Partition-no-partition transition temperature during pearlite-to-austenite transformation
Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
I am currently working on the study of the pearlite to austenite transformation kinetics during fast heating in a Fe-C-Mn pearlitic steel. The aim was to reproduce the morphology of the austenite in the early stage of formation and also the overall transformation kinetics.
Some studies have shown that at low superheating above the eutectoid temperature the austenite formation is not thermodynamically possible without the redistribution of the substitutional alloy elements, Mn in our case. These studies introduced the concept of a partition-no-partition transition temperature (PNTT), defined as the minimum temperature at which the growth of austenite may occur without the redistribution of the substitutional alloy elements. It was shown that this transition temperature depends on the manganese partition (kθα=xMnθ/xMnα) ratio in the initial microstructure.
More details are in the following papers:
[1] Y. Xia, M. Enomoto, Z.N. Yang, et al. Effects of alloying elements on the kinetics of austenitization from pearlite in Fe–C–M alloys. Philosophical Magazine, 2013, 93 (9): 1095–1109.
[2] Z.N. Yang, Y. Xia, M. Enomoto, et al. Effect of Alloying Element Partition in Pearlite on the Growth of Austenite in High-Carbon Low Alloy Steel. Metallurgical Material Transaction 2016, 47A: 1019-1027.
Can MICRESS, coupled with Thermo-Calc, be used to prove that below the PNTT(calculated for the Mn partition ratio in the initial ferrite-carbide microstructure) austenite cannot grow if paraTQ is set for the redistribution of Mn?
Kind Regards,
Pina
I am currently working on the study of the pearlite to austenite transformation kinetics during fast heating in a Fe-C-Mn pearlitic steel. The aim was to reproduce the morphology of the austenite in the early stage of formation and also the overall transformation kinetics.
Some studies have shown that at low superheating above the eutectoid temperature the austenite formation is not thermodynamically possible without the redistribution of the substitutional alloy elements, Mn in our case. These studies introduced the concept of a partition-no-partition transition temperature (PNTT), defined as the minimum temperature at which the growth of austenite may occur without the redistribution of the substitutional alloy elements. It was shown that this transition temperature depends on the manganese partition (kθα=xMnθ/xMnα) ratio in the initial microstructure.
More details are in the following papers:
[1] Y. Xia, M. Enomoto, Z.N. Yang, et al. Effects of alloying elements on the kinetics of austenitization from pearlite in Fe–C–M alloys. Philosophical Magazine, 2013, 93 (9): 1095–1109.
[2] Z.N. Yang, Y. Xia, M. Enomoto, et al. Effect of Alloying Element Partition in Pearlite on the Growth of Austenite in High-Carbon Low Alloy Steel. Metallurgical Material Transaction 2016, 47A: 1019-1027.
Can MICRESS, coupled with Thermo-Calc, be used to prove that below the PNTT(calculated for the Mn partition ratio in the initial ferrite-carbide microstructure) austenite cannot grow if paraTQ is set for the redistribution of Mn?
Kind Regards,
Pina
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Pina,
Thank you for the interesting question. I think that you describe a type of problem which could well be addressed by MICRESS simulations. However, I am not sure whether I correctly understand what is meant by "redistribution" in this case. It could be - like you probably assume - that the PNTT-temperature refers to a transition between para-equilibrium (no redistribution) and local equilibrium or "nple" (no partitioning - local equilibrium -->redistribution). However, at least the first of your references
[1] Y. Xia, M. Enomoto, Z.N. Yang, et al. Effects of alloying elements on the kinetics of austenitization from pearlite in Fe–C–M alloys. Philosophical Magazine, 2013, 93 (9): 1095–1109.
explicitly describes already in the Abstract (I have no access to the rest of the document, unfortunately) that always local equilibrium is assumed. Hence, they obviously talk about a transition between redistribution without diffusion of substitutional elements (at high temperature, if the driving force is sufficient for a phase transformation without change in composition), and redistribution with diffusion of substitutional elements (at low temperatures when the driving force is lower, so that a phase transformation only can proceed at altered compositions. This for me makes perfect sense, however, would describe a transition between a diffusion-limited growth with visible segregation on one hand, and nple on the other hand. And this is something which automatically should come out if you use the nple model (instead of "paratq") in MICRESS.
How do you understand their concept of the PNTT-temperature?
