Storage and Reduction of NO _x Over LNT Catalysts

Abstract

In this paper, fundamental issues of the NSR technology are addressed. The reaction paths and the mechanism involved in the storage of NO _x and the relevance of the proximity between the noble metal and the alkali or alkaline-earth component of the catalyst are discussed first. Then the release of NO _x , the reaction pathways of NO _x reduction by different reducing agents, and the mechanisms involved in the formation of N₂, NH₃ and N₂O are illustrated. The use of the combined technologies LNT + SCR and DPNR is addressed next and key aspects that highlight the synergies of the individual technologies are discussed. Finally, the perspectives of the NSR technology are outlined.

Graphical Abstract

1 Introduction

Environmental policy in industrialized countries forces a drastic reduction of NO _x emissions from diesel engines that operate under lean conditions and have significantly better fuel economy compared to gasoline stoichiometric engines. The EU emission standards for passenger cars are summarized in Table 1 [1]. Since the Euro 2 stage, EU regulations introduce different emission limits for diesel and gasoline vehicles. Diesel have more stringent CO standards but are allowed higher NO _x . However, significant reductions in NO _x limits are introduced for diesel engines with Euro 6 (from 0.18 g/km in Euro 5 down to 0.08 g/km in Euro 6), which makes NO _x emission standards essentially fuel neutral (0.08 g/km in diesel engines vs. 0.06 g/km in gasoline engines). Table 1 also shows that gasoline vehicles are exempted from particulate matter (PM) standards through the Euro 4 stage. Euro 5 and 6 regulations reduce the PM mass emission standard for diesel and introduce PM standards for gasoline cars with direct injection (DI) engines as well, numerically equal to those for diesel. The legislation also introduces a particulate number emission standard (P or PN) at the Euro 5/6 stage. At the time of adoption of the Euro 5/6 regulation, its mass-based PM emission limits could only be met by closed particulate filters (wall through filters). Number-based PM limits will prevent the possibility that in the future open filters are developed that meet the PM mass limit, yet enable a high number of ultrafine particles to pass.

Table 1 EU emission standards for passengers’ cars (Category M), g/km

Table 2 Isotopic distribution of N₂: i) expected in case of statistical coupling of Pt-¹⁵N and Pt-¹⁴N for the reaction 6 ¹⁵NO + 4 ¹⁴NH₃ → 5 (¹⁴N₂ + ¹⁵N₂ + ¹⁴N¹⁵N) + 6 H₂O; ii) measured during NH₃-TPR experiments

The light-duty (LD) NO _x regulations in Europe and US are compared in Fig. 1 [2]: the trend is always for very low NO _x emissions. Tighter regulations are foreseen for developing markets like Brazil, China and India.

Fig. 1

Euro V and Euro VI light-duty NO_x regulatory limits compared to the U.S.: a About 55 to 60 % NO_x control will be needed for a Euro V (2009) diesel to hit the U.S. Bin 8 maximum allowable emission (45 states). For Bin 5 (50 states) nominally 85 to 90 % NO _x control is needed; b For Euro VI (2014), the requirement is 60 to 65 % additional NO _x reduction (reprinted from [2])

While compliance with previous NO _x regulatory limits under lean conditions has been largely achieved without resort to NO _x after-treatment, either selective catalytic reduction (SCR) or NO _x storage reduction (NSR) need to be applied to achieve the most severe NO _x standards. Indeed, with the proven technologies of automotive catalysis, the three-way catalytic (TWC) converter used for stoichiometric gasoline engines requires the absence of oxygen or very little excess oxygen and can reduce some NO _x under lean conditions. An additional driver for the use of de-NO _x after-treatment systems is emerging now in Europe: real world driving emissions. Investigators have found that the NO _x emissions from LD diesels can be 3–5 times higher than the laboratory certification level [3]. Along these lines new test cycles (e.g. worldwide harmonized Light duty driving test cycle, WLTC) are under development that are obtained from real world data and require advanced control for cold start, steeper transients accelerations, and higher speed and load conditions. These will make the removal of NO _x extremely challenging. Finally, the strategy to improve the fuel economy of cars and to reduce CO₂ emissions will motivate more extensive use of diesel vehicles and lean burn DI gasoline vehicles due to the more effective combustion realized under lean conditions.

The SCR approach is based on the reaction between NO and NH₃ [4], which is produced by hydrolysis of an aqueous urea solution injected into the exhausts from an on-board tank, according to the standard SCR reaction:

$$ 4\;{\text{NO + 4 NH}}_{3} {\text{ + O}}_{2} \to 4\,{\text{N}}_{ 2} { + 6}\,{\text{H}}_{ 2} {\text{O}} $$

(1)

The activity at low temperature, where most of NO _x emissions are generated, can be improved by positioning a DOC unit upstream of the SCR converter to oxidize NO to NO₂. This allows the occurrence of the fast SCR reaction, which is considerably faster than reaction (1) at low temperatures [5, 6]:

$$ {\text{NO + NO}}_{ 2} {\text{ + 2 NH}}_{ 3} \to 2 {\text{ N}}_{ 2} {\text{ + 3 H}}_{ 2} {\text{O}} $$

(2)

The first commercial use of SCR NO _x control in heavy duty (HD) trucks employed traditional vanadia-tungsta-titania catalysts. Nowadays Fe- and Cu-zeolites are preferred, in view of better high-temperature durability, since the SCR catalyst receives hot gases during regeneration of the upstream diesel particulate filter (DPF), and in view of higher low temperature activity. Copper zeolite is the preferred catalyst, has the best low-T performance and shows very little sensitivity to NO₂ concentration. Iron zeolite has the best high-T performance but requires NO₂ management in the inlet gas to improve LT performance, as do vanadia catalysts. This in turn, implies a higher precious metal usage in the DOC unit.

The NH₃/urea SCR is recognized as the most effective technology for the abatement of NO _x emissions from HD diesel vehicles, and a growing interest in the same technology is coming also from LD vehicles and passenger cars, in order to combine low fuel consumption and low NO _x emissions.

The NSR, also referred to as lean NO _x traps (LNT), represents an alternative viable technology for the removal of NO _x under lean conditions in vehicles [7–9]. The system is operated in a cyclic manner by alternating a long lean phase (30–90 s) and a short rich phase (3–5 s). The basic concept is that NO _x are trapped onto the catalyst when the engine is running lean and are reduced to produce N₂ when the engine is running rich. A typical NSR catalyst comprises a noble metal such as Pt to oxidize NO, CO and unburned hydrocarbons and to reduce trapped NO _x , and an alkaline and/or alkaline-earth component such as barium and/or potassium to trap the NO _x as nitrites/nitrates. Besides the catalysts typically contain Rh and Ce. These catalyst components are dispersed on a high surface area alumina support. The overall principle of operation of NSR has been described by the following five steps [9]: (i) oxidation of NO to NO₂; (ii) NO and/or NO₂ up-take in the form of nitrite and/or nitrates; (iii) evolution of the reducing agents when the exhaust is switched to the rich condition; (iv) NO _x release from stored nitrites and nitrates; (v) reduction of released NO _x to N₂.

The NSR technology has been commercialized for passenger cars equipped with lean burn engines in the early 2000s. In spite of the cost of the noble metal catalysts, NSR is a favored approach for lighter vehicles (i.e. engines of less than about 2.0 to 2.5 l capacity), mainly due to the fact that big layout modifications are unnecessary and the relatively fixed cost of an on-board urea system. However, since mixed-mode engines greatly reduce low-load NO _x , NSR systems could be operated at temperatures greater than 300 °C with a substantially lower loading of the platinum group metals (PGM). This could make NSR economically attractive for cars with mixed-mode engines of higher capacity as well (i.e. up to 5 or 6 l) [2].

The durability of LNTs to sulphur contamination has always been a major problem of NSR, based on the fact that sulphates are more stable than nitrites/nitrates and tend to poison the catalyst basic sites where NO _x are adsorbed [10, 11]. Accordingly, the catalyst is typically oversized to compensate for degradation during a given period and must be periodically regenerated; this is accomplished by decomposing the sulphates at high temperature under reducing conditions to give SO₂ and H₂O for a total of few minutes every few thousands km. In view of the introduction of ultra-low sulphur fuels, improvements in the catalyst resistance to sulphur poisoning and new control strategy, which limit the desulfation temperature, LNTs are effective to about 80 % NO _x efficiency in real world Light Duty systems [12]. Attention is also paid to limit the H₂S peak amount released during desulfation.

In order to decrease the size and the weight of the after-treatment devices, integrated systems have been developed where existing complementary technologies are combined. In the LNT + SCR configuration, ammonia slipped from the LNT catalyst during the rich phase is adsorbed onto the SCR catalyst bed placed downstream. Such adsorbed NH₃ reacts according to the SCR reactions with upcoming NO _x present in the exhausts which have not been trapped onto the NSR catalyst bed during the lean phase and with NO _x released during the fast lean-rich switch that escape from the NSR catalyst bed. Accordingly, over the LNT + SCR system the NO _x removal efficiency is increased, the ammonia slip is reduced and the PGM usage is lower. This technology was introduced in the US and in Europe in the 2000s [13].

Besides, in order to control soot emissions, technologies like DPF and continuously regeneration trap (CRT) [14, 15] are used in combination with SCR and LNT catalysts. Although changes to engine design will improve engine emissions in order to meet both PM and NO _x regulatory limits, combined systems including PM and NO _x after-treatment can be utilized. Soot treatment, NO _x reduction and CO and hydrocarbons oxidation functions are combined on a single monolith developed by Toyota and known as diesel particulate—NO _x reduction (DPNR) [16]. The commercialized “4 way catalytic converters” are composed of a particulate wall-flow filter coated with an NSR catalyst layer and work under cyclic conditions, as LNT catalyst does. Particulate filter regeneration is supposed to be effective during the lean phase, due to the presence of NO _x and the excess of oxygen in the exhaust gas, similarly to that occurring in the CRT technology (soot combustion by direct and cooperative reactions).

