Study On Multilevel Storage And Performance Of Silk Fibroin Based Memristors

The memristor with multilevel storage can improve the data storage density without changing the size of the storage cell.This is not only an effective solution of adapting the future high-density storage requirements and improving the efficiency of the neuromorphological computing system,but also one of important future direction of memristor.Silk fibroin（SF）,which is one bio-material with good biocompatibility and adjustable structure,shows good comprehensive memristive performance.However,there are still some challenges to prepare SF-based memristor with high-performance,low-power,and multilevel storage.Therefore,in this paper,SF is selected as the main material and compounds with inorganic materials,such as graphene（Gr）,graphene quantum dots（GQDs）or graphite oxide（GO）,to fabricate bio-memristors with multilevel storage through regulating the interface interaction The memristive performance and resistive transition mechanism are systematically studied,and the bionic neural synaptic function is verified.The specific research content is as follows:（1）SF-Gr-based biological memristor with high stability:In order to improve the current switching ratio,repeatability and stability of SF-based memristor,ITO/Gr/SF/Al and ITO/Gr/SF/Gr/Al composite memristor are prepared by layer-b layer（LBL）technology.These memristors can realize bipolar resistance transformation with large storage window.Among them,the V_SETand V_RESETof ITO/Gr/SF/Al memristor are-1.1 V and 2.4 V,respectively;and the retention time of each resistance state exceeds 1200 s.The switching window of ITO/Gr/SF/Gr/Al memristor increased to 10³,V_SETand V_RESETare-1 V and 2 V,respectively.The retention time of each resistance state can reach 2000 s.The resistive transition mechanism of the above memristors is space charge limiting current（SCLC）mechanism.（2）SF-GQDs-based multilevel storage memristor:The ITO/GQDs/SF/GQDs/Al memristor is constructed by combining GQDs and SF.This memristor realizes multilevel storage in a single cycle,and its stability,data retention ability and other performances are further improved.The memristor transforms from a low resistance state to an intermediate resistance state at 1.52 V(V_RESET-1).In addition,the transformation from an intermediate resistance state to a high resistance state occurs at 3.84 V(V_RESET-2).The memristor can be stably cycled for more than 50 times.The three resistance states maintain for more than 2000 s.The multilevel resistive transition is attributed to the interface interaction between SF and GQDs,and controlled by the Schottky emission mechanism.Furthermore,the device can simulate synaptic functions,such as short-term plasticity（STP）and paired pulse facilitation（PPF）,and exhibits the potential to construct bionic synapses.（3）SF-GO based memristor integrated with double resistance states and triple resistance states:In order to expand the application of SF memristor in different scenarios,an ITO/GO/SF/GO/Al composite memristor integrated with reversible double resistance states and triple resistance states in a single cycle is constructed by regulating the interface structure and interaction between SF and GO.Furthermore,these two transition modes can be transformed reversibly.The V_SET,V_RESET-1and V_RESET-2of triple resistance states are-3.9 V,2.4 V,and 4.1 V,respectively,and every resistance state can remain for 10⁴s.In addition,the device can be stored stably for 30 days without performance loss.The resistive transition mechanism of this memristor is dominated by SCLC and Pool Frenkel emission mechanism.The ITO/GO/SF/GO/Al composite memristor exhibits excellent simulation performance of synaptic plasticity,such as STP,PPF and long-term plasticity（LTP）.Based on the above abilities,the ternary quantization artificial neural network has a recognition accuracy of 92.3%for handwritten numerals,and can be applied in image compression and reconstruction.It has good application potential in the field of neural morphological computing.In conclusion,this paper not only improves the memristor performance of SF-based memristor,but also realizes multilevel storage in a single cycle and reversible switching of different memristor transition modes,by regulating the interface structure between SF and graphene based materials.This provides new idea for building other functional bio-memristor.At the same time,this paper verifies the application of multilevel storage SF memristor in bionic neural synapse,handwritten digit recognition and image reconstruction,which lays the foundation for data storage,artificial synapse construction and neural morphology calculation based on memristor with multilevel storage.