The Birth of Molecules

In less than a trillionth of a second, atoms can collide, interact and give
birth to molecules. Wit-h lasers and molecular beams, it is now possible
to witness the motions 7f molecules as one substance changes to another
(

by Ahrned H. Zewail
' n 1872 railroad magnate Leland
Stanford wagered $25,000 that a
galloping horse, at some point in
its stride, lifts all four hooves off the
ground. To prove it, Stanford employed
English photographer Eadweard Muybridge. After many attempts, Muybridge developed a camera shutter that
opened and closed for only two thousandths of a second, enabling him to
capture on film a horse flying through
the air [see illustration at top right].
During the past century, all scientific
disciplines from astrophysics to zoology have exploited high-speed photography to revolutionize understanding
of animal and mechanical motions that
are quicker than the eye can follow.
The time resolution, or shutter
speed, needed to photograph the ultrafast motions of molecules is beyond
any conventional scale. When a molecule breaks apart into fragments or
when it combines with another to form
a new n~olecule,the chemical bonds
between atoms break or form in less
than a trillionth of a second, or one picosecond. Scientists have hoped to observe molecular motions in real time
and to witness the birth of molecules :
the instant at which the fate of the
molecular reaction is decided and the

AHMED H. ZEWAIL is the Linus Pauling Professor of Chemical Physics at the

California Institute of Technology and
has been a faculty member there since
1976. He has pioneered ultrafast laser
femtochernistry to investigate the realtime motion of molecules in reactions.
He received h s B.S. from Alexandria
University in Egypt and h s Ph.D. from
the University of Pennsylvania. Among
the awards he has received are the
1989 King Faisal International Prize, the
Harrison-Howe Award and the John Simon Guggerheim Fellowship. Zewail is a
member of the National Academy of Sciences. T h s article is dedicated "to the
memory of my dear friend Dick Bernstein, a giant in the field of molecularbeam science."

76

final products are determined. Like

Muybridge, they needed. to develop an
ultrafast shutter, but it had to work 10
billion times faster than the 19th-century model.
For the past decade our research
group at the California Institute of
Technolow has been developing techniques toldbserve the dyna&cs of molecules in real time. From 1985 to 1987
we refined our system of advanced
lasers and molecular beams to a point
at which we can now record the motions of molecule's as they form and
break bonds. The reaction can be seen
as it proceeds from reactants through
transition states and finally to products-chemistry as it happens.
ecause transition states exist for
less than a trillionth of a second,
the time resolution should be
shorter-a few quadrillionths of a second, or a few femtoseconds (one is
equal to 10-15 second). A femtosecond
is a smaller unit of time than a tick of
even the finest atomic clock. A femtosecond is to a second what a second
is to 32 million years. Furthermore,
whereas in one second light travels
nearly 300,000 lulometers-almost the
distance between the earth and the
moon-in one femtosecoiid light travels .3 micron-about the diameter of
the smallest bacterium.
Alchemists in ancient Egypt, Greece,
Arabia and Chma did not know about

TIME RESOLUTION of high-speed photography has improved 10 billion times

as the technology has evolved from filming movements of animals to capturing
the birth of molecules. In one second,
a horse gallops 10 meters, as shown in
photographs (top) taken by Eadweard
Muybridge in 1887. In five trillionths
of a second, hydrogen iodide collides
with carbon dioxide, creating carbon
monoxide, hydroxide and iodine. The
illustration (bottom)is based on quantitative observations made at Caltech.

SCIENTIFIC
AMERICAN December 1990

the fundamental importance of this

time scale to the change of one substance to another, although they certainly did recognize the art of the
transmutation. Only in this century
have chemists been able to use a variety of tools to understand the sci-

ence of molecular reaction dynamics.

