In situ Activation of Bimetallic Pd−Pt Methane Oxidation Catalysts

This study investigates possible pathways to reduce water inhibition over Pd−Pt methane oxidation catalysts supported on alumina and a ceria‐zirconia mixed oxide under conditions typical for lean burn gas engines. Spatially resolved concentration and temperature profiles reveal that addition of hydrogen to the lean reaction mixture leads to significant heat production and an increase in methane conversion. Moreover, reductive pulses during operation are not only able to regenerate the catalyst deactivated by water by removal of OH‐groups from the catalyst surface, but even promote its activity after repeated application of pulsing for several hours. X‐ray absorption spectroscopy reveals the formation of a partially reduced PdO−Pd mixture during the pulsing, which explains the increase in activity. This state of high activity is stable for several hours under the tested lean conditions. The results presented in this study suggest an efficient in situ activation strategy to overcome water inhibition of methane total oxidation over Pd−Pt catalysts by careful process control.


Introduction
Natural gas engines may increasingly replace diesel and gasoline engines due to the ubiquity of natural gas as fuel and the lower CO 2 and particulate matter emissions resulting from natural gas combustion. Unfortunately, a small amount of methane always remains after combustion. As CH 4 is a much more potent greenhouse gas than CO 2 , this CH 4 slip -as much as several hundred ppm depending on the vehicle operationmay become a significant obstacle to the widespread use of natural gas as a transportation fuel. [1] Therefore, an efficient after-treatment system must be able to abate this remaining methane. This is a challenge for both, lean combustion and stoichiometric combustion; in the case of the latter the Three-Way Catalyst (TWC) is largely ineffective. [2] The typical lean-burn engine exhaust has several challenging obstacles for effective catalytic methane oxidation. The engine exhaust stream is highly lean (up to approximately 10 % O 2 ), contains significant amounts of water (up to 12-15 vol.%), contains SO x (fuel source dependent) and has only moderate temperatures (typically below 500°C). [3] Supported platinum group metals (PGM) are the favored active catalytic components for the treatment of methane under these conditions, with Pdbased catalysts showing the highest activity. [3,4] Under lean conditions, the active state of Pd is believed to be PdO, although there is an ongoing debate as to whether other phases of Pd or a combination of phases could be more active. [5][6][7][8][9][10][11][12][13][14][15][16][17] In addition to PGM loading, the catalytic activity is affected by the catalyst support, metal dispersion, gas environment and even operating history. [18][19][20][21][22][23] Bimetallic catalysts, wherein a small amount of Pd is replaced with Pt, have been found to be more active, more hydrothermally stable and sintering resistant, and possess somewhat higher sulfur resistance compared to monometallic Pd catalysts. [24][25][26][27][28][29][30][31] Furthermore, research by Epling and Hoflund has shown that mixtures of Pd 0 and PdO are more catalytically active than pure PdO, presumably due to facilitated CH 4 dissociation on metallic Pd. [32] Recent work by Chin et al. shows that the site pair Pd-PdO is effective for CH 4 dissociation to CH 3 and OH. [33] Despite the improved performance from the addition of Pt, inhibition from water vapor and sulfur compounds still pose significant challenges. Water vapor inhibits catalytic activity through competitive adsorption with methane as well as deactivation through formation and accumulation of surface hydroxyl species. [34][35][36][37] Previous research has shown that brief reductive periods during lean operation of Pd-based catalysts can not only regenerate catalysts deactivated by water or sulfur compounds, but can also stabilize their catalytic performance above what it would otherwise be. [18,22,38,39] Similarly, Petrov et al. applied periodic rich phases to continuously regenerate a Pd/zeolite catalyst for lean methane oxidation. [40] In the context of methane abatement in the exhaust of a natural gas engine by a TWC, Bounechada et al. similarly found that periodic rich pulses could recover catalytic activity lost due to water inhibition and that periodic oscillations resulted in higher and more stable catalytic methane conversion over a Pd/Rh-based TWC. [41] Despite some references to the beneficial effects of reductive regenerative periods on Pd-based methane oxidation catalysts that have been deactivated by water in a lean environment, [22,40,42] to our knowledge there has been no substantial systematic investigation on the potential effects and underlying mechanisms by which reductive pulses can benefit lean methane oxidation.
