Communications of the ACM

neutral chemical to act as a base, then adding the block copolymer on top, and heating the whole thing. The polymers would form alternating lines— think of them as red stripes and blue stripes—and one set, say the red, would be washed away. The remaining blue stripes would act as a pattern for the same etching process used in current photolithography to inscribe circuits into silicon.

Photolithography also relies on chemicals—photoresists that are patterned through exposure to light, creating a pattern that is used to define which parts of the wafer to etch. Many modern designs, however, require multiple masks, driving up costs. The machines for focusing the light beams to ever-smaller sizes are also expensive, so DSA could, the thinking goes, create smaller features and do it more cheaply.

A few years ago, DSA was seen as the next big thing, and many chipmakers were investing heavily in it, says Kurt Ronse, director of the advanced lithography research program at the Interuniversity Microelectronics Center (imec) in Leuven, Belgium. The focus was on a block copolymer consisting of polystyrene and polymethyl methacrylate, or PS-PMMA, which was targeted for features with a 28nm pitch, in which lines and spaces are each 14nm wide. At the time, the anticipated next photolithography technology was extreme ultraviolet (EUV), but progress toward making it commercial seemed to be stalled.

Too Many Defects

The hype over DSA did not last, says Ronse. “Very quickly it turned out that one of the big issues was going to be defects,” he says. Sometimes the copolymer would leave a gap, or the stripes would get too close to each other or even cross, a phenomenon called “line edge roughness.” The undercoating might have a pinhole or a slight bump that would throw the process off.

“It uses thermodynamics to form
the patterns,” says Charles Black, di-
rector of the Center for Functional
Nanomaterials at Brookhaven Nation-
al Laboratory in Upton, NY. “It usually
ends up with a defect here or a defect
there, and microelectronics is really in-
tolerant of that.”
Over time, Ronse says, researchers
managed to reduce the level of de-
fects, while at the same time, EUV was
beginning to look more promising.
The combination made it seem less
likely that DSA would provide the low-
cost alternative the industry sought.
“After a couple of years of develop-
ment, the window of opportunity for
28nm pitch was starting to close,”
Ronse says.

Though the hype has faded, he says, work continues. Some of the companies imec works with have ended their DSA research, while others are still going, though they have directed some of their chemistry work toward developing photoresists for EUV. “They have not stopped DSA development, but by putting less effort in it, it definitely slows down,” Ronse says.

While research into DSA is not as ag-
gressive as it once was, it has not
stopped, says Christopher Ellison, as-
sociate professor of chemical engi-
neering at the University of Minnesota.
“I don’t think it’s going away,” he says.
“It’s not growing at a rapid rate.”
Part of the reason for that slowdown
may be that it is not the optical physics
that the chipmaking industry has al-
ways relied upon. “It makes it challeng-
ing for industry to accept because it’s
so radically different,” Ellison says.

Surface Tension

While PS-PMMA may seem less attrac-
tive than it once did because its resolu-
tion limit is 11 nm, chemists are pursu-
ing other block copolymers that could
go even smaller. Some materials have a
high chi, a measure of how incompati-
ble the two polymers are. High-chi co-
polymers can perform the same self-
assembly as PS-PMMA, but at a smaller
size. Ellison’s group has created struc-
tures just four to five nm wide.

The trouble with high-chi copolymers is that it can be difficult to get them to stand up straight. To form parallel strips, they need to orient themselves perpendicular to the surface they are on. Often, however, they will flop over onto their sides, destroying the pattern.

Because self-assembly is a thermodynamic process, it is controlled by differences in surface energy, and the top “surface” is open air. Some researchers, such as Ellison and his colleague C. Grant Willson, a professor of chemical engineering at the University of Texas at Austin, have been working on topcoats, films of chemicals that change the surface energy on top of the copolymer so that it aligns itself correctly.

Paul Nealey, a professor of molecular engineering at the University of Chicago and one of the pioneers of DSA, says there are three or four methods to control the orientation of high-chi materials. Nealey and Karen Gleason, a professor of chemical engineering at the Massachusetts Institute of Technology, Cambridge, MA, published a paper in Nature Nanotechnology in March in which they reported using a technique called initiated chemical vapor deposition to create a topcoat that forced the block copolymer to line up the way they wanted it. The approach allowed them to create features only 9.3nm wide.

Another approach exposes the copolymer to vapor from a solvent to change the balance of energy. A technique explored by IBM mixes in an additive to the copolymer that changes how it responds to surface energy at the air interface.

However they are controlled, block copolymers cannot do everything on their own. Some lithography must be used to inscribe guiding lines for them to follow—that is the “directed” part of DSA. Because the material forms lines much thinner than those inscribed, they require less-advanced photolithography processes. At some resolutions, the guiding pattern could be created by the current state-of-the-art process, 193nm immersion lithography, which uses a liquid to focus the light beam to a