J. Phys. Chem. B 2005, 109, 2331-2337
2331
Temperature Dependence of the Kinetics for the Complete Oxidation of Methane on Palladium and Palladium Oxide† Guanghui Zhu,‡ Jinyi Han,‡ Dmitri Yu. Zemlyanov,§ and Fabio H. Ribeiro*,‡ School of Chemical Engineering, Purdue UniVersity, 480 Stadium Mall DriVe, West Lafayette, Indiana 47907-2100, and Materials and Surface Science Institute, UniVersity of Limerick, Limerick, Ireland ReceiVed: March 14, 2004; In Final Form: June 15, 2004
The kinetics for the complete combustion of methane was studied on a Pd foil in the regions where the oxide and then the metal were the bulk stable phases. The use of a model catalyst allowed the kinetics to be studied at higher temperatures than are possible on supported catalysts since heat and mass transport limitations could be avoided for this nonporous model catalyst. For all reaction conditions, CH4 and O2 reaction orders remained the same at about 0.7 and 0, respectively. With PdO as the stable phase, the water reaction order increased from -1 to 0 and the apparent activation energy (Ea) decreased from 125 to 30 kJ mol-1 as the reaction temperature increased from 600 to 880 K. We propose that as the temperature is increased water desorbs from the sites responsible for combustion and as a result water inhibition and Ea decrease. To investigate the rate of reaction on Pd versus PdO, the rates were measured around the Pd-PdO transition temperature. The turnover rate decreased from 3.0 s-1 to 0.3 s-1 at the transition temperature (907 K with 1.5 Torr O2 and 0.30 Torr CH4) when PdO decomposed to Pd metal, showing that PdO was more active than Pd metal for methane oxidation at this temperature. The reaction orders for Pd metal in the range of 933-1003 K were 0.7, 0, and 0 for methane, water, and O2, respectively, with an apparent activation energy of 125 kJ mol-1. Thus, the turnover rate and Ea changes suggest that the reaction mechanism for methane oxidation on Pd is different from the one on PdO.
1. Introduction Palladium is the most active catalyst for methane combustion.1 Palladium catalysts for methane combustion present an unusual situation in that the thermodynamically stable bulk phase can be either Pd metal or Pd oxide (PdO) depending on the oxygen partial pressure and the temperature. In air, at atmospheric pressure, the stable phase is PdO at a temperature below 1073 K and metallic Pd above it.2 Knowledge of the kinetics on PdO and on Pd is important, for example, in a catalytic combustor where the temperature of operation will span both phases. It is also important in automotive applications where the three-way catalyst will be exposed to oxidizing and reducing conditions. The kinetics at the temperatures where the metal is the stable phase cannot be studied easily with porous supported catalysts since at these conditions the limitations on the transport of heat and mass will mask the kinetics. To study the kinetics at high temperatures we used a nonporous Pd foil as the catalyst since in this case the only transport limitation is on external mass transport, which can be avoided by choosing appropriate reaction conditions. Another advantage of a model system is that the catalyst can be easily examined by surface analytical techniques and yet reactions can be carried out at the same conditions used to test supported catalysts. By comparing the published literature,3-16 it is concluded that the apparent kinetics for methane oxidation changes with the reaction temperature and the thermodynamically stable phase †
Part of the special issue “Michel Boudart Festschrift”. * To whom correspondence should be addressed. E-mail: Fabio@ purdue.edu. Phone: 765-494-7799. Fax: 765-494-0805. ‡ Purdue University. § University of Limerick.
of Pd. We will show how the reaction kinetics at high temperatures, including turnover rate, reaction orders, and activation energies change with temperature. The kinetics on PdO at around 600 K is now well established but the kinetics at higher temperatures is more difficult to find in the literature. We studied the kinetics at temperatures slightly below and above the PdO decomposition temperature so that we could compare the rates on PdO and Pd directly. Measuring the surface area at these conditions proved to be very important as the surface area changes significantly during this transition. There is still debate on which state of palladium (Pd or PdO) is more active for methane oxidation at high-temperature conditions. Lyubovsky et al.14 observed that Pd/R-Al2O3 after in situ hydrogen reduction was more active than oxidized Pd/R-Al2O3 for methane oxidation. Burch et al.17 and Farrauto et al.18 reported that the metal phase was inactive. We found in our study that the turnover rate difference between PdO and