Interfacial energy, morphology and triple junction angle

Exchange about the physics background, diffuse interface theory, etc..
deepumaj1
Posts: 104
Joined: Fri Feb 13, 2015 11:00 am
anti_bot: 333
Contact:

Interfacial energy, morphology and triple junction angle

Post by deepumaj1 » Fri Apr 22, 2016 2:33 am

Hi all,
I was trying to model a gamma-alpha simulation in which I tried to tune up the simulation with experimental results(final ferrite volume fraction). By adjusting the mobility, I'm able to reach the targets, but I'm concerned about the interfacial energy values used. Should the interfacial energy value of gamma-alpha be greater than gamma-gamma and alpha-alpha? And also, should the triple junction angle inside the newly nucleated ferrite grain be greater than 120 degree?

Thanks in advance,
Deepu

Bernd
Posts: 1144
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interfacial energy, morphology and triple junction angle

Post by Bernd » Fri Apr 22, 2016 6:46 pm

Dear Deepu,

gamma-alpha transformation is much more complicated as one may think at first look. The most important role of the interfacial energy, in my understanding, is the question whether there will be wetting of the austenite grain boundaries by ferrite, which has a tremendeous impact not only on morphology but also on the transformation kinetics. If the gamma-alpha phase boundary energy is smaller than the energy of the austenite grain boundaries, wetting should be observed. Whether this happens in reality or not can only be checked by comparison to experiments.

Calibrating the final ferrite fraction by adjusting the interface mobility - in my opinion - is a bit indirect and not fully satisfactory, especially if you assume the system to be equilibrated with respect to carbon diffusion. Alternatively, you should try to use the nple model for redistribution. Assuming the correct redistribution behaviour of the substitutional elements is important to get realistic transformation kinetics. Perhaps, you could get the correct final ferrite fraction automatically, with a sound physical model...

mmendes
Posts: 8
Joined: Sat Nov 19, 2016 12:35 am
anti_bot: 333

Re: Interfacial energy, morphology and triple junction angle

Post by mmendes » Fri Jan 06, 2017 7:54 pm

Hi there,

I am new in the group and I am trying to simulate allotriomorphic ferrite along austenite grain boundaries. I have tried two paths:
1st) using the categorization of alpha grains (first image) and 2nd) decreasing the interfacial energy in order to get a wetting scenario (second image).
With the first option, I couldn't get an "even" thickness of the film in all the boundaries. On the other hand, when I use the second option, the categorization somehow does not work and I always have the formation of interfaces between the alpha grains (as you can see in attachment). I can only get rid of those interfaces if I set the maximum number of nuclei to 1. In this case, despite I get a uniform film in all the boundaries as I want, the fact that there is only 1 nuclei seems physically incorrect to me.
So my questions are: 1) Is there any way to use the categorization in the wetting scenario (with a gamma-alpha interfacial energy lower than the gamma-gamma / alpha-alpha ones) so that I can get a uniform film without the need to place only one alpha nuclei?
2) How can I get a thinner and uniform film along the boundaries if I use the 1st path? I tried to decrease the shield distance, but at some point the categorization was not happening anymore.

Thank you in advance for your help!
Attachments
Allotriomorphic ferrite.jpg
Allotriomorphic ferrite.jpg (300.23 KiB) Viewed 1915 times

Bernd
Posts: 1144
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interfacial energy, morphology and triple junction angle

Post by Bernd » Sat Jan 07, 2017 8:08 pm

Dear mmendes,

Welcome to the MICRESS forum!

"Categorization" means assigning two or more individual grains to the same grain number. In the current MICRESS implementation there are two important requirements for categorization of existing grains: Firstly, the grains must have identical properties, and secondly, they must not touch each other.

The first requirement is a natural one. When setting grains with random orientation, casting them into categories allows for creating groups of grains which afterwards can be "categorized" so that they afterwards behave like one grain. One important restriction here is that new grain after nucleation are treated as so-called "small" grains until they reach full size and subsequently can be treated by the normal phase-field algorithm. This means they have extra properties like their size which prevent from "categorization". Thus, categorization works only after they got full size (or they are forced to be normal grains, see below).
In your "wetting" scenario; I guess, they never reach full size but rather grow in diffuse form along the interface.

The second requirement is a technical one: Once grains have formed interfaces between each other, it is very difficult to categorize them because the already existing interfaces would have to be removed. Thus, setting grains too dense prevents categorization because they touch before getting "big" grains.

