RequestLink
MICRO
Advertiser and
Product
Information

Buyer's Guide
Buyers Guide

tom
Chip Shots blog

Greatest Hits of 2005
Greatest Hits of 2005

Featured Series
Featured Series


Web Sightings

Media Kit

Comments? Suggestions? Send us your feedback.

 

MicroMagazine.com

EDITOR'S PAGE

A powerful one-two punch

Few international conferences in the chipmaking community pack as powerful a one-two informational punch of thought-provoking keynote speeches and in-depth technical presentations as the IEEE International Symposium on Semiconductor Manufacturing. ISSM alternates each year between Japan and Northern California, with the 2003 edition convening in downtown San Jose. This year's event statistics were impressive: nine keynotes and 126 oral and poster papers from 119 different organizations representing 17 different countries. In this era of just-in-time delivery, it was not surprising to hear program committee chair Bruce Sohn of Intel note that 54% of the 286 abstracts submitted were received within 24 hours of the deadline.

The keynoters' affiliations covered many sectors and regions of the semiconductor industry. Speakers represented fabless (Xilinx/U.S.), foundry (SMIC/China), capital equipment manufacturing (Applied Materials/U.S.), chip R&D and production (STMicro/Europe and IBM/U.S.), and automotive electronics (Toyota/Japan). Engineers and scientists from many of the speakers' companies (with the glaring exception of IBM) were in the technical presentation mix as well.

Xilinx chief Wim Roelandts championed the fabless model, positing that the portion of global IC revenues garnered by companies who outsource their manufacturing will rise from 17% to 35% in a few years. A big fan of the economics of 300-mm processing, Roelandts pointed out that they get five times more 90-nm die per wafer at 1.8 times the cost. He also said that Xilinx, with its foundry partner UMC, may fabricate a higher percentage of chips on 300-mm wafers than any other semiconductor player. His claim was borne out a few weeks later during the company's quarterly earnings announcement when he stated that Xilinx makes more than 50% of its field-programmable gate arrays (FPGAs) on 300-mm silicon.

Later that morning, Xiao-Yu Li, a senior manager of product engineering at Roelandts's company, discussed how a UMC-Xilinx team developed a poly gate CD characterization method that takes advantage of FPGA programmability and uses different test-structure configurations as yield and performance indicators. So far, Li said, the method has worked quite well during development and production to help increase product speed and improve yields. What struck me about this presentation was how end-product devices were employed to improve their own creation process. It also drove home that, contrary to some people's beliefs, fabless companies are intimately involved with the making of their own chips, even if they don't have their own production lines.

SMIC headman Richard Chang talked about the burgeoning Chinese semiconductor market as well as his own upstart company, confronting what he called myths with certain realities. Dispelling the common notion that it is difficult to do business in China because of import and export restrictions, he noted the recent approval of a special comprehensive license (SCL) from the U.S. government, which is contingent on his company's pledge not to make any military products. The SCL has greatly reduced the export licensing turnaround times for SMIC's suppliers—from 6 months to a few weeks—allowing the Chinese foundry to increase U.S. chipmaking-tool imports from 40% to 70% over the last few months. Speaking of improved supply lines, Chang stated that Fab 4, SMIC's 300-mm factory under construction in Beijing, will be ready for equipment move-in by the first quarter of 2004, with pilot runs taking place by the second quarter.

One reason SMIC has emerged as the leading Chinese chipmaker is sound manufacturing practices, some of which were presented at ISSM. Ivan Goh, a technical manager and principal engineer at the company's Shanghai facilities, presented an integrated yield-enhancement approach that has helped accelerate yield learning for advanced BEOL processes. The strategy's framework comprises three
segments: the use of products with embedded SRAM and other test keys; debugging, fault detection, and other analytical methods; and a synthesis of the first two approaches into a systematic yield
analysis flow. By unifying and integrating the various techniques, Goh and his colleagues have helped improve SMIC's ability to ramp quickly to profitable yields and have established the company as a world-class manufacturer.

Although downturnitis has put a damper on turnouts, events like ISSM remind us that there's no substitute for attending a well-run, comprehensive technical conference in the flesh.

Tom Cheyney
Editor

tom.cheyney@cancom.com


MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit

Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com.

© 2007 Tom Cheyney
All rights reserved.