1

PHOTOCHEMISTRY

Prof. M.N.R. Ashfold (S305)

Fates of Excited State Molecules
absorption and emission of electromagnetic radiation
Einstein coefficients, absorption probabilities
fluorescence and phosphorescence
internal conversion and intersystem crossing
photodissociation and predissociation
Jablonski diagram

Lasers
requirements for laser action
population inversions
properties of laser radiation
examples of lasers
applications in spectroscopy and photochemistry

Photodissociation and Predissociation
viewed from perspective of
(a) the excited parent molecule, and
(b) the resulting photofragments
Ozone photochemistry, the stratospheric ozone
'deficit' and ozone photochemistry in the troposphere

Femtochemistry
following photochemically induced processes in real
time

2
Absorption and emission of
electromagnetic (EM) radiation
by atoms and molecules

Photochemical processes (e.g. photodissociations) are
initiated by the absorption of EM radiation (i.e. light).

This absorption is induced by the electric (
r
E) field of
the EM radiation [in the case of an electric dipole (or
electric quadrupole) interaction], or the magnetic (
r
B)
field [magnetic dipole interaction].

Electric dipole transitions are typically ~10
5
times
stronger than those carried by a magnetic dipole
interaction, and ~10
7
times more intense than electric
quadrupole transitions. Thus we shall concentrate on
electric dipole transitions.

In the presence of EM radiation a molecule experiences
a perturbation described by the dipolar interaction:


$
h(t) =
$
.
r
E(t) (1.1)

where the time-dependence recognises the oscillatory
nature of the EM field.
$
is the transition dipole moment operator (not the
permanent dipole moment), and
r
E(t) =
r
E
0
cost, where
r
E
0
is the amplitude of the
r
E field
and is the (angular) frequency 2.
3
An electric dipole interaction will connect two states
with wavefunctions
A
and
B
if the matrix element


B A
all space
E t d
*
$
. ( )
r

0 (1.2)

This is often written more compactly in the bra and
ket notation as B
$
.
r
E(t) A 0.

The transition rate (i.e. the transition probability per
unit time interval) is proportional to B
$
.
r
E(t) A
2
.

If we now assume that:
(i) the molecular dimensions are much smaller than ,
the photon wavelength, and
(ii) the interaction is sufficiently weak that it is valid to
use first order perturbation theory,
r
E(t) can be factored out of the matrix element (1.2), and
the transition rate becomes:

, B
$
A .
r
E
0
,
2
, and thus ,
$

BA
.
r
E
0
,
2
(1.3)


where
$

BA
= B
$
A is the transition dipole moment
between states A and B.
$

BA
is a measure of how much the charges within the
molecule rearrange (and in what directions) in making
the transition.
4
Absorption is one of three possible processes by
which resonant radiation can induce transfer of
population between two states A and B.



Rate = B
AB
[A] (
AB
) (1.4)

B
AB
= Einstein coefficient for (stimulated) absorption:

B
AB
=
1
6
0
2
h
, B
$
A ,
2
(1.5)

[A] = number density in state A
(
AB
) = spectral energy density (a measure of light intensity,
with units: J m
-3
s).

5
The other two processes involve emission, i.e.


Spontaneous emission


Rate = A
BA
[B] (1.6)

A
BA
= Einstein coefficient for spontaneous emission

A
BA
=
8
3
3
h
c
B
BA
(1.7)

The
3
factor in the numerator means that, other things being
equal, excited states at higher energies will have a faster
radiative decay rate.


Stimulated emission


Rate = B
BA
[B] (
BA
) (1.8)

B
BA
= Einstein coefficient for stimulated emission

=
g
g
B
A
B
AB
, (1.9)

where g
A
is the degeneracy of state A, etc.
The stimulated photon has the same frequency and
direction as the stimulating photon.
6
Question.
What determines the strength of a spectral transition?

Answer.
The transition probability is determined by B
$
A
2
.
In molecules, this 'electronic' transition strength is sub-
divided by vibrational and rotational fine-structure,
governed by Franck-Condon factors and rotational
linestrengths, respectively.

Symmetry considerations can tell us if
$

BA
= 0.
(recall NCN and CMW Level 3 lectures), e.g.

For atoms: l = 1 (the Laporte selection rule).

The spin quantum number, S, should be conserved, i.e.
S = 0, though this becomes less restrictive if the spin-
orbit coupling is large (e.g. in heavier atoms, or in
molecules containing at least one heavy atom).

For linear molecules (where states are classified by a
quantum number , which defines the axial projection
of the orbital angular momentum, e.g. , , , etc.):
= 0, or 1.

If spin-orbit coupling is important, is no longer a
good quantum number, but the axial projection of the
total electronic angular momentum (, the sum of
and S) is: = 0, or 1.

For centrosymmetric molecules, gu.

These are often called electric dipole selection rules.

7
Absorption: The Beer-Lambert Law

A beam of light (intensity I
0
) passes through a sample of
length l with concentration c.


I
0 I
l
Sample, Concentration c


The intensity, I, of light transmitted through the sample
is given by the Beer-Lambert Law:

I = I
0
exp[-()cl] (1.10)

which is often given in terms of the absorbance, A:

A = log
10
I
I
0
= ()cl (1.11)

() (base e) and () (base 10) are called the molar
absorption coefficient, molar absorptivity or molar
extinction coefficient with dimensions of
(concentration length)
1
, e.g. mol
1
dm
3
cm
1
.

8
Absorption profiles often span a spread of frequencies.
It is thus usual to quote the maximum absorption,
max
.



() is related to the Einstein coefficient (and thus the
transition dipole moment) by:


( )
B
Lh
d
AB
AB
=

ln10
1
2

. (1.12)

L is the Avogadro number and
AB
is the average
frequency of the transition.


