Resist Sensitivity, Source Power, And EUV Throughput

What do improvements in resist sensitivity mean for EUV’s commercial viability?

popularity

In a recent article, I quoted 15 mJ/cm2 as the target sensitivity for EUV photoresists, and discussed the throughput that could be achieved at various source power levels. However, as a commenter on that article pointed out, reaching the 15 mJ/cm² target while also meeting line roughness requirements is itself a challenging problem. Because of the high energy of EUV photons, a highly sensitive resist will capture a relatively small number of individual photons. The stochastic distribution of photons within the exposed feature will therefore inevitably lead to edge roughness. For this reason, Moshe Preil of GlobalFoundries has argued that production resists will need to be much less sensitive — perhaps even in the area of 50 to 100 mJ/cm².

So what do changes in resist sensitivity imply for overall EUV system throughput? I posed the question to lithography expert Chris Mack, and appreciate his prompt and concise answer. As he explained, the total time per wafer consists of the exposure time plus the overhead time. The latter includes time spent accelerating and decelerating the stage, and is typically very small. The exposure time (T^expose) is a function of resist dose to size (E^size), source power, and a number of equipment specific factors (k). For example, each reflective element in the EUV optical path will introduce additional optical losses, so the power actually reaching the resist will be less than the power delivered by the source.

EUV2

If the throughput is the reciprocal of the exposure time, and the overhead time is very small, then

EUV1

So, throughput is approximately linear and is directly proportional to both source power and resist sensitivity. This is very bad news for EUV advocates. If acceptable image quality can only be achieved with 60 mJ/cm² rather than 15 mJ/cm2 resists, then source power will need to quadruple, too. Rather than an 80 W source being able to deliver 50 wph, a 320 W source would be required.

The dual challenges of resist sensitivity and line roughness are helping to motivate research into alternative resist chemistries. Have conventional chemically-amplified resists come to the end of the road? That’s the question I was investigating before turning down this side street, and the subject of my next article.