Unraveling Mössbauer spectra of Fe-based nanostructures
FIRST CALL
APPLICATION ID: ALL15
What we are looking for:
We are seeking innovative research proposals focused on the development of ML algorithms to identify the detailed system features contributing to a specific spectroscopic signal. Our main interests are Mössbauer and Raman spectroscopies, where we can generate a hierarchical workflow of training databases combining from ab initio data to measurements.
The context:
There is an urgent need to replace rare-earth (RE) magnets by sustainable alternatives, Fe nanostructures emerging as versatile and abundant candidates. Our ability to improve their efficiency to the level of REs strongly relies on improving characterization techniques that enable fast but accurate exploration of the widest range of Fe-based systems.
The problem to address:
Our groups at ICV (Adrián Quesada) and ICMM (Silvia Gallego) lead an international consortium involved in the development of an efficient scheme combining theoretical models together with fabrication and characterization tools to accurately explore Fe-based nanostructures in the search for the best permanent magnets. The proper interpretation of spectroscopic signatures is a bottleneck in this procedure. Our aim is to use machine learning (ML) algorithms to enable the identification of individual contributions to the spectroscopic signal, so as to provide an improved estimate of the material features giving rise to the spectrum.
The associated partners: The proposal is part of a wider international research carried by a consortium composed by experimental and theoretical research teams and industrial partners. Our expertise covers different spectroscopies, (both simulation and experimental data acquisition), from those targeted to Fe systems (Mössbauer) to broader scoped ones as Raman or X-ray photoemission. Stakeholders of our research include companies interested in the implementation of ML protocols in their magnet production chains (Encapsulae), and research partners with expertise on the development of spectroscopy techniques (IQF-CSIC, UCLouvain, Antioquia University).
Objectives:
- To provide a targeted ML method for a specific spectroscopy that can be trained on the basis of the appropriate database, enabling the efficient use of spectral signals to discriminate detailed features of magnetic nanostructures.
- To refine the interpretation of spectral signals based on the potential of ML algorithms to detect trends beyond human discrimination.
Expected Outcomes:
- To implement efficient ML algorithms in the exploration of large datasets for the spectral characterization of nanostructured magnetic materials, useful both for research purposes and innovative industrial processes.
- Ideally, the complete automation of the spectral analysis could enable the setup of an open access platform available to the research community, specialized in each type of spectroscopy.
