6.

Spectroscopy fundamentals

This chapter provides a broad overview of the spectroscopic principles required in order
to perform quantitative spectroscopy of atmospheres and to couple the details of the
atmospheric spectroscopy with the radiative transfer processes and also with the
assessment of spectroscopic measurements of atmospheres.

6.1 Einstein A and B coefficients

These provide the relationship between spontaneous

emission by atoms or molecules that are in quantum states
above the ground state, including the thermal emission by
atmospheres (Einstein A coefficient) and induced absorption
and emission, (Einstein B coefficient). The blackbody flux
Figure 6.1. A 2 quantum-
density (eq. 4.7, converted to emitted photons) level system at equilibrium
2π cσ 3dσ where, by definition, the
is Rn (σ ) dσ = c σ /T , and the radiation density (eq. 4.8) number of transitions in
e 2 −1 absorption equals the
8π hcσ 3dσ number in emission.
is ρ (σ ) dσ = c σ /T .
e 2
−1
Consider a macroscopic two quantum-level system, perhaps a volume of a simple gas, at
equilibrium as shown in Figure 6.1. At equilibrium, the number of molecules in each of
the two states does not change macroscopically, although individual molecules may
change state at any time. The populations in each state are determined by Boltzmann
statistics:

−c2σ /T
N1 = N 0 e . (6.1)

At equilibrium, dN1 / dT = dN 2 / dT = 0 , implying

N 0 ρ (σ ) B01 = N1ρ (σ ) B10 + N1 A10 . (6.2)

Applying the Boltzmann factor
N0
ρ (σ ) B01 = ρ (σ )ec2σ / T B01 = ρ (σ ) B10 + A10 , (6.3)
N1
or
A10 8π hcσ 3
ρ (σ ) = = . (6.4)
B01ec2σ /T − B10 ec2σ /T − 1
Therefore
B01 = B10 (6.5a)
A10 = 8π hcσ B10 .3
(6.5b)

This establishes the ratios among the absorption and emission rates for given radiation
densities. Absolute values are given in Section 6.2.

1
Note the very strong wavenumber/frequency σ 3 (or ν 3 ) dependence of A10; emission at
short wavelengths is very fast, and the upper states have very short lifetimes (and broad
line shapes (Section 7.2).

If the upper and lower state degeneracies are different, A10 = 8π hcσ 3 B10 as before, but
g0 B01 = g1B10
.

6.2 Rotational spectroscopy

6.2.1 Diatomic molecules

Figure 6.2. Schematic of a diatomic molecule with internuclear distance r and atomic
masses m1 and m2.

The moment of inertia I for rotation about the center of mass is given by

I = ∑ mi ri 2 = m1r12 + m2 r22 .
(6.6)
since
rm2 rm1
r1 = ;r2 = ,
m1 + m2 m1 + m2 (6.7)
2 2
r m1m
2
r m m2
2
mm
I= 2
+ 1
= r 2 1 2 ≡ µr 2 , (6.8)
(m1 + m2 ) (m1 + m2 )
2 2
m1 + m2

where µ is the reduced mass for rotation (reduced masses for other degrees of freedom of
motion are also possible; reduced masses for modes of vibration are particularly notable).
Note the common case of “hydrogenic” reduced mass: For HX molecules (X=F, Cl, Br, I,
O, …) molecules, the reduced mass for rotation is small, similar to the mass of the
hydrogen atom.

In general, for axis i of a three-dimensional object, I i = ∫ ri 2dmi , where r2 is the distance

from the axis and dmi the differential in mass.

2
The moment of inertia tensor has diagonal elements, e.g., I xx = ∑i mi ( yi 2 + zi 2 ) and off-
diagonal elements, e.g., I xy = I yx = −∑ mi xi yi . The principal axes are the choice of axes
i
that diagonalize the moment of inertia tensor for an object: I xy = I xz = I yz = 0.

