Paul van Gerven
26 October

ASML expects to supply its customers with a new generation pellicle by the end of the year. Raising transmittance from 83 to 90 percent, it will take away some of the pain of having to use one.

You’d think that anything reducing the amount of precious EUV radiation hitting the wafer in an EUV scanner would get the thumbs-down from semiconductor manufacturers. But they made an exception for the pellicle. Even though the membrane that keeps stray particles away from the EUV photomask will inevitably absorb some EUV radiation, thus decreasing throughput, the idea of having their costly masks open and ‘exposed’ was simply unbearable to chipmakers. They insisted on having a way to protect it.

Since there were no takers at the time, ASML was more or less forced to take up pellicle development itself. Four years ago, its efforts reached the industrialization phase. What has happened since then? Raymond Maas, ASML’s product manager for pellicles, fills us in.

Credit: ASML

So, did your pellicle make it into fabs already?

Maas: “Indeed it did: we’ve shipped thousands of them over the past two years, to all of our EUV customers. The films are being manufactured by Teledyne Dalsa in Canada, and then assembled and qualified in Veldhoven. All the required tooling we had to develop in-house, because obviously it didn’t exist before.”

Originally, ASML didn’t think pellicles would be necessary because it predicted the environment inside the scanner would be clean enough. Since chipmakers are currently using pellicles, was that prediction too optimistic?

“The inside of the scanner isn’t the only consideration. Reticles are also handled outside the scanner, during which particles can be introduced. In many cases, our customers choose to use a pellicle because they want to protect the reticle all the time, not just inside the scanner.”

“That said, while we’ve made great strides with reducing particle density inside the scanner, I can’t say it’s perfect. It’s clean enough for some use cases; this mainly depends on the size of the chip. If the chip is relatively small, the mask pattern takes up only a small percentage of the mask’s surface, meaning the chance of a particle landing on a critical part is relatively small. This is one situation in which our customers may opt to work without a pellicle.”

How much light is absorbed by the pellicle?

“Our current polysilicon-based pellicle has a transmittance of 83 percent. EUV masks being reflective, light has to pass through it twice, so that represents a considerable loss. Of course, our customers would prefer not to use it, but it’s just not feasible to obtain the level of cleanliness in every aspect of mask handling.”

“We expect to introduce a new, metal silicide based pellicle by the end of the year, with which we target 90 percent transmittance. Equally important, however, is that this upgrade supports our roadmap, which eventually will take source power up to 400 watts. The pellicle heats up to 600 degrees Celsius at that power level, which the polysilicon couldn’t withstand.”

Now that things are up and running, and other organizations have started pellicle development, shouldn’t ASML be stepping back from this activity?

“We’ll keep driving the roadmap until there’s a pellicle with 95+ percent transmittance that fulfills all the other requirements. We feel a responsibility towards our customers to do that. However, we happily support every effort to come up with improved designs. As we found out ourselves, though, going from lab to fab takes a lot of effort and requires large investments.”

“Now that volumes are going up – perhaps to as much as 10,000 pellicles per year – we don’t consider ourselves to be ideally suited to keep handling the assembly. As announced last year, Mitsui Chemicals will take over in that department. We expect that Mitsui will start handling pellicle assembly by the end of this year.”