Ellipsometry

Matt Brown
Alicia Allbaugh
Electrodynamics II Project
10 April 2001
 Ellipsometry

A method
of probing
surfaces
with light.
Introduction
 History
 Methodology
 Theory
 Types of Ellipsometry
 Applications
 Summary
History
 Fresnel derived his equations which
determine the Reflection/Transmission
coefficients in early 19th century.
Ellipsometry used soon thereafter.
 Last homework assignment
Electrodynamics I.
 Ellipsometry became important in
1960’s with the advent of smaller
computers.
 Methodology

 Polarized light is reflected at an oblique angle

to a surface
 The change to or from a generally elliptical
polarization is measured.
 From these measurements, the complex
index of refraction and/or the thickness of the
material can be obtained.
 Theory

 Determine r = Rp/Rs (complex)

 Find r indirectly by measuring the shape of
the ellipse
 Determine how e varies as a function of
depth, and thickness L of transition layer.
Note: We will focus on the case of very thin films.
In this case, only the imaginary part of r matters.
 z

1 Maxwell’s equations for a wave incident

x On a discontinuous surface. (Gaussian Units)
2
y
e    1   
Ex  H z  H y H x  E y  Ez
c t y z c t z y
e    1   
Ey  Hx  Hz H y  Ez  Ex
c t z x c t x z
e    1   
Ez  H y  H x H z  Ex  E y
c t x y c t y x

Boundary Conditions

E x1  E x 2 E y1  E y 2 e1Ez1  e 2 Ez 2
H x1  H x 2 H y1  H y 2 H z1  H z 2
 Derivation of Drude Equation
Fundamentals of Derivation
 Concept: Integrate a Maxwell Equation along z I  R
over transition region of depth L. Result will be a L
new Boundary Condition. Z
T
X Y
Y
 Fundamental Approximations:
 a. 
 1
L
 b. We assume certain field components ,
which vary slowly along z, are constant.
Incident
beam H p
H x Example: Since Hx+= Hx-, and
/L<<1, Hx1~Hx2.
H x
H p
 Derivation of Drude Equation
Incident
1    H p
H x  E y  Ez beam
c t z y H x

Assumption that E is uniform

With respect to y
z H x
H p
0
1   
H x  E y  Ez
c t z y R
I

Integrate along z over L
L
Z
1   T
L L


c t 0
H x dz   E y dz
0
z
Y
X
 Derivation of Drude Equation
Incident
Assumption that H x varies little:
Since H  H , H  H = constant.
beam H p
x x x1 x2
H x
1  L 
L
H x  dz  Hx
c t 0
c t H x
and H p

L
Substituting
0 dz z E y  E y 2  E y1 L 
H x  E y 2  E y1
c t
Rearrangement yields
L 
Ey2  E y1  Hx
c t
 I  R
1   
H y  Ez  Ex ; e1Ez1  e 2 Ez 2 Z
L
c t x z Y Y
T
X
Integrate
1  
L L


c t 0
H y dz   Ez dz  ( Ex 2  Ex1 )
0
x Dp  e  E p
H y
Dz 
Dp
H y and e z Ez vary
little over L
H y

  ez 
L L L
1 Dz 
0 x Ez dz  0 x e z Ez dz  x e z Ez 0 e z dz
L  
L
1
E x1  E x 2  H y 2  qe 2 Ez 2 where q   dz
c t x 0
e
 Similarly, we now find new B.C. for H x and H y
New complete Boundary Conditions

L  L
E x1  E x 2  H y 2  qe 2 E z 2 E y 2  E y1  Hx e1Ez1  e 2 Ez 2
c t x c t
 p  p 
H x1  H x 2  L H z 2  Ey2 H y1  H y 2  Ex 2 H z1  H z 2
x c t c t

Where I
 R
L
1 L L
q   dz p   dze Z
T
0
e Y Y
0
X
 We now solve Maxwell’s equations with
these new Boundary Conditions
Boundary l 
Condition
Ey2  E y1  Hx
c t
Relate
H and E H  e kˆ  E
Form of E field (to i ( k r t )
I
 R
satisfy Maxwell eq.)
E  E0 e L
Z
T
Continuity ( Einc  R  T )  nˆ  0 Y Y
X
E y1  Einc, y  Ry , E y 2  Ty H p2

L
Einc, y  R y  E y 2 (1  iCos[Y ] e 2 )
c
 Again solve Maxwell’s equations
with these new Boundary Conditions
Note on notation:
l  
Boundary Condition Ex1  Ex 2  H y 2  qe 2 Ez 2 Subscript p refers to
c t x component parallel to
incident plane (x-z plane),
Relate
H and E
H  e kˆ  E and subscript s refers to
perpendicular (same as y)
component.
Form of E field (to
satisfy Maxwell eq.)
E  E0 ei ( k r t ) I  R
L
Continuity Z
(ki  E  kr  R  kt  T )  nˆ  0 Y y T
X
E p1  Einc, p  R p E p2  Tp
i
( Einc, p  R p )Cos[ ]  E2 p (Cos[Y ]  ( e 2 l  Sin[Y ]2 e 2 q))
c
 This results in 4 relations between Einc , R , and E2 .

