Ni-Al precipitation simulations

solid-solid phase transformations, influence of stresses and strains
Post Reply
WTMuser
Posts: 31
Joined: Thu Jan 26, 2012 9:33 am
anti_bot: 333

Ni-Al precipitation simulations

Post by WTMuser » Thu Dec 05, 2013 3:01 pm

Hello,

I am simulating a simple model example describing the growth / dissolution of gamma prime phase particles (Ni3Al) in a single fcc-phase matrix (Ni-Al) in the simple Ni-Al system. The thermodynamic database is TTNi8 and the kinetic is MobNi1. As a test I am defining the original microstructure as shown in the graphs. The gamma prime phase composition is chosen to be very close to the gamma / gamma prime equilibrium (12.1 wt-% Al in gamma prime) at 900 °C (the simulation temperature). In the matrix phase the equilibrium composition would be 6.88 wt-% Al. I am choosing 3 wt-% in the simulation and I would assume that in this situation the gamma prime particles dissolve until the matrix composition has risen to the equilibrium value. But in the simulation there is practically no change of gamma prime phase fraction. And the profiles (the diagrams) suggest that the interface compositions on the matrix side are very close to 3.5 wt-%, which is far away from the equilibrium given by ThermoCalc? So obviously in the simulation there is no driving force to dissolve gamma prime. Which type of equilibrium exists at the interface here?

Another question is why in the interface the phase field parameter is -1. I would expect a gradual change between 0 and 1?

Thank you very much for your comments!

Best regards,
Ralf
Attachments
NiAl_equilibrium.png
ThermoCalc equilibrium calculation (database TTNi8).
NiAl_equilibrium.png (16.75 KiB) Viewed 4961 times
NiAl_cross_section.png
Phase field parameter and concentration profile of a gamma prime particle in the fcc-matrix after 2000 seconds at 900 °C.
NiAl_cross_section.png (26.57 KiB) Viewed 4961 times
NiAl_conc1_mcr.png
Overview of the Ni-Al simulation domain after 2000 seconds at 900 °C. Visible are the gamma prime particles within the fcc-matrix.
NiAl_conc1_mcr.png (28.29 KiB) Viewed 4961 times

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Ni-Al precipitation simulations

Post by Bernd » Fri Dec 06, 2013 7:14 pm

Hi WTMuser,

there are several ways to find out which equilibrium exists at the interface:

1.) Check the initial equilibrium which is given in the .log file at the end of the initialisation part. There you will see the temperature (which should be 1173K) as well as the equilibrium compositions and the deviation from equilibrium (deltaG). Please check whether the temperature for initial equilibrium was set correctly!
2.) During runtime, you can see (representative) results of the relinearisation in .TabLin. In case you did not request any relinearisation (updating) of thermodynamic data during runtime, the values from 1.) will be found there.
3.) You can check the local equilibrium compositions in the .c1pha1 and .c1pha2 outputs. When comparing to the Thermo-Calc results, please check the composition units (at%, wt%).
4.) You can check the driving force in the .driv output.
5.) If there is a substantial driving force but no front movement, check the interface mobility in the .mueS output. There could be a reduction of the mobility e.g. when you specify a high value for the minimum time step (time input section).
6.) If there is no substantial driving force and the equilibrium data seem correct, the diffusion coefficient could be too small or you did just not wait long enough.

These are the points I would look at first.

What you call "phase-field" parameter is probably the .phas output - this is not the phase-field parameter but gives only the phase number and -1 at the interface. The phase-field parameters should be taken from the .frac1 and .frac2 output in your case.

Best wishes
Bernd

WTMuser
Posts: 31
Joined: Thu Jan 26, 2012 9:33 am
anti_bot: 333

Re: Ni-Al precipitation simulations

Post by WTMuser » Tue Jan 14, 2014 3:11 pm

Hi Bernd,

your hints were useful, thank you! When I go on, I would like to be able to simulate the same setup (i.e. a given concentration field) heating it up from very low temperatures. This is for studying the behaviour during the beginning of a heat treatment. However, if a phase interaction between gamma and gamma prime phase is activated the gamma prime would strongly grow at low temperatures because the gamma matrix is supersaturated. This is unwished. I tried to prevent this by setting to no phase interaction between gamma and gamma prime phase, however then I get after some simulation time negative concentrations at the interface. Is this the correct way to make a precipitate phase in the simulation"inert", i.e. just to have it in the domain without changing its size or concentration?

Best regards,
Ralf

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Ni-Al precipitation simulations

Post by Bernd » Wed Jan 15, 2014 9:13 am

Hi WTMuser,

I hope you had a nice start into 2014!

Using "no_phase_interaction" is not the best choice, because I guess you want \gamma' to grow once you reach higher temperatures. In case of no_phase_interaction", the so-called "constant-K" method is applied to allow diffusion through the interface. This can give wrong results (although no negative concentrations, maybe the initial equilibrium is wrong...).

My interpretation why you don't want to allow \gamma' to grow at low temperatures is that it would not happen due to the low diffusivities at these temperature. In principle, this should come out automatically in the simulation, but at low temperatures, the diffusion length of the elements is much smaller than the interface thickness and numerical problems are likely to occur. In my opinion, the best solution is to use a temperature-dependent mobility which allows you to prevent the numerical issues at low temperatures or also to just suppress growth at low temperatures.

Another difficulty which could occur in your case is to a get proper initial equilibrium. Please check in the .log output whether the compositions of the two phases are reasonable. If necessary, put the temperature for the calculation of the initial equilibrium to a higher value.

Best wishes

Bernd

WTMuser
Posts: 31
Joined: Thu Jan 26, 2012 9:33 am
anti_bot: 333

Re: Ni-Al precipitation simulations

Post by WTMuser » Wed Jan 15, 2014 12:19 pm

Hello Bernd,

in fact it seems to be the best solution to use a temperature dependent mobility, then the test simulation runs stable as it should. Our goal is to suppress secondary gamma prime phase precipitation below its gamma prime solvus temperature. This would occur because our input concentration set contains in the matrix actually a mixture of gamma and (secondary) gamma prime because the small cubes cannot be resolved with these measurements.

Best regards,
Ralf

Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Ni-Al precipitation simulations

Post by Bernd » Thu Jan 16, 2014 5:22 pm

Oh, I understand! That means that in reality, you have a structure consisting of the big \gamma' cubes and a mixture of \gamma and fine \gamma' which formed during quenching. And as you use the mixture composition of the latter two for the \gamma matrix in the simulation, all growth kinetics would be wrong as long as you are below the \gamma' solvus temperature.
But, why do you want to simulate the heating process up to \gamma' solvus at all, if you anyway try to achieve that nothing happens?

Post Reply