Abstract:Online training of deep neural networks (DNN) can be significantly accelerated by performing in-situ vector matrix multiplication in a crossbar array of analog memories. However, training accuracies often suffer due to device non-idealities such as nonlinearity, asymmetry, limited bit precision and dynamic weight update range within constrained power budget. Here, we report a three-terminal Ferroelectric-Field-Effect-Transistor based on low thermal budget processes that can work efficiently as an analog synaptic transistor. Ferroelectric polymer P(VDF-TrFE) as the gate insulator and 2D semiconductor MoS2 as the n-type semiconducting channel material makes them suitable for flexible and wearable substrate integration. The analog conductance of the FeFETs can be precisely manipulated by employing a ferroelectric-dielectric layer as the gate stack. The ferroelectric-only devices show excellent performance as digital non-volatile memory operating at +-5V while the hybrid ferroelectric-dielectric devices show quasi-continuous resistive switching resulting from gradual ferroelectric domain rotation, important for their multibit operation. Analog conductance states of the hybrid devices allow linearity and symmetry of weight updates and produce a dynamic conductance range of 104 with >16 reproducible conducting states. Network training experiments of these FeFETs show >96% classification accuracy with MNIST handwritten datasets highlighting their potential for implementation in scaled DNN architectures.

A computationally efficient compact model for ferroelectric FETs for the simulation of online training of neural networks

Compact Modelling of Ferroelectric Tunnel Memristor and Its Use for Neuromorphic Simulation

Ferroelectric FDSOI FET modeling for memory and logic appliations

A Compact Model of Non-Volatile Ferroelectric Tunnel Fet with Ambipolarity for In-Memory-computing Based Edge AI

A Physics-Based Compact Model of Ferroelectric Tunnel Junction for Memory and Logic Design

The Study of Hysteretic Characteristics of FeFETs for Neuromorphic Computing Using a Closed-Form Analytical Model

TCAD modeling of neuromorphic systems based on ferroelectric tunnel junctions

Intrinsic synaptic plasticity of ferroelectric field effect transistors for online learning

Compact Modelling for Co/Bto/Lsmo Ferroelectric Tunnel Junction

An efficient device model for ferroelectric thin-film transistors

A Computationally Efficient Compact Model for Ferroelectric FinFETs Switching with Asymmetric Non-Periodic Input Signals

Back-end and Flexible Substrate Compatible Analog Ferroelectric Field Effect Transistors for Accurate Online Training in Deep Neural Network Accelerators

Device Modeling and Simulation of Ferroelectric Tunnel Junction for Computing-in-Memory Application

Optimal Weight Models for Ferroelectric Synapses Toward Neuromorphic Computing

A Comprehensive Model for Ferroelectric FET Capturing the Key Behaviors: Scalability, Variation, Stochasticity, and Accumulation

The Critical Role of Charge Balance on the Memory Characteristics of Ferroelectric Field-Effect Transistors

Design-Technology Co-Optimizations for Symmetric Linear Synapse Behaviors in Ferroelectric FET Based Neuromorphic Computing

Temperature- and variability-aware compact modeling of ferroelectric FDSOI FET for memory and emerging applications

A van der Waals Synaptic Transistor Based on Ferroelectric Hf<sub>0.5</sub>Zr<sub>0.5</sub>O<sub>2</sub> and 2D Tungsten Disulfide

Device Modeling and Application Simulation of Ferroelectric-FETS with Dynamic Multi-Domain Behavior

Modeling the Coupled Charge Trapping Dynamics in FeFETs for Reliability Characterizations