L
Einc, y  R y  E y 2 (1  iCos[Y ] e 2 )
c
i
( Einc, p  R p )Cos[ ]  E p 2 (Cos[Y ]  ( e 2 L  Sin[Y ]2 e 2 q))
c
p
( Einc, y  R y ) e1 Cos[ ]  E2 y ( e 2 Cos[Y ]  i ( L e 2 Sin[Y ]2  ))
c
p
( Einc, p  R p ) e1  E p 2 ( e 2  iCos[Y ] )
c
 Algebraically eliminate transmission terms.

Example: Parallel components

2
Cos[ ] e 2  Cos[Y ] e1  i ( pCos[ ]Cos[Y ]  ( L  qe 2Sin[ ]2 ) e1e 2
Rp
 
Ep 2
Cos[ ] e1  Cos[Y ] e 2  i ( pCos[ ]Cos[Y ]  ( L  qe 2Sin[ ]2 ) e1e 2

L L
1
where q   dz p   dze
0
e 0

Notice that if we assume p and q terms to be

Proportional to L, the imaginary parts of top and
Bottom are proportional to L

Approximation for when L<< such that terms
in second order of L/ can be neglected.

Rp Cos[ ] e 2  Cos[Y ] e1

Ep Cos[ ] e 2  Cos[Y ] e1
e1  pCos[Y ]  Le 2  qe 2 2 Sin[Y ]2
(1  2iCos[ ] )
e 2Cos[ ]  e1Cos[Y ]
2 2

Rs Cos[ ] e1  Cos[Y ] e 2

Es Cos[ ] e1  Cos[Y ] e 2
Le 2  p
(1  2iCos[ ] )
e 2Cos[ ]  e1Cos[Y ]
2 2
 R p Es R p
Set polarization at 45 degrees. Then 
E p Rs Rs

Using Snell’s Law, e1 Sin[ ]  e 2 Sin[Y]

We get
eCos[ 2  e Cos[Y  e1  e 2
e1  e 2
e 2Cos[ ]  e1Cos[Y ] 
2 2
(e1Sin[ ]2  e 2Cos[ ]2 )
e2
Again, keeping only terms to first order in L/, and using binomial expansion,

Rp Cos[  Y] 4 e 2 e1 Cos[ ]Sin[ ]2

 (1  i )
Rs Cos[  Y]  e1  e 2 e1Sin[ ]  e 2Cos[ ]
2 2

(e  e1 )(e  e 2 )
L

where   p  L(e1  e 2 )  qe1e 2   dz

0
e
 Rp Cos[  Y] 4 e 2 e1 Cos[ ]Sin[ ]2
 (1  i )
Rs Cos[  Y]  e1  e 2 e1Sin[ ]  e 2Cos[ ]
2 2

(e  e1 )(e  e 2 )
L
  p  L(e1  e 2 )  qe1e 2   dz
0
e
Recall that at Brewster’s angle Ep is minimized
So near Brewster’s Angle, we get
This is the
e e

 1 2 (e  e1 )(e  e 2 ) Drude
L
Rp
r  Im[ ] 
Rs 
 e1  e 2 0 e
dz
Equation.
For thin films, we often take e1 to be the dielectric constant
Of air, e 2 to be that of our substrate, and e to be constant
in the film. Then

 e1  e 2 (e  e1 )(e  e 2 )
Rp
r  Im[ ]  L
Rs  e1  e 2 e
Types of Ellipsometry
 Null Ellipsometry

 Photometric Ellipsometry
 Phase Modulated Ellipsometer

 Spectroscopic Ellipsometry
 Null Ellipsometry  rp 
We seek r  Im  
We choose  rs 
our polarizer
orientation
such that the
relative phase
shift from
Reflection is
just cancelled
by the phase
shift from the
retarder.

We know that the relative phase

shifts have cancelled if we can null
the signal with the analyzer
Example Setup
Phase modulated ellipsometer
 Rp 
We seek r  Im 
 Rs 
How to get r,an example.
Phase Modulated Ellipsometry
How to get r,an example.
Phase Modulated Ellipsometry

The polarizer polarizes light to

45 degrees from the incident plane.

 1 1
E  E0 ( sˆ  pˆ )
2 2
 How to get r,an example.
Phase Modulated Ellipsometry
The birefringment modulator
introduces a time varying phase shift.

 1 1
E  E0 ( sˆ  pˆ )
2 2
 E0
E ( sˆ  exp[ i 0 Sin[ 0t ] pˆ )
2
Note that at Sin[ 0t ]  0,
polarizati on is unchanged.
 0 determines the extrema.
 How to get r,an example.
Phase Modulated Ellipsometry
Upon reflection both the parallel
and perpendicular components are
changed in phase and amplitude.

 E0
E ( sˆ  exp[ i 0 Sin[ 0t ] pˆ ) For a discontinuous
2 interface,  p   s .

