H₂-SCR over Low Loaded Platinum-Based Catalysts: Investigations in the Reaction Pathways

Abstract

The pathways and mechanistic aspects of H₂-SCR over precious metal-based catalysts is still under debate. This study focusses on low loaded platinum-based catalysts (0.07–0.3%) in a large temperature range (50–500 °C), with special focus on (i) the role of NH₃ as a possible intermediate species, (ii) the origin of the undesired N₂O emission and (iii) the platinum sites involved in the H₂-SCR deNO_X reactions. Up to 60 °C, the N₂O selectivity was close to 100%, with no influence of the presence of oxygen in the 50–100 °C temperature range. Ammonia formation was observed at relatively low temperatures (from 60 °C), but its reactivity was then limited. All these low temperature reactions were associated with the same platinum sites, probably a mix of edge and face sites. The maximum outlet NH₃ was observed around 100 °C and the role of the NH₃-SCR in the whole H₂-SCR process appeared very limited. On the contrary, the ammonia oxidation by O₂, which started near 120 °C, significantly contributed to the H₂-SCR process and appeared responsible for the second N₂O emission peak (150–500 °C). This reaction did not imply the same platinum sites and appears mainly dependant on the platinum particle size.

1. Introduction

To reduce the CO₂ emissions responsible for the greenhouse gas (GHG) footprint of the automotive sector, three main carbon zero-emission technologies for vehicles are proposed: battery electric vehicles (BEVs), hydrogen fuel cell electric vehicles (FCEVs), and hydrogen internal combustion engines (H₂-ICEs). Achieving the objective of the GHG emission mitigation necessitates that hydrogen production come from non-fossil fuels and renewable energy. The use of hydrogen as a carbon-free fuel in internal combustion engines is gaining significant interest due to its high energy density, its lower sensibility to H₂ quality (compared to fuel cells) and the possibility of the reconversion of current robust and mature diesel and natural gas engine technologies on a possibly hybrid powertrain coupled system. As a result, H₂-ICEs may serve as an attractive transitional solution toward the large-scale deployment of hydrogen-based propulsion systems [1,2].

However, some nitrogen oxides (NO_X) are still emitted by H₂-ICEs as harmful pollutants, with emissions possibly exceeding those of conventional Diesel engines when a stoichiometric air/H₂ mixture is used [3]. Fortunately, NO_X emissions decrease significantly as the air/H₂ ratio increases, dropping from approximately 4000 ppm under stoichiometric mixture to around 100 ppm in lean conditions [4]. Moreover, operating H₂-ICEs under lean conditions not only significantly limits NO_X emissions—potentially achieving a 90% reduction compared to diesel engines—but also enhances thermal efficiency [3]. Nevertheless, achieving a zero-emission H₂-ICE that can compete with clean fuel cells is still a challenge. One attractive option is to develop a robust, passive and cost-effective catalytic for the NO_X selective catalytic reduction by H₂ (H₂-SCR) to convert the remaining NO_X traces into N₂.

According to the literature, the H₂-SCR mechanism can be divided into two distinct routes, namely (i) the NO adsorption/dissociation pathway based on a Langmuir–Hinshelwood mechanism and (ii) the oxidation–reduction pathway [5]. Regardless of the considered mechanism, a typical volcano-shaped curve in NO_X conversion is obtained. This behaviour is attributed to the intensified competition with the H₂ combustion reaction as the temperature increases, particularly above 150 °C on precious metals.

The NO adsorption/dissociation mechanism is primarily associated with noble metal-supported catalysts which have been developed for lean-deNO_X systems for Diesel engines after-treatment [6]. Pt and Pd-based catalysts exhibit the highest activity when hydrogen is used as the reductant. However, the operating temperature range for these catalysts is rather limited (typically between 100 and 300 °C) and their selectivity at low temperatures suffers from the generation of undesirable N₂O emissions, especially in the presence of water [7]. The key elementary step for the NO adsorption/dissociation route involves the NO dissociative adsorption, leading to the formation of adsorbed O and N atoms. The N₂ formation occurs through the recombination of two adsorbed nitrogen atoms (N_ads) on the catalyst surface. The role of H₂ is then to remove adsorbed oxygen atoms (O_ads) from the active sites, thereby cleaning the surface. It is generally admitted that the N₂O formation arises from the reaction between N_ads and NO [5,8]. However, it has also been proposed that N₂O formation can result from the reaction between NO_ads and NO, leading to N₂O and O_ads [9]. Indeed, the N₂O emission via the formation and the decomposition of a (NO)₂ dimer species at a low temperature is supported by the lower energetic barrier of this pathway compared to NO dissociation [10]. Moreover, it was also observed on Rh/SiO₂ that the suppression of H₂ in the reaction mixture led to N₂O formation, while outlet N₂ decreased [11]. It was concluded that NO_ad and O_ad neighbouring species were involved in N₂O formation. The increase in O_ad leads to a decrease in N_ad and also promotes hydrogen combustion [12]. These features illustrate competitions in neighbouring platinum active sites, which must be addressed to gain a deeper understanding of H₂-SCR pathways at low temperatures.

The oxidation–reduction mechanism is based on adsorbed intermediate N-containing species including nitrosonium, nitrites, nitrates or ammonia-like (NH_X) species [13]. Interestingly, this NO reduction mechanism appears similar to the NH₃-SCR pathway, in which the generation of nitrosamine intermediates (NH₂NO) plays a key role in the enhancement of the SCR performance [14]. Whatever the reaction mechanism considered, the reaction pathways remain a topic of ongoing debate and NH_X type species are also reported to play a significant role in the H₂-SCR process. NH_X formation takes place on metallic sites via the reaction between adsorbed nitrogen species from the NO dissociative adsorption and active hydrogen. Subsequently, NH_X compounds react with adsorbed NO species to yield N₂ [15]. At low temperatures, the dissociation of NO remains the rate-determining step (rds) in the H₂-SCR process, while the in situ formation of NH_X species becomes a crucial step reaction at higher temperatures. In fact, N₂ selectivity was proposed to depend on the H_(ad)/NO_X(ad) ratio [16]. As expected, since the NH_X species is generated, the NO_X conversion is favoured with catalysts that exhibit acidic behaviours, which promote the NH₃-SCR activity. The acidic properties of the support have also been reported to influence the H₂-SCR behaviour by impacting the metal oxidation state. Li et al. [17] showed that platinum-based catalysts supported on acidic supports promoted the formation of metallic platinum particles, thereby promoting the dissociative adsorption of NO_X. However, the reaction selectivity was reported to be mainly dependent on the acidity of the support rather than the intrinsic properties of platinum. These observations were attributed to a bifunctional mechanism: at a low temperature (75° C), the formation NH₄⁺ intermediate species occurs via the reaction of NO with H₂ on the metallic platinum surface. These NH_X species subsequently adsorb onto the acidic sites of the support, where they facilitate the selective reduction in NO, ultimately yielding N₂ as the main reaction product. This NH₃-SCR involvement is notably reported in the very recent review from Jabłońska et al. [18].

