Measures to Reduce the N 2 O Formation at Perovskite-Based Lean NO x Trap Catalysts under Lean Conditions

: The net oxidising atmosphere of lean burn engines requires a special after-treatment catalyst for NO x removal from the exhaust gas. Lean NO x traps (LNT) are such kind of catalysts. To increase the efﬁciency of LNTs at low temperatures platinised perovskite-based inﬁltration composites La 0.5 Sr 0.5 Fe 1-x M x O 3- δ /Al 2 O 3 with M = Nb, Ti, Zr have been developed. In general, platinum based LNT catalysts show an undesired, hazardous formation of N 2 O in the lean operation mode due to a competing C 3 H 6 -selective catalytic reduction (SCR) at the platinum sites. To reduce N 2 O emissions an additional Rh-coating, obtained by incipient wetness impregnation, besides the Pt coating and a two-layered oxidation catalyst (2 wt.% Pd/20 wt.% CeO 2 /alumina)-LNT constitution, has been investigated. Though the combined Rh-Pt coating shows a slightly increased NO x storage capacity (NSC) at temperatures above 300 ◦ C, it does not decrease N 2 O formation. The layered oxidation catalyst-LNT system shows a decrease in N 2 O formation of up to 60% at 200 ◦ C, increasing the maximum NSC up to 176 µ mol/g. Furthermore, the NSC temperature range is broadened compared to that of the pure LNT catalyst, now covering a range of 250–300 ◦ C.


Introduction
Lean-burn engines operating with air excess show high emissions of NO and NO 2 , referred to as NO x [1]. To reduce the emissions of these harmful gases additional aftertreatment systems need to be included in the exhaust pathway [2]. Under ideal conditions [3,4] the selective catalytic reduction (SCR) catalyst [5,6] or the lean NO x trap catalyst (LNT), also known as NO x storage and reduction catalyst (NSR) [7], reduce NO x selectively to N 2 and water. However, ideal conditions are rarely encountered due to the varying exhaust temperatures (150-500 • C) [8,9] giving rise to different exhaust gas compositions as a result of incomplete combustion processes [9,10] and side reactions [3,[11][12][13]. Hence, side product formation of secondary pollutants N 2 O and NH 3 occurs due to the use of catalytic after-treatment systems [14,15]. Whereas ammonia is able to reduce NO x to N 2 , N 2 O is less reactive and still being emitted [16,17]. Nitrous oxide is proven to be one of the most dangerous greenhouse gases with a lifetime of 120-150 years and a global warming potential (GWP) which is almost 300 times higher than that of CO 2  To reduce N 2 O formation over Pt-based LNT catalysts at low temperatures there are two options: either conversion of the N 2 O to N 2 or suppressing the C 3 H 6 -SCR side reaction. The former option is quite challenging as the O 2 concentration is two orders of magnitude higher than that of N 2 O concentration [34]. The most common catalysts investigated for this specific application are Rh-based due to their high N 2 selectivities [29,35], but they commonly suffer from a poor low-temperature activity as the light-off temperatures T 50 (temperature of 50% conversion) are around 350 • C [34,36]. Centi et al. [34,36] investigated a zirconia catalyst with 1 wt.% of Rh for its N 2 O conversion performance in different feed gas mixtures and observed a shift of T 50 (~260 • C) to higher temperatures (T 50~3 80-400 • C) in the presence of O 2 and H 2 O due to catalyst deactivation. However, upon removal of O 2 and H 2 O from the feed gas, the good performance (before adding O 2 and H 2 O) at low temperatures could be retrieved [36,37]. The investigations of Beyer et al. [35] and Parres-Esclapez et al. [38] also showed a performance loss in N 2 O conversion for 0.58 wt.% Rh/Al 2 O 3 (T 50 = 341 • C) and 0.5 wt.% Rh/SrAl 2 O 3 (T 50 ≈ 300 • C), caused by the presence of O 2 .
The second approach of SCR reaction inhibition can be achieved with a reduction of available propylene by oxidation. Again, supported noble metal catalysts possess a remarkable ability for the oxidation of hydrocarbons and CO at low temperatures. Palladium is the most common commercially used precious metal [39], for example, in the compositions Pd/CeO 2 [40,41], Pd/Al 2 O 3 [42] and Pd/CeO 2 -Al 2 O 3 [42]. The survey of Shen et al. [41] displays a strong relation between the C 3 H 6 and CO light-off temperatures and the loaded Pd content. It was shown that the light-off temperature T 50 (C 3 H 6 ) shifts from around 225 • C to 140 • C upon increasing the Pd loading from 1 to 7 wt.% [41]. The same shift was observed for the CO oxidation T 50 values, which are commonly at lower temperatures than for propylene oxidation (e.g., T 50 (CO) at~200 • C for 1 wt.% Pd). Even though Faria et al. [43] proposed CeO 2 to enhance the oxidation ability of Pd/Al 2 O 3 due to a better precious metal dispersion, Guimarães et al. [42] observed a T 50 shift to higher temperatures with the incorporation of ceria (T 50 (Pd/Al 2 O 3 ):~375 • C and T 50 (Pd/CeO 2 /Al 2 O 3 ): 490 • C). However, it must be pointed out that both catalysts were pre-treated differently.
Infiltration composites with 20 wt.% loadings of B-site substituted lanthanum strontium ferrates (La 0.5 Sr 0.5 Fe 1-x M x O 3-δ M = Nb, Ti, Zr) on an alumina support showed promising NO x storage capacities (NSC) for the substituting elements Nb, Ti and Zr under laboratory testing conditions [44]. Depending on the perovskite composition a maximal NSC of around 120 to 164 µmol/g was measured in the temperature range from 250 • C to 350 • C. Hereby, an increasing maximum NSC and a temperature-dependent shift to lower temperatures are found in the order of Nb < Ti < Zr [44].
The main focus of this study is to investigate nitrous oxide formation on 2.5 wt.% Pt/ 20 wt.% La 0.5 Sr 0.5 Fe 1-x M x O 3 /Al 2 O 3 (M = Nb, Ti, Zr)-based lean NO x trap catalysts in the lean operation mode. After determining the underlying mechanism of formation in the presence of CO and C 3 H 6 reductives [23], different approaches to reduce or avoid nitrous oxide formation have been examined. The first approach considers an additional precious metal coating of 0.125 wt.% Rh obtained via incipient wetness impregnation onto the LNT catalyst in order to increase the number of active NO x oxidation and reduction sites on the catalyst surface as well as the N 2 selectivity. The second approach comprises the usage of an oxidation catalyst (2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 ) deposited on top of the LNT material in a two-layer configuration within the reactor to decrease the reductive concentrations before passing through the LNT catalyst. The temperature-dependent NO x storage capacity investigations were carried out up to the NO x saturation of the catalysts in order to determine not only the maximum storage capacity of the materials but also any possible side reactions, e.g., N 2 O or NH 3 formation. Figure 1 shows the detected temperature-dependent N 2 O formation curves  (BaO_Pt). The general curve trend is that a profound increase in the amount of N 2 O formed at low temperature (up to~19 to 25 ppm at 250 • C), is followed by a profound decline at temperatures above 350 • C to an almost zero level. These results demonstrate similar behaviour for the perovskite-type oxide-based infiltration composites as is known for other (earth-) alkaline-based LNT catalysts [9,11,16,45]. Nevertheless, a small difference is found between the performance of LSFN_Pt and that of the other catalysts. The Nb-containing catalyst displays a much lower activity as significantly lower N 2 O formation values are found in the temperature range of 250-300 • C when compared to values found for the other catalysts.

