Hello Mr. Bernd,
Could you tell me why can we see a variation of the Al
concentration inside of a grain in solidification of alloy AlCu (Al= 9 % wt. %, Cu = 81 wt. %)?
Could it be due to dendrite formation or at least dendrite like grains?
microsegregation
Re: microsegregation
Dear RAGHAV,
please excuse me for splitting your question off from "other postprocessing programs"! I think, your actual post is not related to this topic...
For practically all alloys, solidification leads to a microsegregation effect, because the dissolved elements (Al in your case) have a different solubility in the solid and liquid phase. This leads to the fact that, during solidification, the composition of the melt is changing. Because the composition of the solid is always related to the composition of the melt (in first approximation by a constant partition coefficient), also the composition of solid will change during solidification. This "solidification history" is not changing afterwards because, typically, diffusion is very slow in the solid. Only a long homogenisation treatment at elevated temperature could remove such a microsegregation pattern.
In solidification processes, microsegregation occurs practically always, independent whether the solidification is cellular or dendritic. It can be only prevented if one enforces a planar solidification front. This is possible for an extremely slow solidification speed combined with a high temperature gradient, or with extremely high cooling rates, when solute trapping occurs (several 1000 K/sec).
Is this what you describe the result of a simulation or of an experiment?
Bernd
please excuse me for splitting your question off from "other postprocessing programs"! I think, your actual post is not related to this topic...
For practically all alloys, solidification leads to a microsegregation effect, because the dissolved elements (Al in your case) have a different solubility in the solid and liquid phase. This leads to the fact that, during solidification, the composition of the melt is changing. Because the composition of the solid is always related to the composition of the melt (in first approximation by a constant partition coefficient), also the composition of solid will change during solidification. This "solidification history" is not changing afterwards because, typically, diffusion is very slow in the solid. Only a long homogenisation treatment at elevated temperature could remove such a microsegregation pattern.
In solidification processes, microsegregation occurs practically always, independent whether the solidification is cellular or dendritic. It can be only prevented if one enforces a planar solidification front. This is possible for an extremely slow solidification speed combined with a high temperature gradient, or with extremely high cooling rates, when solute trapping occurs (several 1000 K/sec).
Is this what you describe the result of a simulation or of an experiment?
Bernd
Re: microsegregation
Hello Mr. Bernd
Actually, this was the result of a simulation.
Thanks for such a explanatary anwer.
MFG
Raghav
Actually, this was the result of a simulation.
Thanks for such a explanatary anwer.
MFG
Raghav
Re: microsegregation
Dear Bernd:
Hello!
Recently I am confused about the solute element distribution during the solidification.As we all know, for practically all alloys solidification leads to a microsegregation effect, because the dissolved elements (Al in your case) have a different solubility in the solid and liquid phase. And there will always be segregation in the dendritic and between dendrites. When I observe the segregation in the dendritic I find that the solute content in the dendritic center is very small which is not inconsistent with the experimental results. And sometimes when I view the solute distribution in the .conc I can see the bright lines in the center of primary dendrites and secondary dendrites. Can you tell me is it right and how to improve this phenomenon?
Besides I find the primary dendrites in my experimental results are more coarse then the simulation results. The primary dendrites in simulation are thin eventhough the subcooling is big. How can I deal with it?
Thank you very much!
Hello!
Recently I am confused about the solute element distribution during the solidification.As we all know, for practically all alloys solidification leads to a microsegregation effect, because the dissolved elements (Al in your case) have a different solubility in the solid and liquid phase. And there will always be segregation in the dendritic and between dendrites. When I observe the segregation in the dendritic I find that the solute content in the dendritic center is very small which is not inconsistent with the experimental results. And sometimes when I view the solute distribution in the .conc I can see the bright lines in the center of primary dendrites and secondary dendrites. Can you tell me is it right and how to improve this phenomenon?
Besides I find the primary dendrites in my experimental results are more coarse then the simulation results. The primary dendrites in simulation are thin eventhough the subcooling is big. How can I deal with it?
