Joshua Joseph May
Published: 2017
Total Pages: 250
Get eBook
The continued development of the chirped pulse amplification technique has allowed for the development of lasers with powers of in excess of $10^{15}W$, for pulse lengths with durations of between .01 and 10 picoseconds, and which can be focused to energy densities greater than 100 giga-atmospheres. When such lasers are focused onto material targets, the possibility of creating particle beams with energy fluxes of comparable parameters arises. Such interactions have a number of theorized applications. For instance, in the Fast Ignition concept for Inertial Confinement Fusion \cite{Tabak:1994vx}, a high-intensity laser efficiently transfers its energy into an electron beam with an appropriate spectra which is then transported into a compressed target and initiate a fusion reaction. Another possible use is the so called Radiation Pressure Acceleration mechanism, in which a high-intensity, circularly polarized laser is used to create a mono-energetic ion beam which could then be used for medical imaging and treatment, among other applications. For this latter application, it is important that the laser energy is transferred to the ions and not to the electrons. However the physics of such high energy-density laser-matter interactions is highly kinetic and non-linear, and presently not fully understood. In this dissertation, we use the Particle-in-Cell code OSIRIS \cite{Fonseca:2002, Hemker:1999} to explore the generation and transport of relativistic particle beams created by high intensity lasers focused onto solid density matter at normal incidence. To explore the generation of relativistic electrons by such interactions, we use primarily one-dimensional (1D) and two-dimensional (2D), and a few three-dimensional simulations (3D). We initially examine the idealized case of normal incidence of relatively short, plane-wave lasers on flat, sharp interfaces. We find that in 1D the results are highly dependent on the initial temperature of the plasma, with significant absorption into relativistic electrons only possible when the temperature is high in the direction parallel to the electric field of the laser. In multi-dimensions, absorption into relativistic electrons arises independent of the initial temperature for both fixed and mobile ions, although the absorption is higher for mobile ions. In most cases however, absorption remains at $10's$ of percent, and as such a standing wave structure from the incoming and reflected wave is setup in front of the plasma surface. The peak momentum of the accelerated electrons is found to be $2 a_0 m_e c$, where $a_0 \equiv e A_0/m_e c^2$ is the normalized vector potential of the laser in vacuum, $e$ is the electron charge, $m_e$ is the electron mass, and $c$ is the speed of light. We consider cases for which $a_0>1$. We therefore call this the $2 a_0$ acceleration process. Using particle tracking, we identify the detailed physics behind the $2 a_0$ process and find it is related to the standing wave structure of the fields. We observe that the particles which gain energy do so by interacting with the laser electric field within a quarter wavelength of the surface where it is at an anti-node (it is a node at the surface). We find that only particles with high initial momentum -- in particular high transverse momentum -- are able to navigate through the laser magnetic field as its magnitude decreases in time each half laser cycle (it is an anti-node at the surface) to penetrate a quarter wavelength into the vacuum where the laser electric field is large. For a circularly polarized laser the magnetic field amplitude never decreases at the surface, instead its direction simply rotates. This prevents electrons from leaving the plasma and they therefore cannot gain energy from the electric field. For pulses with longer durations ($\gtrsim 250fs$), or for plasmas which do not have initially sharp interfaces, we discover that in addition to the $2 a_0$ acceleration at the surface, relativistic particles are also generated in an underdense region in front of the target. These particles have energies without a sharp upper bound. Although accelerating these particles removes energy from the incoming laser, and although the surface of the plasma does not stay perfectly flat and so the standing wave structure becomes modified, we find in most cases, the $2 a_0$ acceleration mechanism occurs similarly at the surface and that it still dominates the overall absorption of the laser. To explore the generation of relativistic electrons at a solid surface and transport of the heat flux of these electrons in cold or warm dense matter, we compare OSIRIS simulations with results from an experiment performed on the OMEGA laser system at the University of Rochester. In that experiment, a thin layer of gold placed on a slab of plastic is illuminated by an intense laser. A greater than order-of-magnitude decrease in the fluence of hot electrons is observed when those electrons are transported through a plasma created from a shock-heated plastic foam, as compared to transport through cold matter (unshocked plastic foam) at somewhat higher density. Our simulations indicate two reasons for the experimental result, both related to the magnetic field. The primary effect is the generation of a collimating B-field around the electron beam in the cold plastic foam, caused by the resistivity of the plastic. We use a Monte Carlo collision algorithm implemented in OSIRIS to model the experiment. The incoming relativistic electrons generate a return current. This generates a resistive electric field which then generates a magnetic field from Faraday's law. This magnetic field collimates the forward moving relativistic electrons. The collisionality of both the plastic and the gold are likely to be greater in the experiment than the 2D simulations where we used a lower density for the gold (to make the simulations possible) which heats up more. In addition, the use of 2D simulations also causes the plastic to heat up more than expected. We compensated for this by increasing the collisionality of the plasma in the simulations and this led to better agreement. The second effect is the growth of a strong, reflecting B-field at the edge of the plastic region in the shock heated material, created by the convective transport of this field back towards the beam source due to the neutralizing return current. Both effects appear to be caused primarily by the difference is density in the two cases. Owing to its higher heat capacity, the higher density material does not heat up as much from the heat flux coming from the gold, which leads to a larger resistivity. Lastly, we explored a numerical effect which has particular relevance to these simulations, due to their high energy and plasma densities. This effect is caused by the use of macro particles (which represent many real particles) which have the correct charge to mass ratio but higher charge. Therefore, any physics of a single charge that scales as $q^2/m$ will be artificially high. Physics that involves scales smaller than the macro-particle size can be mitigated through the use of finite size particles. However, for relativistic particles the spatial scale that matters is the skin depth and the cell sizes and particle sizes are both smaller than this. This allows the wakes created by these particles to be artificially high which causes them to slow down much faster than a single electron. We studied this macro-particle stopping power theoretically and in OSIRIS simulations. We also proposed a solution in which particles are split in to smaller particles as they gain energy. We call this effect Macro Particle Stopping. Although this effect can be mitigated by using more particles, this is not always computationally efficient. We show how it can also be mitigated by using high-order particle shapes, and/or by using a particle-splitting method which reduces the charge of only the most energetic electrons.
