InterfaceKernels for Electrons

[Shane Keniley]

Hello all,

Since I am no longer having convergence issues in 2D, this should probably be a new thread! While running these 2D simulations, it seems like the concentration of electrons in the water layer increase much faster than electrons could possibly be leaving the gas phase. If the density of electrons in the gas phase is only 10^11 m^-3, how could I possibly be seeing 10^17 m^-3 electrons in the liquid phase? I know they'll build up, but it just seems like more electrons are entering the water phase than should be possible.

So my question is about the relationship between BCs on one side of an interface and InterfaceKernels on the other. Unless I'm misunderstanding InterfaceKernels, right now InterfaceAdvection is essentially a kind of "do nothing BC"; e.g. electrons pass through the interface freely, and since the transport coefficients of electrons in water are dramatically lower than in a gas, they just build up at the surface. This makes sense to me. But we also apply the HagelaarElectronBC to the electrons at the interface on the gas side, which includes the possibility of reflection along with a form of diffusion based on thermal velocity rather than the gradient of the density. If that's the case, shouldn't the InterfaceAdvection kernel have the same form as the BC on the gas side? Otherwise we are applying two different conditions on either side of the boundary. As an example, if we allow electron reflection on the gas side but allow pure advection on the water side, aren't we double counting those reflected electrons (they are both reflected and allowed to enter the water)?

I may be misunderstanding the relationship between BCs and InterfaceKernels, so please correct me if I'm wrong. 

[John Haase]

Shane,

I as to the physics question if I'm remembering correctly this paper has an estimate of densities (https://journals.aps.org/pre/abstract/10.1103/PhysRevE.95.053203)

[Alexander Lindsay]

Shane I encourage you to look at the mean_en_out.e file in the tests/1d_dc/gold directory. I've attached a plot of the electron current in the gas and liquid phases. You can see that even though the electron concentration is orders of magnitude greater in the liquid phase, the flux is still continuous at the interface, which is the goal of the interface kernels.

The HagelaarElectronBC sets the condition for what the gas phase electron flux needs to be at the interface. Then the interface kernels just set the liquid phase electron flux equal to that coming from the gas phase. If you open up the interface kernel source files, you'll see that they only apply residuals to one side of the interface (usually intended to be the liquid side).

--
You received this message because you are subscribed to the Google Groups "zapdos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to zapdos-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/zapdos-users/6078ed20-559e-4427-94af-1c60b627adf8%40googlegroups.com.