Lithographically generated LER in photoresist 2...→ It defines the theoretical locations of exposed and unexposed material 0 50 100 150 200 250 300 0 0.2 0.4 0.6 0.8 1 X coordinate

Lithographically Generated Line Edge Roughness in

Photoresist

Adam R. Pawloski, Alden Acheta, Bruno La Fontaine, Scott Bell, Harry J. Levinson

Advanced Micro Devices

January 13, 2005 2Adam R. Pawloski et al.– Advanced Micro Devices

• A few remarks why LER is bad

• At least 5 major factors contributing to LER in photoresist – The quality of the aerial image– The edge roughness of the photomask absorber– Photon statistics– Chemical transformations that occur during image formation– The physical chemistry of the resist surface

• Outlook – What solutions to the LER problem are available? – What role do etch engineers play?

Outline


Why LER is Bad, Bad, Bad

• CD Control and Device Performance→ Across chip linewidth variation (ACLV)!

• LER is CD control at relevant frequencies (we have the tendency to think of LER with respect toward the limit as

, but CD control as )

→ The industry has more than 20 years of data that tells us leakage is reduced by making gate dimensions more uniform across the chip and yield goes up when reducing variations across the wafer

→ Simulation data (and some limited experimental data) show that device fluctuations get much, much worse when LER is large

→ Better CD control = less leakage = low power = $$$

0→y ∞→yx

y


avgy1avgy2

x

)(1 xy )(2 xy

avgavg yyCD 21 −=

CD

{ }2

)()(33

221

−

∑ −−=

N

CDxyxyLWR i

ii σ

( )1

)(33

211

−

∑ −=

N

yxyLER i

avgi

σ

∫×= dffLER PSD LER )(233σ

0

2)(2xx

fY

f −=LERPSD

( )∫ −= −fx

x

ifxavg dteyxyfY0

211 )()( π

Line Edge Roughness Measurement

0.00001

0.00010

0.00100

0.01000

0.10000

1.00000

0.1 1 10 100 1000Frequency (um-1)

Po

wer

Sp

ectr

al D

ensi

ty (n

m2/

um

-1)

EUV 22.2 mJ/cm2EUV 23.1 mJ/cm2EUV 24.0 mJ/cm2EUV 25.0 mJ/cm2EUV 26.0 mJ/cm2EUV 27.0 mJ/cm2EUV 28.1 mJ/cm2EUV 29.2 mJ/cm2EUV 30.4 mJ/cm2EUV 31.6 mJ/cm2EUV 32.8 mJ/cm2193nm Resist at Min LER (ILS>18um-1)


Quality of the Aerial Image

• The aerial image is the intensity distribution of light from the exposure tool incident upon the wafer

→ It defines the theoretical locations of exposed and unexposed material

0 50 100 150 200 250 3000

0.2

0.4

0.6

0.8

1

X coordinate (nm)

Rel

ativ

e In

ten

sity

• Think of an aerial image as modulation

→ Very little in the exposure is completely dark or completely bright

→ It becomes more difficult to generate high contrast aerial images as features get smaller (decreasing pitch)

Mask



• The “steepness” of the aerial image at the edge of the patterned feature plays a large role on resist pattern quality

• The closer an aerial image is to a step function in intensity, the better

→ CD control improves with increasing “steepness” of the aerial image because the resist CD at the pattern edge is less sensitive to local fluctuations in dose

Examples of aerial images for different linewidths at 300nm pitch

-150 -100 -50 0 50 100 1500

0.2

0.4

0.6

0.8

1

x coordinate (nm)

Rel

ativ

e In

ten

sity

100nm125nm150nm175nm200nmideal 100nm

( )

∂∂=

0

)(1

0 xxxI

xIILS

• The aerial image parameter that quantifies profile “steepness” is the image-log-slope (ILS)



• The most significant contributor to LER is the quality of the aerial image

→ Pitch, illumination, line width, flare, fading, etc.

• LER is a minimum as the image-log-slope (ILS) of the aerial image is a maximum

• As the ILS approaches infinity, the magnitude of LER reaches a non-zeroconstant

→ In this regime we begin to see the effects of polymer physics

Further improvements in aerial image quality do not reduce LER further

*A.R Pawloski et al., Proc. SPIE, vol. 5376, 2004


Mask Absorber Roughness

• Since the photomask is made using lithography and photoresists, it also has some magnitude of LER

• Can mask absorber roughness transfer to the wafer as resist LER?