) during Lean Operation Conditions
The temperature-dependent NOx storage capacity investigations were carried out up to the NOx saturation of the catalysts in order to determine not only the maximum storage capacity of the materials but also any possible side reactions, e.g., N2O or NH3 formation. Figure 1 shows the detected temperature-dependent N2O formation curves for 2.5 wt. % Pt/20 wt. % La0.5Sr0.5Fe0.5Ti0.5O3/Al2O3 (LSFT_Pt), 2.5 wt. % Pt/20 wt. % La0.5Sr0.5Fe0.75Nb0.25O3/Al2O3 (LSFN_Pt) and 2.5 wt. % Pt/20 wt. % La0.5Sr0.5Fe0.5Zr0.5O3/Al2O3 (LSFZ_Pt) along with that for the reference catalyst 2.5 wt. % Pt/20 wt. % BaO/Al2O3 (BaO_Pt). The general curve trend is that a profound increase in the amount of N2O formed at low temperature (up to ~19 to 25 ppm at 250 °C), is followed by a profound decline at temperatures above 350 °C to an almost zero level. These results demonstrate similar behaviour for the perovskite-type oxide-based infiltration composites as is known for other (earth-) alkaline-based LNT catalysts [9,11,16,45]. Nevertheless, a small difference is found between the performance of LSFN_Pt and that of the other catalysts. The Nb-containing catalyst displays a much lower activity as significantly lower N2O formation values are found in the temperature range of 250-300 °C when compared to values found for the other catalysts. To examine the impact of reductive gases CO and C3H6 on the N2O formation in more detail, additional temperature-dependent NOx storage measurements were carried out under different synthetic exhaust gas compositions (Table 1). These experiments were conducted only for the LSFZ_Pt. Corresponding results are given in Figure 2. To examine the impact of reductive gases CO and C 3 H 6 on the N 2 O formation in more detail, additional temperature-dependent NO x storage measurements were carried out under different synthetic exhaust gas compositions (Table 1). These experiments were conducted only for the LSFZ_Pt. Corresponding results are given in Figure 2. without propylene (labelled w/o_C3H6) shows a completely temperature-independent behaviour with a constant, almost negligible N2O formation of around 1-2 ppm. The results clearly reveal a N2O evolution dependency on the presence of propylene only, and so do verify a propylene-SCR mechanism. Furthermore, when comparing the progression curves of the lean exhaust gas and that without CO, a higher N2O evolution at 200 and 250 °C is observed for the latter. These results suggest that CO counteracts the undesired N2O formation mechanism at low temperatures.  Table 1.

Impact of Rh on the NOx Storage Capacity and N2O Evolution of the Infiltration Composites
Rhodium catalysts are most effective in reducing N2O emissions due to their excellent N2O decomposition properties at temperatures above 300 °C [34][35][36]. However, Hamada [29] observed a slight C3H6-SCR activity for Rh/Al2O3 in a net oxidising atmosphere already at 200 and 250 °C. Due to this activity, the addition of rhodium to the platinised  Table 1.
Whereas the exhaust gas mixture without CO (labelled w/o_CO) shows a curve progression analogous to that of the lean gas mixture (i.e., with CO and C 3 H 6 ), the feed gas without propylene (labelled w/o_C 3 H 6 ) shows a completely temperature-independent behaviour with a constant, almost negligible N 2 O formation of around 1-2 ppm. The results clearly reveal a N 2 O evolution dependency on the presence of propylene only, and so do verify a propylene-SCR mechanism. Furthermore, when comparing the progression curves of the lean exhaust gas and that without CO, a higher N 2 O evolution at 200 • C and 250 • C is observed for the latter. These results suggest that CO counteracts the undesired N 2 O formation mechanism at low temperatures.

