12.

Nonlinear optics I
What are nonlinear-optical effects and why do
they occur?
Maxwell's equations in a medium
Nonlinear-optical media
Second-harmonic generation
Conservation laws for photons ("Phase-
matching")
Quasi-phase-matching

1
Nonlinear Optics can produce
many exotic effects.

Sending infrared light into a

crystal yielded this display of
green light:

Nonlinear optics allows us to

change the color of a light beam,
to change its shape in space and
time, to switch telecommunica-
tions systems, and to create the
shortest events ever made by
humans.

2
Why do nonlinear-optical effects occur?
Recall that, in normal linear optics, a light wave acts on a molecule,
which vibrates and then emits its own light wave that interferes with
the original light wave.

We can also imagine this

process in terms of the
molecular energy levels,
using arrows for the
photon energies:
3
Why do nonlinear-optical effects
occur? (continued)
Now, suppose the irradiance is high enough that many molecules
are excited to the higher-energy state. This state can then act as
the lower level for additional excitation. This yields vibrations at all
frequencies corresponding to all energy differences between
populated states.

4
Reminder: Maxwell's Equations in a Medium
The induced polarization, P, contains the effect of the medium.
The inhomogeneous wave equation (in one dimension):
2 E 1 2 E 2 P
 2 2  0 2
x 2
c0 t dt
The polarization is usually proportional to the electric field:
 
P  0 E  = unitless proportionality constant
Recall, for example, in the forced oscillator model, we found:
Ne 2 E t 
P t  
m 0 2   2  i
Then, the wave equation becomes:
2 E 1 2 E 2 E
 2 2   0 0  2
x 2
c0 t dt
 2 E 1     2 E 1
  0 0
or   0 since 2
x 2 c0
2
t 2 c0 5
Reminder: Maxwell's Equations in a Medium
 2 E 1     2 E
 0
x 2 c0
2
t 2
But this is the same equation as the usual homogeneous equation,
1 1   
if we define a new constant c:
2
 2
c c0
And, we call the quantity 1  the “refractive index”.

So, we can describe light in a medium just like light in vacuum, as

long as we take into account the refractive index correction.

But this only worked because P was proportional to E…

What if it isn’t? Then P is a non-linear function of E!

6
Maxwell's Equations in a Nonlinear Medium
Nonlinear optics is what happens when the polarization is the result
of higher-order terms in the field:

P   0   (1) E   (2) E 2   (3) E 3  ...

 PLinear  Pnon linear
Then the wave equation must look like this:
 2 E n2  2 E  2 Pnon linear
 2 2  0
x 2
c t dt 2
The linear term can be treated in the same way as before, giving
rise to the refractive index. But the non-linear term is a problem…
 2 E n2  2 E  2
 2

2   2  
     (2)
E 2
    (3)
E 3

x c t
2 2 2 0 0 0 0
dt dt
Usually, (2), (3), etc., are very small and can be ignored.
7
But not if E is big…
The effects of the non-linear terms

What sort of effect does this non-linear term have?

If we write the field as: E (t )  E0 exp(it )  E0* exp(it )

E (t )  E exp(2it )  2 E0  E0*2 exp(2it )
2 2 2
then 0

terms that vary at a new frequency, the 2nd harmonic, 2!

Nonlinearity can lead to the generation of new frequency components.

This can be extremely useful:

Frequency doubling crystal:
1064 nm 532 nm

8
 Sum and difference frequency generation
Suppose there are two different-color beams present, not just one:

E (t )  E1 exp(i1t )  E1* exp(i1t )  E2 exp(i2t )  E2* exp(i2t )

Then E(t)2 has 16 terms:

E(t )2  E12 exp(2i1t )  E1*2 exp(2i1t ) 2nd harmonic of 1

 E22 exp(2i2t )  E2*2 exp(2i2t ) 2nd harmonic of 2

 2E1E2 exp(i 1  2  t )  2E1* E2* exp(i 1  2  t ) sum frequency

 2E1E2 exp(i 1  2  t )  2E1* E2* exp(i 1  2  t ) difference frequency

 2 E1  2 E2
2 2
zero frequency - known as “optical rectification”

This is an awful lot of processes - do they all occur simultaneously?

