Diatomic Molecules

The step from atom to molecule takes us onto a higher,

significantly more complex level of understanding the structure
of matter. Although the properties of atoms play an important
role when describing molecules, we are faced with a
significantly more intricate task than simply adding atomic
properties. In this chapter we identify the most important
molecular phenomena and introduce suitable methods for
understanding them.

Overview
This chapter outlines the basic concepts of molecular physics as exempli-
fied for diatomic molecules. We begin with some energetic considerations in
Sect. 3.1 and introduce in Sect. 3.2 the B ORN -O PPENHEIMER approxima-
tion – the basis of all molecular physics. Molecular rotation and vibration
are treated in Sect. 3.3 followed by an elaboration on dipole transitions in
Sect. 3.4. Elements of the molecular orbital concept are presented in Sect. 3.5
while Sect. 3.6 focusses on angular momentum coupling and the famous
H UND’s cases. While all the previous discussion was focussed on homonu-
clear molecules, the chapter ends by introducing the specificities of heteronu-
clear diatomic molecules in Sect. 3.7. The content of this chapter is essential
for understanding most of the following ones – it is one of the keystones
within these textbooks. The reader should thus familiarize him- or herself
very thoroughly with all topics discussed here.

A broad range of methods is available today for obtaining detailed information

about the structure and dynamics of molecules. A key role plays spectroscopy in all
spectral ranges: from radio frequency (NMR) via the microwave range (EPR and
rotational spectroscopy), the FIR and NIR (vibrations), the visible and ultraviolet
(electronic transitions) to finally X-ray spectroscopy (chemical shifts of inner shell
transitions). Absorption and emission of electromagnetic radiation is used in a wide
variety of methods and has led to a wealth of information without which our present
understanding of molecules would not be conceivable.
Additional information is obtained e.g. from X-ray and neutron diffraction,
which give very direct insight into the spatial structure of molecules. Scattering

© Springer-Verlag Berlin Heidelberg 2015 135

I.V. Hertel, C.-P. Schulz, Atoms, Molecules and Optical Physics 2,
Graduate Texts in Physics, DOI 10.1007/978-3-642-54313-5_3
136 3 Diatomic Molecules

experiments and more recently also ultrafast methods have revealed important struc-
tural information, but they also allow in addition to study the dynamics of molecular
systems, i.e. the evolution of processes within or between molecules when interact-
ing with each other or with photons. We shall treat these themes in some detail in
Chaps. 6–8, and some special aspects will be mentioned in Chap. 10. In the present
chapter we shall develop the key concepts for understanding molecules as exempli-
fied for diatomic molecules, the simplest molecular systems. Based on these con-
cepts we shall explain the most important experimental findings for homonuclear
(A2 ) and heteronuclear (AB) diatomic molecules.

3.1 Characteristic Energies

The large difference in mass between electrons (me ) and atomic nuclei (M)
me
 10−3 . . . 10−5
M
is the basis for the most important approximations in molecular physics. The rele-
vant forces which keep molecules together and are responsible for their interactions
and spectroscopic properties, are again – as for atoms – of purely electromagnetic
nature.1 Since the C OULOMB force acting on electrons is identical to that acting
on nuclei, the velocity of the atomic nuclei is typically much smaller than that of
their electrons. Thus, nuclei stay essentially fixed in space while electrons move
around them very fast. Equilibrium distances RAB are usually found in a rather
narrow range from 0.075 nm to 0.18 nm. For example for the hydrogen molecule
(H2 ) RHH = 0.07417 nm, for oxygen (O2 ) ROO = 0.12074 nm, for nitrogen (N2 )
RNN = 0.10976 nm and for carbon monoxide (CO) RCO = 0.11282 nm. In some
few cases also larger distances are observed, as e.g. in K2 with RKK = 0.3923 nm or
in I2 with RII = 0.2668 nm.
Polyatomic molecules may have a broad variety of geometries. One example of
quite specific symmetry is methane (CH4 ), its atoms being positioned in tetrahedral
form as sketched in Fig. 3.1 with an equilibrium distance2 of RCH = (0.108595 ±
0.00003) nm.

Fig. 3.1 Tetrahedral H

structure of methane H

C
H
H

1 Notwithstanding this fundamental fact, there is a notable, brave quest to observe influences of

weak interaction in special instances by state-of-the-art high precision spectroscopy.

2 In chemical terminology the equilibrium distance is usually called “bond distance” or “bond

length”. The value given here for CH4 is based on state-of-the-art quantum chemical calcula-
tions and comparison with precision infrared and R AMAN spectra of different isotopologues of
CH4−x Dx according to S TANTON (1999).
3.1 Characteristic Energies 137

Fig. 3.2 Molecular z

coordinates for a diatomic
molecule: The nuclear MA
coordinates for atom A and B r3 RA
are indicated by capital r1
letters, the coordinates (here
r 1 , r 2 and r 3 ) of the O
individual electrons by lower y
r2
case letters RB

MB x

3.1.1 Hamiltonian

In this chapter we concentrate on diatomic molecules and start with deriving the
Hamiltonian. We use atomic units (a.u.), and relative coordinates in respect of the
centre of mass (O) of the system as illustrated in Fig. 3.2. Translational motion of
the molecule as a whole does not play a role in the following discussion. It simply
reflects the thermal motion in an ensemble. The kinetic energy of the two nuclei and
the N electrons is3
N 
 
me 2 1 2
Tn = − ∇R and Tr = − ∇ ri , (3.1)
2M̄ 2
i=1

respectively, with the electron coordinates r i , the internuclear distance R = R B −

R A , and the reduced mass M̄, here of the two nuclei
M A MB
M̄ = . (3.2)
M A + MB
The C OULOMB potential for all particles is
N
 N
 N

ZA ZB e2 ZA ZB
V (r, R) = − − + + . (3.3)
|r i − R A | |r i − R B | |r i − r k | R
i=1 i=1 i,k=1
i<k

The sum of these energies gives the total Hamiltonian

 = Tn (R) + Tr (r) + V (r, R),
H (3.4)

where now r is considered to represent the entirety of the coordinates of all elec-
trons. With this follows the S CHRÖDINGER equation
 
Tn (R) + Tr (r) + V (r, R) Ψ (r, R) = W Ψ (r, R). (3.5)

3 Weignore here the kinematic shift (which is anyhow problematic for multi-electron systems) and
assume (with reasonable accuracy) m̄e = me M̄/(me + M̄)  me .
138 3 Diatomic Molecules

In this form it is also valid for polyatomic molecules, if R is taken to represent the
entirety of all nuclear coordinates.

3.1.2 Electronic Energy

Let us try to obtain an estimate for the electronic energy of such a system. For clarity
we switch back to SI units. We identify the extension of the electron orbitals r with
the bond length R0 of the molecule. For molecular hydrogen (H2 ) as an example
one finds r  R0  0.074 nm. The uncertainty relation gives us an estimate for
the momentum of the electrons p = /r, and an average kinetic energy Tr  =
p 2 /(2me ), which we may insert into the well known Virial theorem 2Tr  = −Ve 
for −1/r potentials. Thus, the binding energy of the electrons can be estimated
roughly as
 
We = Tr  + Ve  = −Tr   −p 2 /(2me ) = −2 / 2me r2  7 eV.

Typically, We is found in an energy range of several eV, and the electronic spectra
are expected in the ultraviolet and visible range of the electromagnetic spectrum –
quite comparable to electronic transitions within atoms. More interesting are the
nuclear degrees of freedom. We distinguish vibrations of the atomic nuclei within
a molecule, relative to each other, and rotation of the whole nuclear structure –
considered as fixed in a 0th order approximation.

3.1.3 Vibrational Energy

Since atoms and electrons are bound to each other, the force on the nuclei must be
of the same order of magnitude as that on the electrons. Let us assume this force
 i.e. F = −kR. The (angular) vibrational frequency of the nuclei is
to be harmonic,
then ωv = k/M̄, with the reduced mass M̄ on the order of the nuclear masses.
The frequency of the electron is given by a force constant of similar magnitude so
√
that ωe = k/me . The corresponding
 energies of vibrational motion and electronic
√
energy are Wv = ωv =  k/M̄ and We = ωe =  k/me , respectively. Thus,
for
 the ratio of vibrational
 nuclear motion and electronic energy we find Wv /We 
me /M̄, and since me /M̄  10−2 vibrational energies will be on the order of
magnitude

Wv  me /M̄We  0.1 eV. (3.6)

Transitions are found in the infrared spectral range, e.g. for HCl at ν̄ = 1/λ = ν/c 
3000 cm−1 =  0.37 eV or λ  3.33 µm.
3.2 B ORN O PPENHEIMER Approximation 139

Fig. 3.3 Rotation of a z

diatomic molecule ω

O
R0 ~
1Å

3.1.4 Rotational Energy

Let us consider a diatomic molecule, rotating around an axis perpendicular to the

molecular axis as indicated in Fig. 3.3. We denote the angular momentum operator
of the molecule by N  , the corresponding rotational quantum number is N . The
standard rules for angular momenta apply (Appendix B in Vol. 1). N  2  = 22 is
the squared angular momentum in the first rotationally excited state (N = 1), the
moment of inertia in the ground state is I0 = M̄R02  M̄r2 . Thus, we estimate the
rotational energy:

 
N
2
2 me
WN = 2
 2
= We
2M̄R M̄r M̄
 10−4 –10−3 We  1 meV to 10 meV. (3.7)

The corresponding transitions are in the far infrared and microwave spectral range
(ν̄ = 1 cm−1 to 10 cm−1 ).

Section summary
• We have introduced the molecular Hamiltonian (3.4) and derived some rough
estimates for typical molecular energies, being several eV for the electronic
part, 0.1 eV for the molecular vibration and in the 1 meV –10 meV region
for rotational motion.

3.2 B ORN O PPENHEIMER Approximation

3.2.1 Molecular Potentials

The big difference between electronic and nuclear energies suggests to separate
electronic and nuclear motion. This was first proposed by B ORN and O PPEN -
HEIMER (1927)who expanded the contributions of the nuclei to the Hamiltonian
4
into a series of me /M̄. They found that nuclear vibrations correspond to 2nd, ro-
tation to 4th order, while 1st and 3rd order disappear. The B ORN -O PPENHEIMER
(BO) approximation turns out to be an excellent approximation and forms the basis
of all molecular structure theory.
140 3 Diatomic Molecules

The first step is a product ansatz composed by the wave functions of electrons
φ(r 1 , r 2 , . . . , r N ) and those of atomic nuclei ψ(R),
Ψ (r 1 , r 2 , . . . , r N , R) = φ(r 1 , r 2 , r 3 , . . . , r N )ψ(R) ≡ φ(r)ψ(R), (3.8)
where r refers to the entirety of electronic and R to all nuclear coordinates.
The key idea is to consider – on a timescale relevant for the rapid motion of the
electrons – the nuclei as fixed in space. The electronic part of the Hamiltonian (3.4)
at a fixed value of R is
 
H(el) = H − Tn (R) = Tr + V (r; R) , (3.9)
and the corresponding S CHRÖDINGER equation is now written as
 
(el) φγ (r; R) = Tr + V (r; R) φγ (r; R) = Vγ (R)φγ (r; R).
H (3.10)
For a start, BO approximation considers R simply as a ‘parameter’. The semicolon
in the electronic wave function φγ (r; R) and in the potential V (r; R) is meant to
emphasize this assumption. But in the back of our minds we have to remember
that electronic and nuclear coordinates are, strictly speaking, nonseparable: in the
potential V (r, R) given by (3.3), the r and R coordinates are nicely intertwined –
except for the C OULOMB repulsion terms of which ZA ZB /R can be treated as a
simple additive constant to the electronic energy.
In the framework of the BO approximation, (3.10) is solved for each, fixed value
of R independently. This leads to a set of electronic quantum numbers γ . The elec-
tronic energy Vγ (R) for each set of γ is a continuous function of R, called molec-
ular potential (in the diatomic case), and potential hypersurface (in the general,
multidimensional case).
As the velocities of the electrons are orders of magnitude higher than those of the
nuclei, the electrons orbit many times around the nuclei before these have moved
significantly. Considering R as fixed is thus a very reasonable approximation for
solving (3.10).
All preceding discussion holds for any number of atoms. In the following we
shall, however, concentrate for simplicity on the diatomic case, the generalization
being straight forward. Then R stands for the relative coordinate R = R A − R B
of the two atomic nuclei, and the electronic energy will depend exclusively on the
internuclear distance R = |R|. By solving (3.10) for a range of R, one thus obtains
for each γ a molecular potential Vγ (R), along with the corresponding electronic
wave function φγ (r; R).4
In analogy to atoms, the indices α, β, γ define a set of quantum numbers which
characterize the electronic charge cloud. The electronic wave functions φγ form
again a complete, orthonormal set

φγ  |φγ  = φγ∗  (r; R)φγ (r; R)d3 r = δγ  γ . (3.11)

4 Depending on the required accuracy, r refers usually only to the coordinates r i of the most im-
portant electrons, e.g. to the valence electrons, and one ignores the inner shells.
3.2 B ORN O PPENHEIMER Approximation 141

Fig. 3.4 Schematic overview COULOMB repulsion

of a molecular potential the of nuclei ∝1/R
bonding (full red line) and Vγ (R)
antibonding case (dashed
line). At large internuclear
distances R an attractive, not overlapping
albeit very weak polarization electronic clouds
=> antibonding
interaction is always
dominant VAN DER WAALS
attraction ∝- 1/ R 6
overlapping
electron clouds
Vγ (∞)
=> molecular bonding R

De
R0
united equilibrium separated
atoms distance atoms

3.2.2 General Form of Molecular Potentials

Electronic energies, for diatomic molecules called potentials Vγ (R), may be ob-
tained from (3.10) for any set of electronic quantum numbers as a continuous func-
tion of R. We postpone to Sect. 3.5 a more detailed discussion about how exactly to
compute them. For the moment we just give a qualitative overview. With (3.3) we
may write the potentials derived from the electronic S CHRÖDINGER equation (3.10)
ZA ZB
Vγ (R) = Uγ (R) + , (3.12)
R
now again in a.u. The C OULOMB repulsion of the nuclei dominates at short inter-
nuclear distances R while the term Uγ (R) is due to the electrons and is dominantly
attractive if the molecular electron configuration is bonding.
A typical molecular potential has several characteristic regions as illustrated in
Fig. 3.4, for a bonding and for an antibonding diatomic potential. At large distances
the polarizability of the atomic electron clouds leads to an attractive (albeit usually
very weak) VAN DER WAALS potential ∝ −R −6 , as discussed in Sect. 8.3, Vol. 1.
The repulsive C OULOMB interaction plays an important role in a systematic treat-
ment of molecular orbitals – even though the limit of united atoms (more precisely
atomic nuclei) is never reached. We shall come back to this in Sect. 3.5.4.
In between these limiting cases the interaction of the electron shells is particu-
larly strong. Its nature determines whether two atoms can form a molecule or not:
as we shall proof in Sect. 3.5 a strong overlap of the electronic charge clouds leads
to bonding, little or no overlap to antibonding potentials. As sketched in Fig. 3.4
the bond energy (or dissociation energy) De and the equilibrium distance R0 corre-
spond to the potential minimum, zero energy referring here to completely separated
atoms, Vγ (∞) = 0.
142 3 Diatomic Molecules

The behaviour of the interaction potentials discussed here is of general nature,

and pertains also to large molecules and their coordinates.

3.2.3 Nuclear Wave Functions

In BO approximation, the nuclear wave functions can be evaluated without problems

if the potentials are known. We rewrite the Hamiltonian (3.4):

 = − 1 ∇ 2R + H
H (el) . (3.13)
2M̄
The S CHRÖDINGER equation (3.5) for the molecule as a whole is then

φγ (r i ; R)ψvN (R) = − me ∇ 2R φγ (r i ; R)ψvN (R) + H

H e φγ (r i ; R)ψvN (R)
2M̄
= Wγ vN φγ (r i ; R)ψvN (R), (3.14)

with Wγ vN = W being the total energy of the molecule. The indices vN refer to
the nuclear motion as we shall see in a moment. Now let us have a detailed look
at the components of this S CHRÖDINGER equation. The electronic term is indeed
given by (3.10), here explicitly as H(el) φγ ψvN = Vγ (R)φγ ψvN , since H(el) acts
as differential operator only on the electronic coordinates. However, the kinetic
energy operator for the nuclei, −∇R2 /(2M̄), acts on both factors of the product
φγ (r i ; R)ψvN (R)φγ ψvN . The S CHRÖDINGER equation (3.14) for the whole sys-
tem thus becomes:
 
 me 2
H φγ (r i ; R)ψvN (R) = φγ − ∇ R ψvN + Vγ (R)φγ ψvN (3.15)
2M̄
 
me 2 me
− ψvN ∇ R φγ + 2(∇ R φγ )(∇R ψvN ) (3.16)
2M̄ 2M̄
= Wγ vN φγ (r i ; R)ψvN (R).

So far, the derivation is still completely correct. Now, in B ORN -O PPENHEIMER

approximation one simply neglects ∇R φγ completely, i.e. the terms (3.16) are
dropped: the electronic wave function φγ (r i ; R) changes only very little with
the nuclear distance R (in particular so near equilibrium distance R0 ). Hence,
∇R φγ (r i ; R) is neglected, and even more so the second derivative.
To obtain a more quantitative feeling we consider the following: The magnitude
of ∇R φγ is of the same order of magnitude as ∇r i φγ , since the same regions of the
molecule are involved in the evaluation of the gradients. For clarity we use now SI
units, write the electron momentum pe = ∇r i φγ , and compare

2 2 2 2 p2 me pe2 me
∇ R φγ  ∇ r i φγ  e =  Vγ ,
2M̄ 2M̄ 2M̄ M̄ 2m e M̄
3.2 B ORN O PPENHEIMER Approximation 143

i.e. we neglect an energy on the order of 10−3 . . . 10−5 Vγ . As a matter of fact, one
finds that the B ORN -O PPENHEIMER approximation is a surprisingly excellent ap-
proximation, far beyond what one might expect from this estimate!
Thus, we have to solve for the nuclear motion (now back to a.u.)
 
1 2
φγ − ∇ R ψvN + Vγ (R)φγ ψvN = Wγ vN φγ ψvN .
2M̄
By multiplying this from the left with φγ∗  , integrating over r, and using the or-
thogonality of the electronic wave functions (3.11), we obtain the S CHRÖDINGER
equation for the nuclear wave function ψvN (R) and its eigenvalues Wγ vN :
me 2
− ∇ R ψvN (R) + Vγ (R)ψvN (R) = Wγ vN ψvN (R). (3.17)
2M̄
The Hamiltonian for the nuclear motion is thus

Hn = − me ∇ 2R + Vγ (R). (3.18)

2M̄
The R dependent eigenvalues Vγ (R) of the electronic S CHRÖDINGER equation
(3.10) thus constitute the potential of the S CHRÖDINGER equation for the motion of
the nuclei (3.17).

3.2.4 Harmonic Potential and Harmonic Oscillator

Our next goal is to understand the nuclear motion and describe it in quantum me-
chanical terms. Traditionally, several steps of approximation are taken: again one
tries to separate different types of motion. A first step is to expand the potential into
a TAILOR series and see how far the lowest order approximation leads. To study
the oscillations of the molecule, the potential is expanded around its equilibrium
distance R0 :
 
dVγ  1 2 
2 d Vγ 
Vγ (R) = Vγ (R0 ) + (R − R0 )  + (R − R0 ) + ··· .
dR R=R0 2 dR R=R0
2

At the potential minimum dVγ /dR|R=R0 = 0, and, neglecting higher terms, we ob-
tain a harmonic potential
1
Vγ (R) = Vγ (R0 ) + k(R − R0 )2 (3.19)
2

with the force constant k = d2 Vγ /dR 2 R=R .
0

According to classical mechanics such a potential leads to harmonic oscillations

with an angular frequency

ω0 = k/M̄. (3.20)
144 3 Diatomic Molecules

Fig. 3.5 Harmonic Wv / ħω0

oscillator: Potential energy v=4
(full black line), total energy 4.5
(dashed, black lines) and v=3
eigenfunctions Rv (R) for the
vibrational sates v = 0 . . . 4 3.5
v=2
(full, red lines, shifted in
height for clarity) 2.5
v=1
1.5
v=0
0.5

-4 -2 0 2 (R - R0 ) / l

The harmonic oscillator is probably one of the best treated objects on all levels of
physics education. We just summarize the essential results. In quantum mechanics
one obtains the wave functions Rv (R) and the energy eigenvalues Wv from the
one-dimensional S CHRÖDINGER equation:

2 d 2 M̄ω02
− + (R − R 0 )2
Rv (R) = Wv Rv (R). (3.21)
2M̄ dR 2 2
Introducing a characteristic length l (typically between 0.1 and 0.25a0 )

l = /(M̄ω0 ) and setting x = (R − R0 )/ l, (3.22)

one may write (3.21) in dimensionless form:


d2 2Wv
+ − x Rv (x) = 0.
2
(3.23)
dx 2 ω0
Solutions are the H ERMITE functions hv (x) which are related to the H ERMITE poly-
nomials Hv (x) of degree v by
 
Rv (x) ≡ hv (x) = exp −x 2 /2 Hv (x).

