Martensitic Transformation

[R.Hess]

Hello,

we try to simulate the microstructure evolution during a cooling process in a heat affected zone of a low carbon steel. We expect a martensitic structure because of the high cooling rates (about 500-1000k/s).

My first question is, is it possible to couple TTT Diagrams from ThermoCalc with Micress? Because, i think the "normal" Phase diagrams, calculated by TC, only include stable phases in equilibrium.
Obviously, Martensite is a metastable phase, so i think MICRESS can´t get the necessary informations for martensite.
If it´s not possible, is the only way using the linearised phase-diagram?

In a paper,written by R.G.Thiessen, martensite was modeled like a supersaturated bcc-phase. Is there another way to describe martensite in Micress?

And my last question is, how can I identify Martensite in DP_Micress? I think i will not see the lamellae-structure, which is typically for martensite, is that right?
And furthermore, what about the option "add_to_grain"? In the handbook for micress, i read, that this option is useful for modelling the martensitic transformation. But i really didn´t understand what it means and what this application will do? what further information needs micress, after i wrote

...
add_to_grain
parent_grain
...

?


Best Regards

Raphael

[Bernd]

Dear Raphael,

direct simulation of martensite is impossible in MICRESS due to two reasons: First of all, the martensitic transformation is displacive, i.e. there is no moving phase boundary - the whole grain or parts of it "flip" instantly to another phase state. Secondly, martensite is not an equilibrium phase and as such is neither compatible to the CALPHAD approach of thermodynamic database assessment nor to the local (quasi-)equilibrium approach of MICRESS.
Having said that it becomes clear that only indirect approaches can be used. There are several of those, and depending on which one is used the way how to visualize the martensitic phase in DP_MICRESS would be different.

The first such approach (and the simplest one) I heard of was by Gerhard Pariser from IEHK at the RWTH Aachen who just used a criterion function for martensite formation. With this model it was even not necessary to simulate martensite inside MICRESS: It was done by postprocessing in DP_MICRESS by coloring the regions where martensite was expected as a function of local austenite composition. For easier use, we implemented his multicomponent martensite criterion directly in the (old version of) DP_MICRESS in those times.

An approach which includes Martensite as a "phase" in MICRESS is to use linearized phase diagram descriptions which are taken from other sources and which under certain circumstances can mimic martensite. The problems which remain are from where to take the phase diagram, the conceptual issue to treat it in a pseudo-equilibrium manner, and the problem that the phase transformation will not proceed instantaneously (with speed of sound) like in reality. Furthermore, such approaches will mostly be limited to binary alloys or a pseudo-binary treatment for practical reasons. As martensite is described as a phase in MICRESS, it can easily be visualized as such using DP_MICRESS.

An alternative thermodynamic approach is the "supersaturated" bcc phase which you mentioned. Besides the normal BCC_A1 phase, a second phase in MICRESS is linked to BCC_A1 phase in the database (named BCC_A1_2). In MICRESS 6.3 there is an option to apply a driving-force offset to each phase interaction allowing to favor/disfavor a phase. That means, martensite thermodynamically will behave like bcc but is e.g. less stable. Other properties like diffusivities can be modified (e.g. by a prefactor for each element). This approach has the advantage of being multicomponent and coupled to CALPHAD data, and at the same time reduces the number of parameters which can/have to be chosen freely. Like with the linearized phase diagram approach, martensite is described as a phase in MICRESS and can easily be visualized using DP_MICRESS.

A completely different approach is the use of "add_to_grain" which allows switching between phases instantaneously in MICRESS. By nature, this is not a very physical approach, and all physics enter into the criterion function which is used together with "add_to_grain". Nevertheless, it can be combined with the different thermodynamic approaches described above and overcomes the problem how to simulate "flipping" between phases.

However, all these approaches suffer from being "pseudo-equilibrium" models which do not include the real nature of martensite as a metastable distorted lattice which also includes important elastic and crystallographic effects. This also implies that modelling martensite using the above approaches cannot describe the interior lamellar structure of martensite. These properties perhaps can be included in future (indirect) approaches which do not exist yet (at least in the field of applied phase-field simulations)...

Bernd

[deepumaj1]

Hello Bernd,
I have some questions in the following context:

An alternative thermodynamic approach is the "supersaturated" bcc phase which you mentioned. Besides the normal BCC_A1 phase, a second phase in MICRESS is linked to BCC_A1 phase in the database (named BCC_A1_2). In MICRESS 6.3 there is an option to apply a driving-force offset to each phase interaction allowing to favor/disfavor a phase. That means, martensite thermodynamically will behave like bcc but is e.g. less stable. Other properties like diffusivities can be modified (e.g. by a prefactor for each element). This approach has the advantage of being multicomponent and coupled to CALPHAD data, and at the same time reduces the number of parameters which can/have to be chosen freely. Like with the linearized phase diagram approach, martensite is described as a phase in MICRESS and can easily be visualized using DP_MICRESS.

1. If I want to simulate the martensite formation while cooling, how can I define the supersaturation of carbon in the Martensite BCC_A1_2 phase (Lets say, I want to nucleate this phase to simulate the martensite formation)?
2. Won't diffusion happen in this phase, even if we specify a very low prefactor to the diffusivity?

Deepu

[Bernd]

Hi Deepu,

with this approach, you could define a "supersaturated" martensite in the sense of "metastable": The only free parameter is the shift in Gibbs energy by defining a driving force offset for all interactions with other phases. Unfortunately, it is not possible to define specific deviations in the phase composition for each element.

For diffusion, you are free to define "what you want" for each element, i.e. either just modify the BCC diffusivity by a constant factor, or define your own values. So, if e.g. you nucleate martensite by "switching" a C-rich austenite grain to a modified BCC description using "add_to_grain", you can prevent that the carbon diffuses out rapidly by setting a reduced C diffusivity.

However, I myself have only very little experiences whether or in which cases this procedure can help you for a useful description of martensite...

Bernd