MICRESS Forum

I am interested in developing a phase field model to study the interface of a binary Cu-Ni weld/diffusion couple where the concentration transitions from 100% Cu to 100% Ni similar to what has been achieved in this paper: https://link.springer.com/article/10.10 ... 007-9340-1.

However, instead of having two simulations at 25% and 75% as they did, I am curious if MICRESS can be used with a sliding window to effectively dilute the interface so that the transition from Ni-Cu can be studied over a distance of 100 microns. Are there ways to do this within MICRESS? This would be very helpful for understanding both dissimilar welds and functionally graded materials.

Thanks,
Sean

Hi Sean,

I am not sure whether I understand well what you exactly want to achieve. Should the interface "dilute" by ongoing melting? And how do you want to match the thermal boundary condition with your sliding window?

In principle, it is possible in MICRESS to combine the use of the co-moving frame with the 1d-temperature approach, or alternatively use prescribed temperature profiles (like e.g. from a Rosenthal solution).

Bernd

Thanks for your quick response. The end goal of the simulation would look something like attached where one side of the simulation is 100% Cu and the other is 100%Ni with a thermal gradient as shown at the melting point of each material. For the initial model a linear cooling rate can be used. I have tried specifying compositions for the liquid and solid differently but then when the solid melts, the liquid moves to a uniform composition instead of exhibiting a concentration gradient.

Can the define composition "from file" be useful to achieve this? Alternatively with a sliding window, can that follow the prescribed thermal gradient while also introducing additional Ni into the system to achieve the desired compositional gradient?

Let me know if you would like some additional clarification.
Thanks,
Sean

Hi Sean,

now it is more clear.

You can create the initial concentration gradient, as you say, by reading an initial composition distribution from file. You can build such a file (ASCII, vtk) by whichever means (e.g. Excel), or also quite conveniently by DP_MICRESS (you use "Operations" to calculate the composition field from the temperature field output and afterwards export as ASCII or vtk).

In principle, with the setup you depicted, you can start from pure fcc and melting can begin from both sides simultaneously (the melting points are higher in the middle region). Of course, due to the large diffusion coefficients in LIQUID, the concentration gradient vanishes there (insulation boundary condition).

If the simulation domain would be to large that way, you can also do it single-sided. Then, you can use moving_frame and a 1d-extension to the concentration field in the liquid (you can read in an initial concentration gradient into this field, too). That means, liquid forms at the bottom, and the moving_frame is following the liquid front, eventually moving solid material from the 1d-extension into the simulation domain. It is a bit tricky to properly select the reference phase of the moving frame - there have been some changes on that recently, so that it may depend on the exact MICRESS version which you are using...

Bernd

Bernd