BaTi_0.8B_0.2O₃ (B = Mn, Fe, Co, Cu) LNT Catalysts: Effect of Partial Ti Substitution on NOx Storage Capacity

Abstract

The effect of partial Ti substitution by Mn, Fe, Co, or Cu on the NOx storage capacity (NSC) of a BaTi_0.8B_0.2O₃ lean NOx trap (LNT) catalyst has been analyzed. The BaTi_0.8B_0.2O₃ catalysts were prepared using the Pechini’s sol–gel method for aqueous media. The characterization of the catalysts (BET, ICP-OES, XRD and XPS) reveals that: i) the partial substitution of Ti by Mn, Co, or Fe changes the perovskite structure from tetragonal to cubic, whilst Cu distorts the raw tetragonal structure and promotes the segregation of Ba₂TiO₄ (which is an active phase for NOx storage) as a minority phase and ii) the amount of oxygen vacancies increases after partial Ti substitution, with the BaTi_0.8Cu_0.2O₃ catalyst featuring the largest amount. The BaTi_0.8Cu_0.2O₃ catalyst shows the highest NSC at 400 °C, based on NOx storage cyclic tests, which is within the range of highly active noble metal-based catalysts.

1. Introduction

Diesel engines are a type of lean burn engine, operating at upper stoichiometric air-to-fuel ratios (A/F>14.7/1), which grew in popularity at the end of 20^th century as they offered higher fuel efficiency and less CO₂ emissions respect to gasoline engines [1]. However, these engines show a highly relevant drawback since they generate large amounts of NOx and soot [2,3]. In order to minimize the level of these pollutants, more stringent standards were progressively established all over the world. Nowadays, it is accepted that the current EURO VI standard regarding NOx emissions is not met by just improving the quality of the fuel, by modifying the engine, or by using three-way-catalysts (TWCs). Consequently, alternative catalytic strategies are mandatory in order to avoid the disappearance of vehicles fitted with a diesel engine [4,5].

Two methodologies have been proposed to control NOx emission in lean burn engines: selective catalytic reduction (SCR) and NOx storage and reduction (NSR), also called lean NOx trapping (LNT). LNT technology involves the adsorption of NOx under lean conditions, followed by the periodic regeneration of the catalyst by reduction under rich conditions [6]. The conventional LNT catalysts (fitted in diesel cars) are composed of a platinum-group metal and an alkaline or alkaline-earth oxide (BaO or K₂O) supported on a high surface area material (Al₂O₃, TiO₂, …). It has been found that LNT technology matched with a TWC in a direct-injection spark ignition (DISI) engine can feature interesting results for NOx control emission [7].

Nevertheless, these conventional LNT catalysts present some drawbacks [8], with the high cost of noble metals (mainly Pt) being one of the most relevant. In fact, an interesting challenge has been highlighted in a recent EU report, regarding the need to develop alternatives to the use of critical raw materials such as precious metals [9]. In this line, the potential of perovskite base catalysts is being largely illustrated in the literature for environmental applications [10,11,12].

In previous studies [11,12], titanium was partially substituted by copper in the BaTiO₃ perovskite structure, showing the resulting BaTi_1−xCu_xO₃ perovskites a high activity for NOx storage, which was attributed to the presence of oxygen vacancies (created on the catalyst surface as a consequence of the copper incorporation into the structure) and to the segregation of some phases (mainly BaCO₃ and Ba₂TiO₄, but also CuO). It was also concluded that the BaTi_0.8Cu_0.2O₃ catalyst presents a NOx storage capacity (NSC) at 420 °C in the range of levels reported for noble metal-based catalysts (around 300 µmol/g) [13], and hence could be proposed as a potential component of high-temperature LNT systems for lean burn engines, such as gasoline direct injection engines. Moreover, in the literature, other metals such as Mn, Fe, or Co have been proposed as promising B cations in the perovskite used as catalysts for NOx and soot removal [14,15,16]. Thus, the aim of this paper is to determine the effect of Ti partial substitution by Mn, Fe, and Co in the NSC of the BaTi_0.8B_0.2O₃ LNT catalyst. The results will be analyzed with respect to the performance of the previously studied BaTi_0.8Cu_0.2O₃ catalyst [11,12].

