Thermal Compact Modeling and Resistive Switching Analysis in Titanium Oxide-Based Memristors.

Resistive switching devices based on the Au/Ti/TiO2/Au stack were developed. In addition to standard electrical characterization by means of I-V curves, scanning thermal microscopy was employed to localize the hot spots on the top device surface (linked to conductive nanofilaments, CNFs) and perform in-operando tracking of temperature in such spots. In this way, electrical and thermal responses can be simultaneously recorded and related to each other. In a complementary way, a model for device simulation (based on COMSOL Multiphysics) was implemented in order to link the measured temperature to simulated device temperature maps. The data obtained were employed to calculate the thermal resistance to be used in compact models, such as the Stanford model, for circuit simulation. The thermal resistance extraction technique presented in this work is based on electrical and thermal measurements instead of being indirectly supported by a single fitting of the electrical response (using just I-V curves), as usual. Besides, the set and reset voltages were calculated from the complete I-V curve resistive switching series through different automatic numerical methods to assess the device variability. The series resistance was also obtained from experimental measurements, whose value is also incorporated into a compact model enhanced version.

TiN/Ti/HfO2/TiN memristive devices for neuromorphic computing: from synaptic plasticity to stochastic resonance.

We characterize TiN/Ti/HfO2/TiN memristive devices for neuromorphic computing. We analyze different features that allow the devices to mimic biological synapses and present the models to reproduce analytically some of the data measured. In particular, we have measured the spike timing dependent plasticity behavior in our devices and later on we have modeled it. The spike timing dependent plasticity model was implemented as the learning rule of a spiking neural network that was trained to recognize the MNIST dataset. Variability is implemented and its influence on the network recognition accuracy is considered accounting for the number of neurons in the network and the number of training epochs. Finally, stochastic resonance is studied as another synaptic feature. It is shown that this effect is important and greatly depends on the noise statistical characteristics.