Active Surfaces for CO Oxidation on Palladium in the Hyperactive

Nov 8, 2010 - The active surfaces at the hyperactive state for palladium were ..... As shown in Figure 4B, the gas phase CO2 for spectra 27 and 28 is ...
0 downloads 0 Views 1MB Size
Download PDF
pubs.acs.org/Langmuir © 2010 American Chemical Society

Active Surfaces for CO Oxidation on Palladium in the Hyperactive State Mingshu Chen,* Xin V. Wang, Lihua Zhang, Zhenyan Tang, and Huilin Wan State Key Laboratory of Physical Chemistry of Solid Surfaces, National Engineering Laboratory for Green Chemical Productions of Alcohols-Ethers-Esters, Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, Fujian, China Received August 6, 2010. Revised Manuscript Received October 15, 2010 Hyperactivity was previously observed for CO oxidation over palladium, rhodium, and platinum surfaces under oxygen-rich conditions, characterized by reaction rates 2-3 orders higher than those observed under stoichiometric reaction conditions [Chen et al. Surf. Sci. 2007, 601, 5326]. In the present study, the formation of large amounts of CO2 and the depletion of CO at the hyperactive state on both Pd(100) and polycrystalline Pd foil were evidenced by the infrared intensities of the gas phase CO2 and CO, respectively. The active surfaces at the hyperactive state for palladium were characterized using infrared reflection absorption spectroscopy (IRAS, 450-4000 cm-1) under the realistic catalytic reaction condition. Palladium oxide on a Pd(100) surface was reduced eventually by CO at 450 K, and also under CO oxidation conditions at 450 K. In situ IRAS combined with isotopic 18O2 revealed that the active surfaces for CO oxidation on Pd(100) and Pd foil are not a palladium oxide at the hyperactive state and under oxygen-rich reaction conditions. The results demonstrate that a chemisorbed oxygen-rich surface of Pd is the active surface corresponding to the hyperactivity for CO oxidation on Pd. In the hyperactive region, the CO2 formation rate is limited by the mass transfer of CO to the surface.

1. Introduction Platinum-group metals (Pd, Pt, Rh, etc.) are commonly used as supported catalysts in the chemical industry, for reactions such as hydrogenation, dehydrogenation, petroleum cracking, and so forth. Today, over half of the supply of palladium, and its congener platinum, goes into the three-way catalytic converters used in automobiles, which convert most of the harmful gases from auto exhaust streams (hydrocarbons, CO, and NOx) into less harmful substances (CO2, H2O, and N2). Current generation internal combustion engines work at a high air/fuel ratio, leading to an oxygenrich exhaust stream in the catalytic converter. Hence, for good catalytic converter performance it is essential that, under oxidizing conditions, Pt group metals in three-way converters remain in an active form and should not undergo irreversible change. CO oxidation on Pt-group metals is a well-studied catalytic reaction, and is generally accepted to follow a LangmuirHinshelwood reaction mechanism whereby both CO and oxygen species adsorb on a metallic surface and undergo a *Corresponding author. E-mail: [email protected]. Phone: 86-5922183723. Fax: 86-592-2181888.

(1) Langmuir, I. Trans. Faraday Soc. 1922, 17, 621. (2) Engel, T.; Ertl, G. Adv. Catal. 1979, 28, 1. (3) Somorjai, G. A.; Park, J. Y. Angew. Chem., Int. Ed. 2008, 47, 9212. (4) Goodman, D. W. Chem. Rev. 1995, 95, 523. (5) Campbell, C. T.; Ertl, G.; Kuipers, H.; Segner, J. J. Chem. Phys. 1980, 73, 5862. (6) Eischens, R. P.; Pliskin, W. A. Adv. Catal. 1957, 9, 662. (7) Berlowitz, P. J.; Peden, C. H. F.; Goodman, D. W. J. Phys. Chem. 1988, 92, 5213. (8) Rodriguez, J. A.; Goodman, D. W. Surf. Sci. Rep. 1991, 14, 1. (9) Somorjai, G. A.; McCrea, K. R. Adv. Catal. 2000, 45, 385. (10) Cant, N. W.; Angove, D. E. J. Catal. 1986, 97, 36. (11) Goodman, D. W.; Peden, C. H. F. J. Phys. Chem. 1986, 90, 4839. (12) Szanyi, J.; Goodman, D. W. J. Phys. Chem. 1994, 98, 2972. (13) Szanyi, J.; Goodman, D. W. J. Phys. Chem. 1994, 98, 2978. (14) Su, X. C.; Cremer, P. S.; Shen, Y. R.; Somorjai, G. A. J. Am. Chem. Soc. 1997, 119, 3994. (15) Xu, J. Z.; Yates, J. T. J. Chem. Phys. 1993, 99, 725. (16) Peden, C. H. F.; Goodman, D. W.; Weisel, M. D.; Hoffmann, F. M. Surf. Sci. 1991, 253, 44.

