Study On The Synaptic Bionics Of Memristor Based In Ferroelectric Thin Films

As a new type of semiconductor device,the memristor is a functional device that changes its resistance by controlling current changes to achieve data storage,and it has attracted much attention because of its ability to achieve multi-level storage and learning of brain-like synapses.Although several emerging storage technology have been proposed,their poor performance,poor high temperature stability and easy oxidation under general environmental conditions lead to dispersive and regulated switching voltages,so they still face many difficulties.The ferroelectric thin film memristor studied in this paper achieves controllability by adjusting the change of the barrier through the directional polarization of the ferroelectric domain and then adjusting its resistance.Firstly,we explored the polymorphic storage and neuromodulation of the Si-based traditional ferroelectric material Ba TiO₃(BTO)thin-film memristor and realized algebraic calculation;secondly,we studied the Si-based new ferroelectric material Hf_0.5Zr_0.5O₂(HZO)thin-film memristor,and realized the synaptic interaction Plasticity adjustment;Finally,the flexible memristor based on Bi FeO₃(BFO)ferroelectric materials is explored to solve the pressing bottleneck problem existing in the current flexible memory based on organic matter and memory based on simple metal oxide.The main contents are as follows:Firstly,the polymorphic storage and neuromodulation of silicon-based BTO ferroelectric thin-film memristors and the realization of algebraic calculations.In this work,we have studied the memristor of BTO/La_0.67Sr_0.33MnO₃(LSMO)/Sr TiO₃(STO)/Si heterojunction.The transmission electron microscope(TEM)test showed the excellent microstructure of the memristor,and the piezoelectric response force microscope(PFM)tested the polarization characteristics of the device.Through electrical testing,the memristor can realize the function of multi-state storage.Not only that,the device can also achieve controllability under different parameter pulses.Further,the memristor realized the four arithmetic operations,showing excellent algebraic calculation ability.In terms of cranial nerve simulation,the memristor can also realize the spike-timing dependent plasticity(STDP)function of neurosynaptic learning.The results prove that silicon-based ferroelectric memristors are very promising candidates for future non-volatile memories.Secondly,the ferroelectric memristor based on HZO film has both memristive and neuromorphic functions.In this work,we fabricated a memristor based on HZO on a Si substrate.HZO has the strongest ferroelectricity in the Hf_xZr_1-xO_yseries of films.Through the electrical performance test of the device,a typical current-voltage(I-V)curve of the memristor can be obtained,and the high-low resistance ratio R_OFF/R_ONof the device can reach 10⁴.In addition,we also studied the influence of pulse sequence modulation of different parameters on the conductance of the device,and studied the synaptic learning behavior of the device.The above research results have laid a certain foundation for the development of ferroelectric memristors with synaptic-like behavior.Thirdly,a flexible artificial synapse based on single crystal BFO film.In this article,we research a flexible artificial synapse based on a ferroelectric tunnel junction(FTJ)that uses an independent single crystal BFO film as a functional layer.Flexible artificial synapses based on FTJ can combine the excellent properties of inorganic perovskite films grown on rigid substrates at high temperatures with those of flexible films.The project studied inorganic perovskite ferroelectric thin film memristors on flexible substrates as key functional elements of neuromorphic computing systems,and explored their memristive properties and brain-like synaptic learning properties.In addition,the free-standing tunnel junction exhibits excellent plasticity learning characteristics in both flat and curved states,so it can potentially be used as an artificial flexible electronic synapse for future neuromorphic calculations.