This thesis studies the relativistic laser plasma interaction (a0 = eE0/meωc > 1) using joint experimental and computational approaches, the former using high power, short pulse laser systems and the latter via particle-in-cell (PIC) modeling and high performance computing. Different aspects are explored including determination of criteria for physically meaningful PIC simulations, development of the technology needed for such experiments including new systems for targetry and a new model for understanding dielectric plasma mirrors and, finally, a study of laser based ion acceleration in three different acceleration regimes.
We have shown that, although in the laboratory frame an electron accelerated via direct laser acceleration undergoes a decreased oscillation frequency, the criterion on the time resolution required to model this process actually becomes more restrictive. This is due to the difficulty of resolving the stopping points in the electron trajectory giving rise to dephasing. It is shown that when using the Boris particle pusher, the time step must satisfy Δt << T/a0, where a0 is the dimensionless vector potential. An adaptive time step algorithm based on this criterion is demonstrated that achieves more than an order of magnitude improvement in preserving the constant of the motion γ - px/mec = 1 with only a 1/√a0 increase in the number of time steps required, as opposed to the 1/a0 scaling in the non-adaptive algorithm.
High pulse contrast is crucial for performing many experiments on high intensity lasers in order to minimize modification of the target surface by pre-pulse. Dielectric plasma mirrors are commonly used to enhance the contrast of laser pulses and their subsequent ion production, but have not been modeled extensively. Presented here are novel 2D3V LSP particle-in-cell simulations of liquid crystal plasma mirror operation which include a dielectric model with a population of cold neutral atoms and incorporating multiphoton ionization, low temperature modifications to the collision model, and dimensionality corrections. For the first time, plasma mirror behavior is modeled as a function of laser intensity, with excellent agreement to experiment over three orders of magnitude. In the future, this simulation framework can be used to make predictions about plasma mirror behavior about aspects that are difficult to measure, such as pulse shortening and mode degradation at full power.
Results from a high-contrast, ion acceleration experiment are presented, which used the Draco laser (~3 J, 1021 W/cm2) at 45° angle of incidence on liquid crystal targets. 450 shots were collected in 5 days with each shot having up to 9 diagnostics in use. The data showed predominantly target normal directed ions for all target thicknesses from > 1 μm down to 10 nm, with peak protons energies up to 26 MeV found for the thinnest targets. Target normal ions are often considered to be an indication of the target normal sheath acceleration (TNSA) mechanism, but the TNSA-like signal in this experiment persisted down to target thicknesses < 30 nm, well under a predicted theoretical transition to radiation pressure acceleration (RPA) of ~130 nm for these conditions. By comparison, experimental measures of the transmitted light and particle-in-cell simulations indicate that relativistic transparency was achieved for 30 nm targets. 3D particle-in-cell simulations using LSP reproduce the dominance of target normal acceleration seen in the DRACO experiment with a good quantitative match to experimental peak proton energies. Tracking individual particles in simulations reveals that target normally directed ions are produced by TNSA for thick targets, but by radiation pressure acceleration (RPA) for thin targets. A simple analytical model is derived for RPA-driven target deformation that agrees well with 2D and 3D simulation results, supporting this conclusion. Using these simulations, the amount of energy in accelerated protons can be broken down into TNSA and RPA contributions. For a 300 nm target, TNSA makes up 90% of this energy with RPA contributing 10%, while in a 100 nm target TNSA and RPA each make up 30%, with the remaining due to relativistic transparency effects.