Experimental investigation of NO reduction by H 2 on Pd using planar laser-induced fluorescence

This study investigates the NO reduction by H 2 on a Pd / Al 2 O 3 catalyst in a temperature range of 100 − 300 ◦ C and NO / H 2 ratios from 0.5 − 2, aiming to gain a deeper understanding of the reaction kinetics and its interaction with mass transfer. Planar laser-induced fluorescence (PLIF) is used to visualize the NO distributions over the catalyst, supplemented by end-of-pipe gas analysis of other components. The reduced Pd-based catalyst undergoes a slow deactivation after exposure to the reactive flow, leading to reduced overall NO conversion and decreased selectivity towards N 2 . The NO-PLIF measurements are only conducted on the reduced catalyst without considering the temporal evolutions. Despite the overall NO conversion varying only around 50-65% across all the investigated conditions, the spatially resolved NO distributions reveal three distinct regimes that limit the overall NO conversion: the regime governed by intrinsic reaction rates, the regime constrained by H 2 availability, and the regime restricted by NO di ff usion. These findings, demonstrating the interaction between reaction kinetics and mass transfer over a heterogeneous catalyst, highlightthesignificanceofanalyzingspatiallyresolvedconcentrationdistributionsobtainedthroughPLIFmeasurements. This approach complements the conventional end-of-pipe analysis, o ff ering a more comprehensive understanding of the underlying processes.


Introduction
In recent years, good progress has been achieved in reducing CO 2 emissions by promoting renewable energies, setting emission targets, and implementing policies toward carbon neutrality.Hydrogen-based renewable energy technologies, such as hydrogen combustion and hydrogen fuel cells, are being widely recognized as crucial means to attain a sustainable and zero-carbon future [1][2][3][4].However, when H 2 reacts with O 2 in the presence of N-containing compounds, such as during H 2 combustion with air, NOx emissions may occur due to high reaction/combustion temperatures [5][6][7].(Ultra-)lean combustion has been proposed to suppress NOx emissions, however, this comes at the expense of thermal efficiency [8].As a result, there is a growing demand for efficient and environmentally friendly NOx removal technologies [9].The selective catalytic reduction of NOx with H 2 is considered as one of the most promising options, since H 2 is usually available both as a fuel and as a reductant and because H 2 O is the only product from the H 2 -O 2 reaction [10][11][12].H 2 -SCR, which typically takes place over catalysts with platinum group metals (PGMs) as the active component, has been extensively studied [13][14][15][16].The reaction occurs through multiple pathways: Reaction ( 2) is considered as the most desirable reaction since it converts NO to harmless N 2 and H 2 O.However, the formation of other byproducts cannot be entirely avoided.Reactions (1-5) can be divided into two sub-mechanisms.
The first sub-mechanism includes Reactions (1-3), which determine the selectivity towards NH 3 , N 2 , and N 2 O.The second sub-mechanism involves combustion reactions (R4-5), primarily contributing to the diminishing reduction efficiency.This is caused either by consuming the reducing agent, H 2 , or by oxidizing NO to NO 2 , which requires more H 2 for reduction.Generally, Pt-based catalysts exhibit the highest overall activity among all PGMs, particularly at low temperatures (≤ 200 • C) [17].However, this high activity is often accompanied by high selectivity towards nitrous oxide (N 2 O) [15,16,18], a greenhouse gas with a 298 times higher global warming potential than CO 2 for a 100-year timescale [19].Alternatively, Pd-based catalysts are gaining increasing attention due to their higher N 2selectivity and lower cost compared to Pt-based catalysts [13,14].The primary objective of H 2 -SCR optimization is to achieve both high NOx conversion and high selectivity towards N 2 , which requires a comprehensive understanding of the reaction network.Although experimental results from a wide range of catalyst systems have been reported [13,14,20,21], the gas-phase experimental data from previous literature is often limited to end-of-pipe concentrations.Neglecting the spatial variance along the catalyst bed may lead to a misperceived intrinsic reaction rate, as the overall reaction may be under the mass transfer limitation.Torkashvand et al. demonstrated in their 3D simulation that modeling the final percentages of the unconverted formaldehyde in the catalytic oxidation monolith converter requires considering the local washcoat distributions, rather than solely focusing on elevating the intrinsic reaction rate in the kinetic mechanism [22].Therefore, spatially resolved concentration data is crucial for accurately assessing the reaction kinetics and mass transport phenomena in H 2 -SCR systems.
Gas sampling techniques with capillary probes [23,24], e.g.SpaciMS and SpaciPro, allow the characterization of 1D concentration or temperature profiles within the catalyst.Moreover, in-situ laser-based techniques have also been applied for catalyst studies, as these are not only non-invasive techniques that are free of bias by possible capillaryinduced flow distortion, but commonly offer a higher spatial and temporal resolution.Several pioneering studies with planar laser-induced fluorescence (PLIF) and Raman spectroscopy have been conducted by Mantzaras et al. to investigate catalyst-assisted combustion [25][26][27].Zetterberg and Lundgren et al. have utilized PLIF to conduct comprehensive studies on CO oxidation over noble metal catalysts [28][29][30].More recently, they have combined PLIF with Raman spectroscopy and near-surface mass spectrometry to acquire detailed gas species distributions during the partial oxidation of methanol [31].In our previous studies, HCHO-PLIF and NO-PLIF have been used to study the catalytic oxidation of HCHO and NOx storage and removal, respectively [32][33][34][35][36]. Compared with flames, to some extent, heterogeneous catalytic reactions are even more suitable for applying PLIF.This is because compensating for the quenching effect in flames is challenging due to significant variations in species and temperature.This work investigated the impact of reaction temperature and gas components on the NO reduction by H 2 over a 1 wt.%Pd/Al 2 O 3 catalyst-coated plate in a parallel wall channel using NO-PLIF measurements.To achieve a precise understanding of each sub-mechanism, this study excludes the competition between NO and O 2 for H 2 , by examining the reaction in the absence of O 2 .The focus lies on the selectivity towards the reduction products, NH 3 , N 2 , and N 2 O. Furthermore, by visualizing the local NO concentration distributions, the interaction between the surface reaction kinetics and mass transport is discussed in detail.Three distinct regimes that constrained the overall NO conversion have been uncovered, demonstrating the significance of spatially resolved data, particularly for catalytic systems involving multiple reaction pathways.

