understanding euv resist stochastic effects through...
TRANSCRIPT
Greg Denbeaux [email protected] 1
Understanding EUV resist stochastic effects
through surface roughness measurements
SUNY Polytechnic Institute – Colleges of Nanoscale Science and Engineering
257 Fuller Rd. Albany, NY, USA 12203
February 23, 2020
IEUVI Resist TWG meeting
Greg Denbeaux, Eric Liu, Amir Hegazy, Hyeonseon Choi, Cole
Gregory, Belle Antonovich, Steven Grzeskowiak, Mueen
Mattoo, Robert Brainard
This work is funded by SRC
Greg Denbeaux [email protected] 2
Chemically amplified resists have multiple components
Multiple components
- Current issues with distribution of components
- Upcoming issues with limited number of photoactive (PAG) molecules that can
fit within smaller features
Greg Denbeaux [email protected] 3
Secondary electron interactions in EUV lithography
λ
Incident photon
hν ≈ 92 eV
e-
p+
p+
p+
e-
p+
e-
e-
p+ = hole
e- = electron
References1S. Grzeskowiak, et al., J. Micro/Nanolith. MEMS MOEMS 17 (3), 033501 (2018).2T. Kozawa and S. Tagawa, Jpn. J. Appl.Phys., 49 (3) (2010) 030001.3A. Narasimhan, S. Grzeskowiak, et al., Proc. SPIE, 9779 (2016) 97790F.4J. Torok, et al., J. Photopolym. Sci. and Technol., 26 (5) (2013) 625–634.5P. de Schepper, et al., Proc. SPIE, 9425 (2015) 942507.6T. H. Fedynyshyn, et al., Proc. SPIE, 5039 (2003) 310.
• Photons liberate electrons in the resist and
possibly cause chemistry to occur in the process.
• Further chemistry due to electron-resist
interactions.
• These electron travel away from the EUV
absorption site resulting in blur.
Greg Denbeaux [email protected] 4
Photoresist Exposure Tools
Substrate
Photoresist
Spin Coat
EUV (~ 92 eV) light
Expose
5-80 eV Electrons
or
Expose
ERIC (Electron Resist Interaction Chamber)ROX (Resist Outgassing Exposure Chamber)
Greg Denbeaux [email protected] 5
Outline
1. Resist stochastics
2. Homogeneity measurements for PMMA and a chemically
amplified open source resist
3. Intentionally segregated resists
Greg Denbeaux [email protected] 6
Failure rate for contacts (and other features)
is too large!
Peter De Bisschop,
“stochastic printing failures in
EUV lithography”, JM3 vol 17,
no 4 2018
For this resist and process,
41 nm pitch contacts with
dose 30-40 mJ/cm2
The failure rate is ~ 10-6 !!!
Greg Denbeaux [email protected] 7
How to measure local chemical inhomogeneity on
the ~ 10-50 nm scale?
• We can’t find any suitable metrology technique to measure local
chemical variations
• Would need spatial resolution of 10’s of nm
• Would need speed to make millions or billions of
measurements
• Our goal is to measure the effect of the chemical inhomogeneity
Greg Denbeaux [email protected] 8
Prior work (SPIE 2019) has shown the electron range – for
PMMA at 80 eV to be about 2 nm
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 2 4 6 8 10 12
PMMA – 80 eV
Depth [nm]
Rea
cti
on
Pro
ba
bilit
y
Blur = 1.8 [nm]
where Blur represents the
depth reactions have a 67%
probability of being induced
within this value.
Dose [µC/cm2]
Th
ick
ne
ss
Lo
ss
[n
m]
y = 1.4565ln(x) - 0.3716R² = 0.9966
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0.01 0.1 1 10 100 1000
PMMA – 80 eV
Use fit of data to generate
probability distribution of reactions.