Langmuir 2010, 26(23), 18113–18118

bimolecular reaction.1-18 Previously, we had found that CO oxidation over Pt-group metals, including Pd, Rh, and Pt, exhibits a hyperactive state under oxygen-rich conditions with a rate at least 2-3 orders higher than those under nearstoichiometric reaction conditions (Figure S1, Supporting Information).18 In the hyperactive state, no CO adsorption was detected by infrared reflection absorption spectroscopy (IRAS) on a Pd(110) surface. The term “hyperactive state” was introduced previously, since under these reaction conditions we believed that the rate was limited by the mass transfer of CO to the surface. As discussed in our previous paper, a turnover frequency (TOF) of ∼40 000 (CO2 per second per surface Pd atom) was achieved in a small spot sample,18 the configuration of which improves the mass transfer. The reaction kinetics at realistic catalytic reaction conditions can be described as a low rate region where the reaction rate is determined by CO desorption from the metallic surface, and a hyperactive region where the surface adsorbed CO amount is very low and the mass transfer of the CO to the surface may limit the reaction rate. The hyperactive surface was proposed to be a surface phase that contains primarily chemisorbed atomic oxygen on the metallic Pd surface and a low coverage of CO based on the significant change of CO adsorption by IRAS.17,18 However, there is still a debate regarding the nature of the active surface for CO oxidation on Pt-group metal surfaces. Both metallic and oxide surfaces have been reported to play a key role.17-20 Frenken et al.19,21 stated that the oxide surfaces are (17) Gao, F.; Wang, Y.; Cai, Y.; Goodman, D. W. J. Phys. Chem. C. 2009, 113, 174. (18) Chen, M. S.; Cai, Y.; Yan, Z.; Gath, K. K.; Axnanda, S.; Goodman, D. W. Surf. Sci. 2007, 601, 5326. (19) Hendriksen, B. L. M.; Bobaru, S. C.; Frenken, J. W. M. Surf. Sci. 2004, 552, 229. (20) Zheng, G.; Altman, E. I. J. Phys. Chem. B 2002, 106, 1048. (21) Gustafson, J.; Westerstrom, R.; Balmes, O.; Resta, A.; Van Rijn, R.; Torrelles, X.; Herbschleb, C. T.; Frenken, J. W. M.; Lundgren, E. J. Phys. Chem. C. 2010, 114, 4580.

Published on Web 11/08/2010

DOI: 10.1021/la103140w

18113

Article

Chen et al.

significantly more active for CO oxidation, while Goodman et al.11,17,18,22 and Altmann et al.20 reported the opposite results. Gopinath et al. using molecular beam experiments demonstrated positive influence of the subsurface oxygen toward CO adsorption and oxidation to CO2 at 600-900 K on a Pd surface.23-25 On supported Pt catalysts, it was also found that partially oxidized Pt is significantly more active than a metallic Pt surface covered with CO.26 Additionally, the surface chemistry of oxygen interaction with Pd surfaces has been reported to be very complicated, mainly due to diffusion of oxygen into the subsurface and/or the formation of a bulk and metastable PdxOy depending on the experimental conditions.27-37 Furthermore, there is no oxygen poisoning observed with palladium in an auto converter, even under oxygen-rich conditions. Therefore, from a fundamental perspective, it is important to study in detail the active surfaces for Pt-group metals in oxygenrich conditions with in situ characterization techniques. In this study, we constructed a new IRAS system which enables spectroscopic measurements down to 450 cm-1, a range that covers the vibrational region of most metal oxides, by using a KBr window rather than a CaF2 one (Figure S2). Thus, IRAS measurements can provide a direct in situ probe of the surface phase change under various reaction conditions, including those of realistic CO catalytic oxidation conditions. Our results reveal that the active surface is not a palladium oxide, but a chemisorbed oxygen-rich palladium surface that exhibits hyperactivity for CO oxidation on Pd. Additionally, changes in the gas phase compositions of CO and CO2 at the hyperactive state were also monitored by the infrared absorption intensity of gas phase CO and CO2. These measurements reveal that CO is depleted at the hyperactive state.

