RII Track-4:NSF: Spin-orbitronics in quantum materials for energy-efficient neuromorphic computing

RII Track-4：NSF：量子材料中的自旋轨道电子学用于节能神经形态计算

基本信息

批准号：
2229498
负责人：
Fei Xue
金额：
$ 26.43万
依托单位：
University of Alabama at Birmingham
依托单位国家：
美国
项目类别：
Standard Grant
财政年份：
2023
资助国家：
美国
起止时间：
2023-02-01 至 2025-01-31
项目状态：
未结题

来源：
https://www.nsf.gov/awardsearch/showAward?AWD_ID=2229498&HistoricalAwards=false
关键词：
RII Track NSF Spin orbitronics

项目摘要

The rapid development of artificial intelligence relies on modern computing technologies. However, the existing von Neumann-based technology suffers from high energy consumption for data-intensive tasks. This high energy consumption may limit the future adoption of artificial intelligence technologies. Inspired by the biological brain, neuromorphic computing provides the promising technological capability to tackle this challenge by creating superior energy-efficient hardware for information processing. This research project exploits the superior nonlinear nonvolatile spin-related responses in quantum materials. The key challenge to implementing neuromorphic computing is to create artificial neurons and synapses with great energy efficiency. Spintronics study the interplay between electron spin transport and charge transport, so they naturally couple electronic and magnetic configurations, thus offering non-volatility and nonlinearity. Nonvolatile spintronics memory devices can emulate artificial synapses and nonlinear spin-torque nano-oscillators can emulate artificial neurons. The proposed research activities will provide a unique opportunity for students at the University of Alabama at Birmingham to gain computation skills and collaborate with distinguished scientists at the National Institute of Standards and Technology (NIST). Additionally, as a part of this project, the PI will develop a new advanced physics course about artificial intelligence and how to implement it with physical devices. This Research Infrastructure Improvement Track-4 EPSCoR Research Fellows (RII Track-4) project would provide a fellowship to an Assistant Professor and training for a postdoctoral fellow at the University of Alabama at Birmingham (UAB). Brain-inspired neuromorphic computing offers appealing technology capability for artificial intelligence applications. Spintronics devices, which couple electronic and magnetic configurations, can emulate synapses and neurons in an energy-efficient compact manner. The magnetic random-access memory can serve as the synapses and the spin-torque nano-oscillators can serve as the neurons. The quantum materials provide additional appealing features including more efficient control of magnetization and new functionalities due to the coupling of spin, orbital, and magnetization degrees of freedom. Understanding and modeling microscopic mechanisms with state-of-art first-principles methods of neuromorphic spintronics with quantum materials is the main goal of this proposal. The two main thrusts focus on utilizing the superior spin-orbitronics properties in quantum materials for artificial synapses and neurons. By collaborating with the experts at the NIST, the project will apply first-principles methods to calculate the band structures and spin dynamics in spin-orbit coupled quantum materials. This allows for the understanding of the microscopic mechanism of spin-orbit torque switching and dynamics and provides a pathway to improve the figure of merits. Project outcomes will also include illustration of how to utilize these nonlinear nonvolatile properties of spintronics devices into emulating neurons and synapses for neuromorphic computing. The proposed work will provide the physics foundation for implementing neuromorphic spintronics devices with emerging quantum materials such as two-dimensional materials and topological materials.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

人工智能的快速发展依赖于现代计算技术。然而，现有的基于冯·诺依曼的技术在数据密集型任务中存在高能耗问题。这种高能耗可能会限制人工智能技术的未来采用。受生物大脑的启发，神经形态计算通过创建用于信息处理的卓越节能硬件，提供了有前途的技术能力来应对这一挑战。该研究项目利用了量子材料中卓越的非线性非易失性自旋相关响应。实现神经形态计算的关键挑战是创建具有极高能源效率的人工神经元和突触。自旋电子学研究电子自旋传输和电荷传输之间的相互作用，因此它们自然地耦合电子和磁性配置，从而提供非易失性和非线性。非易失性自旋电子存储器件可以模拟人工突触，非线性自旋扭矩纳米振荡器可以模拟人工神经元。拟议的研究活动将为阿拉巴马大学伯明翰分校的学生提供一个独特的机会，以获得计算技能并与国家标准与技术研究所（NIST）的杰出科学家合作。此外，作为该项目的一部分，PI 将开发一门新的高级物理课程，内容涉及人工智能以及如何使用物理设备实现人工智能。该研究基础设施改进 Track-4 EPSCoR 研究员 (RII Track-4) 项目将为阿拉巴马大学伯明翰分校 (UAB) 的助理教授提供奖学金并为博士后研究员提供培训。受大脑启发的神经拟态计算为人工智能应用提供了有吸引力的技术能力。自旋电子设备结合了电子和磁性配置，可以以节能紧凑的方式模拟突触和神经元。磁性随机存取存储器可以充当突触，自旋扭矩纳米振荡器可以充当神经元。量子材料提供了额外的吸引人的特性，包括更有效地控制磁化以及由于自旋、轨道和磁化自由度的耦合而产生的新功能。该提案的主要目标是利用量子材料神经形态自旋电子学的最先进的第一原理方法来理解和建模微观机制。两个主要方向集中于利用量子材料中优越的自旋轨道电子学特性来制造人工突触和神经元。通过与 NIST 的专家合作，该项目将应用第一原理方法来计算自旋轨道耦合量子材料中的能带结构和自旋动力学。这使得人们能够理解自旋轨道扭矩切换和动力学的微观机制，并提供了提高品质因数的途径。项目成果还将包括说明如何利用自旋电子器件的这些非线性非易失性特性来模拟神经形态计算的神经元和突触。拟议的工作将为利用二维材料和拓扑材料等新兴量子材料实现神经形态自旋电子器件提供物理基础。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响进行评估，被认为值得支持审查标准。