Bernd
Thank you for the interesting question. I think that you describe a type of problem which could well be addressed by MICRESS simulations. However, I am not sure whether I correctly understand what is meant by "redistribution" in this case. It could be - like you probably assume - that the PNTT-temperature refers to a transition between para-equilibrium (no redistribution) and local equilibrium or "nple" (no partitioning - local equilibrium -->redistribution). However, at least the first of your references
[1] Y. Xia, M. Enomoto, Z.N. Yang, et al. Effects of alloying elements on the kinetics of austenitization from pearlite in Fe–C–M alloys. Philosophical Magazine, 2013, 93 (9): 1095–1109.
explicitly describes already in the Abstract (I have no access to the rest of the document, unfortunately) that always local equilibrium is assumed. Hence, they obviously talk about a transition between redistribution without diffusion of substitutional elements (at high temperature, if the driving force is sufficient for a phase transformation without change in composition), and redistribution with diffusion of substitutional elements (at low temperatures when the driving force is lower, so that a phase transformation only can proceed at altered compositions. This for me makes perfect sense, however, would describe a transition between a diffusion-limited growth with visible segregation on one hand, and nple on the other hand. And this is something which automatically should come out if you use the nple model (instead of "paratq") in MICRESS.
How do you understand their concept of the PNTT-temperature?
Bernd
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
Thanks for the replay.
Your explanation of the Partition-No partition Transition Temperature (PNTT) is correct. The pearlite to austenite transformation can proceed without the redistribution of the substitutional element (Mn in our case) only if the activity of the carbon activity in γ at the γ/θ interface is higher than the carbon activity in γ at the γ/α interface. This happens above a given temperature, PNTT.
This transition temperature depends on the Mn concentration in a and q in the initial pearlite microstructure or, in other words, from the partition ratio Kθα=Mnθ/Mnα.
Then I considered an initial microstructure with 3 colonies of pearlite having three different concentration values of Mn in cementite lamellae, 4.5 at% (colony 1 in the attached document) 5.7 at% (colony 2) and 9 at% (colony 3). The concentration of ferrite was set equal to 0.78 at% in all the colonies.
I calculated the PNTT for colony 1 and colony 3 and they are 1013 K and 1028 K respectively.
I set the initial temperature of the simulation equal to 1013 K.
I attach a file with the simulated microstructure and Mn concentration map at 1026 K. I did not expect growth of austenite in colony 3 (PNTT = 1028 K) without Mn redistribution between austenite and the parent phase but it does not seem the case.
Kind Regards
Pina
Thanks for the replay.
Your explanation of the Partition-No partition Transition Temperature (PNTT) is correct. The pearlite to austenite transformation can proceed without the redistribution of the substitutional element (Mn in our case) only if the activity of the carbon activity in γ at the γ/θ interface is higher than the carbon activity in γ at the γ/α interface. This happens above a given temperature, PNTT.
This transition temperature depends on the Mn concentration in a and q in the initial pearlite microstructure or, in other words, from the partition ratio Kθα=Mnθ/Mnα.
Then I considered an initial microstructure with 3 colonies of pearlite having three different concentration values of Mn in cementite lamellae, 4.5 at% (colony 1 in the attached document) 5.7 at% (colony 2) and 9 at% (colony 3). The concentration of ferrite was set equal to 0.78 at% in all the colonies.
I calculated the PNTT for colony 1 and colony 3 and they are 1013 K and 1028 K respectively.
I set the initial temperature of the simulation equal to 1013 K.
I attach a file with the simulated microstructure and Mn concentration map at 1026 K. I did not expect growth of austenite in colony 3 (PNTT = 1028 K) without Mn redistribution between austenite and the parent phase but it does not seem the case.
Kind Regards
Pina
- Attachments
-
- PNTT_Phase&Mn@1026K.docx
- Microstructure and Mn concentration map in at% at 1026 K
- (62.44 KiB) Downloaded 127 times
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Pina,
Did you change "paratq" to "nple" as I suggested? Otherwise, austenite would grow in any case because there is never redistribution with "paratq"...?
Bernd
Did you change "paratq" to "nple" as I suggested? Otherwise, austenite would grow in any case because there is never redistribution with "paratq"...?
Bernd
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
Yes, I changed "paratq" to "nple", as you suggested.
Kind Regards,
Pina
Yes, I changed "paratq" to "nple", as you suggested.
