9.5.

Rotation and Vibration of Diatomic Molecules 349

9.5 Rotation and Vibration

of Diatomic Molecules
Up to now we have discussed the electronic states of ri-
gid molecules, where the nuclei are clamped to a fixed
position. In this section we will improve our model of
molecules and include the rotation and vibration of dia-
tomic molecules. This means, that we have to take into
account the kinetic energy of the nuclei in the Schrö-
dinger equation Ĥψ = Eψ, which has been omitted in
the foregoing sections. We then obtain the Hamiltonian
2 N
!2 ! 1 2 !2 ! 2
Ĥ = − ∇k − ∇
2 k=1 Mk 2m e i=1 i
 
$ %
e2 Z 1 Z 2 ! 1 ! 1 1  Fig. 9.41. Energy E nel (R) of the rigid molecule and total
+ + − +
4πε0 R r ri1 ri2 energy E of the nonrigid vibrating and rotating molecule
i, j i, j i
nucl el 0
= Ê kin + E kin + E pot = T̂k + Ĥ0 (9.73)
can be written as the product of the molecular wave
where the first term represents the kinetic energy of the function χ(Rk ) (which depends on the positions Rk of
nuclei, the second term that of the electrons, the third the nuclei), and the electronic wave function Φ(ri , Rk )
represents the potential energy of nuclear repulsion, the of the rigid molecule at arbitrary but fixed nuclear po-
fourth that of the electron repulsion and the last term sitions Rk , where the electron coordinates ri are the
the attraction between the electrons and the nuclei. variables and Rk can be regarded as a fixed parameter.
This implies that nuclear motion and electronic motion
9.5.1 The Adiabatic Approximation are independent and the coupling between both is ne-
glected. The total energy E is the sum of the energy
Because of their much larger mass, the nuclei in a mol- E nel (R) of the rigid molecule in the nth electronic state,
ecule move much slower than electrons. This implies which is represented by the potential curve in Fig. 9.41
that the electrons can nearly immediately adjust their and the kinetic energy (E vib + E rot ) of the nuclei.
positions to the new nuclear configuration when the
nuclei move. Although the electronic wave functions Note that the total energy is independent of R!
ψ(r, R) depend parametrically on the internuclear di-
stance R they are barely affected by the velocity of the Inserting this product into the Schrödinger equation
moving nuclei. The kinetic energy of the nuclear motion (9.73) gives the two equations (see Problem 9.4)
E kin = 12 Mv2 is small compared to that of the electrons.
We therefore write the total Hamiltonian H in (9.73) as Ĥ0 Φnel (r, Rk ) = E n(0) · Φnel (r, Rk ) (9.75a)
( )
the sum
T̂k + E n(0) χn,m (R) = E n,m χn,m (R) . (9.75b)
H = H0 + Tk
The first equation describes the electronic wave func-
of the Hamiltonian H0 of the rigid molecule and Tk of tion Φ of the rigid molecule in the electronic state
the kinetic energy of the nuclei. Since the latter is small (n, L, Λ) and E n(0) is the total electronic energy of this
compared to the total energy of rigid molecule we can state without the kinetic energy Tk of the nuclei.
regard Tk as a small perturbation of H. In this case the The second equation determines the motion of the
total wave function nuclei in the potential
* el +
ψ(ri , Rk ) = χ(Rk ) · Φ(ri , Rk ) (9.74) E n(o) = E kin + E pot (ri , Rk ) , (9.76)
350 9. Diatomic Molecules

which consists of the time average of the kinetic Inserting the product (9.80) into (9.79) gives, as
energy of the electrons and the total potential energy has been already shown in Sect. 4.3.2, the following
of electrons and nuclei. The total energy equation for the radial function S(R):
$ %
nuc
E n,k = E kin + E n(o) (9.77) 1 d 2 dS
R (9.80)
R2 dR dR
of the nonrigid molecule is the sum of the kinetic energy , -
2M J(J + 1)!2
of the nuclei and the total energy of the rigid molecule. + 2 E − E pot (R) − S=0.
Equation (9.75b) is explicitly written as ! 2MR2
,$ % - For the spherical surface harmonics Y(ϑ, ϕ) we obtain
−!2 −!2 (n)
∆A − ∆B + E pot (Rk ) χn,m (Rk ) the Eq. (4.88), already treated in Sect. 4.4.2:
2MA 2MB
$ %
= E n,m χn,m (Rk ) . (9.78) 1 ∂ ∂Y 1 ∂2Y
sin ϑ + 2 (9.81)
sin ϑ ∂ϑ ∂ϑ sin ϑ ∂ϕ2
In the center of the mass system this translates to
, 2 - + J(J + 1)Y = 0 .
−! (n)
∆ + E pot (R) χn,m (R) = E n,m χn,m (R) While the first Eq. (9.80) describes the vibration of the
2M
diatomic molecule, (9.81) determines its rotation.
(9.79)
where M = MA MB /(MA + MB ) is the reduced mass of
9.5.2 The Rigid Rotor
the two nuclei and the index m gives the mth quantum
state of the nuclear movement (vibrational-rotational A diatomic molecule with the atomic masses MA and
state). MB can rotate around any axis through the center
The important result of this equation is: of mass with the angular velocity ω (Fig. 9.42). Its
rotational energy is then
The potential energy for the nuclear motion in
E rot = 12 Iω2 = J 2 /(2I) . (9.82)
the electronic state (n, L, Λ) depends only on the
nuclear distance R, not on the angles ϑ and ϕ, i. e., Here I = MA RA 2
+ MB RB2 = MR2 where M = MA MB /
it is independent of the orientation of the mol- (MA + MB ) is the moment of inertia of the molecule
ecule in space. It is spherically symmetric. The with respect to the rotational axis and |J| = Iω is its
wave functions χ = χ(R, ϑ, ϕ), however, may still rotational angular momentum. Since the square of the
depend on all three variables R, ϑ, and ϕ. angular momentum

Because of the spherically symmetric potential |J| 2 = J(J + 1)h 2

equation (9.77) is mathematically equivalent to the can take only discrete values that are determined by
Schrödinger equation of the hydrogen atom. The diffe- the rotational quantum number J, the rotational ener-
rence lies only in the different radial dependence of the gies of a molecule in its equilibrium position with an
potential. Analogous to the treatment in Sect. 4.3.2 we internuclear distance Re are represented by a series of
can separate the wave functions χ(R, ϑ, ϕ) into a radial
part depending solely on R and an angular part, depen-
ding solely on the angles ϑ and ϕ. We therefore try the RA RB
product ansatz
B
χ(R, ϑ, ϕ) = S(R) · Y(ϑ, ϕ) . A

The radial function S(R) depends on the radial form S

MA MB
of the potential, while the spherical surface harmo-
nics Y(ϑ, ϕ) are solutions for all spherically symmetric R
potentials, independent of their radial form. Fig. 9.42. Diatomic molecule as a rigid rotor
 9.5. Rotation and Vibration of Diatomic Molecules 351

discrete values with the rotational constant

!
J(J + 1)!2 Be = , (9.86)
E rot = . (9.83) 4πcMRe2
2MRe2
which is determined by the reduced mass M and the
The energy separation between the rotational levels J equilibrium nuclear distance Re . For historical reasons
and J + 1 one writes Be in units of cm−1 instead of m−1 .
(J + 1)!2
∆E rot = E rot (J + 1) − E rot (J) = (9.84) EXAMPLES
2MRe2
increases linearly with J (Fig. 9.43). 1. The H2 molecule has a reduced mass M = 0.5MH =
This result can also be directly obtained from (9.80). 8.35 ×10−28 kg, and the equilibrium distance Re =
For a fixed nuclear distance R the first term in (9.80) 0.742 ×10−10 m ⇒ I = 4.60 ×10−48 kg m2 . The ro-
is zero. Therefore the second term must also be zero, tational energies are
because the sum of the two terms is zero. The kinetic E rot (J) = 1.2 ×10−21 J(J + 1) Joule
energy of a rigid rotor, which does not vibrate, is E kin =
= 7J(J + 1) meV .
E rot = E − E pot , where E is the total energy. The bracket
of the second term in (9.80) then becomes for R = Re The rotational constant is Be = 60.80 cm−1 .
equal to (9.83). 2. For the HCl molecule the figures are M =
In the spectroscopic literature, the rotational term 0.97 AMU = 1.61 ×10−27 kg, Re = 1.27 ×10−10 m
values F(J) = E(J)/hc are used instead of the energies. ⇒ E rot = 2.1 ×10−22 J(J + 1) Joule = 1.21 J(J + 1)
Instead of (9.83) we write meV, Be = 10.59 cm−1 .

