EUV Engineering Test Stand EUV Engineeri
EUV Engineering Test Stand EUV Engineeri
UCRL-JC-137668
February14, 2000
U.S. Department
of Energy
Lawrence
Livermore
National
Laboratory
/
This document was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor the University of California nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for
the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or
represents’that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not
necessarily constitute or imply its endorsement, recommendation, or favoring by the United States
¯ Governmentor the University of California. The views and opinions of authors expressed herein do not
necessarily state or reflect those of the United States Government or the University of California, and
shall not be used for advertising or product endorsement purposes.
This is a preprint of a paper intended for publication in a journal or proceedings. Since changes may be
made before publication, this preprint is made available with the understanding that it will not be Cited
or reproduced without the permission of the author.
OR
ABSTRACT
The Engineering Test Stand (ETS) is an EUVlaboratory lithography tool. The purpose of the ETSis to demonstrate EUV
full-field imagingand provide data required to support production-tool development.The ETSis configured to separate the
!magingsystem and stages from the illumination system. Environmentalconditions can be controlled independently in the
two modulesto maximizeEUVthroughput and environmental control. A source of 13.4 nmradiation is provided by a laser
plasma source in which a YAGlaser beam is focused onto a xenon-cluster target. A condenser system, comprised of
multilayer-coated mirrors and grazing-incidence mirrors, collects the EUVradiation and directs it onto a-reflecting reticle. A
four-mirror, ring-field optical system, havinga numericalaperture of 0.1, projects a 4x-reductionimageonto the waferplane.
This design corresponds to a resolution of 70nmat a kl of 0.52. The ETSis designed to produce full-field imagesin step:
and-scan modeusing vacuum-compatible,one-dimension-long-travel magnetically levitated stages for both reticle and wafer.
Reticle protection is incorporated into the ETSdesign. This paper provides a system overview of the ETSdesign and
specifications.
Keywords:EUVL,lithography, multilayers coatings, optical fabrication, optical design, laser-produced plasma, laser plasma
source, maglev,magneticlevitation, stages, precision engineering
i. INTRODUCTION
ExtremeUltraviolet Lithography (EUVL)is one of the leading candidates for patterning semiconductordevices during the
next decade. The developmentof multilayer coatings ~ for wavelengths between 1 lnm and 14 nmenables the design Of all-
reflecting systems that support multiple generations of microlithographytechnology, capable of feature sizes downto 30 nm.
The development of EUVLtechnology is funded by the EUV, Limited Liability Company(LLC), a consortium
semiconductor manufacturers founded in 1997 and comprised of Advanced Micro Devices, Intel and Motorola. The
Department of Energy Virtual National Laboratory, comprised of Lawrence Berkeley National Laboratory, Lawrence
LivermoreNational Laboratory and Sandia National Laboratories, perform research, development,and engineering leading to
the commercialization of EUVL.A major componentof this effort is the development of the EUVEngineering Test Stand
(ETS), an alpha-class tool designed to demonstrate full-field EUVimaging and provide data to participating equipment
manufacturers to support production-tool development.
EUVimages with resolution as high as 50 nmhave been demonstrated using the 10x microstepper.2 This tool, designed for
proof-of-principle experiments, utilized a simple condenser, small-field projection system and low powerEUVsource. The
ETSincorporates major advances in all subsystems including: a high-power laser-produced-plasma source, a large-solid-
angle condenser, an advancedring-field projection system, predsion scanning stages and an advancedenvironmentalsystem.
* Correspondence: Email: [email protected]; Telephone: 925 294 2137; Fax: 925 294 3870
provides an environmental barrier between the illuminator and main enclosure. The six beams are combined and further
conditioned by a final grazing-incidence element, C4, to illuminate at the reticle and to achieve an effective aperture fill
factor of 0.7.
Condenser Collector, C1
element 04
Pro
system
Gas-jet
assembly-
Laser-produced
plasma
Waferstage
Condensing
¯ optics, C2& C3
Main
enclosure
Spectral
:: purityfilter
Illuminator enclosure--
The ETS projection system, a four-mirror, ring-field design, projects an image of the reflecting reticle onto the wafer within a
print field 1.5 mmwide by 24 mmlong. The projection optics are designed to print 4x-reduction, full-field images in step-
and-scan mode, using a numerical aperture of 0.1. The ETS design requirement for printing 100 nm features is satisfied by
this optical design, which also supports the printing of 70 nm features at a kI of 0.52. The projection system is kinematically
mounted on a rigid metrology frame of invar which is carried on a set of three active/passive isolators to protect it from
ground vibrations.
The reticle and wafer are electrostatically clamped and carried on magnetically-levitated (maglev) stages, which are mounted
on a grounded structure. This design assures that the reaction forces, resulting from stage acceleration, are shunted to ground
without exciting modes in the projection system or support structures. Each stage controls 5 degrees of freedom over short
travel ranges and one long-travel degree of freedom in the scanning direction. A laser metrology system feeds the positions
of the wafer and reticle platens to a high-speed digital control system, designed to achieve precision scanning, while isolating .
the suspended reticle and wafer from ground vibrations induced in the stage bases.
The environmental enclosure is comprised of two primary components, the illuminator enclosure containing the source and
condenser and the main enclosure containing the C4 condensing element, the projection optics box and the scanning stages.
