Picongpu: Field-Ionization: Damp E-Field

[ax3l]

For field ionization, we should try to estimate the local energy invested in a field ionization step.

With that, we can initially try to simply damp the components of the local electric field in order to reduce energy.

This approach will not satisfy a divergence free j equivalent automatically and needs to be verified for each specific example with our charge conservation and local energy plugins. It's likely that for some regions of highly depleted laser of plasma waves this causes more troubles by additional noise (and fake-capacitor-slots) then it helps, but it might be a good approximation for damping the overall field energy appropriately in order to get more realistic penetration depths of ionization fronts.

One can further correct the introduced violation of Gauss Law with charge correction schemes.

ccing @HighIander @n01r

Implementation Concept

• during ionization, accumulate energy in a shared variable for local gridpoint
• after ionization of a frame list, save the "invested" energy to a new FieldTmp slot
• optional: after ionization, smooth the FieldTmp slot with "local ionization energy"
• reduce local E-Field (reduce components proportionally to reduce abs of vector without changing pointing) with energy in FieldTmp slot (just 0.5 eps_0 E^2 * cell vol.)

[n01r]

As a first approach we could just take the ionization energy of the ion and add it to the local FieldTmp value if ionization occurs.

Secondly I thought that for tunneling ionization it might be better to use the value of dE = (ionization energy - barrier height). But as we approach barrier suppression that would mean dE --> 0 which does not sound reasonable.

[n01r]

At the moment, we have a student intern, @jtrojok, who is looking into different papers and implementations of the so-called ionization current, the method that is usually employed to impose energy conservation.

We will then have an internal developer meeting at the end of this week to talk about the implementation.

[n01r]

So, @BrianMarre, perhaps this is connected to your master's thesis and it could be a good first contribution to PIConGPU for you.

[n01r]

This issue will be closed with the merging of #3355