Trikon/mike1st/mark2nd By BEN SULLIVAN Staff Reporter Cramming more transistors onto ever-smaller silicon real estate is the overarching goal of computer chip makers. Now researchers at Chatsworth-based Trikon Technologies Inc., a supplier of chip-making machines, say they can help manufacturers do just that and dramatically improve chip performance in the process. Their method is called Low-k Flowfill, and if testing underway by the likes of IBM, Texas Instruments and Daewoo pans out, analysts say, it will be one of the biggest advances in chip making this decade. “It would absolutely be a significant breakthrough, beyond a doubt,” said Ron Bowman, an engineer at Integrated Circuit Engineering, a computer consultancy in Scottsdale, Ariz. “For the Intels of the world, it could mean instead of making 200 MHz Pentium Pros, they could maybe make 400 MHz chips, or higher.” It would also be a chance for 650-employee Trikon to go from also-ran to heavy-hitter in the chip equipment market. The new method is not homegrown. Last fall Trikon, then known as Plasma & Materials Technologies Inc., bought the British firm Electrotech to beef up its own global marketing. Like Trikon, Electrotech made the machines used to make computer chips. Electrotech brought with it a new way to lay down the crucial insulation layers on a computer chip using a liquid instead of the currently popular gas-deposition method. Insulation protects the maze of wispy leads that connect a chip’s transistors and make up a computer’s brain. Trikon officials knew Electrotech had the liquid “Flowfill” technology and saw it as an attractive supplement to its own gas insulation method. What they didn’t know was that the Electrotech process produced an insulator with a “dielectric constant” far below current standards. And that’s a very good thing. Trikon didn’t learn of that aspect of the technology until after the purchase of Electrotech. Had the purchase fallen through, Trikon spokesman Jeffrey Reynolds said, the British company didn’t want to have tipped its hand to a competitor. A dielectric constant is the measure of a material’s ability to store electricity. The perfect insulator a vacuum has a dielectric constant, or “k,” of 1. Lesser insulators have higher k values. Current insulators used on computer chips have k’s of between 3.5 and 4. Electrotech’s method produced an insulator with a k of 2. The upshot is that signals speeding down a chip’s hair-like leads do so with less electric drag. That means smaller bursts of electricity are needed to power the chip and, as a result, transistors and their leads can be placed in closer proximity to each other. Avoiding “cross talk” in which signals traveling on parallel leads interfere with each other or even jump tracks is considered a major hurdle to putting more transistors on a chip, and lower energy levels lead to less cross talk. “All electronics have this parasitic effect,” Bowman said. If the Flowfill process proves effective in a factory setting, “It’s going to be effective on everybody’s chips.” The new method has drawn the attention of nearly every major chip maker. Trikon has a backlog of orders from companies that include IBM, Texas Instruments, Daewoo and even the venerable Intel Corp. to produce sample chips using the new technology. “Basically they want to know if this is for real,” said Reynolds. “They want to know if the system is reliable and is the process repeatable over a large number of wafers.” Trikon has shifted the bulk of its corporate resources to research, development and marketing of the new method, according to Trikon CEO Greg Campbell, as the market for chip insulation is estimated to reach $3 billion a year by 2000. The company expects to spend $10 million this year developing and promoting the Low-k Flowfill technology. Trikon has already sold one of its Low-k Flowfill machines to a U.S. chip maker for just over $2 million, Campbell said, and he expects further orders to come in as companies finish their analysis of the Flowfill demos. That will help Trikon’s bottom line: Though its revenues have grown steadily, nearing $40 million in 1996, the company has lost money each of the last five years. Competitors Novellus Systems Inc. and Applied Materials Inc. dominate the chip equipment market, with revenues of $460 million and $4 billion, respectively, in 1996. Analyst Daniel Klesken, at Robertson Stephens & Co., says Trikon stands to gain from a bevy new plants being built by chip makers over the next two years. New factories are more likely to incorporate pioneering technology than plants already tooled with older machines, he said. “It appears they have an important new capability that could propel them into some nice markets,” Klesken said. The risk, he said, is that another company will appear with an equally effective insulation method, at which point it would become a marketing slug-fest for acceptance of competing technologies. A prime candidate is Dow Chemical Co. of Michigan, which recently announced a method that spins a polymer insulation coating onto a chip with a dielectric constant of 2.65. Dow would not comment on Trikon’s technology other than to say it expects its own method will eventually compete with Trikon. CEO Campbell is unfazed. “Flowfill is the hottest thing going,” he said. “I just don’t see how (Dow) will get down to a k of 2.”
