Thermomechanical Modeling of The EUV Reticle During Exposure

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

PROCEEDINGS OF SPIE

SPIEDigitalLibrary.org/conference-proceedings-of-spie

Thermomechanical modeling of the


EUV reticle during exposure

Martin, Carl, Engelstad, Roxann, Lovell, Edward

Carl J. Martin, Roxann L. Engelstad, Edward G. Lovell, "Thermomechanical


modeling of the EUV reticle during exposure," Proc. SPIE 4343, Emerging
Lithographic Technologies V, (20 August 2001); doi: 10.1117/12.436683

Event: 26th Annual International Symposium on Microlithography, 2001, Santa


Clara, CA, United States

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use


Thermomechanical modeling of the EUV reticle during exposure
Carl J. Martin,* Roxann L. Engelstad, and Edward G. Lovell
Computational Mechanics Center, University of Wisconsin, Madison, WI 53706

ABSTRACT

Thermal deformations of lithographic reticles during the exposure process may become an important consideration for all
candidate Next-Generation Lithography technologies as these reticles are subject to stringent image placement and flatness
requirements. The reflective reticles used for extreme ultraviolet lithography (EUVL) absorb energy during exposure
producing temperature gradients and thermomechanical distortions that result in pattern placement errors. As throughput
requirements of EUVL are increased, the necessary illumination power levels rise producing higher reticle temperatures. The
use of a low-thermal-expansion substrate material reduces, but does not eliminate, reticle distortions, and the thermal and
structural boundary conditions greatly influence the thermomechanical response. These factors make the accurate predictions
of the reticle thermal and structural response essential to the design of EUVL systems. Previously published analyses
focused on relatively low throughputs, 10 wafers-per-hour (wph), and 200-mm diameter wafer substrates. Proposed
production systems have throughputs of 80 wph and use 6-in. square substrates. Finite element models of current format
EUV reticles have been developed to simulate the reticle’s thermomechanical response to high-throughput exposure heating
and assess the resulting image placement errors. The results of thermal and structural analyses for a variety of EUVL load
and boundary conditions are presented.

Keywords: EUV lithography, EUV reticles, exposure, finite elements, thermomechanical response

1. INTRODUCTION

Proposed extreme ultraviolet lithography (EUVL) systems will be subject to the stringent image placement requirements of
sub-100 nm lithography. Thus the thermal deformations of the reticle during the exposure process must be accurately
assessed. Extreme ultraviolet lithography systems employ reflecting/absorbing reticles that absorb a significant amount of
the incident illuminating energy (approximately 100% in the patterned areas and 35% in the unpatterned areas), and the
required illumination energy rises with increasing throughputs. Even with the use of low thermal expansion substrates the
reticle themomechanical distortions can be significant at proposed throughput rates of 80 wafers-per-hour (wph). In addition,
the thermal response of the reticle is highly dependent on the thermal boundary conditions, most significantly the contact
conductivity between the reticle and the electrostatic chuck. In published papers, estimates of the reticle-to-chuck contact
conductivity in a vacuum have ranged from 3 to 150 W/m2-K providing for a wide range of thermal responses.1 Accurate
modeling of both the thermal and structural boundary conditions is critical for accurate simulations of the EUV reticle
behavior. In this paper, the thermal and structural responses of the EUV reticle are determined under a range of throughput
levels, pattern coverages, and boundary conditions to evaluate reticle distortions under a variety of potential operating
conditions.

EUV reticles consist of multilayer reflective coatings deposited on a flat substrate. Because EUV light is strongly absorbed
by all materials, EUVL systems must utilize reflective reticles and optics systems and must also operate in vacuum. The
patterned areas of the reticle are coated with an EUV absorbing material. Virtually all of the incident radiation is absorbed in
the patterned areas and about 35% is absorbed in the reflective areas producing a significant thermal loading to the reticle. A
low expansion substrate is used to limit thermal distortions. The 6-in. square reticle format used for these analyses, as
illustrated in Fig. 1, utilized a 6.35-mm thick ULETM glass substrate. ULE glass has a near zero coefficient of thermal
expansion at room temperature, and its use significantly reduces thermal deformations in EUVL reticles. The patterned or
active area of the reticle is coated with a Mo/Si multilayer which is reflective to light at EUV wavelengths. This multilayer
consists of 41½ molybdenum/silicon bilayers with a total thickness of 282 nm. A patterned EUV absorbing layer is
deposited on top of the multilayer reflector. The thermal and mechanical properties of the various materials used in the
reticle fabrication are listed in Table 1. The dimensions of the patterned area are discussed in the next section of this paper.
________________________________________
*correspondence: cjmartin@engr.wisc.edu

