En poursuivant votre navigation sur ce site, vous acceptez le dépôt de cookies dans votre navigateur. (En savoir plus)
Portail > Offres > Offre UMR7198-MARTAI-078 - Recherche sur la métallurgie physique : Optimisation des microstructures d'aciers duplex moyen manganèse par apprentissage automatique et méthodes de caractérisation à haut débit (H/F)

Research on physical metallurgy : Optimization of duplex medium manganese steel microstructures by machine learning and high-throughput characterization methods (H/F)

This offer is available in the following languages:
- Français-- Anglais

Date Limite Candidature : vendredi 14 avril 2023

Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler. Les informations de votre profil complètent celles associées à chaque candidature. Afin d’augmenter votre visibilité sur notre Portail Emploi et ainsi permettre aux recruteurs de consulter votre profil candidat, vous avez la possibilité de déposer votre CV dans notre CVThèque en un clic !

Informations générales

Intitulé de l'offre : Research on physical metallurgy : Optimization of duplex medium manganese steel microstructures by machine learning and high-throughput characterization methods (H/F)
Référence : UMR7198-MARTAI-078
Nombre de Postes : 1
Lieu de travail : NANCY
Date de publication : mardi 28 février 2023
Type de contrat : CDD Scientifique
Durée du contrat : 24 mois
Date d'embauche prévue : 3 avril 2023
Quotité de travail : Temps complet
Rémunération : 2.833,40€ gross salary/month adjustable according to the level of experience
Niveau d'études souhaité : Doctorat
Expérience souhaitée : 1 à 4 années
Section(s) CN : Materials, nanomaterials and processes chemistry

Missions

The mission of the researcher is to conduct research in physical metallurgy in the frame of the DIAMS project, funded through DIADEM PEPR Program (operated by ANR / France 2030) and by LABEX DAMAS (operated by ANR / PIA3).

Activités

Duplex Medium Manganese (MedMn) steels belong to so-called 3rd generation AHSS (advanced high-strength steel) for cold-stamping applications. Their ground-breaking properties are achieved thanks a fine “ferritic” matrix and a large amount of retained austenite. The refined matrix explains the superior strength and toughness of these steels while retained austenite explains their good formability thanks to an efficient TRIP effect (Transformation Induced Plasticity).
Contrary to other 3rd generation concepts as carbide free bainitic or Quenching & Partitioning steels, the stabilization of retained austenite in MedMn steels is achieved in the intercritical range by both carbon and manganese partitioning between a ferritic phase and the austenite. This particular microstructure design is governed by an ortho-equilibrium at high temperature but will be affected by the different interactions between recrystallization/cementite precipitation/austenite transformation during heating and displacive transformations during cooling. For instance, a granular microstructure made of recrystallized ferrite and retained austenite is obtained after an intercritical annealing after rolling. If the initial microstructure is fully martensitic, austenite grows at martensite laths during annealing (Austenite Reversion Transformation) and the final microstructure is fibrous with a recovered martensitic matrix. Several studies showed that various size, morphology, topology as well as austenite stability can be achieved. As a consequence, the final properties are highly sensitive to the processing parameters. Even if the basic mechanisms explaining the microstructure morphogenesis are now understood, many efforts remain to be made to understand all the interactions between those individual mechanisms and the associated transformation kinetics (chemical partition, cementite precipitation/dissolution), to explore little-studied areas of design (chemistry/process spaces) and to rationalize the great versatility of experimental observations made on these microstructures. One difficulty comes from the fact that the chemical design space of these concepts is very large in terms of carbon, manganese (up to 10%) but also silicon (affecting the precipitation of carbides), aluminum and chromium (affecting the intercritical domains).

In the absence of physics-based modelling integrating all known mechanisms and allowing a reliable and quantitative link between processes/microstructures/properties, the objective of this project is to investigate first the formation mechanisms of these complex microstructures using high-throughput characterization methods and optimization by machine learning. This optimization can be done on multiple criteria (fraction or thermochemical stability of austenite, presence of cementite, sensitivity to annealing temperature, morphology/topology, ...) which have a known impact on the final mechanical properties.

