7. Nonlinear Optics

Why do nonlinear-optical effects occur?

Maxwell's equations in a medium

Nonlinear-optical media

Second-harmonic generation

Sum- and difference-frequency generation

Higher-order nonlinear optics

The Slowly Varying Envelope Approximation

Phase-matching and Conservation laws for photons

Nonlinear optics isn’t something you see everyday.

Sending infrared light into a crystal yielded this display of green light(second-harmonic generation):

Nonlinear optics allows us to change the color of a light beam, to change its shape in space and time, and tocreate ultrashort laser pulses.

Why don't we see nonlinear opticaleffects in our daily life?

1. Intensities of daily life are too weak.2. Normal light sources are incoherent.3. The occasional crystal we see has the wrong symmetry (for SHG).4. “Phase-matching” is required, and it doesn't usually happen on its own.

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 interfereswith the original light wave.

We can also imagine thisprocess in terms of the molecular energy levels,using arrows for thephoton energies:

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.

Nonlinear optics is analogous to nonlinear electronics, which we can observe easily.

Sending a high-volume sine-wave (“pure frequency”) signal into cheap speakers yields a truncated output signal, more of a square wave than a sine. This square wave has higher frequencies.

We hear this as distortion.

Nonlinear optics and anharmonic oscillators

For weak fields, motion is harmonic, and linear optics prevails.For strong fields (i.e., lasers), anharmonic motion occurs, and higherharmonics occur, both in the motion and the light emission.

Another way to look at nonlinear optics is that the potential of the electron or nucleus (in a molecule) is not a simple harmonic potential.

Example: vibrational motion:

Maxwell's Equations in a Medium

The induced polarization, P, contains the effect of the medium:

020

0

10

t

c t t

BE E

E PB B

2 2 2

02 2 2 20

1

z c t t

E E P

Sine waves of all frequencies are solutions to the wave equation; it’s the polarization that tells which frequencies will occur.

The polarization is the driving term for the solution to this equation.

These equations reduce to the (scalar) wave equation:

“Inhomogeneous Wave Equation”

Solving the wave equation in the presence of linear induced polarization

For low irradiances, the polarization is proportional to the incident field:

0 P E

In this simple (and most common) case, the wave equation becomes:2 2 2

2 2 2 2 20 0

1 1

z c t c t

E E E

0( , ) cos( )z t E t k z EThis equation has the solution:

The induced polarization only changes the refractive index. Dull. If only the polarization contained other frequencies…

where = c k and c = c0 /n and n = (1+)1/2

Using the fact that:2

0 0 01/ c

2 2

2 2 20

10

z c t

E E

Simplifying:

Maxwell's Equations in a Nonlinear Medium

Nonlinear optics is what happens when the polarization is the resultof higher-order (nonlinear!) terms in the field:

What are the effects of such nonlinear terms? Consider the second-order term:

2 = 2nd harmonic!

Harmonic generation is one of many exotic effects that can arise!

(1) (2) 2 (3) 30 ... E E E E

*

*2222

( ) ,

exp(2

exp(

) exp

)

( ) 2

e p

( )

x ( )

2

t

t

E i t E i t

E i t E i tE

Since E

E

Sum- and difference-frequency generation

Suppose there are two different-color beams present:

So:

*1 1 1 1

*2 2 2 2exp( ) e exp( ) exp( ) x (( )) pE i t E i t E i t E i tt E

* *1 2 1 2

2

2

2 *21 1 1 1

2 *22 2 2

* *1 2 1 2 1 2

1 2 1

1

2

1 2

2

2

2

( )

exp(2 ) e

2 exp( ) 2 exp

2 exp( ) 2 exp(

x

exp

( )

2

(2

p( 2

)

) xp

2

e ( )

)

2

E i t E

E E i t E E i

E i t E i t

E E i t E E

E

t

i

t

tt

i

E

E 2nd-harmonic gen

2nd-harmonic gen

Sum-freq gen

Diff-freq gen

dc rectification

Complicated nonlinear-optical effects can occur.

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. Also, if the photon energies coincide with the medium’s energy levels as above the effect will be stronger.

Nonlinear-optical processesare often referred to as:

"N-wave-mixing processes"

where N is the number ofphotons involved (including the emitted one).