9
Widths of Transitions

Absorption may be spread over a range of frequencies
for many reasons:

Spectral congestion due to the overlap of many
rotational or vibrational transitions.
Finite spectrometer resolution.
Doppler broadening; the spread of velocities in a
sample of molecules typically introduces a width of
~10
6
of the transition frequency.
The Doppler broadened width scales as T M.
If either state involved in a transition has a short
lifetime, , the transition will exhibit a homogeneous
linewidth.
This linewidth can be estimated from the energy-
time form of the uncertainty principle:

h t E (1.13)

(c.f. Heisenberg uncertainty principle).
Equating t with , the consequent uncertainty in E

h
E
2


gives a frequency width (full width half maximum)

2
1
(1.14)


(via E = h).


10
Processes that can cause a reduced lifetime
include:

Spontaneous emission - typically a small
contribution to the total linewidth.
Stimulated emission or absorption power
broadening occurs if the light intensity is too high.
Collisions: The extent of collisional broadening in
the gas phase is proportional to pressure, giving a
linewidth of ~0.1 cm
-1
/atm.
Dissociation or Predissociation can be very
significant.

11
Question:
What are the possible decay routes for an electronically
excited molecule?

Fluorescence:
Radiative decay between states of same spin
multiplicity (i.e. S=0).
Phosphorescence:
Radiative decay between states of different spin
multiplicity (i.e. S 0).
Internal Conversion:
Radiationless transition between states of same spin
multiplicity (e.g. S
n
~~> S
m<n
or T
n
~~> T
m<n
).
Intersystem Crossing:
Radiationless transition between states of differing
spin multiplicity (e.g. S
n
~~> T
n
).
Isomerisation
Dissociation (Predissociation)
Collisional Relaxation (Quenching)

Traditionally summarised using a Jablonski diagram.

S
0
= ground singlet (S = 0) state
S
1,....,n


= singlet states of progressively higher energy
T
1,....,n
= excited triplet (S = 1) states.
12
POSSIBLE FATES OF AN ELECTRONICALLY EXCITED MOLECULE:
JABLONSKI DIAGRAM

13
IN THE CASE OF A DIATOMIC MOLECULE




14
DISSOCIATION AND PREDISSOCIATION

The potential energy (PE) curve for the S
1
state
correlates with the excited products A + B*.
S
1
S
0
photoexcitation, as shown, provides insufficient
energy to access these products, however.

Nonetheless, S
1
molecules can break up to form ground
state A + B products if they first undergo an electronic
rearrangement (radiationless transition) to the S
0
(IC)
or T
1
(ISC) states. Such fragmentations, whereby
dissociation is preceded by radiationless transfer to an
electronic state that correlates to a lower energy
dissociation limit, are termed predissociations.

Predissociation can also involve transfer from a bound
S
1
level to a continuum associated with a purely
dissociative electronic state (e.g. state R, for which the
associated PE curve is entirely repulsive).

The timescale of predissociation processes depends on
the strength of the coupling between the excited state
populated by photon absorption (S
1
in this case) and the
various possible dissociating state(s) S
0
, T
1
or R,
depending on the coupling mechanism.

If excitation is to an energy higher than the A + B*
asymptote (or to the repulsive state R) then we can also
observe direct dissociation. Bond breaking then occurs
on a very short timescale (10
-13
10
-14
s).
15
Example of predissociation: The A
2

+
X
2
transition of the SH radical.
See strong variation in lifetime, and thus linewidth, with vibrational state and
isotope, reflecting ease of A state mixing with the repulsive
4

state.

SH SD

/cm
1
/cm
1

v' = 0 3 ns 0.002 260 ns 210
5

v' = 1 5 ps 1 50 ps 0.1





16
Features of Direct Dissociation

Excitation to a repulsive excited state, or to above the
dissociation limit of a bound state.
Very short excited state lifetime, < vibrational period.
Structureless, continuous absorption spectrum.
If excitation is to a single repulsive state, the
absorption spectrum has a Gaussian-like form
(Condon reflection principle).


Reflection of v=0 wavefunction in absorption spectrum

The angular distribution of product recoil velocities
reflects 'perpendicular' or 'parallel' nature of the
excitation step.

17
Quantum Yields
If an excited state decays by more than one process,
e.g. fluorescence and (pre)dissociation,



we can define a quantum yield, , for the various
processes, e.g.
fluor
;

fluor
BA
BA
k
k k
A
A k
= =
+
=
+
photons emitted
photons absorbed
fluor
fluor pre pre

(1.15)

The quantum yield for predissociation = (1
fluor
).

18
Collisional relaxation is another possible decay
process, which can quench fluorescence.

Collision induced processes are second order (c.f.
fluorescence and predissociation, which follow first
order kinetics):

B + Q A Rate = k
Q
[B][Q]

fluor
, and the measured lifetime of B, will thus depend
on the pressure, i.e. if quenching competes just with
fluorescence


1

= + A k
BA Q
Q [ ]. (1.16)

A plot of
-1
vs [Q] is called a Stern-Volmer plot.

Collisional Relaxation Rates

Collisional energy transfer (CET) is usually most
efficient when transferring small amounts of energy,
c.f. spontaneous emission. Typical CET rates are:


Transition
Type
Energy Separation

/ cm
-1
/ kJ mol
-1
Emission
Lifetime
Fraction of
collisions
that cause CET
Rotational 1

0.012 1 day 1
Vibrationa
l
2000

24 50 s 10
-2

Electronic 20000 240 50 ns (few)

There are no selection rules associated with collisional
energy transfer.
19
LASERS: Basic Laser Equations
A beam of light incident on a sample of a two level
system will excite molecules at a rate:

Rate of absorption = B
AB
[A]()
= rate of loss of photons

If the excited state B is significantly populated, then
stimulated emission will add photons to the beam:

Rate of stimulated emission
= B
BA
[B]()
= rate of gain of photons
Thus:
Net rate of loss of photons
= B
AB
[A]() B
BA
[B]()
= B
AB
() ([A]-[B]) (2.1)
(assuming g
A
= g
B
).