Within the main enclosure special seals create three separate zones: the reticle zone containing the reticle stage, the optics
zone containing the projection optics and stage metrology, and the wafer zorie containing the wafer stage. This configuration
provides the flexibility to implement environmental conditions designed to protect the optics from carbon contamination due
to cracking of hydrocarbons on the mirror surfaces and to protect the reticle from particulate contamination.
. (9)
(9)
sop,do
aesuepuoo
selnpom Jese’l
.<
eeaqj,uJoal M 00/.t.
(9)
uJeeBaese-I
pexeldBlnlN
To achieve the required EUVsource power and repetition rate, a 1700 Wlaser driver has been developed by the TRW Inc.
and integrated with the continuous jet target. The laser is a diode-pumped,solid state Nd3+:YAG laser comprised of three
identical laser modules, each of which produces 570 Wof time-averaged 1.06 pmpowerat repetition rates of 166%2000 Hz.
The beamsfrom the three modulesare multiplexed to deliver 285-342mJ/pulse at a combinedrepetition rate of 5000-6000
Hz, or can be synchronizedto deliver 900 mJ/pulse at a repetition rate of 1667-2000Hz. Pulse duration for each individual
chain is -6 ns FWHM and the beamquality, as determined by the focal spot size in the far field, is 1.1 - 1.2 times the
diffraction limit. Laser beampointing is actively controlled at 6000 Hz, achieving a total combinedbeamdrift/jitter of
approximately0.25 times the diffraction-limited spot size. + Output powerstability is also excellent, exhibiting root-mean-
square deviations ranging from 0.6-0.8%, depending on output power.
Wavelength 13.4 nm
Total power incident on 4.4 W over 1.8 sr and 2.5% spectral
condenser bandwidth
Power deliveredto reticle >__0.76
W
Repetitionrate >3000 Hz
Illuminationformat 96 mmarc, <6 mmwide
Partial coherence 0.7
Illuminationuniformity +2%
Illumination stability 1%1-sigma
Sourcemotion < +25t~m
Condenser lifetime <10%reflectanceloss after 10l° pulses
The 1700Wlaser driver has been integrated with the continuous jet source and optimization of the EUVyield, and the nozzle
thermal performanceis in progress. Under1260 Wof incident laser irradiation, the laser-to-EUVconversion efficiency at
13.4 nmis measuredto be 0.54 mJEuv/Jtaser-eV-sr, yielding an output powerof 9.8 Watts integrated over a 2.5%spectral
bandwidthand 2x steradians. As shownin Fig. 3, this represents an increase in source powerby a factor of morethan 7.5
during the past year. Efforts are in progress to satisfy the source powerrequirementfor the ETSby improvingthe laser beam
delivery efficiency to increase laser powerdelivered to the jet target to 1500Wand also by reducing the attenuation of
plasma source emission caused by residual Xe in the source chamber.
The jet assembly has been designed to keep the nozzle throat and mountingassembly cool during operation at full laser
power. In addition, materials used for plasma-facing nozzle and vacuumhardware are chosen to minimize the rates of
erosion and subsequent deposition of eroded material on nearby multilayer-coated optics. Experiments to establish the
reflectance lifetime of the plasma-facing first condenser element as functions of laser pulse energy, average power, and
nozzle-laser separation are in progress.
10
100
o Lase/
200W~
> 0.1
LU
45
2y- W~
~
ka~j
001. .............................
1995 1996 1997 1998 1999 2000 2001
Year
Figure 3. Recentprogress in EUV powergeneration. (solid triangles). In 1999the EUVpowerwas increased by morethan
factor of 7.5 throughthe integration.of the TRW1700Wlaser with the AEScontinuousXejet source. TheeventualEUVpower
requirement(opencircles) for the ETSis 15 Wintegratedover2~ sr and 2.5%spectral bandwidth.
The condensermechanicaldesign (See Fig.4) must maintain the positions of the 19 condenser mirrors to translational
tolerances on the order of 60 micronsand angular tolerances of 40 microradians. Additionally, the C1 assemblyand the C3
assemblymust be easily replaced since they are consumablein the ETSoperation. The stability requirements were met by
isolating the entire condenser from the walls of the vacuumchamber,to minimizemotion from pump-down and to mitigate
acoustic and pumpvibrations. To minimizethermal distortion of the optics, the C1 assemblyis water-cooled and the
individual substrates are madeof silicon for high thermal conductivity. The C3 assemblyhas passive cooling.
The C 1 and C3 assemblies are kinematically mountedand readily replaceable. Pre-alignment stations were designed,
fabricated, and tested in parallel with the illuminator fabrication. Thesealignmentstations are used to pre-align the C1and
C3 assemblies relative to their kinematic mountswhile the illuminator remains operational. As a result, replacing a C1 or C3
assemblyis analogousto replacing a single optic. The pre-alignment of the assemblies has been demonstratedand no re-
alignmentof the illuminator is required whenthese optics are replaced.