Pd depends on the temperature. For example, at 907 K the turnover rate on PdO is about 10 times higher than on Pd. These data are important for understanding the catalyst activity hysteresis18-20 and unstable reactor performance21,22 which occur due to PdO decomposition and reformation. The reaction kinetics was studied at temperatures below 600 K previously3,5 and the reaction order was 0.6-1 for methane, 0 for oxygen, -1 for water, and 0 for CO2. The magnitude of water inhibition on methane oxidation was found to be a function of temperature. At 553 K, water reaction order was about -1;3 above 723 K, the water reaction order was close to 0.4 A strong inhibition effect from CO2 was observed only at high CO2 concentration,3,4 but when both water and CO2 were in the stream, the CO2 effect was negligible and the inhibition was due to water only.4
10.1021/jp0488665 CCC: $30.25 © 2005 American Chemical Society Published on Web 07/27/2004
2332 J. Phys. Chem. B, Vol. 109, No. 6, 2005 The activation energy with values in the range of 25-184 kJ mol-1 was reported for methane oxidation at temperatures where PdO was the thermodynamically stable phase under reaction conditions.3,5,10,13-15 At temperatures below 650 K, the activation energy on PdO was reported to be 125-184 kJ mol-1 when the water inhibition effect was accounted for3,5,10 and 80126 kJ mol-1 when the water inhibition effect was not considered15,16 although these latter values are not activation energies as the rate constants are effectively a function of water concentration (k/[H2O]). Lower activation energies, with values in the range of 24-45 kJ mol-1, were obtained at higher temperatures with PdO still being the thermodynamically stable phase on catalysts of Pd supported on TiO2, γ-Al2O3, R-Al2O3, ThO2, and γ-Al2O3 + SiO2.16 The temperatures where the lower activation energy was observed started between 654 and 820 K,6,16 depending on the type of support and the composition of the methane-oxygen mixture.6,16 Lyubovsky et al.14 reported an activation energy of 74-80 kJ mol-1 in the temperature range of 600-800 K for PdO and about 157-200 kJ mol-1 when Pd metal was the thermodynamically stable phase.13,14 We will show that the apparent reaction orders and activation energy depend on the temperature range when PdO is the bulk stable phase although the reaction mechanism seems to be the same. At even higher temperatures where PdO decomposes to Pd, the reaction kinetics on Pd is different from the one on PdO, suggesting that the reaction mechanism is different. 2. Experimental Methods The experimental setup consisted of a reaction cell that could be operated up to atmospheric pressure and an ultrahigh vacuum (UHV) surface analysis chamber. The reaction cell could be operated as a continuous-stirred tank reactor (CSTR) or a batch reactor. Rates were measured in CSTR mode at high temperature and in batch mode at low temperature. Two gas circulation pumps (Metal Bellows, MB-21) were employed to control the reactant flow rate and to avoid mass transfer limitations. A valve was employed to adjust the flow rate around the catalyst to avoid excessive cooling of the sample. The maximum flow rate was 4600 cm3 min-1. An infrared heater was mounted outside the reactor with a line of sight, through the glass window, to the catalyst. For low-temperature operation (623 K) the infrared lamp was employed to assist resistive heating. The UHV chamber was equipped with an AES, an XPS, a mass spectrometer, and a sputtering gun. The catalyst could be transferred between the reactor and the UHV chamber without exposure to air. Details about this system were described in a previous paper.5 The palladium catalyst was a polycrystalline foil with 0.1 mm thickness and surface area of about 0.5 cm2 (Alfa Aesar, 99.9%). It was spot-welded to stainless steel power pins. The thermocouple wires for measurement and control were spotwelded on the back of the foil. Temperature control was achieved with a temperature controller (Eurotherm model 2408) and TCR power supply (Electronic Measurements Inc.). The cleaning procedure consisted of carrying out the methane oxidation at 773 K for 100 min with a reactant mixture of 4.5 Torr CH4, 18 Torr O2, and inert gases (He, N2) balance to atmospheric pressure, followed by sputtering with 2.0 kV Ar+ and annealing at 873 K for 1 min in vacuum. The surface was regarded as clean when sulfur, phosphorus, and silicon species could not be detected on the surface by XPS. Before each experiment, the sample was sputtered by 2.0 kV Ar+ and then annealed at 873 K for 1 min.
Zhu et al.
Figure 1. Methane conversion as a function of rate of reactant circulation. Reaction conditions: 863 K, 1.5 Torr O2, 0.30 Torr CH4, and inert gas (N2, He) balance to atmospheric pressure.
