Facility for Pulsed Extreme Ultraviolet Detector Calibration
S. Grantham, R. Vest, C. Tarrio and T.B. Lucatorto
National Institute of Standards and Technology, Gaithersburg, MD
Abstract. Extreme ultraviolet lithography, operating at a wavelength of about 13.5 nrn, will require high-repetition-rate
pulsed sources with average power outputs of over 100 W. As the output of the sources approaches this level, the lowpower, continuous-wave calibrations of the detectors used to determine the source output will no longer be valid. The
National Institute of Standards and Technology (NIST) is currently in the process of commissioning a system for the
calibration of Extreme ultraviolet detectors under pulsed conditions similar to those produced by potential sources for
extreme ultraviolet lithography. This system incorporates a laser-produced plasma source based on a Kr droplet target.
The plasma's output is collimated by a grazing incidence parabolic reflector and split by a Mo/Si multilayer beamsplitter
to provide two channels of illumination. Each channel is refocused onto filtered detectors to provide two signals that are
proportional to the plasma Extreme Ultraviolet (EUV) output. Calibration is done by comparing a detector under test to
a detector calibrated at NIST's Synchrotron Ultraviolet Radiation Facility. In this paper we will describe this calibration
facility and the present performance of the two-channel normalization scheme. In addition, conversion of continuous
wave to pulsed calibration will be discussed.
INTRODUCTION
With the ongoing commercialization of EUV
lithography (EUVL), light sources are being developed
to meet the rigorous needs of an integrated circuit
production facility. These needs include high average
EUV power, currently predicted to be >100 W, and a
reliable stable output. The light sources currently
under consideration are multi-kilohertz pulsed, high
peak-powered devices relying on Xe plasmas as an
EUV emitter. This differs greatly from the controlled
conditions present when EUV detectors are calibrated.
EUV detector calibrations typically incorporate a
synchrotron light source which produces radiation that
is quasi-cw because of its high repetition rate in the
100's of MHz. In addition, calibrations are typically
done at both low average and peak powers. A
calibration made under these conditions may not
accurately reflect the performance of a detector under
the conditions present in a potential source for EUVL.
In fact, we have recently found that the responsivity of
a photodiode can falloff well before the onset of largescale saturation effects.1 Furthermore, the electronics
used for recording the output of detectors under pulsed
conditions may vary greatly from those used during
calibration. Amplifiers, AC coupling and reverse
biasing are all typically incorporated under pulsed
detection schemes but are not present during
calibration. Because of this discrepancy of operating
conditions, we are commissioning a system based on a
pulsed EUV source similar to those being considered
for EUVL. This system will be used to calibrate EUV
detectors under pulsed conditions with similar
electronics to those utilized with pulsed sources.
EXPERIMENTAL SETUP
The first part of the system is the light source. The
design of the source is based on systems at Sandia
National Laboratory and the University of Maryland.2'3
The excitation is provided by a Nd:YAG laser with
800 mJ/pulse and approximate pulse length of about
10 ns. Clusters of Kr gas are introduced to the vacuum
through a supersonic expansion. The expansion is
provided by a pulsed valve, specially designed nozzle,
and 2.76 MPa backing pressure of Kr. With roomtemperature gas, submicron clusters are provided by
this scheme. The pulsed valve may be cooled using a
temperature-controlled liquid-nitrogen-filled jacket to
provide larger droplets, which yield a higher
conversion efficiency.
The laser runs at 10 Hz,
however, due to the high heat load on the cryopump,
the valve is run at 1 Hz. This leads to a background
pressure of 10 mPa, which is acceptable for our
purposes.
CP683, Characterization and Metrology for VLSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
2003 American Institute of Physics 0-7354-0152-7/03/$20.00
444
The incoming laser light interacts with Kr clusters
or droplets to form a plasma near the nozzle. A single
grazing incidence parabolic mirror collects light from
the plasma. The collecting mirror is aligned so that the
plasma forms at the focus of the mirror, resulting in a
collimated beam exiting the collection mirror. The
collimated light is relayed onto a multilayer
beamsplitter composed of two Mo/Si mirrors which
form an inverted roof mirror. This beamsplitter serves
two functions: it monochromatizes the EUV light and
splits the collimated beam in two. Parabolic mirrors
that are identical to the collection optic then focus the
two outputs from the beamsplitter. The result is two
separate EUV images of the plasma. One image or
channel is monitored by a Zr-coated photodiode at all
times.
