Researchers lay plans to build 500-GHz transistors
By Paul Kallender , EE Times
Apr 19, 2001 (1:07 PM)
URL: siliconstrategies.com
TOKYO — Japan's Communications Research Laboratory (CRL), working with Fujitsu Laboratories and Osaka University's Graduate School of Engineering Science, has developed what is claimed to be the highest-frequency high-electron-mobility transistor (HEMT) achieved to date. The work is part of a national effort to develop millimeter communications technologies for commercial and network communications applications.
The HEMT, operating at 386 to 398 GHz, opens doors to developing, by mid-decade, such products as automotive collision-avoidance radars at 76- to 77-GHz frequencies, ITS trunk communications, and 100-Mbit/second home LANs at 60 GHz, said Fujitsu Labs Fellow Takashi Mimura. Mimura invented the world's first HEMT, now a standard communications device in mobile phones, back in 1979.
"My big interest is exploring the GHz performance limits. We are gradually approaching those limits, but we have projections of hitting 500 GHz," Mimura told EE Times.
To do that, the group — which is backed by Japan's Ministry of Public Management, Home Affairs, Posts and Telecommunications — will have to shorten the HEMT's gate from its present 25-nm width down to 10 nm, he said.
Narrowing the device's gate width is one of two approaches CRL used to reach the present record-breaking transistor, said Toshiaki Matsui, group leader of the lab's Communication Device Group.
Very-high-frequency HEMTs face two huge performance barriers, Matsui said: electron speed and the distance the electrons must travel. A year ago, the group hit a record 362 GHz by adopting indium gallium arsenide, with a 53 percent indium content, for the free-electron channel layer structure. To overcome the second barrier, the group developed a simple liftoff technology using electron-beam lithography to form 50-nm gates.
Several groups have pushed HEMTs into the mid-300-GHz performance range so far. In 1992, Hughes Electronics hit 340 GHz with a 50-nm gate device. In 1998, Nippon Telegraph and Telecommunications reached 350 GHz using a 30-nm gate.
To break its own record, the CRL-led group optimized the e-beam lithography to form a stable 25-nm gate. It also boosted electron mobility 25 percent in the HEMT's channel layer by increasing the channel's indium content to 70 percent, said Keisuke Shinohara, a CRL researcher with Communications Device Group.
"How far are we ahead? Reducing the gate size to 25 nm makes it the shortest of any transistor metric gate [using a mix of titanium, platinum and gold] in the world," Shinohara said.
Japan has long held ambitions to enter and exploit the virgin territories of higher-frequency communications bands. The country has made it national policy to develop satellite communications on S-band, for example.
Above 20 GHz, where millimeter waves begin and vistas of frequency lay untapped, the Ministry of Public Management, Home Affairs, Posts and Telecommunications has extensive plans to develop a series of national communications infrastructures. The ministry wants to develop a national 4G communications infrastructure linking a national ITS network, together with an office-home national LAN system, to a multimedia mobile-access communications network.
All may link to the ministry's proposed Skynet system, a constellation of up to 200 giant dirigibles anchored at 20 km dotting the sky above Japan that will serve as cheap, extremely low-orbit stationary communications satellites, said Matsui. But those dirigibles remain pies in the sky, with the ministry's ambitious plans temporarily starved of development budget because of the Japanese government's current financial situation.
At the moment, millimeter-wave communications applications are usually limited to military and scientific uses. Most famously, perhaps, the terrain radar tucked into the snub of Tomahawk missiles uses the 94-GHz band. In a less exotic application, millimeter spectrometers measure ozone levels and aid in space research. But attenuation through oxygen severely limits ranges at 60 GHz, as do buildings and natural obstacles. And high-powered, narrow beams are unsuitable for low-cost omnidirectional planar antennas.
Ultrahigh-frequency HEMTs will deliver at least double the gain of present-generation gallium arsenide devices, said Shinohara. A 400-GHz-capable HEMT at 60 GHz achieves about double the gain of the most powerful 150-GHz gallium arsenide models. Redeveloped versions could cut four to eight amplification stages, shrinking chip sizes and costs dramatically, he said.
Fujitsu Ltd. will have to make some strategic decisions about when and how it will develop its own millimeter-wave products, said Mimura. Cash is one critical issue; developing integration strategies to make the HEMTs sufficiently low-cost and producible is another. But if the government plows ahead, he suggested, so will industry.
The integration strategy will probably involve flip-chipping monolithic, microwave integrated circuits (MMICs) with hybrid ICs to combine MMICs' cost advantages with the hybrid approach's low-cost aluminum substrate, said Mimura. "We recently got a flip-chip bonder, so maybe Fujitsu will make them, but it will cost money and take time.
"So many applications and system ideas are being proposed, but none of them are fixed," Mimura added. "What frequency do we use; what demodulation do we use? And so many companies [that are interested] can't yet focus on an absolute technology and equipment development decision. Also, the Japanese economy isn't so good, and we don't have the spare capacity yet."
CRL is confident that it is overcoming front-end issues, namely external, heavy, expensive high-gain antennas, said Masahiro Kiyokawa, a senior researcher at the lab who is responsible for circuit integration. Work on peripheral technologies, such as antennas, predates CRL's four-year HEMT program, he said. While there are international efforts to construct planar antennas suitable for millimeter-wave reception, those efforts still face an uphill task: The higher the frequency, the higher the transmission-line leakage becomes.
"Even if you have an excellent low-noise amp, the receiving antenna comes first. If there are leaks there, the total loss before the signal gets detected begins to impact the degradation. But Matsui and myself are not confined to planar structure," he said.
Instead, CRL is making breakthroughs with optical technologies to make Gaussian beam antennas, he said.
"Matsui and I have made a very efficient antenna, but we haven't done any integration yet. That's our next goal," Kiyokawa said. |
