Modification On Resistive Switching Properties Of The HfO_x Based RRAM By Embedding Metal Nanoparticles

Resistive random access memory(RRAM)is one of the most promising candidates for next generation non-volatile memory due to its simple structure,low power consumption,high switching speed and good down-scaling capability.Among them,HfO_x-based RRAM has been favored by academia and industry due to its low power consumption,good data retention,and fast read/write speed.In addition,with the rapid expansion of information data,traditional computer architecture faces many challenges.Artificial neural network will be the future direction of computer development.Emulating Synaptic behaviors is considered to be the first step to realize neural network.Therefore,in this paper,the effect of Cu nanoparticles on the resistive switching properties of the HfO_x-based RRAM was investigated,and devices embedded with nanoparticles were also used as synaptic electronics to simulate biological synaptic behavior.The detailed research contents and results are as follows.The RRAM with Ti/HfO_x/ZnO/ITO structure was fabricated by micro-fabrication process,the device exhibits bipolar resistive switching features with self-compliance and low power consumption.It was further found that the device is dominated by the Schottky emission mechanism in the high-resistance state and the Poole-Franker emission mechanism for the low-resistance state.And the operation voltages among devices and the resistance states of a single device are not uniform.Furthermore,2-bit storage was realized,but the intermdiate resistance states exhibit poor data retentionthat causes memory failure.To optimize device performance,we embedded copper nanoparticles into HfO_xlayer.The embedded Cu nanoparticles enhance local electric field,which not only reduce the operating voltages with improved uniformity but also increase resistance ratio between the high resistance state and the low resistance state.The embedded Cu nanoparticles also stabilize resistances of states to ensure excellent endurance characteristics.Further studies found that the conduction mechanism of the low resistance state was no longer dominated by Poole-Frenkel emission but was dominated by Ohmic conductive filaments assisted with oxygen vacancies.As a result,the resistance of the low-resistance state was futher decreased to increase the on/off ratio between the high restance state and the low resistance state.In addition,2-bit storage capable of good reproductivity and data retentionwas implemented on the device embedded with Cu nanoparticles.In additional,the devices embedded with Cu nanoparticles have excellent performance and are expected to be used as synaptic electronics.For this reason,a variety of synaptic learning behaviors were realized on a single device embedded with Cu nanoparticles.Short-term memory,i.e.paired-pulse facilitation(PPF)and paired-pulse depression(PPD),was simulated by adjusting the pulse interval time.The transition from short-term memory(LTM)to long-term memory(LTM)was also realized by increasing the pulse amplitudeand width but reducing the pulse interval time.Besides,spiking-rate-dependent plasticity(SRDP)is realized by adjusting the pulse frequencywith 2 Hz as threshold frequency.Moreover,spiking-timing-dependent plasticity is realized by adjusting the interval time between pulses trains and the synaptic weight adjustment range is close to the biological synapses.Finally,long-term potentiation and long-term depression were realized by adjusting the pulse width,and it also achieved extremely low power consumption of repeated potentiation and depression.The device can be applied to large-scale neural networks.