Emerging Lithographic Technologies V, Elizabeth A. Dobisz, Editor,


Proceedings of SPIE Vol. 4343 (2001) © 2001 SPIE · 0277-786X/01/$15.00 515

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
152.4 mm

Scan Direction
Reticle Materials

• Substrate: ULE glass


Thickness: 6.35 mm
Illumination strip
152.4 mm

Field • Multilayer Reflector: Mo/Si 41 ½ bilayers


Height Thickness: 282 nm

• Absorber: Titanium nitride (TiN)


Thickness: 100 nm

Patterned Area
Field Width

Fig. 1. Schematic of EUV 6-in. square reticle with all dimensions in mm.

Table 1. Thermal and structural properties of reticle materials.

Property ULE glass Silicon Molybdenum Titanium Nitride


Density (kg/m3) 2205 2330 10300 5220
Thermal conductivity (W/m-K) 1.31 148 138 19.3
Specific heat (J/kg-K) 766 712 255 522
Young’s modulus (GPa) 67.6 107 272 600
Poisson’s ratio 0.17 0.25 0.25 0.25
Thermal expansion coeff. (ppm/K) 0.02 2.5 5.35 9.4
Emissivity 0.735 0.122* 0.122* 0.189
*emissivity for multilayer reflector stack

2. LOADING AND MODEL DESCRIPTIONS

Exposure in EUVL is achieved by scanning the reticle and wafer through a strip of EUV light. The scanning is bi-
directional, and the illumination strip width is 6 mm. The height of the illumination field is the height of the active patterned
area. The scan velocity and illumination intensity are functions of the wafer throughput, resist sensitivity, camera
attenuation, and handling overhead time. As examples, throughput models for three proposed EUV tools representing ever
increasing throughputs are listed in Table 2. The first, the Engineering Test Stand (ETS), is an alpha-class tool currently
under development with a throughput of 10 wph.2 Most of the published EUVL exposure analyses1,3 have used the
throughput parameters of this tool. The second model, a Beta tool proposed by the EUV LLC,4 has a throughput of 63 wph.
And finally, a proposed system from the EUCLIDES project5 has a throughput of 81 wph. Highlights of the throughput
parameters for all of these systems are contained in Table 2, and full details may be found in the references cited. It is
interesting to note that the EUV power on the reticle is 2.5 times higher for the EUCLIDES system than for the Beta tool.
This is a product of the higher throughput, higher resist exposure dose, and a lower camera attenuation factor. The total
camera attenuation factor is the fraction of the energy entering the camera that reaches the reticle and incorporates all the
optical losses between the reticle and the wafer. All analyses assumed the presence of framing blades so that only the
patterned area was illuminated. Because the EUCLIDES 81 wph throughput best matches the current goals for production
EUV tools, this model will be used for all analyses. Parametric analyses of both the thermal and structural responses to
changes in the applied EUV illumination power are included so that the effects of differences in tool efficiencies may be

516 Proc. SPIE Vol. 4343

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
evaluated. As mentioned previously, the reflecting/absorbing reticles used in EUVL absorb approximately 100% of the
incident illumination in the patterned areas and 35% in the unpatterned areas. The majority of the analyses presented
assumes 100% absorber (pattern) coverage because this produces the most severe thermal loading. A few analyses were
repeated in which one-half of the patterned area was completely covered with absorber and the opposite half was completely
reflective. This case was selected to illustrate how differences in pattern coverage might limit the effectiveness of
magnification corrections.

Table 2. Throughput and thermal loading parameters.

Parameter EUV LLC ETS2 EUV LLC Beta Tool4 EUCLIDES5


Throughput (wph) 10 63 81
Resist exposure dose (mJ/cm2) 10 5 7
Wafer load/unload/align time (s) 16 12 12
Fields per wafer 67 67 89
Stage stepping time per field (s) 1 0.22 0.112
EUV power on wafer (W) 0.023 0.109 0.212
Total camera attenuation factor 0.034 0.072 0.057
EUV power on reticle (W) 0.683 1.51 3.74
Field width at reticle (mm) 130 136 100
Field height at reticle (mm) 98 104 100
Illumination field width (mm) 6 6 6
Scan velocity (mm/s) 37.7 332.7 483
Illumination time per field (s) 3.83 0.453 0.253

