Download pdf - Low damage integration of ultralow-k porous organosilicate ...pesm2014.insight-outside.fr/presentations/Sesssion6-2-PoreStuffing_PESM2014_Liping...Low damage integration of ultralow-k

Low damage integration of ultralow -k porous organosilicate glasses by Pore-Stuffing approach

L. Zhanga,b, J.-F. de Marneffea, M. Heynea, F. Vajdaa, V. D. Rutigliania, L. Wena, Jurgen Bommela, Z. Tokeia, S. de Gendta,b and M. R. Baklanova

a IMEC vzw, 3001 Leuven, Belgiumb Katholieke Universiteit Leuven, 3001 Leuven, Belgium

PESM2014, Grenoble (France)

2© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

Outline

i. Cu/low-k interconnect and plasma induced damage

ii. Pore stuffing - Process flow

iii. Plasma Induced Damage

iv. Integration flow with Pore Stuffing

v. Summary


Cu/low -k interconnect

- Interconnect RC delay will dominate the total time delay as IC scales down.

- Cu/low-k interconnect was introduced to replace Al/ SiO2.

+⋅⋅⋅⋅≈22

20

112

TWLRC εκρ Al → Cu

3.0 → 1.72×10-8 ΩmSiO2 → Low-k4.2 → 3.7-1.8

ITRS, 2011


Porous OSGs

• High porosity• Large pore size

• Low plasma resistance• Low mechanical strength

• Solutions are not known for sub2.2 low-k dielectrics.

PECVDSpin-coating


Plasma Induced Damage (PID)

Radical

n H2OOO Si—CH3OO Si—CH3OO Si—CH3OO Si—CH3O

Moisture uptake leads to increasing of k-value & leaka ge current

OSG before etch

Por

e

OO Si—OHOO Si—OHOO Si—OHOO Si—OHO

3 H2O

b

UV, VUV

OSG post etch

IonRadicalPhoton

plasma

OSG

substrate

PID sidewall

HMHM

PID bottom

- Plasma process is widely used in BEOL, for example: Deposition, Cleaning and Patterning.

- Depth of damage increases with porosity and pore size: L~ a*d*N 0.5 (Random Walk Theory).

N = <number> of collisions before recombination; d = pore diameter; a = distance for a jump; L = depth of penetration.

- Reducing PID is a crucial challenge for porous OSG integration.

Material K-value

SiO2 3.8

Low-k 1.8-2.4

Water 79


Advanced low -k integration

Minimization of Plasma Induced

Damage

Optimization of Low-k materials

Optimization of Plasma Etching

Post processDielectric Recovery

- Engineering of the chemical composition and porous structure.

- Optimization of plasma species or integration proc ess, Post-plasma porogenremoval.

- Restoring the methyl groups by CH 4 plasma or silylation agents, eliminating -OH bonds and physisorbed water by UV irradiation.

→ These approaches are limited and/or complex. New ap proach is needed.


Outline





v. Summary


Damascene integration with Pore -Stuffing

CuCu

Porous low-k

• Conventional Cu/low-k integration flow.

• Damage caused by plasma etching + TaNTa barrier pene tration degrades k eff.

• Cu/Low-k integration flow with polymer pore stuffin g (P4 approach, G. Dubois et al.)

• Polymers suppress the penetration of reactive radic als → lower C depletion

• Supressed TaNTa penetration

• Material: PECVD p-OSG, k value @ 100kHz is 2.0, open porosity is 46%


Polymer Stuffing

- De-wetting issue due to ultra-low surface energy of low-k. A surface activation process by CO2 plasma is used.

- Polymer penetration is driven by capillary force, affected by molecular size, pore size and annealing conditions.

10© IMEC 2013/ CONFIDENTIAL

(a) TOF-SIMS carbon depth profile shows polymers penetrate into bulk low-k until bottom. Surface depletion layer is observed for over removal process.

(b) Ellipsometer Porosity with 0% open porosity, confirming the stuffed state.

LIPING ZHANG

Surface polymer removal

• Solvent need to be adapted to polymer (PGMEA for PMMA, Water for PEG)

• Preferably high surface tension that dos not wet low-k → polymer inside low-k will not be removed.

• Solvent spin cleaning process is optimized to avoid surface polymer depletion.

TOF-SIMSCarbon depth

profile

EllipsometryPorosimetry


Polymer unstuffingPolymer removal by thermal annealing

• Polymer degradation process → no low-k damage

• Limited by BEOL process thermal budget (< 420°C)

Polymer removal by downstream plasma (DSP)

• Low temperature process (250°C), High polymer removal efficiency. DSP causes low-k damage: formation of Si-OH & k-value increase.


Outline





v. Summary


Plasma Induced Damage

1300 1280 1260 1240 1220

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

N

orm

aliz

ed a

bsor

ptio

n (a

.u.)

