Discussion Forum

I am currently simulating the mass-transfer limited dissolution of a metal substrate through an opening in an insulating mask. Think etching through a patterned photoresist film. A very similar problem has already been solved using COMSOL (see www.comsol.com/paper/primary-current-distribution-model-for-electrochemical-etching-of-silicon-throug-26101).

This problem can be treated in a simplified manner in 2D by combining the convection-diffusion equation (cdeq) with the moving mesh (ale) node. Only the liquid domain between two surfaces of constant concentration (c=1 at metal surface, c=0 at diffusion layer boundary) is modelled. This domain extends underneath the insulating layer and with a velocity proportional to the the normal gradient of the concentration field at the metal surface.

I can solve this problem and plot the results for different time steps. However, I would like to obtain the results not for specific time steps but rather for a specific displacement at a chosen boundary point. For example, I would like to plot the results for every 1% of displacement in y-direction at the cavity center. I have solved this problem previously in MATLAB using a boundary element method, where implementing this step is straightforward. Is it possible to adjust the time step during the simulation according to some criterion? To make the time step a variable which is calculated at each step according to a user-defined formula? Or is what I want to do only possible in post-processing? Or not at all?

Edit: I have attached a minimum working example of the type of problem I would like to solve. The 2D geometry consists of a circle surrounded by a square. The concentration is zero along the square and unity along the circle. The circle shrinks with a prescribed normal mesh velocity of kF*(ux*nx+uy*ny), where kF is just a freely chosen proportionality factor. Using a global variable probe I can track the position of the point at the top of the circle over time, and hence calculate the dimensionless radius at this position. What I would like to do is compute shape profiles not over time, but rather for different values of the dimensionless radius.

Grateful for any advice on this topic!