Finally, the pathways and mechanistic aspects of the H₂-SCR over precious metal-based catalysts remain the subject of ongoing debate. Among the usual precious metals (Pt, Pd, Ir), Shao et al. recently confirmed that Pt exhibited the highest catalytic activity but also offered the widest operational temperature window [19]. Consequently, platinum was selected in this study to investigate the reaction pathways of the H₂-SCR over low loaded platinum-based catalysts, with a special focus on (i) the role of NH₃ as a possible intermediate species, (ii) the undesired N₂O emission and (iii) the platinum sites involved in the deNO_X reactions. To address these aspects, platinum loading was limited to 0.07%–0.3 _wt% due to its high cost, and various silica-alumina supports with different silica contents were selected to study the influence of acidity on H₂-SCR activity and selectivity, in comparison with the use of alumina alone. This study is mainly based on catalytic measurements and a statistical cuboctahedron platinum cluster model. To assess the H₂-SCR reaction pathways depending on the temperature, various masses of catalyst were involved, and various associated reactions, such as the NO + H₂ reaction (without oxygen), NH₃-SCR, the NH₃ oxidation by O₂, were separately examined. Very recently, Dong et al. developed a similar approach on Pt/SSZ-13 catalysts [20], varying the platinum loading between 0.01 _wt% and 0.5 _wt%. They proposed that the H₂-SCR reaction over this zeolite-supported catalyst involves hydrogen spillover and the generation of NH_X⁺ species, and that NH₃-SCR and NH_X⁺ oxidation may occur in the whole reaction.

2. Results

2.1. Investigation in Silica-Alumina and Al₂O₃ Supports Properties for H₂-SCR over 0.30 _wt%Pt-Based Catalysts

In order to select the more appropriate support to investigate the H₂-SCR reaction pathways, an alumina support and three silica-alumina supports were selected and impregnated with 0.30 _wt% platinum. Catalytic tests were then first performed with 200 mg of various samples in the 100–500 °C temperature range. The catalytic test set-up and protocols are described in Section 3.3 and Appendix A.

2.1.1. Characterization

The textural properties of the alumina support and the three selected silica-alumina supports with silica weight contents of 5%, 20% and 40% (all provided by Sasol) are reported in

Table A1. Textural properties of the selected supports after hydrothermal treatment at 700 °C, 4 h.

Support	SiO2 Loading (%)	SBET (m2·g−1)	Average Pore Size (Å)	Pore Volume (cm3·g−1)
Alumina	-	157	126	0.50
Siral-5	5	296	83	0.62
Siral-20	20	356	77	0.96
Siral-40	40	422	86	0.91

(Appendix B.1). Pure alumina exhibited the lowest specific surface area (157 m² g⁻¹), while increasing the silica content from 5% to 40% led to a progressive rise in surface area, from 296 m² g⁻¹ to 422 m² g⁻¹.

The acidic properties of the supports were evaluated by pyridine adsorption monitored by infrared spectroscopy, which allows the quantification of both Lewis acid sites (LAS, associated with PyL species) and Brønsted acid sites (BAS, associated with the formation of PyH⁺ species). The corresponding spectra obtained after pyridine adsorption and evacuation at 150 °C are presented Figure A1 (Appendix B.2).

Alumina shows a typical spectrum with characteristic bands associated with the Lewis acid sites (LAS) at 1454 cm⁻¹, 1624 and 1618 cm⁻¹. The doublet with peaks at 1622 and 1618 cm⁻¹ reveals the presence of multiple types of LAS, differing by their acidity strength and Al coordination (Al_IV and Al_VI). Additionally, as expected, no Brønsted acidity was detected on Al₂O₃. The spectrum obtained with Siral-5 is closely similar to that of pure alumina, except that the components observed at 1624 and 1618 cm⁻¹ now merged into a single band centred at 1622 cm⁻¹, which suggests a homogenization of the Lewis acid sites in terms of acidity strength. Increasing SiO₂ content progressively led to a decrease in the intensity of the LAS-related bands and to a shift in the band at 1454 cm⁻¹ to higher frequencies, reaching 1457 cm⁻¹ for Siral-40. This shift indicates a change in the nature of the surface Lewis acid sites, likely due to alterations in their electronic structure and chemical environment. Additionally, a weak ν_19b band at around 1548 cm⁻¹, attributed to pyridinium ions formed by a pyridine interaction with Brønsted acid sites, was evidenced in the spectra of Siral-20 and Siral-40. Note that the intensity of this ν_19b band was very low for the Siral-5 sample.

The quantification of the amount of each LAS and BAS depending on the evacuation temperature is plotted in Figure 1 for each support, normalized by gram of catalyst and determined from ν_19b area, using its molar coefficient. As expected, the increase in the evacuation temperature led to a decrease in the amount of absorbed species. Some pyridine adsorption was still observed on LAS at 450 °C, while Brønsted sites exhibited nearly no adsorption beyond 300 °C. Among the studied supports, Siral-5 exhibited the highest amount of acid sites, with a low proportion of Brønsted sites. In contrast, Siral-20 and Siral-40 exhibited significant Brønsted acidity, which persisted from pyridine evacuation up to 250 °C.

The platinum loading of the 0.30 _wt%Pt/support samples (

Table 1. Specific surface area, platinum loading, mean platinum particle size and dispersion.

