Previous studies of the reaction NO + CO → 1 2N2 + CO2 on a cylindrical single crystal of Pt have revealed a variety of spatiotemporal concentration patterns. Understanding of these patterns has been limited by the fact that they are irregular and that no well-defined structures like target patterns or spiral waves are discernible. In this study we use the Karhunen-Loève (KL) decomposition, a classical tool of statistical pattern recognition, to assist the visualization and analysis of the patterns. We find that behavior that was previously thought of as 'turbulent' can actually be quite low-dimensional. In general, the patterns are locally like travelling waves, with overall structure affected by the inhomogeneity of the catalyst surface. In one case, we find localized regions that are driven to subharmonic resonance by 'more global' oscillations. There are also cases where spatially local and global oscillations apparently coexist on the surface. Finally, the use of KL analyses of simulated data as a guide to the understanding of real data is described and discussed.
All Science Journal Classification (ASJC) codes
- Statistical and Nonlinear Physics
- Physics and Astronomy(all)
- Applied Mathematics