Under normal (equilibrium) conditions [A] > [B], giving
net absorption.

If [B] > [A], a so-called population inversion, the
intensity of the light beam is increased, giving Light
Amplification by Stimulated Emission of Radiation.

20
To obtain laser action, we require some or all of the
following:


A Laser Cavity

A "laser" is normally considered as a light source,
rather than simply providing amplification.
To achieve this, the output of the amplifier is fed back to
its input with mirrors, to make a cavity:



Many passes though the cavity results in a large
amplification. The light has to travel back and forth
along the same path, leading to many of the
characteristic properties of laser light.
21
Properties of Laser Light

1. Low divergence. The light from a laser is very
parallel. Thus:
laser radiation will travel long distances without
spreading useful in LIDAR (see below).
it can be focused to a very small spot, of the order of
the wavelength of the light. This implies small sample
volumes, and high power densities.
2. Coherent, implying a fixed phase relationship between
different parts of the laser beam. This ensures that
applications which rely on interference, like
holography, will work effectively.
3. Monochromatic. The best lasers have a frequency
spread of 1 part in 10
9
, implying very accurate
measurement and resolution of very small splittings.
4. Very short pulses the shortest pulses are of the order
of 10 fs (= 10
14
s), comparable to the time scale of
atoms moving during a reaction (such pulses
necessarily span a wide range of frequencies).
5. High intensities. Tight focusing and a narrow
frequency spread mean that transitions can be excited
very efficiently. Pulsed lasers can provide GW
(10
9
W, the output of several power stations) powers
for short periods. Thus very weak transitions,
including multiphoton transitions, can be excited.
These have different selection rules (like Raman, only
applied to electronic transitions).



22
HeliumNeon Laser




Notes:
This is a simplified energy level diagram, showing
configurations only; a full picture would consider the
term symbols arising from each configuration.
Electron bombardment induces ionisation.
Subsequent recombination of ions and electrons
results in the formation of a wide range of excited
states of neutral He and Ne.
The selection rule for radiative transitions is l = 1.
The 1s2s configuration in He is metastable (i.e. long
lived).
Energy transfer between the He(1s2s) levels and the
Ne(2p
5
5s) states is efficient, as the energy mismatch is
small.
23
Nd:YAG Laser



Notes:
Solid state laser, consisting of Nd
3+
ions doped in a rod
of ytrrium aluminium garnet (YAG).
Flashlamp excitation, plus Q-switching or mode-
locked operation to give short pulse durations.
Fundamental lasing transition is in the infra-red.
Frequency conversion techniques allow this output to
be shifted into the visible and/or ultraviolet.

Frequency Conversion
If an electric field is applied to a substance, e.g. an ionic
crystal, the atoms can move and electron clouds can
distort, generating an electric field. This can be
expressed as a power series in the applied field:

E E E
induced
= + +
1
2
2
L (2.2)


24
where E is the applied field. If we take this to vary as
E
0
cos t then the induced field will contain a component
at twice this input frequency, i.e.

( )
E E t E t
induced
= = +
1
2
0
2 2
1
4
0
2
1 2 cos cos (2.3)

This is second harmonic generation or frequency
doubling. The efficiency depends critically on the
material used and requires high light intensities.

Commonly used non-linear materials include KDP
(KH
2
PO
4
) and BBO (-BaB
2
O
4
), both of which are used
to 'double' the frequency of the output of dye lasers and
Nd:YAG lasers (e.g. 1064 nm 532 nm).

Frequency sum and difference generation is also
possible. In these processes, two different photons (with
frequencies
1
,
2
) are combined to give radiation at the
new frequencies (
1
+
2
) or (
1

2
). Applications
include:

1064 nm (from Nd:YAG) + 532 nm (doubled Nd:YAG) 355 nm
dye laser Nd:YAG (gives 5 m - 195 nm).

It is also possible to split one photon (
p
) into two parts
(
s
and
i
) using Optical Parametric Oscillation such
that
p
=
s
+
i
. The division between
s
and
i
is
variable, so this can give tunable light from a fixed
frequency laser.

25
Excimer Lasers

These use 'molecules' whose ground state is unbound,
but which have strongly bound excited states. The
excited state is an excited dimer (hence excimer) where
the bonding is essentially ionic (e.g. Ar
+
F

).



The excited molecules are formed in an electric
discharge via reactions like:

Ar* + F
2
Ar
+
F

+ F
Ar
+
+ F

+ Ar Ar
+
F

+ Ar
(where Ar* is an electronically excited Ar atom).

Several rare gas halogen mixtures give useful lasing
action; other examples include XeCl (308 nm) and
KrF (248 nm). These lasers typically give a short pulse
of light (30 ns) at repetition rates of 200 Hz.
26
Dye Lasers

To obtain tuneable radiation, we require a lasing
transition that covers a broad spread of wavelengths.
Such is the case for many dye molecules, which contain
long conjugated chains and thus absorb in the visible. A
commonly used dye is Rhodamine 6G, dissolved in a
non-absorbing solvent (e.g. methanol):
N O N
+
O
O
Cl
-
H


Dye lasers are typically pumped by another laser, e.g.
the 532 nm or 355 nm output from an Nd:YAG laser, or
with 308 nm radiation from a XeCl excimer laser. A
selection of ~25 dyes provides coverage of the entire
region 350 - 900 nm.


Not all dyes are useful as laser dyes. As the above
figure shows, for a dye to be useful its fluorescence (and
27
thus lasing) must occur at different (lower) frequencies
than its absorption; this is possible if there is a
significant geometry change between the S
1
and S
0

states.

28
Laser Applications

Laser Spectroscopy
In principle, a laser can be used in a straightforward
absorption experiment, i.e.
Tunable laser Sample Detector
c.f. the teaching lab. spectrometers you have used,
though these use a white light source and select the
required wavelength with a prism or grating.