J(J + 1)!2 In Table 9.7 the equilibrium distances Re and the

Frot (J) = = Be J(J + 1) (9.85)
2hcMRe2 rotational constants are listed for some diatomic mol-
ecules. The figures show that the rotational energies are
within the range of
J F(J)
4 (J + 1) h 2 E rot = (10−6 −10−2 ) J(J + 1) eV .
∆Erot =
MRe2
For a rotational angular momentum J the rotational
ν4 period becomes
α 2πI/!
Trot = √ . (9.87)
3 ∆Erot h 2 J(J + 1)
tgα ∝ =
J + 1 MRe2 Depending on the rotational constant Be they range from
ν3 Trot = 10−14 s to 10√
−10
s. For Be = 1 cm−1 one obtains
0 1 2 3 4 5 6 J
b) −11
Trot = 1.6 ×10 / J(J + 1) s. If an electro-magnetic
2 wave falls onto a sample of molecules it can be ab-
I(ν) sorbed on rotational transitions J → J + 1 resulting in
ν2
absorption lines with frequencies
1
ν1 νrot (J) = [E(J + 1)] − E(J)]/! (9.88a)
0
ν1 ν2 ν3 ν4 νrot or, in wavenumber units cm−1 ,
a) c) νrot (J) = 2Be (J + 1) . (9.88b)
Fig. 9.43. (a) Energy levels of the rigid rotor (b) Separati-
ons ∆E rot = E rot (J + 1) − E rot (J) (c) Schematic rotational The rotational transitions between levels J and
spectrum J + 1 fall into the spectral range with frequencies
352 9. Diatomic Molecules

Table 9.7. Equilibrium di- Molecule Re / pm Be De αe ωe ωe xe

stances and rotational and
vibrational constants in H2 74.16 60.8 1.6 ×10−2 3.06 4401 121.3
units of cm−1 for some Li2 267.3 0.673 9.9 ×10−6 0.007 351.4 2.6
diatomic molecules N2 109.4 2.01 5.8 ×10−6 0.017 2359.0 14.3
O2 120.7 1.45 4.8 ×10−6 0.016 1580.0 12.0
I2 266.6 0.037 4.2 ×10−9 0.0001 214 0.61
H35 Cl 127.4 10.59 5.3 ×10−4 0.31 2990 52.8
D35 Cl 127.4 5.45 1.4 ×10−4 0.11 2145 27.2
ICl 232.1 0.114 4.0 ×10−8 0.0005 384 1.50
CO 112.8 1.931 6 ×10−6 0.017 2170 13.29
NO 115.1 1.705 0.5 ×10−6 0.017 1904 14.08

ν = 109 −1013 Hz, i. e., in the Gigahertz–Terrahertz E

Epot (R)
range with wavelengths between λ = 10−5 −10−1 m.
This spectral region is called the microwave range.

In Sect. 9.6.2 we will see, that only molecules with

a permanent electric dipole moment can absorb or → →
Fr Fz
emit radiation on rotational transitions (except for
very weak quadrupole transitions). Therefore ho-
monuclear diatomic molecules show no rotational
absorption or emission spectra!
Re R
Fig. 9.44. Compensation of centrifugal and restoring force in
the nonrigid rotating molecule
9.5.3 Centrifugal Distortion
A real molecule is not rigid. When it rotates, the
centrifugal force acts on the atoms and the inter- energy E pot (R) is, for R > Re , larger than E pot (Re ) we
nuclear distance widens to a value R where this force have to include the additional energy ∆E pot = 12 k(R −
Fc = −Mω2 R is compensated by the restoring force Re )2 in the rotational energy of the nonrigid rotor. The
Fr = − dE pot (R)/ dR holding the two atoms together, total energy of the nonrigid rotor is then
which depends on the slope of the potential energy
function E pot (R) (Fig. 9.44). J(J + 1)!2 1
E rot = + k(R − Re )2 . (9.91)
In the vicinity of the equilibrium distance Re the 2MR2 2
potential can be approximated by a parabolic function If we express R on the right side of (9.90) by Re and k
(see Sect. 9.4.4). This leads to a linear restoring force with the help of (9.89) we obtain
$ %
Fr = −k(R − Re ) R̂ . (9.89) J(J + 1)!2
R = Re 1 + = Re (1 + x)
MkRe4
From the relation J 2 = I 2 ω2 = M 2 R4 ω2 we obtain:
with x ' 1. This allows us to expand 1/R2 into the
J(J + 1)!2 ! power series
Mω2 R = = k(R − Re )
MR3 ,
1 1 2J(J + 1)!2
J(J + 1)!2 = 1 − (9.92)
⇒ R = Re + , (9.90) R2 Re2 MkRe4
MkR3 -
3J 2 (J + 1)2 !4
which means that the internuclear distance R is wi- + ∓...
dened by the molecular rotation. Since the potential M 2 kRe8
 9.5. Rotation and Vibration of Diatomic Molecules 353

and the rotational energy becomes spin S precesses independently around the z-axis with
a projection
J(J + 1)!2 J 2 (J + 1)2 !4
E rot = − (9.93)
2MRe2 2M 2 kRe6 )Sz * = Ms h . (9.96b)
3J 3 (J + 1)3 !6 Both projections add to the total value
+ ±... .
2M 3 k2 Re10
Ωh = (Λ + Ms )h . (9.96c)
In the case of strong spin-orbit coupling L and S couple
For a given value of the rotational quantum to J el = L + S with the projection
number J the centrifugal widening makes the mo- * el +
ment of inertia larger and therefore the rotational Jz = Ω × h
energy smaller. This effect overcompensates for
(see Sect. 9.3.3).
the increase in potential energy.
The total angular momentum J of the rotating mol-
ecule is now composed of the angular momentum N of
Using the term-values instead of the energies, (9.94) the molecular rotation and the projection Λh or Ωh. For
becomes Ω + = 0 the total angular momentum J of the molecule
is no longer perpendicular to the z-axis (Fig. 9.45).
Frot (J) = Be J(J + 1) − De J 2 (J + 1)2
+He J 3 (J + 1)3 − . . . Since the total angular momentum of a free mol-
ecule without external fields is constant in time,
(9.94) the molecule rotates around the space-fixed direc-
tion of J and for Λ + = 0 the rotational axis is no
with the rotational constants
longer perpendicular to the molecular z-axis.
! !3
Be = 2
, De = , (9.95)
4πcMRe 4πckM 2 Re6 In a simple model, the whole electron shell can be re-
3!5 garded as a rigid charge distribution that rotates around
He = . the z-axis. The rotating molecule can then be described
4πck2 M 3 Re10
as a symmetric top with two different moments of in-
The spectroscopic accuracy is nowadays sufficiently ertia: 1.) The moment I1 of the electron shell rotating
high to measure even the higher order constant H. around the z-axis and 2.) the moment I2 of the molecule
When fitting spectroscopic data by (9.95) this constant,
therefore, has to be taken into account.
→
N
9.5.4 The Influence of the Electron Motion →
J
→
Up to now we have neglected the influence of the elec- L
tron motion on the rotation of molecules. In the axial
symmetric electrostatic field of the two nuclei in the
nonrotating molecule, the electrons precess around the
Λ
space-fixed molecular z-axis. The angular momentum
L(R) = Σli (R) of the electron shell, which depends on
the separation R of the nuclei, has, however, a constant A Λh B z
projection R
)L z * = Λh (9.96a)
independent of R. For molecular states with electron Fig. 9.45. Angular momenta of the rotating molecule
spin S += 0 in atoms with weak spin-orbit coupling the including the electronic contribution
354 9. Diatomic Molecules

(nuclei and electrons) rotating around an axis perpendi- depend on the integer vibrational quantum number v =
cular to the z-axis. Because the electron masses are very 0, 1, 2, . . . .
small compared with the nuclear masses, it follows that They are equally√ spaced by ∆E = !ω. The
I1 ' I2 . frequency ω = kr /M depends on the constant
The rotational energy of this symmetric top is kr = ( d2 E pot / dR2 ) Re in the parabolic potential and on
the reduced mass M of the molecule. The lowest vibra-
Jx2 Jy2 J2
E rot = + + z (9.97) tional level is not E = 0 but E = 12 !ω. The solutions
2Ix 2I y 2Iz of (9.80) with a parabolic potential are the vibrational
with I x = I y = I1 + = I z = I2 . eigenfunctions
From Fig. 9.45 the following relations can be obtained: S(R) = ψvib (R, v) = e−πMω/h R · Hv (R) (9.102)