Which one dominates (if any)? What determines the efficiency?
9
Complicated nonlinear-optical effects can occur.

Nonlinear-optical processes
are often referred to as:

"N-wave-mixing processes"

where N is the number of

photons involved (including
the emitted one).
Emitted-light
photon energy

This cartoon illustrates a 6-wave mixing process.

It would involve the (5) term in the wave equation.

The more photons (i.e., the higher the order) the weaker the effect,
however. Very-high-order effects can be seen, but they require
very high irradiance, since usually (2) > (3) > (4) > (5) …
10
Conservation laws for photons in nonlinear optics
Energy must be conserved. Recall that
the energy of a photon is  . Thus:

1   2   3   4   5   0
Photon momentum must also beconserved.
The momentum of a photon is k, so:
     
k1  k2  k3  k4  k5  k0
  2 n n0
But k0 is related to 0: k0  
 c
So k0 may not correspond to a light
wave at frequency 0!
Satisfying these two relations simultaneously
is called "phase-matching."
Usually, only one (or zero) of the many possible N-wave mixing
11
processes can be phase-matched at a time.
 Phase-matching: an example
Consider the 2nd harmonic generation process:

Energy conservation requires:  in  out

nonlinear
material
  
Momentum conservation requires: k    k    k  2 
2 red photons 1 blue photon

 2
2n    n  2  
c c
n(2 )  n( )
Refractive index

n  2 

n   Unfortunately, dispersion prevents

this from ever happening!
 Frequency
2 12
Phase-matching Second-Harmonic
Generation using birefringence
Birefringent materials have different refractive indices for different
polarizations: the “Ordinary” and “Extraordinary” refractive indices!

Using this, we can satisfy the

phase-matching condition.

Refractive index
ne
For example:
Use the extraordinary polarization no
for  and the ordinary for 2:

 2
no (2 )  ne ( ) Frequency

ne depends on propagation angle, so by rotating the

birefringent crystal, we can tune the condition precisely by
moving the red curve up and down relative to the blue curve. 13
Light created in real crystals
Far from
phase-matching: SHG crystal

Input beam Output beam

Closer to
phase-matching: SHG crystal

Input beam Output beam

Note that SH beam is brighter as phase-matching is achieved. 14

Second-Harmonic Generation

SHG KDP crystals at Lawrence

Livermore National Laboratory

These crystals convert as much

as 80% of the input light to its
second harmonic. Then
additional crystals produce the
third harmonic with similar
efficiency!

15
Difference-Frequency Generation: Optical
Parametric Generation, Amplification, Oscillation
Difference-frequency generation takes many useful forms.

1 1
2 = 3  1 "signal"
3
3 2
"idler"

Parametric Down-Conversion By convention:

Optical Parametric
signal  idler
(Difference-frequency generation) Generation (OPG)

1 1 1
3
3 2 2
mirror mirror
Optical Parametric Optical Parametric
Amplification (OPA) Oscillation (OPO)
16
All of these are (2) processes (three-wave mixing).
Another 2nd-order process: Electro-optics
Applying a voltage to a crystal changes its refractive indices and
introduces birefringence. In a sense, this is sum-frequency
generation with a beam of zero frequency (but not zero field!).

V Polarizer
If V = 0, the pulse
polarization doesn’t
change.
Pockels cell
If V = Vp, the pulse
polarization switches to its
orthogonal state.

The Pockels effect can be described as a (2) nonlinear optical

interaction, where E2 E() E( = 0). Sum frequency is at  + 0 = .
17
The wave equation with nonlinearity

We have derived the wave equation in a medium, for

the situation where the polarization is non-linear in E:

 2 E n2  2 E  2 P NL
 2 2  0
x 2
c t dt 2
linear optics

where P NL   0   (2) E 2   (3) E 3  ...

Usually,  (2)   (3)

In these cases, we neglect the third (and higher) orders.

