Scaling study of phase change memory using carbon nanotube electrodes

Phase change material (PCM);Phase Change Memory;Carbon Nanotube (CNT);Finite Element Model;Resistive Memory;Resistive random access memory (RRAM);Crossbar

Xiong, Feng

关键词: Phase change material (PCM);    Phase Change Memory;    Carbon Nanotube (CNT);    Finite Element Model;    Resistive Memory;    Resistive random access memory (RRAM);    Crossbar;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/49709/Feng_Xiong.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Demands for data storage and computer memory are growing exponentially. It is thus essential to find a new scalable, energy-efficient memory technology. We have been investigating the smallest and most energy-efficient data storage technology, based on phase change materials (PCMs) rather than conventional charge-based memory technology. PCMs are materials which undergo a large change in resistance in the presence of electric and thermal fields. Memory based on PCMs could be ten times denser and one hundred times more energy-efficient than present technologies, enabling instant turn-on computers and mobile devices (smart phones) that could last for weeks on a single battery charge.One drawback of PCM technology is its relatively high programming power, since the PCM bit needs to be heated above its crystallization temperature during operation. If we can reduce the size of the electrodes which controls the size of the PCM bit, we could significantly reduce the programming power. In this work, We have been working to achieve the most energy-efficient memory to date by integrating PCMs with nanoscale electrodes made from carbon nanotubes (CNTs). CNTs are the smallest conductors known today, with diameters ranging from ~1-5 nm.Firstly, We investigated the compatibility of CNTs with PCMs and the feasibility of integrating the two materials. We demonstrated for the first time that CNTs can be used as a heater to induce ultra-narrow phase change regions in PCMs such as Ge2Sb2Te5, while applying currents on the order of 10 μA. We then designed and built lateral PCM memory devices with nanoscale CNT electrodes. We created tiny gaps, roughly 20 nm in width, in the middle of CNTs and placed PCMs inside these gaps. We demonstrate reversible switching with programming currents from 1 to 8 µA, more than 100× lower than state-of-the-art PCM devices, enabled by the very small volume of PCMs addressed with a single CNT. We performed a device scaling study on more than 100 such devices, which suggests that our device is highly scalable with the bit size and memory switching is possible with voltages below 1 V and energy less than fJ/bit. We have since developed a novel lithography-free, self-aligned technique to fabricate sub 40-nm PCM nanowires with CNT electrodes.By covering our CNT device with PMMA (eBeam resist) and then passing current through the CNT, the device gets hot from Joule heating, which evaporates the PMMA directly covering the CNT and creates a narrow trench. We then create nanogap in the CNT and sputter-deposit PCMs to fill the nanogap and trench in PMMA. A PCM nanowire that is self-aligned with the two CNT electrodes is formed, after PMMA lift-off. These devices show excellent characteristic threshold switching behavior with programming currents (~0.1 μA SET, ~1.6 μA RESET) and power dissipation among the lowest reported to date. We also adopted this self-align technique and developed a new approach to achieve individually electrically addressable CNTs using Cu as an etch mask. We also studied AlOx-based resistive random access memory with CNT crossbar electrodes. We showed both metallic and semiconducting CNTs would effectively switch AlOx bits and demonstratedON/OFF ratios up to 105 with programming currents of 1 to 100 nA.From a practical aspect, our research has demonstrated PCM-based memory with 100× lower power consumption than previous state-of-the-art devices. From a fundamental aspect, our study has examined the ultimate limits of PCM devices, down to dimensions of single molecules. Our findings address the potential size and power reduction that are possible for programmable bits of PCM. These results are encouraging for ultra-low power electronics and memory based on programmable PCM with nanoscale carbon interconnects. Such advances could open exciting opportunities from mobile electronics to butt computing, all applications requiring extensive, energy-efficient data storage.