1 – Defect modeling for electronic and optical properties
The presence of defects in solid materials, whether intrinsic (vacancies, substitutions…) or extrinsic (dopants, interfaces…), strongly influences their electronic and optical properties. Modeling these defects is therefore essential for understanding, predicting and optimizing the performance of devices such as photovoltaic cells, sensors or optoelectronic materials.
This research theme is based on theoretical chemistry and materials physics approaches, combining quantum calculations (DFT, GW, etc.) with specific tools developed in-house, such as PyDEF for the analysis of defect formation energies and Hylight for the simulation of luminescence spectra using vibronic coupling.
The aim is to link the nature of defects to their microscopic and macroscopic effects, based on concrete cases synthesized in the laboratory in collaboration (Sb₂Se₃, CuInSe₂…). This work makes it possible to establish fine correlations between structure, defects and properties, to guide the design of high-performance functional materials.
Applications: Photovoltaics, Optoelectronics, Semiconductor electronics, Energy materials, Optical sensors, Molecular functional materials, Light detection and conversion.
Keywords: Crystalline defects, Density functional theory (DFT), Electronic properties, Optical spectroscopy, Defect formation, Multiscale modeling, Semiconductor materials, Photoluminescence, Emerging materials, Ab initio simulation, Structure-properties link, Chemical bonding and bonding index.
People involved: Stéphane Jobic, Camille Latouche
2 – Simulating the spectroscopic properties of inorganic and hybrid materials
Spectroscopy (optical, vibrational, electronic…) is a key tool for characterizing materials, but its interpretation can be complex, especially in inorganic or hybrid systems with correlated electronic effects or strong coupling with vibrations.
Theoretical modelling is used to simulate these spectra in order to offer a detailed reading, assign bands and better understand the underlying mechanisms. Using ab initio approaches (DFT, TD-DFT, post-HF methods, etc.), this work aims to reproduce experimental spectroscopic signatures, analyze the origin of transitions and predict the effects of structural or electronic modifications.
This research applies to a variety of systems: transition metal complexes, inorganic semiconductors, hybrid perovskites and defect materials.
Tools developed in-house, such as Hylight, can be used to simulate vibronically resolved spectra in solids. The aim is to better understand the links between structure, dynamics and spectroscopic response, to guide the development of innovative materials.
Applications: Optical spectroscopy (UV-Vis, IR, Raman, etc.), Materials photophysics, Coordination chemistry, Energy materials (photovoltaics, photocatalysis), Hybrid materials (organic-inorganic), Detection and sensors, Lighting and display materials.
Keywords: Crystal defects, Density functional theory (DFT), TDDFT, Phosphorescence, Electronic properties, Optical spectroscopy, Defect formation, Multiscale modeling, Semiconductor materials, Photoluminescence, Emerging materials, Ab initio simulation, Chemical bonding and bonding index, Structure-properties link,
People involved: Stéphane Jobic, Camille Latouche
3 – Accelerated materials design and synthesis assisted by artificial intelligence
The emergence of artificial intelligence (AI) is opening up new prospects for accelerating the discovery, optimization and synthesis of materials with targeted properties. This research sub-theme aims to exploit machine learning techniques to identify new compounds (oxides, chalcogenides, hybrids…) and optimize both their fabrication processes and their functional properties (electronic transport, photoluminescence, optical response…).
The team develops and applies a variety of models, such as convolutional neural networks (CNN), random forests (Random Forest) and Bayesian optimization, to efficiently explore the space of compositions and experimental conditions. These approaches are closely coupled with data from high-throughput experiments, thanks in particular to a robotized solution synthesis platform set up as part of the PEPR DIADEM program (Hiway-2-Mat).
Particular emphasis is also placed on the automatic identification of phases from X-ray diffraction data, including in situ, to enable real-time monitoring of crystal phase formation and a better understanding of crystallization mechanisms.
By combining algorithmic exploration, predictive modeling, automated structural analysis and experimental feedback, this work helps guide the design of innovative materials for a wide range of applications.
Applications: Photoluminescence, Energy materials, Thermoelectrics, Optics, Emerging multifunctional materials.
Keywords: Machine learning, New materials, Neural networks, Bayesian optimization, High-throughput synthesis, Materials discovery, Predictive modeling, Artificial intelligence, Chalcogenides / oxides / hybrids, Robotic platforms, Automatic phase identification, In situ X-ray diffraction, Crystallography
People involved: Romain Gautier, David Berthebaud, Olivier Hernandez
Current research projects: ANR AI-Unclon, PEPR DIADEME MADNESS project, PEPR DIADEME HIWAY2MAT project, PEPR DIADEME DREAM-BIO project
Major publications:
- “Role of Hydrogen Bonding on the Design of New Hybrid Perovskites Unraveled by Machine Learning” R. Laref, F. Massuyeau, R. Gautier, Small, 20 (5), 2306481 (2024)
- “Perovskite or Not Perovskite? A Deep Learning Approach to Automatically Identify New Hybrid Perovskites from X-ray Diffraction Patterns”, F. Massuyeau, T. Broux, F. Coulet, A. Demessence, A. Mesbah, R. Gautier, Advanced Materials 34, 41, 2203879 (2022)
- “Machine Learning Guided Design of Single-Phase Hybrid Lead Halide White Phosphors”, H. Yuan, L. Qi, M. Paris, F. Chen, Q. Shen, E. Faulques, F. Massuyeau, R. Gautier, Advanced Science, 19 (8), 2101407 (2021)
- “Machine Learning Identification of Experimental Conditions for the Synthesis of Single-Phase White Phosphors
- H. Yuan, R. Génois, E. Glais, F. Chen, Q. Shen, L. Zhang, E. Faulques, L. Qi, F. Massuyeau, R. Gautier, Matter, 12 (1), 3967-3976 (2021)