Catalyst materials
Analogous to one of our previous studies [35], the catalyst was prepared via incipient wetness impregnation (IWI).For this, the catalyst support, γ-Al 2 O 3 (Puralox, SASOL Germany GmbH), was first calcined at 700 • C for 5 hours.Palladium (Pd) was then introduced using an aqueous tetraamminepalladium(II)nitrate solution (abcr, 5.0 wt.% Pd) as a precursor, targeting a 1 wt.%Pd/Al 2 O 3 catalyst powder sample.The received powder was dried at 70 • C for 2 hours and subsequently calcined at 550 • C for 5 hours in static air.The catalyst powder and AlO(OH) (Disperal P2, SASOL) are mixed in a weight-ratio of 9 to 1 and water is added to form an aqueous slurry.The amount of the slurry, containing approximately 50 mg of Pd/Al 2 O 3 catalyst powder, was applied to the upper surface of an 18 mm wide, 25 mm long, 1 mm thick section of a monolithic cordierite substrate (NGK, 3.0 mil, 600 cpsi) by using a syringe.Careful rotation and tilting of the substrate ensured a uniform distribution of the slurry.After drying in a static air environment, the coated plate was subjected to a further calcination at 600 • C for 5 hours in static air to ensure secure adhesion of the washcoat onto the cordierite substrate.
The specific surface area and the total pore volume of the powder catalyst was assessed by N 2 -physisorption using a Belsorp mini II (MicrotracBEL) and subsequent evaluation according to the BET method.The BET surface area of the Pd/Al 2 O 3 powder is 173 m 2 /g, and the total pore volume is 0.48 mL/g.The noble metal dispersion, determined through a CO chemisorption measurement, is 43%.