Kind Regards,
Pina
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Pina,
In order to find out where the problem could be located, I have a few more specific questions:
- how did you create the initial microstructure? Is it by reading phase structure and compositions from file? It seems that the Mn composition is "sharp" while the phase structure is "smooth". Is it the same for the C-compostion? Did you apply "initialisation" of interfaces (which could make compositions and phase fractions inconsistent)?
- do you use coupling to a thermodynamic database or linearized phase diagrams?
- do you use constant heating as thermal boundary condition? If yes, do you use "bottom_temperature" as nple-option?
- is the expected effect of Mn on austenite growth at least qualitatively correct (i.e. does the dissolution of cementite in colony 3 at least start at higher temperature than for colony 1 and 2)?
- did you have a look at the real phase compositions (.c2pha*) of Mn how it is redistributed between all phases? Maybe they show unexpected values which could help finding the error.
Best wishes
Bernd
In order to find out where the problem could be located, I have a few more specific questions:
- how did you create the initial microstructure? Is it by reading phase structure and compositions from file? It seems that the Mn composition is "sharp" while the phase structure is "smooth". Is it the same for the C-compostion? Did you apply "initialisation" of interfaces (which could make compositions and phase fractions inconsistent)?
- do you use coupling to a thermodynamic database or linearized phase diagrams?
- do you use constant heating as thermal boundary condition? If yes, do you use "bottom_temperature" as nple-option?
- is the expected effect of Mn on austenite growth at least qualitatively correct (i.e. does the dissolution of cementite in colony 3 at least start at higher temperature than for colony 1 and 2)?
- did you have a look at the real phase compositions (.c2pha*) of Mn how it is redistributed between all phases? Maybe they show unexpected values which could help finding the error.
Best wishes
Bernd
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
Here are my answers.
- how did you create the initial microstructure? Is it by reading phase structure and compositions from file? It seems that the Mn composition is "sharp" while the phase structure is "smooth". Is it the same for the C-compostion? Did you apply "initialisation" of interfaces (which could make compositions and phase fractions inconsistent)?
I read both the phase structure and the Mn and C composition from file. Both C and Mn have a sharp transition at the interface. Maybe this can be the problem. How can I avoid this sharp transition?
- do you use coupling to a thermodynamic database or linearized phase diagrams?
I use thermo-calc coupling
- do you use constant heating as thermal boundary condition? If yes, do you use "bottom_temperature" as nple-option?
I use constant heating rate but “local_velocity” as nple option.
- is the expected effect of Mn on austenite growth at least qualitatively correct (i.e. does the dissolution of cementite in colony 3 at least start at higher temperature than for colony 1 and 2)?
The dissolution of cementite seems to start at the same temperature in all the colonies but the austenite growth in the ferrite starts at a lower temperature in colony 3 than in the other colonies.
- did you have a look at the real phase compositions (.c2pha*) of Mn how it is redistributed between all phases? Maybe they show unexpected values which could help finding the error.
No. I run again the simulation by setting as output also c2pha2 and c2phas3 ( I did it only for the phase 1).
Kind Regards,
Pina
Here are my answers.
- how did you create the initial microstructure? Is it by reading phase structure and compositions from file? It seems that the Mn composition is "sharp" while the phase structure is "smooth". Is it the same for the C-compostion? Did you apply "initialisation" of interfaces (which could make compositions and phase fractions inconsistent)?
I read both the phase structure and the Mn and C composition from file. Both C and Mn have a sharp transition at the interface. Maybe this can be the problem. How can I avoid this sharp transition?
- do you use coupling to a thermodynamic database or linearized phase diagrams?
I use thermo-calc coupling
- do you use constant heating as thermal boundary condition? If yes, do you use "bottom_temperature" as nple-option?
I use constant heating rate but “local_velocity” as nple option.
- is the expected effect of Mn on austenite growth at least qualitatively correct (i.e. does the dissolution of cementite in colony 3 at least start at higher temperature than for colony 1 and 2)?
The dissolution of cementite seems to start at the same temperature in all the colonies but the austenite growth in the ferrite starts at a lower temperature in colony 3 than in the other colonies.
- did you have a look at the real phase compositions (.c2pha*) of Mn how it is redistributed between all phases? Maybe they show unexpected values which could help finding the error.
No. I run again the simulation by setting as output also c2pha2 and c2phas3 ( I did it only for the phase 1).
