Low-Temperature H2-deNOx in Diesel Exhaust

For NOx removal from diesel exhaust, the selective catalytic reduction (SCR) is the most common abatement technology. However, low engine load scenarios such as city driving and cold start phases demand efficient NOx reduction clearly below 200 °C, which is difficult to achieve with SCR. Hence, this work investigates the potential of the low-temperature NOx reduction with H2 in diesel exhaust. A monolithic Pt/WOx/ZrO2 catalyst, recently reported as highly active, was evaluated in synthetic and real diesel exhaust. The monolith demonstrated high deNOx activity between 130 and 215 °C in the synthetic exhaust including peak conversion of 90% with N2 selectivity up to 85%. CO/HC components were shown to inhibit the H2-deNOx conversion thus requiring a pre-oxidation catalyst in practice. Furthermore, studies performed in an optical reactor indicated strong heat evolution along the monolith. As a result, the reaction kinetics was accelerated with an enhanced consumption of H2 limiting the H2-deNOx efficiency above 200 °C. Stationary tests in diesel engine exhaust substantiated the low-temperature H2-deNOx performance of the monolith including NOx conversions up to 80% at temperatures as low as 135 °C.


Introduction
Combustion engines are the most important propulsion systems for transportation but come increasingly under pressure due to the emission of CO 2 and pollutants. Thus, battery and fuel cell electric vehicles are intensely developed, whereas combustion engines will still be needed in the next decades to guarantee energy supply. Additionally, sustainable fuels based on biogenic and synthetic components are more and more advanced towards the substitution of fossil resources. Furthermore, exhaust gas after-treatment techniques were worldwide introduced to abate pollutants from combustion engines. For stoichiometric gasoline vehicles, three-way catalysts are employed, which simultaneously convert hydrocarbons (HC), carbon monoxide (CO) and nitrogen oxides (NO x ). For diesel exhaust, diesel oxidation catalysts (DOC) are taken to remove HC and CO, while particulate matter is separated by diesel particulate filters (DPF). Additionally, NO x [1] is widely eliminated by the selective catalytic reduction (SCR) employing V 2 O 5 /WO 3 /TiO 2 [2] or Fe-/Cu-zeolite catalysts [3]. SCR uses NH 3 as reducing agent (4 NO + 4 NH 3 + O 2 → 4 N 2 + 6 H 2 O), which is produced on-board from urea, and operates between approx. 220 and 550 °C. However, its limited performance below 200 °C is increasingly an issue closely associated with the engine efficiency, which has been continuously enhanced in the recent years. In particular, low exhaust temperatures reflect engine cold start, city driving, stop-and-go traffic and intermittent operation of the combustion engine in hybrid vehicles. These operation ranges of diesel engines are challenging to meet latest emission limits demanding real driving scenarios. In this context, the reduction of NO x by hydrogen (H 2 -deNO x , Eq. 1) is known as a promising approach to remove nitrogen oxides at low temperatures, whereupon precious metal catalysts are known to be most active [4]. High H 2 -deNO x efficiency below 200 °C was recently demonstrated for the first time in real diesel [5] and lean H 2 combustion engine exhaust using Pt catalysts [6]. Despite this progress towards practical application, crucial constraints remain including the side-production of N 2 O (Eq. 2) [7] and competing consumption of H 2 by O 2 present in excess (Eq. 3). Nevertheless, the H 2 -deNO x reaction can be clearly enhanced by increasing the molar H 2 /NO x ratio. Additionally, the increased amount of H 2 heats up the catalyst due to the exothermic conversions (Eqs. 1-3) thus shifting the range of substantial H 2 -deNO x activity to lower catalyst inlet temperatures [5].