The principal tasks of the researcher will be the following:
1. Bibliography and datamining (experimental data and equilibrium calculations)
Along with the necessary bibliographic study, datamining operations will be initiated on literature results to identify of non-well explored or non-well understood composition/processing spaces and to serve as test/training database for the machine learning developed in step 3. The limits of conventional equilibrium calculations (Thermocalc) for microstructure prediction should be also studied.
2. High-throughput exploration of the manufacturing space (composition and processing conditions)
In this section, we will follow a recently proven method for studying phase transformations in conventional steels, which consists of studying samples containing controlled chemical gradients in order to be able to study a wide range of chemical compositions in the same sample at the same time. The first sub-step will consist in producing the graded samples in a controlled way using the elaboration platforms of the project. Then the phase transformations will be characterized in situ in HEXRD on the different compositions along complex thermal paths using the project's thermomechanical platform ITM. The obtained microstructures will then be characterized post-mortem using EBSD on large domains. Given the very large amount of data (diffractograms or micrographs) that it would be desirable to acquire, the use of automated techniques by AI will be essential to process the tests and observations in reasonable time.
3. Machine learning model for microstructure prediction
The final step of the project will consist in the development of several machine learning tools for microstructure prediction trained on all the different sources of data.
The prerequisite for a robust training of these algorithms is to provide them with homogeneous data (same nature and similar accuracy) and with a balanced dataset representative of the situations on which the models will be designed to make predictions.

In this project, many different types of data will be aggregated and will have to be used to feed the model. Thus, a large part of the work will be dedicated to data engineering (cleaning, filling, curing) so that data of different nature can work together. Indeed, the data from thermodynamic modeling cover a large domain of the parameter space but predict only thermodynamic equilibria. Conversely, the experimental data will necessarily be more fragmented but more reliable and will intrinsically contain temporal evidences. The data mining step should also provide data from the literature but which will be mismatched anyway. It should permit to propose new optimization strategies of these alloys.

[Arlazarov2012] Arlazarov, A., Gouné, M., Bouaziz, O., Hazotte, A., Petitgand, G., & Barges, P. (2012). Evolution of microstructure and mechanical properties of medium Mn steels during double annealing. Materials Science and Engineering: A, 542, 31-39.
[Lee2015] Lee, Y. K., & Han, J. (2015). Current opinion in medium manganese steel. Materials Science and Technology, 31(7), 843-856.
[Callahan2019] Callahan, M., Perlade, A., & Schmitt, J. H. (2019). Interactions of negative strain rate sensitivity, martensite transformation, and dynamic strain aging in 3rd generation advanced high-strength steels. Materials Science and Engineering: A, 754, 140-151.
[Lamari2020] Lamari, M., Allain, S. Y., Geandier, G., Hell, J. C., Perlade, A., & Zhu, K. (2020). In situ determination of phase stress states in an unstable medium manganese duplex steel studied by high-energy X-ray diffraction. Metals, 10(10), 1335.

Compétences

• The researcher should have a good ability to work as a team, in a multicultural and international environment and integrate into a collaborative project involving industrial partners. The person recruited should expect long stays at the different partner sites of the project.
• The researcher should have a good ability to use and develop machine learning tools (implementation, capacity to develop own codes, datamining, use of thermodynamic databases, ..)
• Knowledge in physical metallurgy (microstructure analysis, phase transformation, thermodynamic, deformation mechanisms) would be necessary to conduct the project
• Experience in characterization of microstructures by scanning electron microscopy would be appreciated.

Contexte de travail

• The researcher should have a good ability to work as a team, in a multicultural and international environment and integrate into a collaborative project involving industrial partners. The person recruited should expect long stays at the different partner sites of the project.
• The researcher should have a good ability to use and develop machine learning tools (implementation, capacity to develop own codes, datamining, use of thermodynamic databases, ..)
• Knowledge in physical metallurgy (microstructure analysis, phase transformation, thermodynamic, deformation mechanisms) would be necessary to conduct the project
• Experience in characterization of microstructures by scanning electron microscopy would be appreciated.

Contraintes et risques

/

Informations complémentaires

PhD in Materials Sciences preferred