Low damage integration of ultralow-k porous organosilicate ... ... Low damage integration of ultralow-k

  • View
    1

  • Download
    0

Embed Size (px)

Text of Low damage integration of ultralow-k porous organosilicate ... ... Low damage integration of...

  • 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, Belgium b 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

  • 3© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    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

    2 0

    11 2

    TW LRC εκρ Al → Cu

    3.0 → 1.72×10-8 Ωm SiO2 → Low-k 4.2 → 3.7-1.8

    ITRS, 2011

  • 4© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Porous OSGs

    • High porosity • Large pore size

    • Low plasma resistance • Low mechanical strength

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

    PECVD Spin-coating

  • 5© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Plasma Induced Damage (PID)

    Radical

    n H2OOO Si—CH3 O O Si—CH3 O O Si—CH3 O O Si—CH3 O

    Moisture uptake leads to increasing of k-value & leakage current

    OSG before etch

    P o

    re

    O O Si—OH O O Si—OH O O Si—OH O O Si—OH O

    3 H2O

    b

    UV, VUV

    OSG post etch

    Ion Radical Photon

    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*N0.5 (Random Walk Theory).

    N = 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

  • 6© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Advanced low-k integration

    Minimization of Plasma Induced

    Damage

    Optimization of Low-k materials

    Optimization of Plasma Etching

    Post process Dielectric Recovery

    - Engineering of the chemical composition and porous structure.

    - Optimization of plasma species or integration process, Post-plasma porogen removal.

    - Restoring the methyl groups by CH4 plasma or silylation agents, eliminating -OH bonds and physisorbed water by UV irradiation.

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

  • 7© 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

  • 8© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Damascene integration with Pore-Stuffing

    CuCu

    Porous low-k

    • Conventional Cu/low-k integration flow.

    • Damage caused by plasma etching + TaNTa barrier penetration degrades keff.

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

    • Polymers suppress the penetration of reactive radicals → lower C depletion • Supressed TaNTa penetration

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

  • 9© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    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-SIMS Carbon depth

    profile

    Ellipsometry Porosimetry

  • 11© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Polymer unstuffing Polymer 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.

  • 12© 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

  • 13© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    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

    or m

    al iz

    ed a

    bs or

    pt io

    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

    T h

    ic kn

    es s

    af te

    r et

    ch (

    n m

    )

    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

    Plasma chemistry and etch time

    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.

  • 14© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Fluorocarbon gas discharges

    1300 1250 1200 1150 1100 1050 1000 950

    0.0

    0.2

    0.4

    0.6

    0.8

    1.0

    N

    or m

    al iz

    ed a

    bs or

    pt io

    n (a

    .u .)

    Wavenumber (cm-1)

    PMMA protected Non protected Pristine p-OSG

    Plasma Chemistry Ar/CF4/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.

  • 15© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    Protection Mechanism

    0 50 100 150 200 250 300 350 400 1000

    10000

    0 50 100 150 200 250 300 350 400 1000

    10000

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

    In te

    ns ity

    ( co

    un ts

    /s )

    155 nm

    Carbon depth profile

    In te

    ns ity

    ( co

    un ts

    /s )

    Sputter Time (s)

    Fluorine depth profile 91 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.

  • 16© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    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

  • 17© 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

  • 18© IMEC 2013/ CONFIDENTIAL LIPING ZHANG

    300mm integration flow using PMMA

    Cu 40,1

    15

    2,6

    38,2

    7,5

    3,5

    0

    5

    10

    15

    20

    25

    30

    35

    40