The rotational energy E = Iω2 / 2, classically, and the rotational angular momentum
L2
L = I ω, or, E = . However, quantum mechanics says that L must be quantized in
2I
h nh (nh / 2π )2
units of  = , where h is Planck’s constant: L = ; E=
2π 2π 2I

The Hamiltonian expression for rotation about one axis is

Ĵ 2 Ψ J (J + 1) 2 Ψ
= EΨ = , (6.9)
2I 2I

h2 h2
with solution at energies E = BJ ( J + 1), where B = = 2 . B is the rotational
2 I 8π I
constant. Nonlinear polyatomic molecules have 2 or 3 rotational constants, depending on
the symmetry of the molecule. Expressing B in cm-1:

B (H35Cl) 10.59 ← hydrogenic

B (HF) 20.94 ← hydrogenic
B (OH) 18.87 ← hydrogenic
B (O2) 1.446 ← not hydrogenic

The spectroscopic consequences of having small, hydrogenic, moments of inertia, with

corresponding large rotational constants, are that such molecules (or the corresponding
rotational degrees of freedom in polyatomic molecules having one or more hydrogenic
reduced masses about rotational axes) 1. have smaller rotational partition functions, and
thus distribute the rotational spectra into fewer, stronger transitions; 2. have Boltzmann
population distributions that extend to higher energies. This corresponds to the far
infrared (submillimeter, terahertz) spectral region, where intensities also are stronger
because of the σ3 factor in the blackbody expression. This spectral region is thus
particularly useful for the measurements of gases that are important in stratospheric
chemistry, including OH, H2O, HO2, H2O2, HCl, HF, and HBr. These are not hydrides. In
hydrides, the hydrogen atom has a formal negative charge (e.g., NaH, LiH, LiAlH4).

Jˆ 2 Ψ
= E Ψ has the set of solutions Ψ = PJM (cos θ )ei M φ , which are spherical harmonics,
2I
where J = 0, 1, 2, …. and |M| ≤ J. There are thus 2J + 1 “states” of same energy  for each

J. Their degeneracy (in the absence of magnetic and electric fields m and E which may

3
perturb these states from energy degeneracy when the molecules have magnetic or
electric moments, Section 6.3) is 2J + 1 (Figure 6.3).

This shows why we need to modify our statement for

the Einstein coefficients from Bnm = Bmn to
gn Bnm = gm Bmn . The microscopic probabilities are the
same but the summed quantities are not. Detailed
balance requires the above relationship, e.g., Figure 6.3. Illustration of the
1 × B01 = 3 × B10 . Degeneracy in general is given by degeneracies of rotational energies
for the lowest several rotational
2F +1, where F is the total angular momentum of the states of a diatomic molecule
state (including rotation, orbital, spin, electronic,
nuclear).

2π hc 2σ 3dσ
Reminders: Blackbody radiancy Rσ dσ =
ehcσ / kT − 1
8π hcσ 3
The radiation density ρ (σ ) = hcσ / kT
erg cm-3/cm-1 = erg cm-2.
e −1
hcσ
Since ρ (σ ) = F (σ ) (F = flux), F (σ ) = ρ (σ ) / hσ in cm-3 s-1.
c

Dipole moments, transition moments, line strengths, and intensities

Spectroscopic transitions
are effected by the
interaction of
electromagnetic radiation
with electromagnetic
moments in atoms and
molecules. Important Figure 6.4. Left: a simpler arrangement of masses and charges,
transitions giving rise to the corresponding to a (hypothetical) purely ionically-bound molecule.
spectra of atmospheres Right: The more general case of an arbitrarily shaped molecule
and charge distribution.
arise from electric dipole
moments and (generally much weaker) magnetic dipole moments. Electric quadrupole
and even higher moments can also contribute in limited cases (electric quadrupole-
allowed transitions are significant in the near infrared A band of O2).

The electric dipole moment of a molecule has between one and three components
depending on the symmetry and arrangement of electric charges in the molecule (Figure
6.4). The components of the dipole moment are determined by taking integrals over the
product of charge density and distance from the center of mass:

µr = ∫ er r dr, r = x, y, z. (6.10)

4
 Figure 6.5. An electric dipole
moment of 1 Debye
corresponds to unit plus and
minus charges separated by 1
Å (10-10 m).

The intensity of a transition is proportional to µ 2 : The square of the quantum-mechanical

matrix element which governs intensity is

∫ Ψ1µ Ψ2d Ω = µ0 × (direction cosine matrix element) ,

2 2
(6.11)

where the direction cosine matrix elements describe the relative intensities for transition
among different quantum states. Their squares are the line strength factors. (Additionally
the Boltzmann statistics, Chapter 4, must be included for a complete description of line
intensities.)

