Session 1B:

Substrates

James N. Hilfiker
University of Toledo
August 17, 2009
 1B Outline

ƒ WVASE Navigation
ƒ Substrates
– Data Interpretation
– Metals
– Dielectrics
•Transparent substrate
•Transmission
•Substrate backside reflection
– Semiconductors
2 © 2009 J.A. Woollam Co., All Rights Reserved
Ellipsometry Flow Chart

2
3

© 2009 J.A. Woollam Co., All Rights Reserved

 WVASE At a Glance

Experimental Data:
2 Display data
Model: Set up layers
with n, k and thickness 1

Hardware: 1
Generated Data:
Calculate from model 3 Collect data

Fit: Select fitting method

Graph: show data
and show results
4 5 and results

4 © 2009 J.A. Woollam Co., All Rights Reserved

 Demonstration

ƒ WVASE Navigation: “Demo_1_TOx.mat”

– Navigate windows
– Locate menu for each window
– Basic modeling procedures

© 2009 J.A. Woollam Co., All Rights Reserved

 WVASE32 Short-Cut Keys

ƒ CTRL-D: Defaults
ƒ CTRL-R: Range-Select Data
ƒ CTRL-G: Generate
ƒ CTRL-F: Normal Fit
ƒ CTRL-T: Toggle Through Graphs
ƒ CTRL-SHIFT-T: Toggle graphs in reverse
order.

6 © 2009 J.A. Woollam Co., All Rights Reserved

 Normal or Point-by-Point?

ƒ Use Point-by-Point, when

– all parameters are fixed except “n,k”.
– Final results may be significantly different
than starting “n,k”.

Don’t use Point-by-Point if fitting thickness

ƒ Use Normal Fit,

– Anytime a thickness is also being fit.
– When nominal “n,k” are very close to
expected answer.
7 © 2009 J.A. Woollam Co., All Rights Reserved
 Point-by-Point Fit
Cauchy Fit - Transparent Range Extrapolating Cauchy into Absorbing Region
25 40
Model Fit
Exp E 55°
20 30 Exp E 65°

Ψ in degrees
Ψ in degrees

Exp E 75°
15
20
10 Model Fit
Exp E 55°
Exp E 65° 10
5 Exp E 75°

0 0
300 600 900 1200 1500 1800 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Wavelength (nm) Photon Energy (eV)

Optical Constants
7.0
Index of Refraction ' n'

Cauchy - Extrapolate Transparent Range

6.0 Normal FIT
Point-by-Point FIT
5.0

4.0

3.0

2.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
8
Photon Energy (eV) © 2009 J.A. Woollam Co., All Rights Reserved
 Sample Categories

ƒ Substrate

ƒ Single-Layer film
– 1nm to 20 μm of
material on substrate

ƒ Multi-Layer films
– More than one film
with different n,k

9 © 2009 J.A. Woollam Co., All Rights Reserved

 What does the data “tell” us?

ƒ Study data features to classify sample type.

ƒ Choose appropriate model strategy.

Experimental Data
30

25
Exp E 65°
Exp E 75°
Ψin degrees

20 Data
Datainterpretation
interpretation
15 not
notquantitative
quantitative
10

0
300 600 900 1200 1500 1800
Wavelength (nm)

© 2009 J.A. Woollam Co., All Rights Reserved

Viewing Ellipsometry Data
ƒ GRAPH Window: “Type” Menu
– Psi (Ψ): Amplitude Ratio = Rp/Rs
• Range from 0° to 90°

– Delta (Δ): Phase Difference = φp-φs

• Range of 360°, depends on start.

– “Pseudo” optical functions:

<n>,<k>,<e1>,<e2>
• Transform data assuming 1-reflection
– DOES NOT APPLY TO THIN FILMS.
• Removes angle effects
11 © 2009 J.A. Woollam Co., All Rights Reserved
 “Pseudo” Optical Constants
ƒ <n>,<k> or <e1>,<e2>
– Transform Data assuming single reflection
– Removes Angle of Incidence Effects from Data

where,
⎛1 − ρ ⎞
2

n + ik = ε = ε a sin ϕ × 1 + tan ϕ ⎜⎜
2
⎟⎟ ρ = tan Ψ eiΔ
⎝1 + ρ ⎠ ϕ = ΑΟΙ

© 2009 J.A. Woollam Co., All Rights Reserved

 “Pseudo” Functions

Case 1: Substrate
– “Pseudo” values will match substrate optical constants if
no extra reflections (surface layer, backsides).