In this paper, fundamental issues of the NSR technology are addressed with the aim to provide a critical survey, and to identify open issues and space for future advancement. The reaction paths and the mechanism involved in the storage of NO _x and the role of proximity between the noble metal and the alkali or alkaline-earth component of the catalyst are discussed in Sect. 2. The release of NO _x , the reaction pathways of NO _x reduction using different reducing agents (H₂, CO, hydrocarbons), and the mechanisms involved in the formation of N₂, NH₃ and N₂O are illustrated in Sect. 3. The use of LNT in combination with other technologies like LNT + SCR and DPNR is addressed next and key aspects that highlight the synergies of the individual technologies are discussed. Finally, the perspectives of the NSR technology are outlined.

2 Reaction Pathways and Mechanistic Features in the Storage of NO _x

2.1 The NO _x Storage Mechanism

There is ample evidence in the literature that suggests NO₂ is a precursor in the storage of NO _x ; the NO₂ can be gas-phase, a surface-bound precursor, or an intermediate species.

Fridell et al. [17] suggested that NO₂ is formed from NO oxidation and is stored onto the catalyst surface as nitrates:

$$ {\text{NO + 1/2 O}}_{ 2} \to {\text{NO}}_{ 2} $$

(3)

$$ {\text{BaO + 3 NO}}_{ 2} \to {\text{Ba}}\left( {{\text{NO}}_{ 3} } \right)_{ 2} {\text{ + NO}} \uparrow $$

(4)

In this pathway, hereafter quoted as the nitrate route, Pt catalyses the oxidation of NO to NO₂ and the alkali- or alkali-earth metal compound participates in the NO₂ storage reaction. It is noted that reaction (4) involves the disproportionation of NO₂ to nitrite and nitrate intermediates, followed by the oxidation of nitrites to nitrates:

$$ 2 {\text{ BaO + 4 NO}}_{ 2} \to {\text{Ba}}\left( {{\text{NO}}_{ 3} } \right)_{ 2} {\text{ + Ba}}\left( {{\text{NO}}_{ 2} } \right)_{ 2} $$

(5)

$$ {\text{Ba}}\left( {{\text{NO}}_{ 2} } \right)_{ 2} {\text{ + 2 NO}}_{ 2} \to {\text{Ba}}\left( {{\text{NO}}_{ 3} } \right)_{ 2} {\text{ + 2 NO}} \uparrow $$

(6)

Indeed NO is effectively oxidized to NO₂ over Pt, although Ba has a negative effect on the rate of NO oxidation, and NSR trapping materials sorb more effectively NO₂ compared to NO [9]. Besides both nitrite and nitrate ad-species have been simultaneously observed by several authors by means of FTIR at the preliminary stage of storage from NO₂. Nitrites eventually are oxidized to nitrates, that prevail after a sufficiently long storage period [17–20].

However spectroscopic data in the early LNT literature suggest that the formation of nitrites may precede that of nitrates during NO/O₂ adsorption [21, 22]. Indeed some of us have indicated another pathway for the storage from NO/O₂, on the basis of combined pulse reactor experiments and in situ FTIR measurements accomplished over Pt–Ba/Al₂O₃, Ba/Al₂O₃, Pt/Al₂O₃ catalysts and bare Al₂O₃ [23, 24]. This route, hereafter called the nitrite route, is based on the stepwise oxidation of NO at a Pt site followed by adsorption at a neighboring Ba site to form nitrite ad-species:

$$ {\text{BaO + 2 NO + }}{\raise0.7ex\hbox{$1$} \!\mathord{\left/ {\vphantom {1 2}}\right.\kern-0pt} \!\lower0.7ex\hbox{$2$}}{\text{O}}_{ 2} \to {\text{Ba}}\left( {{\text{NO}}_{ 2} } \right)_{ 2} $$

(7)

In this route, the cooperative action between Pt and nearby Ba sites prevents the over-oxidation of NO to NO₂ due to the trapping of nitrite species; nitrites can be oxidized to nitrates. In addition to the formation of nitrites, the oxidation of NO to NO₂ over Pt may also occur and NO₂ can be stored on Ba sites directly in the form of nitrates via reaction (4). A sketch of the nitrite and nitrate pathways for NO _x storage is given in Fig. 2.

Fig. 2

Sketch of the nitrite and nitrate routes in the process of NO _x storage

It is worth noting that the formation of nitrates during NO/O₂ uptake followed by FTIR collected in situ under static conditions may be biased by the spurious oxidation of NO to NO₂ in the gas phase followed by adsorption of NO₂; this is prevented under operando conditions.

Recently it has been confirmed by FTIR under operando conditions [20] that NO/O₂ adsorption on Pt–Ba/Al₂O₃ at low temperature results exclusively in the formation of nitrites. Indeed, as shown in Fig. 3a, the adsorption at 150 °C results in the formation of chelating nitrites with characteristic bands at 1,350 cm⁻¹ [ν_sym (NO₂)] and at 1,217 cm⁻¹ [ν_asym (NO₂)], that grow during storage. A small band at 1,544 cm⁻¹ is seen only at high exposure times whose assignment is still under debate (monodentate nitrites/bidentate nitrates). Besides, as shown in Fig. 3d, immediately upon admission NO is detected at the reactor exit and the formation of NO₂ at catalyst saturation is at most negligible. It was concluded that at low temperature (150 °C) NO _x are stored only via the nitrite route in the form of nitrites.

Fig. 3

Results of operando FT-IR analysis upon NO/O₂ exposure (1,000 ppm NO in He + 3 % v/v O₂) of the model Pt–Ba/Al₂O₃ sample: a at 150 °C; b at 250 °C; c 350 °C. FT-IR spectra are recorded every 30 s (increasing storage time indicated by the arrow in the Figures) and the subtrahend spectrum is the catalyst before NO/O₂ admission. In d–f the gas phase analysis at the same temperature, respectively, in terms of NO, NO₂ and NO _x is reported

At higher T (250 °C), as shown in Fig. 3b, the band of nitrites at 1,217 cm⁻¹ is observed as well, that increases with storage, goes through a maximum and then almost disappears, while the bands of nitrates at near 1,320 and 1,420 cm⁻¹ increase markedly. The presence of the isosbestic point in the FTIR spectra confirms that nitrites are transformed into nitrates. Besides the oxidation of NO to NO₂ increases and a dead time in the NO breakthrough is observed (Fig. 3e), that has been associated with the occurrence of the nitrite route and with the integral behavior of the reactor. It was concluded that in addition to the nitrite pathway, the nitrate route was also effective at this temperature (250 °C).

At 350 °C both nitrites and nitrates are formed initially (Fig. 3c). The band of nitrites at 1,217 cm⁻¹ disappears soon while the bands of nitrates markedly increase and are the only bands present after 10 min of exposure. Besides, the dead time in NO breakthrough is greater and the oxidation of NO to NO₂ is more significant at this temperature (Fig. 3f). It is noted that the storage of NO₂ does not allow conversion of NO to NO₂ to be monitored using NO₂ evolution until a significant amount of sorption has occurred. It was concluded that at high temperature both the nitrite and nitrate pathways are rather effective and NO _x are stored only in the form of nitrates after prolonged time of exposure.

By combining the results of operando FTIR spectroscopy with those of the synchronous quantitative analysis of the gaseous species, the concentrations of nitrites and nitrates ad-species as function of time were evaluated at the quantitative level, as shown in Fig. 4. Inspection of the Figure indicates that the storage of NO _x at 150 °C from NO/O₂ occurs exclusively through the nitrite route: in fact only nitrites are formed at this temperature, and NO₂ and nitrates are not observed. The nitrite route is faster at higher temperatures as well. However at 200 °C and above the nitrites are oxidized to nitrates and nitrates are also formed directly from NO₂ via the nitrate route. Because of this, a maximum in the concentration of nitrite species is observed. At 350 °C the maximum is smaller and is seen at shorter time of exposure. Still the nitrites prevail over the nitrates for a storage period of about 15 min at 250 °C and of a couple of minutes at 350 °C.

Fig. 4

Evolution versus time of the main bands for different surface species detected by using FT-IR spectroscopy under operando conditions upon NO/O₂ exposure (1,000 ppm NO in He + 3 % v/v O₂) at 150, 200, 250 and 350 °C of the model Pt–Ba/Al₂O₃ sample (reprinted from [20])

The concentrations of the adsorbed species provided by operando FTIR and of the gaseous species provided by the simultaneous on line analysis of the gas phase upon exposure of Pt–Ba/Al₂O₃ to NO/O₂ mixture have been used as direct experimental responses to estimate the kinetic parameters of the steps involved in the process of NO _x storage. At variance, in traditional kinetics the concentration of adsorbed species is estimated in order to gain a satisfactory fitting of the time evolution of the gaseous species. Appropriate reaction scheme, kinetic models and reactor model of the IR cell were developed. Considering the fact that this approach is the result of the joint application of spectroscopic and kinetic methods, it has been termed as spectro-kinetics. The model was found to fit nicely with the time evolution of the concentration of NO and NO₂ in the gas phase and the surface coverage of nitrites and nitrates in the investigated temperature range [25].

The effect of CO₂ and H₂O, that are present in the exhausts, on NO _x storage over Pt–Ba/Al₂O₃ has been addressed recently in our labs [26], and among others by Weiss et al. [27], Epling et al. [28] and by Chaugule et al. [29]. We found that the nitrites pathway is not significantly affected by the presence of CO₂ at low temperature, but is inhibited at higher temperatures. However, the nitrate pathway is inhibited as well, particularly at high temperature, but to a lower extent. The different inhibiting effect of CO₂ has been ascribed to the different relative thermal stability of carbonates, nitrites and nitrates as function of temperature, which has been evaluated by means of dedicated adsorption–desorption experiments. The effect of water was minor, because BaCO₃ is more stable than Ba(OH)₂ [30, 31]. It is concluded that nitrites are formed preferentially during the early period of NO _x storage from NO/O₂ also in the presence of CO₂ and H₂O.