The standard unit of the electric dipole moment µe is the Debye (D, 1 Debye = 1 esu-Å,
as illustrated in Figure 6.5). Magnetic dipole moments arise from circulating electronic
charges in the atom or molecule, take on quantized values, and generally interact much
more weakly with radiation. The standard unit of the magnetic dipole moment µB is the
Bohr magneton. To gauge the relative ranges of strength of electric dipole allowed and
magnetic dipole allowed transitions, (1 Debye)2 = 10-36 erg cm3 and (1 Bohr magneton)2
= 8.60×10-41 erg cm3. For several important atmospheric molecules, as examples, µe
(HCl) = 1.1 D, µe (OH) = 1.7 D, µe (O2) = 0, but µB = 2 Bohr magnetons (quantized to
almost exactly 2, with tiny corrections, including a relativistic correction).

For simple diatomic molecules, with electric dipole moments but without additional
electronic or nuclear moments, we identify µe with µ0 and find intensities proportional to

2 J +1 2 J +1
µ J→J +1 = µ02 ; µ J←J +1 = µ02 .
2J +1 2J + 3 (6.11)

The usual selection rules, determining by symmetries of the quantum mechanic matrix
elements as in Eq. 6.11 for rotational transitions of polar diatomics are ΔJ = ±1 (from the
symmetry of ∫ Ψ a µ Ψ b ). Then, the rotational lines occur at energies of ΔE = 2B, 4B, 6B
(Figure 6.6).

5
Figure 6.6. The lowest
several rotational energy
levels for polar diatomic
molecules.

More generally, the rotational energy levels can be expanded as:

E = BJ (J + 1) − DJ 2 (J + 1)2 + HJ 3 (J + 1)3 + ....

Departures from the simple BJ ( J + 1) are due to centrifugal distortion. By convention,

only the first distortion term, − DJ 2 ( J + 1)2 has a negative sign. D is usually (perhaps
always?) positive. Higher terms can in principle be positive or negative.

Rotational partition functions for diatomics are approximated by QR  T / c2 B , which is

accurate for B = kT and negligible centrifugal distortion. This is a very occasionally
useful approximation but it is mostly a relic of pre-computer days. As an example, the
HITRAN database of line parameters, which we will meet later, tabulates partition
functions for every 1oK. Interpolating assuming QR  T / c2 B gives very accurate values
in such a case.

The transition dipole moment between states m and n is proportional to the matrix
element ∫ Ψ m µΨ n dΩ. It is independent of degeneracy as defined:
2 2 2
Rmn = µJ →J +1 (2 J + 1) = µJ ←J +1 (2 J + 3). (6.12)
Then, quantum mechanics gives:
64π 4σ 3 2 64π 4σ 3 2
Amn = Rmn / gm = µmn , where m is the upper state.
3h 3h
8π 3 2 8π 3 2 8π 3 2 8π 3 2
Bmn = 2 Rmn / gm = 2 µmn , Bnm = 2 Rmn / gn = 2 µnm .
3h c 3h c 3h c 3h c

The standard definition of intensity, S

Treating the induced absorption and emission, with N0 as the total

population, Nl as the population N0fl in the lower state and Nu as
the population N0fu in the upper state:

6
− N l = N 0 ( f l Blu − f u Bul ) ρ (I 0 ,σ ) in cm-2 s-1, or
8π 3
− N l = N 0 f l 2 µlu2 (1− e 2 ) ρ (I 0 ,σ )
−c σ /T

3h c
 8π 3 2
− Nl = N0 fl µluσ (1− e −c2σ /T )F(I 0 ,σ )
3hc
− Nl 8π 3
= fl σ (1− e −c2σ /T ) µlu2 ≡ S.
N0 F 3hc

The above version is in “absorption form.” The entirely equivalent “emission form” can
be derived from spontaneous emission using the Einstein A coefficient.

u
N u
= f u Aul .
No
l
8π 3
After normalizing to blackbody sterradiancy, the result is S = σ (ec2σ / T − 1) f u µul2 .
3hc
8π 3 (e− c2 El / T − e− c2Eu / T ) 2
In general: S = σ R .
3hc Q
fl is the fraction of molecules in the lower state and fu in the upper state. S is in cm (my
preference), or in cm-1/(molecules cm-2), which is more standard usage.