– Data from multiple angles have same “pseudo” values,

because angle is removed.
Experimental Data Experimental Data
45 160 20 21

42 140 0 18

-20 15
Ψ in degrees

39 120
Δ in degrees

Exp < ε1 >-E 70°

< ε1> -40 Exp < ε1 >-E 75° 12

< ε2>
Exp Ψ-E 70°
36
Exp Ψ-E 75°
100 Exp < ε2 >-E 70°
Exp Δ -E 70° -60 Exp < ε2 >-E 75° 9
33 Exp Δ -E 75° 80
-80 6
30 60
-100 3
27 40
0 300 600 900 1200 1500 1800 -120 0
0 300 600 900 1200 1500 1800
Wavelength (nm)
Wavelength (nm)

© 2009 J.A. Woollam Co., All Rights Reserved

 “Pseudo” Functions
Case 2: Thin Films
– “Pseudo” values do NOT give optical constants.
– Data from multiple angles will have
• Same “pseudo” values for thin or absorbing layers.
• Different “pseudo” values for thick layers.
Experimental Data Experimental Data
80 400 200 100
Exp <ε 1>-E 60°
Exp <ε 1>-E 75°
150 Exp <ε 2>-E 60° 50
60 300 Δ in degrees Exp <ε 2>-E 75°
Ψ in degrees

100 0

< ε 2>
< ε 1>
40 200
50 -50

20 Exp Ψ -E 60° 100

0 -100
Exp Ψ -E 75°
Exp Δ-E 60°
0 Exp Δ-E 75° 0
-50 -150
200 400 600 800 1000 1200 200 400 600 800 1000 1200
Wavelength (nm) Wavelength (nm)

© 2009 J.A. Woollam Co., All Rights Reserved

 Bare Substrates

ƒ What do we expect from Psi/Delta?

– Psi and Delta “features” from structure in
optical constants.

– Psi goes to minimum at Brewster angle.

• Minimum is zero for transparent substrates

• Minimum remains larger if substrate is absorbing.

– Delta is 180° or 0° for transparent substrates,

moves away from 0°,180° when substrate is
absorbing or if surface has thin film.

© 2009 J.A. Woollam Co., All Rights Reserved

 Bare Substrates

ƒ What do we expect from “Pseudo” values?

– <n> and <k> same for all angles.

– <n> and <k> same shape as optical constants.

– <k> can be non-zero if there is a surface film.

© 2009 J.A. Woollam Co., All Rights Reserved

 Substrate Classification

Bare Substrates

Dielectric Semiconductor Metal

SiO2 7.0
Silicon 6.0 4.0
Nickel 8.0
1.56

Extinction Coefficient

Extinction Coefficient
6.0 n 5.0 3.5 7.0

Index of Refraction
Index of Refraction

1.54
Index of Refraction

n 5.0
k
4.0 3.0
6.0
1.52
4.0 5.0
1.50 3.0 2.5
3.0 4.0
2.0 2.0 n
1.48 2.0 k 3.0
1.46 1.0 1.0 1.5 2.0
1.44 0.0 0.0 1.0 1.0
0 300 600 900 1200 1500 1800 0 300 600 900 1200 1500 1800 0 300 600 900 1200 1500 1800
Wavelength (nm) Wavelength (nm) Wavelength (nm)

17 © 2009 J.A. Woollam Co., All Rights Reserved

 Metal Substrate
ƒ Psi never close to zero, Delta away from 0°/180°
(typically increasing toward IR).
ƒ Pseudos follow optical constant shape.
Experimental Data Experimental Data
42 160

Exp E 60° 140

39
Exp E 65°
Ψ in degrees

Δ in degrees
Exp E 70°
36 120

33 100 Exp E 60°

Exp E 65°
Exp E 70°
30 80

27 60
200
2.00 400 600 800 1000 1200 1400 1600 200
6.0 400 600 800 1000 1200 1400 1600
Wavelength (nm) Wavelength (nm)
1.90
5.0 Exp E 60°
1.80 Exp E 65°
Exp E 70°
4.0
1.70
<k>
<n>