2.2 Relevance of the Proximity Between the Noble Metal and the Alkali or Alkaline-Earth Component

To investigate the importance of close contact between Pt and BaO during the storage, Cant et co-workers compared the NO _x storage characteristics of four systems: BaO/Al₂O₃, a sequential system with Pt/SiO₂ ahead of BaO/Al₂O₃, a combined system with Pt/SiO₂ and BaO/Al₂O₃ physically mixed, and BaO/Al₂O₃ with Pt deposited on it [32]. They found that storage of NO _x from NO/O₂ by the Pt–BaO/Al₂O₃ sample differed from that of the sequential and combined systems. Indeed, the rate of storage on Pt–BaO/Al₂O₃ is faster, and the total storage capacity under the conditions used is attained more quickly. The implication is that Pt, in close proximity to BaO, either allows uptake of NO or NO₂ in a way that is not feasible when Pt is absent or enables conversion of NO _x to nitrites/nitrate as storage proceeds. Spillover of NO₂ from Pt to BaO/Al₂O₃ is one way in which the latter might occur. These authors also investigated the isotopic exchange between ¹⁵NO and stored ¹⁴NO _x . The rate of exchange was more than five times as fast for Pt/BaO/Al₂O₃ than for the combined Pt/SiO₂ + Ba/Al₂O₃ system, demonstrating spillover of NO _x species between Pt and Ba in close proximity.

Similar results have been reported by Nova et al. [33]. Indeed, comparing the results between the physical mixture made of Pt/γ-Al₂O₃ and Ba/γ-Al₂O₃ and the ternary Pt–Ba/γ-Al₂O₃ sample it clearly appeared that the two systems are characterized by a similar NO _x breakthrough time but the physical mixture presents a higher oxidizing capacity, as revealed by the higher conversion of NO to NO₂ measured at catalyst saturation. In addition, the physical mixture resulted able to adsorb significant amounts of NO _x . It was assumed that the mechanism of NO _x adsorption operating on the physical mixture, i.e. in the absence of interaction between Pt and Ba atoms, is the nitrate route: NO is oxidized to NO₂ on Pt sites (reaction (3)) and then is adsorbed on Ba (and Al sites) via the disproportionation reaction (reaction (4)). NO, which is released during the same reaction is re-oxidized to NO₂ and this results in a significant delay in the NO _x breakthrough also in this case.

Hence, the data on the physical mixture confirm the feasibility of the NO _x storage mechanism proposed in the literature implying the oxidation of NO to NO₂ followed by NO₂ adsorption on the catalyst surface to form nitrates. In this case, no direct interaction is necessary to have an effective NO _x uptake. However, this is not the case of representative NSR catalysts, in which Pt and Ba are on the same support and which must be able to catalyze the reduction of the stored NO_x, as indeed realized in the NSR technology. As a matter of fact, the physical mixtures made of Pt/SiO₂ and Ba/Al₂O₃ or Pt/Al₂O₃ and Ba/Al₂O₃ are active towards the NO _x reduction at considerably higher temperatures compared to the ternary system Pt–Ba/Al₂O₃ [32, 33]. This is consistent with the decomposition of stored NO _x to gaseous NO_x and subsequent reduction by a catalytic reaction on Pt particles located remotely from BaO.

To investigate this interaction further and to better understand its role in the storage mechanism, a study was undertaken by Castoldi et al. [34] using Pt–Ba/Al₂O₃ samples with different Ba loadings, in order to vary the number of Pt–Ba neighboring couples. The CO chemisorption measurements performed by means of in situ FTIR analysis showed the characteristic band of CO linearly adsorbed on Pt sites in all cases; however, upon increasing the Ba loading the band progressively shifts towards lower energy due to the increase of the catalyst basic character. Hence, the data indicate a strong interaction between Pt and the basic oxygen anions of the Ba phase, suggesting that the exposed Pt sites and the Ba component are in close proximity. In addition, it has been shown that when applying a rectangular step feed of NO in He + O₂ (3 % v/v) both the NO _x breakthrough and the storage capacity of the catalyst increase with the Ba content not only because of the amount of Ba, but also the number of Pt-neighboring Ba sites increase, the uptake of NO/O₂ is faster over these sites and a large percentage of Ba participates to NSR [34]. The fraction of Ba that participates in the storage process (“active” Ba) increases linearly with the catalyst Ba content passing through a maximum that represents the best exploitation of the Ba component, while the Pt dispersion decreases from 80 to 40 % [34].

The relevance of the Pt–Ba proximity in the storage of NO _x implies that Ba sites distant from Pt are less reactive in the storage. This eventually accounts for the slower uptake of NO _x which is observed at the later stages of the trapping process, and that was explained invoking a nitrate spillover type mechanism [35, 36]. In fact the adsorption sites near Pt react and saturate first, then the trapping rate slows due to increasing resistance to nitrate surface diffusion as the N-species build around the Pt particles (two dimensional shrinking core model) and/or into the inner core of the Ba particles (three dimensional shrinking core model). The close proximity between Pt and BaO facilitate the removal of stored NO _x as well by lowering the stability of adsorbed species.

3 Reaction Pathways and Mechanism in the Reduction of the Stored NO _x

There are three methods for introduction/generation of the reductant in NSR [9]. The first strategy is the DI of the fuel into the exhausts upstream of the catalyst during the rich event. In the second strategy, the fuel is continuously processed over a reformer or other devices such as plasma or a partial oxidation catalyst to generate reductants, which are more effective in accomplishing the regeneration of the NO _x trap. In the third strategy, the lean burn engine is operated to generate intermittent pulses of rich exhausts to the LNT catalyst that either directly contain the reductant or are processed by the NSR catalyst or by an upstream catalyst to form the reductant. In this case, the regeneration event involves a rich transient originating from the engine.

It is difficult to determine what exact reducing species participates in the NO _x reduction process since the original reductant may undergo substantial changes over the NSR catalyst or the reformed devices added upstream. Besides, extensive heat is evolved due to catalytic combustion of the reductant during regeneration events in some modes of operation. Accordingly, the catalyst operation is highly non isothermal, resulting in complex dynamic temperature distribution profiles. Temperature swings, measured as gas phase temperature changes, can be as high as 50–100 °C at the front of the catalyst and only 2–3 °C at the back of the catalyst [9]. Keeping in mind these aspects, it is obvious how complex the mechanism of reduction is for the stored NO _x .

3.1 The Release of NO _x

It is a common belief that the reduction of stored NO _x over LNT catalysts implies the release of NO_x from the catalyst surface, followed by the reduction to N₂ and other by-products such as NH₃ and N₂O. The process of NO _x release is most commonly discussed in terms of the breakthrough of desorbed and unconverted NO _x during the rich event as schematically shown in Fig. 5. However, the detailed mechanism of the release of NO_x and of their further reduction remains under debate.

Fig. 5

A representative NO _x breakthrough and release profile. The catalyst was Pt/Ba/Al₂O₃,the reactant gas contained 250 ppm NO, 8 % H₂O, 8 % CO₂, 8 % O₂, balance N₂ in the lean/trapping phase and 2,000 ppm H₂, 8 % H₂O, 8 % CO₂, and a balance of N₂ in the regeneration/rich phase. The space velocity was 25,000 h⁻¹. The outlet NO _x concentration was monitored with a chemiluminescence detector (reprinted with permission from [9])

Some authors [37, 38] proposed that the NO_x are released as a result of the heat generated by the exothermic reactions upon the lean/rich switch (thermal release). Indeed, as the rich front passes over the catalyst and the surface temperature increases, stored NO_x can be released because their stability decreases with increasing temperature. However it has been shown that the reduction process, even when accomplished under nearly isothermal conditions (i.e. with small concentration of the reductant), occurs at temperatures well below that at which stored NO _x thermally decompose. Accordingly, under these conditions, the process is not initiated by the thermal decomposition of stored nitrates, but rather by a catalytic pathway involving Pt [33]. Other authors [39, 40] have proposed that the release of NO _x is caused by a decrease in the equilibrium stability of the stored nitrates due to the decrease in the partial pressure of NO and/or of oxygen upon the lean/rich switch. However this contribution is small during typical NSR operation since only a negligible desorption of weakly adsorbed NO _x species was observed in the time range of few seconds characteristic of the rich period, when the NO feed and the O₂ feed were shut down after the storage of NO _x was completed [30, 33]. Finally, the release of NO _x may be promoted by the establishment of a net reducing environment [41, 42].

To account for the reduction of NO _x either the reverse spill-over of NO _x from Ba to Pt is occurring or alternatively the reductant activates and then spills over from Pt sites to the storage component and reacts directly with nitrites and nitrates, so that they are eventually released as NO _x [32, 43, 44]. The subsequent reduction of the released NO _x may occur at the Pt metal sites following a TWC-like chemistry [45]. The transient product distribution displayed by three catalysts having varied dispersions (3.2, 8 and 50 %) was described by Harold and coworkers [35, 46] using a model based on global kinetics and detailed crystallite-scale catalyst features and was explained by the localized stored NO _x gradients in the Ba phase. The rate determining process during the regeneration was found to be the diffusion of stored NO _x within the Ba phase towards the Pt/Ba interface. Indeed the results reported by Coronado and Anderson [47], Cant et al. [32] and Nova et al. [33] clearly show that NO _x decomposition is easier when Pt and Ba are in close proximity.