2. Experimental Section Experiments were carried out in an ultrahigh vacuum (UHV) chamber equipped with a high-resolution-electron energy loss spectroscopy (LK-5000) and basic surface science techniques. A contiguous in situ IRAS reaction cell, capable of pressures from UHV to 1 atm and sample temperatures of T = 80-1100 K, was mounted onto the UHV chamber. A Pd(100) sample purchased from Princeton Scientific Corporation was used. The sample was cleaned by repeated Arþ-sputtering following a subsequent anneal at 1100 K in UHV. The cleanness was confirmed by Auger spectroscopy (AES) and low energy electron diffraction (LEED). The clean sample was then transferred into the reaction cell under UHV for in situ IRAS measurements. The Pd foil with thickness of 0.1 mm (Alfa Aesar) was first cleaned by successively boiling in (22) McClure, S.; Goodman, D. W. Chem. Phys. Lett. 2009, 469, 1. (23) Nagarajan, S.; Thirunavukkarasu, K.; Gopinath, C. S.; Counsel, J.; Gilbert, L.; Bowker, M. J. Phys. Chem. C 2009, 113, 9814. (24) Gopinath, C. S.; Thirunavukkarasu, K.; Nagarajan, S. Chem. Asian J. 2009, 4, 74. (25) Nagarajan, S.; Gopinath, C. S. J. Indian Inst. Sci., 2010, 90, 245. (26) Alayon, E. M. C.; Singh, J.; Nachtegaal, M.; Harfouche, M.; van Bokhoven, J. A. J. Catal. 2009, 263, 228. (27) Campbell, C. T. Phys. Rev. Lett. 2006, 96, 066106. (28) Leisenberger, F. P.; Koller, G.; Sock, M.; Surnev, S.; Ramsey, M. G.; Netzer, F. P.; Kl€otzer, B.; Hayek, K. Surf. Sci. 2000, 445, 380. (29) Gabasch, H.; Unterberger, W.; Hayek, K.; Kl€otzer, B.; Kresse, G.; Klein, C.; Schmid, M.; Varga, P. Surf. Sci. 2006, 600, 205. (30) Zheng, G.; Altmann, E. I. J. Phys. Chem. B 2002, 106, 1048. (31) Kan, H. H.; Shumbera, R. B.; Weaver, J. F. Surf. Sci. 2008, 602, 1337. (32) Wickam, D. T.; Banse, B. A.; Koel, B. Surf. Sci. 1991, 243, 83. (33) Bondzie, V. A.; Kleban, P.; Dwyer, D. J. Surf. Sci. 1996, 347, 319. (34) Teschner, D.; Pestryakov, A.; Kleimenov, E.; Haevecker, M.; Bluhm, H.; Sauer, H.; Knop-Gericke, A.; Schloegl, R. J. Catal. 2005, 230, 186. (35) Ketteler, G.; Ogletree, F.; Bluhm, H.; Liu, H.; Hebenstreit, E. L. D.; Salmeron, M. J. Am. Chem. Soc. 2005, 127, 18269. (36) Han, J.; Zemlyanov, D. Y.; Ribeiro, F. H. Surf. Sci. 2006, 600, 2730 & 2752. (37) Lundgren, E.; Gustafson, J.; Mikkelsen, A.; Andersen, J. N.; Stierle, A.; Dosch, H.; Todorova, M.; Rogal, J.; Reuter, K.; Scheffler, M. Phys. Rev. Lett. 2004, 92, 046101.

18114 DOI: 10.1021/la103140w

Figure 1. In situ IRAS spectra for a Pd(100) surface preoxidized in 10 Torr of 16O2 at 700 K for 5 min, then exposed to 2 Torr of CO at 450 K. From bottom to top, each spectrum was taken over t = 30 s. The inset shows the IR intensity of the band 669 cm-1 as a function of CO exposing time at 450 K. NaOH and HNO3 solution, and then cleaned by acetone. After such chemical treatments, the Pd foil was introduced into the UHV chamber. The high purity CO and O2 (Hong Kong Specialty Gases Co., Ltd.) were further cleaned by a liquid nitrogen trap before being introduced into the reaction cell. Note that the vapor pressures at the liquid nitrogen temperature are about 0.52 atm (395 Torr) for CO and 0.2 atm (152 Torr) for O2, which are much higher than the pressures used in the present experiments. The sample was heated resistively with the temperature measured by a C-type thermocouple. CO oxidation rate was monitored by the pressure change, which was measured by an MKS precise pressure sensor, and also by the IR intensity changes of gas phase CO and CO2. Surface species were characterized by in situ IRAS, using a Bruker vacuum-type infrared spectroscopy, as shown in Figure S2.

