|
ADVANCES IN DEEP
UV OPTICS
Improvements in
UV materials, polishing, coating and testing are being made to meet
stringent application demands.
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/
Optics for the
deep ultraviolet (below 250 nm) is a rapidly growing market, driven
primarily by excimer laser based applications such as microlithography,
micromachining, thin film transistor (TFT) annealing, photorefractive
keratectomy (PRK) and Fiber Bragg Grating (FBG) writing. In many of these
uses, the components experience moderate to high fluences over very large
(multi-billion) pulse counts. Manufacturing high precision optics that can
withstand these conditions presents a number of unique challenges. This
article reviews the latest advances in UV materials technology, optical
polishing and thin film coating that have been developed to overcome these
difficulties.
Materials
One of the main
obstacles to working in the deep UV is the limited number of substrate
materials that are transmissive in this spectral range. The only two
materials that offer a desirable combination of optical, mechanical and
chemical characteristics are fused silica, which transmits down to about
190 nm, and calcium fluoride (CaF2), which can be
used at wavelengths as short as 130 nm. Magnesium fluoride
(MgF2) also operates
in this range, but it is birefringent, which significantly limits its
utility for transmissive optics.
Fused
Silica
Corning
Incorporated (Corning, NY) is a leading manufacturer of fused silica
material produced using the flame hydrolysis process. In this method,
silicon-containing starting materials are passed through a hot flame to
yield pure, amorphous silica glass. Once a boule is produced, it is then
cut up and the pieces graded with respect to parameters such as spectral
transmission, inclusion cross section and index homogeneity. The flame
hydrolysis approach provides a number of advantages for the deep UV,
including high purity (which affects transmittance), a large processing
latitude (so that glass properties can be adjusted) and the ability to
produce large boules of glass (>48 inch diameter and several inches
thick). The process also allows for careful control of glass
stoichiometry, resulting in a material minimized with respect to oxygen
deficiency or excess. Both of these conditions can impact transmittance
and laser damage resistance for microlithography.
Corning has
intensified its efforts in metrology of deep UV materials, primarily in
response to the needs of the microlithography industry, which now utilizes
KrF excimer laser sources at 248 nm, with plans to migrate to ArF lasers
at 193 nm. The ability to make more exacting measurements enabled Corning
to introduce an "excimer grade" material in 1997.
Corning has also
devoted substantial effort to understanding the response of silica glass
to prolonged laser beam irradiation, such as is encountered in
microlithography. The two most important observed effects are color center
formation and densification. Color centers are lattice defects that absorb
light. Densification is a structural rearrangement in the glass that
results in an actual physical contraction after prolonged exposure to
laser light. Both of these can result in image degradation.
This research
has yielded a good qualitative understanding of the factors controlling
color center formation. Specifically, the presence of molecular hydrogen
in the glass has been identified as a key parameter. The densification
process has been found to be intimately dependent on the precise specifics
of the exposure parameters (both sample shape and beam geometry). It has
even been determined that the phenomenon evolves according to a power law
relationship that includes the total number of pulses, the fluence and the
integral square pulse width.
Calcium
Fluoride
CaF2 is a
crystalline material with a transmission that can range from 130 nm to 10
µm. However, the actual deep UV transmission is highly dependent upon
material purity. This was not a problem in the past, as most applications
for CaF2 were in the
infrared. With the emphasis shifting to the UV, material growers such as
Bicron (Solon, OH) have put significant energy into producing higher
purity materials, as well as increasing damage threshold and index
homogeneity, while reducing fluorescence and stress birefringence.
|
Schematic
of the Stockbarger technique for producing CaF2
figure
1
|
Bicron produces
synthetic CaF2 using a
modified form of the Stockbarger method. This involves slowly lowering a
crucible containing molten chemical precursors through a "freeze-zone"
where crystallization takes place (figure 1). The entire process typically
takes 6-8 weeks to produce an ingot.
In order to
improve purity (which also affects damage threshold, color center
formation and fluorescence), Bicron has implemented a number of testing
methodologies (VUV and UV/VIS spectrophotometry, fluorescence
spectroscopy, etc.) to detect contaminants during synthesis of the
precursor materials. These efforts over the past year have enabled Bicron
to lower typical contaminant levels from 10 ppm to under 0.5 ppm.
Further material
purification is performed during the crystal growth process. Chemical
"scavengers" are placed in the crucible which react with any remaining
impurities. While the use of chemical scavengers has been part of the
Stockbarger technique for some time, Bicron has developed a number of new
chemicals whose reaction produces gaseous by-products. These are then
easily removed from the process by vacuum pumps.
Amongst Bicron's
modifications to the traditional Stockbarger furnace are changes to the
freeze-zone to enable tighter control over crystal growth. This also
allows the use of more sophisticated process control instrumentation for
better regulation of temperature, vacuum and the annealing process, which
affect material homogeneity and stress birefringence. The results of these
efforts have been a reduction in the level of stress birefringence from 20
nm/cm (a typical value for the industry) to just 1 nm/cm. In addition,
these innovations have also enabled Bicron to produce larger ingots. These
can now be as large as 36" in diameter by 6" in length; such large sizes
are required for some semiconductor microlithography lenses.
Polishing
Optical
polishing of high damage threshold laser optics for the deep UV also
presents several challenges. From a purely mechanical point of view, it
should be noted that a l/10 flatness specification at 193 nm (which is a
typical value) is over three times tighter than a l/10 flatness at 633 nm.
