|
MEETING
THE CHALLENGE OF DEEP UV OPTICS
By
David Collier and Wayne Pantley, Alpine Research Optics
David
Collier is President and Wayne Pantley is Sales Manager of Alpine Research
Optics, 3180 Sterling Circle, Boulder, CO 80301, 303-444-3420, FAX
303-444-1686, E-mail AROcorp@AROcorp.com, http://www.optics.org/arocorp/
Applications
such as microlithography and photorefractive keratectomy (PRK) are creating a
growing demand for high precision optical components that operate in the deep
UV (below 250 nm). Some of this need is being met with components fabricated
from CaF2
, a crystalline material that transmits from 130 nm to 10 µm. Historically, CaF2
has been used mostly for its infrared transmission, so adaptations in how bulk
material and finished components are produced must be made in order to meet the
special requirements of deep UV operation. This article will review those
performance requirements and then examine the new technologies being developed
by both material and component producers.
Background
The
single most important and technically demanding application for UV optics is
microlithography, used to produce integrated circuits (ICs). The latest
microlithography production systems use excimer lasers, with output at 248 nm.
However, a move to excimer lasers operating at 193 nm, and possibly 157 nm, is
already underway in order to achieve smaller circuit features. While fused
silica components are adequate for 248 nm, refractive optical systems at 193 nm
often use at least some CaF2.
And at 157 nm, CaF2
is virtually the only useful transmissive material.
In
the microlithographic process, an excimer laser is used to illuminate a circuit
pattern (called a reticle), that is then projected at a high reduction ratio on
to a photoresist covered silicon wafer. Presently, most ICs have feature sizes
of about 0.35 µm, with 0.25 µm geometries just coming on-line. The use of 193
nm and 157 nm systems will allow feature sizes smaller than 0.1 µm. Needless to
say, producing a lens system that delivers this kind of resolution is extremely
difficult. In order to meet this resolution challenge, a typical
microlithography projection lens has over 25 elements, with some being more
than 1 foot in diameter. In fact, it's not uncommon for a projection lens to
weigh over 100 pounds.
Excimer
lasers only operate in the pulsed mode, so each exposure actually consists of a
number of pulses. The projection optics typically experience fluences in the
0.1 mJ/cm2 to 1 mJ/cm2 per pulse range. Some of the other optics in the system
are subjected to fluences in the 1 mJ/cm2 to the several mJ/cm2 per pulse
range. Since a production line system might involve several million pulses per
day, the high replacement cost means that the optics must be able to withstand
several billion pulses over their lifetime.
Material
Production
In
order to meet the size, quality and quantity requirements of microlithography,
material producers are engaging in an ongoing effort to improve CaF2
. For example, Bicron (Solon, OH) has specifically focused their development
efforts for "excimer grade&qupt; CaF2
on the following areas:
Improved
deep UV transmission
Higher damage threshold
Lower fluorescence
Lower index homogeneity
Reduced stress birefringence
Increased physical size
Higher quantity production
The
naturally occurring, mined CaF2
material that is acceptable for most IR applications is completely inadequate
in regard to these parameters. Instead, synthetic CaF2
is typically produced using the Stockbarger method. In this approach, a
crucible containing molten material is slowly lowered through a "freeze-zone"
where crystallization takes place. The entire process typically takes 6-8 weeks
for a batch.
|
Schematic of the
Stockberger CaF2 fabrication method.
figure 1
|
Bicron
has adapted the Stockbarger technique (figure 1) to improve the parameters just
identified. Specifically, the furnace has been modified to create a freeze-zone
that allows tighter control over crystal growth, and to enable the use of more
sophisticated process control instrumentation for better regulation of
temperature, vacuum and the annealing process. These latter factors are
particularly important, as they govern material homogeneity and stress
birefringence. The results have been quite dramatic. Over approximately the
past year, stress birefringence of production parts has been reduced from a
value of 20 nm/cm (a typical value for Stockbarger produced CaF2
crystals) to just 1 nm/cm. These innovations have also enabled Bicron to scale
up their production process; they now routinely produce CaF2
ingots of 24" in diameter by 6" in length, with the capability to go as large
as 36" in diameter.
The
primary factor affecting the transmission, damage threshold, color center
formation and fluorescence of the final crystal is the purity of the materials
used. To this end, Bicron utilizes a variety of testing methodologies (VUV and
UV/VIS spectrophotometry, fluorescence spectroscopy, etc.) to detect about 50
different contaminants during production of the precursor materials from which
the CaF2
is created. At each process step, specific impurities are targeted for chemical
elimination. Ongoing development and rigorous application of these techniques
at Bicron over the past year has lowered typical contaminant levels from 10 ppm
to under 0.5 ppm.
Further
material purification is also performed during the crystal growth process.
