In a follow up to our post last week, James C. Ellenbogen writes to provide insight and personal perspective on the world’s first programmable nanoprocessor, achieved as the product of a collaboration between Harvard and MITRE, with the team at MITRE comprising Shamik Das, James Klemic, and Ellenbogen.
The nanocomputer system we demonstrated is programmable. We showed that multiple logical and arithmetic functions can be programmed and implemented on the same circuit “tile,” which is built from nanometer-scale (i.e., molecular scale) electronic devices integrated on that same scale. The nanodevices on the tile are non-volatile, so the system incorporates memory. Further, the ultra-tiny, ultra-low power system is quite scalable: the device performance and architecture are such that multiple tiles can be linked together to produce more complex and capable nanoprocessors. Additionally, consistent with the Foresight vision, these are designed to be nanocontrollers, tiny computers intended to govern the function of other tiny mechanisms.
Thus, as I first stated in a presentation at the 6th Foresight Conference in November 1998, “There will be nanocomputers and they will be molecular electronic computers.” So now there are.
For me, though, the journey to the realization of the first nanocomputer begins back in 1986, when I first heard Eric on public radio talking about nanocomputers. It took almost 25 years from that initial point of inspiration, but I believe that we have taken a major step on the road to engineered nanosystems governed by other engineered nanosystems. I hope that you find the potential for this innovation to be as exciting as I do.
James C. Ellenbogen, Ph.D.
Chief Scientist, Nanosystems Group and Emerging Technologies Division
The MITRE Corporation
Dr. Ellenbogen also writes to clarify the speed of their nanoprocessor, as reported last week, and the potential for further shrinking the dimensions of the nanoprocessor:
You have a link there that suggests the potential for 2 Terahertz switching speeds. That link seems to suggest that our entire nanoprocessor might be sped up to that very high rate. The supplementary material for our paper in Nature does mention the potential for 2 THz switching, but that is for individual nanodevices. That may be misleading, though, if one applies this figure to the entire system. As part of our design simulation process at MITRE, we calculate that with the present devices and the present interconnect strategy, the entire system probably can only be accelerated to operate at about 100 MHz. This is because of all the intrinsic resistances and capacitances in the interconnects and the way they combine in a tiny network. From our point of view this 100 MHz rate of operation will be OK, though, because it is well suited to a tiny controller. The 100 MHz rate is faster than most of the other tiny systems we want to sense and control.
Some better news, though, is that we do not feel quite so constrained when it comes to making the system much smaller in area and much more dense. We calculate that it is probably possible to scale the footprint of the system down by a factor of 600 to 1200 (i.e., 1000 in round numbers). That would provide the present computational functionality of our nanoprocessor tile in a footprint of only approximately 1 sq. micron, instead of the present 1000 sq. micron (e.g., 1 micron by 1 micron, instead of approx. 30 micron by 30 micron).
MITRE also provided a press release describing the work MITRE-Harvard Team Develops First Programmable Nanoprocessor:
MCLEAN, Va., February 10, 2011 — The world’s first programmable nanoprocessor has been developed and demonstrated by an interdisciplinary collaboration between teams of scientists and engineers working at The MITRE Corporation and Harvard University.
The groundbreaking prototype computer system is described in a paper published today in the journal Nature. The system represents a significant step forward in the complexity of computer circuits that can be built from nanometer-scale (i.e., molecular scale) components. It also represents an advance because the ultra-tiny nanocircuits can be programmed electronically to perform a number of different basic arithmetic and logical functions.
The versatile, ultra-tiny circuits are assembled on a tiny “tile” from sets of precisely engineered and fabricated germanium-silicon wires, surrounded by insulating shells of metal oxides, but still having a total diameter of only 30 nanometers. The novel architecture of the tiles allows a number of them to be connected to assemble even more capable nanoprocessors that could, for example, control a complex electromechanical system.
An additional feature of the advance is that the circuits in the nanoprocessor operate using very little power because their component nanometer-scale wires contain transistor switches that are “nonvolatile.” Unlike transistors in conventional microcomputer circuits, once the nanowire transistors are programmed they remember without any additional expenditure of electrical power.
“Because of their very small size and very low power requirements, these new nanoprocessor circuits are building blocks that can control and enable an entirely new class of much smaller, lighter weight electronic sensors and consumer electronics,” according to Shamik Das, lead engineer in MITRE’s Nanosystems Group and chief architect of the nanoprocessor.
Other members of the development team at MITRE—a pioneer in the nanotechnology field since 1992—included nanotechnology laboratory director James Klemic and James Ellenbogen, chief scientist of the Nanosystems Group. The MITRE team collaborated with a five-person team at Harvard, led by Charles Lieber, a world-leading investigator in the field of nanotechnology, especially for nanowire-based innovations such as the new nanoprocessor.
Ellenbogen, who has worked for nearly two decades toward the development of computers integrated on the nanometer scale, including prior collaborations with Lieber, added that, “This new nanoprocessor represents a major milestone toward realizing the vision of a nanocomputer that was first articulated more than fifty years ago by physicist Richard Feynman.”