At a n~olecularlevel a reaction starts
when two molecules-call them A and
B-are brought close enough together
so that they begin to interact. As the
molecules move even closer, other molecular species, which are neither A nor
B, are formed until a new, stable molecule, C, is created.
One of these transitory species represents the stage at which the reaction proceeds irreversibly to products.
This stage, technically called the transition state, is here denoted as A s - - B.
Hence,

The converse process is also possible:

when molecule C is energized, it can
pass through a transition state to form
molecules A and B.
In any chemical reaction the motions

of the electrons and nuclei of atoms

determine how the molecules interact,
and those interactions in turn create
the forces that govern the reaction's
dynamics. The molecular motions can
be rigorously delineated by the laws of
quantum mechanics. In many cases,
Newtonian, or classical, mechanics can
simply give the transfer of energy and
momentum during the course of the
reaction, just as effectively as it describes the collision between two objects. If investigators can determine
how molecular motions change during the critical transition phase, we can
understand how new chemical bonds
form and old ones disappear.
In practice, chemists' do not keep
track of every possible motion of every
electron and nucleus in a nlolecular
system. Instead they have discovered
clever ways to represent the state of a
system. Each state possesses a certain

amount of energy. The potential energy

of a molecule when the atoms are a
certain distance apart can be represented as a point on a graph. When all the
states and their corresponding potential energies are graphed, they form a
surface that has mountains and valleys.
Molecular systems will spontaneously
move from high-energy states (mountains) to low-energy states (valleys),but
they require energy to move from valleys to mountains. The bottom of a valley indicates a stable state; the slopes
around the valley represent the region
of transition states, the different configurations for the atoms in proximity.
Chemists technicallv define a transition
state as a saddle point on a potentialenergy surface.
If a molecular system is composed
of two atoms and if the system's potential energy depends only on the
distance that separates the atoms, the

SCIENTIFICAMERICAN December 1990

77

FEMTOSECOND PHOTOGRAPHY captures the dynamics of molecular reactions in real time. The laser system
(above)generates a pump pulse and a
probe pulse. As shown in the illustration (lefi),the pump pulse and the probe
pulse leave the laser system at the same
time. The probe pulse is diverted so
that a time delay of a few femtoseconds
is introduced between the two pulses.
When the pump pulse hits a molecule in
the molecular beam, it initiates a chemical reaction. The probe pulse strikes the
molecule a few femtoseconds later. The
molecule then emits a spectrum of light,
which can be analyzed to determine
the dynamics of the atoms in real time.

potential-energy surface reduces to a

curve. If the system is governed by the
attraction of oppositely charged atoms
at "loiig" distances and by the repulsion of atonlic nuclei at "short" distances, the curve will have a single well
at "intermediate" distances, indicating
that the system is stable there.
For molecular systems composed of
many atoms that can move in many
ways, the potential-energy surface can
be complicated and multidimensional. Nevertheless, the surface is very useful for describing every possible path
that molecules may follow-from reactants (valleys)through transition states
(mountains)to products (valleys).
One can estimate how long it takes
for molecular systems to go through
the mountains-that is, the duration of
the transition state-by applying ideas
from Newtonian mechanics. Heavy molecules or atoms typically emerge from
chemical reactions at speeds of 1,000
meters per second. If molecules A and
B are close enough to be in a transition
state, then the distance between them
is on the order of that of a molecular bond, approximately . l nanometer
(10-l0meter). The ratio of this distance
to the recoil velocity yields the amount
of time during whch A and B form
78

a transition state-100 femtoseconds.

For the "simplest" of all chemical reactions-a light hydrogen atom approaching a hydrogen molecule-the
transition state is very short-lived,
about 10 femtoseconds. Thus, the time
scales for transition states range from
about 10 to 100 femtoseconds.

or more than 100 years chemists have studied reaction mechanisms and molecular kinetics in
hope of understanding reactivity. In
the late 19th century, Swedish scientist
Svante Arrheilius took an important
step in describing the macroscopic dynamics of chemical reactions. He determined how the rate of a reaction depends on temperature. His famous exponential equation reveals that the rate
of a reaction increases as the amount
of applied heat increases.
But such equations, whch describe
the dynamic properties at the macroscopic level, yield information about
microscopic molecular dynamics only
if a number of assumptions are made.
Nevertheless, many aspects of molecular dynanlics can be deduced from
macroscopic observations, and Arrhenius's equation remains important to
t h s day.

