Issue



New transistor technology could rearrange semiconductor cleanrooms


11/01/2005







BY HANK HOGAN

AUSTIN, Texas - A recent announcement from SEMATECH (www.sematech.org) about what are known as mid-bandgap transistors may mean looming contamination-control challenges for semiconductor cleanrooms. Mid-bandgap transistors could be part of advanced manufacturing processes within 18 to 24 months.

In September, SEMATECH’s ATDF subsidiary, which develops manufacturing processes, announced it had entered into a deal with Thunderbird Technologies Inc. (www.tbird.com) of Morrisville, N.C., to optimize mid-bandgap transistors that are intended to drop into the existing process and boost device performance with minimal manufacturing impact.

“Our device is designed to make use of the good cleaning technology that’s been developed to date,” says Michael Dennen, executive vice president of Thunderbird. “We haven’t seen a lot of things that would be incompatible with our device.”

Mid-bandgap transistors use a thin layer of metal as a gate electrode, replacing the polysilicon that is the standard electrode today.

Because of differences in the material, the result is a transistor that Thunderbird says has improvements in circuit performance, power dissipation, and manufacturing cost.

Some metals are contaminants and will poison integrated circuits, leading to cross-contamination that can crash yields and bring an entire line down. But that’s not the case here, according to ATDF’s technology officer Shu Ikeda. He notes that mid-bandgap transistors make use of titanium nitride (TiN) or equivalent metals.

“They are popular already and evidence no contamination problems. TiN is a well known material used for the back end of the line,” says Ikeda. All process steps are the same for mid-bandgap transistors as for the standard process, Ikeda adds. To ensure no cross-contamination, fabs might clean wafers with a wet solution, if needed, or the metal gate tools might be dedicated for that purpose.

Clean transport challenges

But as Thunderbird’s Dennen points out, most processes are set up to move wafers directly from gate oxidation to a polysilicon deposition tool. Metal deposition tools are typically located quite a distance away from the oxidation area, in part to minimize cross-contamination and in part because today metallization comes many steps after gate oxidation. So, building a mid-bandgap transistor within current cleanroom layouts requires moving wafers from the oxidation area to a far corner of the fab while the contamination-sensitive gate oxide is exposed.

Successfully building high-performance mid-bandgap devices requires paying particular attention to cleanliness during this transportation.

Minimizing the problem will require rearranging the tools within a fab and possibly modifying automated wafer-handling systems. But that won’t happen until semiconductor manufacturers are convinced that mid-bandgap transistors give the performance punch needed to allow the industry to continue to follow Moore’s law. Mid-bandgap technology has that promise and could allow the use of inexpensive bulk silicon-based technologies for several years, reasons cited by ATDF for pursuing the approach.

What’s more, unlike more exotic technology solutions, mid-bandgap transistors could appear fairly soon. ATDF’s Ikeda says that if current research and development proves their performance to be reliable, mid-bandgap devices could go into production starting middle of next year.


As shown in this simulation, Thunderbird Technologies’ Fermi-FET, a mid-bandgap transistor (top), conducts better than a standard transistor (bottom) under the same conditions. This increased performance could also mean new contamination-control concerns, particularly during transport of wafers. Artwork courtesy of Thunderbird Technologies, Inc.
Click here to enlarge image

null