Ultrafast Optics

Definition: the generation, measurement, and application of

ultrashort laser pulses

The birth and history of ultrafast optics

The uncertainty principle and long vs. short pulses

Ultrashort pulses = mode-locking

Prof. Rick Trebino
Georgia Tech
Generating ultrashort pulses
www.physics.gatech.
edu/frog
Measuring ultrashort pulses

Ultrafast spectroscopy

Medical imaging
The birth of ultrafast technology

Bet: Do all four hooves

of a galloping horse ever
simultaneously leave the
ground?
Leland Stanford Eadweard Muybridge

Palo Alto, CA 1872

Time Resolution:
1/60th of a second
Harold Edgerton: strobe photography
“How to Make
Apple sauce
at MIT”
1964
Harold
Edgerton
MIT, 1942

“Splash on a
Glass”
Curtis Hurley
Junior High
School student
1996

Time resolution: a few microseconds

Pulsed pumping: s pulses

Pumping a laser medium with a short-pulse flash lamp yields a

fairly short pulse. Flash lamp pulses as short as ~1 µs exist.

Unfortunately, this yields a pulse as long as the excited-state

lifetime of the laser medium, which can be considerably longer
than the pump pulse.

Since solid-state laser media have lifetimes in the microsecond

range, it yields pulses microseconds to milliseconds long.

I(t)

Long and potentially

complex pulse
Q-switching: ns pulses

Q-switching involves: Output pulse intensity

Preventing the Gain

laser from lasing saturation
(by adding 100%

Cavity Gain
Cavity Loss
massive loss)
until the flash
lamp is finished
flashing, and
0%
Abruptly allowing Time
the laser to lase.

The pulse length is limited by how fast we can switch and the
round-trip time of the laser and yields pulses 10 - 100 ns long.
Ultrafast optics vs. electronics
–6
10

–9
10
Timescale (seconds)

Electronics

–12
10

Optics
–15
10

1960 1970 1980 1990 2000

Year

No one expects electronics to ever catch up.

The metric system
We’ll need to really know the metric system because the pulses are
incredibly short and the powers and intensities can be incredibly high.

Prefixes:

Small Big
milli (m) 10-3 Kilo (k) 10+3
micro (µ) 10-6 Mega (M) 10+6
nano (n) 10-9 Giga (G) 10+9
pico (p) 10-12 Tera (T) 10+12
femto (f) 10-15 Peta (P) 10+15
atto (a) 10-18 Exa (E) 10+18
 Timescales
It’s routine to generate pulses < 1 picosecond (10-12 s) long.
Researchers generate pulses a few femtoseconds (10-15 s) long.

Computer Camera One Age of Human existence

10 fs light clock cycle flash month pyramids
pulse 1 minute Age of universe

10-15 10-12 10-9 10-6 10-3 100 103 106 109 1012 1015 1018
Time (seconds)
1 femtosecond 1 picosecond

Such a pulse is to a minute as a minute is to the age of the universe.

Such a pulse is to a second as 5 cents is to the US national debt.

Ultrafast set-ups can be very sophisticated.
Unbelievably high intensities!

National Ignition Facility

(under construction)

192 shaped pulses

>1 MJ total energy
~ 10 Petawatts
Continuous vs. ultrashort pulses of light
A constant and a delta-function are a Fourier-Transform pair.

Irradiance vs. time Spectrum

Continuous beam:

time frequency

Ultrashort pulse:

time frequency
 Long vs. short pulses of light
The uncertainty principle says that the product of the temporal and
spectral pulse widths is greater than ~1.

Irradiance vs. time Spectrum

Long pulse

time frequency

Short pulse

time frequency
Ultrafast laser media
Solid-state laser media have broad bandwidths and are convenient.
Laser power
Generating short pulses = Mode-locking
Locking vs. not locking the phases of the laser modes (frequencies)

Intensity vs. time

Random Light bulb
phases

Time

Intensity vs. time

Locked Ultrashort
phases pulse!

Time
The saturable absorber

Like a sponge, an absorbing medium can

only absorb so much. High-intensity spikes
 (I ) burn through; low-intensity light is absorbed.
The absorption coefficient vs. input intensity:
0
0
 (I) 
1
2 0 1  I Isat

0 I sat I
The effect of a saturable absorber
First, imagine raster-scanning the pulse vs. time like this,
where k indicates the round-trip number:
Intensity

Short time (fs)

k=1
k=2
Ro

k=3
u nd
tri

k=7
ps
(k
)

Notice that the weak pulses are suppressed, and the

strong pulse shortens and is amplified (by the laser gain).

After many round trips, even a slightly saturable absorber can yield
a very short pulse. Alas, most absorbers recover slowly…
 The gain
Initial unsaturated
saturates, loss
Saturated
too. This gain and
is good. Initial unsaturated
gain
loss

Amplifying the pulse

uses up the stored gain < loss gain > loss gain < loss
energy and hence
reduces the gain.