Starting from a broad data set of systematic fundamental experiments, this study presents a method for in situ activation of PdÀ Pt/Al 2 O 3 and PdÀ Pt/CeO 2 À ZrO 2 À Y 2 O 3 À La 2 O 3 (in the following referred to as PdÀ Pt/CZ) methane oxidation catalysts under different temperatures and gas compositions. Specifically, periodic reductive pulses of hydrogen are demonstrated to result in two distinct effects dependent upon the operating conditions. Further knowledge of the processes on the catalysts were gained by spatial profiling of the gas species concentrations and gas phase temperature using a capillary technique. With our study we aim at contributing to more efficient lowtemperature methane oxidation over Pd-based catalysts and understanding the factors governing high catalytic activity.

Results and Discussion
Light-off trends at varying feed composition. Figure 1  For the PdÀ Pt/CZ catalyst under baseline operating conditions (A) -no H 2 O or H 2 in the feed -the light-off temperature is 360°C and complete CH 4 conversion is achieved at 450°C. The addition of water to the feed (B) has a well-known significant inhibitory effect on the catalytic activity, [9,37,43] increasing T 50 bỹ 100°C and the complete conversion temperature to~550°C. This is a result of surface hydroxyl formation on top and at the periphery of the noble metal particles, possibly in the form of Pd(OH) 2 . [44] These hydroxylated sites are considered catalytically inactive. [36,37] Addition of 1 % H 2 to the feed promotes CH 4 oxidation. The conversion obtained with the feed containing H 2 (curve C) has a T 50 of 325°C and a complete conversion temperature of 380°C. Note that at lower temperatures (< 300°C) the feed with H 2 has a slightly lower CH 4 conversion. This may be a result of inhibition by the water generated from H 2 oxidation, resulting in blockage of active sites. When H 2 O and H 2 are present (D), T 50 decreased by 60°C from that of the wet feed without H 2 (B). This is attributed to a temperature increase from the exothermic H 2 oxidation. The same general trends were also observed for the PdÀ Pt/Al 2 O 3 catalyst (Figure 1b). For this catalyst, the addition of H 2 results in a T 50 35°C lower compared to dry conditions and 45°C lower under wet conditions.
To examine the impact of H 2 in more detail, light-off experiments for different concentrations of H 2 in the presence of water were conducted. The dry feed with 1 % H 2 is provided for comparison ( Figure 2). The general trend is a shift of the light-off curves to the left with the addition of H 2 . It is interesting to note the rather insensitive dependence of the CH 4 conversion on H 2 concentration between 0.2 % and 2 % for the wet feed. However, by 5 % the light-off curve shifts sharply to lower temperature. This nonlinear dependence of the CH 4 conversion on the H 2 concentration is caused by competition between the beneficial impact of heat generation and the detrimental impact of hydroxyl formation. Moreover, there is a shift from the typical sigmoidal shape at low H 2 feed concentrations to an unusual convex curve when 5 % H 2 are   added. This conveys the much lower activation energy for H 2 oxidation compared to CH 4 oxidation.
Spatially resolved species concentration and temperature profiles (so-called spatial profiling, hereinafter mentioned as SpaciPro) provide further insight. Figure 3 shows the concentration measurements along the catalyst. When O 2 is present, CH 4 is completely oxidized to CO 2 and no CO was detected during these experiments. SpaciPro measurements under steady state conditions consistently suggest that only minor amounts of H 2 reach the front heat shield (FHS, used to minimize temperature gradients) zone or the catalyst itself. Similar behavior was observed for the PdÀ Pt/CZ catalyst. In one light-off experiment, the capillary connected to the mass spectrometer (MS) was placed in-between the FHS and the catalyst. According to the MS data, 94 % of the H 2 reacted with oxygen before it reached the catalyst; as the temperature increased this value approached 99 %. The remaining H 2 was quickly converted to water upon contact with the catalysts. This behavior is attributed to a combination of effects. With its small size, H 2 is able to back-diffuse against the convective current, which can potentially lead to apparent pre-catalyst consumption due to mixing. In addition, the MS settings could cause a reaction of H 2 with O 2 inside the ionization chamber, which may result in misleading high H 2 conversion. Thus, it is likely that H 2 reaches the catalyst but cannot be detected due to experimental hurdles. As H 2 oxidation is a highly exothermic reaction, the temperature profiles discussed later clarify this open question and confirm that H 2 in fact is converted at the catalyst inlet.
