Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

VI.

Geometrical Optics

GEOMETRICAL OPTICS

Introduction
The description of light propagation in the limit when 0 is known as geometrical or
ray optics. In this limit, diffraction effects disappear and the normal to the wavefront (or
phase front) completely describes the propagation of light. The normal to the wavefront
is known as the optical ray hence the term ray optics.
It can be shown that light takes a path through an optical system that minimizes the
optical path length (i.e. the physical length multiplied by n) between two points. This is a
restatement of Fermats principle;
Light travels the path which takes the least time
this principle guides the design of optical
systems that rely on refraction and reflection
(i.e. contain lenses and mirrors).
E.g. Law of Refraction:
A

d-x

i
ni
O

nt
b

t
B

The time t taken for light to propagate from A to B is given by

t=

ni
n
(AO) + t (OB)
c
c

n
= i
c

n
a 2 + (d x )2 + t
c

b2 + x2

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

The minimum time can be found by evaluating

dt
= 0;
dx

ni (d x )
nt x
dt
=
+
= 0
dx
c b2 + x2
c a 2 + (d x )2

n
n
i sin i + t sin t = 0
c
c
ni sin i = nt sin t
as expected

NOTE: It can be shown that this is a minimum by proving

d 2t
dx 2

>0.

N.B.
(i)

For an optical system, aberrations indicate the accuracy of the geometrical

optics approximation. A diffraction limited optical system is one for
which geometrical optics accurately describes the propagation of light
except for diffraction effects.

(ii)

A stigmatic optical system is one which focuses an object to an image;

Optical
System

Image
(blurred by
diffraction
effects)

Object

S and P are conjugate points.

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Refraction at Curved Surfaces

Consider an object O at a distance so from a curved surface that refracts the light to
produce an image I at a distance si from the surface at the vertex point V;
A

li

lo

n1

n2

so

si

Since O and I are conjugate points, the optical distances OVI and OAI must be equal, i.e.

n1l o + n2 li = n1s o + n2 si
For fixed image and object distances (si and so) we get

n1 l o + n2 li = constant
This is the equation of a Cartesian oval, and describes the shape required by the curved
surface to image the object at point I. In general, the surface is somewhere between an
ellipsoid or hyperboloid, since these two surfaces are required to produce an image or
object at infinity, e.g.

n1 < n2

hyperbolic

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

n1 > n2

ellipsoidal

NOTE: Objects and images can be reversed in each case (conjugate points).

Unfortunately, these aspheric surfaces are difficult and expensive to manufacture, so

spherical surfaces are usually employed. The error that this introduces is insignificant
provided that the optical rays do not strike the interface too far from the axis of the
system so that a spherical surface is a good approximation to the optimal aspheric
surface;

n1 < n2
Ideal aspheric surface

C (Center of curvature for spherical surface)

Spherical surface

i.e. for small angles , refraction

from spherical and aspheric surfaces
cannot be distinguished.

Paraxial Approximation.

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Paraxial/Gaussian Optics
Consider refraction at an interface with a radius of curvature R in the paraxial limit
(sin , tan , cos 1) ;
A
1

x
O

li

lo

n1

n2

so

si

n1 < n2
Now, Snells Law in the paraxial limit reads

n11 = n2 2
And 1 and 2 are given by

1 = [ ( + )] = +
2 = [ + ( )] =
with sin =

x
.
R

Hence, Snells law becomes

x
n1 +
R

= n 2
R

Now, for small angles li si and l o so so that =

n1 n2 n2 n1
+
=
s o si
R

x
x
and = ;
so
si

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

There are conventions associated with this equation describing refraction of light at a
curved interface:

so;

+ve if O on incident side

-ve if O on transmission side
+ve if I on transmission side
-ve if I on incident side
+ve if C on transmission side
-ve if C on incident side

si ;
R;

This equation also allows the imaging of a lens to be described, since a lens is simply two
curved dielectric interfaces.