Table 3.1 H ERMITE v hv (x)

functions for vibrational
1 −x 2 /2
states v = 1 . . . 4 0 √
4πe

√
1 √ 2 −x 2 /2
4 π xe

√ 1√ e−x
2 /2
2 (2x 2 − 1)
2 π

√ 1√ xe−x
2 /2
3 (2x 2 − 3)
3 π

√1√ e−x
2 /2
4 (4x 4 − 12x 2 + 3)
2 6 π
3.2 B ORN O PPENHEIMER Approximation 145

In normalized form they may be derived from

(−1)v   dv  
hv (x) =  √ exp x 2 /2 v
exp −x 2 . (3.24)
v
2 v! π dx

The corresponding energy eigenvalues of the harmonic oscillator are

Wv = ω0 (v + 1/2) for v = 0, 1, . . . . (3.25)

The wave functions for the four lowest levels are sketched in Fig. 3.5 and sum-
marized in Table 3.1. Note that they alternate between being symmetric and anti-
symmetric in respect of the equilibrium distance R0 . We also emphasize the finite
extension of the ground state wave functions (v = 0, zero point oscillation), which
is a pure G AUSS function. The corresponding zero point energy of the ground state
is Wmin = ω0 /2.
Figure 3.6 shows an example for a rather high vibrational quantum number v =
20. One sees that in this case the probability density at the boundaries increases
substantially – corresponding fully to the longer time a classical harmonic oscillator
spends at the classical turning point.

Fig. 3.6 Harmonic oscillator Wv / ħω0

21.0
in the v = 20 state: potential
energy (full black lines on the v = 20
left and right), total energy
(dashed black line) and
eigenfunction (red dashed).
The full red line gives the
square of the wave functions
and hence the density 20.5
probability as a function of
internuclear distance R -6 -4 -2 0 2 4 (R - R0 ) / l

3.2.5 M ORSE Potential

It is instructive to compare the ideal harmonic potential with a realistic molecule.

We choose CO for whose ground state an experimentally very well determined, so
called RYDBERG -K LEIN -R EES (RKR) potential is available (we shall come back
to the RKR method in Sect. 3.4.6). In Fig. 3.7 the experimental data points are
indicated by crosses. For each measured vibrational state one pair of such data point
exists, one for the inner and one for the outer turning point. Some vibrational levels
are indicated.
The harmonic approximation of this potential, the dashed red line in Fig. 3.7(a),
is obviously of very limited value, and only allows a description of the lowest vibra-
tional levels (v = 0, 1). There are a number of approaches for an analytic approxi-
146 3 Diatomic Molecules

mation to reality. Often the potential introduced by P.M. M ORSE (1929) is used:
 
VM (R) = De e−2a(R−R0 ) − 2e−a(R−R0 ) . (3.26)

This M ORSE potential offers already some flexibility with three parameters De
(bond energy), R0 (equilibrium distance) and a (some kind of stiffness of the po-
tential). It reproduces the limits correctly and is rather simple to handle in compu-
tations. For CO the fit to the experimental data by a M ORSE potentials (full red line
in Fig. 3.7) looks surprisingly good.5
As indicated in Fig. 3.7 one usually defines zero energy for the separated atoms in
the electronic ground state, i.e. Vγ (∞) = 0, and obtains for the potential minimum
Vγ (R0 ) = −De . Alternatively one also uses the form
 2
VM (R) = De 1 − e−a(R−R0 ) ,

which differs from (3.26) just by a shift of zero energy to −De . The true ground
state bond energy required to dissociate the molecule (from its vibrational ground
state v = 0 as indicated in Fig. 3.7) is in any case

D00 = De − ω0 /2.

V(R) / eV R / nm

0.10 0.14 0.18 0.22 0.26 0.3 0.10 0.14 0.18

0 0
harmonic (a) ( )
-2 -2
v = 28 Lennar -Jones
-4 -4
v = 28
v = 27
-6 -6
De = 11.11eV
-8 -8
0
v =1 D0 = 10.91eV
- 10 -10

v=0

Fig. 3.7 Potential of the CO ground state. Crosses give the experimentally determined points of
the RKR potential according to F LEMING and R AO (1972), M ANTZ et al. (1971). A M ORSE po-
tential has been fitted to it (full, red line) with a bond energy De = 11.108 eV and an equilibrium
distance R0 = 0.11282 nm. This potential may be compared to (dashed red lines) (a) a harmonic
potential of equal vibrational frequency in the ground state or (b) a 12, 6 L ENNARD -J ONES poten-
tial

5 However, experimental precision is so far advanced that the M ORSE potential does not suffice
state-of-the-art requirements. The accuracy of the data points given in the literature is in some
instances as good as 1 ppm.
3.2 B ORN O PPENHEIMER Approximation 147

Table 3.2 Potential parameters for some characteristic diatomic molecules: Reduced mass M̄,
equilibrium distance R0 , dissociation energy D00 = De − ω0 /2 in respect of the ground vibrational
state, its vibrational energy ω0 , and its dipole momenta Dγ v
Molecule M̄/ u R0 / nm D00 / eV ω0 / eV Dγ v /10−30 C m
H+
2 0.504 0.1052 2.651 0.2714
H2 0.504 0.07414 4.478 0.5156
D2 1.0071 0.07415 4.556 0.37095
7 Li1 H 0.8812 0.15957 2.4287 0.16853 19.6256
1 H35 Cl 0.9796 0.12746 4.433 0.3577 3.6979
N2 7.0015 0.109768 9.759 0.28888
14 N16 O 7.466 0.115077 6.497 0.23260 0.52943
O2 7.997 0.120752 5.115 0.19295
12 C16 O 6.8562 0.112832 11.09 0.26573 0.3662
Na2 11.4949 0.30788 0.720 0.01955
23 Na35 Cl 13.870 0.23608 4.23 0.0448b 30.025
Cl2 17.4844 0.1987 2.479 0.0687
As far as not otherwise mentioned, according to H UBER and H ERZBERG (1979)
a L OVAS et al. (2005)
b R AM et al. (1997)

For general orientation and further reference, Table 3.2 summarizes the relevant
potential parameters for some characteristic diatomic molecules.
The harmonic approximation to the M ORSE potential is
 
VM (R) = −De 1 − a 2 (R − R0 )2 + · · ·

and with (3.19) and (3.20) the stiffness a is related to the spring constant k close to
the potential minimum and to the fundamental frequency ω0 by
 
k = 2De a 2 and ω0 = k/M̄ = a 2De /M̄. (3.27)

The anharmonicity of the potential can be reflected as a decrease of the spring

constant k with increasing v. Consequently angular frequency ωv and vibrational
energy ωv decrease with v: the difference between neighbouring vibrational levels
is smaller for higher energies. For the same reason the average nuclear distance
increases with vibrational excitation:
   
R(v + 1) > R(v) > · · · > R0 .

This is also the physical cause for thermal expansion of solids: with increasing
temperature the average vibrational energy increases, hence the average value of v
increases as well, and so does R.
148 3 Diatomic Molecules

Note, however, that the number of bound states in such a “potential well” is
always finite – even though the energy spacing between neighbouring levels gets
very small close to the dissociation limit. We recall that, in contrast, the number of
bound electronic states in an atom is infinite. This is entirely due to the particular
nature of the 1/r C OULOMB potential. All other (bonding) potentials support at
most only a finite number of bound states.

3.2.6 VAN DER WAALS Molecules

True chemical bonding due to formation of molecular orbitals will be discussed

in Sect. 3.5. There are, however, also other, weaker forces by which atoms can be
attracted to each other. C OULOMB attraction between atoms and polar molecules
is one of them. Another important, attractive interaction is the so called VAN DER
WAALS (vdW) interaction, or dispersion interaction, which has already been in-
troduced in Sect. 8.3, Vol. 1 and was briefly mentioned in Sect. 3.2.2. It acts even
between neutral atoms and molecules without a dipole, and is caused by mutual
polarization of the electron charge clouds. The vdW interaction can be estimated
according to (8.93), Vol. 1 as V (R) ∝ −αA αB R −6 with αA and αB being the po-
larizabilities of the two interacting atoms A and B. It is approximately additive and
acts pairwise. The vdW interaction leads to (among other things) deviations from
the ideas gas law. As well known, the state equation of real gases is given by the
VAN DER WAALS equation
 
a
p + 2 (V − b) = RT . (3.28)
V

While the parameter b (the so called co-volume) reflects the finite extension of
atoms and the repulsive part of the intermolecular potentials and decreases the avail-
able volume, the term a/V 2 (the so called cohesive pressure) is attributed to VAN
DER WAALS forces at large distances which effectively increase the external pres-
sure p.
The vdW interaction is about a factor 102 to 103 smaller than typical bond ener-
gies. For example, for the non-bonding rare gas system Ar· · · Ar it amounts to about
0.08 eV in the minimum, while Cl2 (in the periodic table directly next to Ar) is
chemically bound, albeit very weakly, with De  2.48 eV. VAN DER WAALS inter-
action is typically of the same magnitude as thermal energies at room temperature

 203.6 cm−1 .
kB × 293 K  0.025 eV = (3.29)

Hence, to study it in detail one has to work at very low temperatures. Helium, for
instance, may be liquified at low temperatures (4.3 K) in spite of its closed 1s 2
shell. At sufficiently low temperature He gas forms clusters or little droplets. At any
rate, rare gases have a strong tendency to form atomic clusters at low temperatures.
A popular method to generate such objects is adiabatic expansion of the rare gases
in dense atomic beams.
3.2 B ORN O PPENHEIMER Approximation 149

V(R) / 10-4 Eh He2

1.5
Tang-Toennies
1.0 Lennard-Jones (12,6)

0.5

0.0
De = 0.348
- 0.5
4 8 10 12 R / a0
R 0 = 5.62

Fig. 3.8 Interaction potential between two He atoms. We compare the nearly exact TANG -T OEN -
NIES potential (TANG et al. 1995, red line) with a L ENNARD -J ONES 12, 6 potential (3.30) fitted to
it (black line). Note that only two free parameters De = 0.34784 × 10−4 Eh ( = 0.9465 meV) and
R0 = 5.62a0 determine the L ENNARD -J ONES potential. Both potentials are – on the scale used
here – nearly indistinguishable

In this manner diatomic, very weakly bound combinations of practically all

atoms may be generated, even if they do not form chemically bound molecules.
For these so called VAN DER WAALS molecules, in particular for rare gas pairs,
the interaction is usually approximated very well by a so called L ENNARD -J ONES
potential, also called 12, 6 potential:
De 2De
V (R) = 12
− . (3.30)
(R/R0 ) (R/R0 )6
It contains only two free parameters, R0 and De , and is thus not suitable for mod-
elling potentials of chemically bound molecules – as documented by the dashed red
line in Fig. 3.7(b). However, 12, 6 potentials are quite appropriate to describe VAN
DER WAALS molecules. In the normalization used here, R0 is the equilibrium dis-
tance (also called VAN DER WAALS contact distance) and De is the depth of the
potential well.
For the example of the He-He system, Fig. 3.8 documents excellent agreement
between the TANG -T OENNIES potential, which has been determined experimentally
with extremely high precision (TANG et al. 1995), and the 12, 6 potential fitted to it..
The graphical accuracy of Fig. 3.8 hardly allows to see any difference. The experi-
mental data are, however, much more precise and allow decisive conclusions about
the He2 ‘molecule’. For many years it was not clear whether the simplest rare gas
dimer forms a bound state at all. To obtain an estimate we recall the one dimensional
potential box. According to (2.52) in Vol. 1 the lowest state of a particle with mass
M̄ has the energy W1 = h2 /(8M̄L2 ). For such a state to exist the potential depth
must at least be W1 , or De L2 × 8M̄/ h2 > 1. With De = 0.348 × 10−4 (all in a.u.)
the effective extension of the He-He potential would have to be L > 6.24a0 , which
appears not unrealistic in view of Fig. 3.8. This is of course only a very rough es-
timate which cannot replace an exact computation. In diffraction experiments with
150 3 Diatomic Molecules

Table 3.3 VAN DER WAALS Element vdW radius/ nm

radii of some important atoms
H 0.120
He 0.140
C 0.170
N 0.155
O 0.152
P 0.180
S 0.180

ultracold He atomic beams the existence of the He2 molecule has been established
beyond any doubt. According to G RISENTI et al. (2000) the presently most accu-
+0.3
rate value for the bond energy of its one existing state is D00 = (1.1−0.2 ) mK, i.e.
−7
 10 eV(!), its average size is R = (5.2 ± 0.4) nm(!). Clearly, He2 is a quite
pathological case of a molecule!
By a combination of VAN DER WAALS contact distances between different part-
ner atoms one estimates the so called VAN DER WAALS radii of individual atoms.
Table 3.3 presents examples for some important elements. A graphical overview has
already been communicated in Fig. 3.3, Vol. 1.

Section summary
• We have introduced the B ORN -O PPENHEIMER approximation as fundamental
for our quantitative understanding of molecules. It is based on the fact that
electrons move much faster than nuclei. In a classical picture, electrons circle
on their orbits many times before the nuclei have significantly changed their
position R.
• Hence, the electronic S CHRÖDINGER equation (3.10) can, to a very good
approximation, be solved for the electronic coordinates only, with R being
treated as a freely variable parameter. The electronic energy Vγ (R) for a set
of electronic quantum numbers γ is a function R, called molecular potential
(hypersurface for polyatomic molecules).
• Characteristic shapes of Vγ (R) are shown for the diatomic case in Fig. 3.4.
Nuclear motions occurs on these potentials.
• The harmonic potential and the harmonic oscillator were found to be only
a very crude approximation, describing the near equilibrium region and the
lowest vibrational levels of a diatomic molecule reasonably well.
• The M ORSE potential (3.26) was shown to be a useful analytic form for a
chemically bound diatomic molecule, characterized by dissociation energy
De , equilibrium distance R0 and molecular stiffness a related to the funda-
mental frequency ω0 by (3.27).
• Even if there is no chemical bond, neutral atoms experience a long range at-
tractive force caused by mutual polarization, the so called VAN DER WAALS
interaction ∝ −1/R −6 . Based on this and a short range repulsion term ∝
1/R 12 the L ENNARD -J ONES (12, 6) potential is used to describe non-bonding
3.3 Nuclear Motion: Rotation and Vibration 151

Z Z M
Θ Θ R = RA – R B
MA ζ
R A = (M B / M ) R
O
Y O Y
Φ
X
X Φ
MB R B = (MA / M ) R

Fig. 3.9 The ‘reduced nuclear particle’: one transforms the two body problem by introduction of
a reduced molecular mass M̄ = MA MB /M (with M = MA + MB ) into an effective one particle
problem whose motion is described by the relative nuclear position coordinate R

interactions, e.g. between rare gas atoms. He2 was seen to be a very special
case of a weakly bound molecule with De  10−7 eV.

3.3 Nuclear Motion: Rotation and Vibration

3.3.1 S CHRÖDINGER Equation

With our present understanding of molecular potentials it is possible to study the nu-
clear motion in some more detail. To avoid confusions, we switch back to SI units
in the following discussion. The reduced mass M̄ = MA MB /M moves on a spher-
ical symmetric electronic potential Vγ (R). By introducing the relative coordinate
R = R A − R B instead of the individual nuclear coordinates R A = R MB /M and
R B = R MA /M (with M = MA + MB ) one replaces the vibrating nuclear mass, so
to say by one ‘reduced nuclear particle’ as illustrated in Fig. 3.9.6 This leads again to
a one particle S CHRÖDINGER equation very similar to that for the hydrogen atom.
Merely the potential 1/r has to be replaced by Vγ (R), the electronic eigenenergy
which now depends on the distance R between the nuclei. In complete analogy to
(2.108) in Vol. 1 we write (3.17) now:
 
n ψγ vN (R) = H
H R + H rot + Vγ (R) ψγ vN (R) = Wγ vN ψγ vN (R) (3.31)
   2
R = −  1 ∂ R 2 ∂ and H
2
with H rot = N .
2M̄ R 2 ∂R ∂R 2M̄R 2
The quantum number v characterizes the radial motion (vibration). The angular
momentum operator N  of the reduced nuclear particle is a conserved quantity –
just as   commutes according to the usual angular
L for the electron of the H atom. N

6 The electron coordinates too are transformed and referred to the molecular axis (in Fig. 3.9 the ζ

axis).
152 3 Diatomic Molecules

momentum rules with the Hamiltonian. Thus, [H n , N

z ] = 0 and [H
n , N
 2 ] = 0 holds.
 2 is described in analogy to the H atom by
In the position representation N
  
 2 1 ∂ ∂ 1 ∂2
N = − 2
sin Θ + 2
sin Θ ∂Θ ∂Θ sin Θ ∂Φ 2
and its eigenfunctions are the spherical harmonics YN MN (Θ, Φ)

 2 YN MN (Θ, Φ) = 2 N(N + 1)YN MN (Θ, Φ),

N (3.32)

with eigenvalues 2 N(N + 1), where N is an integer. In analogy to the electron

orbital angular momentum in atomic hydrogen, N and MN are now the quantum
numbers of the molecular orbital angular momentum, i.e. of the molecular rotation.
We characterize the angular coordinates of the molecular axis with capital Greek
letters Θ and Φ (to be distinguished from lower case θ and ϕ, which we shall
continue to use for electronic coordinates).
The wave function of the nuclear motion ψγ vN (R) may thus be written as a
product of spherical harmonics YN MN (Θ, Φ) and a radial part:

ψγ vN (R) = R −1 Rγ vN (R)YN MN (Θ, Φ). (3.33)

Insertion into (3.31) leads to the one dimensional S CHRÖDINGER equation


2 d 2
− + Veff (R) Rγ vN (R) = Wγ vN Rγ vN (R) (3.34)
2M̄ dR 2
for the radial motion (vibration), characterized by the vibrational quantum num-
ber v. The effective potential is here

2 N(N + 1)
Veff (R) = + Vγ (R). (3.35)
2M̄R 2
The reduced mass M̄ moves in this effective potential. In principle, one may solve
(3.34) for each N numerically – as usual for reasonable physical boundary con-
ditions. One can thus obtain rotational-vibrational energies Wγ vN and the radial
eigenfunctions Rγ vN (R). It is, however, useful to simplify the problem even fur-
ther.

