Formation of nitrous oxide over Pt-Pd oxidation catalysts: Secondary emissions by interaction of hydrocarbons and nitric oxide

The interaction of hydrocarbons (HC) and nitric oxide (NO) over noble metal catalysts for exhaust gas after-treatment of lean-operated combustion engines can lead to secondary emissions, namely the formation of nitrous oxide (N 2 O), which is a strong greenhouse gas calling for N 2 O reduction concepts. By means of a series of light-off tests over state-of-the-art Pt-Pd oxidation catalysts, this study identifies the most critical catalyst operation regimes that should be avoided in order to minimize N 2 O levels. Especially unsaturated HCs react with NO to form significant amounts of N 2 O between 150 ◦ C and 350 ◦ C; an increasing HC/NOx ratio generally promotes N 2 O formation, whereas the NO oxidation reaction is increasingly inhibited. Since low space velocities and fast catalyst heating allow for minimizing N 2 O levels, active heating of catalytic converters during cold start and phases of low exhaust gas temperatures may efficiently reduce the formation of N 2 O in real-world applications.


Introduction
After decades of research on exhaust gas after-treatment systems, modern catalytic converters ensure an efficient emission abatement for combustion engines running on diesel, gasoline, and natural gas [1][2][3][4].While the first after-treatment systems comprised only a single oxidation catalyst, state-of-the art exhaust tailpipes combine several converters and technologies to account for conversion of all exhaust gas species and continuously tightening emission legislation calls for even more complex after-treatment systems.In addition to economic considerations, especially the limited space in mobile applications drives the development of innovative compact catalytic converters that possibly merge different technologies in a single device.This compact design results in competitive adsorption on the catalyst surface and promotes interactions of the many different exhaust gas species, which can ultimately lead to the formation of additional pollutants, so-called secondary emissions, such as ammonia (NH 3 ) [5][6][7], nitrous oxide (N 2 O) [8][9][10], and even very toxic compounds like hydrogen cyanide (HCN) [11][12][13].In contrast to earlier regulations, current and upcoming legislation will introduce very stringent limits for secondary emissions, among which N 2 O is considered particularly critical, since it has a more than 300 times higher global warming potential than that of carbon dioxide (CO 2 ) and exhibits an ozone depleting effect [14][15][16].
N 2 O is a pollutant that is relevant for all kinds of combustion engine applications [16], as it forms as an undesired byproduct in stoichiometric exhausts over three-way catalysts (TWC) [7,[17][18][19] and in the exhaust tailpipe of lean-burn engines alike [14,20].In the latter case, especially deNOx-converters for selective catalytic reduction (SCR) of nitrogen oxides (NOx) using ammonia (NH 3 ) [9,[21][22][23] or hydrocarbons (HCs) [20,[24][25][26] as reducing agent are prone to N 2 O evolution.The interaction of nitric oxide (NO) and HCs over noble metal based catalysts was subject to several mechanistic studies aiming at unraveling the complex reaction pathways causing formation of N 2 O during deNOx reactions over oxidation catalysts.In short, three different mechanisms are considered: First, surface isocyanate species (R-NCO) were reported as the most relevant intermediates during NOx reduction via HC-SCR, whose reaction with NO ultimately results in N 2 O formation [27][28][29].Second, organic nitro compounds or derivatives and related species were suggested as most relevant intermediate compounds [30][31][32][33], e.g. with N 2 O (and ketone) formation from oximes (R 1 R 2 -C--N-OH) [33].Third, several studies [24,[34][35][36][37] suggested dissociative NO adsorption on the noble metal surface and subsequent oxygen reduction by hydrocarbon species as mechanism for HC-deNOx, with N 2 O forming by a combination of an adsorbed (undissociated) NO molecule and an adsorbed N atom [34,36].