3. Results 3.1. Reduction of Surface Palladium Oxides. As discussed earlier, one of the most interesting questions regarding CO oxidation on Pt-group metals is whether the catalytically active surface is reduced (metallic) or oxidized. In situ infrared spectroscopy studies of CO oxidation have shown absorption features characteristic of CO bound to a metallic surface under near-stoichiometric reaction conditions (O2/CO = 0.5).2-5,18 Many studies had shown that the reaction rates are significantly higher under oxygen-rich conditions on Pd and Pt surfaces.2,5-8,11-14,16-18 Using in situ scanning tunneling microscopy (STM) and mass spectroscopy, it was supposed that an oxide surface on Pd corresponds to the higher reaction rate.19,20,38 Here, we employed in situ IRAS to investigate the stability of a palladium oxide under pure CO conditions. The Pd(100) surface is preoxidized in 10 Torr O2 at 700 K for 5 min, and an IR spectrum is obtained at room temperature. As shown in Figure 1, two IR bands at 669 and 615 cm-1 were well resolved. The vibrational modes of PdO have been studied using far-infrared spectroscopy (FIR) by Kliche39 and Raman by McBride et al.40 Two longitudinal modes (LOs) at 672 and 622 cm-1 were observed by FIR, and two transverse modes (TOs) at 651 and 445 cm-1 were observed by Raman spectroscopy with polarized scattering. These (38) Hendriksen, B. L. M.; Bobaru, S. C.; Frenken, J. W. M. Topic Catal. 2005, 36, 43. (39) Kliche, G. Infrared Phys. 1985, 25, 381. (40) McBride, J. R.; Hass, K. C.; Weber, W. H. Phys. Rev. B 1991, 44, 5016.

Langmuir 2010, 26(23), 18113–18118

Chen et al.

phonons were assigned to the stretching modes of the [OPd4] tetrahedral, 672 cm-1 (LO: A2u, FIR), 651 cm-1 (TO: B1g, Raman), 622 cm-1 (LO: E2u, FIR) and 445 cm-1 (TO: Eg, Raman). Accordingly, the observed two bands at 669 and 615 cm-1 for the preoxidized Pd(100) surface by IRAS can be assigned to the vibrations of Pd-O of a palladium oxide. Note that, only LO modes were observed in IRAS.41 The vibration modes for the surface chemisorbed oxygen on a Pd surface were reported to be below 500 cm-1,42,43 which is located near the cutoff of the IR spectrometer using KBr as windows and a beam splitter prism. The preoxidized palladium surface was then exposed to 2 Torr of CO at room temperature. Although the reduction of a palladium oxide by CO can be neglected at room temperature, it is obvious at 450 K as indicated by the IR peak intensity of the palladium oxide, which decreases with increasing the exposure time. This is more clearly evidenced in the plot of the IR band intensity as a function of the reduction time, as shown in the inset in Figure 1. The reduction rate is low at the beginning, then increases significantly after about 100 s with the IR peak intensity for PdO decreasing almost linearly. The observation of an induction period for PdO reduction suggests that a thick oxide surface is more difficult to be reduced than that of a surface characterized by the coexistence of oxide and metallic Pd regions, consistent with previous temperature-programmed reduction (TPR) results.44 Since the IR bands at 669 and 615 cm-1 for PdO overlap with the bending modes of gas phase CO2 at 668 and 616 cm-1, respectively (Figure S3), isotopic 18O2 was used to further characterize the reduction of the preoxidized Pd(100) surface in a CO/O2 mixture. Pd18O shows two main IR bands at 648 and 590 cm-1, red-shifts of 31 and 25 cm-1, respectively, from those of Pd16O (Figures S3 and S4). For the TO modes, B1g and Eg of the [OPd4] tetrahedral in crystalline PdO, isotopic shifts of 38 and 25 cm-1, respectively, were observed by Raman.40 As shown in Figures S3 and S4, the IR bands of Pd18O can be distinguished from those of gas phase 18 OC16O with an intense peak at 663 cm-1. Here 18OC16O is the main product for C16O oxidation in 18O2, regarding the fact that CO was not dissociatively adsorbed on Pd metal surfaces. Note that there are two small peaks at 648 and 644 cm-1 for gas phase 18 OC16O, which may affect the measurement of PdO. The Pd(100) surface preoxidized by 2 Torr of 18O2 at 700 K for 5 min was then exposed to 27 Torr of C16O and 18 Torr of 18O2. At 450 K, the oxide surface was reduced quickly, as indicated from the decrease of the IR intensity of the band at 648 cm-1 in Figure S4. The oxide was almost completely reduced at the beginning 300 s at 450 K. The complete reduction after 1500 s was confirmed by pumping out the gas CO, O2, and CO2, and measuring a spectrum under UHV, which shows a non-apparent amount of oxide residue. This temperature is lower compared to those for CO oxidation on Pt-group metals of between 450 and 550 K. The results demonstrate that even the preoxidized surface will be reduced to metallic state under CO reaction conditions rich with oxygen. Numerous studies have shown that, under stoichiometric reaction conditions, chemisorbed CO is the dominant species on the metallic surface and that the rate is limited by the desorption of CO.2,4-9,11-13,17,18 Thus, it can be concluded that the active surface is metallic Pd for CO oxidation with a CO/O2 ratio near the stoichiometric one of 2/1, whether starting with a (41) Frank, M.; Wolter, K.; Magg, N.; Heemeier, M.; Kuhnemuth, R.; Baumer, M.; Freund, H. J. Surf. Sci. 2001, 492, 270. (42) Imbihl, R.; Demuth, J. E. Surf. Sci. 1986, 173, 395. (43) Stuve, E. M.; Madix, R. J. Surf. Sci. 1984, 146, 155. (44) Hinojosa, J. A., Jr.; Kan, H. H.; Weaver, J. F. J. Phys. Chem. C 2008, 112, 8324.