Impact of Rh on the NO x Storage Capacity and N 2 O Evolution of the Infiltration Composites
Rhodium catalysts are most effective in reducing N 2 O emissions due to their excellent N 2 O decomposition properties at temperatures above 300 • C [34][35][36]. However, Hamada [29] observed a slight C 3 H 6 -SCR activity for Rh/Al 2 O 3 in a net oxidising atmosphere already at 200 and 250 • C. Due to this activity, the addition of rhodium to the platinised infiltration composites (2.5 wt.% Pt/20 wt.% La 0.5 Sr 0.5 Fe 1-x M x O 3-δ /Al 2 O 3 ) seems to be a possibility to reduce N 2 O emission at 250 • C, corresponding to the temperature of maximal N 2 O formation, by a more selective catalytic reduction reaction (see Figure 1).
Since Abduhlhamid et al. [20] reported a low catalytic activity of Rh in NO oxidation, a combined Pt-Rh coating on the infiltration composites was used to maintain a good NSC, while decreasing the N 2 O evolution. According to the commonly commercially used Pt to Rh ratio, the Rh amount was set to 0.125 wt.%, leading to a Pt-Rh ratio of 1:0.05. Figure 3 compares obtained data for the NSC and N 2 O formation of as-prepared LSFT_Pt and LSFT_PtRh. Starting with a comparison of the NSC ( Figure 3A) of the bare Pt-coated and that of mixed Pt-Rh coated samples, a general shift of the adsorption curve by 50 • C towards higher temperatures occurs due to the addition of 0.125 wt.% Rh. As a result, the Pt-Rh combination shows a higher NSC than bare Pt in the range of 300-450 • C.
Additionally, the maximum storage temperature increases from 142 µmol/g for LSFT_Pt (250 • C) to 151 µmol/g for LSFT_PtRh (300 • C).The expectation that N 2 O formation for LSFT_PtRh is decreased with slightly increasing the NSC is refuted by the results shown in Figure 3B. Here it can be seen that with additional Rh coating, a higher N 2 O formation over the entire temperature range occurs for LSFT_PtRh. While the average amounts of N 2 O formed in the moderate to high-temperature range continue to converge, ending at values of 1-3 ppm, the difference in the low-temperature range is substantial. While the difference is about 15 ppm at 200 • C, it is reduced to 2 ppm at temperatures above 250 • C. a combined Pt-Rh coating on the infiltration composites was used to maintain a good NSC, while decreasing the N2O evolution. According to the commonly commercially used Pt to Rh ratio, the Rh amount was set to 0.125 wt. %, leading to a Pt-Rh ratio of 1:0.05. Figure 3 compares obtained data for the NSC and N2O formation of as-prepared LSFT_Pt and LSFT_PtRh. Starting with a comparison of the NSC ( Figure 3A) of the bare Pt-coated and that of mixed Pt-Rh coated samples, a general shift of the adsorption curve by 50 °C towards higher temperatures occurs due to the addition of 0.125 wt. % Rh. As a result, the Pt-Rh combination shows a higher NSC than bare Pt in the range of 300-450 °C. Additionally, the maximum storage temperature increases from 142 µmol/g for LSFT_Pt (250 °C) to 151 µmol/g for LSFT_PtRh (300 °C).The expectation that N2O formation for LSFT_PtRh is decreased with slightly increasing the NSC is refuted by the results shown in Figure 3B.
Here it can be seen that with additional Rh coating, a higher N2O formation over the entire temperature range occurs for LSFT_PtRh. While the average amounts of N2O formed in the moderate to high-temperature range continue to converge, ending at values of 1-3 ppm, the difference in the low-temperature range is substantial. While the difference is about 15 ppm at 200 °C, it is reduced to 2 ppm at temperatures above 250 °C.