Catalyst	SBET (m2 g−1)	Pt Loading from ICP Analysis (wt%)	Mean Pt Particle Size (nm) *	Pt Dispersion (%)
0.30 wt%Pt/Al2O3	144	0.39	7.0/-	19
0.30 wt%Pt/Siral-5	218	0.34	10.4/10.6	13
0.30 wt%Pt/Siral-20	245	0.30	11.8/-	11
0.30 wt%Pt/Siral-40	232	0.32	13.4/-	10
0.15 wt%Pt/Al2O3	149	0.13	-/4.5	27
0.15 wt%Pt/Siral-5	225	0.16	11.4/11.2	12
0.15 wt%Pt/Siral-20	239	0.15	-/10.9	12
0.15 wt%Pt/Siral-40	230	0.13	14.3/13.5	9
0.07 wt%Pt/Siral-5	269	0.07	8.2/8.4	15
0.30 wt%Pt/Siral-5 w/o hydrotreatment	224	0.23	1.9/-	56
1% wtPt/Al2O3 (commercial) **	142	0.90	-/2.6	44

* From H₂ chimisorption/from TEM (examples of TEM images and particle size distribution are available in Appendix C). ** reduced under 10% H₂ at 600 °C for 1 h.

) shows that the expected amount of platinum was deposited whatever the considered support. Hydrogen chemisorption measurements and TEM were also performed to evaluate the platinum dispersion. Table 1 shows that the higher platinum dispersion was observed on alumina (19%). The platinum dispersion decreased with the silica content, from 13% on siral-5 to 10% on Siral-40. These results are consistent with those previously reported by Shibata et al. [21] who also observed a decrease in platinum dispersion with the increase in the silica content in SiO₂-Al₂O₃ supports.

2.1.2. H₂-SCR Behaviour

The 0.30 _wt%Pt supported catalysts were evaluated in H₂-SCR in the 100–500 °C temperature range (m_cata: 200 mg). NO_X conversion, N₂ and N₂O yields are reported in Figure A2 (Appendix B2). For all catalysts, the NO_X conversion reached 100% at 100 °C. The NO_X conversion then decreased with the temperature increase, except for 0.30Pt/Siral-5 which maintained a complete NO_X conversion until 200 °C. The N₂ and N₂O formations in the 100–200 °C temperature range were dependent on the nature of the support. The major product was N₂O for 0.30Pt/Al₂O₃, 0.30Pt/Siral-20 and 0.30Pt/Siral-40 (mean selectivity around 78%, 54% and 58%, respectively). On the contrary, the 0.30Pt/Siral-5 sample showed N₂ as the major product (mean N₂ selectivity: 58%). Figure 2 illustrates that NO_X conversion and especially N₂ selectivity were mainly dependant on acidic properties at 250 °C. Note that ammonia emission was not observed during these tests involving 200 mg of catalyst. These results are in accordance with Zhang et al. [22] which showed by diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) that LAS can contribute to NH₃ adsorption, enhancing the N₂ selectivity and deNO_X performances by assisted NH₃-SCR. However, this particular aspect about the ammonia formation is further examined in the next Section 2.2 by decreasing the involved weight of the catalyst.

Finally, these preliminary results showed that Siral-5 support offered the best compromise between NO_X conversion and N₂ selectivity. This support exhibited the higher acidity, mainly as Lewis sites with traces of Brønsted sites, a high specific surface area compared to pure alumina, and an intermediate platinum dispersion (12%, slightly higher than on the two other alumina-silica supports). The Siral-5 support was therefore selected for further investigations about the deNO_X pathways involved in the H₂-SCR reaction.

2.2. Investigation of the H₂-SCR Behaviour Depending on the Involved Amount of Catalyst (0.30 _wt%Pt/Siral-5)

To investigate the involvement of intermediate species and by-side reactions in the H₂-SCR process, the mass of the catalyst was varied to increase the space velocity (same feed flow and composition). Starting from 200 mg, additional catalytic tests were performed decreasing the amount of the catalyst to 100 mg, 30 mg and 5 mg. Moreover, the minimum catalytic test temperature was decreased to almost 60 °C to insure NO_X conversion below 100% in the low temperature range. The resulting NO_X conversion profiles and the calculated yields in N₂, N₂O and NH₃ are reported in Figure 3 depending on the temperature.

Considering the initial catalyst mass of 200 mg (Figure 3a), a high NO_X conversion was recorded at low temperature, reaching 86% at 30 °C. However, the main product was N₂O, with a yield of 71%. The NO_X conversion remained higher than 95% across the 50–250 °C temperature range, with maximum N₂ yield (66%) obtained around 250 °C. At higher temperatures, NO_X conversion decreased due to the competition with hydrogen oxidation by O₂ (see thereafter Section 2.3.4). No outlet ammonia was detected whatever the considered temperature in this experiment (200 mg of catalyst).

As the catalyst mass decreased (Figure 3a–d), the decrease in the number of available active sites resulted in a narrower operating temperature range. The NO_X conversion was maintained above 90% in the 40–240 °C temperature ranges for 100 mg and 200 mg of involved catalyst, and this activity window decreased to 85–240 °C and 110–210 °C for 30 mg and 5 mg of catalyst, respectively.

Interestingly, decreasing the catalyst mass significantly influenced the distribution of the products. Particularly, Figure 3c,d shows the detection of NH₃ centred around 100 °C, with an increasing amount recorded as the involved mass of the catalyst decreased. This observation indicates that NH₃ is mainly formed at the beginning of the catalyst bed before further reaction. Note that a significant ammonia concentration was recorded at the lowest tested temperature when using 5 mg of catalyst (20 ppm at 60 °C). This raises important questions regarding the role of NH₃ as an intermediate or by-product, which warrants further investigations. Additionally, decreasing the amount of the catalyst also revealed that N₂O emissions split into two distinct broad peaks, one below and one above 100 °C. Surprisingly, the maximum N₂ yield remained around 65% regardless of the mass of catalyst ranging from 200 mg to 5 mg. However, the temperature at which this yield was obtained was shifted to lower values, from 250 °C to 120 °C, as the catalyst mass was decreased.

Finally, these catalytic results clearly illustrate the participation of various reactions depending on the temperature. Further investigations are required to (i) determine at which temperature NH₃ is produced, according to Equation (1), especially in comparison with the NO_X conversion on-set temperature; (ii) elucidate the subsequent reactions of NH₃ with other compounds present in the reaction mixture since NH₃ may react with the remaining NO_X (Equations (2) and (3)) and/or O₂ (Equations (4) and (5)) to form N₂ and/or N₂O. Additionally, the H₂ reactivity also deserves particular attention. Note that the possible reaction between H₂ and N₂O was also checked using 200 mg of catalyst: in a rich mixture (900 ppm N₂O + 1% H₂), the N₂O conversion started below 100 °C and reached 100% at 250 °C (100% N₂ selectivity), but the conversion became negligible with the addition of 2% O₂ in the mixture (lean mixture). These results show no N₂O reactivity in oxidative media.

