This 3-D rendered image represents a large-scale and complex nanosystem integrating multiple emerging nanotechnologies for logic, memory, and sensing. It realizes a new 3-D computing architecture for high-performance and energy-efficient electronic systems that can capture and process massive amounts of data.A three-dimensional nanoelectronic system made up of stacked layers of carbon nanotube transistors and random-access memory cells could ease a computing communication bottleneck, according to researchers from Massachusetts Institute of Technology and Stanford University [Shulaker et al., Nature (2017), doi: 10.1038/ nature22994].
Computing operations involving vast volumes of data will soon become more than current systems based on silicon metal-oxide-semiconductor field-effect transistor (MOSFET) technology can handle. Simply improving existing devices will be insufficient, so a new transformative technology solution will be needed.
Max M. Shulaker and his colleagues believe that a combination of carbon nanotube field-effect transistors (CNFETs) and resistive random access memory (RRAM) could be the answer.
“Circuits today are two-dimensional, since building conventional Si transistors involves extremely high temperatures of over 1000 ?C,” comments Shulaker. “If you build a second layer of Si circuits on top, that high temperature will damage the bottom layer of circuits.”
Instead, the new design relies on layers of millions of CNFETs and RRAM cells fabricated on top of each other at much lower temperatures without any damage. By assembling layers of CNFETs to perform computing right on top of RRAM cells to store data, along with data input and output devices, the team create a ‘vertically integrated’ three-dimensional nanoelectronic system.
“The devices are better: logic made from CNTs can be an order of magnitude more energy-efficient compared to today’s logic made from Si, and, similarly, RRAM can be denser, faster, and more energy efficient compared to DRAM,” points out co-author H.-S. Philip Wong.
The interconnections within and between chips are also improved by the three-dimensional architecture. Consequently, the nanosystem can capture massive amounts of data every second, store it on the chip itself − rather than in a separate memory device − and process the captured data in situ.
To demonstrate the capabilities of the design, the researchers integrated a layer of CNFET chemical vapor sensors into the nanosystem, each of which is connected directly to an underlying memory cell. The sensors can write their data into their associated memory cells in parallel, generating the capacity to handle massive amounts of data simultaneously.
The prototype nanosystem successfully classified common substances including nitrogen gas, lemon juice, vodka, wine, and beer. But, point out the researchers, the layer of chemical vapor sensors could readily be replaced with other forms of input/output or computational systems.
Timothy M. Swager of Massachusetts Institute of Technology believes the work realizes the dream of integrating organic nanoelectronic materials into high performance computational devices.
“Shulaker and coworkers demonstrate the potential of this advance for the large scale integration of sensor devices and in doing so provide a powerful platform for next generation gas and biological sensors,” he comments.
John A. Rogers of Northwestern University agrees that the work represents a milestone not only in densely integrated, nanotube-enabled electronics but also in schemes for three-dimensional, heterogeneous integration.
“These findings, together with the recent report of nanoscale nanotube transistors from IBM researchers, represent powerful evidence that nanotubes may have an important role to play in future integrated circuit technologies,” he comments.
This article was originally published in Nano Today (2017), doi: 10.1016/j.nantod.2017.08.005.