Langmuir 2010, 26(23), 18113–18118

Article

metallic or an oxidized surface. The reduction of PdO by CO in CO atmosphere and even in a CO/O2 mixture may be due to the fact that the rate of the reaction of CO with the surface oxygen species including chemisorbed oxygen and oxide is faster than the activation/dissociation of O2 under near-stoichiometric reaction conditions and at certain temperature. Recent, in situ ambient pressure X-ray photoelectron spectroscopy (APXPS) and mass spectrometry studies of CO catalytic oxidation of CO on a Pt(110) surface under various pressures of CO and O2 (with an O2/CO ratio up to 10) reveal that there is no surface oxide formation on Pt and are consistent with the Langmuir-Hinshelwood reaction mechanism.45 3.2. CO Oxidation on a Pd(100) Surface Using 18O2. We carried out experiments similar to those of the previous study,18 in which CO oxidation rate is monitored by the pressure change in a batch reactor as shown in Figure S1. Here, a constant resistive heating current was used to control the initial reaction temperature at 525 K for an O2/CO ratio of 2. In this case, the abrupt change of the sample temperature during CO oxidation can also serve as an indicator for the significant gain of the reaction rate. A liquid nitrogen trap for partially pumping out the reaction product, CO2, and also for further cleaning the reaction gas, and a reservoir of CO/O2 mixture for continuous feeding of CO/ O2 were directly connected to the reaction cell (Figure S2). Starting with an oxygen-rich condition, the 18O2/CO ratio will increase as the reaction proceeds. Here the isotopic 18O2 was used to better evaluate whether PdO was formed under the reaction conditions, as mentioned above. The results are shown in Figure 2. Within the first 120 s, the temperature is almost constant at 525 K, and the total pressure decreases virtually linearly. As the critical point is approached, the total pressure decreases dramatically, indicating a high reaction rate. Simultaneously, a sharp increase of the IR intensity of gas phase CO2 was evidenced by in situ IRAS (see Figure S5). Note that the critical point is characterized by a temperature jump and a sharp pressure drop, as indicated in Figure 2A, which was described in detail previously.18 The main IR band of Pd18O is 648 cm-1. However, there are two weak bands at 649 and 644 cm-1 for gas phase 18OC16O, as shown in Figure 2B, which results in additional complexity in evaluating the surface changes during reaction. IR spectra for Pd18O and gas phase 18OC16O are displayed in Figure 2B. The two spectra were added together, leading to a sum spectrum as shown by the dotted line in Figure 2B. As indicated, when summing these two spectra together, the intensity of the shoulder at ∼648 cm-1 becomes larger, i.e., IPdO þ ICO2. Thus, the ratio of the average intensities of bands 648 and 644 cm-1 (IsurfaceþICO2) to the band of 677 cm-1 at the left shoulder (ICO2) can be used to evaluate the surface change of Pd(100) during CO oxidation, as indicated in Figure 2B. This ratio is plotted as a function of the reaction time in Figure 2C. It is almost constant at about 0.97 for the entire reaction period, when starting with an 18O2/CO ratio of 2, including the low active region and the hyperactive region. The ratio of 0.97 is similar to the value obtained from gas phase CO2, and is apparently lower than that for the sum spectrum of PdO and gas phase CO2. Thus, these in situ IRAS measurements during CO oxidation using isotopic 18O2 experimentally reveal that palladium oxide is not formed during the hyperactive state. In fact, after the hyperactive state the single crystal surface was cooled down to room temperature under oxygen-rich conditions, and then pumped down to a vacuum. The acquired IR spectrum (45) Chung, J. Y.; Aksoy, F.; Grass, M. E.; Kondoh, H.; Ross, P.; Liu, Z.; Mun, B. S. Surf. Sci. 2009, 603, L35.