R1

R2

n1

n2

n3

For the first surface we have

n 2 n2 n1 n1
=

si
R1
so
And if we have a thin lens (i.e. d << image and object distances), then the image distance
si becomes the object distance for the second surface (i.e. s 'o = si ) , hence

n 2 n3 n3 n 2
+
=
si s 'i
R2

n
n n2
n
2 = 3 3
si
s 'i
R2

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

So, if n1 = n3 = 1 and n2 = n as is the case for a lens in the air, we have

n 1 1
1 n 1

=
+
R1 s o
s ' i R2
Thus the focal length, f, of the lens is given by

1
1
1 1
1

=
+ = ( n 1)
f
s o si
R1 R2
the lens-makers equation

N.B.
(i)

The same conventions apply to the above equation.

(ii)

If the lens is thick then the analysis proceed as outlined above with
s ' o = d si .

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

The Ray Transfer Matrix Method

__

a method of ray tracing using 2 x 2 matrices operating on column vectors

describing the height x above the optical axis and the angle of inclination with
respect to the optical axis.

Consider two reference planes surrounding an optical system;

1

2
Optical
System

x1

x2
Optical
Axis

Reference plane #1

Reference plane #2

The optical system can be represented by a 2 x 2 matrix so that

x 2 A B x1
=

2 C D 1
A B
is the ray transfer matrix (RTM).
Where (xi, i) is the ray vector and
C D
In other words, we have

x 2 = Ax1 + B1
2 = Cx1 + D1 .

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Several examples follow:

(i)

Propagation in dielectric medium

e.g. free space

x2

1
x1

d
#1

#2

Clearly:

x 2 = x1 + d1
2 = 1
1 d

RTM =
0 1

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

(ii)

Geometrical Optics

Single refracting surface

1
2

x1
O

x2
2

So

Si

n1

n2
#1

#2

For a thin surface the two reference places are essentially coincident so that

x2 = x1
The relationship between 1 and 2 can be found using

n 2 n1 n1 n 2
=
+
R
s o si
x
x
x
since 1 = 1 and 2 = 2 = 1 .
so
si
si
Hence,
n 2 n1
x1 = n11 n 2 2
R
n n1
n
x1 + 1 1
2 = 2
n2
n2 R

RTM = (n 2 n1 )

n2 R

10

0
n1

n2

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

(iii)

Geometrical Optics

A lens

By either multiplying two RTMs for the two carves surfaces of the lens or by following
the object-image procedure used above, it is straight forward to show that the RTM for a
lens is given by
1

RTM = 1
f

N.B.
(i)

For a sequence of optical elements, surfaces and propagation distances

described by RTMs M1, M2.Mi, the ray vector emerging from the optical
system is given by

x i +1
x

= (M i )(M i 1 )...(M 2 )(M 1 ) 1

1
i +1
where ( x1 , 1 ) is the incident ray vector.

(ii)

For an imaging optical system described by

A B

= (M i )(M i 1 )...(M 2 )(M 1 )

C D
The position of the image is given by B = 0.

(iii)

The power, , of a refracting surface is defined by

n n
= 2 1
R

units of diopters.

So the power of a lens is given by

lens = 1 + 2 =

1
n 1 1 n
1 1
=
+
= (n 1)

R1
R2
R1 R2 f
as expected

11

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Reflective surfaces
Consider a reflecting surface with a radius of curvature R;
A

'
Si

So

In the paraxial approximation we have

=

x
x
x
, = , and = ,
so
si
R

and for the angles;

OAC: = [ + ( )] =
OAI: 2 = [ + ( )] =
2( ) =
2 = +
Hence, the focal length of the curved mirror is given by
1
2
1 1
= =
+
f
R s o si
In this case, the sign conventions are:
R;
so ;
si ;

+ve for concave surface (f +ve)

-ve for convex surface (f ve)
+ve for O in front of mirror
-ve for O behind mirror
+ve for I in front of mirror
-ve for I behind mirror
12

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

A mirror can also be described in the RTM method by modifying the RTM of a spherical
refracting surface;
1

n
RTM(refracting surface) =
2 n1 )
n R

0
n1

n2

Now, for a mirror the refracted ray is reflected which can be taken into account by setting
n2 = n1 (ray travels in opposite direction back into medium with index n1 ).
Thus,

