1 of 1

Using ParMOO to Drive Materials Design

Research Details

    • FASTMath-funded software ParMOO (https://github.com/parmoo/parmoo) was originally designed for solving multiobjective optimization problems involving computationally expensive simulations
      • Deployed here in a time-expensive synthesis and characterization loop
    • Traditionally, material scientists choose manufacturing conditions via trial/error:
      • Expensive, in terms of time and raw material requirements
      • Involves tradeoffs between maximizing yield and minimizing unwanted byproducts
    • ParMOO generates CFR configurations in a closed, operator-free loop:
      • ParMOO sends candidate designs directly to CFR controller
      • CFR performs material synthesis and sends output to NMR
      • NMR characterization results sent to ParMOO

Scientific Achievement

FASTMath-funded multiobjective optimization software ParMOO is backbone for self-driving continuous-flow chemistry lab for materials design and manufacturing 

​

Significance and Impact

By integrating multiobjective optimization with a continuous-flow reactor (CFR) and nuclear magnetic resonance (NMR) spectroscopy, we save significant time and resources in materials design and discovery

FASTMath (PI Munson)�Argonne SDL (PI Stevens)

Setup: lab computer (left) with ParMOO generates design, CFR (middle) performs synthesis, NMR (right) processes output, which is returned to ParMOO

Tyler Chang, Stefan Wild (MCS)�Trevor Dzwiniel, Kris Pupek, Joe Libera (AMD)

ParMOO dynamically updates yield and byproduct �surrogates, then uses these models to balance tradeoffs