Scribd Logo
Upload

Welcome to Scribd!

  • 29 views

    Advances in LPC Monitoring in CMP Slurries

    Uploaded by

    Average Joe
    1) Advances in monitoring large particle counts (LPCs) in CMP slurries can improve yields by detecting statistically significant differences in LPC levels that correlate with defects. 2) Ideal LPC monitors would be inexpensive, point-of-use without dilution, and accurately find LPCs to maximize yield return on investment. 3) Proper dilution and measuring sufficient particle counts are needed to avoid coincidence errors and accurately detect even low LPC levels, respectively. Tens of thousands of particles preferably hundreds of thousands need to be counted.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Advances in LPC Monitoring in CMP Slurries

    Uploaded by

    Average Joe
    29 views20 pages
    1) Advances in monitoring large particle counts (LPCs) in CMP slurries can improve yields by detecting statistically significant differences in LPC levels that correlate with defects. 2) Ideal LPC monitors would be inexpensive, point-of-use without dilution, and accurately find LPCs to maximize yield return on investment. 3) Proper dilution and measuring sufficient particle counts are needed to avoid coincidence errors and accurately detect even low LPC levels, respectively. Tens of thousands of particles preferably hundreds of thousands need to be counted.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    1) Advances in monitoring large particle counts (LPCs) in CMP slurries can improve yields by detecting statistically significant differences in LPC levels that correlate with defects. 2) Ideal LPC monitors would be inexpensive, point-of-use without dilution, and accurately find LPCs to maximize yield return on investment. 3) Proper dilution and measuring sufficient particle counts are needed to avoid coincidence errors and accurately detect even low LPC levels, respectively. Tens of thousands of particles preferably hundreds of thousands need to be counted.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    29 views20 pages

    Advances in LPC Monitoring in CMP Slurries

    Uploaded by

    Average Joe
    1) Advances in monitoring large particle counts (LPCs) in CMP slurries can improve yields by detecting statistically significant differences in LPC levels that correlate with defects. 2) Ideal LPC monitors would be inexpensive, point-of-use without dilution, and accurately find LPCs to maximize yield return on investment. 3) Proper dilution and measuring sufficient particle counts are needed to avoid coincidence errors and accurately detect even low LPC levels, respectively. Tens of thousands of particles preferably hundreds of thousands need to be counted.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 20

    Advances in LPC* Monitoring

    in CMP Slurries
    Statistics are a Bear

    Mark Bumiller
    Particle Sizing Systems

    (*LPC = Large Particle Counts, typically > 1 µm)


    LPC vs. Defects/Scratches (Yield)*

    • Slurries differing in LPC


    by a minimum
    of 1.8×105particles/g
    slurry had statistically
    different predicted scratch
    events at the 95%
    confidence level.
    • LPC from PSS AccuSizer
    780
    – LE400 (extinction) sensor

    *Remsen, E. et al., Analysis of Large Particle Count in Fumed Silica Slurries and Its Correlation with 
    Scratch Defects Generated by CMP, Journal of The Electrochemical Society, 153 (5) G453‐G461(2006)
    LPC Correlates to Yield
    *

    **

    * R. Nagahara et al, VMIC‐MIC, 1996

    **from Y.‐J.Seo et al., Effects of slurry filter size on the chemical mechanical 
    Polishing (CMP) defect density, Materials Letters, 58(2004) 2091‐2095
    The Wish
    • Please give me an ideal LPC monitor:
    • Inexpensive • Yield ROI?
    • Point of Use? • Yes, LPCs generated
    • No Dilution? • Coincidence (statistics)
    • Finds LPC’s • Statistics matter
    CMP Filtration Monitoring
    • Incoming slurry supply line
    Supply Line; Particles/ml greater or equal to 1.01 um
    • At point of use
    • Change filters based on 24 hr

    particle count, not just on


    expected life time = better
    yield, lower cost
    tote              mixer           day tank     
    manifold
    filter

    POU filter POU filter

    AccuSizer Mini AccuSizer Mini

    polishing tools
    Effect of Pumping (Shear Stress)

    Increased circulation

    S/C = scratches
    Effect of Pumping (Sedimentation)

    • Top vs. bottom of 
    tank
    • Supply line LPC > 
    1.01 µm
    • Online FX results

    New tote

    • From day tank
    Effect of Slurry Pumping (in the Tool)

    • Real time monitoring


    of LPC counts
    downstream of a
    peristaltic pump
    provides feedback on
    pump performance
    and generated LPCs.