Performance Study of a Mixed Composition of 80 vol. % LNT (Bottom) and 20 vol. % Oxidation Catalyst (Top)
The second approach to avoid N2O evolution deals with the oxidation of propylene before it reaches the LNT catalyst. Here, the choice for the oxidation catalyst was set to 2 wt. % Pd/20 wt. % CeO2/Al2O3 (Ce_Pd) as this composition has advantageous properties for both oxidation of propylene and CO. Next to the outstanding and well-established oxidation properties of Pd, CeO2 is expected to promote the oxidation ability of the catalyst due to the easy change between Ce 3+ and Ce 4+ oxidation states [46,47]. Furthermore, former studies on combined Pd and CeO2-based oxidation catalysts showed a more homogeneously dispersed palladium coating and thus sinter stable Pd particles [43] and a performance shift to lower temperatures depending on the coated Pd amount [41].  The second approach to avoid N 2 O evolution deals with the oxidation of propylene before it reaches the LNT catalyst. Here, the choice for the oxidation catalyst was set to 2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 (Ce_Pd) as this composition has advantageous properties for both oxidation of propylene and CO. Next to the outstanding and well-established oxidation properties of Pd, CeO 2 is expected to promote the oxidation ability of the catalyst due to the easy change between Ce 3+ and Ce 4+ oxidation states [46,47]. Furthermore, former studies on combined Pd and CeO 2 -based oxidation catalysts showed a more homogeneously dispersed palladium coating and thus sinter stable Pd particles [43] and a performance shift to lower temperatures depending on the coated Pd amount [41].
In this experiment first, the activity of the Ce_Pd catalyst composition for CO and C 3 H 6 oxidation has been investigated in separate (w/o_CO and w/o_C 3 H 6 ) and co-feed (full lean gas mixture; see Table 1) experiments and the resulting light-off curves are shown in Figure 4. The CO oxidation occurs in the temperature range of 70-185 • C for the separate (w/o_C 3 H 6 ) feed mixture and the propylene oxidation is observed between 185 and 280 • C for the feed gas mixture without CO. With these feed conditions the T 50 values are estimated to be 157 and 252 • C for CO and propylene, respectively. Due to the co-feed of 500 ppm CO and 200 ppm C 3 H 6 , the light-off curves of both reducing gases have converged and show similar start and end temperatures for the oxidation curves. In principle, the light-off curve of CO was shifted on average by 40 • C to higher temperatures (T 50 (CO) = 191 • C), while the C 3 H 6 light-off curve was shifted by around 20 • C to lower temperatures (T 50 (C 3 H 6 ) = 232 • C) than in the separate feed experiment. With that, the results of the full lean gas feed experiment indicate a kind of promoting role of CO for C 3 H 6 oxidation and an improved performance in the low-temperature range of 200-300 • C.  Table 1. In this experiment first, the activity of the Ce_Pd catalyst composition for CO and C3H6 oxidation has been investigated in separate (w/o_CO and w/o_C3H6) and co-feed (full lean gas mixture; see Table 1) experiments and the resulting light-off curves are shown in Figure 4. The CO oxidation occurs in the temperature range of 70-185 °C for the separate (w/o_C3H6) feed mixture and the propylene oxidation is observed between 185 and 280 °C for the feed gas mixture without CO. With these feed conditions the T50 values are estimated to be 157 and 252 °C for CO and propylene, respectively. Due to the co-feed of 500 ppm CO and 200 ppm C3H6, the light-off curves of both reducing gases have converged and show similar start and end temperatures for the oxidation curves. In principle, the light-off curve of CO was shifted on average by 40 °C to higher temperatures (T50(CO) = 191 °C), while the C3H6 light-off curve was shifted by around 20 °C to lower temperatures (T50(C3H6) = 232 °C) than in the separate feed experiment. With that, the results of the full lean gas feed experiment indicate a kind of promoting role of CO for C3H6 oxidation and an improved performance in the low-temperature range of 200-300 °C.
In the following, the reactor filling changed to a layered two component system. Therefore, 2 mL (20 vol. %) Ce_Pd was fixed on top of 8 mL NOx storage active infiltration composite (see Figure 5A), keeping the reactor filling constant at a total filling volume of 10 mL, to effectively change the gas mixture by the conversion of propylene and CO to CO2 and water before entering the LNT catalyst (80 vol. %). Figure 5B reveals the temperature-dependent NSC for the single LSFZ_Pt sample and the layered Ce_Pd-LSFZ_Pt system. The comparison shows that the layered catalyst system keeps the general Gaussian-type curve progression with a maximum at 250 °C as already observed for the pure LSFZ_Pt sample. The maximal performance for bare LNT and the layered catalyst system are similar, as the average values are 171 µmol/g and 176 µmol/g. However, by the addition of the oxidation catalyst the stored NOx amount is increased at temperatures above 250 °C until 400 °C, resulting in a performance enhancement of 15-22%. Due to the increase an almost equal performance at 300 °C (169 µmol/g) In the following, the reactor filling changed to a layered two component system. Therefore, 2 mL (20 vol.%) Ce_Pd was fixed on top of 8 mL NO x storage active infiltration composite (see Figure 5A), keeping the reactor filling constant at a total filling volume of 10 mL, to effectively change the gas mixture by the conversion of propylene and CO to CO 2 and water before entering the LNT catalyst (80 vol.%). Figure 5B reveals the temperature-dependent NSC for the single LSFZ_Pt sample and the layered Ce_Pd-LSFZ_Pt system. The comparison shows that the layered catalyst system keeps the general Gaussian-type curve progression with a maximum at 250 • C as already observed for the pure LSFZ_Pt sample. The maximal performance for bare LNT and the layered catalyst system are similar, as the average values are 171 µmol/g and 176 µmol/g. However, by the addition of the oxidation catalyst the stored NO x amount is increased at temperatures above 250 • C until 400 • C, resulting in a performance enhancement of 15-22%. Due to the increase an almost equal performance at 300 • C (169 µmol/g) than for 250 • C occurs thereby broadening the high-performance window of the LNT catalyst. In conjunction with the increased NO x storage capacity, a decrease in the formed N 2 O amount can be observed in Figure 5C due to the addition of the oxidation catalyst. The decrease occurs in the range of 200-350 • C and accounts for between 60% (200 • C) and 13% at the maximum evolution temperature.
Catalysts 2021, 11, 917 8 of 16 than for 250 °C occurs thereby broadening the high-performance window of the LNT catalyst. In conjunction with the increased NOx storage capacity, a decrease in the formed N2O amount can be observed in Figure 5C due to the addition of the oxidation catalyst. The decrease occurs in the range of 200-350 °C and accounts for between 60% (200 °C) and 13% at the maximum evolution temperature.

Catalyst Characterisation
The infiltration composites Pt/La0.5Sr0.5Fe0.5Ti0.5O3-δ/Al2O3 (LSFT_Pt), Pt/La0.5Sr0.5Fe0.75Nb0.25O3-δ/Al2O3 (LSFN_Pt) and Pt/La0.5Sr0.5Fe0.5Zr0.5O3-δ/Al2O3 (LSFZ_Pt) were already introduced as potential LNT materials in a previous work and extensively characterised by BET, X-ray diffraction (XRD), selected area electron diffraction (SAED) and transmission electron microscopy (TEM) [44]. The analysis of the specific surface areas (SSA) of the infiltration composites showed that with the use of the incipient wetness impregnation method the loss of the SSA, starting from γ-Al2O3, could be kept low despite the multi-step synthesis route. As a result, an SSA between 99 and 106 m 2 /g was measured for the three nanocomposite compositions at a perovskite loading of 20 wt. % (equivalent prepared Pt/BaO/Al2O3 displayed an SSA of 80 m 2 /g). The varying specific surface areas depending on the substituting B-site cations in the perovskites could be explained by the different crystallisation behaviour due to the material composition. SAED measurements performed on the freshly prepared samples showed that the degree of crystallisation of the perovskites decreased in the order Ti > Nb > Zr, thus contradicting the decreasing order of the SSA. Furthermore, the successful in situ crystallisation of the expected perovskite compositions could be confirmed by additional ACOM-TEM images besides the SAED measurements. The results identified the infiltration composites as an ultrafine, uniform mixture of platinum, perovskite and alumina particles.
As the titanium-containing infiltration composite has shown the best crystallisation behaviour and NSC of all investigated infiltration composite mixtures, this material composition was chosen to investigate the impact of the additional 0.125 wt.% Rh coating. The resulting Pt/Rh/La0.5Sr0.5Fe0.5Ti0.5O3-δ/Al2O3 (LSFT_PtRh) specimen shows the same wellmatched perovskite composition due to the synthesis procedure using a perovskite precursor solution and has an SSA of 104 m 2 /g. Thus, the impact of the Rh-coating is marginal in the case of the morphological material properties.