The reaction temperature must be raised above the PdO decomposition temperature to study the reaction kinetics of methane oxidation on palladium metal. At an oxygen pressure of 30 Torr, the PdO decomposition temperature is higher than 1000 K. Considering possible mass transfer limitations and the equipment tolerance to high temperatures, the oxygen pressure in this study was lowered to less than 3 Torr so that the metal phase could be studied at a temperature lower than 930 K. Except where noted, the turnover rate was calculated using the geometric Pd metal surface area assuming an average Pd surface atom density for a polycrystalline foil of 1.27 × 1015 atoms cm-2.23 This assumption was not critical for measurement of reaction orders and activation energy because the surface area did not change during reaction, but it was critical for the calculation of turnover rates. The PdO surface area was measured by a surface exchange experiment with labeled oxygen (18O isotope), which was done by exposing the oxidized foil to 5 Torr 18O2 at 598 K for 12 s. These conditions were designed based on results from Au-Yeung et al.24 to ensure that the exchange between the 16O in the PdO and the 18O2 isotope occurs mostly at the surface, without appreciable diffusion into the bulk. The reference point for oxygen coverage was made by assuming that a foil exposed to O2 at room temperature will form an oxygen layer with 0.25 ML coverage at saturation. This coverage was based on the wellestablished coverage for a Pd(111) single crystal25,26 and the fact that a foil is composed of mostly (111) planes. The oxygen exchange proved to be an effective method to measure PdO surface area.5,27 When Pd metal was the stable phase, the geometric surface area of Pd metal foil was employed as the surface area under reaction condition. 3. Results 3.1. System Test. A stainless steel foil with the same area as the palladium foil was used to test background activity. The turnover rate was less than 0.15 s-1 and thus not substantial as compared to the turnover rate of 9.3 s-1 on a Pd foil at the same condition (6 Torr O2, 1.5 Torr CH4, and 1023 K). The condition tested (high temperature) was the one where background reactions would be most significant. Thus, we expect that the background contribution to the rate on a Pd foil was not significant. We have also tested if the rate measurements were free of mass and heat transfer limitations. The palladium foil was nonporous, thus there were no internal mass and heat transfer limitations. The temperature of the foil was kept constant by a temperature controller based on the feedback from a thermocouple spot-welded to the foil surface; the surface temperature could therefore be controlled directly. Thus, the only concern with transport limitations was external mass transfer. Figure 1
Methane Oxidation on Palladium Foil
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2333
TABLE 1: Comparison of Reaction Orders and Activation Energy at Different Temperatures and Chemical States of Pd Catalysts reaction order
bulk state of foil
CH4
H2O
O2
temperature range/K
activation energy/ kJ mol-1
PdOa PdO Pd
0.7 0.6 0.7
-0.9 0 -0.1
0.2 0 -0.1
558-623 783-873 933-1003
125 30 125
a
Data from Monteiro et al.5
Figure 3. Reaction order dependence for Pd foil at 973 K on CH4 (0.28-0.88 Torr CH4, 2.3 Torr O2, 0.68-0.74 Torr H2O), O2 (2.3-4.6 Torr O2, 0.42 Torr CH4, 0.75 Torr H2O), and H2O (0.05-0.21 Torr H2O, 0.44 Torr CH4, 2.3 Torr O2). Balance to atmospheric pressure was completed with N2 and He. Turnover rate was calculated based on the geometric Pd metal surface area assuming an average Pd surface atom density of 1.27 × 1015 atoms cm-2.
Figure 2. Reaction order dependence for Pd foil at 863 K on CH4 (0.06-0.14 Torr CH4, 0.53-0.58 Torr O2, 1.3 Torr H2O), O2 (0.651.32 Torr O2, 0.06-0.066 Torr CH4, 1.3 Torr H2O), and H2O (0.181.31 Torr H2O, 0.06 Torr CH4, 0.58 Torr O2). Balance to atmospheric pressure was completed with N2 and He. The turnover rate was calculated based on the geometric Pd metal surface area assuming an average Pd surface atom density of 1.27 × 1015 atoms cm-2.
shows the methane conversion as a function of gas circulation rate. Under test conditions, the methane conversion stayed constant when the circulation rate was raised above 3800 cm3 min-1. Thus, when the circulation rate was above this value, the reaction was free of external mass transfer limitations. The circulation rate was kept higher than 3800 cm3 min-1 in our experiments. 3.2. Reaction Orders and Activation Energy. Table 1 summarizes the reaction orders and activation energies obtained in this study under fuel lean conditions on both the oxide phase and the metal phase. The data on reaction orders and activation energies are shown in Figures 2-4. The activation energies in Figure 4 were obtained at the temperature ranges where the indicated phases were stable. 3.3. Characterization. The chemical state of the foil surface was examined before and after reaction by X-ray photoelectron spectroscopy (XPS). The clean metallic Pd is characterized by the Pd 3d5/2 peak at 335.0 eV (Figure 5), which is in agreement with the one in the literature.28 After measurement of reaction orders at 863 K and activation energy in the range of 783-873 K, palladium oxide was observed on the foil surface. The characteristic binding energy on the Pd 3d5/2 line observed at 336.9 eV, which corresponds to PdO,28 is shown in Figure 5. Note that if a mixture of Pd and PdO would be present, the Pd 3d5/2 peak corresponding to the metal or to the oxide would have shown broadening. An example of the XPS Pd 3d core level when Pd metal was the thermodynamic stable phase is shown in Figure 5 after reaction at 903 K with 0.76 Torr O2, 0.15 Torr CH4, and inert gases (N2, He) balance to atmospheric pressure; the binding energy of Pd 3d5/2 was 335.0 eV. Analyzing the metallic sample after reaction is difficult because cooling under reaction conditions will cause sample oxidation
Figure 4. Arrhenius plot for the combustion of methane over Pd foil (open square, 0.3 Torr CH4; open circle, 0.46 Torr CH4). Balance to atmospheric pressure was completed with O2, H2O, N2, and He. Turnover rate was calculated based on the geometric Pd metal surface area assuming an average Pd surface atom density of 1.27 × 1015 atoms cm-2.