2. Results and Discussion

2.1. Characterization of the Catalysts

Table 1. Nomenclature and basic characterization data.

Catalyst	Nomenclature	SBET (m2/g)	Bexp (wt%)/ Bnom (wt%)	B/Ba+Ti+B 1	Olattice/Ba+Ti+B 2
BaTi0.8Mn0.2O3	BTMnO_2	13	5.2/5.4	0.08	1.5
BaTi0.8Fe0.2O3	BTFeO_2	7	4.7/4.8	0.09	1.8
BaTi0.8Co0.2O3	BTCoO_2	5	4.9/4.9	0.13	1.4
BaTi0.8Cu0.2O3	BTCuO_2	12	4.9/5.0	0.07	1.4
BaTiO3	BTO_ref	9	--	--	2.0

¹ B/Ba+Ti+B nominal = 0.1, ² O_lattice/Ba+Ti+B nominal = 1.5.

presents the nomenclature and the basic characterization data of the catalysts: B metal content (measured by ICP-OES), BET surface area (obtained by applying the BET equation to N₂ adsorption data) and XPS. As it can be observed, the ICP-OES results reveal that all the metals added during the synthesis processes are present in the catalysts. Besides, the BET surface areas of the catalysts are low (as correspond to solids with negligible porosity, as mixed oxides with perovskite structure are [11]) and they range from 5 to 13 m²/g.

Concerning the XRD results, Figure 1a shows the XRD patterns of the catalysts that reveal a perovskite like structure (the diffraction peaks observed at 2θ: 22.3°; 31.4°; 38.8°; 45.2°; 51.0°; 56.1°; 65.8°; 74.9°; for (100), (110), (111), (200), (210), (211), (220), and (310) lattice planes, correspond to the standard JCPDS for tetragonal perovskite structure: 5-626 [17]) as the major crystalline phase for all the catalysts. Based on the splitting of peak around 51°, it seems that the perovskite structure is tetragonal for BTO_ref and BTCuO_2, but it changes to cubic for the other catalysts [11,12]. The magnification of the main peak of the diffractograms (31.5°), included in Figure 1b, clearly shows a shift to a lower angle value, respect to the BTO reference which is more evident for BTMnO_2, BTCoO_2, and BTFeO_2 (suggesting a modification of the perovskite structure) [11,12] than for BTCuO_2 (as the tetragonal structure is preserved for this catalyst), even though a decrease in the peak intensity is featured. Additionally, other minority phases are also identified by XRD, that is, mainly Ba₂TiO₄ and BaCO₃ (formed by the carbonation of segregated barium oxide during samples atmospheric exposure) but also CuO for BTCuO_2, and only BaCO₃ for BTMnO_2, BTFeO_2, and BTCoO_2 catalysts. As it has been previously reported [11,12], the existence of segregated phases on a metal substituted perovskite proves that the metal has been incorporated into the perovskite structure. Therefore, the shift of the main diffraction peaks ascribed to the perovskite structure and the segregation of minority phases identified in the XRD patterns indicate that Ti is successfully substituted by Mn, Fe, Co, and Cu in the perovskite framework. It is worth mentioning that: i) for BTCoO_2 catalysts, other cobalt phases (BaCoO₃ perovskite and Co₃O₄) are also identified as minority segregated phases in the XRD diffractogram, suggesting that cobalt has been introduced into the perovskite lattice in a lower extent than Cu, Fe, and Mn; and ii) Ba₂TiO₄, which has been suggested as an active phase for NOx storage [11,12], is only detected for BTCuO_2 catalyst.

To verify the structural modifications in the perovskites suggested by XRD patterns, Raman spectroscopy was used. According to literature [18,19], only the tetragonal structure of BaTiO₃ perovskite, which belongs to space group P4mm, presents first-order Raman-active modes with bands at, approximately, 180 cm⁻¹, 265 cm⁻¹, 305 cm⁻¹, 520 cm⁻¹, and 720 cm⁻¹, corresponding to irreducible representations ((A1(LO)), (A1(TO)), (B1), (A1, E(TO)), and (A1, E(LO)), respectively. On the one hand, the Raman spectra, shown in Figure 2, confirm that BTCuO_2 preserves the original tetragonal structure of the raw perovskite, as suggested by XRD, as it features the main bands previously indicated. In spite of this, BTCuO_2 spectrum shows broader peaks than BTO spectrum, pointing out that copper incorporation distorts the original tetragonal structure. On the other hand, BTFeO_2 and BTMnO_2 catalysts show an almost flat spectrum, ascribed to perovskite cubic structure, which does not show active modes in Raman spectroscopy. Finally, some Raman peaks are identified in the BTCoO_2 spectrum which are ascribed to the presence of minority phases such as BaCoO₃ and Co₃O₄, also identified by XRD. This result supports that a lower degree of cobalt is incorporated into the perovskite framework of the catalyst.