3.3.2 Rigid Rotor

One exploits again the different time and energy scales: The molecule oscillates
rapidly in comparison to the rotation, as we have seen. Thus, as a good first ap-
proximation, one may assume the nuclear distance R to be fixed. As sketched in
Fig. 3.10 the molecule may assume any arbitrary polar Θ and azimuthal angle Φ
in space. One may identify this fixed value of R with the equilibrium distance in
3.3 Nuclear Motion: Rotation and Vibration 153

Fig. 3.10 Rigid rotor in z

space R Θ MA

ω
y
MB Φ
x

the lowest vibrational state. The S CHRÖDINGER equation (3.31) may then indeed
be fully separated, and we obtain the eigenvalue equation for the rigid rotor:

N2
rot YN MN (Θ, Φ) =
H YN MN (Θ, Φ)
2M̄R 2
= WN YN MN (Θ, Φ). (3.36)

Using (3.32), the rotational energy thus becomes

2
WN = N(N + 1) = hcBN(N + 1) (3.37)
2M̄R 2
with the rotational constant (unit cm−1 )

2 1 2 1
B= = , (3.38)
2M̄R 2 hc 2I hc
I = M̄R 2 being the molecular moment of inertia. The resulting energy scheme is
sketched in Fig. 3.11. In wavenumbers the rotational energies are simply

F (N) = WN / hc = BN (N + 1). (3.39)

The rotational constant B is (with the reduced mass known) a very sensitive measure
for the bond length R0 of the molecule. Table 3.4 summarizes typical values of
important molecules. The subscript in I0 , B0 and ω0 indicates that the constants
given here refer to the lowest vibrational state. We see that spectra from the infrared
to the microwave spectral region are expected.
We must emphasize at this point that the image of a diatomic molecule as a rotat-
ing dumbbell has now to be replaced by the quantum mechanical picture: The spher-

Fig. 3.11 Term scheme of N WN / Bhc

the rigid rotor
4 + 20
8
3 – 12
6
2 + 6
4
1 – 2
0 + 0
154 3 Diatomic Molecules

Table 3.4 Rotational and vibrational constants and corresponding absorption wavelengths for sev-
eral characteristic diatomic molecules. Note that only molecules with a finite dipole moment are
infrared active
Molecule 2 (2I0 )−1 / eV B0 / cm−1 λrot ω0 / eV λvib / µm IR active
H+
2 3.641 × 10−3 29.4 170 µm 0.2714 4.568 no
H2 7.356 × 10−3 59.32 84 µm 0.5156 2.405 no
D2 3.708 × 10−3 29.90 167 µm 0.37095 3.342 no
LiH 9.184 × 10−4 7.4065 675 µm 0.16853 7.357 yes
HCl 1.2945 × 10−3 10.440 479 µm 0.3577 3.466 yes
N2 2.467 × 10−4 1.9896 2.51 mm 0.2888 4.292 no
O2 1.7827 × 10−4 1.4377 3.48 mm 0.19295 6.426 no
CO 2.384 × 10−4 1.9225 2.60 mm 0.26573 4.666 yes
NO 2.103 × 10−4 1.696 1 2.95 mm 0.23260 5.330 yes
Na2 1.913 × 10−5 0.15427 3.24 cm 0.01955 63.4 no
NaCla 2.6938 × 10−5 0.21725 2.30 cm 0.0448 27.68 yes
Cl2 3.02 × 10−5 0.243 2.05 cm 0.0687 18.0 no
As far as not otherwise mentioned according to H UBER and H ERZBERG (1979)
a For the isotopologue 23 Na 35 Cl according to R AM et al. (1997)

N= 0 N = 1 py N = 1 pz N = 1 px

N = 3 MN = 3

real states

Fig. 3.12 The rigid rotor in space: the probability for finding the molecular axis aligned at Θ and
Φ, is given by the absolute square of the spherical harmonics (or linear combinations of these);
here shown for N = 0 and N = 1, the latter having been excited by linearly polarized light; also
shown is the distribution for N = 3, MN = 3 as an example where the molecule has been excited
by circularly polarized light

ical harmonics YN MN (Θ, Φ) describe the probability amplitudes for the angles Θ
and Φ, and |YN MN (Θ, Φ)|2 sin ΘdΘdΦ is the probability to find the molecular axis
aligned between (Θ, Φ) and (Θ + dΘ, Φ + dΦ). We have discussed this already
in detail in Sect. 2.5.3, Vol. 1. As |YN MN |2 does not depend on Φ this probability
is rotational symmetric around the Z-axis. Alternatively the real basis YN |MN | (Θ)
may represent the excitation probabilities more conveniently – if linearly polarized
light has been used for excitation into the NMN states. In Fig. 3.12 the correspond-
ing probabilities are shown for N = 0 and N = 1. The image of a rotating molecule
3.3 Nuclear Motion: Rotation and Vibration 155

Table 3.5 Rotational temperatures for some diatomic molecules

Molecule H2 D2 HCl N2 O2 CO Cl2
Trot / K 93 47 15 2.89 2.08 2.77 0.4

cannot be recognized in these probability distributions. In particular, the rotational

ground state with N = 0 has an isotropic angular distribution. This is also true for
higher N if all 2N + 1 substates are equally populated, e.g. in the thermal equilib-
rium.
Only for large values of N and MN the wave function may reflect the classical
image of a molecule rotating around the z-axis, depicted in Fig. 3.3. Such a wave
function (e.g. N = 3, MN = 3) can be prepared by absorption of several circularly
polarized photons. If the molecule is excited by linearly polarized light, states with
positive and negative MN are populated with equal probability – independent of
which coordinate system is used to describe this situation – and the expectation
value of Nz disappears, Nz  ≡ 0. In summary, a truly rotating molecule requires
rather special excitation conditions.

3.3.3 Population of Rotational Levels and Nuclear Spin

We want to address now the population probabilities of rotational states in diatomic

molecules. Apart from their significance for molecular spectroscopy they play an
important role in statistical mechanics and thermodynamics. Of course they depend
on the rotational energies and on the degeneracies of the levels. They are a function
of temperature and provide the basis for the derivation of specific heat capacities.
If we compare the rotational constants in Table 3.4 with the thermal energy, say
at room temperature, kB T  0.025 eV, we see that rotational energies (3.37) are
usually small in comparison to thermal energies. Thus, at room temperature, (3.29),
many rotational levels are populated. We define a characteristic rotational temper-
ature Trot for a molecule by

2
kB Trot = = Bhc. (3.40)
2I

This rotational temperature plays an important role when determining the specific
heat capacity as we shall see in a moment. Typical values are given in Table 3.5.
In analogy to n levels in the atomic case ( being the orbital angular momen-
tum), the degeneracy of rotational levels is gN = 2N + 1. The population probability
of a rotational level N is given by a B OLTZMANN distribution
   
gN WN (2N + 1) N (N + 1)Trot
w(N, T ) = exp − = exp − , (3.41)
ZN kB T ZN T
156 3 Diatomic Molecules

w(N) w(N)
ortho H2
CO 0.6
0.06

0.04
(a) 0.4 (b)

0.2 para H2
0.02

0.00 N 0.0 N
0 10 20 30 0 1 2 3 4 5

Fig. 3.13 Relative population of rotational levels N in diatomic molecules at room temperature
(293 K) in thermodynamic equilibrium: (a) CO – nuclear spin statistics is irrelevant in this case
since the two nuclei are distinguishable particles; (b) for H2 (nuclear spin I (1 H) = 1/2) – here
states with even and odd N are populated differently (red: para H2 ↑↓, grey: ortho H2 ↑↑), see
text

with the partition function

∞

ZN (T ) = (2N + 1)e−WN /kB T (3.42)
N =0
 ∞ T
→ (2N + 1)e−N (N+1)Trot /T dN = for T Trot .
0 Trot

As examples we consider carbon monoxide and molecular hydrogen.

For the heteronuclear molecule CO the atoms are distinguishable, and nuclear
spin statistics does not play a role. Inserting the (very low) rotational tempera-
ture Trot = 2.77 K into (3.41) we obtain the population probabilities depicted in
Fig. 3.13(a). We notice that due to the degeneracy (2N + 1) of rotational levels the
energetically lowest level N = 0 is by no means the most populated one.
In contrast to heteronuclear molecules, for homonuclear molecules one has to
account for the indistinguishability of atomic nuclei. If the constituent nuclei are
fermions, the PAULI principle demands the total wave function (nuclear spin ×
nuclear rotation × electronic state) to be antisymmetric in respect of nuclear ex-
change. For bosons the wave function has to be symmetric. In respect of the rota-
tional wave function YN MN (Θ, Φ), exchange of two nuclei is equivalent to inversion
at the centre of mass. Exchange symmetry is thus identical with parity and is given
by (−1)N , as indicated by + and − in Fig. 3.11. Correspondingly, in the case of
two fermions the nuclear spin function of the molecule must be odd for even N ,
while it must be even for odd N . For bosons the opposite holds. For bosons with
nuclear spin I = 0, as e.g. in the case of He or O, the nuclear spin function is
always symmetric, thus, the remainder of the wave function has to be symmetric
too.
Particularly clear is the case of H2 , being a fermion pair with the nuclear spins
I = 1/2. The electronic ground state is characterized by 1 Σg+ (see Sect. 3.6) which
3.3 Nuclear Motion: Rotation and Vibration 157

Fig. 3.14 Relative fraction 1.0

of ortho and para H2 in a gas
of hydrogen molecules as a
ortho H2 (↑↑)
function of absolute 0.75
temperature; red: para – grey:
ortho

w(T )
0.5

para H2 (↑↓)
0.25

0.0
0 100 200 T /K

is symmetric in respect of reflection at a plane through the nuclear axis but also in
respect of exchange of the nuclear coordinates. The nuclear spin function may be a
singlet or a triplet (analogous to the electron states in excited atomic He, see Chap. 7,
Vol. 1). To obtain an overall antisymmetric wave function for even rotational states
(N = 0, 2, 4 . . . ) the nuclear spin function must be antisymmetric, i.e. be a singlet.
One speaks of para hydrogen (p-H2 ). In contrast, ortho hydrogen (o-H2 ) is char-
acterized by rotational states with N = 1, 3, 5, . . . belong to nuclear spin triplets.
The population density (3.41) has to be multiplied then by the spin state degeneracy
gS = (2S + 1), i.e. with gS = 3 and = 1 for triplet and singlet states, respectively.
The resulting populations for p-H2 and o-H2 at room temperature are depicted in
Fig. 3.13(b).
Transitions between p-H2 and o-H2 are very improbable under normal conditions
in a gas. One may almost speak of two different species which are stable over days,
once generated. In thermodynamic equilibrium, the respective probabilities w(T )
to find either species is shown in Fig. 3.14 as a function of temperature. For very
low temperatures only para H2 is found (rotational ground state) while at higher
temperatures the ratio between p-H2 and o-H2 approaches the value 1:3 reflecting
the ratio of the spin state degeneracy gS .
One can generate para H2 from normal H2 gas by cooling it below 20 K and
bringing it into contact with a catalyzer, e.g. Fe III containing substances, which
support collision induced singlet-triplet transitions (magnetic interaction), and thus
achieving thermodynamical equilibrium. As shown in Fig. 3.14 for low tempera-
tures this equilibrium corresponds to pure para-H2 . After this process one heats the
gas carefully up again without catalyzer(!). Only states of the para system will then
be populated thermally, e.g. states with N = 0, 2, 4 . . . . It takes many hours, even
days before the thus prepared p-H2 (without catalyzer) returns into thermodynamic
equilibrium.
As already mentioned, nuclear spin statistic plays an important role in the inter-
pretation of rotationally resolved electronic or R AMAN spectra. We shall come back
to this in Sect. 5.6.6.
158 3 Diatomic Molecules

3.3.4 Specific Heat Capacity

At this point, a brief discussion of CV (T ) for diatomic molecules is thus in order.

The relevance of the above discussion for the specific heat capacity CV of molecules
is evident: For very low temperatures T Trot rotation cannot be excited and thus,
does not contribute to the specific heat capacity. Hence the temperature dependence
of this macroscopically relevant quantity is a direct consequence of the rotational
structure, while at higher temperatures molecular vibration enters.
In the following discussion we assume to be well above the critical temperature
(1.71), Vol. 1 for B OSE -E INSTEIN condensation, so that B OLTZMANN statistics de-
scribes the molecular gas well. In statistical thermodynamics one derives the char-
acteristic observables from the partition function
∞
 
Wi
Z(T ) = gi exp − (3.43)
kB T
i=1

where gi is the degeneracy of level i with the energy Wi . The summation is over all
energy levels, if necessary, it has to be replaced by an integration and gi becomes
the density of states. The average internal energy of a system with N molecules
may be expressed in terms of the partition function as
∞
N W
− i N kB T 2 ∂Z ∂ ln Z
U  = Wi gi e kB T = N = kB T 2 .
Z Z ∂T ∂T
i=1

From this one derives the molar heat capacity (specific heat capacity per mol at
constant volume). Setting N kB = R (general gas constant) it is given by
  
∂U   ∂ 2 ∂ ln Z ∂ RT 2 ∂Z
CV = = RT = . (3.44)
∂T V =const ∂T ∂T ∂T Z ∂T

From this expression we first derive the specific heat of an ideal gas due to trans-
lational motion. The partition function of a freely moving atomic particle without
internal energy is given by the M AXWELL -B OLTZMANN distribution:
 ∞    
mv 2 π kB T 3/2
Z(T ) ∝ v 2 exp − dv = .
0 2kB T 2 m

When inserting this into (3.44) we obtain the well known contribution from

CV 3
translational motion, = , (3.45)
R 2
reflecting the equipartition law with an average energy kB T /2 per molecule for each
of the three translational degrees of freedom.
3.3 Nuclear Motion: Rotation and Vibration 159

3.5 Cv / R 3.5
H2
CO vi ration
3.0 thermal 3.0
T/K
eq ili ri m (c)
2.5 2.5 2.5
CO rotation 0 2000 4000 6000
(a)
2.0 H2 2.0
( )
3:1 ortho-para
mixt re
1.5 1.5
0 200 400 0 10 20

Fig. 3.15 Molar heat capacity CV for diatomic molecules (in units of the general gas constant R).
At very low temperature only the translational degrees of freedom are active (CV = 1.5R). The
rotational contribution rises up to R; (a) for H2 thermal equilibrium conditions are compared with
a mixture of ortho to para H2 of 1:3 as found at higher temperatures; (b) shows the rotational
contribution for CO and (c) illustrates how CV rises again at higher temperatures due to vibrational
excitation, in the limit again by R

Next we evaluate the rotational contribution to CV for a diatomic molecule. If

spin statistics does not play a role (heteronuclear molecules) the partition function is
given by (3.42). If it does, one has to include the nuclear spin degeneracy factor gS :

∞
  
N (N + 1)Trot
ZN (T ) = gS (N )(2N + 1) exp − .
T
N =0

For example, for the hydrogen molecule in thermal equilibrium we set gS (N ) =

2 − (−1)N , while for para H2 we have to set gS (N ) = (1 + (−1)N )/2 and for ortho
H2 gS (N ) = 3(1 − (−1)N )/2. To evaluate (3.44) we have to resort to numerical
evaluation. Fortunately, since the rotational energies increase ∝ N 2 , higher rota-
tional levels are progressively less populated and one obtains reasonable accuracy
with a moderate number of N in the sum. In the limit of high temperatures, T Trot ,
rotational excitation of a diatomic molecule always contributes R.
The results are depicted for H2 and CO in Fig. 3.15(a, b). H2 with Trot = 93 K
features a rotational contribution which becomes noticeable at rather high tempera-
tures. CV is shown for H2 in thermal equilibrium (i.e. in the presence of a catalyzer)
and for a mixture 1:3 of ortho to para H2 – corresponding to normal H2 at room
temperature quickly cooled without a catalyzer. Both functions of course converge
for higher temperatures and have a limiting value of CV = R, corresponding to an
internal energy for the two degrees of rotational freedom of 2 × kB T /2 per molecule.
In contrast, CO with Trot = 2.77 K reaches this limit already at rather low tempera-
tures.
160 3 Diatomic Molecules

In Fig. 3.15(c) the contribution of vibrational excitation to CV is shown for the

example of CO. Using the harmonic oscillator model the partition function is
  
∞
 1 ω0 exp(− kω
BT
0
)
Zv (T ) = exp − +v = ω
, (3.46)
v=0
2 kB T 1 − exp(− kB T0 )

and with (3.44) the contribution of vibrational excitation to the molar heat capacity
is
 
CV Tvib 2 exp(−Tvib /T ) ω0
= , with Tvib = . (3.47)
R T [1 − exp(−Tvib /T )]2 kB
In the limit of high temperatures T Tvib the denominator becomes (Tvib /T )2 and
again CV → R (we remember the two degrees of freedom for vibration mentioned
in Sect. 1.3.3, Vol. 1, i.e. per molecule 2 × kB T /2 energy).
For multi-atomic molecules with several vibrational degrees of freedom, (3.47)
is the starting point for evaluating the specific heat capacity. One has to sum (3.47)
over all vibrational frequencies. If the molecule possesses a quasi-continuum of
vibrational levels (e.g. a large organic molecule) one has to multiply (3.47) with the
density of states and to integrate over all frequencies. For solid state materials the
lattice contribution to the specific heat is derived in just the same manner (theories
of E INSTEIN and D EBYE).

3.3.5 Vibration

Vibrational energies are typically two to three orders of magnitude larger than ro-
tational energies (see Table 3.4), so that one may separate the two forms of motion
in a 1st order approximation as indicated above. Typically, the anharmonicity of the
potentials is not too dramatic, so that for low vibrational excitation one may indeed
assume that the average bonding lengths is constant, R = R0 , and the rotational
energy derived for the rigid rotor can just be inserted into the radial S CHRÖDINGER
equation (3.34) for the nuclear motion. One replaces

2 2
N (N + 1) in (3.35) by WN = N (N + 1) = B0 hcN(N + 1).
2M̄R 2 2M̄R02
The total energy Wγ vN of the rotating oscillator may thus be understood as the sum
of this vibrational energy Wγ v in the potential Vγ (R) and the rotational energy WN .
One has to solve the radial wave equation without centrifugal potential:

2 d 2
− + Vγ (R) Rv (R) = Wγ v Rv (R). (3.48)
2M̄ dR 2
More explicitly, the total energy of a ro-vibrational state with vibrational and
rotational quantum numbers v and N , respectively, is given by

Wγ vN = Vγ (R0 ) + WvN . (3.49)

3.3 Nuclear Motion: Rotation and Vibration 161

Fig. 3.16 Qualitative, W

schematic picture of the V2(R)
electronic energies
W2vN
(potentials) Vγ (R) and the
total energy R0(2)
Wγ vN = Wγ + WvN for some V1(R)
bound molecular states
W1υN

V0(R) R0(1)

W0υN R

R0(0)

Fig. 3.17 Blow up from one vibration

electronic state: vibrational
Wv = (v+1/2)ħω0
and rotational energies of a
rotation
harmonic oscillator with rigid v =3
rotor (for clear visibility the WN = B h c N ( N+1)
distances between rotational
levels are massively enlarged) v =2

v =1
N= 4
3
2
v =0 ħω0 /2 1
0
potential minimum

Vγ (R0 ) refers here to the minimum of the potential in the electronic state γ , while
in the most simple approximation (harmonic oscillator + rigid rotor) the energy of
the nuclear motion is given by
 
1
WvN = Wv + WN = ω0 v + + B0 hcN(N + 1). (3.50)
2
Qualitatively we obtain the picture shown in Fig. 3.16: for bound electronic states
(ground state and several excited states), γ = 0, 1, 2 . . . , the molecular potentials
(γ )
Vγ (R) are sketched with their equilibrium distances R0 . The nuclear motion oc-
curs in these potentials. The energies of vibration and rotation WvN = Wv + WN are
just added to the respective electronic energies at equilibrium distance.
On an enlarged scale Fig. 3.17 illustrates for one electronic state the vibrational
and rotational energies WvN = Wv + WN according to (3.50). Real values for vibra-
tional ω0 and rotational B0 hc = 2 /(2I0 ) energies have been reported in Table 3.4
for a selection of molecules.
162 3 Diatomic Molecules

Table 3.6 Nonlinearity parameters for our set of diatomic reference molecules: ωe xe (Anhar-
monicity), αe (change of B with vibrational state) and centrifugal term De in comparison to the
harmonic vibrational frequency ωe
Molecule ωe / cm−1 ωe xe / cm−1 Be / MHz αe / MHz De / kHzb
H+
2 2321.7 66.2 905400 50370
H2 4401.21 121.33 1824330 91800
D2 3115.50 61.82 912660 32336
LiH 1405.65 23.20 225258 6491
HCl 2990.946 52.8186 317582 9209
N2 2358.57 14.324 59906 519
O2 1580.19 11.98 43100 477
CO 2169.756 13.288 57908 524.8 184
NO 1904.2 14.075 50121 534 34
Na2 159.124 0.7254 4638.0 26.19
NaCla 364.684 1.776 6537.37 48.709 9.3506
Cl2 559.7 2.67 7319.5 45.5
If not mentioned otherwise according to H UBER and H ERZBERG (1979), L OVAS et al. (2005)
a For the isotopologue 23 Na35 Cl according to R AM et al. (1997)
b Not to be confused with the minimum potential energy De

Of course, the simple model of a harmonic oscillator with a rigid rotor is only a
beginning of a realistic description. For a given potential Vγ (R) one may (still as-
suming a rigid rotor) solve the radial S CHRÖDINGER equation (3.34) without prob-
lems numerically – using the standard recipes for stable solutions as for the bound
electron in atoms. Solutions Wv found in that manner may be expanded for N = 0
and any anharmonic potential into a series of the type
   
1 1 2
G(v) = Wv / hc = ωe v + − ω e xe v + + ··· . (3.51)
2 2
For spectroscopic reasons one typically gives the term positions in wavenumbers,
with the harmonic oscillation frequency (in wavenumbers)

ωe = ω0 / hc = ω0 /2πc = ν0 /c, (3.52)

and the so called anharmonicity constant ωe xe ωe .

Table 3.6 gives anharmonicity constants ωe xe for the characteristic molecules
known already from Table 3.4. The manner in which these parameters are written
may appear somewhat arbitrarily. The terminology is historic and in agreement with
the standard compendium on molecular physics (H ERZBERG 1989): ωe is not an
angular frequency, ωe xe is one parameter (it is rather small indeed, justifying the
series expansion (3.51)), and the parameter αe as well as De (written in calligraphic
fonts for distinction) will be explained in the next subsection. (The latter has nothing
to do with the bond energy De introduced in Sect. 3.2.5.)
3.3 Nuclear Motion: Rotation and Vibration 163

Fig. 3.18 Schemati Fc=ħ 2 N(N+1) /μRc3

illustration of centrifugal
distortion

Rc Fr =−k(Rc−R0)

For a M ORSE potential according to Sect. 3.2.5 one finds (here without proof)

2 ω02 2 hc
hcωe xe = = a2 or ωe xe = ωe2 . (3.53)
4De 2M̄ 4De

We recall again, that in any potential well the number nmax of bound states is fi-
nite. Specifically for a M ORSE potential one finds that nmax ≤ 2De /ω0 − 1. For
instance, in the case of CO the number of bound vibrational states is 41.

