Low-Temperature NO_x Reduction by H₂ on Mo-Promoted Pt/ZrO₂ Catalysts in Lean Exhaust Gases

The samples were prepared by means of incipient wetness impregnation using monoclinic ZrO₂ as a support (Saint-Gobain NorPro). First, the Mo promoter was introduced by an aqueous solution of (NH₄)₆Mo₇O₂₄ · 4 H₂O (Merck) with different concentrations to obtain Mo loads between 0 and 10 wt%. After drying for at least 4 h at 90 °C in air, each sample was impregnated with a Pt(NO₃)₂ solution (Chempur), resulting in a Pt proportion of 0.25 wt.% referred to the bare support. Note that preliminary investigations, shown in section S1, indicated no significant re-dissolution of the Mo precursor during the subsequent impregnation with the aqueous Pt solution. Pt activation of the 2 g samples was carried out in a flowing mixture of 10 vol% H₂ and 90 vol% N₂ (400 ml/min, STP) while ramping the temperature to 300 °C at a rate of 1.7 K/min and holding that temperature for 30 min. Finally, the samples were calcined for 5 h at 500 °C in static air. In this paper, the Pt-containing catalysts are denoted as Pt/xMo/ZrO₂, where x refers to the loading of Mo.

The characterization of the catalysts was performed prior to the H₂-deNO_x investigations, since exposure to the lean model exhaust gas at a maximum temperature of 280 °C is unlikely to alter the samples calcined at 500 °C before. This aspect was checked in preliminary studies.

Powder X-ray diffraction (PXRD) was performed on a D8 Discover (Bruker-AXS) with a Bragg–Brentano configuration, Fe-filtered Co-Kα radiation and a VANTEC-1 detector. Diffractograms were taken from 10° to 80° in 2θ mode with a step width of 0.5° and an integration time of 250 s. The PXRD patterns were evaluated using the Powder Diffraction Files (PDF) database.

N₂ physisorption was carried out on a TriStar II (Micromeritics). Respective sample was degassed in vacuo (10^–1 mbar) at 350 °C for 16 h to remove adsorbed components, then the N₂ adsorption isotherms were taken at − 196 °C. The BET surface area (S_BET) was determined from the adsorption data recorded at p/p₀ ratios from 0.05 to 0.20.

Laser Raman spectroscopy (LRS) was conducted with an inVia Raman microscope (Renishaw) equipped with a Nd:YAG laser (532 nm, 100 mW), a grating with 1800 lines per mm and a CCD array detector. The spectra were collected under ambient conditions from 10 to 1800 cm⁻¹ at a resolution of 1.6 cm⁻¹, an exposure time of 120 s and a laser power of 10 mW, accumulating 3 scans per spectrum.

Temperature-programmed desorption of CO (CO-TPD) was performed on a home-made laboratory rig. A granulated sample (300 mg, 125–250 µm) was introduced into the U-shaped quartz glass tube reactor (i.d. 8 mm), packed as a fixed bed and pre-treated in 2 vol% O₂ (He balance) at 450 °C for 30 min (400 ml/min, STP). Then, it was cooled to − 196 °C with liquid N₂ to reduce spill-over effects from Pt to the promoter or support during the CO exposure. Saturation with CO was performed by applying pulses of 200 ppm CO in He for 9 s each followed by intermediate purging with He for 3 min (400 ml/min, STP). After the final flushing with He, the TPD was started (150 ml/min, STP) by heating the sample to 550 °C at a rate of 20 K/min. The temperature was recorded by a K-type thermocouple located directly above the sample. Desorbing CO and CO₂ were monitored using a non-dispersive infrared (NDIR) spectrometer (X-Stream Enhanced XEGK, Emerson). Blank CO-TPD studies of the bare Mo/ZrO₂ samples provided CO desorption profiles very similar to the Pt/xMo/ZrO₂ samples without significant amounts of CO₂. The evolution of CO₂ upon TPD is ascribed to the oxidation of CO by the MoO_x/ZrO₂ surface. The blank CO_x profiles were subtracted from the respective TPD traces of the Pt catalysts. The resulting CO_x desorption profiles were integrated to quantify the available Pt sites (n_a(Pt)), assuming a CO/Pt adsorption stoichiometry of 1 [17]. From n_a(Pt) and the total abundance of platinum (n_Pt), the Pt dispersion (D_Pt = n_a(Pt) / n_Pt) was estimated, while the mean size of Pt particles (d_Pt = 6 V_Pt/(a_Pt ∙ D_Pt)) was calculated assuming spheres; V_Pt is the volume of a Pt atom present in the bulk metal (15.10 ų), a_Pt represents the surface area of a Pt atom located on a polycrystalline surface (8.07 Ų) [17].

