Down to 20nm width photoresist patterns fabricated by using a dry - - PowerPoint PPT Presentation

▶
down to 20nm width photoresist patterns fabricated by
SMART_READER_LITE
LIVE PREVIEW

Down to 20nm width photoresist patterns fabricated by using a dry - - PowerPoint PPT Presentation

Mar 17, 2023 39 likes •241 views

Down to 20nm width photoresist patterns fabricated by using a dry plasma trimming A. DE LUCA 1 . E. Dien 2 . P. France 2 and M. Heitzmann 1 CEA-LETI 1 and ST-Microelectronics 2 . Grenoble. France . 2005 AGENDA Introduction Different

slide-1
SLIDE 1

Down to 20nm width photoresist patterns fabricated by using a dry plasma trimming

  • A. DE LUCA1. E. Dien2. P. France2 and M. Heitzmann1

CEA-LETI1 and ST-Microelectronics2. Grenoble. France .

slide-2
SLIDE 2
  • A. DE LUCA – ICNT 2005 -San Francisco

2

2005 AGENDA

  • Introduction

– Different approaches to reduce the gate width – State of the art : resist trimming process

  • Description of the experiment
  • Results and discussion

– Preliminary results – CL2-O2 chemistry results – Why a hardening step is needed ? – The Bi-trimming process

  • Conclusion
slide-3
SLIDE 3
  • A. DE LUCA – ICNT 2005 -San Francisco

3

2005

Different approaches to reduce the gate width

Trimming resist  Hard mask trimming without resist  Hard mask trimming under the resist  Polysilicon trimming  FOX process

Silicon active layer Polysilicon gate Hard mask Resist Gate dielectric

INTRODUCTION

Best dimensions control with dry etching Gate Morphology is important

slide-4
SLIDE 4
  • A. DE LUCA – ICNT 2005 -San Francisco

4

2005

State of the art : which chemistry should de used ?

INTRODUCTION

 O2 is used to burn the photoresist  Another gas is necessary to brake the resist burning kinetic (Cl2. HBr. CF4. Ar. others ?)

10 20 30 40 50 60 70

  • 80
  • 60
  • 40
  • 20
  • HCl. dense
  • HCl. iso
  • HBr. dense
  • HBr. iso

CD Bias (nm)

O2 % in Gas Feed

  • Cl2. dense
  • Cl2. isolated

Erwine Pargon thesis (LTM / CNRS) V(HBr/O2) > V(Cl2/O2) µ(HBr/O2) > 0 > µ(Cl2/O2)

10 20 30 40 50 60 70

  • 10
  • 5

5 10 15 Cl2 HCl HBr

O % in Gas Feed

Dense-Iso CD Bias difference (nm)

Trim rate (isolated) > Trim rate (dense) Trim rate (isolated) < Trim rate (dense)

2

CD Bias as = CD after trimming – CD before trimming

slide-5
SLIDE 5
  • A. DE LUCA – ICNT 2005 -San Francisco

5

2005 DESCRIPTION OF THE EXPERIMENT

CD Si-Bulk HTO ~ 100 nm Ep

Sumitomo NEB22 E-Beam resist Thickness before trimming : ~160-200 nm Gate lithography level 12 E-beam chips checked CD before and after trimming :

  • 40 nm. 50 nm. 65 nm et 75 nm for isolated patterns
  • 50 nm for dense lines

CD Bias = CD before trimming – CD after trimming Microloading = CD Bias isolated line/ CD Bias dense lines

slide-6
SLIDE 6
  • A. DE LUCA – ICNT 2005 -San Francisco

6

2005

  • NEB22 resist status :

Hard mask etching (60 nm) : EpNEB22 must be up to 60 nm Active layer etching : EpNEB22 must be up to 120 nm  Ep NEB22 trimmed ≤ 80 nm => Vcr ≤ 1.3 nm.s-1 (for trimming time of 60 s)

RESULTS AND DISCUSSION

Vcr (nm. S-1)

0.5 1 1.5 2 2.5 3 10 20 30 40 50 Cl2-O2 (40%) Cl2-O2 (30%)

