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

This article aims to improve the low-temperature H2-deNOx performance of the active Pt/ZrO2 catalyst using MoOx as a promoter. For this purpose, a systematic series of Pt/ZrO2 samples were prepared with a Pt content of 0.25 wt% and Mo loads from 0 to 10 wt%. The samples were physico-chemically characterized by means of powder X-ray diffraction, N2 physisorption, temperature-programmed desorption of CO and NH3, Raman spectroscopy and diffuse reflectance infrared spectroscopy using NH3 as probe molecule, while the H2-deNOx efficiency was investigated in a lean synthetic exhaust. The Pt/ZrO2 catalyst with a Mo load of 3 wt% showed the best performance, including H2-deNOx between 80 °C and 150 °C, a maximum NOx conversion of 90% and N2 selectivity up to 78%. Isolated MoOx species predominately present at Mo loads below 4 wt% were found to act as structural promoter by stabilizing the BET surface area, while also providing smaller Pt particles and more active Pt sites, respectively. By contrast, the aggregated Mo oxide moieties found at higher Mo loads exhibit a clearly weaker promotional effect. The structure–activity-selectivity correlations also suggest that the promoter additionally enables a SCR-related mechanistic pathway to be followed, including the spill-over of NHx species from the Pt sites to strong Lewis acid sites in the case of highly dispersed MoOx entities followed by reaction with NOx.


Introduction
For the reduction of CO 2 , which is the most significant greenhouse gas, combustion engines are progressively being replaced by fuel cell and battery electric vehicles. However, mobile and stationary combustion engines are still essential to maintain transportation and the energy supply in the long term. Thus, one special challenge lies in replacing fossil with sustainable fuels such as biofuels or "e-fuels" originating from power-to-X strategies. Moreover, H 2 is a CO 2 -neutral fuel, particularly when it is produced from H 2 O electrolysis powered by wind and solar energy. Combustion engines are most efficient under lean burn conditions, but the output of pollutants such as hydrocarbons, CO and NO x remains an issue [1]. Consequently, automotive pollution control techniques are required to meet international emission standards. To remove NO x from O 2 -rich exhaust gases, SCR (Selective Catalytic Reduction) and NO x storage catalyst (NSC) technologies are well established. These operate effectively between approx. 175 °C and 500 °C. However, both procedures perform only to a limited extent at the lower temperatures relevant for cold starts and low load engine operation such as city driving. NO x reduction with H 2 (H 2 -deNO x ) using Pt and Pd catalysts is the only available technique (2 NO + 2 H 2 → N 2 + 2 H 2 O) providing outstanding conversion in the lowtemperature regime of lean engine exhaust gases (< 175 °C). However, one current issue of H 2 -deNO x is the formation of by-products like N 2 O and NH 3 as exemplarily known for Pt/ Al 2 O 3 , Pt/H-ZSM-5 and Pt/SiO 2 catalysts [2][3][4]. Marked N 2 selectivity above 90% has been reported for Pt/MgO-CeO 2 [5,6], Pt/La 0.7 Sr 0.2 Ce 0.1 FeO 3 [7] and Pd/LaCoO 3 catalysts [8]. Additionally, WO x -promoted Pt catalysts supported by TiO 2 [9], HZSM-5 [10] and ZrO 2 [11] have been shown to be highly active, including the preferential formation of N 2 , while Mo-modified Pt/SiO 2 and Pt/Al 2 O 3 samples [12] also exhibited improved N 2 selectivity. Moreover, Pt catalysts modified with metal oxide promoters such as 1st and 2nd group elements (Na, K, Ba, etc.) [13,14] and transition metals (Co, Mn, W, Mo, etc.) [11,15,16] enhanced N 2 formation during H 2 -deNO x .
The present article aims to improve the low-temperature H 2 -deNO x performance under O 2 -rich conditions using Pt/ ZrO 2 , which was recently reported to be a highly active catalyst. For this purpose, MoO x was evaluated as a potential promoter [12], requiring a systematic series of Pt/ZrO 2 catalysts with different Mo loads. The prepared samples were physico-chemically characterized and studied in terms of their lean H 2 -deNO x activity and selectivity. From these investigations, structure-activity relations were derived to reveal the determining properties of the Mo promoter, which control the H 2 -deNO x activity leading to N 2 .