Even if the net NO _x release at the lean/rich interface represents a small loss under many of the operating conditions, this release increases at higher temperatures and can result in not attaining the required emissions standards. Therefore, one of the most important issues is understanding the pure contributes that cause the NO _x release and the selectivity to N₂. Although the proposed pathways of NO _x release are relatively well accepted, decoupling the release of NO _x from their subsequent reduction to N₂ or other products is not possible during typical NSR operation. Indeed, in the presence of a reducing agent the NO _x release is immediately followed by reduction of the released NO _x , which is a very fast reaction. However, recently the release of stored NO _x has been decoupled from the reduction of the released NO _x and measured by performing NO isotopic exchange experiments under temperature programming, using labelled gaseous ¹⁵NO and unlabelled ¹⁴NO _x stored on the Pt–Ba/Al₂O₃ catalyst. The exchange between gaseous ¹⁵NO and stored ¹⁴NO _x was previously used by Cant et al. [32] to probe the forward and reverse spillover of NO _x species between Pt and BaO.

NO isotopic exchange experiments have shown that [48]: (i) Pt is necessary for the exchange because this is observed over Pt–Ba/Al₂O₃ but not over Ba/Al₂O₃; (ii) the NO isotopic exchange implies the release of stored NO _x and is observed from a temperature well below that of thermal decomposition of stored NO _x , thus pointing to a Pt-catalysed process. The exchange is observed from about 50 °C for nitrites stored from NO/O₂ at 150 °C and from about 250 °C for nitrates stored from NO/O₂ at 350 °C; (iii) all stored NO _x have been exchanged after T programming up to 350 °C and subsequent hold at this temperature for half an hour so that both nitrites and nitrates migrate between Pt and Ba sites.

The mechanism of NO isotopic exchange is sketched in Fig. 6. Upon interaction of unlabelled stored nitrite or nitrate species with reduced Pt sites unlabelled NO is released in the gas phase and PtO species are formed. PtO species oxidize labelled NO present in the gas phase to give labelled nitrite and nitrate adsorbed species restoring the reduced Pt sites. The NO isotopic exchange is a redox process. The reduced Pt sites activate NO _x stored species that are destabilized and eventually released as NO in the gas phase. It is worth to note that during typical NSR operation stored NO _x upon activation could either react with the reducing agents present in the exhausts to give N₂O/N₂/NH₃ or produce released but unconverted NO.

Fig. 6

Sketch of the mechanism of NO isotopic exchange/NO _x release

It is worth noting that in the NO isotopic exchange process labelled NO represents the reducing agent so that the released unlabelled NO cannot be reduced further. For this reasons, in NO isotopic exchange experiments the release of stored NO _x could be decoupled from its further reduction and the process of NO _x release could be characterized unambiguously.

3.2 Pathways of the Reduction of Stored NO _x Using H₂

The regeneration of the NO _x storage component is a crucial phase of the NSR process. The regeneration has been studied by several authors using H₂ as the reductant: the formation of NH₃, N₂O and N₂ was observed depending on temperature, H₂ concentration, time of exposure to the reductant.

Harold and coworkers [49] on the basis of measurements performed with different catalyst length sections proposed that there are two primary competing routes to the desired N₂ product: a direct route from the reduction of stored NO _x by H₂ (H₂ + NO _x  → N₂) and a sequential route of stored NO _x reacting with H₂ forming NH₃ and then NH₃ reducing NO _x stored downstream to give N₂ (H₂ + NO _x  → NH₃; NH₃ + NO _x  → N₂). Evidence for the direct route and the sequential route with ammonia as intermediate has also been provided by means of in situ measurements for the transient intra catalyst species (H₂, NH₃, N₂, NO _x ) using spatially resolved capillary inlet mass spectrometry (SpaciMS), which provides a spatiotemporal resolution during NSR catalyst regeneration as shown in Fig. 7 [50]. While the data clearly prove that intermediate NH₃ plays a significant role in LNT catalyst regeneration, it was further recognized that experimental and numerical work are needed to more clearly understand the partitioning between the two suggested regeneration pathways, the direct route and the sequential route with ammonia as intermediate.

Fig. 7

Species transients at different locations through the catalyst for a H₂, b N₂ and c NH₃ at 200 °C. The legend indicates the specific measurement location in millimeters relative to the catalyst inlet face. (reprinted with permission from [50])

Lindholm et al. [31] showed that N₂O and NH₃ can be formed when reducing NO _x with H₂ and that NO reduction, NH₃ formation and N₂O formation depend on the H₂ concentration. A delay in the ammonia signal was observed and was explained by the SCR-type reaction between nitrites/nitrates adsorbed on the storage component and ammonia; when the coverage of nitrites/nitrates has decreased NH₃ is observed in the gas phase.

Ribeiro et al. [51, 52] proposed that the regeneration of Pt–Ba/Al₂O₃ monolith NO _x trap with hydrogen involves a localized reaction front of the reductant that travels through the catalyst bed with complete regeneration of the trapping sites. The process was limited by the supply of the hydrogen atoms, irrespective of the source of hydrogen (H₂ or NH₃), and occurred through a complex pathway where at first nitrogen and ammonia were formed according to a parallel scheme. NH₃ further reacts with the stored NO _x leading to the selective formation of N₂.

In order to elucidate the mechanism governing the reduction of the NO _x species stored onto a model Pt–Ba/Al₂O₃ LNT catalyst in the form of nitrates, TPD (temperature programmed desorption), H₂-TPSR (temperature programmed surface reaction), NH₃-TPSR, H₂-ICSC (isothermal concentration step change) and NH₃-ICSC experiments have been performed by Lietti et al. [33, 53, 54] under nearly isothermal conditions by using a small concentration of the reductant. The data in Fig. 8 show that the reduction of stored NO _x is active at temperatures well below that of NO _x thermal decomposition (near 120 °C in panel B vs. near 330 °C in panel A) and accordingly the NO _x reduction is a Pt-catalyzed process that does not require the thermal release of stored NO _x as the preliminary step. Besides, Fig. 8 shows that the reduction of nitrates by H₂ occurs from much lower temperature in the presence than in the absence of H₂O (near 50 °C in panel C vs. near 120 °C in panel B). The promoting effect of water might be explained considering that water increases the H₂ spillover rate over oxide surfaces [55], thus enhancing the reduction of nitrates stored far away from Pt sites. However specific effects of water on the mobility and reactivity of adsorbed nitrates can be suggested as well considering that water affects the features of nitrate ad-species, as reported by Szanyi and coworkers [56]. This is in line with mechanism involving the NO _x spillover from Ba to Pt sites. From Fig. 8 it is also clear that the reduction of nitrates by H₂ results in the formation of ammonia as the main product when accomplished at low temperatures (panel C) as opposed to when it’s carried out at high temperatures (e.g. 350 °C). Indeed, in this case N₂ represents the most abundant reaction product and NH₃ is detected in small amounts and only after N₂ evolution (not shown in Fig. 8). Finally, Fig. 8 shows that the reduction of nitrates by ammonia is slower since the onset of this reaction is observed at higher temperature than the one with hydrogen (near 125 °C in panel D vs. near 50 °C in panels C), and is fully selective to nitrogen (panel D).

Fig. 8

a TPD in He from r.t. to 600 °C (10 °C/min); b H₂-TPSR (2,000 ppm H₂) from r.t. to 500 °C (10 °C/min); c H₂ + H₂O-TPSR (2,000 ppm H₂ + 1 % v/v H₂O) from r.t. to 500 °C (10 °C/min); d NH₃ + H₂O-TPSR (1,000 ppm NH₃ + 1 % v/v H₂O) from r.t. to 500 °C (10 °C/min). Storage conditions: 1,000 ppm NO in 3 % v/v O₂ + He at 350 °C over Pt–Ba/Al₂O₃ catalyst

These data were described by the following molecular reaction scheme:

$$ {\text{Ba}}\left( {{\text{NO}}_{ 3} } \right)_{ 2} {\text{ + 8 H}}_{ 2} \to 2 {\text{ NH}}_{ 3} {\text{ + BaO + 5 H}}_{ 2} {\text{O}} $$

(8)

$$ 3 {\text{ Ba}}\left( {{\text{NO}}_{ 3} } \right)_{ 2} {\text{ + 10 NH}}_{ 3} \to 8 {\text{ N}}_{ 2} {\text{ + 3 BaO + 15 H}}_{ 2} {\text{O}} $$

(9)

where reaction (8) is faster than reaction (9).

This scheme together with the integral behavior of the laboratory micro-reactor accounts for the temporal evolution of the products observed during the reduction of the stored nitrates with H₂ and with NH₃ in the investigated T range 150–350 °C. It also implies the full consumption of H₂ and the formation of a hydrogen front travelling along the reactor axis, with ammonia formation that occurs at the H₂ front. Although the data could be explained at the quantitative level by the consecutive pathway provided by reactions (8) and (9) [57] a parallel + consecutive reaction scheme, where N₂ is also formed directly from nitrates, cannot be ruled out.

A schematic of the regeneration pathway proposed by Ribeiro et al. [51, 52] (and by Phil et al. [58]) is depicted in Fig. 9. The Figure illustrates the propagation of the reductant front along the catalyst bed with complete regeneration of the trap. Once NO _x is released it reacts with H₂ over Pt to form NH₃, N₂ and N₂O, the selectivity to the individual species depending on the local NO _x /H₂ concentration ratios. In the beginning of the front where the H₂ level is high compared to the NO _x NH₃ will be mostly formed. The NH₃ formed will further react with more NO _x to give either N₂ or N₂O if the temperature is sufficiently high. If the N₂O is formed behind the front, it will be reduced to N₂ by H₂ or NH₃. As the front approaches the end of the catalyst the supply of NO _x starts to deplete and is insufficient to react with the NH₃ formed upstream, leading to the NH₃ breakthrough. Before the reaction front reaches the end of the bed, most of the non-selective products will react further to form N₂ and maintain the high selectivity of the overall NSR catalyst.

Fig. 9

Schematic of the reduction mechanism for a Pt/BaO/Al₂O₃ monolith regenerated with H₂. Bottom panel illustrates a zoomed-in picture of the reaction front (reprinted with permission from [52])

3.3 Pathways of NO _x Reduction Using CO and Hydrocarbons

The regeneration of LNT catalysts has also been studied using CO, propane, propene and other hydrocarbons as reductants [56, 59–61]. It has been observed that H₂ and CO are highly efficient reductants compared to C₃H₆, which is somewhat less efficient. C₃H₈ did not show any NO _x reduction ability for NO _x stored on Pt/Ba/Al₂O₃ except at 350 °C.