Due to diffraction, surface imperfections have more impact at shorter
wavelengths, and a surface quality (scratch and dig) specification of 10-5
is also not unusual for these components.
|
Schematic
of subsurface damage.
figure 2
|
Subsurface
damage (SSD) has been identified by researchers at Lawrence Livermore as a
major factor in limiting laser damage resistance. Subsurface damage
consists of fractures and scratches that occur during the grinding and
polishing process which become partially or totally hidden by the
polishing redeposition layer (figure 2). The polishing redeposition layer
is a thin layer of material that flows while the material is being worked
and covers the surface. It is theorized that SSD lowers damage threshold
by reducing fracture strength, providing a place for light absorbing
contaminants to reside, allowing atoms at or near the fractures to be more
easily ionized (by changing their chemical or electronic environment) and
causing local modulations in the electromagnetic field. SSD is minimized
by using a sequence of successively finer grinding and polishing steps,
and ensuring that each step removes sufficient material to eliminate any
damage caused by the previous step.
Another
potential source of laser damage is residual polishing compound that may
be left on the surface of a part, incorporated into the surface layer
itself, or deposited in microfractures and surface defects. Especially in
the latter two situations, such residual material is virtually impossible
to remove through cleaning. The damage then occurs when this polishing
material absorbs light and heats up. Ceria (CeO2), one of the
most commonly used polishing compounds, absorbs strongly at short
wavelengths, making its use particularly problematic. As a result, Alpine
Research Optics (ARO) utilizes alternative polishing compounds, which are
less absorbing at short wavelengths, for the fabrication of deep UV
optics.
Polishing of
CaF2 in particular
is rendered even more difficult by the fact that it is anisotropic,
hygroscopic and very prone to chipping and fracturing. Small particles
break loose from the edges and are then dragged over the soft surface of
the component during polishing leading to sleeks (sleeks are small surface
scratches). Thus, meeting the tight surface quality specifications (10-5)
typical of UV laser optics systems usually requires 50% to 100% longer
polishing times than for fused silica. But unfortunately, the longer the
polishing time, the more difficult it is to hold a given flatness or
surface figure. The bottom line is that fabricating CaF2 components
requires a very delicate balancing act between the various process
parameters (polishing time, spindle speed, spindle pressure, etc.) in
order to simultaneously meet both shape and surface quality
specifications.
Every
manufacturer involved in fabricating CaF2 components has
evolved their own techniques for working with the material. At ARO, we use
pad polishing, as opposed to conventional pitch laps, because we have
found that these speed up the process, making it easier to achieve tight
surface quality specifications.
Probably the
most important step we have taken, however, is to environmentally isolate
the polishing area for CaF2 from the rest
of our production area. This prevents the possibility of
cross-contamination between our standard polishing procedures and those
for CaF2.
Thin Film
Coating
One of the
difficulties in producing coatings for use with deep UV excimer lasers is
that damage mechanisms are still not completely understood. With high
power Nd:YAG lasers, damage usually refers to catastrophic failure; this
means that part of the film is actually destroyed or burned. The
occurrence of this damage is directly related to peak beam fluence. In
contrast, coatings exposed to high pulse counts of moderate fluence
excimer laser beams usually experience a gradual decrease in performance
before the onset of catastrophic failure. This type of damages appears to
be related to total accumulated flux, the operating atmosphere (due to
photo-oxidation) and other still unidentified factors. Because damage is
not a well defined event, it has become standard in both the
microlithography and PRK industries to define a 5% decrease in
reflectivity as "failure" for a mirror.
The primary
challenge has been at 193 nm, where, once again, efforts are hampered by
the small selection of transmissive materials. At ARO, this has led us to
utilize fluoride materials for both the high and low index layers at this
wavelength. Furthermore, we've been able to determine that several
operational parameters are critical for the production of long lived
coatings. Variations in temperature, pressure and deposition rate can
cause fluoride materials to deposit with inconsistent refractive index,
resulting in spectral performance problems or mechanically unstable films.
Accurate process data logging for each coating run, combined with actual
lifetime and damage results from end users has enabled us to optimize the
parameters of the coating process.
We have also
found that conventional e-beam evaporation is the most effective method
for producing these coatings. Ion-assisted deposition, which has proved
capable of delivering higher density, higher damage threshold coatings at
longer wavelengths, has drawbacks in the deep UV. Specifically, when used
with fluoride materials, this technique can result in the deposition of
oxides that are absorptive at short wavelengths; any absorption in the
film dramatically reduces its lifetime.
The results of
our efforts are demonstrated in testing conducted by a Sematech funded
group at MIT Lincoln Laboratory. The charter of this team, led by Dr.
Mordechai Rothschild, is to characterize many of the components of next
generation microlithography systems, including optical materials, optical
coatings, photoresists and photomasks (reticles). In the latest results of
their ongoing testing, one of ARO's 193 nm high reflectors has withstood
over 1.2 billion pulses at a fluence of 15 mJ/cm2/pulse without
showing any significant decrease in reflectivity. In a second test, an
antireflection coated optic has accumulated over 1.8 billion pulses at 4
mJ/cm2/pulse with no
detectable change in performance.
Conclusion
Just a few years
ago, optics for 193 nm that could withstand multi-billion pulse exposures
were a dream. Intense development efforts by materials producers, optics
fabricators, coating houses, laser manufacturers and industry research
groups have now made them a viable, commercial reality. As the needs of
cutting edge, deep UV applications continue to expand, expect optical
components to keep pace.
Figure
Captions
1. Schematic of
the Stockbarger technique for producing CaF2.
2.
Schematic of subsurface damage.
|