Catalyst Characterisation
The infiltration composites Pt/La 0.5 Sr 0.5 Fe 0.5 Ti 0.5 O 3-δ /Al 2 O 3 (LSFT_Pt), Pt/La 0.5 Sr 0.5 Fe 0.75 Nb 0.25 O 3-δ /Al 2 O 3 (LSFN_Pt) and Pt/La 0.5 Sr 0.5 Fe 0.5 Zr 0.5 O 3-δ /Al 2 O 3 (LSFZ_Pt) were already introduced as potential LNT materials in a previous work and extensively characterised by BET, X-ray diffraction (XRD), selected area electron diffraction (SAED) and transmission electron microscopy (TEM) [44]. The analysis of the specific surface areas (SSA) of the infiltration composites showed that with the use of the incipient wetness impregnation method the loss of the SSA, starting from γ-Al 2 O 3 , could be kept low despite the multi-step synthesis route. As a result, an SSA between 99 and 106 m 2 /g was measured for the three nanocomposite compositions at a perovskite loading of 20 wt.% (equivalent prepared Pt/BaO/Al 2 O 3 displayed an SSA of 80 m 2 /g). The varying specific surface areas depending on the substituting B-site cations in the perovskites could be explained by the different crystallisation behaviour due to the material composition. SAED measurements performed on the freshly prepared samples showed that the degree of crystallisation of the perovskites decreased in the order Ti > Nb > Zr, thus contradicting the decreasing order of the SSA. Furthermore, the successful in situ crystallisation of the expected perovskite compositions could be confirmed by additional ACOM-TEM images besides the SAED measurements. The results identified the infiltration composites as an ultrafine, uniform mixture of platinum, perovskite and alumina particles.
As the titanium-containing infiltration composite has shown the best crystallisation behaviour and NSC of all investigated infiltration composite mixtures, this material composition was chosen to investigate the impact of the additional 0.125 wt.% Rh coating. The resulting Pt/Rh/La 0.5 Sr 0.5 Fe 0.5 Ti 0.5 O 3-δ /Al 2 O 3 (LSFT_PtRh) specimen shows the same well-matched perovskite composition due to the synthesis procedure using a perovskite precursor solution and has an SSA of 104 m 2 /g. Thus, the impact of the Rh-coating is marginal in the case of the morphological material properties. specimen can be observed. A comparable behaviour pattern was also observed in the NSC experiments [44]. Assuming that the weaker NSC and N 2 O evolutions depend on the perovskite composition, this behaviour could be due to a slowed ad-species spill-over from the precious metal to the storage material. Hodjati et al. [48] postulated a reversible opening mechanism for the storage of NO x on ABO 3 perovskites (A = Ba, Ca, Sr; B = Sn, Ti, Zr) according to Equations (6) and (7).

N 2 O Formation Investigations
In this case, the incorporation of the niobium cations may have led to a more stable perovskite lattice compared to Ti 4+ (0.74 Å) [49] and Zr 4+ (0.86 Å) [49] due to the equal ionic radii of Fe 3+ (0.78 Å) [49] and Nb 5+ (0.78Å) [49]. The high redox stability of niobium oxides at temperatures below 900-1000 • C discussed by Gervasini [50] could also contribute to a lower crystal lattice opening probability. Under this assumption, the Nb-containing perovskite would be less reactive with respect to NO x uptake and the ad-species would have to dwell longer on the platinum, thus inhibiting the overall activity of the precious metal with respect to NO x storage and the N 2 O forming side reaction. By a following survey concerning different feed gas compositions (Figure 2), the underlying N 2 O formation mechanism for our infiltration composites was clearly identified as a C 3 H 6 -SCR. Interestingly, a kind of inhibiting effect is observed, when CO and C 3 H 6 are dosed together. The resulting lower N 2 O evolution at 200 and 250 • C might be explained by an increased availability of active (reduced) Pt sites due to an additional PGMO x to PGM 0 transformation by CO leading to more likely formed N*-and O*-ad-species and so a lower N 2 O formation (see Equations (3) and (5)).
To the best of our knowledge, only Masdrag et al. [23] have so far investigated the influence of different reducing agents on N 2 O formation on an LNT catalyst (Pt/CeO 2 -ZrO 2 ) in lean operation. In their study, they defined a temperature-dependent reductive influence with CO > C 3 H 6 ≈ 0 at 200 • C and C 3 H 6 > CO ≈ 0 at 300 • C in lean operation mode. [23].
Hence, the catalyst investigations in our study and in the study of Masdrag et al. were conducted entirely in a net oxidising atmosphere (see Table 1), the Pt/La 0.5 Sr 0.5 Fe x M 1−x O 3−δ / Al 2 O 3 (M = Nb, Ti, Zr) infiltration composites show a pronounced N 2 O formation by the C 3 H 6 -SCR reaction at 200 • C (see Figure 1), while Pt/CeO 2 -ZrO 2 was completely inactive (reference [23], Figure 7A,B). It seems that the Pt particles (2.5 wt.%) in the composites can be reduced more effectively by C 3 H 6 at low temperatures and O 2 excess than in the LNT catalyst studied by Masdrag et al. (2.12 wt.% Pt, CeO 2 content unknown) [23]. This difference can most likely be explained by the influence of the support materials and/or the Pt particle sizes (Pt particle size Masdrag et al.: 6.2 nm, this study: 2-4 nm).
Since the focus of C 3 H 6 -SCR catalyst studies is mostly on NO x conversion and N 2 selectivity, no literature is known at this time that explicitly addresses the undesired N 2 O formation as a function of different support materials and/or PGM particle size.