Szailer et al. [56] reported that during the reduction by CO of NO _x stored on Pt/Al₂O₃ and Pt–Ba/Al₂O₃ at low temperature and under dry conditions stable isocyanates form at the Pt particles and spill over the basic oxide component of the catalyst, resulting in consumption of a large fraction of gas-phase CO and formation of a significant amount of CO₂ but not of N₂. At temperatures corresponding to the onset of stored nitrate decomposition these NCO species can readily react with the NO _x released by the catalyst to give N₂ and CO₂ according to the reaction:

$$ {\text{NCO + NO}}_{x} \to {\text{N}}_{ 2} {\text{ + CO}}_{ 2} $$

(10)

N₂ formation was significantly enhanced by adding water to the NCO covered catalysts because NCO was hydrolyzed to NH₃ and CO₂ so that the formed NH₃ could react with NO _x and form the intermediate that eventually generated N₂ and H₂O.

Forzatti et al. [62] showed that the reduction of nitrates stored onto Pt–Ba/Al₂O₃ by CO under dry conditions does not involve the thermal release of stored NO _x as the preliminary step, is seen from a temperature where Pt-carbonyls disappear and CO is consumed while N₂ is observed from a slightly higher temperature. It was found that the reduction of nitrates by CO occurs via a consecutive reaction scheme with formation of NCO species followed by reaction of these species with nitrates/nitrites and that the reaction of NCO species to give nitrogen is slower than the reduction of nitrates with CO to give NCO species. Large amounts of stable NCO species were formed and retained onto the catalyst surface at the end of the rich phase and were oxidized to give N₂ during the subsequent lean phase upon admission of O₂ and NO/O₂. Lower amounts of isocyanate species were left at the end of the reduction over Pt-K/Al₂O₃ than over Pt–Ba/Al₂O₃ likely due their higher mobility [63, 64]. It is worth noting that the C and N balances during the rich and the lean phases of the NSR cycle are consistent with the formation of stable NCO species and their consumption to give N₂ respectively. Amiridis and coworkers [61] confirmed that the reaction of surface isocyanates with O₂ and NO accounts for a significant fraction of the overall N₂ production. The oxidation of isocyanates/cyanates with O₂ to give N₂ and CO₂ was observed previously over Pt–Rh/Ba/Al₂O₃ and Ag/Al₂O₃ catalysts by Daturi and coworkers [65, 66] using FT-IR spectroscopy and by Schoulten and coworkers [67] over Pt–Ba/Al₂O₃ performing lean–rich cycling experiments. Also Toops and coworkers [68] found that Ba–NCO is directly derived from the reduction of Ba nitrate by CO, i.e., Ba–NCO species represent a critical intermediate in the reduction of stored NO_x with CO. Under dry cycling conditions with CO as the reductant, N₂ is mainly generated via NCO reaction with NO + O₂ after the switch to lean phase conditions, rather than being formed during the rich phase. However, the evolution of DRIFT spectra under both temperature-programmed and isothermal reaction modes revealed that H₂O is the most reactive species with respect to isocyanate of those tested (H₂O, O₂, NO, NO/O₂).

The role of surface isocyanate species as intermediate in the formation of N₂ is also supported by previous literature reports where isocyanates formed on several HC-SCR catalysts were found to be highly reactive towards NO₂ and NO + O₂ to yield N₂ [69–72]. It is worth noting that NCO species are stabilized in the regeneration of NSR catalyst by CO due to the interaction with the alkali or alkaline-earth component of the catalyst.

However, different routes may be active in the reduction of stored NO _x species by CO under wet conditions. Indeed the onset temperature for the CO consumption is observed at temperatures slightly lower under wet than under dry conditions and is very close to the onset temperature of the WGS reaction, the NO _x removal efficiency is higher, and NH₃ and H₂ are seen among the products [73]. FTIR showed that bidentate nitrates are transformed with temperature into ionic nitrates while NCO species are observed only in small amounts. The hydrolysis of the surface NCO species to give ammonia and carbon dioxide has also been reported by Daturi and coworkers [65] and the formation of small amounts of NCO species during cycling experiments in the presence of water was observed by Amiridis and coworkers [61]. Therefore, it is concluded that hydrogen formed through the water gas shift (WGS) and/or ammonia formed by hydrolysis of NCO species could be involved in the reduction of stored NO _x by CO in the presence of water. Recently Harold and co-workers [74] provided evidence that the surface NCO species are important reaction intermediates at moderate to higher temperatures (≥250 °C) during the cyclic reduction of NO _x by CO under wet conditions. Current on-going spatiotemporal studies have been anticipated to identify the best routes to N₂.

Propylene and n-heptane were also used in the regeneration of LNT catalysts under dry conditions [35, 75]. Both propylene and n-heptane were found to react well below the onset temperature of nitrate thermal decomposition. Hence, also in this case, as in the case of H₂ and CO, the reduction of the stored NO _x does not imply the preliminary thermal decomposition of the NO _x adsorbed species. The reduction of the stored nitrates is accompanied by the evolution of N₂ that represents the main product, and then of minor amounts of ammonia.

However, in reality the NSR is performed in the presence of water and it is commonly accepted that the reduction of stored NO _x occurs thanks to the production of H₂ via steam-reforming of the hydrocarbon employed as reductant. Then n-C₇-TPSR run in the presence of water has been carried out and the onset of the nitrate reduction was observed at temperatures very close to that measured under dry conditions (200–250 °C), i.e. the reaction was not significantly affected by the presence of water. Also the distribution of reduction products was very similar.

Dedicated n-heptane steam-reforming (SR) experiments have been performed, which showed the formation of H₂ and CO₂ above 250 °C although with poor reactivity. In the SR experiments, the production of H₂ is observed from 250 °C. However, the amount of H₂ produced via the SR reaction was not sufficient to justify the large amount of N₂ produced. Besides no N-containing surface species, in particular isocyanates, could be detected by in situ FT-IR. Accordingly, a different pathway should likely be considered. The pathway may imply cleaning of the Pt sites by the reductant followed by migration-decomposition of the stored NO _x at the reduced Pt sites.

3.4 Mechanism of the Reduction of Stored NO _x by NH₃

As already mentioned, ammonia is seen among the products during reduction of stored NO _x in particular at low T when hydrogen is used as reductant and it has been suggested in the literature as intermediate in the production of nitrogen. However, the mechanism of the reduction of stored NO _x by NH₃ has not been completely clarified so far. A study has been carried out recently in our labs to identify the possible mechanism for N₂ formation when ammonia represents the reducing agent of the stored NO _x .

Considering that the NSR catalysts have been derived from classical three-way catalysts by adding an alkaline component to the active wash-coat layer, and that during the regeneration of NSR systems NO _x are released in the gas phase as NO, one can suppose that a TWC-like reduction chemistry occurs. Accordingly, the following set of elementary reactions can be considered for the reaction between NH₃ and NO over Pt-based catalysts, as suggested in the literature [76]:

$$ {\text{NO + Pt}} \to {\text{Pt - NO}} $$

(11)

$$ {\text{Pt-NO + Pt}} \to {\text{Pt-N + Pt-O}} $$

(12)

$$ {\text{NH}}_{ 3} {\text{ + Pt}} \to {\text{Pt - NH}}_{ 3} $$

(13)

$$ {\text{Pt-NH}}_{ 3} {\text{ + Pt}} \to {\text{Pt-H + Pt-NH}}_{ 2} $$

(14a)

$$ {\text{Pt-NH}}_{ 2} {\text{ + Pt}} \to {\text{Pt-H + Pt-NH}} $$

(14b)

$$ {\text{Pt-NH + Pt}} \to {\text{Pt-N + Pt-H}} $$

(14c)

$$ {\text{Pt-N + Pt-N}} \to {\text{N}}_{ 2} {\text{ + 2 Pt}} $$

(15)

$$ {\text{Pt-O + 2 Pt-H}} \to {\text{H}}_{ 2} {\text{O + 3 Pt}} $$

(16)

In this scheme gaseous NO dissociates at Pt sites cleaned by the reductant to form Pt-N and Pt-O species (reactions 11 and 12). Activation of ammonia leads to the formation of Pt-H species (reactions 14) that scavenge the O-ad-atoms leading to the formation of reduced Pt sites and water (reaction 16). The recombination of adsorbed nitrogen atoms follows to form N₂ (reaction 15) while reaction of adsorbed nitrogen and hydrogen atoms produces NH _x and ultimately NH₃ (reverse reactions 14 and 13).

The mechanism of the reduction by ammonia of nitrates stored onto a Pt–Ba/Al₂O₃ catalyst has been investigated recently using labelled nitrates and unlabelled ammonia [77, 78]. Since in real automobile operation NO will also be present, it is of interest to study how its presence modifies the reaction network for stored NO _x reduction to N₂, by feeding NO together with the reductant in the regeneration period. To do this, NH₃-TPSR and NH₃-ICSC of stored nitrates in the presence and absence of NO in the gas phase have been accomplished. The temperature programmed reaction (TPR) between gaseous labelled ¹⁵NO and unlabelled ¹⁴NH₃ over Pt–Ba/Al₂O₃ was also studied for comparison purposes.

