YIP: Multi-Functional Ferroelectric Nanoelectronics for Intelligent Sensing and Computing
Abstract
Project Summary/Abstract (Approved for Public Release)Artificial intelligence (AI) demands exceptional data collection and processing capabilities. Despite significant progress in algorithm development, particularly with artificial neural networks, a fundamental obstacle persists due to the limitations of the conventional von Neumann hardware architecture. In this architecture, the segregation of sensing, computing, and memory units leads to substantial challenges including data transfer overhead, heat generation, large footprint, and compromises in speed, power efficiency, and scalability. This well-recognized predicament is commonly referred to as the "von-Neumann bottleneck." To overcome this obstacle, a promising solution involves embracing non-von Neumann architectures for sensing and computing. These architectures integrate sensing, computing, and memory units in a manner that emulates the highly efficient physical structure of the human brain. This hardware paradigm, commonly referred to as "in-memory sensing and computing," requires the development of capable multi-functional materials. The quest for such materials has long been a prominent demand within the field.Ferroelectrics have emerged as highly competent materials for non-von Neumann computing, offering inherent multi-functionality in sensing, computing, and memory operations. They possess distinct advantages such as high-speed operation, low power consumption, and exceptional endurance, distinguishing them from other candidates for multi-functional operations. At the nanoscale, the realization of ferroelectric electronics enables higher storage density, faster data processing, reduced power consumption, and enhanced scalability. Furthermore, there is a strong motivation to explore ferroelectrics with new or improved functionalities as fundamental components for brain-inspired sensing and computing. By investigating and designing novel functionalities at the elemental level, we can broaden our understanding and enhance our capabilities in creating a wide range of brain-like functions.This proposal aims to develop a new class of nano-ferroelectrics that enable compact, efficient, and concurrent sensing, computing, and memory capabilities through nanoscale van der Waals stacking and sliding. A significant aspect of this approach is its capability to design and seamlesslyintegrate quantum properties, including topology, magnetism, and electron correlations, into ferroelectric physics#a task that is often challenging. This integration establishes a powerful and synergistic connection between these fields, unlocking vast potential for the development of innovative intelligent hardware applications driven by the unique attributes of quantum physics. The outcomesof this research, encompassing design principles, characterization techniques, and fundamental insights, will offer valuable guidance to the scientific community in the development of new nano-ferroelectrics. These discoveries will empower the engineering community to effectively utilize these materials in #in-memory sensing and computing# applications. These advancements align with the objective pursued by the ONR to achieve cognitive dominance by harnessing digital tools and AI for faster and more informed decision-making based on distilled information and data.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 24, 2024
- Source ID
- N000142412102
Entities
People
- Qiong Ma
Organizations
- Boston College
- Office of Naval Research
- United States Navy