NANOFABRICATION
NANOIMPRINT LITHOGRAPHY
Dr. Nikos Kehagias
Head of Nanofabrication division
Catalan Institute of Nanotechnology (ICN)
B
Barcelona
l
Spain
Email: nikolaos.kehagias.icn@uab.es
nikolaos kehagias icn@uab es
Outline
• Nanofabrication techniques:
¾ Resolution and limits
¾ Alternative nanolithography techniques
¾Nanoimprint lithography: issues, challenges, potentials
• Nanoimprint lithography applications:
¾ Examples of passive photonic devices
¾ Functional materials
¾ 2D PhC devices for enhanced light extraction
• Nanometrology
¾ Non destructive techniques
Nanotechnology: enabling multi-billion
dollar industry
Approach towards Nanotechnology
Key Requirements of Lithography for
Manufacturing ICs*
• Critical Dimension Control
– Size of features must be controlled within wafer and wafer‐to‐wafer
Size of features must be controlled within wafer and wafer to wafer
• Overlay
– For high yield, alignment must be precisely controlled
For high yield, alignment must be precisely controlled
• Defect Control
– Other than designed pattern, no additional patterns must be imaged
• Low Cost
– Tool, resist, mask; fast step‐and‐repeat 30‐40% of total semiconductor
manufacturing cost is due to lithography (Masks, resists, metrology)
– At the end of the roadmap, μP will require 39 mask levels
*ITRS 2005/6, Lithography
NANOFABRICATION
METHODS
Fabrication methods for small structures
Decrease in minimum feature size with
time (Moore’s
(Moore s law)
Nanopatterning techniques
• Nanolithography techniques
• EUV/UV lithography
lith h
• Electron beam lithography
• Focused ion beam lithography
• X-ray
X ray lithography
• Alternative lithography techniques
•Template assisted self assembly techniques
• Micro-contact printing
• Nanostensil technique (nano-mask
(nano mask lithography)
• Ink-jet lithography
• Nanoimprint lithography
Nanopatterning techniques
Patterningg time for 10% of a 4” wafer as function of obtainable line width for different
lithography techniques. The arrows and the question mark in the NIL bar indicate
that faster imprint times may be obtainable by optimizing the imprint process.
Template assisted self assembly techniques
““Self-assembly”
lf bl ” refers
f to the
h deposition
d off an organized
d layer
l onto a
substrate with a high-degree of control and/or ordering.
Colloidal self assembly
Set-up
Material: Polymethyl metacrylate with
a mean diameter of 368nm
(<5% polydispersity)
Substrate: Glass
Concentration: 4% wt in de-ionized
water
Acoustic vibration: white noise (40-
4kHz)
Drawing speed: stepping motor at
1.3mm/hr
Colloidal self assembly
Cross section SEM images: 3D ordering
Stochastic resonance-like behaviour
L – noise level
L40 - best
Standing wave formation
at high noise level leads to
locally suppression (or
optimization) of noise
vibration
Local but uncontrollable
i
increase i llattice
in tti ordering.
d i
Block copolymer self assembly
Requirements for graphoepitaxy
Patterned sidewall material
Mesa width ~ 50 nm - Would like to test both PS
Groove width
(or as narrow as possible) and PMMA wetting walls
40 – 200 nm
- Fab-friendly materials ideal
Wall height ~ 30 nm Base must have surface OH
groups to allow us attach
Neutral
eu a po
polymer
y e bbrush
us
Substrate
Patterned Silicon dioxide Patterned HSQ
Designed Nanostructures via Templating
Silicon-based trenches and Angled lamellae nanostructures
aligned nanostructures
nanostructures.
Nanostensil Lithography
The Nanostensil
Th N t il ttechnique
h i iis a patterning
tt i method
th d b
based
d on
shadow mask evaporation
A thin membrane is used as a solid mask to transfer the patterns
from the membrane to the substrate during the evaporation
Full-wafer stencils Side view stencil
Nanostensil Lithography
Main advantages:
• No resist, development or baking
• Non contact
• Re usable
• Micro and nanostructuring in a single step
• High flexibility of materials (metals, oxides, SAMs)
Challenges:
• Clogging occurs due to the accumulation of
deposited material on top and inside the membrane
apertures.
• Blurring
• Contamination of stensil
• Stiffness of stensil
Dip-pen Lithography
Ink transfer using a coated AFM tip by capillary effect.
