LDRD Seminar: June 25, 2019
Three Argonne researchers will discuss their Laboratory-Directed Research and Development (LDRD) sponsored work at the LDRD Seminar Series presentation Tuesday, June 25, 2019, at 12:30 p.m. in Building 212, Room A157. All are welcome to attend.
Visit the LDRD website to view upcoming seminars.
“Industrial-Scale Brain Mapping at a National Laboratory,” by Neuroscience Researcher Bobby Kasthuri (BIO)
The Kasthuri lab at the University of Chicago and Argonne National Laboratory is pioneering new techniques for brain mapping of the fine structure of the nervous system at industrial scale. These developments include large-volume automated electron microscopy for mapping neuronal connections at the nanoscale, synchrotron source X-ray microscopy for mapping the cellular composition of entire brains, and combining both with cell type-specific labeling for multiscale, multimodal brain maps. These tools have been applied to brains from octopuses and squids to primates and mice in the service of answering the questions: How do brains grow up and age, and how do brains differ across individuals and across species, and how can we being to reverse brain function in silico?
Bobby Kasthuri is the first neuroscience researcher at Argonne and an assistant professor in the Department of Neurobiology, University of Chicago. He has an M.D. from Washington University School of Medicine and a D.Phil. from Oxford University where he studied as a Rhodes scholar. As a postdoctoral fellow, Kasthuri developed an automated approach to large volume serial electron microscopy (‘connectomics’). Currently, the Kasthuri lab continues to innovate new approaches to brain mapping including the use of high-energy X-rays from synchrotron sources for mapping brains in their entirety. The Kasthuri lab is applying these techniques to in service of answering the question: How do brains grow up, age, and degenerate?
“Deep Learning and Dynamic Sampling for Smart Data Acquisition in Scanning Electron Microscopy,” by Materials Scientist Charudatta Phatak (MSD)
In conventional point-based scanning microscopy for imaging or spectroscopy, each pixel measurement can take up to a few seconds, which can translate into several hours of data acquisition time for large image sizes. This is often true for energy-dispersive X-ray spectroscopy (EDX) in a scanning electron microscope (SEM) or scanning transmission electron microscope (STEM) which is widely utilized in materials science for determining elemental compositions. Due to longer dwell times, the sample is exposed to high energy electrons which can result in radiation damage.
A dynamic sampling method based on supervised learning algorithm and convolutional neural networks (CNNs) for data acquisition in a SEM demonstrated results for two modalities: EDX mapping and secondary electron imaging. For EDX mapping, we developed a method using CNNs that uses a dictionary for training and classification of the EDX spectra. This method can achieve high-quality elemental maps with as low as 5 percent sampling. For SE imaging, a method using deep neural networks predicted the optimal sampling locations based on a set of training images, and then reconstructed the final image. The network can be pretrained using generic images available online, and was capable of reconstructing images with high signal to noise ratio with sampling as low as 20 percent.
The impact of this project is that our methods can be used for beam-sensitive materials and the high-throughput materials characterization needed for applications such as additive manufacturing.
Charudatta Phatak is a materials scientist at Argonne. He received his Ph.D from Carnegie Mellon University in materials science and engineering in 2009. His research interests focus on exploring magnetic frustration and physical curvature effects in patterned nanostructures, understanding grain boundary effects in solid-state fuel cell electrolytes, and lithium-ion batteries. He is also interested in developing machine learning methods for electron microscopy and data curation for microscopy data. He has authored over 60 publications and reviews in the field.
“A Conveyer Belt of Nanoliter to Picoliter Droplets for Hard X-ray Pump-Probe Experiments,” by Principal Mechanical Engineer Kamlesh Suthar (AES)
Acoustic levitation of the sample provides a contact-free sample environment and circumvents possible contamination from interaction. Investigation of liquid samples by X-ray can be challenging when unwanted interaction is possible within container. A sample management system (SMS) is developed that will provide a high degree of control for designer quantities of precious aqueous/liquid systems such as nanoparticles (NP), bio-molecules, fuels, and photocatalytic compounds. The SMS has targeted two critical areas in sample management that impose significant challenges to current experimental efforts: sample consumption and material damage. Sample consumption is a significant concern for free jets and continuous flow devices. The levitator can create and hold droplets on-demand to allow delivery of a controlled amount of sample to be delivered into the interaction region with nearly perfect utilization efficiency. We are addressing the need for on demand refreshing of systems that are permanently altered or damaged by the combination of X-ray and laser exposure.
Existing tools offer levitation at different frequencies. Droplet levitation can be achieved via automatic loading into the acoustic trap using micro-dispensers. The first laser pump X-ray probe SAXS experiments in an acoustic levitator will be discussed. Results from a pump-probe experiment where colloidal solution of 15-nm gold NPs were excited using high power laser and the lattice expansion was measured using time-resolved SAXS. The excellent droplet stability enables repeatability of the measurements with and without the application of the external stimuli.
Kamlesh Suthar is a principal mechanical engineer working in the mechanical engineering department in the APS Engineering Support division. Suthar’s research includes development of sample management systems for X-ray community, energetic nanomaterials, additive manufacturing, acoustic levitation, and multiphysics finite element analysis. He has been with Argonne since 2010, and received his M.S. and Ph.D from Western Michigan University.