Three-dimensional thermal and structural finite element models of the reticle were developed for the transient simulations.
The geometry of the reticle and the exposure scanning allowed for the use of symmetry, reducing the model size by half, as
illustrated in the typical finite element mesh shown in Fig. 2. The ULE glass substrate was modeled using three-dimensional
solid elements, and the multilayer reflector and absorber layers were modeled with two-dimensional shell elements. All
modeling and analysis was performed using the commercial finite element code ANSYS®. For the thermal analyses, it was
assumed that the back surface of the reticle was in contact with the chuck; consequently, the contact resistance at this
interface was simulated with a convective boundary condition over the surface. The interface contact conductivity for the
flat, electrostatic chuck proposed for use in EUV tools has been estimated to be 150 W/m2-K1, but a range of values was
explored in this paper due to the uncertainty of this estimate. The chuck temperature was assumed to be a uniform 20 oC.
Radiative boundary conditions were applied to the front surface of the reticle. This surface was assumed to radiate to a black
environment maintained at a constant 20 oC, and appropriate emissivities were used for each surface material. The radiative
contribution to the cooling of the reticle is rather small in comparison to conduction to the chuck. This may be illustrated by
linearizing the radiation component and expressing it as a convective boundary using the following:

qrad = ε σ ( T 4-T04) = ε σ (T 2+T02)(T+T0)(T-T0) ~ 4ε σ T03 (T-T0) (1)

where ε is the surface emissivity and σ is the Stefan-Boltzmann constant. The expression 4ε σ T03 can be thought of as an
equivalent convection coefficient hrad in the linearized radiation model. For the TiN absorber at 20 oC, the value of hrad was
1.08 W/m2-K , or over two orders of magnitude less than the reticle-to-chuck contact conductance.

For the structural analyses, the electrostatic chuck was assumed to constrain the bottom surface of the wafer to be completely
flat. The frictional forces between the reticle and chuck were ignored allowing the reticle to expand freely in-plane. All in-
plane motion was restricted at a single point in the center of the reticle, and thus all displacements presented are with respect
to the reticle center. The constraint of in-plane rigid body motions was required for the finite element analysis, but in practice
the electrostatic chuck does not have a specific point of in-plane constraint. The displacements calculated in all the analyses
presented in this paper are the result of exposure heating only, and any reticle distortions resulting from fabrication and
mounting need to be added vectorially to these results.

Proc. SPIE Vol. 4343 517

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Symmetry Plane Patterned Area

Fig. 2. Typical finite element mesh for EUCLIDES reticle.

3. EXPOSURE SCAN SIMULATION

The effects of the exposure heating may be categorized into two types of response. The first is the immediate and local effect
due to the scanning column of EUV illumination. The illumination time for the scanning of one field is 0.25 s in the
EUCLIDES tool. Because of the short response times and localized nature of the response, any distortions produced by the
short-term illumination energy would be very difficult to correct. The second type is the long-term global thermal response
of the reticle to the exposure of many device wafers. The transient response times for this type of heating are measured in
minutes and are generally global magnifications that are readily correctable. In this section, the short-term, localized effects
of the illumination energy will be evaluated to determine their significance with respect to total mask distortions. The
EUCLIDES exposure parameters, as listed in Table 2, along with full absorber coverage were used in these analyses. The
applied thermal loading simulated the scan of the EUV illumination column across the length of the mask. The scan velocity
was 483 mm/s, and the heat flux under the 6 mm × 100 mm illumination field was 6.21 mW/mm2. In the finite element
simulation, the heat flux was moved across the elements in the patterned area (at the scan velocity) to simulate the exposure
illumination. Results from the thermal simulation near the completion of one scan are illustrated in Fig. 3. The maximum
temperature rise of 0.4 oC occurs near the trailing edge of the illumination field, and temperature rises can only be seen near
the heated surface of the reticle. Little heat diffusion can be observed around the perimeter of the mask, and only a small
amount is visible through the thickness due the relatively low thermal conductivity of the ULE glass and the short time
duration of the scan. The thermomechanical displacements produced by a single illumination scan were found to be less than
0.05 nm at the mask level. These analyses indicate that the short-term, local heating by the EUV source does not produce
significant distortions.

(oC)

Fig. 3. Transient thermal response of the EUV reticle after the first scan. Results are for EUCLIDES 81 wph exposure
parameters with 100% absorber coverage, and the illumination scans from right to left on the reticle.

