About the different components in diffusion couple

[tatalemon]

Hi, I am now simulating a diffusion couple which is composed of two equilibrium alloys, the interdiffusion process between the two alloys is the target of the simulation. The schematic diagram is shown as follows,
[img]

Fig.1.jpg (13.96 KiB) Viewed 5082 times

[/img]

I divide my simulation into two steps,
For the first step, the separate two equilibrium alloys were obtained via simulation of precipitation process.
For the second step, the interdiffusion process of the diffusion couple will be simulated.

Well, the problem is that the two alloys have different kinds of alloy elements and phases, respectively.
Alloy 1: Component：0, 1, 2 Phase: 1(matrix), 2(precipitation)

Alloy 2: Component : 0,1,2,3 Phase: 3(matrix), 4(precipitation)

However, the two alloys use the same thermodynamic database in MICRESS. I have included all the components and
phases in the database, then the composition for different phases need to be input in the script file,
as shown in the following,
...
# Initial concentrations
# ======================
# How shall initial concentrations be set?
# Options: input equilibrium from_file [phase number]
input
# Initial concentration of component 1 in phase 0 ? [at%]
0.0000
# Initial concentration of component 1 in phase 1 ? [at%]
15.0
# Initial concentration of component 1 in phase 2 ? [at%]
18.0
# Initial concentration of component 1 in phase 3 ? [at%]
9.0
# Initial concentration of component 1 in phase 4 ? [at%]
2.0
# Initial concentration of component 2 in phase 0 ? [at%]
0.0000
# Initial concentration of component 2 in phase 1 ? [at%]
11.0
# Initial concentration of component 2 in phase 2 ? [at%]
6.0
# Initial concentration of component 2 in phase 3 ? [at%]
15.0
# Initial concentration of component 2 in phase 4 ? [at%]
7.0
# Initial concentration of component 3 in phase 0 ? [at%]
0.0000
# Initial concentration of component 3 in phase 1 ? [at%]
0.0000
# Initial concentration of component 3 in phase 2 ? [at%]
0.0000
# Initial concentration of component 3 in phase 3 ? [at%]
12.0
# Initial concentration of component 3 in phase 4 ? [at%]
5.0
...
So my problem locates in the component 3, whose content is zero in phase 1 and phase 2.
Which will lead to the following unexpected result in the first step of my simulation (i.e. simulation of precipitation process) ,
“ Phase 2 disapeared at 100.5047 s “
The simulation can not continue, several groups of new interfacial kinetic data are tried , but it didn’t work. Unless a small value at least 5.0 in phase 1 and 2 were given, then the simulation can continue.

Another attempt has been performed , the same problem would appear.
Alloy 1: Component：0, 1, 2 Phase: 1(matrix)
Alloy 2: Component : 0,1,2,3 Phase: 2(matrix), 3(precipitation)

“ Phase 3 disapeared at 100.2067 s “

I guess that as long as there are different kinds of components in the two sides of the
diffusion couple, in other words, the composition of some components is set to be
zero (as the same database is used in MICRESS), then the problem would appear.

Do you have some good recommendation to resolve my confusion?
Hope for your answer!

[Bernd]

Dear tatalemon,
what you are doing is a quite challenging task. To better understand your approach and to be able to analyse which are the specific problems in your case, I have two fundamental questions:

1.) Which is the way you join your diffusion couple? Do you join the MICRESS results of the first step as ASCII files and read them in for the diffusion couple simulation?
2.) Are the two microstructures which you join built up of different phases, or are perhaps phase 1 and 3 (and also phase 2 and 4 respectively) identical?

Generally, if the component 3 has solubility in phases 1 and 2, you should never put the composition to 0 as this may lead to numerical issues in the Thermo-Calc subroutines. Instead you should put a small value like 1.E-3 at%.

Bernd

[tatalemon]

Dear Bernd,

1) I join my diffusion couple by inputting two grains in the script file,
firstly, one big rectangular grain is inputted, then the other smaller rectangular
grain is inputted, the latter partially covers the former one. Thus there exists
a interface between them, and the two grains just represent the two alloys which
make up the diffusion couple together.

...
# Grain input
# ===========
# Type of grain positioning?
# Options: deterministic random from_file
deterministic
# NB: the origin of coordinate system is the bottom left-hand corner,
# all points within the simulation domain having positive coordinates.
# Number of grains at the beginning?
2
...

2) The two microstructures which I join are built up of different phases.
3) I have put the composition of component 3 from 0 at.% , 1e-3 at.%, 1.0 at.% ......until 4.0 at.%, but
the same problem always appears "Phase 2 disapeared at 100.5047 s" , until I put it as 5.0 at.% or more , it will work well.

I really don't know what the problem is, looking forward to your answers.

best regards,
tatalemon

[Bernd]

Ok,

that means phases 0, 2 and 4 are not yet existing, and there is a phase boundary between the two sides of the diffusion couple.
But this means also, that the phase boundary may move if the two sides are not (yet) at the equilibrium compositions. And then, phase one or two may vanish, right?