A comprehensive overview on effective H 2 -deNO x catalysts under O 2 -rich conditions was demonstrated in recent articles and reviews [8][9][10][11][12]. Promising catalysts investigated on the powder level are Pt/MgO-CeO 2 [13], Pd/LaCoO 3 [14] and Pt/WO 3 /TiO 2 [15] revealing high activity below 200 °C with maximal NO x conversions above 80% and corresponding N 2 selectivities between 80 and 90%. Additionally, WO x -promoted Pt/ZrO 2 showed NO x conversion of 98% at temperatures as low as 80 °C with N 2 selectivities up to 90% [16]. Different pathways of the H 2 -deNO x reaction were reported in the literature. For Pt/WO x /ZrO 2 [17] and Pt/SiO 2 /Al 2 O 3 [12], the platinum atoms are considered to act as the sole active site, on which NO dissociatively adsorbs followed by recombination of two N atoms to yield N 2 , whereas N 2 O originates from combination of N and NO adsorbates [18]. Moreover, for Pt/WO x /SiO 2 /Al 2 O 3 [12] and Pt/W/HZSM-5 [19], NH x intermediates formed at the interface of Pt and the respective substrate were suggested to participate in a SCR-related reaction. Also, dissociation of H 2 on the Pt sites and migration of the formed H atoms to NO x surface species located on the substrate was proposed for Pt/MgO-CeO 2 [13].
The present article addresses the low-temperature H 2 -deNO x conversion with special focus on the effect of CO and HC using a diesel engine test bench and a monolithic Pt/ WO x /ZrO 2 catalyst. For comprehensive understanding of the results gained in the diesel exhaust line, tests on a laboratory rig were performed including a model exhaust with defined CO and C 3 H 6 proportions. Additional studies were made with an optical reactor to identify reaction zones inside the catalytic monolith upon H 2 -deNO x .

Catalyst Preparation
The catalyst used in this study was a recently reported WO x -promoted Pt/ZrO 2 catalyst developed for H 2 -deNO x (1) in lean H 2 combustion engines [5,6]. For clarity, the preparation is briefly described here. The powder catalyst was coated onto two monolithic honeycomb substrates (cordierite, 600 cpsi, 5.66″ × 3″) resulting in a catalyst load of 80 g/l each. The coating was processed by Umicore. For the H 2 -deNO x tests in synthetic diesel exhaust, 1″ × 2″ cores were extracted from a honeycomb, while for the investigations in real diesel exhaust, the full-size monoliths were employed and fitted to a homemade metal housing (5.66″ × 6″). For the studies with the optical reactor, a recently reported monolithic Pt/WO x /ZrO 2 model catalyst, also coated by Umicore, was used implying a 400 cpsi cordierite honeycomb and a loading of 100 g/l [16]. Cores with a 1.4 × 1.4 cm square base and lengths between 1.2 and 6 cm were taken from the full size brick.

Catalyst Characterization
N 2 physisorption was performed on a TriStar II (Micrometrics). Prior to the analyses, respective sample was outgassed in vacuum (10 −1 mbar, 16 h) at 350 °C to remove adsorbed species such as H 2 O. The BET surface area (S BET ) was obtained from adsorption data at -196 °C and p/p 0 ratios between 0.05 and 0.20.
Temperature-programmed desorption of CO (CO-TPD) was used to quantify the available Pt sites of the catalyst powder. A sample mass of 200 mg was charged into a U-shaped glass reactor [20] and pretreated for 15 min at 500 °C in 5 vol.% O 2 (He balance) to remove adsorbed species. At the same temperature, the sample was flushed with He and then exposed to a mixture of 3000 ppm CO and He (balance) for 15 min to reduce the Pt surface. After this, it was cooled to -196 °C in flowing He and the sample was saturated with CO by dosing a blend of 750 ppm CO and He (balance). The low adsorption temperature was taken to avoid spill-over of CO from the Pt sites to the support as previously reported for similar catalysts [21,22]. Subsequently, it was flushed with He until no more CO was detected in the gas phase. Note that for all these steps, the total gas flow was kept at 400 ml/min (STP). Finally, the TPD was started by increasing the temperature at a rate of 20 K/min while supplying a He flow of 150 ml/min (STP). The temperature was recorded by a K-type thermocouple inside the sample in middle of the sample chamber through a notch in the reactor. Gaseous CO and CO 2 were continuously monitored by non-dispersive infrared spectroscopy (NDIR, X-Stream Enhanced XEGK, Emerson). It should be mentioned that in TPD CO, which desorbs below 100 °C, is originated from the support as shown by blank experiments made with the bare support. Additionally, above approx. 100 °C CO desorbing from the Pt sites is totally converted into CO 2 by subsequent reaction with the support. This supposition is in line with the literature [23,24] and was confirmed by the blank experiments indicating only minor desorption of CO 2 from the bare support. Thus, for the determination of the available Pt sites, the CO 2 signal detected in TPD was integrated and the contribution of the minor CO 2 signal of the blank TPD was subtracted. With the resulting molar amount of CO 2 and assuming a CO/Pt stoichiometry of 1 [25] the total number of available Pt sites (n a (Pt)) was derived. The dispersion of Pt (D Pt ) was calculated from n a (Pt) and the total number of Pt atoms (n Pt ) known from the incipient wetness impregnation (D Pt = n a (Pt)/n Pt ). The mean Pt particle size (d Pt ) was estimated by supposing perfect spheres, i.e. d Pt = 6V Pt /(a Pt ·D Pt ), whereas V Pt is the volume of a Pt atom in the bulk (15.10 Å 3 ) and a Pt represents the surface area of a Pt atom located on a polycrystalline surface (8.07 Å 2 ) [25]. All samples were characterized in powder form prior to deposition on the monolith.

