The Asian Flu and Industry Standards -- Turning Lemons Into Lemonade

The advanced process control Holy Grail travels slowly over the bumpy road of de facto standards.

Staff -- Semiconductor International, 8/1/1998

Advanced process control (APC) can be viewed as being composed of four parts, two addressing control of the wafer state and two the control state.

The first part is devoted to model-based control, either in situ or in-line. The second is for run-to-run control, which refers to the time base. The last two pertain to equipment fault detection, which can be done with trace analysis, by taking telemetry from the system and analyzing it to determine whether it is possible to predict a fault, identify a fault and recommend for corrective action.

Taking a panoramic view of the APC landscape, it is obvious that there are few standards for integrating process or metrology tools into a process tool cluster and for integrating this metrology/process tool cluster with CIM architecture. There is the SEMI Equipment Communications Standard (SECS), which specifies a group of messages and the respective syntax and semantics for messages relating to semiconductor manufacturing equipment control, with the SECS/GEM HSMS communications protocol being worked upon, with few promising results.

For example, there are no standards that will allow someone to build a lithocell management system with four CD-SEMs attached to it, four overlay tools, plus two thin-film thickness measurement tools, then attach it to a server with the right software to get integrated data handling, integrated analysis and integrated feedback control for APC.

Thus, there is no standard way to get into a stepper to modify the recipes, that is, no process tools control standards. There are no metrology tool integration standards. Finally, there is no standard way to get that data into the cell controller or the CIM system. So the industry is trying to meet integration problems, without standards to help solve them.

To be sure, there is a prototypical standard, the SEMATECH CIM framework, but it is far from completion.

The standards creation process recalls the story of the mountain that struggled to give birth a mouse.

When it is decided that the industry needs a certain standard, a committee is formed to study what has been done and determine what processes must take place to solve the problem requiring the standard. The first step is to find out and study de facto standards that may be out there. Equipment or software manufacturers with a unique technology and/or with a large market share are often the originators of de facto standards, created to solve problems their systems faced.

The standards committee studies these de facto standards carefully, presuming that there is little point in creating a different standard that will clash with the one already fielded by a company with a large market share. The committee members may modify it, amplify it somewhat, or simply document it well so other manufacturers and software providers will be able to hook to it and architect their own products to work with it. To say this process takes time states the obvious.

Behind the doors of the meeting rooms, in-fighting over de facto standards can get intense. Sometimes a small, aggressive company can get disproportionate representation on a standards committee and pushes to make its version the standard. Sometimes the result is a superior standard, but is not quite the one that Gigantic, Inc., wants because it might help its competition, so it refuses to use it, declaring itself the de facto standard for the industry on the basis of its large market share and fielded systems.

Going on with the process after looking at the various cost, flexibility, and extendibility alternatives, the committee generates a proposed standard. If there already was a de facto standard, it is careful to incorporate it. Then the standard is verified, which means going to the organizations that will use it and getting suggestions and then going to yet another step, refining the standard.

The final level is consensus, which may result in the standard going back for yet further refinement.

Although the process makes sense, it fails to recognize that technology moves faster than the standards cycle. Because of this, de facto standards are often driven by economics and may result from something as common as the tool acceptance process. It is not unusual to have metrology tools for example, generate their own test vehicles to enable the buyer of the system to accept the tools.

Another problem the industry faces is that there can be numerous de facto standards applying to the same thing. An example of this is the evaluation of a CD-SEM. Generally, each buyer has a de facto process for evaluating the system, each one different. The system manufacturer provides a modicum of standard evaluation procedures to help the customer gain some commonality with others, thus assisting in the system evaluation, but the reality is that de facto evaluation standards are quite common.

The Asian economic situation is making the industry take a closer look at 300 mm and 193 nm lithography. Some think 300 mm may not meet the ROI models originally envisioned. The move to 193 nm could mean that a scanner may cost as much as $14 million, about twice the cost of its 248 counterpart. Both Intel and Samsung are considering extending the 258 nm DUV process with higher numerical apertures. Now, using scanner systems capable of 0.73 NA, Samsung believes it can go down to 0.13-µm, putting more devices into a 200 mm wafer. As things stand, 193 nm lithography may be considerably delayed, giving DUV standards a chance to mature.

When technology is pushed further than originally envisioned, standards must become more stringent, but the results are worthwhile. While unlikely that such a fast-moving industry as this will get away from the de facto standard, the enforced hiatus in growth and development the industry is experiencing, courtesy of the Asian flu, could be a time when it is possible to turn lemons into lemonade.

The current slow-down offers an opportunity to firm up and develop much needed standards. This is a time when the industry must come together and consolidate its gains and prepare for the future when the current malaise passes, and it finds itself in the middle of another boom.