The temperature-programmed desorption of NH₃ (NH₃-TPD) was also carried out on a home-made rig. A granulated sample (500 mg, 125–250 µm) was introduced into the quartz glass tube reactor (i.d. 8 mm), fixed with quartz wool and pre-treated with 2 vol% O₂ in N₂ at 450 °C for 30 min (500 ml/min, STP). After cooling to 130 °C in a N₂ flow, the sample was saturated with 1000 ppm NH₃ in N₂ (500 ml/min, STP). Weakly bound NH₃ was removed by flushing with N₂ until the outlet NH₃ fraction was below 2 ppm. For the final NH₃-TPD, the sample was heated to 550 °C at a rate of 10 K/min in N₂ (500 ml/min, STP). The temperature was measured by two K-type thermocouples located directly in front of and behind the catalyst bed, respectively. Desorbing NH₃ was monitored by means of NDIR spectroscopy (X-Stream, Emerson). Assuming a molar NH₃ to acid site stoichiometry of 1, the number of surface acid sites was estimated from the integrated NH₃ traces.

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was conducted using NH₃ as a molecular probe. The investigations were performed on a Tensor 27 FTIR spectrometer (Bruker) equipped with Praying Mantis reflectance optics (Harrick Scientific) and a MCT detector. The sample compartment was continuously purged with N₂ to avoid the diffusion of air. Each sample was introduced into the heatable stainless steel IR cell (Harrick Scientific) equipped with ZnSe windows. After pre-treatment with 2 vol% O₂ in N₂ at 450 °C for 30 min (200 ml/min, STP), the sample was cooled to 180 °C, 130 °C, 90 °C and 50 °C in a N₂ flow, a background spectrum being collected at each temperature. Subsequently, the sample was exposed to 1000 ppm of NH₃ (N₂ balance) for 30 min at 50 °C. After flushing with N₂ for another 30 min at 50 °C to remove physisorbed NH₃, an IR spectrum was recorded. The temperature was then increased to 90 °C and the sample was purged again with N₂ for 30 min before another spectrum was taken. The same procedure was adopted at 130 °C and 180 °C. The total gas flow was always kept at 200 ml/min (STP). Spectra were collected from 800 to 4000 cm⁻¹ with a resolution of 1 cm⁻¹, and 200 scans were accumulated per spectrum, resulting in an acquisition time of approx. 5 min.

Catalytic investigations were performed on a laboratory bench using a lean model exhaust gas. The granulated samples (200 mg, 125–250 μm) were introduced into the quartz glass tube reactor (i.d. 8 mm), fixed with quartz wool and heated in an N₂ flow to 280 °C for 20 min. Subsequently, the synthetic exhaust, composed of 200 ppm NO, 2000 ppm H₂, 5 vol% O₂, 10 vol% H₂O and N₂ (balance), was dosed using independent mass flow controllers (Bronkhorst). The total flow was kept at 400 ml/min (STP), corresponding to a space velocity (GHSV) of 160,000 h⁻¹. Once a steady state was reached at 280 °C, the temperature was linearly reduced to 60 °C at a rate of 1.7 K/min. Note that this test procedure provided a quasi-steady state, as proven by preliminary studies. The temperature was measured using two K-type thermocouples located directly in front of and behind the catalyst bed. The difference between inlet and outlet temperature was always below 8 K. All H₂-deNO_x data are referred to the averaged temperature. To investigate the impact of CO on the H₂-deNO_x performance of Pt/xMo/ZrO₂, 200 ppm CO was added to the lean model exhaust. For reference purposes, SCR experiments with and without 2000 ppm H₂ were also performed by adding 100 ppm NO and 100 ppm NH₃ to the model exhaust. After pre-treatment at 280 °C, each temperature was kept constant for 45 min to achieve steady-state conditions excluding NH₃ storage effects.

