IEEE Spectrum February, 2014 - 38

Ordinarily, this approach would work well in only two dimen-
sions. That's because even if we focused the laser deep into the
resist, photons would be absorbed not only close to the focal re-
gion but also throughout the entire beam cone, below and above
the focus. To make the curing process work in three dimensions,
we alter the strategy slightly, pairing the "photoinitiator"-the part
of the photoresist that absorbs photons-with a longer-wavelength
laser. If the combination of the two is just right, we can create a sys-
tem in which the photoinitiator must absorb two photons instead
of one to become excited.
If two photons are needed, the absorption response isn't linear.
Instead, it scales with the square of the light intensity; if we dou-
ble it, we can get four times the response. That helps confine the
effect of a laser: If a laser beam is tightly focused, the exposure will
be sharply confined to a small volume around the center of the
focus. Then, to draw an arbitrary 3-D shape, we need only move
the focus around, by either moving the sample or the laser beam.
All by itself, however, this technique isn't quite enough to get us
meta-atoms small enough to interact with light. The limitation is
the Abbe diffraction barrier-a characteristic of microscope optics
that limits the spatial resolution of a lens and thus how closely
you can put two adjacent features or lines. For an 800-nm laser-
which works well with common photoresists-and for high-end
microscope lenses, you're limited to a lateral distance of about
300 nm because of Abbe.
For decades, this limitation appeared to be fundamental. But
about five years ago, we proved there was a way past it. The idea
got its start with physicist Stefan Hell, now at the Max Planck Insti-
tute for Biophysical Chemistry, in Göttingen, Germany. In the early
1990s, he proposed a way to break the diffraction barrier by using
a second laser operating with a different frequency.
Through a process called stimulated emission depletion, this
second laser can cause an excited molecule to spit out a photon
and relax back to a lower energy state. That's useful for lithogra-
phy, because, in effect, it gives us an eraser to go with our pen. The
writing beam will have a hot spot in the center, and the erasing
beam will have a different cross section: a specially shaped focus
with zero intensity right where the writing laser is at its maximum
[see illustration, "Writing Through Resist"]. When both beams are
38

FEB 2014

north amErican

SPEctrUm.iEEE.orG

used, everything outside the very center of the writing beam can
be de-excited, and the photoresist there will remain unexposed.
Hell's group was interested in using this technique to cause
stained cells and other biological structures to fluoresce in as tiny
a spot as possible, so they could be imaged at very high resolution
under the microscope. (His team has since done this to great effect:
In 2009, they showed they could resolve features under the micro-
scope as small as 6 nm-just a few atoms across-using visible light,
well under the Abbe diffraction barrier.)
But adapting the two-laser approach to curing photoresists
wasn't easy. When our group at KIT started to look into the pos-
sibility in 2008, there were no photoresists that had been specifi-
cally developed to support the approach. One complication we
encountered early on was that photoinitiators are designed to be
efficient; after they're hit by photons, they react in almost no time
at all, too fast for a second laser beam to de-excite the molecule and
stop the polymerization reaction. It took about a year of research
and some trial and error, but we eventually found a photoinitiator-
a dye molecule that had been used only sparingly for lithography
up until that point-that would do the job.
With that photoinitiator in hand, we found we could print struc-
tures with a lateral resolution-the distance between two adjacent
features-of about 175 nm, about 40 percent finer than what can
be achieved at the Abbe limit.
That's quite an improvement. But in principle, spatial resolu-
tions of a few tens of nanometers (beyond that you're starting to
reach the molecular scale) should be possible using 800-nm light.
Pushing the resolution of this lithography down to such a scale
will require more work. The problem is no longer optics; it's the
photoresists that we use. For reasons that are still being explored-
perhaps the diffusion of the photoresist molecules themselves-
attempts to make smaller structures typically result in ill-defined
features. If you try to make two features closer together than 175 nm,
for example, you can end up curing areas that aren't supposed to
be part of the final shape.
Still, the resolution is now fine enough to allow us to create ar-
tificial materials that can operate in the visible part of the spec-
trum. Many of them would have been impossible to make even
five years ago.
continued on page 57

left: karlsruhe institute of teChnology; right: karlsruhe institute of teChnology/oPtiCs letters

gone in 5 micrometers: the polymer-based metamaterial above, made up of rods stacked in a woodpile-like arrangement, was the first to hide the
presence of a 3-d object at visible wavelengths of light. this 5-micrometer-thick "carpet cloak" can make a 0.5-μm-high gold bump [parallel lines in
center of top right image] appear as if it were a flat metal mirror for all polarizations of light and from nearly all angles of incidence.

http://SPEctrUm.iEEE.orG

Table of Contents for the Digital Edition of IEEE Spectrum February, 2014