SCIENTIFIC
AMERICAN December 1990

Some 40 years ago techniques were

introduced to study chemical intermediates and fast kinetics for the first
time. Ronald G. W. Norrisli and George
Porter of the University of Cambridge
and Manfred Eigen of the Max Planck
Institute for Pliysical Chenlistry were
able to resolve chemical events that
lasted about a thousandth of a second.
This time scale was ideal for intermediates but too long for transition states.
To probe the dynamics of niolecules as they collide, investigators developed the new methodology of molecular-beam technologies in the 1960s.
By releasing molecules into a vacuum
and collimating them with a skimmer,
chemists created beams of molecules
that did not collide with one another.
When one beam containing molecules
of, say, A is crossed with another introducing molecules of B, n~oleculesof A
collide with those of B to form a product, C. The full collision (A+ B C ) is
called a bimolecular reaction. The investigators then detected molecule C
as it was created in different amounts
and in different directions depending
on the nature of the forces between A
and B. They used the precollision conditions of the reactants and the postcollision attributes of the products to
deduce the dynamics of the collision.
Conversely, a beam of molecules of
C could be bombarded with light from
a laser, energizing the molecules and
brealung their chemical bonds to produce fragments A and B. This dissociation reaction (C A+ B) could then

be analyzed by measuring the attributes of the fragments. Since the development of molecular beams aiid
other techniques that resolve the quantum states of products, chemists have
learned much about the elementary dynamics of reactions. As testimony to
the importance of molecular-beam science and reaction dynamics, the 1086
Nobel Prize in chemistry and the 1988
Welch Award were given for contributions in these fields.
In molecular-beam experiments, the
sequence of events in a reaction, including the transition states, could not
be observed in real time. Chemist Sture
Forsen of Lund University came up with
an insightful analogy that illustrates
the importance of understanding transitory stages in the dynamics. He compared the scientific community to a
theater audience watching a drastically
shortened version of a classical drama.
The audience is shown only the opening scenes of, say, Hamlet and its finale. Forsen writes, "The main characters are introduced, then the curtain
falls for change of scenery and, as it
rises again, we see on the scene floor a
considerable number of 'dead' bodies
and a few survivors. Not an easy task
for the inexperienced to unravel what
actually took place in between."
In an attempt to probe transition
states more directly than can be done
with conventional molecular-beam experiments, chemists turned to new
time-averaged techmques aiid developed methods that record the absorption, emission or scattering of light by
transition states. Like molecular beams,
these methods do not resolve the
events in real time, but they do provide
more direct possibilities for examining
transition states and dynamics.
In 1979 our research team at Caltech
first attempted to enter the field of ultrafast chemistry by using lasers and
molecular beams. In those days the
time resolution was tens of picoseconds, too gross a time scale for viewing the motion of reactions through
transition states. We concentrated on
developing techniques to study how energy moves in molecules prior to reactions. By 1984 time resolution in molecular-beam experiments had improved
to three picoseconds. That advance allowed us to measure directly the rate
of reaction as a molecule changed from
one quantum state to another. Still, we
needed a technology that could record
the spectrum of transition states in
real time with femtosecond resolution.
To build such a femtosecond, molecular "camera," we combined the technology of molecular beams with that of
ultrashort laser pulses. The molecular

beams allowed us to isolate the reactions in a vacuum; the ultrashort laser

pulses gave us the ability to resolve the
dynamics with the needed time resolution. During the past five years, advances in laser technology at AT&T
Bell Laboratories and other institutions
have made it feasible to reach the critical femtosecond time resolution. At
the moment, pulses as short as six
femtoseconds can be acheved. With
such ultrashort pulses, we can obtain
a "shutter speed" of approximately 10
femtoseconds.

lie principles involved in ultrafast, molecular photography have

some similarity to those applied
by Muybridge. The key to h s work was
a special camera shutter that exposed
film for only .002 second. To set up
the experiment, Muybridge spaced 1 2
of these cameras half a meter apart
alongside a horse track. For each camera he stretched a string across the
track to a mechanism that would trigger the shutter when a horse broke
through the string.
With this system, Muybridge attained
a resolution in each picture of about
two centimeters, assuming the horse
was galloping at a speed of about 10
meters per second. (The resolution,
or definition, is simply the velocity of
the motion multiplied by the exposure
time.) The speed of the motion divided by the distance between cameras
equals the number of frames per second-20 in this case. The motion within a picture becomes sharper as the
shutter speed increases. The resolution
of the motion improves as the distance
between the cameras decreases.
Two aspects of high-speed photography are relevant to femtosecond, molecular photography. First, both types
of photography break up a continuous
motion into a series of snapshots, or
frames. Thus, one can slow down a fast
motion as much as one likes so that
the eye can see it. Second, both methods must produce enough frames in
rapid succession so that the frames
can be reassembled to give the illusion
of a continuous motion. The change in
position of an object from one frame
to the next should be gradual, and at
least 30 frames should be taken to provide one second of the animation.
In femtosecond, molecular photography, the definition of the frame and
the number of frames per second must
be adjusted to resolve the elementary
nuclear motions of reactions and, most
important, ephemeral transition states.
The frame definition must be shorter
than .l nanometer. Because the speed
of the molecular motion is typically