The presence of H 2 affects the catalyst during both pretreatment and operation. Discussed in more detail in the following section, spatial temperature measurements during transient operation show a significant temperature rise as a result of the exothermic oxidation of H 2 . This in turn leads to an increase in CH 4 oxidation rate. That the relationship between H 2 feed concentration and CH 4 oxidation light-off is nonlinear underscores contributions from the reaction exothermicity and the aforementioned sensitivity to H 2 O.
The enhancement in light-off behavior due to the reductive pretreatments may be ascribed to separate interactions between H 2 and the support and Pd complexes. Bozo et al. showed that CZ plays an active role in CH 4 oxidation, augmenting the supply of oxygen. [20] Reductive pretreatments of H 2 were shown to create surface oxygen vacancies in the CZ, allowing electron transfer to the Pd and resulting in higher catalytic activity. Xiong et al. and Mahara et al. reported that reductive treatments of PdO complexes resulted in the formation of metastable PdOÀ Pd 0 domains which are more active for CH 4 oxidation. [5,45] Effect of pulses: continuous O 2 stream while pulsing H 2 . The catalysts were exposed to pulses of H 2 , with all other gases kept constant, for the duration of the SpaciPro measurements. Figure 4 shows the CH 4 conversion and effluent temperature for PdÀ Pt/CZ under steady-state operation and during H 2 pulses for dry and wet conditions. For a dry feed, introduction of 1 % H 2 results in an initial decrease in conversion due to the H 2 O generation and its subsequent adsorption. For a wet feed, the additional water generated by the H 2 oxidation has little effect. As discussed above in the context of Figure 3, most of the H 2 appeared to be converted before reaching the catalyst during steady-state operation. The MS data suggest a similar behavior during H 2 pulsing experiments. SpaciPro temperature probes show the temperature at the front of the catalyst by 85°C higher during the 1 % H 2 pulse, regardless of initial catalyst temperature or the absence/presence of water. The measured temperature rise, evident in Figure 5, points to the fact that most of the H 2 reaches the monolith, in contrast to the concentration profile obtained from evaluation of the SpaciPro data. The temperature increase during H 2 pulsing leads to a cycle average increase in the CH 4 conversion under 2 % and 3 % H 2 pulses, and no net change for the 1 % pulses. In all cases, the H 2 pulsing is unable to prevent the long-term decline in the CH 4 conversion over several hours.
As previously stated, spatial temperature profiles for the 1 % and 2 % H 2 feed concentrations explain the effects of H 2 pulsing. The initial temperature profile (time = 0 s, before H 2 pulse initiation) shows that the temperature increases along the monolith length due to the incomplete conversion of CH 4 . Over the next 30 s the pulsed H 2 oxidizes and results in a dramatic temperature increase at the front of the monolith. A maximum temperature rise of~85°C is observed for the 1 % H 2 pulse while a~155°C increase is encountered when pulsing 2 % H 2 . After the H 2 pulse ends, the heat diffuses down the catalyst. The heat from H 2 oxidation remains well after the produced H 2 O, resulting in the temporary boost to CH 4 oxidation. With respect to the experimental data shown in Figure 3b, the temporal temperature increase around the catalyst inlet ( Figure 5) that originates form the high exothermicity of the H 2 oxidation clearly indicates that H 2 actually reaches the catalyst and is not consumed before.

Effect of pulses: switching between O 2 and H 2 streams.
