In the CM³ Lab, we make scientific and technical advances in materials, mechanics, and manufacturing.

From the scientific perspective, we investigate processing-structure-property relationships in engineering structural materials.

From the technical perspective, we build software infrastructure for autonomous materials laboratories.

For the processing-structure relation, we use the discrete element method for powder dynamics (using FLOW-3D AM), thermo-fluid model for melt pool (using FLOW-3D AM), and CALPHAD-informed phase field model for solidification (using Thermo-Calc and MOOSE).

For the structure-property relation, we mainly work with two multiscale modeling frameworks. The first framework involves concurrent multiscale modeling based on a coarse-grained atomistic method (using PyCAC). The second framework concerns sequential multiscale modeling, which links density functional theory (using VASP), atomistic simulation method (using LAMMPS), phase-field model (using PFDD and MOOSE), discrete dislocation dynamics (using OpenDiS), and crystal plasticity model (using ρ-CP).

We establish processing-structure-property relationships in materials using machine learning (ML), with a focus on neural networks and Gaussian processes.

We develop ML interatomic potentials and foundation potentials for atomistic simulations.

We also employ generative large language models for domain-specific applications in STEM.

We develop other software components for autonomous materials design, such as multi-agent framework, multi-objective optimization, and experimental scheduler.

• PyCAC

  - Python-scripted concurrent atomistic-continuum (CAC) simulation environment

• PFDD (code & overview)

  - Phase-field dislocation dynamics

• Code repositories on GitHub