Jz2 = Ω 2 !2 where the functions Hv (R) are the Hermitian polyno-

mials. Some of these vibrational eigenfunctions of the
Jx2 + Jy2 = N 2 !2 = J 2 − Jz2 (9.98) harmonic oscillator are compiled in Table 4.1 and are
. /
= J(J + 1) − Ω 2 !2 . illustrated in Fig. 4.20.
Although the real potential of a diatomic molecule
Inserting this into (9.97) gives the term values F(J) = can be well approximated by a parabolic potential in the
E rot (J)/hc of the rotational levels vicinity of the potential minimum at R = Re , it deviates
. / more and more for larger |R − Re | (see Fig. 9.38). This
F(J, Ω) = Be J(J + 1) − Ω 2 + AΩ 2 (9.99) figure also illustrates that the Morse potential is a much
better approximation. Inserting the Morse potential
with the two rotational constants . /2
E pot (R) = E D 1 − e−a(R−Re ) (9.103)
! !
A= , Be = . (9.100) into the radial part (9.80) of the Schrödinger equation
4πcI1 4πcI2
allows its exact analytical solution (see Problem 9.5).
The term AΩ 2 , which does not depend on J, is generally The energy eigenvalues are now:
added to the electronic energy Te of the molecular state, $ % $ %
1 !2 ω20 1 2
since it is constant for all rotational levels of a given E vib (v) = !ω0 v + − v+
electronic state with quantum number Ω. It is therefore 2 4E D 2
also not influenced by the centrifugal distortion. (9.104)
The ground states of the majority of diatomic mol- with energy separations
ecules are 1 Σ-states with Λ = Ω = 0. For these cases
A = 0 and (9.99) is identical to (9.94). ∆E(v) = E vib (v + 1) − E vib (v) (9.105a)
, -
!ω
= !ω 1 − (v + 1) ,
2E D
9.5.5 Vibrations of Diatomic Molecules
where E D is the dissociation energy of the rigid mol-
For a nonrotating molecule, the rotational quantum ecule. The vibrational levels are no longer equidistant
number J in (9.80) is zero. The solutions S(R) of (9.80) but separations decrease with increasing vibrational
are then the vibrational wave functions of the diatomic quantum number v, in accordance with experimental
molecule. For J = 0 they solely depend on the radial observations.
form of the potential energy E pot (R). For a parabolic The term-values Tv = E v /hc are
potential, the vibrating molecule is a harmonic oscilla-
tor, which has been already treated in Sect. 4.2.5. The Tvib (v) = ωe (v + 12 ) − ωe xe (v + 12 ) (9.105b)
result obtained there was the quantization of the energy with the vibrational constants
levels.
ω0 !ω20 hc
The energy levels of the harmonic oscillator ωe = , ωe xe = = ω2e .
2πc 8πcE D 4E D
E(v) = (v + 12 )hω (9.101) (9.105c)
 9.5. Rotation and Vibration of Diatomic Molecules 355

The vibrational frequency Fig. 9.46. Vibrating rotor

0
ω0 = a 2E D /M (9.106) R(t)

corresponds to that of a classical oscillator with the S

restoring force constant kr = 2a2 E D . From measure- Re
ments of kr (for instance from the centrifugal distortion
of rotational levels) and the dissociation energy E D the
constant a in the Morse potential can be determined.
With the more general expansion of the potential
! 1 $ ∂ n E pot %
E pot (R) = (R − Re )n (9.107) frequency is higher than the rotational frequency by
n! n
n
∂R Re one to two orders of magnitude, the molecule un-
dergoes many vibrations (typically 5−100) during
the Schrödinger equation can only be solved numeri-
one rotational period (Fig. 9.46). This means that the
cally. We will, however, see in Sect. 9.5.7 that the real
nuclear distance changes periodically during one full
potential can be very accurately determined from the
rotation.
measured term values of the rotational and vibrational
levels.
EXAMPLES
Note: 1. For the H2 molecule, ωe = 1.3 ×1014 √ s−1 ⇒ Tvib =
−14 −13
4.8 ×10 s, while Trot = 2.7 ×10 J(J + 1) s.
• The distance between vibrational levels decreases 2. For the Na molecule, ωe = 4.5 ×1012 √ s−1 ⇒ Tvib =
with increasing v, but stays finite up to the disso- 1.4 ×10−12 s, while Trot = 1.1 ×10−10 J(J + 1) s.
ciation energy. This means that only a finite number
of vibrational levels fit into the potential well of
a bound molecular state. This is in contrast to the Since the total angular momentum J = I · ω of
infinite number of electronic states in an atom such a freely rotating molecule has to be constant in time,
as the H atom. Here the distance between Ryd- but the moment of inertia I periodically changes,
berg levels converges with n → ∞ towards zero the rotational frequency ω has to change accor-
(see (3.88)). This different behavior stems from the dingly with a period Tvib . Therefore the rotational
different radial dependence of the potentials in the energy
two cases. J(J + 1)!2
• One has to distinguish between the experimen- E rot =
exp 2m R2
tally determined dissociation energy E D , where
the molecule is dissociated from its lowest vibration also varies periodically with a period Tvib .
level, and the binding energy E B of the potential
well, which is measured from the minimum of the Because the total energy E = E rot + E vib + E pot
potential (Fig. 9.41). The difference is has to be constant, there is a periodic exchange of
exp rotational, vibrational and potential energy in the
E D = E B − 12 !ω .
vibrating rotor (Fig. 9.47).

The rotational energy, considered separately, is

9.5.6 Interaction Between Rotation
the time average over a vibrational period. This time
and Vibration
average can be calculated as follows:
Up to now we have looked at the rotation of a non- The probability to find the nuclei within the interval
vibrating molecule and the vibration of a nonrotating dR around the distance R is
molecule. Of course a real molecule can simul-
taneously rotate and vibrate. Since the vibrational P(R) dR = |ψvib (R)| 2 dR .
356 9. Diatomic Molecules

E always equal to Re (Fig. 9.48). Therefore, the rotatio-

nal constant Bv of the rotating harmonic oscillator also
Erot
depends on v. For the more realistic anharmonic poten-
tial, both )R* as well as )1/R2 * change with v. While
Evib
)R* increases )1/R2 * decreases with increasing v.

In order to express the rotational term values by

a rotational constant similar to (9.86) or (9.95) we
introduce, instead of Be , the rotational constant
Epot 1
! ∗ 1
Bv = ψvib (v, R) 2 ψvib (v, R) dR
t 4πcM R
(9.110a)
Fig. 9.47. Exchange between rotational, vibrational and
potential energy during a vibrational period averaged over the vibrational motion. It depends on the
vibrational quantum number v.
For a Morse potential we then obtain
The quantum mechanical expectation values of R and
1/R2 are then Bv = Be − αe (v + 12 ) (9.111a)
1
∗
)R* = ψvib Rψvib dR , and (9.108) where αe ' Be . In a similar way an average centrifugal
1 constant
∗ 1
* +
1/R2 = ψvib ψvib dR . !3
1
R2 ∗ 1
Dv = ψvib ψvib dR (9.110b)
This gives the mean rotational energy, averaged over 4πckM 2 R6
one vibrational period can be defined, which is related to De by
1
J(J + 1)!2 ∗ 1
)E rot (v)* = ψvib (v) 2 ψvib (v) dR . Dv = De − βe (v + 12 ) with βe ' De . (9.111b)
2M R
(9.109) For a general potential, higher order constants have to
Note: be introduced and one writes
Even for a harmonic potential the expectation value of Bv = βe − αe (v + 12 ) + γe (v + 12 )2 + . . . (9.112a)
1/R2 depends on the vibrational quantum number v. It
increases with v although )R* is independent of v and Dv = De + βe (v + 12 ) + δe (v + 12 )2 + . . . . (9.112b)

= 〈R〉 1 Bv
Re2 = = 〈R〉
R2 Be
v=6
v=4
5
4
3
3
2
b a 2

1 1

0 0
0.8 1 1.2
Fig. 9.48. Mean internuclear distance )R*
and rotational constant Bv ∝ )1/R2 * for the
a) Re R b) Re R harmonic (a)and anharmonic (b)potential
 9.5. Rotation and Vibration of Diatomic Molecules 357