The different effects of the B cations, that partially substitute Ti on the perovskite structure, seem to be related with their ionic radius. Fe, Mn, and Cu (as M²⁺) have ionic radii larger than Ti⁴⁺ causing the distortion of the raw tetragonal perovskite structure. As Co²⁺ presents the most similar ionic radius to Ti⁴⁺, the formation of the stable BaCoO₃ perovskite is also allowed and, consequently, a lower fraction of cobalt is inserted into the BaTiO₃ perovskite framework to partially replace titanium.

XPS provides valuable information about the catalysts surface composition. All the XPS spectra and contributions assignment are featured in Figure A1 and

Table A1. Binding energy and assignment of the Cu2p, Co2s, Fe2p, and Mn2p transitions.

XPS Transition	Binding Energy (eV)	Assigned Species
Cu2p3/2	932.9	Cu(II) oxide surface species [25,26]
	934.8	Lattice Cu(II) [11]
	940.8	Cu(II) satellite [25]
	943.3	Cu(II) satellite [25]
Co2s	925.28	Co (II) oxide species [24]
	928.88	Co(III) oxide species (Co3O4)
Fe2p3/2	710.12	Fe(II) oxide species [24]
	712.09	Fe (III) oxide species [24]
	718.15	Fe (III) Satellite [25]
Mn2p3/2	641.46	Mn(III) oxide species (BaMn8O16) [27]
	642.90	Mn(IV) oxide species (BaMn8O16, BaMnO3-x) [27]
	644.98	Mn satellite [26,28]

, in the Appendix A. Table 1 shows the data related to metal (Mn, Fe, Co, or Cu) distribution presented as B/Ba+Ti+B (B= Mn, Fe, Co, Cu) ratio, whilst the data related to lattice oxygen, is shown as O_lattice/Ba+Ti+B (B= Mn, Fe, Co, Cu) ratio. It can be observed that for Mn, Fe, and Cu, the B/Ba+Ti+B XPS ratio is lower than the corresponding nominal value (0.1), which supports that these metals have been partially introduced into the perovskite structure. The BTCuO_2 catalyst presents the lowest value, so, the highest percentage of metal inside the perovskite lattice, whilst for Co, a B/Ba+Ti+B ratio higher than the nominal is found due to the presence of BaCoO₃ and Co₃O₄ segregated phases. The O_lattice/Ba+Ti+B XPS ratio (calculated from the area for O1s peak corresponding to lattice oxygen) for all catalysts is lower than the corresponding value for the BTO_ref perovskite (2.0), evidencing the creation of oxygen vacancies in the perovskite structure to compensate the imbalance in positive charge due to the partial substitution of Ti⁴⁺. Note that the BTCuO_2 catalyst presents the lowest O_lattice/Ba+Ti+B ratio and, consequently, the largest amount of surface oxygen vacancies. This result seems to be explained considering that BTCuO_2 catalyst presents a positive imbalance larger than the other catalysts due to the highest difference between the oxidation state of Ti⁴⁺ and the Cu⁺² (Fe and Mn appear mainly as Fe(II) and Mn(III) but Fe(III) and Mn(IV) have been also identified by XPS). Additionally, surface oxygen vacancies are also created because Ti⁺⁴ cannot achieve a higher oxidation state as other B cations (as Mn or Fe in BaMn _1−x Cu_xO₃ and BaFe_1-x Cu_xO₃ [20,21]) do. Finally, it is remarkable that BTCuO_2 catalyst preserves the original tetragonal structure (shown by XRD and Raman results), but with a high degree of distortion, which causes the presence of a larger amount of oxygen vacancies.

