Applications of high efficiency

FELs for EUV lithography

Patrick Naulleau
Center for X-ray Optics
Lawrence Berkeley National Laboratory
Physics & Applications of High Efficiency Free-Electron Lasers Workshop, Apr 11-13, 2018, UCLA

Berkeley Lab | MSD Materials Sciences Division 1

 Outline

• Semiconductor industry trends

• Introduction to EUV Lithography
• Future power needs
• Key FEL source requirements

Berkeley Lab | MSD Materials Sciences Division 2

 Semiconductor industry is huge
economic driver
$59 Billion
Semiconductor
R&D (2017) $412 Billion
Semiconductor
device market
(2017) $2.0 Trillion
Global electronics
market (2017)
SIA, www.persistencemarketresearch.com

Berkeley Lab | MSD Materials Sciences Division 3

 “By 2020, [expected] cost of between $15
and $20 billion for a leading-edge fab”
EE Times

Berkeley Lab | MSD Materials Sciences Division 4

 Industry growth enabled by Moore’s Law:
transistors double every two years
1.E+11
Number of Transistors

Intel 62 core Xeon

1.E+09 Intel Itanium2

Intel Pentium 4
1.E+07
Intel Pentium
1.E+05 Intel 386
Intel 8086

1.E+03 Intel 8008

1970 1980 1990 2000 2010 2020

Berkeley Lab | MSD Materials Sciences Division 5

 The other half of Moore’s Law:
density increase at shrinking cost
Normalized cost

1.00
per transistor

0.10

0.01
2000 2004 2008 2012 2016
Data from Intel (http://www.pcworld.com/article/2887275/intel-moores-law-will-continue-through-7nm-chips.html)

Berkeley Lab | MSD Materials Sciences Division 6

 128 GB microSD
$59.99

Would have cost

$256 billion in 1970

Berkeley Lab | MSD Materials Sciences Division 7

 Lithography drives shrink
Mask
Illumination

Demagnifying Wafer
lens

Res  /NA
Berkeley Lab | MSD Materials Sciences Division 8
 Moore’s Law driven by wavelength shrink
1.E+11
Number of Transistors

White

438 nm

365 nm
1.E+09

1.E+07

13.5 nm
193i nm
193 nm
248nm
1.E+05

1.E+03
1970 1980 1990 2000 2010 2020

Berkeley Lab | MSD Materials Sciences Division 9

How does EUV lithography work?

Berkeley Lab | MSD Materials Sciences Division 10

EUVL: optical lithography at  = 13.5 nm
Reticle

Illuminator Projection
Intermediate optics
focus

Source
Plasma

Collector
Wafer
Berkeley Lab | MSD Materials Sciences Division 11
Near term source power requirements
Wafer Throughput wafer/h 145
Total wafer time sec 24.8
Stage motion overhead sec 18
Wafer exposure time sec 6.8
Wafer diameter mm 300
Wafer fill factor % 89%
Required
Resist Sensitivity
Required Power at Wafer
mJ/cm^2
W
15
1.38
source power
POB reflectivity (0.66^6) % 8.27% = 250W
Mask reflectivity % 62%
Illuminator reflectivity (0.66^4) % 18.97%
Overfill efficiency % 75%
Pellicle efficiency % 76%
Total Optical Efficiency % 0.55%

Required collected source power W 250

Berkeley Lab | MSD Materials Sciences Division 12

Berkeley Lab | MSD Materials Sciences Division 13
LPP source status

ASML, SPIE Advanced Lithography Conference 2018

Berkeley Lab | MSD Materials Sciences Division 14

 Future source power needs?

Berkeley Lab | MSD Materials Sciences Division 15

Key variable assumptions in 250W
number

• Throughput = 145 Wafers/hr

• Wafer size = 300 mm
• Dose = 15 mJ/cm^2

Berkeley Lab | MSD Materials Sciences Division 16

Source power requirements expected
to grow significantly in the future
Double patterning
• Throughput
= 220 wafers/hr
• Stage overhead
= 12 seconds

1.6x more power

Berkeley Lab | MSD Materials Sciences Division 17
Source power requirements expected
to grow significantly in the future
450-mm wafers
• Throughput
= 105 wafers/hr
• Stage overhead*
= 12 seconds

1.5x more power

* Normalized to 300-mm wafer
Berkeley Lab | MSD Materials Sciences Division 18
Source power requirements expected
to grow significantly in the future

Dose
• 15 mJ/cm2
likely not
enough in
the future

Berkeley Lab | MSD Materials Sciences Division 19

Source power requirements expected
to grow significantly in the future

Dose
• 15 mJ/cm2
likely not
enough in
the future

𝐿𝑊𝑅 ∝ 1 𝑑𝑜𝑠𝑒
2-4x more power
Berkeley Lab | MSD Materials Sciences Division 20
 Depending on resist performance,
future power needs could range
from 500W to 2000W

Berkeley Lab | MSD Materials Sciences Division 21

 Other requirements

Berkeley Lab | MSD Materials Sciences Division 22

Size: about the size of a shipping container

• Allowable
source footprint:
~ 10m x 3m

Berkeley Lab | MSD Materials Sciences Division 23

Reliability: Require > 95% uptime

Even better uptime required

if driving multiple tools
Berkeley Lab | MSD Materials Sciences Division 24
Source power stability

Need <1% with 1 ms

integration window
• Implies rep rate
>1kHz x FPN^2
• 30% pulse noise
=> rep rate > 900kHz

FPN = fractional pulse noise

Berkeley Lab | MSD Materials Sciences Division 25
Pulse length

• Multilayer BW limits
require pulse > 2.5 fs
• Longer is better to avoid
optics damage issues

Berkeley Lab | MSD Materials Sciences Division 26

Bandwidth
Multilayer to the 11th power
• Multilayer mirrors require
bandwidth
< +/-0.14nm (1%)
• Narrower bandwidth =
greater effective optical
throughput
• +/-0.02nm BW would
provide 28% effective
power boost

Berkeley Lab | MSD Materials Sciences Division 27

Polarization
Horizontal
Vertical
• Linear polarized light
causes significant
imaging anisotropy
• If it can be
manipulated,
polarization can be
viewed as asset

Berkeley Lab | MSD Materials Sciences Division 28

Coherence

• No longitudinal
coherence needed
• No lateral coherence
needed (coherence
must be destroyed)

Berkeley Lab | MSD Materials Sciences Division 29

 Coherent Partial coherence
Defocus

Berkeley Lab | MSD Materials Sciences Division Data courtesy of K. Goldberg, A. Wojdyla, LBNL
 Summary

• EUV is on its way

• We need creative solutions
to carry the technology well
into the future

Berkeley Lab | MSD Materials Sciences Division 31

Berkeley Lab | MSD Materials Sciences Division