Experiments with the Laboratory Test Bench and the Optical Reactor
The catalytic H 2 -deNO x investigations in the diesel model exhaust gas were performed on a laboratory test bench using the 1″ × 2″ cores. Respective monolith was placed in the tubular quartz glass reactor slightly larger than the honeycomb catalysts, heated by an external oven, and was wrapped by quartz wool to avoid bypassing. Temperatures were recorded by K-type thermocouples located directly in front of and behind the honeycomb.
were used. The latter is exclusively presented for NO x conversions above 20% to minimize error propagation. As N 2 was not directly measured, its selectivity was calculated from that of N 2 O [S(N 2 ) = 1 -S(N 2 O)]. Its formation in H 2 -deNO x has been shown in previous experiments by GC/TCD analysis using Ar as balance [16].
The laboratory facility with the optical reactor was used to investigate reaction zones inside the monolith by measuring the local heat temperature. The reactor consists of CaF 2 , which is permeable for UV/VIS and IR radiation, and reveals a square base of 1.4 × 1.4 cm with a height of 9 cm. The temperature was adjusted by an oven placed in front of the CaF 2 reactor, whereas reference temperatures were taken by K-type thermocouples located directly in front of and behind the monolith, which was introduced into the optical reactor. The thermal imaging camera PYROVIEW 380 M (DIAS Infrared) was positioned close to the optical reactor such that the whole length of the monolith was scanned. The camera collected a spectral range from 3 to 5 μm corresponding to temperatures between 100 and 300 °C, whereas the thermal images were recorded and processed using the software PYROSOFT Compact. Supply and analysis of the gases were conducted as in the laboratory tests described above. The total flow was 3 l/min (STP) and constant inlet temperatures between 105 and 165 °C were adjusted. After reaching the desired temperature in a gas flow consisting of 5 vol.% O 2 in N 2 , the reactive gases (y(NO) = 2000 ppm, y(H 2 ) = 6000 ppm) were added and thermal images were recorded after reaching a stationary temperature profile. In the H 2 -deNO x investigations, the length of the monolith was varied between 1.2 and 6.0 cm corresponding to space velocities from 18,000 to 90,000 h −1 . While the NO x conversion was determined as described above the H 2 conversion was calculated from the H 2 O proportion monitored by the FTIR analyzer [X(H 2 ) = 1 − y(H 2 O) out /y(H 2 ) in ].