A laser gives the advantages of higher resolution,
sensitivity, and long path lengths.

Direct absorption measurements like these are limited
by detector noise (particularly for conventional
spectrometers) and fluctuations in the light source
(normally the limiting factor in laser absorption).
Because it is difficult to measure small changes in a
large signal, absorption measurements are rarely the
method of choice for laser spectroscopy. The smallest
change in intensity that can be measured is typically
~10
5
of the incident light.
(But note cavity ring down spectroscopy
http://www.chm.bris.ac.uk/pt/laser/laserhom.htm)

Greater sensitivity can often be obtained by measuring
a side-effect produced by the absorption of light, rather
than the loss of light itself. Such methods can detect
changes equivalent to the absorption of the order of
10
17
of the incident light!
29
Laser Induced Fluorescence (LIF)
The spontaneous fluorescence from excited molecules
can be detected in a very sensitive manner as the signal
off-resonance is very small - the noise level of the best
detectors is just a few photons/sec.

This means only a few molecules need be excited and
concentrations as low as 100 molecules/cm
3
(=10
19
M)
can be detected. This makes LIF a method of choice for
electronic spectroscopy. The technique does have some
limitations, however:
1. Scattered laser light (e.g. from dust particles,
imperfections in windows)
2. Fluorescence may be quenched (recall possible
fates of excited state molecules: (pre)dissociation,
IC, ISC, collisional energy transfer)
3. Absorption and LIF spectra may not be identical.
4. Not useful when spontaneous fluorescence is slow.
Given the
3
factor in A, this means that infra-red
spectra cannot be taken this way.
(IR detectors are also relatively insensitive.)

LIF is well suited for use with small localised samples
(e.g. jet-cooled molecular species, including free
radicals, in a supersonic beam), and for spatially
resolved concentration profiling.
(see http://www.chm.bris.ac.uk/pt/western/cmwspec.html)



30
Laser Identification, Detection and Ranging, LIDAR

Remote sensing of, e.g., pollutant emission from a
chimney is possible by firing a laser into the sky and
collecting light coming back from the laser beam from
fluorescence, the Raman effect or scattering from dust
particles. Analysis of this light enables identification of
the species present, their concentrations and their
positions (from the time delay).

Multiphoton Spectroscopy

In sufficiently intense radiation
fields molecules can absorb more
than one photon at once, giving a
resonance condition E = nh
rather than the usual E = h.
Two photon absorption can be
viewed as a two step process in
which a virtual state comprising a
molecule + a photon is prepared.
This lasts only as long as the
photon takes to pass the molecule.
Absorption then takes place from
the virtual state. Raman
spectroscopy can be viewed in a
similar way, i.e. as (spontaneous)
emission from a virtual state, and
has analogous selection rules.












Virtual State
A
B

Multiphoton processes are inherently improbable. The
table below shows the fraction of molecules excited for a
31
typical, strong n-photon transition for a laser producing
just 5 kW in a 10 ns pulse and focussed to a 1 mm
2
spot.

The n-photon absorption
probability scales as I
n
;
multiphoton processes can
therefore be enhanced by
focusing the light into a small
(wavelength limited) spot.

n Fraction excited

1 0.1
2 10
-10
3 10
-20


Multiphoton Selection Rules
The selection rules for an n-photon transition are
essentially those for n single photon transitions in
succession, giving different overall rules:

1 photon 2 photon
Atoms
l = 1 l = 0, 2
Linear
Molecules
= 0, 1 = 0, 1, 2
(Linear)
Molecules
u g , u u
/
,
g g
/

g g , u u ,
g u
/

Character
Tables
Look for x, y, z x
2
, y
2
, z
2
, xy, yz, xz

The S = 0 rule is unchanged for multiphoton
transitions.


Multiphoton Ionisation (MPI)

32
Having absorbed two or more UV photons, a molecule
will usually have a total energy such that absorption of
one further photon will excite the molecule above its
ionisation limit. Ionisation of the molecules in excited
state B with further photons from the same laser pulse
thus becomes probable.
Ion detection is easy and sensitive, thus MPI is the
preferred method of monitoring multiphoton
absorption. The ionisation step is usually independent of
wavelength, so the spectrum reflects the initial,
resonant, multiphoton absorption step only. Hence the
name resonance enhanced multiphoton ionisation, or
REMPI. When used with pulsed lasers, REMPI can
give mass selectivity simply by measuring the arrival
time of ions at a negatively biased detector ~20100 cm
distant from the interaction region.

Advantages Disadvantages
Different selection rules
allows observation of a
wider range of electronic
states
Needs high power,
tuneable lasers
High energy states
accessible using just
visible/UV photons
High intensities
required can cause
other, unintended
processes in the sample
Can be mass selective Less sensitive than
single photon methods
33
Raman Spectroscopy

Standard spectroscopies depend on a resonance
between the energy of the photon (h) and the energy
change within the molecule, i.e. E = h. However, any
electric field applied to a molecule will cause a
redistribution of the electron density so as to induce a
temporary dipole in the molecule. Such induced dipoles
are generally proportional to the applied electric field, .
= (2.4)
The proportionality constant, , is the polarizability.

Electromagnetic radiation, which contains an oscillating
field, will thus induce an oscillating dipole in the
molecule.
The Raman effect arises because changes as the
molecule rotates or vibrates. The frequencies of the
molecular motion and that of the electric field beat
together and radiation can be observed at the sum and
difference frequencies.
(Recall cosAcosB = {cos(AB)+cos(A+B)})

Quantum mechanically, we
think of the combination of
a molecule and a photon
(that is not being absorbed)
as a virtual state. Its energy
is equal to that of the
molecule plus that of the
photon:


Virtual State
E=h


34
The lifetime of this virtual state is just the time it takes
for the photon to pass by the molecule. Nonetheless, the
virtual state still has many of the properties of a real
level. For example, it can emit though the complete
absorption emission process is better envisaged as an
(in)elastic scattering of the photon by the molecule.
This is the Raman effect.
Any difference in the
energies of the incident
and scattered photons
must match an energy
separation between levels
of the molecule.