Catalyst Preparation
The samples were prepared by means of incipient wetness impregnation using monoclinic ZrO 2 as a support (Saint-Gobain NorPro). First, the Mo promoter was introduced by an aqueous solution of (NH 4 ) 6 Mo 7 O 24 · 4 H 2 O (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(NO 3 ) 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% H 2 and 90 vol% N 2 (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/ZrO 2 , where x refers to the loading of Mo.

Catalyst Characterization
The characterization of the catalysts was performed prior to the H 2 -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. N 2 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 N 2 adsorption isotherms were taken at − 196 °C. The BET surface area (S BET ) was determined from the adsorption data recorded at p/p 0 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% O 2 (He balance) at 450 °C for 30 min (400 ml/min, STP). Then, it was cooled to − 196 °C with liquid N 2 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 CO 2 were monitored using a non-dispersive infrared (NDIR) spectrometer (X-Stream Enhanced XEGK, Emerson). Blank CO-TPD studies of the bare Mo/ZrO 2 samples provided CO desorption profiles very similar to the Pt/xMo/ZrO 2 samples without significant amounts of CO 2 . The evolution of CO 2 upon TPD is ascribed to the oxidation of CO by the MoO x /ZrO 2 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 Å 3 ), a Pt represents the surface area of a Pt atom located on a polycrystalline surface (8.07 Å 2 ) [17].
The temperature-programmed desorption of NH 3 (NH 3 -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% O 2 in N 2 at 450 °C for 30 min (500 ml/min, STP). After cooling to 130 °C in a N 2 flow, the sample was saturated with 1000 ppm NH 3 in 1 3 N 2 (500 ml/min, STP). Weakly bound NH 3 was removed by flushing with N 2 until the outlet NH 3 fraction was below 2 ppm. For the final NH 3 -TPD, the sample was heated to 550 °C at a rate of 10 K/min in N 2 (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 NH 3 was monitored by means of NDIR spectroscopy (X-Stream, Emerson). Assuming a molar NH 3 to acid site stoichiometry of 1, the number of surface acid sites was estimated from the integrated NH 3 traces.
Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was conducted using NH 3 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 N 2 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% O 2 in N 2 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 N 2 flow, a background spectrum being collected at each temperature. Subsequently, the sample was exposed to 1000 ppm of NH 3 (N 2 balance) for 30 min at 50 °C. After flushing with N 2 for another 30 min at 50 °C to remove physisorbed NH 3 , an IR spectrum was recorded. The temperature was then increased to 90 °C and the sample was purged again with N 2 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 H 2 -deNO x Studies
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 N 2 flow to 280 °C for 20 min. Subsequently, the synthetic exhaust, composed of 200 ppm NO, 2000 ppm H 2 , 5 vol% O 2 , 10 vol% H 2 O and N 2 (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 H 2 -deNO x data are referred to the averaged temperature. To investigate the impact of CO on the H 2 -deNO x performance of Pt/xMo/ZrO 2 , 200 ppm CO was added to the lean model exhaust. For reference purposes, SCR experiments with and without 2000 ppm H 2 were also performed by adding 100 ppm NO and 100 ppm NH 3 to the model exhaust. After pre-treatment at 280 °C, each temperature was kept constant for 45 min to achieve steady-state conditions excluding NH 3 storage effects.
Gas-phase analysis was performed using a Multi-Gas 2030 FTIR spectrometer (MKS Instruments), detecting NO, NO 2 , N 2 O, NH 3 and H 2 O, whereas O 2 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 ) and N 2 (S(N 2 )) were calculated. Since N 2 is not detectable using FTIR, S(N 2 ) was derived from the mass balance of N including the measured species NO, NO 2 , NH 3 and N 2 O, i.e. S(N 2 ) = 1-(2·y(N 2 O) out + 2·y(NH 3 ) out ) /(y(NO x ) in -y(NO x ) out ).

H 2 -deNO x Performance of the Pt/Mo/ZrO 2 Catalysts
The most important H 2 -deNO x features of the Pt/xMo/ZrO 2 catalysts are demonstrated in Fig. 1, indicating that the maximum NO x conversion and N 2 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/ZrO 2 displays the highest H 2 -deNO x activity and N 2 selectivity, including The performance of Pt/3Mo/ZrO 2 is similar to that of the highly active WO x -modified Pt/ZrO 2 catalysts, which also provide NO x conversions up to 90% between 100 °C and 180 °C, but are characterized by higher N 2 selectivity (80-90%) [11,18]. A N 2 selectivity above 90% was reported for Pt/MgO-CeO, albeit using a markedly higher H 2 proportion of 1 vol% [6]. Note that for the WO x -promoted catalysts, a rise in S(N 2 ) with increasing H 2 concentration was also demonstrated [9,11,18], while at the same time the selectivity of H 2 to deNO x drastically decreases.