Experiments on the Diesel Engine Test Bench
The H 2 -deNO x investigations in diesel exhaust were performed by employing an emergency power generator (YTO YD385D, Yangdong) with an engine displacement of 1.5 l and a maximal power output of 11 kW providing an exhaust stream of approx. 880 l/min (STP). The exhaust line consisted of a commercial cordierite-based diesel oxidation catalyst (DOC, 5.66″ × 3″, 400 cpsi, precious metal loading: 30 g/ft³, Pt/Pd mass ratio: 6:1), a commercial DPF (SiC, HJS Technologies), an exhaust cooler and the monolithic H 2 -deNO x catalyst. The DOC was positioned close to the engine to achieve CO and HC light-off (approx. 200 °C) removing all carbon containing pollutants. The pressure loss of the particulate filter was continuously monitored (IDPS 200, Schneider Messtechnik), whereas filter regeneration was performed before each H 2 -deNO x experiment by using an external oven (550 °C, 4 h, static air). The exhaust cooling was made with a gas recirculation cooler (Scania) designed for truck engines enabling stationary inlet temperatures at the H 2 -deNO x catalyst between 100 and 180 °C. The required H 2 was dosed by a mass flow controller (Bronkhorst) and was injected centrally counter-streamwise behind the particulate filter. A butterfly mixer [26] was used to homogenize the H 2 distribution in the gas stream. Prior to the H 2 -deNO x experiments, the exhaust compositions were checked by the FTIR analyzer and lambda probe described above. During the tests, the NO x proportions were recorded by NO x probes (UniNO x , Continental) placed in front of and behind the H 2 -deNO x catalyst. The H 2 fractions present in the exhaust gas after the H 2 -deNO x catalyst were checked by online gas chromatography coupled with thermal conductivity detection (GC/TCD INFICON 3000 Micro GC, Inficon). The exhaust temperature was continuously monitored in front of and behind each aftertreatment device by K-type thermocouples.

Catalyst Properties
The Pt/WO x /ZrO 2 sample prepared to washcoat the honeycomb substrates was the same as reported earlier and displays a very similar H 2 -deNO x performance and catalyst properties [5]. The most important physico-chemical properties of the catalyst powder are briefly summarized in Table 1.
In accordance with a recent study on related Pt/WO x /SiO 2 / Al 2 O 3 catalysts, a key feature driving the H 2 -deNO x activity is the high Pt dispersion [12], which was indeed achieved for Pt/WO x /ZrO 2 . As a consequence, the Pt loading of the coated monoliths amounted to approx. 6 g/ft³ only, which is a very low precious metal content compared to state-of-theart diesel oxidation catalysts and three-way catalysts.

Effect of CO and HC on H 2 -deNO x in Synthetic
Diesel Exhaust  (Fig. 1) showed NO x conversions of more than 90% in the low-temperature regime including a N 2 selectivity up to 85%. The temperature window for significant NO x reduction lay between 130 and 215 °C. However, outside of this optimal operation range, the N 2 O selectivity increased significantly reaching up to 50%. On the contrary, NH 3 formation was not observed at all. For the scenario without active pre-DOC, proportions of CO (150 ppm) and HC were additionally adjusted using  Fig. 3, the addition of 150 ppm CO (without C 3 H 6 ) to the synthetic model exhaust had a strong effect on the H 2 -deNO x reaction leading to complete loss of activity below 125 °C but having no effect on the NO x reduction above 140 °C decreasing the maximal N 2 selectivity only slightly from 85 to 75%. Additionally, the light-off temperature of the H 2 -deNO x reaction (125-140 °C) correlates with the light-off temperature of the CO oxidation. Therefore, the loss of low-temperature activity in the presence of CO and C 3 H 6 is attributed to the presence of CO in the exhaust gas but is not involved in the reduced H 2 -deNO x activity above the CO light-off temperature. This finding is in line with literature attributing the loss of low-temperature efficiency to the blockage of active Pt sites by CO, while at temperatures above the light-off temperature of CO sufficient Pt sites are available for simultaneous NO x reduction and CO adsorption and oxidation [16,27].
The sole presence of 50 ppm C 3 H 6 (without CO) in the exhaust gas led to a significantly decreased H 2 -deNO x activity in the whole temperature range (Fig. 4). The maximal NO x conversion of approx. 60% was observed at 125 °C with a significant N 2 O selectivity of more than 30%. In contrast to the experiments with only CO, no clear correlation between C 3 H 6 light-off temperature and H 2 -deNO x activity becomes apparent. Incomplete C 3 H 6 oxidation to CO was not detected. Similar to the experiments with CO supply, the reduced H 2 -deNO x activity is attributed to the blockage of active Pt sites by C 3 H 6 or its intermediate species [28] occurring during conversion to CO 2 and H 2 O.