Introduction to line shapes and spectral saturation

Multiplying S by the column density n of

molecules, in cm-2, gives the integral over the
wavenumber-dependent optical thickness τ of the
line or, in the weak limit of Beer-Lambert
absorption or emission, the line’s equivalent width
(W) in cm-1 (Figure 6.7). Figure 6.7. The integral of the optical
thickness of a line is an area which is an
upper limit to the line’s equivalent
width, it’s actual area.

6.2.2 Polyatomic molecules

Linear polyatomics - vibrational angular momentum, l-type doubling

Degenerate bending modes of vibration in a linear polyatomic can occur in the two planes
orthogonal to the linear axis. For bending vibrational quantum numbers of two or greater
the interaction of rotation and vibration gives rise to a splitting of the degenerate
vibrational levels known as l-type doubling. Splitting of lines in CO2, for example,

7
increases its global warming potential since the lines become more efficient absorbers
and emitters of radiation.

Symmetric top molecules

Two rotational constants are the same. This usually (not always) requires 4 atoms at
minimum.

3-fold axis
Symmetric top, but rare
(astrophysical H3+)

Not symmetric tops

Prolate: A > B = C

Oblate: A = B > C (much

less common)

A is the largest rotational constant, C the smallest.

E (prolate) = BJ(J +1) + (A − B)K 2 − DJ J 2 (J +1)2 − DJK J(J +1)K 2 − DK K 4 + ....

Selection rules: ΔJ = ±1, ΔK = 0 (The K-distributions are like discrete diatomic

manifolds)
1/2
π 1/2 ⎛ T ⎞ ⎛ T ⎞
Partition function: Q ≅ , s ≡ symmetry number “the number of
s ⎜⎝ c2 A ⎟⎠ ⎜⎝ c2 B ⎟⎠
different values of the rotational coordinates which all correspond to one orientation of
the molecule, remembering that identical atoms are indistinguishable.” (from Mayer and
Mayer, as quoted by Davidson).

Asymmetric top molecules

The have quite complex energy expressions which are normally determined iteratively by
diagonalizing a parameterized Hamiltonian expression in a basis set of symmetric top
wave functions. The rotational partition function is approximated by:

8
 1/2 1/2 1/2
⎛ T ⎞ ⎛ T ⎞ ⎛ T ⎞
Q≅π 1/2
⎜⎝ c A ⎟⎠ ⎜⎝ c B ⎟⎠ ⎜⎝ c C ⎟⎠ .
2 2 2

Some of the rotational degrees of freedom for nonlinear polyatomics can be hydrogenic,
as for HX molecules. Examples include HO2, H2O, H2O2, HOCl, HOBr, and NH3. For
HO2, the rotational constants are A = 20.357 cm-1, B = 1.118 cm-1, C = 1.056 cm-1. A has
a hydrogenic value.

A rigid planar molecule (or any rigid planar object) will have IC = IA + IB (easy to
demonstrate using the Pythagorean theorem). The inertial defect measures the departure
from this, defined as Δ = IC- IA - IB. Non-zero inertial defect is caused by vibrational
averaging and/or (for tetratomic or greater molecules) out-of-plane bending vibrations.

6.3 Vibrational spectroscopy

6.3.1 Diatomic molecules

Potential energy curves:

Figure 6.8 Idealized potential energy curve for a bound electronic state of a
diatomic molecule.

The potential well pictured in Figure 6.8 is for a particular electronic state (a bound state
in this case), which could well be the electronic ground state.

9
The bound-state potential curve is approximately harmonic (i.e., a parabola) near the
bottom. A better, and still analytic, potential is the Lennard-Jones (or 6-12) potential:
⎡⎛ b ⎞12 ⎛ b ⎞6 ⎤
V ( r ) = 4a ⎢⎜ ⎟ − ⎜ ⎟ ⎥ .
⎢⎣⎝ r ⎠ ⎝ r ⎠ ⎥⎦
It has nice analytic properties for calculations of spectroscopic and thermodynamic
properties. Also note the Morse potential, in Bernath:
V ( r ) = D (1 − e − β ( r −re ) ) . In general, though, accurate potentials are non-analytic, and are
2

derived either from inversion of spectroscopic data or from complex theoretical

calculations.