CD SEM images illustrating typical line quality for absorber lines from binary mask fabrication


Mask Absorber Roughness

0

0.2

0.4

0.6

0.8

1

1 10 100Spatial Frequency (µm-1)

LE

R T

ran

sfer

Fu

nct

ion

KrF 0.85NAArF 0.85NAArF 1.1NAEUV 0.1NAEUV 0.25NA

• The LER transfer function from the mask absorber to the resist was derived by Naulleau and Gallatin (Appl Opt, vol. 42, pp. 3390-7, 2003) for EUV imaging

• Following these procedures, the LER transfer function was calculated for a variety of exposure tools from KrF through EUV

• As the resolving power of the optical system improves (moving to higher NA and shorter wavelength) more mask LER is transferred to the wafer leading to resist LER

→ Low frequency roughness in resist

• Negligible contributions for most optical lithography processes and masks


Photon Statistics

• Decreasing the exposure wavelength, the exposure dose, or the quality of the aerial image increases the probability that photon statistics (shot noise) contribute to LER

• At EUV, shot noise could contribute to increasing the magnitude of LER for fast resists (fewer, higher energy photons at EUV)

• For optical lithography, shot noise effects are small, if at all present, and likely are overshadowed by other contributions to resist LER

→ However, photon statistics beget photoacid molecule statistics!

0

1

2

3

4

5

6

7

8

9

10

0 5 10 15 20 25 30 35

Resist sensitivity (mJ/cm2)

LER

3 (n

m)

EUV - ILS=15 um-1

EUV - ILS=20 um-1

EUV - ILS=25 um-1

EUV - ILS=30 um-1

193 - ILS=15 um-1

193 - ILS=20 um-1

Results from stochastic simulation assuming Poisson statistics to describe the arrival of photons at the resist.

The resist contribution is simulated using a 40nm gaussian blur to describe photoacid diffusion


The Chemical Composition of Resist

• The resist film is made up mostly of polymer

• A photo-acid generator produces photoacid catalyst upon exposure

• A quencher (base) provides profile control, environmental stability and lower LER

• Other additives, like dissolution inhibitors and surfactants are added to improve dissolution behavior and contrast

OH

O

O

X Y Z

S+ S

O

CF2CF2CF2CF3

O

O-

NH3C(CH2)7 (CH2)7CH3

(CH2)7CH3

Phenolic polymers typical of

248nm resists

Methacrylate polymers typical of

193nm resists

O

O

O

O

OO

X Y

Triphenylsulfonium 1-perfluorobutane

sulfonate PAG

Trioctylamine base

Uncle Joe’s secret sauce, wing of bat, pixie dust, etc.???


Latent Image Formation at the Pattern Edge

• Photogenerated acid catalyzes the removal of blocking groups from the polymer backbone, converting to hydrophilic, base soluble functional groups (acids and alcohols)

• Reaction catalysis is coupled with thermal diffusion during the post-exposure bake (PEB)

Nominally unexposed

Nominally exposed

Extent of polymer

deprotection

Transition region

Photoacid diffusion

and reaction catalysis

• Photoacid statistics and gradients are important

• The final distribution of reaction products at the end of the PEB defines latent roughness prior to development

→ Which polymer chains dissolve and which remain


Remaining resist

Resist developed

away

Extent of polymer

deprotection

Transition region

Aqueous base developer

• During development, polymer chains dissolve into solution

• The dissolution mechanism for resist is also based on chemical reaction (acidic functional groups are converted to ions allowing solubil ity in aqueous base while remaining insoluble in water)

Resist Dissolution at the Pattern Edge

• Do polymer chains dissolve individually, or do they come out at aggregates, possibly contributing to roughness?

• How do the thermodynamics of the resist-developer interface contribute to roughness formation?

• Are these chemical mechanisms limited to high frequency roughness, or are there cooperative, low frequency effects?


