Image by Ross Harder

MAUI: Modeling, Analysis, and Ultrafast Imaging

A DOE LDRD project researching the combination of light source imaging and molecular dynamics modeling through the common language of data analysis. In conjunction with the Argonne National Laboratory Integrated Imaging Initiative.

Investigators

Tom Peterka (MCS), Ian McNulty (NST), Nicola Ferrier (MCS), Ross Harder (XSD), Todd Munson (MCS), Sven Leyffer (MCS), Subramanian Sankaranarayanan (NST), and Haidan Wen (XSD)

Postdocs

Kiran Sasikumar (NST), Mathew Cherukara (XSD), Andrew Ulvestad (XSD), and Youssef Nashed (MCS)

Scientific Opportunity

Integrating ultrafast time-resolved imaging with large-scale molecular dynamics modeling and in situ data analysis and visualization in order to design, conduct, and understand spatiotemporal measurements can provide crucial insights for energy research. The temporal behavior of in situ externally stimulated materials beyond equilibrium can lead to breakthroughs, for example, in heat dissipation of next-generation semiconductors, conversion of wasted heat into electricity in thermoelectric materials, and electrochemical processes across liquid-solid interfaces in water purification. All these diverse applications share a common behavior: they transport energy through phonons (sound waves that carry heat) in a time-evolving crystal lattice. We anticipate that our integrated approach to predict, image, and analyze phonon dynamics can be applied to other externally stimulated (for example, heated, pressurized, laser-pumped, acid-dissolved, or electromagnetically induced) systems measured by various imaging techniques including x-ray, electron, and optical microscopy.

Context

Computing capability in molecular dynamics is growing exponentially at leadership computing facilities. Likewise, new analysis and visualization techniques for 4D and higher-dimensional data are being researched. Recently, experimentalists have begun to conduct imaging in time-evolving lattice dynamics. demonstrated a ``pump-probe'' experiment, whereby an infrared laser pumps (pulses) the material sample followed by x-ray diffraction probes at various times after the pump event. Never before, however, have all three efforts---forward modeling through atomistic simulations, multimodal imaging, and reverse modeling through reconstruction and analysis---been combined in an interoperable method that provides iterative feedback from each component to the others. In the context of the APS upgrade for transformational sciences, this proposal addresses the need for high speed, high-volume data processing for novel time-resolved imaging; and it aligns with other Argonne strategic directions in hard x-ray sciences and advanced computing.

Benefits

Lattice vibrations in individual nanoparticles affect phase transitions, bond softening/hardening, ferroelectricity, solid/liquid interfaces, heat dissipation, phononic local structure, phase front propagation, and spectrometry. Understanding such phenomena can enable energy applications such as photocatalysis, photonics, thermoelectrics, semiconductor design, groundwater photo remediation, and heat transfer in battery solute-solvent interfaces. Moreover, integrating forward modeling and inverse reconstruction with experiment will enable improved utilization of facilities at the APS and CNM. The most basic properties of the measured images, for example, where and how frequently to measure, can be aided by atomistic modeling and near-real-time data analysis. Likewise, the veracity of simulated results, crucial for the outcome of the experiments, will be improved by timely analysis of reconstructed images. Although we will concentrate on x-ray coherent diffractive imaging (CDI), we anticipate that other imaging methods (such as optical and electron microscopy) used to observe time-resolved in situ experiments can benefit from the product of our research.

Approach

We will focus on two energy materials problems: (1) heat dissipation in aqueous solution of nanoparticles upon laser heating for water purification, and (2) heat transport in nonaqueous battery electrolytes (alkyl carbonates) containing diamond nanoparticles for efficient thermal management. In the context of these problems, we will combine three individual components, molecular dynamics modeling, ultrafast imaging, and data analysis/visualization, through the development of common data models and data adapters that will facilitate the data movement between components and will innovate each individual component.