Greg Denbeaux [email protected] 9
exposed unexposed
Partial exposure
Conventional aerial imageOur top-down “aerial image”
Using low energy electrons
exposed
unexposed
Can tune the range of
partial exposure – MUCH
sharper than available
conventional exposures
Resulting line edge roughness
(LER) is a combination of aerial
image contrast, photon
stochastics, with fundamental
material inhomogeneities and
deprotection stochastics
With a dramatically sharper exposure
contrast, and better metrology across
large areas, we can measure the resist
non-uniform response over large areas
for statistics far beyond 3s
Resist measurements
Greg Denbeaux [email protected] 10
Top down exposures provide much sharper contrast
to separate exposure contrast and statistics from material
response
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5 10 15 20
Re
acti
on
s/vo
lum
e (
no
rmal
ize
d)
distance (nm)
0.3 NA EUV imaging(lateral aerial image)
0.5 NA EUV imaging(lateral aerial image)
80 eV electrons(top down)
Slope can be
adjusted by varying
electron energy
Greg Denbeaux [email protected] 11
Cartoon of exposure and sample for measurement
Low energy electron exposure to limit range,
increase vertical contrast and to provide the
same chemical response as EUV photons
Pits and bumps should occur due to
local resist sensitivity or solubility
variations
Greg Denbeaux [email protected] 12
After exposure and development there are
multiple sources of roughness
A. Original roughness after spin coating
B. Added roughness due to development – local dissolution rate variations
C.Added roughness due to exposure (electron) statistics
D.Added roughness due to acid statistics and range (for chemically amplified
resists)
E. Final result of local variations in sensitivity due to local resist chemical
inhomogeneity
Greg Denbeaux [email protected] 13
Outline
1. Resist stochastics
2. Homogeneity measurements for PMMA and a chemically
amplified open source resist
3. Intentionally segregated resists
Greg Denbeaux [email protected] 14
0
0.2
0.4
0.6
0.8
1
1.2
0 10 20 30 40 50 60 70
RM
S (n
m)
Position
Measurement Position vs RMS
2.60μC/cm2 7.79μC/cm2
Avg. RMS = 0.63nm Avg. RMS = 0.84nmNX
Avg. RMS = 0.54nm
Each data point is the result
of an AFM scan
Electron exposure spots
causing roughness
change are about 1 cm
wide
Roughness variations across the exposure spot
(PMMA resist – not chemically amplified)
• Exposure times are a few
seconds with 80 eV
electrons
• Typical electron spacing at
these doses is about 2 nm
apart (~ 1013 electrons/cm2
incident)
Greg Denbeaux [email protected] 15
Post PAB
Post PAB +
DevelopedPost PAB + Exposure + Developed
Si Substrate
Photoresist
Dose (2.60μC/cm2) Dose (7.79μC/cm2)
PMMA Resist AFM scans
Greg Denbeaux [email protected] 16
0
0.5
1
1.5
2
2.5
0 10 20 30 40 50 60
RM
S (n
m)
Position
Measurement Position vs Nor. RMS3.46μC/cm2 7.79μC/cm2
Avg. RMS = 1.75 nm Avg. RMS = 1.80nm
NX
Avg. RMS = 1.07nm
Roughness variations across the exposure spot (OS2 resist – chemically amplified)
Greg Denbeaux [email protected] 17
Post PAB
Post PAB +
Developed
Post PAB + Exposure + Developed
Si Substrate
Photoresist
Dose (3.46μC/cm2) Dose (7.79μC/cm2)
OS2 Resist AFM scans
Greg Denbeaux [email protected] 18
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
PAB No dose Low dose High dose
RM
S r
oughness (
nm
)
PMMA
A. Original roughness after spin coating
B. Added roughness due to development – local dissolution rate variations
C. Added roughness due to exposure (electron) statistics
D. Added roughness due to acid statistics and range (for chemically amplified
resists)
E. Final result of local variations in sensitivity due to local resist chemical
inhomogeneity
A
A+B
A+B+C+E
Roughness increases through development and exposurePMMA results
Greg Denbeaux [email protected] 19
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
PAB No dose Low dose High dose
RM
S r
oughness (
nm
)
PMMA
CAR
A. Original roughness after spin coating
B. Added roughness due to development – local dissolution rate variations
C. Added roughness due to exposure (electron) statistics
D. Added roughness due to acid statistics and range (for chemically amplified
resists)
E. Final result of local variations in sensitivity due to local resist chemical
inhomogeneity
A
A+B
A+B+C+D+E
Roughness increases through development and exposureComparing CAR to PMMA
Greg Denbeaux [email protected] 20
Outline
1. Resist stochastics
2. Homogeneity measurements for PMMA and a chemically
amplified open source resist
3. Intentionally segregated resists
Greg Denbeaux [email protected] 21
Separating the multiple effects is a challenge
A. Original roughness after spin coating
B. Added roughness due to development – local dissolution rate variations
C. Added roughness due to exposure (electron) statistics
D. Added roughness due to acid statistics and range (for chemically amplified
resists)
E. Final result of local variations in sensitivity due to local resist chemical
inhomogeneity
So, we started formulating resists to intentionally segregate to see the effects
Copolymer (PHS/TBA)
Standard Intentionally mix in
homopolymers
Polyhydroxystyrene
(PHS)
Polystyrene
(PS)
Greg Denbeaux [email protected] 22
First test – varying the polystyrene
homopolymer content
- A-4 sample (4% PS)
29% PHS/TBA (60/40)
15% PAG
4% PS
52% PHS
- A-8 sample (8% PS)
29% PHS/TBA(60/40)
15% PAG
8% PS
48% PHS
- A-12 sample (12% PS)
29% PHS/TBA (60/40)
15% PAG
12% PS
44% PHS
Large segregation effects,
possibly related to the
polystyrene content
~ 100-200 nm features!
