Why a number of individual steps ultimately proved so difficult, and how that will play out for future chips.
EUV systems are beginning to ship to large foundries in volume, setting the stage for one of the biggest leaps in technology the semiconductor industry has ever witnessed.
ASML has emerged as the sole supplier in this market, but it has taken an entire ecosystem to develop EUV. It has taken billions of dollars of investment by ASML, along with enormous cash infusions by Intel and TSMC, contributions from universities and research houses around the globe, and complementary technologies such as photoresists and atomic-level etch by other companies.
But EUV has turned out to be so complex that it has left many scientists and engineers shaking their heads in disbelief. So why did the industry end up backing this technology?
Semiconductor Engineering posed this question to Hans Meiling, ASML’s vice president of service and product marketing for EUV. His answer: “There is 20-plus years of history here. You see a big CO² laser cutting steel. Then you have plasma, the conversion process from infrared into EUV light. And then you have an electrical discharge source. People used water molecules, but they didn’t get enough energy out of that. So they thought, ‘Let’s move to another material. Tin has a high conversion.’ You can look it up in the Periodic Table. ‘Maybe this is a good candidate.’ It started in a university laboratory. And then someone else is using a high-powered laser. Look at the U.S. Defense Department in the 1990s. They were looking at very big lasers. So people started combining things. If you need hundreds of watts of source power, is there a way of combining the knowledge?”
The plasma power source turned out to be such a big issue that many experts were convinced that EUV would never materialize.
“We’ve been ready for EUV for many years already, and every year it was always one more year,” said Meiling. “But we’re beyond that. Predictability is now a lot higher. We’ve overcome the main hurdles. It’s physics. It’s surface chemistry. It’s optics that must stay intact. It all comes together in one vessel. It’s not only the power. It’s availability.”
Still, a lot of things have happened since EUV was supposed to show up, first at 45nm, and then at 28, 16/14nm. For one thing, chipmakers have come to grips with the fact that ever since the end of classical scaling at 90nm, just shrinking features doesn’t necessarily improve performance and lower power. And even with EUV, the price per transistor is no longer decreasing at the same rate as it did prior to finFETs and double patterning.
To achieve improvements in performance, power and cost reduction, chipmakers have developed new architectures, different packaging options, introduced new materials, and even come up with lithography alternatives such as directed self-assembly and multi-beam e-beam.
Whether all of these developments would have progressed this far had EUV hit its initial deadline is speculation. But given all of these developments, it does raise some interesting questions:
• How much of the semiconductor industry will take advantage of EUV for developing chips, and how much will that affect the cost?
• Will EUV be used just for making cuts and holes in critical metal layers, or will it ultimately replace immersion lithography.
• At what process nodes will EUV play a significant role?
• How many semiconductor applications will require smaller nodes, and will it be as standalone chips with more regular structures, such as chiplets in a package, or will it continue to be complex SoCs?
The answers to all of these questions have a direct bearing on the future of this massive development effort. The challenges in developing EUV were enormous. But much has changed since the project was started, from the size of individual markets to the pace of change and the level of customization required in those markets. This is no longer the only game in town.
So while EUV is desperately needed at the most advanced nodes, how many companies will move to those nodes and with what volumes is uncertain at this point.
Moore’s Law: A Status Report
The ability to shrink devices will continue for at least four more nodes as EUV begins to ramp, but it’s just one of a growing number of options.
Why EUV Is So Difficult
One of the most complex technologies ever developed is getting closer to rollout. Here’s why it took so long, and why it still isn’t a sure thing.
More EUV Mask Gaps
Pellicles and inspection remain problematic.
Inside EUV Resists
Inpria CEO talks about EUV resist challenges