both on the surface and inside the BaNiO 3 perovskite lattice. Note that the BFN4-E catalyst presents the highest percentage of surface nickel (around 90% vs. around 40-50% for the other two catalysts). Figure 2b shows the band deconvolution of the Fe2p 3/2 transition. The lower energy band at ca. 710 eV indicates the presence of Fe(III), which is confirmed by the satellite peak of this oxidation state at 718 eV. The band with the highest binding energy, detected between 712 and 713 eV, indicates the presence of Fe(IV) [18,25]. Table 4 shows the Fe(IV)/ Fe(III) surface ratios, calculated from the corresponding peak areas. Note that, for the raw perovskite BaFeO 3 , Fe should be present as Fe(IV), according to stoichiometry to compensate the positive and negative charges of ions in the perovskite structure. However, as both Fe(III) and Fe(IV) coexist on the catalyst surface, a positive charge imbalance is generated which has to be compensated by the formation of oxygen vacancies. The data indicate that Fe(III) is present on the surface of all catalysts and, consequently, these perovskites should present oxygen vacancies. Figure 2c shows the spectra for the O1s transition for all catalysts. According to the literature [18,25], the O1s spectrum is usually deconvoluted into three bands corresponding to three oxygen species: (i) lattice oxygen (O L ) (around 530 eV) (ii) surface oxygen species (adsorbed oxygen, hydroxyl and carbonate groups) around 531 eV and (iii) water adsorbed on the surface (at around 533 eV). Note that these bands are detected for all catalysts, being the main band corresponding to the surface oxygen species. Additionally, Table 4 includes the O L /(Ba + Fe + Ni) ratio, calculated from the atomic percentages obtained by XPS and the area of O L peak. The corresponding nominal ratio (calculated from the stoichiometric composition of the BaFe 1−x Ni x O 3 catalysts) is 1.5, but all catalysts present an O L /Fe + Ba + Ni ratio lower and far to the nominal one, which indicates that the perovskite structure presents a very low proportion of lattice oxygen, that is, it is a very distorted structure. The use of EDTA as additional chelating agent during sol-gel process seems to promote the distortion of the structure, as BaFeO 3 obtained avoiding the use of EDTA [16,18] presents a higher O L /Fe + Ba + Ni ratio, i.e. , a less distorted structure. Thus, the main oxygen band is the O ads and OH − /CO 3 2− bands as the formed perovskite is non-stoichiometric (due to the presence of Fe(III) and Ni(II)). Therefore, there is a positive charge imbalance that is compensated by the presence of oxygen vacancies. In the presence of Ni, the amount of oxygen vacancies seems to decrease as O L /Fe + Ba + Ni ratio increases probably as the BaNiO 3 perovskite presents a less distorted structure.

Reducibility
The reducibility of the catalysts was determined by Temperature Programmed Reduction with H 2 (H 2 -TPR) following the procedure described in Sect. 2.1. H 2 -TPR profiles (Fig. 3) show several reduction peaks between 200 °C and 900 °C, which corresponds to the reduction of Ni and Fe. Hosseinpour et al. [26] analysed the Fe 2 O 3 reduction and observed three reduction peaks corresponding to the reductions of Fe 2 O 3 to Fe 3 O 4 (partial reduction of Fe(III) to Fe(II)) at 480 °C, and from Fe 3 O 4 to FeO at 615 °C and finally from Fe(II) to Fe (0) at 825 °C. On the other hand, Bai et al. [27] studied the reduction of nickel perovskites and observed several reduction peaks: (i) at 230-240 °C, attributed to the reduction of NiO species present on the perovskite surface, with a weak interaction Ni-perovskite; (ii) between 410 °C and 500 °C, appear the reduction peaks due to NiO x species on surface with an intense interaction with perovskite surface and subsurface, and (iii) peaks at temperatures above 510 °C, attributed to reduction of Ni(II) present on the bulk perovskite structure.
For BF-E sample, two well overlapped peaks between 200 °C and 480 °C are observed. According to the literature  [26], these peaks correspond to the partial reduction of Fe(IV) to Fe(III) and to the reduction of Fe(III) to Fe(II). BFN2-E presents a reduction peak between 230 °C and 240 °C which could be assigned to the reduction of NiO species with a weak interaction with the perovskite [27] as, for this sample, XRD peaks corresponding to NiO have been detected (Fig. 1). For BFN4-E, the reduction profile shows three peaks around 400 °C that should correspond not only to the Fe reduction but also to the reduction of NiO x species on surface and subsurface of perovskite, that is to the reduction of NiO x species with more intense interaction with perovskite [27]. Thus, Fe and Ni reduction peaks seem to be overlapped for this sample. Finally, two well defined peaks appear on the BFN8-E catalyst: (i) the peak at a lower temperature (approximately at 430 °C) corresponds to the reduction of nickel on the perovskite surface and also to the reduction of Fe(IV) to Fe (III) and; (ii) the highest temperature peak (about 480 °C) corresponds to the nickel of the BaNiO 3 perovskite and the reduction of Fe (III) to Fe (II). The absence of low-temperature Ni reduction peak for BFN4-E and BFN8-E samples is in agreement with the absence of XRD NiO peak in the diffractogram of these two samples (Fig. 1). catalysts. According to the literature [28], perovskite-type mixed oxides evolve O 2 at three temperatures that correspond to three types of oxygen: (i) the peak with a maximum below 400 °C is assigned to desorption of oxygen adsorbed on surface vacancies, called α-oxygen; (ii) the peaks with a maximum between 400 °C and 700 °C correspond to the desorption of oxygen coming from lattice defects such as dislocations, which is called α'-oxygen and, finally and; (iii) the peaks with a maximum above 700 °C correspond to the desorption of lattice oxygen, denoted as β-oxygen, which is related with the reducibility of the B cation perovskite and informs about the mobility of oxygen through the lattice.