Gas-phase analysis was performed using a Multi-Gas 2030 FTIR spectrometer (MKS Instruments), detecting NO, NO₂, N₂O, NH₃ and H₂O, whereas O₂ was monitored with an LSU 4.9 lambda sensor (Bosch). Inlet concentrations were checked employing the reactor bypass. To assess the catalysts, NO_x conversion (X(NO_x) = 1- y(NO_x)_out/y(NO_x)_in), maximum NO_x conversion (X_max) and a temperature of X_max (T(X_max)) were used. Additionally, the selectivity of N₂O (S(N₂O) = 2·y(N₂O)_out/(y(NO_x)_in – y(NO_x)_out)), NH₃ (S(NH₃) = y(NH₃)_out/(y(NO_x)_in – y(NO_x)_out)) and N₂ (S(N₂)) were calculated. Since N₂ is not detectable using FTIR, S(N₂) was derived from the mass balance of N including the measured species NO, NO₂, NH₃ and N₂O, i.e. S(N₂) = 1- (2·y(N₂O)_out + 2·y(NH₃)_out) /(y(NO_x)_in – y(NO_x)_out).

The most important H₂-deNO_x features of the Pt/xMo/ZrO₂ catalysts are demonstrated in Fig. 1, indicating that the maximum NO_x conversion and N₂ selectivity at X_max increase when the Mo load (w(Mo)) grows to 3 wt%, and decrease at higher Mo contents. As a result, Pt/3Mo/ZrO₂ displays the highest H₂-deNO_x activity and N₂ selectivity, including X_max of approx. 90%, T(X_max) of 130 °C and S(N₂) of 78% at X_max. For clarity, the H₂-deNO_x performance of Pt/3Mo/ZrO₂ is shown in Fig. 2. This reveals a broad activity window with deNO_x above 30% between 105 °C and 200 °C. Furthermore, Pt/3Mo/ZrO₂ strongly suppresses the NH₃ formation (maximum S(NH₃) = 2%), an effect that is markedly pronounced for the 0.25Pt/ZrO₂ reference (maximum S(NH₃) = 50% at 160 °C). The catalysts with Mo proportions above 3 wt% do not produce NH₃ at all, while a N₂O selectivity of 22% still remains.

The performance of Pt/3Mo/ZrO₂ is similar to that of the highly active WO_x-modified Pt/ZrO₂ catalysts, which also provide NO_x conversions up to 90% between 100 °C and 180 °C, but are characterized by higher N₂ selectivity (80–90%) [11, 18]. A N₂ selectivity above 90% was reported for Pt/MgO-CeO, albeit using a markedly higher H₂ proportion of 1 vol% [6]. Note that for the WO_x-promoted catalysts, a rise in S(N₂) with increasing H₂ concentration was also demonstrated [9, 11, 18], while at the same time the selectivity of H₂ to deNO_x drastically decreases.

Additional H₂-deNO_x investigations performed with 200 ppm CO in the model exhaust led to a drastic decline in the catalytic activity of the Pt/xMo/ZrO₂ catalysts, especially at low temperatures (Fig. S2). Since CO strongly covers active Pt sites at low temperatures [19], significant deNO_x activity is only observed above 135 °C, when CO is almost completely oxidized and free Pt sites become accessible for the NO_x reduction. Note that this experiment was performed with decreasing temperatures. Contrary, when heating up the sample the CO light-off might be slightly shifted to higher temperatures potentially compromising the NO_x conversion. However, even above the CO light-off temperatures, the presence of CO causes a diminished NO_x conversion due to the competing adsorption on the active Pt sites; for instance, the 0.25/3Mo/ZrO₂ sample only achieves a X_max of 50% appearing at 163 °C. CO and HC were reported as having a similar effect on the H₂-deNO_x reaction in the case of the Pt/W/ZrO₂ catalysts [1, 18]. Thus, for engines operating with hydrocarbon-based fuels such as gasoline and diesel, an oxidation catalyst upstream of the H₂-deNO_x stage is mandatory to ensure that the H₂-deNO_x is efficient at low temperatures.