2.2. Catalytic Activity

For the analysis of the activity of the catalysts for NO to NO₂ oxidation and NOx adsorption/desorption, temperature programmed reaction (TPR-NOx) experiments were carried out. These experiments also allow the selection of the optimal temperature for isothermal NOx storage experiments that have been carried out in order to determine the NOx storage capacity (NSC, which is the amount of NOx stored (in µmol) per gram of catalyst). The results obtained as explained in the Materials and Methods section are presented in Figure 3.

Figure 3a features the NOx conversion profiles for all catalysts. It has been considered that positive values of NOx% conversion indicate that NOx adsorption is taking place, while negative values correspond to a NOx desorption process. According to this, at temperatures lower than 500 °C, approximately, the NOx conversion profiles represent NOx adsorption profiles and, at temperatures higher than 500 °C, these represent NOx desorption profiles. An analysis of the NOx conversion profiles reveals that BTCoO_2 but, mainly, BTCuO_2 catalysts show NOx adsorption/desorption activity, this performance being consistent with the presence of Ba₂TiO₄ segregated phase, which has been suggested as an active phase for NOx adsorption [11,12].

Before analyzing the NO₂ generation profiles shown in Figure 3b, it is worth mentioning that NO₂ is the main compound involved in NOx adsorption processes [8,11,12] and, for this reason, it has to be considered that the NO₂ registered by the analyzers is only the evolved NO₂, that is: i) below 500 °C, it is the fraction of NO₂ generated which is not stored; and ii) above 500 °C, it represents the NO₂ that is being desorbed. Therefore, the NO₂ generation shown in Figure 3b cannot be considered as a straight representation of the total NO₂ generated and, consequently, any conclusion regarding NO to NO₂ oxidation activity of the catalysts must be drawn from the combination of Figure 3a,b.

Thus, all the BaTi_0.8B_0.2O₃ perovskite catalysts increase the rate of NO₂ generation percentage at low temperature as the %NO₂ generated is higher than that shown by the BTO perovskite used as a reference (BTO_ref in Figure 3a,b). However, a deeper analysis of the data reveals some significant differences in the NO₂ profiles of the catalysts. Firstly, BTFeO_2 and BTMnO_2 catalysts feature Gaussian-shape NO₂ generation profiles with maxima at around 470 °C. Considering that almost any significant NOx adsorption/desorption activity is observed for these catalysts, it can be suggested that they are mainly active for NO to NO₂ oxidation (however, NOx adsorption capacity cannot be totally ruled out due to the intrinsic characteristics of TPR experiment). Secondly, although BTCoO_2 catalyst presents a similar type of NO₂ generation profile, it shows the highest NO oxidation activity, which seems to be related to the presence of Co₃O_4, as metal oxides are active for the NO to NO₂ oxidation reaction [11,12,14,15,16]. In addition, the low intensity NOx conversion peaks observed for BTCoO_2 catalyst in Figure 3a, indicates a low NOx adsorption activity that, according to literature [14,15,16], could be due to the presence of minority segregated phases and oxygen vacancies. Finally, a different NO₂ generation profile with two maxima and a minimum (at ca. 421, 507, and 441 °C, respectively) are clearly identified for BTCuO_2 catalyst. It is worth indicating that the temperature of the minimum NO₂ generation perfectly matches with the temperature of the maximum NOx conversion observed for this catalyst in Figure 3a. This result points out that, at this temperature, BTCuO_2 shows higher NOx adsorption rate than NO oxidation rate as Ba₂TiO₄ phase, which is active for NOx adsorption [11,12], has been identified.

As a summary, TPR- NOx results reveal that only BTCuO_2 presents the NOx conversion and NO₂ generation profiles expected for LNT catalysts [11,12]. Thus, even though all the catalysts present active sites for NO-to-NO₂ oxidation, such as oxygen vacancies and surface metal oxides, only the catalyst containing copper shows the presence of the Ba₂TiO₄ segregated phase, which is active for NOx adsorption [11,12].

In order to determine the NSC, NOx storage experiments at 400 °C (the minimum temperature for NOx adsorption in TPR-NOx profiles) have been carry out for the three perovskites in which Ti has been substituted in a larger degree, that is, BTCuO_2, BTFeO_2, and BTMnO_2. The NSC values (shown in

Table 2. NSC data at 400 °C for BTO reference, BaTi_0.8B_0.2O₃ catalysts and for some reference noble metal-base catalysts.

