PF simulation of Eutectic solidification

dendritic solidification, eutectics, peritectics,....
omid
Posts: 60
Joined: Thu Oct 01, 2015 11:19 am
anti_bot: 333

PF simulation of Eutectic solidification

Post by omid » Tue Oct 20, 2015 11:04 am

Hello there

I am totally new to the Micress and am trying to find out how to deal with this tool through manuals, examples and forum topics. I already have bunch of problems and hope someone can help me!
I want to model the solidification process of Fe-C eutectic alloy. I need some distinct data like Diff.-coefficient Prefactor for diffusion of element 1 in phase 1? or its activation energy! :shock: I have not found such a specific data in literature! Does anyone know how can I find them? Actually FeC is a well known and widely used alloy, in the case when I am supposed to model a strange alloy which has a very narrow application, what should I do? :cry:
Furthermore, I am trying to use GES file which I myself have generated it in thermocalc, and have pasted it in the GES library, but whenever I try to follow a file like AlCu_dri, as soon as it tries to read the GES file, it says there is no such a file or Thermo-Calc coupling not available :? I need an emergency hand to overcome these challenges!
Thanks in advance!

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: PF simulation of Eutectic solidification

Post by Bernd » Tue Oct 20, 2015 5:27 pm

Dear Omid,

welcome to the MICRESS Forum.

Please don't worry, the problems you describe can be easily solved. Of course, it would be much easier if you had the opportunity to have a MICRESS training course before starting work. But we will try to help you as much as possible.

Generally, I would advise you always to start from an already existing example when you build up your problem. This could be e.g. the AlCu_dri for directional solidification which you already mentioned in your post.
When you need to specify diffusion data, there are different options how to do this. In case of binary Fe-C only one single term exists per phase. You can specify either an Arrhenius description (like in AlCu_dri), read them from an ASCII file as function of temperature, or read obtain them completely from a mobility database. The choice depends on whether and in which form you have literature data available, and whether there is a mobility database for the system in quest. If you just know one single diffusion coefficient value from literature (which should be easy to find for bcc and fcc, but not for the melt in case of Fe-C), you simply specify it as "Prefactor" and set the Activation energy to 0 (like it has been done in the AlCu_dri example). If you know an Arrhenius description (i.e. diffusion coefficient plus temperature dependency) from literature, you can input both, Prefactor and Activation energy.
If you have a diffusion database like MOBFE2 available, the simplest way is just to use these data. This has e.g. been done in Gamma_AlphaTQ_dri and also works for complex steels with many components. Please note that in case of a binary system the keywords "multi" and "diagonal" are the same because there is only one single diffusion term per phase. Furthermore, for the liquid phase, the MOBFE2 like other diffusion databases only contains a dummy value of 1.0E-6 cm2/s, because there are no experimental data available. Your diffusion data input could look like that:

# How shall diffusion of component 1 in phase 0 be solved?
diagonal d
# Diff.-coefficient: (in case you prefer another value than 1.E-5cm2/sec)
# Prefactor? (real) [cm**2/s]
2.00000E-05
# Activation energy? (real) [J/mol]
0.0000
# How shall diffusion of component 1 in phase 1 be solved?
diagonal g
# How shall diffusion of component 1 in phase 2 be solved?
diagonal g
#
# How shall the interval for updating diffusion coefficients
# data be set?
# Options: constant from_file
constant
# Interval for updating diffusion coefficients data? [s]
5.0000
#

With the identifier "g" you get an Arrhenius description which is used "globally", i.e. without extrapolation of the concentration dependency. In case you want to use diffusion coefficients from database, please take care to append mobility data to the .ges5 file when creating it using Thermo-Calc. You can use the script "HOWTO_FeCMn.TCM" from the Examples directory and adapt it for the used databases, elements and phases.
This .ges5 file should be correctly found when you run MICRESS if you use a Thermo-Calc coupled version of MICRESS (MICRESS_TQ8_x64.exe, in case you use Windows) and if the path to this file is specified correctly in the driving file (MICRESS input file). If you put the .ges5 file to the GES_Files directory (where also the AlCU.ges5 is located), and replace AlCU by the new name of your file (e.g. FeC) it should work. If you get the message that the file is not found, probably the path is wrong. The path can be given either absolute or relative starting from the directory where the _dri.txt file is located. The path is case-sensitive.
If you get the message that Thermo-Calc coupling is not available, you probably used a no_TQ-executable...