Dip-pen lithography of SAMs
Introduction to Nanoimprint
Proposed by S.Y. Chou (Minnesota Uni., USA) in 1995
(Appl. Phys. Lett., 67, 3114 (1995))
Idea: a nanometer-size pattern is transfered not by electron, ion or
other beams, but by a stamp via mechanical contact between the
stamp and a substrate with a polymer.
Advantages:
• Cost
C efficient
ffi i
• High throughput
• High
Hi h resolution
l i
• Simple
• Flexible
Fl ibl
Current Fabrication methods
Electron Beam Lithography Extreme UV Lithography Scanning Probe
Lithography
Ad t
Advantages Ad t
Advantages Ad t
Advantages
• Very accurate control of • Extreme UV is 10‐14nm • Very good control of
pattern with direct writingg
p wavelength source
g pattern and resolution
p
• No mask needed • Resolution approaching Approximately
• Highly automated 30nm 10nm possible
• 5nm resolution possible • High Throughput • Highly automated
Disadvantages Disadvantages Disadvantages
•Very low throughput • Mask fabrication is difficult • Very slow process
Less than 10 wafers per • Reflective optics can be • Instrument can be
hour expensive costly
• Expensive CaF instead of SiO2 optics • Time of process
Hardware cost 6‐10 Million
Hardware cost 6‐10 Million • Cost of EUV startup ~50‐60
• Cost of EUV startup 50‐60 eliminates
million industrial feasibility
Alternative method:Nanoimprint Lithography
• Nanoimprint lithography (NIL) is simple in comparison to alternatives
• High throughput capabilities
• Low cost for a next-generation technology (No need for small λ laser sources
and optics)
• High cost in master mold, but all other molds can be made from this master
Lithography Resolution Cost (M $) Throughput Feasible
(nm)
248 nm 90 8 √ √
193 nm 45 20 √ √
157 nm 32 50 √ √
EUV
EUV 16 100 √ √
Ebeam 10 5‐10 x √
Imprint 14 1 √ √
R&D machines can be 1 master Æ10.000 sub mastersÆ100 million disks
purchased for 100k $250.000Æ $1.000 eachÆ10c per disk for mask cost
Nanoimprint enables multiple billion dollar
industries
MEMS/NEMS
Displays Wireless Com.
Com
NIL
Data storage Etc.
Biotech Semicond. IC’s
Pharmaceutical
Nanoimprint Lithography (NIL)
Stamp (Si
(Si, Quartz
Quartz, etc) Advantages
Resist (polymer, monomer) • Resolution (sub 10 nm)
Substrate • Fast (sec/cycle)
• Low
L costt ($0.2M
($0 2M vs $25M)
• Simple
Imprint • Flexible (UV, heat)
(Pressure +heat or UV light)
Applications
Release • Semiconductors
(
(cool
lddown ) • Optics
O ti
• Bio
• Organic electronics
RIE of residual layer • Sensors
High resolution Complex patterns Functional devices
Multimode NIL
Thermal Reversal NIL Reverse UV NIL
Transparent stamp with metal protrusions
UV light, pressure, heat
Inking Whole layer transfer
Development
Single NIL tool capable of multiple modes pattering/fabrication
Highly versatile, yet simple, nanofabrication tool
Step & Stamp/flash NIL
NPS 300 Nano imPrinting Stepper
• Thermal + UV nanoimprinting
• Up to 300 mm wafers
• Sub‐20 nm features
S b 20 f t
• 250 nm overlay accuracy
• Automatic alignment
~ 10 nm holes in polymer
T. Haatainen et al., VTT 2001
Step and Stamp nanoimprint lithography
In liquid alignment:
Pre- and post-exposure.
Defects caused due to material
failure in small features with large
S.V. Sreenivasan et. al., Semiconductor Fabtech, 25th ed., 111, 2005. feature height.
Roll-to-Roll NIL
Bendable Ni stamp
AFM images of stamp and imprint
• Printing speed from 0.3 to 20 m/min
• Line depth of 151 – 112 nm
• Min feature at 5m/min is 50 nm
Min feature at 5m/min is 50 nm
Courtesy of T. Mäkelä et al., VTT, Finland
Roll to roll NIL
Se Hyun Ahn et al., ACS Nano, 3, 8, 2304, 2009 Adv. Mater. 2008, 20, 2044–2049
Polymer stamps
State of the art of NIL techniques
Smallest/ Min Largest Overlay t align,
largest pitch wafer Accuracy t print,
features in (nm) printed (nm) t release,
same print (mm) t cycle
NIL 2 nm, Minutes,
10s, Min,
N/A 14 200 500
10-15 min
Full cycle 2.5
8 nm, min
i with,
ith 20 s
SSIL 50 nm/5 µm 50 200 250
without full
auto
collimation.