Especially the catalyst operation conditions influence the emerging N 2 O levels during HC-SCR over noble metal based catalysts [24,25,38,39], aside from catalyst formulation [35,40] and material-related catalyst properties like noble metal particle size [41], alloying state [42], or sulfation [43] that may change due to degradation, hereby causing changes in the catalytic activity for hydrocarbon, carbon monoxide (CO), and NOx conversion.Since in the context of HC-SCR the type of hydrocarbon used as reductant plays a particular role, a wide variety of hydrocarbons has been suggested for NOx removal.Among other species, particularly ethylene (C 2 H 4 ) [35], propylene (C 3 H 6 ) [26,36,44], propane (C 3 H 8 ) [36,40], n-octane (C 8 H 18 ) [45], and even aromatic compounds like toluene (C 7 H 8 ) [46] have been tested for HC-SCR, with the latter one being proposed as reductant that avoids N 2 O production over noble metal catalysts.In general, unsaturated hydrocarbons are more reactive, which allows for significant HC-NOx interactions already at comparably low temperatures and an onset of N 2 O formation around 130-200 • C, depending on the catalyst formulation [26,35,40,47,48].In contrast, saturated hydrocarbons with their lower reactivity are typically activated at slightly higher temperatures, which consequently shifts the onset of NOx-reduction as well [36,40,47].Moreover, both short-chain species like C 3 H 6 as well as long-chain species such as dodecane (C 12 H 26 ), are known to inhibit NO oxidation to nitrogen dioxide (NO 2 ) over Pt-Pd oxidation catalysts, since during hydrocarbon oxidation NO 2 is preferentially consumed compared to oxygen (O 2 ) [49,50].In contrast, hydrocarbons can also impede an overoxidation of the platinum surface; the consumption of surface oxygen by hydrocarbons can be particularly beneficial for NO oxidation [50] that preferentially takes place over reduced noble metal surface sites [51,52].These observations are also highly relevant in the context of SCR.As already mentioned above, hydrocarbon and NO conversion rates as well as possible byproduct formation strongly depend on the adsorbed surface species, which in turn are heavily influenced by the gas species concentrations and by the species residence time or in terms of the macroscopic operation conditions the gas hourly space velocity (GHSV), respectively.
A successful suppression of N 2 O formation requires both, a detailed knowledge of the formation pathways as well as a profound database on the conditions that support N 2 O formation.Hence, our present study revisits the topic of N 2 O formation caused by HC-NOx interactions by a comprehensive measurement campaign that comprises kinetic tests under a wide variety of conditions of two different diesel oxidation catalysts (DOC) that are used for heavy-duty applications and passenger car systems, respectively.Based on our results we consider the effect of hydrocarbon type (unsaturated vs. saturated), temperature, space velocity, and the HC/NOx ratio as the most relevant factors on N 2 O formation.The final objective of this study is to provide guidance for optimization of operating conditions to minimize N 2 O emissions.

Catalyst materials
Within this study, two different commercial diesel oxidation catalysts purchased from commercial sources served as materials for catalytic testing.Both samples consist of a cordierite substrate (400 cpsi) that is coated with a Pt-Pd washcoat (noble metal ratio 3:1).The first catalyst sample (DOC-1) exhibits a total noble metal loading of 12 g/ft 3  (corresponding to approx.0.42 g/L) and is representative for an aftertreatment system used in commercial diesel vehicles for heavy-duty applications, whereas the second catalyst sample (DOC-2) has a substantially higher noble metal loading of 100 g/ft 3 (corresponding to approx.3.53 g/L), which is representative for oxidation catalysts that are used in lean-burn passenger car systems and therefore light-duty applications.Herewith, we chose two different industrially relevant catalyst systems that were subject to extensive testing under welldefined conditions in a lab-scale catalyst testing bench.Cylindrical lab-size samples with a diameter d of 2.54 cm and a length l of 5.00 cm were obtained by cutting the full-size commercial catalysts.In order to ensure a stable catalyst during the performance tests, a degreening procedure that consisted of an exposure to 10 vol.-%H 2 O in air (60 000 h − 1 ) for 4 h at 700 • C was conducted for both samples prior to these tests.

Catalyst testing bench
The kinetic data were obtained using a synthetic exhaust gas labbench catalyst testing unit that was already described in detail in an earlier study [10].The setup essentially consists of a furnace-heated quartz glass tubular reactor (plug-flow configuration; inner diameter: 2.54 cm, length: 100 cm) that contains the lab-size catalyst sample (d = 2.54 cm, l = 5.00 cm).Mass flow controllers and a controlled evaporation mixer precisely dose reaction gases (originating from gas bottles) and steam (from a pressurized water tank).The only liquid hydrocarbon used in this study, C 12 H 26 , is dosed by bubbling nitrogen (N 2 ) gas through a heated stainless-steel tank filled with C 12 H 26 ; herein, the C 12 H 26 concentration is controlled by varying the inert N 2 flow and the temperature of the tank.Two type N thermocouples approx. 2 mm upand downstream the sample are used to control and monitor the inlet and outlet temperatures and a Fourier-transform infrared (FTIR) spectrometer (MG2030, MKS Instruments) continuously analyzes the effluent reaction gas stream.

