Anisotropy of Interface Mobility

[CharMIC]

Hej Bernd,

I recently found out the MICRESS manual that for cubic systems, 0.05 is a typical value Anisotropy of interfacial mobility. But in one of the discussion thread you have mentioned that 0.2 value is typically good for cubic metallic systems.

Which one is correct since I have used 0.2 up to now for both interracial stiffness and mobility values.

Also I want to know what is a good value for stiffness and mobility anisotropic coefficient if the crystal have hexagonal system?

Are these only depend on the crystal symmetry or it also depend on the material composition?

BR
Chamara

[Bernd]

Dear Chamara,

The answer to this question is not as easy because it depends on the way how you are using MICRESS. From a theoretical point of view, 0.05 is the better value. You should also use such a value when you quantitatively try to simulate an anisotropic system, e.g. a dendrite tip. However, this requires that you work with a very high resolution.
However, in most technical application cases, you will not be able to work with such a high resolution, because you want to include multiple components and phases and certainly a larger simulation domain with more dendrites. Then, anisotropy will anyway not be quantitative, and you will chose the value for numerical purposes. Under such conditions I propose to use a value of 0.2 because it reduces the relative effect of grid artifacts.

For hexagonal systems, I have no experience. Perhaps, Janin can give you some hints here...

Bernd

[CharMIC]

Thanks Bernd.

I hope Janin will give some input for me soon

[janin]

Hi Chamara,
First, I like to say that 0.05 is a good approximation for the physical kinetric anisotropy in cubic systems, but I woukd recommend to search in literature for alloy specific anisotropic data obtained from MD-simulations.

Anyway you should be aware that the kinetic anisotropy should hardly play any role in diffusion-controlled solidifcation processes. If local equilibrium is reached at the interface, the morphology will be determined by the static anisotropy, while the kinetic underscooling should be negligible, and hence the effect of the interfacial mobility and its anisotropy should vanish.

However, as Bernd mentioned, in practical technical simulations the controlling tip curvature is often not well resolved and here it might make sense to use a calibrated numeric interfacial mobility and a numeric kinetic anisotropy to pragmatically reproduce the desired morphology.

Furthermore, I like to mention that in former times we still had problems with modelling static anisotropy, and therefore in old examples often used the kinetic anisotropy for compensation , but this is hardly necessary anymore. Nowadays, I would no longer recommend to use high kinetic ansisotropic coefficients such as 0.2 without very good reason.

In principal, the same is true for hcp symmetry, but hcp alloys can be very different from each other. Mostly you will not only see the six arms in the the basal planes, but several more in other directions. If you simulate in 3D you will have to adjust multiple anisotropy coeffients in order to obtain the correct orientations. In a former project, I modelled oriention selection in Mg-Al alloys starting from MD-simuations anisotropy data of pure Mg.
This work ( including a proposed anisotropic formulation) is published in:
J. Eiken: Phase-field simulation of microstructure formation in technical magnesium alloys
International Journal of Materials Research 101 (4), 503-509

Anisotropy coefficients of alloys also depend on concentration, however this dependency can still be neglected in alloys as long as the concentration of the solute is small. We have not yet modelled the concentration dependency in Micress, but might do so in future if there is demand.

Regards,
Janin

[CharMIC]

Hej Janin

Thanks a lot for the detailed explanation.

BR
Chamara