Alternating between O 2 and H 2 pulses has a dramatic effect on catalytic performance. Figure 6 shows the CH 4 conversion for PdÀ Pt/CZ over a period of about 4 h during lean-rich (O 2 /H 2 ) pulsing. Figures 6a and 6b correspond to a dry feed carried out at 295°C and 315°C, respectively. The experiments demonstrate the sustained CH 4 conversion achieved with the H 2 /O 2 pulsing, in contrast to the gradual conversion decline over the same period for the H 2 pulsing in a constant 10 % O 2 (Figure 4). At 290°C (Figure 6a) the conversion is sustained at~5-22 % from its~30 % initial level while at 315°C (Figure 6b) the conversion is sustained at~5-70 % from its~60 % initial level.
Spatially resolved temperature profiles shown in Figure 7 for the 315°C dry feed experiment reveal temperature changes along the catalyst bed consistent with the slowing and accelerating of the CH 4 oxidation reaction. Between 0 s and 30 s, H 2 is admitted while between 31 s and 210 s O 2 is admitted. During the H 2 feed the temperature drops due to the lack of O 2 . Upon the admission of O 2 heating of the entire catalyst is observed. Unlike in the case of H 2 pulses in a constant O 2 flow for which a significant hot spot was observed in the front region, the temperature profile during H 2 /O 2 pulses are much more gradual and have the same shape as the initial temperature profile.
A more detailed look at the CH 4 conversion and effluent monolith temperature for the 315°C dry-feed and 400°C wet-feed experiments ( Figure S9) reveals additional information. For the dry feed experiment the initial decrease in conversion is likely due to the rapid consumption of H 2 by stored O 2 . This results in the generation of water and the rapid inhibition of CH 4 oxidation by surface hydroxyls. An even larger drop in performance quickly follows this initial decrease as gaseous O 2 is fully removed from the system and CH 4 conversion ceases. At the end of the H 2 pulse and the start of the O 2 pulse, the CH 4 conversion returns to a value slightly higher than that at the start of the cycle. Over many cycles, this slight increase in activity can result in substantial reactivation of the catalyst. More complex behavior is seen in the experiments conducted in the presence of water ( Figure S9b). A brief spike in CH 4 conversion is observed when the feed is switched from lean to rich. This is followed by a steep drop in conversion as the available oxygen is consumed. Finally, as oxygen reenters the system, the conversion overshoots its value at the start of the cycle before rapidly deactivating in the lean phase.
Experiments carried out over a range of conditions indicated that the first pulse typically resulted in the largest change in the CH 4 conversion. The ultimate change depends on the initial state of the catalyst. For example, in Figure 8a, for a dry feed at 290°C the H 2 pulse led to a sharp decrease from an initial conversion of 30 % to just~3 %. Subsequent H 2 /O 2 pulsing did not restore the conversion, with conversion ranging from 3-25 % and a cycle-average value of 21 %. Under wet conditions, the first pulse significantly raised CH 4 conversion. At 400°C, the conversion rose from an initial value of 30 % to 75 % and an ultimate cycle-average value of 57 % (Figure 6c). At 430°C, the conversion jumped from 55 % to 98 % and a cycleaverage value of 96 % (Figure 6d). Again, these conversions were sustained for several hours as the catalyst remained under lean/rich switching.  The difference in behavior between wet and dry conditions can be explained by two competing factors. The first is the catalysts sensitivity to water. For a dry feed, the H 2 O resulting from the reductive H 2 pulse inhibits CH 4 oxidation over PdO. At higher temperature the H 2 pulse is more effective in removing surface hydroxyls. The processes on the catalyst can be summarized considering the following four reactions [Eqs. (1-4)]: For the dry feed, the only source of H 2 O is the oxidation of CH 4 , so the surface coverage with hydroxyls is low. The admission of H 2 leads to a large increase in the hydroxyl coverage as a result of reactions [Eq.  2)], which is more effective at higher temperature.