Please tell me whether you succeeded to start your MICRESS simulation or whether you need more help!

Best wishes and good luck

Bernd

omid
Posts: 60
Joined: Thu Oct 01, 2015 11:19 am
anti_bot: 333

Re: PF simulation of Eutectic solidification

Post by omid » Wed Oct 21, 2015 10:45 am

Dear Bernd

Thanks for the warm welcome and the comprehensive reply. You are right, it's much better to have a training course, however I already have registered in one and am waiting for the course! ;)
I have tried a combination of AlCu and TQ_Eutectic to see what happens! I adjusted the number and position of nuclei and disabled the further nucleation, at least it showed me a directional growth! :P I have changed the data in "# Concentration data" based on your advise, but I still have some question about needed materials for the "# Phase interaction data" part! For example I am supposed to adjust the surface energy and Kinetic coefficient mu between phases ! do you have any advise in this regard? Where can I find such a distinct data like surface energy between phases liquid and geraphite? :shock: :shock: :shock:
Moreover as I run my setup, it goes ahead and tries to perform the simulation but I get these errors:
Grain number = 29
Updating of diffusion data from database...
Thermo-Calc error 1611 MICRESS error 22 phases 1/ 1
Thermo-Calc error 1611 MICRESS error 22 phases 1/ 1
Thermo-Calc error 1611 MICRESS error 22 phases 1/ 1

this error does not force the running to stop, but there is another strange error:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Warning in spfAddPhInkr, t=0.1042344E-02
Wrong fractions at nTupelp= 112
sum =0.1797693+309
x,y,z = 71 1 920
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

The interesting point is although it's running but in the post processing file there is just the data for time step=0!!! It seems that it doesn't save any data :?
do you have any Idea what these errors are?

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: PF simulation of Eutectic solidification

Post by Bernd » Wed Oct 21, 2015 3:36 pm

Dear Omid,

nice to hear that you got the simulation running.

To find realistic data for interfacial energies is really a big problem. Although there have been a number of intents to measure interfacial energies for low-melting alloys, experimental values are still very difficult to find. For graphite, the problem is even worse, because the behavior of this phase (and probably also the surface energy) strongly depends on trace elements. In most cases, we estimate the values, or we calibrate them by comparing experimental microstructures with simulated ones. This can be done e.g. for the secondary arm spacing which depends on the interfacial energy between the liquid and the primary phase (unfortunately, one needs to know or assume to know the diffusion coefficients, which is also difficult to find for the liquid phase :( ). When I estimate, a typical start value is 1.E-5 J/cm2 for liquid/metal interfaces. For faceted interfaces or intermetallic phases I typically use by a factor of 10 higher values. Another source for obtaining interfacial energy values could be the "broken-bond" model or MD and ab initio simulations.

The kinetic coefficient in most cases is not physically motivated but is chosen such that diffusion limited growth is obtained. It then can be calibrated in the way which is described here.

Error 22 occurs when trying to calculate the diffusion coefficient of phase 1 from database. It is difficult to tell the reason, it could be a negative or otherwise extreme carbon composition or some problem with the consistency of the .ges5 file.

The warning about an extremely high fraction sum in an nTupel (Triple or higher junction) shows that the simulation is diverging completely. In such cases, it is important to check from the beginning where things start going wrong. Essentially it can be everything from a too high interface mobility to wrong diffusion data. Please check the initial linearisation data in the .log output (if the initial equilibrium is wrong, everything afterwords will fail). Start with a small interface mobility and increase step by step. The diffusion coefficients which have been calculated from database can be checked in the .diff output. If you don't succeed, please paste the input file in your next post, and I will check for "obvious" errors.

If a simulation is not running properly, it will typically not advance, and if the time for the first output is not reached, you cannot see anything in the post-processing with Display_MICRESS. It is therefor wise to chose a sufficiently small tab_log interval ("selection of the outputs"). Then you already get a lot of information via the tab_log related ASCII file output (.TabL, .TabP, .TabT, .TabTQ). This could be e.g. a very small time step value which does not let the simulation advance. It can also help to define an extra output at a very early stage.

