Standing Waves, Beats,

and Group Velocity

Superposition again

Standing waves: the

sum of two oppositely
traveling waves
Beats: the sum of
two different
frequencies

Group velocity: the speed of information

Going faster than light...

Group-velocity dispersion
Superposition allows waves to pass
through each other.
Otherwise they'd get tacked up
while overlapping.
 Now well add waves with different
complex exponentials.
It's easy to add waves with the same complex exponentials:

E ( x, t ) E1 exp i(kz t ) E2 exp i(kz t ) E3 exp i( kz t )

( E1 E2 E3 )exp i (kz t )

where all initial phases are lumped into E1, E2, and E3.
Note
the plus
But sometimes the complex exponentials will be different. sign!

E( x, t ) E1 exp i(k1z 1t ) E2 exp i(k2 z 2t ) E3 exp i(k3 z 3t )

For such cases, well have to add the terms the old-fashioned way.
Adding waves of the same frequency, but
opposite direction, yields a standing wave.
Waves propagating in opposite directionscounter-propagating waves:

E ( z, t ) E0 exp i (kz t ) E0 exp i (kz t )

E0 exp(ikz )[exp(i t ) exp(i t )]
2 E0 exp(ikz )cos( t )

Taking the real part of the field complex amplitude:

E ( z, t ) 2 E0 cos(kz )cos(t ) or 2 E0 sin(kz)cos(t )

depending on whether E0 is real or imaginary.

Standing waves occur inside lasers, where beams bounce back and
forth.
 l
A Standing Wave
Nodes
The points where
the amplitude is
always zero are
called nodes. E
The points where
the amplitude
oscillates
maximally are z
called anti-
nodes.
Note that the
nodes and anti-
nodes are each
separated by l/2.
Anti-nodes
A Standing Wave: Experiment
3.9 GHz microwaves

Mirror

Input beam

Reflected beam

The same effect

occurs in lasers.

Note the node at the mirror at left, where there is a boundary condition
that the electric field = 0.
 Mirror Mirror
Standing Waves in a Laser
A laser has two mirrors, each with the
same boundary condition: E = 0. z
not cos(kz) 0 L
The left mirror means using sin(kz):
m
E ( z, t ) 2 E0 sin(kz )cos(t )

But the right mirror means that

sin(kL) = 0. k can take many values:

kmL = mp or (2p/lm)L = mp
m(lm/2) = L or lm = 2L/m

In other words, L = an integral

number (m) of half wavelengths.
0 z L

And the possible frequencies are: nm = c/lm = mc/2L

 Beams Crossing at an Angle k1 k cos z
x k sin x

Fringes
k2
k2 k cos z
z
k sin x
k1 L

E1 E2 E0 exp i(kz cos kx sin t ) exp i(kz cos kx sin t )

E0 exp i(kz cos t )exp(ikx sin ) exp(ikx sin )

E( x, z, t ) 2 E0 expi(kz cos t ) cos(kx sin )

I ( x, t ) 4 I0 cos2 (kx sin ) 2 I0 1 cos(2kx sin )

Fringe spacing: L = 2p/(2ksin) or L = l/(2sin)

Laser Beams Crossing at an Angle
Finite-size (laser) beams yield fringes only where the beams overlap.

Fringe spacing Interference fringes

L = l/(2sin)
x
k1

k2
Pulsed Waves Crossing at an Angle
Big angle: small fringes.
Small angle: big fringes.
The fringe spacing, L: L
Large angle:
L l / (2sin )
2
As the angle decreases to
zero, the fringes become
larger and larger, until finally,
at = 0, the intensity pattern
becomes constant.
Small angle:
L
L = 0.1 mm is about the minimum
fringe spacing you can see: 2
sin l / (2L)
0.5m / 200m 1/ 400 rad 0.15
very small!
Two Point Sources Emitting Spherical
or Circular Waves

Different separations. Note the different node patterns.

The farther apart the sources, the smaller the node spacing in angle.
When two cosines (or sines) of different
frequency combine, the result is beats.
E(t ) E0 exp(i1t ) E0 exp(i2t )

1 2 1 2 1 ave
Let: ave and
2 2 2 ave

E(t ) E0 exp i(ave t t ) E0 exp i(avet t )

E0 exp(iave t )[exp(it ) exp(it )]
2 E0 exp(iave t )cos( t )
E (t ) 2 E0 cos(ave t )cos( t ) taking E0 to be real.