NO-PLIF over catalyst-coated plate
The experimental setup for the NO-PLIF measurements over a 1 wt.%Pd/Al 2 O 3 catalyst-coated plate is illustrated in Fig. 1, which is similar to that described in our previous studies [32][33][34][35][36]. 0.5 mg Pd/Al 2 O 3 catalytic powder is coated on the top surface of an 18 mm wide, 25 mm long plate, which is placed in a ceramic holder to form a 2 mm high parallel wall channel in a quartz pipe.The channel is sealed, and a uniform temperature field is achieved by the heating control unit.The gases are introduced into the reactor using mass flow controllers (MFCs) in combination with 3/2-way solenoid valves, which allow for the control of flow directions -either into the reactor or bypassing it.A heated water tank is connected into the gas line, allowing water vapor to be carried along when N 2 flows through.By controlling the heating temperature and the mass flow of the N 2 carrier gas, the concentration of H 2 O in the inlet gas mixture can be accurately regulated.
After placing the sample in the reactor, the calcined catalyst-coated plate undergoes an additional in-situ oxidation at 450 • C for 12 hours with a 10% O 2 in N 2 gas flow.This step eliminates any potential contamination that might have occurred during the reactor assembly process.After purging with inert gas, the sample is then reduced by 10% H 2 in N 2 at 400 • C for 2 hours.A final purging with inert gas is then performed to ensure that no H 2 remains in the system.
All experiments are conducted at atmospheric pressure with a total flow rate of 1 L/min (NTP ≡ 1 atm & 20 • C).Considering the dimensions of the parallel wall channel and the given flow rate, the experimental conditions are comparable to those of a 100 cpsi monolith catalyst operating with a gas hourly space velocity of 60000 h −1 .A deand reactivation test is performed to ensure that all PLIF measurements are carried out with the catalyst in the same status.The influences of temperature (ranging from 100 to 300 • C) and inlet H 2 concentration (varying from 200 to 800 ppm) are investigated in this study, with a fixed NO concentration of 400 ppm and H 2 O concentration of 1%, balanced with N 2 .In this study, H 2 O in the feed gas has been used as the dominant quencher to mitigate the effects of spatially non-uniform fluorescence quenching.Further details are provided in the next section.Selecting H 2 O as the dominant quencher and N 2 as the carrier gas, rather than other combinations of gases, such as CO 2 and Ar that are chemically fairly inert under the conditions applied throughout this study [37] but do not belong to the products, limits the ability to directly measure H 2 O and N 2 formation.Nevertheless, their formation can still be accurately calculated through mass balance.This choice closely aligns with real-world conditions in industrial applications, particularly in the rapidly advancing field of H 2 -based renewable energy technologies.
The detection of NO is accomplished by exciting the P 2 (12.5) + Q 12 (12.5)line in the NO A 2 Σ + → X 2 Π (0,0) band with a UV laser at 226.67 nm [32], and collecting fluorescence through an image-intensified charge-coupled device (ICCD) camera (IRO 25 & Imager QE, LaVision) with a band pass-filter centred at 248 nm (FWHM = 10 nm, peak transmission 15%, LaVision).The 10 Hz excitation laser is generated by a frequency-doubled Nd:YAG (Quanta-Ray Pro, Spectra-Physics) pumped dye laser (PrecisionScan, Sirah) sum-frequency mixed with the third harmonic output of the same Nd:YAG laser.A Fourier-transform infrared spectrometer (FTIR) (MG2030, MKS) and a mass spectrometer (MS) optimized for H 2 quantification (HSense, MS4) are connected to the exhaust of the reactor to monitor the NO, N 2 O, NH 3 , H 2 O and H 2 concentrations at the outlet and can be alternatively switched to the bypass connection for measuring the inlet concentrations.