Carbon monoxide was observed during wet experiments ( Figure S10), particularly over the PdÀ Pt/Al 2 O 3 catalyst. The formation of CO occurred only when the H 2 stream was engaged. CO concentration increased with increasing temperature; at 430°C nearly 130 ppm of CO was produced during the switching pulse by the PdÀ Pt/Al 2 O 3 catalyst. The PdÀ Pt/CZ catalyst produced only 30 ppm of CO under identical experimental conditions. Under dry conditions, CO formation was below 10 ppm if its formation was observed at all. While under highly lean conditions (10 % O 2 ) the dominating process is the total oxidation of methane to carbon dioxide [Eq. (4)], CO formation is likely the result of steam reformation between the water and methane [Eq. (5)]: The improvement in CH 4 oxidation performance by applying rich pulses is long-lasting and persists between experiments and after numerous pretreatments. For illustration of this observation, the catalyst was exposed to the dry conditions of the experiment in Figure 6b. After an initial decrease in the conversion to 50 %, a gradual increase in conversion was observed over a several hour period of pulsing (2000 ppm CH 4 , bal. Ar with 30 s/180 s of 1 % H 2 /10 % O 2 at 315°C). Steady state operation on the following day resulted in 15 % higher initial CH 4 conversion compared to the previous day's experiment. Furthermore, it took several hours before the deactivation seen during the prior day was reached. Initial steady-state SpaciPro measurements and light-off curves showed CH 4 conversion of 20 % over the PdÀ Pt/CZ catalyst at 315°C. After several experiments with O 2 /H 2 switching pulses, a CH 4 conversion as high as 80 % was observed. Figure 8a shows the light off-curve of a fresh PdÀ Pt/CZ catalyst and subsequent light-off curves after varying durations of H 2 /O 2 switching pulses, while Figure 8b shows the behavior of the PdÀ Pt/Al 2 O 3 catalyst after four hours of pulsing. After 10 hours of H 2 /O 2 pulsing, the PdÀ Pt/CZ catalyst has experienced a 30°C shift of the T 50 to 330°C. The PdÀ Pt/Al 2 O 3 catalyst shows an even more pronounced shift in the T 50 , from 350°C to 270°C, as well as the development of a non-monotonic light-off curve. This effect has been demonstrated with H 2 /O 2 pulses as short as 10 s and separated by lean-phases as long as 600 s, suggesting that the pulse duration, quantity of H 2 involved and pulse separation time can be optimized for a maximal CH 4 oxidation activity.
Our results suggest that PdO particle size/dispersion and temperature effects play a minor role and cannot explain the enhanced CH 4 oxidation in the experiments described before. Transmission electron microscopy (TEM) measurements of PdÀ Pt/CZ and PdÀ Pt/Al 2 O 3 reveal no significant change in PGM dispersion between freshly prepared samples and samples that experienced H 2 /O 2 pulsing (30 s/180 s) for 4 h at 315°C for either catalyst (Figures S3-S5). These findings are similar to results of Mahara et al.: [45] Reduction of PdO/Al 2 O 3 below 415°C does not affect the particle size of PdO. The previously discussed spatial-temperature measurements of H 2 /O 2 pulses (Figure 7) do not show the large exothermicity responsible for the H 2 pulsing and steady-state effects (Figure 8). Hence, we attribute the enhanced CH 4 oxidation seen during H 2 /O 2 switching results to the contributions of several, simultaneous, independent effects occurring during the reductive treatment; namely: (i) removal of accumulated hydroxyl species; (ii) formation of metastable PdOÀ Pd complexes; (iii) creation of oxygen vacancies in the catalyst supported on CZ. These are described next.
The inhibitory effects of hydroxyl surface species accumulation have already been discussed; removal of these species can be clearly concluded from the SpaciPro results during the H 2 /O 2 pulse experiments. H 2 is observed to decrease in concentration along the length of both catalysts concurrent with the release of H 2 O from the catalyst. Removal of these hydroxyl species also explains the observed brief boost in CH 4 conversion under wet conditions at 400°C ( Figure S9). By keeping water from the catalyst surface, the catalyst can briefly take advantage of the higher temperatures before O 2 completely leaves the system. That the catalysts do not lose activity over several hours under H 2 /O 2 pulsing conditions implies that the rate at which H 2 removes the OH species is far faster than they can accumulate.
