Dresden 2026 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 85: Vacuum Science & Technology: Theory and Applications II
O 85.5: Talk
Thursday, March 12, 2026, 16:15–16:30, HSZ/0401
K-Means Clustering of Time-Resolved Diffraction: Dynamics of Laser-Excited Si(111)-(7x7) — •Jonas D. Fortmann1,2, Alexander Neuhaus1, Pascal Dreher1,3, Birk Finke1, Christian Brand1, and Michael Horn-von Hoegen1 — 1Uni Duisburg-Essen, Duisburg, Germany — 2MAX IV, Lund, Sweden — 3Uni Würzburg, Würzburg, Germany
We present an unsupervised analysis workflow that leverages k-means clustering to extract characteristic dynamics from time-resolved diffraction data, using ultrafast reflective high energy electron diffraction (URHEED) measurements of laser-excited Si(111)-(7x7) as a case study. Delay-dependent intensity profiles of diffraction features are clustered by trajectory shape to reveal characteristic dynamics without prior model assumptions. When combined with a simple population model, relative integral-intensity changes of clusters yield lifetimes of coupled phonon systems and trace the excitation pathway to surface electronic states. The method is robust to practical limitations (e.g., camera light sensitivity), scales to large datasets, and is directly transferable to other diffraction modalities (ULEED, UTEM, and pump-probe X-ray diffraction), where clustering of intensity evolutions can provide an objective map of structural dynamics.
Keywords: clustering; k-Means; phonon dynamics; electron diffraction; silicon