10 nm 50 300 50 20
SFIL 25 nm/ µm wafers/hr
UV-NIL 9nm/100µm 12 200 20 20s/step
3 wafers/hr
Soft
S ft UV
UV- 4-5
4 5 min
i ca.
25nm/ 20µm 150 200 1-50µm 12wafer/hr
NIL
NIL issues and complications
NIL
NIL metrologies critically needed:
t l i iti ll d d Blazed gratings
• Critical Dimension measurements of sub‐50 nm features
• Quantify fidelity of imprint pattern transfer
• Feedback on pattern quality needed to engineer and optimize NIL
db k li d d i d i i
NIL materials science :
• Resist material selection is done empirically Tearing of pillars
• Guidelines for imprinting functional materials?
Guidelines for imprinting functional materials?
• Imprinted nanostructures may have different properties
• Possible orientation and anisotropic properties
• Low temp and low pressure
• Minimal shrinkage
• Mechanical strength and tear resistance
• Mold fill Æ Viscosity
• Tg
• Tg for thermoplatic resist (imprint usually done 70
for thermoplatic resist (imprint usually done 70‐80°C
80 C above T
above Tg)
NIL issues and complications
Template
• Usually fabricated from Si, quartz, or nickel
• Critical dimension control
• Critical dimension control
• Defect free fabrication & Inspection
• Adhesion and use of antisticking coating on template
• Cleaning & re‐use
• designing for imprint uniformity –> Uniform residual layer
Courtesy of Dr C Gourgon (CNRS‐LTM)
Courtesy of Dr C. Gourgon (CNRS LTM)
Overlay accuracy
Moiré concentric circles
• NIL has no distortion due to lens (since no lens is used)
NIL has no distortion due to lens (since no lens is used)
• Smaller error budget for template pattern placement Aligned
• Mask/template distortion due to pressure and/or
temperature & defects
Misaligned
Principles of NIL
T‐P vs. time diagram of NIL process
Stamp
Polymer layer (a)
Substrate
Heat
Apply pressure (b)
Cool down
Demolding
Separate (c)
Residual layer (d)
Etch residual
(e)
layer
Viscosity dependance on MW, P and T
MW dependance
MW dependance
⎧Μ , M < Mc
η 0 ∝ ⎨ 3.4± 0.2 Temperature
⎩Μ , M > Mc dependance
C1 (T − Tg ) η (T )
log aT = − = log 0
Pressure dependance C 2 + T − Tg η 0 (Tg )
Δ ln η − 2 Δ ln η William Landel Ferry equation
≈ −4 × 10
ΔP (bar ) ΔT
Squeeze flow theory during a typical NIL process
z
Stamp z wi
S S/2
h pr
Polymer
l x vy=0 vy((z))
y h(t)
Substrate
N N N
Continuity equation: ho ∑ (s i + wi ) = h f ∑ (si + wi ) + h pr ∑ (wi )
i =1 i =1 i =1
Navier Stokes equation: ∇p = η 0 ∇ 2 u
Residual Polymer height Estimated imprinting time
1 1 2 F pr ηο s 2 ⎜⎛ 1
1 ⎞
⎟
= 2 + t tf = − 2
2
h (t ) h0 n0 Ls 3 ⎜ 2
2 P ⎝ h f ho ⎟
⎠
H. Schift and L.J. Heyderman, Nanorheology“.
Chapter 4 in, Alternative Lithography“, ed. C. Sotomayor‐Torres. Kluwer Academic (2003).
Polymers used in NIL
Material Glass Molecular Viscosity Solvent
transition weight
g
Temperature
PMMA 105 oC 75k 10 ± 2 Anisole
(mPas)
mr-L 6000 40 oC 7k 2,4 ±1 PGMEA/
mr-NIL 6000 (mPas) Anisole
mr-II 7000
mr 60 oC 120k 4±2 PGMEA
(mPas)
mr-I 8000 115 oC 120k 5±2 PGMEA
((mPas))
Polystyrene 100 oC 50k - Toluene
(PS)
Typical refractive index values for polymer are between 1.3 ‐ 1.6
Functional polymers
Polymers with embedded NC’s (CdSe, CdSe, etc.),
NP (Au, TiO
( , 2 etc.), Dyes (Rhodamine etc.)