3.3.6 Non-Rigid Rotor

So far we have used the rigid rotor model. The radial S CHRÖDINGER equation (3.34)
was solved for fixed nuclear distance R ≡ R0 , with the centrifugal term 2 N (N +
1)/(2M̄R 2 ) considered to be constant. This assumption allows one to treat rotation
and vibration as separate, uncoupled forms of motion. Rotation is just an additive
constituent of the total energy Wγ νN . For a real molecule one has to account for the
effective potential (3.35)

2 N(N + 1)
Veff (R) = + Vγ (R)
2M̄R 2
as a function of R and to solve the full radial S CHRÖDINGER equation (3.34). This
leads to further modifications of the energy.
One characteristic effect is centrifugal distortion by rotation, which leads to a
coupling of rotation and vibration. Without explicitly solving the radial equation we
try to estimate this effect. Rotational energy (3.37) leads to a centrifugal force

dWN 2 N(N + 1)
Fc = − = .
dR M̄R 3
As sketched in Fig. 3.18 it is balanced by the restoring force of the harmonic poten-
tial,

Fr = −k(Rc − R0 ) = −M̄ω02 (Rc − R0 ),

164 3 Diatomic Molecules

with the force constant k = M̄ω02 according to (3.20). At the new equilibrium posi-
tion Rc we must have Fc (Rc ) + Fr (Rc ) = 0, hence

2 N(N + 1)
M̄ω02 (Rc − R0 ) = and
M̄Rc3
 
2 N(N + 1) 2 N (N + 1)
Rc ≈ R0 1 + > R 0 with 1.
M̄ 2 ω02 R04 M̄ 2 ω02 R04

The molecule stretches with higher rotational energy and the rotational constant
becomes smaller. The overall rotational energy of the stretched molecule is the sum
of rotational energy and potential (stretch) energy:

2 N(N + 1) M̄ω02
WN = + (Rc − R0 )2 .
2M̄Rc2 2

Inserting Rc , expanding around R0 and neglecting terms of higher order leads to

2 4
WN = N(N + 1) − N 2 (N + 1)2 + · · ·
2M̄R02 2M̄ 3 ω02 R06
 Bhc N (N + 1) − De hc N 2 (N + 1)2 . (3.54)

To obtain an estimate we insert B from (3.38):

1 4 1 4(Bhc)3 4B 3 4B 3
De = = = = .
hc 2ω02 (M̄R02 )2 hc 2 ω02 (Wv / hc)2 ωe2

Comparison for CO, NO and NaCl on the basis Table 3.6 shows rather good agree-
ment with the spectroscopically determined values of De .
A precise treatment of the S CHRÖDINGER equation (3.34) requires at least one
more term. Typically one writes:

Wγ vN / hc = Te electronic term (3.55)

   
1 1 2
+ ωe v + − ω e xe v + vibration with anharmonicity
2 2
+ Be N(N + 1) − De N 2 (N + 1)2 rotation with stretch correction
 
1
− αe v + N(N + 1) vibrational-rotational coupling.
2
(γ ) (0)
The electronic term corresponds to Te = (Vγ [R0 ] − V0 [R0 ])/ hc. The last term,
vibrational-rotational coupling, may be understood as a change of the rotational con-
stant B due to the anharmonicity of the vibration and the resulting increase of R.
3.3 Nuclear Motion: Rotation and Vibration 165

Often one extends the expansion even further:

F (N) = Bv N(N + 1) − De N 2 (N + 1)2 (3.56)

   
1 1 2
with Bv = Be − αe v + + γe v + + ··· (3.57)
2 2

describes rotation, while the vibrational terms are written as

     
1 1 2 1 3
G(v) = ωe v + − ω e xe v + + ω e ye v + + ··· . (3.58)
2 2 2

The spectroscopic accuracy today is often so high that expansions up to the 6th
or even 10th power of (v + 12 ) become meaningful (see e.g. M ANTZ et al. 1971;
L E ROY 1970).

3.3.7 D UNHAM Coefficients

Somewhat more general, today the eigenvalues of the S CHRÖDINGER equation

(3.34) are often written as a series expansion in both quantum numbers v and N :

WvN  1 i k
= Yik v + N(N + 1) . (3.59)
hc 2

This formulation has first been used by D UNHAM (1932) already in the early
years of quantum mechanics. D UNHAM coefficients Yik – not to be confused with
the spherical harmonics – have become more and more accepted during the past
decades. For comparing different spectroscopic literature it is useful to note the fol-
lowing equivalences and relations:

Y10  ωe ∝ 1/M̄ 1/2 Y11  −αe ∝ 1/M̄ 3/2

Y20  −ωe xe ∝ 1/M̄ Y21  γe ∝ 1/M̄ 2
Y30  ωe ye ∝ 1/M̄ 3/2 Y02  −De ∝ 1/M̄ 2 (3.60)
Y40  ωe ze ∝ 1/M̄ 2 Y12  −βe ∝ 1/M̄ 5/2
Y01  Be ∝ 1/M̄ Y03  −He ∝ 1/M̄ 3 .

Although these are approximations, they are sufficient for most comparisons. The
exact expressions are given in the original work of D UNHAM (1932). We only note
here that due to anharmonicity, the vibrational terms are slightly displaced in respect
of the potential minimum (M ANTZ et al. 1971) by

G(v) + Y00
with Y00 = Be /4 + αe ωe /12Be + (αe ωe )2 /144Be3 − ωe xe /4.
166 3 Diatomic Molecules

Important is also the dependence on the reduced mass M̄, which allows one to com-
pare spectra for different isotopologues.

Section summary
• We have developed an understanding of nuclear motion. In summary, energy
levels of a diatomic molecule may be expressed as a sum of electronic, vi-
(γ )
brational and rotational energy, Wγ vN / hc = Vγ [R0 ]/ hc + ωe (v + 12 ) +
B0 N (N + 1) to 1st order approximation (rigid rotor with rotational constant
B0 and harmonic oscillator with eigenfrequency ωe , both given in wavenum-
bers).
• We have discussed higher order approximations, obtaining reasonable es-
timates for the changes due to centrifugal stretch and anharmonicity in a
M ORSE potential. Even more general one expands the total energy (3.59) as
a series in powers of [v + 1/2]i [N(N + 1)]k with the so called D UNHAM
coefficients.
• We also have made a brief excursion into statistical thermodynamics in
Sect. 3.3.3, acquainting ourselves with the population of rotational energy
levels at low temperatures. Based on this information we have derived in
Sect. 3.3.4 specific heat capacities CV (T ) for diatomic molecules as a func-
tion of temperature T . At very low temperature only the kinetic degrees of
freedom can be excited and CV = 3/2R, while at temperatures T  Trot =
hcB/kB the specific heat capacity rises to 5/2R. At still higher temperatures,
around T  Tvib = ω0 /kB vibration can be excited and the specific heat ap-
proaches its limiting value CV = 7/2R.

3.4 Dipole Transitions

Which transitions can be induced by electromagnetic radiation in molecules? To

answer this question we essentially follow the treatment of atomic transitions in
Chap. 4, Vol. 1, specifically as outlined in Sect. 4.3.4, Vol. 1 for E1 transitions.
We have, however, to account now explicitly for the interaction of N electrons (i)
of charge −e and the charges +eZk of several nuclei (k) with the electromagnetic
field. Hence, the relevant electric dipole operator is now given by
N N
  N 
 nu  

D(R, r) = Zk R k − ri 
· e = ZR − r i · e, (3.61)
k=1 i=1 i=1

where again r represents all electronic, R all Nnu nuclear coordinates, and e the
unit polarization vector of the radiation. In principle this expression holds for any
number of nuclei. Specifically for a diatomic molecule with Nnu = 2, the electric
field acts on a distance weighted, average charge (see Fig. 3.9)

 = (ZA RA + ZB RB )/(RA + RB ).
Z (3.62)
3.4 Dipole Transitions 167

The transition probability into a state |a = |γ vN M from a state |b = |γ  v  N  MN 
is proportional to the squared dipole transition matrix element

Dba = γ v N MN |D|γ vN M = Ψb∗ (r, R)
     
D(R, r)Ψa (r, R)d3 Rd3 r.

By inserting (3.33) we obtain


 ∗

Dba = YN  M  (Θ, Φ)R −1 R∗γ  v  N  (R)φγ∗  (r; R)
D(R, r)
N

× φγ (r; R)R −1 Rγ vN (R)YN M (Θ, Φ) d3 Rd3 r. (3.63)
The evaluation of this matrix element is not a completely trivial task. We thus post-
pone electronic transitions to Chap. 5, and begin with rotational and then vibrational
transitions within one electronic state (γ  = γ ). These spectra in the infrared and
microwave region may be observed by absorption (possibly also by induced emis-
sion). Spontaneous emission spectra are not observable due to the ν 3 factor in the
E INSTEIN coefficients Aab ∝ Bab ν 3 .

3.4.1 Rotational Transitions

In pure rotational transitions the vibrational state v and the electronic state γ remain
unchanged. With d3 R = R 2 dR sin ΘdΘdΦ and eR = R/R being the unit vector
of the relative nuclear coordinate, one may separate the dipole transition matrix
element (3.63) for a transition N  ← N into a radial and an angular part. We make
use of the molecular symmetry around the internuclear axis – i.e. we let the electron
coordinates r i = (ξi , ηi , ζi ) refer to the molecular axis (ζi  eR ). With (3.61)–(3.63),
the radial part of the integration can be cast into
    N

 2 3  2
Dγ v = 
ZR − ζi φγ (r; R) d r i Rγ v (R) dR.
  (3.64)
i=1

The contributions for the other two components ξi and ηi disappear in this symmetry
when averaging over all r i . Obviously

Dγ v = −eDγ v is the permanent dipole moment

of the molecule in the state |γ v (see e.g. Appendix F.2 in Vol. 1, specifically
Eq. (F.10)). With this abbreviation, the angular part of the integration in the dipole
transition matrix element is given by

DN  N = Dγ v YN∗  M  (Θ, Φ)eR · eYN MN (Θ, Φ) sin ΘdΘdΦ. (3.65)
N

The important message from this expression is: pure rotational spectra can only
be observed for molecules with a permanent dipole moment, but not for homonu-
clear molecules such as H2 or N2 . Table 3.4 explicitly emphasizes the infrared active
168 3 Diatomic Molecules

heteronuclear molecules. The selection rules are derived in analogy to those for 
and m in the case of atoms, described in detail in Sect. 4.4, Vol. 1. We just have
to replace the electronic dipole transition operator D = e · r there, by 
D = Dγ v e R · e
here. The permanent dipole moment of the molecule D = −eDγ v eR is parallel to
the molecular axis eR . As in the atomic case, the three components of the molecular
unit operator eR are now expressed in terms of the renormalized spherical harmon-
ics eR = {C1−1 (Θ, Φ), C10 (Θ, Φ), C11 (Θ, Φ)}. We can now evaluate (3.65), which
leads to selection rules in full analogy those for atomic transitions between different
m states:
N = ±1 and MN = 0, ±1. (3.66)
For absorption, N  = N + 1 ← N , the transition energy in wavenumbers is
WN +1 − WN  
ν̄rot = = B (N + 1)(N + 2) − N(N + 1) = 2B(N + 1). (3.67)
hc
For induced emission, N  − 1 ← N  , one finds correspondingly

ν̄rot = (WN  − WN  −1 )/ hc = 2BN  . (3.68)

The spectral lines for a rigid rotor all have the same distance 2B in wavenumbers.
The measurement of such a spectrum is, however, not trivial since it involves a broad
spectral range from microwave to the FIR spectral range. Tabulated values usually
refer to a number of different measurements.
In Fig. 3.19 we show an artificially synthesized spectrum of rotational lines for
CO in the vibrational ground state v = 0, as derived from transition frequencies
published by L OVAS et al. (2005). It is quite instructive to consider the intensities
in these line spectra. The absorption probability RN  MN N MN for specific transition
|N  MN  ← |N MN  between the orientation states is proportional to the square of
the rotational dipole matrix element (3.65), which according to (4.79) and (C.28) in
Vol. 1 is (for a polarization vector e = eq )
 2
RN  MN N M ∝ |Dγ v |2 N  MN |C1q |NMN  (3.69)
  2  2
   N 1 N N 1 N
= |Dγ v | 2N + 1 (2N + 1) ×
2
.
0 0 0 MN q MN
The C1q (Θ, Φ) are the renormalized spherical harmonics characterizing the polar-
ization (q = 0, ±1) of the incident light, and N  = N ± 1 is the rotational quantum
number of the upper state. The 3j symbols (: : :) have been introduced in Ap-
pendix B, Vol. 1.
In view of the fact that many rotational states are populated, as discussed in
Sect. 3.3.3, we also have to include induced emission in our evaluation of the line
spectrum. The population probabilities for the specific rotational substates |N MN 
and |N  MN  are given by the B OLTZMANN factor7 exp(−N (N + 1)Trot /T ), where

7 Note that for each individual state |NM

N  the statistical weight is gN = 1, while the factor 2N + 1
used in (3.41) refers to all MN states of a rotational level N .
3.4 Dipole Transitions 169

Fig. 3.19 The rotational

100%
absorption spectrum of CO in
the FIR and mm spectral
range, synthesized from the

relative transmission
data given by L OVAS et al.
(2005). The lines have a
distance of 2B = 3.84 to
3.76 cm−1 corresponding to
the rotational constant
B0 = 1.9225 cm−1 according
to Table 3.4. The absorption
has been derived from (3.71)
0 20 40 60 80 100
ν rot / cm-1

T is the temperature of the sample and Trot the rotational temperature of the
molecule studied (see Table 3.5).
Now, the probability RN  MN N MN for the absorption process |N  MN  ← |N MN 
is exactly equal to the induced transition probability RN MN N  MN for |N  MN  →
|N MN . On the other hand, the B OLTZMANN factor for the upper rotational level
is somewhat smaller than for the lower states, so that for each individual rotational
state the relative net absorption for the incident radiation I is

I (N  MN N MN )  2
∝ |Dγ v |2 N  MN |C1q |NMN  (3.70)
I
      
× exp −N(N + 1)Trot /T − exp −N  N  + 1 Trot /T .

For the overall absorption we have to sum over all upper and lower orientation
states. With (3.69) we obtain the relative absorption signal for a single rotational
line N  ← N :
I (N  N )  I (N  M  NMN )
= N
I 
I
MN MN
      
∝ |Dγ v |2 exp −N(N + 1)βr − exp −N  N  + 1 βr
  2   2
   N 1 N N 1 N
× 2N + 1 (2N + 1) × .
0 0 0 
M N q MN
MN MN

Exploiting the orthogonality relation (B.42), Vol. 1, and with (B.53), Vol. 1 for eval-
uation of the remaining 3j symbol we finally obtain a simple expression for the
overall absorption probability:

I (N  N ) N + 1  −N (N+1)Trot /T 
∝ |Dγ v |2 e − e−(N +2)(N +1)Trot /T . (3.71)
I 3
This is presented in Fig. 3.19 for the vibrational ground state of CO.
170 3 Diatomic Molecules

Fig. 3.20 Centrifugal

/ cm-1
distortion in CO. Shown is a
quadratic fit to the 3.84
CO

/ (N +1)
spectroscopic data from v=0
L OVAS et al. (2005). The 3.83
ordinate intercept gives 2B,

(WN+1 - WN )
while the parameters De and 3.82
H in (3.72) are derived from
the slope and the slight 3.81
curvature of the line
connecting the data points 0 400 800 1200 (N+1)2

3.4.2 Centrifugal Distortion

In a more precise analysis of the data one notices, however, that with increasing
N the distance between neighbouring lines decreases slightly. This documents the
centrifugal distortion discussed in Sect. 3.3.6. For a quantitative evaluation of the
experimental material we write the rotational energy

WN / hc = BN (N + 1) − De N 2 (N + 1)2 + HN 3 (N + 1)3 .

With this, one easily verifies

WN +1 − WN
= 2B − (4De − 2H)(N + 1)2 + 6H(N + 1)4 . (3.72)
hc(N + 1)

Figure 3.20 shows this expression, plotted as a function of (N + 1)2 , for CO in the
v = 0 state. To a first approximation this gives a straight line with the ordinate inter-
cept 2B while the slope is −4De . The parameters derived from this fit correspond
to the values given in Table 3.6. The curvature of the fit even allows one to derive
the parameter H = 0.1715(4) Hz.
Today rotational lines are determined with highest precision by F OURIER trans-
formation IR spectroscopy (FTIR). An example is shown in Fig. 3.21. We shall
describe FTIR spectroscopy in Sect. 5.3.2 in detail.

Fig. 3.21 Single rotational

transition line, N = 1 ← 0, H 35Cl v = 0
N=1 ← 0
for H35 Cl in the v = 0 state,
recorded by F OURIER F= 5/2 ← 3/2
transform spectroscopy in the
FIR adopted from K LAUS et
al. (1998). The extreme
precision allows to resolve 3/2 ← 3/2 1/2 ← 3/2
even the hyperfine structure
(35 Cl has a nuclear spin 3/2) 625.90 625.92 625.94
frequency / GHz
3.4 Dipole Transitions 171

3.4.3 S TARK Effect: Polar Molecules in an Electric Field

We are now prepared to treat the S TARK effect in polar molecules, i.e. to evaluate the
energies of rotational levels in an external static electric field. In Sect. 8.2, Vol. 1 we
have discussed the S TARK effect in atoms at some lengths. We have found there that
it is a weak effect for low lying electronic terms, while for highly excited RYDBERG
states it may become substantial due to interaction with other, near neighbouring
RYDBERG states. For polar molecules we expect the largest S TARK effect for the
lowest rotational levels, since the energies (3.37) rise quadratically with the rota-
tional quantum number, and hence neighbouring levels interact less and less as N
increases.
Let the external electric field E be parallel to the molecular z-axis (E  eR ). With
Vel (R) = eE · eR Dγ v and (3.69) the interaction matrix elements (8.53), Vol. 1 are

γ vN  MN |Vel |γ vN MN  = eE Dγ v N  MN |C10 (Θ)|N MN . (3.73)

Using the same formulas (8.58) and (8.59), Vol. 1 as in the atomic case we have for
each rotational N > 0 two nonvanishing matrix elements

γ vN  MN |Vel |γ vN MN  (3.74)

⎧
⎪
⎪ (N +1)2 −MN2
⎨ for N  = N + 1
(2N +1)(2N +3)
= eE Dγ v δMN MN δN  N ±1 × 
⎪
⎪ N 2 −MN
2
⎩ for N  = N − 1
(2N −1)(2N +1)

while for N = 0 only N  = 1 is meaningful. The diagonal matrix elements disappear

and the S TARK effect becomes quadratic – just as in the atomic case with already
removed  degeneracy. In analogy to (8.63), Vol. 1 we obtain

(0)
 N  MN |C10 |N MN 2
WN MN = WN MN − WN MN = |eE Dγ v |2 . (3.75)
WN − WN 
N

Inserting (3.74), the two terms in the sum give

|eE Dγ v |2
WN MN = f (N, MN ), (3.76)
Bhc
1 N(N + 1) − 3MN2
with f (N, MN ) =
2 N(2N + 3)(2N − 1)(N + 1)
and = −1/6 for N = 0.

Figure 3.22 illustrates this expression for N = 4.

172 3 Diatomic Molecules

Fig. 3.22 S TARK splitting in |M | = 0

a polar molecule 5 1

103 f(N,M )
N=4 2
0
3
-5
-10 4

To obtain a feeling for the order of magnitude of the S TARK effect, we consider
again CO as an example. In its ground state it has a permanent dipole moment
|D| = eDγ v = 0.3662 × 10−30 C m and its rotational constant is B0 = 1.9225 cm−1 .
For an electric field strength of 1 kV / cm, which is still conveniently achievable in
the laboratory, one obtains a relative S TARK shift

WN MN |eE Dγ v |2
= 2 f (N, MN )  10−6 f (N, MN ). (3.77)
WN MN B0 h2 c2

Considering the very high precision of microwave spectroscopy this is experimen-

tally quite detectable.
In electromagnetic radiation fields the dynamic S TARK effect leads to signifi-
cant splittings already at rather moderate intensities I . In the spirit of Sect. 8.4.1,
Vol. 1 we identify E 2 in (3.77) with the average field E 2  = I /ε0 c and obtain
WN MN /WN MN  3.5 × 10−10 f (N, MN ) × I / W cm−2 for CO. In the field of a
short pulse lasers the S TARK splitting of such a molecule becomes substantial. For
the N = 4, MN = 4 state in CO, we find e.g. a lowering of the energy by about 3 %
at a very moderate intensity of I = 1010 W cm−2 .
We finally mention that the S TARK effect is also the basis for aligning molecules
in high electric fields or by laser pulses.