Kind Regards,
Pina
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Pina,
My feeling is that the definition of the initial compositions could be a source of problems. Given the different diffusivity - and thus different segregation behavior - of the dissolved elements Mn and C, we need different strategies for defining the initial concentration distribution. For Mn, I believe, you should have a constant average compositions in the whole colony, and not only in the region of the cementite, if we assume e.g. a segregation band as the origin of the altered Mn composition and negligible Mn diffusion during formation of pearlite. For C, instead, we assume faster diffusion, and we have a fixed value of 25 at% in cementite. Thus, we need a smooth concentration profile of C with a value in pure ferrite close to the equilibrium composition.
Unfortunately, MICRESS currently doesn't allow directly mixing the sources of input for the initial concentrations for different elements. Therefore, we need a two-step procedure:
1.) In a first step, you define only the composition of the ferrite phase (phase number 2), and set cementite to equilbrium (using the "equilibrium 2" option). You can break the simulation directly after writing the output for t=0, and convert the .conc1 output (C-distribution) to vtk (e.g. using DP_MICRESS).
2.) In the second step, you use the "from_file" option as you did before, but now you can read the C-distribution from the .vtk-file you just created. And, as explained above, for Mn I suggest using a continuous "segregation band" which also extends over the ferrite regions.
Regarding the nple-model, using "bottom_temperature" is numerically much more stable than "local_velocity" and recommendable in your case. Please note that you cannot use this when holding isothermally, and the temperature trend must be in accordance with the expected phase transformation directions of the interfaces which use "nple" option.
Bernd
My feeling is that the definition of the initial compositions could be a source of problems. Given the different diffusivity - and thus different segregation behavior - of the dissolved elements Mn and C, we need different strategies for defining the initial concentration distribution. For Mn, I believe, you should have a constant average compositions in the whole colony, and not only in the region of the cementite, if we assume e.g. a segregation band as the origin of the altered Mn composition and negligible Mn diffusion during formation of pearlite. For C, instead, we assume faster diffusion, and we have a fixed value of 25 at% in cementite. Thus, we need a smooth concentration profile of C with a value in pure ferrite close to the equilibrium composition.
Unfortunately, MICRESS currently doesn't allow directly mixing the sources of input for the initial concentrations for different elements. Therefore, we need a two-step procedure:
1.) In a first step, you define only the composition of the ferrite phase (phase number 2), and set cementite to equilbrium (using the "equilibrium 2" option). You can break the simulation directly after writing the output for t=0, and convert the .conc1 output (C-distribution) to vtk (e.g. using DP_MICRESS).
2.) In the second step, you use the "from_file" option as you did before, but now you can read the C-distribution from the .vtk-file you just created. And, as explained above, for Mn I suggest using a continuous "segregation band" which also extends over the ferrite regions.
Regarding the nple-model, using "bottom_temperature" is numerically much more stable than "local_velocity" and recommendable in your case. Please note that you cannot use this when holding isothermally, and the temperature trend must be in accordance with the expected phase transformation directions of the interfaces which use "nple" option.
Bernd
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
Many thanks for your suggestions. I am now running a new simulation with the corrected initial concentrations.
I will let you know the results.
Kind Regards,
Pina
Many thanks for your suggestions. I am now running a new simulation with the corrected initial concentrations.
I will let you know the results.
Kind Regards,
Pina
Re: Partition-no-partition transition temperature during pearlite-to-austenite transformation
Dear Bernd,
Now, the kinetics of austenite formation and the austenite morphology are in better agreement with the experimental one, probably thanks to the new initial concentration data. However, austenite in colony 3 is still starting to grow at 1026 K when austenite should not grow without the partition of Mn (PNTT = 1028 K).
Kind Regards,
Pina
Now, the kinetics of austenite formation and the austenite morphology are in better agreement with the experimental one, probably thanks to the new initial concentration data. However, austenite in colony 3 is still starting to grow at 1026 K when austenite should not grow without the partition of Mn (PNTT = 1028 K).
Kind Regards,
Pina
- Attachments
-
- Phase__1033K.png (20.38 KiB) Viewed 5086 times
-
- Phase_1030K.png (19.99 KiB) Viewed 5086 times
-
- Phase_1026K.png (15.88 KiB) Viewed 5086 times
-
- Phase_1023K.png (12.77 KiB) Viewed 5086 times