Impact of Layout Dependent Flare on Printing in EUV Lithographyeuvlsymposium.lbl.gov/pdf/2005/pres/33 2-MA-32 Kamm.pdf · 2015. 11. 24. · Strong layout-dependent flare effects on

Copyright © Infineon Technologies 2005. All rights reserved.Page 1 “Infineon Technologies “2005 EUVL Symposium”

Impact of Layout Dependent Flare on Printing in EUV Lithography

F.-M. Kamm, M. Bender, S. Schwarzl, S. Trogisch, S. WurmInfineon Technologies, Munich/Dresden, Germany

4th Int. EUVL Symposium, San Diego, 2005


Outline

Impact of FlareImpact of Flare

ExperimentExperiment

Summary/ConclusionsSummary/Conclusions

Discussion & SimulationsDiscussion & Simulations

Printing ResultsPrinting Results


Outline







Impact of Flare

45 nm hp 32 nm hp 22 nm hp

5.5 3.9 2.7 DRAM

1.6/2.0 1.2/1.4 0.8/1.0 MPU/ASICCD control nm (3σ)

CD Control

CD Control

Major challenge in EUVL for High Volume Production

ITRS 2004 Up-date

Flare Flare

Light scattering by surface roughness in imaging systemsStrong flare effects in EUVL due to short wavelength (∝ 1/λ2) Detrimental impact on contrast, image quality, process latitude, throughputFlare is usually non-uniform in reflective EUV systems due to

- non-uniform mirror fabrication processes- non-uniform pattern densities of mask layouts

Non-uniform flare ⇒ CD variations

Layout depending printing effects have to be known in order to minimize CD variations by adequate compensation methods


Outline







The Micro Exposure Tool (MET) in Berkeley

ProcessConditionsProcess

Conditions

45° dipole and 0° quadrupole illumination used for printingResist MET-1K (XP3454) from Rohm&Haas on 4“ wafers.Target thickness 125 nm, PAB 130°C (60s), PEB 130°C (90 s).14 x 14 FEM with 5% dose steps and 50 nm focus steps.CD measurement by SEM and software SuMMIT (EUV Technology)

From synchrotron

Scanner modules

Reticle stage

MET

Wafer stage and height sensor

Pupil-fill monitor

λ = 13.5 nmNA = 0.3σ = adjustablered. = 5xres. = 30 nmk1 = 0.7

λ = 13.5 nmNA = 0.3σ = adjustablered. = 5xres. = 30 nmk1 = 0.7

by courtesy of P. Naulleau/VNL(σi=0.3, σ0=0.6) (σi=0.2, σ0=0.7)


1x wafer scaleBlue = Etched ML (dark)

White = ML Reflector (bright)

Flare Test Pattern on Etched Multilayer Binary Mask (absorberless)

Brightfield(155µm x 200µm)

Distance from brightfield [µm]

100nm

70nm

50nm

30nm

30nm

50nm

70nm

100nm

vertical dense lines

horizontal dense lines

1:2 vertical posts

1:2 horizontal posts

155 0x

1:2 vertical holes

1:2 horizontal holes

4FF

2F4F

Etched Multilayer Mask

Etched Multilayer Mask

No absorber stackNo shadowing High contrast

No absorber stackNo shadowing High contrast

ML

S

X-section of etched ML mask


Outline







CD vs. Distance for 100 nm Coded Dense Lines

Significantly stronger flare impact on horizontal lines than on vertical.Horizontal lines completely erased close to bright field (x = 0 - 2 µm)Residues of horizontal resist lines collapsed up to 20µm off the bright fieldOpposite curvatures of line width dependencies for horizontal and vertical lines close to dark field

100nm Horizontal and Vertical Dense Lines vs. Distance from Bright Field

0

20

40

60

80

100

120

0 20 40 60 80 100 120 140 160Distance x from Bright Field [µm]