Thank you very much!
Re: microsegregation
Dear Shenyz,
this is an interesting observation. I believe the bright lines are due to solute trapping which is highest at the center of the dendrite tip because the normal velocity is highest. So far, the effect is real. However, due to the unphysical thickness of the interface in our phase-field approach, the effect is exaggerated. It would be interesting to try out whether it is smaller when you use "atc" correction (in phase diagram input if you have chosen "redistribution_control" in the corresponding phase interaction data). If you give it a try, please tell us about the outcomes. (Edit: please see next post!)
Dendrites in simulation can change morphology for many reasons. One important point is whether they grow at the correct tip undercooling. If the interface mobility is too high (compared to the numerical equivalent for diffusion limited growth), dendrites tend to be too thin (and the other way round). You should try to use "mob_corr" (also in phase diagram input) if possible to get close to diffusion limited growth without the need to calibrate the interface mobility.
Other parameters like the interface energy and anisotropy as well as averaging options for dG also have strong influence on dendrite morphology. Unfortunately, the problem of dendrite growth is too complex to be able to make quantitative morphology predictions for dendrite arms without calibration to experiments...
Bernd
this is an interesting observation. I believe the bright lines are due to solute trapping which is highest at the center of the dendrite tip because the normal velocity is highest. So far, the effect is real. However, due to the unphysical thickness of the interface in our phase-field approach, the effect is exaggerated. It would be interesting to try out whether it is smaller when you use "atc" correction (in phase diagram input if you have chosen "redistribution_control" in the corresponding phase interaction data). If you give it a try, please tell us about the outcomes. (Edit: please see next post!)
Dendrites in simulation can change morphology for many reasons. One important point is whether they grow at the correct tip undercooling. If the interface mobility is too high (compared to the numerical equivalent for diffusion limited growth), dendrites tend to be too thin (and the other way round). You should try to use "mob_corr" (also in phase diagram input) if possible to get close to diffusion limited growth without the need to calibrate the interface mobility.
Other parameters like the interface energy and anisotropy as well as averaging options for dG also have strong influence on dendrite morphology. Unfortunately, the problem of dendrite growth is too complex to be able to make quantitative morphology predictions for dendrite arms without calibration to experiments...
Bernd
Re: microsegregation
Dear Sheniz,
my answer was too fast - there is a much simpler explanation for the bright line inside the dendrite: Due to the highest curvature at the tip, equilibrium concentrations are shift
Example: W-composition in directional dendrites of Ni-superalloy:
If resolution is low, these lines can be just one grid cell in thickness.
Bernd
my answer was too fast - there is a much simpler explanation for the bright line inside the dendrite: Due to the highest curvature at the tip, equilibrium concentrations are shift
Example: W-composition in directional dendrites of Ni-superalloy:
If resolution is low, these lines can be just one grid cell in thickness.
Bernd
You do not have the required permissions to view the files attached to this post.
Re: microsegregation
Dear Bernd,
Hello!
I have one short question:
How are the partition coefficients introduced into simulations during solidification? Is it possible to find out from any output, the calculated constants for partition coefficients?
Regards,
Hello!
I have one short question:
How are the partition coefficients introduced into simulations during solidification? Is it possible to find out from any output, the calculated constants for partition coefficients?
Regards,
Re: microsegregation
Dear Atur,
When using coupling to thermodynamic database, you always get linearisation data for the given scope (e.g. average over the interface). These values are written to the .TabLin output in tabular form. This you can use for calculating Partition coefficients.
Kαβ=cβ/cα
In MICRESS, rather than normal partition coefficients, "differential" partition coefficients are used:
Kαβ=dcβ/dcα
This type of partition coefficient is written also to the .TabLin output in case "diagonal" redistribution is used.
Bernd
When using coupling to thermodynamic database, you always get linearisation data for the given scope (e.g. average over the interface). These values are written to the .TabLin output in tabular form. This you can use for calculating Partition coefficients.
