No abstract

Memristors are emerging as unique electrical devices with potential applications in memory, reconfigurable logic and biologically inspired computing. Due to the novelty of these devices, the complete details of their switching mechanism is not yet well established. In this work, the switching mechanism of our solution-processed titanium dioxide-based memristor is investigated by studying how variations in the device area and film thickness affect electrical behavior and correlating these behavioral changes to proposed switching mechanisms. The conduction path of the switching is also investigated through electrical characterization of devices both before and after physically cutting the devices in half, as well as through infrared imaging of the devices during operation. The results suggest that the electrical behavior of these devices is dominated by a localized, charge-based phenomenon that exhibits a dependence on device area.

Memristors were fabricated on flexible polyethylene terephthalate substrates consisting of an oxide film generated through hydrolysis of a spun-on sol-gel. X-ray photoelectron spectroscopy, spectroscopic ellipsometry, transmission electron microscopy, and electron energy loss spectroscopy measurements indicated that the oxide films were amorphous TiO2 with a significant fraction of organic material, causing a heterogeneous surface morphology. Despite the morphology and the organic material, these memristors exhibit switching similar to "traditional" memristors. Also, current-voltage (I-V) measurements suggest that this switching was not due to the electric field in the memristors. Additionally, thermal imaging measurements and I-V measurements performed after sectioning the memristors suggest that conduction occurred via localized conduction pathways. Capacitance-frequency and conductance-frequency measurements indicate an additional dielectric loss mechanism was present prior to switching to the higher current state. This loss mechanism is attributed to dipoles present in the organic components of the oxide films from the original sol-gel.

Background: This paper presents a reduced mathematical model using a practical numerical formulation of the thermal behavior of an integrated power electronics module (IPEM). This model is based on the expanded lumped thermal capacitance method (LTCM), in which the number of variables involved in the analysis of heat transfer is reduced only to time. Method of Approach: By applying the LTCM, a simple, nonspatial, but highly nonlinear model is obtained for the case of the IPEM Generation II. Steady and transient results of the model are validated against results from a three-dimensional, transient thermal analysis software tool, FLOTHERM™ 3.1. The electrothermal coupling was obtained by implementing the reduced-order thermal model into the SABER™ circuit simulator. Two experimental setups, for low- and high-speed thermal responses, were developed and used to calibrate the reduced model with actual data. Results: The comparison of the LTCM model implemented in a Generation II IPEM with FLOTHERM 3.1 results compared very favorably in terms of accuracy and efficiency, reducing the computational time significantly. Additional validations of the reduced thermal model were made using experiment data for the low-speed thermal response at different constant powers, and good agreement was demonstrated in all cases. A comparison between SABER™ simulations, which incorporated the proposed LTCM, and the fast thermal experimental response results is also presented to validate the dynamic electrothermal model response, and excellent agreement was found for this case. Conclusions: The good agreement found for all three cases presented, the three-dimensional, transient numerical formulation, and the low- and high-speed experimental data indicates that reduced electrothermal models are an excellent alterative for design methodologies of new generations of IPEMs.