O 2 Desorption
In the O 2 -TPD profiles, the observed peaks correspond to the desorption of α-oxygen, α'-and β-oxygen are detected. BF-E catalyst presents two O 2 desorption peaks: (i) one between 200 °C and 300 °C, which corresponds to the desorption of oxygen adsorbed on surface vacancies (α-O 2 ) and (ii) another more intense and defined peak with a maximum over 720 °C, which corresponds to the desorption of oxygen from the lattice (β-O 2 ). These results confirm the presence of oxygen vacancies on BaFeO 3 perovskite (the main crystal phase in BF-E), that have been created to compensate the positive charge imbalance due to the presence of Fe (III). BFN2-E and BFN4-E present also a peak, between 200 °C and 300 °C, which is more intense than the observed for BF-E, especially for BFN4-E. This means that BFN4-E presents a higher oxygen mobility at low temperatures which is expected to help the soot oxidation. This is because when Fe (III) is partially substituted by Ni(II), the positive charge imbalance increases, so, more oxygen vacancies are generated to achieve the electroneutrality. Note that for these samples, the high temperature peak (corresponding to the desorption of lattice oxygen) appears at higher temperatures because it also includes, the desorption of lattice oxygen coming from the BaNiO 3 perovskite. This shift suggests that the desorption of oxygen coming from BaNiO 3 perovskite is more difficult than the coming from BaFeO 3 perovskite. Finally, BFN8-E catalyst presents not only a high temperature peak similar to BFN4-E, (which corresponds to the desorption of lattice oxygen from both perovskites) but also three peaks assigned to the desorption of oxygen chemisorbed in the lattice defects of hexagonal BaNiO 2.67 .

NO to NO 2 Oxidation (NO x -TPR)
The activity of catalysts for the oxidation of NO to NO 2 was studied by Temperature Programmed Reaction (NO x -TPR) experiments carried out using a gas mixture which simulates the composition of the exhaust gases of a diesel vehicle (500 ppm NO x and 5% O 2 in N 2 ). Figure 5a shows the NO x conversion profiles as a function of temperature. Considering Eq.
(1), a positive conversion value indicates that the amount of NO x detected at the reactor outlet is lower than that in the inlet, i.e., the NO x is being adsorbed on the catalyst. On the contrary, a negative value indicates that NO x is being desorbed. All catalysts show a certain NO x adsorption capacity, which decreases in the presence of Ni. Note that, as observed for other perovskite-based catalysts [15][16][17][18][19], the peak corresponding to NO x desorbed is larger than the peak corresponding to NO x adsorbed due to the amount of NO x adsorbed at room temperature before starting the NO x -TPR experiment which is not registered.
In Fig. 5b the NO 2 generation profiles as a function of temperature during NO x -TPR experiments are shown. The percentage of NO 2 generation (Eq. (2)) represents the fraction of NO 2 at the outlet of the reactor. However, these profiles cannot indicate the real value of NO 2 generated because, as indicated by Fig. 5a, a fraction of the NO 2 is being adsorbed on the catalyst surface, so, it cannot be detected. Note that NO 2 generation profiles indicate that all catalysts are active for the oxidation of NO to NO 2 at temperatures below 450 °C, but the oxidation capacity is not directly related to the nickel content as, except for the BFN4-E catalyst, nickel-containing catalysts have a lower activity than the nickel-free catalyst (BF-E). Thus, BF-E and BFN4-E catalysts are the catalysts presenting the highest activity to generate NO 2 with a maximum of around 38% at 400 °C. These two samples present also higher NO x adsorption capacity, so, their activity for NO to NO 2 oxidation should be much higher. In fact, BFN4-E catalyst is more active for NO x adsorption than BFN2-E and BFN8-E, probably because it presents the highest proportion of surface nickel detected for this sample and the highest oxygen mobility at low temperatures (see Table 3 and Fig. 4).