2, ), y ( )
Polymers with embedded NP (Au, TiO2, etc.)
Surface modification of polymers (nanoparticle deposition,
change of the polymer surface tension, etc.)
h f th l f t i t )
[ -6.63V -> -3.96V ] -9.02V -> -1.99V [V]
-2.00
5.00
-3.00
Di‐block co‐polymers (PS‐b‐PMMA) -4.00
-5.00
[ µV ]
-6.00
-7.00
-8.00
-9.00
0
0 [ µV ] 5.00
Conductive polymers (polypyrrole, polyaniline etc.)
NIL process challenge: imprint quality
control
Fundamental process challenges
Critical Dimensions
Critical dimensions (CD)
Critical dimensions (CD)
Width Height
Slope
Residual layer
1 μm
Residual layer thickness and uniformity
over large areas (> 300mm) Residual layer
100 nm
Nanoimprint lithography process
Stamp Stamp with different size protrusions
Polymer layer
Substrate
Heat
Imprint
C l down
Cool d
Stamp bending
Separate
Etch residual
layer Different filling factors Æ lead to
inhomogeneous residual layer
Photonic circuit
Combination of variable scale features on the same stamp
Mathematical model
The resist movement is determined by the 2D pressure distribution P(x,y,t) calculated from the following
problem:
{ 3
}
∇ [D( x, y ,t ) + h( x, y )] ∇P ( x, y ,t ) = 12η
∂D( x, y ,t )
∂t
, ( x, y ) ∈ Ω f , t ∈ (0,T ], P ( x, y ) = 0, ( x, y ) ∈ Ω / Ω f ,
(1)
t
D( x, y ,t ) = d0 − ∫ Vst (ζ ) dζ + δ st ( x, y ,t ) + δ sb ( x, y ,t ),
0
where d0 is the initial resist thickness; h is the stamp relief height; δst and δsb are the normal displacement
of the stamp and substrate surface, respectively; Vst is the stamp velocity; T is the duration of the
imprinting process; η is the dynamic viscosity of the resist; Ω is the considered domain of the stamp; Ωf is
the part of Ω, in which all cavities are filled with the resist.
q
Equations ((1)) is derived from 3D Navier-Stokes equations
q with the understanding
g that the resist motion is
largely directed along the substrate surface.
For the calculation of δst and δsb, the stamp and the substrate are represented as semi-infinite regions (an
elastic medium bounded by a plane). In this case, the elastic normal displacement is described by the
following expression:
1− σ 2 P ( x ′, y ′,t ) dx ′dy ′
δ ( x, y ,t ) =
πE ∫∫
Ω ( x − x ′) + ( y − y ′)
2 2
, ( x, y ) ∈ Ω, t ∈ (0,T ],
where
h σ is
i Poisson's
P i ' ratio
ti and
d E is
i modulus
d l off elasticity.
l ti it
Experimental parameters
Polymer used: PMMA and mr-I8030E
Initial polymer thickness: 340 nm and 318 nm
Imprinting
p g temperatures:
p 180 oC - 200 oC
Dynamic Viscosity : 2×104 Pa⋅s @ 180°C and 3×103 Pa⋅s @ 200°C.
Chirped grating structures stamp was used
Stamp relief: ~300 nm
Stamp design
Simulation parameters:
• stamp velocity:1 nm/s,
• duration of the imprinting process: 268 sec
• grid size:128×128 pixel
Instruments used:
• Dektak profilometer (Veeco instruments)
• Reflectometer
Resist PMMA 75K.
Imprinting parameters: the stamp cavities depth - 300 nm, the initial resist thickness - 340 nm,
the imprint temperature - 190°C, the resist viscosity - 104 Pa⋅s.
Experiment Simulation
500
400
experiment Comparison
p of measured and
simulation
H, nm
300 simulated values of resist
thickness
200
Accuracy
100
1.5% 1.3% 2
2.6%
6% 2
2.1%
1%
0.5% 0.5% 1.5%
1 9% 2
1.9% 2.4%
4% 2 4%
2.4%
0% 0%
0
0 2 4 6 8 10 12
zone number
Resist mr-I 8000 (Micro Resist Technology GmbH).