The results of the TPR run between ¹⁵NO and ¹⁴NH₃ are shown in Fig. 10, panel A. The reaction is observed starting near 100 °C, with formation of single- and double-labeled N₂O, and of single labeled N₂. Above 160–180 °C the conversion of NO is complete, a drop in the concentration of N₂O is observed and the reaction becomes fully selective to N₂. All N₂ isotopes are seen above this temperature, with abundances that do not change significantly with temperature. Notably, the relative abundance of the di-nitrogen isotopes observed at T > 160–180 °C, where ¹⁵NO is no longer observed among the products, compares well with that calculated in the case of statistical coupling of N-adspecies formed via dissociation of gaseous ¹⁵NO and gaseous ¹⁴NH₃ according to the stoichiometry:

Fig. 10

a TPR ¹⁴NH₃ (1,000 ppm) + ¹⁵NO (1,000 ppm); b TPSR run with ¹⁴NH₃ (1,000 ppm) + ¹⁴NO (1,000 ppm) after ¹⁵NO _x adsorption at 350 °C (1,000 ppm ¹⁵NO + O₂ 3 % v/v in He); c TPSR run with ¹⁴NH₃ (1,000 ppm) after ¹⁵NO _x adsorption at 350 °C (1,000 ppm ¹⁵NO + O₂ 3 % v/v in He)

$$ 6\,^{ 1 5} {\text{NO + 4}}\,^{ 1 4} {\text{NH}}_{ 3} \to 5 { }\left( {^{ 1 4} {\text{N}}_{ 2} { + }^{ 1 5} {\text{N}}_{ 2} { + }^{ 1 4} {\text{N}}^{ 1 5} {\text{N}}} \right){\text{ + 6 H}}_{ 2} {\text{O}} $$

(17)

as shown in Table 2.

Table 3 Isotopic distribution of N₂: (i) expected in case of statistical coupling of Pt-¹⁵N and Pt-¹⁴N for the reaction 3 Ba(¹⁵NO₃)₂ + 10 ¹⁴NH₃ → 8 (¹⁴N₂ + ¹⁵N₂ + ¹⁴N¹⁵N) + 3 BaO + 15 H₂O; (ii) measured during NH₃-ICSC of stored nitrates at 200 °C and at 350 °C, and during NH₃-TPSR of stored nitrates

The reduction of labeled nitrates with unlabeled ammonia has also been investigated, and results are shown in Fig. 10, panel C. The reaction is detected from slightly above 150 °C, is fully selective to N₂ with an isotopic distribution that does not change with temperature. Note that NO has never been detected among the products. The Pt-¹⁴N species originated from ¹⁴NH₃ and the Pt-¹⁵N species produced from labelled nitrates participate in the reaction in a ratio 10/6 in line with the stoichiometry constraint provided by the reaction:

$$ 3 {\text{ Ba}}\left( {^{ 1 5} {\text{NO}}_{ 3} } \right)_{ 2} { + 10}\,^{ 1 4} {\text{NH}}_{ 3} \to 8 { }\left( {^{ 1 4} {\text{N}}_{ 2} \,{ + }\,^{ 1 5} {\text{N}}_{ 2} \,{ + }\,^{ 1 4} {\text{N}}^{ 1 5} {\text{N}}} \right){\text{ + 3 BaO + 15 H}}_{ 2} {\text{O}} $$

(18)

leading to the following abundance of N₂ isotopes, calculated assuming the statistical coupling of Pt-¹⁴N and Pt-¹⁵N species: ¹⁴N₂ = 39 %, ¹⁴N¹⁵N = 47 % and ¹⁵N₂ = 14 %. These numbers compare reasonably well with those measured during the ¹⁴NH₃-TPSR shown in Fig. 10, panel C and also with ¹⁴NH₃-ICSC experiments of stored labelled nitrates (Table 3). It is concluded that in the reduction of stored nitrates by ammonia, N₂ is produced primarily by statistical coupling of adsorbed N-atoms coming from stored NO _x and gaseous NH₃.

The reaction of ¹⁴NH₃ with ¹⁵NO (TPR experiment, Fig. 10 panel A) is faster than the reaction of ¹⁴NH₃ with the stored labelled nitrates (TPSR experiment, Fig. 10 panel C). In fact, the onset of the reaction was observed at lower temperatures in the former case (near 100 vs. near 150 °C) and the reaction proceeds till the full consumption of the limiting reactant gaseous ¹⁴NO [77]. Besides when both gaseous NO and stored nitrates are present (Fig. 10 panel B) the reaction involving gaseous species prevails [77]. Indeed, when gaseous ¹⁴NO is present unlabelled N₂O and unlabelled N₂ are formed, and when gaseous ¹⁴NO has been depleted, ¹⁴NH₃ reacts with stored labelled nitrates to give ¹⁴N¹⁵N and ¹⁵N₂ although unlabelled N₂ is by far the most abundant isotope. The reactivity of NH₃ with gaseous ¹⁴NO seems to be slightly inhibited in the presence of stored nitrates. In fact, the onset of the NH₃ + NO reaction is seen at higher temperature in the presence of stored nitrates (near 120 vs. near 100 °C). Based on these results it is concluded that gaseous NO is more reactive towards NH₃ than stored nitrates and that the activation of the stored species represents the rate determining step in the reduction of stored nitrates. Besides, when NO is not present in the gas phase, the Pt sites are cleaned by the reducing agent and accordingly reducing conditions prevail. The primary route in the formation of di-nitrogen is the statistical coupling between N-adspecies formed upon release of ¹⁵NO from stored nitrates and upon decomposition of ¹⁴NH₃.

The reduction of labelled nitrites stored from ¹⁵NO/O₂ at 150 °C on Pt–Ba/Al₂O₃ with unlabelled ammonia was also investigated [79]. Under temperature programming, the reaction is seen near 120 °C. At low T (T < 160 °C) single labelled ¹⁵N¹⁴N represents the main product (with minor amounts of ¹⁵N¹⁴NO). Then, also the other di-nitrogen isotopes are detected. The reaction was also accomplished under isothermal conditions at 150 °C by applying a step change in the concentration of ¹⁵NO in the feed. Again single labelled ¹⁵N¹⁴N represents the main product initially (with minor amounts of ¹⁵N¹⁴NO). Then also the other N₂ isotopes are detected. In both cases NO was never detected among the products.

The abundance of single labelled nitrogen ¹⁵N¹⁴N obtained during TPR of ¹⁵NO + ¹⁴NH₃, TPSR of stored labelled nitrite with unlabelled NH₃ and isothermal concentration step change experiments for the reduction of stored labelled nitrites by unlabelled NH₃ at low T (150 °C) is shown in Fig. 11a–c, respectively. It appears that the abundance of single labelled nitrogen ¹⁵N¹⁴N is significantly higher than that expected in the case of statistical coupling of N-ad-atoms formed from decomposition of gaseous ¹⁴NH₃ and decomposition of gaseous ¹⁵NO or from release of stored nitrite and its further reduction (around 50 % to the best) either at low temperature (TPR and TPSR experiments) and at short time on stream (ICSC experiment at 150 °C).

Fig. 11

¹⁵N¹⁴N relative abundance during a TPR ¹⁴NH₃ (660 ppm) + ¹⁵NO (1,000 ppm); b TPSR run with ¹⁴NH₃ (1,000 ppm) after ¹⁵NO _x adsorption at 150 °C (1,000 ppm ¹⁵NO + O₂ 3 % v/v in He); c ICSC run with ¹⁴NH₃ (1,000 ppm) after ¹⁵NO _x adsorption at 150 °C (1,000 ppm ¹⁵NO + O₂ 3 % v/v in He)

Accordingly, the statistical coupling of N ad-species cannot be considered here as the unique route responsible for the formation of N₂. Another pathway involving the preferential coupling of NO- and NH₃-related species must be invoked to explain concentrations of single labeled ¹⁵N¹⁴N species significantly higher than 50 % (Fig. 12; [79]). It is worth noting that this pathway is also accompanied by the formation of N₂O, i.e. an oxidized N-containing species whose presence suggests that the coupling between NO- and NH₃-related species operates under slightly oxidizing conditions. Indeed, the presence of PtO sites prevents the NO dissociation.

Fig. 12

a TPSR run with ¹⁴NH₃ (1,000 ppm) after ¹⁵NO _x adsorption at 150 °C (1,000 ppm ¹⁵NO + O₂ 3 % v/v in He), b TPR run with ¹⁴NH₃ (660 ppm) and ¹⁵NO (1,000 ppm) in He, c TPR run with ¹⁴NH₃ (660 ppm), ¹⁵NO (1,000 ppm), and O₂ (3 % v/v) in He over Pt–Ba/Al₂O₃ catalyst. (reprinted with permission from [79])

Isotopic labelled experiments also provide information on the pathways involved in the formation of N₂O [77–79]. As shown in Fig. 12, this species is formed in limited amounts during the reduction of stored nitrites by NH₃ (panel A) while amounts higher by one order of magnitude have been observed during the reaction of ¹⁵NO (1,000 ppm) present in the feed gas with ¹⁴NH₃ (panel B). Since unlabelled N₂O was not observed, ¹⁵NO must be directly involved in the formation of N₂O. Accordingly and in line with literature proposals [80, 81] it was suggested that N₂O formation involves either the coupling of two adsorbed NO molecule [reaction (19)] or the recombination of an adsorbed NO molecule with an adsorbed NH _x fragment originated from NH₃ adsorption [reaction (20)]:

$$ 2\,^{ 1 5} {\text{NO}}_{{ ( {\text{a)}}}} \to {}^{15}{\text{N}}^{ 2} {\text{O}}_{{ ( {\text{g)}}}} {\text{ + O}}_{{ ( {\text{a)}}}} {\text{ + s}} $$

(19)

$$ ^{ 1 4} {\text{NH}}_{\text{x(a)}} { + }^{ 1 5} {\text{NO}}_{{ ( {\text{a)}}}} \to \,^{ 1 4} {\text{N}}^{ 1 5} {\text{NO}}_{{ ( {\text{g)}}}} {\text{ + x H}}_{{ ( {\text{a)}}}} $$

(20)

The formation of N₂O is strongly affected by temperature. Indeed above 180 °C the N₂O concentration is always negligible in the absence of oxygen. It was suggested that at high temperature NO dissociation was favored because Pt is kept in a reduced state by ammonia, so that N₂O formation is prevented due to the lack of adsorbed NO species. Besides N₂O could decompose onto reduced Pt sites to give N₂ and PtO [82].