Attempts to Reduce N 2 O Evolution
The first attempt to reduce the N 2 O formation on LNT catalysts under lean conditions was the usage of an additional Rh coating on the infiltration composites. By the addition of 0.125 wt.% Rh the NO x storage capacity above 300 • C is slightly improved ( Figure 3A). Though an auxiliary effect of the Rh coating at higher temperatures seems to be plausible, as the studies of Kubiak et al. [51], Andonova et al. [52] and Castoldi et al. [53] showed appreciable NO oxidation abilities and NO x storage capacities for Rh-based BaO/Al 2 O 3 catalysts above 350 • C. Figure 3B compares the N 2 O evolution on LSFT_Pt and LSFT_PtRh. The observed N 2 O formation increase caused by the Rh addition could be related to an interplay between the generally high light-off temperatures for NO oxidation and N 2 O decomposition of Rh-based catalysts on the one hand, and the slight NO x adsorption capacity at low temperatures proceeding via the nitrite route on the other hand. The latter usually proceeds by a subsequent transfer of activated oxygen from the PGM to the storage material, allowing the adsorption of NO in the form of nitrite adsorbents. Consequently, Kubiak et al. [51] observed a low NO x adsorption of 0.106 mmol/g for Rh/BaO/Al 2 O 3 at 150 C in 1000 ppm NO + 3 vol.% O 2 balanced with helium. Due to the feed gas composition used here, which contains reducing agents during the lean phase (propylene and CO), it can be assumed that the nitrite pathway is inhibited by the partial reduction of the noble metal sites. Thus, the available amount of activated oxygen would decrease and consequently lead to the recombination of the weakly bound NO-ad species at the PGM and storage material sites and the formation of N 2 O. However, the dramatic increase of about 500% of evolving N 2 O at 200 • C cannot only be related to additional N 2 O formation over 0.125 wt.% Rh, hence indicating a strong impeding interaction between the Rh and Pt concerning NO oxidation at low temperatures in the presence of reductives.
The second attempt deals with the use of a layered oxidation and LNT (ratio 1:4) catalyst as illustrated in Figure 5A. The chosen oxidation catalyst 2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 (Ce_Pd) showed a promising oxidation ability for propylene and CO in separate and co-feed experiments (Figure 4), thus seeming able to suppress N 2 O production via SCR reaction. In general, Ce_Pd showed a similar performance compared to the results of Shen et al. [41] and Burch and Millington [30], who observed a complete activation of their investigated Pd/CeO 2 -ZrO 2 and Pd/Al 2 O 3 catalysts for propylene oxidation in a temperature range of around 125 • C (T 50 = 203 • C) and 100 • C (T 50 = 250 • C). The oxidation activity for both-CO and C 3 H 6 -raises from 0 to 100% by a temperature increase of around 100 • C in all feed gas experiments. Interestingly, due to the co-feed of CO and propylene, a convergence of the light-off curves (LO) was observed, shifting the CO-LO to higher and the C 3 H 6 -LO to lower temperatures. Similar experiments on Pt/Al 2 O 3 [54][55][56] or Pt/CeO 2 [57] oxidation catalysts have shown that under co-feed conditions both LO curves commonly shift to higher temperatures as the two reducing gases compete for the active sites and thus inhibit the oxidation of the other gas. Only Lang et al. [58] observed a similar behaviour for 1 wt.% Pd/CeO 2 -ZrO 2 as for the herein investigated 2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 with a more pronounced approximation of the light-off curves for CO and propylene (T 50 (CO) shift from 164 to 229 • C and T 50 (C 3 H 6 ) shift from 226 to 232 • C) and an improved C 3 H 6 conversion at temperatures above~240 • C (T 85 (C 3 H 6 ) ≈ 261 • C shifts to~250 • C) due to the CO and propylene co-feed. Lang et al. [58] explained this behaviour by a possible alternating oxidation mechanism and/or lower activation barriers due to the available ceria compared to alumina-based catalysts (CO reaction order Pd/Al 2 O 3 : 1st order and Pd/CeO 2 -ZrO 2 : 0th order). Related to the mentioned literature, we presume a combination of both theories to be the reason for the observed curve shifts due to the CO and C 3 H 6 co-feed. The observed shift of the CO-LO to higher temperatures seems to be caused by the general inhibition of the CO oxidation by competing propylene molecules and/or oxidation intermediates for the active catalyst sites. Instead, the promotion of the C 3 H 6 oxidation by co-dosed CO might be explained by either an alternating oxidation mechanism dependent on the Pt supporting oxides or by more available Pt 0 sites due to the prior or simultaneously elapsing CO oxidation consuming the oxygen species bound on the precious metal (PtO x ) [56]. Hazlett et al. [56] already observed that the addition of propylene to the feed gas resulted in a strong increase in multiple precious metal-bound surface species (triple bound adsorbates, carbonyl and dicarbonyl species) on Pt-and Pdbased alumina catalysts competing with the single precious metal-bound CO. In this case, C 3 H 6 needs, on average, three times the amount of available active catalyst sites (Pt 0 ) compared to CO for the oxidation reaction, which is commonly described as a Langmuir-Hinshelwood mechanism using surface co-adsorbed reactants [54]. However, the results in Figure 4 reveal that the catalyst offers the necessary oxidation activity in the desired temperature range of 200-500 • C to be used as an oxidation catalyst top layer for our approach to reduce C 3 H 6 concentrations in the feed gas of the LNT catalyst. The resulting NO x storage and N 2 O evolution curves for the layered system are compared to the bare LNT material (LSFZ_Pt) in Figure 5B,C. Due to the addition of Ce_Pd an increase of 15-22% of the NO x storage capacity (NSC) in the temperature range of 250-400 • C is obtained. In parallel to the NSC enhancement, a N 2 O formation decrease can be observed at 200 to 350 • C. The results thus show a N 2 O evolution diminishment of 13-60% in the general N 2 O formation window. Consequently, an oxidation catalyst layered on top of the LNT seems to be a suitable means to suppress N 2 O formation during the lean storage phase of the LNT due to a competing propylene-SCR. Nevertheless, nitrous oxide emissions were not reduced to a zero level which might be caused by the chosen configuration with only a small oxidation catalyst layer on top of the LNT materials. To enhance the effect on NSC and N 2 O evolution, the optimal thickness of the oxidation catalyst has to be found.