    Monitoring LPC Counts of Silica Slurry in a Loop


    No Dilution? Avoid Coincidence Errors
    • Must avoid coincidence error for best data
    • Only one particle in measurement zone
    • Collimated beam: 0.5 - 400 µm 60 ml/min
    • 10,000 particles/mL
    Advancement: Focused Beam
    • Focused beam
    • ~106 particles/mL
    • 103 improvement
    • FX: 0.7- 20 µm
    • FX Nano .15 – 10 µm
    • But dilution may still
    be required to avoid
    coincidence
    Check for Coincidence Error
    • Measure at several different concentrations (dilutions)
    • Should have linear response @ size of interest

    Error @ 1 µm

    No error @ 5 µm
    Advancements in Autodilution
    Internal Fluidics
    2013                              2016
    Dilution Fluidics:
    HEADING2 Stage Dilution
    Variable sample loop volume
    = VL= 1, 3, or 5 mL

    Volume = V1= 30 mL

    Goodbye coincidence errors


    Tailor Measurement to Slurry

    Plus many controls fixed in firmware


    Parameters saved for each slurry
    Distribution Statistics: Standard Error
    SE=s/(√n) where
    s = standard deviation
    n= number of particles counted

    So n = s2/SE2

    If the sample has a standard deviation of 2 and we


    can live with a 2% standard error, then the number
    of particles, n, that needs to be measured is:

    n = 22/ (0.02)2 = 10,000


    ISO 13322-1 Approach*

    n* with admissible error of 5% as a function of the geometric standard deviation


    GSD of the sample. Here, the probability, P, is taken as P = 0.95

    *ISO 13322-1, Particle Size Analysis—Image Analysis Methods—Part 1: Static Image Analysis Methods, available at www.iso.org
    H. MASUDA & K. GOTOH, Study on the sample size required for the estimation of mean particle diameter,
    Advanced Powder Technol., 10(2), 1999, pp. 159-173
    Experimental Confirmation

    # Sized D10 (µm) D50 (µm) D90 (µm)


    442 8.5 13.82 21.65 D90
    D50
    2579 8.3 13.82 20.99 D10
    5213 8.73 13.82 21.71
    11364 9.47 15.39 25.63
    13196 10 16.24 25.02
    16748 17.41 24.07 44.11
    37688 17.14 29.42 50.45

    NEED to count tens of thousands,


    preferably hundreds of thousands of
    particles to achieve statistical accuracy
    at tails of the distribution (LPC)
    ?
    LPC Sensitivity

    • Spike silica slurry w/1 µm


    PSL standard
    • Dilute slurry A 200:1
    • Add 1.44 mL diluted PSL
    stock at 1.74 × 107/mL
    • Measure on AccuSizer FX
    Point of Use system
    • Note increase of ~100,000
    particles/mL @ expected
    size = excellent recovery
    Ceria CMP Slurry
    • Spike ceria A slurry
    w/ 106/mL 1 µm PSL

    • Mix Slurry A & B

    99% Ceria A + 1%
    Ceria B

    99% Ceria A + 1%
    Ceria B

    90% Ceria A +
    10% Ceria B
    Conclusions
    • It’s worth the investment: increased yields
    • Point of Use monitors detect LPCs where
    they cause troubles
    • Need autodilution, tailored to the slurry
    • Need to measure 104 particles to always
    find LPCs
    • Can’t beat statistics
    – Sorry, you won’t win this jackpot
    Booth 1100
    [email protected]

    You might also like