The combination of
saturable absorption
and saturable gain
Laser
yields ultrashort
pulse
pulses even when the intensity
absorber recovers
slowly. time
The passively mode-locked dye laser:
ps pulses

Pump Saturable
beam absorber

Gain medium

Passively mode-locked dye lasers yield pulses as short as a few

tens of fs.
Ultrafast dye lasers were big in the 80’s and early 90’s. But dyes
are messy and a hassle. We now use solid-state lasers.
Passive mode-locking via
Kerr lensing: fs pulses

Just as absorption varies with intensity, a medium’s other optical

characteristic, the refractive index, will also change with
intensity.

n( I )  n0  n2 I

The nonlinear refractive index is the basis of most ultrafast lasers

today.
A lens and a lens
x
(x) = n k L(x)
A lens is a lens because the
phase delay seen by a L(x)
beam varies quadratically
with x:

Typical laser beam with a

Gaussian intensity profile:
I0 exp(-x2/w2) x (x) = n(x) k L

Now what if L is constant, In both cases,

but n varies quadratically a quadratic
with x: variation of the
n(x)
phase with x
yields a lens.
Kerr-lens
mode-locking
Titanium:Sapphire not only
lases, but it has a large n2!

Placing an aperture at the

focus favors a short pulse:

Losses are too high for a low-

Large n2 intensity cw mode to lase, but
not for a high-intensity fs pulse.

Kerr-lensing is the mode-locking mechanism of the Ti:Sapphire laser.

Dispersion is critical in ultrafast optics.
Dispersion, the dependence of the refractive index on
wavelength, has two effects on a pulse, one in space and the
other in time.
Angular dispersion disperses a beam in space (angle):

Group-velocity dispersion (GVD) disperses a pulse in time:

vg(blue) < vg(red) Both effects play major

roles in ultrafast optics.
What exactly does a pulse look like after
it passes through a prism?
Prisms cause all kinds of interesting distortions in the pulse.
But notice that the blue emerges first and red last.

Angularly
Input
dispersed
pulse
pulse

Prism

Negative GDD!
Pulse compressor
This device has negative group-delay dispersion and hence can
compensate for propagation through materials (i.e., for positive chirp).

Angular dispersion yields

negative GDD.

The additional prisms are required to put the pulse back together
again.
A simpler, two-prism pulse compressor

Reflecting the pulse back through the first two prisms also works
and is easier to set up.

Uncompressed input pulse

Compressed output pulse

Mirror

This design is particularly convenient inside a laser.

The Ti:Sapphire laser, including
dispersion compensation
Adding two prisms compensates for GDD in the Ti:Sapphire crystal
and mirrors.

Ti:Sapphire Slit for

cw pump beam
gain medium tuning

Prism dispersion
compensator

This is currently the workhorse laser of the ultrafast optics community.

Typical properties: 800 nm wavelength, 100 fs pulse length, 1 nJ
pulse energy, 100 MHz rep rate
The Dilemma

In order to measure
an event in time,
you need a shorter one.

To study this event, you need a

strobe light pulse that’s shorter.
Photograph taken by Harold Edgerton, MIT

But then, to measure the strobe light pulse,

you need a detector whose response time is even shorter.

And so on…

So, now, how do you measure the shortest event?

Pulse Measurement in the time domain:
The Intensity Autocorrelator
Crossing beams in a nonlinear-optical crystal, varying the
delay between them, and measuring the signal pulse energy
vs. delay, yields the Intensity Autocorrelation, A(2)().
Pulse to be
measured

The signal field is E(t) E(t-).

Beam So the signal intensity is I(t) I(t-)
SHG
splitter
crystal
E(t–) SHG
crystal Detector

Esig(t,)
Variable E(t)
delay, 

The Intensity A(2)      I (t ) I  t    dt
Autocorrelation:

 Autocorrelations of very complex pulses
As the intensity Intensity Autocorrelation
increases in
complexity, its
autocorrelation
approaches a
broad smooth
background and a
coherence spike.
Retrieving the
intensity from its
autocorrelation is
also equivalent to
the 1D phase-
retrieval problem!

This proves that retrieving the intensity from the autocorrelation

is fundamentally impossible!
The Spectrogram of a waveform E(t)
It involves measuring the spectrum of the product: E(t) g(t-)

E (t ) g(t-)
Example:
Linearly
Field amplitude

chirped
Gaussian
pulse

g(t-) gates out a Esig(t,)

piece of E(t),
centered at .