From the above said, several approaches can be deducted:
1.) If in the wetting scenario resolution is high enough or the relation of interfacial energies is chosen accordingly such that the ferrite grains can reach full size (at least one cell needs to have phase fraction 1.0), they can be categorized and will not form grain boundaries when growing together.
2.) Using the option "kill_metastable" in the nucleation input, grain are forced to get "big" after the end of the shield time. Then, they can be categorized even if they are have not yet reached full size.
3.) Using the option "add_to_grain" with sub-option "new_set", it is possible to directly nucleate several instances of the same grain. The disadvantage of this option is that the "small grain" model is not working correctly, so that nucleation may be a bit more difficult. Furthermore, the first one typically grows a bit faster than the subsequent instances.

My personal preference would be method 3 if it works for you. Then, you are free to use both of your scenarios without need of categorization. Method 2 should also be quite good when choosing a good value for the shield time.

Good luck
Bernd

mmendes
Posts: 8
Joined: Sat Nov 19, 2016 12:35 am
anti_bot: 333

Re: Interfacial energy, morphology and triple junction angle

Post by mmendes » Tue Jan 10, 2017 1:32 am

Dear Bernd,

First of all, thank you so much for you prompt and clear reply!
I checked the .log file of my simulation in the "wetting scenario", and the alpha grains do reach a full size. However, by checking the .korn file I saw that the grains touch each other before the full size to be reached. So according to your explanation I assume that this is the reason why the categorization cannot happen. Then I tried to do your suggestions number 2 and 3.
2.) By using "kill_metastable" I almost got what I want. I could see that some categorization in fact happened and some alpha grains merged. However, I don't know why the grains were categorized to 3 different grain numbers. As a consequence, I still have the interfaces between them. I thought that the problem was with the gamma grains orientation, but after increasing the gamma_alpha interfacial energy (in order to no longer stay in the wetting scenario), all the grains were categorized to the same grain number. How the grain number for categorization is chosed by the code? And how can I get the categorization to occur completely (i.e. to have only 1 grain number) in my wetting scenario?
3.) With this option I could get the morphology that I want, but even though I set the maximum number of nuclei to 10, when I open the .phas file there is a nucleation of only one nuclei. Therefore, with this "add_to_grain" option I got the same behavior as when I just set the maximum number of nuclei to 1 without using this option. Does this "add_to_grain" always "overwrite" the nucleation information in the next command lines? And how can I get several instances of the same grain in order not to have only 1 nuclei?

Thank you! :)

Bernd
Posts: 1144
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interfacial energy, morphology and triple junction angle

Post by Bernd » Tue Jan 10, 2017 12:45 pm

Dear mmendes,

@2) Unfortunately, the "categorize" algorithm is not complete. When we implemented it, the main purpose was to reduce the overall number of grains and thus to reduce memory usage and increase performance. Completeness was no issue. In fact, "categorizing" most of the grains is technically much easier than doing it for really all of them. Apart from the limitation that grains should not touch, there is a second one: If both grains are connected to a third grain which has a (internal) grain number which is between the numbers of the two grains, they are also not categorized. The reason is that the order of the numbering would change what causes technical trouble. Perhaps this was the case for you.

@3) The above said is the main reason why I proposed method 3. From what you say I guess that here all 10 grains are nucleated but do not grow. You can see it from the messages in the Screen or .log output, or you can activate "out_nucleation" in order to get an extra output at nucleation time and check the .intf result (which shows interfaces). The .phas output shows only those nuclei which reach a majority phase fraction.
Probably the last instance of the "new_set" is the one which really grows (I mentioned this higher stability of the last one before) while the other ones are not stable enough. The reason is that the grain properties are overwritten always by the next one with identical grain number, so only the last one can make use of the "small grain" model, shields, etc.
There are tricks to increase stability in this situation, like e.g. to slightly increase the initial fraction of the seed. You can achieve that by increasing the initial radius to a value between 0 and Deltax (the fraction can be calculated as volume of the sphere with radius r divided by Deltax3).
The use of multiple instances of the same grain is still a new feature, and we are not sure about how to treat this more consistently in future...

Bernd

mmendes
Posts: 8
Joined: Sat Nov 19, 2016 12:35 am
anti_bot: 333

Re: Interfacial energy, morphology and triple junction angle

Post by mmendes » Thu Feb 02, 2017 3:39 am

Hi Bernd,

Thank you for helping.

By increasing the initial radius I could solve that problem ;) I would like to ask you some other questions:
1) It is possible to make the nuclei which appear in the triple junction to have different orientations along the 3 grain boundaries when it grows? This way my allotriomorph alpha grains will not merge at the triple junctions, which is what happens in reality. So far I could only get a "single film" which might turn into a single alpha grain given time. Whenever I try to place the nuclei on the grain boundaries, I cannot make them stop growing when they touch the triple junction. Therefore, the nuclei placed in a smaller GB will grow in my wetting scenario and will overtake the tiple junction, while the nuclei placed in a longer GB will be still growing there. Eventually, when those grains touch each other I have an interface in the middle of the GB, not in the triple junction. Do you have a suggestion on how I can deal with this problem?
2) I am also trying to simulate a needle-like alpha phase, but also without success so far. I started with a pretty simple domain, with one interface and 2 austenite grains, but somehow the ferrite grains grow equally to both sides. Even with a high resolution (650 cells and grid spacing 0.1) I couldn't get the needle shape. Do you know what I am doing wrong? Please, find below part of my input file.

