My experience with lithography suggests the following (since I have not seen it with my own eyes).
The "circles" more than likely needed to be squares or off square rectangles. Therefore,
1. Lens resolution is lacking (I doubt this since it would never have made it off the R&D shelf).
2. DUV source is unstable and "mis-pulsing" (I find this hard to believe also since I doubt Cymer would release an unstable system).
3. The operating conditions of the "chemistry" involved in this process (Resist and Developer). This is extremely likely. While variations and instabilities caused by such parameters as thickness, temperature, humidity, airborne elemental contamination, exhaust, soft/PEB/hard bakes parameters may have improved relative to the stability of the actual chemistry, they are still huge variables that can affect the process.
If I can go back to the g-line and i-line chemistries that are supposedly well established, I can cite cases where the incoming materials from the vendors have been the cause of certain process problems. I am sure that under extemely controlled laboratory conditions, the performance of the DUV chemistries are stellar. However, it is my experience that the suppliers of these chemistries are extremely quick to get these chemicals to market and have not done testing over a large range of manufacturing or operating conditions. Batch to batch consistency may not be there are other unknown parameters are not being controlled. I could give you stories that would curl your hair. Ultimately, the end user becomes the test bed for all this information as the chemistries are exposed to all industry conditions through its customer base.
Likewise, end users sometimes do not process under the ideal or reccommended operating conditions because of facility limitations, processing limitations, equipment limitations, or their zeal to get a process implemented before it is completely characterized.
In this case, slow and sure wins the race. I have seen wonderful SEMs (Scanning Electron Micrographs) of DUV features down below 0.25u which to say the least, are impressive. The capability exists for quality DUV processing under given conditions. I do not believe these conditions are clearly understood nor do I believe the resist chemistry is as reproducible as we are being led to believe. This will improve over time as more and more companies experiment with DUV in production and feed back the deficiencies. Such is the life of the "lead sled dog".
On a personal note, sour grapes intended, the quality of some of the Engineers leaves something to be desired. There are too many lithographers coming onto the scene that lack the basic SPC and Design of Experiments skill sets which may exacerbate this issue. As the industry grows, I see 2 things happening. First, we are putting the experienced individuals (processing experience complete with all the problems they have run into over the years with equipment, people and materials) out to pasture for younger blood or promoting these individuals out of the hands on ranks. We are being replaced by less experienced individuals that have become "false" experts on certain equipment. Any fool can be trained to run a piece of equipment but it takes experience to determine what experiments need to be run, generate the right data, and interpret the data CORRECTLY, to achieve the end result.
I also find that companies are unwilling to pay for the experience, provide the additional training (if necessary), or give the appropriate amount of time to do the implmentation correct the FIRST time. Everyonei s a a damn hurry to do things on the fly. More frustrating are the APD (Advanced Process Development) groups that define the process for manufacturing on their own and then "dump" it over the wall. Manufacturing winds up reinventing the wheel for production use because APD lives in their own sheltered and tightly controlled fantasy world.
If I have offended any newbie engineer or APD engineer, you have a small apology. However, you should become a little more introspective in the future and examine the consequences of some of your actions. New is not necessarily better and most expensive does not make it better either. Some of us are finally being listened to and have made management aware that DUV does NOT have to be implemented across the board YET. The old technology works just as well so a marriage of both is both justified and cost effective.
Andrew |