Calibration of PLIF signal
All measurements are conducted with the laser power in the linear LIF regime.The fluorescence intensity is given by Equation ( 1) [38], where C exp is an experimentally derived constant, B 12 is the Einstein coefficient for photon absorption, I v is the local laser energy, N NO is the number density of NO molecules, f 1 is the Boltzmann factor, A 21 is the spontaneous emission coefficient and Q 21 is the collisional quenching rate.Therefore, the local absolute NO concentration under reactive conditions, C NO , can be calculated by comparing the PLIF image of the reactive flow to a reference image captured under non-reactive conditions in the same system at an identical temperature.Disregarding minor differences in the laser absorption by NO above the catalytic plate, the local laser intensity distributions for both the target and reference images can be considered as identical.Therefore, the ratio of the local laser energy (I v ) between the two images is equivalent to the ratio of the total laser beam energy (I beam ) employed during the acquisition of these images.The local NO concentration under reactive conditions is given by Equation (2).
Herein, the subscript re f stands for the values under the reference condition.C NO,re f is determined by the setting of MFCs and additionally monitored by the FTIR.I beam and F are the measured values.The determination of the total collisional quenching rate, Q 21 , requires the information on the local concentration for each gas component according to Equation (3).
σ Qi is the cross section for collisional quenching of NO A 2 Σ + (v ′ = 0) by species i, ⟨v i ⟩ is the average collision velocity of the collision pair, NO and species i, and N i is the number density of species i.However, obtaining the local gas concentrations with NO-PLIF solely is not feasible due to the unknown local selectivity among the multiple reaction pathways.Thus, similar to our previous studies, a dominant quencher, H 2 O for this study, has been added into the flow to keep the total collisional quenching rate approximately uniform in the entire channel.Paul et al. reported comprehensively on the quenching of NO A 2 Σ + based on a wide range of experimental data, and developed models that provide physically plausible fits to the measurements [38][39][40].The predicted quenching cross sections for NO fluorescence are summarized in Table 1.
Table 1.Cross sections for collisional quenching of NO A 2 Σ + (v ′ = 0) calculated from the model by Paul et al. [40].400 ppm H 2 , 1% H 2 O, and balance N 2 .The first cycle starts immediately after the pretreatment.As time progresses, a gradual decrease in NO conversion is observed.However, the products NH 3 and N 2 O exhibit an increasing yield over time.The decreased NO conversion is attributed to the reduced N 2 formation.Between t = 1 min and t = 30 min, despite only a slight decrease in NO conversion from 63% to 60%, the selectivity to N 2 significantly dropped from 65% to 42%.A deactivation test was also conducted at a lower temperature (T = 100 • C) (data not shown).Compared to the higher temperature, the aging effect at lower temperature was found to be much less pronounced.The gradual decline in performance is likely attributed to the evolution of structure and chemical state of Pd nanoparticles during the reaction.The redispersion of metallic Pd on the support and the subsequent formation of species, including Palladium ion (Pd 2+ ), partially oxidized Palladium (Pd δ+ ), and Palladium oxide (PdO), have been reported in previous studies, as a result of the reduced catalyst being exposed to reactive flows [41][42][43].
In contrast to the reversible deactivation observed during NO reduction by CO on Pt-based catalysts, where the catalyst can be reactivated by flowing inert gas [44], the aging effect on the Pd-based catalyst can only be reversed through a reduction process with H 2 .Such observation provides support to the notion that the performance drop of the Pd catalyst is linked to the transition from a metallic state to an oxidized state.Previous studies have also reported that the reduction of oxidized Palladium by H 2 can efficiently occur at temperatures as low as 100 • C [45,46].A total of four cycles were performed at T = 300 • C. Between each cycle, the catalyst plate was reduced with 400 ppm H 2 in N 2 for 30 minutes, followed by purging with N 2 until H 2 was not detectable.Notably, a high level of reproducibility was observed between each cycle, as shown in Fig. 2. Consequently, instead of performing a complete pretreatment involving both oxidation and reduction pripor to each subsequent measurement conducted under different conditions, the catalyst was only subjected to a 30-minute reduction process with 400 ppm H 2 in N 2 , followed by purging with N 2 .Furthermore, considering the mild changes in NO conversion over an extended period, this study does not explore the temporal evolution of the 2D NO distributions.All NO-PLIF measurements were consistently conducted at t = 5 min with an average of 300 consecutive laser shots (data acquisition time = 30 s, according to 10 Hz repetition rate).Therefore, the catalyst plate has only been exposed to the reactive flow for less than 6 minutes before undergoing the shortened reduction process.