The term value of a rotational-vibrational level can then and also to the coefficients an in the general potential
be expressed as the power series expansion (9.114) [9.10].
.
T(v, J) = Te + ωe (v + 12 ) − ωe xe (v + 12 )2
/ 9.5.8 Rotational Barrier
+ ωe ye (v + 12 )3 + ωe z e (v + 12 )4 + . . .
. The effective potential for a rotating molecule (see
+ Bv J(J + 1) − Dv J 2 (J + 1)2
/ (9.80))
+ Hv J 3 (J + 1)3 ∓ . . . . (9.113a)
eff (v) J(J + 1)!2
E pot (R) = E pot (R) + (9.117)
For a Morse potential this series is reduced to 2MR2
includes, besides the potential E pot (R) of the nonrota-
T Morse (v, J) = Te + ωe (v + 12 ) (9.113b)
ting molecule, a centrifugal term that depends on the
− ωe xe (v + 12 )2 + Bv J(J + 1) rotational quantum number J and falls of with R as
− Dv J 2 (J + 1)2 1/R2 (Fig. 9.49). For a bound electronic state this leads
eff
to a maximum of E pot (R) at a distance Rm , which can
where only five constants describe the energies of all be obtained by setting the first derivative of (9.117) to
levels (v, J) up to energies where the Morse potential zero. This distance
is still a good approximation. , -1/3
J(J + 1)!2
Rm = (9.118)
M( dE pot / dR)
9.5.7 The Dunham Expansion
depends on the rotational quantum number J and on the
In order to also reproduce the rotational-vibrational slope of the rotationless potential.
term values T(v, J) of a rotating molecule for a more The minimum of the potential is shifted by the ro-
general potential (9.107) tation of the molecule from Re to larger distances and
! the dissociation energy becomes smaller.
E pot (R) = an (R − Re )n , (9.114)
Energy levels E(v, J) above the dissociation energy
n
E D can be still stable, if they are below the maximum of
with
$ %
1 ∂ n E pot
an = . E
n! ∂Rn Re hc
/ cm−1 J = 275
9,000 Predissociation
Dunham introduced the expansion
!! . /k 8,000 J = 250
T(v, J) = Yik (v + 12 )i J · (J + 1) − Λ2
i k 7,000
(9.115)
6,000
where the Dunham coefficients Yik are fit parameters
chosen such that the term values T(v, J) best repro- J = 200
5,000
duce the measured term values of rotational levels in v = 10, J = 200
J = 150
vibrational states of the molecule. 4,000 v = 25, J = 150
With (9.115) the energies of all vibrational-
3,000 D = 6,000 cm−1
rotational levels of a molecule can be described by
a set of molecular constants. These constants are re- 2,000 J = 120
lated to the coefficients in the expansion (9.113a) by
the relations 1,000 J=0

Y10 ≈ ωe , Y20 ≈ −ωe xe , Y30 ≈ ωe ye 0

2 3 4 5 6 7 8 9 R/Å
Y01 ≈ Be , Y02 ≈ De , Y03 ≈ He (9.116)
Fig. 9.49. Effective potential curves of the rotating Na2
Y11 ≈ −αe , Y12 ≈ βe , Y21 ≈ γe molecule for different rotational quantum number J
358 9. Diatomic Molecules

E variety of molecular states, with energies depending on

Ekin( A + B) the electronic, the rotational and vibrational structure
v, J
of the molecule, the matrix elements of molecules are
more complicated than those of atoms. In this section
En(R) ED we will discuss their structure and the molecular spectra
derived from them.
For spontaneous emission (fluorescence spectra) the
emission probability of a transition |i* → |k* is given
R by the Einstein coefficient Aik . According to (7.17) Aik
is related to the transition dipole matrix element MiK by
Fig. 9.50. Predissociation of a molecule through the rotational
3
barrier 2 ωik
Aik = |Mik |2 . (9.119a)
3 ε0 c3 !
the potential barrier. However, due to the tunnel effect For the absorption or stimulated emission of radiation
(Sect. 4.2.3) molecules in these levels can dissociate by the transition probability Pik = Bik w(νik ) is proportio-
tunneling through the barrier (Fig. 9.50). This effect is nal to the spectral energy density w(ν) of the radiation
called predissociation by tunneling. The tunnel proba- field. In Sect. 7.2 it was shown that Pik is given by
bility depends exponentially on the width of the barrier
π 2 2
and on the energy gap between the maximum of the Pik = 2 E 02 2ψk∗ ε· pψi dτ 2 2 (9.119b)
barrier and the level energy. 2!
The predissociation rate can be determined by mea- where ε = E/|E| is the unit vector in the direction of
suring the width δE = h/τ of levels with a lifetime τ. If the electric field E of the electromagnetic wave, incident
the predissociation rate is large compared to the radia- onto the molecules. The transition probability therefore
tive decay of a level, the lifetime τ is mainly determined depends on the scalar product ε· p of electric field vector
by predissociation. Measuring τ(v) for all levels above and electric transition dipole of the molecule.
the dissociation limit gives information on the form and
heights of the potential barrier.
The dissociating fragments have a kinetic energy 9.6.1 Transition Matrix Elements
The dipole matrix element for a transition between two
E kin = E(v, J) − E D (J = 0) ,
molecular states with wave functions ψi and ψk is
which is shared by the two fragments according to their 11
masses. Mik = ψi∗ pψk dτel dτN . (9.119c)

9.6 Spectra of Diatomic Molecules

When a molecule undergoes a transition
E i (n i , Λi , vi , Ji ) ↔ E k (n k , Λk , vk , Jk )
between two molecular states |i* and |k*, electromagne-
tic radiation can be absorbed or emitted with a frequency
ν = ∆E/h. Whether this transition really occurs de-
pends on its transition probability, which is proportional
to the absolute square of the dipole matrix element Mik
(see Sect. 7.1). The relative intensities of spectral lines
can therefore be determined if the matrix elements of Fig. 9.51. Illustration of nuclear and electronic contributions
the transitions can be calculated. Because of the larger to the molecular dipole moment
 9.6. Spectra of Diatomic Molecules 359

The integration extends over all 3(Z A + Z B ) electronic function χ(R); which only depends on the nuclear coor-
coordinates and over the six nuclear coordinates. Often dinates. Inserting (9.120, 9.121) into (9.119) the matrix
only one of the electrons is involved in the transition. In elements is written as
this case the integration over dτel only needs to be perfor- 11
med over the coordinates of this electron. The vector p Mik = Φi∗ χN,i
∗ ∗
( pel + pN )Φk χN,k dτel dτN .
is the dipole operator, which depends on the coordina- (9.122a)
tes of the electrons, involved in the transition and on the
nuclear coordinates. In Fig. 9.51 it can be seen that Rearranging the different terms gives
! 1 ,1 -
p = −e ri + e(Z A RA + Z B RB ) = pel + pN ∗ ∗
i Mik = χi Φi pel Φk dτel χk dτN (9.122b)
(9.120) 1 ,1 -
∗ ∗
where pel is the contribution of the electrons and pN + χ1 pN Φi Φk dτel χk dτN .
that of the nuclei.
We distinguish between two different cases (Fig. 9.52):
Note that for homonuclear molecules Z A = Z B but
RA = −RB . Therefore pN = 0! • The two levels |i* and |k* belong to the same elec-
tronic state (Φi = Φk ). This means that the dipole
Within the adiabatic approximation we can separate transition occurs between two vibrational-rotational
the total wave function ψ(r, R) into the product levels in the same electronic state Φi . In this case
the first term in the sum (9.122b) is zero because the
ψ(r, R) = Φ(r, R) × χN (R) (9.121) integrand r|Φi |2 is an ungerade function of the elec-
of electronic wave function Φ(r, R) of the rigid mol- tron coordinates r = {x, y, z}. The integration from
ecule at a fixed nuclear distance R and the nuclear wave −∞ to +∞ therefore gives zero.