Emitted-lightfrequency

Induced polarization for nonlinear optical effects

Arrows pointing upward correspond to

absorbed photons and contribute a factor

of their field, Ei; arrows pointing downward

correspond to emitted photons and contri-

bute a factor the complex conjugate of

their field:

(5) *0 1 2 3 4 5E E E E E P

Solving the wave equation in nonlinear optics

2 2 2

02 2 2 2

1

z c t t

E E P

Recall the inhomogeneous wave equation:

Because it’s second-order in both space and time, and P is a nonlinear function of E , we can’t easily solve this equation. Indeed, nonlinear differential equations are really hard.

We’ll have to make approximations…

Take into account the linear polarization by replacing c0 with c.

Separation-of-frequencies approximation

The total E-field will contain several nearly discrete frequencies,1, 2, etc.

So we’ll write separate wave equations for each frequency, considering only the induced polarization at the given frequency:

2 2 2

02 2 2 2

1

z c t t

1 1 1E E P where E1 and P1 are the E-field and polarization at frequency 1.

2 2 2

02 2 2 2

1

z c t t

2 2 2E E P where E2 and P2 are the E-field and polarization at frequency 2.

etc.

This will be a reasonable approximation even for relatively broadband ultrashort pulses

The non-depletion assumption

We’ll also assume that the nonlinear-optical effect is weak, so we can assume that the fields at the input frequencies won’t change much. This assumption is called “non-depletion.”

As a result, we need only consider the wave equation for the field and polarization oscillating at the new “signal” frequency, 0.

2 2 2

02 2 2 2

1

z c t t

0 0 0E E P

where E0 and P0 are the E-field and polarization at frequency 0.

We’ll write the pulse E-field as a product of an envelope and complex exponential: E(z,t) = E (z,t) exp[i(t – kz)]

We'll assume that the new pulse envelope won’t change too rapidly. This is the Slowly Varying Amplitude Approximation.

If d is the length scale for variation of the envelope, SVEA says: d >>

The Slowly Varying Envelope Approximation (SVEA)

2

020

/ /~ 20 0 0 0E E z E z E

kz d z

~ 20 0 00 0

E E Ek E

z d

22

20 0

0 0 0

E Ek k E

z z

Comparing E0 and its derivatives:

We’ll do the same in time:

If is the time scale for variation of the envelope, SVEA says: >>

where is one optical period,

The Slowly Varying Envelope Approximation (continued)

2

020

/ /~ 20 0 0 0E E t E t E

t T t

0

~ 20 0 00 0

E E EE

t T

22

20 0

0 0 0

E EE

t t

Comparing E0 and its time derivatives:

And we’ll do the same for the polarization:

P (z,t) = P (z,t) exp[i(t – kz)]

If is the time scale for variation of the envelope, SVEA says: >>

where is one optical period,

The Slowly Varying Envelope Approximation (continued)

2

020

/ /~ 20 0 0 0P P t P t P

t T t

0

~ 20 0 00 0

P P PP

t T

22

20 0

0 0 0

P PP

t t

Comparing P0 and its time derivatives:

Computing the derivatives:

The Slowly Varying Envelope Approximation (continued)

222

2 22 exp[ ( )]0 0

0 0 0 0 0

E Eik k E i t k z

z z z

0E

222

2 22 exp[ ( )]0 0

0 0 0 0 0

E Ei E i t k z

t t t

0E

222

2 22 exp[ ( )]0 0

0 0 0 0 0

P Pi P i t k z

t t t

0P

Neglect all 2nd derivatives of envelopes with respect to z and t.Also, neglect the 1st derivative of the polarization envelope (it’s small compared to the 02P0 term). We must keep E0’s first derivatives, as we’ll see in the next slide…

xxx x

exp[ ( )]00 0 0 0

Eik E i t k z

z z

0E

Now, because k0 = 0 / c, the last two bracketed terms cancel.And we can cancel the complex exponentials, leaving:

The Slowly Varying Envelope Approximation

Substituting the remaining derivatives into the inhomogeneouswave equation for the signal field at 0:

22 20 0 0 0

0 0 0 0 0 0 0 0 0 0 02 2

22 exp[ ( )] exp[ ( )]

E i Eik k E E i t k z P i t k z

z c t c

20 0 00 0 0 02

22

E i Eik P

z c t

Slowly VaryingEnvelope Approximation

20 0 0 0

00

1

2

E Ei P

z c t k

Dividing by 2ik0:

2 2 2

02 2 2 2

1

z c t t

0 0 0E E P

x x

Including dispersion in the SVEA

20 0 0 0

00

1

v 2g

E Ei P

z t k

We can include dispersion by Fourier-transforming, expanding k() to first order in , and transforming back. This replaces c with vg:

We can include GVD also expanding to 2nd order, yielding:

2220 0 0 0 0

02 20

1

v 2 2g

E E Ei d ki P

z t d t k

We can understand most nonlinear-optical effects best by neglecting GVD, so we will, but this extra term can become important for very very short (i.e., very broadband) pulses.

