How a non-interacting grain boundary influences

[chika]

Hello all,

I would like your help interpreting a growth behavior I observe when a grain boundary is defined without mutual interaction.

Setup (minimal case):
1. Two grains of the same phase A separated by an A/A grain boundary.
2. A small spherical B phase nucleus is placed near that boundary.
3. I set no_phase_interaction for A/A (i.e., no interaction considered for the A–A pair).
4. I consider interactions only for A/B (standard interfacial energy/mobility defined for A–B).

Observation:
Despite disabling A/A interaction, phase B grows preferentially toward the A/A grain boundary.

For your reference, I’ve attached the input file and an excerpt of VKT files showing the result.

Results.in

Results_1.zip

I would be grateful if you could explain why the A/A boundary still appears to bias B’s growth, based on the basic MICRESS equations described here: https://docs.micress.de/7.3/

To be specific, could you clarify the following points?

1. Meaning of no_phase_interaction for A/A:
When this option is chosen for a pair (here, A–A), are the corresponding interfacial terms (e.g., interfacial energy and mobility, gradient-energy cross terms) removed entirely or effectively set to zero? Are there any residual coupling terms—arising from the multi-phase constraint (Σφi = 1), anti-cycling/penalty terms, or numerical regularization—that could still influence growth near an A/A interface?

2. Triple-line/dihedral-angle interpretation:
In a region where phases A, A, and B meet, how does MICRESS treat the local “triple junction” when one pair (A–A) is set to no interaction? If σAA = 0 while σAB > 0, should we expect B to “wet” the A/A boundary, and under what conditions would this occur according to the MICRESS formulation?

3. Gradient-energy coefficients for same-phase grain boundaries:
Even with no_phase_interaction for A–A, does the representation of a same-phase grain boundary (two A grains) still carry any effective interfacial penalty via gradient-energy coefficients or phase-field potential terms? If so, how are those coefficients determined for an A–A pair with no interaction?

4. Input recommendations:
If the intent is to avoid B being preferentially attracted toward an A/A boundary when A/A interaction is disabled, are there recommended input settings or flags (e.g., interaction models, exclusions, or other options) to achieve that behavior?


Thank you very much for your guidance.

Best regards,
Chika

[Bernd]

Hi Chika,

Janin would be the best person to answer your questions, but she currently is unavailable, so I will try to give you a reasonable answer:

On the phase-field level, using "no_phase_interaction" basically means that the interface mobility is set to 0 and the interfacial energy remains undetermined. For the corresponding (binary) interface this is the same as if you would use "phase_interaction", but set the interface mobility to 0. That means that both, the interface term and the bulk (=driving force) term, are omitted, and the interface is perfectly pinned and cannot move.

At triple junctions, if the other 2 interfaces are mobile, this is different, because the phase-field parameter can be changed indirectly via the other interfaces. Even when using the "multi_obstacle" potential, triple point term are omitted in case of "no_phase_interaction", because no data is available on the interface energy/stiffness of the immobile interface. That means we would expect equilibrium angles of 120° like in the "double_obstacle" case, provided that the mobile interfaces can adjust accordingly. Please note that there are always interactions between the mobile and immobile interfaces, even if triple junction terms are completely omitted.

However, in MICRESS it is also possible to explicitly include triple junction terms in such a case by choosing

no_phase_interaction junction_force

Then, also for triple junctions, the outcome should be identical to "phase_interaction" with an interface mobility of zero. In that case you are required to provide the missing value of the interface energy, and such you should observe triple point angles according to Young's law (provided that the mobile interfaces are able to adjust accordingly).

On the level of thermodynamics, "no_phase_interactions" means that

a) using linearized phase diagrams, the element redistribution is achieved everywhere via the phase diagrams of the other phase interactions

b) using TQ-coupling, the element redistribution at triple junctions is achieved via the thermodynamic description of the other phase interactions. Once the other phase interactions are unavailable (in the TQ-case when the triple point grid cell transforms to an immobile binary interface), a "constant-K" approximation is used which is based on the last data from the other interfaces.

[chika]

Hi Bernd,

Thank you for your very detailed explanation, it has given me a much better understanding of this issue.

Best regards,
Chika