Electron Acceleration and Radiation Generation from Relativistic Laser-Plasma Interactions and Statistical Methods at High Repetition-Rate

Abstract

This dissertation explores the interaction between high-intensity lasers and plasmas to accelerate electrons and produce radiation via experimental and computational efforts. The laser pulses used in this dissertation have ultrashort duration ($<100$ fs), near-infrared to mid-infrared wavelength (0.8 $mu m$, 2 $mu m$, or 3.9 $mu m$), millijoules of energy, and high repetition rates (480 Hz or 20 Hz). The plasma sources applied are from solid-density targets (overdense) or gaseous targets (underdense). With the high-repetition-rate capability, statistical methods are employed to optimize certain aspect of the experiments and to interpret the physics.

MeV-level attosecond electron bunches from the interactions between ultrashort pulses (30 fs, 0.8 $mu m$, 12 mJ) and solid targets (fused silica and copper). Attosecond electron bunches are only observed at grazing incidence , and the bunch duration is measured in acf{PIC} simulations. The effects of carrier-envelope phase, preplasma density profile, laser intensity, and the focal spot size are analyzed.

Surface acs{HHG} and corresponding phenomena are studied using femtosecond mid-infrared laser pulses (2 $mu m$, 1.6 mJ, 67 fs) interacting with solid targets (fused silica and silicon). Experimental measurements of the acs{HHG} spectra and the beam divergence are reported. The power-law scaling of harmonic efficiency vs. harmonic order is examined. The intensity of polarization of the harmonics are measured when the driving laser pulses are polarized in linear and circular directions. The scaling of harmonic efficiency vs. laser intensity is investigated.

Characteristic x-ray emissions from laser-solid interactions are presented. Laser pulses with various wavelengths and pulse energies are used to interact with a molybdenum target with various preplasma density profiles. The study is performed both experimentally with hundreds of thousands of laser shots, and computationally with acs{PIC} simulations scanning over the 4-dimensional parameter space consisting of laser wavelength, pulse energy, preplasma profile, and x-ray emission.

Statistical methods are used to improve the focus of laser beams in high numerical aperture. A method that optimizes the focus of a high-power laser without attenuation is demonstrated experimentally using near-infrared (0.8 $mu m$) and mid-infrared (2 $mu m$) laser pulses, where the second harmonic generation at full intensity in a low-pressure gas provides a figure of merit for optimizing the laser wavefront via a genetic algorithm.

Coherent control of the dynamics of laser-wakefield acceleration driven by ultrashort ($sim 100$ fs) mid-infrared ($sim 3.9~mu$m) laser pulses is demonstrated, where plasma densities up to $3times 10^{19}cm^{-3}$ (or $40%$ of the critical density at $lambda=3.9 mu m$) are used. MeV-level, collimated electron beams with non-thermal, peaked energy spectra are generated. Optimization of electron beam qualities is realized through adaptive control of the laser wavefront using a deformable mirror and a genetic algorithm. The improvement in the electron beam quality is explained by acs{PIC} simulations using the optimal wavefront.

Applications of machine learning techniques in relativistic laser-plasma experiments are explored beyond optimization purposes. With the trained supervised learning models, the beam charge of electrons produced in a laser wakefield accelerator is predicted given the laser wavefront change caused by a deformable mirror. Feature importance analysis on the trained models shows that specific aberrations in the laser wavefront are favored. The quality of the measured data is characterized, and anomaly detection is demonstrated. The model robustness against measurement errors is examined by applying a range of virtual measurement error bars.