and cooling after evacuation will decompose any oxide formed during reaction. Because the sample cooled in a few seconds in the reaction mixture once the power was turned off, we adopted this procedure. Compared with the Pd 3d5/2 peak obtained on a clean Pd foil, a slight broadening toward higher binding energy for the Pd 3d5/2 peak was obtained on the foil after the 903 K experiment, which indicates the presence of oxide on the surface but not significant bulk oxidation. To predict the temperature at which PdO decomposed to Pd, we used the expression obtained by Warner,2 log P ) 31.905 - 6.29 log T - 14 510/T ( 0.006, where P is O2 pressure (atm) and T is temperature (K). The phase predicted by this expression agreed with the one measured by XPS. 3.4. Rates on Pd and PdO. With the objective of deciding which species (Pd or PdO) is the most active, kinetic data including turnover rates, reaction orders, and activation energies were obtained on palladium oxide and palladium metal at temperatures around the PdO decomposition temperature. Figure 6 shows the nominal turnover rates under 1.5 Torr O2 and 0.30
2334 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Zhu et al. TABLE 2: Comparison of Activation Energy and Turnover Rates on Palladium Oxide Catalysts catalyst Pd foil Pd foil Pd/Si-Al2O3 Pd/Al2O3 Pd/ZrO2 Pd/ZrO2 Pd/ZrO2
Ea/kJ mol-1
TORa/s-1
ref
125 170-184 150 170 185 N/A
3.8 5.3 0.1b 0.07-0.16b 0.1-0.7b 0.5-3.0b 0.3b
this work 5 3 3 3 10 44,45
a TOR calculated at 598 K, 16 Torr CH4, 1 Torr H2O, and N2 balance to 800 Torr. Reaction orders were assumed to be 1 for CH4, 0 for O2, and -1 for H2O. b For the plug flow reactor, partial pressures for reactants and products are the average of the values of the inlet and exit concentration.
Figure 5. Comparison of the XPS Pd 3d core level scan for clean metal foil, foil after lean reaction at 903 K, and foil after lean reaction at 863 K. Reaction conditions: 0.76 Torr O2, 0.15 Torr CH4, 0 Torr H2O, and inert gases (N2, He) balance to atmospheric pressure.
the use of the activation energy and reaction orders on PdO obtained in this study, the turnover rate was calculated as 3.0 s-1 at 907 K with 1.5 Torr O2 and 0.30 Torr CH4. The Pd metal surface area after reaction with 0.76 Torr O2, 0.15 Torr CH4 at 923 K was measured by the 18O exchange method and found to be 0.9 times the area of the clean foil. Since the value of the surface area obtained is close to the one for the Pd metal surface area of a clean surface, it is going to be assumed that when Pd is the stable phase then its surface area is the same as the surface area of a clean Pd metal. On the basis of this assumption, a turnover rate of 1.2 s-1 on Pd was obtained at 973 K with 2.3 Torr O2 and 0.46 Torr CH4. With the use of the activation energy and reaction orders on Pd obtained in this study, the turnover rate was calculated to be 0.3 s-1 at 907 K, 1.5 Torr O2, and 0.3 Torr CH4. Thus, the turnover rate (per Pd atom) at 907 K is 10 times higher on the oxide than on the metal. 4. Discussion
Figure 6. Arrhenius plot for palladium in the metal and oxide phases. Reaction conditions: 1.5 Torr O2, 0.30 Torr CH4, 0 Torr H2O, and inert gases (N2, He) balance to atmospheric pressure. Rates were not corrected for surface area changes; it was assumed that the area was constant with 1.27 × 1015 sites cm-2.
Torr CH4 on both palladium oxide and palladium metal at temperatures around the PdO decomposition temperature (907 K). The rates are called nominal because we have calculated the rates in the figure using the same surface area for both phases. As we will show later on, the surface area changes substantially when the phase transition occurs and it will be important to calculate the rate per unit of surface area also. The rates on Pd metal were measured at 973 K with 2.3 Torr O2 and 0.46 Torr CH4. The higher methane concentration for the experiment with Pd metal as the active phase was used to increase the conversion and thus facilitate CO2 detection. The O2 pressure was also raised to keep the ratio of O2 to CH4 constant. For rate comparison, all the turnover rates on Pd metal were corrected to 1.5 Torr O2 and 0.30 Torr CH4 using the reaction orders obtained in this study: 0.7 for CH4, 0 for O2 and H2O. The difference in the nominal turnover rate at the temperature where the transition Pd to PdO occurs is about a factor of 80 higher on PdO. The rates will be calculated per unit of surface area for a comparison of the intrinsic rate. The surface area was measured by 18O2 isotope exchange for the metal and the oxide.27 A PdO surface area 8 times higher than the one on a clean foil was obtained after reaction with 0.76 Torr O2, 0.15 Torr CH4 at 843 K. The turnover rate corresponding to 0.76 Torr O2, 0.15 Torr CH4 at 843 K was 1.5 s-1 after surface area correction. With
Comparison of turnover rates on PdO with the rates in the literature is presented in Table 2. The value for the turnover rates on Pd foil is the highest among all palladium catalysts reported. The reported lower rates can be due to interference from the support and the presence of impurities. Because the kinetics on the oxidized foil has no interference from the support and the absence of impurities could be verified directly by XPS, this rate can be used as a benchmark. The availability in the literature for rates with the foil in the metallic state is much smaller. The turnover rate estimated from the data in Lyubovsky et al.14 was in the range of 0.016-0.16 s-1 at 907 K with 0.30 Torr CH4 in air, which compares well with our value of 0.3 s-1. In addition to the change in rates with the state of Pd (PdO or Pd), it is evident from Table 1 that the orders and apparent activation energies also change with temperature and the state of Pd. We will discuss in turn the kinetics on PdO in the range of 600-873 K and then on Pd in the range of 925-1000 K. 4.1. Kinetics on PdO. The reaction order was 0.6-0.7 for methane, about 0 for oxygen in the range of 598-973 K and changed for water from about -1 to 0 as the temperature increased. The zero order in O2 indicates that this reaction occurs over a completely oxidized surface. The positive order in methane indicates that methane has to compete for sites on the surface, and the negative order in water indicates that water can effectively block these sites. As the temperature is increased, water inhibition disappears. Although kinetic models for the combustion of methane have been proposed,10,29,30 there is no model that can explain all the facts satisfactorily. The following sequence of steps can explain
Methane Oxidation on Palladium Foil
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2335
in the simplest way the facts we observed here. It is designed only as a guideline for the actual mechanism.