The second effect concerns the active metal Pd. Both Mahara et al. and Xiong et al. found that reductive conditions partially reduce PdO complexes into metastable PdOÀ Pd structures that were reported to show higher catalytic activity than pure PdO. [5,45] Xiong et al. went further and showed distinct regions of PdO and Pd on the same particle in high resolution TEM images. [5] Although we cannot see this in our TEM images, XAS spectra of PdÀ Pt/CZ and PdÀ Pt/Al 2 O 3 samples which were pretreated by H 2 /O 2 pulsing reveal reduced Pd coexisting with PdO ( Figure 9). These samples were pulsed for four hours at 315°C and were exposed to oxidizing conditions for over a minute before being cooled in Ar to room temperature. To determine the amount of metallic Pd and PdO before and after the pulse treatment, a linear combination fitting (LCF) of the Pd and PdO reference spectra was performed in the range of À 20 to 30 eV around the Pd K-edge. The results obtained show that the fresh samples are fully oxidized after preparation. After pulsing, however, the fraction of metallic Pd is~55 % on PdÀ Pt/ Al 2 O 3 , while the fraction of Pd is only~14 % on PdÀ Pt/CZ. Additionally, TEM revealed more pronounced alloying between the Pd and Pt on the Al 2 O 3 support than on the CZ support ( Figure S3). This is true not only in the fresh state, but also after H 2 /O 2 pulsing ( Figure S4) and is consistent with the findings of Xiong et al. who found that the Pt helped maintain the reduced state of Pd under oxidizing conditions. [5] We attribute the decrease in the light-off temperature of these catalysts predominantly to this effect. Furthermore, we attribute the larger improvements over the PdÀ Pt/Al 2 O 3 catalyst, as seen in Figure 8, to the increased amount of metallic Pd in combination with pronounced alloying between Pd and Pt. Apparently the activity can be maximized by the right Pd/PdO ratio, especially if promoted or at least stabilized by Pt. According to the found PdO/Pd ratios we can speculate, that CH 4 oxidation still takes place according to a Mars-van-Krevelen mechanism on PdO as frequently claimed. [7][8][9]16,17] The presence of metallic Pd may destabilize the PdÀ O bond, which could ultimately result in a significant increase of the catalytic activity.
The final of the three effects concerns the PdÀ Pt/CZ catalyst but not the PdÀ Pt/Al 2 O 3 catalyst. The reductive pulses of H 2 result in oxygen vacancies on reducible the ceria-zirconia support. Bozo et al. found that the CZ support played an active role in CH 4 oxidation for a Pd/CZ catalyst. [20] In the absence of gas phase oxygen, lattice oxygen was easily removed and the subsequent vacancies promoted electron transfer from the support to the Pd as well as increased lability and mobility of further lattice oxygen atoms. Hence, we can assume a similar behavior for our catalyst sample for two reasons. First, the H 2 pulses utilized in this experiment are much more reductive than the CH 4 pulses utilized by Bozo et al. [20] in their studies; second, SpaciPro results during pulsing revealed the release of H 2 O during the reductive pulse. However, in the current study we are unable to distinguish what percentage of those released species are surface hydroxyl species and which are the result of reactions with the lattice oxygen. For this, further investigations are needed, including more advanced methods such as isotopically labelled experiments. Finally, we believe that these CZ support effects explain why PdÀ Pt/Al 2 O 3 exhibits a higher degree of metallic Pd and subsequent increased CH 4 oxidation activity. The increased mobility of lattice oxygen of the CZ support allows either the easy reoxidation of metallic Pd to PdO or hinders the reduction of PdO to metallic Pd in the first place when compared to a support such as Al 2 O 3 .

Effect of pulses: O 2 pulses.
