SAN JOSE, Calif. — ASML unveiled the NXE:3400B, the extreme ultraviolet lithography scanner that it hopes makes it into volume-production fabs, and gave the first close-up look at its follow-on. At the same event, Carl Zeiss and Samsung described high-resolution systems to inspect masks at the same wavelength as EUV light.
The systems were disclosed at the SPIE Advanced Lithography conference, where Globalfoundries and Imec both detailed work on 5-/7-nm nodes using a mix of EUV and current 193-nm scanners.
The talks suggested that EUV commercial use is coming into focus. However, with real hurdles still ahead and no announcements of major purchase orders, skeptics still had plenty of breathing room.
The production-worthiness of the new ASML system is based on “a bunch of assumptions” including photo-resist materials effective at light exposures of 20 mJ/cm2, said Chris Mack, chief technology officer of Fractilia. The fact is that current resists need 30 mJ/cm2, he said.
What’s more, the road map still depends on a variety of other ASML and third-party tools arriving in synch, including pellicles to keep contaminating particles off of EUV wafers. “If the schedules slip a year, we will miss the 5-nm window,” he said.
Meanwhile, new issues are popping up. For example, although feature sizes are shrinking, the roughness of their edges are not — taking up as much as 20% of a finely printed line. Mack is one of many in the industry seeking ways to use statistics to at least predict the roughness, if not smooth it out.
A flexible illuminator is the key ingredient of the 3400B. Click to enlarge. Source: ASML.
For its part, ASML is shipping the first 3400B this week. The system is the company’s first rated to produce 125 wafers/hour, the throughput target for use in production fabs.
The company hit the goal not by nudging its light source up to the target level of 250 W, but by speeding up wafer handling inside the machine, which should be able to hit 210 W. Initially, the 3400B is expected to run at 148 W to produce 104 wafers/h.
Most, if not all, of the 14 EUV systems installed in the field are expected to take the upgrade, a process that could take two years. Each upgrade takes about three months, a complex process compared to open-heart surgery in part because it involves breaking the vacuum that holds the light source and other key components.
ASML’s road map includes light-source upgrades, in part because it knows that users need headroom to make up for underperforming resist chemicals and optics that will degrade over time. But the company is mum on just how and when those upgrades will arrive, presumably because it has not yet gotten consensus among its handful of top customers.
In a separate talk, ASML gave the first detailed look at its second generation, a scanner announced in November with upgraded optics co-designed with Carl Zeiss. It will be nearly twice the size of the 3400 and print 8-nm features, 40% smaller than those made by the 3400.
New optics from Zeiss are the central focus of the new scanner. Other components, including the light source, will have evolutionary upgrades.
Zeiss has already started ordering some components and building new grinding machines for the human-sized optical module. It requires two new inspection systems, also human-sized, that, for the first time, sit inside the vacuum chamber.
Unlike the 3400, the new system will expose two 16.5 x 26-mm half-sized fields at a time. That’s because it uses separate magnification levels on the x- and y-axis. To hit throughput targets, it will move through the system wafers at twice and masks at four times the acceleration rate of the 3400.
The system will have a 30-nm focus budget, half that of the 3400B. It aims to start commercial life kicking out more than 180 wafers/hour while printing pitches from 22 to 14 nm. The system is also being designed to deliver EUV light doses up to 60 mL/cm2.
Next page: Fabs try mix-and-match lithography
ASML’s next-gen system dwarfs its already huge 3400B. Click to enlarge. Source: ASML.