EUVL - A Reality in the Making · o To resume power scaling and demonstrate feasibility of 50W o To...

EUVL - A Reality in the Making The Reality of Laser Assisted Discharge Plasma EUV Light Sources (S49)

2012 International Workshop on EUV and Soft X-ray Sources Dublin, October 2012

©Copyright USHIO and XTREME

XTREME’s LDP* Concepts – A Quick Refresher *Laser assisted Discharge Plasma

Liquid Tin Bath

Capacitor Bank

Tin Film

Cooling

Trigger Laser

EUV in 2π

Tin Supply Disc

- +

Plasma 200,000K


Intermediate Focus (IF)

Plasma Tin Supply

Power Supply

Vacuum

Cooling

Foil Trap

Rotating Wheels

Trigger Laser

Source Head Capacitor Banks

Collector Mirror

Baffles

Far Field (FF)

Far Field (FF) image àPower measurement

XTREME’s LDP* Concepts – A Quick Refresher *Laser assisted Discharge Plasma


o  Far Field images 35cm after IF

o  Power refers to power after IF

o  Power was actually measured after IF aperture o  Not calculated from plasma power

Where Were We Last Year ?

30 W 100% DC

37 W 50% DC


o  STILL … #1 Issue: Source o  Power x Duty Cycle (= average power) à Throughput o  Dose stability à CD uniformity à Yield o  Availability / Uptime

o  #2 Issue: Mask

o  Defect density and contamination mitigation o  Inspection tools and sources for inspection tools

o  #3 Issue: Resist

o  Line edge roughness (LER) o  Sensitivity o  Resolution

The Biggest Headache ?

o  Will EUV light sources ever scale ?

o  Could EUV light sources be turned into a product ?

o  Do we have our priorities right ?

o  Conclusions

Will EUVL Ever Be A Reality ?


o  Last July, XTREME has resumed power scaling experiments on Ushio 3 integrated system to investigate short term scalability

Could Physics Be Integrated Into A Viable Technology ?

Plasma Engineering

Peak EUV emission

Trigger pulses

Laser Engineering

Trigger laser

Tin Flow Engineering

Tin delivery wheel

Thermal Engineering

Thermal shield

Ushio 3 (at XTREME)

Control Engineering

HVPS Engineering

Resonant Charging Device 2B

Constant HV Supply 2Resonant Charging Device 2A

Resonant Charging Device 1BLDPConstant HV Supply 1 Resonant Charging Device 1A

DC

HV-Cable

EUVDetector

Dose Control processor

> 30m

Controller

EUV (n-1)

Reference Energy

drivers

HV (n)

Real time dose control

Resonant charging PPS


1

Measuring Collectable EUV Power 1 - At plasma collectable in-band power with pinch camera calibrated energy monitor 2 - Before IF NFST camera can be moved in the EUV beam 3 - Behind IF FFST (external sensor – XTREME only) 4 – Inside scanner Energy Sensor at reticle level

2 3 4


Since Last Year …


Just In:


LDP Has Made Steady Progresses

37 W 74 W

Last year This year

Integrated source Ushio 3 Burst mode


o  Why would anyone pursue a given technological path if it does not scale in the long term ? à A test stand (Obelix II) has been specially built to allow XTREME to validate LDP long term scalability

Could LDP Scale Beyond 250 W ?

Power scalability: o  Reprate

scalability

o  Pulse energy scalability

o  Conversion efficiency optimization

o  Collection efficiency optimization


First, What It Means To Scale LDP


o  LDP’s reprate long term scalability is proven BEYOND the requirements for 3300B (250W) o  Interlaced low energy pulses

experiments

LDP RepRate Scalability

9.8 µs

EUV signal

Charging voltage

0 10 20 30 40 50 60 70 80 90 1000

500

1000

1500

2000

2500

3000

3500

0

500

1000

1500

2000

2500

3000

3500

EUV

pow

er (W

/2πs

r)

frequency (kHz)

colle

ctab

le E

UV

pow

er (W

/2πs

r)

NXE:3300B equivalent Repate

Beyond NXE:3300B

Experiment Frequency = 1/(ΔT between pulse 1 and 2)


o  LDP’s long term pulse energy scalability is proven BEYOND the requirements for NXE:3300B (250W)

LDP Pulse Energy Scalability

0 5 10 15 200

100

200

300

400

500

Inba

nd E

UV

ene

rgy

per p

ulse

[m

J/2π

]

Electrical energy on capacitor bank [J]

NXE:3300B (250W) equivalent pulse energy

Beyond NXE:3300B

OBELIX II (SoMo)


o  Tailoring the laser pulse train allows XTREME to engineer the plasma emission

Single trigger laser pulse Double trigger laser pulse

Plasma Engineering

Δτ

Peak EUV emission

Peak EUV emission


o  Tailoring the laser pulse trains yield improvement in the emission spectrum (1.3X) and the out of band EUV content

Engineering The Emission Spectrum

Spectral purity was also proven not to change with increasing input electrical energy Can we further tweak the Sn spectrum to get more in band (2%) EUV ?