En

Eel
2 (R)

Rotational
levels E'(J')

v' = 3 ED
'
v' = 2 Vibrational
v' = 1 states E'(v')
v' = 0

Re'
Eel
1 (R)

Rotational
levels E''(J'')
Eel el
2 − E1 ED
''
v' ' = 3
v' ' = 2
Vibrational
v' ' = 1 states E''(v'')
v' ' = 0

Fig. 9.52. Rotational and vibrational le-

vels in two different electronic states of
Re'' Re' R a diatomic molecule
360 9. Diatomic Molecules

Since the electronic wave functions Φ are

Homonuclear diatomic molecules have no dipole-
orthonormal, i. e.,
1 allowed vibrational-rotational spectra. This
Φi∗ φk dτel = δik (9.123) means they do not absorb or emit radiation on
transitions within the same electronic state. They
the integral over the electronic coordinates in the may have very weak quadrupole transitions.
second term in the sum (9.122b) is equal to one.
The matrix element then becomes
Note:
1
Mik = χi,N pN χk,N dτN . (9.124) The molecules N2 and O2 , which represent the major
constituents of our atmosphere, cannot absorb the in-
frared radiation emitted by the earth. Other molecules,
The integrand solely depends on the nuclear such as CO2 , H2 O, NH3 and CH4 do have an electric
coordinates, not on the electronic coordinates! dipole moment and absorb infrared radiation on their
• Transitions between levels in two different electro- numerous vibrational-rotational transitions. Although
nic states. In this case the integral over the electronic they are present in our atmosphere only in small concen-
coordinates in the second term of (9.122b) is zero be- trations they can seriously perturb the delicate energy
cause the Φi , Φk are orthonormal. The second term balance between absorbed incident sun radiation and
is therefore zero and the matrix element becomes the energy radiated back into space by the earth (green-
house effect). If their concentration is increased by only
1 ,1 - small amounts this can increase the temperature of the
Mik = χi∗ Φi∗ pel Φk dτel χk dτN atmosphere at the earth’s surface (greenhouse effect).

= χi∗ Mikel (R)χk dτN . (9.125) The structure of the vibration-rotation-spectrum and
the pure rotation spectrum can be determined as follows.
Since the interaction potential between the two
We will now discuss both cases separately. atoms is spherically symmetric, we choose spherical
coordinates for the description of the nuclear wave
function χN (R, ϑ, ϕ).
If the interaction between rotation and vibration is
9.6.2 Vibrational-Rotational Transitions sufficiently weak we can write χN as the product
All allowed transitions (vi , Ji ) ↔ (vk , Jk ) between two χN (R, ϑ, ϕ) = S(R)Y JM (ϑ, ϕ) (9.127)
rotational-vibrational levels in the same electronic state
form for vi + = vk the vibrational-rotational spectrum of of the vibrational wave function S(R) in (9.102) and the
the molecule in the infrared spectral region between rotational wave function Y JM (ϑ, ϕ) for a rotational level
λ = 2 − 20 µm. For vi = vk we have pure rotational with angular momentum J and its projection M · ! onto
transitions between rotational levels within the same the quantization axis, which is a preferential direction
vibrational state, which form the rotational spectrum in the laboratory coordinate system. For absorbing tran-
in the microwave region with wavelengths in the range sition the quantization axis is, for instance, the direction
0.1−10cm. of the incident electromagnetic wave, or the direction
The dipole matrix element for these transitions is of its E-vector.
according to (9.120) and (9.124) With R = |RA − RB | and RA /RB = MB /MA
1 (Figs. 9.42 and 9.51) and p̂ = p/| p| the dipole moment
Mikrot vib = e χi∗ (Z A RA + Z B RB )χk dτN . (9.126) can be written as
MB · Z A − MA · Z B
For homonuclear diatomic molecules with Z A = Z B pN = p̂ · | pN | = e · R · p̂
MA + MB
and MA = MB is RA = −RB and therefore the integrand
is zero ⇒ Mikrot vib = 0. = C R p̂ . (9.128)
 9.6. Spectra of Diatomic Molecules 361

The volume element in spherical coordinates is z Fig. 9.53. Orienta-

2 tion of molecular
dτN = R dR sin ϑ dϑ dϕ . dipole moment p in
This gives the matrix element a space-fixed coor-
1 pz dinate system
Mik = C · Sv∗i (R)Svk (R)R3 dR (9.129)
→
1R θ p
M M
× Y Ji i Y Jk k p̂ sin ϑ dϑ dϕ .
ϑ,ϕ
px φ
The first factor describes the vibrational transition vi ↔ py
vk . If the harmonic oscillator functions are used for
x
the vibrational functions S(R) the calculations of the
y
integral shows that the integral is zero, unless
∆v = vi − vk = 0 or ±1 . (9.130)
The + sign stands for absorbing, the minus sign for
which gives
emitting transitions. Transitions with ∆v = 0 are pure
rotational transitions within the same vibrational level. ε̂· p = (9.131c)
This selection rule means that for the harmo- 3 $ %
4π 0 −εx + iε y 1 εx + iε y −1
nic oscillator only transitions between neighboring p εz Y1 + √ Y1 + √ Y1 .
vibrational levels are allowed. 3 2 2
For anharmonic potentials, such as the Morse po- Inserting this into the second integral in (9.129) and
tential, higher order transitions with ∆v = ±2, ±3, . . . extracting the components of the space fixed unit vec-
are also observed. Such overtone-transitions are, howe- tor ε out of the integral gives for the angular part of the
ver, much weaker than the fundamental transitions with transition probability integrals of the form
∆v = ±1. 1
M M
The second integral in (9.129) describes the rota- Y Ji i Y1∆M Y Jk k dΩ with ∆M = 0, ±1
tional transitions. It depends on the orientation of the
molecular dipole moment p in space. with the result that these integrals are always zero,
The amplitude of the radiation emitted into the di- except for ∆J = Ji − Jk = ±1.
rection k in space is proportional to the scalar product This selection rule is readily understandable, be-
of k · p and the intensity is the square of this amplitude. cause the absorbed or emitted photon has the spin
For absorbing transitions it is proportional to the scalar s = ±1h and the total angular momentum of the system
product E · p of electric field amplitude and molecular photon + molecule has to be conserved.
dipole moment p. For the projection quantum number M the selection
With the orientation angles Θ and φ of p̂ = p/| p| rules are analogue to that for atoms:
against the space-fixed axis X; Y ; Z we obtain the ∆M = 0 for linear polarization of the radiation and
relation (Fig. 9.53) ∆M = ±1 for circular polarization.
ε̂· p = p(εx sin Θ cos φ + ε y sin Θ sin φ + εz cos Θ)
(9.131a) Note:
where εi is the ith component of ε̂ = E/|E| against The angle ϑ is measured against the molecular axis in
the space fixed axis i = X, Y, Z. The angles can be the molecular coordinate system, while Θ and φ are the
expressed by the spherical surface harmonics Y JM : angles between the molecular dipole moment and the
3 3 space fixed quantization axis (see above).
4π 0 8π ±1
Y = cos Θ ; Y = ∓ sin Θ · e±iφ
3 1 3 1 In order to save indices in spectroscopic literature
(9.131b) the upper state (vk , Jk ) is always labeled with a prime as
362 9. Diatomic Molecules

(v2 , J 2 ), whereas the lower state (vi , Ji ) is labeled with J''

a double prime as (v22 , J 22 ). Transitions with 10

∆J = J 2 − J 22 = +1 8

are called R-transitions, those with 6

2 22
∆J = J − J = −1 P branch 4 R branch

are P-transitions.
2
All allowed rotational transitions appear in the spec-
trum as absorption- or emission lines (Fig. 9.54). All
rotational lines of a vibrational transition form a vi- ν0 ν
brational band. Its rotational structure is given by the Fig. 9.55. Fortrat diagram of the P- and R-branch of
wavenumbers of all rotational lines vibrational-rotational transitions

ν(v2 , J ↔ v22 , J 22 ) (9.132)

= ν0 + Bv2 J 2 (J 2 + 1) − Dv2 J 22 (J 2 + 1)2
. /
− Bv22 J 22 (J 22 + 1) − Dv22 J 222 (J 22 + 1)2
where ν0 is the band origin. It gives the position of 2,700 2,800 2,900 3,000
a fictious Q-line with J 2 = J 22 = 0. Since this line does
Fig. 9.56. Vibrational-rotational absorption of the H35 Cl
not exist in rotational-vibrational spectras of diatomic and H37 Cl isotopomers in the infrared region between
molecules, there is a missing line at ν = ν0 (Fig. 9.54). λ = 3.3−3.7 µm
Since the rotational constant Bv = Be − αe (v + 12 )
generally decreases with increasing v (αe > 0 for most
molecules) it follows that Bv2 < Bv22 . Plotting ν(J = J 22 ) the P lines are on the low frequency side. In Fig. 9.56
for P- and R transitions as a function of ν gives the the vibration-rotation spectrum of HCl is shown with
Fortrat-diagram shown in Fig. 9.55. The R-lines are on the P- and R-branch. The lines are split into two
the high frequency side of the band origin ν0 while components, because the absorbing gas was a mixture
of the two isotopomers of HCl with the two atomic
isotopes 35 Cl and 37 Cl. Since the rotational and vibra-
J'
tional constants depend on the reduced mass M the
v' = 1 lines of different isotopomers are shifted against each
4
other.
R(3)
3
E 9.6.3 The Structure of Electronic Transitions
P(4) R(2)
2 We will now evaluate the matrix element (9.125) for
P(3) R(1)
1 electronic transitions. The electronic part Mikel (R) de-
0
pends on the internuclear distance R, because the
electronic wave functions Φ depend parametrically
P(1)
P(2)

v' ' = 0 J''