1,000 meters per second, the shutter

resolution must be in a time range of
better than 100 femtoseconds.
The conceptual idea behind these
femtosecond experiments is rather simple. A first laser flash, called the pump
pulse, hits a molecule in isolation, initiating a reaction and setting the experimental clock at zero. A second flash,
called the probe pulse, arrives several
femtoseconds later and records a snapshot of the reaction at that particular instant. Like the cameras in Muybridge's experiment, a femtosecond,
molecular camera records successive
images at different times to obtain information about different stages of the
reaction.
To produce time delays between the

STATE OF MOLECULAR SYSTEMS can

be identified by the light that the molecule absorbs. When atoms in a molecule are relatively close together, they
tend to absorb long wavelengths of light
(red, for example). When the atoms
are farther apart, they tend to absorb
short wavelengths of light (blue, for instance). The change in the spectrum is
the fingerprint of the atoms in motion.
TIFIC

AMERICAN December 1990

79

$?b..

(A/4,k+
-- @
@i,If, p\,-:.;

o= -"

c o o
d

-- - - - -- - - - - - -

----------------------------

00

&dU,,

20 FEMTOSECONDS
BEFORE REACTION

START OF
REACTION

100 FEMTOSECONDS
AFTER REACTION

180 FEMTOSECONDS AFTER REACTION

SEPARATION BETWEEN I AND CN (NANOMETERS)

SEPARATION BETWEEN I AND CN (NANOMETERS)

FEMTOSECOND DYNAMICS of cyanogen iodide (ICN) can be tion, because the exact momentum and position of the atoms
best described in terms of potential-energy curves. In clas- cannot be determined simultaneously,as specified by quansical mechanics, when molecules of ICN in a low-energy state tum mechanics. As shown in the diagram at the right, it
( a ) interact with a laser beam, light excites ICN to a high- is most likely, for example, that iodine is separated from
energy state (b).The molecule then dissociates into iodine cyanide by about .85 nanometer at 180 femtoseconds (d).
and cyanide, and the distance between iodine and cyanide At that instant, however, there is a small chance that they
increases ( c and d ) . The diagram at the left is a sirnplifica- might be separated by .7 nanometer or even 1.0 nanometer.

pump pulse and the probe pulse, our

research group takes advantage of the
large but finite speed of light. Initially
we tune the optical system so that both
pulses reach the apparatus at the same
time. We then divert the probe pulse
so that it travels a longer distance than
does the pump pulse before it reaches
the molecular beam. If the probe pulse
travels one micron farther than the
pump pulse, it will be delayed 3.33
femtoseconds, because light travels at
300 million meters per second.
Accordingly, pulses that are separated by distances of from one to 100 microns resolve the motion during periods of from 3.33 to 333 femtoseconds.
Hence, a spatial detour of microns,
whch can be easily accomplished in
the laboratory, can be used to obtain
a delay of femtoseconds. (A shutter
speed of a few femtoseconds is beyond
the capability of any camera based on
mechanical or electrical devices.)
When the probe pulse h t s a molecule, it does not then transmit an image to a detector, like light reflecting
off a scene into a camera. Instead the
probe pulse interacts with the molecule, and then the molecule emits a
spectrum of light. The spectrum varies
in intensity and color. The variation depends on the wavelength of the original probe pulse, the atoms in the orig80

inal molecule and the state of the

molecule when it was hit by the probe
pulse. Free molecules A and B have a
different spectrum from that of A. B
in the transition state, and the transition state's spectrum differs from that
of the product, C.
We could therefore distinguish the
spectral signals from each of these
states. We could deternline the timing
of these signals by knowing the time
that had elapsed between the impacts
of the pump pulse and the probe pulse.
Based on these principles, we were able
to reconstruct and shtdy an image of
elementary reactions in real time.
'hen my colleague and friend
Richard Bernstein of the University of California at Los Angeles learned about the femtosecond,
molecular camera, he was very enthusiastic about the development, and we
discussed the exciting possibililies the
technique created. Over champagne at
his house in Santa Monica, he suggested the designation "femtochemistry."
Since 1985 we have investigated the
femtochemistry of cyanide reactions,
atn~osphericreactions and salt reactions. These examples illustrate the
broad scope of the method and the potential for gaining insights into the dynamics of various chemical processes.