Catalyst	NSC (µmol/g)	Temperature (°C)	Lean Cycle Time (s)
BTMnO_2	83	400	300
BTFeO_2	99	400	300
BTCuO_2	269	400	300
1%Pt/20%BaO/Al2O3 [10]	150	350	120
2.2%Pt/16.3%BaO/Al2O3 [11]	400	350	240
2.2%Pt/20.8%BaO/Al2O3 [12]	240	300	240

) have been obtained during the 10th NOx storage cycle at which the catalysts achieve a stable performance (see Materials and Methods section for more details). Figure 4 shows, as an example, the NO, NO₂, and NOx profiles during NSC experiments at 400 °C, corresponding to the BTCuO_2 catalyst.

Data on Table 2 reveals that the three catalysts present a measurable NSC, but, in agreement with TPR-NOx results, BTCuO_2 catalyst is the most active one. The characterization results previously discussed allow us to justify the high NSC shown by copper perovskite. On the one hand, the incorporation of Cu into the perovskite lattice distorts the raw tetragonal structure, generates the largest pool of oxygen vacancies, and promotes the segregation of mainly Ba₂TiO₄ and BaCO₃, but also CuO as segregated phases that seem to be the active sites for both NO to NO₂ oxidation and NOx storage [11,12]. On the other hand, the insertion of Mn and Fe causes a structural change from tetragonal to cubic, but only BaCO₃ appears as segregated phase and a lower amount of oxygen vacancies respect to Cu. Consequently, a lower NO to NO₂ oxidation activity and NSC is shown by these two catalysts. Finally, it is important to underline that the NSC of BTCuO_2 is within the range of values reported for noble metal/alkali or alkali earth base catalysts (Table 2). Moreover, the BTCuO_2 perovskite does not incorporate any noble metal, and therefore it could be a cheaper alternative to current catalysts based on noble metals. Additionally, as this catalyst works at 400 °C, presenting an acceptable NOx storage capacity, it could be proposed as a component of high-temperature LNT for lean burn gasoline engines (GDI gasoline direct injection) which need catalysts working between 400 and 500 °C.

3. Materials and Methods

Four BaTi_0.8B_0.2O₃ catalysts (being B Mn, Fe, Co, or Cu), named BTMnO_2, BTFeO_2, BTCoO_2, and BTCuO_2 respectively, were prepared by the sol–gel method as previously described [11]. Summarizing, first of all, the hydrolysis of titanium isopropoxide (Ti) was carried out, dissolving the resulting species in an aqueous solution of citric acid (CA) (Ti:CA = 1:2) and hydrogen peroxide (Ti:H₂O₂ = 2:1), and obtaining the citrate–peroxo–titanate (IV) complex. Subsequently, NH₃ was used in order to adjust the pH to 8.5, and the addition of an stoichiometric (BaTi_0.8B_0.2O₃) amount of barium (Ba:Ti = 1:1) and metals precursors (barium acetate and Fe, Co, Cu, and Ni nitrates), took place. During 5 h, until the obtention of a gel, the temperature of the mixture remained 65 °C. Afterwards, a temperature of 90 °C or 24 h was used to dry the sample, which was in the end calcined at 850 °C for 6 h.

To measure the metal content in the samples by ICP-OES, a Perkin-Elmer device model Optima 4300 DV was used. An Autosorb-6B instrument from Quantachrome served to determine, by N₂ adsorption at −196 °C, the BET surface area of the samples. To identify different phases and crystalline structures, X-ray diffraction (XRD) and Raman spectroscopy were employed. XRD tests were performed with a Rigaku Miniflex II powder diffractometer, using Cu Kα (0,15418 nm) radiation with the 2θ angle in the range 20 to 80°, with a step of 0.025° and a time per step of 2 s. Raman scattering spectra were obtained on a Jobin-Ivon dispersive Raman spectrometer (model LabRam) with a variable power He:Ne laser source (633 nm) in the range of 100–1000 nm. To register the XPS spectra, a K-Alpha photoelectron spectrometer by Thermo-Scientific, with an Al Kα (1486.6 eV) radiation source, was used in the following conditions: 5 × 10⁻¹⁰ mbar pressure in the chamber and setting the C1s transition at 284.6 eV, and the binding energy (BE) and kinetic energy (KE) values then determined with the peak-fit software of the spectrophotometer, to regulate the BE and KE scales.

