Helena Silva (PI, University of Connecticut) and Chung Lam (Co-PI, IBM Watson Research Center)
The objective of this project is to investigate thermoelectric transport in nanometer scale phase-change bridge memory devices and to devise improved phase-change memories that utilize the asymmetries introduced by thermoelectric effects such as multi-bit per cell operation.
Phase-change memory has been the subject of intensive research due to the continued need for higher capacity storage and expectations of the end of current memory scaling in the near future. Our findings point to the significance of Thomson Effect (thermoelectric effect in uniform materials) in phase-change processes. A better understanding of thermoelectric transport and phase-changes at the micrometer and nanometer scales can be applied towards thermoelectric devices for clean power generation and cooling, higher efficiency solar cells that combine photovoltaic with thermoelectric generation and on-chip cooling for micro/nanoelectronics. This project is carried out in partnership with IBM Watson Research Center.
This research consists of detailed analysis of Thomson Effect at nanometer scale using phase-change memory devices. The role of electronic convective and electron-phonon scattering components in the thermoelectric transport in these structures is investigated. Students learn about nanoelectronic devices, phase changes, thermoelectric transport, nanofabrication, electrical and physical characterization, instrumentation and numerical modeling. This project allows investigation of basic physical phenomena relevant to a promising memory device technology and offers a rich program for educational and outreach activities at all levels. This collaborative research strengthens an important relationship between University of Connecticut faculty and students and IBM scientists.
Project Overview This project was a collaboration between the University of Connecticut’s Nanoelectronics Laboratory and IBM Watson Research Center’s Advanced Memory Group to study thermoelectric effects in nano-scale phase-change memory (PCM) devices. Phase change memory is an emerging technology for higher speed non-volatile memory which can lead to significant energy savings in memory and computation. Currently the main difficulties with implementation of phase-change memory are the high power required for programming and limited reliability and endurance of the devices due to the high-temperatures involved (up to ~ 900 C). In this project we sought to better understand thermoelectric effects in PCM devices which determine the internal temperature distributions and hence impact both power consumption and device reliability. Thermoelectric effects (Peltier, Seebeck and Thomson effects) are very significant in PCM devices due to large current densities, high temperatures and very large temperature gradients due to the small scales (> 10 k/nm). We have performed experimental and computational studies on Ge2Sb2Te5 PCM devices. The devices were fabricated at IBM under a Joint Study Agreement initiated for this project. One UConn graduate student spent on average 2 days/week at IBM working on the fabrication processes. The devices were electrically characterized at the UConn. Computational studies using obtained experimental parameters were also performed at UConn. Outcomes Our work has shown that there is a clear dependence of the heating profile of the cells on the current polarity, leading to a preferable operation polarity (negative polarity – defined as positive current flowing from the narrow, nano-scale heater to the common electrode). This preferable polarity operation results in lower power consumption and lower thermal gradients within the phase-change element which are thought to be related to failure of the cells due to element segregation within the compound material (GST). These results can be taken into account in designing more efficient PCM devices and memory arrays. As part of this project we have developed a general technique for high-speed and high-temperature electrical characterization of metastable amorphous and crystalline phases in phase-change materials. We have also developed efficient computational models that enable physics-based simulations of phase-change and polycrystalline materials which add significant accuracy and flexibility to device simulations for phase-change memory which are required for a better fundamental understanding of phase-change materials as well as advances toward larger scale implementations of phase-change memory technology. Both experimental and computational results are expected to be applicable to other high-temperature small-scale electronic devices such as other resistive memories or micro/nano scale thermoelectric generators for on-chip energy harvesting or thermal control. Work related to this project resulted in the publication of 1 book chapter, 13 journal articles and 5 conference proceedings and in 22 conference presentations at international conferences. Broader Impacts During the five years the project partially supported the education of 12 graduate students, 2 of which were minority students. These students received the remaining support through fellowships and university teaching assistantships. The project also supported 9 undergraduate students, 5 of which were minority students. All students were trained in the Nanoelectronics Laboratory in experimental and computational techniques for research on electronic devices and materials. Graduate students have worked at IBM's state-of-the-art micro/nanofabrication facilities learning about the techniques involved in fabrication of modern electronic devices and benefitting from frequent interactions with IBM's engineers and scientists. Most of the undergraduate students involved with this project are now pursuing graduate degrees. During this project the Principal Investigator has offered a senior undergraduate/graduate class on memory devices with a significant focus on PCM, which strongly benefitted from the knowledge and experience gained through this project. The project has greatly benefitted from this academic-industry collaboration and has led to successful integration of research and education of graduate and undergraduate students, as well as numerous outreach activities to high-school students and prospective undergraduate students and their families. This collaboration is now continuing through a new Joint Study Agreement (approved in December 2014) for further joint studies on phase-change memory materials and devices.