1
RTM(mirror) = 2

NOTE: This RTM is similar to that for a lens,

1
RTM(lens) = 1

R
, except that the last element is negative to account for the fact that
2
2 is measured in the opposite sense to 1 after reflection.

since f =

ALSO: If the incident ray propagates in the +ve direction, then the reflected ray
propagates in the ve direction; i.e., if it propagates a distance d in the ve
direction the appropriate RTM is
1 d

0 1
Thus, RTMs preserve directionality.
BUT: Usually, one unfolds the reflection, replacing the mirror RTM with an
equivalent lens RTM, thus preserving the positive propagation direction
(e.g., for laser cavity stability analysis).

13

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Aberrations
Aberrations in optical systems fall into two categories: those associated with a violation
of the paraxial or Gaussian approximation (e.g. spherical aberration), and aberrations due
to dispersion in refracting systems (i.e. chromatic aberrations).
(i)

Non-chromatic aberrations
We will limit the discussion to axial aberrations (i.e. spherical aberration) and
ignore off-axis aberrations such as coma and astigmatism. Spherical aberration is
the focusing of rays with different x values (distances from the optical axis) to
different focii;

effect varies as x 4

14

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

There are several solutions to this problem;

a) Use aspheric optics expensive (although becoming less so)
b) Restrict x using an aperture stop (i.e. pupil)

c) Spherical aberration is reduced if refraction at each lens surface is minimized,

i.e.

better than

refraction at both surfaces

refraction at one surface

A lens which has equal refraction at both surfaces is called a best-form lens;

R1 < R2

15

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

d) Use a compound lens system to correct for aberration, e.g. the tessar;

different refractive indicies for the

elements of the system.

(ii)

Chromatic aberrations
From the lens-maker equation

1
1
1

= (n 1)

f
R1 R2
It is clear that f is a function of n and hence wavelength;
white
light

nblue > nred

blue light

red light

The solution is to use an achromatic doublet;

white
light
red light

blue light
n2

n1
_

Realistically, the doublet is only exact for two wavelengths for which the
compound lens is designed.
16

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Waveguides
_

an important topic in modern optics; e.g., fiber optics, semiconductor lasers ...

We will limit the discussion to a one-dimensional waveguide;

no

n1
2

t
i

n2

n3

For TIR at the n1 n2 interface we require n2 > n1 and 2 c , i.e.

n
sin 2 sin c = 1
n2

But, since 2 = 90 o t and sin 90 o t = cos t , we get

n
cos t 1
n2
Now, for the injection of light into the waveguide by end coupling, no sin i = n2 sin t ,
so that for confined propagation of light down the waveguide we require

n
n
sin i 2 1 1
no
n2

NOTE: This result is good for one and two dimensional planar waveguides and optical
fibers. It leads to the definition of the numerical aperture (NA) of a guide;
NAguide = no sin i =

n2 2 n12

where i is the acceptance angle.

17

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

A similar condition exists for the n2 n3 interface. The more restrictive

condition applies to the guide as a whole.
If n1 = n3 the guide is symmetric.

The allowed propagation modes of a planar waveguide can be found by finding the
conditions under which successive reflected waves interfere constructively with each
other;
n1
E

n2

n3

A
wavefronts

all indicated angles equal to 2

Including the phase changes upon TIR and A and B ( 3 and 1 respectively) we require
k (AB CD ) + 1 + 3 = 2m

m = 0,1, 2, ...
Notice that 1 and 3 are summed since the wavefront BD must be in-phase with
wavefront AC after TIR at A and B.