Testing conditions and procedure
The experimental testing matrix comprises a systematic variation of the gas composition, the gas hourly space velocity (GHSV), and the reaction temperature at ambient pressure.A typical exhaust gas mixture relevant for lean-burn applications was used, excluding CO 2 that is considered playing a subordinate role [49], but including the unavoidable exhaust gas component water (H 2 O), which is known as a potential inhibitor due to a blockage of noble metal surface sites [53,54].In addition to a base gas feed of 150 ppm NO, 10 vol.-% O 2 , 10 vol.-%H 2 O and balance N 2 , a HC concentration of 150 ppm, 375 ppm, 750 ppm, 1500 ppm or 7500 ppm was added, which corresponds to a HC/NOx ratio of 1.0, 2.5, 5.0, 10.0, and 50.0, respectively (molar concentrations).Considering dynamic operation conditions of real-world applications, particularly in the mobility sector, these HC/NOx ratios replicate an industrially relevant range of conditions that can occur during engine operation.In order to cover different types of hydrocarbons that can be present in exhaust gasessaturated and unsaturated species, long-chain versus short-chain HCstests were performed using either C 2 H 4 , C 2 H 6 , C 3 H 6 , C 3 H 8 , or C 12 H 26 ; note that only one hydrocarbon was dosed at a time.Table 1 summarizes the gas composition variation.Using C 2 H 4 as model compound, GHSV values of 30 000 h − 1 , 60 000 h − 1 , and 90 000 h − 1 were chosen to uncover the influence of the space velocity on N 2 O formation.Performing light-off runs with a temperature ramp of 5 information on temperature and ramp rate effects.

Temperature-dependent evolution of gas species
Fig. 1 shows the temperature-dependent conversion and species concentration trends as obtained during a light-off test in 150 ppm NO, 750 ppm C 2 H 4 , 10 vol.-% O 2 , 10 vol.-%H 2 O, bal.N 2 (gas mix 1c, 5 • C/ min, GHSV = 60 000 h − 1 ).In line with earlier research [35], the simultaneous dosage of NO and C 2 H 4 in a lean gas mixture results in considerable N 2 O formation.Under the conditions chosen, the N 2 O concentration reaches a maximum of 48 ppm at 190 • C over DOC-1 and 57 ppm at 166 • C over DOC-2, respectively.Overall, DOC-2 exhibits a higher oxidation activity, both for NO and C 2 H 4 , which does not only result in higher N 2 O concentrations but also seems to broaden the temperature window in which N 2 O is formed.Regarding the trends of the gas phase species concentrations, the onset of C 2 H 4 and NO conversion gives rise to the N 2 O signal.Notably, N 2 O formation is favored over NO 2 formation below ~225 • C and the maximum N 2 O concentration is detected at the maximum NO conversion and once the catalyst fully converts C 2 H 4 .A further temperature increase during the light-off test results in a continuous shift of the product selectivity from N 2 O to NO 2 , which goes along with an overall declining NO conversion over both catalyst samples.Notably, the NO conversion is less than 60% over DOC-1 in the temperature window between 250 • C and 350 • C, whereas the NO conversion decreases from about 80% at 250 • C to approximately 62% at 350 • C over DOC-2.Simultaneously, the N 2 O level drops substantially and only 1 ppm and ~1.7 ppm N 2 O are found over DOC-1 and DOC-2, respectively, at the maximum temperature of 350 • C. Note, that no NH 3 formation was observed during the light-off tests.
From a mechanistic point of view, N 2 O evolves as a byproduct of NOx reduction by C 2 H 4 [25], with the global reaction Eqs. ( 1) and ( 2) being possible N 2 O formation pathways: (1) Hence, the simultaneous activation of both C 2 H 4 and NO is a prerequisite not only for the SCR of NO to N 2 , which would typically be desired in the context of NOx removal, but also for N 2 O formation, which explains the correlation of the NO and C 2 H 4 conversion curves with the N 2 O concentration curve in Fig. 1.In this respect, varying hydrocarbon inlet concentrations have an impact on both the temperature evolution due to the exothermicity of C 2 H 4 oxidation as well as the surface chemistry taking place on the noble metal.These aspects are discussed in detail in the following Section 3.2.