Adding cosine waves of two different frequencies yields the product

of a rapidly varying cosine (ave) and a slowly varying cosine ().
 When two cosines or sines of different
frequency interfere, the result is beats.
In phase Out of phase In phase Out of phase In phase
Indiv-
idual
waves

Sum

Envel- 2p/
ope

Inten-
sity
time
When two light waves of different frequency
interfere, they also produce beats.
E( z, t ) E0 exp i(k1 z 1t ) E0 exp i(k2 z 2t )

k1 k2 k1 k2
Let: kave and k
2 2
2
and: ave 1 and 1 2
2 2
So:
E( z, t ) E0 exp i(kave z kz avet t ) E0 exp i(kave z kz avet t )

E0 exp i(kave z ave t )exp i(kz t ) exp[i(kz t )]

2 E0 exp i(kave z avet ) cos(kz t )

So: E ( z, t ) 2 E0 cos(kave z avet ) cos(kz t ) taking E0 to be real.

 Traveling-Wave Beats

In phase Out of phase In phase Out of phase In phase

Indiv-
idual
waves

Sum

Envel-
ope

Inten-
sity
z
 Its usually very difficult to see optical
Seeing Beats: beats because they occur on a time
Pulses scale thats too fast to detect.
This is why we said earlier that beams of
However, a sum of different colors dont interfere, and we
many frequencies will only see the average intensity.
yield a train of well-
separated pulses: Call the resulting pulse shape APulse.

In phase Out of phase

Indiv-
idual
waves
E ( z, t ) 2 E0 cos(kave z ave t ) APulse (kmin z mint )
Sum
Pulse separation: 2p/min
2p/max
I ( z, t ) APulse (kmin z mint )
2
Irrad-
iance
time
Group Velocity
Light-wave beats (continued):

E (z,t) cos(kavezavet) APulse(kminzmint)

Carrier wave
This is a rapidly oscillating wave: cos(kavez avet)
Pulse
with a slowly varying amplitude: APulse(kminz mint) Amplitude

The phase velocity comes from the rapidly varying part: v = ave / kave

What about the other velocitythe velocity of the pulse amplitude?

Define the group velocity: vg min /kmin

Taking the continuous limit,

we define the group velocity as: vg d /dk
The group velocity is the velocity of
the envelope or irradiance.

The carrier wave propagates at the phase velocity.

And the pulse envelope propagates at the group velocity:

E(t ) APulse ( z v g t ) exp[ik ( z v t )]

Or, equivalently, because the irradiance is proportional to the

square of APulse, it propagates at the group velocity:

E (t ) I Pulse ( z v g t ) exp[ik ( z v t )]
Group velocity is not equal to phase velocity
if the medium is dispersive (i.e., n varies).
Evaluate the group velocity for the two-frequency case:

c0 k1 c0 k2
vg
k n1k1 n2 k2

where k1 and k2 are the k-vector magnitudes in vacuum.

c0 k1 k2 c0
If n1 n2 n, vg v phase velocity
n k1 k2 n

If n1 n2 , v g v c0 / n
Phase and Group The phase velocity, v, is that of the
high-frequency oscillations. The group
Velocities velocity, vg, is that of the pulse envelope.

In vacuum Unrealistic

Most
common Possible
case

Unrealistic
Rare
Calculating the Group Velocity, vg d /dk
Now, is the same in or out of the medium, but k = k0 n, where k0 is
the k-vector in vacuum, and n depends on the medium.
So it's easier to think of as the independent variable:

v g 1/ dk / d using the
product rule

Using k = n() / c0, calculate: dk/d = (n + dn/d) / c0

So:
dn c0 dn
v g c0 / n / 1
d n n d

Thus vg = v when dn/d = 0, such as in vacuum.