CH4 + * / CH4*
(1)
CH4* + [ f .... (rds)
(2)
H2O + * / H2O*
(3)
In this mechanism, * represents a site on the PdO surface and [ represents an oxygen vacancy. Methane first adsorbs on the surface of PdO (step 1) and the C-H bond activation requires an oxygen vacancy (step 2). The breaking of the C-H bond (step 2) is the rate-determining step. Combustion of CH4 and CD4 gave a normal isotope effect (KH/KD > 1, K is the reaction rate constant), suggesting that the rate-determining step in the catalytic sequence involves breaking a carbon-hydrogen bond.31 The rate of H-D exchange for the gas mixture of CH4/CD4/O2 was found to be only one-tenth of the combustion rate at 573 K,31 indicating that the C-H bond activation could be treated as an irreversible step. Isotope studies31-33 showed that during methane oxidation, a fraction of the O in the products came from oxygen stored in the bulk of the PdO while the other fraction of oxygen came from gas phase O2. It is not known whether gas phase O2 transformed to lattice O before it was consumed by methane combustion. Thus, at least part of the oxygen (if not all) in the products came from lattice oxygen. Oxygen scrambling was observed when a 16O2/18O2 mixture was contacted with Pd16O/ZrO2 catalyst at 573 K24 but was not observed during reaction.31 This difference indicates that either methane oxidation consumed the adsorbed oxygen or the oxide surface was blocked by other species. Ciuparu et al.34 found that water could inhibit the oxygen exchange between the gas phase and the PdO surface, indicating that the slower exchange could be attributed to water-blocking exchange sites (step 3). The most abundant surface intermediate (MASI) in the mechanism is assigned to adsorbed water (H2O*). From steps 1-3 above the rate can be written as
r ) k2 K1 [L]
[CH4][[] 1 + K3[H2O]
(I)
where [L] is the concentration of active sites, [H2O*] + [*] ) [L]. The concentration of oxygen vacancies ([) will be regarded as constant (C). The concentration of vacancies on the surface is probably very small and is due to the fluctuations in the transport of oxygen from the surface to the bulk of the solid. It is thus a function of temperature but not of O2 pressure. The equilibrium constant for water K3 should decrease as temperature increases. At low temperature when K3[H2O] > 1, the rate expression can be simplified to the following expression
r)
[CH4] k2K1 C [L] K3 [H2O]
(II)
which shows a reaction order of 1 for CH4 and a -1 reaction order for H2O. At high-enough temperature K3[H2O] < 1 and expression I simplifies to
r ) k2 K1C [L] [CH4]
(III)
A decrease of activation energy for methane oxidation on PdO was reported from 75-95 kJ mol-1 to 24-45 kJ mol-1 as the temperature increased, and the transition temperatures were
in the range of 650-720 K.6,16 Those reports support the value for activation energy in this study of 30 kJ mol-1 for methane oxidation, obtained in the range of 783-873 K. The apparent activation energy for eq II is ∆H1 + Ea2 - ∆H3, while the apparent activation energy for eq III is ∆H1 + Ea2. As ∆H3 is negative, the activation energy is higher at low temperatures when water has an inhibition effect. We can then calculate an approximate value for the heat of adsorption of water (∆H3) of 90 kJ mol-1 by difference of the two apparent activation energies. We assumed that the constant C was not a function of temperature for this estimation. 4.2. Kinetics on Pd. Less is known about the kinetics on Pd metal. The reason is that this phase is stable only after decomposition from PdO, which occurs at high temperatures. The high temperatures involved imply that the rates on supported catalysts can be heat and mass transfer limited. The rates on Pd metal are, however, important in practice when these catalysts are used in catalytic combustors because the operation temperature will exceed the decomposition temperature of the oxide. One question of importance in determining the mechanism is what is the coverage of oxygen on the surface of Pd under reaction conditions? Since the reaction order in O2 is 0, the surface must be covered with oxygen under reaction conditions even though the bulk is metallic. We can estimate the coverage under reaction conditions based on the values for the heat of adsorption as a function of coverage obtained by Ertl and Rau35 on Pd(110). They reported the heat of O2 adsorption to be 80 kcal mol-1 when the coverage produced a LEED image