Lean NO x Trap Catalyst Preparation
All investigated LNT storage materials were synthesised by a four-step synthesis pathway using incipient wet impregnation (IWI) processes ending up with a milling and granule formation step to give the materials the necessary geometry for the NO x adsorption tests. A detailed description of each single preparation step can be found elsewhere [44].
The platinised titanium perovskite-type oxide-based catalyst was further impregnated with 0.125 wt.% rhodium by another IWI coating step. Therefore 50 g of the LSFT_Pt composite was infiltrated with a solution of 0.174 g Rh(NO 3 ) 3 hydrate (36 wt.% Rh content, Merck, Darmstadt, Germany) dissolved in 60 mL distilled water. The wet powder was dried at 140 • C and calcined at 550 • C for 1 h using a heating rate of 175 • C/h. Afterwards, an analogous milling and granulation procedure as for the Pt only containing infiltration composites was carried out to examine their NSC and N 2 O evolution at lean conditions. The composite 0.125 wt.% Rh-2.5 wt.% Pt on 20 wt.% La 0.5 Sr 0.5 Fe 0.5 Ti 0.5 O 3−δ /Al 2 O 3 , was abbreviated as before but with the additional element symbol of rhodium resulting in LSFT_PtRh.
Additionally, a comparable 2.5 wt.% Pt/20 wt.% BaO/Al 2 O 3 sample, condensed as BaO_Pt, was synthesised by IWI. A 100 g synthesis batch was prepared by impregnating  2 were dissolved in 120 mL distilled water. After impregnation and homogenisation, the wet powder was dried at 120 • C overnight and calcined at 700 • C for 3 h. Similar to all former specimen, the composite powder was additionally coated with 2.5 wt.% Pt, grounded and formed to granules.

Oxidation Catalyst Preparation
50 g of the used 2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 oxidation catalyst was prepared via incipient wetness impregnation. First, 28 g of an aqueous (NH 3 ) 4 Pd(NO 3 ) 2 solution (10 wt.% Pd content, Alfa Aesar, Karlsruhe, Germany) were further diluted with 32 mL distilled water. The created solution was infiltrated on 49 g of commercial 20 wt.% ceria containing alumina powder (Puralox SCFa-160/Ce20, Sasol, Hamburg, Germany). After drying at 140 • C overnight, a final calcination was performed at 550 • C for 1 h using a heating rate of 175 • C/h. The catalyst powder was milled down and granulated in the way as described before. The catalyst is abbreviated as Ce_Pd.

NO x Adsorption and N 2 O Evolution Measurements as a Function of Temperature
Temperature-dependent NO x adsorption measurements were performed in a laboratory exhaust gas test bench under constant lean conditions representing the exhaust gas composition on a lean burn engine at a lambda situation of λ = 1.5. In general, this was realised by the lean gas mixture (see Table 1) and a gas hourly space velocity set to 80,000 h −1 . In order to examine the dependence of the N 2 O formation on the dosed reductives, additional NO x storage measurements with altered gas compositions, either without CO (w/o_CO) or without propylene (w/o_C 3 H 6 ) were dosed (see Table 1). A more detailed description of the used test bench could be found elsewhere [44]. In brief, all gas concentrations were controlled by dynamic gas flow controller (MFC) of type 5850 from Brooks Instrument, LLC (Hatfield, PA, USA). The inert carrier gas stream of N 2 , O 2 and CO 2 was heated up to 200 • C before the water vapour was added, dosed by a gear type pump from GATHER Industrie (Wülfrath, Germany) and vaporised by an evaporator from LINSEIS Messgeräte (Selb, Germany). The gas mixture was passed through a furnace (DLERH, Horst, Lorsch, Germany) and heated to the respective adsorption temperature (200-450 • C in 50 • C steps) in advance. The further gas flow towards the reactor with the sample was carried out in heated pipes and shortly before entering the reactor the reactive gases NO, CO and propylene were mixed into the carrier gas stream. To guarantee a constant temperature during the measurement, the titanium reactor was wrapped with a heat jacket (Horst, Lorsch, Germany) and the temperature was controlled by thermocouples (NiCrNi) at two points in the gas stream. The first thermocouple was placed before the reactor entrance and the second directly within the sample. The sample holder was filled with 10 mL of granules of a sole infiltration composite (see Figure 6) or a top-layered 20 vol.% oxidation catalyst on 80 vol.% infiltration composite setting (see Figure 5A). In both cases, the catalysts were covered up-and downstream with 100 mg quartz wool and a titanium sieve. The outlet gas composition was analysed using a Fourier-transformed infrared spectrometer (FT-IR) of the type MultiGas 2030 from MKS instruments to monitor the formation of C-and N-containing components (e.g., NO, NO 2 , N 2 O, propylene, CO) during the measurements.
vol. % oxidation catalyst on 80 vol. % infiltration composite setting (see Figure 5A). In both cases, the catalysts were covered up-and downstream with 100 mg quartz wool and a titanium sieve. The outlet gas composition was analysed using a Fourier-transformed infrared spectrometer (FT-IR) of the type MultiGas 2030 from MKS instruments to monitor the formation of C-and N-containing components (e.g., NO, NO2, N2O, propylene, CO) during the measurements. The NOx storage capacity was measured between 200 and 450 °C in 50 °C steps by saturating the sample with NOx. To control the gas composition before and after this adsorption phase, the gas mixture was switched via solenoid valves through a bypass, which directed the gas around the reactor. Afterwards, the stored NOx was desorbed by heating up the sample to 550 °C within the same gas mixture but without NO dosing. Finally, the specific molar storage capacity was calculated from the NOx desorption peak.