The LRS investigations of the Pt/xMo/ZrO₂ samples (Fig. 3) indicated that the ZrO₂ carrier is exclusively present in the monoclinic phase as indicated by the peaks at 176–184 cm⁻¹ (B_g), 310 cm⁻¹ (A_g), 335 cm⁻¹ (B_g), 382 cm⁻¹ (B_g), 474 cm⁻¹ (A_g), 536 cm⁻¹ (B_g), 560 cm⁻¹ (A_g), 618 cm⁻¹ (B_g) and 635 cm⁻¹ (A_g) [20‐22]. This interpretation is substantiated by PXRD, which shows the reflexes of monoclinic ZrO₂ (PDF-04-004-4339) (Fig. S3). Moreover, the samples with low Mo loadings (≤ 3 wt% Mo) provide additional LRS peaks at 860 cm⁻¹ (Mo-O-Mo vibrations) and 930 cm⁻¹ (M=O vibrations), ascribed to highly dispersed Mo oxide species [23, 24]. These peaks are shifted from 860 to 889 cm⁻¹ and from 930 to 960 cm⁻¹ when the Mo proportion is raised to 7 wt% reasonably suggesting aggregated MoO_x entities, which likely exist in the form of large clusters [23]. For the highest Mo content of 10 wt%, exceeding the estimated theoretical monolayer of MoO₃ (equivalent to 8 wt% Mo) [25], the formation of crystalline MoO₃ occurs. This is shown by the LRS peaks at 748 cm⁻¹ (ν_s(O-Mo-O)) and 947 cm⁻¹ (ν_as(O-Mo-O) [23, 24] and is also suggested by the PXRD reflex at 26.9° (Fig. S3), which is attributed to the (112) plane of β-MoO₃ (PDF-04-007-2607).

Figure 4 (left) illustrates that the MoO_x promoter steadily increases the BET surface area of the samples when the Mo proportion rises from 0 wt% (65 m²/g) to 4 wt% (101 m²/g). This effect is ascribed to the inhibition of the sintering of ZrO₂ by MoO_x, which is known to reduce the number and mobility of defects on the support surface [26]. However, when Mo loadings increase further, the BET surface area continuously decreases, probably due to the aggregation of the MoO_x species, which have a lower specific surface area compared to ZrO₂ [27].

CO-TPD studies show that the number of active Pt sites is significantly increased by promoting the bare Pt/ZrO₂ (Fig. 4, right), going through a maximum of 7.1 μmol/g at a Mo load of 3 wt%. Table S4 indicates that the Pt dispersion obviously follows the same trend including maximum dispersion of 56% for the catalyst with 3 wt% Mo (14% for bare Pt/ZrO₂), while the estimated Pt particle size is decreased to a minimum of 2 nm for the same sample. It is important to note that the clear decrease in the number of available Pt sites at Mo loadings above 3 wt% cannot be entirely attributed to the decline in the BET surface area (Fig. 4, left). Since the increase in the molybdenum content is connected with the aggregation of the Mo oxide species (Fig. 3), it may be supposed that already during the impregnation of the ZrO₂ support Mo clusters form at the higher Mo loads. Then, electrostatic repulsion of the positively charged Mo cores of these clusters and Pt²⁺ cations may occur during subsequent treatment with the Pt solution. With this assumption it may be speculated that the Pt cations are more concentrated on the reduced surface area of the bare ZrO₂ and at the interface to the Mo clusters, respectively, leading to agglomeration of the Pt particles upon the following activation and calcination.

Previous mechanistic studies of WO_x-promoted Pt/ZrO₂ catalysts using in-situ DRIFTS [11] and elementary kinetic modelling [28, 29] indicated that the H₂-deNO_x reaction mainly occurs on the Pt sites. With this molecular mechanistic understanding [28, 29], four neighbouring active Pt sites (*) are required for the N₂ formation, including the adsorption of 2 NO molecules, the dissociation of the resulting NO adsorbates (NO*, Eq. (1)) and the recombination of the two N adsorbates (N*) (Eq. (2)). By contrast, the production of N₂O only requires 3 adjacent active sites, since N₂O originates from a NO and a N adsorbate (Eq. (3)). Consequently, the number of active sites is assumed to affect both the catalytic activity and the N₂ selectivity. The formation of the side-product NH₃ is not considered here in detail due to its minor yield when using the Mo-promoted Pt/ZrO₂. For the same reason, the possible formation of N₂O by the oxidation of NH₃ tends to be negligible, as substantiated in Sect. 3.3. Basically, NH₃ is formed by the step-by-step addition of a H atom to NH_x adsorbates [30].