Influence of C 2 H 4 /NOx ratio
The data shown in Fig. 2 illustrate that the N 2 O emission profiles strongly depend on the C 2 H 4 /NOx ratio.In particular, increasing C 2 H 4 / NOx ratios promote N 2 O formation over a broad temperature range (Fig. 2a, b) and substantially increase the maximum N 2 O concentration (c(N 2 O) max ).At the same time, T(c(N 2 O) max ), which is the temperature where the maximum N 2 O concentration is observed, drops over both catalysts (Fig. 2c, d).Herein, T(c(N 2 O) max ) is the gas temperature measured at the catalyst inlet and therefore remains unaffected by the exothermicity of the hydrocarbon oxidation.However, the exothermicity could be one reason for the drop of T(c(N 2 O) max ) with increasing C 2 H 4 /NOx ratios: As illustrated in Fig. 2e and f  P. Lott et al. which exhibits a generally lower catalytic activity that can be attributed to its lower noble metal loading, achieved only 77% C 2 H 4 conversion at T(c(N 2 O) max ) = 179 • C, while producing 50 ppm N 2 O (Fig. S2).Furthermore, the overall trends depicted in Fig. 2d (as well as in Fig. S1d for a ramp rate of 15 • C/min) point to a minimum T(c(N 2 O) max ) of approximately 165 • C ( ± 3 • C).In particular, an increase of the C 2 H 4 /NOx ratio from 5.0 to 10.0 over DOC-2 does not result in a drop in T(c(N 2 O) max ) (Fig. 2d) as observed for DOC-1 (Fig. 2c).Under consideration of the more pronounced heat evolution that occurs when the hydrocarbon concentration increases (Fig. 2e, f for light-off tests with a ramp rate of 5 • C/min and Fig. S1e, f for those with 15 • C/min), our data do not suggest a simple direct correlation between the hydrocarbon conversion and the maximum N 2 O formation rate, but rather point to a more complex interplay of different phenomena.On the one hand, temperature effects that are influenced by the exothermicity and consequently depend on the hydrocarbon concentration are particularly relevant in the low-temperature regime, since during the light-off already minor changes of the temperature can have a strong impact on the conversion of the hydrocarbon.On the other hand, with increasing temperature also the oxidation of NO to NO 2 comes into play, which is subject of the following paragraph.
Regarding the NO conversion profiles, which are plotted in   1 into account, we can conclude that at low temperatures the catalytic HC-SCR reactions dominate the NOx profiles.In contrast, the NO oxidation becomes the dominant catalytic process at higher temperatures, with NO 2 as the main reaction product (Fig. S3).Based on spatially resolved species concentration profiles obtained by Irani et al.
[49], we can expect that above 300 • C the major share of hydrocarbons is converted in the front half of the monolith.Nevertheless, the remaining amount of hydrocarbons that are still present in the second half and whose conversion may require most or even the entire length of the monolith is sufficient to inhibit NO oxidation [49,50], which explains the trend reversal regarding the hydrocarbon concentration impact on the NOx conversion depicted in Fig. 3a and the decreasing overall NO 2 concentrations that can be observed at higher temperatures, e.g. as exemplarily shown in Fig. 3b for 325 • C.
In summary, the results discussed above suggest that N 2 O evolution can be circumvented twofold: Either the hydrocarbon concentration is kept as low as possible, simply resulting in a shortage of HC-species that could interact with NO, which consequently minimizes N 2 O formation, or the hydrocarbon concentration is kept so high that the exothermicity generated by the total oxidation of C 2 H 4 provides sufficient thermal energy to heat the catalyst beyond the critical temperature range.In the latter case, the low NO to NO 2 oxidation needs to be considered, which does not only originate from an inhibition due to the presence of C 2 H 4 , but which also occurs due to the thermodynamic NO-NO 2 equilibrium that is increasingly shifted toward NO with rising temperature despite the presence of excess oxygen [55].

Influence of space velocity
Since an increase of the space velocity results in faster passage of reactive gases through the monolith and reduces the time of pollutantcatalyst interaction, the hydrocarbon conversion and consequently also the N 2 O selectivity are shifted to higher temperatures (Fig. S4).During a light-off with DOC-2 in 150 ppm NO, 750 ppm C 2 H 4 , 10 vol.-% O 2 , 10 vol.-%H 2 O, bal.N 2 , for instance, T 90 (C 2 H 4 ) increases from 153 • C at 30 000 h − 1 to 171 • C at 90 000 h − 1 .Similarly, an increasing GHSV shifts the maximum N 2 O selectivity towards higher temperatures.Although beyond the simple shift in temperature both the selectivity profiles and the absolute N 2 O concentrations essentially remain unchanged, the total amount of N 2 O that is formed during the light-off test substantially increases with increasing space velocity, irrespective of the catalyst sample.Obviously, the activity of the catalyst is sufficient to activate all C 2 H 4 and NO molecules not only at low space velocity (30 000 h − 1 ) but also at high space velocity (90 000 h − 1 ), hence, despite a constant molar fraction of NO in the inlet gas flow, the total amount of NO converted to N 2 O varies with the space velocity.In order to allow for a direct comparison, the N 2 O amount in milligram (mg) produced per catalyst volume in liter (L) during each light-off test was calculated using the N 2 O signal versus time; further details can be found in the Fig. S1.
