VIEWPOINT

Tool suppliers need IC makers to help make the 'Year 2000 Problem' a top priority

For a business to continue operating profitably into the next millennium, electronic commerce transactions must successfully handle the transition from December 31, 1999, to January 1, 2000. If a business sells computer hardware, computer software, or any type of date-sensitive product, it is obligated to make certain that such products recognize the "00" at the end of the six-digit date as 2000 and not 1900.

Understanding the Year 2000 Problem is fairly easy. But fixing it will be hard for everyone, especially for companies that supply equipment to the semiconductor industry.

This is the case because a number of other issues have been put on the plates of equipment suppliers by their customers. As a result, the Year 2000 Problem has not received the attention it needs. Suppliers understand the situation and want to apply resources to solving this problem, but the demands already placed on their scarce resources have prevented them from doing the work at a rate that will avoid a last-minute panic. This rush could have been prevented, and it can possibly be mitigated even now, but time is running out.

Suppliers need guidance and relief that will help them make the Year 2000 Problem a higher priority than it is at present. That help has yet to come, and I will explain why shortly. First, though, we need to examine the demands confronting the supplier community in order to understand why guidance and relief are needed.

These demands present suppliers with five major challenges. The first challenge is important only when combined with the others. This challenge is the ebb and flow of the marketplace. Right now, the semiconductor market is soft, and the supplier community is dealing with all of the obvious difficulties arising from this problem, such as poor book-to-bill ratios and order cutbacks. But this chapter of our industry's development contains four added challenges that, taken together, stretch resources beyond the breaking point for all but the very largest equipment makers.

The second challenge is tied to the first one. Money is tight enough and lithography technology is advanced enough that IC manufacturers are determined—and rightly so—to milk 200-mm wafer fabs for their very last technology node. As a result, device makers are asking suppliers to develop 200-mm equipment with processing capabilities that the SIA, in its 1997 National Technology Roadmap for Semiconductors (NTRS), said would be used only for wafers measuring 300 mm and larger. Suppliers have thus been forced to establish additional R&D programs that retrofit 300-mm technology into 200-mm equipment.

The third challenge is the cost of R&D. As fabs have grown in their operational sophistication and processing capabilities, so have the costs of research and development. It was in part these high costs that precipitated the formation of consortia such as Sematech. In these consortia, collaboration on precompetitive technologies led to standardization in many process areas, which has helped both the supplier and customer save R&D dollars. Today, however, the IC manufacturers through these same consortia are asking the supplier community to not only fund its own research but also donate very costly, highly advanced equipment for performance testing. These high costs have caused many suppliers to be very cautious about participating in consortia development work. The downside of not participating is that semiconduc-tor manufacturers often brand such suppliers as nonresponsive. Hence, some suppliers take part in consortia R&D as a form of self-defense, even though participating places them in a financial bind.

The fourth challenge is that pesky—and apparently risky—300-mm conversion. The conversion's development costs are sufficiently high to absorb any company's entire R&D budget. The reason for the high costs is not that the industry is incapable of figuring out how to process the larger substrates. The problem is developing technology that does the job efficiently enough to make the transition affordable.

Finally, as if those four challenges aren't enough to make even the flintiest CEO flinch, the fifth challenge comes stomping onto our front porch and pounding on our door. This is a challenge the likes of which none of us has seen before: our utter dependence on electronic commerce and manufacturing systems built on a computerized foundation that will misinterpret dates beyond the year 1999. You hear people call this the "Millennium bug." That's too benign a term. The Year 2000 Problem is not about a software bug. It's much bigger. It's about business continuation. The other challenges will neither receive nor require much attention unless we fix this big problem. If we don't fix it, our profit-making engines could grind to a halt.

As Figure 1 (p. 52) shows, supplier R&D budgets grew dramatically in 1994, fell in 1995, and began growing again in 1996 under the influence of consortia guidance to prepare for 300-mm fabs. But, as the market has softened, so has the ability of suppliers to maintain high levels of R&D spending. With the Year 2000 Problem added to multiple technology-node development demands, suppliers need help in setting priorities. The list of requirements is simply too long and too costly for suppliers to swallow in such a short period. But their customers, the IC manufacturers, insist, "We want it all, and we want it now." There is a conflict here, and people of vision and leadership must step forward and resolve it.

Figure 1: Supplier R&D spending as a percentage of revenues grew in 1994, declined in 1995, and began to grow again in 1996 to prepare for the 300-mm wafer transition. (Source: Paine Webber)

With all the industry councils, consortia, and conferences so active in the semiconductor manufacturing industry, how has all of this happened? What happened to generate all of the R&D stress and confusion, and how does that connect to the Year 2000 Problem?

Tracing the confusion is not difficult. The ink was still wet on the NTRS when leading-edge device companies began pushing suppliers to hit smaller technology nodes sooner and to do it on 200-mm wafers. The 1997 NTRS may have inadvertently misled suppliers. The NTRS simply missed the forecast when the roadmap dictated which technology node would be implemented with the 300-mm wafer size. Furthermore, consensus, guidance, and clarity on 200-mm tech-node requirements from established consortia have not been adequate. Had better guidance been available, the suppliers would not now be rushing to provide 0.18-µm-and-better design rule capability on 200-mm wafers. With so many consortia efforts focused on 200-mm technology, why do suppliers still have deep and important questions about how far 200-mm technology will be pushed? Why is there still confusion about which technology node will be accomplished on 200-mm wafers? These pressing questions must be answered at the industrywide level that is clearly the domain of consortia.