Bernd

omid
Posts: 60
Joined: Thu Oct 01, 2015 11:19 am
anti_bot: 333

Re: PF simulation of Eutectic solidification

Post by omid » Wed Oct 28, 2015 10:10 am

Dear Bernd
I am trying to model a ternary eutectic solidification, and based on your advises I know how deal with surface energy between phases and mobility, However I have bunch of questions about my new model, which is also the main goal of my PhD work. As you said before, there are not the exact experimental data available for numerical parameters, but I have no Idea about the rational intervals for entropy of fusion for each component inside the other phases (I already have one liquid phase and three distinct solid phases). and also have no idea for linearisation of the diagram's data. I have tested the binary diagrams and all the needed information are clear and all questions like the slopes and concentration at the reference points are known. But I have already no vision about these questions in a ternary system! I guess I should extract the data of binary diagram when Micress asks me about the slope at a reference point in the section of phase interaction 1/2. Kindly have look at these questions and let me know your idea ;)
# Input of the phase diagram of phase 0 and phase 2:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear :D
# Temperature of reference point? [K]
????? :shock:
# Entropy of fusion between phase LIQUID and 2 ? [J/(cm**3 K)]
??????? :?
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
?????? :o
# Reference point 2: Concentration of component 1 in phase 2 ? [at%]
??????? :o
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
????? :o
# Reference point 2: Concentration of component 2 in phase 2 ? [at%]
?????? :shock:
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
??????? :shock:
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
?????????? :shock:
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
????? :shock:
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
?? :cry:

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: PF simulation of Eutectic solidification

Post by Bernd » Wed Oct 28, 2015 7:06 pm

Dear omid,

Obviously, you tried to simulate a ternary system with three phases (like e.g. the peritectic reaction in Fe-C-Mn), but you do not have the corresponding thermodynamic database. If you had the database (or find a Calphad description in literature), it would be quite easy (see example Delta_Gamma_dri.txt). But without that, you need to describe the complete phase diagram of this system "by hand" defining a linearized phase diagram in a consistent way, which is not trivial.

A ternary phase diagram is 3-dimensional, and all equilibrium lines are now planes. You need 6 such planes to describe the pair-wise interaction for all 3 phase pairs (one solidus and one liquidus plane for both). For each phase pair you need a reference tie line (defined by two reference points, each given with two concentration values). All tie lines must be defined at the reference temperature. Furthermore, each equilibrium plane must be defined by two slopes (in the direction of the two dissolved components), starting at the corresponding reference point.
So, what you need to have is not only the information about the three tie lines at the reference temperature, but also intersections of the phase diagram in both concentration directions for all 6 reference points to determine the slopes. All that is not easy to find in literature...

Compared to that, it is relatively easy to define the entropy of fusion. Typically, these values are close to 1.0 J/(cm**3 K) for metal-liquid interfaces. For solid-solid interfaces the entropy of fusion is about one order of magnitude smaller. In order not to produce or destroy entropy in a closed reaction cycle, the sum of dS0/1 and dS1/2 must be equal to dS0/2. So, you could use e.g. dS0/1=0.9, dS0/2=1.0, and dS1/2=0.1 J/(cm**3 K).

In your case I would try to find an existing thermodynamic description in form of a Calphad assessment or a database.

Bernd

omid
Posts: 60
Joined: Thu Oct 01, 2015 11:19 am
anti_bot: 333

Re: PF simulation of Eutectic solidification

Post by omid » Thu Oct 29, 2015 9:51 am

Dear Bernd,

Many thanks for your constructive reply, now I have a good sight of fusion data, but about the linearisation, I actually didn't get completely what to do, can you please do a favor for me? I have attached a 3D phase diagram. Kindly show me how to adjust those data for i.e phase 1 and 2. This is a sample diagram and kindly show in this diagram, where are the needed data located!

