The bleeding edge: After 10 years of study, researchers say they have finally created the world's first functional graphene-based semiconductor. It should prove helpful in quantum and traditional computing and will allow Moore's Law to continue, much to Jensen Huang's dismay.
Earlier this month, researchers from the Georgia Institute of Technology in Atlanta published a paper in Nature. The study discusses producing epigraphene from silicon carbide (SiC). Semiconducting epitaxial graphene (SEC), also known as epigraphene, has far greater electron mobility than silicon.
According to Georgia Tech Regents Professor of Physics Walt de Heer, electrons can move 10 times faster than traditional silicon-based transistors. This exponential boost means that chips using epigraphene could potentially hit cycles in the terahertz range.
Making epigraphene builds on a process that has produced graphene for half a century. First, two SiC chips are stacked inside a graphite crucible and placed inside an argon quartz tube wrapped with copper tubing. High-frequency current is sent through the copper coil, heating the graphite crucible through induction to 1,000°C for about an hour.
As silicon evaporates from the surface of the SiC chips, it is replaced with carbon, forming a two-dimensional (single-atom) graphene layer. The wafer produced is charge-neutral, so when removed from the tube, it instantly becomes doped by oxygen. They then release the oxygen doping by heating the graphene to 200°C in a vacuum, creating epigraphene on a silicon carbide substrate.
According to de Heer, the process is relatively inexpensive.
"The (SiC) chips we use cost about $10 [US], the crucible about $1, and the quartz tube about $10," the professor explained to IEEE Spectrum.
Scientists have produced semiconducting graphene since 2008 by heating SiC in a vacuum. However, it lacked a measurable bandgap, so transistors cannot turn on and off. De Heer's and his team's modified method eliminates this problem.
Previous efforts to produce bandgaps have involved modifying a substrate with graphene nanoribbons or nanotubes. These methods have not yielded successful results because they require high precision when depositing the ribbons on the substrate.
"There has been some success with graphene nanoribbons, but in principle, this technology is very similar to semiconducting carbon-nanotube technology which has not been successful after 30 years of nanotube research," says de Heer.
Researchers have had more success creating a bandgap by deforming the graphene (wrinkling). However, this produces a bandgap of only 0.2 electron volts, which de Heer says is too small to be practical. By comparison, silicon has a bandgap of 1.12 eV. Georgia Tech's method creates a bandgap of 0.6 eV, enough for logic switching while operating at cooler temperatures.
"Our research is distinct from these other approaches because we have produced large areas of semiconducting SEC on defect-free, atomically flat SiC terraces," de Heer said. "SiC is a highly developed, readily available electronic material that is fully compatible with conventional microelectronics processing methods."
While science has successfully produced functioning, highly mobile semiconducting epitaxial graphene, SEC processors in quantum or regular computers are still a distant vision. For one, de Heer says it requires further study to determine if it is more suitable than the superconductors used in contemporary quantum computers.
As for silicon computing, the team already knows SEC is a superior semiconductor with far lower resistance. Therefore, faster speeds and cooler operating temperatures are achievable. However, there is no easy way to incorporate SEC into traditional silicon electronics. To reap the benefits the material has to offer requires a complete paradigm shift in current manufacturing practices.
"I compare this work to the Wright brothers' first 100-meter flight," says de Heer. "It will mainly depend on how much work is done to develop it."
Image credit: Chris McKenney