Laser beam
C4 mirror "~b~ ~
Laser beamdelivery
& diagnostics
C1 Assembly
Jet Assembly
C2 Assembly
Spectralfilter C3 Assembly,
Assembly
Successful operation of the illuminator requires simultaneousOperationof the laser producedplasmasource, alignmentof all
19 elementsof the condenseras well as the spectral purity filter, and environmentalprotection for the optics.
4. PROJECTION SUBSYSTEM
The ETSprojection systemis designed to demonstratethe printing of full-field imagesat 100 nmresolution in step-and-scan
mode.Having a numerical aperture of 0.1, the projection" system also meets the requirements for printing of 70 nmdense
lines and Spacesat a ki value of 0.52.
The projection optics is a 4-mirror ring-field design5’6, with a 4x reduction, that is shownschematically in Fig. 6. The
numerical aperture is 0.1, which readily supports printing 100 nmdesign rules at an EUVwavelength of 13.4 nm, The
optical design consists of four mirrors labeled M1, M2,M3,and M4in Fig. 6. Three of the mirrors are aspheric and M3is a
sphere. The aperture stop is located at M3. The overall length from the mask plane to the wafer plane is 1075 mm.The
design is rotationally symmetric, and the well-corrected field is an annulus extending over a radial range from 208 mmto
214 mmin the maskplane. The used field is a 30" arc of this annulus, whichcorresponds to a chord length of 104 mmat the
mask. Imageswill be printed by synchronously scanning the wafer at a quarter of the velocity of the mask, so that the
movingimage will appear stationary on the wafer. Of the full rotationally symmetricaspheric parent, only a region around
the actual used clear aperture was manufactured, to minimize the weight of the substrates and improve their mounting
stability.
o
M1
I Wafer
200 mm
The optical design was optimized with the aim of simultaneously satisfying several design constraints. These design goals
included wavefrontquality, reduction ratio, telecentricity, field curvature, imagedistortion, aspheric departure, and control of
incidence angles on the mirrors. This last constraint was to allow uniform, rather than graded, multilayer coatings on the
mirrors, and was achieved by using one multilayer d-spacing on the aspheres and a slightly larger d-spacing on M3wherethe
average angles of incidence are higher. The aberration correction is dominated by the behavior of astigmatism. By
controlling the high-order aspheric terms on the mirrors, both 5th and 7th order astigmatismwas brought to zero at the center
of the ring field. The design wavefronterror at this point is 0.004 wavesand 0.020 wavesat top and bottomof the ring-field,
for a wavelengthof 13.4 nm. The distortion was controlled so that the magnification as a function of field radius has a
stationary point at the center of the ring field. This "balanced"distortion minimizesthe blurring of the scannedimagecaused
by imagedistorion. It is found that although the static distortion field has a maximum magnitudeof 16 nm, scanning through
this field leads to a scan blur that reduces the Strehl ratio by only 0.007, with a residual dynamicdistortion of less than 2 nm.
Thedesign is telecentric at the wafer, but since the maskis reflective, the imagingat the maskis necessarily oblique. This
characteristic imparts stringent flatness requirements on the mask, but it does allow a residual magnification control of
+20 ppmto be achieved simply by varying the conjugate plane locations.
The projection optics box (POB)is madeupof an assemblyof four optics each supported in optic cells, 10 remotely actuated
adjustments, a main structure, a host of sensors and a control system. The mechanical design of the POBis driven by the
need to achieve and maintain precise alignment amongthe optics and to isolate the optics from forces and movements that
woulddistort their figure. The design philosophy attempts to keep the POBimagingsystem completelypassive other than the
actuated adjustments, the objective being to avoid active thermal or mirror positioning control. This is achievedby designing
for high resonant frequencies, using low CTE(coefficient of thermal expansion) material, and taking advantage of the long
thermal time constants inherent with large thermal masses and radiative heat transferjbetween the POBand ETSchamber.
~ udials
|
o
M2Substrate I Prel°ad spring
Button attached to
~t
rate
Bipod flexure
4.2.2.
n Projection Optics Box Structure Desig
The POBstructure appears in Figure 8. The POI3structure is a Super Invar weldmentconsisting of three horizontal plates
and bent side walls. Openingsin the sides allow access to the inside first for weldingthe structure then for assemblingthe
manycomponentswithin the POB.The major design goals are: 1) to provide high dimensional stability, 2) to provide access
to assembledoptical cells within the POB,3) to facilitate coarse alignmentof the optics, and 4) to be relatively compact,
aspect that ripples through the design of the ETSand systemmetrologyfacilities and contributes to goodstructural dynamics
and low temperature gradients. The POBstructure is 514.3 mm(20.25 in.) wide by 549.4 mm(21.63 in.) deep by 618.5
(24.35 in.) tall and weighs approximately191 kg (420 pounds)fully assembled.
Finite element analysis (FEA)has been used throughout the design of the POBstructure, optic mountsand actuation systems.
This model was used to Comparewith the experimental modal analysis of the assembled POBsystem for free-free boundary
conditions. Table 3 showsthe correlation to be very good. In addition, the animated modeshapes of the FEAmatch those of
the experiment. As expected, all themodesof the unconstrained system are very lightly damped,in the range from 0.1%to
0.6%of critical damping.
