||Since long time ago, the nature being has been the source of human’s senses of invention principles productions, in which bionic technology has been well developed in recent years. Lithium aluminum oxide (LiAlO) thin film Resistance Random Access memory (RRAM) reveals an excellent multi-bites memory. It is considered for application of artificial neural network and analog storage to emulating memory rules in the brain.|
In this thesis, two different RRAMs, investigated are that in one the insulation layer is lithium aluminum oxide, and the other one is aluminum oxide, fabricated by Multi-target magnetron sputtering. The RRAM upper and lower electrodes are platinum and titanium nitride, respectively. In operating process of this LiAlO RRAM the reset process reveals two reset stages. Applying the ionic diffusion and the titanium nitride attraction an ion model of lithium-ion, is discussed about the phenomenon of the two-phase wide resistance value. And utilizing oxygen ions model the two-stage redox Reset resistance of lithium wires are explained.
The experiment using sputtering platinum on the titanium nitride electrodes can prevent lithium ions diffusing into the TiN electrode, that HRS distribution is reduced to one Order. However, the distribution of LRS on titanium nitride becomes wider than that one without Pt layer on titanium nitride electrode. Since the Pt electrode blocks oxygen ions and lithium ions diffusing into the titanium nitride electrode, causes preventing of the instability of LRS.
Both DC and Pulse Voltage have been applied to perform lithium aluminum oxide RRAM, and reveal the HRS resistance changing continuously, but alumina RRAM does not. Applied with Pulse voltage on lithium aluminum oxide RRAM can induce a bionic brain behavior of Spike-Timing-Dependent Plasticity (STDP), which is one of the most important characteristics of biological brain and memories. the LRS of LiAlO RRAM changes gradually after applying a continuous DC voltage. From the fitting of their I-V curves the conductive mechanism is identified. From its resistance of Ohmic conduction region changes after numbers of I-V scans, a Multi-filament model is proposed to explain the Multi-Reset electrical mechanism.