A good example: second harmonic generation

18
Second Harmonic Generation: SHG


(2)


z=0 z=L

In this process, we imagine that one laser (at frequency ) is used

to illuminate a nonlinear medium.

intensity of the fundamental

As this field propagates
through the medium, its intensity
intensity will be depleted
and the intensity of the 2nd intensity of the 2nd harmonic
harmonic wave (initially
zero) will grow.
propagation distance z
19
 Describing the 2nd harmonic wave
We are interested in the behavior of the field that oscillates at 2;
that is, the 2nd harmonic. We can assume that this field is of the
form:

E 2   z , t   A2   z  e
 i  2  t
ik 2  z
e  c .c .
where we require that the amplitude A2(z) is slowly varying, and
also that it vanishes at the input facet of the nonlinear medium:

A2   z  0   0

Furthermore, the wave vector of this wave is related to the

refractive index of the nonlinear medium at frequency 2:

2
k 2  n  2 
c
Our goal is to determine A2(z). 20
What equation must the 2nd harmonic obey?
The 2nd harmonic wave must obey the wave equation, of course.

 E2  n  2    2 E2
2
2
 2 P (2)
   0
z 2
 c   t 2
dt 2

As we have seen, the 2nd-order polarization results from the

field at frequency  - the fundamental. Putting in the spatial
dependence explicitly:
2 
 t   2 0  2 
E e 
 i t  ik  z 2
P 0

this is the k of the

the amplitude of the incident field incident field:
(the one at frequency ) 
k   n  
c
P  2   t   2 0  2  E 0 2 e 
i 2 k z  2  t 

21
 Plugging in to the wave equation…
Plug our assumed forms for E(z,t) and P(2), to find:

  2 A2 A2 2  2 
2
 i  k 2  z  2 t 
  2 jk2  k2 A2  n 
2
 A2  e
 z z  c 
2

2  2
 2 
2

E0 2 e 
i 2 k z  2t 

c2
Slowly Varying Envelope Approximation (SVEA):

 2 A2 A2
 k2
z 2
z

So we neglect the second derivative of A2.

22
Solving the wave equation in second order

The nonlinear wave equation becomes:

 2
A2 8  2
E0  z  e e
2 i 2 k z  ik2  z
2ik2 
z c 2

Now, we could find a similar first-order differential equation for E0,

and then solve the two coupled equations.
But, instead of doing that, let’s see if we can gain some physical
insight by making another simplifying assumption:
Assume: The incident field is not significantly depleted
by the conversion process. That is, E0 does
not decrease very much with increasing z.
E0 is independent of z.
In this case, we can easily integrate both sides of this equation.
23
 Integrate both sides
A2 4i   2 2 2 i2 k z ' k2 z '
z z

0 z ' dz '  k2 c 2 E0 0 e dz '

This is just A2(z).

Define the 'phase mismatch' k  2k  k2

We can do the integral on the right side: Note, this is just:
2 2
z
1 2 n  n2
 dz '  eik  z  1   2
ik  z '
e
j k 4
 n  n2 
0


Thus we’ve arrived at a result!

exp ikz   1
A2  z   E0  2

k 24
 The solution
The intensity of the second harmonic radiation is proportional to | A2 |2.

I 2  z   A2  z   I 0
2 2 sin 2
 k  z 2 
 k 
2

 I0 z
2 2 sin 2
  where   k  z 2
 2
1
The intensity of the 2nd 0.8  sin 2  
harmonic is proportional to the  2 
0.6   
square of the intensity of the
fundamental. 0.4

It also depends sensitively on 0.2

the product of k and z. 0
-10 -8 -6 -4 -2 0 2 4 6 8 10
 = dimensionless phase mismatch  25
 Phase matching for a (2) process