DOI: 10.1021/la103140w

18115

Article

Chen et al.

Figure 2. (A) Plots of the total pressure and reaction temperature as a function of the reaction time for CO oxidation in a batch reactor. The total pressure is 45 Torr (CO:O2 ∼ 1:2) at the beginning. The reactor is connected with a reactant gas reservoir to continually feed CO/O2 with a rate much slower than the reaction rate, and also connected to a liquid nitrogen trap to partially remove the formation product CO2. (B) IR spectra for a PdO surface, gas phase 18OC16O, and the sum spectrum of PdOþ18OC16O. (C) The ratio of the average IR intensity of the band at 648 and 644 cm-1 to that of 677 cm-1 as a function of the reaction time.

Figure 3. Plots of the total pressure and reaction temperature changes as a function of the reaction time for CO oxidation on a polycrystalline Pd in a batch reactor. The initial CO:O2 ratio is 1:2. (A) with continuous feeding of CO/O2 mixture, and (B) without continuous feeding of CO/O2 mixture.

does not present much PdO as compared to the IR band intensity for a PdO surface (Figure S5). 3.3. CO Oxidation on Polycrystalline Palladium. The experiments were also carried on a polycrystalline Pd metal surface to imitate a system that is more closely mimicked the exposed facets of a realistic catalyst. In the previous experiment, we used a constant current to maintain the initial sample temperature at 525 K for CO oxidation on Pd(100) under oxygenrich conditions, as shown in Figure 2. A temperature rise of more than 100 K was observed at the hyperactive state, indicated the substantial increase of the reaction rate at the critical point. Note that CO oxidation is an exothermic reaction with a standard enthalpy of 283 kJ/mol. In the following experiments, the reaction temperature is controlled constantly at 500 K to better evaluate the catalytic active surfaces for CO oxidation on Pd under oxygen-rich conditions. Two experiments, with and without continuous feeding of CO/O2 mixture during the reaction, were performed with in situ IRAS to monitor the surface state. The pressure and temperature changes as a function of the reaction time were compared in Figure 3. Sharp pressure drops occur in 18116 DOI: 10.1021/la103140w

Figure 4. In situ IRAS for CO oxidation on a polycrystalline Pd in a batch reactor with continuous feeding of CO/O2 mixture. The initial CO:O2 ratio is 1:2. (A) All spectra were subtracted by the clean surface background. (B) The spectra before and during the critical point are subtracted by the first spectrum that reaches the reaction temperature, while those after the critical points were subtracted by the spectrum just before the critical point.

both experiments, at the 20th and 17th spectra for with and without the CO/O2 feed, respectively. Here each spectrum takes about 40 s to acquire. The experiment with continuous CO/O2 feed achieves the sharp pressure drop, e.g., hyperactive state, in a longer time. This is because the feeding of CO/O2 delays the O2/ CO ratio reaching the critical point. Additionally, due to the continuous feeding, the pressure increases slightly for the last six spectra, which may result from the complete uptake of CO while leaving the overplus O2 from the feeding. In contrast, for the second experiment, without continuous feeding of CO/O2 mixture, it reaches the hyperactive state in a shorter time with the pressure continuously decreasing after the critical point. The pressure decrease after the hyperactive state is due to removing the gas phase CO2 by the liquid nitrogen trap, as indicated by the IR intensity of gas phase CO2 in Figure S6. Figure 4A shows the CO2 formation, gas phase CO decrease, and intensity change of adsorbed CO for the experiment of Figure 3A. There is a substantial increase of CO2, decrease of gas phase CO, and vanishment of adsorbed CO peak at 1970 cm-1 Langmuir 2010, 26(23), 18113–18118

Chen et al.