518 Proc. SPIE Vol. 4343

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
After determining that the short-term, local effects of the exposure illumination were negligible, longer-term global heating
simulations were developed. The models used in these analyses were similar to those of the previous study, but fewer
elements were required through the substrate thickness. The thermal loading was modified so that a uniform, average heat
flux was applied over the entire patterned area. The average heating applied the same total energy as the illumination scan
but averaged the heat load over the entire scan and stage stepping times. Analysis demonstrated that the two illumination
methods produced identical long-term thermal responses, and the average heating approach greatly reduced computational
requirements. The reticle temperatures immediately after the exposure of the 81st wafer are illustrated in Fig. 4. A reticle-to-
chuck contact conductivity of 150 W/m2-K was assumed for this analysis, and the maximum temperature rise was computed
to be 1.92 oC. The temperature was highest in the center and dropped off quickly outside the patterned area. The transient
responses at four locations on the reticle are plotted in Fig. 5. The reticle reaches a quasi-steady state condition in about 500
s after which the temperatures oscillate around a mean during exposure and wafer change cycles. The oscillations were
greatest at the heated surface and were 0.48 oC at the top surface at the center of the reticle. In addition, there was a
significant through-thickness temperature gradient that reached a maximum of 0.88 oC at the reticle center. In comparison,
the temperatures, oscillations, and gradients at the corners of the patterned area were significantly lower.

(oC)

Fig. 4. Temperature contours in the EUV reticle after 1 hr. of exposure illumination with EUCLIDES 81 wph exposure
parameters and 100% absorber coverage. The maximum temperature rise is 1.92 oC.

22.0

Center – top surface


Center – bottom surface
21.5 Corner – top surface
Corner – bottom surface
Wafer change
Temperature (oC)

21.0
Corner

Center
20.5

20.0
0 200 400 600 800 1000
Time (s)

Fig. 5. Long-term thermal response of the EUV reticle to the exposure of multiple device wafers at a throughput of 81 wph.

Using these temperatures, the associated mechanical deformations were calculated. The boundary conditions assumed the
reticle was chucked completely flat with no restraint to in-plane motions (frictionless). The in-plane motions were, however,
arbitrarily restrained to zero at the center of the reticle to facilitate a solution. The in-plane and out-of-plane displacements
(IPD and OPD) of the patterned region of the reticle resulting from the global heating are illustrated in Fig. 6. The maximum

Proc. SPIE Vol. 4343 519

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
in-plane displacement in the patterned area was 1.3 nm with respect to the fixed center point, and the maximum OPD was
0.19 nm. Because of the applied boundary conditions, the resulting OPD was predominantly due to thermal expansion
through the thickness of the reticle.

(nm) (nm)

(a) In-plane displacements (IPD) (b) Out-of-plane displacements (OPD)

Fig. 6. Thermally-induced in-plane and out-of-plane displacements of the EUV reticle after 1 hr. of exposure illumination
with EUCLIDES 81 wph exposure parameters and 100% absorber coverage.

As seen in Fig. 6(a), most of the in-plane displacements are simple magnifications due to the global heating and resulting
thermal expansion of the reticle. Thus magnification corrections could be expected to significantly reduce the placement
errors associated with exposure heating; a magnification correction reduces the maximum IPD to 0.28 nm from the
uncorrected 1.33 nm. This level of correction is probably not attainable due to the thermal transients associated with the
illumination and wafer change cycles illustrated in Fig. 5. Analyses of the reticle immediately after a wafer change identified
a maximum IPD of 1.13 nm indicating a transient reticle expansion of 0.20 nm during the exposure of a single device wafer.
In order to predict the effectiveness of a steady state magnification correction, a correction was calculated using the average
of the IPD’s before and after a wafer change. Using this correction the maximum corrected IPD was 0.37 nm. These
corrections were based upon thermal expansion around a fixed center point. The undefined in-plane restraint associated with
the electrostatic chuck would certainly complicate and may well limit the effectiveness of these linear magnification
corrections.

Because of the uncertainty in many variables involved in determining the exposure illumination energy, the sensitivities of
the thermal and structural responses with respect to the heat load were calculated. In these analyses the applied heat load was
scaled linearly while throughput rates and overhead times were held constant. This type of scaling would accompany
changes in resist exposure dose or camera attenuation. The results, illustrated in Fig. 7, indicate there were linear variations
in both the thermal and structural responses with changes in the applied loading. This was to be expected because the
nonlinear radiation effects are relatively small, and thus the inherent linearity of the response allows the extension of these
results over a wider variety of throughput conditions.