Engineering Pulse Energy With Lasers

+32%

3100


o  CCE (Collect. Conv. Eff.) = Conversion Efficiency x Collection Efficiency

Engineering Collectable Conversion Efficiency

+78%

+46%

3100




o  Conclusions



o  Ushio 1 source at IMEC initial availability was very disappointing

The Not So Long Ago Early Days Were Painful


o  To stabilize the technology, over H1 2012, XTREME reallocated resources, launched a major re-design effort and upgraded Ushio 1

Turning Technology Into A Product

Pictures not for hand out

Tin heating

Tin delivery

Tin flow

Tin collection

Optical transmittance

Thermal shielding

…


After Upgrade, Ushio 1 Uptime Has Steadily Increased …


Upg

rade

… And Utilization (7x24) Is High

3,000 wafers have been printed so far

Courtesy


o  Power at IF is stable over the collector life

Long Collector Lifetime Is Achieved

Collector: ∼3.5 months (∼1,500 hours production time)


o  Source Heads (SH) are no more the primary source of downtime

Lifetimes Have Increased

SH v 2.2 SH v 2.2.1 SH v 2.3


o  Ushio 2 & Ushio 4 light sources (3100) are integrated to NXE:3300B to support scanner development

o  U2 (20 kW configuration) & U4 (50 kW configuration) are being upgraded as well

o  U2 has now printed its first wafer

U2 & U4 @ ASML

Courtesy 11-Jul-’12 | public | slide 20

>70,000 wafers cycled through scanner body>4,200 reticle exchanges performed

Current MTBI ~8 hrs 13wks average, last week 14hrs MTBI70000

60000

50000

40000

30000

20000

10000

1146 1150 1202 1206 1210 1214 1218 1222

0

Num

ber

of

exp

ose

d w

afe

rs

Extensive reliability testing ongoing on multiple NXE:3300B systems.




o  Conclusions



Can we really talk about power without talking about

dose stability ?!?!?

POWER à THROUGHPUT &

DOSE STABILITY à YIELD

Power, Power, Power !!! … Simply More Power ?


o  Small ΔDose à Small ΔCD

o  Large ΔDose à

Large ΔCD

ASML dose specifications for EUV sources: 3 σ < 0.2%

Why Does Dose Stability Matter ?

BF


o  Dose stability specification: 3 σ < 0.2%

LDP Dose Stability at 20kW


How To Ensure Tight Dose Stability ?

LDP Technology For HVM


o  Tin delivery is continuous

à No temporal discretization

à No missing target

à No missed pulse

LDP Stability = Continuous Tin Delivery

Trigger laser

Tin continuous supply disc


o  Laser focus, Tin and plasma are always at the surface of the wheel

o Plasma position remains stable with power scaling

LDP Stability = Plasma Position Stability

à Stable laser focus à Stable plasma

position

à Stable Far Field image

à No dose variation caused by plasma position instability

Plasma position stability @ 3 kW operation

Plasma position stability @ 20 kW operation

x,y

z

x’,y’ IF

Plasma

Coll.


o  Trigger laser stability makes Tin vaporization and plasma production highly reproducible

à Trigger laser intensity stability

à Tin vaporization repeatability

LDP Stability = Trigger Laser Energy Stability

20 kHz operation


o  The jitter between trigger laser pulses is kept << 1 ns

LDP Stability = Trigger Laser Timing Stability

à No timing instability

à Tin vaporization

is highly reproducible

20 kHz operation


o  The amount of Tin vaporized (with each trigger pulse) is highly repeatable

LDP Stability = Tin Flow Uniformity

Not OK OK




o  Conclusions



o  To drastically improve and stabilize the reliability of XT’s 3100 source at IMEC to enable Affiliate Chipmakers to develop their EUV process

o  To prove LDP long term

scalability

XTREME’s 2012 Objectives

∼ IMEC Today ∼ NXE:3300 B >> NXE:3300 B

þ Done

þ Done

0 5 10 15 200

100

200

300

400

500

Inba

nd E

UV

ene

rgy

per p

ulse

[m

J/2π

]

Electrical energy on capacitor bank [J]

Upg

rade


o  To resume power scaling and demonstrate feasibility of 50W

o  To upgrade XT’s 3100 source at

IMEC for higher power

XTREME’s 2012 Objectives

o  Objective 1

o  Objective 2

o  Objective 3

o  Objective 4

☐ Soon

þ Done 74 W

o  EUV is a reality in the making supported by recent progresses of LDP

o  No more claims. Results are in: o  LDP is scalable in the long term o  74W power after IF was demonstrated on an

integrated source o  LDP technology is now being turned into a

viable product and high uptime is achieved

o  Next is to demonstrate 50 W @ 100% DC

o  The night is always darker before dawn … but the EUV revolution is around the corner

Conclusions


o  XTREME would like to acknowledge this work has been possible thanks to a very valuable and fruitful collaboration with Fraunhofer ILT

o  XTREME would also like to thank NEDO for their continued support

Acknowledgments

THANK YOU VERY MUCH FOR YOUR ATTENTION


XTREME technologies GmbH www.xtremetec.de


EUVL - A Reality in the Making · o To resume power scaling and demonstrate feasibility of 50W o To...

Documents

Transcript of EUVL - A Reality in the Making · o To resume power scaling and demonstrate feasibility of 50W o To...