Article

at the hyperactive state deduced from the IR band intensities. The disappearance of adsorbed CO at the hyperactive state is consistent with our previous observation on Pd(110).18 The absence of gas CO after the hyperactive state indicates the depletion of CO at the hyperactive state, and that the reaction rate at the hyperactive state is limited by CO diffusion, which confirmed our previous suggestion.18 Due to CO depletion, the IR intensity of gas phase CO2 decreases as it is quickly removed by the liquid nitrogen trap. Similar results were observed for the second experiment of Figure 3B, as shown in Figure S6A. The low wavenumber region between 750 and 550 cm-1 is displayed to characterize the active surface under realistic CO oxidation conditions in Figure 4B. As discussed in section 3.1, the IR band of PdO is convoluted with the IR bending mode of CO2. Even using an isotopic 18O2, it is still complicated to distinguish IR bands of PdO from the bending modes of gas phase CO2. Hence, in this section the reaction temperature is reduced to 500 K, allowing for the CO2 formation rate to be moderated to facilitate the effect of the liquid nitrogen trap. That is, the partial pressure of CO2 in the reaction cell is depressed. With such improvements, the spectra after the hyperactive state are subtracted by the spectrum just before the sharp pressure drop. As shown in Figure 4B, the gas phase CO2 for spectra 27 and 28 is almost completely eliminated. By comparison of spectrum 28 with the spectrum of PdO/Pd for a polycrystalline Pd oxidized at 500 K in 10 Torr of O2 for 10 min and the spectrum (PdO) for Pd(100) oxidized at 700 K in 10 Torr of O2 for 5 min, the surface oxide formed during CO oxidation at 500 K under oxygen-rich conditions is negligible. For the experiment without continuous feeding of CO/O2 mixture (Figure S6B), the elimination of gas phase CO2 occurs at spectrum 21. A nonapparent amount of PdO is formed after the hyperactive state.

4. Discussion 4.1. Surface Oxide. For a two-dimensional (2D) palladium oxide on Pd metal surface, if a 2D network of Pd-O runs parallel to the metal surface, one may expect that the in-plane Pd-O vibration would be IR inactive regarding the IR selection rule. However, such a 2D Pd-O network is not simply isolated above the metal surface. It closely locates on the Pd surface, leading to direct bonding of O and Pd to the underlying Pd surface atoms. In fact, a trilayer structure of√O-Pd-Owas proposed for the √ surface oxide of Pd(100)-( 5 5)R26.7°-O.46 Thus, there should be some perpendicular component of the νPd-O mode. The active vibrational modes for such surface oxides, such as SiO2/Mo(112), were well examined by both experimental and theoretical studies, in which the significant intensity of the Si-O-Mo mode was observed by IRAS and high-resolution electron energy loss spectroscopy (HREELS).47,48 Therefore, we believed that there are IR active modes for any type of palladium oxides, including the surface 2D oxide and bulk-like oxide. 4.2. Active Surface. Catalytic oxidation of CO and CH4 on Pd-catalysts exhibits many unusual phenomena, such as oscillation and extreme sensitivity to the pretreatment history of the catalyst.43-46 These have been attributed to transitions between the metal and oxide surfaces. Using STM, temperature-programmed desorption (TPD), and LEED, Altman et al.49 carried out detailed studies of the mechanism of oxide formation and distinguished (46) Kostelnik, P.; Seriani, N.; Kresse, G.; Mikkelsen, A.; Lundgren, E.; Blum,  V.; Sikola, T.; Varga, P.; Schmid, M. Surf. Sci. 2007, 601, 1574. (47) Chen, M. S.; Santra, A. K.; Goodman, D. W. Phys. Rev. B 2004, 69, 155404. (48) Todorova, T. K.; Sierka, M.; Sauer, J.; Kaya, S.; Weissenrieder, J.; Lu, J. L.; Gao, H. J.; Shaikhutdinov, S.; Freund, H. J. Phys. Rev. B 2006, 73, 165414. (49) Zheng, G.; Altman, E. I. Surf. Sci. 2000, 462, 151; 2002, 504, 253.

