Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Resistive Random Access Memories (ReRAMs) have been researched intensively in the last past decades as a promising alternatives technology for the next-generation non-volatile memory (NVM) devices. ReRAMâs excellent performance properties such as high switching speed, excellent scalability, and low power consumption together with accessible fabrication techniques has raised increasing interests. One ReRAM device in its most basic structure consists of a thin layer of transition metal oxide layer (TMO) sandwiched between two metal layers (Top and Bottom Electrodes). ReRAMs could be categorized into different groups regarding the implemented materials and their performance. The concept of switching mechanism is straightforward; the as-fabricated devices appeared to be highly resistive (HRS); when the top electrode (TE) is biased, the resistance state of the devices switched to the lower resistance level (LRS) in an operation called SET. This process is commonly reversible through the application of proper negative external stimuli, which can change back the resistance state of the device from LRS to HRS. Oxygen vacancies are recognized as the main elements controlling the device performance. During the set process, the oxygen vacancies rearrange to form conductive bridges between the TE and bottom electrode (BE). Further application of negative biases can partially/ fully dissolve the conductive filaments and led to HRS by disconnecting the two metallic electrodes from one the other. The excellent ReRAMs performance is enabled due to their switching mechanism which is based on relocation and movement of nano-scale oxygen vacancies, but the stochastic nature of the formation of filament also makes it hard to control this phenomenon precisely. As a result, the reported variation issue in the key switching parameters of ReRAMs lessens the reliability of this technology and hinders its commercialization. To suppress the variability hurdle in ReRAMs, it is crucial to have a better understanding of the switching mechanism. Hence, the interaction between deposited materials in the as-fabricated devices and during electrical measurements under external stimuli needs to be intensely studied to obtain an in-depth knowledge of the behavior of the oxygen vacancies. The first purpose of this research is to overcome the variation in the key switching parameters of stand-alone ReRAMs to be used in cross-point structures. In this work, we implemented two main approaches to improve the reliability and uniformity of HfO2-based ReRAMs, the structural engineering, and post-fabrication thermal treatment v and studied the effect of each method on the performance of HfO2-based fabricated ReRAMs. ReRAMs due to their simple structure can be implemented to the cross-point configuration. However, the sneak current through unwanted neighbor cells significantly decreases the system efficiency, deteriorates the read margin, and limits the maximum size of a cross-point array. To overcome the sneak path issue, different strategies could be implemented; among all, serially connecting each memory element to an additional selection device in a 1S1R configuration is an active way to introduce selectivity to the cross-point arrays. In the second part of this work, we introduced a novel one-selector one-resistor (1S1R) configuration, which eliminates the need for the physical wiring and provides valuable information on isolated selector/resistor and the integrated 1S1R.
Mihai Adrian Ionescu, Elison de Nazareth Matioli, Pirouz Sohi, Riyaz Mohammed Abdul Khadar