|
Sponsored by the Air Force Office of Sponsored Research Program Monitor Dr. Ali Sayir |
|
Participating Universities and Personnel |
|
Georgia Institute of Technology |
Role: Cation intercalation biomimetic synapses, neurons and axons, materials synthesis, device fabrication, characterization and modeling. The Doolittle group will investigate materials in which Li and other cations diffuse and drift in response to electric fields, light and thermal stimulus. This unique set of materials allows for short, medium and long term plasticity and voltage gated hysteretic switching in neuromorphic devices all within the same materials system. Some of the many goals of the Doolittle effort will be to team with other experts in theory (Lee) , spectroscopic characterization (Piper), and thermal science (Graham) to understand the role of interface layers and their electrochemistry, ion solubility in contacts and poorly understood phase changes to explain the complex and opportunistic behavior of these remarkable materials and devices. Doolittle will also team with Raychowdhury and Vogel to implement these cation devices in neuromorphic circuits. |
|
Dr. W. Alan Doolittle |
|
Joseph M. Pettite Professor Electrical and Computer Engineering 777 Atlantic Dr. Atlanta, GA 30332-0250 alan.doolittle@ece.gatech.edu |
|
Dr. Samuel Graham |
|
Eugene C. Gwaltney, Jr. School Chair and Professor in the Woodruff School of Mechanical Engineering He holds a courtesy appointment with the School of Materials Science and Engineering and a joint appointment with the Energy and Transportation Science Division at Oak Ridge National Laboratory. 771 Ferst Dr Atlanta, GA 30332
|
|
Role: The Graham group will develop electro-thermal models to guide the design, formation, and analyze the performance synaptic materials and devices for neuromorphic computing. This work will require the measurement of thermophysical properties of synaptic materials for use in the models. The group will perform Raman, SEM, TEM, SIMS, and HERMES in order to characterize the relevant structural and chemical features of the materials to better understand thermophysical properties. Results will be compared to DFT calculations to provide additional information. To verify the predictions of the models, the Graham group will perform both Raman and transient thermoreflectance imaging to characterize the device thermal response. The final goal will be the creation of accurate models that can be used in the development of neuromorphic components. |
|
Binghamton University |
|
Dr. Louis Piper |
ProfessorPhysics DepartmentBinghamton University (a State University of New York)Director of the Materials Science & EngineeringDirector of Institute for Materials Research (IMR)Binghamton University – State University of New YorkPO Box 6000, Binghamton, New York 13902-600 |
Role: Biomimetic material and device characterization via advanced spectroscopic means. Our research focuses on x-ray spectroscopy techniques, which involve the photoexcitation and subsequent relaxation of electrons to probe the chemical configuration and electronic structure of bulk and/or surface regions of materials. These studies are performed within our in-house UV/X-ray photoemission spectrometer and at dedicated Synchrotron facilities across the U.S. (e.g. National Synchrotron Light Source and Advanced Light Source) and abroad. Our most commonly used techniques include soft (1486eV) and hard (3-6KeV) x-ray photoemission spectroscopy (XPS) and soft (O K-edge and TM L3,2-edges) x-ray absorption/emission spectroscopy (XAS/XES). More recently we have been implementing synchrotron x-rays to obtain x-ray of absorption of transition metal K-edges. XAS of transition metals gives information in the form of X-ray Absorption Near Edge Structure (XANES) and Extended X-Ray Absorption Fine Structure (EXAFS). Both of these sets of information not only give information about an absorbing atom's chemical state but also to the chemical coordination environment. The XANES region can be very sensitive to transition metal oxidation state. The backscattering nature of photoelectrons results in the EXAFS region, which can resolve the local structure information. With these elaborate technologies, we can look into the quantum nature of adaptive oxide materials and map out the density off electron states vs energy (the quantum mechanical view) as the device transitions from one stat/phase to another. |
|
Dr. Wei-Cheng Lee |
|
Assistant Professor Department of Physics, Applied Physics, and Astronomy Binghamton University – State University of New York PO Box 6000, Binghamton, New York 13902-600
|
Role: First principals modeling for clarification of advanced spectroscopic techniques and materials exploration. Dr. Lee is focusing on advanced “beyond DFT” calculation methods to explain and predict quantum mechanical behavior such as the density of electronic states relationships to structure, phase, domain size and chemistry. He plays a vital role in not only explaining the spectroscopic data from the Piper group but also modeling the ion movement, conduction and energy relationships in the adaptive oxide materials. |
|
Dr. Eric Vogel |
|
Professor, School of Materials Science and Engineering |
Role: Anion filamentary biomimetic synapses, neurons and axons, materials and devices. synthesis, device fabrication, characterization and modeling. Memristors used as electrical synapses consist of a Metal-Insulator-Metal nanostructures: Two metallic electrodes and one dielectric The switching from high resistance to low resistance state is caused by the creation of a localized conductive path (also known as conducting filament). The origin of the resistive switching (or insulator to metal transition) is well accepted as due to the formation of a filament in the active layer when a voltage is applied between the two electrodes. However, the mechanism is still not fully explained in the literature. As such, we are interested to study the formation of CFs in the state-of-the-art materials (for e.g. HfO2 and HfTiOx) by examining the effects of 1) applied voltage, 2) oxygen vacancies/ ions (in the active layers), 3) local heating, and 4) metal-electrode interfaces. In this work, a new strategy is used to design, prepare, and characterize the memristor, and to compare its performance for different applications. Our aim is to understand the atomic scale behavior of a memristor, including the ionic conduction and transport when forming the CF across the metal-dielectric, and to develop an efficient memristor with high endurance and stability. |
|
Dr. Arijit Raychowdhury |
|
Professor School of Electrical and Computer Engineering,
|
Role: Biomimetic Circuit, System and Architecture fabrication, characterization and modeling. Dr. Raychowdury holds a key gatekeeper position within CEREBRAL. He is responsible for ensuring all the materials science efforts are directed toward practical neuromorphic uses and is developing new architectures and algorithms that will utilize the unique properties found in the CEREBRAL material systems. He is pioneering both the surprisingly beneficial role stochastic variation (which is inherent in the adaptive oxides ) has on the learning process and developing methods to use adaptive oxides enabling autonomous learning. |