Line

Wid

th[n

m]

horizontalvertical

Best focus, dose 25mJ/cm²

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Dis

tanc

e x

from

Brig

ht F

ield

[µm

]15

0

100

5

0

0


Vertical lines erased close to bright field (x ≤ 2µm) and pattern collapse up to ~ 15µm off the bright field

Horizontal resist lines completely erased up to 25 µm off the bright field; residues/pattern collapse up to 60µm off the bright fieldBest focus, dose 25mJ/cm²


0

20

40

60

80

0 20 40 60 80 100 120 140 160Distance x from Bright Field [µm]

Line

Wid

th [n

m]

horizontalvertical

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Dis

tanc

e x

from

Brig

ht F

ield

[µm

]40

2

0

10

0 CD vs. Distance for 70 nm Coded Dense Lines


Vertical lines completely erased up to 15µm off the bright field, residues/pattern collapse up to 20µm off the bright field

Horizontal lines completely erased up to 25 µm off the bright field, residues/pattern collapse up to 80µm off the bright fieldBest focus, dose 25mJ/cm²


0

20

40

60

0 20 40 60 80 100 120 140 160

Distance x from Bright Field [µm]Li

ne W

idth

[nm

]

horizontalvertical

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Brig

ht F

ield

Dis

tanc

e x

from

Brig

ht F

ield

[µm

]80

5

0

20

0 CD vs. Distance for 50 nm Coded Dense Lines


Vertical Dense Lines vs. Distance from Bright Field

0

20

40

60

80

100

120

0 20 40 60 80 100 120 140 160

Distance x from Bright Field [µm]

Line

Wid

th [n

m]

100nm70nm50nm30nm

Horizontal Dense Lines vs. Distance from Bright Field

0

20

40

60

80

100

120

0 20 40 60 80 100 120 140 160


Line

Wid

th [n

m]

100nm

70nm

50nm

Horizontal line widths significantly smaller than vertical ones for all distances.

h/v difference smaller for smaller features (smaller pattern distance in vertical direction).

Different curvatures close to darkfield of vertical and horizontal 100 & 70nm lines

No pattern below 40 nm dense (resist limit)

(best focus, 25mJ/cm² (100/70/50nm DL), 22.5mJ/cm² (30nm horiz. line)

Overview for 30-100 nm Coded Dense Lines


100nm Coded Horizontal Hole Diameters (1:2) vs. Distance from Bright Field

0

50

100

150

200

250

0 20 40 60 80 100 120 140 160

Distance from Bright Field [µm]

Dia

met

er [n

m]

short diam. long diam.

100nm Coded Vertical Posts (1:2) vs. Distance from Center of Bright Field

0

50

100

150

0 20 40 60 80

Distance from Center of Bright Field [µm]

Dia

met

er [n

m]

short diam. long diam.short diam. hor. post long diam. hor. post

100nm Coded Vertical Hole Diameters (1:2) vs. Distance from Bright Field

0

50

100

150

200

250

0 20 40 60 80 100 120 140 160

Distance from Brightfield [µm]

Dia

met

er [n

m]

short diam. long diam.

CD vs. Distance for 100 nm coded (1:2) Holes and PostsHoles Posts

Quadrupole illumination.Large CD difference for vertical and horizontal features.Posts: both diameters strongly reduced due to bright surrounding.Opposite curvatures of hole and post plots.H/v difference consistent with line results.


Outline

Introduction- Motivation of the ExperimentIntroduction- Motivation of the Experiment

Small-Field Printing ExperimentsSmall-Field Printing Experiments



Mask LayoutMask Layout



Settings:αs = 1.00, σs = 25 µmShort range:αs = 0.55, σs = 12 µm

Long range:αl = 0.45 σl = 100 µm

0

20

40

60

80

100

120

0 20 40 60 80 100 120 140 160


Line

wid

th [n

m]

measured data

simulation

0

0

0

0

0

0

0

measured data

simulation

( )⋅Θ⋅+Θ−⋅⇒ 0)1()()( IxIxI

The CD vs. distance dependency - is related to the lateral arrangement of layout pattern and bright field- can be fitted for vertical lines by superposition of short and long range