3.0 3.0
Max. IPD
Max. Temperature Rise (oC)

2.5 2.5
Max. Temp Rise
Max. IPD (nm)

2.0 2.0

1.5 1.5

1.0 1.0

0.5 0.5

0.0 0.0
-30 -20 -10 0 10 20 30

Change in Applied Heat Load (%)


Fig. 7. Sensitivities of the structural and thermal responses to changes in the applied heat load. Baseline loading was from
EUCLIDES 81 wph exposure parameters.

520 Proc. SPIE Vol. 4343

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
The structural analysis boundary conditions assumed that the back surface of the reticle was held flat. In order to evaluate the
electrostatic clamping forces required to do this, the normal stresses in the reticle at the reticle-to-chuck interface were plotted
(Fig. 8). Because this is a free surface, these stresses indicate the local clamping pressures required to maintain the flat back
surface condition. Tensile (positive) stress values indicate the local clamping force required, and compressive (negative)
stresses indicate the pressures against the chuck. The maximum stress of 100 Pa was well below the 5 to 10 kPa chuck
clamping pressures proposed to prevent sliding during reticle stage acceleration.

σz (Pa)

Fig. 8. Normal stress (Pa) in the reticle at the reticle-chuck interface resulting from exposure thermal distortions. Tensile
stresses indicate local electrostatic clamping pressures required to maintain flat back surface.

4. EFFECTS OF VARIATIONS IN PATTERN COVERAGE

The analyses presented thus far have assumed 100% absorber coverage in the patterned area which maximized the energy
input to the reticle. Any reductions in absorber coverage would therefore reduce the temperature rise and thermal distortions.
Patterning, however, will generally reduce the fraction of absorber coverage nonuniformly over the surface of the reticle and
would affect the displacement field differently than a simple reduction in heat input. The extreme case where one-half of the
patterned area had 100% absorber coverage (a dark field) and the other half had no absorber (a bright field) was modeled to
evaluate the potential effects of variations in pattern coverage. The thermal results of this simulation are illustrated below in
Fig. 9, where the right half of the patterned area is the dark field. The temperature contours indicate there was little heat
conduction laterally in the reticle, and the maximum temperature rise of 1.87 oC was only 0.04 oC less than the full dark-field
results. The uncorrected in-plane displacement vectors, plotted in Fig. 10(a), indicated the maximum IPD was nearly
identical to the full dark-field case, but that the displacements were much larger on the dark-field half of the reticle which
was characterized by a greater temperature rise. The displacements after linear magnification corrections in the X and Y
directions are shown in Fig. 10(b). The maximum IPD after correction of 0.60 nm was over twice the maximum IPD after
correction of the full dark-field case. The asymmetric heating of half-dark / half-bright load case reduced the effectiveness of
the linear magnification corrections, but this case represents an extreme in pattern variation.

(oC)

Fig. 9. Temperature contours in a half-dark / half-bright field reticle after 1 hr. of exposure illumination with EUCLIDES
81 wph exposure parameters. The dark field is on the right-hand side of the patterned region. The maximum
temperature rise is 1.87 oC.

Proc. SPIE Vol. 4343 521

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
(nm) (nm)

(a) Uncorrected in-plane displacements (IPD) (b) IPD after magnification correction

Fig. 10. (a) Thermally-induced in-plane displacement vectors of a half-dark / half-bright field reticle after 1 hr. of exposure
illumination. (b) The resultant distortions after independent linear magnification corrections in the X and Y
directions.