NO + 5/2 H₂ →NH₃+ H₂O

(1)

4 NH₃ + 4 NO + O₂ →4 N₂ + 6 H₂O

(2)

4 NH₃ + 4 NO + 3 O₂ → 4 N₂O + 6 H₂O

(3)

2 NH₃ + 3/2 O₂→ N₂ + 3 H₂O

(4)

2 NH₃ + 2 O₂ → N₂O + 3 H₂O

(5)

To address these questions, an individual and rigorous investigation of the various reactions which are supposed to intervene is necessary. This specific study under various reaction mixtures was performed with 30 mg of 0.30 _wt%Pt/Siral-5 sample.

2.3. Investigation on H₂-SCR Reaction Pathways: H₂-SCR, (NO + H₂) Reaction, NH₃-SCR, NH₃ Oxidation Reactions

2.3.1. H₂-SCR vs. (NO + H₂) Reaction Behaviour

To investigate ammonia formation in the H₂-SCR process, a catalytic test was performed without oxygen in the reaction mixture (NO + H₂ reaction,

Table 2. Reactional mixtures depending on the type of catalytic test (total flow: 150 mL min⁻¹).

Catalytic Test	NO (ppm)	NH3 (ppm)	H2 (%)	O2 (%)	N2
H2-SCR	400	-	1	2	balance
NO + H2 reaction	400	-	1	-	balance
NH3-SCR	400	400	-	2	balance
NH3 oxidation	-	400	-	2	balance
H2 + O2 reaction	-	-	1	2	balance

). Particular attention was paid to determine the initial ammonia emission temperature (Equation (1)) in relation to the NO_X conversion onset temperature.

Figure 4a,b compare NO_X conversions and yields in N₂, N₂O and NH₃ depending on the reaction mixture, i.e., NO + H₂ + O₂ and NO + H₂. Both figures show that the NO_X conversion started at the same temperature with or without O₂ in the feed stream, and both NO_X conversion curves remained superposed until reaching full NO_X conversion at around 100 °C. Similarly, the N₂O yield curves were also superposed until reaching their maximum (56% at 80 °C), indicating that O₂ is not involved in the “low temperature” behaviour. Furthermore, the NH₃ emission also began at the same temperature of 70 °C for both reaction mixtures. Given the excess of reductant in the NO + H₂ mixture, NO_X conversion remained complete within the 115–500 °C temperature range. Since NO_X conversion reached 100%, NH₃ was the main product with a maximum yield of 80% (at 375 °C), while the N₂ yield was between 7% and 13% in the 200–500 °C temperature range. Conversely, in the presence of O₂ (H₂-SCR, Figure 4a), the obtained ammonia can then react along the catalytic bed as illustrated in Figure 3, either with NO_X or O₂ (Equations (2)–(5)). Both reactions can influence the distribution of reaction products. These strongly temperature-dependent reactions are evaluated in the two following sections.

2.3.2. H₂-SCR vs. NH₃-SCR Reaction

Since NH₃ was formed under the H₂-SCR condition, the NH₃-SCR reaction (Figure 4c) may occur along the catalytic bed. However, considering the measurements obtained in H₂-SCR with 5 or 30 mg of catalyst (Figure 3c,d), the simultaneous presence of NO_X and NH₃ was restricted to a narrow temperature range, from approximately 70 °C to 100–150 °C (the oxidation by O₂, evaluated in the next section, is then more likely).

According to the results reported in Figure 4c, NOx reduction by NH₃ significantly started at approximately 100 °C, which also corresponds to the drop in NH₃ emission under the H₂-SCR conditions (Figure 4a). Consequently, ammonia was not involved in low temperature NO_X conversion (below 100 °C) during the H₂-SCR process. The conversion of NO_X by NH₃ (NH₃-SCR, Figure 4c) reached a maximum at around 200 °C, a temperature at which the NH₃ conversion became complete. Thereafter, the conversion gradually declined, becoming negligible at approximately 300 °C. At higher temperatures, the amount of detected NO_X exceeded the inlet NO_X, which can be attributed to the ammonia oxidation by O₂ into NO_X (orange line). However, the major product of the NH₃-SCR reaction was clearly N₂O, while the N₂ yield remained around 11–18% in the 200–500 °C temperature range.

2.3.3. H₂-SCR vs. NH₃ Oxidation (By O₂)

As previously mentioned, ammonia consumption during the H₂-SCR reaction can also be associated with its oxidation by oxygen. As shown in Figure 4d, NH₃ oxidation by oxygen (NH₃ + O₂, Table 2) started slowly at around 100 °C, and then increased significantly from 150 °C, reaching 100% at 190 °C. The ammonia conversion remained complete up to 500 °C. This reaction led mainly to N₂, with a maximum yield of 80% recorded when the ammonia conversion reached 100%.

The N₂O emission started to increase from 150 °C, reaching a maximum close to 30% at around 240 °C. The obtained N₂O yield profile is fairly similar to that observed during the H₂-SCR reaction at similar temperatures. This similarity suggests that NH₃-oxidation by O₂ would largely contribute to the second wave of N₂O formation observed at high temperatures during H₂-SCR.

Outlet NO_X species were recorded from 200 °C during the NH₃ oxidation test. This temperature also corresponds to the decrease in NO_X conversion observed during the H₂-SCR reaction. This suggests that the NH₃ oxidation by O₂ may contribute to the decrease in the H₂-SCR efficiency at high temperatures, by the potential re-oxidation of NH₃ into NO. However, another plausible explanation is the insufficient availability of H₂ at these high temperatures.

In summary, results in Figure 4d show that the oxidation of NH₃ by O₂ appears to play a major role in the H₂-SCR behaviours, especially at temperatures above 150 °C. This contribution could likely explain the high N₂ selectivity observed in the H₂-SCR reaction and the formation of the second N₂O phase at high temperatures. In contrast, the NH₃-SCR appears to be favoured only in a narrow temperature range, between 100 °C and 150 °C.

