Describing Litho-Constrained Layout by a High-Resolution Model Filter Photomask Japan 2008 April 17, 2008 Min-Chun Tsai

Predictable Success

Outline Problem statement & review of prior-arts. Our proposal – High Resolution Model. first principle approach.

Gauge line pinching sensitivity in HRM. Gauge line-end bridging sensitivity in HRM. Finding the optimal HRM. Results & Usages. Conclusions.

2

Problem statement & review of prior-arts Problem Statement Need to mitigate lithography issues during design stage. Has to be predictable. Has to be fast with minimum foundry process information.

1. Post-OPC approach Use full blown mask-synthesis flow. Very foundry specific, extremely long run time. Need tremendous amount of foundry information. Foundry recipes are not matured till late in technology developments. It needs design data to iterate. Before recipe maturing, majority can be fixed in opc refining.

3

Review of prior-arts 2. Pattern-Match A design-rule complementary approach. It applies only to pre-known difficult patterns. Can not be used to find new hot-spots. Needs detailed analyzing and complete description of the pattern to avoid flagging false violations.

3. Contour-Curve-Fitting Attempt to fit contour using compact modeling. Fit well only in 1D layout, but not in 2D patterns. Hot-spots are caused by 2D pattern interactions.

4

Our Proposal - HRM optical system

P ⊗ ∫ 193nm

OPCed mask

layout / mask

silicon/wafer OOPS !

Shape recovered

P ⊗ ∫

Had a High Resolution Optics, P’

P ' ⊗ ∫ 5

First principle approach Rayleigh criterion:

R = k1 × (λ / NA) Industry experience: Technology node 500,350@365 @wavelength, nm 350@248 K1 required >= 0.65 RET/OPC

No

180@248

130@248

130@193 0.5 – 0.65

90@193 < 0.5

Weak

Aggressive

Current 65nm node, k1 of 0.4 was achieved by aggressive opc. First principle calculation: a 120nm wavelength stepper could bring k1 back to 0.65. 6

Gauge line pinch sensitivity* Baseline: extract the intensity profile from a known lithosensitive post-OPC pattern. Gauge: contrast gradient

*Published in DAC 2007, Tsai, Zhang and Tang 7

Gauge line-end bridging sensitivity Reasons of bridging: low NILS/ high MEEF in tight spacing Gauge for line-end bridging sensitivity: ∆CD/CD

black:193, red:140, orange: 110,

| X n − X n −1|< 0.1nm

green:70 (nm) 8

Finding the optimal HRM Baseline: extract ∆CD/CD on one known “litho-sensitive” and two known “friendly” patterns

9

Wavelength effects On a small testchip – 165 different patterns. Hot-Spot Rank vs. Wavelength

Rank

Range

90

0.070

80 High Rank <---> > Low Rank

0.060

70 0.050

60 50

0.040

40

0.030

30

0.020

20 0.010

10 0

0.000

140

130

120

110

100

Wavelength, nm

90

80

70

need further understandings 10

Results 80mm2 65nm MT1 layout containing 7.95 millions candidate sites HRM: λ = 90nm, sensitivity threshold=0.1 Methods

# hot-spots

%missing

#overshoots

%overshoots %die size

Post-OPC

1582

---

---

---

HRM

1810

0%

228

14.4 %

0.023%

64X speed up with less than 15% overshoots. 11

Usages – guide design fix

Before

After

Confirmed by post-OPC flow. 12

Usages: chip health & opc recipe Gauge chip litho health. OPC recipe optimization.

13

Conclusions A fast high resolution modeling technique was proposed to extract and quantify litho-constrained layout. This technique is a move towards “mask-synthesis flow independent model”. It can provide a robust and predictable way to enhance litho friendly design. This method is in its infancy, and needs future developments to capture all types of litho-constrained patterns.

14