Is ArF the Final Wavelength?

Will Conley (Source: Freescale Semiconductor )

The lithography prognosticators of the early 1980s declared the end of optics for sub-0.5 µm imaging. However, significant improvements in optics, photoresist and mask technology continued through the mercury lamp lines (436, 405 and 365 nm) and into laser bands of 248 and 193 nm.^1,2 As each wavelength matured, innovative optical solutions and further improvements in photoresist and process technology have demonstrated that extending imaging resolution is possible, thus further reducing k₁.
 
Our industry will continue to focus on the most cost-effective solution. What continues to motivate lithographers to discover new and innovative lithography solutions? The answer is cost. The development of new tooling, masks and even photoresist platforms impacts cost. The switch from KrF to ArF imaging materials had a significant impact on process integration. The requirements stated in the International Technology Roadmap for Semiconductors (ITRS) for current and future technology nodes are very aggressive. Therefore, it is likely that high numerical aperture (NA), in combination with enhancement techniques, will continue further for aggressive imaging solutions.

There is an interesting bifurcation in progress among the device manufacturers. On one side, there are the NAND flash/DRAM manufacturers that will continue to push half-pitch at a very aggressive rate. It would appear that flash/DRAM manufacturers' rate of shrink might be faster than what is stated in the ITRS.³ Flash/DRAM manufacturers represent more than ~60-65% of the exposure tool suppliers' customer base in terms of tool purchases. Logic manufacturers are about half a generation behind in half-pitch and thus could be forced into accepting imaging solutions that are less than ideal.

Lithography and more importantly “imaging solutions” are driven by economics. The technology might be extremely innovative and “fun,” but if it's too expensive it may never see the light of scanner. The industry continues to focus on process tricks and creative science to solve problems in the most cost-effective manner. There are many examples of this, such as new rinses to reduce pattern collapse, surface conditioning for reducing line edge roughness, new materials to assist in shrinking contact holes, and surface treatment of ArF resist materials to reduce or minimize slimming.

Over the past several years, there have been several interesting publications for manufacturing and integration challenges for sub-0.3 k₁.^4,5 However, there are still limitations to the rate of shrink, and other methods such as double patterning have been proposed. Double patterning can enable more aggressive shrink rates through pitch doubling, but also at a significant increase in cost. This cost increase can have a different effect based on your product mix, and also affect what technology you believe is best for your company.

We have taken a hypothetical look into the cost of technology for <1000 wafers per mask set, considering ArF optical, EUV and maskless (ML2) lithography. Based on our model, we find that ML2 at ~10 wph is the most cost-effective solution, with optical a close second. Of course, any increase in wafers per hour for maskless would reduce the cost even further. The cost of the mask for each technology is a major contributor to the overall cost of the product. In this case, double exposure cannot be considered because of the overwhelming cost compared with the other options.

We also investigated the cost of technology for >5000 wafers per mask, considering the same technology options. In this situation, we see that optical is more cost-efficient and ML2 at 5 wph is the most expensive. In this case, the cost of the mask is much less significant because it is disbursed over the additional wafers. We can imagine that those IDMs that run >10,000 or even 100,000 wafers per mask set are able to further reduce this cost contribution. However, in this case, a single-exposure optical solution is the most cost-effective. There is another interesting aspect to this work: the cost of double patterning ArF vs. EUV. There are some similarities, but in a best-case scenario where the cost of the mask is less and the throughput is double, EUV is more cost-effective.

Different product sets can and will require different imaging solutions. There is no one solution that will fit all. The answer to the question of whether ArF will be the final wavelength will depend on the cost of each technology and the availability to maintain the shrink path. Regardless of the next imaging technology beyond single-exposure water immersion, it only has to cost less than double patterning at 1.3 NA.