Vibrational Raman Scattering

Carl Hall
PHYS 251b
Final Paper
April 18, 2012
Introduction
The study of combustion science is fraught with diculties, particularly with the availability of
experimental measurement techniques. One of the challenges is to perform a true measurement
of a reacting gas ownot just measure the perturbation applied by the measurement technique.
For many ows of interest, this requires that no physical probes are inserted into the ow eld.
Non-intrusive laser-based techniques are a common alternative, and one powerful technique is
spontaneous Raman scattering. The Raman scattering phenomenon is strongly amenable to exper-
imental implementation, and allows the researcher to use one laser to measure multiple chemical
species concentrations simultaneously. Additionally, the level of quantum mechanical theory re-
quired understand this phenomenon was reached in this course.
Raman scattering is the inelastic scattering of electromagnetic radiation from the electron cloud
of a chemical species. The incident and scattered light dier by an energy shift, due to the gain
(or loss) of energy to the molecule. To limit the scope of this discussion, the type of chemical
species considered is a diatomic molecule, and only the vibrational energy storage mode is inves-
tigated. This simplies the Raman shift to the dierence between vibrational energy levels, as
shown schematically in Figure 1. Also shown in the gure is the elastic analog to Ramantermed
Rayleigh scattering.
The analysis of Raman scattering will rst be examined using classical theory, followed by a
quantum mechanical treatment. And to conclude, the experimental attributes of this particular
technique are discussed.
Classical Treatment
The scattered light is predominately electric dipole emission induced by the incident electric eld.
A schematic of the radiation anisotropy and coordinate system is shown in Figure 2. Classically,
the dipole radiation from a point source can be expressed as
I

() | p|
2

4
o
sin
2
(1)
where p is the dipole moment,
o
is the incident light frequency, and I

is the irradiance in units

of power per solid angle. The dipole moment is dened as the distribution of charge in the electron
1
Figure 1: Simplied schematic of vibrational Raman and Rayleigh scattering [3]
cloud,
p =

i
e
i
r
i
(2)
where the summation is over all electrons. Dipole radiation emission is not uniform with azimuthal
angle and is maximized for an angle of

2
, as shown in Figure 2. Calculation of the amount of
scattered light is then reduced to evaluating the square of the induced dipole moment.
A more tractable form of the electric dipole is expressed in terms of the polarizability of the
electron cloud,
p =
o

E (3)
where is a second rank tensor. For a symmetric diatomic molecule, the polarizability is shown
schematically in Figure 3.
To simplify the vector problem, we consider only light polarized along z,
p
z
=
o

zz
E
z
. (4)
Additionally, the behavior of the polarizability for small vibrations (as shown in Figure 3) suggests
a rst-order Taylor series expansion is reasonable,

zz

o
+
_

Q
_
o
Q+O
_
Q
2
_
. (5)
2
Figure 2: Distribution of radiation emission from an oscillating electric dipole [3]
where Q is the nuclear coordinate. The forms of the nuclear coordinate and incident electric eld
can be simplied to,
Q = Q
o
cos (
v
t) (6)
E
z
= E
o
cos (
o
t) (7)
where
v
is the harmonic oscillator frequency. Combining the expressions, the variation of the
induced dipole can be expressed in terms of a modulated electric eld amplitude,
p
_

o
cos (
o
t) +
Qo
2
_

Q
_
o
[cos ((
o

v
) t) + cos ((
o
+
v
) t)]
_

o
E
o
. (8)
The rst term is Rayleigh scattering, and is dependent on the mean polarizability. This ra-
diation is at the frequency of the incident light,
o
, and is hence elastically scattered light. The
second two terms describe Raman scattering, and dier in frequency from Rayleigh radiation by
the vibrational frequency,
v
. The lower frequency component is termed Stokes radiation, and
is where the incident energy promotes the molecule to a higher vibrational state. The higher fre-
quency component is termed Anti-Stokes, and is where the molecule loses energy to the emitted
photon.
The dependency of both Raman and Rayleigh scattering on the polarazability is clear: Rayleigh
is dependent on the mean polarazibility, and Raman on the polarazability gradient. This has the
implication that chemical species with particular symmetries may not be Raman active, but that
all species are Rayleigh active.
Quantum Mechanical Treatment
Time dependent perturbation theory is commonly used to solve Raman scattering problems. The
perturbation Hamiltonian is the induced electric dipole,
H

= p

E (9)
3
Figure 3: Schematic of the polarizability of a vibrating symmetric diatomic molecule [1]
where p is now the dipole operator. After expanding the perturbed wavefunctions, the dipole
transition becomes,
p
nm
= p
0
nm
+p
1
nm
+. . . (10)
where,
p
0
nm
=

