Despite the massive development (design, assembly) and operational (energy, human, maintenance) costs, the fastest supercomputers cannot match the energy efficiency of the human brain. In fact, the human brain is about 132,000 times more energy efficient than the fastest supercomputer. However, there exists a notable gap in understanding the material processing, microstructure and property relationship of switching mechanisms in these material systems, as they involve intricate changes in physical and chemical behavior. In the non-volatile memory switching devices, the resistance depends on the current and voltage history, and thus, it can be controlled by external factors, remembered by the system. Non-volatile switching memory devices have recently gained interest to provide a unique potential to enable the realization of human brain like neuromorphic computing efficiency. By mimicking how human beings process information, the proposed brain-inspired computing engine has the capability to provide high-precision and parallel operations; hence, significantly reduce power consumption and design area.  This workshop brings together world-class researchers presenting various aspects of their current research in the area of memristors and their applications in neuromorphic computing.

CMOS technology has been the mainstream hardware technology for the development of ubiquitous information technology so far. In the era of ‘big data’ and ‘Internet of Things’ nowadays, the traditional computing architecture based on CMOS hardware has become increasingly inefficient to support Artificial Intelligence (AI) and Machine Learning (ML), which necessitates some emerging technologies, such as memristive technology. Memristive devices have become a promising candidate to enable bio-inspired computing with much improved efficiency and throughput. Such computing can be implemented on a Resistive Neural Network with memristive synapses and neurons or a Capacitive Neural Network with memcapacitive synapses and neurons. I will first briefly introduce the traditional drift memristors for deep learning accelerators with supervised online learning. I will then focus on diffusive memristors and discuss how to use such dynamical devices to enable neuromorphic computing related applications, including timing selector, fully memristive neural networks with unsupervised learning, and memcapacitive neural networks capable of associative learning and efficient dot product computing. 

Non-volatile resistive memory devices offer the potential for low power, highly scalable implementation of non-von Neumann computing architectures. Over the past 10 years, my research group has focused on fabrication and integration strategies for CMOS-compatible, non-volatile memory devices (aka: memristors). Recently we have been exploring the potential of these devices to act as neuronal synapses in neural networks and for in-memory computing applications. One challenge with introducting novel memory devices is integration with traditional CMOS processing and manufacturing. In this talk I will describe the unique technological aspects of resistive random access memory (RRAM) based non-volatile memory devices, our integration approach using 65nm CMOS in the Albany Nanotech 300mm foundry, as well as demonstration of this novel hardware for neuromorphic and in-memory computing applications. 

PCM (phase-change memory) is an important class of data storage, operating based on the Joule heating-induced reversible switching of chalcogenide alloys. Recently, hardware implementation of deep learning has made a remarkable stride with the emerging non-volatile memory technology, and PCM is among the most promising candidates to accelerate this technological trend. In this talk, state-of-the-art research on PCM-based neuromorphic computing will be reviewed along with the promises and challenges as compared with other memristor technologies based on transition metal oxides, ferroelectrics, and magnets. Additionally, how best the nanoscale PCM device can be implemented for neuromorphic computing will be discussed in the context of down-scaling and energy efficiency.

Resistive Random Access Memory (RRAM) Memristive Devices have gathered tremendous research interests for high density data storage and neuromorphic applications. This talk will cover some of the challenges with two-terminals (2T) RRAMs and propose strategies to integrate gate in RRAMs to result in three-terminals (3T) RRAM devices. The talk will present experimental data on 3T RRAMs and discuss applications development in the domain of neuromorphic computing where gate-control can provide opportunities for implementing advanced training algorithms.

A non-volatile switching memory platform has recently attracted great interest as it provides a unique potential to enable the realization of human brain like neuromorphic computing efficiency.  Memristors are novel nanoscale non-volatile resistive memories with a high resistance ratio above 1000 between high resistance state (ON state) and low resistance state (OFF state), good temperature tolerance, long-term durability, high tunability with nanosecond pulses, which are very attractive for the neuromorphic computing.  In order to better understand the material processing, microstructure and property relationship of switching mechanisms in the memristor devices, computational methodologies and tools were developed to predict I-V characteristics of oxide-based memristor devices in an 1T1R configuration subject to externally biased voltages.  A multiphysics model based on coupled partial differential equations for electrodynamic and thermal transport phenomena were solved for the high- and low-resistance states upon formation, growth and destruction of a conducting filament (SET state), and those of an insulating gap (RESET state), which represent the migration of oxygen vacancy in an oxide exchange layer.  Two discrete models for four modules (ON, OFF, SET and RESET) were implemented in a commercial multiphysics finite element software, COMSOL.  Various model parameters including the material properties for two oxide materials (TaO_x and HfO_x) were determined from a series of sensitivity study.  Model predictions were compared with experimental data measured with the two oxides.  Once validated, the model could further be used to identify critical parameters (material type, morphology and doping in the metal oxides as well as surrounding electrodes) affecting the device performance and durability of the memristor devices during long-term cyclic operations.