Kαβ=cβ/cα
In MICRESS, rather than normal partition coefficients, "differential" partition coefficients are used:
Kαβ=dcβ/dcα
This type of partition coefficient is written also to the .TabLin output in case "diagonal" redistribution is used.
Bernd
Re: microsegregation
Dear Bernd,
Thank you very much for your reply. At concentration solver, diagonal was selected.
I went into TabLin output (I provided as attachment) but I am a bit confused to choose which data to extract here for plotting/finding the change in K for each solute in my system. For instance, can you help me to elaborate understanding what ph1 and ph2 means or chaw c0/ph1 - K/ph1 represents? Also, I wonder why ph1 and ph2 values are alternating between 0,1 and 2.
I would appreciate your help on elaborating the output
.
Kind regards
Thank you very much for your reply. At concentration solver, diagonal was selected.
I went into TabLin output (I provided as attachment) but I am a bit confused to choose which data to extract here for plotting/finding the change in K for each solute in my system. For instance, can you help me to elaborate understanding what ph1 and ph2 means or chaw c0/ph1 - K/ph1 represents? Also, I wonder why ph1 and ph2 values are alternating between 0,1 and 2.
I would appreciate your help on elaborating the output
. Kind regards
You do not have the required permissions to view the files attached to this post.
Re: microsegregation
Dear Atur,
In the .TabLin file the linearisation parameters are given which are used for redistribution of the elements between the different phases. They are exactly the same as those written to the .log-file when a certain phase interaction is initialized. The corresponding equations can be found in this paper which resumes the different redistribution models which are available in MICRESS.
Linearisation parameters are calculated for each phase pair ph1/ph2 by obtaining a quasi-equilibrium (parallel tangent construction) at the temperature T0. c0/ph1 and c0/ph2 are the equilibrium compositions for each element, and m/ph2 and m/ph2 the corresponding slopes of the pseudo-phase diagram (plotted for the driving force dG instead of temperature). The classical partition coefficients are defined as K=c0(ph2)/c0(ph1) for each element. This is what you can use for monitoring the partition coefficients in your case. In your simulation, you have phases 0, 1, and 2 - probably you are interested partitioning between liquid (phase 0) and the primary phase (ph1).
In the paper, furthermore, the (differential) partition coefficients are defined for each element as K12=m(ph2)/m(ph1) for multibinary extrapolation and K12=dc(ph1)/dc(ph2) for diagonal redistribution (keeping compositions in ph2 constant for all other elements). K12 and K21 are also given in the .TabLin file for each element (as K/ph1 and K/ph2) for the case of diagonal redistribution.
Bernd
In the .TabLin file the linearisation parameters are given which are used for redistribution of the elements between the different phases. They are exactly the same as those written to the .log-file when a certain phase interaction is initialized. The corresponding equations can be found in this paper which resumes the different redistribution models which are available in MICRESS.
Linearisation parameters are calculated for each phase pair ph1/ph2 by obtaining a quasi-equilibrium (parallel tangent construction) at the temperature T0. c0/ph1 and c0/ph2 are the equilibrium compositions for each element, and m/ph2 and m/ph2 the corresponding slopes of the pseudo-phase diagram (plotted for the driving force dG instead of temperature). The classical partition coefficients are defined as K=c0(ph2)/c0(ph1) for each element. This is what you can use for monitoring the partition coefficients in your case. In your simulation, you have phases 0, 1, and 2 - probably you are interested partitioning between liquid (phase 0) and the primary phase (ph1).
In the paper, furthermore, the (differential) partition coefficients are defined for each element as K12=m(ph2)/m(ph1) for multibinary extrapolation and K12=dc(ph1)/dc(ph2) for diagonal redistribution (keeping compositions in ph2 constant for all other elements). K12 and K21 are also given in the .TabLin file for each element (as K/ph1 and K/ph2) for the case of diagonal redistribution.
Bernd