Marketing novel technology: An historical lesson
10/01/1997
David K. Lam
Semiconductor manufacturing continues to rely on new technologies to produce ever more challenging device structures. The next decade will see many new materials and processes reaching high-volume production. With different equipment and materials companies proposing competing solutions, often with conflicting claims, it is difficult for end-users to decide which new technology will truly be needed and which will end up as historical footnotes. New-technology vendors must help customers overcome the confusion and uncertainty that invariably accompany new process releases.
Based on my personal experience of introducing plasma etching to volume IC production at the start of the 1980s, some general observations follow.
Case history
In January 1974, on the very first day of my first job, I walked into the Semiconductor Research and Development Laboratory of Texas Instruments (TI) in Dallas to find it abuzz with excitement. The entire Semiconductor Group was bubbling with enthusiasm over the installation of the new 3-in. production line. However, the transition from 2-in. wafers resulted in serious yield problems, and teams were dispatched to examine each section of the production line.
Young and green, never having worked in a semiconductor fab before, I anxiously took pencil and notebook in hand and approached every operator to ask what he or she was doing. It was humbling to realize how little I'd known, but it gave me an early understanding of what was needed in the production end of the business.
There was phenomenal growth in semiconductor manufacturing in the early 1970s, but it was generally recognized that in order to make smaller and more powerful ICs, the industry would need advanced photolithography and dry etching.
At that time, etching of silicon wafers was done in a sink filled with liquid chemicals — wet etching. Wet etching expenses included large quantities of liquid chemicals and rinse water, and disposal of the spent chemicals. The main limitation of wet etching is line "undercutting," which prevents the reliable production of linewidths below 4 to 5 µm. With wet etching, the industry would not be able to advance to 64K DRAM complexity and feature sizes with acceptable yield.
Plasma etching, on the other hand, uses low pressures and high-frequency electromagnetic fields to partially ionize small amounts of gases into electrons and ions. The electrons act as a catalyst to produce chemically active molecules and molecular fragments, which in turn etch away the unprotected material on top of the wafer. This gas-solid chemical reaction could produce fine lines of 1 µm or smaller, with almost acceptable speed, thereby opening up the possibility of a whole new era.
By the late 1970s, scientists and engineers in research and development laboratories around the world had demonstrated the capability of plasma etching. It promised to push the technology envelope of IC processing to the next higher level of integration. The "unbelievable" 64K DRAM was going to be a reality. Yet, plasma etching was not exactly an easy sell.
Evangelical challenges
1. Customers lack sufficient understanding to make an informed decision. First of all, plasma etching was totally different from wet etching. None of the experience gained by engineers and operators in wet etching was readily translated into a quick understanding of the new method. From this standpoint, plasma etching was a revolutionary technology based on a very complex chemical-physical phenomenon and only a handful of academic and industrial labs were conducting any research in the field. Customers needed an education in the basic science of the new technology, and in its true capabilities and limitations.
Back in 1980, the most common wafer sizes in production were 3 and 4 in. So, when Lam Research introduced its single-wafer etcher, it had to fight the perception of low throughput. A marketing effort was then put together to engage the industry into using not wafers/hour but in.2/hour as the way to measure throughput, since single-wafer was actually better at 6-in. wafers and beyond. At the same time, Lam Research announced that its first product would be 6-in. compatible, further focusing the attention on 6-inch wafers where single-wafer had an advantage. Customers needed to understand that the new technology required new metrics for comparison, with new success criteria.
2. Technology complexity and early hardware problems delay customer confidence. Secondly, a plasma etcher was much more expensive and complex to use than a wet-etch station. Early-generation plasma etchers had equipment and process problems, and often broke down. Keeping a plasma etcher humming requires multidisciplinary expertise in vacuum engineering, RF engineering, electrical engineering, mechanical engineering, and process chemistry.
The main cause of the lack of customer confidence in plasma etching was process inconsistency due to the inherent complexity of plasma chemistry. Commercial single-wafer etching was pioneered by Tegal in the 1970s, but mere single-wafer processing didn't achieve reproducibility. The best, if not the only, way was through automation, which may not have provided exactly what was wanted but made the etch results predictable.
Automation came in different shapes and forms, including mechanization and computer-control. To achieve etch consistency, Lam Research implemented the following innovations on its first product in 1980 — the AutoEtch 480: single-wafer etching with double loadlocks, robotic arms, microprocessor system control, and EPROM recipe storage. Building customer confidence in the new technology was crucial to eventual acceptance.
3. Competing specsmanship and marketing pitches incite further confusion. Thirdly, there were more than 10 serious plasma etcher vendors in those days. Fab managers were often confused about the relative merits and claims of each, and few people had real production experience in plasma etching at that time. To make things worse, there were constant debates over planar etching vs. reactive ion etching (RIE), batch vs. single-wafer, and loadlocked vs. nonloadlocked.
My early fab-floor experiences allowed me to communicate clearly with end-users. By emphasizing that the new etcher would meet the simultaneous needs of process operators, technicians, engineers, and managers, the confusion of competing features was eliminated.
Epilogue
About 10 years elapsed from the time when plasma etching showed promise to its wide acceptance in production. But, if recent technology adaptation is any guide, the next major innovation in the industry will take a much shorter time. Technology innovations have turned dreams into reality, accelerated industry growth, and even spawned whole new industries.
A few months ago, I met with TI's CEO, Tom Engibous, and told him how grateful I was to TI for making Lam Research able to compete at the forefront of the market. He was visibly puzzled until I told him how that first yield problem had made me much more aware of the challenges production managers face every day. I told him that the experience of what the customer truly needed eventually enabled Lam Research to introduce successfully an important new technology to the industry.
 |
David K. Lam, Sc.D., is president and CEO of Caliper Technologies Corp., 1275 California Ave., Palo Alto, CA 94304; ph 650/842-0700, fax 650/842-1970, www.calipertech.com.