2.3.4. H₂-Reactivity

The H₂ conversion curve during the H₂-SCR test on 0.30 _wt%Pt/Siral-5 is reported in Figure 5. Considering the various reactions involving H₂ (Equations (1) and (6–8)) and the outlet gas composition, an estimation of the hydrogen consumption in the H₂-SCR was calculated and reported in Figure 5a. As expected, due to the large amount of H₂ compared to NO_X, only a small percentage (<5%) of the inlet hydrogen was used to reduce NO_X, with the predominant reaction being the H₂ oxidation by O₂ regardless of the temperature.

H₂ + 1/2 O₂→H₂O

(6)

2 H₂ + 2 NO → N₂ + 2 H₂O

(7)

H₂ + 2 NO → N₂O + H₂O

(8)

5/2 H₂ + NO → NH₃ + H₂O

(9)

It is important to note that the H₂ inlet concentration was 1% compared to 400 ppm NO. At 50 °C, the H₂ conversion was very limited, while the NO_X conversion had already reached 10% (a theoretical H₂ conversion of 0.2% is required to convert 10% of the inlet NO into the N₂O, Equation (8)). The H₂ conversion curve (black line) showed a more significant increase from 80 °C, which also corresponded to significant ammonia outlet detection. The highest ratio of hydrogen utilization toward NO_X reduction was observed at 100 °C, when there was sufficient H₂ activation, but limited O₂ activation. Maximum ammonia emission was also observed at this temperature. The initial increase in N₂ yield followed the increase in H₂ conversion. For temperatures above 250 °C, the NO_X conversion decreased, indicating that the available reductant for NO_X reduction became insufficient, likely due to competition with the H₂ oxidation by O₂.

The H₂ + O₂ reaction (without NO) was subsequently examined on 0.30%Pt/Siral-5. The H₂ conversion curves with or without NO in the feed stream (H₂-SCR vs. H₂ + O₂ reactions) are compared in Figure 5b. The H₂ conversions at 50 °C were 4% and 82% in H₂-SCR (black line) and H₂ + O₂ reactions (red line), respectively (shift of T₅₀ from 28 °C to 98 °C). These results clearly indicate that the presence of NO in the feed stream inhibited the hydrogen conversion by oxygen at a low temperature. These specific statements are discussed in the next section which is a focused on the low-temperature behaviours considering the N₂, NO, O₂ interactions over platinum-based catalysts.

2.4. Discussion

Many studies have examined the interactions of O₂, H₂ and NO with a platinum surface in the past. Although reported heat of adsorption or sticking coefficient values vary significantly depending on the considered platinum-based sample, the general trends clearly indicate that, at room temperature, oxygen shows the strongest interaction with platinum. O₂ adsorption is dissociative on platinum, with a reported heat of adsorption ranging from 300 to 370 kJ mol⁻¹ [23] to 210 kJ mol⁻¹ [24]. For supported platinum species, adsorption enthalpies are typically in the 210–280 kJ mol⁻¹ range [24,25,26], mainly depending on the platinum dispersion. Notably, Frank et al. [8] reported a significantly lower adsorption enthalpy of 97 kJ mol⁻¹. They explained this value by the fact that the heat of adsorption is lower than the activation energy required for O₂ desorption because the dissociative oxygen adsorption is an activated process. Only in the case of non-activated adsorption, the heat of adsorption can be directly compared to the desorption energy. Interestingly, it has also been reported that the oxygen coverage of Pt(111) surface is limited to 0.25 [27], in accordance with the fact that the saturation coverage of oxygen has been reported to be lower than that of hydrogen [28]. Indeed, a complete hydrogen surface coverage of platinum via H₂ dissociative adsorption can be obtained [29]. Again, reported heats of adsorption for H₂ on platinum vary significantly across the studies, but they appear always lower than those of O₂, typically in the 56–87 kJ mol⁻¹ range [8,27,29,30]. No significant trend with platinum crystallite size was observed over a range of 1.3 to 22.3 nm Pt deposited on SiO₂ [30].

Concerning the NO adsorption on platinum, both molecular and dissociative adsorption have been proposed at room temperature. Fiorin et al. [26] reported that NO initially adsorbs molecularly on Pt(111), and dissociatively on Pt(211) (stepped surfaces) and Pt(411) (unit cell with two terraces), with a higher amount of dissociated NO observed on Pt(411) compared to Pt(211), This aligns with previous studies showing that step and/or terraces sites enhance the reactivity of NO [31,32]. The calculated adsorption energies at the fcc threefold sites are around 170–180 kJ mol⁻¹ [23,33,34] and smaller at the bridge site (140 kJ mol⁻¹) and at the atop site (62 kJ mol⁻¹) [33]. Consequently, as mentioned by Dhainaut et al., the competition between H₂ and NO for adsorption on Pt(111) was largely in favour of NO, especially considering that the probability for finding two adjacent sites for the H₂ dissociative adsorption would be very low [35].

Finally, considering O₂, H₂ and NO, it can be concluded that O₂ exhibits the strongest interactions with platinum, but with a limited maximum coverage (0.25 on Pt(111)) which allows the co-adsorption of other molecules. This limited coverage by oxygen is consistent with the fact that NO conversion at low temperatures appeared unaffected by the presence of oxygen (Figure 4a,b). Considering the competition between H₂ and NO, the NO adsorption is favoured, which is consistent with the fact that the presence of NO inhibited the H₂ + O₂ reaction at a low temperature (Figure 5b). Additionally, in the presence of O₂ and NO, the hydrogen adsorption should be limited, which could suggest an Eley–Rideal type mechanism for the NO_X reduction. High nitric oxide and oxygen coverage is also consistent with the high N₂O selectivity observed at a low temperature (close to 100% below 70 °C), probably via the formation of the (NO)₂ dimer species as mentioned in the Introduction section [10].

The possible participation of NO_X species stored on the Siral-5 support was also examined through NO_X-TPD experiments. The NO_X desorption profiles recorded with the Siral-5 support and the 0.15 _wt%Pt/Siral-5 catalyst are reported Figure 6. Despite its acidic behaviour (Section 2.1.1, Appendix B.2), the Siral-5 support is able to store significant amounts of NO_X species, around 950 µmol g⁻¹. The NO_X desorption from Siral-5 support without platinum started near 200 °C and showed a maximum at 425 °C (blue curve). The addition of platinum mainly led to a shift in the starting desorption temperature down to 180 °C, and the whole profile was shifted to approximately 30 °C to lower temperatures compared to the support alone (red curve). Consequently, NO_X adsorption clearly occurred on the silica-alumina support, but these species appear stable at 100 °C and below. Therefore, we can assume that they do not intervene to a large extent in the deNO_X activity at a low temperature.