5. RESPONSE SENSITVITY TO RETICLE-TO-CHUCK CONTACT CONDUCTANCE

Previous published analyses1 have indicated that the thermal response of the EUV reticle to exposure heating was extremely
sensitive to the value of reticle-to-chuck contact conductance. The uncertainty in any estimates of contact conductance would
be extremely high given the current uncertainties in the design of both the reticle and chuck. Analytical models of the heat
conduction in contacting surfaces indicate that the joint contact conductance is dependent on the thermal conductivities,
roughnesses, and microhardnesses of the contacting surfaces as well as the contact pressure.6 Reference 1 states a measured
contact conductance of 150 W/m2-K for a flat electrostatic chuck, but it is clear that that in actuality this value would be
highly dependent on both the final wafer and chuck design specifications. Nevertheless, this value was used as a baseline in
these analyses. Because of the uncertainties involved with this estimate, an investigation of the sensitivities of the thermal
and structural responses to changes in the reticle-to-chuck conductance was carried out by repeating the analyses using a
range of values. The results, summarized in Fig. 11, indicated that there were only small reductions in temperature and IPD
attained by increasing the contact conductance above the 150 W/m2-K baseline value. As the contact conductance was
reduced, the temperature and IPD increased more rapidly, especially when the contact conductance fell below 100 W/m2-K.
From these analyses, the advantages of higher values of contact conductance were two-fold: reduced temperature rise and
IPD and also reduced sensitivity to variations in contact conductance. All of these analyses have assumed a constant contact
conductance across the entire interface, but in real systems there is the likelihood of considerable nonuniformity due to
variations in contact pressure and surface finish across the interface. The factors that govern contact conduction will need to
be considered when wafer and chuck standards are defined so that acceptable values and variations in conductance are
obtained.

7.0 7.0
Max. Temperature Rise (oC)

6.0 6.0
Max. IPD
Max. IPD (nm)

5.0 5.0
Max. Temp Rise
4.0 4.0

3.0 3.0

2.0 2.0

1.0 1.0

0.0 0.0
0 100 200 300 400 500
Reticle-to-Chuck Contact Conductance (W/m2-K)

Fig. 11. Sensitivities of the structural and thermal responses to changes in the reticle-to-chuck contact conductance.

522 Proc. SPIE Vol. 4343

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
6. CONCLUSIONS

Finite element simulations have been used to evaluate the thermal and mechanical behavior of a typical EUV reticle during
the exposure process. Both the local effects of the EUV illumination scan, and the global effects of long-term multiple wafer
exposures were evaluated. The local, short-term effects of a single scan were found to be negligible. The maximum long-
term temperature rise in the reticle was 1.92 oC. The maximum in-plane and out-of-plane displacements were calculated to
be 1.33 nm and 0.19 nm, respectively. Magnification corrections reduced the effective distortions to less than 0.4 nm. The
effect of variation in pattern coverage was analyzed using the extreme case of a half dark-field / half bright-field mask. The
maximum temperature rise and IPD were similar to the dark field cases, but the applied magnification corrections were
slightly less effective. The thermal response of the reticle was found to be highly sensitive to the reticle-to-chuck contact
conductance especially for values below 100 W/m2-K. The factors the influence the interface thermal conductivity such as
clamping pressure and surface finishes should be considered when developing reticle and tool standards. All the results
presented in this paper include only thermal deformations produced by exposure heating and are at the mask level. Reticle
distortions produced by fabrication and mounting must be vectorially added to these results. In addition, both the thermal
and structural responses are directly scalable with power input, i.e., doubling the input power would double the resulting
temperature rises and deformations.

ACKNOWLEDGMENTS

This research has been funded in part by DARPA, International SEMATECH, and the Semiconductor Research Corporation
(SRC). Computer support has been provided by International SEMATECH, Intel, and Microsoft. The technical assistance of
N. Wester (Intel / International SEMATECH) and R. S. Mackay (International SEMATECH) is gratefully acknowledged.

REFERENCES

1. S. E. Gianoulakis and A. K. Ray-Chaudhuri, “Thermal management of EUV lithography masks using low expansion
glass substrates,” SPIE Emerging Lithographic Technologies III, Vol. 3696, pp. 598-605, 1999.
2. C. Gwynn, et al., Extreme Ultraviolet Lithography, A White Paper, Extreme Ultraviolet Limited Liability Co.,
Livermore, CA, 1998.
3. S. E. Gianoulakis and A. K. Ray-Chaudhuri, “Thermal-mechanical performance of extreme ultraviolet lithographic
reticles,” Journal of Vacuum Science and Technology B, Vol. 16, No. 6, pp. 3440-3443, 1998.
4. C. Gwynn, et al., Extreme Ultraviolet Lithography, A White Paper, Extreme Ultraviolet Limited Liability Co.,
Livermore, CA, 1999.
5. J. Benschop, U. Dinger, D. Ockwell, “EUCLIDES: first phase completed!,” SPIE Emerging Lithographic Technologies
IV, Vol. 3997, pp. 34-47, 2000.
6. M. M. Yovanovich, “New contact and gap correlations for conforming rough surfaces,” AIAA 16th Thermophysics
Conference, AIAA –81-1164, 1981.

Proc. SPIE Vol. 4343 523

Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 13 Dec 2020


Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

You might also like