N.Y. Start-Up May Give Japan a Race in Chips : Technology: Hampshire Instruments appears to be out front in commercializing the new chip-making process of X-ray lithography.

The complicated and volatile business of making the machines that make computer chips has become a critical battleground in the United States’ ongoing technological competition with Japan--and the Japanese are winning.

Over the past decade, scores of U.S. equipment and materials firms have been bought by Japanese companies or lost their edge in the market. That has led many in the industry to call for restrictions on Japanese purchases of high-tech companies and government assistance for chip-technology development. This battle cry recently became even more shrill with accusations that Japanese chip-equipment firms were withholding their best technology from American companies.

The frighteningly fast technological changes that make the chip-equipment business so treacherous also create opportunities, however, and an American start-up firm believes that it is poised to leap-frog the Japanese competition in the most difficult and important corner of the industry.

Exploiting technology that was first developed in the American nuclear weapons programs, tiny Hampshire Instruments of Rochester, N.Y., appears to be out front in commercializing a controversial chip-making technique known as X-ray lithography.

It’s not clear if or when the approach being pursued by Hampshire will win out. Some believe that continued improvements in the existing optical lithography techniques will render X-ray a non-issue for at least the next decade. Such electronics powerhouses as International Business Machines and the major Japanese chip companies are pursuing X-ray technologies of their own--just in case it does become important.

But Hampshire has already sold a handful of its multimillion-dollar systems, and if the product catches fire, the company stands a chance of joining market leaders Nikon and Canon as a major player in the lithography business.

Japan supposedly has all the advantages in the industry--cheap capital, close links between manufacturers and suppliers and a willingness to invest for the long term. Even so, it seems that there’s still room for clever entrepreneurs with innovative approaches to financing, the know-how to tap technology resources such as the Energy Department’s national weapons development laboratories and plenty of old-fashioned determination.

“In the semiconductor capital-equipment business, you have to be either very committed to success or crazy,” says Moshe J. Lubin, the engaging 52-year-old physics professor who founded Hampshire with two partners in 1984. “To develop a significant new tool takes six or seven or eight years. And it’s not enough to have a technological edge. You need to understand (the context), how technologies displace other technologies.”

The dynamics of technical change is one of Lubin’s favorite subjects. His speech is peppered with historical references to “displacement”--the diesel engine pushing out the steam engine, jet airplanes overtaking propeller-driven ones. He especially likes the aircraft analogy because it provides an example of a technology that America pioneered, lost to overseas competitors, then regained.

In chip-making technology, as in other areas of electronics, the United States has fallen from uncontested leader to faltering runner-up in less than a decade. In 1980, the nine largest chip-equipment companies in the world were American. Today, four of the top five are Japanese.

In the all-important lithography segment, two Japanese firms--Nikon and Canon--hold an estimated 70% of the market. That’s in a technology that was developed entirely in the United States, primarily at Bell Laboratories and at a once-high-flying company called GCA.

Lithography machines, known in industry parlance as “steppers,” carry out the central task in the chip-making process: drawing electrical circuits onto slabs of silicon called wafers.

As its name suggests, lithography is based on photographic processes. A wafer is first covered with a thin film of silicon dioxide and a layer of chemical photoresist. Then, light is shined through a stencil-like glass “mask” containing a layer of circuit patterns, exposing the corresponding areas of the resist-covered wafer.

The exposed resist is developed, which softens it and allows it to be etched away by acids--along with the silicon dioxide below it. The unexposed resist is then removed, leaving the circuit layer engraved into the silicon dioxide and the wafer. The process is repeated for the next layer.

For ever-more-complex chips, ever-finer circuit lines need to be etched into the silicon. Currently, most advanced chips are being made with 0.8 micron line widths, but the next generation of devices is expected to require line widths of less than 0.5 microns. (One micron is equal to 1-millionth of a meter.)

The problem with optical lithography is that the length of the light waves begins to interfere with the process at such narrow line widths. X-rays have much shorter wavelengths, and X-ray steppers thus have the potential to allow line widths as narrow as 0.1 micron or below. They also have better variation in depth of focus, enabling them to speed the process by etching multiple layers at the same time.

But X-rays are hard to handle and require masks that are the actual size of the circuit. (Optical steppers use masks five times or even 10 times larger, with lenses projecting a reduced image onto the wafer). Moreover, continued advances in optical lithography, revolving around new types of light sources, masks and better resist materials, promise to expand the capabilities of this method far beyond what was once believed possible.

“X-ray has missed its window,” says Martin Lepselter, the former director of Bell Laboratories’ pioneering efforts in chip-building technology and now head of a start-up company that has developed a chip-making machine using a technology known as electron beam. “You might sell a few of them, but there isn’t a single DRAM (dynamic random access memory chip) manufacturer who is planning to do anything but optical lithography in this decade.”

Canon and Nikon confirmed that they expect their optical lithography systems to be the dominant technologies for the 1990s. Critics also note that X-ray lithography, which has been pursued in various forms for many years, sometimes seems to fit the well-worn high-tech joke: “It’s the technology of the future, and always will be.”

Lubin, though, maintains that the skeptics are wrong, and he says Hampshire has 10 commercial orders--at $4.5-million apiece--to prove it. “Optical is at the end of its life, though it could be a long end,” Lubin acknowledged. “We now have to show in earnest that (X-ray) is lower cost in a real manufacturing environment. Then the displacement will begin in earnest.”

Hampshire is banking on a technological edge in lasers--which it uses to create the X-rays--to propel it to the top of this emerging market. While IBM and other big companies are pursuing systems that use expensive atom smashers known as cyclotrons to produce X-rays, Hampshire is using laser technology developed under government contracts at the University of Rochester, where Lubin and Hampshire’s two co-founders once worked.

In fact, Hampshire is doing what many consider vital to the complex problem of maintaining America’s high-tech might: commercializing top-notch technology that has been developed at national laboratories under Defense Department and Energy Department contracts. The company has joint development programs with Lawrence Livermore Laboratory and Sandia National Laboratory and has contracts from the Defense Advanced Projects Research Agency, the National Institute of Standards and the government-subsidized chip research consortium Sematech.

Hampshire has also exploited another form of government assistance: low-interest loans from the state of New York, which was eager to support a promising high-tech start-up venture. When Hampshire gets an order, the state will lend it the money needed to fill the order at interest rates 4 percentage points below commercial rates.

That makes Hampshire’s cost of capital--a key factor in international competitiveness--comparable to that of Japanese companies and far below what prevails in the United States, Lubin said. With the government contracts, state assistance, venture capital funding and early development contracts with American Telephone & Telegraph, McDonnell Douglas and Digital Equipment, Hampshire has been able to invest more than $70 million in bringing its system to market.

The company announced last week that it had completed installation and testing of its first commercial system at AT&T; Bell Laboratories. Cypress Semiconductor, a small but highly efficient chip manufacturer, has agreed to buy a machine to see how the technology stacks up against new optical lithography techniques.

The confident Lubin says Hampshire hopes to go public next year.

Proud though he is of the firm’s entrepreneurial verve, he recognizes that Hampshire couldn’t have done what it’s done without government help.

“The government’s role is to provide industry with technological choices,” he says. It’s been done before, in agriculture and airplanes and even nuclear power, Lubin adds, and there’s no reason it can’t be done in chip-making equipment.