2.4.1. Identification of the Main Active Sites Responsible for the Low Temperature H₂-SCR Activity and the NH₃-SCR Activity

Varying the catalyst, and particularly the platinum dispersion, can reveal a possible sensitivity of a reaction pathway to the size and shape of metal Pt particles. With this aim, several catalysts were prepared. For the four considered alumina and silica-alumina supports, a new series of catalysts with 0.15% platinum loading were prepared. Additionally, the platinum loading was also decreased down to 0.07% for the selected Siral-5 support. Their mean platinum particle sizes were determined and reported in Table 1. However, probably due to the rather severe hydrothermal treatment, the obtained platinum size range remained relatively limited. To enlarge it, additional catalysts were considered: a 0.30%Pt/Siral-5 sample treated only under pure N₂ at 700 °C (without hydrotreatment at 700 °C), and a commercial reduced 1%Pt/Al₂O₃. This allowed for a broader distribution of platinum particle sizes (Table 1).

All these 11 catalysts were evaluated in H₂-SCR and NH₃ oxidation tests. To avoid full NO_X conversion at a low temperature, catalysts with platinum loading until 0.30% were evaluated involving 30 mg, while 5 mg were used for the 1% Pt/Al₂O₃ catalyst. To study the initial low temperature H₂-SCR behaviours, the selected temperature was 60 °C for which the NO_X conversions were between 31% and 73%, while the ammonia production was evaluated at 100 °C. For the NH₃ oxidation tests, a temperature of 180 °C was selected. The corresponding catalytic data are reported in

Table A2. Platinum characterization and catalytic behaviours of the various selected catalysts.

Catalyst	SBET (m2 g−1)	Pt Loading * (wt%)	Mean Pt Particle Size (nm) **	Pt Dispersion (%)	H2 SCR Test ***: NOx Conv. at 60 °C (%)	H2 SCR Test ***: N2Ooulet at 60 °C (ppm)	H2 SCR Test ***: NH3 oulet at 100 °C (ppm)	NH3 Oxidation Test ***: NH3 Conv. at 180 °C (%)
0.30wt%Pt/Al2O3	144	0.39	7.0/-	19	95	140	140	56
0.30wt%Pt/Siral-5	218	0.34	10.4/10.6	13	63	95	149	41
0.30wt%Pt/Siral-20	245	0.30	11.8/-	11	73	129	80	99
0.30wt%Pt/Siral-40	232	0.32	13.4/-	10	76	113	141	55
0.15wt%Pt/Al2O3	149	0.13	-/4.5	27	57	85	136	13
0.15wt%Pt/Siral-5	225	0.16	11.4/11.2	12	31	48	60	40
0.15wt%Pt/Siral-20	239	0.15	-/10.9	12	42	64	69	38
0.15wt%Pt/Siral-40	230	0.13	14.3/13.5	9	33	49	116	33
0.07wt%Pt/Siral-5	269	0.07	8.2/8.4	15	33	52	76	13
0.30wt%Pt/Siral-5 w/o hydrotreatment	224	0.23	1.9/-	56	37	56	71	18
1wt%Pt/Al2O3 (commercial) **	142	0.90	-/2.6	44	46	76	149	9

* From ICP analysis. ** mean value from H₂ chimisorption and TEM analysis when both pieces of data were available. *** Catalytic tests were performed with 30 mg of catalyst, except for the 1%Pt/Al₂O₃ sample (5 mg).

(Appendix C).

The turnover frequency (TOF), which corresponds to the deNO_X activity (mol s⁻¹ g⁻¹) divided by the amount of surface platinum sites (mol g⁻¹) is usually suitable to evaluate which kind of platinum sites are involved in a reaction when plotted depending on the metallic particle size. With this aim, Figure A4a,b report the evolution of the surface site concentration and distribution depending on the particle size of a perfect fcc truncated octahedron (from [36]). The turnover frequency (TOF) for the deNO_X activity at 60 °C was plotted in Figure A5 depending on the platinum particle size. This figure shows a general increase in the TOF with the particle size increase, but the points are rather dispersed, and the trend does not match well with any evolution of specific platinum sites.

Then, since the considered catalysts exhibited large ranges of specific surface areas and platinum loadings, the influence of the platinum density was examined by plotting the NO_X conversion rate at 60 °C (mol s⁻¹ g_cata⁻¹) depending on the amount of platinum per square metre (Figure 7a). This figure shows that a high platinum density favours the NO_X conversion. It can be then assumed that interactions between platinum particles are beneficial, probably via the mobility of the surface species. It should be noted that no relevant relationship between the NO_X conversion rate at 60 °C and the acidity of the catalysts was obtained, which may be attributed to (i) the relatively narrow range in support acidities or (ii) a metal-intensive catalytic reaction.

To take into account the platinum density previously evidenced, the NO_X conversion rate was expressed in mol s⁻¹ m⁻² and normalized by the amount of platinum (mol_Pt g⁻¹). Results are plotted in Figure 8a, depending on the platinum particle size. Practically, the total amount of surface platinum globally varies with the square of the particle size. Consequently, the total surface site normalized by mass of platinum decreases with the particle size, as also illustrated in Figure A4a considering the addition of all the sites. However, the profile reported in Figure 8a does not respect this evolution and the NO_X conversion rate appears depending mainly on specific sites, for which the proportion varies with the particle size. Figure A4 shows the surface site evolution calculated for corners, edges, (100) faces and (111) faces. Clearly, corners cannot be responsible for the NO_X conversion at 60 °C. In contrast, the trend observed in Figure 8a appears consistent with those reported for edges, (100) faces and (111) faces. However, Figure A4 predicts the maximum surface concentration for edges and (111) faces for platinum particle sizes around 2 nm, which does not perfectively match the evolutions reported in Figure 8a. Then, this suggest that NO_X conversion probably depends on a mix of these face and edge sites.

As expected, since the major product at 60 °C was N₂O, Figure 8b shows that same platinum sites are involved in N₂O formation. Note that the observed trend takes into account all the various supports, with no specific behaviour depending on the oxide support. This observation indicates that the reaction at low temperature is governed predominantly by the platinum phase, rather than the support properties.