20 0 0

0v 02

Ei P

z k

Transforming to a moving co-ordinate system

Define a moving co-ordinate system: zv = z

tv = t – z / vg

20 0 0 0 0

0v v v 0

1 1

v v 2g g

E E Ei P

z t t k

0 0 v 0 v

v v

E E z E t

z z z t z

Transforming the derivatives:

0 0 v 0 v

v v

E E z E t

t z t t t

0 0 0

v v

1

vg

E E E

z z t

0 0

v

0E E

t t

The SVEA becomes:

Canceling terms, the SVEA becomes:

We’ll drop the sub-script (v) to simplify our equations.

The time deriva-tives cancel!x x

Integrating the SVEA

Usually, P0 = P0 (z,t), and even this simple equation can be difficult to solve (integrate). For now, we’ll just assume that P0 is a constant, and the integration becomes trivial:

20 0 0

002

Ei P

z k

20 0

0 00

( , )2

E z t i P zk

And the field amplitude grows linearly with distance.

The irradiance (intensity) then grows quadratically with distance.

We choose 0 to be the sum of the input ’s:

The k-vector of light at this frequency is:

But the k-vector of the induced polarization is:

Unfortunately, kp may not be the same as k0!

So we can’t necessarily cancel the exp(-ikz)’s…

But the E-field and polarization k-vectors aren’t necessarily equal.

1 2 3 4 5pk k k k k k

0 1 2 3 4 5

0 0 0 0 0/ ( ) /k c n c

Phase-Matching

That kp may not be the same as k0 is the all-important effect of phase-matching. It must be considered in all nonlinear-optical problems.

If the k’s don’t match, the induced polarization and the generated electric field will drift in and out of phase.

20 0 0

00

exp( )2

Ei P i k z

z k

where:0Pk k k

0 0 0( , ) ( , ) exp[ ( )]z t E z t i t k z 0E0 0( , ) ( , ) exp[ ( )]z t P z t i t z 0P Pk

Integrating the SVEA in this case over the length of the medium (L) yields:

The SVEA becomes:

20 0

0 00 0

exp( )( , )

2

Li k z

E L t i Pk i k

Phase-Matching (continued)2

0 00 0

0 0

exp( )( , )

2

Li k z

E L t i Pk i k

20 0

00

exp( ) 1

2

i k Li P

k i k

20 0

00

exp( / 2) exp( / 2)exp( / 2)

2

i k L i k Li P i k L

k i k

20 0

00

exp( / 2) exp( / 2)exp( / 2)

2

i k L i k Li P i k L

k i k

2

0 00

0

sin( / 2)exp( / 2)

k Li P i k L

k k

( , ) sin( / 2) /E L t k L k So:

L

I

k large

k small

2 2( , ) sin ( / 2) /I L t k L k

Sinusoidal Dependence of SHG Intensity on Length

Large k Small k

Notice how the intensity is created as the beam passes through the crystal, but, if k isn’t zero, newly created light is out of phase with previously created light, causing cancellation.

The Ubiquitous Sinc2(kL/2)

20 0

00

sin( / 2)exp( / 2)

2 / 2

k Li P i k L L

k kL

0 0 0 exp( / 2) sinc( / 2)2

ic P i k L L k L

E0(L, t) P0 L sinc (k L / 2)

Phase Mismatch almost always yields a sinc2(k L / 2) dependence.

20 0

0 00

sin( / 2)( , ) exp( / 2)

k LE L t i P i k L

k k

Recall that:

I(L, t) P0

2L2 sinc2 (k L / 2)

Multiplying and dividing by L/2:

I

k

More of the Ubiquitous Sinc2(kL/2)

The irradiance (intensity):

To maximize the irradiance,we must try to set k = 0. This is phase-matching.

k

E0

E0(L, t) P0 L sinc (k L / 2)

I(L, t) P0

2L2 sinc2 (k L / 2)

The field:

Phase-matching = Conservation laws for photons in nonlinear optics

Adding the frequencies:

is the same as energy conservation if we multiply both sides by h-bar:

Adding the k’s conserves momentum:

So phase-matching is equivalent to conservation of energy and momentum!

1 2 3 4 5 0

1 2 3 4 5 0k k k k k k

1 2 3 4 5 0

1 2 3 4 5 0k k k k k k