The oxidation of the Pd(111) surface was studied by in situ XPS during heating and cooling in 0.4 mbar O 2 . The in situ XPS data were complemented by ex situ TPD results. A number of oxygen species and oxidation states of palladium were observed in situ and ex situ. At 430 K, the Pd(111) surface was covered by a 2D oxide and by a supersaturated O ads layer. The supersaturated O ads layer transforms into the Pd 5 O 4 phase upon heating and disappears completely at approximately 470 K. Simultaneously, small clusters of PdO, PdO seeds, are formed. Above 655 K, the bulk PdO phase appears and this phase decomposes completely at 815 K. Decomposition of the bulk oxide is followed by oxygen dissolution in the near-surface region and in the bulk. The oxygen species dissolved in the bulk is more favoured at high temperatures because oxygen cannot accumulate in the near-surface region and diffusion shifts the equilibrium towards the bulk species. The saturation of the bulk "reservoir" with oxygen leads to increasing the uptake of the near-surface region species. Surprisingly, the bulk PdO phase does not form during cooling in 0.4 mbar O 2 , but the Pd 5 O 4 phase appears below 745 K. This is proposed to be due to a kinetic limitation of PdO formation because at high temperature the rate of PdO seed formation is compatible with the rate of decomposition.

The oxidation of the Pd(111) surface was studied by in situ XPS during heating and cooling in 3×10 The surface was completely covered with the 2D oxide between 600 K and 655 K. Depth profiling by photon energy variation confirmed the surface nature of the 2D oxide. The 2D oxide decomposed completely above 717 K. Diffusion of oxygen in the palladium bulk occurred at these temperatures. A substantial oxygen signal assigned to the dissolved species was detected even at 923 K. The dissolved oxygen was characterised by the O 1s core level peak at 528.98 eV. The "bulk" nature of the dissolved oxygen species was verified by depth profiling. During cooling in 3×10 -3 mbar O 2 , the oxidised Pd 2+ species appeared at 788 K whereas the 2D oxide decomposed at 717 K during heating. The surface oxidised states exhibited an inverse hysteresis. The oxidised palladium state observed during cooling was assigned to a new oxide phase, probably the ( 67 67 × )R12.2° structure.

The reaction between CH4 and O2 (1:5) was studied by in situ XPS during heating and cooling in a 0.33 mbar reaction mixture. During heating, the reaction rate exhibited an activity maximum at 650 K, whereas no activity maximum was found during the subsequent cooling ramp. This kinetic hysteresis was assigned to the spectroscopically observed difference in the surface oxidation state. During heating, the reaction rate approached the 650 K maximum in the stability range of bulk PdO seeds among the otherwise Pd5O4 2D oxide covered surface. On the other hand, no PdO seeds were formed during cooling, most likely due to kinetic limitations of PdO nucleation on a passivating surface oxide layer containing less oxygen than Pd5O4.

The reactivity of several Pd-O species toward CO oxidation was compared experimentally, making use of chemically, structurally and morphologically different model systems such as single-crystalline Pd(111) covered by adsorbed oxygen or a Pd(5)O(4) surface oxide layer, an oriented Pd(111) thin film on NiAl oxidized toward PdO(x) suboxide and silica-supported uniform Pd nanoparticles oxidized to PdO. The oxygen reactivity decreased with increasing oxidation state: O(ad) on metallic Pd(111) exhibited the highest reactivity and could be reduced within a few minutes already at 223 K, using low CO beam fluxes around 0.02 ML s(-1). The Pd(5)O(4) surface oxide on Pd(111) could be reacted by CO at a comparable rate above 330 K using the same low CO beam flux. The more deeply oxidized Pd(111) thin film supported on NiAl was already much less reactive, and reduction in 10(-6) mbar CO at T > 500 K led only to partial reduction toward PdO(x) suboxide, and the metallic state of Pd could not be re-established under these conditions. The fully oxidized PdO nanoparticles required even rougher reaction conditions such as 10 mbar CO for 15 min at 523 K in order to re-establish the metallic state. As a general explanation for the observed activity trends we propose kinetic long-range transport limitations for the formation of an extended, crystalline metal phase. These mass-transport limitations are not involved in the reduction of O(ad), and less demanding in case of the 2-D Pd(5)O(4) surface oxide conversion back to metallic Pd(111). They presumably become rate-limiting in the complex separation process from an extended 3-D bulk oxide state toward a well ordered 3-D metallic phase.