The Defect-Mediated Mechanism of the High-Temperature Oscillatory NO + CO Reaction on Pt{100} As Revealed by Real-Time in-situ Vibrational Spectroscopy J. H. Miners † and P. Gardner* ,‡ Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6 D-14195, Germany, and Department of Chemistry, Faraday Building, UMIST, Manchester M60 1QD, U.K. ReceiVed: February 4, 2000; In Final Form: May 22, 2000 At a partial pressure of 10 -7 mbar, the reaction between NO and CO on Pt{100} exhibits oscillatory behavior in two distinct temperature regimes. Oscillations in the high-temperature regime (380-411 K) are accompanied by a phase transition from the (1 × 1) surface to the hex surface. Using infrared reflection-absorption spectroscopy (IRAS) and a novel method of data acquisition, we show that during the oscillation cycle, the only molecular species present on the surface is atop CO, adsorbed on the (1 × 1) phase at very low coverage (∼0.03-0.007 ML). Furthermore, the minimum in the CO coverage coincides with the maximum reaction rate, as measured by the partial pressure of CO 2 . From a comparison of these data with previously published LEED and PEEM studies of the same system, it can be seen that the high-reaction-rate branch of the oscillatory cycle coincides with the maximum area of the surface in the hex phase. This is in contrast to previously proposed mechanisms, which assume that the (1 × 1) surface is the active phase. Since the hex surface is inactive for NO dissociation, we conclude that defects on the hex surface, created during the (1 × 1)-hex phase transition and known to be active for NO dissociation, are responsible for the high-reaction-rate branch. Removal of these defects by annealing provides the means by which the reaction returns to the low-rate branch of the cycle. This annealing process also accounts for the observation that the period of oscillation decreases with temperature. 1. Introduction Kinetic oscillations in heterogeneous catalysis were redis- covered nearly 30 years ago, during studies of CO oxidation on polycrystalline Pt surfaces at atmospheric pressure. 1-3 More recent studies, in particular of the CO + O 2 and CO + NO reactions, have focused on single-crystal surfaces where the full range of modern surface science techniques can be applied. Kinetic oscillations in the CO + NO reaction on Pt{100} were first reported by Singh-Boparai and King. 4 Since then, four separate regions of existence have been identified. 5-8 Further work by Veser et al. 9-12 determined that the lower-temperature oscillatory regime exists for the range of partial pressure ratios 0.8 < P NO :P CO < 2.5 and that only the unsynchronized local pattern-forming oscillations are observed, whereas the upper- temperature regime exists in the range 0.8 < p NO :p CO < 1.8 and exhibits sustained oscillations. Fink et al. have mathematically modeled the reaction and have reproduced both the steady-state behavior and the lower- temperature kinetic oscillations by assuming island formation and a vacant site requirement for NO dissociation. 8 The driving forces are considered to be the surface explosion and the subsequent availability of vacant sites for NO dissociation. Veser et al. 10 subsequently noted, however, that the model did not adequately describe the oscillations in the higher-temperature regime, where their predictions were in contradiction with their experimental results of an inverse relationship between tem- perature and the period of oscillation. More recently, Hopkinson and King 13 have formulated an alternative model which predicts essentially the same behavior in the lower existence region but not in the upper one, where the reaction mechanism is very similar to that of the oscillatory CO + O 2 reaction on the same surface. The driving force is considered to be a combination of the surface phase transition, the existence of two rate branches, and the hysteresis associated with the change in coverage of the (1 × 1) phase. The essential steps in their proposed mechanism for the high-temperature oscillatory regime are as follows: the higher sticking coefficient of CO, compared to NO, leads to a buildup of CO coverage on the hex surface and a local lifting of the reconstruction. The (1 × 1) islands initially grow via the trapping of CO from the surrounding hex surface. As the (1 × 1) islands grow, direct adsorption from the gas phase begins to play a more prominent role, and since P NO > P CO , NO adsorption dominates. NO immediately dissociates and reacts with the CO, resulting in O-covered (1 × 1) islands. These are not stable, and the reconstruction reforms. This results in a reduction of the reaction rate, and so, the CO coverage on the hex phase increases again. This model suffers, however, from an inability to describe the temperature-related behavior of the oscillatory frequency, similar to that of Fink et al. 8,14 The weakness in these models stems largely from a lack of direct information concerning the nature and concentration of the species adsorbed on the surface during the reaction. Infrared spectroscopy offers the possibility of remedying this situation at least in part. Infrared reflection-absorption spectra have been measured for such systems under globalsi.e., nontemporally resolvedsconditions. 15-17 Spectra with temporal resolution were, however, recorded in the transmission mode by Schu ¨th and Wicke on supported polycrystalline platinum under 1 atm * Corresponding author. E-mail: Peter.Gardner@umist.ac.uk. Tel: ++44 161 200 4463. † Fritz-Haber-Institut der Max-Planck-Gesellschaft. ‡ UMIST. 10265 J. Phys. Chem. B 2000, 104, 10265-10270 10.1021/jp000465b CCC: $19.00 © 2000 American Chemical Society Published on Web 10/12/2000