From the above tests investigating the effect of CO and HC it is inferred that an active DOC positioned in front of the H 2 -deNO x stage is mandatory for practice, since CO and HC strongly inhibit the NO x reduction, particularly at low temperatures. Thus, the H 2 -deNO x studies made in real

Evaluation of Reaction Zones Inside the Catalyst
Since the H 2 -NO x and H 2 -O 2 conversions (Eqs. 1-3) are strongly exothermic, hot areas inside the catalyst can be formed potentially causing local auto-acceleration of the reactions. Thus, temperature profiles inside the catalyst were evaluated to identify such reaction zones. For this purpose, the optical reactor was used allowing the measurement of surface temperatures along the outer catalyst channels. These fundamental studies were performed with the monolithic Pt/WO x /ZrO 2 model catalyst, as for this sample an elementary kinetic model is available [17,29]. To investigate the effect of NO on the heat evolution and formation of reaction zones a monolith with a length of 6 cm (GHSV: 18,000 h −1 ) was taken, while dosing a gas mixture of 0 or 2000 ppm NO, 6000 ppm H 2 , 5 vol.% O 2 and N 2 as balance. The thermal images were taken under stationary conditions adjusting gas inlet temperatures between 105 and 165 °C (Fig. 5). In the absence of NO, the H 2 oxidation started at an inlet temperature of 125 °C as indicated by a temperature increase by 30 °C at the central upstream part of the monolith. With inclining inlet temperature, the reaction zone increasingly ranges over the entire width of the monolith, while the largest axial extension was 2.0 cm for the inlet temperature of for 165 °C resulting in a rise of 40 °C up to 205 °C (Fig. 5.) Note that for all inlet temperatures the strongest heat evolution, corresponding to the highest rate of H 2 oxidation, occurred at the central front side of the monolith. Contrary, the presence of NO shifted the H 2 lightoff to higher temperature (approx. 145 °C) with a temperature rise of 25 °C while also shortening the axial extension of the reaction zone, e.g. from 1.8 cm to 1.4 cm at the catalyst inlet temperature of 155 °C. Also, the maximal temperature detected at a catalyst inlet temperature of 165 °C was only 195 °C indicating an inhibiting effect of NO on the H 2 oxidation. This is in line with the elementary kinetics of the lean H 2 -deNO x reaction showing predominant coverage of the Pt sites by NO below 200 °C. As a consequence, the adsorption and reaction of H 2 on the catalyst is limited thus shifting the H 2 oxidation to higher temperatures [17,29].
Furthermore, the influence of the length of the monolith (1.2 cm, 3 cm, 6 cm) on the temperature profile upon H 2 -deNO x reaction was investigated retaining the total flow as well as gas composition. Note that the different lengths correspond to space velocities ranging from 18,000 h -1 to 90,000 h -1 . NO x and H 2 conversions were again determined using the NO x and H 2 O proportions at the reactor outlet taken by the FTIR spectrometer. As a result, the thermal images (Fig. 6) show higher temperatures the shorter the monoliths are. For instance, at the inlet temperature of approx. 145 °C using a monolith with 6 cm a maximal temperature of ca. 170 °C appeared, while for the samples with 3 and 1.2 cm length temperature rose to 180 and 190 °C, respectively. The differences are attributed to heat transport by thermal conduction and convection as well as the heat capacity, which increases with catalyst mass and length, respectively. Heat conduction predominates in the laminar flow regime of the honeycomb (Re ≈ 150), whereas the heat transport from the catalyst to the environment directly correlates with the contact area [30]. As with reducing the length from 6 to 1.2 cm the surface area of the cuboid honeycomb is lowered from 321 to 64 cm², it is clear that the heat transition from the monolith to the surroundings is drastically lowered. It should be mentioned that the heat conductivity of the solid fraction of the catalyst is higher compared to the gas flow as indicated by the thermal conductivity coefficients (k), which amount to 1.3-2.5 W/mK for cordierite, 1.5-2 W/mK for ZrO 2 as the main washcoat component, but only 0.02 W/ mK for N 2 as the main gas component [31]. Additionally, the heat capacity of the honeycomb is substantially decreased when using the shorter monoliths, which consequently take up less heat. For instance, the heat capacity of the 1.2 cm honeycomb is only 20% of that with 6 cm length due to its lower mass. As a result of the reduced heat transfer and heat capacity of the shorter catalysts, the temperature is increased inside the catalyst accelerating the H 2 oxidation and shifting the light-off temperature to lower inlet temperatures as shown in Fig. 7.  Fig. 6 Moreover, Figs. 5 and 6 demonstrate a parabolic radial distribution of the temperature inside the monoliths. This effect is obviously related to the heat transfer from the centre of the honeycomb to the environment of the optical reactor. This cooling effect is very likely supported by the gas, which flows in the outer catalyst channels along the edges of the reactor. Since no catalytic reaction takes place on the walls and no backmixing occurs inside the gas stream due to the laminar flow regime, the gas streaming at the edges acts as a cooling film and so it enhances the radial temperature distribution.