Catalytic activity measurement
To evaluate the impact of temperature and inlet gas composition on the catalytic activity and the product selectivity, NO-PLIF measurements, along with end-of-pipe analysis of the effluent gas stream, are performed in the channel reactor over the 1 wt.%Pd/Al 2 O 3 catalyst-coated plate.As previously mentioned, all data are collected 5 minutes after the reactive flow is initiated.The data acquisition time is only 30 seconds, which is significantly shorter than the time scale of the aging process.As a result, no temporal evolution is considered in the following data analysis.
Figure 3 illustrates the impact of temperature on the conversions and product yields based on the end-of-pipe measurements at the reactor outlet.The inlet gas mixture consists of 400 ppm NO, 400 ppm H 2 , 1% H 2 O, and balance N 2 .When the temperature is increased from 100 to 200 • C, the NO conversion remains approximately 50%; however, the H 2 conversion rises significantly from 42% to 96%.Within this temperature range, the N 2 formation remains consistently around 20 ppm.Nevertheless, there is a remarkable increase in NH 3 formation with the increased temperature, along with a corresponding decrease in N 2 O formation, particularly noticeable in the temperature range of 100 to 150 • C. Between 200 and 300 • C, nearly full conversion of H 2 is achieved in the channel reactor, indicating that H 2 diffusion is not the limiting factor for its consumption under these conditions, likely due to the large diffusion coefficient of H 2 , which is approximately 4 times larger than that of NO.In the meantime, NO conversion gradually increase from 49% to 62%, accompanied by an increased N 2 production from 25 to 61 ppm. Figure 3(b) displays the contribution of each competing reaction pathway (R1-3) to the overall NO conversion, which is referred to as the NO reaction path share.It is important to note that the formation of one NH 3 molecule requires one NO molecule (R1), whereas the formation of one N 2 or one N 2 O molecule needs two NO molecules (R2, R3).As a result, at T = 300 • C, despite the higher NH 3 concentration at the outlet, only 43% of NO consumption result in NH 3 formation, whereas 49% of NO consumption account for N 2 formation.In the temperature range of 100 to 300 • C, it is observed that only the NO reaction path share for N 2 formation (R2) is monotonically increasing.Considering the stoichiometric coefficients of H 2 in the chemical equations (R1-3), it becomes evident that the apparent reaction rate of NH 3 formation (R3) should be enhanced by an excess of H 2 , which is in contrast to the formation of N 2 O (R1).However, at T = 100 • C, although more than half of the H 2 remains after passing through the reactor, over 70% of NO are consumed for the formation of N 2 O. On the other hand, at T ≥ 150 • C, the NO reaction path share for N 2 O formation (R3) remains less than 15%, irrespective of the low H 2 concentration at outlet.The results indicate that the formation of N 2 O is much more favored at lower temperatures, while the low temperatures inhibit the formation of NH 3 .Similar changes in selectivity with varying temperatures have been reported in other literature [17,21].The NO reaction path share for NH 3 formation (R1) decreases from 67% at 200 • C to 43% at 300 • C. The inhibited NH 3 formation at higher temperature is likely due to the more favored N 2 formation under the conditions with a limited amount of H 2 .
The impact of the inlet H 2 concentration on the catalyst performance has been investigated at T = 300 • C, and the results are shown in Fig. 4. As expected from the stoichiometric relation, the formation of NH 3 significantly increases with the higher H 2 concentration at the inlet.Meanwhile, the NO reaction path share for N 2 O formation (R3) decreases from 24% with 200 ppm H 2 at the inlet to 0% with 800 ppm H 2 at the inlet.However, even though the coefficient ratio between NO and H 2 is 2 for N 2 O formation (R3) and 1 for N 2 formation (R2), with 400 ppm NO and 200 ppm H 2 at the inlet, the formation of N 2 remains the dominant reaction path.Besides the reaction kinetic reasons, the much slower diffusion of NO compared to H 2 also contributes to a reduced local NO/H 2 ratio over the catalyst plate.The stoichiometric coefficient ratio between NO and H 2 is 0.4 for NH 3 formation (R1).According to this stoichiometric relation, it is theoretically possible to achieve full conversion of H 2 with a gas mixture consisting of 400 ppm NO and 800 ppm H 2 .However, when the inlet H 2 concentration is raised to 800 ppm, the H 2 conversion reduces from full conversion to 72%.This reduction is partially attributed to the limited diffusion rate of NO, which will be further discussed in the later analysis of the NO spatial distribution.