3 on R. In many cases the dependence on R is weak
and we can expand Mikel (R) into a Taylor series
2 $ %
1 dMikel
0 Mikel (R) = Mik (Re ) + (R − Re ) + . . . .
dR Re
P R Fig. 9.54. P and R rota- (9.133)
tional transitions between
the vibrational levels v22 = In a first approximation only the first term, independent
ν0 ν 0 and v2 = 0 of R, is considered, which can be regarded as an average
 9.6. Spectra of Diatomic Molecules 363

of Mik (R) over the range of R-values covered by the el

Mik /Debeye
vibrating molecule. In this case the constant Mik (Re )
can be put before the integral over the nuclear coor- 1 +
10 Σu −1 Σ g+
dinates. Using the normalized nuclear wave functions
χN = S(R) · Y(ϑ, ϕ) and the vibrational wave functions
ψvib = R · S(R) the matrix element becomes
1 3
Σ g+ − 3 Σu+
Mik = Mikel ψvib∗
(vi )ψvib (vk ) dR (9.134) 9.5
1
M M
· Y Ji i Y Jk k sin ϑ dϑ dϕ
Atomic value
9
where Mh is the projection of the rotational angular
momentum J onto a selected axis (for instance, in the
1
Πu −1Σ g+
direction of the E-vector or the k-vector of the incident
electromagnetic wave for absorbing transitions, or in the 0.15 3 6 9 R / nm
direction from the emitting molecule to the observer for
Fig. 9.57. Dependence of electronic transition dipole on in-
fluorescent transitions). ternuclear distance R for several transitions of the Na2
molecule
Note:
This approximation of an electric transition dipole mo-
depends on the rotational angular momenta and their
ment independent of R is, for many molecules with
orientation in space. This factor determines the spatial
a strong dependence Mikel (R), too crude (Fig. 9.57). In
distribution of the emitted radiation.
such cases the second term in the expansion (9.133) has
An electric dipole transition in fluorescence can only
to be taken into account.
take place if none of these three factors is zero.
The probability of absorbing transitions depends
Since the probability of spontaneous emission is
according to (9.119b) on the scalar product of the
proportional to the square |Mik |2 , the intensity of
electric field vector E and the dipole moment p
a spectral emission line
2 2 Pik ∝ |E · Mik | 2 .
I(n i , vi , Ji ↔ n k , vk , Jk ) ∝ 2 Mikel 2 2 (9.135)
Since only the last factor in (9.135) depends on the
· FCF(vi , vk ) · HL(Ji , Jk ) orientation of the molecule in space, i. e., the direc-
is determined by three factors. tion of Mik against the electric field vector E, only the
The electronic part |Mikel |2 gives the probability of Hönl–London factor differs for spontaneous emission
an electron jump from the electronic state |i* to |k*. It and absorbing transitions. For the intensity I = ε0 E 2
depends on the overlap of the electronic wave functions of the incident electromagnetic wave we obtain with
Φi and Φk and their symmetries. E = ε· |E| the transition probability
21 22
The Franck–Condon factor 2
2 2 el
22 2 2
Pik = ε0 E · Mik (Re ) × 2 ψvib · ψvib · dR22
2 2 vi vk
21 22
2 2
FCF(vi , vk ) = 22 ψvib (vi ) · ψvib (vk ) dR22 (9.136) 21 22
2 M M 2
× 22 Y Ji i ε̂ · p̂Y Jk k sin ϑ dϑ dϕ22 . (9.138)
is determined by the square of the overlap integral of
the vibrational wave functions ψvib (vi ) and ψvib (vk ) in
the upper and lower electronic state. a) The General Structure
The Hönl–London factor of Electronic Transitions
21 22
2 M M 2
HL(Ji , Jk ) = 22 Y Ji i Y Jk k sin ϑ dϑ dϕ22 (9.137) Molecular electronic spectra have structures as shown
in (Fig. 9.58).
364 9. Diatomic Molecules

excitation mechanism. Generally, the energy of the up-

E per electronic state is for T = 300 K large compared to
the thermal energy kT . Therefore the thermal popula-
tion is negligible. Optical pumping with lasers allows
Ei the population of single selected levels. In this case the
fluorescence spectrum becomes very simple because it
(v', J') is emitted from a single upper level. In gas discharges,
Emission many upper levels are excited by electron impact and
Electronic the number of lines in the emission spectrum becomes
Absorption
transitions (UV / vis)
very large.
The absorption spectrum consists of all allowed
R transitions from populated lower levels.
Their intensity, as given in Sect. 7.2, is given by
Iikabs = gi Ni w(ν)Bik . (9.139b)
EK At thermal equilibrium the population distribution
(v'', J'') follows a Boltzmann distribution
Rotational-vibrational Pure rotational
transitions transitions Ni = gi e−Ei /kT . (9.140)
Fig. 9.58. Schematic representation of the structure of
molecular transitions
b) The Rotational Structure
of Electronic Transitions
All allowed transitions Ji22 ←→ Jk2 between the ro-
tational levels Jk2 of a given vibrational level v2 in the The wavenumnber of a rotational line in the electro-
upper electronic state and Ji22 of v22 in the lower elec- nic spectrum of a diatomic molecule corresponding to
tronic state form a band. In absorption or fluorescence a transition (n i , vi , Ji ) ↔ (n k , vk , Jk ) is
spectra such a band consists of many rotational lines. 4 5
νik = (Te2 − Te22 ) + Tvib (v2 ) − Tvib (v22 ) (9.141)
Transitions with ∆J = 0 form the Q-branch, those 4 2 22
5
with ∆J = Jk2 − Ji22 = +1 the R-branch and with + Trot (J ) − Trot (J )
∆J = −1 the P-branch. Q-branches are only present where Te gives the minimum of the potential curves
in transitions where the electronic angular momentum E pot (R) of the electronic states |i* or |k*, Tvib is the
changes by 1h, (e. g., for Σ ↔ Π transitions) in order to term value of the vibrational state for J = 0 and Trot (J)
compensate for the spin of the absorbed or emitted pho- the pure rotational term value.
ton. Electronic transitions with ∆Λ = 0 (e. g., between The rotational structure of a vibrational band is
two Σ-states) have only P and R branches. then (similarly to the situation for vibrational–rotational
The total system of all vibrational bands of this transitions within the same electronic state) given by
electronic transition is called a band system. The total
number of lines in such a band system depends not only νik = ν0 (n i , n k , vi , vk ) + Bv2 J 2 (J 2 + 1) (9.142)
on the transition probabilities but also on the number of − Dv2 J 22 (J 2 + 1)2
populated levels in the lower or upper electronic state. . /
− Bv22 J 22 (J 22 + 1) − Dv22 J 222 (J 22 + 1)2 .
The intensities of the lines in the emission spectrum
are proportional to the population of the emitting upper In contrast to (9.132), the rotational constant Bv2 in the
levels and to the transition probability Aik : upper state can now either be larger or smaller than Bv22 in
the lower electronic state. This depends on the binding
Iikem = gk Nk Aik (9.139a)
energies and the equilibrium distances Re in the two
where gk = (2Jk + 1) is the statistical weight of the states. The Fortrat-Diagrams shown in Fig. 9.59 has
level. The number of emitting levels depends on the a different structure for each of the two cases.
 9.6. Spectra of Diatomic Molecules 365

At those J-values where the curve ν(J) becomes

vertical, the density of rotational lines within a given
1 GHz
spectral interval has a maximum. The derivative dν/ dJ
changes its sign. For the case Bv22 > Bv2 the positions ν(J)
of the rotational lines increase for R-lines before the
maximum and then decrease again (Fig. 9.59a). The po-
sition νh of this line pileup is called the band head. For
Bv22 > Bv2 the R-lines show a band head at the high fre-
quency side of the band, while for Bv22 < Bv2 the P-lines
accumulate in a band head at the low frequency side
(Fig. 9.59b). The line density may become so high, that
even with very high spectral resolution the different li-
nes cannot be resolved. This is illustrated by Fig. 9.60,
which shows the rotational lines in the electronic tran-
sition of the Cs2 molecule around the band head, taken
with sub-Doppler resolution.
In molecular electronic spectra taken with photogra- Fig. 9.60. Band head of the vibrational band v2 = 9 ← v22 =
phic detection and medium resolution where only part 14 of the electronic transition C 1Πu ← X 1Σg+ of the Cs2
of the rotational lines are resolved, a sudden jump of the molecule, recorded with sub-Doppler-resolution
blackening on the photoplate appears at the band head
while the line density gradually decreases with incre-
asing distance from the band head. The band appears
shadowed (Fig. 9.61) to the opposite frequency side of
the band head. For Bv22 > Bv2 the band is red-shadowed
and the band head is on the blue side of the band, while
for Bv22 < Bv2 the band is blue-shadowed and the band
head appears on the red side.
In cases where the electronic transition allows Q-
lines, their spectral density is higher than that of the
P- and R-lines. For Bv2 = Bv22 all Q-lines Q(J) have the
same position. For Bv2 > Bv22 their positions ν(J) increase
with increasing J (Fig. 9.59a) while for Bv2 < Bv22 they
decrease (Fig. 9.59b).
3,805