Imprinting parameters: the stamp cavities depth - 300 nm, the initial resist thickness - 318 nm,
the imprint temperature - 180°C, the resist viscosity - 2×104 Pa⋅s.
600
500
400
H, nm
300
200
100 simulation
experiment
0
-4200 -4000 -3800 -3600 -3400 -3200 -3000
x, μm
(a) (b)
600 600
500 500
400 400
H, nm
H, nm
300 300
200 200
100 simulation 100 simulation
experiment experiment
0 0
-4200 -4000 -3800 -3600
x, μm
-3400 -3200 -3000
(c) -4200 -4000 -3800 -3600
x, μm
-3400 -3200 -3000
(d)
((a)) The
Th optical
ti l microscopy
i i
images off the
th test
t t structure
t t i
imprinted
i t d in
i the i t att 180°C. Horizontal
th resist H i t l color
l lines
li
indicate zones of profilometer measurements of resist thickness. White isolines specify the calculated
distribution of the stamp/substrate deformation (numbers signify the elastic displacement in nanometers). (b)-
(d) Comparison of measured and simulated profiles of resist thickness for the test structure.
Resist mr-I 8000 (Micro Resist Technology GmbH).
Imprinting parameters: the stamp cavities depth - 300 nm, the initial resist thickness - 318 nm,
the imprint temperature - 200°C, the resist viscosity - 3×103 Pa⋅s.
600
500
400
H, nm
300
200
100 simulation
experiment
0
-4400 -4200 -4000 -3800 -3600 -3400 -3200
x, μm
(a) (b)
600 600
500 500
400 400
H, nm
H, nm
300 300
200 200
100 simulation 100 simulation
experiment experiment
0 0
-4400 -4200 -4000 -3800
x, μm
-3600 -3400 -3200
(c) -4400 -4200 -4000 -3800
x, μm
-3600 -3400 -3200
(d)
((a)) The
Th optical
ti l microscopy
i i
images off the
th test
t t structure
t t i
imprinted
i t d in
i the i t att 200°C. Horizontal
th resist H i t l color
l lines
li
indicate zones of profilometer measurements of resist thickness. White isolines specify the calculated
distribution of the stamp/substrate deformation (numbers signify the elastic displacement in nanometers). (b)-
(d) Comparison of measured and simulated profiles of resist thickness for the test structure.
Viscosity estimation for resist mr-I 8000 at 180°C.
the resist dynamic viscosity = 3 ×103 Pa⋅s the resist dynamic viscosity = 10 4 Pa⋅s
500 500
400 400
H, nm
H, nm
300 300
200 200
-4800 -4600 -4400 -4200 -4000 -3800 -3600 -3400 -3200 -3000 -2800 -4800 -4600 -4400 -4200 -4000 -3800 -3600 -3400 -3200 -3000 -2800
x, μm x, μm
the resist dynamic viscosity = 2 ×104 Pa⋅s
500
400
H, nm
300
200
-4800 -4600 -4400 -4200 -4000 -3800 -3600 -3400 -3200 -3000 -2800
x, μm
The best fit of simulation results to the experimental data.
the resist dynamic viscosity = 3 ×104 Pa⋅s the resist dynamic viscosity = 10 5 Pa⋅s
500 500
400 400
H, nm
H, nm
300 300
200 200
-4800 -4600 -4400 -4200 -4000 -3800 -3600 -3400 -3200 -3000 -2800 -4800 -4600 -4400 -4200 -4000 -3800 -3600 -3400 -3200 -3000 -2800
x, μm x, μm
NIL Potentials
Intel microprocessor-Brief history
Intel microprocessor-Fabrication steps
35 nm
35 nm Three dimensional Si stamp for NIL applications
3D nanofabrication techniques
C
Conventional methods:
ti l th d
Electron beam Focused ion beam Two photon polymerization
Non‐conventional methods:
Combination of NIL and X‐ Combination of lithographic Reverse NIL
ray Lithography steps and wet etching
3D nanofabrication techniques
Direct patterning of three dimensional structures by NIL
Transistor Metal T‐gate with 3D‐Hot embossing of Triangular Profile Imprint
90 nm wide foot undercut structures
M. Li et. al N. Bogdanski et. al Z. Yu et. al
3D nanofabrication techniques
Towards three dimensional photonic crystals
Woodpile‐like structure
Determistic defect
Reverse UV NIL technique
Selective Transfer mode
3D woodpile like structures
1 Layer 2 layers
1 μm
10 μm
3 layers
4 μm
