105 / 2023-04-13 21:13:59

Quasistatic simulations of long-distance laser wakefield acceleration

simulations,laser wakefield acceleration,plasma

极端光下的基础物理学 > 粒子加速-海报

Laser-driven plasma wakefield acceleration (LWFA) is a promising technology for compact sources of high-energy electron beams. With advent of more powerful laser systems, new parameter regions for LWFA become accessible, including relatively low plasma densities that provide higher energies and charges of accelerated bunches. Numerical simulations of a plasma accelerator operating at low plasma densities are computationally demanding because of the large ratio of laser frequency w₀ to the plasma frequency w_p, especially if the driver is channeled and the acceleration distance is long.

If the frequency ratio w₀/w_p is large, it is especially advantageous to use a quasistatic code and the envelope model of the laser pulse for numerical simulations. The envelope equation increases the temporal and spatial scales to be resolved by a factor of w₀/w_p. Accordingly, it is possible to increase the simulation grid step and reduce the simulation time. The quasistatic approximation additionally allows increasing the grid step along the longitudinal coordinate by the ratio of Rayleigh length to plasma wavelength, i.e., by another w₀/w_p times. However, in order to correctly simulate strongly depleted laser pulses, the time step of the grid must be further reduced by a factor of w₀/w_p. Thus, the quasistatic approximation combined with the envelope model for the laser pulse gives a w₀²/w_p² times speedup of calculations compared to simulations without simplifying assumptions by the particle-in-cell (PIC) method.

As an example, we consider the acceleration of electrons in a long plasma channel with a single drive laser pulse, the parameters of which correspond to the discussed project XCELS. The maximum electron energy reaches 91 GeV at a plasma density of 3 10¹⁵ cm^-3 and a bunch charge of 50 pC. The quasistatic code (LCODE) allows to simulate the driver evolution with high accuracy up to the termination of acceleration, which is confirmed by controlling the energy balance of the system. A typical run requires about 10³ CPU hours, and the use of the reduced models gives a speedup of 6 orders of magnitude compared to the traditional PIC method.