# Data for phase 2:
# -----------------
# [identical phase number]
# Simulation of recrystallisation in phase 2?
# Options: recrystall no_recrystall [verbose|no_verbose]
no_recrystall
# Is phase 2 anisotrop?
# Options: isotropic anisotropic faceted antifaceted
antifaceted
# Crystal symmetry of the phase?
# Options: none cubic hexagonal tetragonal orthorhombic
none
# Number of type of facets in phase 2
1
# kin. anisotropy parameter Kappa?
# only one value for all facets/phases
# 0 < kappa <= 1
0.8000000
# Number of possible orientations of a facet 1
2
# 1 -th normal vector facet 1 ? 3*
1.000000
0.000000
0.000000
# 2 -th normal vector facet 1 ? 3*
1.000000
0.000000
1.000000
# Should grains of phase 2 be reduced to categories?
# Options: categorize no_categorize
no_categorize
#
# Orientation
# -----------
# How shall grain orientations be defined?
# Options: angle_2d euler_zxz angle_axis miller_indices quaternion
angle_2d
#

Thank you so much for your attention.

Bernd
Posts: 1144
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interfacial energy, morphology and triple junction angle

Post by Bernd » Thu Feb 02, 2017 9:03 pm

Dear mmendes,

I need some more information to answer your questions:

@1) What is the physical reason why the ferrite grains when growing along an interface should not cross the triple junction? Is it just anisotropy of the phase itself, or does it depend on the orientation relation to a specific austenite grain?

@2) With needles you mean ferrite grains which start from the interface and grow into the bulk of the austenite grains? And they should grow only into one austenite grain and not into the other?
If this is what you intend, then the issue consists of two different problems, namely first to create a needle-like morphology (where the antifaceted anisotropy could be one of the options), and second to restrict growth to one side only. For the second effect you probably would need to take orientation relations into consideration.

Bernd

mmendes
Posts: 8
Joined: Sat Nov 19, 2016 12:35 am
anti_bot: 333

Re: Interfacial energy, morphology and triple junction angle

Post by mmendes » Thu Feb 02, 2017 10:25 pm

Dear Bernd,

Thanks for answering.

1) It is because the orientation relationship between austenite grains must vary. So if the alpha grain cross the triple junction I will have the same alpha grain (with the same orientation) along 2 or 3 austenite grain boundaries which have different misorientations.
So far, my alpha grains are crossing the tiple junction and touching each other in the middle of the grain boundary. Then I end up having two alpha grains with different orientations in the same grain boundary, which is a bit unrealistic. So if I set the nuclei on the interface, is there a way that I can stop the grain to grow when it reaches the triple junction? And if I set the nuclei in the triple junction is there a way it grows assuming a different orientation in each grain boundary?

2) Exactly. The ferrite grains should grow only into one austenite grain. From what you said, I am not getting the needle shape because of the second issue that you mentioned (given that I am using the antifaceted anisotropy already). In this case, how can I take orientation relation into consideration? Could you be a bit more clear on where in the input file I should set this?

Thank you so much!

Bernd
Posts: 1144
Joined: Mon Jun 23, 2008 9:29 pm

Re: Interfacial energy, morphology and triple junction angle

Post by Bernd » Fri Feb 03, 2017 12:17 am

Dear mmendes,

the misorientation models in MICRESS allow taking into account the effect of the misorientation of different phases or grains on the corresponding interface mobility and energy. As soon as you define both phases of an interface (or the one phase which forms grain boundaries) as anisotropic, you will be asked in the phase interaction data input about misorientation model input. The simplest one is a step model which only distinguishes between small angle and high angle boundaries with a simple factor on interfacial energy and/or interface mobility. With that one and the "parent_relation" option of the nucleation input it is quite simple to favor growth of a ferrite needle into one grain (with parent relation of 0°) compared to the other one which is has a high angle boundary.
For ferrite growth along the interface, it is more complex, because an austenite-austenite grain boundary with a certain misorientation is replaced by two austenite-ferrite interfaces which have different misorientations. So, wetting would be specific for different austenite grain boundaries.
I guess that both effects, i.e. growth of ferrite along the grain boundary and needle growth inside the austenite, appear under the same conditions. To find out how this exactly works it would be interesting to check with the experimental data which orientation relations are selected and lead to the effects you describe.

Bernd

Post Reply