3,577

3,371

3,159

2,977

2,820

v'' 2 1 0 0 0 0
v' 0 0 0 1 2 3

Fig. 9.61. Photographic recording of the band structure in

the electronic transition 3Πg ← 3Πu of the N2 molecule.
The wavelengths of the band heads are given in Å = 0.1 nm
above the spectrum (with the kind permission of the late
Fig. 9.59a,b. Fortrat-diagram for P, Q and R branches in Prof. G. Herzberg [G. Herzberg: Molecular Spectra and
electronic transitions: (a) Bv22 > Bv2 (b) Bv22 < Bv2 Molecular Structure Vol. I (van Nostrand, New York, 1964)])
366 9. Diatomic Molecules

c) The Vibrational Structure it follows that the electronic transition takes place at
and the Franck–Condon Principle a nuclear distance R∗ where the kinetic energies of the
vibrating nuclei in the upper and lower state are equal,
2 22
The vibrational structure of electronic transitions is go- i. e., E kin (R∗ ) = E kin (R∗ ). This can be graphically
verned by the Franck–Condon factor (9.136), which in illustrated by the difference potential
turn depends on the overlap of the vibrational wave 22
U(R) = E pot 2
(R) − E pot (R) + E(v2 ) (9.144)
functions in the two electronic states. In a classical
model, which gives intuitive insight into electronic tran- introduced by Mulliken (Fig. 9.63). The electron jump
sitions, the absorption or emission of a photon occurs from one electronic state into the other takes place at
within a time interval that is short compared to the vi- such a value R∗ , where Mulliken’s difference potential
brational period Tvib of the molecule. In a potential intersects the horizontal energy line E = E(v22 ), where
diagram (Fig. 9.62) the electronic transitions between
the two states can be then represented by vertical ar- U(R∗ ) = E(v22 ) .
rows. This means, that the internuclear distance R is In the quantum mechanical model, the probability
the same for the starting point and the final point of the for a transition v2 ↔ v22 is given by the Franck–Condon
transition. Since the momentum p = hν/c of the absor- factor (9.136). The ratio
bed or emitted photon is very small compared to that of
the vibrating nuclei, the momentum p of the nuclei is ψ 2 (R)ψvib
22
(R) dR
P (R) dR = 6 vib2 (9.145)
conserved during the electronic transition. Also, the ki- ψvib (R)ψvib22
(R) dR
netic energy E kin = p2 /2M does not change. From the
energy balance gives the probability that the transition takes place in the
interval dR around R. It has a maximum for R = R∗ .
2 22
hv = E 2 (v2 ) − E 22 (v22 ) If the two potential curves E pot (R) and E pot (R)
2
= E pot (R) + E kin 2 22
(R) − [E pot 22
(R) + E kin (R)] have a similar R-dependence and equilibrium distan-
2 ces Re2 ≈ Re22 the FCF for transitions with ∆v = 0 are
= E pot (R∗ ) − E pot
22
(R∗ ) (9.143) maximum and for ∆v += 0 they are small (Fig. 9.62a).

'

∆v > 0
U(R) difference
potential

''

Fig. 9.62. Illustration of the Franck-Condon principle for

vertical transitions with ∆v = 0 (a) and ∆v > 0 in case of Fig. 9.63. Illustration of the Mulliken-difference potential
potential curves with Re22 = Re2 and Re2 > Re22 22 (R) − E 2 (R) + E(v2 )
V(R) = E pot pot
 9.6. Spectra of Diatomic Molecules 367

The larger the shift ∆R = Re2 − Re22 the larger becomes transferred to the Rydberg electron, which then gains
the difference ∆v for maximum FCF (Fig. 9.62b). sufficient energy to leave the molecule (Fig. 9.65). The
situation is similar to that in doubly excited Rydberg
atoms where the energy can be transferred from one ex-
9.6.4 Continuous Spectra
cited electron to the Rydberg electron (see Sect. 6.6.2).
If absorption transitions lead to energies in the upper However, while this process in atoms takes place wi-
electronic state above its dissociation energy, unbound thin 10−13 −10−15 s, due to the strong electron-electron
states are reached with non-quantized energies. The ab- interaction, in molecules it is generally very slow (bet-
sorption spectrum then no longer consists of discrete ween 10−6 −10−10 s), because the coupling between
lines but shows a continuous intensity distribution I(ν). the motion of the nuclei and the electron is weak.
A similar situation arises, if the energy of the upper In fact, within the adiabatic approximation it would
state is above the ionization energy of the molecule, be zero! The vibrational or rotational autoionization
similarly to atoms (see Sect. 7.6). of molecules represent a breakdown of the Born–
In the molecular spectra the ionization continuum is, Oppenheimer approximation. The decay of these levels
however, superimposed by many discrete lines that cor- by autoionization is slow and the lines appear sharp.
respond to transitions into higher vibrational-rotational In Fig. 9.65 an example of the excitation scheme of
levels of bound Rydberg states in the neutral elec- autoionizing Rydberg levels is shown. The Rydberg le-
tron. Although the electronic energy of these Rydberg vels are generally excited in a two-step process from
states is still below the ionization limit, the additio- the ground state |g* to level |i* by absorption of a pho-
nal vibrational-rotational energy brings the total energy ton from a laser and the further excitation |i* → |k* by
above the ionization energy of the non-vibrating and a photon from another laser. The autoionization of the
non-rotating molecule (Fig. 9.64). Rydberg level |k* is monitored by observation of the
Such states can decay by autoionization into a lower resultant molecular ions. A section of the autoioniza-
state of the molecular ion, where part of the kinetic tion spectrum of the Li2 -molecule with sharp lines and
energy of the vibrating and rotating molecular core is a weak continuous background, caused by direct pho-
toionization, is shown in Fig. 9.66. The lines have an
asymmetric line profile called a Fano-profile [9.11].
The reason for this asymmetry is an interference effect
between two possible excitation paths to the energy E ∗
in the ionization continuum, as illustrated in Fig. 9.67:

Auto ionization
M*
|k〉 M+
|f〉
neutral ion

M*(k ) → M+ ( f) + Ekin(e− )

|i〉

|g〉
Fig. 9.65. Two-step excitation of a molecular Rydberg le-
Fig. 9.64. Excitation (1) of a bound Rydberg level in the vel |k*, which transfers by auto ionization into a lower level | f *
neutral molecule and (2) of a bound level in the molecular of the molecular ion. The difference energy is given to the free
ion M+ electron
368 9. Diatomic Molecules

Fig. 9.66. Section of the auto ioniza-

tion spectrum of the Li2 molecule
8,000

6,000
Γ = 0.02 cm−1
⇒ τeff = 1.3 ns
Ion signal

Γ = 0.034 cm−1
4,000 ⇒ τeff = 0.76 ns
Γ = 0.027 cm−1
⇒ τeff = 0.96 ns

2,000

0
42,136.6 42,136.8 42,137.0 42,137.2 42,137.4 42,137.6 42,137.8
Wavenumber (cm −1)

1. The excitation of the Rydberg level |k* of the neu- sonance destructive on the other constructive, resulting
tral molecule from level |i* with the probability in an asymmetric line profile.
amplitude D1 with subsequent autoionization, Continuous spectra can also appear in emission, if
2. The direct photoionization from level |i* with the a bound upper level is excited that emits fluorescence
probability amplitude D2 . into a repulsive lower state. Such a situation is seen in
excimers, which have stable excited upper states but an
When the frequency of the excitation lasers is tu-
unstable ground state (Fig. 9.68). For illustration, the
ned, the phase of the transition matrix element does not
emission spectrum of an excited state of the NaK alkali
change much for path 2, but much more for path 1, be-
molecule is shown in Fig. 9.69. This state is a mixture of
cause the frequency is tuned over the narrow resonance
of a discrete transition. The total transition probability