Gaussian flare distributions (1-D)

Double Gaussian Flare Model

Long range flare causes CD non-uniformities over distances > 100µm (1/e drop over ~30µm distance)

Distance x from Bright Filed [µm]Li

ne w

idth

[nm

]0

0

0

0

0

0

0

measured data

simulation

⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛

⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

′⋅′⋅

⋅⋅Θ⋅+Θ−⋅⇒ ∫∞

∞−

⎟⎟⎠

⎞⎜⎜⎝

⎛ −′−

xdexIxIxI s

xx

ss

2

21

0 )(2

1)1()()( σρπσ

α⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛

⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

′⋅′⋅

⋅+′⋅′⋅

⋅⋅Θ⋅+Θ−⋅⇒ ∫∫∞

∞−

⎟⎟⎠

⎞⎜⎜⎝

⎛ −′−∞

∞−

⎟⎟⎠

⎞⎜⎜⎝

⎛ −′−

xdexxdexIxIxI ls

xx

ll

xx

ss

22

21

21

0 )(2

1)(2

1)1()()( σσ ρπσ

αρπσ

α


Comparison to Uniform Pattern Density

8.5µm

Elbows in bright field (LBNL subfield)

Area below PW Graph of horizontal DL : 0.90201

Area below PW Graph ofvertical DL: 0.90498

Process windows of 60nm dense lines in LBNL subfield differ by only 0.3%

No h/v shadowing-effects with etched multilayer mask.Basically same h/v pattern density over flare range.→ Result: same h/v CD and process window for brightfield elbow pattern.

Example for 60nm: - h/v CD-difference = - 0.02 nm ± 1.21 nm (1σ) for best focus/dose- PW difference = 0.3% (area below PW graph)


Observed Asymmetry

Horizontal and vertical lines have different orientation with respect to brightfield and direction of EUV incidence.Observed h/v asymmetry cannot be explained by shadowing effects, mask CD difference or aberrations.

Indication for asymmetric PSF (e.g. due to forward scattering):– Smaller flare intensity in the “upper” part of the pattern (vertical)

→ wider lines and posts, smaller holes

– Higher flare intensity in the “lower” part of the patterns (horizontal)→ smaller lines and posts, wider holes

Further verification– experimentally by printing with mask rotated by 180°– by simulation

vertical

horizontal


Outline

Introduction- Motivation of the ExperimentIntroduction- Motivation of the Experiment

Small-Field Printing ExperimentsSmall-Field Printing Experiments



Mask LayoutMask Layout



Summary/Conclusions

Strong layout-dependent flare effects on critical dimensions

Stronger impact for smaller dimensions

CD vs. distance dependency can be described by short-range (12 µm) and long-range (100 µm) contributions

Horizontal lines significantly stronger affected than vertical lines for the given layout

CD effects on elongated holes and posts consistent with those on lines

H/V difference presumably due to:

– different pattern positions with respect to bright field and incident EUV radiation

– long range forward scattering of EUV radiation

“Flare-OPC” will be required to take long-range effects into account.


Acknowledgement

S. Hirscher, K. Lowack, W.D. Domke, A. GrahamInfineon Technologies, Munich/Dresden/Erlangen,

Germany

A.R. Pawloski, B. La FontaineAdvanced Micro Devices, Sunnyvale, CA

C. Holfeld, U. DerschAdvanced Mask Technology Center, Dresden, Germany

F. Letzkus, J. ButschkeIMS-Chips, Stuttgart, Germany

This work was partly financially supported by the Federal Ministry of Education and Research of the Federal Republic of Germany (Project No 01 M 3064A). The authors are responsible for the content of the paper.

Documents

Impact of Layout Dependent Flare on Printing in EUV Lithographyeuvlsymposium.lbl.gov/pdf/2005/pres/33 2-MA-32 Kamm.pdf · 2015. 11. 24. · Strong layout-dependent flare effects on