SCIENTIFIC
AMERICAN December 1990

We began work five years ago on the

dissociation of cyanogen iodide:

The goal was to detect the fragment

CN (cyanide) as a nascent product. Our
graduate students and postdoctoral
fellows were working hard to reach this
goal, and we were all determined to
acllieve results by the end of 1985. We
built a two-pulse laser system, and by
borrowing some equipment to help
compress the pulses, we were able to
get the experiment running. In December we reported that we had detected
nascent cyanide molecules. We realized
that the transition states of cyanogen
iodide were just beyond the time resolution of our apparatus.
To improve the time resolution, we
decided to build a better system of molecular beams and lasers. We set up our
experiment in a newly rehabilitated,
dust-free room that once housed the
X-ray machnes from Liiius Pauling's
days at Caltech. Shortly after Thanksgiving in 1986, we opened the facility.
We were very excited about the possibility of being able to observe transition states directly.
A key feature of these experiments
was the ability to make the definition
of the frame adequate enough to re-

solve the motion. The detector was

made sensitive enough to probe the
cyanide molecule when it was free or in
the process of separating from the iodine atom. The sensitivity was achieved
by using probing laser pulses of different colors in accordance with the spectrum of the molecules. We were thus
able to observe the spectral frames
changing with time as the reaction progressed from an intact ICN molecule to
an iodine atom and molecular fragment, cyanide. We enjoyed discussing
the results late into the night.
In 1987 we reported on the probing
of the reaction of ICN to iodine and
cyanide with enough time resolution to
be able to observe the transition state
{ I CN } directly. We obtained the
real-time dynamics of the reaction and
then deduced detailed features of the
potential-energy surface [see illustration on opposite page].
That experiment made it possible to
clock the breaking of an elementary
chemical bond oil a femtosecond time
scale. The findings generated enthusiasm in the scientific community and inspired many more theoretical and experimental studies. The science writer
Isaac Asimov provided an interesting
description of the experiment. He compared the experiment with sticking a
pin in a balloon only four billionths of
an inch across-the size of an average
molecule.
Whle we were still building the new
femtosecond laser facility in 1986,
Bernstein visited us at Caltech, and we

-..

SODIUM
IODIDE

We wanted to understand t h s reaction

at the elementary level. We broke the
bond between hydrogen and iodine
with a pump pulse. We then observed
with probe pulses that the hydrogen attacks the carbon dioxide and sticks to
it for hundreds of femtoseconds. The
hydrogen atom then strips one of the
oxygen atoms from the carbon. The hy-

droxide molecule finally emerges five

picoseconds after the start of the reaction, as we reported in 1987 and earlier
this year.
The technique also allowed us to
view the dynamics of the transitory collision complex HOCO in real time. We
were able to relate this complex to the
potential-energy surface representing
hydrogen's approach to carbon dioxide. We are still examining the many
subtle differences between our results
and theory.
ust as physicists like to deal with
the hydrogen atom for simpliciv,
chemical physicists like to understand elementary molecular systems. The simplest of all chemical reactions is that involving two atoms. I
thought it would be interesting to direct some efforts at unraveling the dynamics of salt molecules such as sodium iodide:

Alkali halide reactions were the prototype for the "alkali age" of molecular-beam experiments, and it is insightful to study them in the "femtosecond age." Furthermore, researchers at
the University of Toronto had investigated the emission spectra of the sodium iodide molecule during its transition to atoms of sodium and iodme.
What intrigued us about sodium iodide is a rather interesting potentialenergy surface for the interaction of