The Zr filter effectively attenuates the
fundamental laser light by a factor of 10~6 and also cuts
out long wavelength UV that is reflected by the
beamsplitter.4
A picture of the vacuum chamber
containing the aligned EUV optics and detectors is
shown in figure 1.
FIGURE 1. Photograph of the EUV optical setup for
calibration of EUV detectors under pulsed conditions. The
image shows a plasma position and the collector,
beamsplitting and condensing optics along with the detectors
used in both channels.
To calibrate the response of a detector, one must
have a working standard and understand the behavior
of both the standard and the detector under test (DUT).
Standards are calibrated at the NIST EUV Detector
Radiometry Beamline at the Synchrotron Ultraviolet
Radiation Facility (SURF III). This facility carries
out calibrations with quasi-cw synchrotron radiation at
power levels on the order of nanowatts.5'6 The
conditions for detector calibration at SURF III are
significantly different than those used in the system
described here and in all other pulsed EUV systems.
The performance of EUV photodetectors under pulsed
conditions has been a matter of increased study
recently1'7 and work is ongoing at NIST to determine
the effects of intensity, pulse duration and bias
electronics on the performance and saturation behavior
of photodetectors. Currently, this system uses the
direct transfer of a cw calibration of a working
standard to a detector under test, this is a relative
calibration and doesn't address any changes in the
responsivity of the working standard due to the pulsed
nature of the radiation used in the calibration. Below
we will discuss plans for modifying this technique to
address differences in pulsed and cw detector
performance.
the signals from both channels reflect any changes
from shot to shot due to plasma nonuniformity or Kr
gas transmission. To evaluate the performance of this
technique a trial was done to examine the
proportionality of the two channels.
The system was aligned with a filtered
photodetector at the focus of each channel. The laser
was run for 1000 shots with varied output energy.
This generated a pulsed Kr plasma with a varying size
and EUV output. The results of this trial are shown in
figure 2. The first graph shows the collected charge of
the working standard channel vs. the collected charge
in the normalizer channel. Looking at the data plot a
majority of the deviation from the linear fit is at the
extremes of the data. The deviation on the low-level
signal end of the data is from small signal noise due to
operation in a plasma environment. The deviation
from the linear fit on the high-level signal end is due to
the saturation of the detector in the normalizer channel
of the setup. Despite these limitations this plot is fitted
with a linear function with good agreement. The
residuals of this fit are shown in the second graph, and
show that the residual has a normal distribution with a
relative standard deviation of 1.5% from the linear fit.
We feel increased electrical shielding of detectors
will help reduce the uncertainty of this experimental
setup. In addition, a new alignment technique is
currently being developed, which will further optimize
the alignment of the parabolic mirror setup. Small
misalignments can cause portions of the collecting
optic to be illuminated nonuniformly and will
PRELIMINARY RESULTS
One of the key features of the design of this
calibration system is that both the normalizing and test
channels use the same EUV image of the source. This
minimizes the uncertainty of the calibration because
445
Integrated Charge Normalization
Integrated Charge Residual
40
r y = 0.0472480052 + 0.58395039x I R= 0.99907297
35
I 30
o
25
20
10
0
0
10
20
30
40
50
60
-0.1
70
Normalizer Charge (nC)
-0.05
0
0,05
Normalized Residual
0.1
FIGURE 2. Plots of results from proportionality trial. Plot on the left shows collected charge of the working standard channel
versus the collected charge of the normalizer channel along with linear fit of the data. The plot on the right shows the residuals
from the linear fit to the data. The result is a normal distribution with a standard deviation of 1.5%.
therefore reduce the proportionality in the signals from
the two channels. Alignment will be done using a
reflective ball placed with its center coinciding with
the center of the plasma. This ball can be placed by
using a Hartmann8 technique with the focused light
from the Nd:YAG laser, which initiates the plasma.
By placing an opaque plate with a patterned array of
holes in the collimated output of the laser, a collimated
output pattern is produced that follows the original
path of the laser. When the center of the reflective ball
is placed at the focus of the laser (which also coincides
with the plasma position), the reflected light is retroreflected back towards the incoming laser light.