The second condition that has fueled the current R&D confusion and resulting overload is the 300-mm equipment development push. Directives from collaborations such as I300I and the Japanese consortium Selete have caused suppliers to invest precious development funds too early, dramatically lengthening the elapsed time between supplier R&D investment and customer equipment purchases. In addition, IC manufacturers such as TI, Intel, and IBM learned their lessons about leading the charge to the next wafer size. They are not eager to repeat the experience by spearheading the 300-mm conversion. Now most IC manufacturers want to be "fast followers" and are pushing suppliers to make the financial jump first. They want to finance this shift with supplier money. This is all being acted out on a stage where the economics and timing of the 300-mm transition are now much fuzzier as a result of the 200-mm tech-node push and the Asian financial meltdown. Manufacturers are delaying ramp-ups to 300-mm fabs, and suppliers will have to wait even longer to recover their R&D investments.

We are seeing an extremely demanding and confusing research agenda pressed upon the supplier community. Meanwhile, the Year 2000 time bomb ticks away quietly somewhere in the industry's agenda—but not at the top. The implications of the Year 2000 situation are profound. If your company, your customers, and your suppliers do not correct all software and hardware that create errors beyond 12/31/99, then you will very likely experience a number of serious difficulties. Unfortunately, all of this is occurring when the semiconductor market is soft for other reasons. And the market could soften even more if consumers lose confidence and quit buying hardware and software until they are sure that it will operate smoothly after the turn of the century.

While the Year 2000 Problem is not an overwhelmingly complex technical problem, it is a formidable project management problem. Considerable management and programming talent must be redirected to solve this one. Some businesses have established formal program management offices. Locating software and time-sensitive hardware, finding the date references, changing what needs to be changed, and then testing for performance in millions of application-specific settings—those are the hard tasks confronting us.

Solving this will be costly and difficult, but it can be done efficiently if the industry takes the right steps. The first job is to keep the lawyers at bay. As is too often the case, what starts as a technical and business problem deteriorates into a legal problem. As many people know, when lawsuits happen, only the lawyers win. That would be especially true with this Year 2000 situation. If we liken the problem to a technological natural disaster, then we need to prepare ourselves for some level of destruction, or even casualties in the form of failed businesses. Where disaster strikes, lawyers usually follow. If we understand that to a great extent this is a no-fault situation, then we can keep post-2000 scavenging to a minimum.

No single industry segment insisted on using six-digit dates. All of us did it out of habit, convenience, ignorance, and necessity. It is foolish to now act as if specific parties are to blame. We all ran down this blind alley together, and we will have to run out together. Laying blame and attempting to profit from the situation through litigation will only line the pockets of lawyers and close down innocent businesses.

Sematech has been successfully raising industry awareness about the problem for approximately a year. The consortium is also tracking the status of industry suppliers in an attempt to reassure its member companies that most tool vendors will be in business beyond 1999 and that all of their process tools will handle the Year 2000 changeover.

To do this, Sematech is diligently tracking suppliers' progress in becoming "Year 2000 ready." Arguably, the most significant contribution of Sematech in this area is its consensus suite of test scenarios. This continually evolving test suite reduces the remediation possibilities to a more manageable size and provides suppliers with a much-needed set of tests to best enable them to meet their customers' expectations.

SEMI/Sematech, the consortium of U.S. suppliers to the semiconductor industry, has written a "Year 2000 Readiness Guideline" that provides its membership with a clear explanation of the Year 2000 Problem. The guideline is designed to help members develop a formal plan for correcting the problem in their products and in their business systems. The guideline will help any company, regardless of size, correct the problem. The Web-based document is linked to other useful sites. Its six sections include a detailed section explaining the legal obligations and risks associated with this problem. Since the problem is both pervasive and no-fault in nature, the guideline will be shared globally. SEMI/Sematech will cooperate with SEMI, known for its excellent trade shows and education programs, to help distribute the document. SEMI/Sematech will also continually update the guideline to make certain that it stays relevant and useful.

As you can see, consortia are sounding the call to fix the Year 2000 Problem, and even taking the lead in doing so. But the solution is still not high enough on the industry's agenda. IC manufacturers need to help suppliers rearrange their R&D priorities, providing relief in some process development areas so that the suppliers can move the Year 2000 Problem to the top of the to-do list. Will they?

In summary, if IC manufacturers continue to press suppliers with marginally reasonable demands to deliver on all of the aforementioned priorities without access to extraordinary financial support, some, perhaps many, suppliers will go under. This scenario may lead to equipment gaps, consolidations, and associated reductions in competition. I urge IC manufacturers to press on only if this is their goal. If it is not, then semiconductor manufacturers will have to provide better guidance through their consortia to reduce the overwhelming demands on the suppliers' discretionary research and remediation budgets.

Suppliers are not asking for bags of money from outside sources to solve this problem. But they do need relief from confusing and impractical demands. This relief must come in the form of clearly established priorities. And the only "hard" deadline in any of the priorities I have mentioned is the Year 2000 fix. Tool-development R&D is needed to keep the industry on its productivity curve as dictated by Moore's Law. However, if the semiconductor industry refuses to make the Year 2000 fix its top priority, we may not completely fall off Moore's curve, but we may see a short and very exciting Year 2000 bungee jump that could have long-term implications for our industry. Let's hope we have the courage and the maturity to make the right choices.

Rick Scott is the director of software programs for SEMI/Sematech. He has worked for more than 25 years as a manufacturing and industrial engineer and has held engineering management positions at Sematech, Mead, and Texas Instruments. He is also founder and manager of an industrial engineering firm.

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit