Physicists at the March Meeting reported on their work looking for ways to keep the exponential growth in computing power from tapering off by researching possible new materials and techniques.
While computing power has exponentially increased over the last several decades, many in the field have worried that it could saturate at some point. Moore’s Law predicts that the number of transistors that fit on a computer chip, and its corresponding processing power, will double every 18 to 24 months. Though not a law of nature, it has been generally accurate over the last forty years. However as transistors have shrunk, many are worried about hitting a point where the transistors can’t scale down any further and will start to become less efficient.
Graphene, thin sheets of hexagonally arranged carbon atoms, has garnered a lot of attention for its remarkable electrical properties. Some physicists say that it could be the material that in the future will replace silicon as the semiconducting basis of transistors.
“Over the past few years researchers have been looking into different materials…and graphene is considered one of the most promising,” said Helen Xiangyu Chen of Stanford University.
At the meeting, Chen announced that her team has made a significant leap forward in developing a workable graphene transistor. They have been able to successfully integrate graphene interconnects into commonly used complementary metal-oxide semiconductor (CMOS) technology and achieve processing speeds as fast as a gigahertz.
Also at the meeting, Phaedon Avouris of the IBM Research Center announced that his team has taken an important step towards overcoming a major impediment researchers have encountered in graphene-based transistors. In order to function, semiconductors need a band gap where current completely stops flowing when in the “off” position. Creating a graphene transistor with the necessary band gap has so far proven difficult, but Avouris said that they are getting close. He and his team built a transistor out of two layers of graphene where only 1 percent of current flows in the off position. Though it’s still not quite a large enough band gap to be used in digital electronics, it is about twenty times more effective than single layers of graphene.
Just as important as the transistors that do the computing is the wiring that connects to each transistor. The surging field of topological insulators shows a lot of promise for physicists working to solve the fundamental problem of energy dissipation in integrated circuits.
Shoucheng Zhang of Stanford University said that while the engineering of transistors has improved, the fundamental materials for the connecting circuit haven’t changed much. Electrical wiring on small scales loses a lot of energy as heat when flowing electrons collide.
“We’re facing a situation where fundamental new ways of thinking are required,” Zhang said, “The engineering [of circuits] has been progressing, but the fundamentals have not changed.”
Physicists say that topological insulators do much to reduce lost energy. Using the quantum Hall effect, flowing electrons organize along the surface of a specially prepared layered semi-conducting material. When they do, the charges all flow in a single direction, preventing charges from colliding with each other and cutting out nearly all resistance, much like a superconducting material.
At the meeting, physicists announced that for the first time they have been able to synthesize a material from mercury telluride with these properties that don’t require powerful magnetic fields or ultra cold temperatures. Also from Stanford, Yi Cui presented a material, derived from bismuth selenide, made into tiny nano-ribbons that could, with more development, function as wires between transistors.
The problem of overloading wires has led some researchers to think that the days of electronics are numbered. Many see photonics as the next step in computer processing, where data is sent and received using photons rather than electrons.
“The wiring within machines is becoming a bigger problem than the logic…It’s a basic scaling problem–once you’ve filled the space with wiring, you can’t send any more information,” said David Miller of Stanford University, “The energy benefit of optics is something we’re only beginning to start to see.”
At the meeting, researchers offered an overview on the current state of the field. Recent research suggests that silicon and germanium each are a promising medium to send photonic computing signals. More development is needed to improve the optic cables and to shrink transmitters and receivers before silicon photonics replaces electronic computer chips. However physicists developing these new computer systems say that photonics would take up less space and require significantly less energy than traditional electronics.
“The last century I consider to be the century of electronics. This century I consider to be the one of photonics,” said David Lockwood of the National Research Council in Canada.
©1995 - 2017, AMERICAN PHYSICAL SOCIETY
APS encourages the redistribution of the materials included in this newspaper provided that attribution to the source is noted and the materials are not truncated or changed.
Editor: Alan Chodos