This has not been posted...
Intel money moves Micron even faster
By Jack Robertson
BOISE, Idaho -- Steve Appleton got major backing in his strategy to take Micron Technology Inc. to the top of the global DRAM market. Intel Corp.'s half-billion dollar investment in the Idaho DRAM producer will give it the cash to continue what the Micron CEO called "the most accelerated production technology of any DRAM producer."
Intel said it was investing in Micron to ensure an adequate supply of next generation Direct RDRAMs, which the chip giant has specified for its upcoming 500-MHz to 1-GHz processor generations. But the deal could also solve some major problems that have cropped up with Direct RDRAM parts.
Micron, already working hard to move out in front of the DRAM industry by a year or more with its die-shrink technology (see story in the Oct. 15 publication), has now targeted the end of next year to start early production of 0.15-micron DRAMs. That would put it neck-and-neck with Intel on the Semiconductor Industry Association's national technology roadmap timetable.
"As far as Micron is concerned, we have brought DRAMs back to parity with logic devices on the SIA Roadmap," Appleton claimed in an interview in his office here with SBN. The latest roadmap revision (in late 1997) put logic devices one generation ahead of DRAMs on the technology curve. Appleton felt that assessment likely was still true for other DRAM firms which he believed trail Micron by at least one generation in feature size.
What was most surprising in Micron's technology blitz to 0.15-micron processing was its decision to use the company's existing 248-nm-wavelength laser steppers for production. Most chip technologists until now have felt that the more costly and production-untested 193-nm argon fluoride laser lithography systems would be required for 0.15-micron line widths.
But Micron already has proven that current 248-nm deep-UV steppers can pattern 0.15-micron features sizes in production by using phase shift masks, according to Mark Durcan, vice president of process development. "This gives us a very quick jump on 0.15-micron production, as well as providing a big equipment cost savings by sparing us the high investment expense of 193-nm systems," he pointed out.
Another big advantage that Micron has over the rest of the DRAM industry, according to Appleton, was that it was one of the few DRAM producers -- perhaps the only one -- already using chemical mechanical polishing (CMP). This planar surface polishing, which is widely used now by logic chip makers for higher productivity and chip performance, means that all of his company's memory chips, including Direct RDRAM, can be produced on smaller die.
"Our big investment in CMP is now paying off," declared Appleton, while his cash-strapped DRAM rivals may be hard-pressed to upgrade to CMP anytime soon.
A major goal of Intel's investment, the Micron CEO believed, was to tap Micron's expertise in reducing DRAM cost and die size. Both have been identified by prospective OEM customers as major problems now for Direct RDRAM.
Micron will produce an initial Direct RDRAM chip that will be much smaller than those turned out by rival memory firms, predicted the company's Durcan. But until the final design is frozen, he wasn't sure just what that die size would be. Initial Direct RDRAM parts will be produced on Micron's emerging 0.18-micron production process, which Appleton believed was one shrink generation ahead of his competitors.
Micron also is tackling what it called another Direct RDRAM obstacle: the very high heat dissipation caused by the high-density, high-speed memory. "It's a light bulb," summed up Appleton. Whichever chip maker masters that heat problem, he predicted, will quickly become the market champion in the next-generation device.
Although Intel's big investment has turned Micron into a key Direct RDRAM player, the Idaho memory maker still is pursuing other wideband memory devices such as SLDRAM and double-data-rate SDRAMs.
Ironically, news of Intels investment and its link to Direct RDRAM suddenly has spurred major interest in Micron's DDR-SDRAM development. "OEMs came out of the woodwork the day after the Intel announcement, wanting to be sure Micron was continuing its DDR-SDRAM program," remarked Michael Sadler, vice president of sales.
DDR-SDRAMs should find a home in servers and high-performance workstations, he predicted, while Direct RDRAM, powered by Intel's strong backing, will move into desktop PCs.
Late October was a busy time for Appleton. Besides working on the Intel deal, the litigious CEO decided to fire a shot across the bow of the Taiwanese DRAM industry. A dumping suit was filed against all Taiwan DRAM makers because "they continued to invest irrationally in building added capacity in the face of a three-year DRAM market downturn," he said.
Appleton conceded that all global DRAM companies, including Micron, are now selling chips at below cost -- technically, one definition of dumping under U.S. law. However, he said that Micron singled out the Taiwanese vendors because they alone embarked on a major capacity buildup that threatened to make the market glut even worse.
South Korean DRAM rivals were forced by the Asian financial meltdown largely to curtail their fab expansion, although the Micron CEO charged that Samsung was now adding capacity at its Austin plant. By contrast, he said, the Japanese DRAM suppliers "are acting rationally and cutting back semiconductor capital investments."
The Micron dumping suit covered all DRAMs produced in Taiwan, even those sold under foundry agreements to chip makers in other countries, as well as Taiwanese DRAMs included in memory modules produced anywhere in the world. That would hit Toshiba, Mitsubishi, and Fujistu hard, since they use Taiwanese foundries to produce up to 50% of their DRAMs. Alliance Semiconductor Corp. also was cited because the U.S. supplier has its DRAMs produced there.
Appleton also provided more details of Micron's strategy for upgrading the three fab complexes acquired from Texas Instruments Inc. They will all be equipped with a 0.21-micron processing capability in 9 to 15 months, he said, using the proven technology from the company's Boise fabs. "Once that has been achieved," he said, "we will move the acquired fabs to 0.18-micron capability as quickly as possible."
Micron will be spared much of the expense of upgrading its joint venture fabs in Japan and Singapore because its partners "have their own financing syndicates that will fund the transition investment at these fabs." The Boise company also will have access to Italian government financing for its Avezzano fab, but details of this arrangement are still being worked out.
"It takes time and we haven't included or excluded the Italian government assistance in our capital spending plans," Appleton said. Micron is replacing the entire 6-inch wafer line at Avezzano with its deep-UV 0.18-micron 8-inch wafer process.
As far as other new technology is concerned, Appleton reported that Micron is now in early development of lithography systems built around 193-nm argon fluoride lasers. But it has no definite plans now on when it will start using these new-generation tools. It is conducting this R&D with the IMEC research center in Belgium, the same outfit that's working on this generation system for ASM Lithography, Micron's stepper vendor.
Appleton said that Micron still is looking at 300-mm wafers, but will not make any moves to the next generation wafer "until we are convinced it is cost-effective." He believed that the time eventually will come when a 300-mm fab "makes sense, but that isn't any time soon."
semibiznews.com
|