The POBwas assembled in a two-stage process. The first step was a mechanical assembly of the mirrors using a CMM, and
the secondstep was optical ¯alignmentusing a visible-light point-diffraction phase-shifting interferometer (also known,as
Sommargren interferometer ).7 The optical alignment was based on a detailed and rigorous procedure, 8and finds the rigid-
body adjustmentsof the mirrors required to correct for wavefront aberrations and image distortion measfired at manypoints
across the ring field. This methoddeterminedwhich of the rigid-body degrees of freedomshould be used in the alignment
process: Theseare a small subset of the total 24 possible degrees of freedomof the four mirrors and are those whichaddress
the most linearly independent set of aberration modes. Eight degrees of freedomwere identified (tip, tilts and piston on
minors M2and M4; x y translation on M3)that could be used to correct all misalignment modes, providing that the
magnitudeof the aberration after mechanical assembly did not exceed about 50 nm, rms (this is referred to as the capture
tolerance of the alignment). This choice of compensatorsset the specification on the accuracyof the initial alignmentand on
the design of the mounts.To achieve this accuracy required knowingthe location ’of the aspheric surface (in all six degrees
of freedom) relative to measurable fiducials on each of the substrates. This was achieved with the use of precisely
characterized masks that were commonto both interferometric surface figure measurementsand CMM measurementsof the
mirror substrates, as well as by makingcontact measurements of the mirrors just outside their clear aperture. The estimated
errors in locating the aspheric surfaces within the POBwere of the order 5-10/amin translation and 100 grad in orientation.
The mirrors were assembledin the POBaccording to a recompensationof the original design, based on the measuredfigure
errors Of the mirrors. This recompensation identified positions and orientations of the optics that would give optimum
performance for scanned images. ,
The primary metrology tool used during mechanical assembly is a coordinate measuring machine (CMM).However, the
CMM is not accurate enough to determine the x-y location of the vertex of a very mild asphere to within the capture
tolerances. The vertex must be located with the interferometer and related back to physical datumsurfaces on the optic. This
information combinedwith CMM measurementsaround the clear aperture of the optic is sufficient to establish the optical
coordinate system. Since the optical surfaces are not readily accessible in the assembledPOB,three tooling balls on each
optic cell are located in the optical coordinate system. Figure 9 showsthis mappingprocess for the M4optic. The tooling
balls are very accessible in the assembled POBas Figure 9 shows. A geometric modelthat incorporates the optical design,
mappinginformation and the measuredlocations of the tooling balls computesthe position errors of the optics and the
required mechanical adjustments to correct the errors. With just a few iterations, the mechanical alignment converges to
approximately5 micronsin z and 10 microns in x and y, not including the positioning error in the CMM and the uncertainty
of the vertex location.
Uponmechanical assembly, the first measuredwavefro.nt through the projection optics had a wavefront aberration of only
5.0 nmrms. An alignment correction was computed, based on wavefront and distortion measurementsmade at 15 field
points. In the correction the maskand wafer degrees of freedomwere allowed to vary, in addition to the eight projection
optics adjustments, in order to locate the best conjugate planes. Prior to makingthe correction, alldegrees of freedomwere
calibrated by makingan adjustment of each degree of freedomin turn and comparingthe observed aberration change to the
modeled aberration sensitivity. As well as calibrating the magnitudes and signs of the adjustments, cross coupling of
adjustments was also found and accounted for. The alignment correction required adjustment of all degrees of freedomin
order to correct the. field-dependent wavefront aberration and image distortion. The correction was performed and the
aberrations were corrected as expected. The final wavefront was well corrected across the field, with a meanwavefront
aberration of 1.3 rim, and a minimum aberration of 0.95 nmnear the center of the field. The alignment also removed40 ppm
of magnification error and reduced the mean image-plane distortion from 350 nmto 150 nm. A plot of the wavefront
aberration, measuredat 45 points across the field, is shownin Fig. I0.
;L ’ .....
t-
-10- ....................... 0.9 nm
-60 -40 -20 0 20 40 60
-x (mm)
Figure10. Measuredrmswavefront
error as a functionof field locationat the maskplane,as measured
in the visible-light
alignmentinterferometer.
5. STAGE SUBSYSTEM
The wafer and reticle stage subsystem consists of the reticle and wafer stages, metrology for each stage, and control
electronics. The stage support structures (bridge), vibration isolation system, and isolated frame of the ETSmain chamber
are critical to stage performanceand are included in all stage subsystemanalysis. Thecombinationof the wafer stage, reticle
stage, and bridge provide a stable, stiff platform that allows the wafer and reticle to be positioned to the tight tolerance
required for EUVL.The wafer stage must track the reticle stage during scanning of the reticle stage through the EUVfield.
Tight tolerance is placed on the tracking performance to assure minimuminfluence to image position repeatability and
distortion. A feedbackcontrol systemcontrols the synchronizationof the reticle and wafer stages for scanning the die site
during EUVexposure. Throughput,structural stability, optical capability, and wafer exposurecharacteristics determinestage
subsystemspecifications for acceleration, scan speed, settle time, and dynamicstability. Magnetically levitated (maglev)
stages have been chosen as the lowest-risk approach to addressing the performanceand vacuumrequirements. Testing of the
prototype waferand reticle stages indiCates that the stage systemwill meet ETSperformancespecifications.