Fig. 4 summarizes the N 2 O production for both catalyst samples that were operated at different space velocities, C 2 H 4 /NOx ratios, and with different temperature ramps during the light-off.The comparison of the subfigures a-d of Fig. 4 reveals some general trends that are valid for both catalyst samples.In addition to the already discussed higher total N 2 O amounts formed during catalytic tests at higher GHSV, the data clearly demonstrate that the N 2 O evolution strongly correlates with the HC/NOx ratio: the higher the hydrocarbon concentration in the inlet gas flow, the higher is the N 2 O production.This trend is valid up to a C 2 H 4 / NOx ratio of 5 over DOC-1 and of 10 over DOC-2, respectively.However, the N 2 O amount formed drops significantly when using a ratio of 50, for instance when comparing the light-off with 5 • C/min and a GHSV of 60 000 h − 1 using mix 1d and mix 1e by a factor of approximately 4 over DOC-1 and by a factor of almost 10 over DOC-2.This behavior is due to the pronounced heat evolution when exposing the catalyst to high hydrocarbon amounts , namely 1500 ppm C 2 H 4 for mix 1d versus 7500 ppm C 2 H 4 for mix 1e, which mostly avoids the presence of critical temperature regimes along the catalyst sample that would allow for N 2 O evolution.Notably, the more active sample DOC-2 oxidizes C 2 H 4 at even lower temperatures, hereby passing the critical temperature regime that allows for N 2 O formation even faster.If operated at 30 000 h − 1 , both samples tested are capable of fully converting 7500 ppm C 2 H 4 at inlet temperatures as low as 150 • C. The absence of an N 2 O signal for these light-off tests suggests that under these conditions the high exothermicity enables the hydrocarbon conversion taking place so fast that the HC-NOx interactions that could result in N 2 O formation are suppressed.
While the N 2 O concentration curves remain mostly unchanged when varying the heating ramp (Fig. 2 and Fig. S1), the comparison of the total N 2 O amounts formed during the light-off tests reveals a striking reduction of the N 2 O quantity when choosing higher ramp rates.Notably, a faster passage of the critical temperature window, e.g. when using 15 • C/min (Fig. 4c, d) instead of 5 • C/min (Fig. 4a, b), is the most efficient measure to reduce the overall N 2 O formation over the catalyst.Regarding real-world applications, this corresponds to engine operation phases with high loads that contribute to passing the critical temperature range as fast as possible.In addition, active heating of the exhaust tailpipe may be a suitable approach to circumvent longer periods of temperature regimes enabling N 2 O formation.
In terms of total N 2 O amounts formed during the light-off tests, the effect of catalyst loading on N 2 O formation is most pronounced at a C 2 H 4 /NOx ratio of 1.0 and seems to reduce with rising hydrocarbon concentrations.For instance, the light-off for DOC-2 in gas mix 1a using a space velocity of 60 000 h − 1 and a ramp rate of 5 • C/min yields about 680 mg(N 2 O)/L(cat) and consequently produces a 2.3 times larger quantity than DOC-1 with only 295 mg(N 2 O)/L(cat) under identical conditions.In contrast, this factor drops to about 1.7 in gas mix 1d.The different behavior of the two catalyst samples becomes even clearer when plotting the N 2 O amount formed during a single light-off in relation to the total amount of NO dosed during the light-off as depicted in Fig. 5. Within the accuracy of the data, changes of the space velocity do not cause major changes of the NO to N 2 O conversion over DOC-1 (Fig. 5a).In contrast, related to the NO amount dosed during a lightoff test, a more pronounced N 2 O evolution is found over DOC-2 with decreasing space velocity (Fig. 5b).
While the very different noble metal loading, namely 12 g/ft 3 for DOC-1 and 100 g/ft 3 for DOC-2, can clearly be identified as reason for the different behavior of the two systems, we can only speculate about the microkinetic and microscopic reasons at this point.These comprise a different surface site availability for adsorption and reaction that is presumably higher over DOC-2, differences in the noble particle size that have a direct impact on the redox-behavior of the catalysts, and   diffusion limitations in the washcoat.In fact, contradictory findings have been reported on the role of the particle size on the selectivity and activity and thereby related phenomena like competition for surface sites and redox-behavior, and their exact role is at least partly still under debate in the context of HC-SCR [39, [56][57][58].While the higher noble metal loading of DOC-2 clearly promotes the activation of NO and C 2 H 4 already at low temperatures (Fig. 1) and causes an overall higher N 2 O production compared to DOC-1 (Fig. 4), the enhanced NO to N 2 O conversion at low space velocity that was only observed over DOC-2 (Fig. 5b) may be explained by different particle sizes that could lead to different turnover frequencies (TOF) regarding hydrocarbon oxidation, NO to NO 2 oxidation, and NO to N 2 O conversion.Based on the data shown in Fig. 5, one may speculate about different adsorption behavior over the two samples, i.e. by provision of more surface sites due to the higher loading of DOC-2 or by different adsorption energies and surface coverages as the consequence of variations of the particle size, morphology, and oxidation state [42,[59][60][61][62][63].Notably, also the various suggestions on the HC-SCR reaction mechanism and accompanying N 2 O formation pathways that were proposed in literature may originate from slight differences of the catalyst state or morphology, which call for future fundamental studies with well-defined model catalysts exploiting bulk-and surface-sensitive operando characterization techniques.Hence, we renounce giving a final explanation for the different behavior on a microkinetic and microscopic level at this point and rather focus on the role of hydrocarbons on N 2 O formation over the commercial oxidation catalysts in the following paragraph, which is of utmost importance in the light of real-world applications.

