Neuromorphic Simulation And Robustness Of Organic Semiconductor Memristor

Modern computing based on the Von Neumann architecture linearly processes digital information and executes programs.The functional separation between memory and processor results in limited computing speed.Researchers are attempting to design computer chips that are massively parallel and efficient like the human brain,and the development of brain like computers requires artificial synapses with real synaptic adaptability.The first condition to simulate the plasticity of biological synapses is that the conductance of artificial synaptic devices can be simulated continuously.Memristor with reconfigurable history-dependent resistive switching behaviors is capable of emulating the synaptic functions of biological synapses,which is standing out as one of the most promising technologies to construct an analogue neural network for neuromorphic computing.The memristor has a simple device structure whith low power consumption,and the size of the device is easy to compress,so it has great potential for multi-dimensional,high-density integration and low-cost manufacturing;Moreover,the historical operation-dependent plasticity feature of memristors has great advantages in realizing learning and memory functions and building intelligent neural networks.Compared with inorganic semiconductor memristor,organic semiconductor memristor has developed rapidly due to its advantages of low processing cost,tunable functionalities,and mechanical flexibility.Although organic semiconductor memristor has made great progress in improving the device performance at present,and has realized a series of complex computing tasks based on the high-performance organic semiconductor memristor,including ultra-low power parallel computing,the establishment of artificial neural networks,etc.,there are few researches on environmental tolerance and device robustness,especially in extreme/harsh working environment.Such as the automobile industry,energy exploration,aerospace and other areas of stability issues still need to be addressed.This dissertation focuses on neuromorphic simulation,working mechanism analysis and robustness of organic semiconductor memristors.First of all,the memristor needs to have smooth gradient memristor curves and adjustable conductance can enable the device to have the capability of neuromorphic simulation.Therefore,it is necessary to find suitable materials to construct memristor which can be applied to neuromorphic simulation.For organic materials,many factors such as intermolecular force,volatility,molecular size and main side chain action should be taken into account.Secondly,there is no unified physical working mechanism at present,so it is necessary to further reveal the resistive mechanism of various memristors and master the regulation law and methods of memristor characteristics.The charge distribution and transport of organic memristors are complicated,so it is necessary to expand the characterization methods,explore the mechanism of organic memristors,and develop a powerful tool to explain the mechanism of organic memristors.Finally,high robustness of memristor is the prerequisite for its application.Organic materials are unstable,and devices cannot work in the harsh environment of high humidity,high temperature and low temperature,leading to a great loss of its application prospects.Explore the robustness of organic memristor in the harsh working conditions,and realize the preparation of high robustness organic memristor.The main research content is as follows:1.The n-type organic semiconductor hexadecafluoro copper phthalocyanine(F₁₆Cu Pc)—a type of metal phthalocyanines with low-lying lowest unoccupied molecular orbital and highest occupied molecularorbital energy level was used to prepare memristor.The mechanism of proton conduction in the device was demonstrated by X-ray photoelectron spectroscopy（XPS）.The basic functions of artificial synapses were simulated using its smooth and gradual memristor curve,including excitatory postsynaptic current（EPSC）,paired-pulse facilitation（PPF）,short-term potentiation（STP）to long-term transition potentiation（LTP）.2.As the working mechanism of F₁₆Cu Pc-based memristor is proton conduction mechanism in high humidity environment,the device performance of F₁₆Cu Pc memristor under different humidity conditions was further studied,and the effect of ambient humidity on the memristor was systematically tracked.The devices without encapsulation can storage in an ambient（90%relative humidity）over 45 days,in air environment（60%relative humidity）for one year,and immersed in water for 96 h,exhibit no obvious change in resistive memristive behavior.These results demonstrate the feasibility of organic semiconductor F₁₆Cu Pc-based memristors for artificial neural networks,especially in future extreme humidity environments.3.Organic conjugated polymer materials in organic materials are suitable for low-cost solution processing because of their good solubility.The memristor device was fabricated by using organic polymer poly[2-methoxy-5-（3’,7’-dimethyloctyloxy）-1,4-phenylenevinylene]（MDMO-PPV）.It is found that the continuous incremental changes in conductivity of devices provide the basic ability for synaptic simulation and neural morphology calculation due to their nonlinear transmission characteristics.Using this simple memristor,a series of synaptic behaviors,including excitatory postsynaptic current（EPSC）,paired-pulse facilitation（PPF）,spike-rate-dependent plasticity（SRDP）,spike-time-dependent plasticity（STDP）,short-term potentiation（STP）to long-term potentiation（LTP）,“learning–forgetting–relearning”process have been successfully simulated.Further,utilizing X-ray photoelectron spectroscopy（XPS）to explain the device working mechanism,and also clearly demonstrated the dynamic process of ion migration in the device by time-of-flight secondary-ion mass spectrometry（To F-SIMS）.To F-SIMS also provides clear and intuitive evidence for explaining the mechanism of memristors.4.MDMO-PPV have planar structures due to the presence of delocalizedπsystems,which facilitate stacking interactions and formation of crystal domains.The strongπ-πinteraction between the polymer segments effectively enhances the energy transfer and makes the molecule more stable.Based on this,the device performance of MDMO-PPV based memristor at different temperatures was explored.The electrical performance was tracked under test temperature ranging from 77 K to 573K.The results show a robust memristive response and MDMO-PPV based memristor has good robustness in both low and high temperature environments.The robust memristive response achieved at extreme temperatures may open opportunities for the next generation of potential applications of information technology and pave the way for potential applications in extreme or harsh environments.5.To explore the device performance of F₁₆Cu Pc/MDMO-PPV heterostructure memristor in high humidity and high temperature environment.The transport property in F₁₆Cu Pc/MDMO-PPV heterostructure memristor should be a joint effect of several conduction mechanisms,such as tunneling and SCLC mechanism.Tracking its memristive response,the devices without encapsulation can storage in an ambient（70%relative humidity）up to 3 weeks.Afterwards,the device was kept at70%RH to explore the memristor characteristics at different temperatures.The device still has good memristor performance in the temperature range of 50-300℃.F₁₆Cu Pc/MDMO-PPV heterostructure memristor have successfully simulated a series of basic synaptic behaviors.In summary,the robust memristor based on organic heterostructure provides a new solution for overcoming humidity and temperature environments in the future.