Virtual State
E=h
E=h



The short lifetime of the virtual state means that the
Raman effect is very weak; its observation requires an
intense, monochromatic light source (i.e. a laser) and a
monochromator that allows effective separation of the
weak Raman shifted radiation from the incident light.

Raman Selection Rules

In order that a molecule shows a rotational Raman
spectrum, its polarizability must be sensitive to its
orientation. This will be true for all molecules except
spherical tops. Raman spectroscopy is thus useful for
determining structural parameters for centrosymmetric
molecules like H
2
, O
2
and N
2
that are not amenable to
study by microwave spectroscopy (which requires the
molecule to have a permanent dipole moment).

35
The selection rules for allowed Raman transitions can be
obtained by considering the Raman effect as two
successive electric dipole transitions. Two successive
J = +1 transitions gives J = +2 overall (a) below.
Such Raman scattered light has lower frequency than the
exciting light. Such red-shifted light is known as Stokes
radiation. The reverse (b) is also possible:

(a)
J
J+1
J+2
(b)
J
J1
J2


The resulting radiation is of higher frequency than the
exciting line, and is known as anti-Stokes radiation. The
nett selection rule for a rotational Raman spectrum is
thus J = 0, 2, giving a Raman spectrum like:

Frequency
Stokes Lines
Anti-Stokes lines
Rayleigh Line

36
The intense central feature, the Rayleigh line, typically
dominates the spectrum. This light has been scattered by
the molecule but not changed in frequency.

We can apply similar logic to determine the selection
rules for allowed vibrational Raman transitions. For any
given molecule, the allowed one-photon (infrared)
transitions transform in the same way as x y and z in the
relevant character table. Allowed vibrational Raman
transitions must have the symmetry of one of the possible
combinations (products) of x y and z, i.e. xy, xz, yz, x
2
, y
2

and z
2
.
This leads to the mutual exclusion principle which states
that, for molecules with a centre of symmetry, no
vibrational modes can be both IR and Raman active.
(This follows because x, y and z always have u symmetry
so their products must always have g symmetry. The
analogy with the selection rules for 2-photon absorption
should be obvious.)

Additional information can be gleaned from Raman
spectra excited with plane (or linearly) polarized light, i.e.
with oscillating in one plane. (Most laser radiation is
plane polarised).
Features in Raman spectra that show the same
polarization as the exciting light are associated with
totally symmetric vibrational modes.
All other Raman active vibrations show (at least some)
loss of polarization, or depolarization.
37
PHOTODISSOCIATION AND PREDISSOCIATION

What can we learn about photofragmentation and how can we measure it?

Observables Experimental Methods

Frequency of absorption Absorption Spectroscopy
(cavity ring down spectroscopy,
laser induced fluorescence (LIF)
multiphoton ionisation (MPI)).
Excitation spectra Photofragment excitation methods.
Excited state lifetimes Linewidths in absorption, fluorescence decay,
ultrafast lasers.
Product yields and Nascent product detection (LIF, MPI), imaging,
branching ratios translational spectroscopy methods.
Product vibration/rotation (as for product yields)
Product translation Translational spectroscopy (inc. Doppler profiles).
Recoil anisotropy Ion imaging, Doppler profiles.
Rates of product formation Ultrafast methods.
38
PRODUCT YIELDS AND BRANCHING RATIOS:

Translational spectroscopy.
(http://www.chm.bris.ac.uk/pt/laser/mike.htm)

If a diatomic AB is dissociated as a result of absorption
of photon of frequency , energy conservation requires:

h + E
int
(AB) = D
0
(AB) + E
int
(A) + E
int
(B) + E
trans
(A) + E
trans
(B)
(3.1)

where D
0
(AB) = AB bond dissociation energy
E
int
(AB) = internal excitation of AB parent molecule
E
int
(A,B) = internal (electronic) excitation of A,B products
E
trans
(A,B) = translational energy of A,B products.

For experiments employing a jet-cooled molecular beam of
parent molecules, E
int
(AB) ~ 0.

The translational (kinetic) energies of A and B are
related by momentum conservation (in centre-of-mass
(cm) frame)

m
A
v
A
= m
B
v
B


(3.2)

Thus the speed of A (in the cm frame) will be given by:

v
A
=
( )
2
0
2
1 2
[ ( ) ( ) ( )]
int int
/
h D A B E A E B
m m m
A B A

+

(3.3)

{CHECK YOU CAN DERIVE THIS!}
39
If we carry out pulsed photolysis (with a laser) and
measure the time, t
A
, it takes for product A to travel a
known distance, d, from the photolysis volume to a
detector:













we can derive information about D
0
(AB) and the
internal energy distribution in the partner fragment B.


40
Example: Near UV photolysis of HI into H + I at 266 nm.

The I atom can be formed in either its ground (
2
P
3/2
) or
spin-orbit excited (
2
P
1/2
) state. These have different
energies. The partner H atoms thus have different
recoil velocities:




H + I(
2
P
1/2
)












H + I(
2
P
3/2
)



41
Early time peak is associated with H + I(
2
P
3/2
)
products v
H
= 17.6 km/s.
Analysis gives D
0
(HI) = 24630 20 cm
-1
.
Note that relative areas of the fast and slow peaks
vary with , the angle between the electric vector of
the photolysis light and the axis along which the
recoiling H atoms are detected.
Analysis indicates that ~52% of the I atoms formed in
the 266 nm photolysis of HI are formed in their
excited (
2
P
1/2
) spin-orbit state.

