Rice researchers have developed a new kind of memory chip from silicon oxide using nanotechnology that could increase maximum storage space on chips and open the way for 3-D memory storage.
Flash memory, the current standard, uses an applied voltage to allow electrons to flow from a source through a gate and into a drain. The transistor begins conducting when electrons leave the gate. Two-terminal silicon chips, which contain a source and drain, stand out from standard three-terminal flash memory, Chemistry Professor Jim Tour said. Without the need for a third terminal, these silicon nanocrystal devices can be built into three-dimensions, which allows for increased density and memory space, Tour said.
"[The silicon nanocrystal devices are] faster than flash memory by two to three orders of magnitude," Tour said.
When acted on by an initial voltage pulse, silicon oxide creates a 3-5 nanometer filament - about 1/200 the size of a human hair - that moves atoms in a faster process than flash memory, Tour said. Silicon nanocrystals form underneath these wires. While Tour's research group has recorded speeds up to 50 nanoseconds, Tour said because lab equipment limits the measurements, the speeds are probably faster. Flash memory runs at speeds of one microsecond, or roughly 5 percent of the silicon chip speeds.
According to Moore's Law, the number of transistors per square inch on an integrated circuit should double every 18 months until the transistors reach a maximum density. To bypass the problem of transistor density, scientists have also built in more functionality for each transistor, allowing each transistor to perform more efficiently.
Scientists predict flash technology will reach its limit at around 20 nanometers because of constraints associated with storing charge, Tour said.
"[Silicon nanocrystals devices] give you the ability to stay on Moore's Law for several more generations," Tour said.
Last year, Tour's research group coated silicon oxide, an insulator, with strips of carbon and used electric current to break and reconnect the strips, analogous to turning a circuit on and off. Tour said many of the group members assumed the carbon caused the switching behavior.
Graduate student Jun Yao, primary author for an article on the silicon chips that ran in the Aug. 31 online edition of Nano Letters, suggested silicon caused the switching after performing routine experiments to isolate the effects of the carbon.
"At that time we didn't know what was going on," Yao said. "Silicon oxide is insulating and the only conducting material was carbon."
To confirm carbon caused the switching, Yao removed the coating and placed only the silicon oxide between two layers of silicon that acted as electrodes. To his astonishment, the switch worked even in the absence of carbon.
"It was a big surprise to me," Yao said. "Without the carbon we found the effects were still working."
Yao also faced the challenge of explaining metal filaments contained in the silicon oxide. Persuading his group that silicon caused the switching took more than half a year.
Yao replaced the carbon surrounding the silicon oxide with various materials and experienced the same result.
"I just thought the picture was so clear, but no matter what I did, people could always find another argument," Yao said.
To better observe the processes inside the silicon oxide, Yao came up with the idea to cut out a small segment of silicon oxide and locate the switch. He placed a carbon nanotube on top of the silicon oxide and sliced the silicon oxide into pieces 10 nanometers thick and observed the pieces under a transmission electron microscope.
Yao determined that applying a charge stripped oxygen atoms from the silicon oxide to form a conductive pathway of silicon nanocrystals that turn the chip into an "on" state. The path can be broken to put the chip into an "off" state. Tour said the silicon nanocrystals also allow for multiple activation states, turning the chip to high, medium, low and off. Newer flash memory designs have the same capability but require more space, and traditional flash memory designs had only on and off states.
"We're now looking forward to utilizing this silicon oxide," Yao said.
Tour said the silicon nanocrystal research has funding from the Army Research Office, Navy and Air Force. Tour is in the process of talking with large-scale manufacturers to build 1,000-bit silicon nanocrystal memory chips and showcase their capability.
Although Yao said he enjoyed the publicity the project has received, including a front-page article in the Aug. 30 issue of The New York Times, he said research is more than finding results.
"At the beginning, people didn't believe me, and if I did not stick to that, nobody would know silicon oxide alone could do this kind of thing." Yao said. "When I look back, the excitement is not about the research itself, but about the experience."
Yao said researchers should never be afraid to defend their results, even when traditional thinking suggests otherwise.
"What's important is that you stick to what you believe," Yao said. "Truth is truth."