Study On Construction And Synaptic Functions Of Chitosan Electrolyte-Based Biomimetic Memristor

With the development of artificial intelligence,the demand for low-power and high-speed information processing is increasing.The shortcomings of the traditional von Neumann computing architecture are becoming increasingly prominent,which cannot meet the existing demand.Inspired by the brain,the artificial neural network based on memristors has low-power and high-speed processing.However,traditional rigid materials have poor biocompatibility and cannot fit with skin tissue,which limits the development of memristors in implantable bionic systems.Biomaterial-based memristors have the advantage of flexible biocompatibility,which provides an effective scheme for the realization of high-fidelity synaptic bionics.The practical application of memristors still has problems such as poor reliability,high energy consumption,and single synaptic function.Therefore,there is an urgent need to develop a high-performance biomaterial-based biomimetic memristor to realize efficient information processing.For this reason,this study focuses on the reliability,low energy consumption,and multiple synaptic functions of biomimetic memristors.The interpenetrating network chitosan electrolyte was synthesized,and the construction strategy of highly reliable digital memristors was established.Chitosan electrolyte with Li ion conductivity was synthesized,and a high-speed and low-power bionic memristor was constructed through reasonable device structure design.The larger number of functions endowed to the bionic memristor by introducing interface engineering.The highly stable CS-rGO electrolytes were prepared by the in situ hydrothermal reduction method.The obtained CS-rGO electrolyte has an interpenetrating network structure consisting of hydrogen bonds and covalent bonds.The effect of the doping content of rGO on the mechanical properties of CS-rGO electrolyte was studied.Compared to the CS matrix,the tensile strength of the interpenetrating network CS-rGO increased by 1.84 times,the ion conductivity increased by 2 orders of magnitude at room temperature,and the cell viability was close to 100%.The LiTFSI was added to the CS-rGO network to prepare the CSrGO-LiTFSI with high ionic conductivity.The ionic conductivity of the obtained electrolyte was higher than 10-5 S·cm-1,which realizes the rapid migration of Li ions.Employing CS-rGO electrolyte as an active layer,a highly reliable digital memristor was constructed.The interpenetrating network structure of CS-rGO improves the stability of metal ion transport channel and solves the problem of poor reliability of memristors.The memristor shows stable bipolar resistive switching up to 100 consecutive cycles,reproducible multistate storage with six different memory states,and a low programming power of 9.4μW.Besides,the fabricated digital memristor exhibits deformation stability,high biocompatibility with HEK293 cells and skin adhesion,which promotes its application in the field of wearable information storage.CS-rGO-LiTFSI electrolyte/MoS2 heterojunctions have been employed to construct the energy-efficient bionic memristor.The impulse response of the bionic memristor is reduced to ten millivolts with competitive femtojoule-level consumption,exceeding the biological level by orders of magnitude.Under an electric field,the ion intercalation-induced structural evolution of MoS2 regulates the conductance switching of the memristor.Due to the good controllability and reversibility of the intercalation process,the memristor exhibits fast switching of 500 ns and electrical stability.The memristor is capable of parallelly processing signals transmitted from multiple pre-neurons,thus realizing logic operations and spatiotemporal rules.The Nafion cation exchange membrane was introduced as the interface layer,and the bionic memristor with an electrolyte/Nafion/MoS2 structure was constructed.Interface engineering is used to modulate the ion transport behavior under the electric field,and the construction method of the memristor-based filter is established to realize multiple synaptic functions.The 7 × 7 memristor array was developed to simulate the human visual recognition system.The unipolar SVDP integrates the synaptic weight enhancement and suppression behaviors,resulting in a contrast between the target signal and non-path pixels as high as 7.0.The unipolar SVDP behavior plays a key role in the signal filtering process.The development of synaptic functions provides a new idea for the application of memristors to efficient information processing.