use of dsa to rectify lcdu of euv exposed contact...

© IMEC 2013

USE OF DSA TO RECTIFY LCDU OF EUV EXPOSED CONTACT HOLES

ROEL GRONHEID, ARJUN SINGH

TODD R. YOUNKIN

© IMEC 2013

OUTLINE

DSA

▸ Grapho-epitaxy: EUV NTD resist + homo-polymer blend

- Limitations of this process: ~20nm

- Challenges to implement BCP into similar flow

▸ Chemo-epitaxy:

- Honeycomb process flow + results

- Data on CDU improvement and critical contributors to CDU in current

process

2 ROEL GRONHEID - SPIE 2013 - SAN JOSE

© IMEC 2013

USE OF (BLOCK CO) POLYMERS TO

REPAIR CDU

K. Maruyama et al. – EUVL Symposium 2012


Poly(A-block-B)

Cylinder (hole) CD is determined by polymer chain length

© IMEC 2013

DSA SHRINK BEHAVIOR THROUGH DOSE

ROEL GRONHEID - SPIE 2013 - SAN JOSE 4

DSA blended agent requires closed pre-pattern and saturates ~ 25-30 nm.

May be used as CDU enhancement?

48P80 40hp

© IMEC 2013

Resist NTD PTD

Shrink None Shrink A, Std

FT Shrink A, FT+ Shrink A, FT++ None

Image

Esize

(mJ/cm2) 23.5 20.8 (11 %) 20.4 (13 %) 17.0 (27 %) 17.0

CD (nm) 32.5 30.2 29.2 27.3 30.4

CDU (nm) 1.4* 1.3* 1.3 1.0 --

* Missing holes observed; Believed to arise from NTD pre-pattern

Optimization = Shrink FT > Resist FT >> Resist Anneal > Shrink Anneal

Process optimization yielded ~10-25% Esize Gain at 30 hp vs. NTD

Process does not scale well below ~25nm CD; material modification may

be required for these dimensions

38P60 30hp

Key Parameter = DSA Shrink Agent FT


© IMEC 2013

GRAPHO-EPITAXY FLOW USING HOMO-

POLYMER BLENDS

K. Maruyama et al. – EUVL Symposium 2012


use block copolymer instead of homo-polymer

better CDU improvement?

© IMEC 2013

GRAPHO-EPITAXY WITH BCP


193nm pre-pattern

Pst-Litho Pst-HardBake Pst-DSA

+ Wet Dev

SEM image

CD (nm) 53.8 58.9 14.5

Change (nm, %) -- +5.1, +9.5% -44.4, -75.4%

Net Shrink (nm, %) -- -- -39.3, -73.0%

CD Range (nm) 3.4 3.3 0.5

1 2 3

© IMEC 2013

PROPOSED CHEMO-EPITAXY FLOW

anneal

Etch X-PS

and strip

resist

Coat

BCP

Coat X-PS

+EUV resist

Pattern

hexagonal hole

array by EUV


PMMA

removal

© IMEC 2013

HEXAGONAL CONTACT ARRAYS DENSE

Center-to-center distance same in

all directions

9

p

p

Honeycomb reticle is available and prints target structures

p=54nm; exposed on NXE:3100

Mask patterns

ROEL GRONHEID - SPIE 2013 - SAN JOSE

© IMEC 2013

PROCESS OPTIMIZATION

Large L0 BCP materials are not easy to work with

▸ Material preparation/tuning proves more challenging than for

smaller L0

▸ Larger L0 BCP have smaller process windows for neutral layer,

film thickness, pre-pattern pitch, ...

▸ For cylindrical PS-b-PMMA operational L0 window ~30-60nm

- Little overlap with

processing window

for EUV patterned

contact holes

10

p54nm


© IMEC 2013

EUV CDU REPAIR WITH HONEYCOMB 27NM HALF PITCH

11

Post-litho Post-BCP anneal Post-PMMA

removal

1s CDU: 2.1-2.5nm 1s CDU: 1.6-1.7nm

Residual CDU to be improved by

process and material optimization


© IMEC 2013

HEXAGONAL ARRAY THROUGH PITCH


Pitch 50 Pitch 52 Pitch 54 Pitch 56

Pre-pattern

BCP


Pre-pattern

BCP

• For current BCP material

we see best performance

for 54 – 58 nm pitch

• BCP performance is

mostly independent of pre-

pattern hole CD

• It’s possible that it works

for tighter pitches but pre-

pattern not good enough

(yet)

• Smaller L0 BCP materials

generally work better:

process benefits expected

to increase on NXE:3300

12

© IMEC 2013

SQUARE ARRAY THROUGH PITCH



Resist only

BCP

• Annealed at 275 ℃ for 5 minutes

• Best performance for 54 – 56 nm pitch

• DSA process works, but pattern quality is clearly impacted by

forced snapping to square gridded layout

13

© IMEC 2013

HEXAGONAL CONTACT ARRAYS MISSING HOLES

Only purple holes are printed to

mimic missing patterns

p

p

14

p=54nm

Mask patterns


© IMEC 2013

HEXAGONAL ARRAYS WITH MISSING

HOLES


Pitch 54 Pitch 56 Pitch 58

• Repair works, but success rate is not yet sufficient

• Expected to improve with process optimization and material tuning

15

© IMEC 2013

SUMMARY

300mm wafer processes for DSA-based hole patterning for EUV

LCDU rectification have been set up at imec

Grapho-epitaxy based on homo-polymer blends works, but only

to keep relative CDU performance in combination with shrink

Grapho-epitaxy based on BCP requires very large initial hole CD

and may not be attractive path for dense hole patterns

Chemo-epitaxy DSA flow with BCP looks promising, but does

not yet meet requirements

More work is needed to understand how BCP material and

process impacts CD uniformity


use of dsa to rectify lcdu of euv exposed contact...

Documents