Since ammonia appeared to play a key role as the intermediate species in the H₂-SCR pathway, its formation was also tentatively considered. Due to the very narrow temperature range between the ammonia formation and its subsequent reactivity, a temperature of 100 °C was selected for evaluation, as it generally corresponds to the maximum of ammonia emission. The NH₃ formation rates (mol_NH3 s⁻¹ m⁻²) normalized by the amount of platinum (mol_Pt g⁻¹) are plotted in Figure A6 depending on the platinum particle size. Once again, it appears that the ammonia emission at 100 °C followed the same trend as the NO reduction at 60 °C (mainly into N₂O), indicating that the same platinum sites were involved. However, points in Figure A6 are more scattered, likely due to the partial reactivity of ammonia even at 100 °C.

Ammonia was previously shown to react mainly with oxygen (Section 2.3.3). Therefore, the ammonia oxidation was also investigated depending on the platinum particle size. With this aim, the NH₃ + O₂ reaction was examined at 180 °C, a temperature for which the NH₃ conversions were between 9% and 56% (5 mg of catalyst were used for 1% Pt loaded catalyst while 30 mg were used for the sample with lower platinum loadings). As previously performed for the H₂-SCR tests, the NH₃ conversion rates at 180 °C (mol s⁻¹ m⁻²) normalized by the amount of platinum (mol_Pt g⁻¹) were plotted depending on the platinum particle size (Figure A7). Unlike the previous obtained trends, points appear scattered, and a general increase can be observed, especially when the 0.3Pt/Al₂O₃ catalyst (d_Pt = 7 nm) is not considered. This result suggests that ammonia oxidation does not occur on the same platinum sites compared to the previous considered H₂-SCR reaction. It is assumed that catalytic behaviour is multifactorial (specific surface area, acid properties, platinum dispersion…). Particularly, Figure 2 (H₂-SCR over 0.30Pt/support catalysts at 250 °C, m_cata = 200 mg) showed that NO_X conversion in H₂-SCR and especially N₂ selectivity appeared mainly dependant on the acidic properties of the support. More precisely, Figure A2 shows that the more acidic support maintained a high level of NO_X conversion over a wider temperature window. In this relatively high temperature range, the NH₃ + O₂ reaction was demonstrated to be the main one responsible for the observed behaviour (Figure 4). However, despite extensive calculations involving the NH₃ + O₂ conversion rates at 180 °C and the acidity behaviour, no clear correlation has been obtained. The only general trend obtained for the ammonia oxidation behaviour at 180 °C is an increase in the NH₃ conversion rate normalized by the amount of platinum (mol_NH3 s⁻¹ mol_Pt⁻¹) with the increase in the platinum particle size (Figure A8). This indicates that large platinum particle sizes appear more active in ammonia oxidation by oxygen. Considering the platinum site distribution (Figure A4), this trend does not match the contribution of corners and edges, but rather suggests the involvement of faces. In addition, a large particle size could also be associated with a higher metallic state. Note that the catalysts exhibiting the larger platinum particle sizes are not associated with any specific support (Table 1).

2.4.2. Overview of the H₂-SCR Reaction Pathways Depending on Temperature

In order to have an overview of the varied reactions involved in the H₂-SCR process, the main conversion/yields curves are compared in Figure 9 depending on the reactional mixture.

Oxygen from the gas phase had no influence on the low temperature NO conversion (50–100 °C). This observation strongly suggests that NO oxidation was not involved in the catalytic activity, in contrast to the oxidation–reduction mechanism. In accordance with this assumption, no NO₂ was detected in the gas phase. In addition, O₂ in the gas phase did not affect the N₂O yields (Figure 5) and the selectivity in N₂O remained close to 100% up to 60 °C. A few ppm of ammonia were detected from this temperature. The reactivity of ammonia was then very low. Indeed, the NH₃-SCR activity was observed from around 70 °C and became significant only from 100 °C. Consequently, below 100 °C, the dissociative mechanism (including the formation of NH₃) is more probable than the oxidation–reduction one.

Since the NO_X conversion in H₂-SCR reached 100% from 100 °C, the availability of NO_X for the NH₃-SCR reaction was limited to a very narrow temperature window (mostly below 100 °C, while the activity in NH₃-SCR was still limited). As a result, the maximum outlet NH₃ was observed around 100 °C and the role of the NH₃-SCR in the whole H₂-SCR process appeared very limited. On the contrary, the ammonia oxidation by O₂, which started near 120 °C, significantly contributed to the H₂-SCR process. This reaction appeared responsible for the second N₂O emission peak (150–500 °C, Figure 5), and also facilitated the N₂ formation as the main product. The optimal N₂ yield was observed when H₂ conversion reached 100%, around 150 °C, which can be attributed to the optimal ammonia formation followed by its oxidation by O₂. These results are in accordance with the very recent work of Dong et al. on Pt/SSZ-13 catalysts [20]. They propose that the NH₃-SCR route, resulting mainly in N₂O formation and limited N₂ selectivity occurs mainly below 150 °C, while the oxidation of the generated NH_X⁺ species at a higher temperature favours N₂ formation. These various reaction pathways are finally summarized in Figure 10.

3. Materials and Methods

3.1. Catalysts

Alumina and three silica-alumina supports with silica weight loadings of 5%, 20% and 40% (denoted as Siral-5, Siral-20 and Siral-40, respectively), all provided by Sasol, were selected for this study. These supports were in the form of fine powders with grain sizes below 100 μm. Prior to use, they were firstly hydrothermally treated at 700 °C for 4 h (synthetic air flow + 10% H₂O; heating rate: 5 °C min⁻¹). Platinum was then added using the incipient wetness impregnation method. The desired amount of metal precursor (Pt(NH₃)₂(NO₂)₂; Sigma Aldrich, Darmstadt, Germany) was dissolved in ultrapure water and then added dropwise to the support, which had been previously moistened with ultrapure water, to reach Pt loadings of 0.07, 0.15 and 0.30 _wt%. The mixture was thereafter stirred magnetically for 30 min to ensure a homogeneous distribution of the precursor. The solvent was then evaporated at 80 °C using a sand bath. The resulting powder was dried overnight at 120 °C in an oven and then treated at 700 °C under a nitrogen atmosphere (60 mL min⁻¹) for 4 h to decompose the nitrate precursors and stabilize the metal on the support surface [37]. Finally, the catalyst was hydrotreated at 600 °C for 4 h under synthetic air with 10% H₂O.