3.4.4 Vibrational Transitions

To evaluate the dipole transition matrix elements (3.63) for vibrational excitation
(or de-excitation) we follow Sect. 3.4.1 – still within one electronic state. We have
to replace the permanent dipole moment Dγ v = −eDγ v by the transition dipole
moment and (3.65) now becomes

Dγ v ←v = R∗γ v  (R)Dγ (R)Rγ v (R)dR
 (3.78)
  
N

 2 3
 −
with Dγ (R) = ZR ζi φγ (r; R) d r i . (3.79)
i=1

Once more, we have exploited the symmetry of the diatomic molecule and chosen
electronic coordinates with ζi  eR : the dipole moment, −eD, and its derivatives are
parallel to the molecular axis. We may thus expand Dγ (R) around the equilibrium
3.4 Dipole Transitions 173

distance R0 :

∂ Dγ 
Dγ (R) = Dγ (R0 ) + (R − R0 ) + · · · . (3.80)
∂R R0

This has to be inserted into (3.78) for a transition v  ← v. The first (constant) term
disappears due to the orthogonality of the radial wave functions Rv (R). Thus, the
linear term dominates for dipole induced (E1) vibrational transitions:
 
∂ Dγ 
Dv  ←v = R∗v  (R)(R − R0 )Rv (R)dR + · · · . (3.81)
∂R R0

The evaluation leads to the following vibrational selection rules:

• Vibrational transitions are only possible if ∂ Dγ /∂R|R0 = 0. Since in homonu-

clear molecules the dipole moment is zero, and also its derivative disappears, vi-
brational transitions within one electronic state are not possible. Diatomic gases
(H2 , N2 , O2 , etc.) are transparent in the infrared spectral region. In contrast, CO,
HCl etc. are strong IR absorbers since ∂ Dγ /∂R|R0 is large in these cases.
• For a pure harmonic oscillator v = ±1 holds.8 The spectroscopy in this case
is relatively simple, since the energy difference between two vibrational states is
always independent of v:
 
G = ωe (v + 1 + 1/2) − (v + 1/2) = ωe .

When the demand for precision is higher, one has to account for the anharmonic-
ity of the potential and possibly also has to include higher terms in the series ex-
pansion (3.80). In consequence the selection rule v = ±1 does no longer strictly
apply, transitions with v = ±2, ±3, . . . become weakly allowed, and the separa-
tion of the energy levels changes with v.
The population of vibrational states in thermodynamic equilibrium is determined
by the B OLTZMANN distribution:
   
g(v) Wv Tvib ω0
Nv = exp −  exp − with Tvib = . (3.82)
Zv kB T T kB

Here Zv is the partition function, for the harmonic oscillator given by (3.46). Nv
decreases monotonically with v, since in contrast to rotation the vibrational states
of a diatomic molecule are non-degenerate, i.e. g(v) = 1 holds. We note that at
room temperature for small diatomic molecules ω0 kB T  0.025 eV holds (cf.
Table 3.4 for some examples). Hence, at normal conditions and thermodynamic
equilibrium essentially only the ground vibrational state (v = 0) is populated.

8 One readily verifies this for the lowest levels by inserting the H ERMITE functions from Table 3.1
into (3.81).
174 3 Diatomic Molecules

P branch R branch

band origin
19
N'' 20
transmission

20
N' 19

1 0
1← 0 2← 0
0 1

×100 CO
2000 4000 v' = 1 ← v'' = 0

2050 2100 2150 2200 2250

_
ν / cm-1

Fig. 3.23 Vibration-rotation band for CO, in the electronic ground state for the v  = 1 ← v  = 0
transition, at a temperature T = 293 K, simulated with HITRAN (ROTHMAN et al. 2009) data for
the R and P branch; the inset shows also (weak) second harmonics lines, v  = 2 ← v  = 0

3.4.5 Vibration-Rotation Spectra

However, pure vibrational transition are not allowed! Simultaneously (3.65) has to
be applied (with Dγ v replaced now by Dγ v  ←v ). And as no E1 transitions occur with-
out change of the rotational quantum number N (parity conservation). In summary,
we have now three selection rules:

v = ±1 (as well as ± 2 in the anharmonic case – very weak),

N = ±1, and MN = 0, ±1. (3.83)

According to H ERZBERG (1989), the upper levels are designated with v  N  , the
lower ones with v  N  . As an introductory example, Fig. 3.23 shows the infrared
absorption spectrum of CO in the ground state (v  = 1 ← v  = 0). One recognizes a
typical band structure with many lines. The inset (not rotationally resolved) gives a
feeling for the importance of higher harmonics, showing the v  = 2 ← v  = 0 band
which amounts to less than 1 % of the total absorption. The spectrum (so called
sticks spectrum) shown has been generated synthetically from data of the HITRAN
data bank. In this impressive collection of molecular spectra more one finds nearly
three million spectral lines of presently 47 molecules with 120 isotopologues – a real
treasure for analysts and spectroscopists, who want to search for traces of molecules,
e.g. in the earth atmosphere.
3.4 Dipole Transitions 175

The band structure shown in Fig. 3.23 is caused by a combination of vibra-

tional and rotational transitions. With (3.56)–(3.58), the difference energies ν̄ (in
wavenumbers) for N  v  ← N  v  transitions are
         
ν̄ N  v  ← N  v  = F N  + G v  − F N  − G v  . (3.84)
The most important characteristics of the vibration-rotation bands may already be
recognized without any higher order terms in (v + 1/2) and N . In the following
we just account for the anharmonicity. In principle, the vibration-rotation spectra
comprise three ‘branches’ of transitions – we write them ν̄ = P (N), ν̄ = Q(N) and
ν̄ = R(N ). Of these, however, the Q branch is forbidden for pure vibration-rotation
spectra in diatomic molecules:

1. P branch with N = −1 (i.e. N  = N  − 1):

   
P N  = ωe − 2ωe xe v  + 1 − 2BN  for N  = 1, 2, 3, . . . . (3.85)
2. Q branch with N = 0 forbidden for diatomic molecules):
   
Q N  = ωe − 2ωe xe v  + 1 for N  = 0, 1, 2, 3, . . . . (3.86)
3. R branch with N = +1 (i.e. N  = N  + 1):
     
R N  = ωe − 2ωe xe v  + 1 + 2B N  + 1 for N  = 0, 1, 2, . . . . (3.87)

Note that due to anharmonicity the band origin is not exactly at ωe , as one recog-
nizes in Fig. 3.23 for the CO spectrum: ωe = 2169.756 cm−1 according to Table 3.6,
however, the origin (between P and R branch) is located at 2143.24 cm−1 , corre-
sponding to (3.86).
Figure 3.24 illustrates schematically how the band structure arises due to rotation.
In respect of a (hypothetical) pure vibration spectrum (Q branch) with Q(N) = ωe
the lines in the R branch have higher (R(N ) > ωe ), those in the P branch lower
energies (P (N ) < ωe ). The Q branch with N = 0 is missing due to the overall
parity conservation. Somewhat more precisely: this is due to the dipole moment
Dγ = −eD(R)eR being parallel to the molecular axis. Its changes are, according
to (3.78)–(3.81), responsible for vibrational excitation – one speaks about “par-
allel” transitions. They are the only ones possible in a diatomic molecule. How-
ever, already for a triatomic linear molecule this may be different as we shall see in
Sect. [Link]
The intensity distribution in the vibration-rotation absorption bands may be de-
rived in complete analogy to our considerations for pure rotational transitions in
Sect. 3.4.1. It arises again essentially from the thermal population. However, in-
duced emission usually does not play a role since initially the final vibration state is
practically unpopulated.

9 Also, in transitions between different electronic states a

Q branch may be possible – if the photon
angular momentum is transferred to the electronic charge cloud. We come back to this aspect in
Sect. 5.4.4.
176 3 Diatomic Molecules

Fig. 3.24 How the bands N' = 5

arise: P , (Q) and R branches
in a vibration-rotation 4
v' = 1
spectrum, the Q band is 3
forbidden for a diatomic 1 2
0
molecule within one
electronic state
Δ N = −1 ΔN = 1

ΔN = 0
N'' = 5
4
v'' = 0
3
1 2
Q branch 0
P branch R branch
_
ν

As an example we show in Fig. 3.25 the infrared absorption spectrum of HCl,

which has a large dipole moment is thus infrared active. The spectrum is again
based on a simulation with the help of the HITRAN data bank. It also illustrates that
isotopologues have to be considered when evaluating such spectra.
If higher accuracy is required, one has to account for contributions from higher
order terms in (v +1/2) and N to the energies F (N ) as well as to G(v). Also, the de-
pendence of Bv on the vibrational state according to (3.57) has to be included. Thus,

P branch R branch
2626
2652

3073
3059
2678

3045
2703

band origin

3030
2728

3015
2752

2866
transmission

2906

2998
2776

2844
2799

2981
2926
2822

the deeper minima HCl

2963

originate from H35Cl,

2945

the lesser minima from H37Cl

2600 2700 2800 2900 3000 3100

_
ν / cm-1

Fig. 3.25 Vibration-rotation spectrum of HCl, simulated according to HITRAN (ROTHMAN et al.
2009) for absorption at room temperature (sticks spectrum)
3.4 Dipole Transitions 177

the spectra become quite complicated, since the term distances are no longer equal.
When evaluating experimental data one uses a nice trick to separate the rotational
constants of the upper and lower vibrational states: in suitably chosen differences
of spectral lines, one or the other of these constant drops out. One finds from (3.56)
and (3.58):
 
R(N − 1) − P (N + 1) = 4B  − 6De (N + 1/2) − 8De (N + 1/2)3 (3.88)
  
R(N ) − P (N) = 4B − 6De (N + 1/2) − 8De (N + 1/2)3 . (3.89)

By plotting the data correspondingly, one obtains by suitable fits the four parameters
B  and De for the upper, and B  and De for the lower state – in the same manner as
discussed in Sect. 3.3.6 for pure rotational spectra.
By measuring several vibrational states one can obtain in a similar manner the
anharmonicity. In a so called B IRGE -S PONER plot the differences between experi-
mentally determined energies of adjacent vibrational lines (band origin) are plotted
as a function of (v  + 1). With (3.58) up to second order
     
G v  + 1 − G v  = ωe − 2ωe xe v  + 1 + · · · , (3.90)

holds – which is of course identical to (3.86). The slope of this curve gives the anhar-
monicity ωe xe , the intercept with the ordinate yields the ground state frequency ωe .
Specifically for a M ORSE potential one may estimate the bond energy with the thus
determined parameters from (3.53)

ωe2
De = hc .
4ωe xe
As for the pure rotational lines, different isotopologues have different vibration-
rotation frequencies. In addition to the change
 of the moment of inertia I0 , now the
shift of the vibrational frequency ω0 = k/M̄ is significant. As a rule, the (abso-
lute) isotope shift of the vibrational transitions is larger than that for the rotational
absorption line.

3.4.6 R YDBERG -K LEIN -R EES Method

Before turning to the ab initio (i.e. the quantum mechanical) computation of molec-
ular potentials we briefly want to introduce a standard method to determine poten-
tials directly from the experimentally measured spectra. The method developed by
RYDBERG, K LEIN and R EES already in the early years of quantum mechanics, the
so called RKR method, uses measured vibration-rotation spectra in a semiclassical
ansatz. It tries, so to say, to invert the solutions of the S CHRÖDINGER equation for
energies and nuclear wave functions. One has to know the vibrational term ener-
gies G(v) and the rotational constants B(v) as good as possible, for as many v as
possible. They are then considered as continuous functions of v and one determines
178 3 Diatomic Molecules

from these the classical turning points Rmax (v, N = 0) and Rmin (v, N = 0). From
these one may then construct the potential as illustrated in Fig. 3.7. One may show
rigorously that
 
Rmax = f 2 − f/g + f and Rmin = f 2 − f/g − f (3.91)

holds. The functions f and g are derived from

 v
1  dv 
f = (Rmax − Rmin ) = √ √ (3.92)
2 2hcM̄ vmin G(v) − G(v  )
  √ 
1 1 1 2hcM̄ v B(v  )dv 
g= − = √ . (3.93)
2 Rmax Rmin  vmin G(v) − G(v  )

Evaluation of these integrals is not completely trivial. The interested reader is re-
ferred to the literature (e.g. M ANTZ et al. 1971; F LEMING and R AO 1972, where
one also finds references to the original work).

Section summary
• E1 transitions between vibrational and rotational states within one electronic
state are studied by absorption in the infrared and microwave region, respec-
tively. Spontaneous emission is not observed at these frequencies as a conse-
quence of the ν 3 factor for spontaneous emission.
• They require a permanent molecular dipole moment (i.e. do not occur in
homonuclear molecules). Vibrational transitions depend on the change of the
dipole moment with distance.
• Pure rotation spectra in polar molecules obey selection rules for the angular
momentum, N = ±1 and its projection MN = 0, ±1. To 1st order the lines
are equally spaced by 2B, with B being the rotational constant. Centrifugal
distortion adds small terms ∝ (N + 1)2 and more.
• A quadratic S TARK effect in polar molecules is weak but observable due to
the high accuracy of microwave spectra. In intense laser field it becomes im-
portant.
• In harmonic approximation, only v = ±1 transitions are allowed. Some
weak higher harmonic lines may be observed due to anharmonicity.
• Vibration-rotation bands have, according to (3.85)–(3.87) in principle three
branches, P , Q, and R for N = −1, 0, and 1, respectively (in absorption).
The Q branch is not allowed in diatomic molecules within one electronic state.
• Relatively simple formulas (3.88)–(3.90) can be used for extracting anhar-
monicities from observed spectra. The RYDBERG -K LEIN -R EES method in-
verts in principle the S CHRÖDINGER equation and derives the molecular po-
tential from measured vibration-rotation spectra.
3.5 Molecular Orbitals 179

3.5 Molecular Orbitals

3.5.1 Variational Method

Let us now discuss the electronic part of the S CHRÖDINGER equation (3.10). It is
most instructive to explain the basic concepts used in typical quantum chemical
methods for the example of H+ 2 , the simplest of all molecules. The extension to
multi-electron systems is relatively straight forward, even though it may become
rather elaborate. With the coordinates defined in Fig. 3.26 the Hamiltonian (again in
a.u.) becomes:

el = − 1 ∇ 2r − 1 − 1 + 1 .
H (3.94)
2 rA rB R
The variational principle has been used already in Sect. 7.2.5, Vol. 1. Even in its sim-
plest form, it provides a good first approximation for the molecular potentials, i.e. for
the electronic energies as a function of R, and for the wave functions. One chooses
a suitably parameterized “trial function” φ(r i ; R) and rewrites the S CHRÖDINGER
equation Hφ = W φ as
 ∗
φd3 r
φ H
W = min  ∗ 3 . (3.95)
φ φd r

In this formulation φ(r; R) doesn’t even need to be normalized. The best eigenvalue
W is obtained by variation of φ (i.e. by changing the parameters which define φ)
such, that W becomes a minimum. Formally, the “functional” W (φ) is minimized
!
by searching for ∂W (φ)/∂φ = 0.

MOs from LCAO

As a trial wave function for the molecular orbitals (MO) one often uses a linear
combination of atomic orbitals (AOs), called MO from LCAO:


2
φ(r; R) = ci Φi (r j ). (3.96)
i j =1

We have to sum over a suitable number i of atomic orbitals Φi (r j ) as well as over

both atomic nuclei j = A, B. In the most simple case the Φi are eigenfunctions
of the H atom, however, localized now at the two different atoms and rewritten

Fig. 3.26 Coordinates for e-

the H+2 molecule
rA rB

r
A B

O R
180 3 Diatomic Molecules

such that the coordinates refer to the same origin O as indicated in Fig. 3.26. For
different masses MA and MB , the origin O is shifted towards the heavier mass. With
M = MA + MB the electronic coordinates are
r B = r − (MA /M)R and r A = r + (MA /M)R. (3.97)
(In the case of several electrons r, r A and r B represent all electron coordinates.)
With the LCAO “trial” wave function (3.96) the trial energy becomes
( ∗ ∗ (
c Φ H  c k Φk d 3 r
" =  (i i ∗ i ∗ ( k 3
> Wtrue ,
i c i Φi k c k Φk d r

where Wtrue is the true eigenvalue. Since the coefficients ci are just (complex) num-
bers, we may interchange summation and integration and obtain
( ( ∗
c ck Hik
" = (i (k i∗ with the (3.98)
i k ci ck Sik

molecular Hamiltonian in the atomic basis Hik = Φi∗ H Φk d3 r, (3.99)

and the so called overlap integral Sik = Φi∗ Φk d3 r. (3.100)

Note, that for normalized AOs, Φi , we have Sii = 1, but Sik vanishes only for or-
bitals i = k which are localized on identical atoms. We have to find now the mini-
mum of ", which is as close as possible to the true energy eigenvalue W within the
given class of functions. At this minimum,
∂" ∂"
= ∗ =0
∂ci ∂ci
must hold for all i. We rewrite (3.98) as
 
ci∗ ck Hik = " ci∗ ck Sik ,
i k i k

and differentiate in respect of ci∗ . With ∂"/∂ci∗ = 0 we obtain


(Hik − "Sik )ck = 0 (3.101)
k

after some reordering for the optimal set {ck }. This homogeneous system of linear
equations has nontrivial solutions only if
det(Hik − "Sik ) = 0. (3.102)
This is called the characteristic equation for ". The solutions "γ (i.e. the roots of
a polynomial) are the sought-after energies of the electronic system. To obtain the
MOs one inserts a specific solution "γ (for a given state γ ) into the system of linear
γ
equations (3.101). The solutions are the coefficients {ck } to be inserted into (3.96).
3.5 Molecular Orbitals 181

Fig. 3.27 Overlap (hashed)

ΦA ΦB
of 1s atomic H orbitals
localized at the two nuclei A
and B of the H+ 2 molecule,
respectively A R B

3.5.2 Specialization for H+

2

The most simple approach to the lowest MOs is to superpose just two 1s atomic
hydrogen orbitals, each centred at one of the two protons:
 1/2
1 A B
on proton A: ΦA = Φ1s (rA ) = e−rA /a0
a0 π
 1/2
1 A B
on proton B: ΦB = Φ1s (rB ) = e−rB /a0
a0 π

Since ΦA and ΦB are normalized, SAA = SBB = 1 holds and we just have to com-
pute the overlap integral

∗
S(R) = SAB (R) = Φ1s (rA )Φ1s (rB )d3 r (3.103)

indicated in Fig. 3.27. This two centre integral has limiting values S = 0 and
R→∞
S = 1.
R→0
To compute Hik we exploit the fact that ΦA and ΦB are eigenfunctions of the H
atoms. We indicate this in the Hamiltonian (3.94):
A
=H

el = − 1 2e − 1 − 1 + 1 .
H
2 rA rB R
B
=H

We define matrix elements Hik between two orbitals (where i, k = A or B):


Hik = Φ1s ∗ el Φ1s (rk )d3 r.
(ri )H

For symmetry reasons HAA = HBB and HAB = HBA . Explicitly

 
∗ A Φ1s (rA )d3 r A − Φ1s
∗ 1 1
HAA = Φ1s (rA )H (rA ) Φ1s (rA )d3 r A +
rB R
W1s Φ1s (rA )

1 ∗ 1 1
= W1s + − Φ1s (rA ) Φ1s (rA )d3 r A = W1s + − C(R). (3.104)
R rB R
182 3 Diatomic Molecules

Fig. 3.28 Energy scheme for H AA – H AB

H+2 : increase and decrease of 1– S
the 1s energies for the 1σu∗
ɕu =1σ*u
and 1σg MO, respectively W1s W1s
A B
ɕg = 1σg
H AA + H AB
1+ S

The so called C OULOMB integral C(R) is positive and −C(R) represents simply
the interaction of the electron charge distributed around nucleus A with the positive
charge of nucleus B – it disappears for large R and compensates for small R the
C OULOMB repulsion of the nuclei. The matrix element HAB = HBA is called the
resonance integral:
  
∗ el Φ1s (rB )d3 r ∗ 1 1 
HAB = Φ1s (rA )H = Φ1s (rA ) − + HB Φ1s (rB )d3 r
R rA
W1s Φ1s (rB )
 