The influence of the different temperature profiles on the H 2 -deNO x reaction at different space velocities becomes apparent when comparing the corresponding NO x and H 2 conversions (Fig. 7). The shift of NO x conversions to lower temperatures when shortening the catalyst from 6 to 3 cm, hence increasing the GHSV from 18,000 to 36,000 h −1 , corresponds to the higher temperatures of the monolith accelerating the H 2 -deNO x reaction. Additionally, the H 2 -O 2 conversion, which is more temperature sensitive than the H 2 -deNO x reaction due to its faster reaction kinetics [17], increases more strongly than the NO x conversion; for instance at the inlet temperature of 125 °C the H 2 conversion inclines from 30% (18,000 h −1 ) to 50% (36,000 h −1 ), while the NO x conversion only grows from approx. 40 to 50%. When increasing the space velocity even higher from 36,000 to 90,000 h −1 by shorting the catalyst from 3 to 1.2 cm the NO x conversion as well as H 2 conversion clearly decrease for all inlet temperatures. As the reaction zone for the H 2 -deNO x reaction is within the first 2 cm of the monolith according to the thermal images (Fig. 6), this decrease in catalytic activity is attributed to the lower residence time (τ) inside the monolith with a length of only 1.2 cm (τ(l = 1.2 cm) = 47 ms, τ(l = 6 cm) = 235 ms). The fraction of H 2 taking part in the H 2 -deNO x reactions (Eqs. 1 and 2) reaches up to 50% at temperatures of highest NO x conversion, i.e. between 125 and 135 °C. However, above 135 °C the proportion of H 2 reacting with O 2 (Eq. 3) is strongly increased and at catalyst inlet temperatures above 160 °C this fraction exceeds 80% similar to reports on other H 2 -deNO x catalysts [8,10,32].

H 2 -deNO x Efficiency in Real Diesel Exhaust
The potential of the low-temperature H 2 -deNO x efficiency in real diesel exhaust was evaluated with the two consecutive 5.66″ × 3″ monoliths by systematically varying the H 2 supply. The H 2 -deNO x catalyst was positioned downstream to the DOC, the particulate trap and the exhaust cooler. The engine was operated stationarily at constant engine speed resulting in a space velocity of 22,000 h -1 mimicking city driving and stop-and-go traffic. The air-fuel ratio was set to λ = 2.4 (y(H 2 O) = 5.7 vol.%, y(CO 2 ) = 6 vol.%, y(O 2 ) = 12.0 vol.%) and λ = 1.9 (y(H 2 O) = 7.1 vol.%, y(CO 2 ) = 7.5 vol.%, y(O 2 ) = 10.0 vol.%) by adjusting the load to approx. 60% and 85% respectively. The temperature at the H 2 -deNO x catalyst inlet was adjusted by the exhaust cooler to below 200 °C typical for cold start conditions.
