Description du sujet de thèse

From sound (acoustic emission – AE), it is possible to detect various atomic scale processes (fracture, defect motion, structure transformation) in a material during deformation. But can we achieve the categorization of different types of such processes by sound? Can we also predict the crystalline structure itself by the induced sound and anticipate the micro-deformation, or even extrapolate the behaviour of other materials?

The goal of the project is to use machine learning (ML) to classify AE signals from microscopic samples captured during deformation. CASCADE-RAPIDE will employ micromechanical experiments to systematically study the AE signal properties under various conditions. ML algorithms and perceptual testing will be employed to analyse and classify the experimental data.

The aim of the PhD project is to investigate extreme deformation speeds, to analyse experimental AE signals based on wave properties by ML, and to provide insights on rapid deformation mechanisms at small scales. To reach this objective, CASCADE-RAPIDE will use a unique in situ setup, that couples micromechanical testing and AE signal detection at a wide strain rate regime (over 8 orders of magnitude). This device operates inside a scanning electron microscope (SEM), where the surface and microstructure of small pillars can be observed during deformation. The collected AE data will be processed by advanced ML-algorithms and human perceptual testing for classification.

Requirements:
- Master's degree, (or to be close to acquiring it) in experimental physics/materials science
- Skills in mechanics of materials and mechanical testing, electron microscopy characterization is a plus
- Affinity for programming, open to data science
- Good level of English (French is a plus)
- Ability to work in a team

Contexte de travail

1. Compression tests will be performed on FIB-milled micropillars at various strain rates, measured on different materials.
2. Artificial intelligence techniques will be employed to learn the long-term dependencies in the time domain, and in light of the deformation information.
3. Application of human perceptual testing to classify AE signals.

The coaching team: The experiments will be carried out at the LGF CNRS (Mines Saint Etienne) under the direction of Szilvia Kalacska and Guillaume Kermouche. The machining work by FIB will be carried out in the premises of Manutech-USD (St. Etienne). The ML part will be done in Cergy-Pontoise under the supervision of Alexandre Pitti and Péter D. Ispánovity.

The Host Institution: The research groups at LGF CNRS (SURF and PMM) are experts in the fields of metallurgy, mechanics of materials and surface functionality (i.e. against corrosion). They are committed to designing the new generation of metallic materials in relation to the use of new manufacturing processes and surface treatments. The LGF laboratory is involved in the Manutech network, which aims to put the Lyon-St-Etienne region at the forefront of surface manufacturing and tribology.

What we offer: Cutting-edge training in mechanics and materials; a stimulating and enriching research and teaching program, a large international network with the best scientists in the field (in France, Switzerland, Germany, etc.). The student will have access to the latest micromechanical equipment (Alemnis ultra-high-speed deformation setup) coupled with versatile characterization (in situ high resolution EBSD, in situ Acoustic Emission) techniques, and a dual beam FIB/SEM system dedicated to micro-sample preparation. Furthermore, the student will likely be involved in a large-scale facility measurement at the ESRF to study the evolution of dislocation structures by X-ray diffraction tomography.

Contraintes et risques

-