A modelof the ETSstructures, stages, and controls was created to predict performanceof the stage subsystemand to serve as
a tool in formulating and comparing control algorithms. It incorporates information from the ETSsystem structural
dynamicsmodeland the optical modelto yield system performancepredictions. Early modelresults, along with validation
experiments~contributed to ETSdesign, decisions in the areas of stage control, vibration isolation, and overall machine
configuration.
5.1 Maglev Stages
The vacuumenvironment required-for EUVL requires a stage design that minimizes contamination of the object reticle,
target wafer, and machineoptics. 1D-long-travel maglevstages based on demonstratedconcepts9’1° have been chosen for this
application. Maglevoffers the advantages of low powerand low particle generation, and has been demonstratedin the fine
1~
position stage used in the IOXMicrostepper.
CZZtarget
cable ~..~, metrology tray
coarse stage
Platen
cable stage
interferometers
on m etrolo D" traY
actuator
Figure11. Waferstage modelincludinginterferometry.
The waferstage (Figure 11) consists of the stage base, off-axis coarse stage (the beamof the maglevstage), levitated platen,
cable stage, and wafer chuck. The reticle stage (Figure 12) is identical to the wafer stage without the off-axis coarse stage
and is mountedin an inverted position. Each stage includes an electrostatic chuck for holding a wafer or reticle. The
monolithic platen is magnetically suspendedfrom the beamand carries image diagnostic and alignment sensors.
beam
stage
actuator
" -.......
iaterferometers
on
metrolagyn’ay.......
¯,...
, ~. :cable stage,
" cable,
platen
cry
The stage controller is implementedin software that runs on several real-time embedded
computers. The controller compares
the measuredpositions of the two stages with their desired positions and attempts to drive these errors to zero by a
combination of feedback and feedforward strategies. The hardware and software used to control the ETSstages is a
combination of commercially available electronics, custom electronics, and software routines and drivers specifically
configured to perform tasks necessary to interface to stage systemsensors and actuators, and execute the necessary control
algorithms. Position accuracy of the stages during EUVexposure is dependent on manyfactors. These contributing factors
include: the stiffness of the overall structuresand stages, the accuracy and precision of all feedback sensors, associated
electronics, and the control strategy used to manipulatethe sensor based information,
5.3 Performanceresults
Stage performancehas been evaluated using a prototype wafer stage using a configuration that approximatesthe structural
properties of the ETSknownas the Stage DevelopmentSystem (SDS). The ETSrequires that the jitter of each stage
referenced to the POBalways be less than 10nmrms and the mean tracking error must always be less than +_3nm.
Performanceresults (Figure 13) from the SDSshowjitter less than 4.5nmrms and meantracking error less than 1.5rim are
achievable for the ETS.
6. ENVIRONMENTAL SUBSYSTEM
The ETSenvironmental system is engineered to minimizeparticulate and chemical contamination of sensitive lithographic
componentssuch as optics and the reticle. The ETSMainChamberis divided into three environmental zones. (See Fig. 14)
These zones are called the Reticle Zone, the Optics Zone and the Wafer Zone. Each zone is an individually controllable
vacuumenvironment.
particle exclusion
...... RelJcle Zone
,~ zone seal
contamination
prevention.... Optics Zone
Resist Cutg&~sing’~indo w
\
outgassingcontrol ...... Wafer Zme zoneseal
Main Chamber
Figure 14. Schematicof ETSenvironmentalsystem.
In the Reticle Zone, conditions are maintained to allow operation of thermophoretic~2 reticle protection via the Klebanoff-
Rader "pellicle." The Klebanoff-Rader approach affords robust protection (> 10 6) of the reticle from particles during
lithographic operation. Other virtues of the thermophoretic protection schemeare high EUVtransmittance and intrinsic
cleanliness. The Reticle Zone is isolated from the Optics Zonevia a slot that alIows entry and exit of EUVradiation. The
slot acts as a differential pumpingstage, allowing an elevated pressure in the Reticle Zone (needed for thermophoretic
protection ) while maintaining a reduced pressure in the Optics Zone(needed for high EUVtransmission).
In the Optics Zone, the chief environmental concern is carbon and oxygen deposition on the optics caused by EUV
"cracking" of residual hydrocarbon and water molecules in the gas-phase environment. The ETSis being constructed from
materials and componentsthat have been carefully selected and screened for low hydrocarbonoutgassing. The Optics Zone
will allow contamination prevention methodsto be deployed, for examplethe use of ethanol in a "Gas Blend" to prevent
optic oxidation.
The Wafer Zone is a completely separate vacuumenvironment with no vacuumcommunicationto the other two zones. The
purpose of the Wafer Zone is :to prevent photoresist outgassing from contaminating optics in the Optics Zone. A Resist
Outgassing Windowresides above the wafer and allows the entry of EUVradiation while intercepting any EUV-induced
photoresist outgassing.