FREE ATOMS OF
SODIUM AND IODINE

UPPERCURVE

\
0

SODIUM AND IODINE

IN A TRANSITION STATE

planned to probe the collision and

bonding of two molecules, that is, a bimolecular reaction. At first we could
not find a way to detect the start of
these reactions: the time zero. Although
bimolecular reactions last less than a
picosecond, the two reactant molecules
must first travel to meet each other-a
journey that lasts about a million times
longer than the reaction itself.
We found a solution to this problem
when we studied the investigations of
our colleagues at Orsay, France, and at
the University of Southern California.
By anchoring the two reagents together
with a weak van der Waals bond, we
could avoid the long time journey and
establish the time zero. Fortunately,
we had guidance from investigators at
U.S.C. who had performed time-integrated studies on the reaction of hydrogen bromide and carbon dioxide.
For our fkst real-time experiments
on these types of bimolecular reactions, we combined hydrogen iodide
with carbon dioxide to produce iodine,
carbon monoxide and hydroxide:

LOWERCURVE

.5
1.0
1.5
SEPARATION BETWEEN Na AND I (NANOMETERS)

FEMTOSECOND MOTIONS of the salt sodium iodide (NaI) reveal the molecular dynamics of the chemical bond linldng
sodium and iodine. In the potential-energydiagram at the left,
the ionic (attractive)curve intersects the covalent (repulsive)
curve. As a result, sodium iodide can break up into sodium
and iodine, acting in a covalent manner, or sodium iodide can

-2

2
4
TIME DELAY (PICOSECONDS)

exist in a high-energybound state, alternating between covalent and ionic behavior. If sodium iodide dissociates into
atoms, then the intensity of part of the spectrum will jump in
steps, as shown in the experimental results (top at right). If
sodium iodide is in its high-energybound state, then the intensity of part of the spectrum will oscillate (bottom at right).
SCIENTIFICAMERICAN December 1990

81

sodium and iodine atoms. If these atoms are brought together, they repel
one another, particularly at short &stances. On the other hand, if the same
elements are brought together as oppositely charged ions of sodium and iodine, they attract each other. In fact,
at a short enough distance, a stable salt
composed of sodium and iodine ions
will be formed. When the sodium and
iodine form a covalent bond, they share
electrons to create a stable energy state.
When the atoms form an ionic bond,
the sodium atom donates an electron
to the iodine atom to create charged
ions that attract.
Nah~re,however, does not work exactly in this way. Actually, the atoms
behave as if they are both ionic and covalent. The true potential-energym e s
are therefore composed of an ionic and
a covalent curve [see illustration on preceding page]. The ionic curve crosses
the covalent curve at a certain distance.
In this region of interaction, the molecular system has a mixture of covalent and ionic characteristics; in other
words, the molecule has a certain probability of being covalent and a certain
probability of being ionic.
Theorists have divided the potentialenergy curves into two parts: a lower
and an upper. The lower curve, which
represents low-energy states, is composed of the ionic curve at short distances and the covalent curve at long
distances. The upper curve depicts the
opposite situation: it is covalent at
short distances and ionic at long distances. The upper curve also represents a state high in energy.
The behavior represented by the lower surface usually dominates when the
sodium and iodine atoms are brought
together slowly so that they begin to
interact. In t h ~ case,
s
an electron passes from sodium to iodine to create a
stable n~oleculeof sodium iodide. If,
however, the atoms come together too
rapidly, they can jump, with a certain
probability, from the lower to the upper surface. The upper surface no longer represents the covalent repulsive
(or slightly bound) curve and instead
has a poirlt of least energy (a minimum). Molecules in this potential minimum will thus be somewhat stable.
(More specifically, the molecules enter
a quasi-bound state.)
We hoped to view the femtosecond
dynamics of the bond in sodium iodide
as it breaks to form atoms of sodiunl and iodine. In particular, we focused on the jump from the upper potential surface to the lower one, or vice
versa. We observed the motion as the
molecules change from being covalent
to being ionic and as they go through
82

SCIENTIFIC
AMERICAN
December l !