Doppler spectroscopy can yield the same information.
An H atom (or any atom or molecule) at rest will absorb
radiation at a frequency
0
corresponding to the line
centre of the transition between two of its quantum
states. The effect of motion along the propagation axis
of the probe radiation, z, (with velocity component v
z
) is
to shift the frequency of absorption to a new frequency
given by:

=
0
1
|
\

|
.
|
v
c
z
. (3.4)

The +/ corresponds to motion respectively parallel and
anti-parallel to the propagation direction of the probe
radiation. =
0
=
0
v
z
/c is called the Doppler
shift (recall p. 9)




42

Doppler profile for H atoms from the 266 nm photolysis of HI

The Doppler profile of the H atoms resulting from HI
photolysis at 266 nm can be divided into two
components, one wider (i.e. bigger Doppler shift,
faster, associated with H + I(
2
P
3/2
) products) than the
other (associated with H + I(
2
P
1/2
) products).

Note: the two profiles have different shapes.

This, like the observation that the relative intensities of
the two peaks in the H atom TOF spectrum depends on
the relative orientation of the laser polarisation axis and
the detector axis, reflects the fact that the distribution of
fragment recoil velocities is anisotropic.

43
Anisotropy Parameters

Consider fragmentation of a diatomic molecule, e.g. HI.

Assume axial recoil, i.e. that atoms recoil along a
direction parallel to the breaking bond.
(When might this assumption fail?)

For a diatomic, the transition moment must either lie
parallel ( = 0) or perpendicular ( = 1) to the bond.

The preferred direction of recoil is thus either parallel
or perpendicular to .

Recalling eq. 1.3, the absorption probability depends on
the dot product of E
0
and (i.e. the projection of E
0
onto
). Thus there will be a spatial correlation between the
fragment recoil velocity, v, and E
0
.

The measured (LAB) frame distribution of recoil
velocities about E
0
is obtained by averaging the square
of the projection of E
0
onto

over the (random in space)
distribution of .

44
This leads to:

P(v,) =
f v
v
( )
4
2

{1 +

2
(3cos
2
1)}. (3.5)

is the angle between E
0
and v,
f(v) is the speed distribution, and
is the anisotropy parameter. lies in the range +2 to -1.

If = +2, P(v,) cos
2
parallel transition
= 1, P(v,) sin
2
perpendicular transition


Fragment recoil velocity distributions characterised by
intermediate values of can arise if:
axial recoil approximation is invalid
(pre)dissociation occurs over a timescale long
compared with the rotational period
overlapping parallel and perpendicular absorption.

45
MEASURING VALUES

The most direct methods use a position sensitive
detector to record a 2-D projection of the 3-D velocity
distribution; the full 3-D distribution can be
reconstructed using a mathematical transform.

Photofragment Ion Imaging involves:
(i) photolysis using linearly polarised laser pulse.
(ii) ionisation of (quantum state selected) fragment of
interest, by REMPI, within a few nanoseconds of fragment
creation.
(iii) the resulting ion cloud expands (with a speed and
angular distribution characteristic of the dissociation
event), and is simultaneously accelerated (by an electric
field) so as to impact on the position sensitive detector.


Schematic diagram of a position sensitive detector

46
Image of the H atoms from 266 nm photolysis of HI:

The same recoil anisotropy also reveals itself in the
different Doppler profiles of the fast and slow H atoms
arising in this photolysis.

Idealised Doppler profiles for fragments following photodissociation
with different values of . The forms shown are appropriate for colinear
photolysis and probe laser beams.
47
Both sets of results can be understood in terms of
excitation to the repulsive potential energy curves
associated with the various excited states of HI that
arise from the electron promotion:

*(LUMO)
n.b.
(HOMO).

The presence of the heavy I atom means that and S
are not good quantum numbers, but that is.

The experiments reveal that photo-excitation of HI
molecules at 266 nm results in population of two excited
states: one with = 1, that fragments to ground state H
+ I(
2
P
3/2
) products; the other with = 0, that yields
spin-orbit excited products H + I(
2
P
1/2
).



48
OZONE PHOTOCHEMISTRY

Pivotal to stratospheric chemistry is the creation and
destruction of the ozone layer, which protects the
Earth's surface and everything on it from all harmful
solar radiation in the wavelength range 240 - 310 nm.
(AJO-E lectures, Level 2)

O
3
is produced by O
2
photolysis at < 242 nm:

O
2
+ h 2 O (4.1)

O + O
2
+ M O
3
+ M (4.2)

[O
3
] is limited by two further reactions:

O
3
+ h O + O
2
(4.3)

O + O
3
2 O
2
(4.4)

(4.1) - (4.4) constitute the Chapman scheme.

Species like Cl, NO, H, OH (henceforth X) provide an
alternative path for reaction (4.4):

X + O
3
XO + O
2
(4.5)

XO + O X + O
2
(4.6)

X acts as a catalyst.

49
What do we know about ozone photochemistry?

The electronic spectrum of O
3
shows broad intense
absorption in the UV (210-305 nm, the Hartley band)
and weak structured absorption in the region 305-345
nm (Huggins band). Both are shown below, the latter
on much expanded horizontal and vertical scales.




Both absorptions are now thought to be part of the same
1
B
2

~
X
1
A
1
transition, involving a short lived excited
state whose equilibrium geometry involves two OO
bonds of different length (i.e. not strictly C
2v
symmetry.

50
What states of O and O
2
are formed when O
3
absorbs a
UV photon and dissociates?

Time-of-flight (TOF) spectroscopy of the O atom
products resulting from O
3
photolysis at 266 nm
shows:

O
3
+ h(266 nm) O(
3
P) + O
2
( X
3
); ~ 0.1 (4.3a)
O(
1
D) + O
2
( a
1
); ~ 0.9 (4.3b)

where = quantum yield (i.e. the number of reactant
molecules producing specified primary products for
each absorbed photon, recall p. 17).