Vielen Dank im Voraus ;)
Attachments
ternfig1.gif
example 3D diagram
ternfig1.gif (17.47 KiB) Viewed 5421 times

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: PF simulation of Eutectic solidification

Post by Bernd » Thu Oct 29, 2015 12:52 pm

Dear omid,

Your sample phase diagram does not contain all the information which is needed for constructing a linearized phase diagram description in MICRESS. Essentially, it only shows the liquidus planes for the 3 interactions between 3 solid phases and the melt.
What you need else is at least the 3 corresponding solidus planes and the tie-lines between the solidus and liquidus planes. Then you could chose a reference temperature for which you select 3 tie-lines (in contrary to binary systems, the tie-lines are not defined only by temperature, so you have many choices). Around the intersections of the tie-line with the solidus and liquidus planes (=equilibrium compositions) you now can "span up" the corresponding plane in the direction of each composition:
Tliquidus= Tref+m1liq*(c1-c1eq) + m2liq*(c2-c1eq)
Tsolidus= Tref+m1sol*(c1-c1eq) + m2sol*(c2-c1eq)
Tref, c1, c2, m1liq, m1sol, m2liq, m2sol are the parameters you specify in the linearized phase diagram description of each phase pair. For getting good estimates for the values of m, you need to plot an intersection of the phase diagram in direction of both elements.

If you furthermore want to include the solid-solid interactions, you need three more description of phase pairs (6 planes, 3 tie-lines, 12 slopes)...

Are you sure that no Calphad description of your ternary system is available?

Bernd

omid
Posts: 60
Joined: Thu Oct 01, 2015 11:19 am
anti_bot: 333

Re: PF simulation of Eutectic solidification

Post by omid » Fri Oct 30, 2015 9:57 am

Dear Bernd,

I have tried to find such a Calphad but seems that it's not available. I am working on Mo-Si-B alloy and do have a 2D diagram for liquid projection at a constant temperature. I have asked a friend of mine who is working on this alloy experimentally, and he is trying his bests to get such a data. But I, myself have not seen any appropriate diagram in the literature. I have also checked the thermocalc databases and found that also there are some databases which contain these elements but they are not able to predict the right reaction and phases. The main focus of me is reaction of liquid in the ternary eutectic point in which the liquid decomposes in to Mo3Si, Mo5SiB2 and B (SS). Here I have also a problem to define the number of dissolved constituents! Actually I have a solid solution of B, and the 2 other phases are stoichiometric! I didn't get the main purpose of:
# Concentration data
# ==================
# Number of dissolved constituents? (int)
:roll:
Kindly have look at the 2D diagram which I have and let me know your idea!
Moreover, I found another imaginary diagram and tried to draw some plates and one tie line, could you show me in this diagram what should I look for, in order to linearize? This photo does not have a high quality, if it is not suitable, do you have any other diagram so that you can tell me how can I deal with the diagram?
Attachments
Hero-m_demo1.png
Hero-m_demo1.png (51.11 KiB) Viewed 5416 times
Liquidus T2 VA 25 Ink_GH Mai 2015_25K (1).jpg
Liquidus T2 VA 25 Ink_GH Mai 2015_25K (1).jpg (498.21 KiB) Viewed 5416 times

Bernd
Posts: 1179
Joined: Mon Jun 23, 2008 9:29 pm

Re: PF simulation of Eutectic solidification

Post by Bernd » Mon Nov 02, 2015 11:42 am

Dear omid,

the phase diagrams you show have definitively been obtained from a Calphad description of the Mo-Si-B system. Thus, the good news is that the system has already been assessed. The question is just whether this assessment is freely available, or whether it is commercial and can only be used with Fact-Sage or Pan-Rome databases. I found a publication from 2005 (https://www.sciencedirect.com/science/a ... 01864?np=y) of an assessment which you should look up. Perhaps it is too old (Y. Yang and Y. A. Chang: Intermetallics 13 (2005) 121–128.). I would advise you to ask people from the Calphad community about public assessments of Mo-Si-B. I will send you a contact by PM.

"Number of dissolved constituents" just means how many elements (without the matrix element" are in your alloy system. This is 2 in your case. Typically, you would chose Mo as matrix element, and Si and B as dissolved constituents.

Bernd

Post Reply