with a (1 × 3) pattern, 77 kcal mol-1 for a (1 × 2) pattern, 62 kcal mol-1 for a c(2 × 4) pattern, and 48 kcal mol-1 for a c(2 × 6) pattern. However, they did not assign a corresponding coverage for each LEED pattern; we will attempt an assignment in what follows. The heat of adsorption at constant coverage was obtained by Ertl and Rau using the Clausius-Clapeyron equation assuming that the constancy of the LEED pattern implied a constancy of oxygen coverage as the temperatures were varied. A coverage of 0.23 ML was reported for the (1 × 3) LEED pattern36 and 0.48-0.5 ML was reported for the c(2 × 4) LEED pattern.36,37 Jo et al.38 reported that the (1 × 3) and (1 × 2) LEED patterns observed by Ertl and Rau35 corresponded to the streaky pattern they observed with coverage between (2 × 3)-1D and c(2 × 4). A coverage of 0.2 ML for the (2 × 3)-1D pattern and a coverage of 0.3 ML were reported for the streaky pattern.39 For Ni(110) which has the same crystal structure as Pd(110), a coverage of 0.33-0.35 ML was reported for the (2 × 1) structure.40,41 Considering that the same LEED pattern corresponds to the same saturation coverage of oxygen for single crystals with the same surface orientation, we assign the streaky (1 × 2) LEED pattern on Pd(110) to 0.3 ML. The c(2 × 6) LEED pattern was observed in our lab with coverage in the range of 0.6-0.7 ML. On Rh(110), the oxygen coverage for the structure of c(2 × 2n) (n ) 3, 4, and 5) was proposed to be (n - 1)/n.42 Considering that Rh(110) and Pd(110) are fcc metals, the c(2 × 6) surface on Pd(110) could correspond to the same oxygen coverage of 0.67 ML as on Rh(110). Thus, we made the following assignments for the heat of adsorption with coverage in the Ertl and Rau paper: 80 kcal mol-1 at 0.23 ML, 77 kcal mol-1 at 0.30 ML, 62 kcal mol-1 at 0.50 ML, and 48 kcal mol-1 at 0.67 ML. With these assignments for the coverage we can calculate from the Clausius-Clapeyron plots in the Ertl and Rau contribution four sets of values of coverage versus O2 pressure for a given temperature. With the help of the Freundlich equation, θ ) Const × (PO2)γ where θ is the equilibrium oxygen coverage and γ is a constant, one can then
2336 J. Phys. Chem. B, Vol. 109, No. 6, 2005
Figure 7. Freundlich isotherm from the data of Ertl and Rau.35
calculate the coverage as a function of O2 pressure (Figure 7). The equilibrium oxygen pressure for 1 ML coverage at 973 and 863 K were extrapolated to be 0.8 and 0.04 Torr. Thus, the surface of Pd(110) should be covered by adsorbed oxygen at 973 and 863 K when exposed to 2.3 Torr O2 while the bulk phase remains in the metallic state. Thus, since the surface is covered with oxygen, it is easy to understand that the reaction order in O2 is 0. The reaction mechanism, however, is not similar to the one on PdO since the turnover rate and apparent activation energy are substantially different on these two surfaces. 4.3. Comparison of Turnover Rates in the Metal and Oxide States. Another important question is what is the rate on the metal compared to the rate on the oxide when the catalyst passes through the decomposition temperature? At 907 K, the turnover rate decreased from 3.0 to 0.3 s-1 in 1.5 Torr O2 and 0.30 Torr CH4 when PdO decomposed to Pd metal, indicating that the bulk phase PdO is more active than Pd metal for methane oxidation at this temperature. Because the activation energy is different on each phase, one can calculate the temperature where the turnover rate at the transition temperature will be the same; the approximate temperature is 1105 K. Thus, at temperatures below 1105 K, PdO is more active than Pd metal for methane oxidation and vice versa at temperatures above 1105 K. McCarty20 studied methane combustion on Pd/γ-Al2O3 with an almost fixed methane pressure (2.3-3.8 Torr) but different oxygen pressures (7.6-315 Torr). McCarty found that methane conversion decreased 60% when PdO decomposed to Pd at 930 K, while the conversion did not decrease when decomposition occurred at temperatures above 1100 K, indicating the turnover rate on Pd metal surpassed the one on PdO at temperatures above 1100 K. This result agrees with our projection. One apparent disagreement is the data of Lyubovsky et al.14 They observed a methane conversion increase when the oxygen pressure in the reaction mixture was reduced and PdO decomposed to Pd. In their experiment, PdO was forced to decompose to Pd by decreasing the oxygen pressure but in