For a continuous
interface,  p   s

 E0
E (rs exp[ i s ]sˆ  rp exp[ i p  i 0 Sin[ 0t ]])
2
 How to get r,an example.
Phase Modulated Ellipsometry

E0
E (rs exp[ i s ]sˆ  rp exp[ i p  i 0 Sin[ 0t ]])
2

 E0
E (rs exp[ i s ]sˆ  rp exp[ i p  i 0 Sin[ 0t ]])
2
 How to get r,an example.
Phase Modulated Ellipsometry
Photomultiplier Tube measures intensity.
2  2rs rp 
2 
I  E  (rs  rp ) 1  2 2 Cos[   0 Sin[ 0t ]] 
2
 r r 
 s p 
Lockin Am plifier Where    p   s

Cos[   0 Sin[ 0t ]]  Cos[]J 0  0   2 J 2  0 Cos[2 0 ]  ...

 Sin[]2 J1  0 Sin[ 0t ]  2 J 3  0 Sin[3 ot ]  ...

Note: The J’’s are the Bessel Functions

 2rp  rs  rp rp
a 2
2 , since is small.
 rp  rs rs
1   
 rs  We orient the birefringe nt
Lockin Am plifier
modulator to set  0 to a
zero of J 0 :  0  0.765

1  2aJ1  0 Sin[]Sin[0t ]

I  rs  r 
2


2
p     

 Where    p   s
 2aJ 2 0 Cos[ ]Cos[ 2 0 t ] ...
 rp  rp
At the Brewster Angle, a Sin[]  2 Sin[]  2 Im[ ]  2 r
 rs  rs

1  4 J1  0 rSin[0t ]

I  rs  r 
2 2
 

 2aJ 2  0 Cos[]Cos[20t ]  ...
p
 How to get r,an example.
Phase Modulated Ellipsometry

1  4 J1  0 rSin[0t ]

I  rs  r 
2 2
 

 2aJ 2  0 Cos[]Cos[20t ]  ...
p

V 0  2 J1  0 r
V2 0   J 2  0 Cos[]
We find the Brewster angle by adjusting until V2 0  0,
Which is where     2 .
Now we can use a calibration procedure to
Find the proportionality of V 0 to r
Applications

 Determining
the thickness
of a thin film

 Focus of this
presentation
 Applications - Continued
 Research
 Thin films, surface structures
 Emphasis on accuracy and precision

 Spectroscopic
 Analyze multiple layers
 Determine optical constant dispersion relationship
 Degree of crystallinity of annealed amorphous silicon

 Semiconductor applications
 Solid surfaces
 Industrial applications in fabrication
 Emphasis on reliability, speed and maintenance
 Usually employs multiple methods
Ellipsometry
 Ellipsometry can measure the oxide depth.
 Intensity doesn’t vary much with film depth
but  does.
Other Methods
 Reflectometry

 Microscopic Interferometry

 Mirau Interferometry
Reflectometry
 Reflectometry
 Intensity of reflected to incident (square of
reflectance coefficients).
 Usually find relative reflectance.
 Taken at normal incidence.
 Relatively unaffected by a thin dielectric
film.
 Therefore not used for these types of thin films.
Ellipsometry
 Ellipsometry can measure the oxide depth.
 Intensity doesn’t vary much with film depth
but  does.
Reflectometry
Reflectometry
 Can be more accurate for thin metal films.
Microscopic Interferometry
 Uses only
interference
fringes.
 Only useful for
thick films and/or
droplets
 Thickness h>/4
Mirau Interferometry
 Accuracies to 0.1nm

 x is less than
present ellipsometry
 At normal incidence.

 Kai Zhang is
constructing one for
use at KSU.
 Ellipsometry
 Allows us to probe the surface structure of
materials.
 Makes use of Maxwell’s equations to
interpret data.
 Drude Approximation
 Is often relatively insensitive to calibration
uncertainties.
Ellipsometry
 Accuracies to the Angstrom
 Can be used in-situ (as a film grows)
 Typically used in thin film applications

 For more information and also this

presentation see our website:
html://www.phys.ksu.edu/~allbaugh/ellipsometry
 Bibliography
1. Bhushan, B., Wyant, J. C., Koliopoulos, C. L. (1985). “Measurement of
surace topography of magnetic tapes by Mirau interferometry.” Applied
Optics 24(10): 1489-1497.
2. Drude, P. (1902). The Theory of Optics. New York, Dover Publications, Inc.,
p. 287-292.
3. Riedling, K. (1988). Ellipsometry for Industrial Applications. New York,
Springer-Verlag Wein, p.1-21.
4. Smith, D. S. (1996). An Ellipsometric Study of Critical Adsorption in Binary
Liquid Mixtures. Department of Physics. Manhattan, Kansas State University:
276, p. 18-27.
5. Tompkins, H. G. (1993). A User's Guide to Ellipsometry. New York,
Academic Press, Inc.
6. Tompkins, H. G., McGahan, W. A. (1999). Spectroscopic Ellipsometry and
Reflectometry: A User's Guide. New Your, John Wiles & Sons, Inc.
7. Wang, J. Y., Betalu, S., Law, B. M. (2001). “Line tension approaching a first-
order wetting transition: Experimental results from contact angle
measurements.” Physical Review E 63(3).