Wb (Watt)

CF4-O2 (30%) HBr-O2 (30%)

  • Equipment status :

Wb=0 or Wb ≥ 50 Watt Wb = O Watt

  • P = 10 mTorr
  • Ws = 300 Watt
  • Trimming time : 60 s

Preliminary results

slide-7
SLIDE 7
  • A. DE LUCA – ICNT 2005 -San Francisco

7

2005

  • Different gases for O2 dilution : Cl2 or CF4 or HBr

1.5 1.2 0.9 20 HBr-O2 1.25 0.8 0.6 30 CF4-O2 1.1 0.25 0.08 120 Cl2-O2 Microloading Trim etch rate (nm.s-1) Vertical etch rate (nm.s-1) Process time (s) Chemistry 20 40 60 80 100 120 40

50

75 Cl2-O2 CF4-O2 HBr-O2

% of valid patterns Initial pattern size (nm)

  • P = 10 mTorr
  • Ws = 300 Watt
  • Q = 200 sccm
  • CDafter trim : 44 nm -> 18-22 nm

RESULTS AND DISCUSSION

Preliminary results

slide-8
SLIDE 8
  • A. DE LUCA – ICNT 2005 -San Francisco

8

2005

  • Wb = 0 Watt :
  • Resist consumption is lower : << 1.3 nm. S-1
  • The best chemistry is Cl2-O2 :
  • Resist consumption is the lowest : only 0.2 nm. S-1
  • Best throughput after trimming

RESULTS AND DISCUSSION

Preliminary results

"The most slowly you trim. the highest throughput you have"

slide-9
SLIDE 9
  • A. DE LUCA – ICNT 2005 -San Francisco

9

2005

Fixed parameters :

  • P = 10 mTorr
  • Ws = 300 Watt
  • Wb = 0 Watt
  • Q = 200 sccm
  • t trim = 60 s

0.85 0.15 0.8 30 1.1 0.3 0.1 40 1.3 0.35 0.185 50 Microloading Trim etch rate (nm.s-1) Vertical etch rate (nm.s-1) % of O2

RESULTS AND DISCUSSION

CL2-O2 chemistry results : O2 concentration impact

 30% d’O2 < "microloading = 1"< 40 % d’O2

slide-10
SLIDE 10
  • A. DE LUCA – ICNT 2005 -San Francisco

10

2005

20 40 60 80 100 120 40 50 75

50% of O2 40% of O2 30% of O2 % of valid patterns Initial pattern size (nm)

 % of valid patterns decrease with O2 "The most slowly you trim. the highest throughput you have"  40% of O2 seems to be the best compromise between the yield and the microloading

RESULTS AND DISCUSSION

CL2-O2 chemistry results : O2 concentration impact

slide-11
SLIDE 11
  • A. DE LUCA – ICNT 2005 -San Francisco

11

2005

Pattern before trimming After 60 s trimming After 90 s trimming After 120 s trimming

RESULTS AND DISCUSSION

CL2-O2 chemistry results : trimming time impact

slide-12
SLIDE 12
  • A. DE LUCA – ICNT 2005 -San Francisco

12

2005

10 20 30 40 50 60 70 80 20 40 60 80 100 120 CD (nm) Trimming time (s)

20 40 60 80 100 20 40 60 80 100 120 140

% of valid patterns Trimming time (s)

Limit width after trimming for a yield of 100% : 22-23 nm

 40-45 nm : 60 s process  50-55 nm : 90 s process

18 nm !

RESULTS AND DISCUSSION

CL2-O2 (40%) chemistry results : trimming time impact

slide-13
SLIDE 13
  • A. DE LUCA – ICNT 2005 -San Francisco

13

2005

Gate width reduction :

  • Goal : reduce isolated line width from 40 nm to 20 nm
  • 12 E-Beam chips characterized on 20 processed wafers
  • Trimming without hardening step during 60 s
  • CD values before /after trimming :
  • 43nm (40) / 24 nm
  • 48 nm (50) / 32 nm
  • 72 nm (75) / 56 nm

RESULTS AND DISCUSSION

CL2-O2 (40%) chemistry statistical results

Active zone reduction :

  • Goal : reduce isolated and dense line widths from 50 nm to 30 nm
  • 5 E-Beam chips characterized on 19 processed wafers
  • Trimming without hardening step during 90 s
  • CD values before /after trimming :
  • 45nm (iso-50) / 24 nm
  • 48 nm (résH-50) / 29 nm
  • 48 nm (résV-50) / 29 nm
slide-14
SLIDE 14
  • A. DE LUCA – ICNT 2005 -San Francisco

14

2005

To reach CD lower than 20 nm with a high output Why patterns fall down ?