I 2  z   I 0 z
2 2 sin 2
 k  z 2 
 k  z 2 
2
To summarize:
 SVEA and zero-depletion approximations give lowest order solution.
 Intensity of SHG radiation is proportional to the square of the input intensity.
 Intensity of SHG radiation grows quadratically with propagation distance.
 Intensity of SHG is very sensitive to phase mismatch - maximum when k = 0
1
SHG intensity is most
0.8  sin 2   efficient for || < 1
 2 
0.6    If  = 1, then sin2/2 = 0.71.
2
0.4
|| < 1 corresponds to k 
L
0.2
If the SHG medium is too thick for a
0
-10 -8 -6 -4 -2 0 2 4 6 8 10 given k, conversion efficiency suffers.
 26
 What does phase matching mean?
When k = 0, this means that n() = n(2). The phase velocity of the
fundamental and 2nd harmonic are equal.  = 2 2.

phase-matched: amplitude

propagation distance

When k is not zero, the phase velocity of the fundamental and 2nd
harmonic are different, and   2 2. As z increases, the 2nd
harmonic wave gets increasingly out of phase with the fundamental.
This is why
amplitude

k L << 1 is
not phase-matched: the important
condition to
satisfy.
27
propagation distance
Materials and configurations for (2) NLO
There are a number of materials commonly used for SHG or
other frequency conversion effects based on (2).

• KDP: potassium di-hydrogen phosphate

• BBO: beta-barium borate
• LiNbO3: lithium niobate
• etc.

A non-linear crystal inside the

laser cavity to produce UV light:
LiNbO3 crystals

This is a “VECSEL”: a
“vertical external cavity
surface emitting laser”

28
 optic
SHG illustration axis
k
vector
Example of matching n() and n(2) in a nonlinear medium: 
e-ray
refractive indices for BBO
1.7
o-ray
1.6545 no
For  = 1064 nm, at this angle,
refractive index

ne for  = 22.78º no() = ne(2) and thus k = 0.

1.6
What if we changed the angle
slightly? For example: 23º.
ne for  = 0º
Then no() is unchanged. But
1.5 ne(2) = 1.6542. And thus:
4
532 nm 1064 nm
0.4 0.8 1.2 1.6 k   n  n2   4150 m-1
wavelength (m) 
sin 2 
For a crystal of thickness = 1 mm:   k  z 2  2.1 and so  0.18
 2
29
What if the phase matching is not perfect?
sin 2  k  z 2 
intensity of the fundamental (decreasing?)
I 2  z   I 0 z
2 2

 k  z 2 
2
intensity

intensity of the 2nd harmonic (increasing quadratically if k = 0)

If the phase mismatch is not

propagation distance z precisely zero, then how does the
second harmonic intensity behave?

The SHG intensity oscillates as a function of propagation distance:

k = 0 (quadratic)
2 intensity

decreasing peak
SHG crystal
signal with
increasing k

propagation distance 30
Another way to boost the SHG efficiency
Why does the signal oscillate?
SHG crystal
If phase matching condition is not perfect, then
after a certain length (called the ‘coherence
length’ Lcoh), the fundamental and 2nd harmonic
walk out of phase with each other.

At that point, the process reverses itself, and the fundamental grows
while the 2 beam diminishes. This process then oscillates.

What if, at z = Lcoh, we could flip the sign of

(2)? This would change the phase of E2 by
. Instead of cancelling out as it propagates
beyond Lcoh, E2 would be further enhanced.

In some cases, we can control the sign of LiNbO3

(2) by changing the crystal structure.
31
 Quasi-phase matching
Flipping the sign of (2) once each coherence length is known as
“quasi-phase matching.” It has recently become a critically important
method for efficient second harmonic generation.
(Length)2 quasi-
phase phase The process of fabricating a
matching matching material where the sign of
(2) flips back and forth is
no phase known as “periodic poling”.
matching

A photo of PPLN:
periodically poled
lithium niobate
32
 SHG at a surface
Another method of minimizing  = k z / 2 : use a very small value of z.
For example, at a surface or an interface.

“surface second harmonic generation”

- a very sensitive probe of surfaces
 = 1 + 2 (but very weak!)

Applications:
• measuring the orientation of molecules at a liquid surface
• studying buried interfaces, e.g., silicon/insulator
33