0  Time (t)

The spectrogram tells the color and intensity of E(t) at the time, .
Spectrograms for Linearly Chirped Pulses
Spectrograms for linearly chirped pulses
Negatively chirped Unchirped pulse Positively chirped
pulse
Negatively chirped Unchirped pulse
Positively chirped
Frequency

Time
SHG FROG tra c e --e x pa n de d
FROG tra c e --ex pa n d e d

60
60
Frequency

50
50
1
40
40

30
30

0
20
20

10
10

10 20 30 40 50 60
10 20 30 40 50 60

Delay

Like
Like a musicalscore,
a musical score,thethe spectrogram
spectrogram visually
visually displaysdisplays the
the frequency
frequency vs.thetime.
vs. time (and intensity, too).
Properties of the spectrogram
Algorithms exist to retrieve E(t) from its spectrogram.
The spectrogram essentially uniquely determines the waveform intensity,
I(t), and phase, (t) [and, equivalently, S() and ()].
There are a few ambiguities, but they’re trivial.

The gate need not be—and should not be—much shorter than E(t).
Suppose we use a zero-width gate pulse:
It would gate out an infinitely short chunk of the pulse, which
would have an infinitely broad spectrum—no color (phase)
information at all!

The spectrogram resolves the dilemma! It doesn’t need the shorter

event! It temporally resolves the slow components and spectrally
resolves the fast components.
Spectrogram pulse retrieval is equivalent to the 2D Phase Retrieval
Problem—a well-behaved problem!
Frequency-Resolved Optical Gating
(FROG)
FROG is simply a spectrally resolved
Pulse to be autocorrelation, which is a spectrogram.
measured

Beam I FROG ( , )  Spectrum  Esig (t , ) 

splitter

E(t–) Camera
SHG
crystal Spec-
trometer

Variable E(t) Esig(t,)= E(t) E(t-)

delay, 

FROG can use any fast nonlinear-optical process.

SHG FROG is the most sensitive version.
One of the shortest events ever created!

FROG traces

A 4.5 fs pulse!

Baltuska,
Pshenichnikov,
and Weirsma,
J. Quant. Electron.,
35, 459 (1999).
FROG measurement of the ultrabroadband
continuum
Retrieved intensity
Ultrabroadband continuum was created by and phase
propagating 1-nJ, 800-nm, 30-fs pulses through
16 cm of microstructure fiber.

Spectrogram

This pulse has a time-bandwidth product

of ~ 4000, and is the most complex
ultrashort pulse ever measured.
Ultrafast laser spectroscopy: why?
Most events that occur in atoms and molecules occur on fs and ps time
scales. The length scales are very small, so very little time is required
for the relevant motion.
Fluorescence occurs on a ns time scale, but competing non-radiative
processes only speed things up because relaxation rates add:

1 1 1
 
 ex  fl  nr

Biologically important processes utilize excitation energy for purposes

other than fluorescence and hence must be very fast.
Collisions in room-temperature liquids occur on a few-fs time scale, so
nearly all processes in liquids are ultrafast.
Semiconductor processes of technological interest are necessarily
ultrafast or we wouldn’t be interested.
The simplest ultrafast spectroscopy
method is the excite-probe technique.
This involves exciting the sample with one pulse, probing it with
another a variable delay later, and measuring the change in the
transmitted probe pulse average power vs. delay:

Change in probe
pulse energy
Probe
pulse

Epr(t–) Sample 0 Delay, 

medium
Detector

Variable Eex(t)
delay,  Excite Esig(t,)
pulse

The excite and probe pulses can be different colors.

This technique is also called the pump-probe technique.
Ultrafast excite-probe measurements in DNA
DNA bases undergo photo-oxidative damage, which can yield
mutations. Understanding the photo-physics of these important
molecules may help to understand this process.

Transient absorption at 600 nm of protonated guanosine in acidic

(pH 2) and basic (pH 11) aqueous solution.
Pecourt, et al., Ultrafast Phenomena XII, p.566(2000)
Beyond ultrafast spectroscopy: controlling
chemical reactions with ultrashort pulses
You can excite a chemical bond with the right wavelength, but the
energy redistributes all around the molecule rapidly (“IVR”).

But exciting with an intense, shaped ultrashort pulse can control the
molecule’s vibrations and produce the desired products.
Ultrashort in time is also ultrashort in
space
Novel imaging techniques yield ~1-µm resolution, emphasizing
edges of objects. They include optical coherence tomography
and multi-photon imaging.

Two-photon fluorescence emission from a focused pulse:

One-photon fluorescence from a

beam entering from the right

Two-photon fluorescence from

an identical beam entering from
the left
SHG and THG imaging simultaneously
Images of a Zebrafish larva

20 mm 20 mm

THG (blue) shows edges: the Muscle fibers exhibit strong SHG
larva skin and boundary of (green) due to crystalline nano-
somite and notochord. structure.
Sun and coworkers, Opt. Expr. Nov. 2003
Low-coherence Reference
Michelson
interferometry Interferometer
Sample
When the interferometer paths Source
are equal, the intensity fringes
are the strongest. The Detector
accuracy is the coherence
time/c.
High-coherence Source Low-coherence Source
/2
Coherence
Output intensity

Output intensity
Length

Mirror Displacement Mirror Displacement

 OCT measurements of a live tadpole

Dorsal Ventral

Reflectance 1 mm

Boppart, et al., Dev. Biology 177 (1996)