Spatially resolved NO distribution over catalyst plate
Figure 5 shows the NO-PLIF images acquired over the 1 wt.%Pd/Al 2 O 3 catalyst-coated plate in the 2 mm high channel under selected conditions.The position of the sample plate is marked as the pink line at the bottom of each sub-figure.The spatial resolution of the PLIF images is 22 µm/pixel.Uniform NO distribution is observed before the catalyst plate, indicating that no gas-phase reaction happens under the given conditions, and the streamwise diffusion is not significant.Under the influence of surface reactions, diffusive and convective mass transfer, NO concentration gradients appear in both wall-normal and streamwise directions above the catalyst plate.The concentration gradients vary at distinct locations and also under different conditions.For a better comparison of quantities, the NO concentration profiles extracted from Fig. 5 are plotted in Fig. 6 and 7.The profiles in the streamwise direction represent averaged NO concentrations over the entire channel height, while the profiles in the wall-normal direction show the local NO concentration distributions at x = 5 mm (dashed lines) and 20 mm (solid lines).The local net NO consumption rate can be evaluated from these spatially resolved profiles using the following two approaches: (1) By dividing the channel into small finite elements in the streamwise direction, the local net NO consumption rate is determined by the mass difference between the NO flowing into and out of each small finite element.Since the mass flow rate is constant for all conditions, regardless of thermal expansion due to variations in reaction temperature, the gradient of the averaged NO concentration in streamwise direction directly reflects the local NO mass loss.It is to note that, within this discussion, the term "concentration" refers to the mole fraction in units of ppm, rather than the molar concentration expressed in mol/L.(2) Considering that the flow velocity at the wall boundaries is zero, the conductive mass transfer in the vicinity of the catalytic surface becomes negligible.Therefore, the local net NO consumption rate can reasonably be approximated as equal to the NO mass flux driven by diffusive mass transfer near the catalytic surface, which can be evaluated by the local NO concentration gradient in the wall-normal direction.However, it should be noted that the differences in gas density and diffusion coefficient stemming from temperature variations should be taken into consideration when comparing cases with different temperatures.In addition to the local NO consumption rate, the NO concentration profiles in the wall-normal direction provide valuable information for assessing diffusive mass transfer.
When comparing T = 100 and 200 • C, the NO concentrations at the outlet are approximately at the same level.However, as depicted in Fig. 6 (a), at T = 100 • C, the NO conversion along the catalyst plate shows minor variation, whereas at T = 200 • C, a much faster NO conversion is observed upstream, but a slower one downstream.This pattern is also evident in the wall-normal concentration gradients, where at x = 5 mm (dashed lines), a much larger concentration gradient is observed at T = 200 • C, while at x = 20 mm (solid lines), the concentration gradient at T = 100 • C becomes more pronounced.The observed results can be related to the difference in the overall H 2 conversion as shown in Fig. 3 (a).At lower temperature (T = 100 • C), the formation of N 2 O, which consumes much less H 2 , is favored over NH 3 formation.Consequently, 58% of H 2 from the feed gas remain after flowing through the reactor, indicating the availability of H 2 for reacting with NO downstream.At upstream, subtle concentration gradients of NO are observed both in the streamwise and wall-normal directions, indicating a relatively low local NO conversion rate.In the vicinity of the catalyst surface, the relatively high local NO concentration at x = 5 mm implys that NO mass transfer does not predominantly restrain the NO conversion rate here.Considering the availability of H 2 and the significantly faster diffusion of H 2 in comparison to NO, the primary constraint on the local NO conversion rate is rooted in the intrinsic reaction rates.A gradual decrease in the concentration gradient of reactants is typically anticipated along the streamwise direction.This expectation arises from the diminishing rate of reactant consumption correlated with the declining reactant concentrations.However, a slight anomaly emerges This rise in the net NO consumption rate, even with a reduced gas number density resulting from the elevated temperature, serves as a distinct indication of the pronounced temperature-dependent increase in intrinsic reaction rates.Moreover, in the upstream region, the notable NO concentration gradient in the wall-normal direction provides clear evidence that the local NO consumption rate in this area is strongly influenced by NO diffusion.However, downstream, the diminished NO concentration gradients imply a lessened limitation due to NO diffusion.On the other hand, at T = 200 • C, the H 2 conversion increases to 96% as the NO reaction path share for NH 3 formation reaches its maximum, as shown in Fig. 3.It is very likely that the major H 2 consumption has already occurred upstream by forming NH 3 , leading to a significantly reduced NO conversion rate downstream due to the lack of H 2 .
As the reaction temperature raised from 200 to 300 • C, an enhanced NO conversion is observed downstream, despite the further increase in H 2 conversion.This enhancement can be attributed to the higher selectivity towards N 2 formation at the elevated temperature.Since the formation of N 2 (R2) consumes less H 2 compared to NH 3 formation (R1), a larger amount of H 2 remains available for reaction downstream, thereby maintaining the NO conversion rate along the entire channel.As shown in Fig. 6 (b), the NO concentration gradients at T = 200 and 300 • C are similar at both x = 5 and 20 mm.Considering that the diffusion coefficient is proportional to T 1.5 and gas density is proportional to T −1 , a faster diffusion-driven NO mass flux at the catalytic surface can be expected at the higher temperature, when such wall-normal concentration gradients remain similar.Evidently, the local net NO consumption rate, which is approximately equal to this diffusion-driven mass flux, is expected to be higher along the entire channel at T = 300 • C.This faster conversion is attributed not only to the higher intrinsic reaction rate but also to the selectivity shift towards the more efficient NO consuming reaction pathway (from R1 to R2).
Figure 7 compares the NO concentration profiles at T = 300 • C with varying inlet H 2 concentrations.As shown in Fig. 4, a full conversion of H 2 is reached with either 200 ppm or 400 ppm H 2 at the inlet.NO concentration gradients at x = 5 mm are very similar under these two conditions; however, with less H 2 in the feed gas, the NO conversion downstream is greatly suppressed.The significant reduction in the downstream NO concentration gradients bear resemblance to the situation depicted in Fig. 5 (b), with 400 ppm H 2 at T = 200 • C. In both cases, the overall NO conversion is constrained by the availability of H 2 downstream.When the inlet H 2 concentration increases from 400 to 800 ppm, the H 2 conversion reduces, indicating that more H 2 is available to react, especially at downstream position.Nevertheless, the NO distributions in both streamwise and wall-normal directions maintain a similar shape along the entire channel.The fact that the NO conversion rate remains unchanged despite having more potential reacting partners suggests that the NO conversion under these conditions is limited by the NO diffusive mass transfer.The diminished NO concentration in the vicinity of the catalytic surface further supports this proposition.However, with a larger amount of H 2 feeding, the local H 2 concentration becomes much higher, resulting in a significant increase in selectivity towards NH 3 formation.
The local gas concentration distributions play a crucial role in heterogeneous catalytic reactions, particularly when multiple reaction pathways are involved.At low reaction temperatures, e.g.T = 100 • C, the NO conversion is primar-ily limited by the low intrinsic reaction rate, particularly in the reaction path of NH 3 formation.Consequently, both NO and H 2 undergo limited overall conversions.However, the concentration gradients in the wall-normal direction indicate that the overall reaction rate is still affected by diffusion.At elevated temperatures with a limited quantity of H 2 in the feed gas, remarkable variations in NO concentration gradients are observed between the upstream and downstream regions.This observation suggests that NO conversion is predominantly consumed upstream, and the reaction progression is impeded downstream due to the lack of H 2 .Importantly, this phenomenon is not only observed under conditions with a NO/H 2 ratio greater than 1, it is also applicable under conditions favoring NH 3 formation, which leads to rapid upstream H 2 consumption.A NO diffusion-limited regime emerges at elevated temperatures with a low NO/H 2 ratio.Despite an excess of H 2 and the elevated intrinsic reaction rates, the overall conversions of both NO and H 2 are hindered by the diffusion rate of NO.For all cases, due to the significant difference in the diffusion coefficients between NO and H 2 molecules, the local NO/H 2 ratio at the catalytic surface tends to be higher than the averaged NO/H 2 ratio on the cross-sections.This leads to a selectivity towards those reaction paths that are favored by higher concentrations of H 2 .Quantitative comparisons of the chemical reaction timescales, diffusive and convective mass transfer timescales, using dimensionless numbers such as the Damköhler number, would offer a clearer understanding of how the controlling factors shift among all conditions [47].Future numerical and experimental investigations will continue to enhance this discussion after revealing local H 2 distributions and the detailed reaction kinetics.