18

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Now,
AB =

d
.
cos 2

And,

CD = CB sin 2
with,
CB = EB EC = d tan 2

d
tan 2

1
CD = d sin 2 tan 2
tan 2

d
sin 2 2 cos 2 2
cos 2

Thus,

kd
1 sin 2 2 + cos 2 2 + 1 + 3 = 2m
cos 2

So the dispersion relation for the waveguide is

4n2 d
cos 2 + 1 + 3 = 2m
o
We will restrict the discussion to a symmetric guide (n3 = n1 ) so that 3 = 1 . The TIR
restriction on 2 is then
n
sin 2 1
n2

n
cos 2 1 1
n2

19

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

The maximum number of possible modes, mmax, is then given by

mmax

2dn 2

n
1 1
n2

n
2dn2
The quantity, V =
1 1
o
n2

+ 1

n
= n2 kd 1 1 ,
n2

is known as the normalized film thickness for a planar waveguide and the V number for
fibers. Hence,
m max

V + 1

N.B.
(i)

It is usually the transverse electric (TE) wave that propagates down the
waveguide, i.e. the polarization is ;

n1

n2

n1

The phase change for TIR is therefore equal to , which is given by

n cos 2 in1
r = 2
n2 cos 2 + in1
n sin 2
where = 2
n1

20

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Now, for complex numbers of the form

a 2 + b 2 e i

a + ib
=
a ib
where tan =

2 i

a + ib
we have
a ib

= e 2i

a +b e

b
a

So, since r = 1 for TIR and the analogy gives = 2 , we get

n1

tan =
2 n2 cos 2
2

n
2
1
sin 2
n

2
1
= 2 tan
cos 2

(ii)

The above discussion is valid for a general 2-D rectangular waveguide

(two orthogonal co-ordinates);

21

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

Solution for Symmetric Guide

_

Propagation vector formalism.

The ray or wavevector k for propagation along a waveguide can be decomposed into a
component along the guide (z-axis) and one perpendicular to the plane of the guide (xaxis);
x

n2k

... k =

= n2 k sin 2

= n2 k cos 2

2
here
o

2 = n2 2 k 2 2

Notice that this propagation vector formalism implies that for the allowed modes of a
symmetric waveguide

n1k < < n2 k

So that the effective propagation constant, , allows an effective guide index of
refraction, N, to be defined by
N =

the limits on N being n1 < N < n2

The introduction of and enables the phase shift upon TIR to be rewritten;

sin 2 2 n1
n

2
= 2 tan 1
cos 2

22

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

= 2 tan

Geometrical Optics

2 n 2k 2
1

2 n12 k 2 is equal to the decay constant of the evanescent wave, 1 , at the

n1 n2 interface since

The factor

2 n12 k 2 = kn1 = 1
The above definition allows the dispersion relation for a symmetric guide,

2n2 kd cos 2 + 2 = 2m ,
to be rewritten as

d 2 tan 1 1 = m

1

tan (d m ) = 1
2

But, tan ( A B ) =

tan A tan B
, hence after some algebra (!);
1 m tan A tan B

m even

m odd

d
1d = d tan

2
d
1d = d cot
2

(1)

with 1d also given by

12 = 2 n12 k 2 = n2 2 n12 k 2 2
(1d )2 = V 2 (d )2

(2)
since V = kd n 2 2 n12

The simultaneous solution of equations 1 and 2 give the allowed propagation modes of
the symmetric planar waveguide.

23

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

Geometrical Optics

N.B.
(i) The mode solutions are usually obtained graphically (or computationally),
e.g. for even m;

(1d)2 = V2 (d)2

1d

d
d tan

a circle!

V
m=0
m=2

m=4

Vc

(ii)

For a symmetric guide there will always be at least one propagating

kd
mode (m = 0) since kd tan intercepts the origin and V > 0
2
always.

(iii)

The guide has a high frequency cut-off when d = Vc , which

represents the case when 2 = c and TIR terminates;
n2 kd cos 2 = kd n2 2 n12

n
n2 1 sin 2 = n2 1 1
n2
2

n
sin 2 = 1 sin c
n2

24

Physics 425 Modern Optics

Lecture Notes W. Andreas Schroeder

(iv)

Geometrical Optics

A single-mode guide is one in which only the m = 0 mode

propagates, i.e. Vc < . For a particular wavelength, o , this places
restrictions on the width, d, of the guide;
Vc <
kd n2 2 n12 =

d<

2d
n2 2 n12 <
o

o
2 n 2 2 n12

This formula is also approximately valid for optical fibers (d = core

diameter);

core

n2
cladding

n1

25