In the presence of oxygen (3 % v/v) in the feed stream, the nitrous oxide concentration is greater (Fig. 12, panel C) because NO dissociation is prevented over the oxygen covered Pt surface and this increases the concentration of NO ad-species. The presence of oxygen in the feed changes the nitrous oxide distribution during the TPR of ¹⁴NH₃ + ¹⁵NO: in the absence of oxygen ¹⁵N₂O prevails over ¹⁴N¹⁵NO while in the presence of oxygen the concentration of the double-labelled nitrous oxide is negligible while that of single-labelled nitrous oxide is significantly greater. Indeed adsorbed oxygen drives reaction (19) from right to left, while it was reported to favour ammonia activation yielding reactive NH _x adsorbed intermediates [83]; this eventually explains the formation in greater amounts of ¹⁴N¹⁵NO.

As previously discussed, the presence of NO in the gas phase during the regeneration period could modify the reaction network for NO _x reduction to N₂. To study the behaviour of a Pt–BaO/Al₂O₃ catalyst during NO _x reduction when NO is also fed together with H₂ in the regeneration period, Pereda-Ayo et al. [82] employed isotopically labelled techniques, storing ¹⁴N unlabelled nitrates that were reduced during the subsequent rich period, feeding labelled ¹⁵NO together with H₂. All N₂ isotopes were detected, namely ¹⁵N₂, ¹⁵N¹⁴N and ¹⁴N₂. Three different routes for N₂ formation were deduced (as illustrated in Fig. 13): (i) in route 1, the incoming hydrogen reacts with stored nitrates to form ¹⁴NH₃ which further reacts with nitrates stored downstream to form ¹⁴N₂; (ii) in route 2 the incoming ¹⁵NO reacts with hydrogen to form ¹⁵NH₃ upstream of the hydrogen front, it travels through the catalyst bed and reaches the regeneration front where it gets involved in the reduction of stored nitrates to form ¹⁵N¹⁴N; (iii) in route 3 incoming ¹⁵NO reacts with hydrogen to form ¹⁵N₂.

Fig. 13

Schematic of the proposed regeneration mechanism of the LNT with hydrogen. Storage was achieved with ¹⁴NO and regeneration was accomplished with H₂ in the presence of ¹⁵NO (reprinted with permission from [82])

4 Combined Systems: LNT + SCR and DPNR

4.1 LNT + SCR

It has been shown that NH₃ is formed during regeneration of LNT catalysts, while it is well known that SCR catalysts store large quantities of NH₃ under reaction conditions [84]. Hence, combining the LNT catalyst with a downstream SCR catalyst offers a potential of capturing NH₃ generated over the LNT and using it to convert NO _x that slips through the LNT [85]. These LNT + SCR coupled systems result in higher NO _x removal and lower NH₃ slip compared to the LNT-only systems. The LNT + SCR concept has been commercially demonstrated on the Mercedes E320 Bluetec vehicle [13] which utilized relatively high-loaded PGM LNT in combination with an Fe-based zeolite SCR catalyst. This approach is variously referred to as an in situ, hybrid, booster, or passive LNT + SCR system.

Most of the fundamental studies have involved model catalysts tested in simple laboratory gas mixtures and with hydrogen as the reducing agent. For example, Corbos et al. [86, 87] showed that the NO _x removal efficiency can be greatly improved under lean–rich atmosphere if a NSR catalyst is physically mixed with Cu-ZSM-5. Indeed, Fig. 14 shows the NO _x removal efficiency during the lean–rich cycles for the different catalyst mixtures (Pt–RhBa and SCR catalysts). The NO _x removal activity of Pt–RhBa catalyst shows a volcano type curve and reaches a maximum of 50 % at 300 °C. Figure 14 points out that mixing the Pt–RhBa catalyst with AgAl catalyst has no significant influence on the NO _x removal activity as compared to mixing with bare α-Al₂O₃, regardless of temperature. On the other hand, the NO _x removal activity was improved in 200–300 °C temperature range by mixing Pt–RhBa catalyst with CoAl or CuZSM-5. A remarkable difference is observed at 250 and 300 °C. At these temperatures, the activity of Pt–RhBa + CuZSM-5 mixed catalyst is much higher than that obtained with Pt–RhBa + CoAl [85, 86, 88].

Fig. 14

NO _x removal efficiency (%) during the lean–rich cycles for physically mixed catalysts (diamond) Pt–RhBa + α-Al₂O₃, (cross symbol) Pt–RhBa + AgAl, (square) Pt–RhBa + CuZSM-5 and (triangle) Pt–RhBa + CoAl. Reaction conditions: lean 100 s, 1 % CO₂, 500 ppm NO, 10 % O₂, 0.13 % CO/H₂ (75/25), 167 ppm C₃H₆ and rich 10 s with 8.53 % CO/H₂ (75/25), 100 ppm NO, 1 % CO₂ (reprinted with permission from [86])

In most of these studies, NSR experiments have been performed by alternating long lean phases with short rich periods characterized by high concentration of the reductant. Accordingly, significant temperature effects may have occurred. In view of this, only a qualitative analysis of the data has been generally attempted.

A systematic investigation of the hybrid arrangements LNT + SCR has also been accomplished in our labs [89, 90], which combines reactivity measurements performed under nearly isothermal conditions (i.e. with low concentration of NO and of the reductant) and in situ FT-IR spectroscopy to analyse the species formed in the gas phase and at the catalyst surface during the lean–rich cycles, covering the temperature range 150–350 °C and also considering the presence of H₂O and CO₂ in the exhausts. Pt–Ba/Al₂O₃ LNT catalyst and Fe-ZSM-5 SCR catalyst have been tested alone in a typical NSR adsorption-reduction cycle. Then both LNT + SCR dual bed and LNT/SCR physical mixture have been tested in the same conditions. NO _x removal efficiency, N₂ production during the lean and rich periods, NH₃ oxidation upon O₂ admission during the lean phase and NH₃ slip were determined and explained at the quantitative level for all the mentioned arrangements and the results obtained during lean–rich experiments performed at 250 °C are reported in Table 4. From this quantitative analysis, and the FT-IR evidences, it has been possible to provide a picture of the working mechanism of these combined catalytic systems, depicted in Fig. 15. During the lean phase, the storage of NO _x (NO _x stored in Table 4) occurs primarily on Pt–Ba/Al₂O₃ and results in the formation of nitrate species, well recognized by FTIR at the end of the lean phase (bands at 1,416, 1,328 and 1,035 cm⁻¹). Moreover, the FT-IR study showed that the Fe-ZSM-5 SCR catalyst is able to adsorb large amounts of ammonia, the trapping of NH₃ being most favoured at low temperatures. Accordingly, during the rich phase of a NSR cycle the ammonia produced in the stored NO _x reduction can be adsorbed by Fe-ZSM-5 SCR catalyst of the combined LNT + SCR catalytic system (both physical mixture or dual bed).

Table 4 Results of lean–rich experiments performed over LNT (Pt–Ba/Al₂O₃), LNT + SCR (Fe-ZSM-5) physical mixture and LNT/SCR dual bed arrangements

Fig. 15

Sketch of the behavior of hybrid LNT-SCR catalytic systems, physical mixture and dual bed

As it clearly appears from the sketch of Fig. 15, and the data of Table 4, the NO _x removal efficiency (NO _x removed in Table 4) is always higher for both hybrid LNT + SCR systems due to the contribution of N₂ produced by the SCR reactions over the Fe-ZSM-5 catalyst during the lean phase (N₂ lean) in addition to NO _x stored onto the LNT catalyst during the lean phase (NO _x stored in Table 4). For all investigated systems (LNT, LNT + SCR physical mixture and LNT + SCR dual bed) the amounts of NO _x removed correspond to amounts of NO _x stored plus those of N₂ lean.

The ammonia species adsorbed during the rich phase are completely consumed by the SCR reactions upon subsequent admission of NO/O₂ to produce N₂ (N₂ lean in Table 4). Until the ammonia is present on the surface of the SCR catalyst, the only observable NO _x surface species are Fe³⁺(NO) indicating that NO is directly bonded at Fe sites (band at 1,878 cm⁻¹). After adsorbed ammonia is completely consumed, the increase of NO⁺ (peak at 2,132 cm⁻¹) and nitrate (main peaks at 1,628 cm⁻¹ and at 1,573 cm⁻¹) species is observed together with the erosion of the band related to Fe³⁺(NO) species. This indicates that at first SCR reaction takes place and then small amounts of NO _x are stored at the catalyst surface as nitrates [91].

When the hybrid LNT + SCR systems are considered, in the reduction of nitrates stored over Pt–Ba/Al₂O₃, N₂ is formed during the rich phase (N₂ rich in Table 4) together with NH₃, that is trapped onto the Fe-ZSM-5 catalyst in the form of coordinated NH₃ (absorptions in the region 3,500–3,100 cm⁻¹ and at 1,616 cm⁻¹) and of protonated NH₄ ⁺ (absorptions in the region 3,000–2,250 cm⁻¹ and at 1,447 cm⁻¹).

In the NSR cycle, NH₃ trapped onto Fe-ZSM-5 during the rich phase is oxidized during the subsequent lean phase immediately after O₂ admission to give N₂, N₂O, and NO _x (N₂ + N₂O + NO _x at O₂ admission in Table 4) and upon subsequent NO admission to give N₂ according to standard and fast SCR reactions (N₂ lean). Therefore, the amount of NH₃ stored during the rich phase over the Fe-ZSM-5 particles is obtained by the sum of amounts of N₂ + N₂O + NO _x at O₂ admission and of amounts of N₂ lean; this sum is always higher for the physical mixture than for the dual bed. It turns out that the trapping of NH₃ is most favoured when the particles of Pt–Ba/Al₂O₃ and of Fe-ZSM-5 are evenly distributed in the catalyst bed, as in the case of the LNT + SCR physical mixture. Trapping the NH₃ onto the Fe-ZSM-5 particles prevents the reaction of NH₃ with NO _x stored downstream onto Pt–Ba/Al₂O₃ particles to give N₂ during the rich phase (N₂ rich in Table 4), that in fact is lower for the physical mixture than for the dual bed.