The resulting powder was then subjected to pelleting, grinding and sieving to achieve a controlled grain size between 400 and 200 μm. Additionally, a commercial 1%Pt/Al₂O₃ catalyst (Alfa Aesar, reduced, 300 m² g⁻¹) was also evaluated after a treatment under 10% H₂ at 600 °C for 1 h.

3.2. Characterizations

Elemental analyses were performed using an Agilent 5100 apparatus (Santa Clara, CA, USA). Samples were previously mineralized in a HNO₃-HCl mixture, assisted by a microwave heating.

Textural behaviours were assessed by nitrogen physisorption performed with a TRISTAR 2 set-up (Micromeritics, Norcross, Georgia, USA). The sample was first degassed under vacuum at 250 °C for 2 h to remove previously adsorbed gases or contaminants. Subsequently, the sample was then exposed to increasing doses of nitrogen gas at −196 °C. Once adsorption was completed, the pressure was gradually reduced. The resulting adsorption and desorption isotherms were analyzed to determine the material’s specific surface area and pore size distribution.

Acidic properties were evaluated by means of pyridine adsorption monitored by FTIR spectroscopy. IR spectra were recorded in a Nexus Nicolet spectrometer (Themo Fisher, Waltham, MA, USA) equipped with a DTGS detector (Deuterium TriGlyceride Sulphur) and KBr beam splitter with a resolution of 4 cm⁻¹ and 64 scans. The spectra presented were normalized to a disc of 10 mg cm⁻². After sample activation at 450 °C under vacuum, pyridine was adsorbed at 50 °C (P = 200 Pa) and desorption was performed up to 450 °C. From pyridine adsorption, the Lewis and Brønsted acid sites (BAS and LAS, respectively) concentration are determined by the evolution of the ν_19b band area (1456 or 1546 cm⁻¹, respectively) using its molar coefficients previously reported [38].

The platinum particle mean sizes were estimated from transmission electron microscopy measurements. Micrographs were recorded on a JEOL 2100 instrument (Peabody, MA, USA) operated at 200 kV with a LaB6 source. For grid preparation, the sample was dispersed in ethanol under an ultrasonic field, and a drop of the upper suspension was deposited on the grid. The mean particle size was determined by considering generally between 300 and 600 particles for each catalyst. Hydrogen chemisorption measurements were also performed using a Micromeritics Autochem 2920II instrument (Norcross, GA, USA) to evaluate the accessible metallic surface. A catalyst sample weighing between 200 and 500 mg was placed in a U-shaped quartz reactor between two quartz wool beds. Downstream of the reactor, a metallic water trap containing magnesium perchlorate was installed to prevent water from reaching the detector. The sample was in situ pretreated for 1 h at 500 °C (5 °C min⁻¹) under a hydrogen flow (30 mL min⁻¹) to ensure a metallic state. Following this, the sample was purged under argon at the same temperature and flow rate (30 mL min⁻¹) for 30 min and subsequently cooled to 30 °C. The measurements were then performed at 30 °C by pulsed injections (0.0551 mL) of 10% H₂/Ar until saturation (20 pulses). After a purge under argon at 30 °C, a second series of pulses was performed to subtract the possible physisorbed hydrogen contribution. Hydrogen consumption was monitored using a thermal conductivity detector (TCD). Calculations for dispersion and particle size were based on the model developed by Le Valent et al. [39]. Due to the low platinum loading, the reported results correspond to a mean value of at least three measurements.

3.3. Catalytic Tests

The same set-up was used to evaluate various reactions such as H₂-SCR, ammonia formation under NO + H₂ (without O₂), NH₃-SCR, NH₃ oxidation by O₂, and NO oxidation. The corresponding mixtures are depicted in Table 2. For all experimental conditions, the total flow rate was 150 mL min⁻¹ regulated by electronic mass-flow controllers.

Various controlled amounts of catalyst, ranging from 5 to 200 mg, were used for the catalytic tests. All tests were performed maintaining a constant catalytic bed volume, corresponding to that obtained with 200 mg of catalyst. For catalyst masses below 200 mg, silicon carbide (SiC) with the same grain size as the catalyst was added to maintain the bed volume. The catalyst was then placed on a bed of quartz wool inside the reactor.

Before each test, the sample underwent a pretreatment until 500 °C for 30 min (heating rate: 5 °C min⁻¹) using the same reactional mixture as for the subsequent catalytic test. Catalytic behaviours were then evaluated by decreasing the temperature from 500 °C to 50 °C, with stepwise measurements at constant temperatures (even if results are later discussed in order of increasing temperature). The continuous monitoring of the feed gas and effluent stream compositions was performed using an online MKS 2030 Multigas infrared analyser for NO, NO₂, N₂O and NH₃. Additionally, an OmniStar mass spectrometer (Pfeiffer Vacuum, Asslar, Germany) was used to record the H₂ signal. N₂ formation was calculated with the assumption that only NO, NO₂, N₂O and NH₃ were formed as N-compounds. All equations employed for calculating conversions, yields and selectivity are displayed in Appendix A.

4. Conclusions

The pathways and mechanistic aspects of the H₂-SCR over low loaded platinum-based catalysts (0.07–0.3%) was investigated over a large temperature range (50–500 °C), with a special focus on the role of NH₃ as a possible intermediate species, the origin of the undesired N₂O emission and the platinum active sites involved in the deNO_X reactions. It was found that oxygen from the gas phase had no influence on the low temperature NO_X conversion (50–80 °C), with the same N₂O yields close to 100% up to 60 °C. Ammonia formation was observed at relatively low temperatures (starting from 60 °C) but its reactivity was then limited. All these low temperature reactions were associated with the same platinum sites, i.e., mainly the platinum faces and/or edges. The maximum outlet NH₃ was observed around 100 °C. In opposition to conclusions frequently reported in the literature, the role of the NH₃-SCR in the whole H₂-SCR process appeared limited to a very narrow temperature range (more or less below 120 °C). On the contrary, the H₂-SCR process at a high temperature appeared governed by the ammonia oxidation by O₂, which started near 120 °C and contributed to the second N₂O emission peak (150–500 °C). This reaction does not involve similar active sites to the initial H₂-SCR reaction route and appeared mainly dependant on the platinum particle size.