Conclusions
Spatially resolved NO concentration distributions were acquired through NO-PLIF measurements conducted over a Pd/Al 2 O 3 catalyst plate during the NO reduction by H 2 .In combination with end-of-pipe gas analysis, this study has explored the interaction between surface reaction kinetics and mass transfer under diverse temperature and NO/H 2 ratio conditions.
The end-of-pipe gas analysis revealed at lower temperature (T = 100 • C), the formation of N 2 O was favored, while NH 3 formation was suppressed.It is also noteworthy that under specific conditions, full H 2 conversion can be achieved, suggesting that the diffusive mass transfer of H 2 is generally not a limiting factor for the overall NO conversion.Spatially resolved NO distributions unveil three distinctive regimes imposing limitations on the overall NO conversion: the regime governed by intrinsic reaction rates, the regime constrained by H 2 availability, and the regime restricted by NO diffusion.The overall conversion, local concentration gradient, and product selectivity are all influenced by the combined effects of kinetics and mass transport.It is essential to highlight that, under distinct reaction conditions, different NO concentration distributions can arise even under a similar overall NO conversion, and differing product selectivities can emerge despite the similarity in shapes of the 2D NO distributions.These findings illustrate the importance of analyzing spatially resolved NO concentration distributions obtained via PLIF measurements, in addition to solely relying on the end-of-pipe analysis.Moreover, considering the envisaged shifts in the local selectivity along the catalyst bed based on diffusion properties, acquiring spatially resolved data for all relevant gas components becomes essential when studying catalytic systems involving multiple reaction pathways.Future investigations will extend to exploring concentration distributions of species beyond NO, contributing to a more comprehensive comprehension of the H 2 -SCR system.