1 ∗ 1
= W1s + S(R) − Φ1s (rA ) Φ1s (rB )d3 r
R rA

1
HAB = W1s + S(R) − K(R) = HBA . (3.105)
R

K(R) is a kind of exchange integral. For not too small R one finds HAB < 0, since
S(R) decreases with increasing R and the term −K(R) dominates.
In summary, with Sii = 1 we obtain for the characteristic equation (3.102) the
determinant
 
 HAA − " HAB − "S 
 
 HAB − "S HAA − "  = 0. (3.106)

It has two solutions for ":

HAA − HAB
"u = and (3.107)
1−S
HAA + HAB
"g = . (3.108)
1+S

With HAB < 0 this leads to the energy scheme sketched in Fig. 3.28.
Inserting these values for " into the system of Eqs. (3.101) allows one to compute
the coefficients cA and cB :

(g) (g) 1 (u) (u) 1

cA = cB = √ and cA = −cB = √ . (3.109)
2(1 + S) 2(1 − S)
3.5 Molecular Orbitals 183

Fig. 3.29 The two lowest

| g |2 | u |2
LCAO-MOs for H+ 2
constructed from Φ1s (rA ) and
Φ1s (rB ). The finite electron
g u
density between the atomic
nuclei A and B for the φg -MO
A B A B
leads to molecular bonding

H2+

We thus find two different (lowest) LCAO-MOs for H+ 2 : one which is symmetric in
respect of inversion r → −r, the so called even or “gerade” state (g) and one which
changes its sign upon inversion, the odd or “ungerade” state (u):

1  
φg = √ Φ1s (rA ) + Φ1s (rB ) (3.110)
2(1 + S(R))
1  
φu = √ Φ1s (rA ) − Φ1s (rB ) . (3.111)
2(1 − S(R))

Figure 3.29 illustrates these two MOs (more about the g − u symmetry in
Sect. 3.5.4). The electron charge density becomes

−e  ∗ ∗
 
−e|φg,u |2 = Φ1s (rA ) ± Φ1s (rB ) Φ1s (rA ) ± Φ1s (rB ) (3.112)
2(1 ± S)
−e  2  2 
= Φ1s (rA ) + Φ1s (rB )
2(1 ± S)
e  ∗ ∗

∓ Φ1s (rA )Φ1s (rB ) + Φ1s (rB )Φ1s (rA ) ,
2(1 ± S)
overlap of the atomic orbitals

where the sign ± refers to the (g) and (u) state, respectively. These two wave func-
tions (MOs) have cylinder symmetry around the molecular axis. We summarize their
key properties:

• The φg = 1σg orbital with the energy "g (R) has even inversion symmetry, i.e. the
sign of the wave function does not change upon inversion r → −r. It describes
the electronic ground state of H+ 2 and is a bonding orbital.
• The φu = 1σu∗ orbital with the energy "u (R) has odd inversion symmetry, i.e. the
sign of the wave function changes upon inversion r → −r. It is an antibonding
orbital, indicated by labelling it *.
184 3 Diatomic Molecules

The negative sign of HAB is responsible for the bonding of the symmetric orbital,
hence with (3.105) eventually the exchange integral is

∗ 1
K(R) = Φ1s (rA ) Φ1s (rB )d3 r. (3.113)
rA
Crucial for the bonding of an orbital is the overlap of Φ1s (rA ) and Φ1s (rB ), i.e. the
second term in the charge density (3.112). This is illustrated in Fig. 3.29.
For H+2 the energies "g and "u according to (3.107) and (3.108) may explicitly be
computed as a function of R. This leads to a first approximation for the potentials,
the equilibrium distance R0 , and the bond energy De . If we insert the integrals
(3.104) and (3.105) we find:

HAA ± HAB W1s + 1

− C(R) ± [(W1s + R1 )S(R) − K(R)]
"g,u (R) = = R
1±S 1 ± S(R)
1 C(R) K(R)
"u (R) = W1s + − + (3.114)
R 1 − S(R) 1 − S(R)
1 C(R) K(R)
"g (R) = W1s + − − . (3.115)
R 1 + S(R) 1 + S(R)
For the φg orbital, the finite charge density between the atoms is clearly visible in
Fig. 3.29. This leads to bonding of the H+2 molecule.
The two centre integrals S(R), C(R) and K(R) can be evaluated analytically in
elliptic coordinates. We just communicate the results (in a.u.):

1 2 −R
S(R) = 1 + R + R e , K(R) = [1 + R]e−R
3
1 
C(R) = 1 − (1 + R)e−2R .
R
As documented in Fig. 3.30 one obtains with this most simple LCAO ansatz
(dashed red line) already a reasonable guess for the potential in the bonding ground
state 1σg 2 Σg+ state, with R0  0.132 nm and De  1, 77 eV. The potential for the
repulsive, antibonding state is also plotted in Fig. 3.30. Interestingly, for the H+
2
molecule a complete analytical solution is possible (the only molecule at all for
which this can be done). The Hamiltonian Hel can be separated in confocal elliptic
coordinates. One obtains for the ground state R0 = 0.106 nm and De = 2.79 eV, the
experimental values R0  0.106 nm and De  2.65 eV are very close (full red line
in Fig. 3.30).

3.5.3 Charge Exchange in the H+

2 System

The symmetry properties of the H+

2 molecule give rise to a number of remarkable in-
terference phenomena which may directly be observed experimentally. They relate
3.5 Molecular Orbitals 185

W (R)/Eh 2Σ +
W (R )/eV
u exact

3
0.10
1σu* LCAO
2
0.05 1σg LCAO
1
0.00
1 2 3 4 5 R / a0
-1
- 0.05
R0 -2
- 0.10 De 2Σ + exact
g
-3

Fig. 3.30 Potentials for the H+ 2 molecule: dashed red lines give LCAO orbital energies in the sim-
plest form. The full red lines show for comparison the exact potentials according to S HARP (1971).
For the bonding ground state (1σu and 2 Σg+ ) LCAO gives only a rough first order approximation.
For the repulsive state, exact (2 Σu+ ) potential and approximation (1σu∗ ) are practically identic. The
energy zero has been fixed for the dissociated atoms in their ground state

to the fact that the molecular states may have even or odd symmetry. In both cases
the probability to find the electron at one or the other proton is equal. Quantum me-
chanically the two positions can, strictly speaking, not be distinguished. However,
if the nuclei are separated by a collision or a photoinduced dissociation process, at
some point during the separation process the electron has to ‘decide’ at which of the
nuclei it wants to remain for good.

Charge Exchange in the Collision Process H+ + H

The classical example are collisions between a proton and a hydrogen atom, first
investigated by L OCKWOOD and E VERHART (1962). A fast proton beam passes
through a target gas of H atoms. If the two atomic nuclei come close enough (“close
encounter”), for a short time an H+ 2 molecule is formed. At this point one can no
longer distinguish which nucleus carries the electron, the charge cloud may thus be
exchanged between the two protons. Schematically one may distinguish two pro-
cesses:
  " H + H+ elastic collision (a)
H + H+ → H+ 2 # H+ + H charge exchange (b) (3.116)

This is illustrated in Fig. 3.31 by ‘snapshots’ as seen in the centre of mass sys-
tem. After the collision one detects the charge exchange by detecting the newly
formed, fast H atoms, which have been scattered into a specific (small) angle θ . In
this particular experiment θ = 3◦ . In the experiment one first selects all particles
scattered at the angle θ by an aperture, thus discriminating scattered from unscat-
tered particles. The respective signal, S(a) + S(b) , is detected by a particle multiplier
(see Appendix B.1). Then one deflects from the scattered particle beam all protons
with the help of an electric field, and detects only the fast H atoms (signal S(b) ). In
186 3 Diatomic Molecules

Fig. 3.31 Charge exchange H e-

in H+ + H collisions
(a) H+
schematic, with the momenta pA pB
of the particles p A,B and
p A,B before and after the H2+ p'
t A
process; prior to collision (a), ( )
the electron charge is attached p'B e-
to one of the protons; during
the collision (b) temporarily H H+
an H+
(c) ( ) p'A
2 molecule is formed;
after the collision the system e- p'A
p'B e-
is found either in p'B
configuration (c) or (d) H+ H

this manner one determines that a close encounter has happened. The probability
for charge exchange during such an interactions is thus we = S(b) /(S(a) + S(b) ).
At the time of the close encounter one can, in principle, not distinguish whether
the system H+ +H moves on the potential curve belonging to φg (1σg ) or to φu (1σu ).
We have a situation very similar to a “double slit interference experiment”. The
probability amplitudes for both possibilities have to be added and this leads to typi-
cal interference structures.
Let us have a closer look at this process. Prior to the collision (t → −∞, or
R → ∞) the electron is localized at proton A, as sketched in Fig. 3.31(a). With
(3.110) and (3.111) we may express this in the molecular H+ 2 picture:

1
Φ1s (rA ) ≡ √ (φg + φu ). (3.117)
2

If we want to describe the temporal evolution of the system, we have to account for
the different time dependence of the two states involved:
   
"g (R) "u (R)
φg (R, t) = φg exp −i t and φu (R, t) = φu exp −i t .
 

The energies "g,u (R) for g and u states (i.e. the potentials shown in Fig. 3.30) split
as the two protons approach each other. Correspondingly, the initial wave function
(3.117) evolves with time:

1  
φ(t)  √ φg exp(−i"g t/) + φu exp(−i"u t/) .
2

With Wgu = "g (R) − "u (R) and W = ("g (R) + "u (R))/2 we rewrite this:
  
1 Wgu t Wgu t iW t
φ(t)  √ (φg + φu ) cos + i(φg − φu ) sin exp − . (3.118)
2 2 2 
3.5 Molecular Orbitals 187

Fig. 3.32 Electron exchange we

probability we in H + H+ 20.1 3.92 1.57 0.78 keV
h h h
collisions experimentally 1.0 ⎯⎯ ⎯⎯ ⎯⎯
observed by L OCKWOOD and 〈aWgu〉 〈aWgu〉 〈aWgu〉
E VERHART (1962); plotted is 0.8
the probability of charge
exchange as a function of the
0.6
inverse relative velocity
(kinetic energy of the proton
0.5 to 50 keV) 0.4
h
⎯⎯
〈aWgu〉
0.2
h H+ + H → H + H+
⎯⎯
〈aWgu〉 3º exchange scattering
0
0 1 2 3
1/v / 10-6 m-1 s

The phase factors sin(Wgu t/2) and cos(Wgu t/2) appear and disappear in opposi-
tion and φ(t) oscillates between

1 1
√ (φg + φu )  Φ1s (rA ) and √ (φg − φu )  Φ1s (rB ).
2 2

Correspondingly, during the close encounter, the electron ‘oscillates’ between pro-
ton A and proton B back and forth.
Strictly speaking, the ansatz (3.118) is valid only if the energy splitting Wgu
between the g and u state is constant. But for a rough guess we may replace Wgu t
by an average value

Wgu t → Wgu t  Wgu tcol = Wgu  × a/v (3.119)

where tcol = a/v is an effective interaction time, a an effective interaction length,

and v the relative velocity of the interacting particles. In more detail, we expect to
obtain a reasonable approximation with
 ∞  ∞
1
Wgu t = Wgu (R)dt = Wgu (R)dR = aWgu /v. (3.120)
−∞ v −∞

Somewhat oversimplified we have identified the relative velocity of the particles

with v = dR/dt, i.e. we assume a straight line trajectory along the internuclear axis.
After the collision the wave function (3.118) of the system may again be recast into
a superposition of the atomic orbitals Φ1s (rA ) and Φ1s (rB ):
   
aWgu  1 aWgu  1
lim φ(t) = Φ1s (rA ) cos π + iΦ1s (rB ) sin π .
t→∞ h v h v
188 3 Diatomic Molecules

The probability to find the electron after the collision at A or B is given by the square
of the respective amplitudes. The probability for electron exchange is thus
 
aWgu  1
we = sin2 π . (3.121)
h v

Correspondingly, we expect an oscillatory behaviour of the exchange probability

as a function of 1/v. This is exactly what one observes in the experiment, as doc-
umented in Fig. 3.32 by the original data. We see very pronounced maxima and
minima for the exchange probability, even though, due to finite angular resolution,
the minima and maxima do not reach 0 and 1, respectively, as predicted by (3.121).
The model predicts maxima of charge exchange for
   
πaWgu  1 1 1 h 1
= n+ π i.e. for = n+ . (3.122)
h v 2 v aWgu  2

In the experiment one reads between the maxima or between the minima a differ-
ence of (1/v)  6.6 × 10−7 m−1 s on the inverse velocity scale (with only a slight
variation). This corresponds to aWgu  = h/(1/v)  6.27 eV nm. From the poten-
tial energy diagram Fig. 3.30 one estimates an average distance between 1σg and
1σu∗ potential of about Wgu   10 eV and an effective interaction length of about
12a0 = 6.3 nm – in plausible agreement with the experiment. We note, however, that
the first maximum is observed at a phase  2.4 and not at π/2  1.57 as predicted
by (3.122) – which shows the limitations of such a simple model.

Photo-Dissociation of H+ 2 Induced by Ultrashort Laser Pulses

Another very nice example for such charge oscillations is a more recent experi-
ment by K LING et al. (2006), studying the laser induced dissociation of D+ 2 (see
also K REMER et al. 2009, for H+ 2 ). State-of-the-art ultrafast (FWHM 5–7 fs), in-
tense (1 × 1014 W cm−2 ) laser pulses (at λ  800 nm) are used with only a few
oscillation cycles to first ionize D2 molecules, and then to dissociate the D+ 2 sys-
tem. We cannot enter into the details of this sophisticated experiment. It combines
a velocity map imaging (VMI) detection system for measuring the kinetic energy
of the D+ fragment ion with phase stabilizing technique for ultra short pulses. Fig-
ure 3.33(a) and (b) show some experimental results together with a model calcula-
tion in Fig. 3.33(d) and (e).
The measured D+ ion signal is plotted in Fig. 3.33(a) as a function of the kinetic
energy. While for this overview spectrum the carrier envelope phase φc is not sta-
bilized, the asymmetry determination shown in Fig. 3.33(b) is only possible with
phase stabilized pulses. We have already mentioned in Sect. 1.4.1 the importance
of this carrier envelope phase φc for very short pulses. It is again illustrated in the
electric field profiles sketched in Fig. 3.33(c), where the electric field
 
∝ exp −(t/τG )2 cos(ωc t − φc )
3.5 Molecular Orbitals 189

measured asymmetry calculated relative

02
0.2 electron delocalization
12
kinetic energy of D+ / eV

(a) (b) -5 0 5 10 15
10 0.5
laser pulse
8
0
amplitude (d)
6
4 0
t / fs
2
0 - 0.2
signal 0 1 2 3 4 ϕc / π - 0.5 A
(c) A
B (e)
laser pulse B
shape charge asymmetry (schematic)

Fig. 3.33 Experiment and model calculation for the dissociation of D+ 2 by phase stabilized, ultra-
short laser pulses (5 fs) according to K LING et al. (2006) (for details see text)

is shown. The light is linearly polarized. In the experiment one selects preferentially
such dissociation processes for which the molecular axis is aligned parallel to the
laser field. Let us think these molecules to be align parallel to the vertical axis in the
paper plane. All three pulses have the same carrier frequency ωc and the same pulse
duration characterized by a Gaussian as discussed in Sect. 1.4.1. They differ only
in respect of the carrier envelope phase. While for φc = 0, 2π etc. the field points
at the maximum of the carrier ‘up’ in respect of the molecule, its direction points
‘down’ for φc = π, 3π etc. For φc = π/2, 3π/2 etc. the highest field strengths are
of equal magnitudes in up and down direction.
The key issue is now, to observe whether the phase has an influence on the dis-
sociation process. The results shown in Fig. 3.33(b) give clear evidence that it does!
Plotted is the asymmetry of the measured ion signal (SA − SB )/(SA + SB ) for ions
which leave the D+ 2 in the direction ‘up’ (SA ) or ‘down’ (SB ), respectively. This
asymmetry (colour code) is plotted as a function of both, the kinetic energy (verti-
cal axis) and the phase φc (horizontal axis). The influence of the phase is amazingly
clear for kinetic energies between ca. 2 eV and 8 eV, while for lower energies no
asymmetry is observed.
These processes too may be understood on the basis of the potential energy di-
agram Fig. 3.30 for the H+ 2 molecule. We cannot analyze here the quite complex
reaction processes which lead to dissociation and generate the rather substantial rel-
ative kinetic energy of the two nuclei. But obviously, the laser pulse produces a
range of wave-packets which describe the two dissociating atomic nuclei with dif-
ferent initial kinetic energies. Low kinetic energies may be attributed to processes
where only the repulsive 1σu potential of the D+ 2 is populated. One may describe this
fully within the B ORN -O PPENHEIMER approximation and speaks about an “adia-
batic” process: the two atomic components of the molecule just separate on the 1σu
potential. Since in this orbital (3.111) the electron has equal probability to be local-
ized at atom A or B, no asymmetry is observed. This is different for higher kinetic
190 3 Diatomic Molecules

energies. Under the influence of the laser field the 1σu and 1σg are strongly coupled
and the negative charge oscillates between the two states.
Very similar to the charge exchange collisions discussed above, this implies os-
cillation of the electron density between the two atoms A and B, indicated in the
(very schematic) sketch Fig. 3.33(e) as a function of time for the separating protons.
The result of corresponding model calculations is shown in Fig. 3.33(d) and verifies
these considerations quantitatively. In the example shown (φc = 0), for large times
the electron charge is found preferentially at atom B (down). The detector then de-
tects preferentially D+ ions ejected up (ion A). This depends strongly on the carrier
envelope phase φc as shown in the 2D plot Fig. 3.33(b). In the model calculation
too, for φc = π the directions A (up) and B (down) a just reversed (not shown here)
and one expects preferentially to observe the D+ ion in direction B. This is verified
in the experiment.

3.5.4 MOs for Homonuclear Molecules

In the following we treat the building-up principles for molecular orbitals for di-
atomic molecules. This concept corresponds to the filling of the n shells in the
atomic case (Sect. 3.1 in Vol. 1), and leads to rules which may be considered a
periodic system for molecules.

Symmetry and Angular Momentum

Atomic electrons move in a spherically symmetric potential. Thus the electronic
wave functions (orbitals) may be separated into a radial and angular part – as we
have done it so far very successfully:

φnm (r) = Rn (r)Ym (θ, ϕ).

In contrast, linear molecules have to be described in cylindrical symmetry. The po-

tential in the electron Hamiltonian (3.94) does explicitly depend on the polar angle
θ (via rA and rB ) and the atomic ansatz does no longer work.
Hence,  el , 
2 2
L no longer commutes with the Hamiltonian, [H L ] = 0, and the
orbital angular momentum quantum number  is no longer a good quantum number.
This does not really surprise us, as we know this situation already from the S TARK
effect (see Sects. 8.2.3 and 8.2.8 in Vol. 1). We may actually view the H+ 2 molecule
as a particular, limiting case of the S TARK effect: an H atom in the (very strong)
electric field of a proton!
Here, as in Sect. 8.2.8, Vol. 1, the projection of L onto the molecular axis (taken
el , L
as z-axis) is still a conserved quantity, i.e. [H z ] = 0. Thus, the eigenvalues of
z are still m  as in the atomic case, and m remains a good quantum number.
L
Adapted to the natural symmetry one writes the electronic wave functions for
diatomic molecules in cylindrical coordinates (ρ, z, ϕ), with ρ = r sin θ and
3.5 Molecular Orbitals 191

z = r cos θ (again in a.u.):

(±λ)
√
φel (ρ, z, ϕ) = ∓φγ λ (ρ, z) exp(±iλϕ)/ 2π (3.123)
z φ
with L
(±λ) (±λ)
(ρ, z, ϕ) = ±λφel (ρ, z, ϕ).
el

We use the standard phase convention and assume that φγ λ (ρ, z) is normalized to
unity. The new quantum number introduced here is λ = |m |, since the sign of m
has no influence on the energy (just as in the case of the S TARK effect). The angular
part of the electronic wave functions may thus be constructed as introduced by (D.6),
Vol. 1 for the real representations of the spherical harmonics.
The quantum number λ is used to characterize one electron wave functions
(MOs) for diatomic molecules. It thus replaces, so to say, the quantum number 
for atoms. Following the atomic designation s, p, d etc. for orbitals the following
notation is used:

λ 0 1 2 3 ... (3.124)
σ π δ φ
Correspondingly, to characterize the total molecular state with several electrons one
uses capital Greek letters. The quantum number for the projection of the total angu-
lar momentum on the molecular axis is called Λ (Lambda), its values are designated
as follows:
Λ 0 1 2 3 ... (3.125)
Σ Π  Φ
(
Since angular momenta( are added vectorially, Λ ≤ λi holds, just corresponding
to the atomic case L ≤ i .
Another allowed symmetry operation with homonuclear molecules is (as we have
already discussed above) inversion at the centre of mass, r → −r. In cylinder co-
ordinates this corresponds to z → −z and ϕ → ϕ + π . The states are thus distin-
guished according to their parity “g” (even) and “u” (odd):

even: φg (r) = φg (−r)

odd: φu (r) = −φu (−r).

