NU 2017-210

Inventors

Mark Hersam*

Vinod Sangwan

Hong-Sub Lee

Short Description

Multi-terminal memtransistors for artificial neuromorphic circuits

Background

The field of neuromorphic computing is emerging as a leading area for next-generation integrated circuit technology. Historically, progress in this field has been enabled by aggressive scaling in silicon transistor technology and corresponding algorithm optimization. The use of memristors, two-terminal passive circuit elements developed for use in non-volatile resistive random-access memory, has been considered for neuromorphic computing as it mimics the function  of biological neurons better than conventional silicon transistors due to physically similar ion-mediated switching.  They possess higher endurance and faster read/write times than flash memory and can provide multi-bit data storage. Two-terminal memristors have also demonstrated capacity for basic neural functions; however, synapses in the human brain outnumber neurons by more than a thousand-fold.  The development of multi-terminal memristors is needed to perform complex functions such as heterosynaptic plasticity. Previous attempts to move beyond two-terminal memristors did not achieve memristive switching in the transistor. 

Abstract

Northwestern researchers have developed a multi-terminal hybrid memristor and transistor (that is, a memtransistor) using polycrystalline monolayer molybdenum disulfide (MoS2) in a scalable fabrication process.  The switching is achieved by a defect migration in monolayer MoS2 near the electrical contacts, as verified by physical modeling. The two-dimensional MoS2 memtransistors show gate tunability in individual resistance states by four orders of magnitude, as well as large switching ratios, high cycling endurance and long-term retention of states. In addition to conventional neural learning behaviour of long-term potentiation/depression, six-terminal MoS2 memtransistors have gate-tunable heterosynaptic functionality, which is not achievable using two-terminal memristors. They have captured and demonstrated technologically relevant device metrics, including endurance, retention, and large-area scaling.  Furthermore, the seamless integration of a memristor and transistor into one multi-terminal device enables complex neuromorphic learning and the study of the physics of defect kinetics in two-dimensional materials.  This device concept may serve as a foundational circuit element of brain-inspired computing.

Applications

• Artificial neuromorphic circuits with more than two synapses
• Heterosynaptic plasticity
• Gate-tunable crossbar memristor arrays
• Parallel computing via multi-terminals

Advantages

• Multi-terminal memtransistor
• More advanced neuromorphic circuits mimicking real neurons better       

Publication

Sangwan VK, Lee HS, Bergeron H, Balla I, Beck ME, Chen KS & Hersam M (2018) Multi-terminal memtransistors from polycrystalline monolayer molybdenum disulfide. Nature, 554: 500-504.

IP STATUS

US patent and PCT applications have been filed.

Schematic of the MoS2 memtransistor