J
o u r n a l P r e -p r o o f Journal Pre-proof S. Wan et al. / Applications in Energy and Combustion Science 00 (2023) 1-14 2

Figure 1 .
Figure 1.Schematic of the experimental setup for NO-PLIF measurements with controlled dosing of H2O vapor in the gas mixture.

C
NO = C NO,re f I beam,re f I beam F F re f (4) 3. Results and discussion 3.1.Catalyst deactivation experiment In Fig. 2, the FTIR-measured outlet concentrations of NO, NH 3 , and N 2 O, along with the calculated N 2 formation, are plotted for the aging test conducted at 300 • C. The inlet gas mixture used for this test consists of 400 ppm NO,

Figure 3 .
Figure 3. Results with an inlet gas mixture of 400 ppm NO, 400 ppm H2, 1% H2O, balanced with N2 in the temperature range of 100 -300 • C. (a) Formation of the reaction products and conversion of the reactants.(b) Reaction shares of different NO reaction pathways.R1: Formation of NH3, R2: Formation of N2, R3: Formation of N2O.

Figure 5 .
Figure 5. PLIF-measured NO distributions over the 1 wt.%Pd/Al2O3 catalyst-coated plate (marked as the pink line) in the 2 mm high parallel wall channel.

J 10 Figure 6 .
Figure 6.NO concentration profiles extracted from PLIF images captured in the temperature range of 100 -300 • C with an inlet gas mixture of 400 ppm NO, 400 ppm H2, 1% H2O, balanced with N2.(a) Streamwise direction averaged over the channel height.(b) Wall normal direction.

Figure 7 .
Figure 7. NO concentration profiles extracted from PLIF images captured at 300 • C with an inlet gas mixture of 400 ppm NO, 200 -800 ppm H2, 1% H2O, balanced with N2.(a) Streamwise direction averaged over the channel height.(b) Wall normal direction.
Considering that only σ Q for H 2 O exhibits a significant decrease from 100 to 300 • C, and NH 3 exhibits the largest σ Q among the products from NO, the maximum error caused by the non-uniform local quenching effect in this study can be estimated by assuming a 400 ppm NH 3 local production at 300 • C. In this worst-case scenario, the value of the last term in Equation (2),A 21 + Q 21 A 21 + Q 21,re f, amounts to 1.032.This implies that when computing the local NO concentration using Equation (4) without accounting for the non-uniform local quenching effect, the calculated local NO concentration can have a maximum discrepancy of 3.2% compared to the actual value.