Even though LCAO-MOs are not the very best approximation for estimating R0 and
De , they provide a good 1st order guide for the construction of molecular orbitals.
Important is in this context that the atomic orbitals involved in constructing a par-
ticular MO must all have the same symmetry in respect of the molecular axis. Only
in this case the overlap integral does not disappear Sik = 0.
For the construction of MOs from atomic s and p electrons the following possi-
bilities exist (for compact writing we use somewhat loosely |s, |px , |py , |pz  for
AOs and |σg,u  and |πg,u  etc. for MOs; antibonding orbitals are designated with an
192 3 Diatomic Molecules

y
(a) σg (ns) y x (b) x
+ + πu (np)
+ py ± p y + +
+ z – –
+ z + z
s±s +
– z y
– σ*u (ns) – x
+ +
π*g (np)
– –
–
z –
+ z
y x
(c) σg (np)
pz ± pz – y
– x
– + – σ*u (np)
+ z –
– + –
+ +
+ z

Fig. 3.34 Construction of molecular orbitals from |s and |p atomic orbitals: (a) |s ± |s,
(b) |py  ± |py  (y ⊥ plane), (c) |pz  ± |pz . Red shaded are the positive, dark-grey the negative
areas of the wave function. In each case the upper MOs are bonding, the lower ones antibonding
(denoted by *)

asterisk *):
 
|s + |s → |σg  and |s − |s → σu∗
 
|pz  + |pz  → σu∗ and |pz  − |pz  → |σg 
 
|py  + |py  → |πu  and |py  − |py  → πg∗ .

The πu and πg∗ orbitals are each twofold degenerate, as they may also be obtained
from |px  + |px  and |px  − |px , respectively. Figure 3.34 illustrates this orbital
construction scheme. Antibonding orbitals always have a nodal plane perpendicular
to the molecular axis (indicated in Fig. 3.34 by dashed lines). It is also important to
note, that the bonding σ orbitals have even symmetry, while the bonding π orbitals
have odd symmetry.
It should be noted at this point that such schemes of molecular or atomic orbitals
as sketched in Fig. 3.34 may be somewhat misleading at times. They typically rep-
resent the magnitude of the wave function (or its absolute square) at given angles θ
and ϕ in space seen from the origin, for a fixed value of the radial distance. There-
fore, they do not necessarily reflect the spatial extension of the orbitals which would
require at least a contour plot such as shown in Fig. 3.29.

Correlation Diagrams
To obtain an overview of the relative position of the energy levels one may consider
the two limits: “united atom” on the one hand, where the two nuclei are arbitrarily
close to each other with a united charge and “separated atoms” on the other hand,
with infinite distance of the two constituents and no interaction. In between, the
3.5 Molecular Orbitals 193

Fig. 3.35 Correlation of united molecular separated

atomic orbitals and molecular atom orbital atoms
orbitals between united atom
and separated atoms. Positive
regions of the wave functions p 3p σ*↔ σu* 2p p+p
are pink shaded, negative
ones grey. Antibonding MOs
are designated by an
asterisk *. Depending on d 3d π* ↔ π*g 2p p-p
whether the molecular nature
of the MOs or the LCAO
view is more relevant one
writes them in one or the
other notation p 2p π ↔ π u 2p p+p

s 2s σ ↔ σg 2p p-p

p 2p σ*↔ σu* 1s s-s

s 1s σ ↔ σg 1s s+s

Fig. 3.36 Correlation united molecule separated

diagram for the orbital atom (R = 0) atoms (R → ∞)
energies of diatomic 3 σu*
molecules. The abscissa 3p π 1π*g σ*u 2p
corresponds to the nuclear 3p
3p σ* π*g 2p 2p
distance, the ordinate reflects 3 σg πu
(very schematically) the
3s σg 2p
3s σ
potential energies. The 1 πu 2σu* σ*u 2s
2s
dashed lines indicate the true 2p π
MO energies at equilibrium 2p σg 2s
2p σ* 2 σg
distance for a number of
2s 1σu*
specific molecules. The
2s σ σ*u 1s
notation of the MOs 1s
corresponds on the left with 1σg σg 1s
the united atom, on the right 1s
1s σ Li2 F2
with the separated atoms and H2 He2
in the middle one just counts B2 N2 O2
the different symmetries and C2
+ +
λ from bottom to top R N2 O2

molecular orbitals are formed as just discussed. We summarize the two perspectives
in Fig. 3.35.
The energetic association of the MOs to united and separated atoms leads to so
called correlation diagrams, which allow a semi-quantitative discussion of the ener-
getics for the different MOs. Figure 3.36 shows schematically the potential energy
194 3 Diatomic Molecules

Fig. 3.37 Ordering of 3 * 3 *

energies for molecular
orbitals. For the lighter 1 *
g 1 *g
2p 2p 2p 2p
molecules the scheme 3 g 1
(a) holds, for O2 and heavier 1 3 g
molecules the ordering within
the 2p levels changes and (b)
is valid. Dashed levels 2 * 2 *
2s 2s 2s 2s
indicate the degeneracy of a 2 g 2 g
state. The total number of
states in the isolated atoms A
and B is identical to the total 1 * 1 *
1s 1s 1s 1s
number of MOs 1 g 1 g
A B A B
(a) ( )

as a function of the nuclear distance for the general homonuclear, diatomic case. On
the left and on the right one marks the energies of the corresponding atomic states
(ignoring the C OULOMB repulsion of the two nuclei). The connections between the
respective energies for united atom and separated atoms leads to a prediction about
the trend for potential energies of the molecular orbitals. A few rules have to be
observed:

1. Only orbitals with the same quantum number λ are connected.

2. Parity must also be conserved (g ↔ g and u ↔ u).
3. Potential curves with equal symmetry don’t cross each other.

The non-crossing rule 3 has already been derived in Sect. 8.1.6, Vol. 1.
From the correlation diagram Fig. 3.36 one may read the energies for different
MOs as a function of internuclear distance. All MOs correlated to the 1s, 2s and
2p AOs in the limit of separated atoms are shown for the general case. For infinite
distance they correspond to the three respective atomic energies, in the case of the
united atom to eight. Also indicated in Fig. 3.36, by vertical dashed lines, are the
orbital energies for some important diatomic molecules from the first row of the
periodic system of elements.
If one is interested in the details for a specific molecule one has, of course, to
draw such a correlation diagram starting with the correct numerical data for this
particular system. One finds e.g., that for light atoms up to N2 the 1πu orbital is
energetically lower than the 3σg orbital, while for O2 and larger molecules this
reverses as summarized in Fig. 3.37(a) and (b), respectively.

Filling the Orbitals

Just as in atoms, the orbitals for different electrons must differ by at least one quan-
tum number as a consequence of the PAULI principle. When building up molecules,
each of the singly degenerate orbitals σg,u may be filled with 2, each of the two fold
degenerate πg,u orbitals with 4 electrons. When filling the orbitals correspondingly,
one obtains the H+ 2 molecule, the H2 and He2 etc.; a kind of periodic system of
3.5 Molecular Orbitals 195

Fig. 3.38 Filling the 3 σu*

molecular orbitals with 1π*g
electrons for the smallest, 3 σg
homonuclear diatomic 1πu
molecules; shown is the
2σu*
occupation of the MOs in the 2 σg
electronic ground state; the
arrows indicate the spin
orientation of the electrons 1σu*
1σg
+ +
H2 H2 He2 He2 Li2 Be2 B2 C2

molecules emerges. Slightly different notations are used in the literature: e.g. the
σu∗ 2px MO is also referred to as 2px σu∗ or simply just as 3σu∗ . The highest occupied
molecular orbital in any given molecule is called HOMO, the lowest unoccupied
molecular orbital is called LUMO.
Figure 3.38 illustrates schematically how the lowest orbitals are filled for the
molecules H+ 2 up to C2 . As in the periodic system of elements, there are some
irregularities, here e.g. for B2 .
The electron configuration of some further molecules is given in Table 3.7 along
with a summary of bonding properties. The 1σg orbital is bonding and the 1σu∗
orbital is antibonding. Thus, H+ +
2 , H2 and He2 are stable molecules, since they own
more bonding than antibonding electrons in their ground state. Among these, H2
has most bonding electrons, and consequently the smallest equilibrium distance R0
and the highest bond energy De .
More general, the strength of a bond may be estimated from the so called bond
order, which is defined as the difference between the number of bonding and anti-
bonding electrons divided by 2, i.e. (n − n∗ )/2. H+ +
2 and He2 have bond order 1/2,
for H2 it is 1, and He2 with the bond order 0 should not form a stable molecule at
all – according to this simple rule. As we have already discussed in Sect. 3.2.6 one
observes indeed only a very weakly bonded VAN DER WAALS molecule.
The same scheme can be used to derive the periodic system of molecules for
other homonuclear diatomic molecules. As mentioned above, the energetic ordering
of the MOs changes between N2 and O2 (see correlation diagram Fig. 3.36 and term
scheme Fig. 3.37). As documented in Table 3.7 the bond lengths beyond Li2 are all
larger than for H+ +
2 , H2 and He2 due to the more extended valence orbitals (n = 2).
One particularly interesting point is the fact that O2 has two unpaired electrons,
a consequence of H UND’s rules which we have introduced already for atoms (see in
particular Chap. 7 in Vol. 1). It says, that among otherwise identical electron config-
urations, states with the highest total spin have the lowest energies: electrons are first
filled into all energetically degenerate orbitals before one of the orbitals is doubly
occupied. The electrons of the singly occupied orbitals have the same orientation. In
the case of oxygen they form a triplet (S = 1). Thus, O2 is paramagnetic, in contrast
e.g. to N2 .
For an approximative treatment of the potentials for multi-electron molecules one
usually may confine the efforts to a few outer electrons, the valence electrons, i.e.
to the highest occupied orbitals. The other electrons are typically strongly localized
196 3 Diatomic Molecules

Table 3.7 Filling the MO shells in the most simple homonuclear, diatomic molecules (electron
configuration in the ground state). The bond order is (n − n∗ )/2, R0 is the bond distance. Disso-
ciation energies De and term energies Te refer to the potential minimum; usually Te = 0 for the
neutral ground state, see (3.55)
n−n∗
Molecule Electron config. 2 State De / eV R0 / nm Te / cm−1
H+
2 (σg 1s)1 1/2 2Σ +
g 2.65 0.1052 125 443
H2 (σg 1s)2 1 1Σ + 4.48 0.074 0
g
He+
2 (σg 1s)2 (σu∗ 1s)1 1/2 2Σ +
u 2.47 1.08 178 400
He2 (σg 1s)2 (σu∗ 1s)2 0 1Σ +
g 0.00095a 0.297a 0
Li2 He2 (σg 2s)2 1 1Σ + 1.07 0.267 0
g
Be2 He2 (σg 2s)2 (σu∗ 2s)2 0 1Σ +
g not observed
B2 He2 (σg 2s)2 (σu∗ 2s) . . . (πu 2px )2 (σg 2pz ) 2 3Σ −
g 3.0 0.159 0
C2 Be2 (πu 2px )2 (πu 2py )2 2 1Σ + 6.32 0.1243 0
g
N+
2 Be2 (πu 2px )2 (πu 2py )2 . . . (σg 2pz ) 2 1/2 2Σ +
g 8.85 0.112 125 744
N2 Be2 (πu 2px )2 (πu 2py )2 . . . (σg 2pz )2 3 1Σ + 9.90 0.1098 0
g
O2 N2 (πg∗ 2px )(πg∗ 2py ) 2 3Σ −
g 5.21 0.121 0
F2 N2 (πg∗ 2px )2 (πg∗ 2py )2 1 1Σ +
g 1.66 0.141 0
Ne2 N2 (πg∗ 2px )2 (πg∗ 2py )2 . . . (σu∗ 2pz )2 0 not observed
a See Sect. 3.2.6

to the atomic nuclei and contribute only little to the molecular bonding. Of course,
here as with atomic structure calculations the general rule holds, that in a rigorous
computation the results become the better the more MOs (which are or possibly
might be occupied) are accounted for.

Section summary
• We have learned how molecular orbitals can be constructed as linear super-
positions of atomic orbitals (MO from LCAO). Using the variational method
to minimize the energy, one has to determine the Hamiltonian matrix (3.99)
and the overlap integrals (3.100) in the atomic basis. This leads to the char-
acteristic equation (3.102) from which the eigenenergies of the orbitals are
derived.
• We have detailed this for the simplest molecule, H+ 2 . By including only the
two 1s AOs localized on either of the protons, the two energetically low-
est MOs are found; they have even (1sσg ) and odd (1sσu∗ ) inversion symme-
try (r → −r). The 1sσg is bonding, the 1sσu∗ antibonding (repulsive poten-
tial). A surprisingly good first guess for the molecular potential is obtained as
shown in Fig. 3.30.
• These two molecular potentials give rise to interesting interference phenom-
ena in charge exchange collisions and in ultrafast photo-dissociation.
• The systematic construction of MOs for a range of homonuclear, diatomic
molecules starts by writing the wave function (3.123) in cylinder coordinates,
the key quantum number λ being the projection of the angular momentum
3.6 Construction of Total Angular Momentum States 197

onto the internuclear axis. We have constructed a few characteristic MOs from
s and p atomic orbitals (Fig. 3.34).
• MOs must have a finite electron density in between the atoms to be bonding.
• Correlation diagrams for MOs give an overview of the energetics; they con-
nect the energies of the united atom with those for separated atoms (Fig. 3.36).
Based on this one may derive some kind of periodic system for homonu-
clear, diatomic molecules (Table 3.7), which allows to make predictions about
the strength, symmetries and spin properties of the molecular bonding. As a
prominent example, the electronic ground state of molecular oxygen is found
to be a triplet, hence O2 is paramagnetic.

3.6 Construction of Total Angular Momentum States

3.6.1 Total Orbital Angular Momentum

For the total orbital angular momentum too, only the z-component L z is a conserved
quantity. It is obtained as the sum of the z-components of the individual MOs:
 
z =
L zi with eigenvalues ML =
L mi . (3.126)
i i

The designation Λ = |ML | with the term notation Σ, Π ,  etc. for Λ = 0, 1, 2 has
already been introduced above. We note here in addition, that molecular orbitals
have a well defined reflection symmetry in respect of planes through the molecular
axis. This reflection symmetry must remain conserved when composing the total
wave function. When combining several MOs to an overall molecular state positive
and negative mi may contribute in different combinations, so that the states are
somewhat more complex than suggested by (3.126).

3.6.2 Spin

As in the atomic case the total spin is the sum of the individual spins of the electrons
 

S= 
S i with | S| = S(S + 1)
i

and results in a multiplicity 2S + 1 of the overall electronic states. The projection

of 
S onto the molecular axis is called Σ = Ms with positive and negative values in
contrast to Λ. For each S there are 2S + 1 values

Σ = S, S − 1, S − 2, . . . , −S.

Caution: do not mistake this spin quantum number Σ for a Σ state (total angular
moment projection Λ = 0)!
198 3 Diatomic Molecules

In analogy to the term designation for atoms, molecular states are denoted
2S+1
Λg,u . (3.127)

We recall briefly and symbolically the construction of spin states which has been
discussed extensively in Vol. 1 (here for one and two electron systems):

Doublet Singlet Triplet

One electron Two electrons
S= 1
2, MS = ± 12 S = 0, MS = 0 S = 1, MS = 1, 0, −1
1/2
|↑ = |χ1/2  |↑↑ = |χ11 
(3.128)
|↑↓−|↓↑ |↑↓+|↓↑
√ = |χ00  √ = |χ10 
2 2
−1/2
|↓ = |χ1/2  |↓↓ = |χ1−1 

Of course, as in the atomic case, the PAULI principle has to be observed, i.e. the
total wave function must be antisymmetric. For a system with two active electrons –
e.g. for the H2 molecule – this implies – as for the He atom: for singlet states (two
antiparallel spins with one antisymmetric spin function, 2nd column in Eq. (3.128))
the spatial electron wave function must be symmetric; for triplet states (two parallel
spins, three symmetric spin functions, 3rd column in Eq. (3.128)) the spatial wave
function must be antisymmetric.

3.6.3 Total Angular Momentum

The projection of the total electronic angular momentum J e of a molecular state

onto the molecular axis (z-axis) is usually designated as Ω. Figure 3.39 indi-
cates schematically how Ω is constructed as sum of orbital angular momentum  L
(Λ when projected onto the z-axis) and total spin 
S (Σ when projected onto the
z-axis):
Ω = |Λ + Σ|. (3.129)
Thus, instead of the indices g or u according to (3.127), for more complex states one
often finds Ω as an index.
An additional complication comes with the rotation of the molecule: all relevant
angular momenta have to be combined to a total angular momentum  J of the whole
system, applying the general rules for angular momentum coupling (see Appendix B
in Vol. 1). Depending on the coupling between orbital angular momentum, spin and
nuclear rotation several different possibilities exist – quite similar to the atomic case
where we had to couple the orbital angular momentum, electron spin and nuclear
spin. For molecules all these complications exist too, and rotation makes them even
a little bit more complex. In Table 3.8 we summarize the angular momenta and
3.6 Construction of Total Angular Momentum States 199

3∆ Ω
3 Λ Σ =1
3∆ Ω
2 Λ Σ=0
Ω Σ = −1
3∆
1 Λ

Fig. 3.39 Electronic angular momentum coupling in diatomic molecules for the example of Λ = 2
( state) with S = 1 (triplet): only the components of the angular momenta in the direction of the
molecular axis are good quantum numbers

Table 3.8 Angular momenta and quantum numbers of diatomic molecules

Kind of angular momentum Operator Quantum number
Total projection onto z-axis
Orbital angular momentum 
L L Λ
Electron spin 
S S Σ
Electron total angular momentum  L +
Je =  S Je Ω
Nuclear rotation 
N N 0
Total angular momentum  L +
J = 
S +N J Ω
Total without spin  
K =L+N K Λ

quantum numbers used in the following, essentially according to the notation of

H ERZBERG (1989).10

3.6.4 H UND’s Coupling Cases

According to H ERZBERG (1989) one distinguishes several, so called H UND’s cou-

pling cases, which are of key importance for the interpretation of rotational bands
in electronic spectra. Here we can give only a brief first introduction and refer the
interested reader to specialized literature, e.g. to H OUGEN (2001).

H UND’s Case (a) As illustrated in Fig. 3.40(a) one assumes the coupling between
molecular rotation and electronic orbital angular momentum  L or spin 
S to be very
weak. In contrast, the coupling of the orbital angular momentum to the internuclear
axis is strong and the spin couples to the thus generated internal magnetic field
parallel to the molecular axis. The situation corresponds to the electric analogue to
the PASCHEN -BACK effect. As for the not rotating molecule we have Ω = |Λ + Σ|,
and Ω together with the nuclear rotation N  forms the total angular momentum  J.
Values of J < Ω are not possible and we have:

J = Ω, Ω + 1, Ω + 2, . . . . (3.130)

10 The finite number of letters in the alphabet limits the choices, unfortunately. H OUGEN (2001)
e.g. use the letters R instead of N , and N instead of K.
200 3 Diatomic Molecules

(a) (b) J S (c) J N (d)

J N L K

K N N

Ω Ω
Λ
Λ Σ Je
L S L L S

Fig. 3.40 H UND’s cases (a), (b), (c) and (d) for coupling angular momenta in diatomic molecules

Because of (3.129) in principle J may also be half integer (if the spin is half integer).
In this coupling case the total energy (3.49) for a 2S+1 ΛΩ state is found to be (here
without proof):
 
Wγ vN / hc = Te + G(v) + F (J ) with F (J ) = Bv J (J + 1) − Ω 2 . (3.131)

The electronic term energy Te may in addition be split by fine structure (cf.
Eq. (6.36), Vol. 1):
Te = T0 + aΛΣ. (3.132)
Apart from the lowest state and the missing values J < Ω there is no difference
between (3.131) and the rigid rotor (3.37), if one replaces N by J . For each fine
structure level of an electronic state a specific ladder of rotational states exists.

