SI CHINA     SI JAPAN
Login  |  Register          Free Newsletter Subscription
Subscribe
Email
Print
Reprint
Learn RSS

TI Unveils 90 nm Process

Peter Singer, Editor-in-Chief -- Semiconductor International, 3/1/2002

Texas Instruments (TI, Dallas) took the wraps off its next-generation 90 nm process technology, which features transistors with gate lengths as small as 37 nm. The company made the announcement as part of its introduction of new ASIC libraries and design kits based on the 90 nm process. Although the 90 nm process is not expected to be in production until 2004, which is in accordance with the latest revision of the International Technology Roadmap for Semiconductors (ITRS), it is unlikely that changes will be made to the process now that it has been announced.

The company recently produced the first integrated silicon using the process, although it was based on larger gate lengths of 52 nm. The process is expected to go through a prototyping phase early next year, followed by qualification in the second half of 2003, initial production ramp in 4Q03 and high-volume production in early 2004.

"There's a level of confidence that we have this capability, both from a standpoint of process and from a completeness of design tools and models," said Hans Stork, vice president of silicon technology research and development. "People can start their designs with it, and they will get products when they need them."

The 90 nm process includes up to nine layers of copper interconnect integrated with a low-k dielectric, organosilicate glass (OSG). TI decided to use the CVD-deposited Coral material developed by Novellus, which has a dielectric constant of 2.8. "We have certainly evaluated other materials along the way, but the choice to use Coral took place in the middle of last year," Stork said.

 

1. Six-level metal interconnect cross section of TI’s 90 nm process, showing integration of copper, with OSG film dielectric on metal levels 1 through 4, and FSG film on metal levels 5 and 6. (Source: Texas Instruments)

 

2. A close-up of Figure 1, with specific layers defined. (Source: Texas Instruments)

For the gate dielectric materials, TI will stay with a plasma-nitrided oxide (PNO) for the core transistors, which will be scaled to 1.3 nm for the first time. "We've been able to get several learning cycles already on this nitrided oxide," he explained. In terms of the more exotic high-k gate dielectrics, such as hafnium or zirconium silicates or oxides, issues related to process integration, mobility, threshold voltages and reliability have yet to be resolved, he added. "We have some efforts in that area as well and are very well along the learning curve, but we don't expect it to be ready for this generation."

When asked if it might be possible for changes to the process to be made, considering it won't be in production until 2004, Stork said it was unlikely. "It would really have to be a breakthrough in the next six months, and a really significant breakthrough for that to be a possibility."

TI's 90 nm process is being developed for both 200 and 300 mm production. Development work is now underway in the 200 mm Kilby wafer fab, and volume production will be in the 300 mm DMOS6 wafer fab in Dallas, said Peter Rickert, director of process technology development and TI fellow. Based on the company's experience with its sub-0.13 µm process, which is in qualification now, Stork said he did not expect there to be much of a delay between when the process would be available in 300 mm after 200 mm. "There might be a quarter difference but not much more," he said.

 

3. Transistor cross section of TI’s aggressive 37 nm transistor, to be used in microprocessor-class applications. (Source: Texas Instruments)
Previous process generations typically utilize a single pair of CMOS transistors optimized to support all the circuit functions on a chip. However, TI's 90 nm process makes it possible to use a collection of transistors that are "tuned" for different functions on a single chip to meet a variety of performance, density and power consumption requirements. This is done through adjustments to the transistors' gate length, threshold voltage, gate oxide thickness or bias conditions. "For example, in the low leakage library which has a nominal gate length of 60 nm, we have some macros that would take it up to 70 nm to achieve a standby threshold leakage decrease," Rickert explained. The result is that transistors with the highest performance can be used in performance-critical functions such as signal processing, whereas transistors with lower power consumption can be used to support functions with more stringent active and standby power requirements. TI expects this capability to reduce system power by 2-3× in future products.

For additional information on wafer processing, go to www.semiconductor.net/wafer

Email
Print
Reprint
Learn RSS

Talkback

We would love your feedback!

Post a comment

» VIEW ALL TALKBACK THREADS

Related Content

Related Content

 

By This Author

SPONSORED LINKS



 
Advertisement
SPONSORED LINKS

More Content

  • Blogs
  • Podcasts
  • Videos

Blogs

Podcasts

Videos

Advertisements





NEWSLETTERS
Plug in and get the latest SI news, trends and industry updates delivered free, directly to your inbox!

SI NewsBreak and Special Reports (Weekdays)
Wafer Processing Report (Monthly)
Lithography Report (Monthly)
Metrology Report (Monthly)
Clean Processing Report (Monthly)
Packaging Report (Twice Monthly)
©2008 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy
Please visit these other Reed Business sites