Indeed, the Pt/xMo/ZrO₂ catalysts show a clear increase in the maximum NO_x conversion when the quantity of available Pt sites increases (Fig. 5, left), and the N₂ selectivity at X_max also tends to rise in line with the number of Pt sites (Fig. 5, right). The impact of the active sites on the activity as well selectivity is also reflected by the turnover frequency of NO (TOF), which expresses the NO conversion rate per available Pt site (TOF = ṅ(NO_x)_in ∙ X(NO_x)/n_a(Pt)). The TOF exemplarily calculated for 130 °C provides similar values for bare Pt/ZrO₂ and the Pt/xMo/ZrO₂ catalysts (Table S4) substantiating that the number of the active Pt centers drives the H₂-deNO_x activity. In contrast to that, the N₂ formation rate referred to the available Pt sites (ṅ(N₂)/n_a(Pt)) grows significantly from 5.9∙10³ s⁻¹ (bare ZrO₂) to 1.4∙10⁴ s⁻¹ and 1.5∙10⁴ s⁻¹ when the Mo load is increased to 3 and 4 wt%, but declines for the further increasing Mo contents (Table S4). This trend supports the importance of the active Pt centers in controlling the N₂ selectivity of the H₂-deNO_x reaction. From the above shown relations it is inferred that highly dispersed MoO_x species act as structural promoters, driving the H₂-deNO_x activity and N₂ selectivity by stabilizing the BET surface area and providing higher numbers of accessible Pt sites and smaller Pt particles, respectively. However, aggregated Mo oxide moieties existing at Mo loads of 4 wt% and above clearly attenuate this promoting effect.

It was shown in Sect. 3.1 that the MoO_x promoter strongly affects the H₂-deNO_x activity as well as selectivities towards N₂, N₂O and NH₃. One notable feature is the decreasing NH₃ formation in the presence of the promoter, with complete suppression at a Mo content of 4 wt% and beyond accompanied by increased N₂O selectivity (Fig. 1).

The decrease in the NH₃ production may be related to the enhanced NH₃ oxidation activity of the Mo-modified catalysts; a similar effect of the WO_x promoter was observed for the related Pt/W/ZrO₂ catalysts [31]. Indeed, the NH₃ oxidation studies show that the conversion of NH₃ is shifted to lower temperatures when bare Pt/ZrO₂ is compared with Pt/4Mo/ZrO₂ (Fig. S5). For instance, the temperature at which a NH₃ conversion of 50% is achieved (T₅₀) is shifted markedly from 240 °C for bare Pt/ZrO₂ to 185 °C for Pt/4Mo/ZrO₂. Moreover, in the additional presence of 2000 ppm H₂ the NH₃ oxidation activity of Pt/4Mo/ZrO₂ is further enhanced, as illustrated by the shift in T₅₀ to approx. 150 °C. Pt/ZrO₂ shows a similar effect (Fig. S5). This H₂-assisted NH₃ oxidation may be due to the reaction of H₂ with oxygen bound to Pt, resulting in a larger number of free Pt sites, which are available for the adsorption and subsequent conversion of NH₃. Additionally, the light-off temperatures of the NH₃ oxidation in the presence of H₂ are in line with the temperature range in which NH₃ is observed during the H₂-deNO_x reaction, i.e. no NH₃ appears in H₂-deNO_x above the NH₃ light-off temperature. Moreover, the conversion of NH₃ in the presence of H₂ is continuously shifted to lower temperatures when the Mo loading grows, including N₂O selectivities between 40 and 50% for all the Pt/xMo/ZrO₂ catalysts (Fig. S6).

Furthermore, the reduced NH₃ formation during H₂-deNO_x on the Mo-promoted catalysts may also be explained by an additional deNO_x mechanism, which involves NH_x species formed on the Pt sites reacting with NO_x [9, 32‐34]. Thus, NH₃ was taken as a reducing agent to evaluate the potential of the catalysts for a SCR-related pathway of this kind. Interestingly, Fig. 6 (left) indicates superior activity (without H₂) of the promoted Pt catalysts with Mo loads of 2, 4 and 7 wt% compared to Pt/ZrO₂, while the highest deNO_x is achieved for the sample with 4 wt% Mo. By contrast, the Pt-free 4Mo/ZrO₂ sample reveals no performance at all (Fig. 6, left), substantiating the importance of Pt for the SCR reaction [30]. As expected from the literature [30], the SCR reaction on Pt catalysts also leads to the strong evolution of N₂O, which clearly represents the major product for the samples with Mo loads of 2 and 4 wt%; for instance, at 170 °C, the N₂O selectivity amounts to 71% and 83%, respectively (Fig. 6, right). The addition of 2000 ppm H₂ to the SCR feed gas also significantly decreases the N₂O selectivity of the Pt/xMo/ZrO₂ catalysts (Fig. 6, right). As an example, for Pt/4Mo/ZrO₂ it declines from 83 to 31% at 170 °C (X(NO_x) = 90%). This performance is close to the H₂-deNO_x reaction on the same catalyst, showing a N₂O selectivity of 25% and a NO_x conversion of 80% at 150 °C.

