Keeping clean in the extreme
By Hank Hogan
Despite the high technology, the semiconductor industry is conservative, preferring to stick to what is known and works rather than trying something new-unless, of course, there’s no alternative. That’s the case, the thinking goes, with extreme ultraviolet (EUV) lithography. EUV uses 13.5 nm wavelength photons in the place of today’s state-of-the-art 193 nm, which can only print 65 nm features using a slew of optical tricks.
“It makes lithography simple again,” says Vivek Bakshi, a senior member of the technical staff at SEMATECH (Austin, TX) in speaking about EUV. “That’s why with all the delays and all the technical challenges and all the differences, we’re still sticking with EUV.”
Lately, there’s been some encouraging news when it comes to EUV processing and contamination. One involves the mask blanks on which circuit layers are written. A second involves the EUV light source, which allows scanners to print circuits on wafers using masks.
The mask blank success comes in the detecting and cleaning of particles measuring only 10 nm, an achievement of SEMATECH’s Mask Blank Development Center (Albany, NY). That size is important since it’s about half the critical features in the 22 nm technology generation, at which EUV is likely to be used.
An EUV mask is painstakingly constructed by depositing multiple layers on a blank. A particle sitting on the blank will distort any pattern written into the layers above it and thereby render the mask useless. So it’s vital that the blank be clean.
The problem is that 10 nm is very small and that makes it difficult to take the first step in contamination control, notes Abbas Rastegar, also a senior member of the technical staff at SEMATECH. “First, you should see the defect,” he says.
To do this, Rastegar and others used a defect inspection tool from Lasertec Corp. of Yokohama, Japan. The inspection tool is based on reflectivity of 266 nm light and is specified to find 95 percent of all particles 30 nm and larger in size-far bigger than the 10 nm particles being sought. To get around this limitation, the group marked likely defects and then confirmed their size using an atomic force microscope. They then cleaned the mask blank using a cleaning tool from HamaTech AG of Sternenfels, Germany. After cleaning, they again scanned the blank, verifying that the defects were removed without harm to the blank.
This operation would be easier using a light source with a smaller wavelength, such as 13.5 nm. The second EUV-related advance may help with that. Recent announcements have shown that improvements in performance have lasers in the running to create the plasmas that, in turn, produce the EUV photons. Laser-based systems might show up in EUV scanners or EUV-based metrology or inspection tools.
One of the advantages of the laser-based approach is that optics can stand off at a greater distance, allowing these expensive components to be better protected against the heat, debris, and associated contaminants from the plasma. As Bakshi explains, “We have more distance and we can put in more debris mitigation devices.”
He adds that lasers still must be improved before they can be used. However, he says, real progress in this regard has recently been made, with lasers now approaching the power levels needed to make them competitive as a plasma producing source. It may be necessary to gang several lasers together to finally achieve the required performance.