Pif = |D1 + D2 | 2
E D1Π
therefore changes with the frequency of the excitation 3
Π
laser because the interference is on one side of the re-
Continuous
fluorescence
|k〉
E*
Discrete
D12 σ lines
D2
D1
1 2

3 + R
Σ
σd Laser excitation
|i〉 X1Σ
a) b) 1/q −q ε
Fig. 9.67. (a) Interference of two possible excitation pathways
to the energy E ∗ in the ionization continuum (b) Resultant
Fano-profile with asymmetric line shape. σd is the absorption Fig. 9.68. Level scheme of the NaK molecule with excitation
crosssection for direct photoionization and discrete and continuous emission spectrum
 9.6. Spectra of Diatomic Molecules 369

a) E E'pot (R) b)

v' = 7
2
I∝ ∫ ψ v'ψ v''dr
Airy function
630 625
v'' Discrete emission
spectrum
U(R)

v''
Continuous
emission spectrum

v''
E''pot (R)
0
ν / cm−1 R λ / nm 695 675 655 635
Fig. 9.69. (a) Vibrational overlap and Franck-Condon factor for the continuous emission (b) Measured emission spectrum of
the NaK molecule

a singlet and a triplet state, due to strong spin-orbit coup- overlap integral between the vibrational wave function
ling. Therefore transitions from this mixed state into of the bound level in the upper electronic state with
lower singlet as well as triplet states becomes allowed. the function of the unstable level in the repulsive po-
While the emission into the stable singlet ground state tential above the dissociation energy of the lower state
X 1Σ shows discrete lines, the emission into the weakly which can be described by an Airy function. The fre-
bound lowest triplet state 3 3Σ shows, on the short wave- quency ν = E 2 (R) − E 22 (R) and the wavelength λ = c/ν
length side, a section of discrete lines terminating at of the emission depends on the internuclear distance R
bound vibrational-rotational levels in the shallow poten- because the emission terminates on the Mulliken po-
tial well of the a 3Σ state and, on the long wavelength tential of the repulsive lower state (dashed blue curve
side, a modulated continuum terminating on energies in Fig. 9.69). The number q = v2 − 1 of nodes in the
above the dissociation limit of the a 3Σ state. The inten- fluorescence spectrum gives the vibrational quantum
sity modulation reflects the FCF, i. e., the square of the number v2 of the emitting level.
370 9. Diatomic Molecules

S U M M A R Y
• For the simplified model of a rigid diatomic monotonically decreases with increasing R the
molecule, the electronic wave functions ψ(r, R) state is unstable and it dissociates.
and the energy eigenvalues E(R) can be ap- • The vibration of a diatomic molecule can be de-
proximately calculated as a function of the scribed as the oscillation of one particle with
internuclear distance R. The wave functions are reduced mass M = MA MB /(MA + MB ) in the
written as a linear combination of atomic orbitals potential E pot (R). In the vicinity of Re the po-
(LCAO approximation) or of other suitable basis tential is nearly parabolic and the vibrations can
functions. be well-approximated by a harmonic oscillator.
• In a rotating and vibrating molecule the kinetic The allowed energy eigenvalues, defined by the
energy of the nuclei is generally small compa- vibrational quantum number v, are equidistant
red to the total energy of a molecular state. This with a separation ∆E = !ω. For higher vibratio-
allows the separation of the total wave func- nal energies the molecular potential deviates from
tion ψ(r, R) = χN (R)Φ el (r, R) into a product of a harmonic potential. The distances between vi-
a nuclear wave function χ(R) and an electro- brational levels decrease with increasing energy.
nic function Φ el (r, R), which depends on the A good approximation to the real potential is the
electronic coordinates r and only contains R as Morse-potential, where ∆E vib decreases linearly
a free parameter. This approximation, called the with energy. Each bound electronic state has only
adiabatic or Born–Oppenheimer approximation, a finite number of vibrational levels.
neglects the coupling between nuclear and elec- • The rotational energy of a diatomic molecule
tron motion. The potential equals that of the rigid E rot = J(J + 1)h 2 /2I is characterized by the ro-
molecule and the vibration and rotation takes tational quantum number J and the moment
place in this potential. of inertia I = MR2 . Due to the centrifugal
• Within this approximation the total energy of force Fc the distance R increases slightly
a molecular level can be written as the sum with J until Fc is compensated by the resto-
E = E el + E vib + E rot of electronic, vibrational ring force Fr = − dE pot / dR and the rotational
and rotational energy. This sum is independent energy becomes smaller than that of a rigid
of the nuclear distance R. molecule.
• The electronic state of a diatomic molecule is • The absorption or emission spectra of a diatomic
characterized by its symmetry properties, its to- molecule consists of:
tal energy E and by the angular momentum and a) Pure rotational transitions within the same vi-
spin quantum numbers. For one-electron systems brational level in the microwave region
these are the quantum numbers λ = l z /h and b) Vibrational-rotational transitions within the
σ = sz /h of the projections l z of the electronic same electronic state in the infrared region
orbital angular momentum and sz of the spin s c) Electronic transitions in the visible and UV
onto the internuclear z-axis. For multi-electron region
systems L = Σli , S = Σsi , Λ = L z /h = Σλi , and • The intensity of a spectral line is proportional to
Ms = Σσi = Sz /h. Although the vector L might the product N · |Mik |2 of the population density N
depend on R, the projection L z does not. in the absorbing or emitting level and the square
• The potential curves E pot (R) are the sum of mean of the matrix element Mik .
kinetic energy )E kin * of the electrons, their po- • Homonuclear diatomic molecules have neither
tential energy and the potential energy of the a pure rotational spectrum nor a vibrational-
nuclear repulsion. If these potential curves have rotational spectrum. They therefore do not absorb
a minimum at R = Re , the molecular state is sta- in the microwave and the mid-infrared region, un-
ble. The molecule vibrates around the equilibrium less transitions between close electronic states fall
distance Re . If E pot (R) has no minimum, but into this region.
!
 Problems 371

• The electronic spectrum consists of a system equal to the square of the vibrational overlap
of vibrational bands. Each vibrational band integral.
includes many rotational lines. Only rotatio- • Continuous absorption spectra arise for transi-
nal transitions with ∆J = 0; ±1 are allowed. tions into energy states above the dissociation
The intensity of a rotational transition de- energy or above the ionization energy. Continuous
pends on the Hönl-London factor and those emission spectra are observed for transitions
of the different vibrational bands are determi- from bound upper states into a lower state with
ned by the Franck-Condon factors, which are a repulsive potential.

P R O B L E M S
1. How large is the Coulomb repulsion of the nuclei 5. Show that the energy eigenvalues (9.104) are
in the H+ 2 ion and the potential energy of the obtained when the Morse potential (9.103) is
electron with wave function Φ + (r, R) at the inserted into the Schrödinger equation (9.80).
equilibrium distance Re = 2a0 ? First calculate 6. What is the ionization energy of the H2 mol-
the overlap integral SAB (R) in (9.13) with the ecule when the binding energies of H2 and H+ 2
wave function (9.9). What is the mean kinetic are E B (H2 ) = −4.48 eV and E B (H+2 ) = −2.65 eV
energy of the electron, if the binding energy is and the ionization energy of the H atom
E pot (Re ) = −2.65 eV? Compare the results with E Io = 13.6 eV?
the corresponding quantities for the H atom. 7. Calculate the frequencies and wavelengths for
2. How large is the electronic energy of the H2 mol- the rotational transition J = 0 → J = 1 and
ecule (without nuclear repulsion) for R = Re and J = 4 → J = 5 for the HCl molecule. The in-
for the limiting case R = 0 of the united atom? ternuclear distance is Re = 0.12745 nm. What is
3. a) Calculate the total electronic energy of the H2 the frequency shift between the two isotopomers
molecule as the sum of the atomic energies of H35 Cl and H37 Cl for the two transitions? What is
the two H atoms minus the binding energy of H2 . the rotational energy for J = 5?
b) Compare the vibrational and rotational energy 8. If the potential of the HCl molecule around Re
of H2 at a temperature T = 300 K with the energy is approximated by a parabolic potential E pot =
of the first excited electronic state of H2 . k(R − Re )2 a vibrational frequency ν0 = 9 ×
4. Prove that the two separated equations (9.75) 1013 s−1 is obtained. What is the restoring force
are obtained when the product ansatz (9.74) is constant k? How large is the vibrational amplitude
inserted into the Schrödinger equation (9.73). for v = 1?