The Investigation On Filtering Function For Neuromorphic Computing Based On Porphyrin Memristors

Since the birth of computer,the performance potential of Von Neumann Architecture has been almost exhausted after more than half a century of development.The level of intelligence,energy efficiency and other aspects have been difficult to meet the needs of future applications with the limitation of computer performance.Human brain,as one of the most complex systems in the universe,can achieve far more advanced intelligence than computer at about 20W power consumption and 10Hz low frequency.It is a model of high performance and low power computing hardware in nature.In addition,some of the neurons in the human brain can be turned off when they are not in use,moreover the human brain can selectively filter information,so that the overall energy consumption remains ultra-low when processing various large data.This process usually controls synaptic plasticity and removes irrelevant information by inhibiting inhibitory neurons which receive signals of different frequencies of neurons.In order to deal with characteristic of tremendous data,complex structure and large background noise,it is necessary to have filtering function at the unit level in the hardware implementation of brain-like system.Based on the previous studies of ITO/Zn TPP/Al₂O_3-x/Al organic memristors,the potentiation and inhibition phenomena of devices based on Matthew-like effect,the electrical performance regulation of porphyrin heterojunction devices and the synaptic signal filtering function are deeply studied in this paper.It will provide technical reserve for the theoretical development and practical application of organic memristors.?1?The potentiation and inhibition characteristics of ITO/Zn TPP/Al₂O_3-x/Al diode memristor was explored.It was found that the conductance of the device would increase under large scanning voltage or positive pulse?10 V?,while the current of the device would gradually decrease under negative?-10 V?and small positive pulse?4 V?,thus realizing the simulation of potentiation and inhibition in synapses.In particular,the device showed obvious voltage threshold dependence in the positive voltage test which was consistent with the"Matthew effect"in sociology,where the stronger the strong,the weaker the weak.It has greatly expanded the application field of organic memristor in the simulation of synaptic potentiation and inhibition.?2?A metalloporphyrin heterojunction memristor ITO/Co TPP/Zn TPP/Al₂O_3-x/Al was designed,and the effect of heterojunction thickness on electrical properties was investigated by fixing one material and adjusting the thickness of the other.After comparing the electrical performance under positive and negative scanning voltage,we found that adjusting the thickness of the heterojunction can control the direction of rectification of the device and rectification ratio.This provides hardware support for simplifying circuit structure and enriches the diversity of biological synaptic energy simulation.?3?Based on the previous studies on the inhibitory function of synapse enhancement by Zn TPP memristor and the dynamic regulation of critical voltage and rectification ratio in heterojunction devices,we combined the two methods to apply porphyrin heterojunction devices to realize the simulation of signal filtering function.Through changing different voltage amplitude and sampling time on the device especially exploring the input pulse signal voltage and pulse width,we realized on the basis of the Matthew effect of porphyrin memristor in the combination of voltage and pulse width signal filtering model.This provides more possibilities for practical circuit simplification and filtering simulation with organic memristors.