1.60 3.0
Exp E 60°
1.50 Exp E 65°
Exp E 70° 2.0
1.40
1.30 1.0
200 400 600 800 1000 1200 1400 1600
200 400 600 800 1000 1200 1400 1600
Wavelength (nm)
Wavelength (nm)
© 2009 J.A. Woollam Co., All Rights Reserved
 Dielectric
Substrate
ƒ Psi flat and smooth, goes close to zero at θB.
ƒ Delta = 0°,180° - except for surface films.

0.5nm roughness

© 2009 J.A. Woollam Co., All Rights Reserved

 Dielectric
Substrate
ƒ <n> – follows shape of substrate index.
ƒ <k> = 0 – except for surface films.

Experimental Data Experimental Data

1.70 0.08

1.65 Exp E 50° Exp E 50°

Exp E 60° 0.06
Exp E 60°
1.60 Exp E 70° Exp E 70°
<n>

<k>
0.04
1.55

0.02
1.50

1.45 0.00
0 300 600 900 1200 1500 1800 0 300 600 900 1200 1500 1800
Wavelength (nm) Wavelength (nm)

© 2009 J.A. Woollam Co., All Rights Reserved

 Dielectric
Substrate
ƒ Backside Reflections:
– Extra “intensity”, but no “phase”.
– Cause Shift in Psi and <n> versus angle.
Experimental Data Experimental Data
18 1.48
Exp E 45°
Exp E 50°
15 Exp E 55°
Exp E 45° 1.46
Exp E 50° Exp E 60°
12
Ψ in degrees

Exp E 55°
Exp E 60°

<n>
9 1.44

6
1.42
3
0 1.40
200 400 600 800 1000 200 400 600 800 1000
Wavelength (nm) Wavelength (nm)

© 2009 J.A. Woollam Co., All Rights Reserved

 Semiconductor Substrate
ƒ Psi goes to zero at θB except where absorbing.
ƒ Delta between 0°and 180°when absorbing
ƒ Pseudos follow shape of optical constants.
Experimental Data Experimental Data
50 180
Exp E 55° 150
40 Exp E 65° Exp E 55°
Ψ in degrees

Δ in degrees
Exp E 75° 120 Exp E 65°
30 Exp E 75°
90
20
60
10 30

0 0
7.00 300 600 900 1200 1500 1800 6.00 300 600 900 1200 1500 1800
Wavelength (nm) Wavelength (nm)
6.0 5.0
Exp E 55° Exp E 55°
5.0 Exp E 65° Exp E 65°
4.0
Exp E 75° Exp E 75°
<n>

<k>

4.0 3.0
3.0 2.0
2.0 1.0
1.0 0.0
0 300 600 900 1200 1500 1800 0 300 600 900 1200 1500 1800
Wavelength (nm) Wavelength (nm)
© 2009 J.A. Woollam Co., All Rights Reserved
 Substrate Modeling Summary

Bare Substrate

Dielectric
•Cauchy, Sellmeier
Semiconductor Metal
•Roughness •Direct Fit
•Crystalline values
•Transmission for “k” •Ignore oxide/
•Back Reflections?
•Surface Oxide roughness

23 © 2009 J.A. Woollam Co., All Rights Reserved

 Metal Substrate - Optical Behavior
ƒ Free-carrier absorption
– Strongly Absorb low frequency light waves
– “k” increases in IR spectral range
ƒ Optical properties depend on microstructure
4.0 8 3.0 18

Extinction Coefficient, k
Extinction Coefficient, k
Index of refraction, n

Index of refraction, n
3.5 n
Nickel 7 2.5 n 15
k 6 k
3.0 2.0 12
5
2.5 1.5 9
4
2.0 1.0 6
3
1.5 2
0.5 Aluminum 3

1.0 1 0.0 0
0 300 600 900 1200 1500 1800 0 300 600 900 1200 1500 1800
Wavelength (nm) Wavelength (nm)

