|
|
|
'Exotic' Materials Come To Fruition San Francisco -- It took decades for them to come this far. After years of unfulfilled promises or technological hurdles, exotic semiconductor materials like SiGe, SOI, PZT, SBT, BST appear to be establishing a beachhead on the product front. IBM spent three decades and hundreds of millions of dollars developing silicon germanium (SiGe) -- theoretically proposed 40 years ago -- as a high-frequency analog IC substrate. Earlier this year, the PC and services giant made concrete its business plan for SiGe parts. Under study since the 1960s, silicon-on-insulator (SOI) materials, devices and circuit technologies found a home only in radiation-hardened circuits for military and space environments such as satellites, a small niche market. Now, chip makers are conducting research into SOI for use in digital devices. The Semiconductor Industry Association (SIA) has also responded by notching up the presence of silicon-on-insulator in the SIA industry roadmap. Meanwhile, ferroelectric memory technology has caught the eyes of U.S., Japanese and Korean memory makers, with new start-ups joining the fray as well. NEC, Fujitsu, Toshiba, Mitsubishi, Siemens, Hyundai, Matsushita and Motorola all have R&D programs involved in ferroelectric memory. SiGe For initial SiGe product targets, IBM aims to replace gallium arsenide (GaAs), another exotic semiconductor material that has found a niche in the telecommunications IC space (EN, Feb. 9). GaAs parts are seen as too expensive and SiGe has been gaining steam as a low-power, lower-cost successor to the material. Nortel has ordered SiGe ICs from IBM for a wireless telephone application, and these IBM parts will replace GaAs parts in the devices (EN, Nov. 24, 1997). Nortel is said to be working on a SONET equipment application that would utilize them as well. A number of equipment vendors are said to have experimented with silicon germanium and claim to have a production-worthy reactor, according to industry sources, including Applied Materials and Advanced Semiconductor Materials International (ASMI). A Japanese equipment vendor is said to be talking to potential customers as well. Meanwhile, Leybold has a license from IBM for the ultra-high-vacuum chemical vapor deposition (UHV-CVD) epitaxial reactor and a system it has already installed at Big Blue. The two companies have worked on the technology for a number of years. Two Japanese semiconductor companies have approached Leybold with interest in working with the equipment vendor. SOI The silicon-on-insulator (SOI) wafer market took a big step toward entering the fab line when it was revealed earlier this year that Ibis Technology Corp., a longtime SOI supplier, received three purchase orders for its Ibis 1000 oxygen implanters (EN, March 23). IBM would later confirm that it bought two of the implanters from Ibis along with licensing rights to supplement Ibis' SOI wafer production with that of its own. IBM now has three implanters. The purchase order makes a strong case that SOI wafers are nearing commercial production, according to some industry observers. The sales signal that IBM is taking SOI out of the R&D stage, if not putting it into full production. In an investment analysts briefing on Feb. 24, IBM outlined its roadmap to bring gigahertz-level microprocessors to market. In order to make this possible, the company ticked off three key technologies under development that held this promise. Copper wiring, much discussed of late, topped the list. (See related story, this page.) IBM said copper offered 40 percent better conductivity than aluminum. Second, IBM placed SOI, where Big Blue forecast 20 percent performance improvement due to reduced capacitance, a by-product of having a thin oxide layer under the transistors. Third was advanced lithography. Motorola is said to be using the material at some level for wireless communication chips in pilot scale operations. In addition, Hyundai and the Digital Equipment division of Compaq Computer are said to be looking into SOI for device applications. Meanwhile, Canon disclosed plans for sales of SOI wafers to generate some $200 million in annual revenues for the company in four to five years. With Mitsubishi and Shin-Etsu Handotai already behind Ibis and its chief rival, Soitec, Canon's comments make a further case for a ramping up of SOI capacity. PZT, SBT, BST No one questions the advantages of ferroelectric material or its potential usefulness, but integration and manufacturing issues have plagued the materials, which include lead zirconate titanate (PZT), strontium bismuth tantalate (SBT) or barium strontium titanate (BST). The key feature of these materials is that they are non-volatile. Unlike typical volatile memory used in PCs, non-volatile memory retains data even when power is shut off, therefore eliminating the need to be constantly refreshed in order to hold information. Ferroelectric memory has gotten a bad rap -- poor cost versus yield in manufacturing. However, Ramtron International Corp.'s Advanced Materials group disclosed on April 15 that it has developed a new high-endurance ferroelectric capacitor structure based on an optimized PZT materials composition, possibly eliminating this problem. Things are heating up elsewhere in ferroelectric memory. Last year, ATMI, a semiconductor materials company, formed a joint development program with Texas Instruments to develop high-density, non-volatile ferroelectric memory technology for smartcard ICs (EN, Nov. 17, 1997). ATMI was the prime contractor under a Defense Advanced Research Projects Agency-sponsored consortium, with IBM, Micron Technology, Texas Instruments and Varian Associates. The company also established a new division, Emosyn, as a fabless semiconductor start-up to compete against the likes of Motorola, Siemens, STMicroelectronics and Hitachi in the pursuit of the smartcard IC market. Meanwhile, Motorola announced its own joint venture for FeRAM with Japan's Matsushita. Also, NEC expects to bring FeRAM products to market after 2000, according to a spokesperson. With a low-cost, high-yield form of ferroelectric memory, the RAM, ROM and EEPROM on a smartcard could all be replaced, according to Ramtron, a ferroelectric memory pioneer. Memory takes up an estimated 80 percent of a smartcard. |