Argonne National Laboratory

The Postdoctoral Programs Office and the Postdoctoral Society of Argonne present the next Postdoctoral Research Seminar Wednesday, January 16, from noon to 1 p.m. in the Building 200 Auditorium.

Please bring your lunch and join us for some networking and science. Cookies will be provided.

Director’s Postdoctoral Fellow, Michael Burke (CSE), will present "Multi-Scale Informatics for High-Accuracy Modeling of Complex Systems: From Electrons to Energy Devices."

Abstract:
Complex chemical networks, which pervade combustion, atmospheric, materials, and biological systems, are inherently multi-scale phenomena. Molecular interactions control the rate of isolated elementary reactions; interactions among simultaneous elementary reactions control global system reaction rates; and global system reaction rates control global observables in ways often tightly coupled with transport phenomena. Complex chemical networks across diverse disciplines often contain the same theme: data from a single scale rarely inform models sufficiently.

This talk presents a new informatics-based, multi-scale approach for creating complex network models. The approach, multi-scale informatics, integrates information from a wide variety of sources and scales: ab initio quantum chemistry calculations of molecular properties, rate measurements of isolated reactions, and global measurements of complex multi-reaction systems. The key feature of the approach is two-way information flow—theory guides interpretations of complex experimental data, and experimental data provides tight constraints on the theoretical parameters. The resulting “model” is capable of uncertainty-quantified predictions of behavior at any scale informed by data at all scales.

Several key features of the present approach are demonstrated for the hydrogen peroxide decomposition mechanism, which contains highly debated reactions that contribute substantially to uncertainties in simulations of advanced engines. For example, the approach was used to resolve apparent inconsistencies in available data for the OH + HO2 = H2O + O2 reaction, where previous experimental interpretations suggested an anomalous temperature dependence that defied theoretical expectations. Further applications of the technique in low-temperature oxidation chemistry (relevant to advanced HCCI engine concepts) and solid electrolyte interphase formation (relevant to Li-ion batteries) will be briefly discussed. Future directions that incorporate electron correlation corrections, turbulent reactive systems, and robotic scientific “communities” will also be explored.