doing so the ratio of O2 to CH4 decreased to a value equal or lower than the stoichiometric ratio. If the conditions are changed from lean to rich, we have reported that the catalyst can have a rate 18 times higher at fuel rich conditions on PdO.27 Thus, the rate difference may have been influenced by a change in the O2 to CH4 ratio instead of the change of Pd to PdO. Another interesting observation that can be made based on the nominal rates presented in Figure 6 is that under the adiabatic conditions of catalytic combustion on a gas turbine the catalyst maintains its temperature at the Pd-PdO decomposition temperature.43 The explanation for this behavior can be made based on a decrease by a factor of about 80 in the nominal rate when the transition from PdO to Pd occurs. It suggests that on a
Zhu et al. supported catalyst the rate per unit of volume of catalyst will also decrease substantially since the particle size on combustion catalysts is large enough to imply that the chemistry on it will be similar to the one on a foil. During operation on a typical light off experiment, the temperature in the catalyst increases faster than in the gas phase due to a higher heat generation than heat loss. When the transition temperature is reached and the transformation PdO-Pd is triggered, the rate will decrease and thus the heat production will be smaller. The lower heat generation will not be able to sustain the catalyst temperature and the temperature will decrease below the transition point. The sample will be oxidized and the rate, and consequently the temperature, will increase again causing the catalyst to cross the transition point, thus repeating the cycle. In practice, these cycles are not observed because they probably occur on small domains with a different time period, and thus on average a constant temperature is observed. It is interesting to observe that the change in the overall rate is caused by a change in surface area (factor of 8) and by a less active phase (factor of 10). The difference in activation energy for the reaction on Pd and PdO also explains the observation that a Pd catalyst after reduction will appear inactive for the combustion of methane at temperatures around 600 K until the sample oxidizes to PdO, as first observed by Farrauto et al.18 and Burch et al.17 5. Conclusion The absence of porosity on a Pd foil model catalyst allowed the measurement of kinetics at high temperature without the interference of heat and mass limitations. Under reaction conditions where PdO was the stable phase, the water reaction order changed from -1 to 0 and the apparent activation energy decreased from 125 to 30 kJ mol-1 as the temperature increased. These results suggest that at high temperatures the amount of water adsorbed becomes small and thus does not inhibit the reaction. The difference in the apparent activation energy between low and high temperature corresponds to the heat of adsorption of water. At even higher temperatures, PdO decomposes to Pd. We estimated that the oxygen surface coverage when Pd is the stable bulk phase is one monolayer which explains the zero-order reaction in oxygen. The turnover rate decreases by a factor of about 10 during decomposition, indicating an intrinsically lower reaction rate on Pd. It is important to stress that the relative rates of combustion on PdO compared to Pd are dependent on the reaction temperature. In addition, the surface area decreases by a factor of about 8 during the PdO to Pd transition which shows that the rates per unit of volume of a supported catalyst will decrease substantially during decomposition. This explains the observation that under industrial conditions the catalyst keeps its temperature at the transition point. The apparent activation energy is higher and the turnover rate is lower on Pd than on PdO suggesting that a different reaction mechanism is present for each phase. Acknowledgment. We gratefully acknowledge support from the Office of Basic Energy Sciences, Chemical Sciences, U.S. Department of Energy, Grant DE-FG02-03ER15408. We thank Professor W. N. Delgass for insightful discussions. References and Notes (1) Anderson, R. B.; Stein, K. C.; Feenan, J. J.; Hofer, L. J. E. Ind. Eng. Chem. 1961, 53, 809. (2) Warner, J. S. J. Electrochem. Soc. 1967, 114, 68. (3) Ribeiro, F. H.; Chow, M.; Dalla Betta, R. A. J. Catal. 1994, 146, 537. (4) Burch, R.; Urbano, F. J.; Loader, P. K. Appl. Catal., A 1995, 123, 173.