How complicated does it get? These are potential energy curves for O2 and its molecular
ions from P.H. Krupenie, The spectrum of molecular oxygen, J. Phys. Chem. Reference
Data 1, 423-543, 1972. States can be bound or purely repulsive (for example the 3Πu state
of O2). States dissociate to atomic
species in different atomic electronic
states. Dissociation of the B 3 Σ −g
state, for example, leads to a ground
state (3P) atom plus an O(1D) atom,
a significant driver of
photochemistry in the upper
stratosphere and mesosphere
through the formation of OH
through O(1P) + H2O → 2OH.
(O(1D) at lower altitudes is
predominantly from O3 photolysis.)

Figure 6.9. Potential curves for O2 and its ionic forms

from P.H. Krupenie, The spectrum of molecular oxygen,
J. Phys. Chem. Reference Data 1, 423-543, 1972.

The vibrational energy levels (Hermite polynomials are the wave functions for the
harmonic approximation to the vibrational potential) are approximately

10
 E = ω (ν +1 / 2), ν = 0,1, 2,..., where ω is the vibrational constant. It is related to the
Hooke’s law restoring force and the reduced mass for vibration, µv: ω ∝ k / µ v . The
lowest vibrational state, for v = 0 has energy ω/2; this is the vibration zero-point energy,
required by the Heisenberg uncertainty principle. In general, vibrations also have
distortions and terms coupling vibration to rotation (cf. Bernath, Eq. 7.28):
E = ω (v + 1/ 2) − ω e X e (v + 1/ 2) 2 + BJ (J + 1) − DJ 2 (J + 1) 2 − α e (v + 1/ 2)J (J + 1) +

Vibrational transitions: Fundamentals, overtones and hot bands

Torqueing between vibration states is accomplished through d µ / dr instead of µ (see

2
dµ
why?) and intensities are ∝ + higher-order terms (µ here is dipole moment; sorry
dr re
about any confusion with the µv of vibrational energy levels). This explains, for example,
why OH vibrational transitions are weak while rotational transitions are strong: µ is large,
while d µ / dr is small near re (it has zero derivative near re).
!

Figure 6.10 Schematic of the vibrational band types for a diatomic molecule.

Vibration band types: The fundamental transition is v = 0 → 1 in absorption,

11
v = 0 ← 1 in emission. Overtone transitions have Δv > 1. Hot band transitions are
between states with v ≥ 1. This gets much more complicated very quickly with
polyatomics (discussed later) since the number of fundamentals goes as 3N − 6
( 3N − 5 for linear molecules), where N is the number of atoms in the molecule.

The vibrational partition function is Qv = 1 + e −c2ω /T + e −2c2ω /T + ..., ignoring distortion.

(Where did the ω / 2 zero-point energy go? Got it? The choice of the zero of energy for
Boltzmann statistics is arbitrary, as long as it is consistent, since any offset occurs in
both numerators and denominators.) Often there is only one upper vibrational state that is
thermally accessible (i.e., which has a non-negligible Boltzmann factor).* Then
1
Qv ; . Again, this is now mostly a pre-computer curiosity, except for narrow
1 − e − c2ω / T
range interpolation.
*
This is less true for polyatomics, with ozone and CO2 being important cases. It is also
less true for hot conditions (astrophysical, combustion, plasmas).

6.3.2 Polyatomic molecules

For a nonlinear N-atomic molecule there are 3N-6

vibrational degrees of freedom (= number of vibrational
fundamentals). This comes from 3 degrees of freedom,
for the 3 spatial directions, from each atom, -3 for
molecular translation, -3 for molecular rotations. For a
linear polyatomic there are 3N-5, plus vibrational angular
momentum, which gives rise to l-type doubling.

Example for triatomic isosceles XY2 molecule (C2v

symmetry) S1=ν1: symmetric stretch; S2 = ν2: bend; S3
= ν3: asymmetric stretch (From Herzberg, Infrared and
Raman Spectra).

Examples (HITRAN2008):
ν1 ν2 ν3
H2O 3652 cm 1595 cm 3756 cm-1
-1 -1

O3 1103 cm-1 701 cm-1 1042 cm-1

also, O3 2ν1 @ 2201; 2ν1 +ν2 2886; 3ν1 @ 3278 L

Figure 6.11. Vibrations for a
CO2 has 4 fundamentals (since it is linear): symmetric triatomic molecule with C2v
and asymmetric stretch and doubly-degenerate bending: symmetry, from Herzberg,
Infrared and Raman Spectra.

