ligo-t1000126-v1 rand dannenberg 1 some defect counting microscope results – lessons learned dark...

LIGO-T1000126-v1Rand Dannenberg

1

Some Defect Counting Microscope Results – Lessons

LearnedDark Field Particle Counting on

LMA Micromap Sample&

Tinsley Scatter Master Sample


2

LMA Micromap Sample

• 3” diameter fused silica.

• 20X objective, 99 frames analyzed - 26.8 mm2 total area.

• Histogram bin size varied: 2 um, 0.5 um, 0.1 um.

• Same Data manipulated from single map for fixed microscope, contrast & analysis settings.Same Data manipulated from single map for fixed microscope, contrast & analysis settings.

• Goal was to understand binning.


3

LMA Micromap Sample – Binning: the result can be made nearly independent of the bin size by dividing by the bin size. Below, the results are NOT normalized to a bin size.

All 20x Objective, bin size varies

0

100

200

300

400

500

600

700

0 5 10 15 20 25 30 35

Equidiameter (um)

De

fec

t D

en

sit

y (

co

un

t/m

m2 )

20X - 2 um bins

20X - 0.5 um bins

20X - 0.1 um bins


0.01

0.1

1

10

100

1000

0 5 10 15 20 25 30 35

Equidiameter (um)

De

fec

t D

en

sit

y (

co

un

t/m

m2

)

20X - 2 um bins

20X - 0.5 um bins

20X - 0.1 um bins

scannedtotalA

bEcountsbEDensity

_

),(),(


4

LMA Micromap Sample – Binning: By dividing by the bin size (in microns) the defect density is normalized to a bin size of 1 micron. Larger bin sizes then serve to integrate the data and reduce noise.


0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 5 10 15 20 25 30 35

Equidiameter (um)

De

fec

t D

en

sit

y (

co

un

t/m

m2 )

No

rma

lize

d t

o 1

um

bin

20X - 2 um bins / bin size

20X - 0.5 um bins / bin size



0.01

0.1

1

10

100

1000

10000

0 5 10 15 20 25 30 35

Equidiameter (um)

De

fec

t D

en

sit

y (

co

un

t/m

m2 )

No

rma

lize

d t

o 1

um

bin

20X - 2 um bins / bin size



bA

bEcountsbENormDensity

scannedtotal

1),(),(_

_


5

LMA Micromap Sample – Conversion of the bin-normalized defect density to a fraction of area covered is done by multiplying by the average defect size S in a bin range at each equidiameter E. The b is the bin size. This produces a peak at the most important equidiameter. Note the peak position is slightly dependent on the bin size. The micromap

sample has mostly small defects 1.1 um < E < 1.5 um, discernable with the smaller bin sizes. Bin sizes that are too big give higher peak values (the 2 um bin has a peak near 3.5 um equidiameter).

),(1),(

),(

2

)(

2

1),(

_

2

bESbA

bEcountsbEFC

bEEbES

scannedtotal


0

0.0001

0.0002

0.0003

0.0004

0.0005

0.0006

0.0007

0.0008

0 5 10 15 20 25 30 35

Equidiameter (um)

Fra

ctio

n C

ove

rag

e N

orm

aliz

ed t

o 1

um

bin

Fraction Coverage Norm to 1 um bin (2 um bin)

Fraction Coverage Norm to 1 um bin (0.5 umbin)

Fraction Coverage Norm to 1 um bin (0.1 umbin)


6

LMA Micromap Sample – Integrating over equidiameter then gives the cumulative coverage fraction.

Very weakly bin size dependent.

EE

bSbDensitybFCbbEFCCumulative

00

),(),(),(),(_

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0 5 10 15 20 25 30 35

Equidiameter (um)

Cu

mu

lati

ve

Co

ve

rag

e F

rac

tio

n

Cumulative Coverage Fraction - 2 um bin

Cumulative Coverage Fraction - 0.5 um bin

Cumulative Coverage Fraction - 0.1 um bin


7

Tinsley Scatter Master Sample

• 1” x 3” microscope slide.

• 20X objective, 99 frames analyzed - 26.8 mm2 total area.

• 5X objective, 24 frames analyzed – 102 mm2 total area.

• Histogram bin size fixed at 0.5 um.

• Contrast (exposure time) must change on changing magnification,Contrast (exposure time) must change on changing magnification,

• Analysis settings (threshold + restrictions) must also change.Analysis settings (threshold + restrictions) must also change.


8

Tinsley Scatter Sample – Defect density normalized to 1 micron bin.

Objective Lens varies, bin size fixed

0

50

100

150

200

250

0 5 10 15 20 25 30 35

Equidiameter (um)

Def

ect

Den

sity

(co

un

t/m

m2)

No

rmal

ized

to

1 u

m b

in

5X Obj (count/mm^2)/(bin size)



0.01

0.1

1

10

100

1000

0 5 10 15 20 25 30 35

Equidiameter (um)

Def

ect

Den

sity

(co

un

t/m

m2)

No

rmal

ized

to

1 u

m b

in



Note that the 5X objective is more sensitive tolarger defects, the 20X objective is moresensitive to smaller ones, while in the mid-rangethey are in closer agreement.


9

Tinsley Scatter Sample – Coverage Fraction normalized to 1 micron bin.


0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

0 5 10 15 20 25 30 35

Equidiameter (um)

Fra

cti

on

Co

ve

rag

e N

orm

aliz

ed

to

1 u

m b

in

Fraction Coverage Norm to 1 um bin (0.5 um bin)- 20x

Fraction Coverage Norm to 1 um bin (0.5 um bin)- 5x

The peak range varies with the magnification. 5X is 13-16 um while at 20X the peak range is 7 – 11 um.


10

Tinsley Scatter Master Sample – Integrating over equidiameter then gives the cumulative coverage fraction.

Strongly magnification dependent.

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0 5 10 15 20 25 30 35

Equidiamter (um)

Cu

mu

lati

ve C

ove

rag

e F

ract

ion

Cumulative Coverage Fraction - 5X Obj

Cumulative Coverage Fraction - 20X Obj


11

Conclusions

• The results thus far seem to be more strongly magnification dependent, and less strongly bin size dependent.

• For a sample type, the bin size and magnification should be standardized before comparing with a scatter measurement.

• How to standardize will come with experience in analyzing real samples that are routinely generated.

ligo-t1000126-v1 rand dannenberg 1 some defect counting microscope results – lessons learned dark...

Documents