Much bigger than the desired
rate of segregation effects
Greg Denbeaux [email protected] 23
Why does segregation occur –
vary thermal processingHigher dose -> -> -> -> -> -> -> -> -> -> lower dose No dose
100C PAB
100C PEB
120C PAB
120C PEB
140C PAB
140C PEB
No PAB
120C PEB
Similar results regardless of
bake temperature
Segregation does not appear to
be a thermal process
Possibly occurring during
coat step as solvent
evaporates and local
concentrations increase
Greg Denbeaux [email protected] 24
Simpler approach to segregated samples
1. 92% PHS/TBA, 8% PAG
2. 96% PHS/TBA, 4% PAG
3. 98% PHS/TBA, 2% PAG
We removed the intentionally segregating
components of the PS and PHS and used
only the copolymer and PAG
Three new samples
After development but no exposure, there were already surface profile
variations due to local dissolution rates
Greg Denbeaux [email protected] 25
Analysis approaches
We tried looking at variations in the power spectral density (PSD)
We tried identifying features based on their height above the average
We tried Fourier Transform approaches to filter for selected feature spatial
frequencies (their size)
The best approach we found was simply the histogram of the height measurements
across the sample, directly from the AFM
Greg Denbeaux [email protected] 26
Histogram results
0
10
20
30
40
50
60
70
80
90
100
-3 -1 1 3
Pro
bab
ilit
y (
AU
)
surface height (nm)
8% PAG sample, outside of exposure spot
0.0001
0.001
0.01
0.1
1
10
100
-7 -5 -3 -1 1 3 5 7P
rob
ab
ilit
y (
AU
)
surface height (nm)
8% PAG sample outside of exposure spotcompared to Gaussian
Resist data is higher
than the Gaussian
due to pits in the
surface
Resist data is higher
than the Gaussian
due to bumps on the
surface
The result looks reasonable…
on a linear scale
On a log scale, you can see the
deviations from the expected
Gaussian profile, even for low rates
of occurrences across the surface
Greg Denbeaux [email protected] 27
What about in the exposed regions
The images look similar by eye, but
the histogram shows the increased
number of bumps in the surface in
the higher PAG concentration
samples
Greg Denbeaux [email protected] 28
Conclusions
• Multi-component resists have signatures of local chemical inhomogeneity
• This shows up as local resist solubility rates (unrelated to exposure)
• This shows up as local resist sensitivity variations (related to exposure)
• All exposure processes also have exposure stochastics
• Since the EUV exposure process is based primarily on electron and hole
chemistry, electrons are a suitable proxy for the exposure process
• Top down electron exposures followed by surface topography measurements
can detect local chemical segregation effects
Greg Denbeaux [email protected] 29
Acknowledgements
Students at SUNY Polytechnic
Institute who have work on this
project:
Grad students
Eric Liu
Amir Hegazy
Steven Grzeskowiak
Amrit Narasimhan
Undergrad students
Hyeonseon (Sunny) Choi
Cole Gregory
Belle Antonovich
Professor Robert Brainard
Thanks to SRC for funding this
research project