Speaker
Description
Dielectric laser acceleration (DLA) enables compact, chip-scale accelerators. Recent demonstrations of multi-stage acceleration, alternating-phase focusing and interaction length in the millimeter range in dielectric nanostructures have verified the scalability of DLA concepts but the compactness places extreme demands on beam diagnostics due to submicron apertures, sub-fs bunch lengths and strong non-linear dynamics. To support the development of multi-stage structures and precise matching between stages, we explore diagnostic concepts that use the dielectric structures themselves. Tailored gratings can encode beam properties—such as bunch length—into emitted radiation, providing compact, high-resolution, on-chip diagnostics. Additionally to these hardware approaches, we propose to apply a machine-learning–based virtual diagnostic for DLA experiments. A neural network trained on 6D tracking simulations reconstructs key interaction parameters from the post-DLA electron beam and intermediate diagnostics and enables real-time optimization even in strongly non-linear regimes. Combining integrated dielectric diagnostics with ML-based reconstruction provides a scalable strategy for precise characterization and control of next-generation dielectric laser accelerators.
Funding Agency
The presented work is funded by the German Federal Ministry of Research, Technology and Space (Grant Nos. FKZ: 05K22RDC and 05K25RD1).
| In which format do you inted to submit your paper? | LaTeX |
|---|