The Photoacid Gradient at the Pattern Edge

Increasing base concentration

Increasing base

concentration

Adding base to the resist

formulation increases the

gradient in photoacid

concentration at the edge of the

resist

Ex: The action of adding base to

reduce LER

*A.R. Pawloski et al., Arch Interface Conf. Proceedings, Tempe, AZ 2004


Physical Chemistry of the Resist Surface

• LER can be altered by the physical chemistry of the surface of the resist in contact with developer or a rinse liquid

Resist after DI water

rinse

Resist after surface

conditioning rinse (Air Products)

Increased Fading (decreasing aerial image quality) à

• Polymer molecules at the surface of the resist adapt conformations to minimize the energy of interaction across the resist/developer or resist/rinse liquid interface

• If methods are found to control this thermodynamic interaction, it may be possible to reduce mid and high frequency roughness

1.00E-05

1.00E-04

1.00E-03

1.00E-02

1.00E-01

1.00E+00

1.00E+01

0.1 1 10 100 1000Frequency (µm-1)

PS

D (

nm

2/µm

-1)

Control, 10s, AL00333 s04

J09, 10s, ARP01 s09

0ppm AMag

0 ppm 3 ppm 6 ppm 9 ppm 11 ppm

Reduction in LER!1.00E-05

1.00E-04

1.00E-03

1.00E-02

1.00E-01

1.00E+00

1.00E+01

0.1 1 10 100 1000Frequency (µm -1)

PS

D (

nm

2/µm

-1)

Control, 10s, AL00333 s04

J09, 10s, ARP01 s09

11ppm AMag


Strategies for LER Reduction

• Minimization of LER requires a holistic approach

• Critical performance needed from suppliers, designers, and process

Minimum Impact of LER

on Device Performance

Design Devices with Increased

Tolerance to LER

Minimize LER Transfer

Through Etch

Minimum Impact of LER

on Device Performance

Design Devices with Increased

Tolerance to LER

Minimize LER Transfer

Through Etch

Minimize Generation of Lithographic LER in Resist

Use Litho-FriendlyLayouts

Optimize Aerial Image

Quality

Minimize Mask Absorber

Roughness

Expose Resist Above Shot Noise Limits

Optimize Resist

Process

Design and Formulation of Resist Materials with

Reduced LER

Reduced photospeed

Large ILSFocus control

Low aberrations

Minimumvibrations

Dose control

Optimum illumination

Optimum mask lithography

Optimum pattern transfer into absorber

Single pitch

Full OPC

Large photoacidconcentration gradients

Favorable thermodynamicsPAB, PEB, and development

RinseTransparency

New polymer designsPAG and base loading

Environmental stability

Accurate and Precise Metrology

Quantification methods

Standards

Film stack

Resist thickness

Low flare


Pre- and Post- Etch LER14

mJ/

cm2

16 m

J/cm

218

mJ/

cm2

20 m

J/cm

222

mJ/

cm2

Resist Poly

Resist Poly


0.001

0.01

0.1

1

0.1 1 10 100 1000

Frequency (um-1)

LE

R P

SD

(nm

2/u

m-1

)

Post Litho - Resist

Post Etch - Poly

14mJ/cm2

0.001

0.01

0.1

1

0.1 1 10 100 1000

Frequency (um-1)L

ER

PS

D (n

m2/

um

-1)

Post Litho - Resist

Post Etch - Poly

16mJ/cm2

0.001

0.01

0.1

1

0.1 1 10 100 1000

Frequency (um-1)

LE

R P

SD

(nm

2/u

m-1

)

Post Litho - ResistPost Etch - Poly

18mJ/cm2

0.001

0.01

0.1

1

0.1 1 10 100 1000

Frequency (um-1)

LE

R P

SD

(nm

2/u

m-1

)

Post Litho - Resist

Post Etch - Poly

20mJ/cm2

Post-Etch LER

After etch, LER in poly can be reduced from the LER of the

resist mask

Smoothing by good etch

processes may be able to affect more than just

the high frequency

components


Acknowledgements

Sincere appreciation is expressed to the following individuals for their assistance in many facets of this work:

• Karen Turnquest, Uzo Okoroanyanwu, Dave Kyser,Jongwook Kye, Jeff Schefske, Bryan Choo, Cyrus Tabery, Rex Ugalde, Cal Gabriel (AMD)

• Prof. Grant Willson, Andrew Jamieson, Tim Michaelson(Univ. Texas, Austin)

• Patrick Naulleau (CXRO, LBNL)

• Chuck Masud (ASML)

• Koji Toyoda (Canon)

• Manny Jaramillo, Peng Zhang, Maud Rao, Leslie Barber(Air Products)


Trademark Attribution

AMD, the AMD Arrow Logo and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other product names used in this presentation are for identification purposes only and may be trademarks of their respective companies.

Documents

Lithographically generated LER in photoresist 2...→ It defines the theoretical locations of exposed and unexposed material 0 50 100 150 200 250 300 0 0.2 0.4 0.6 0.8 1 X coordinate