The SPICE Model For Single Domain And Multi-domain Ferroelectric Transistors And Applications

Ferroelectric field effect transistors have attracted wide attention in low power consumption,low operating voltage,storage,and neural network circuits due to the negative capacitance,multi-conductance adjustable and nonvolatile.In this paper,the SPICE model of single-domain and multi-domain ferroelectric transistors were presented based on physics theory.And we analyzed the electrical properties and circuit applications of ferroelectric transistors.The main research contents were shown as follows:Firstly,SPICE model of the transistor was established for the negative capacitance based on Landau-Khalatnikov theory and the ferroelectric film was considered to be a single domain.The accuracy of the model was compared after it was established.The result showed that our model had higher precision,wider bias conditions,and no initialization conditions.The electrical performance of ferroelectric transistors was simulated by the model.The results showed that the gate amplification of ferroelectric transistors increased with the thickness of ferroelectric thin films.When the thickness of ferroelectric thin films reached a certain thickness,hysteresis would occur.This phenomenon,as an unstable state of energy,should be avoided in the design of digital circuits.The simulation of subthreshold swings showed that ferroelectric transistors can implement Sub-60mV/decade.The ferroelectric transistor had a larger saturated channel current through transfer characteristic simulation.These excellent properties provided a solution to reduce the power consumption of circuit.Based on the excellent electrical properties of ferroelectric transistors,we simulated the functional circuit and obtained the following results: Considering the thickness of the ferroelectric thin film which was from 10 nm to 30 nm.In the simulation of the ring oscillator,we found that the lower damping coefficient can achieve a lower propagate delay.Thus the damping coefficient was set to 0.01 ? ?m.At a low operating voltages,the propagation delay of ferroelectric transistors was lower than that of MOSFETs,the DC power consumption and the dynamic power consumption of ferroelectric transistors was reduced compared to MOSFETs which could achieve low operating voltages and low power consumption.It was found that the SRAM based on ferroelectric transistors had lower read-write delay and larger noise margin than the 6T-SRAM based on MOSFETs by the simulation of 6T-SRAM,which was due to the gate amplification of ferroelectric transistors.It had a higher saturation current for fast charge and discharge,and the DC inverting power consumption of the six ferroelectric transistors was lower than that of the MOSFETs.The hysteresis behavior of ferroelectric transistors was used to form a Schmitt trigger that operates at an operating voltage of 0.2V.Considering the more practical situation,the Preisach theory was used to establish the multi-domain model of ferroelectric transistors.The tanh function was used to express the hysteresis behavior of ferroelectrics.The method of modifying the unsaturated hysteresis loop was proposed.Multi-domain ferroelectrics were established by using Verilog-A language.The SPICE model of the capacitor was combined with MOSFETs to create a ferroelectric transistor model.The synaptic circuit based on ferroelectric transistor was proposed by using the tunable electrical properties of the multiconductance obtained by simulation,and it was applied to the simulation of the Temporal rule neural network circuit.The results showed that the IRIS data set can be correctly classified after being trained many iterations.