Propylene and CO Oxidation Measurements on 2 wt. % Pd/20 wt. % CeO2/Al2O3 as a Function of Temperature
The measurements of the propylene and CO oxidising ability of the Ce20_Pd catalyst were carried out on the same laboratory gas test bench as the NOx storage measurements (see Figure 6) at a gas hourly space velocity of 80000 h −1 with three different feed gas compositions shown in Table 1. For the measurements, 2 mL of the catalyst was filled into the reactor and the gas mixtures were constantly passed through the sample. For propylene the temperature range of 165-295 °C and for CO the range between 60 and 215 °C was investigated. The heating rate in all cases was 2 °C/min. The cycles were repeated three times each. The resulting conversion curves are calculated with the following equations: CO conversion [%] = (COinlet − COoutlet)/COinlet × 100 (8) C3H6 conversion [%] = (C3H6inlet − C3H6outlet)/C3H6inlet × 100 (9) The NO x storage capacity was measured between 200 and 450 • C in 50 • C steps by saturating the sample with NO x . To control the gas composition before and after this adsorption phase, the gas mixture was switched via solenoid valves through a bypass, which directed the gas around the reactor. Afterwards, the stored NO x was desorbed by heating up the sample to 550 • C within the same gas mixture but without NO dosing. Finally, the specific molar storage capacity was calculated from the NO x desorption peak. The measurements of the propylene and CO oxidising ability of the Ce20_Pd catalyst were carried out on the same laboratory gas test bench as the NO x storage measurements (see Figure 6) at a gas hourly space velocity of 80,000 h −1 with three different feed gas compositions shown in Table 1. For the measurements, 2 mL of the catalyst was filled into the reactor and the gas mixtures were constantly passed through the sample. For propylene the temperature range of 165-295 • C and for CO the range between 60 and 215 • C was investigated. The heating rate in all cases was 2 • C/min. The cycles were repeated three times each. The resulting conversion curves are calculated with the following equations:

Conclusions
This study focuses on the undesirable N 2 O formation on newly developed Pt/La 0.5 Sr 0.5 Fe 1−x M x O 3 /Al 2 O 3 (M = Nb, Ti, Zr) lean NO x trap (LNT) catalysts under laboratory gas test bench conditions. In addition to the investigations on the formation mechanism of the undesired by-product, which caused a decreased NO x storage capacity (NSC) for the LNT materials, attempts were made to reduce these emissions. To this end, two approaches were pursued, firstly an additive Rh coating on the LNT catalyst and secondly an oxidation catalyst-LNT layered system.
The mechanism of N 2 O evolution was identified as a low-temperature C 3 H 6 -SCR. Nitrous oxide formation occurred due to the competing partial reduction of PtO x to Pt 0 by C 3 H 6 during the general NO x oxidation and storage procedure on LNT catalysts in net oxidising atmosphere. The investigations showed that even low concentrations of propylene (200 ppm), as generally present in lean exhaust gases, lead to high N 2 O formation of 20 ppm on average at low temperatures.
The additional Rh coating on the Pt/La 0.5 Sr 0.5 Fe 0.5 Ti 0.5 O 3 /Al 2 O 3 nano composite resulted in a dramatic increase in N 2 O formation in the range 200-250 • C. The comparison with the platinised nano composite sample leads to the assumption that this result can only be related to an additional N 2 O formation at the Rh particles. It seems that the Rh nitrites route for NO x storage is prevented by Pt 0 and Rh 0 particles. Thus the necessary oxygen spill-over is omitted causing a recombination of NO* ad-species. Although the additional Rh coating is unsuitable for the reduction of N 2 O formation, it should be mentioned that a slightly better NSC for the Pt and Rh coated sample was observed in the range of 300-450 • C.
In contrast, the catalyst installation with an oxidation catalyst connected upstream of the LNT showed promising results. With a replacement of 20 vol.% of the LNT by the oxidation catalyst (2 wt.% Pd/20 wt.% CeO 2 /Al 2 O 3 ), 13-60% less N 2 O emissions were measured in the low temperature range. The results thus show that the oxidation of the propylene is the most effective way to reduce N 2 O formation. Nevertheless, nitrous oxide formation could not yet be completely suppressed, which may be due to the catalyst configuration with a low oxidation catalyst content. To further improve the effectiveness of the oxidation catalyst, the palladium content and the layer thickness have to be optimised.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the complexity of the analysis which needs guidance for reproduction.