the crossing region. The sodium atom

gave an electron to the iodine atom
at a distance of .7 nanometer. The sodium atom had, in effect, employed its
electron as a harpoon to reel in the iodine atom.
To perform the experiment, we first
sent a laser pulse to excite the NaI
molecules. When the sodium separates
about .25 nanometer from the iodine,
the bond begins to break. A second
pulse is then released to probe the reaction at a point when the s o & m iodide bond is completely broken and
a sodium atom and an iodine atom
are formed. The probe pulse excites
the sodium atom and causes it to emit
yellow light. By probing the motion
from the moment the bond breakstime zero-until the birth of free sodium atoms, we can count, in real
time, the n~mnberof sodium atoms that
have appeared on the lower curve. But,
more important, by detecting the quasi-bound transition species, Na I,
whlch has different spectral properties
from those of free sodium, we can observe the transition from Na I to
free sodium and iodine atoms.
Some of the sodium iodide molecules on the upper curve, when reaching the crossing area, will jump to the
lower curve, and their bond will break
to form sodium and iodine atoms. The
molecules that do not jump but instead remain on the upper curve do
not break a bond. They continue to vibrate until they do jump. Hence, pulses of sodium atoms should be evident
only after each round-trip in the well of
the upper curve. This phenomenon is
exactly what we have observed. The results show the motion of the molecule
during the breaking of a chemical bond
and give the details of the potential energy governing the motion of sodi~un
and iodine atoms.
Laboratories in the U.S. (at IBM) and
in Germany (at Freiburg University) are
now applying femtosecond techniques
to observe the dynamics of many different types of reactions in the gas
phase. At Caltech, we are continuing
to probe elementary reactions and to
examine more complex reactions that
involve multiple births (for example,
AM--AB+A--A+B+A). We hope to
study the simplest of all reactions, the
combination of atomic hydrogen with
molecular hydrogen.
We are also exploring large molecular systems with the aim of answering
questions about selective reactivity. For
example, if a n~oleculewith two identical bonds is energized, do the two
bonds break simultaneously or consecutively? To address this general and
important question, we studied the dis-

-..

...

sociation of C21,F, into iodine and

C2F4.(In C,F4, a double bond forms
between the carbon atoms.) From realtime studies we found that the reaction
proceeds consecutively, despite the molecular equivalence of the two carboniodine bonds. Even more remarkable is
the discrepancy in the time scale for
bond breakage-the first bond breaks
in less than half a picosecond, whereas
the second bond takes 100 times longer to break!
The abllity to view molecular dynamics also suggests new ways of controlling reactions. There are several theoretical schemes for achieving control,
and already experimentalists are exploring some new possibilities. The prospect exists for fine-tuning the motion
and reactivity of molecules. If successful in the coming decades, laser-customized chemistry may be developed.

uybridge invented high-speed

photography for the fun of it,
and today the technique has
gone far beyond what he or Stanford
could imagine. Although one cannot
predict the future, we are certain of the
importance and beauty of understanding the science of molecules and their
reactions. The door is open for much
experimental and theoretical research
and for unexpected findings. It is my
hope that readers will share in the
fun and excitement that we have experienced in learning about the basics
of alchemy-the transmutation of one
substance to another-by observing the
ultrafast motions of molecules.

FURTHER READING
STATE-TO-STATE
REACTION DYNAMICS.
kchard N. Zare and Richard B. Bernstein in Physics Today, Vol. 33, No. 11,
pages 43-50; November, 1980.
INVESTIGATION
OF ULTRAFAST
PHENOMENA. Charles V. Shank in Science, Vol.
233, No. 4770, pages 1276-1280; September 19, 1986.
THE NOBELPRIZEFOR CHEMISTRY
in Les
Prix Nobel 1986. Almquist & Wlksell International, 1987.
REAL-TIME LASER FEMTOCHEMISTRY:
VIEWINGTHE TRANSITION
FROM REAGENTS TO PRODUCTS. Ahmed H. Zewail and Richard B. Bernstein in Chemical and Engineering News, Vol. 66, No.
45, pages 24-43; November 7, 1988.
LASERFEMTOCHEMISTRY.
Ahmed H. Zewail in Science, Vol. 242, No. 4886,
pages 1645-1653; December23,1988.
EXPOSING MOLECULAR
MOTIONS.
Ian W.
M. Smith in Nature, Vol. 343, No. 6260,
pages 691-692; February 22, 1990.
ULTRAFAST
MOLECULAR
REACTION DYNAMICS.
L. Khundkar and Ahmed H.
Zewail in Annual Review of Physical
Chemistry, Vol. 41, pages 15-60; 1990.