0
n
| p|
0
m
_
(11)
p
1
nm
=

1
n
| p|
0
m
_
+

0
n
| p|
1
m
_
. (12)
Only up to rst order is kept for simplicity. The rst order correction describes Raman scattering
[2], and can be evaluated as,
p
1
nm
=
1

r
_
E
o
p
0
nr
p
0
rm

rm
+
0
+
p
0
nr
E
o
p
0
rm

rm

0
_
e
i(
0

rf
t)
+
1

r
_
E
o
p
0
nr
p
0
rm

rm

0
+
p
0
nr
E
o
p
0
rm

rm
+
0
_
e
i(
0
+
rf
t)
(13)
where the index r refers to virtual states. For one particular virtual state there exists four terms,
only two of which apply to Raman scattering (Stokes and Anti-Stokes). Evaluating the electric
dipole transition integrals is the most dicult part of this calculation. The crux of theoretical
Raman scattering is application of reasonable assumptions to force the problem to be solvable.
Using the alternative form of the dipole operator, the integrals may be expressed in terms of
the polarizability,
p
mn
=
0

n
|

E|
m
_
. (14)
4
Using the common assumption that the wavelength of incident light is much longer than the length
scale of the molecule, the electric eld may be treated as constant over the integration domain,
p
mn
=
0

n
||
m

E. (15)
Now the problem is reduced to simplication of the polarizability into a reasonable problem.
Relaxing the assumption of a xed angular orientation of the molecule made earlier, the func-
tional form of the wavefunction includes the nuclear coordinate as well as rotational coordinates.
Using the knowledge that rotational frequencies are much higher than the vibrational frequencies,
the vibration movement may be separated from the rotational ne structure. This adiabatic
assumption then allows the separation of rotational and vibrational coordinates,
(Q, , ) = V (Q) R(, ) (16)
which transforms the integral into,
p
mn
=
0
V
n
| R
n
||R
m
|V
m

E (17)
where V is the vibrational (harmonic oscillator) wavefunction, and R is the rotational wavefunction.
Further extending the adiabatic assumption, the behavior of the electron distribution may be
approximated as dependant locally on the nuclear positions only. This implies that the polaraz-
ability may be separated into an absolute distribution (based on the nuclear coordinate) and a
rotational-orientation factor,
(Q, , ) = (Q) f (, ) . (18)
This assumption was developed by Platzek and others, and greatly simplies the integration into,
p
mn
=
0
V
n
||V
m
R
n
|f|R
m

E (19)
=
0
V
n
||V
m
f
nm

E (20)
where the induced dipole is now a clear vibrational eect, with a rotational ne-structurecomponent.
Substituting the Taylor series expansion for polarizability, the dipole can then be reduced to,
p
mn
=
0

V
n
|
0
+
_

r
_
0
r|V
m
_
f
nm

E (21)
p
mn
=
0

0
V
n
|V
m
f
nm

E +
0
_

Q
_
0
V
n
|Q|V
m
f
nm

E. (22)
From this form the selection rules can be observed. For Rayleigh scattering (from the rst term),
the initial and nal states must be the same, supporting the known elastic behavior of the transition.
For Stokes scattering (the second term), the selection rules for harmonic oscillator wavefunctions
imposes that the dierence in nal and initial states must be 1.
From knowledge of the Boltzmann distribution for a gas, the population of molecules at partic-
ular vibrational energy levels drops o exponentially with increasing vibrational quantum number.
This implies that the Stokes Raman lines (that may start at the ground level) are stronger than
the Anti-Stokes Raman line (that may start at the rst vibrational level).
5
Figure 4: Typical experimental setup for Raman scattering measurements [3]
Experimental Application
Experimentally, the input quantity is the energy of the laser pulse(s) and the detected quantity is
the number of photons emitted. The typical experimental setup is shown in Figure 4. To relate the
pulse energy to the scattered photons, the energy is rst expressed in terms of the Area Irradiance,
e
i
=
_
P dt =
__
I
A
dAdt = I
A
tA (23)
where the duration and cross-sectional geometry of the pulse are kept constant through the exper-
iment. This, with the known form of Irradiance, allows the relation of laser pulse energy to the
electric eld amplitude,
e
i
|

E
i
|
2
(24)
The number of scattered photons can be related to the scattered Irradiance,
n
p
= N
n
e
o

o
=
N
n

__
I

ddt (25)
which has the known form,
I

() = N
n
| p|
2

4
o
sin
2
(26)
Combining equations 25 and 26, the expression for the number of photons emitted is,
n
p
= N
n
1

o
_
dt
_
d| p|
2

4
o
sin
2
(27)
N
n
1

o
t| p|
2

4
o
_
dsin
2
. (28)
6
Lumping all the geometric and time factors into an experimental calibration factor, the number of
photons detected is proportional to,
n
p
N
n
|
n
||
m
|
2
e
i

3
o
. (29)
This shows that the signal collected depends linearly on the number of molecules in the measurement
zone (number density), linearly with incident energy, and scales to the third power of incident laser
frequency. These qualities are experimentally very convenient, as linear signal dependencies greatly
simplify data reduction, and that there is a strong advantage for using shorter wavelength laser
sources.
Conclusion
The strong advantage of using Raman scattering as a measurement technique lies in the frequency
shift. The Raman signals are at dierent frequencies compared to the incident light, greatly re-
laxing the care needed to avoid reections from windows and surfaces near the measurement zone.
Additionally, the shift is dependant on the structure of harmonic oscillator energy levels. This
implies that the Raman signals from dierent molecules will lie at dierent frequencies, allowing
the use of one excitation laser to probe multiple chemical species.
References
[1] Long, D. A., The Raman Eect: A Unied Treatment of the Theory of Raman Scattering by
Molocules, Bradford: John Wiley & Sons, 2002.
[2] Long, D. A., Raman Spectroscopy, New York: McGraw-Hill, 1977.
[3] Eckbreth, A. C., Laser Diagnostics for Combustion Temperature and Species, Gordon and
Breach Publishers, 1996.