In the tests using an air-fuel ratio of λ = 2.4 the maximal NO x conversion clearly increased from approx. 5 to 60% when raising the H 2 fraction from 1000 ppm to 1 vol.% (Fig. 8). This finding was expected from previous studies [5,17] showing enhanced H 2 -deNO x by increasing availability of hydrogen on the active Pt sites of the catalyst. Simultaneously, the temperature of the maximal NO x conversion decreased from 165 to 145 °C. This is associated with the heating of the monoliths due to the exothermic H 2 -O 2 reaction accelerating with growing H 2 fraction. Such heat evolution inside the monoliths has been shown by the investigations with the optical reactor (Figs. 5 and 6) and is substantiated by the strongly increasing outlet temperatures, referred to the inlet, exemplarily rising by approx. 80 °C for 1 vol.% H 2 (Fig. 9b). As for the temperature of the maximal NO x conversion, the heat development in the catalyst also caused a shift of the light-off temperature of the H 2 oxidation to lower inlet temperatures (Fig. 9a). For instance, in the experiment with 5000 ppm H 2 , the GC-TCD analysis indicated almost total conversion of H 2 at 155 °C, while with 1 vol.% full consumption already occurred at 145 °C.
However, the H 2 -deNO x investigations showed continuous decrease in the NO x conversion when the inlet temperature exceeds 150 °C regardless of the proportion of H 2 (Fig. 8). This effect is related to the consumption of H 2 by the competing reaction with O 2 , which is accelerated faster with increasing temperature than the NO reduction in agreement with the elementary kinetics of lean H 2 -deNO x on Pt/  [17]. Consequently, the H 2 availability for the reduction of NO x is progressively diminished with temperature thus limiting the operation window of the H 2 -deNO x conversion.
In the tests using an air-fuel ratio of λ = 1.9 (Fig. 10) significantly higher NO x conversions between 35% (y(H 2 ) = 2000 ppm)) and 80% (y(H 2 ) = 1 vol.%) were observed in comparison to experiments at λ = 2.4 (Fig. 8). This is attributed to the lower O 2 content decreasing the H 2 oxidation rate thus increasing the amount of hydrogen available for the NO x reduction [17]. However, it is noteworthy that the increase in the H 2 proportion from 0.5 to 1 vol.% had only very little effect on the maximal NO x conversion, which is explained by the heat evolution inside the monoliths increasing the outlet temperature from 135 °C at the inlet to 215 °C at the outlet for 1 vol.% H 2 and hereby shifting the catalyst outside the optimal H 2 -deNO x operation temperature.

Conclusion
The potential of the low-temperature NO x reduction using H 2 as reducing agent (H 2 -deNO x ) was investigated in diesel engine exhaust. The low-temperature range (< 200 °C) is of particular relevance for cold start and city driving scenarios, which are increasingly important to cover real driving emissions. A monolithic Pt/WO x /ZrO 2 catalyst, recently  shown as highly active, was firstly evaluated in synthetic diesel exhaust in form of a 1'' core and was then tested on the level of two consecutive 5.66″ × 3″ bricks in real diesel engine exhaust.
High NO x conversions between 130 and 215 °C with maximal conversion of more than 90% and N 2 selectivity up to 85% were achieved in the model exhaust, whereas CO and HC strongly inhibited the H 2 -deNO x reaction due to competing adsorption and blocking of active Pt sites. Thus, an active pre-DOC is mandatory to keep the low-temperature efficiency of the H 2 -deNO x catalyst.
Stationary tests in real diesel engine exhaust using fullsized bricks substantiated the high H 2 -deNO x performance under city driving conditions including NO x conversions up to 80% at temperatures as low as 135 °C. Simultaneously the exhaust gas temperature was increased up to 80 °C by the strongly exothermic H 2 -O 2 and H 2 -NO x reactions. In future applications, this effect might be used to heat up the SCR stage, preferentially positioned downstream to the H 2 -deNO x unit, to enhance the SCR conversion. It might be speculated that such a faster SCR light-off can reduce the H 2 demand, particularly during cold start phases, since the H 2 is only required for the NO x reduction at low exhaust temperatures. Therefore, the combination of conventional SCR with H 2 -deNO x seems highly promising for substantial NO x reduction at all engine operating scenarios and should be investigated in further studies.
From experiments in an optical reactor, the catalyst inlet is identified to be the main reaction zone during H 2 -deNO x giving the opportunity for further catalyst optimization. Generally, this concept of a combined H 2 -deNO x catalyst with a downstream SCR unit may be useful for other lean combustion engines as well such as gas engines and H 2 engines having the potential to keep the NO x output within future emission limits. However, further catalyst development remains necessary to lower the N 2 O selectivity for all engine operation points.