Adjustment of the reflective ball so that reflected
pattern coincides with the hole-pattern on the plate
ensures that the center of the ball lies at the focus of
the laser and the plasma position. Using the same
Hartmann technique, a focused cw laser beam can be
aligned to the ball so that when the ball is removed,
the focus of the cw laser coincides with the focus of
the Nd:YAG laser and the diverging light fills the
aperture of the EUV collection optic. Collimating the
reflected laser light with the collection optic will then
ensure alignment of the focus of the optic with the
center of the plasma. This collimated light can then be
used to align the beamsplitter and condenser optics.
This new alignment technique will minimize the
uncertainty due to these aberrations and optimize the
throughput of the system.
FUTURE GOALS
This system uses a transfer standard that has been
calibrated on a synchrotron radiation source whose
emission is similar in wavelength but very dissimilar
in temporal characteristics. For this reason we are
currently conducting experiments which not only
address the temporal dependence of a photodetector's
performance but also its saturation behavior under a
variety of conditions. This work is designed to
complement work which we have previously done
comparing CW and pulsed response with visible light
illumination.1
We will further examine the
performance of photodetectors under various intensity
conditions with pulsed radiation to determine the
spatial dependence of saturation behavior. In addition,
experiments are underway to determine the detector
performance with higher bias voltages than considered
before. Also, the dependence of saturation behavior
on indirect and direct bandgap transitions will be
explored.
With this research we plan to develop a model that
will be an accurate predictor of detector performance
under various powers, spot-sizes, pulse durations and
bias conditions. This will allow a detector user to
accurately measure an EUV output with a detector
which has a cw calibration despite operating
conditions that differ from the conditions present
during the calibration of the device. Finally, we are
currently implementing an absolute cryogenic
446
radiometer (ACR)9 for the calibration of detectors in
the EUV.
This will significantly reduce the
uncertainty of EUV detector calibrations done at
SURF III, which currently utilize noble gas
transmission as a reference standard.
3. Parra, R, McNaught S.J. and Milchberg H.M., Rev. Sci.
Inst. 73 (2), 468-475 (2002).
ACKNOWLEDGMENTS
4. Berger, K. W. and Campiotti, R. H. "Absolute dosimetry
for extreme ultraviolet lithography", Proc. SPIE vol.
3998, 2000, 838-845 (2000).
extreme ultraviolet lithography," Emerging Lithographic
Technologies III, edited by V. Vladimirsky, Proc. SPIE
vol. 3676, pp. 669-678 (1999).
The authors would like to thank our NIST
colleagues Keith Lykke, and Maria Dowell for their
helpful advice in the construction of the facility
described here. In addition, we would like to thank
Enrique Parra, Stuart McNaught, and Howard
Milchberg of the University of Maryland for access to
their experience and laser-produced plasma data as
well as their aid in the construction of the pulsed valve
and it's temperature control system. This system
would not have been possible without their help.
5. Furst, M. L, Graves, R. M., Canfield, L. R., and Vest, R.
E., Rev. Sci. Inst. 66, 2257-2259 (1995).
REFERENCES
8. Optical shop testing^ edited by D. Malacara, Wiley
(1992).
6. R. E. Vest, L. R. Canfield, M. L. Furst, R. M. Graves, A.
Hamilton, L. R. Hughey, T. B. Lucatorto, and R. P.
Madden, "NIST programs for radiometry in the far
ultraviolet spectral region," in Ultraviolet Atmospheric
and
Space
Remote
Sensing:
Methods
and
Instrumentation //, edited by G. R. Carruthers and K. F.
Dymond, Proc. SPIE 3818, 15-26 (1999).
7. R. Stuik, and F. Bijkerk, "Linearity of P-N junction
photodiodes under pulsed radiation," Nucl. Inst. Meth. A
489/1-3,370-378(2002).
Vest, R. and Grantham, S., submitted to Applied Optics.
9. Shaw, P., Lykke, K., Gupta, R., O'Brian, T, Arp, U,
White, H., Lucatorto, T., Dehmer, J., and Parr, A., Appl
Opt. , 3 8 , 1 , 18-28(1999).
2. Kubiak, G. D., Bernardez, L. J., Krenz , K., and . Sweatt,
W. C, "Scale-up of a cluster jet laser plasma source for
447