Biography:

Dr. J. Joshua Yang is a professor of the Department of Electrical and Computer Engineering at the University of Southern California. He was a professor of the ECE department at the University of Massachusetts Amherst between 2015 and 2020. He spent about 8 years at HP Labs between 2007 and 2015, leading the emerging devices team for memory and computing. His current research interest is Post-CMOS hardware for neuromorphic computing, machine learning and artificial intelligence, where he published several pioneering papers and holds 118 granted and about 60 pending US Patents. He is the Founder Chair of the IEEE Neuromorphic Computing Technical Committee, a recipient of the Powell Faculty Research Award and a recipient of UMass distinguished faculty lecturer and UMass Chancellor's Medal, the highest honor of UMass. He was named as a Spotlight Scholar of UMass Amherst in 2017.He is a Clarivate™ Highly Cited Researcher in the field of Cross-Field and serves on the Advisory Boards of a number of prime international journals and conferences. Dr. Yang is also an IEEE fellow for his contributions in resistive switching materials and devices for nonvolatile memory and neuromorphic computing.

Prof. Cady obtained his BA and Ph.D. from Cornell University in Ithaca, NY. He is currently an Empire Innovation Professor of Nanobioscience in the College of Nanoscale Science & Engineering at SUNY Polytechnic Institute, and is the Interim Vice President of Research. Prof. Cady has active research interests in the development of novel biosensor technologies and biology-inspired nanoelectronics, including hardware for neuromorphic computing. He has published over 150 peer reviewed scientific papers and is an inventor on 11 patents. His research has been supported by the NIH, NSF, AFRL, ARL, DOE, ONR, SRC, as well as multiple industry partners.

Dr. Ahn is currently an Assistant Professor of Electrical Engineering at the University of Texas at San Antonio. Previously, he served as a Sr. Process Engineer at Apple, Inc. He received his Ph.D. at Stanford University in 2015. He is the author of over 75 peer-reviewed research papers in nanoelectronics. Dr. Ahn has been the recipient of numerous awards and honors, including the AFRL ML-RCP award (2022), the AFOSR grant in Quantum Electronic Solids (2019), and the NSF Eager grant (2019). He is currently serving as the concentration chair for Electronic Materials and Devices at UTSA. Dr. Ahn has also served as the technical committee member of the IEEE Electron Devices Society (EDS) for optoelectronic devices. His research interest includes nanoscale materials and devices for emerging computing paradigm.       

Dr. Rashmi Jha is a Professor in Electrical Engineering and Computer Science (EECS) Department at the University of Cincinnati, Cincinnati, USA. She worked as a Process Integration Engineer for Advanced CMOS technologies at IBM Semiconductor Research and Development Center, East Fishkill, NY, USA prior to moving to the academia. She finished her Ph.D. and M.S. in Electrical Engineering from North Carolina State University, Raleigh, North Carolina, USA in 2006 and 2003, respectively, and B.Tech. in Electrical Engineering from Indian Institute of Technology (IIT) Kharagpur, India in 2000. She has been granted 13 US patents and has authored/co-authored several publications. She has been a recipient of Summer Faculty Fellowship Award from AFOSR, USA in 2021 & 2017, CAREER Award from the National Science Foundation (NSF), USA in 2013, IBM Faculty Award in 2012, and IBM Invention Achievement Award in 2007. She is the director of Microelectronics and Integrated Computing Systems with Neuro-centric Devices (MIND) laboratory at the University of Cincinnati. Her current research interests lie in the areas of Advanced CMOS and Beyond CMOS Devices, Artificial Intelligence, Neuromorphic and Brain-Inspired SoC, Cybersecurity with emphasis on Microelectronics, and Neuroelectronics.

Dr. Sangwook Sihn is a Senior Research Engineer at the University of Dayton Research Institute working in the Air Force Research Laboratory as an on-site contractor for more than 20 years in the field of analysis and development of advanced materials and structures.  He has conducted researches on the mechanical, thermal, electronic and electromagnetic analyses of advanced materials including, but not limited to, fiber-reinforced polymeric matrix composites and nanocomposites, carbon nanotubes, graphene and graphites, foams and phase change materials to be used for efficient mechanical, thermal, electronic and electromagnetic applications.  He has developed various multiscale computational models and softwares to be used for the analysis and design of multifunctional advanced materials and their structures.  He has contributed to 5 book chapters, 40 journal publications, and more than 100 conference presentations and online tutorials.