7. OPTICS FABRICATION
7.1 Specifications
The ETSprojection optics substrates have stringent specifications on both figure and finish, which present substantial
challenges for optics manufacturers. Wedivide these specifications into three categories that correspond to the key
functional requirements of the imaging system: figure, Mid-Spatial Frequency Roughness (MSFR),and High-Spatial
FrequencyRoughness(HSFR). 13 Each category is defined as the integral of the two-dimensional powerspectral density of
surface errors over specific ranges of spatial frequency. The definitions of each category and their respective functional
requirements are given in Table 5. Because the specifications are tied to a specific optical design and performance
requirement, it is important to note that these specifications relate to absolute accuracy, as opposedto the precision with
which the optical surface matches a reference artifact. The ETShas been designed to achieve diffraction-limited imaging
with goodcontrast and high reflectivity. The specifications for the ETSsubstrates that support these requirementsare listed
in Table 5 under the heading’Set 2 Specifications’.
Table 5. Specifications and measurementsfor ETSoptical substrates.
Error Term Spatial Set 1 Spec Set 1 (ave) Set 2 Spec
¯ Functional
(Relates to Absolute Frequency Intermediate Measured ETS
Metric
Accuracy) Definition (nm rms) (nm rms) (nm rms)
Resolution
Figure (CA)"l- 1 mm" 0.50 0.50 0.25
Distortion
Mid-spatial frequency Flare "t
lmm-l_l~m 0.50 0.27 0.20
roughness (MSFR) Contrast
7.2 Metrol0gy
The measurement of errors in the three key specification categories correspondsto the ranges of spatial frequencies typically
sampledby three key types of instruments. For measuring absolute figure accuracy over the full clear aperture, we have
designed and. constructed a novel interferometer, the Phase-Shifting Diffraction Interferometer14 (PSDI). It employsnearly
perfect spherical wavesgenerated by propagating both the reference and test wavesthrough small diffracting apertures. Our
detailed analysis of both systematic and non-repeatable error sources supports the certification of surfaces with 0.25 nmrms
absolute accuracy when measuring aspheric ETSsubstrates. Measurements of MSFRand HSFRare performed using a
commercial phase-measuring interferometric microscope and a commercial atomic force microscope, respectively, The
accuracy of these roughness measuringinstruments has been validated by comparingthe powerspectral density (PSD)of the
surface errors calculated from the measured profiles with PSDsdetermined from angle-resolved light scattering
~5
measurementsperformed at the AdvancedLight Source.
Simultaneouslymeetingthe stringent ETSfigure and finish specifications poses extremechallenges for optics manufacturers.
Three of the four substrates are aspheric (order of 5 lam departure), whichtypically compounds
the difficulty in attaining low
roughnesswhiLe attaining figure. At the beginningof the project, the Set 2 specifications were well beyondthe state of the
art in optical fabrication. During the past three years, however,the vendor(s) that we have been workingwith have carried
out substantial developmentsand are currently fabricating surfaces that meet those specifications. In order to support the
ETSdevelopmentschedule; we are acquiring two sets of substrates, one to final ETSspecifications referred to as the Set 2
Precision:0.38 nmrms
Prec|sion:0.37 rim rms
-- AverageRadial 2-D PSD
Accurec~rms
1.o * :
6n rn:.. ’:.
m .112 mm I[,1
Precision: 0.45nmrms
Accuracy:0.56 nm rms ¯ A~ 0,50 .... nm
,(s
+i
~.’
14 1111 mm Irl
14 - 156 mm- I~
Figure15. Measured surfaceerrors for the ETSSet 1 optics. Figure16. PSDsfrommetrologydata for the Set 1 ETSoptics.
Precisionrefers to the direct instrumentreadingandaccuracy (coutesyof Eric Gulliks0nof LBNL)
incorporatesknown systematicerrors.
-.-
The Set 2 substrates for the ETSare nearing completionto the more stringent specifications. The PSDIsfor measuringeach
of the four substrates have been upgradedto achieve lower systematic errors. In-process measurementsof the spherical M3
substrate and the aspheric M4substrate are shownin Figure 17. M3showsa figure error of 0.26 nmrms while the M4was
measured to be 0.19 nmrms. These are the direct readings from _the instrument and are currently being analyzed for
systematic errors. Both of these substrates will undergocontinued processing for further improvement.The remaining Set 2
substrates are also under fabrication with final delivery and coating scheduledfor mid-2000.
Multilayers are deposited in DC-magnetronsputtering systems as described previously] 6 To preserve the figure of the
projection optics, thickness control to about 0.1%rmsis necessary. All four projection optics of the first set weresuccessfully
coated withhigh-reflectance Mo/Simultilayers within a -+0.1%peak-to-valley (P-V) tolerance across the clear aperture
(shadedareas in Fig. 18 (a)-(d)). 17 In fact, the thickness variation is less than 0.1%P-V(-+0.05%P-V)on all four optics. This
¯represents a thickness difference that is commensurate with the size of only one atomfor the full 40-bilayer Stack across the
clear aperture of the optics, the largest of whicti is 160mmwide. The dashedlines in Fig. 18 (a)-(d) showthe best quadratic
fit to the data. Aquadratic, or spherical, figure error inducesa tilt and a focal shift that are easily compensated
duringoptical
alignment. Non-spherical figure errors, however,induce wavefront errors that cannot be corrected. The difference between
the best quadratic fit and the measuredthickness distribution for all four optics cumulatesto less than 0.07 nmrms of non-
correctable aspheric errors. This is well within the current Specification of the optics and, therefore, the error will produce
negligible multilayer-induced wavefront errors. The peak reflectance was measured to be 65.3-+0.3% at a centroid
wavelengthof 13.33__.0.03nmas measuredat their nominalangles of incidence.