Fragment kinetic energy / kcal mole
-1
51
Note that both (4.3a) and (4.3b) are spin-allowed if we
assume that dissociation occurs from a singlet excited
state. The alternative possibilities, i.e.

O
3
+ h O(
3
P) + O
2
( a
1
) (4.3c)
O(
1
D) + O
2
(X
3
), (4.3d)

though both energetically allowed following UV
excitation, would be spin-forbidden.

__________________________

Ozone is topical because of:
the well documented Antarctic ozone hole and the
general thinning of the ozone layer due to Cl catalysed
variants of (4.5) and (4.6), and
increasing levels of ground level O
3
associated with
photochemical smog.

At the more fundamental level there have been two
longstanding questions regarding ozone concentrations
and its role in the atmosphere.

52
The 'Ozone Deficit' Problem

Notwithstanding ozone depletion via (4.5) and (4.6),
current models predict lower O
3
concentrations in the
upper stratosphere (> 40 km) than is actually observed.
Is something missing in the Chapman scheme?
Some of the O
2
(X
3
) fragments formed via (4.3a) when
O
3
is photolysed at < 226 nm are formed with so much
vibrational energy (v 26) that the reverse of (4.4) can
lead to re-generation of O
3
.

O
2
( X
3
, v 26) + O
2
O
3
+ O (4.7)




Including this reaction in the model leads to ~10%
increase in the predicted tropical stratospheric ozone.
53
Tropospheric Ozone

Ozone in the troposphere is a pollutant and one of the
major components of photochemical smog.

Tropospheric ozone arises primarily from NO
2

photolysis:

NO
2
+ h ( < 400 nm) NO + O(
3
P) (4.8)

O + O
2
+ M O
3
+ M (4.9)

NO + O
3
NO
2
+ O
2
(4.10)

Invoking the steady state approximation to both O
atoms and O
3
and solving for [O
3
] yields:

[O
3
] =
k NO
k NO
4 8 2
4 10
.
.
[ ]
[ ]


(Check you can do this!)


54
What happens to O
3
in the troposphere?
It will absorb that fraction of the solar flux that reaches
down through the ozone layer (in stratosphere) and
photodissociate via the various forms of reaction (4.3).

Ground state O(
3
P) atoms are rather unreactive.

Electronically excited O(
1
D) atoms, in contrast, are
highly reactive. O(
1
D) atom reactions control the
formation of tropospheric OH radicals via, e.g.

O(
1
D) + H
2
O 2 OH. (4.11)

Reactions involving the OH radical drive most of the
chemistry occurring in the troposphere.

The O(
1
D) production rate depends on:
the solar flux,
the ozone absorption cross-section,
the O(
1
D) quantum yield.

Each process has a strong wavelength dependence.

55
How does (O(
1
D)) vary with wavelength?
We have already seen that it has a value of ~0.9 at
= 266 nm, and that the partner fragment is the spin-
allowed product O
2
( a
1
) - eq. (4.3b).

(O(
1
D)) the vertical axis of the plot below has a
value of ~0.9 at all photolysis wavelengths 310 nm,
which corresponds to the energetic threshold for
reaction (4.3b).


At longer wavelengths (O(
1
D)) falls dramatically, but
not to zero as NASA recommend (1994) in their
database for stratospheric ozone modelling. Why?
56
Recent experiments have identified two hitherto
unrecognised contributions to the O(
1
D) product yield
following excitation of O
3
at > 310 nm.

(i) Absorption by internally (vibrationally) excited O
3

molecules and dissociation via (4.3b). The Boltzmann
equation ensures that population of these vibrationally
excited levels is low, but this is partially compensated by
more favourable Franck-Condon factors.

(ii) Photo-excitation of O
3
molecules in their ground (v=0)
level, and subsequent predissociation on a triplet excited
state surface leading to O(
1
D) + O
2
(X
3
) - rxn. (4.3d).
Being spin-forbidden, the transition probability is much
lower, but this is compensated in part by the fact that the
v = 0 level of O
3
is the most populated.


The nett effect of proper inclusion of the effects of the
absorption tail at long wavelength has been to double the
calculated daytime tropospheric [OH] (~10
6
cm
-3
) !
57
FEMTOCHEMISTRY

Questions.

On what timescale does a molecule dissociate once it
has absorbed a photon and been excited to an unstable
electronically excited state?

How does the molecular geometry distort during this
dissociation process?

Can we use some property of a light pulse (frequency,
phase) to control the way in which a molecule
fragments, or a reaction proceeds?

We can obtain an idea of the timescales by assuming
that the evolution from excited molecule to products
involves the internuclear separations changing by ~ 0.2
nm, and that the atoms are moving at ~ 10
3
m s
-1
, i.e. the
process takes ~ 10
-13
s.

To investigate the evolution from reactants to products
we need to be able to interrogate the system several
times within this period, i.e. we require femtosecond
(10
-15
s) time resolution.

From the energy-time form of the Uncertainty Principle

E.t ~ h .t ~ 1/2 . (1.14 again)

A 100 fs laser pulse will have a frequency uncertainty
(full width half maximum) of (at least) 1.6 x 10
12
s
-1
, or
>52 cm
-1
. It is only 30 m long, and very delicate!
58
When thinking about molecular excitations using such
pulses it is customary to use time dependent quantum
mechanics, and to think in terms of wavepackets - a
superposition of stationary state wavefunctions with a
well defined initial phase relationship to each other.

It is simplest to concentrate on unimolecular processes,
e.g. photodissociation. We could answer the posed
questions, in principle, by exciting the molecules with a
very short pulse of light and looking at the emission
from molecules in the act of dissociation.

Wave packet propagation describing photofragmentation. The
laser photon transfers the ground state wavefunction to the excited
state at t
0
. During fragmentation, an excited ABC molecule could
fluoresce to various vibrational levels of the ground state. The
spectrum of this emission (bottom left) consists of a series of
broadened bands spaced according to the ground state
vibrational levels.
59
Unfortunately, the quantum yield of such emission

= k
emission
/k
dissociation
~ 10
7
s
-1
/ 10
13
s
-1
~ 10
-6



(recall eq. 1.15) is so low as to be barely detectable.

