Self-heating at high densities

[DanRRRR]

I remember Stuart supported a version of Epoch code which somehow handled the problems associated with high densities. Was the method used there similar to the one in the article Y.Sentoku, A.Kemp, Journal of Computational Physics V.227 p.6846 (2008) and did the method really help with self-heating ?

[Status-Mirror]

Hi Dan,

Apologies for the delay. Are you referring to my hybrid EPOCH code, described in Morris et al (2021)? This code was used to model hot electron propagation in cold, dense targets. We avoid self-heating by replacing the macro-particles of the cold, dense solid with a background fluid - it is the motion of these background macro-particles which drives self-heating. Additional equations are used to describe the current response of the background as hot-electron macro-particles move through it.

The downside of the hybrid-PIC code is that because we use a field-solver which assumes a background fluid, we can no longer model vacuum laser propagation, or the interaction of a laser-pulse with a pre-plasma. Instead, we are forced to inject hot electrons based on the behaviour expected after interaction with a laser.

The "reduced PIC" method of Sentoku and Kemp seems like a method half-way between my hybrid-PIC code and a regular PIC simulation. Their background is described using both normal macro-particles, and special ones which don't feel the fields, acting only collisionally. The combined density of the two macro-particle types is equal to the physical density. The normal macro-particle population is kept at a density high enough to resolve the kinetic effects, but low enough to ensure self-heating doesn't run away. This idea has also been used by in a paper by Wu et al (2018), where they call it a "layered density" method.

So yes, similar but different. Sentoku and Wu replace their cold background with a population of macro-particles which interact only collisionally, while my hybrid code replaces the background with a fluid model.

Hope this helps,
Stuart

[DanRRRR]

Thanks Stuart,

What specific density and speed dependent quantity in EPOCH drives and defines kinetic effects characteristic lifetime and hence the self-heating rate and time step? If this lifetime becomes too fast compared to the simulation time it has no sense to divide the timestep to resolve its characteristic time and is just wise to restrict it on some reasonable value in the places where it is too fast. It will succeed to relax processes to their steady-state values during simulation but just not unreasonably fast. This is often used method in plasma and atomic physics.

This method will not affect pre-plasma, as it has lower density and there kinetic effects lifetime will remain unchanged from its true value.

Do you plan to implement this (Wu-Kemp-Sentoku have done the same) method in EPOCH? Would be great. Otherwise standard model at high densities requires a lot of particles per cell and huge number of cells both making simulations time and resource consuming.

Another possible solution would be to use implicit scheme like in PICNIC (open source PIC code on GIT) . Have you considered this? That would potentially give 1-2 orders speedup of simulations. Other PIC codes currently considering to implement this and the EPOCH could appear late on their holiday of life