A single experiment was carried out at 315°C wherein H 2 was shut-down completely and O 2 was pulsed over the catalyst. Under these dry conditions, CH 4 became a reducing agent for 30 s. Figure 10 shows the longterm pulse behavior of the PdÀ Pt/CZ catalyst during these pulses. Under O 2 pulses, CH 4 oxidation was stabilized between 85 % and 30 %. While these bounds remained constant through the pulses, sharper activity changes were observed as O 2 turned on and off -catalytic activity decreased faster during the reductive CH 4 pulse and rose higher when gaseous O 2 returned as the experiment progressed. This activity was balanced out by a sharp increase in activity -similar to the spike seen under wet H 2 /O 2 pulsing at 400°C ( Figure S9b) -that decreased as time went on. This experiment suggests that the strength of H 2 as a reductive agent plays a role at least in the low temperature range in the improvements seen during H 2 /O 2 pulsing experiments, but that some benefits can be obtained from weaker reducing agents or simply shutting off O 2 periodically.

Conclusions
While in the lively debate concerning the active phase of Pd during methane oxidation a consensus has formed around PdO as the active phase under typical lean conditions, recent research has suggested that some amounts of metallic Pd might possibly boost catalytic activity. Our research joins the latter corpus of work and suggests a process controlled increase in methane oxidation activity. In situ pulses of H 2 reversed catalytic deactivation of bimetallic PdÀ Pt methane oxidation catalysts and maintained high activity over several hours that persisted into future experiments. The presence of H 2 O resulted in a much higher methane oxidation activity during the rich pulse while methane activity dropped to zero under dry conditions. XAS revealed exclusively PdO on fresh samples, while a combination of metallic Pd and PdO was found for samples after pulsing. This suggests that a metastable PdOÀ Pd structure is the predominant cause behind the increased activity of pulsed catalysts, either due to an enhanced activity via the Mars-van-Krevelen mechanism or because the formed PdÀ PdO species is less prone to water poisoning. Moreover, pulse-induced removal of surface hydroxyl species and the formation of lattice oxygen vacancies contribute to the increased activity. PdÀ Pt/Al 2 O 3 had a higher concentration of alloyed PdÀ Pt and a higher percentage of metallic Pd after pulsing, resulting in a larger shift in light-off temperature compared to pulsed PdÀ Pt/CZ samples.
The effects of H 2 were also dependent upon the gas environment of the catalyst. In the presence of constant O 2 flow, pulses of additional H 2 reacted over the Pd to form H 2 O, resulting in a high temperature region at the front of the catalyst. Under dry conditions, the inhibition due to the H 2 O negated the improvements from the increased temperature. Under wet conditions modest improvements were observed as the catalyst was already saturated. Under steady-state conditions, however, the raised catalytic temperature promoted the activity, resulting in light-off curves significantly shifted to lower temperatures.
Experiments varying the pulse durations of H 2 /O 2 switching revealed that the beneficial effects could still be obtained with shorter reductive pulses and longer oxidative periods. An experiment with O 2 on/off pulsing showed that CH 4 could stabilize catalytic activity to a lesser degree than H 2 pulsing accomplished. Future experiments concerning the strength of different reductive agents as well as the duration and frequency of reductive pulses could result in an optimized method for maintaining high methane oxidation under wet conditions at low operating temperatures, which is a key step in making natural gas processes sustainable and competitive.

Experimental Section
Powder catalyst samples were prepared by incipient wetness impregnation of Al 2 O 3 or CeO 2 -ZrO 2 -Y 2 O 3 -La 2 O 3 using tetraammineplatinum(II)nitrate and tetraamminepalladium(II)nitrate as precursors. After coating cordierite substrates with the received powder, the monolith samples were tested under model exhaust gas conditions in a quartz glass tubular reactor. For spatially resolved experiments the setup could be equipped with a movable thin thermocouple or a thin capillary connected to a mass spectrometer (HPR-20, Hiden Analytical) placed within one of the monolith's middle channels. For all experiments, also during spatially resolved experiments, the end-of-pipe gas composition was analyzed with an MG2030 FTIR (MKS). TEM data were obtained at the LEM (KIT, Germany) using an FEI OSIRIS ChemiSTEM microscope and Pd Kedge XAS data were obtained in transmission mode at the ROCK beamline (SOLEIL, France). Comprehensive experimental details are provided in the Supporting Information.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57