Soot Oxidation (Soot-NO x -TPR)
Soot oxidation tests were carried out in Temperature Programmed Reaction (Soot-NO x -TPR) conditions using the same gas mixture described in Sect. 2.2 and Printex-U in loose contact, as model soot. As BF-E and BFN4-E are the most active catalysts for NO 2 generation (see Fig. 5) and, considering that NO 2 is much more active than O 2 for soot oxidation [15][16][17][18][19], these two catalysts have been selected to carry out several consecutive cycles of Soot-NO x -TPR, in order to determine their stability in reaction conditions. Additionally, they have been used in isothermal soot oxidation reaction at 450 °C. Figure 6 shows the soot conversion profiles (Soot-NO x -TPR) for BF-E and BFN4-E catalysts. Note that, at temperatures below 600 °C, which is the interval of interest for the removal of soot from diesel engines [29], the two catalysts present conversion values higher than the observed in the absence of catalysts (blank in Fig. 6), therefore, both catalysts are active confirming that the NO 2 generated is very effective for soot oxidation [15][16][17][18][19].
From these profiles, the temperature at which 50% conversion is reached (T50%), was determined and the data is included in Table 5, where also the selectivity to CO 2 (calculated by Eq. (4) based on the amount of CO and CO 2 coming from soot oxidation) is shown.
As expected, the two catalysts show a high selectivity to CO 2 [30,31] and a lower T50% value than in the absence of catalyst (blank). To obtain more information about the performance of catalysts, four consecutive soot oxidation cycles were carried out, and Fig. 7 compares the values of T50% (°C). BF-E presents the highest deactivation during four cycles since the T50% (°C) continuously increases. BFN4-E shows a similar deactivation but only after the first cycle, being T50% values almost constant for the following cycles. Thus, it seems that the partial substitution of Fe by Ni in this catalyst enhances the catalytic stability [32].
BFN4-E shows the most stable performance during soot oxidation reaction because, as revealed by the soot conversion profiles at 450 °C features in Fig. 8, this catalyst presents a higher soot oxidation rate (25 µmol/min) than the BaFeO 3 catalyst (11 µmol/min), minimizing the accumulation of soot during cycles and, thus, the deactivation. These  soot deposits remain at the end of the experiment due to the high thermal stability of the carbon black used as model soot [33]. It is remarkable that, as revealed by XPS results above discussed ( Fig. 2 and Table 4), BFN4-E catalyst presents the highest proportion of surface Ni(II) species that are able to increase the soot oxidation rate and, hence, helps to avoid the catalyst deactivation.
In order to confirm that the deactivation of BF-E is due to the accumulation of unreacted soot during reaction cycles, BF-E and BFN4-E catalysts used in isothermal soot oxidation at 450 °C were characterized by TEM. Thus, the images feature in Fig. 9 evidence that BF-E accumulates more soot than BFN4-E because of the lower soot oxidation rate.

Conclusions
In this work, a series of BaFe 1−x Ni x O 3 (x = 0, 0.2, 0.4 and 0.8) catalysts were synthesized to determine the effect of partial substitution of Fe by Ni on the NO to NO 2 oxidation and on the NO 2 -assisted diesel soot oxidation activity.
From the above discussed results, the following conclusions can be drawn:  Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.

References
1. Santos NDSA, Roso VR, Malaquias ACT, Baêta JGC (2021) Internal combustion engines and biofuels: Examining why this robust combination should not be ignored for future sustainable