Electrical and Computer Engineering ETDs

Publication Date

Summer 7-28-2026

Abstract

Modern beam-based materials research demands electron sources with increasingly precise time resolution, high brightness, and stability. For example, there are various instruments across the U.S. dedicated to ultrafast electron diffraction (UED), but fewer are dedicated to ultrafast electron microscopy (UEM), which demands beam stability. Modeling femtosecond electron bunches with ultra-low emittance < 50 nm-rad inside a 1.6-cell S-band (2856 MHz) rf photoinjector with full 3D electromagnetic simulations can require high-performance computing (HPC) environments due to the scale disparity between the macroscopic cavity geometry and the femtosecond-scale bunch kinematics. To enable optimization in a desktop computing environment, this work presents a streamlined 3D electromagnetic Particle-in-Cell (PIC) simulation framework optimized for 400-femtosecond bunch regimes. The covariance method was employed to study emittance properties for electron bunch counts ranging from thousands to millions and has successfully resolved highly transient, pure rf phase-space rotations, such as a localized energy-spread minimum occurring at the gun exit iris. To overcome the computational cost of these simulations, a machine-learning-based Bayesian optimization approach was deployed to construct multi-objective Pareto fronts from sparse datasets. Because the simulation software is scalable from desktop to HPC facilities, the framework is ready for experiments at the National Energy Research Scientific Computing Center (NERSC) at Lawrence Berkeley National Laboratory (LBNL).

Document Type

Dissertation

Degree Name

Electrical Engineering

Level of Degree

Doctoral

Department Name

Electrical and Computer Engineering

First Committee Member (Chair)

Manel Martinez-Ramon

Second Committee Member

Edl Schamiloglu

Third Committee Member

Mark Gilmore

Fourth Committee Member

Yu-Lin Shen

Download

Share

COinS