ν1 (symmetric) 1286 cm-1

ν2 (greenhouse) 667 cm-1; doubly degenerate
ν3 (strong, asymmetric) 2349 cm-1

12
The symmetric stretch is Raman active rather than infrared active, that is, the transition is
allowed by the anisotropy of the polarizability rather than by a dipole moment derivative
(since it is, by symmetry, non-existent). The polarizability is responsible for molecular
(Rayleigh) scattering. Its anisotropy causes part of the scattering (4% for air) to be
inelastically scattered. The scattered photons gain or lose energy (becoming bluer or
redder) while the molecules lose or gain rovibrational energy, respectively. More later on
the atmospheric importance (you could check out our paper Ring effect studies: Rayleigh
scattering, including molecular parameters for rotational Raman scattering, and the
Fraunhofer spectrum, K. Chance and R.J.D. Spurr, Appl. Opt. 36, 5224-5230, 1997,
available at http://cfa-www.harvard.edu/atmosphere/publications.html).

The full story for CO2 is, of course, much more complicated. See, for example, Herzberg,
Infrared and Raman Spectra, Table 56 and Figure 84.

For a list of characteristic vibrational band frequencies – see Table 7.5 in Bernath (e.g.,
CO ∼ 1700 cm-1).

(Potential surfaces - later)

6.4 Nuclear spin

Definitions
Nuclear spin I = integral: Bosons; they obey Bose-Einstein statistics
Nuclear spin I = half-integral: Fermions; they obey Fermi-Dirac statistics

The total wave function Ψ must be either symmetric or Nucleus Spin

antisymmetric with respect to the interchange of the (I)
coordinates of the two identical (in type and molecular H ½
2
location) nuclei, for bosons (integral nuclear spin, I) or D (= H) 1
14
fermions (half-integral nuclear spin), respectively. This is a N 1
consequence of the Pauli exclusion principle, which states 12
C 0
that two fermions may not simultaneously occupy the same 16
O 0
quantum state. Bosons may, leading to some interesting 35
Cl 3/2
physical consequences (neutron stars, Bose-Einstein 37
Cl 3/2
condensates – BECs). Half-integral spin values were
discovered, accidentally, in 1922 by Stern and Gerlach during magnetic deflection
studies of a beam of (nuclear spin ½) silver atoms.

Bosons: PABΨ = +Ψ
Fermions: PABΨ = -Ψ,

Where PAB is the permutation operator and Ψ is the wave function, which is the product
of an electronic, a vibrational, a rotational, and a nuclear spin wave function:

Ψ = Ψ elect Ψ vib Ψ rot Ψ spin .

13
Ψelect is symmetric with respect to the permutation for the electronic ground state of H2
(the more usual case); Ψvib is symmetric (always), and Ψrot are symmetric with respect to
permutation of A and B for even J and asymmetric for odd J.

For H2, where the nuclei (protons) A and B have spins of I = 1/2 (and are thus fermions),
the two possible spin wave functions for each proton may be called α and β, where α (A)
is the state where proton A has angular momentum +1/2 (in units of  ) along the axis of
quantization (the internuclear axis), and β (A) has angular momentum -1/2 along the axis.
Possible nuclear spin wave functions for the molecule are:

α(A)α(B)
β(A)β(B)
[α(A)β(B) + β(A)α(B)]/√2
[α(A)β(B) - β(A)α(B)]/√2

The first three are symmetric with respect to permutation of A and B and the last one
asymmetric with respect to permutation of A and B. To make PABΨ = -Ψ, required for
fermions, asymmetric nuclear spin wave functions (1-fold degenerate) may only go with
symmetric rotational wave functions (even J). Conversely, symmetric nuclear spin wave
functions (3-fold degenerate) may only go with asymmetric rotational wave functions
(odd J).

The states of higher spin degeneracy are called ortho states, and those of lower spin
degeneracy are called para states. Thus for H2, the nuclear spin degeneracies are:

gN = 1 for even J (para)

gN = 3 for odd J (ortho).

For homonuclear diatomics in general, both bosons and fermions, the ortho/para ratio is
(I+1)/I. When PABΨelect = +Ψelect (the more usual case), this is the ratio of even J to odd J
spin degeneracies for bosons and of odd J to even J spin degeneracies for fermions. Thus
14
N2 (I = 1) has gN = 2, J even and gN = 1, J odd.