These investigations clearly show that the Pt/xMo/ZrO₂ catalysts are substantially active in the SCR reaction, whereas the role of the Pt component obviously lies in the activation of NH₃, since the SCR reaction is totally inhibited in the absence of Pt. Therefore, it is deduced that the NH_x intermediates (x = 1 – 3) originating from the H₂-deNO_x reaction [30] also participate in the NO_x reduction. Additionally, it is possible that these NH_x species may spill over to the acid sites of the Mo/ZrO₂ support, so that the NO-NH_x reaction occurs at the interface between the Pt particles and the substrate [32, 33]. Mechanistic and kinetic studies showed that below 200 °C, NO is the most abundant adsorbate on the Pt sites of a related Pt/W/ZrO₂ catalyst under very similar reaction conditions [29]. Moreover, the basic NH_x species are believed to preferentially coordinate to the acid sites on the support. To gain a deeper understanding of how NH₃ interacts with Mo/ZrO₂ and NH_x may spill over from the Pt particles, DRIFTS studies were performed with NH₃ as the molecular probe, along with NH₃-TPD examinations.

DRIFT spectroscopy was conducted to check the NH_x adsorbates formed during the adsorption of NH₃. The Pt/xMo/ZrO₂ samples were exposed to 1000 ppm NH₃ at 50 °C followed by flushing and heating to 130 °C in a N₂ flow, then the spectra were recorded. Figure 7 demonstrates the spectral range of the NH_x deformation vibrations, revealing a weak DRIFTS band at approx. 1600 cm⁻¹ (σ_as) and an intense band at approx. 1185 cm⁻¹ (σ_s) for all the samples. Both bands are related to the NH₃ molecularly coordinated to Lewis acid sites, whereas some contribution could be provided from NH₃ adsorbed on Pt sites [35‐37]. Particularly, the intense σ_s(NH₃) band located between 1180 cm^-1 and 1270 cm⁻¹ is significantly increased for the samples containing Mo, the highest intensity being detected for the Mo load of 4 wt% (Fig. 7, inset). This indicates that the highly dispersed Mo oxide entities existing in the samples with the low Mo loads contribute additional Lewis acid sites to the support, probably in the form of Mo⁶⁺. However, the σ_s(NH₃) vibration markedly decreases above 4 wt% Mo, suggesting that the number of Lewis acid sites is lower at high Mo loads, likely associated with the stronger aggregation of the MoO_x entities. Note that the shift in the σ_s(NH₃) band from 1186 cm⁻¹ (Pt/ZrO₂) to 1235 cm⁻¹ (Pt/7Mo/ZrO₂) is in good accordance with the literature (ZrO₂: 1160 cm⁻¹, MoO₃: 1270 cm⁻¹), indicating a stronger Lewis acidity of the MoO_x moieties [35]. This interpretation is supported by the fact that the σ_s(NH₃) band of the Mo-promoted samples does not decline when the temperature is raised from 50 to 180 °C, while it is significantly reduced for bare Pt/ZrO₂ (Fig. S7).

In contrast to bare Pt/ZrO₂, the Mo-promoted Pt/ZrO₂ samples show a weak DRIFTS band at 1660 cm⁻¹ (σ_s(NH₄⁺) and a prominent signal at approx. 1445 cm⁻¹ (σ_as(NH₄⁺)). Both bands are attributed to NH₄⁺ species originating from the reaction of NH₃ with Brønsted acid sites [38]. Moreover, the band areas, particularly that of the 1445 cm⁻¹ feature, increase with the content of Mo (Fig. 7, inset), indicating a growing number of Brønsted acid sites. This may be primarily attributed to the increasing amount of aggregated MoO_x species detected by LRS (Fig. 3); these are known to form Mo-OH-Zr groups that can act as proton donors [39, 40]. However, the intensity of the σ_as(NH₄⁺) band substantially decreases when the temperature is raised from 50 to 180 °C, indicating that the thermal stability of the NH₄⁺ species is relatively low, in line with the literature (Fig. S7) [39, 41].