Methane Oxidation on Palladium Foil (5) Monteiro, R. S.; Zemlyanov, D.; Storey, J. M.; Ribeiro, F. H. J. Catal. 2001, 199, 291. (6) Hayes, R. E.; Kolaczkowski, S. T.; Li, P. K. C.; Awdry, S. Chem. Eng. Sci. 2001, 56, 4815. (7) Cullis, C. F.; Nevell, T. G.; Trimm, D. L. J. Chem. Soc., Faraday Trans. 1 1972, 1, 1406. (8) Cullis, C. F.; Willatt, B. M. J. Catal. 1984, 86, 187. (9) Ahlstroem-Silversand, A. F.; Ulf Ingemar Odenbrand, C. Appl. Catal., A 1997, 153, 157. (10) Fujimoto, K.-i.; Ribeiro, F. H.; Avalos-Borja, M.; Iglesia, E. J. Catal. 1998, 179, 431. (11) van Giezen, J. C.; van den Berg, F. R.; Kleinen, J. L.; van Dillen, A. J.; Geus, J. W. Catal. Today 1999, 47, 287. (12) Ciuparu, D.; Katsikis, N.; Pfefferle, L. Appl. Catal., A 2000, 216, 209. (13) Lyubovsky, M.; Pfefferle, L. Appl. Catal., A 1998, 173, 107. (14) Lyubovsky, M.; Pfefferle, L. Catal. Today 1999, 47, 29. (15) Hicks, R. F.; Young, M. L.; Lee, R. G.; Qi, H. J. Catal. 1990, 122, 280. (16) Cullis, C. F.; Willatt, B. M. J. Catal. 1983, 83, 267. (17) Burch, R.; Urbano, F. J. Appl. Catal., A 1995, 124, 121. (18) Farrauto, R. J.; Hobson, M. C.; Kennelly, T.; Waterman, E. M. Appl. Catal., A 1992, 81, 227. (19) Lyubovsky, M.; Pfefferle, L.; Datye, A.; Bravo, J.; Nelson, T. J. Catal. 1999, 187, 275. (20) McCarty, J. G. Catal. Today 1995, 26, 283. (21) Lyubovsky, M.; Smith, L. L.; Castaldi, M.; Karim, H.; Nentwick, B.; Etemad, S.; LaPierre, R.; Pfefferle, W. C. Catal. Today 2003, 83, 71. (22) Furuya, T.; Sasaki, K.; Hanakata, Y.; Ohhashi, T.; Yamada, M.; Tsuchiya, T.; Furuse, Y. Catal. Today 1995, 26, 345. (23) Anderson, J. R. Structure of Metallic Catalysts; Academic Press: New York, 1975. (24) Au-Yeung, J.; Bell, A. T.; Iglesia, E. J. Catal. 1999, 185, 213. (25) Zheng, G.; Altman, E. I. Surf. Sci. 2000, 462, 151. (26) Conrad, H.; Ertl, G.; Kueppers, J.; Latta, E. E. Surf. Sci. 1977, 65, 245.
J. Phys. Chem. B, Vol. 109, No. 6, 2005 2337 (27) Monteiro, R. S.; Zemlyanov, D.; Storey, J. M.; Ribeiro, F. H. J. Catal. 2001, 201, 37. (28) Wagner, C. D.; Naumkin, A. V.; Kraut-Vass, A.; Allison, J. W.; Powell, C. J.; Rumble, J. R. J. NIST X-ray Photoelectron Spectroscopy Database. In NIST Standard Reference Database, Version 3.4; NIST: Gaithersburg, 2003; http://srdata.nist.gov/xps. (29) Ciuparu, D.; Lyubovsky, M. R.; Altman, E.; Pfefferle, L. D.; Datye, A. Catal. ReV.sSci. Eng. 2002, 44, 593. (30) Burch, R.; Hayes, M. J. J. Mol. Catal. A 1995, 100, 13. (31) Au-Yeung, J.; Chen, K.; Bell, A. T.; Iglesia, E. J. Catal. 1999, 188, 132. (32) Ciuparu, D.; Altman, E.; Pfefferle, L. J. Catal. 2001, 203, 64. (33) Mueller, C. A.; Maciejewski, M.; Koeppel, R. A.; Tschan, R.; Baiker, A. J. Phys. Chem. 1996, 100, 20006. (34) Ciuparu, D.; Pfefferle, L. Catal. Today 2002, 77, 167. (35) Ertl, G.; Rau, P. Surf. Sci. 1969, 15, 443. (36) Goschnick, J.; Wolf, M.; Grunze, M.; Unertl, W. N.; Block, J. H.; Loboda-Cackovic, J. Surf. Sci. 1986, 178, 831. (37) He, J. W.; Memmert, U.; Griffiths, K.; Norton, P. R. J. Chem. Phys. 1989, 90, 5082. (38) Jo, M.; Kuwahara, Y.; Onchi, M.; Nishijima, M. Chem. Phys. Lett. 1986, 131, 106. (39) Tanaka, H.; Yoshinobu, J.; Kawai, M. Surf. Sci. 1995, 327, L505. (40) Smeenk, R. G.; Tromp, R. M.; Saris, F. W. Surf. Sci. 1981, 107, 429. (41) Baro, A. M.; Binnig, G.; Rohrer, H.; Gerber, C.; Stoll, E.; Baratoff, A.; Salvan, F. Phys. ReV. Lett. 1984, 52, 1304. (42) Murray, P. W.; Leibsle, F. M.; Guo, Q.; Bowker, M.; Thornto, G. Phys. ReV. B 1993, 47, 12976. (43) Garten, R. L.; Betta, R. A. D.; Schlatter, J. C. Catalytic Combustion. In Handbook of Heterogeneous Catalysis; Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH: Weinheim, 1997; Vol. 4, p 1668. (44) Mueller, C. A.; Maciejewski, M.; Koeppel, R. A.; Baiker, A. J. Catal. 1997, 166, 36. (45) Muller, C. A.; Maciejewski, M.; Koeppel, R. A.; Baiker, A. Catal. Today 1999, 47, 245.