Additional H 2 -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/ZrO 2 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/ZrO 2 sample only achieves a X max of 50% appearing at 163 °C. CO and HC were reported as having a similar effect on the H 2 -deNO x reaction in the case of the Pt/W/ZrO 2 catalysts [1,18]. Thus, for engines operating with hydrocarbon-based fuels such as gasoline and diesel, an oxidation catalyst upstream of the H 2 -deNO x stage is mandatory to ensure that the H 2 -deNO x is efficient at low temperatures.

Structural Promotion of the Pt/Mo/ZrO 2 Catalysts
The LRS investigations of the Pt/xMo/ZrO 2 samples (Fig. 3) indicated that the ZrO 2 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][21][22]. This interpretation is substantiated by PXRD, which shows the reflexes of monoclinic ZrO 2 (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 MoO 3 (equivalent to 8 wt% Mo) [25], the formation of crystalline MoO 3 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-MoO 3 (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 m 2 /g) to 4 wt% (101 m 2 /g). This effect is ascribed to the inhibition of the sintering of ZrO 2 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 ZrO 2 [27].
CO-TPD studies show that the number of active Pt sites is significantly increased by promoting the bare Pt/ZrO 2 (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/ZrO 2 ), 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 ZrO 2 support Mo clusters form at the higher Mo loads. Then, electrostatic repulsion of the positively charged Mo cores of these clusters and Pt 2+ 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 ZrO 2 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/ ZrO 2 catalysts using in-situ DRIFTS [11] and elementary kinetic modelling [28,29] indicated that the H 2 -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 N 2 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 N 2 O only requires 3 adjacent active sites, since N 2 O 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 N 2 selectivity. The formation of the side-product NH 3 is not considered here in detail due to its minor yield when using the Mo-promoted Pt/ZrO 2 . For the same reason, the possible formation of N 2 O by the oxidation of NH 3 tends to be negligible, as substantiated in Sect. 3.3. Basically, NH 3 is formed by the step-by-step addition of a H atom to NH x adsorbates [30].
Indeed, the Pt/xMo/ZrO 2 catalysts show a clear increase in the maximum NO x conversion when the quantity of available Pt sites increases (Fig. 5, left), and the N 2 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/ZrO 2 and the Pt/xMo/ZrO 2 catalysts (Table S4) substantiating that the number of the active Pt centers drives the H 2 -deNO x activity. In contrast to that, the N 2 formation rate referred to the available Pt sites (ṅ(N 2 )/n a (Pt)) grows significantly from 5.9•10 3 s −1 (bare ZrO 2 ) to 1.4•10 4 s −1 and 1.5•10 4 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 Fig. 4 Relation of BET surface area (left) and specific number of available Pt sites (right) to the Mo content of the Pt/xMo/ZrO 2 samples; the curve fit is just a guide for the eyes Fig. 5 Relation of X max (left) and N 2 selectivity at X max (right) with the specific number of available Pt sites of the Pt/xMo/ZrO 2 catalysts; the line is just a guide for the eyes Pt centers in controlling the N 2 selectivity of the H 2 -deNO x reaction. From the above shown relations it is inferred that highly dispersed MoO x species act as structural promoters, driving the H 2 -deNO x activity and N 2 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.

Acid Promotion of the Pt/Mo/ZrO 2 Catalysts
It was shown in Sect. 3.1 that the MoO x promoter strongly affects the H 2 -deNO x activity as well as selectivities towards N 2 , N 2 O and NH 3 . One notable feature is the decreasing NH 3 formation in the presence of the promoter, with complete suppression at a Mo content of 4 wt% and beyond accompanied by increased N 2 O selectivity (Fig. 1).