It would be much better to use two short laser pulses,
one to excite the parent molecule, the second (after a
well defined time delay) to probe loss of the parent
and/or product build up. Pump-probe experiments of
this type are now performed. Consider two examples:

1. ICN + h (~307 nm) I + CN

2. NaI + h (~310 nm) Na + I

(The monitored species in each case is underlined).

Both studies employed ~40 fs pump and probe laser pulses.
The excitation bandwidth is ~1.2 nm (~130 cm
-1
FWHM).
60




Femtosecond dynamics of ICN dissociation.
One fs laser pulse is used to prepare a wavepacket of ICN* molecules on
the excited state potential energy surface at t
0
. The build up of CN(X)
fragments is detected by laser induced fluorescence as a function of
time delay using a second fs laser pulse tuned to resonance with the
CN(B-X) transition. The build-up time so derived (205 30 fs) gives a
measure of how long it takes for the CN fragments to be essentially
independent of the partner I atom.
61


Femtosecond dynamics of NaI dissociation.
Bottom: experimental observations of wave packet motion, made by
detecting the activated [NaI]*
#
complexes or the free Na atoms. Top:
PE curves (left) and QM calculations (right) showing the wavepacket as
it evolves in time and space. The corresponding changes in bond
character are also noted: covalent (at 160 fs), covalent/ionic (at 500 fs)
and back to covalent (at 1.3 ps).
( from Zewail, J. Phys. Chem. 100, 12701, (1996)).
62
A more involved example -
The fragmentation of gas phase Fe(CO)
5
molecules.

Fe(CO)
5
Fe + 5CO.

Only these 'asymptotic' products are detected in a
traditional experiment involving nanosecond laser
pulses. However, the detailed evolution of the various
bond fissions, and thus the mechanism, can be studied
by femtosecond pump-probe spectroscopy.
Laser 1 (UV) is used to excite the Fe(CO)
5
molecules.
Laser 2 probes (by ionisation) the subsequent build up
and decay of the various products, in real time.



Two colour (400 nm/800 nm) fs transient ionisation spectra of the
Fe(CO)
5
parent molecule and the various Fe(CO)
n
(n < 5) fragments.
The inset shows the expanded transients around time delay zero for
species with 3 n 5, with estimated time delays indicated.
63
Controlling Chemical Reactions; 'Coherent Control'

The aim of this very new area of gas phase chemistry is
to take a molecule and, by judicious choice of the
frequency and/or phase of the exciting light pulse,
control what products are formed.
e.g. multiphoton absorption and ionisation using 80 fs,
800 nm pulses.

Fe(CO)
5
Fe
+
, FeCO
+
, Fe(CO)
2
+
, Fe(CO)
3
+
, .. etc

The [Fe
+
]:[Fe(CO)
5
+
] ratio in the product ions can be
varied by ~70, just by changing (with a feedback
controlled evolutionary algorithm) the distribution of
amplitudes and phases within the exciting laser pulse.


Fs pulses are modified in a computer controlled pulse shaper. The pulse
is expanded (in area) and split into 128 component. Each passes
through a different pixel in a liquid crystal spatial modulator. The
refractive index of each pixel is computer controlled, thereby
introducing selected phase delays into each component. The various
components are then recombined and directed onto the sample which
is in the form of a molecular beam. Ionic fragments from the molecular
dissociation are recorded with a mass spectrometer. This signal is used
directly as feed back in the controlling evolutionary algorithm to
optimise the product branching ratios.
64
A more chemical example:
Acetophenone photolysis by multiphoton excitation with
50 fs 800 nm laser pulses.
(Levis et al, Science 292, 709 (2001)).


A: Mass spectrum of ions following MPI of acetophenone using 50 fs
800 nm transform limited laser pulses.
B: Relative yields of C
6
H
5
CO
+
and C
6
H
5
+
ions, and their optimised ratio.
C: Optimisation of the inverse ratio (N.B. the C
6
H
5
COCH
3
bond is ~15
kJ mol
-1
stronger than the C
6
H
5
COCH
3
bond.
D: Optimisation of C
6
H
5
CH
3
+
yield (a dissociative rearrangement).

65
Coherent control of chemical reactions can be
understood (qualitatively) as follows:
The pulse shaper modifies the time varying electric field
of the laser pulse (both its phase and its frequency
distribution) iteratively, in order that the amplitudes of
the different interfering vibrational modes in the excited
molecule add up in a given bond (or set of bonds) at
some well-defined time after the photo-excitation event.
This results in rupture of that particular bond (or
promotes the sought after molecular rearrangement).
66

Level 3 Photochemistry Course

Prof. M.N.R. Ashfold

Further reading:

Einstein coefficients, absorption probabilities, fates of
electronic excitation, Jablonski diagram:
P.W. Atkins and J. de Paula, 'Physical Chemistry', 7th edn.
16.3; 17.1-17.4.

Lasers and their applications:
D.L. Andrews, 'Lasers in Chemistry'.
P.W. Atkins and J. de Paula, 'Physical Chemistry', 7th edn.;
17.5, 17.6.

Ozone Photochemistry:
General: M.J. Pilling and P.W. Seakins, 'Reaction Kinetics' 8.4
Ozone deficit: Science, 265, 1831-8 (1994)
Tropospheric ozone: Science, 280, 60 (1998)

Femtochemistry:
'The Chemical Bond, Structure and Dynamics' (ed. A.H. Zewail)
Chap. 9
A.H. Zewail, J. Phys. Chem. 100, 12701 (1996)
Advances in Chemical Physics, Vol. 101, (1997)
R.J. Levis et al, Science, 292, 709, (2001)