E2CDA: Type II: 2D Electrostrictive FETs for Ultra-Low Power Circuits and Architectures

世纪有望在新的电子系统上蓬勃发展，这些系统消耗的电力如此节省，以至于即使是环境能量收集也可以满足其所有需求。尽管传统的硅基CMOS技术在过去四十年中不断革新我们的日常生活，但由于MOSFET的基本物理特性的根本限制，它似乎无法实现这一目标。因此，该项目的目标是在各个层面上进行创新：从材料工程到器件物理，从电路设计到建筑实施，以实现同时提供能源效率和高性能的电子系统。这些系统可以用于植入式医疗设备、柔性电子设备、自供电远程传感器以及物联网中的各种设备等。然而，科学和技术的成功只有传播到社会的各个部门才能有效地发挥杠杆作用。 因此，教育推广将是该项目的一个组成部分，将产生的知识传播给本科生、研究生和代表性不足的群体，以便产生真实的和持久的影响。研究所将举办讲习班和研讨会，编写课程材料，并与社区学院和四年制大学的教师建立教育合作关系，以传播研究成果。从技术角度来看，该项目旨在实验性地实现一种基于二维（2D）材料动态能带结构工程的新型、陡斜率、可扩展性强且节能的晶体管，并利用所提出器件的独特特性实现超低功耗电路和架构。二维材料是具有强的层内共价相互作用和弱的货车德瓦尔斯层间相互作用的层状化合物。 这种晶体管，称为二维电致伸缩场效应晶体管（2D-EFET）的工作原理的电压感应应变转换。它使用电致伸缩材料作为栅极氧化物，其响应于施加的栅极偏置而膨胀，从而转换2D沟道材料上的平面外应力。这种应力减小了半导体2D材料的连续层之间的层间距离，并动态地将其带隙减小到零，即将其转换为半金属。因此，该器件在关断状态下以大带隙工作，而在导通状态下以小带隙或零带隙工作。由于这种换能机制，发生内部电压放大，导致陡峭的开关。陡峭的开关实现了积极的电压缩放，这与适当的器件-电路协同设计一起，导致超低功耗和鲁棒的电路操作。这些特性对于高度并行的面向吞吐量的架构和固有的低电压设计（例如由能量收集供电的物联网节点）都非常有吸引力，从而满足了计算频谱两端的需求。采用2D-EFFECT作为电路和架构设计师的标准元素，预期会在能效方面获得哪些收益？的工具包被发现是足够大的，为某些领域，是变革。

The 21st century promises to thrive on novel electronic systems that consume power so frugally that even ambient energy harvesting can supply all of their needs. Conventional silicon-based CMOS technology, despite having repeatedly revolutionized our everyday life for the last four decades, appears incapable of delivering on this objective due to fundamental limitations in the underlying physics of MOSFETs. The goal of this project is, therefore, to execute innovation at all levels: from materials engineering to device physics to circuit design to architectural implementations in order to achieve electronic systems that simultaneously offer energy efficiency and high performance. These systems can then be used for implantable medical devices, flexible electronic appliances, self-powered remote sensors and also for various devices involved in the Internet-of-Things and so on. Scientific and technological successes, however, will not be effectively leveraged unless disseminated to all sections of society. Educational out-reach will, therefore, be an integral part of this project to transmit the knowledge generated to undergraduates, graduates and under-represented groups in order to have real and lasting impact. The PIs will organize workshops and seminars, develop course materials and establish educational collaborations with instructors from community colleges and 4-year universities to disseminate the research results. From technical stand point, this project aims to experimentally realize a novel, steep slope, aggressively scalable and energy efficient transistor based on dynamic band structure engineering in two-dimensional (2D) materials, and to exploit the unique characteristics of the proposed device to enable ultra-low power circuits and architectures. The 2D materials are layered compounds with strong intra-layer covalent interaction and weak van der Waals inter-layer interaction. This transistor, referred to as the two-dimensional electrostrictive field effect transistor (2D-EFET) works on the principle of voltage induced strain transduction. It uses an electrostrictive material as a gate oxide that expands in response to an applied gate bias and thereby, transduces an out-of-plane stress on the 2D channel material. This stress reduces the inter-layer distance between the consecutive layers of the semiconducting 2D material and dynamically reduces its bandgap to zero i.e. converts it into a semi-metal. Thus the device operates with a large bandgap in the OFF state and a small or zero bandgap in the ON state. As a consequence of this transduction mechanism, internal voltage amplification takes place which results in steep switching. The steep switching enables aggressive voltage scaling, which, with proper device-circuit co-design, leads to ultra-low power and robust circuit operation. These properties are very attractive for both highly parallel throughput-oriented architectures and inherently low-voltage designs, such as IoT nodes powered by energy harvesting, thereby addressing needs across both ends of the computing spectrum. The expected gain in energy efficiency from adopting 2D-EFETs as a standard element of the circuit and architecture designer?s toolkit is found to be large enough, for some domains, to be transformative.