Influence of hydrocarbon
By replacing C 2 H 4 by other hydrocarbons that can occur in the exhaust gas of lean-burn applications, namely C 2 H 6 , C 3 H 6 , and C 3 H 8 as short-chain compounds and C 12 H 26 as representative long-chain species, we evaluated their tendency to cause N 2 O formation in the same temperature window and under identical operating conditions as described above.Notably, the species tested herein can be considered as potential reductant to reduce NO over noble metal catalysts, albeit with varying temperature regimes.
Simultaneous exposure of the catalysts to NO and either ethane (C 2 H 6 ) or propane (C 3 H 8 ) during a light-off up to 350 • C did not result in N 2 O formation over either of the catalyst samples.In line with previous results that indicate an increasing reactivity with increasing chain length [64,65], C 2 H 6 exhibits a higher stability than C 3 H 8 .While DOC-1 was unable to convert any of the C 2 H 6 , irrespective of the inlet concentration, and also DOC-2 shows only minimal C 2 H 6 conversion even at 350 • C, DOC-1 and DOC-2 convert approx.6% and 15% C 3 H 8 , respectively, in gas mix 3d at 350 • C and a space velocity of 60 000 h − 1 (cf.Figs.S5-S8).The overall higher activity for hydrocarbon conversion over DOC-2 can be attributed to its higher noble metal loading.While HC-SCR can be neglected in the investigated temperature regime, both catalysts oxidize NO to NO 2 , showing the typical increase in the NO oxidation activity with rising temperature until the thermodynamic equilibrium benefits NO 2 decomposition to NO [10,53].Although in the low-temperature regime neither C 2 H 6 nor C 3 H 8 are converted and also at elevated temperatures above 300 • C the conversion rates are very low, the inhibition of the NO to NO 2 oxidation is more pronounced when C 3 H 8 is present in the gas stream.Hence, an actual conversion of the hydrocarbon does not seem to be a prerequisite for NO oxidation inhibition, but rather a mere adsorption and storage of the hydrocarbon can take place, which can efficiently block active surface sites that are then unavailable for NO adsorption and reaction.
Analogous to C 2 H 4 , also C 3 H 6 formed significant amounts of N 2 O during the light-off tests over a broad range of conditions (Fig. 6).N 2 O formation presumably proceeds via the following global reaction Eq.(3): This is a well-known phenomenon and has previously been reported as an undesired side-reaction during the selective catalytic reduction of NO using C 3 H 6 as reductant [20,26,29,36].
Overall, the results discussed so far suggest that in the temperature window that is subject to this study, only the presence of unsaturated hydrocarbons gives rise to N 2 O formation, whereas saturated hydrocarbons are hardly converted and therefore do not cause secondary emissions.However, our current results with C 12 H 26 levels that are relevant in real-world applications suggest N 2 O formation between 150 • C and 350 • C if NO and C 12 H 26 interact on the Pt-Pd catalysts, similar to earlier results by Kamasamudram et al. [20] and despite the saturated nature of C 12 H 26 .Compared to the total amounts formed in the presence of C 2 H 4 or C 3 H 6 , C 12 H 26 yields substantially lower N 2 O levels (Fig. 6).We assume that this behavior is due to the catalytic decomposition of C 12 H 26 .Earlier mechanistic investigations that typically combine experimental insights with kinetic models suggested that the conversion of C 12 H 26 in the presence of O 2 takes place via the scission of longer HC-chains into HC-species with shorter chain length [66][67][68][69][70]. Subsequently, the reaction of unsaturated fragments that can occur among these short-chain species with NO results in formation of low N 2 O levels.Exposure to gas mix 1b at a GHSV of 60 000 h − 1 and a ramp rate of 15 • C/min, for instance, results in a maximum N 2 O concentration of approximately 4 ppm over both samples (Fig. S9).Although this is an almost negligible level compared to the tests with unsaturated hydrocarbons, the accumulated N 2 O emissions over time are still relevant in the light of the enormous greenhouse potential of N 2 O.