Otherwise, since n usually increases with , dn/d > 0, and:
vg < v
Calculating Group Velocity vs. Wavelength
We more often think of the refractive index in terms of wavelength, so
let's write the group velocity in terms of the vacuum wavelength l0.
dn dn d l0
Use the chain rule:
d d l0 d
d l0 2c0 2c0 l02
Now, l0 = 2pc0/, so:
d 2
(2c0 / l0 ) 2c0
2

dn
Recalling that: v g c0 / n
d
2c0 dn l02
we have: v g c0 / n
l0 d l0 2c0

dn c0 l0 dn
v g c0 / n l0 / 1
d l0
n n d l0
Can vg > v or even c0 in absorbing regions?
For normal dispersion, dn/d is positive. vg = c0 / (n + dn/d)
So vg < v for these frequencies.

Regions of Anomalous dispersion dn/d is negative

Refractive index, n

where dispersion is
anomalous (near a
Normal Normal Normal
dispersion dispersion dispersion resonance).
1
So vg can exceed
0 1 2 3
v or even c0 for
Infrared Visible Ultraviolet X-ray
these frequencies!

Unfortunately, absorption is strong in these regions.

Also, dn/d is only steep when the resonance is narrow, so only a

narrow range of frequencies has vg > c0. Frequencies outside this
range have vg < c0. Communication uses pulses of light, which are
necessarily broadband, and therefore break up into a mess.
Superluminal (vg > c0) Light in Laser
Amplifiers: Not Really Faster Than c

All real-world Output pulse without

amplifiers saturate amplifier saturation
(the gain is depleted
Amplifier
during the pulse by
the pulse).

Output pulse with

Pulse amplifier saturation
Amplification of only peak
the front of the pulse without Pulse peak
gives the appearance saturation with saturation
of superluminal group
velocity. This is not interesting.
Actual Superluminal Light
To exceed c0, we need a region of negative dn/d over a fairly large
range of frequencies. And the slope should not vary muchto avoid
pulse break-up. And absorption should be minimal.
One trickwhich worksis to excite the medium in advance with a laser
pulse, yielding gain (instead of absorption), which inverts the curves.
Excite two nearby resonances, which will have a region in between with
minimal absorption and near-linear negative slope:

2
Slow Light (vg << c0) and Stopped Light
Fabricating a material with
extremely high, positive dn/d
yields a very slow light speed.
Refractive index, n

A false-colored scanning electron

1 microscope image of a photonic
1 2
crystal composed of a periodic array
of holes etched in a silicon slab. It has
an extremely high dn/d.

Fabricating a medium with two

nearby, very strong absorptions Researchers have generated a
can yield a very high, positive beam with vg = 25km/hr and have
dn/d. actually been able to stop light!
Prisms, Phase Velocity, and Group Velocity
Phase-fronts are always perpendicular to the propagation direction.
Equivalently, the phase delay is the same for all paths through the prism.
What about the pulse front (intensity peak)? Because vg < v, the
pulse front will lag behind the phase-fronts in the thicker part of the
prism, causing the pulse to tilt when light traverses a prism.

Phase-fronts Pulse-
front tilt

Pulse Prism

Phase-fronts
Pulse-front

Interestingly, angular dispersion always causes pulse-front tilt.

Group velocity dispersion (GVD) is the
variation of group velocity with frequency.
Because n() is curved, dn/d also varies with frequency, and so
will the group velocity. This is GVD. GVD lengthens a pulse in
time. The more material, the longer the pulse.

A chirped pulse

Short Not so
pulse short
pulse

vg(blue) < vg(red)

Because short pulses have large ranges of frequencies, GVD is a

bigger issue for them than for nearly monochromatic light.
Group-velocity dispersion is undesirable
in telecommunications systems.

Train of input telecom pulses GVD turns short pulses into

long ones.

All materials have

positive GVD in the
visible and near-IR.

Many km of fiber

Fiber must be very

carefully designed to
compensate for GVD.
Train of output telecom pulses
Prisms and Group Velocity Again
What does the pulse transmitted by a prism really look likein color
this time?

Pulse-front tilt
Pulse Prism

Note that the blue precedes the red! Negative GVD!

Pulse This device has negative GVD and
hence can compensate for
Compressor propagation through materials.

In principle, only one prism is

required. But the other three
are needed to put the pulse
back together.

Its routine to stretch and then compress short light pulses by factors
of >1000.
Adjusting the prism maintains alignment.
Any prism in the compressor can be translated perpendicular to the
beam path to add glass and reduce the magnitude of negative GVD.

Conveniently, this does The output path is

not misalign the beam. independent of prism
position.

Input beam Output beam