24 © 2009 J.A. Woollam Co., All Rights Reserved

 Metal Substrate - Reality
ƒ Surface shifts resulting optical constants.
– Surface oxidation, roughness or contamination.
ƒ No backside reflection
Optical Constants
3.3 4.0

3.0

Extinction Coefficient ' k'

3.5
Index of Refraction ' n'

2.7

2.4 3.0

2.1 2.5
1.8
ta_no surface layer,n 2.0
1.5 ta_with 2 nm surface oxide,n
ta_no surface layer,k
ta_with 2 nm surface oxide,k
1.2 1.5
200 400 600 800 1000
Wavelength (nm)

25 © 2009 J.A. Woollam Co., All Rights Reserved

 Ideal Metal - Fitting Procedure

1. Add any metal as

substrate in model.
2. Fit “n” and “k” directly.
3. Normal fit.
4. Save material file
- Tabulated: Yes

26 © 2009 J.A. Woollam Co., All Rights Reserved

 Demonstration

ƒ “Demo_2_copper-bulk.mat”
– Fit optical constants of Cu
– Save optical constants and
compare with published results

© 2009 J.A. Woollam Co., All Rights Reserved

 Fitting Procedure
–Metal substrate with surface complications

1. Assume no surface layers.

2. Fix nominal surface layer (thickness and O.C.).

– Oxide (e.g., Al2O3) or surface roughness (srough.mat)

3. Measure film in-situ in vacuum.

28 © 2009 J.A. Woollam Co., All Rights Reserved

 Examples: Opaque
Substrates
ƒ “Example_1_Nb.dat”
– Determine optical constants with:
• Ideal Surface
• 2nm Roughness (use “srough.mat”)
– Compare Resulting Optical Constants

© 2009 J.A. Woollam Co., All Rights Reserved

 Opacity Limit
ƒ Consider metal a substrate if no light returns from
bottom of layer.
– Typically 50-100nm thick.

Experimental Data
42
10nm

40
20nm
30nm Ni on Si
40nm
Ψ in degrees

50nm
38 100nm
200nm
36

34

32

30
300 600 900 1200 1500 1800
Wavelength (nm)

30 © 2009 J.A. Woollam Co., All Rights Reserved

 Demonstration
Generate Data Window |Range menu

ƒ Build Model with “SiO2.mat”

substrate and al.mat film.

Generate:
ƒ Ellipsometry Data
*300nm to 1000nm
*70°
ƒ Transmission Intensity Data
*300nm to 1000nm
*0°

What is the effect of increasing Au layer thickness?

© 2009 J.A. Woollam Co., All Rights Reserved
 Metal Substrate Review

ƒ Ideal surfaces
– Direct Fit of n and k.
ƒ With surface complications
– Ignore and treat the samples as ideal
– Add oxide or roughness with fixed n, k and
nominal thickness
– Monitor the thickness growth in-situ
ƒ Thick opaque metal films are modeled the
same as substrates

32 © 2009 J.A. Woollam Co., All Rights Reserved

Break!!

33 © 2009 J.A. Woollam Co., All Rights Reserved

 Dielectrics Substrate

ƒ Dielectrics have low refractive indices

(often below 2)

ƒ Often transparent in visible and infrared

– k equal to zero or very small (<10-4)

ƒ Non-idealities Index of Refraction, N

2.2
– Surface roughness
n' '
Index of refraction
2.0
– Backside reflection SiO2
1.8 Si3N4

1.6

1.4
300 600 900 1200 1500 1800
Wavelength (nm)

34 © 2009 J.A. Woollam Co., All Rights Reserved

 Demonstration
ƒ “Demo_3_BK7_SE.dat”
– Use Same Approach as with Metals

– Are the Optical Constants Correct?

– Generate Transmission compared to

“Demo_3_BK7_T.dat”

Two Problems
•Noisy Index
•Absorption and Surface are Correlated.