The decrease in the NH 3 production may be related to the enhanced NH 3 oxidation activity of the Mo-modified catalysts; a similar effect of the WO x promoter was observed for the related Pt/W/ZrO 2 catalysts [31]. Indeed, the NH 3 oxidation studies show that the conversion of NH 3 is shifted to lower temperatures when bare Pt/ZrO 2 is compared with Pt/4Mo/ZrO 2 (Fig. S5). For instance, the temperature at which a NH 3 conversion of 50% is achieved (T 50 ) is shifted markedly from 240 °C for bare Pt/ZrO 2 to 185 °C for Pt/4Mo/ZrO 2 . Moreover, in the additional presence of 2000 ppm H 2 the NH 3 oxidation activity of Pt/4Mo/ZrO 2 is further enhanced, as illustrated by the shift in T 50 to approx. 150 °C. Pt/ZrO 2 shows a similar effect (Fig. S5). This H 2 -assisted NH 3 oxidation may be due to the reaction of H 2 with oxygen bound to Pt, resulting in a larger number of free Pt sites, which are available for the adsorption and subsequent conversion of NH 3 . Additionally, the lightoff temperatures of the NH 3 oxidation in the presence of H 2 are in line with the temperature range in which NH 3 is observed during the H 2 -deNO x reaction, i.e. no NH 3 appears in H 2 -deNO x above the NH 3 light-off temperature. Moreover, the conversion of NH 3 in the presence of H 2 is continuously shifted to lower temperatures when the Mo loading grows, including N 2 O selectivities between 40 and 50% for all the Pt/xMo/ZrO 2 catalysts (Fig. S6).
Furthermore, the reduced NH 3 formation during H 2 -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][33][34]. Thus, NH 3 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 H 2 ) of the promoted Pt catalysts with Mo loads of 2, 4 and 7 wt% compared to Pt/ZrO 2 , while the highest deNO x is achieved for the sample with 4 wt% Mo.
By contrast, the Pt-free 4Mo/ZrO 2 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 N 2 O, which clearly represents the major product for the samples with Mo loads of 2 and 4 wt%; for instance, at 170 °C, the N 2 O selectivity amounts to 71% and 83%, respectively (Fig. 6, right). The addition of 2000 ppm H 2 to the SCR feed gas also significantly decreases the N 2 O selectivity of the Pt/xMo/ZrO 2 catalysts (Fig. 6, right). As an example, for Pt/4Mo/ZrO 2 it declines from 83 to 31% at 170 °C (X(NO x ) = 90%). This performance is close to the H 2 -deNO x reaction on the same catalyst, showing a N 2 O selectivity of 25% and a NO x conversion of 80% at 150 °C.
These investigations clearly show that the Pt/xMo/ ZrO 2 catalysts are substantially active in the SCR reaction, whereas the role of the Pt component obviously lies in the activation of NH 3 , 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 H 2 -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/ZrO 2 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/ZrO 2 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 NH 3 interacts with Mo/ZrO 2 and NH x may spill over from the Pt particles, DRIFTS studies were performed with NH 3 as the molecular probe, along with NH 3 -TPD examinations.
DRIFT spectroscopy was conducted to check the NH x adsorbates formed during the adsorption of NH 3 . The Pt/ xMo/ZrO 2 samples were exposed to 1000 ppm NH 3 at 50 °C followed by flushing and heating to 130 °C in a N 2 flow,  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 NH 3 molecularly coordinated to Lewis acid sites, whereas some contribution could be provided from NH 3 adsorbed on Pt sites [35][36][37]. Particularly, the intense σ s (NH 3 ) 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 Mo 6+ . However, the σ s (NH 3 ) 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 (NH 3 ) band from 1186 cm −1 (Pt/ZrO 2 ) to 1235 cm −1 (Pt/7Mo/ZrO 2 ) is in good accordance with the literature (ZrO 2 : 1160 cm −1 , MoO 3 : 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 (NH 3 ) 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/ZrO 2 (Fig. S7).