Chemical "scavengers" are placed in the crucible during crystallization; these
are selected to react with any remaining impurities. In addition to the
scavengers traditionally used in the Stockbarger technique, Bicron has
developed a number of new chemicals that are designed to react and produce
gaseous by-products. These can then be easily removed from the process by
vacuum pumps.
Component
Fabrication
Meeting
the performance demands of microlithography objectives, as well as other
excimer laser optics, has also required development in optical polishing
techniques. Typical component tolerances for flat excimer laser optics are 1/10
wave flatness and 10-5 surface quality. Keep in mind that a 1/10 flatness
specification at 193 nm is over three times flatter than that same
specification at 633 nm, so these are very tight tolerance optics. Surface
imperfections also take on much more importance at shorter wavelengths due to
diffraction.
The
challenge of meeting these specifications in CaF2
is made even more difficult by the fact that it is an inherently difficult
material to process. CaF2
is hygroscopic, soft and very prone to chipping, particularly at the edges.
Material that chips off during polishing may then be dragged over the soft
surface of the component, leading to sleeks (sleeks are small surface
scratches). The result is that the polishing time needed to achieve a given
surface quality on CaF2
is typically 50% to 100% longer than for glass.
Manufacturers
involved in polishing CaF2
have each developed their own techniques for working with the material. At ARO,
our solution has involved a number changes to our standard polishing methods.
First and foremost, we have environmentally isolated the polishing area for CaF2
from the rest of our production area. This prevents small airborne particulates
from our glass and silica polishing operations from settling on the optics and
causing scratches. Fine debris from CaF2
polishing (mostly salt crystals from fluoride material) must also be
continuously evacuated from the production area.
The
use of dedicated polishing equipment is also necessary because CaF2
requires different polishing compounds than glass and silica. Diamond and
sapphire grits are used at the initial stage because their hardness speeds
material removal. Unfortunately, these compounds also cause sleeking, so their
use requires a delicate timing balance. The edge bevels of CaF2
must even be smoothed or polished to minimize edge chipping effects. We have
also found pad-polishing speeds up the process as compared to conventional
pitch laps. This is because polishing compounds stays on the surface of the
polyurethane pads, whereas it becomes embedded in pitch laps.
Testing
Extensive
effort is also being expended to improve the testing of excimer grade CaF2
optics, especially under the conditions encountered in microlithography. A
group at MIT's Lincoln Laboratory (funded by Sematech), led by Mordechai
Rothschild and Vladimir Liberman, is at the forefront of these efforts.
|
Schematic of the MIT
Lincoln Lab excimer laser optics test setup.
figure 2
|
In
the Lincoln Lab test setup (figure 2), the combined output from two excimer
lasers operating at 193 nm is split into a total of 12 separate beams. Each
beam is then sent through three 80 mm long samples; both CaF2
and fused silica are tested simultaneously. A long path length is helpful
because this accentuates bulk material effects, making them easier to separate
from surface effects. Each laser operates at 400 Hz, but by combining their
output, an overall repetition rate of 800 Hz is attained; approximately 70
million pulses are delivered over the course of a day. Testing is not conducted
seven days a week (due to maintenance downtime), so the total exposure is about
one billion pulses per month. Exposure fluences range from 0.3 mJ/cm2
to 3 mJ/cm2
per pulse along various beam lines.
One
of the most interesting results obtained so far is that transmission actually
increases after an initial period of irradiation, a phenomenon termed "surface
transmission recovery." This occurs because the excimer laser is actually
ablating some of the contaminants left on the surface from the polishing
process (hydrocarbons adhere well to the surface of CaF2
, making it harder to clean than glass).
Testing
conducted over the past two years has helped manufacturers to achieve a steady
improvement in material quality; the bottom line result is that transmission
degradation of excimer grade CaF2
is not a major problem at fluences below 1 mJ/cm2
per pulse for today's material. At higher fluences, results are less clear.
Because the testing conditions cannot be accelerated, the group also hasn't
been able to accurately quantify total lifetime performance yet. In fact,
probably the most important conclusion drawn from the testing to date is that
initial material measurements do not necessarily correlate with extended use
performance. Consequently, the consumer cannot simply purchase components for
excimer laser applications based on manufacturer's nominal specifications. The
long term characteristics of the product must be established.
Microlithography
advances have created an increasing demand for high volumes of large, high
quality excimer laser optics. The benefits of the innovations made to meet this
need are trickling down to other applications for deep UV optics as
manufacturers learn to produce better components with improved economy.
However, testing to date has proven that the end user must look beyond vendor's
printed specifications to understand long term performance.
Figure
Captions
1
Schematic of the Stockbarger CaF2
fabrication method.
2 Schematic of the MIT Lincoln Lab excimer laser optics testing setup.
|