Langmuir 2010, 26(23), 18113–18118

three states involving four surface oxygen phases on Pd(111) and four states with five surface oxygen phases on Pd(100). The key process involves chemisorbed oxygen, surface oxide, and bulk oxide with increasing O2 dosing amounts and treating temperature. Our results evidently showed that the surface oxide and bulk oxide are not presented on the Pd(100) and polycrystalline Pd for CO oxidation at ∼525 K, even under oxygen-rich conditions. Regarding the facts that a very high rate (a TOF of more than several thousands) was observed at the hyperactive state, and an insignificant amount of CO adsorption was evidenced by in situ IRAS, it can be concluded that the active surface is a chemisorbed oxygen-rich surface for the hyperactive state. This can be reasoned by considering mass transfer limitations at the hyperactive state. Due to the mass transfer limit of CO, CO is deficient at the near-surface region, resulting in the absence of an adsorbed CO IR peak. However, the mass transfer coefficients for CO and O2 are similar,50 and the partial pressure of O2 is significantly higher than that of CO in the gas phase at the hyperactive state. That is, the amount of O2 diffused into the near surface region is significantly more than that of CO. Additionally, for CO oxidation, the stoichiometric of CO:O2 is 2:1. This will result in oxygenrich conditions at the near surface region. As mentioned above, no surface oxide or bulk oxide formed at the hyperactive state. Thus, the oxygen-rich conditions at the near surface region would lead to a chemisorbed oxygen-rich surface. Overall, the Pd surface is dominant by adsorbed CO on metallic surfaces at near-stoichiometric conditions. With increasing the oxygen partial pressure or decreasing the CO partial pressure, it may change to an adsorbed oxygen dominant surface, leading to a hyperactive state. Note that such a switch from CO dominance to oxygen dominance depends not only on the O2/CO ratio but also on the reaction temperature; that is, the higher the reaction temperature, the lower the O2/CO ratio required to achieve the hyperactive state.18 Such an oxygen-rich environment would be expected to result in the formation of a palladium oxide at last. However, the formation of PdO can be neglected at the hyperactive state and under oxygen-rich conditions. This can be understood from the facts √ that √ a chemisorbed oxygen-dominant surface, Pd(100)( 5 5)R26.3°-O, was found to produce a considerable kinetic barrier against surface PdO and bulk PdO formation, and also on a Pd(110).51-53 This means the formation of palladium oxide from a chemisorbed oxygen-rich surface is a relatively slow process. However, the on-coming CO that reacts with the surface oxygen species is a faster one, as observed from the hyperactivity. The palladium oxide will not be present as long as the amount of the on-coming CO is high enough. All together, the chemisorbed oxygen-rich surface is an active surface corresponding to the hyperactivity.

5. Conclusions Using the newly built in situ IRAS system combined with isotopic 18O2, we have observed the formation of PdO on a Pd(100) surface in O2. The so-formed PdO can be reduced by CO and even by a CO/O2 mixture at 450 K. The in situ IRAS results evidently demonstrate that no palladium oxide was formed on the Pd(100) and polycrystalline Pd during CO oxidation at 500525 K, even under oxygen-rich conditions and for the hyperactive (50) Seader, J. D.; Henley, E. J. Separation Process Principles; Wiley: New York, 1998. (51) Huang, W. X.; Zhai, R. S.; Bao, X. H. Appl. Surf. Sci. 2000, 158, 287. (52) Stierle, A. Int. J. Mater. Res. 2009, 100, 1308. (53) Westerstroem, R.; Weststrate, C. J.; Gustafson, J.; Mikkelsen, A.; Schnadt, J.; Andersen, J. N.; Lundgren, E.; Seriani, N.; Mittendorfer, F.; Kresse, G.; Stierle, A. Phys. Rev. B 2009, 80, 12543.

DOI: 10.1021/la103140w

18117

Article

Chen et al.

state. These facts conclude that the chemisorbed oxygen-rich Pd surface is the active surface for the observed hyperactivity. The gas phase CO is depleted at the hyperactive state, and the hyperactive rate observed is still limited by the mass transfer of CO.

China (973 program: 2005CB221401, 2010CB732303), the Major Project of Chinese Ministry of Education (No: 309019), the Ph.D. Programs Foundation of the Chinese Ministry of Education (No. 200803841011), and the Natural Science Foundation of Fujian Province, China (2008J0168).

Acknowledgment. The authors thank Sean M. McClure for valuable discussion and advice. We gratefully acknowledge the financial support by the National Natural Science Foundation of China (20873109), the National Basic Research Program of

Supporting Information Available: Hyperactivity for CO oxidation on Pd; In situ IRAS technique; IR spectra for CO oxidation on Pd surfaces. This material is available free of charge via the Internet at http://pubs.acs.org.

18118 DOI: 10.1021/la103140w

Langmuir 2010, 26(23), 18113–18118