The catalytic activity of the samples (80 mg of catalyst diluted in 300 mg SiC) was tested using two different experiments in a fixed-bed quartz reactor at atmospheric pressure and under a gas flow (500 mL/min): i) temperature programmed reaction (TPR-NOx) tests (10 °C/min, 800 °C) in a gas mixture of 500 ppm NOx and 5 % O₂ and ii) NOx storage cyclic tests at 400 °C, with a gas mixture composed of: i) for lean (storage) cycle (5 min), 500 ppm NOx and 5 % O₂ balanced with N₂, and ii) for rich (regeneration) cycle (3 min), 10% H₂ balanced with N₂. To achieve the stability of the catalysts, and then determine the NSC, 10 consecutive storage–regeneration cycles were accomplished. The gas composition was controlled by specific NDIR-UV gas analyzers for NO, NO₂, CO, CO₂, and O₂ (Rosemount Analytical Model BINOS 1001, 1004, and 100).

NOx conversion profiles as a function of temperature were obtained using the next equation

$$\mathrm{NOx} \mathrm{conversion} (\%)=\frac{{\mathrm{NOx}}_{\mathrm{in}}-{\mathrm{NOx}}_{\mathrm{out}}}{{\mathrm{NOx}}_{\mathrm{in}}}\times 100$$

where ‘NOx_in’ is the concentration of NOx (=NO + NO₂) feed to the reactor and ‘NOx_out’ is the concentration of NOx that leaves the reactor.

The percentage of NO₂ generated during TPR was determined with the equation

$${\mathrm{NO}}_2(\%)=\frac{{\mathrm{NO}}_{2,\mathrm{out}}}{{\mathrm{NO}}_{2,\mathrm{in}}}\times 100$$

where ‘NO_2out’ is the concentration of NO₂ that leaves the reactor.

The NSC was obtained as the difference between the NOx signal when the reactor is unfilled and the NOx signal when the reactor is full of catalyst with

$$\mathrm{NOx} \mathrm{storage}={{\displaystyle \int }}_{\mathrm{t}0}^{\mathrm{tf}}{\mathrm{NOx}}_{\mathrm{inlet}}\left(\mathrm{t}\right)-{\mathrm{NOx}}_{\exp }\left(\mathrm{t}\right)\mathrm{dt}$$

where ‘NOx_inlet’ is the concentration of NOx (=NO + NO₂) measured when the reactor is empty, and ‘NOx_exp’ is the concentration of NOx during the NOx storage test.

4. Conclusions

From the analysis of the effect of Ti partial substitution by Mn, Fe, Co, or Cu on the NOx storage capacity (NSC) of the BaTi_0.8B_0.2O₃ lean NOx trap (LNT) catalyst, the following conclusions have been obtained:

• In BaTi_0.8B_0.2O₃ perovskites, Ti is partially substituted by Mn, Fe, Cu and, to a lower extent, by Co.
• The perovskite structure is modified or changed due to the insertion of B metal into the lattice:

  ○

  (i) For the BTCuO_2 catalyst, the tetragonal structure of the raw perovskite is distorted, a larger amount of oxygen vacancies is generated and Ba₂TiO₄ and BaCO₃ appear as main minority segregated phases, but CuO is also detected.

  ○

  (ii) For Mn, Fe, and Co, the tetragonal structure changes to cubic, a lower amount of oxygen vacancies are formed and BaCO₃ appears as segregated phase in the three catalysts. BTCoO_2 presents also BaCoO₃ and Co₃O₄ segregated phases due to a lower degree of Co insertion into the framework.

• Due to the described modifications, all the BaTi_0.8B_0.2O₃ catalysts are active for the NO oxidation to NO_2, which takes place on oxygen vacancies and metal oxide sites, but only the BTCuO_2 catalyst (for which Ba₂TiO₄ segregated phase is identified), presents a significant NOx storage capacity. In fact, at 400 °C, the BTCuO_2 catalyst features the highest NSC which is close to that shown by platinum base catalysts.