In contrast to bare Pt/ZrO 2 , the Mo-promoted Pt/ZrO 2 samples show a weak DRIFTS band at 1660 cm −1 (σ s (NH 4 + ) and a prominent signal at approx. 1445 cm −1 (σ as (NH 4 + )). Both bands are attributed to NH 4 + species originating from the reaction of NH 3 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 (NH 4 + ) band substantially decreases when the temperature is raised from 50 to 180 °C, indicating that the thermal stability of the NH 4 + species is relatively low, in line with the literature (Fig. S7) [39,41]. NH 3 -TPD investigations were carried out to evaluate the NH 3 uptake capacity of the xMo/ZrO 2 supports (Fig. 8). The exposure to NH 3 took place at 130 °C, corresponding to substantial H 2 -deNO x activity in the case of Pt/3Mo/ZrO 2 , which is the best catalyst of the samples tested. Figure 8 (inset) shows the highest NH 3 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 ZrO 2 and the samples with higher Mo fractions adsorb clearly smaller NH 3 quantities. For comparison, a NH 3 -TPD study was also performed with the Pt/2Mo/ZrO 2 catalyst showing no significant effect of the Pt component on the NH 3 -TPD profile (Fig. S8). Moreover, the samples with 2-4 wt% Mo also exhibit larger amounts of strongly bound NH 3 , desorbing above 300 °C only. Based on the DRIFTS studies, this strongly adsorbed ammonia is primarily coordinated to Lewis acid sites, whereas the NH 3 desorption at lower temperatures (approx. 210 °C) is related to the decomposition of NH 4 + entities originating from the reaction of NH 3 with Brønsted acid sites [39,41]. When the LRS (Fig. 3) and NH 3 -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 NH 3 desorption at 300 °C and above. The results of the DRIFTS and NH 3 -TPD studies are correlated with the activity of the Pt/xMo/ZrO 2 catalysts in H 2 -deNO x and SCR to assess how NH x species contribute to an additional deNO x pathway. The relation of the σ s (NH 3 ) DRIFTS band area (Fig. 7) and the SCR activity of the Pt/ xMo/ZrO 2 catalysts at 170 °C (Fig. 6) indicates that the deNO x performance improves with a rise in the amount of NH 3 bound to Lewis acid sites (Fig. 9). From this relation it is deduced that the NH 3 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 NH 4 + species play a minor role: the sample containing 7 wt% Mo shows by far the largest σ as (NH 4 + ) 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/ ZrO 2 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/ZrO 2 substrate [30,33].
Moreover, the area of the σ s (NH 3 ) DRIFTS bands correlates with the maximum N 2 selectivity in the H 2 -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 NH 3 -TPD experiments (Fig. S9). These relations suggest that the introduction of moderate Mo loads (2-4 wt%) to the Pt/ZrO 2 catalyst creates strong Lewis acidity at highly dispersed MoO x moieties, thus enabling an additional deNO x pathway to form, providing high N 2 selectivity that simultaneously extends to the H 2 -deNO x mechanism on the Pt sites stated above [28,29]. On this SCR-related pathway, NH x or NH 3 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 NH 3 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 N 2 and H 2 O as the major products. Therefore, the outstanding activity and N 2 selectivity of the best Pt/3Mo/ ZrO 2 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 SCRtype reaction of NO with NH 3 , featuring high N 2 selectivity.

Conclusions
The present study shows that Mo is an effective promoter for Pt/ZrO 2 , enhancing the lean H 2 -deNO x activity and N 2 selectivity at low temperatures. The best catalyst exhibits a Mo load of 3 wt% revealing high H 2 -deNO x efficiency between 80 °C and 150 °C with a maximum NO x conversion of 90% and a N 2 selectivity of up to 78%. Thus, this Mopromoted Pt/ZrO 2 sample shows a performance similar to the best reported H 2 -deNO x catalysts, which are Pt/W/ZrO 2 and Pt/MgO-CeO 2 [6,11].
Our studies have shown that highly dispersed Mo species have a particularly beneficial promoting effect on Pt/ZrO 2 catalysts, which can be attributed to the enhancement of Pt dispersion, affecting both NO x conversion and selectivity towards N 2 . However, at Mo loadings above 4 wt%, the dispersed Mo species aggregate and there is a decrease in the number of Pt sites. As a result, there is a drastic decrease in the H 2 -deNO x activity and N 2 selectivity. Moreover, our experiments provide strong evidence that the Mo promotion of Pt/ZrO 2 enables an additional deNO x mechanism involving strong Lewis acid sites at the MoO x entities. It is assumed that NH 3 or NH x species formed on the Pt sites during H 2 -deNO x spill over to the strong Lewis acid sites at highly dispersed MoO x moieties, followed by a reaction with NO at the interface between the Pt sites and MoO x . Thus, our study highlights the particular importance of acidity in enhancing the activity and selectivity of lean H 2 -deNO x catalysts. In particular, strong Lewis acid sites can trap active NH x species, creating an additional deNO x pathway at the interface of Pt and the promoter/support, whereas Brønsted acid sites are likely of minor importance for NH x activation. Thus, in our future research we aim to examine whether the acid promotion of Pt/ZrO 2 is a universal means of enhancing the H 2 -deNO x reaction, particularly to produce selective N 2 formation. For this purpose, other acid promoters such as tungsten, sulphur and phosphorus will be systematically investigated, in which context the adjustment of the respective promoter structures is the key to achieving high Lewis acidity.