’ , ’ i ¯ , ’ I ’ ’ ’ i , , , I , , , i ’ ’ , I , ’ ’ I I. i J
Z
1.005
t-- 1,000
I (b) M2 optic
O, 990
Target
- ¯- - Measured
.
I Added figure error
[ = 0.061 nm RMSI I Targ°*
I IAdd*d
i ur’*
orr
,--*--Measured L = 0.050 nm RMS
. t , t . . ~ , t , . t . . . . i . . . i . , . i . . . i , . . i . , ¯ n . . , i . , .
-40 -20 0 2O 4O -80 -60 -40 -20 0 . 20 40 60 80
I I i l ’ .I i ’ , ’ i ’ ¯ ~ I ’ i
(d) M4optic
¯ " ¯ tl,-.,
+0.1% ±0.1%
O. 995
figure error
i Aded l - Target
0.990 [,- ....Target
Measta’ed I[ [ -- 0.018 nm RaMSI [--.--Measured
=
[ [Added figure
= 0.070 errorl
nm RMS
Figure 18. Normalizedthickness distribution of all four projection optics. The ¯solid circles are the measuredvalues, the
solid line is the targeted thickness profile, the dotted lines are the boundariesof the +-0.1%tolerance zone, andthe
dashed lines are second degree polynomial fits to the measureddata points.
Thecondenseroptics require multilayer coatings with a thickness gradient to accommodate a variation in ray angles across
the optic. The thickness control tolerance across the large (260 mmdiameter) deeply concaveC1optic and the small C3
7optics was ±0.4%.Twofull sets of six C1optic elementsand two full sets of six C3 optics plus few spares were coatedJ
Figure 19 (a) and (b) showthe averagenormalizedthickness profiles for six C1andten C3elements, respectively. (Note:
thei’e were two multilayer designs for the C3 elements, one optimized for an incident angle of 10.645° and the other
optimizedfor 11.850°. Onlythe results of the 10.645°-designare shownhere. The results for the 11.850°-designare very
similar). Thethickness distribution lies within a ±0.25%
zone, well within the specifie d ±0.4%tolerance zone, For the six C1
elements, the averagepeak reflectance was 66.4%,and the peak position (centroid) was 13.29 nmat all radial positions
between55 mmand 130 mm.For the C3 elements, the average peak position (centroid) Value was 13.277 nmand 13.291
for the 11.850"-and10.645°-design,respectively, with an averagereflectance of 66%.
(a) Average of
1.06 |.,i,1,,11,,.i,,,,111,,i..,, i,,,,i,, i 1.008 , i, ii ii ,. |1 i. i, ,i ,d ii i1,, ,1,1 i i i i
r.~
r~
rrl f..rd
z 1.04 z
w ~ 1.004
::=
[.- 1.02
:E
...1
y.I.oo ~. l.O00
rTl
tN0.98
..2
.< I go, I 0.996
r,,, 0.96 I ..... I ©
O 7 0.992
z ...! .... i .... i .... I .... | .... i .... i ....
0.94
50 60 70 80 90 100 110 120 130 -15 -10 -5 0 5 10 15
RADIAL
POSITION,r [mm] RADIALPOSITION,r [mm]
Figure 19. Averagenormalizedthickness profiles for (a) a set of six C1elements and (b) ten C3 optical elementsto be used
10.645°. Thecircles are the measuredvalues, the solid lines are the targeted thickness profile, the dashedcurves are second
degree polynomialfits to the measuredpoints, and the dotted lines represent-the boundariesof +0.4%tolerance zones.
1.00 illJl,.,.
~ .... O. 10 ’ ’ ’ ’ i ’ ’ ’ ’ I , ’ , , [ ’ ’ i ,
0.80
---M
It 0.08
Productof reflectance
......... Productof reflectancewith perfect match
J~M4I"
]
Z 0.60 0.06
.<
0.40 0.04
¯
0.20 0.02
9. SUMMARY
TheEUVETSis in the subsystemtesting phase. Theprojection system has been aligned and characterized using the
development(set 1) optics, whilethe final ETSoptics (set 2) are in an advanced
stage of fabrication. Multilayercoatings
both projection and condensingsystemsmeetor exceedspecifications for uniformityand wavelengthmatching.Thedrive
laser has demonstratedperformance meetingor exceedingall ETSspecifications and continuesto performat that level after
approximately one year of operation. The ETSilluminator has been assembled, aligned and characterized at EUV
wavelengths usinga lowpowerlaser, achievinga pupil fill in close agreement with’the design. Environmentaldata is being
acquired using the ETSilluminator, in which the gas-blend system has been installed. Developmentalmaglev scanning
stages have demonstrated stage synchronization jitter and meanposition error better than ETSspecification. Vacuum-
compatible ETSstages are in fabrication and near completion.
ACKNOWLEDGMENTS
The authors are indebted to the membersof the subsystem teams and technology teams, whosehard work is responsible for
advancingthe EUVEngineeringTest Stand to the state of developmentreported in this’paper.