When PABΨelect = -Ψelect, (I+1)/I is the ratio of odd J to even J spin degeneracies for
bosons and of even J to odd J spin degeneracies for fermions. Thus 16O2 (I = 0), which
has a 3 Σ −g ground state, has only odd rotational states (16O17O and 16O18O have both).

Thus, there are ortho and para forms of H2, and of other molecules as well. This affects
the rotational parts of the spectrum very significantly. The nuclear spin degeneracies
must be included in the Boltzmann factors and partition functions:
bJ = g N (2 J + 1)e − c2 EJ / T
∞
Qrot = ∑ bJ .
J =0

14
Polyatomics have a straightforward continuation of these permutation principles (e.g.,
for CO2, H2O, O3, NH3), but we will not develop them here.

In general, ortho and para states of molecules behave as independent species,

unconnected to one another by electromagnetically-allowed transitions. If an equilibrium
mixture changes temperature, equilibrium may only be re-established by other, chemical,
means, where dissociation and reformation is permitted.

Hyperfine structure is due to nuclear spins ≥ 1/2 and gives (small) additional
corrections to the energy levels. This includes “dipolar” contributions due to the
interaction of the nuclear spin magnetic moments with internal magnetic fields.
Hyperfine structure also includes contributions from nuclear quadrupole moments:
Nuclear spins ≥ 1 have nuclear quadrupole moments, Q, giving rise to somewhat larger
(but still small) energy corrections through Q : ∇E.

Nuclear spins always add to the total angular momentum.

Figure 6.12. Cartoon of the nuclear

spins associated with the HCl molecule.

6.5 Electronic spectroscopy

6.5.1 Electronic orbital angular momentum, electronic spin angular momentum

For a diatomic molecule, the electronic state is given by 2 S+1 Λ, e.g., 2Π, 3Σ, 1Δ, etc. Λ is
the quantized orbital angular momentum projected onto the intermolecular axis. Λ = 0 ⇒
Σ state, Λ = 1 ⇒ Π state, Λ = 2 ⇒ Δ state, etc. No half-integral orbital angular
momentum values are allowed. L is the total electron orbital angular momentum. S is the
electron spin angular momentum (half-integral spin angular momentum values are
allowed; the electron has spin ½). 2S + 1 is the multiplicity. There may be additional
complexities (due to additional symmetries), e.g., the O2 ground state is 3 Σ−g , where g
stands for gerade (German for even, as opposed to ungerade). A gerade state is even with
respect to inversion. A state may only (and must be) g or u only if it is a state of a
homonuclear diatomic molecule. The superscript “-” refers to reflection in a plane
containing both nuclei. This symmetry only applies to molecules in Σ states, because they
are non-degenerate.

15
Hund’s coupling cases

How do angular momenta couple? In diatomic molecules, they couple according to

Hund’s cases. Angular momenta must be positive quantities. Angular momenta A and B
(A ≥ B) couple to give resulting totals A - |B| to A + |B|, in integer steps (e.g., 1 + 1/2 = 1/2
or 3/2). Figures 6.13 and 6.14 illustrate the two most common of the numerous Hund’s
cases. In Hund’s case (a) the orbital angular momentum and the spin angular momentum
are strongly coupled to the molecular axis with quantized projections Λ and Σ (which
exists only for Λ>0). The vector sum Ω = |Λ + Σ| couples to the rotational angular
momentum O to give the total angular momentum J. Λ, Σ, and Ω give, e.g., the 2 Π1/2 and
2
Π3/2 states of OH and ground rotational NO. In Hund’s case (b), the spin angular
momentum is uncoupled from the molecular axis and is coupled to the vector sum of Λ
and the rotational angular momentum O to give the total angular momentum J. With
increasing O, the spin angular momentum in NO, for example, rapidly uncouples from
the molecular axis to approach case (b).

Figure 6.13. Hund’s case (a), where the orbital angular momentum
and the spin angular momentum are strongly coupled to the
molecular axis with quantized projections Λ and Σ. The vector sum
Ω = |Λ + Σ| couples to the rotational angular momentum O to give the
total angular momentum J.