© 2009 J.A. Woollam Co., All Rights Reserved

 Using a Dispersion Relation

ƒ Dispersion Relation: Mathematical description of

optical constants as a function of wavelength

ƒ Advantages:
– Optical Constants smooth and “noise-free”.
– Reduce the numbers of fit parameters.
– Better for interpolating and extrapolating

36 © 2009 J.A. Woollam Co., All Rights Reserved

 Cauchy Model
A sets index range
1.65

n(λ)=A+B/ λ 2+C/λ 4
1.62
n'

1.59
Index of refraction '

1.56

1.53 k(λ)=0
1.50 B and C give
dispersion shape
1.47

1.44
200 400 600 800 1000 1200
Wavelength (nm)

ƒ Empirical relationship
ƒ Describe index (n) as a function of wavelength (λ)
ƒ Only for TRANSPARENT materials (k=0 or very small)
37 © 2009 J.A. Woollam Co., All Rights Reserved
 Cauchy.mat Layer

Cauchy
Parameters

Bn Cn
n(λμm ) = An + +
λμm 2
λμm 4

Urbach
Parameters

⎛ ⎛ 1 1 ⎞⎞
Exponent ⎜⎜ 1240 ⎜⎜ − ⎟⎟ ⎟
⎟
⎝ λnm BandEdge ⎠ ⎠
k (λnm ) = k amp e ⎝

38 © 2009 J.A. Woollam Co., All Rights Reserved

 Fitting Procedure

Index of Refraction, n:
1. Add “cauchy.mat”
2. Fit An and Bn parameters to match Ψ.
3. Add Cn if needed.
4. Add “srough.mat” and fit thickness to match Δ
Extinction Coefficient, k:
5. Append Transmission Intensity data.
6. Fix surface layer thickness (and Cauchy parms)
7. Fit “k” directly.

39 © 2009 J.A. Woollam Co., All Rights Reserved

 Example
ƒ “Example_2_float glass_SE.dat” and
“Example_2_float glass_t.dat”
– Append files together in Exp. Data window
– Assume thickness = 1mm.
– Fit with Cauchy, Roughness, and “k” direct fit.

What if our thickness estimate is incorrect?

© 2009 J.A. Woollam Co., All Rights Reserved

 Transparent Substrates Review

ƒ Fit Ψ using Cauchy Index dispersion

(An, Bn and Cn parameters).
ƒ Add Roughness to fit Δ
ƒ Append Transmission data and direct
fit extinction coefficient (k)

41 © 2009 J.A. Woollam Co., All Rights Reserved

 Ideal geometry
Only collect Front Surface reflection

DE
TE
AN CT
N

O
IO
R UT

AL R
AT

YZ
LA INP
IZ

ER
PO

Surface of
interest

42 © 2009 J.A. Woollam Co., All Rights Reserved

 Actual Geometry (Backside reflections)

Front surface beam

ƒ Collect First reflected
1st backsurface beam
beam, but additional
reflections (from back) 2nd backsurface
may overlap. beam
detector
aperture
ƒ Backside reflections
are ‘incoherent’

Surface of interest
43 © 2009 J.A. Woollam Co., All Rights Reserved
 Suppress Backside Reflections

ƒ Roughen backside with grinder, sandpaper,

or sandblaster

ƒ Index-match backside to scatter/absorb light

– translucent Scotch tape*.
ƒ Spatially separate Front and Back reflections
– Focusing optics
– Wedge or Thick substrate.
*R.A. Synowicki, “Suppression of backside reflections from transparent substrates”
Phys. Stat. Sol. (c) 5 No. 5 (2008) 1085-1088.
44 © 2009 J.A. Woollam Co., All Rights Reserved
 Software Correction

ƒ Correct in software
– Measure at multiple angles to avoid error
ƒ From “Experimental Data” window
– “Change Data Type”

45 © 2009 J.A. Woollam Co., All Rights Reserved

 Model Options Window

ƒ Number of backside reflection:

– Default: 3
– Can be small for thick substrates

46 © 2009 J.A. Woollam Co., All Rights Reserved

 Fitting Procedure
- Substrates with backside
reflections

1. Change Data Type (Backside Corrected).

2. Add Cauchy layer to Model.
3. Fit An and Bn. Fit Cn if needed.
4. Open Model-Option menu and fit
“# of backside reflection”
5. Add surface roughness to fit Δ.