With regard to the total N 2 O amounts formed when using either C 2 H 4 or C 3 H 6 , not only the HC/NOx ratio but also the noble metal loading and the ramp rate seem to strongly influence the N 2 O formation.Over DOC-1, the interaction between NO and C 2 H 4 results in significantly lower N 2 O amounts than the NO-C 3 H 6 interaction if a HC/NOx ratio of 1.0 (mix a) or 2.5 (mix b) is chosen (Fig. 6a, c).In contrast, the evolving N 2 O quantities become higher in gas mix c and d the presence of C 2 H 4 .With a HC/NOx ratio of 50, this trend is reversing again and the NO-C 3 H 6 interactions produce more N 2 O.These trends are in line with the trends found when plotting the N 2 O amount in relation to the NO amount dosed (Fig. S10).The behavior in mix e can be attributed to the earlier onset of the exothermicity effect when feeding the more reactive C 2 H 4 : During the light-off with 5 • C/min, full C 2 H 4 conversion is observed already at an inlet temperature of 173 • C, which corresponds to an outlet temperature of 350 • C. On the other hand, C 3 H 6 activation requires higher temperatures and lights-off rapidly above 205 • C (Fig. S11).Although the higher reactivity of C 2 H 4 may also contribute to temperature profiles along the catalyst sample that benefit N 2 O formation when using moderate HC/NOx ratios, also chemical processes taking place on the catalyst surface can come into play.In their mechanistic investigation on HC-SCR utilizing the technique of temporal analysis of products (TAP), Burch et al. [34] pointed out that a single C 3 H 6 molecule is capable of reacting with nine oxygen atoms that are adsorbed on the platinum surface.Since reduced platinum sites are considered a prerequisite for nitrogen oxide adsorption, the affinity of the hydrocarbon towards oxidation is of utmost importance.Considering the higher reactivity of C 2 H 4 , we can speculate about a faster consumption of adsorbed oxygen atoms that are then available for NO adsorption and subsequent partial reduction to N 2 O.With respect to the surface oxygen concentration, however, the impact of the HC reactivity could be less important at low hydrocarbon concentrations and moderate catalytic activity for oxidation reactions.Instead, the maximum number of consumed O atoms per HC molecule -9 for C 3 H 6 versus 6 for C 2 H 4 -may dominate the surface species coverage.Such an overall more efficient consumption of oxygen atoms over DOC-2 would not only explain the overall higher N 2 O levels over DOC-2, but would also explain the different behavior of DOC-1 and DOC-2 in gas mix a and b.While DOC-2 with its high noble metal loading sufficiently catalyzes the hydrocarbon oxidation also at low concentrations, the removal of O surface atoms is less efficient over the less active DOC-1, possibly due to the lower local heat evolution as a result of less efficient exothermic hydrocarbon oxidation.
In addition to the overall N 2 O amount formed during light-off, the influence of the HC/NOx ratio that was already extensively discussed for C 2 H 4 (cf.Section 3.2) seems to change when using C 3 H 6 as reductant instead.As illustrated in Fig. 7, an increasing C 3 H 6 concentration does not only go along with an increasing maximum N 2 O concentration, but the maximum N 2 O value is shifted towards higher temperatures for higher C 3 H 6 inlet concentrations.While these observations are in line with earlier results reported by Acke et al. [29], who investigated NOx reduction with propene over Pt/ZSM-5 catalysts, the trends observed when using C 3 H 6 are clearly inverted compared to the trends observed for the kinetic tests with C 2 H 4 (Fig. 2c, d).Since the temperature T(c (N 2 O) max ) reported herein always represents the inlet temperature that is measured approx. 2 mm upstream the catalyst sample, we do not assume a bias of the temperature data due to a different exothermicity of the hydrocarbon oxidation reactions.Hence, the reason for this difference is most likely of a mechanistic nature.As the hydrocarbon surface coverage is assumed to govern HC-deNOx and the accompanying N 2 O formation [29], one may speculate about a different adsorption behavior of C 2 H 4 and C 3 H 6 and consequently different surface coverages with hydrocarbon species that compete with oxygen and NO for surface sites [64,71].In particular, Acke et al. [29] assumed the formation of an inactive surface layer with C 3 H 6 species, which may explain the shift of T (c(N 2 O) max ) towards higher temperatures when increasing the C 3 H 6 inlet concentration.Also, differences in the power of the reducing agent are of high relevance as reported by Burch et al. [34], since the ability of the reductant to react with adsorbed oxygen atoms ensures the availability of a reduced platinum surface for NO adsorption and reaction.With respect to HC-SCR regimes and NO to NO 2 oxidation when C 3 H 6 is fed (Fig. S12), we found an analogous behavior to C 2 H 4 , even if the temperature regimes that benefit either of the two processes are slightly shifted compared to the gas composition containing C 2 H 4 .These results confirm earlier results in the context of C 3 H 6 -NO-interactions in excess oxygen that point to an inhibition of NO oxidation over platinum-based oxidation catalysts in the presence of C 3 H 6 [49].