It is worth noting that the oxidation of NH₃ trapped onto the Fe-ZSM-5 particles after admission of O₂ was observed only in the case of the physical mixture and was ascribed to desorption of ammonia from the Fe-ZSM-5 particles followed by reaction over the Pt–Ba/Al₂O₃ in close proximity. The selectivity to nitrogen for this reaction increases significantly with temperature and is almost complete at 350 °C. The adsorption of NH₃ over the zeolite catalyst during the rich phase and its consumption during the subsequent lean phase is documented by the FT-IR study. No other intermediate species have been detected.

In conclusion, due to the fact that ammonia, formed in the reduction of nitrates by hydrogen over the LNT catalyst, can be effectively stored onto the SCR catalyst, the NO _x removal efficiency is increased and the ammonia slip is reduced over the coupled LNT + SCR systems. The ammonia slip is reduced to the best over the dual bed arrangement (LNT ahead of SCR). The synergy becomes more important with the aged NSR catalyst where generally higher NH₃ yields are observed in comparison with the fresh NSR catalyst [91].

The class of combined LNT + SCR catalysts also comprises the system developed by Sato and coworkers [92] consisting of a solid acid SCR catalyst and Pt/OSC (oxygen storage component) such as CeO₂/ZrO₂ as NSR catalyst.

4.2 DPNR

The DPNR (Diesel Particulate-NO _x Reduction) technology aimed at the simultaneous reduction of PM and NO _x was developed by Toyota [16, 93]. The catalytic converter for DPNR is a porous ceramic wall-flow filter coated with a NSR catalyst. Like the NSR technology, this system accomplishes the removal NO _x under cyclic lean/rich conditions; in addition the simultaneous soot abatement occurs, mainly under lean conditions due to the presence of NO _x and the excess oxygen in the exhaust gas. Of note, Toyota researchers also proposed that active oxygen species, which are formed during NO _x adsorption, are effective in soot oxidation under rich conditions as well.

Castoldi et al. [94] showed that under cycling conditions, i.e. alternating lean/rich phases, the Pt–Ba/Al₂O₃ NSR system is able to simultaneously remove NO _x and soot.

Soot combustion was primarily ascribed to the NO₂ formed upon NO oxidation over Pt, and to the occurrence of the NO recycling to NO₂ [95]. Moreover, based on a comparative study of Pt–Ba/Al₂O₃ and Pt/Al₂O₃, a role of the stored NO _x in the oxidation of soot has been suggested [96], which is in agreement with the results of Suzuki and Matsumoto [16] who noted a higher combustion activity of soot when a NO _x trap material is incorporated in the catalyst. Actually, there is a general consensus on the fact that the NO _x storage function of the LNT catalysts positively affects the combustion of soot by providing an additional path for oxidation of the carbonaceous material in addition to the well-established gas-phase reactions involving O₂ and NO₂. Kustov et al. [97] reported that the stored nitrates may decrease the onset temperature of soot oxidation when nitrate decomposition occurs in a proper temperature range, due to the release of NO₂ in the gas phase. Along similar lines, Sullivan et al. [98] reported that the presence of a NO _x trapping component like BaO in Pt/SiO₂ systems does not promote per se the particulate combustion, but favors soot combustion due to the increase in the NO₂ gas phase concentration, observed upon nitrates decomposition during LNT catalyst regeneration.

However more recently, some of us investigated the reactivity of stored nitrates during TPD (temperature programmed desorption)/TPO (temperature programmed oxidation) experiments carried out over Ba- and K-based LNT systems and suggested that stored nitrates take a direct part in soot oxidation, which indeed is observed below the temperature threshold for nitrates thermal decomposition [96, 99, 100]. In particular a redox mechanism occurring between nitrates, acting as oxidant agents toward soot, and soot particles, acting as reducing centres, has been suggested. This hypothesis is in agreement with the results obtained by Ito et al. [101] over Cs-MnO _x –CeO₂ catalysts and by Sanchez et al. [102, 103] over lanthanum supported catalysts containing noble metals.

Several authors reported that the activity of LNT catalysts for soot combustion is higher in the presence of K rather than of Ba [104–106]. A direct role of the alkali or alkaline-earth component in the oxidation of soot has been pointed out, which seems to be superimposed to the gas phase soot combustion activity by NO₂. It has been suggested [107, 108] that potassium may favor the soot oxidation through the formation of low melting point compounds, thus improving the mobility of the active surface species and favoring the contact between the soot and the catalyst which has been claimed to be a key factor in the oxidation of soot [109]. More recently, Castoldi et al. [110] investigated the intrinsic reactivity of alkaline (Na, K, Cs)—and alkaline/earth (Ca, Ba, Mg)—based catalysts in the oxidation of soot, and prepared samples in which the active elements were deposited directly onto the soot particles (‘‘full contact’’ conditions). The reactivity in the soot combustion of the selected elements has been correlated with their electronegativity, with the overall activity ranking (Cs ≥ K > Ba > Na > Ca ≫ Mg) which follows the electropositivity order of the investigated elements. Notably, under these conditions, the performance of the poorly reactive Ba-containing sample became more comparable to that of the very active K-based system.

It is worth noting that, when considering the loose contact systems the correlation between electropositivity and activity in the soot combustion is not more apparent. This indicates that under these conditions other parameters play a major role in the activity, thus confirming that the high activity of K-based systems, even under loose conditions, is likely due to the high mobility of the K surface species. It has been noted that NSR catalysts can suffer for the loss of the alkali or alkaline-earth metal oxide in view of its surface mobility. Indeed Krishna and Makkee [104] found Pt-K/Al₂O₃ catalyst to be more active but less stable than Pt–Ba/Al₂O₃.

Finally, several studies have pointed out the negative effect of soot on the NO _x storage behavior of the LNT catalysts [99, 100, 105, 106]. Sullivan et al. [111] showed that the presence of soot decreases the NO _x storage capacity of a Pt–Ba/SiO₂ system at 400 °C; in particular they hypothesized that soot may compete with Ba sites for reaction with NO₂ by offering another reaction path for the utilization of NO₂ rather than the desired formation of surface nitrates. Along similar lines, Artioli et al. [99] observed that soot decreases the NO _x storage capacity of Pt–Ba/Al₂O₃ system and favors the decomposition and reduction of the stored nitrates, while soot is being oxidized. A direct reaction between the stored nitrates and soot was suggested, that has been explained with the surface mobility of the adsorbed nitrates. More recently Klein et al. [112, 113] attributed the decrease in the NO _x storage capacity of Pt–Ba/Al₂O₃ catalysts to the destabilization of stored nitrates via a direct surface reaction with carbon particles.

In conclusion, very few investigations deal with the effect of soot on the regeneration of LNT catalysts. Nakatani et al. [93] observed that the NO_x conversion efficiency of DPNR catalysts is almost similar to that of the soot-free NSR catalysts. This is confirmed by investigations performed within our group [96, 99, 100].

5 Perspectives

The development of successful catalytic technologies for the control of NO _x emissions from vehicles under lean conditions over the last 20 years has been a remarkable technical achievement. Catalytic lean NO _x removal converters are of commercial importance and extensive research has been done on several aspects of their formulation, synthesis and performance. Presently SCR and NSR represent the top contenders for lean NO _x removal. SCR is a more mature technology that requires an on board urea tank so that it is used mostly for HD and high segment vehicles but it is considered as well for LD vehicles and passenger cars. NSR does not require big layout modification because it uses the exhaust as the reducing agent and is the favored approach for low segment vehicles.

In spite of the fact that the purpose of this paper was not to compare one technology with the other, but to address the reaction process fundamentals of the NSR technology, still a comparison between NSR and SCR technologies is worthwhile in order to enlighten analogies and differences in the associated chemistries and catalyst functionalities.

History teaches us that the advances in technology will come mostly from the application and that fundamental insight of such advances usually comes later. Experimental and theoretical tools are now available that allow us to better understand the details of the catalytic chemistry of NSR. Among these special importance has been (and it is expected that more will be in the future) the use of spatially resolved analytical tools such as capillary inlet mass spectrometry (SpaciMS) to provide a spatiotemporal resolution during transient storage and regeneration processes occurring over NSR catalysts, of isotopic experiments to address the mechanism of NO _x release and of N₂ formation and of spectro- kinetics to combine the analysis of gas phase and surface concentrations.

It is noted that coupled LNT + SCR systems, that use the LNT to generate ammonia during the rich period of the cycle and then store and use ammonia in a downstream SCR catalyst, are of special interest nowadays. They are being optimized to ensure high NO _x removal efficiencies while reducing the PGM usage and are moving onto much more vehicle testing. Ammonia generation from the NSR and finding optimum system configurations are at the forefront [12].

Recently a very interesting phenomenon that achieves high NO _x conversion was observed over NSR catalysts by Toyota researchers [114]. This phenomenon occurs when continuous short cycle injections of hydrocarbons are supplied at a predetermined concentration in lean conditions. The system that uses this concept was named Di-Air (diesel NO _x after-treatment by adsorbed intermediate reductant) because it was concluded that such a phenomenon is likely due to the generation of reactive intermediates formed by adsorbed NO _x over the base catalysts and HCs partially oxidized. It is recommended that the Di-Air concept should be studied more deeply by the scientific community.

Besides, the nitrite route, which was found to dominate at low temperatures during the storage of NO/O₂ over LNT catalysts, suggests we should revisit the chemistry of NO oxidation to NO₂ since in this case the reaction proceeds through a stepwise oxidation to form nitrites that are stabilized against further reaction to form NO₂ by the interaction with nearby Ba sites.

Finally, it has been shown that nitrites and nitrates stored species exhibit different stability and reactivity during the reduction process. These aspects could be investigated in details and the possible implications and/or challenges explored.