Issue with nucleation

[materialmoop]

Hello,

I am trying to simulate an austenite to martensite transformation. I have attached my input for reference. However, when I check the results, I don't think any nucleation is happening. Could anyone tell me why that might be? I am a beginner, so I'm sure I'm missing something obvious

Thanks

[Bernd]

Dear materialmoop,

Welcome to the MICRESS forum. I interpret your input file in the way that you want to model the martensite phase basically as a bcc-phase without diffusion, right? I think you should also account for the lower thermodynamic stability of martensite (as compared to bcc) by using a positive dG-offset on the 1/2 interaction.

As the liquid phase does not exist neither in your initial microstructure nor in the nucleation conditions, the phase interaction data for 0/1-interfaces (liqud-fcc) and the seed type 1 for nucleation of phase 1 from the liquid have no effect. This is not an error, but removal would simplify your input file.

Seed type 2 is designed to nucleate phase 2 on interfaces between phases 1 and 2, and thus is inactive as long as there is no phase 2 present. Given that the initial microstructure is pure fcc, and there are no further seed types, this means phase 2 cannot nucleate at all!

Furthermore, you are using "atc mob_corr" for Si in the 1/2-interaction, which means diffusion control. However, you have not defined diffusion of Si in any phase. Thus, even if phase 2 would nucleate, it could not move because the interface mobility for diffusion contolled growth is set to 0. Perhaps, you wanted to apply "atc mob_corr" to C instead which would make sense because C is a fast diffuser in fcc.

Best wishes
Bernd

[materialmoop]

Hi Bernd,

Thank you for your suggestions as they helped a lot! However, I still have a bit of the same problem. In addition, I am now getting an error related to diffusion (Micress error 22; perhaps my database is wrong?). Could it be that I also need an interaction with phase 0/phase 0? It appears as though the microstructure begins to change, but then goes back to austenite. Or, am I overlooking something else?

Many thanks

[Bernd]

Hi materialmoop,

Now you changed everything to the worse. You removed nucleation of phase 2, and you assigned the background phase 0 to BCC...

If you want to simulate a solid state transformation like fcc-->martensite (at least this is what I understood), you should not use phase 0 for one of the solid phases. The reason is that phase 0 is always isotropic and automatically present everywhere as background to the initial grain settings. Then, phase 0 is included but not used. A good template for that would be the Application "Example A004_Gamma_Alpha_TQ.dri".

However, as phase 0 is the background phase, you must provide that your initial microstructure completely overwrites the domain. If you want to start with an initial grain structure of phase 1 (fcc) only, you can do that (like in A004_Gamma_Alpha_TQ.dri) with deterministic initial grains using the Voronoi construction, or also by random initial microstructures. In any case, you must make sure that there is no remaining liquid phase present. This can be done by making the grain radii sufficiently large. Please check with DP_MICRESS before proceeding to anything else!

Then, when it comes to nucleation, you should have at least one seed_type for nucleation of phase 2 (on interfaces, triple junctions or in the bulk). It is important that the temperature interval and the checking interval of the seed_type are chosen such as to allow nucleation at the expected temperatures. The grain radius of the new phase (martensite) should be typically chosen as 0, which means that only a "small grain", i.e. one grid cell with a fraction of 2 times the minimum phase fraction, is set. If you chose large grain sizes (bigger than the grid size), you risk numerical problems, because the new phase automatically will assume the composition of the background (to avoid mass balance violation).

The interface thickness should be chosen to 2.5-4 cells, higher values unnecessarily reduce your effective resolution.

I will not be at work during the next week. Please note that answers from my side to follow-up questions will take longer.

Best wishes
Bernd