Electrothermal Analysis of Memristors and Thermistors

The high demands of reducing power consumption for electronic applications and decreasing the current and voltage range have increased the technological development of the non-volatile Resistive Random Access Memory (RRAM) based on Memristive devices. In this thesis, a full 3D electrothermal model is adopted to study the thermal and signal integrity of a 1Diode-1Resistor RRAM crossbar array. In fact, the 3D integration suffers from two main issues which are the voltage drop along the interconnects and thermal crosstalk between the memory cells. Several solutions are investigated based either on new biasing schemes, new materials (Nickel (Ni), Copper (Cu), and Carbon nanotubes (CNT)), and new architectures integrations (reverse architecture (1D1R-1R1D) and complementary resistive switching (CRS)). Based on these proposed structures with the choice of the new materials and the bias management, the electrothermal performances show the benefits of solving the issues related to the large-scale monolithic 3D RRAM integration. All the results are compared with the reference architecture (1D1R) and also the physical model is validated against other models in the literature. Indeed, the second approach of the thesis is to study the electrothermal behavior of macroscopic graphene strips (Graphene Nano-Platelets (GNPs)) as a heater element which is the Thermistor. Where an equivalent electrical resistivity is studied, by means of models and experimental characterization based on three different materials with different percentages of graphene nanoplatelets. To this end, the electrical resistivity shows a negative temperature coefficient in a wide temperature range [-40, +60] °C, which allowed these materials to be used for thermistor and temperature sensor applications.