Japan reports advances in single-electron devices
By Yoshiko Hara and Paul Kallender
EE Times
(04/13/01 13:22 p.m. EST)
TOKYO — The push to develop single-electron devices has gotten a shot in the arm from two recent contributions from Japan. Toshiba Corp. is reporting that it has developed a single-electron transistor (SET) that operates at room temperature, while Nippon Telegraph and Telephone says it has developed what it claims is the world's first single-electron charge-coupled device.
Most existing SETs work only at extremely low temperatures. Toshiba, however, has demonstrated a room temperature-operational SET that functions as a transistor, a memory device and as part of a circuit, said Ken Uchida, a researcher at the company's Advanced LSI Technology Laboratory (Yokohama, Japan).
Toshiba's SET operation depends on manipulating an electron's behavior under what is known as the Coulomb Blockade Effect. This phenomenon relies on the simple principle that two electrons cannot enter one quantum hole. If one electron is in one quantum hole, the next electron cannot enter because of mutual repulsion. But when a quantum dot confining an electron becomes small enough, the electron's repulsive energy becomes strong enough to demonstrate SET operation at room temperature.
"To move a single electron at room temperature, the size of the quantum dot needs to be less than 10 nanometers [in diameter], but it was difficult for current technology to realize such a small size. This was a major obstacle to room-temperature operation," Uchida said.
Peaks and troughs
Toshiba engineers used a chemical-treatment process instead of a lithographic approach to create the small quantum dots. When fabricating the MOSFET-like device on a silicon-on-insulator (SOI) wafer, the engineers used alkaline solutions to undulate the surface of the channel.
This process formed a jagged, nanometer-scale channel consisting of razorlike peaks and crevicelike troughs.
"We estimate that the troughs are less than 2.5 nm thick," Uchida said. "It is difficult for electrons to enter the troughs, and these areas function as if they are insulators." As a result, the peaks of the silicon are like small "rooms" that can contain only one electron, he said.
The SET device shows both switching and memory functions. This structure is similar to that of a MOSFET device, consisting of a source, a drain and gate electrodes. When voltage is applied at the gate, electrons flow through the channel, but one by one from room to room, sequentially.
When a high gate voltage is applied, the channel is filled with electrons. Most electrons flow through the relatively large rooms; however, even the small rooms can take a single electron. But when the voltage is lowered, some electrons are stranded in some of the rooms, Uchida explained. These remaining electrons influence electron flow in the form of performance changes before and after the voltage is applied, a function that can be used as nonvolatile memory.
When the gate voltage is increased, the drain current oscillates, which is a characteristic of SET operation based on the Coulomb Blockade Effect, Uchida said. If the peak of the current oscillation is assumed as "on" and the valley as "off," the peak-to-valley current ratio (PVCR) is more than two digits, the highest ratio achieved, according to Uchida. The previous best was only 1 or 1.1, he said. PVCR is an essential characteristic for performance as a switching device.
Toshiba engineers have verified that the SET-pMOS circuits can be programmed by using the nonvolatile memory function incorporated in the SET. This suggests that SET-based logic circuits will be programmable with multiple functionality. Toshiba has demonstrated a logic circuit with two parallel SETs with memory functions that showed four logic operations, including NOR and AND.
Better still, the device process is fully compatible with the CMOS process. "In this way we have fabricated SETs that are operating practically with CMOS on a chip," said Junji Koga, another research scientist at Toshiba's Advanced LSI Technology Laboratory.
Uchida said he hopes that Toshiba can develop a prototype LSI sometime around 2010. "We don't think SETs will replace all CMOS LSIs, but SETs can be used as the core function blocks and CMOS can still be useful as peripherals," he said.
NTT development
While Toshiba breaks new ground on SETs, researchers at NTT's Basic Research Laboratories (Atsugi, Japan) have produced a potential breakthrough in atomic-level circuitry, by creating a single-electron CCD.
The device, developed by Yasuo Takahashi, group leader of NTT's Silicon Nanodevices Research Group, and his associate Akira Fujiwara of the same laboratory, consists of two closely packed silicon-wire MOSFETs on an SOI wafer that can act as a CCD and perform single-electron sensing.
The scientists figured out how to generate and store holes in the channels of either of the MOSFETS and found that holes could be transferred between the two and their exact position monitored. In the closely packed array of two silicon-wire MOSFETS defined by polysilicon gates, single holes can be stored. The MOSFETS share a 15- to 20-nm-thick T-shaped silicon wire connected to three large n-type electrodes consisting of one source and two drains, NTT reported.
"We found that when we applied a positive voltage to the substrate and a negative voltage to the upper gate, we could draw electrons to the substrate and also obtain a positive-charge hole," Takahashi said. "We could, therefore, divide the positive and negative elemental charges in the space and keep them separate and at the same position."
With the CCD function the device captures a small number of charges in the silicon channel under either gate, and can transfer the charges between the two gates. In the single-electron sensing function, Takahashi and Fujiwara also found they could use the device to sense the precise number of stored charges on each silicon channel. The scientists believe that the sensing capability will permit them to regulate the number of stored charges.
The charge-sensing scheme is based on a discovery Takahashi and Fujiwara call an electron-hole (e-h) system, in which the silicon wire is separated into a region that stores charges (the hole) and a region where the sensing-electron current flows. If the electron is stored, the system uses the hole-sensing current. To perform the separation, a voltage slope is introduced within the silicon wire, perpendicular to the wafer.
The e-h system is significant for three reasons, Takahashi said. First, because the structure is so tiny, the existence of a hole changes the electron-channel potential and raises the electron current.
Second, although the electron and the hole exist in the same silicon channel, they do not recombine, sometimes for up to several hundreds of seconds, due to the small overlap of wave functions between the electrons and the holes.
This e-h recombination, the researchers found, can be controlled by adjusting the voltage slope.
And third, when the number of stored holes decreases by only one, that decreases the number of electrons that pass through the silicon channel, suppressing the e-h recombination. In this way, e-h recombination lifetimes are dependent on the number of stored holes, so that regulation of the number of stored holes is possible.
But the most important feature of the device, which at the moment rests on a bulky 90-nm long, 20-nm diameter silicon-wire channel, is that it does not rely on tunnel barriers, which until now have been deemed essential for single-electron devices, Takahashi said.
No tunneling barriers
"This device enables us to transfer a single electron or a single hole. Everyone thought single-electron tunneling was necessary to transfer a single electron or a hole. It is usually very difficult to fabricate a small island and a tunnel capacitor on both sides of the island. In our device, we do not need a SET or any tunneling barriers," Takahashi said.
The absence of tunnel barriers could yield important advantages over single-electron pump devices, Takahashi said. One is ease of fabrication, because the single-electron CCD is based on silicon. "We can transfer single charges in either direction just by arranging the gate electrodes on silicon wafers," he said.
The second advantage is speed, according to Fujiwara. The electron-transfer speed theoretically maxes out at 1 terahertz or about a 1-picosecond delay, Takahashi said. By comparison, the maximum operation speed of single-electron transistors is slower than 100 GHz, or 10 picoseconds.
Another advantage, Fujiwara said, is that the device's single-electron sensing ability means it already has a de facto electrometer, obviating the need to construct one separately.
But parlaying this breakthrough into manufacturable devices may take 15 years, the scientists said, adding that they have not completely investigated the recombination process. Before any major refinements are possible, the laboratory has to fully calibrate the device, measuring the various scales of effects through different electric fields, Fujiwara said. |