Wavenumber cm-1

ArCF4_30s with PMMA ArCF4_30s without PMMA ArNF3_20s with PMMA ArNF3_20s without PMMA ArSF6_20s with PMMA ArSF6_20s without PMMA Pristine p-OSG

Si-Me

67,4

79,3

54,5

94,1

21,2

25,3

56,2

26,1

0

50

100

150

200

250

300

Thi

ckne

ss a

fter

etch

(nm

)

Plasma chemistry and etch time

Non-damaged layer Damaged layer

7,2

11,1

2,4

9,8

-0,4

11,2

15,8

18,1

0

50

100

150

200

250

300


Non-damaged layer Damaged layer

• FTIR enable tracing water uptake and Si-CH3 loss (PID).

• Based on loss of Si-Me and thickness, Equivalent Damaged Layer (EDL) is calculated.


Fluorocarbon gas discharges

1300 1250 1200 1150 1100 1050 1000 950

0.0

0.2

0.4

0.6

0.8

1.0

N

orm

aliz

ed a

bsor

ptio

n (a

.u.)

Wavenumber (cm-1)

PMMA protected Non protected Pristine p-OSG

Plasma ChemistryAr/CF 4/CH2F2

• Fluorocarbon plasma is normally used for low-k etching. Due to protection of surface polymerizing effect, low damage is also obtained with blanket test.

• However, this does not apply to patterned test, where high sidewall damage is observed.

• With Pore stuffing, sidewall damage is greatly improved even with O2 containing plasma.

• Trench etch condition is quite different from blanket film etching.


Protection Mechanism

0 50 100 150 200 250 300 350 4001000

10000

0 50 100 150 200 250 300 350 4001000

10000

protected without burn-out protected with burn-out unprotected without burn-out pristine

Inte

nsity

(co

unts

/s)

155 nm

Carbon depth profile

Inte

nsity

(co

unts

/s)

Sputter Time (s)

Fluorine depth profile91 nm

• Surface roughness is improved, indicating no internal etching (ten times lower than without protection).

• TOF-SIMS shows reduced F penetration after etch for pore stuffing, compared to porous low-k.


VUV induced damage

• Polymers with various VUV absorption were stuffed to check their VUV protection against Xe light.

• Slightly different Si-CH3 loss is observed with stuffing materials.

• 147nm Xe VUV light (CCP) was used for VUV damaging test, using a MgF2 glass window.

• FTIR results show loss of Si-CH3, increasing of Si-H and water uptake.

O3-SiCH3 + eV → O3-Si· + CH2(H)

O3-SiCH3 + eV → O3-SiH


Outline


ii. Pore stuffing- Process flow



v. Summary


300mm integration flow using PMMA

Cu40,1

15

2,6

38,2

7,5

3,5

0

5

10

15

20

25

30

35

40

45

SiC SiO2 SiN

CHx loss

C=O loss

Hard Mask type

PM

MA

loss

(%

)

Hard mask induced polymer loss

lowk + PMMA

Substrate

SOC

SOGBARC

PR

SiN Cap layer

Hard Mask stackSiC

370°CSiO2 60°C Si3N4

180°C

Temperature should be < 200°C and no oxidizing gas

• Narrow spacing hanging trench structure is formed in order to evaluate inter-Cu line effective k-value.

• Polymer is removed after metallization and CMP.


Integrated k -valueNo stuffing

SylilationPMMA stuffing

Post-CMP unstuffing

K values : 2.89 ± 0.24 2.64 +/- 0.20• TEM pictures are taken to simulate integrated k value.• Pore stuffing enable lower integrated k value.• In the case of post CMP unstuffing, the sharp interface between barrier and

low-k reveals no penetration of barrier.• Residual damage is tentatively attributed to PMMA removal process


Outline


ii. Pore stuffing- Process flow



v. Summary


Summary

• Following by surface active, spin-coating, thermal drive-in and surface cleaning, > 90% pore-stuffing of a 2.0 porous OSG was achieved.

• Reduced plasma induced damage is observed, based on various plasma tested on both blanket and patterned samples. VUV induced damage is also improved.

• Single damascene Cu/Low-k integration flow on 300mm substrate has been developed incorporating pore stuffing approach.


Plasma induced damage

3800 3600 3400 3200 3000

0.00

0.02

0.04

0.06

0.08

0.10

Nor

mal

ized

abs

orpt

ion

(a.u

.)

Wavenumber cm-1


(a)

water and -OHCHx

1300 1200 1100 1000

0.0

0.2

0.4

0.6

0.8

1.0

Si-Me

Nor

mal

ized

abs

orpt

ion

(a.u

.)

Wavenumber cm-1


(b)

Si-O-Si


Plasma induced Damage

3,14

3,66

2,94

3,57

2,322,53

2,89

4,31

2,092,26 2,11

2,242,08 2,15 2,25 2,31

1,5

2

2,5

3

3,5

4

4,5

k va

lue

at 1

00kH

z


Non protected PMMA protected

k-value

at 100KHz

• PID increases with etching time or etching depth.

• With Pore-Stuffing, PID penetration rate is improved by 80%.

• With formation of front and back electrodes, Ckis measured and k value could be extracted. With Pore stuffing, k value can be well controlled during plasma process.