Contact 

Rémi Dingreville is a Distinguished Member of Technical Staff at Sandia National Laboratories and currently serves as the CINT Co-Thrust Leader for the In-Situ Characterization and Nanomechanics thrust. Rémi employs and combines various theoretical and computational techniques (molecular dynamics, cluster dynamics, phase field, mean field, data mining) to understand and characterize materials aging and performance in solid matter. His research emphasizes designing materials with enhanced functionality through understanding and control of interface and defect phenomena.

Expertise

• Scientific machine learning (SciML): Developing physics-informed deep learning frameworks (operator learning, UNets, autoencoders) to create AI surrogates for complex materials simulations.
• Multiscale materials modeling: Bridging atomic and mesoscale models using molecular dynamics, phase-field modeling, dislocation dynamics, and crystal plasticity to understand materials reliability in extreme environments
• Active learning and autonomous discovery: Utilizing machine-learning-guided approaches, such as Bayesian optimization and genetic algorithms, to discover novel materials with improved performance in extreme environments
• Materials reliability and manufacturing: Characterizing process-structure-performance linkages in advanced manufacturing
• Nanomechanics: Investigating the stability and response of grain boundaries, interfaces, nanostructured materials extreme environments including irradiation, corrosion, and mechanical loading
• Phase-Field Modeling: Expert-level simulation of microstructure evolution using the MEMPHIS phase-field simulator and development of Accelerated Phase Field capabilities via recurrent neural networks for the simulation of evolving microstructural problems
• Atomistic Simulations: High-fidelity modeling of nanomechanics problems
• Physics-Informed Machine Learning: Development of physics-informed ML models for uncertainty quantification and materials "Digital Twins" for both metallic and composite materials.
• Machine Learning for Data Analysis: Multi-modal machine-learning data analysis for experimental and simulation data including scanning probe microscopy, electron microscopy (TEM/SEM), and spectroscopy on the experimental side, enabling high-throughput nanometrology and the prediction of physical properties directly from multi-dimensional datasets
• Active-Learning/Optimization: Integrating multiscale simulations with high-dimensional optimization workflows (e.g., Bayesian optimization, genetic algorithms) to explore vast design spaces. These capabilities are applicable to autonomous synthesis (self-driving labs), combinatorial material screening, and process optimization

Capabilities

• Atomic Force Microscopy-based Modulated NanoIndentation (MoNI)
• Mesoscale Multi-Phyisics PHase Field Simulator (MEMPHIS)
• Nanomechanics Virtual Laboratory

Education