Figure 6.14. Hund’s case (b). The spin

angular momentum is uncoupled from the
molecular axis and is coupled to the vector
sum of Λ and the rotational angular
momentum O to give the total angular
momentum J.

There is no perfect coupling except when symmetry requires it; otherwise only total
angular momentum is an exact quantum number. Also, for a given molecule in a given
electronic state, the closest Hund’s case may change as rotational state changes as angular
momenta uncouple and recouple as strengths of interactions change.

16
Where are angular momentum coupling and perturbations in the spectra due to them
important? In free radicals: OH, O2 (!), NO, HO2, ClO, …. In OH (B = 18.87 cm-1), the
spin-orbit interaction is so strong that the first transitions are at 61 cm-1 rather than at 38
cm-1 (except for fine structure “maser” transitions, as seen in astrophysical OH masers).
Also, the OH spectrum looks totally unlike that of an evenly-spaced diatomic spectrum.

6.5.2 Electronic transitions

Electronic transitions often, although not always, occur in the visible and ultraviolet
portions of the spectrum because of the larger energy levels, and thus energy level
differences encountered for electronic states as compared to vibrations and rotations.

For electronic transitions

| µu→l |2 = qv′v′′ | Re |2 S J ′′ /(2 J ′ + 1), and | µl →u |2 = qv′v′′ | Re |2 S J ′′ /(2 J ′′ + 1), where S J ′′ is a
rotational line strength term (Bernath) ⇒ Hönl-London factor.

S J ′′ is like earlier rotational terms, but now involving Λ′, Λ′′.

Remember | µ J → J +1 |2 = µ02 ( J + 1) /(2 J + 1) and

| µ J ← J +1 |2 = µ02 ( J + 1) /(2 J + 3) for simple rotational spectra.
J + 1 is the Hönl-London factor in this simple case

qv′v′′ is the Franck-Condon Factor or vibrational overlap: | ∫ Ψ ∗v′Ψ v′′dτ |2 (show vertical
transition here)
Re is the electronic transition dipole moment: Re = ∫ Ψ′el µe Ψ′′el dτ . See Bernath chapters 9
and 10 for copious details.

8π 3 (e− c2 El / T − e− c2Eu / T )
Then, S (cm) = σ gl | µl →u |2
3hc Q(T )
8π 3
(e − c2 El / T
− e− c2 Eu / T )
= σ qv′v′′ | Re |2 S J ′′ .
3hc Q(T )

Sum rule: ∑S
J′
J ′′ ∝ 2 J ′′ + 1

“Sums of the line strengths of all transitions to or from a given rotational level are
proportional to the statistical weight of that level.” Herzberg, Spectra of Diatomic
Molecules, pp. 208-209.

Finally, what we mostly do with electronic spectra is to use cross sections Σ(σ ), in cm2,
often with temperature dependence and occasionally with pressure dependence, to
calculate spectra.

17
References

6.1 Bernath, Section 5.2 for development of the quantum mechanical Hamiltonian
operator (total energy operator for a conservative system) for angular momentum
problems (developed for atomic spectra, but just as applicable to rotational angular
momentum) and its solutions.

6.2.1 Bernath; For incredible detail, go to Herzberg, Spectra of Diatomic Molecules.

6.2 Bernath, Chapter 6; Townes and Schawlow and Herzberg, Infrared and Raman
Spectra (for much detail) on linear polyatomics - vibrational angular momentum, and l-
type doubling.

Bernath, Chapters 6 and 7; Davidson, Chapter 11, for more detail; Herzberg, Infrared and
Raman Spectra, for yet more detail on symmetric top molecules; also Mayer, J.E., and
M.G. Mayer “Statistical Mechanics,” Wiley, New York, 1940 for symmetry number
definition.

6.3 Bernath Chapter 6 and Chapter 7, Section 1 (vibrations of diatomics); Herzberg,

Diatomics for great detail.

6.4 Bernath, Sections 7.2 and 9.4. Fuller treatment can be found in a standard statistical
mechanics reference. I use Davidson Statistical Mechanics, which is now very old, but
still excellent.

Townes and Schawlow: Chapter 8 for magnetic hyperfine and Chapter 6 for nuclear
electric quadrupole.

6.51 Bernath, Chapter 9, Townes and Schawlow Chapter 7.

Problems

6.1. In class: Demonstrate that g0 B01 = g1B10 using detailed balance of populations.

18