Future fundamental studies with a focus on mechanistic insights are highly desirable for uncovering the origin of the different behavior of C 2 H 4 and C 3 H 6 in detail and may additionally clarify the exact N 2 O formation pathway that is still under debate.However, even without final mechanistic insights on a microkinetic level, our current data under conditions relevant for real-world applications clearly demonstrate that the type of hydrocarbon is one of the key parameters governing N 2 O formation from NO over noble metal catalysts.When designing operation modes and exhaust tailpipes for lean-burn combustion engines that aim at avoiding N 2 O evolution, these findings need to be considered.

Conclusions
In view of the growing awareness of secondary emissions evolving from modern catalytic converters for emission control, this study investigates the formation of N 2 O over Pt-Pd oxidation catalysts as a consequence of interactions between hydrocarbons and NO.In a series of kinetic light-off tests between 150 • C and 350 • C under conditions relevant for lean-burn engines, the interaction of NO with the unsaturated hydrocarbons C 2 H 4 and C 3 H 6 resulted in pronounced formation of N 2 O even beyond the point of their full conversion.Herein, N 2 O is the outcome of an undesired side reaction of the selective catalytic reduction of NO using hydrocarbons (HC-SCR).In contrast, the saturated hydrocarbons C 2 H 6 and C 3 H 8 were barely converted below 350 • C, hence, no N 2 O was formed.C 12 H 26 gave rise to small N 2 O quantities, which we attribute to the C 12 H 26 decomposition mechanism via chainsplitting and involvement of unsaturated short-chain fragments.Variations of the HC/NOx ratio suggest an increasing N 2 O formation with increasing HC-content, which can be attributed to an efficient removal of surface oxygen atoms from the noble metal, hereby creating free surface sites for the adsorption and conversion of NO.Once the exothermicity of the hydrocarbon oxidation is capable of heating the catalyst beyond the critical low-temperature regime that benefits N 2 O formation, N 2 O formation drops.For instance, a HC/NOx ratio of 50 results in significantly lower N 2 O formation than a HC/NOx ratio of 10.Passing through the critical low-temperature window as fast as possible, i.e. by faster heating, was identified as most effective measure to reduce N 2 O formation.In terms of real-world applications, a fast temperature increase above the critical temperature range regarding N 2 O formation, e.g. by electrical heating of catalytic converters, could be a feasible option.In addition, the presence of very high hydrocarbon concentrations that are oxidized over the catalyst and cause significant heat evolution could be exploited, since the pronounced exothermicity can sufficiently heat the catalyst to skip the critical temperature range in which N 2 O formation can occur.
Moreover, the hydrocarbon concentration and the reactivity of the hydrocarbon have an impact on the oxidation of NO to NO 2 .When using C 2 H 4 as reductant, for instance, the HC-SCR reaction regime dominates up to approximately 250 • C and is followed by a temperature regime above 300 • C that is dominated by NO oxidation.While rising HC concentrations promote NO conversionmainly due to undesired reduction to N 2 Ohydrocarbons inhibit NO oxidation to NO 2 , which needs to be considered when designing a modern and highly efficient overall exhaust tailpipe.In summary, these results underscore that already minor variations in the reaction conditions can tremendously change the dominant reaction mechanism and the corresponding product species concentrations.In addition, a higher noble metal loading facilitates hydrocarbon activation at low temperatures, governs an earlier onset of exothermicity, and corresponds to increasing N 2 O amounts formed during the light-off test.
From a fundamental point of view, the roles of the noble metal particle size, alloying, and oxidation state that were beyond the scope of our present study deserve to be investigated in more detail during future studies focusing on surface and bulk phenomena, possibly supplemented by microkinetic modeling that can help to understand the complex interactions between NOx and hydrocarbons.With regard to industrial relevance, an increasing use of biofuels and synthetic fuels produced from renewable feedstocks that exhibit a different molecular composition compared to conventional fuels can require adaptions regarding the combustion technology [72].Ultimately, such changes could make the prevention of secondary emission evolution in the exhaust tailpipe even more important.Hence, comprehensive considerations taking the fuel and the combustion technology into account are not only a prerequisite for developing novel catalyst materials with high activity and selectivity for future applications, but can also help refining operation procedures in real-world applications.On the long run, this scenario may call for further measures to mitigate N 2 O formation, for instance by fast external heating.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Fig. 3a for DOC-2, a transition regime between 250 • C and 300 • C is observed where the conversion curves intersect.At temperatures below 250 • C, an increasing C 2 H 4 inlet concentration promotes NO removal, which is exemplarily illustrated in the top part of Fig. 3b at 200 • C. Irrespective of the HC/NOx ratio, the NO conversion rises to a maximum and drops continuously afterwards (Fig.3a).On the other hand, high C 2 H 4

P
.Lott et al.
P.Lott et al.