This work was performed under the auspices of the U.S. Department of Energy by the University of California Lawrence
Livermore National Laboratory under Contract No. W-7405-ENG-48 and Sandia National Laboratories under Contract No.
DE-AC04-94AL85000.Funding was provided by the Extreme Ultraviolet Limited Liability Corporation under a
Cooperative Research and Development A~eement.
REFERENCES
1. C. Montcalm,S. Bajt, P. B. Mirkarimi, E. Spiller, F. J. Weber,and J. A. Folta, "Multilayer reflective coatings for
extreme-ultraviolet lithography," in EmergingLithographic Technologies//, Y. Vladimirsky, Ed., Proceedings of SPIE
Vol. 3331, 42-51 (1998).
2. J.E.M. Goldsmith, K. W. Berger, D. R. Bozman,G. F. Cardinale, D. R. Folk, C. C. Henderson, D. J. O’Connell, A. K.
Ray-Chaudhuri, K. D. Stewart, D. A. Tichenor, H. N. Chapman,R. Gaughan, R. M. Hudyma,C. Montcalm, E. A.
Spiller, J. S. Taylor, J. D. Williams, K.. A. Goldberg, E. M. Gullikson, P. Naulleau, and J. L. Cobb, "Sub-100-nm
Imaging with an EUV10x Microstepper," in Emerging Lithographic Technologies III, Yuli Vladimirsky, Editor,
Proceedings of SPIE Vol. 3676, pp. 264-271 (1999).
3. G.D. Kubiak, L. J. Bernardez, K. Krenz, W. C. Replogle, W. C. Sweatt, D. W. Sweeney, R. M. Hudyma,and H.
Shields, "High-powersource and illumination system for extreme ultraviolet lithography," Proceedingsof the SPIEvol.
3767, 136 (1999).
4. G.D. Kubiak, L. J. Bernardez, K. Krenz, and W.C. Sweatt, "Scale-up of a cluster jet laser plasma source for extreme
ultraviolet lithography", Proceedingsof the SPIEvol. 3676, 669 (1999).
5. D.W. Sweeney, R. Hudyma,H. N. Chapman,and D. Shafer, "EUVoptical design for a 100 nmCDimaging system,"
Proceedings of SPIE 3331 2- I0 (I998).
6. R.M. Hudyma,H. N. Chapman,D. W. Sweeney, and D. R. Shafer, "Reflective optical imaging system with balanced
distortion," USpatent 5,973,826(1999).
7. G.E. Sommargren,in OSATrends in Optics and Photonics, Extreme Ultraviolet Lithography Vol. 4, edited by G. D.
Kubiak and D. R. Kania (Optical Society of America, WashingtonDC), pp. 108-112 (1996).
8. H.N. Chapmanand D. W. Sweeney, "A rigorous methodfor compensation selection and alignment of microlithographic
optical systems," Proceedings of SPIE 3331, 102-113(1998).
9. M.E. Williams, P. Faill, S. P. Tracy, P. Bischoff, and J. Wronosky,"MagneticLevitation ScanningStages for Extreme
Ultraviolet Lithography", ASPE14th annual meeting, MontereyCA., November(1999).
10. J. B. Wronosky,T. G. Smith, J. R. Darnold, "Developmentof a Wafer Positioning System for the Sandia Extreme
Ultraviolet Lithography Tool", NASA 3ra International Symposiumon Magnetic SusPension Technology, Tallahassee,
Fla., December(1995).
11. M.Williams, P. Faill, et.al., "Six Degree of Freedom Mag-LevStage Development", Proceedings of the SPIE
Conference, March, Vol. 3051 (1997).
12. S.J. Choi, D.J. Rader and A.S. Geller, "Massivelyparallel simulations of Browniandynamicsparticle transport in low
pressure parallel-plate reactors," J. Vac. Sci. Tech. AVol. 14, 660(1996).
13. J. S. Taylor, G. E. Sommargren,D. W. Sweeney, and R. M. Hudyma,"The fabrication and testing of optics for EUV
projection lithography", EmergingLithographic TechnologiesII, Proceedingsof SPIE ~,ol, 3331, pp. 580-590(1998).
14. G. Campbell, G. E. Sommargren, D. W. Phillion, "Measurement of aspheres using phase shifting diffraction
interferometry", paper presented at the OSAAnnualMeetingand Exhibit, October 20-24, Rochester, NY(1996).
15. E. M. Gullikson, "Scattering from normal-incidence EUVoptics", EmergingLithographic Technologies II, Proceedings
of SPIEvol. 3331, pp. 72-80 (1998).
16. D. G. Stearns, R. S. Rosen, and S. P. Vernon, "Fabrication of high-reflectance Mo-Si multilayer mirrors by planar-
magnetronsputtering," J. Vac. Sci. TechnologyA 9, 2662-2669(1991).
17. C. Montcalm,R. F. Grabner, R. M. Hudyma,M. A. Schmidt, E. Spiller, C. C. Walton, M. Wedowski,and J. A. Folta,
"Multilayer coated optics for an alpha-class extremeultraviolet lithography system," in EUV,X-Ray, and NeutronOptics
and Sources, S. P. Vernonand K. Goldberg, Eds. Proceedings of SPIEVol. 3767, 210-216(1999).