My questions are:
- Do you expect to be at equilibrium without component three? Did you try this with a simplified database with two elements and two phases or with Thermo-Calc?
- If there are precipitations at at later stage, they will need component 3 - so there must be a solubility of this element in phases 1 and 3, and component 3 cannot be 0...
- Which boundary conditions have you chosen for the simulation domain?

Perhaps, it would be easier to understand the problem if I'd know which elements and phases you are using...

Bernd

[tatalemon]

Hi Bernd,

Sorry for late reply, I have forgetten to tell you that the interface mobility of the phase boundary between the two sides of the diffusion couple

has been set to be a very small value, such as 1e-18, in the first step simulation for precipitation process, which is far more smaller than the interface mobility

between the phase 1 and 2 or 3 and 4,i.e., the phase interaction between phase 1/3,1/4, 2/3 and 2/4 has been set to be very small. So the phase boundary

will not move in the first period of simulation for precipitation process, then when both sides of the diffusion couple reach equilibrium compositions, we start the

second period of simulation. for interdiffusion between the diffusion couple, the phase interaction between phase 1/3,1/4, 2/3 and 2/4 will be set to an

appropriate value.

Furthermore, I have tried a simplified simulation with two elements and two phases as you suggest, the two single phase makes up the two sides of the diffusion

couple, respectively. Precipition is not included. That works well.

However, when I tried to simulate the precipitation process in one of the diffusion couple. I encountered another problem, just shown as in the following picture:

fig.2.jpg (31.37 KiB) Viewed 5060 times

If I use the above mentioned method, it is difficult to distinguish the phase 1 in the two sides of diffusion couple in MICRESS. Then my simulation can not continue.
Do you have some good solutions to this problem and also previous problems for different phases and different components ?

Sincerely hope for your answer.
tatalemon

[Bernd]

Dear tatalemon,

let me try once more to explain what I understood from the information which you have given:

1.) You aim is to simulate the diffusion couple made up of two sides:
side 1: two phase microstructure of phase 2 in phase 1
side 2: two phase microstructure of phase 4 in phase 3

2.) You want to obtain the two sides by microstructure simulation of precipitation of phase 2 in 1 and phase 4 in 3, respectively. Instead of "stiching" together the results of two individual simulation, you want to simulate both "in one" by separating the two side using a mobility of 0.

3.) In the second step, after each side is at its own equilibrium, you want to put a finite mobility between the two matrix phases (1 and 3) and simulate the diffusion couple.

Is this correct?

One problem which might occur is that once there is a diffuse interface between phases 1 and 3, there will be diffusion through the interface even when the mobility is set to 0! What can be done to avoid that is to set the number of initializations of the interface (time input data) to zero.
The other problem of setting initial compositions of component 3 to zero on one side I already mentioned - you should use a small value instead!
Apart from these two problems the approach should work!

I do not understand your problem with the "simplified simulation" with two phases and two elements. What do you mean with

"it is difficult to distinguish the phase 1 in the two sides of diffusion couple in MICRESS. Then my simulation can not continue."?

Bernd

[tatalemon]

Hi Bernd,

haha, You are right, my method is just as you have understood.

I just do as you suggest to avoid the two problems you have mentioned, it now works well.

As for the "simplified simulation" , it has been a mistake I have made, I have correctted.

Thanks a lot for your kind help~

I am interested in the method you have mentioned by " "stiching" together the results of two individual simulation"

and "join the MICRESS results of the first step as ASCII files and read them in for the diffusion couple simulation"

Could you please tell me further details about it or give me some references ? How to realize it ?

Thanks in advance.

tatalemon

[Bernd]

Hi tatalemon,

I think in the case that both microstructures are formed under identical conditions, and phase 1 and phase 3 are not identical, your method is probably the best.

"Stiching" microstructures would mean the following procedure:
1.) Make individual simulations with the same set of components
2.) The Micress result files are transformed to vtk or ascii format. The respective .phas outputs and the .conc* outputs are "stiched" in z direction by simple concatenation (of course, the header has to be corrected in case of vtk files)
3.) The combined files are read in into a larger MICRESS domain, the diffuse interfaces are recovered by "initial phase-field iterations to match the diffuse concentration fields inside each side.

Advantages:
- microstructures can be simulated under different conditions
- the two matrix phases (phase 1 and 3) can be identical but with different composition
Disadvantages:
- By using the .phas output to reconstruct the grain and phase structure, fine structures like small grains are lost
- As the composition fields are "diffuse" and the .phas is "sharp", even when using initialisation steps, slight inconsistencies could arise which could lead to numerical issues in case of phases with restricted solubility.

With the next MICRESS version 6.2, a further method of joining simulated microstructures will be available: Using a modified restart functionality it will be possible to replace initially defined grains by complete microstructures from .rest files, if the simulations for creating the .rest files have been done with the same number of element and phases. Thus, it will be possible not only to join several different results, but also to "fill" them into arbitrary shapes!

Bernd

[tatalemon]

Dear Bernd,

Thanks for your answers,

I think I will have a try with your method,

It is really nice to talk with you.

Thanks again for your help.

regards,

tatalemon