If you have problems with "Composition Sets"

helpful tips or ideas for unconventional solutions
Post Reply
Posts: 1502
Joined: Mon Jun 23, 2008 9:29 pm

If you have problems with "Composition Sets"

Post by Bernd » Fri Apr 01, 2016 11:41 pm

Dear all,

Dealing with "Composition Sets" is one of the most difficult topics you may encounter not only when using MICRESS with coupling to thermodynamic databases, but also using other software like DICTRA. Compostion sets are a Calphad modelling construct of different stable or metastable phase compositions for the same phase description in the database, which may coexist but which are often known as completely different phases by metallurgists. Perhaps the most prominent example in our comunity are the MC type carbides which in the Calphad description are identical to the fcc phase. Composition sets are named with the phase name label and a number which is omitted for the basic set (like FCC_A1 and FCC_A1#2).
Composition sets can cause trouble with all simulation tools which locate a phase at a specific location like MICRESS or DICTRA. The problem is that Thermo-Calc does not provide any rules for the numbering: If calculation reveals e.g FCC_A1 and FCC_A1#2 as stable phases, you cannot know which of them is the carbide and which the fcc. The only way to distinguish is to look at the composition. And, moreover, there is no specified way to define limits in the composition space which clearly distinguish these phases. This makes it especially difficult in multicomponent systems where the carbides can split up further into different carbides (NbC, VC, etc.) and nitrides (TiN), and fcc can split into different metallic phases (copper, nickel, etc.) and/or into ordered/disordered phases (gamma and gamma' phase, in TCNI database the phase name then is FCC_L12).
In MICRESS, it is important that we link each "MICRESS phase" to one of the composition sets of the "Thermo-Calc phases", and assure that they do not change number during simulation. This can sometimes be very difficult or even impossible. According to my experience, this is achieved best using the following rules (but please share also your own experiences and tricks):

1.) MICRESS does not create new composition sets itself or gives hints when a phase is getting into a miscibility gap. Composition sets should be created before by using the Thermo-Calc software and must be stored in the binary .ges5 file which is used with MICRESS. They will show up in the phase list when MICRESS is executed, and should be assigned to the corresponding MICRESS phases.

2.) Composition sets can be created in different ways using the Thermo-Calc software. In most cases they will appear automatically when creating a .ges5 file, e.g. using the "HOWTO.TCM" script which is available in the MICRESS examples directory for the FeCMn case. Whether or not this happens depends on the used database. Another possibility is to make a Thermo-Calc equilibrium calculation using the console mode which automatically creates the composition sets which are required under these specified conditions. Switching to the GES module and saving the .ges5 file assures that these composition sets are selected. Finally, composition sets can be explicitly created or changed in the GES module of Thermo-Calc using the "amend phase description" or "set major components" commands.

3.) Apart from the number, each composition set is characterized by a starting composition which is used for the initial equilibrium or always if trouble occurs. This composition is listed in the .log file and in the screen output at the end of the initialisation part ("Start Composition for iteration of quasi-equilibrium") when MICRESS is executed. You should look at these start compositions when deciding which of the composition sets to assign to which MICRESS phase. In difficult cases you should remember that you can define these start composition youself using the explicit way of creating composition sets mentioned in 2). There, you can define which elements should be "major constituents" on which sublattice, and thus decide which composition set should be a TiN, NbC or fcc phase.

4.) Defining appropriate start conditions is important but not sufficient to prevent your phase equilibria from switching to the "wrong" composition when other composition sets are more stable under given local conditions. Therefore, in MICRESS, you can define composition limits (keyword "Limits") for elements and phases which do not allow the phase to adopt an unwanted composition. If a quasi-equilibrium in MICRESS results in a composition which lies outside the allowed range, the quasi-equilibrium results are rejected, and an error message ("micress error 1xxyy", xx: phase number, yy: element number) is written. This demonstrates that your limits are "working". But sometimes MICRESS is unable to find the correct equilibria which results in excessive error messages of this type...

5.) Apart from defining optimal start composition and composition limits, the best you can do is to avoid metastable conditions which force your composition set to switch to wrong compositions:
- avoid large driving forces which may be the result of too low interface mobility or too high nucleation undercoolings
- don't check for nucleation of a phase when you are still outside the stability range. This not only may lead to a lot of error messages (error 20), but also to a wrong initialisation of a phase interaction which cannot easily be healed afterwards.
- check whether other phases are missing in the simulation which brings you to strongly metastable thermodynamic regions
- "global" relinearisation may help to avoid trouble coming from problems lwith local compositions

6.) Apart from the above said, you sometimes may also need a portion of luck...


Post Reply