47 © 2009 J.A. Woollam Co., All Rights Reserved

 Demonstration

ƒ “Demo_4_glass disk roughened.dat” and

“Demo_4_glass disk not roughened.dat”
– Compare psi and delta

– Compare n and k after correct modeling

© 2009 J.A. Woollam Co., All Rights Reserved

 Example
ƒ “Example_3_CaF2_50degree.dat”
ƒ “Example_4_CaF2.dat”
– Fit each data set and save result as
Material File for Comparison.

Can we get the correct answer from one angle

of incidence?

© 2009 J.A. Woollam Co., All Rights Reserved

 Identifying Backside Reflections

ƒ Shifts in ψ and <n> vs angle.

Experimental Data
45 1.48
E 60° Exp E 45°
38 E 65° Exp E 50°
E 70° Exp E 55°
Eb 60° 1.46 Exp E 60°
30 Eb 65°
Ψ in degrees

Eb 70°
23

<n>
1.44

15
1.42
8

0 1.40
400 500 600 700 800 900 1000 200 400 600 800 1000
Wavelength (nm) Wavelength (nm)

50 © 2009 J.A. Woollam Co., All Rights Reserved

Break!!

51 © 2009 J.A. Woollam Co., All Rights Reserved

 Semiconductor Substrates

ƒ Optical constants of crystalline semiconductors

are often well-known
– consistent from wafer to wafer in the UV-NIR region
– Doping primarily important in Infrared.
ƒ Large Index, Strong Absorption above band gap
ƒ Non-Ideality
– Surfaces often oxidized: Very sensitive to surface layers.
– Watch for backside reflections if double-side polished
(Only affects data below bandgap – longer wavelengths).

52 © 2009 J.A. Woollam Co., All Rights Reserved

 Semiconductor Fitting Procedure
Procedure #1: Known Semiconductor
1. Fix optical constants at published values.
2. Fit native oxide thickness.

Procedure #2: New Semiconductor

1. Use Transparent Substrate approach to fit data
at longer wavelengths (where transparent).
2. Fix surface thickness (oxide or roughness).
3. Direct fit for “n,k” at all wavelengths.

53 © 2009 J.A. Woollam Co., All Rights Reserved

 Examples
PROCEDURE #1
“Example_5_Si wafer.dat”
si_jaw.mat, ntve_jaw.mat

PROCEDURE #2
“Example_6_GaAs wafer.dat”
cauchy.mat, gaas-ox.mat

© 2009 J.A. Woollam Co., All Rights Reserved

 Substrate Modeling Summary

Bare Substrate

Dielectric
•Cauchy, Sellmeier
Semiconductor Metal
•Roughness •Direct Fit
•Crystalline values
•Transmission for “k” •Ignore oxide/
•Back Reflections?
•Surface Oxide roughness

55 © 2009 J.A. Woollam Co., All Rights Reserved

 Dielectric
Substrate
ƒ Cauchy or Sellmeier for “n” (match Ψ)
ƒ Srough (match Δ)
ƒ Fit “k” directly (match Transmission),
otherwise assume k=0.
ƒ Backside Reflections
Suppress:
– Roughen, Tape, Spatially Separate
– If Anisotropic, important to suppress!
Model:
– Change Data Type (Backside Corrected)
© 2009 J.A. Woollam Co., All Rights Reserved
 Semiconductor Substrate

ƒ FIX tabulated “n,k” for crystalline

semiconductors.
ƒ Fit thickness of surface oxide.

ƒ If “n,k” not known, fit index only in

transparent region (below bandgap)
with surface roughness or oxide.
Then, fix surface and fit “n,k” directly.

© 2009 J.A. Woollam Co., All Rights Reserved

 Metal Substrate

ƒ Fit “n,k” directly.

ƒ Correlation prevents measurement of
surface layer, so either:
– Assume no surface layers.
– Fix surface layer thickness (and n,k) at
nominal values.

© 2009 J.A. Woollam Co., All Rights Reserved

 Outline

ƒ WVASE Navigation
ƒ Substrates
– Data Interpretation
– Metals
– Dielectrics
•Transparent substrate
•Transmission
•Substrate backside reflection
– Semiconductors
59 © 2009 J.A. Woollam Co., All Rights Reserved