NH₃-TPD investigations were carried out to evaluate the NH₃ uptake capacity of the xMo/ZrO₂ supports (Fig. 8). The exposure to NH₃ took place at 130 °C, corresponding to substantial H₂-deNO_x activity in the case of Pt/3Mo/ZrO₂, which is the best catalyst of the samples tested. Figure 8 (inset) shows the highest NH₃ uptake for the samples with Mo contents of 2 and 3 wt% implying that there is a significant increase in the total amount of acid sites, whereas bare ZrO₂ and the samples with higher Mo fractions adsorb clearly smaller NH₃ quantities. For comparison, a NH₃-TPD study was also performed with the Pt/2Mo/ZrO₂ catalyst showing no significant effect of the Pt component on the NH₃-TPD profile (Fig. S8). Moreover, the samples with 2–4 wt% Mo also exhibit larger amounts of strongly bound NH₃, desorbing above 300 °C only. Based on the DRIFTS studies, this strongly adsorbed ammonia is primarily coordinated to Lewis acid sites, whereas the NH₃ desorption at lower temperatures (approx. 210 °C) is related to the decomposition of NH₄⁺ entities originating from the reaction of NH₃ with Brønsted acid sites [39, 41]. When the LRS (Fig. 3) and NH₃-TPD results are combined, it can be inferred that the highly dispersed MoO_x species that are particularly present in the samples with moderate Mo loads (2–4 wt% Mo) result in stronger Lewis acid sites, as indicated by the pronounced NH₃ desorption at 300 °C and above.

The results of the DRIFTS and NH₃-TPD studies are correlated with the activity of the Pt/xMo/ZrO₂ catalysts in H₂-deNO_x and SCR to assess how NH_x species contribute to an additional deNO_x pathway. The relation of the σ_s(NH₃) DRIFTS band area (Fig. 7) and the SCR activity of the Pt/xMo/ZrO₂ catalysts at 170 °C (Fig. 6) indicates that the deNO_x performance improves with a rise in the amount of NH₃ bound to Lewis acid sites (Fig. 9). From this relation it is deduced that the NH₃ species which are coordinated to strong Lewis acid sites, as primarily found on highly dispersed MoO_x species, participate in the SCR reaction. In contrast, the Brønsted acid sites and resulting NH₄⁺ species play a minor role: the sample containing 7 wt% Mo shows by far the largest σ_as(NH₄⁺) band, but significantly lower SCR activity compared to the sample with a Mo load of 4 wt%. Considering the lack of SCR activity in the Pt-free 4Mo/ZrO₂ sample, it may be further assumed that the reaction of NO adsorbed on Pt and NH_x bound to Lewis acid sites takes place at the Pt sites or the interface of Pt and the Mo/ZrO₂ substrate [30, 33].

Moreover, the area of the σ_s(NH₃) DRIFTS bands correlates with the maximum N₂ selectivity in the H₂-deNO_x investigations (Fig. 9). A similar correlation is obtained regarding the amount of ammonia desorbed from strong acid sites above 300 °C in the NH₃-TPD experiments (Fig. S9). These relations suggest that the introduction of moderate Mo loads (2–4 wt%) to the Pt/ZrO₂ catalyst creates strong Lewis acidity at highly dispersed MoO_x moieties, thus enabling an additional deNO_x pathway to form, providing high N₂ selectivity that simultaneously extends to the H₂-deNO_x mechanism on the Pt sites stated above [28, 29]. On this SCR-related pathway, NH_x or NH₃ species formed on the Pt sites are expected to spill over to strong Lewis acid sites in the case of highly dispersed MoO_x entities. The reaction between these activated NH₃ or NH_x species and the NO probably adsorbed on the Pt sites is assumed to take place at the interface between the Pt particles and the MoO_x species, forming N₂ and H₂O as the major products. Therefore, the outstanding activity and N₂ selectivity of the best Pt/3Mo/ZrO₂ catalyst is attributed not just to the high Pt dispersion, but also to the presence of highly dispersed MoO_x species, forming strong Lewis acid sites and thus enabling an SCR-type reaction of NO with NH₃, featuring high N₂ selectivity.