  • The trimming process
  • The chamber pumping and the wafer dechuck after process
  • Wafer moving
  • Scanning Electron Microscopy observations
  • Hard mask and active zone etching
  • Sample cleavage for SEM observations

Which solution can be used ?

  • A resist hardening as post-trimming step

RESULTS AND DISCUSSION

Why a hardening step is needed ?

slide-15
SLIDE 15
  • A. DE LUCA – ICNT 2005 -San Francisco

15

2005

  • Description of the hardening step
  • Ws = 1500 Watt
  • Wb = 0 Watt
  • P = 5 mTorr
  • Q = 150 sccm
  • Process time = 60 s
  • Results
  • 46 nm : 60 s  90 s process
  • 56 nm : 90 s  120 s process
  • Drawbacks
  • Impossible to increase the trimming time without decreasing the yield
  • Limit after trimming and hardening always at 22-23 nm
  • A second trimming step is needed

RESULTS AND DISCUSSION

Why a hardening step is needed ?

slide-16
SLIDE 16
  • A. DE LUCA – ICNT 2005 -San Francisco

16

2005

40 40 % of O2 5 10 5 10 Pressure (mTorr) HBr Cl2-O2 HBr Cl2-O2 Chemistry 1500 300 1500 300 Ws (W) Wb (W) 60 40 60 90 Time (s) 2nd hardening 2nd trimming 1irst hardening 1irst trimming Parameters

RESULTS AND DISCUSSION

The Bi-trimming process

slide-17
SLIDE 17
  • A. DE LUCA – ICNT 2005 -San Francisco

17

2005

  • SEM CD : 6 patterns on 12 E-Beam chips / 5 wafers

51.1 30.8 20.4 78.1 51.7 45.8 Mean 50.2 30.1 20 76.5 56.5 44.6 H585-P10 53.7 32.8 22.2 79.2 58.8 46.6 H585-P09 51.9 32 21.8 77.2 57.6 46.2 H585-P08 48.6 28.8 18.5 78.1 58.7 45.7 H585-P07 51.1 30.2 19.3 79.4 58.7 46 H585-P06 75 nm 50 nm 40 nm 75 nm 50 nm 40 nm Pattern size CD after trimming (nm) CD before trimming (nm)

 CD ~ 26.5 nm with CDtrim2 ~ 5-6 nm

RESULTS AND DISCUSSION

The Bi-trimming process

slide-18
SLIDE 18
  • A. DE LUCA – ICNT 2005 -San Francisco

18

2005 RESULTS AND DISCUSSION

The Bi-trimming process

slide-19
SLIDE 19
  • A. DE LUCA – ICNT 2005 -San Francisco

19

2005

  • Wb = 0 Watt and Cl2-O2 chemistry
  • Resist consumption is lowest ~ 0.2 nm. S-1 << 1.3 nm. S-1
  • Best throughput after trimming and microloading between isolated and

dense lines near 1

  • Cl2-O2 process
  • 40% of O2 seems to be the best compromise between a high yield and a

microloading near 1

  • It’s possible to have microlading close to 1
  • Limitations : it’s difficult to trim under 22-23 nm without resist hardening
  • A process with two trimming steps is necessary to reach resist width lower

than 20 nm

  • The bi-trimming Cl2-O2
  • Only possibility of going down below 20 nm with good outputs
  • Microloading to be characterized
  • Good resist behavior during etching of a hard mask of 100 nm
  • 1

CONCLUSION

slide-20
SLIDE 20

Thank you for your attention