H UND’s Case (b) If the orbital angular momentum is zero, or if the electron spin
is coupled only very weakly to the orbit, there is no coupling of the spin to the
nuclear axis. As illustrated in Fig. 3.40(b) we have in that case practically just a
rigid rotor, whose angular momentum N  couples with the projection of the orbital
angular momentum   The quantum number K replaces now the rotational
Lz to K.
quantum number. For Σ states it is even identical with N . Only at the very end
one has to account for fine structure coupling. With the usual rules the total angular
momentum becomes:

J = K + S, K + S − 1, . . . , |K − S|. (3.133)

The energy of the rotational states is determined essentially by the nuclear rotation,
possibly showing a small (2S + 1) fold FS splitting. For the particularly simple case
of a 2 Σ state the rotational energy becomes
γ 1
F (K) = B(v)K(K + 1) + K for J = K +
2 2
γ 1
= B(v)K(K + 1) − (K + 1) for J = K − , (3.134)
2 2
where the small constant γ is due to FS interaction.
3.6 Construction of Total Angular Momentum States 201

H UND’s Case (c) If the coupling of the orbital angular momentum to the internu-
clear axis is weak, the spin-orbit coupling of the electron cloud is not broken. As
shown in Fig. 3.40(c) an electronic total angular momentum  J e is formed, which as
a whole couples to the internuclear axis. The projection onto z is again called Ω.
This component of the electronic angular momentum finally couples again with the
 to the overall total angular momentum 
rotation N J . In its final result this case
is hardly different from H UND’s case (a) and the energy terms and possible angu-
lar momentum quantum numbers J are indeed described by (3.131) and (3.130),
respectively.

H UND’s Case (d) Figure 3.40(d) shows the rather simple vector diagram for the
case that the coupling of the orbital angular momentum  L with the internuclear axis
is very weak (e.g. for highly excited electrons). On the other hand one assumes that
the coupling of the molecular rotation N to 
L is strong. One then observes primarily
the energy spectrum of the rigid rotor,

F (N) = B(v)N (N + 1),

 with 
where, however, the coupling of N L leads to a (2L + 1) fold splitting of each
rotational level.

H UND’s Case (e) Finally, it is also possible that  L and S couple strongly with each
other and form an electronic total angular momentum  J e . However, the interaction
with the internuclear axis is weak. This leads to a situation very similar to case (d).
Except that the spitting of the rotational levels is now (2Je + 1) fold. In H ERZBERG
(1989) one still reads about this case that – albeit thinkable – it has never been
observed. This has changed by now, e.g. it has clearly been detected in the case of
moderately high lying RYDBERG states of diatomic molecules such as O2 . Also,
when discussing low energy collision processes between atoms one has to account
for such a possibility, e.g. if collision induced fine structure transitions are observed.
In the spectroscopic practice there are, of course, all kinds of transitions between
the five pure cases introduced here. Higher precision requires, finally, also the in-
clusion of hyperfine interaction – if the atomic constituents have a nuclear spin. In
Fig. 3.21 we have already seen an example of that. The analysis of such interactions
follows essentially that for atoms (see Chap. 9 in Vol. 1). H UND’s cases have to be
extended correspondingly.

3.6.5 Reflection Symmetry

For a range of questions it is important to know the reflection symmetry of the elec-
tronic states with respect of a plane through the molecular axis. For single orbitals
this is rather clear and one can extend the considerations presented in Appendix D,
Vol. 1 – replacing the quantum numbers q used in Appendix D.4, Vol. 1, or M
202 3 Diatomic Molecules

in (D.14), Vol. 1 by λ. For multi-electron systems, the construction of states with

well defined symmetries becomes somewhat more involved and we restrict the dis-
cussion here to a few important examples.
We denote the reflection symmetry operator in respect of an ab plane as σ̂v (ab).
In cylindrical coordinates these reflections correspond to coordinate replacements:

σ̂v (xz) : ϕ → −ϕ (3.135a)

σ̂v (yz) : ϕ →π −ϕ (3.135b)
σ̂v (xy) : z → −z. (3.135c)

Inversion (which determines g or u symmetry) is achieved by

ı̂ : z → −z and ϕ → ϕ + π. (3.136)

For the simplest MOs these symmetries can be gleaned from Fig. 3.34.
It is instructive, however, to elucidate this on the basis of wave functions for the
orbitals. Typically, one uses real combinations of (3.123) as introduced with (D.6)
in Vol. 1. For σ orbitals (λ = 0) we write (somewhat laxly in kets)

φσ (ρ, z)
|σ  = φel (ρ, z, ϕ) = √ , (3.137)
2π

while there are two types of π orbitals (λ = 1):

φπ (ρ, z)   cos ϕ
|πx  = √ exp(+iϕ) + exp(−iϕ) = φπ (ρ, z) √ (3.138a)
2 π π
φπ (ρ, z)   sin ϕ
|πy  = √ exp(+iϕ) − exp(−iϕ) = φπ (ρ, z) √ . (3.138b)
2 πi π

The angular part is normalized to unity in all cases. By applying (3.135a) and
(3.135b), we see that σ orbitals have always positive reflection symmetry (no de-
pendence on ϕ):

σ̂v (xz)|σ  = +|σ  and σ̂v (yz)|σ  = +|σ .

In contrast, the πx,y orbitals have opposite reflection symmetries:

σ̂v (xz)|πx  = +|πx  and σ̂v (yz)|πx  = −|πx 

σ̂v (xz)|πy  = −|πy  and σ̂v (yz)|πy  = +|πy .

For multi-electron systems, we have to account for the indistinguishability of the

electrons and the PAULI principle. We illustrate this now for a few examples.
3.6 Construction of Total Angular Momentum States 203

Example H2 Molecule
In the most simple case we only have to fill one orbital with two electrons, as e.g.
the 1sσ orbital for the ground state of the H2 molecule. With the notations (3.127)
and those given in Fig. 3.35 (close to the united atom), the total electronic wave
function is written:
1 +   0  
 Σ = χ φ1sσ (ρ1 , z1 ) · φ1sσ (ρ2 , z2 ) . (3.139)
g 0

Since the spatial part is obviously symmetric in respect of exchange of the electrons,
the spin function |χSMS  must be antisymmetric, i.e. be a singlet. As for both elec-
trons λ = 0 holds, also the total orbital angular momentum remains Λ = 0, i.e. we
describe a Σ state. Obviously this function is also symmetric in respect of inver-
sion, and finally, positive reflection symmetry for both 1sσ orbitals leads to overall
positive reflection symmetry. This is indicated by the superscript + sign. The bound
ground state of H2 is a singlet X 1 Σg+ state (the letter X is just used for numbering
the states, and one start with X for the ground state, the letters A, B, . . . follow for
excited states).
Next we keep one electron in the 1sσ orbital and bring the other one in the next
higher 2pσ ∗ orbital (σu∗ 1s in MO notation). The configuration |1sσ 2pσ ∗  may be
combined to two different spatial functions:
  √
φ1sσ (ρ1 , z1 )φ2pσ (ρ2 , z2 ) ∓ φ1sσ (ρ2 , z2 )φ2pσ (ρ1 , z1 ) / 2. (3.140)

Both wave functions have again Σ + reflection symmetry (they inherit this property
from the constituents), but both change their sign upon inversion (due to the involve-
ment of a 2pσ ∗ orbital). We also see that one of the functions is antisymmetric, the
other symmetric in respect of exchange of the two electrons. The spin function thus
has to be triplet or singlet, respectively. Closer inspection shows that the resulting
|3 Σu+  state is repulsive (one bonding and one antibonding orbital). The other state
does not lead at all to a low lying molecular state.
Next we have a look at the configuration |1sσ 2pπ (two bonding orbitals with
λ1 = 0 and λ2 = 1). We may construct from this e.g. an odd (u), bonding state,
symmetric in respect of electron exchange. With Λ = λ1 + λ2 = 1 we obtain from
(3.138a):
1   0   √
 Πu = χ |1sσ 2pπx  + |2pπx 1sσ  / 2 (3.141)
0
 0   iϕ 
= χ0 φ1σg (z1 ρ1 )φ1πu (z2 ρ2 ) e 2 + e−iϕ2
   √
+ φ1σg (z2 ρ2 )φ1πu (z1 ρ1 ) eiϕ1 + e−iϕ1 /(2 π ).

With (3.136) we verify that we have indeed constructed a (u) state: for both electrons
φ(zρ) = φ(−zρ) and replacing both azimuthal angles ϕ → ϕ + π brings an overall
minus sign.
Reflection symmetry in this case is positive in respect of the xz plane (πx like –
just replace both ϕ → −ϕ). In contrast, rotation of the molecule around the z-axis
by π/2 generates a state whose reflection symmetry is πy like, i.e. negative. The
electronic problem is, however, completely symmetric in respect of the z-axis: both
204 3 Diatomic Molecules

states |1 Πu+  and |1 Πu−  are degenerate. Thus, one usually omits the symmetry des-
ignation ± for molecules with Λ = 0. We shall, however, discuss the limitations of
this view in Sect. 3.6.6.

Example O2 Molecule
What is, however, the situation for Σ states with Λ = 0? They may e.g. be con-
structed from two π orbitals with opposite orientation. According to Table 3.7 the
ground state of O2 is an example. The system is well bonded by the electrons in the
completely filled N2 of the inner electrons. In addition, two πg∗ orbitals have to be
filled (configuration (πg∗ 2p)2 with λ1 = λ2 = 1). We first note that the overall spa-
tial wave function is in any case even (g) in respect of inversion. We start again with
the complex orbitals (3.123) and use now the MO notation |φel  = |πg∗ 2p±1 .
(±λ)

According to H UND’s rules we expect the lowers state to be a triplet. If we ignore

M
spin orbit interaction the three symmetric spin wave functions |χ1 S  lead to three
degenerate states. The spatial wave function must be antisymmetric in respect of
electron exchange, hence:
3 −   MS  ∗    √
 Σ = χ π 2p−1 π ∗ 2p1 − π ∗ 2p1 π ∗ 2p−1 / 2
g 1 g g g g
 #
  1 1  i(ϕ2 −ϕ1 ) 
= χ1MS φ1πg∗ (z1 ρ1 )φ1πg∗ (z2 ρ2 ) √ e − e−i(ϕ2 −ϕ1 )
π 2 2i
 MS   √
= χ 1
φ1π ∗ (z1 ρ1 )φ1π ∗ (z2 ρ2 ) sin(ϕ2 − ϕ1 ) /(π 2).
g g
(3.142)

The terms e±i(ϕ2 −ϕ1 ) make sure that the angular momenta compensate each other,
so that
 
z1 + L
(L z2 )3 Σg− ≡ 0

and Λ = λ1 − λ2 = 0. For inversion symmetry we note that again for both electrons
φ(zρ) = φ(−zρ), but when replacing both azimuthal angles ϕ → ϕ + π the π is
compensated and the sign does not change. We have indeed constructed a Σg state.
Reflection in respect of the xz plane (i.e. ϕ1 → −ϕ1 and ϕ2 → −ϕ2 ) changes the
sign, hence, the overall reflections symmetry of the state is negative. It is important
to note that in this case rotation around the z-axis by an arbitrary angle δ (i.e. ϕ1 →
ϕ1 − δ and ϕ2 → ϕ2 − δ) does not change anything about the eigenfunction! This
is obviously due to the fact that the two orbital angular momenta just compensate
each other. Thus, for Σ states reflection symmetry is a good quantum number which
characterizes the state. This statement also holds for the singlet state complimentary
to (3.142):
1 +   0   √
 Σ = χ φ1π ∗ (z1 ρ1 )φ1π ∗ (z2 ρ2 ) cos(ϕ2 − ϕ1 ) /(2 π). (3.143)
g 0 g g

It has positive (g) inversion symmetry and positive reflection symmetry, again in-
dependent of the alignment of the reflection plane through the molecular axis. The
exchange symmetry for the two position coordinates (1 and 2) is now positive (cos
function). Hence, the spin function must be antisymmetric and we have a singlet
3.6 Construction of Total Angular Momentum States 205

state – as it turns out its electronic energy is significantly higher than that of the
|3 Σg−  state.
In contrast, the Π ,  . . . etc. states are degenerate in respect of the ± symmetry.
Still with the same electron configuration (πg∗ 2p)2 we construct the remaining two
singlet states, with spatial wave functions which are again symmetric in respect of
electron exchange:
1   0  ∗    √
 g = χ π 2p−1 π ∗ 2p1 + π ∗ 2p1 π ∗ 2p−1 / 2
0 g g g g
 #
 0  1  i(ϕ2 +ϕ1 ) 
 
= χ0 φ1πg (z1 ρ1 )φ1πg (z2 ρ2 ) √
∗ ∗ e +e −i(ϕ2 +ϕ1 )
2 πi
   √
= χ00 φ1πg∗ (z1 ρ1 )φ1πg∗ (z2 ρ2 ) sin(ϕ2 + ϕ1 ) /(2 π). (3.144)

This is a 1 g state with Λ = λ1 + λ2 = 2 due to the identical signs of ϕ2 and ϕ1 . It

has negative reflection symmetry in respect of the xz plane. However, in this case
there exists another, orthogonal but energetically degenerate state 1 g when the
previous one is rotated by π/4 around the z-axis (i.e. ϕ1 → ϕ1 +π/4 and ϕ2 → ϕ2 +
π/4), for which |1 g  ∝ cos(ϕ2 +ϕ1 ) holds. Its reflection symmetry is now positive,
while the electron exchange symmetry as well as inversion symmetry remains also
positive.
In total, we have identified 6 states based on the configuration (πg∗ 2p)2 . They il-
lustrate very clearly why only for the Σ states reflection in respect of a plane through
the molecular axis is independent of the alignment of this plane. We shall come back
to the electronic states of O2 in Sect. 3.6.9. We shall see there that we have found
indeed the lower lying states X 3 Σg− , a 1 g and b 1 Σg+ – in this energetic ordering,
and again in agreement with H UND’s rules (highest multiplicity has lowest energy,
for equal multiplicity highest total angular momentum is lowest, see Sect. 7.3.3 in
Vol. 1).

3.6.6 Lambda-Type Doubling

At the end of this discussion we allude to a spectroscopically important phe-

nomenon, the so called lambda-type doubling. In principle, it is relevant for H UND’s
cases (a) and (b) whenever Λ = 0 (i.e. for Π ,  etc. states) and the rotational
quantum number is large. So far we have neglected the coupling between rota-
tion and electronic orbital angular momentum. Such coupling arises within diatomic
molecules for states with Λ = 0 which – without rotation – are doubly degenerate
and may have positive or negative reflection symmetry as just discussed.
Let us consider the most simple example of a 2 Π state which arises from a single,
filled π orbital in the valence shell. The energy of 2 Π + and 2 Π − state is completely
identical as long as the molecule does not rotate. The x- and y-axes may be defined
completely arbitrary (perpendicularly to the molecular axis z) and the πpy and πpx
orbitals are fully equivalent. However, if the molecule rotates, this symmetry is re-
moved. Assume it rotates around its y-axis. We may then quite intuitively imagine
206 3 Diatomic Molecules

that the py orbital (dumbbell, rotationally symmetric around the y-axis) is hardly
influenced by this rotation. In contrast, the electron in a px orbital (dumbbell, per-
pendicular to the y- and z-axis) is exposed quite evidently to a centrifugal force.
This leads to a splitting which is called Λ-type doubling. Even if this is an ener-
getically very small contribution, it is well observable with modern spectroscopic
precision. The effect may become quite significant in dissociation processes where
e.g. diatomic molecules are ejected from a larger complex. The rotational distribu-
tion of these fragments may then have a pronounced asymmetry in respect of the
two molecular rotation axes.

3.6.7 Example H2 – MO Ansatz

We come back once more to the computation of electronic wave functions of di-
atomic molecules and focus in some detail on the H2 molecule with its two equiv-
alent electrons. We shall use what we have worked out in Sect. 3.5.2 for H+ 2 , and
summarize in passing the findings of Sect. 3.6.5. The overall wave function must
antisymmetric in respect to electron exchange (PAULI principle). For the two active
electrons the total spin is S =S1 + 
S 2 (with Σ = 0 or 1 projected onto the molecu-
lar axis), the total orbital angular momentum is L= L1 +  L2 (with Λ = |λ1 ± λ2 |).
The spatial wave functions may be constructed from even or odd orbitals according
to (3.110). In analogy to the He atom, the total wave functions of H2 are
singlets with MS = 0 : φS (1, 2) = φ+ (1, 2)χ00 (1, 2), and (3.145)

triplets with MS = −1, 0, 1 : φT (1, 2) = φ− (1, 2)χ1MS (1, 2) (3.146)

with symmetric φ+ (1, 2) and antisymmetric spatial function φ− (1, 2).
For the lowest states we include in our evaluation only the 1σg and 1σu MOs. For
brevity we write the respective spatial wave functions φg (i) and φu (i), with i = 1 or
2 depending on which electron is placed into this orbital. The ground state (3.139)
with two electrons in the bonding 1σg orbital is written in short notion:

X 1 Σg+ : φX (1, 2) = φg (1)φg (2)χ00 . (3.147)

If one of the electrons is in the antibonding 1σu MO one may obtains as indicated
in (3.140)
1   M
b3 Σu+ : φb (1, 2) = √ φg (1)φu (2) − φg (2)φu (1) χ1 S , (3.148)
2
for which a detailed calculation shows that it is indeed the lowest lying, repulsive
triplet state and correlates in the separated atom limit with both atoms in the 1s
ground state. In contrast, the second combination with one electron in a 1σu MO,
1  
1
Σu+ : φ3 (1, 2) = √ φg (1)φu (2) + φg (2)φu (1) χ00
2
will even be higher in energy according to H UND’s rules.
3.6 Construction of Total Angular Momentum States 207

This holds a fortiori for

1
Σg+ : φ4 (1, 2) = φu (1)φu (2)χ00

where both electrons are in antibonding orbitals. In the separated atom limit, both
combinations do not correlate with the atoms in their 1s ground state and only con-
tribute to higher lying bound and unbound states.
In a similar manner the higher lying molecular states are constructed from more
complex MOs. The principles are those treated in Sect. 3.6.5. An overview of the
H2 potentials computed with quantum chemical methods gives Fig. 3.41. It also in-
cludes its ions H− +
2 and H2 . As already mentioned, the states are usually ‘labelled’
with capital letters, the ground state with X, the next higher with A, B, C etc. In
a number of cases, due to the historical development, this is not followed conse-
quently. Often one uses the lower case alphabet for states which have been found
later. As H2 is of particular importance (and its potentials form a rather complex
manifold) a section of the energy diagram is shown once again in Fig. 3.42 on an
enlarged scale.
Note that no stable negative ion (anion) H− −
2 exists. The lowest energy of H2 lies
−
above the bound, neutral ground state of H2 . Thus, H2 can decay spontaneously into
H2 + e− – if it is at all ever formed in the first place. One observes such states as
resonances in the scattering of low energetic electrons by H2 . The observed feature
looks very similar to autoionization which we have treated in Chap. 7, Vol. 1. We
shall come back to this in Chap. 8.
One more, closely related, process should be mentioned at this point, the so called
predissociation. We note the quite remarkable potential maxima for a number of H2
states, as seen in Fig. 3.42 e.g. for the states designated with I , i and h. They are
a consequence of avoided crossings – which generate local maxima in the poten-
tials. These maxima make it possible to form vibrational states which lie above the
dissociation limit. Although these states are strictly speaking not stable states, they
still live for some time since the potential barrier prevents immediate decay. The
probability for dissociation of such molecular states by “tunnelling” depends on the
height and width of the barrier.
A few words on how to calculate such potentials are in order at this point. For the
H2 molecule one may in principle follow the same procedure as in the H+ 2 case, and
use suitably chosen trial functions as approximate solutions of the S CHRÖDINGER
equation. The electronic Hamiltonian is now (in a.u.)
 
   1 1
H = H1 + H2 + + (3.149)
r12 R

i = − 1 ∇ 2i − 1 − 1 .
with H
2 rAi rBi

‘New’ as compared to H+ 2 is in particular the repulsive term with r12 , the dis-
tance of the two electrons, while rAi and rBi are the distances of electron i from
atomic nucleus A and B, respectively, and R is again the distance between the nu-
208 3 Diatomic Molecules

24 24

+ D' 1 Π u 4p 22
22
H2 V Π u 4f
1

2Σ+ +
20 D'' 1 Π u 5p u 2pσu B'' 1Σu 4pσ 20
+
+ g 3Σg 3dσ
B' 1Σu3pσ
+ H++H(1s)
18 X 2Σg 1sσg
+ 5ℓ 4ℓ
e 3Σu2pσ
d 3Π u 3p H(1s)+H(3ℓ)
+
16 m 3Σu 4fσ 3 16
3Σ+
i Π g 3d
h g 3sσ H(1s)+H(2ℓ)

14 14
-
H(2s)+H (1s 2)
D 1Π u 3p and J,j 1,3∆ g 3d δ
potential energy V(R ) / eV

12 I Π g 3d 12
1
B Σu 2pσ
+
C 1 Π g 2p m 3Σu 4fσ
+ +
C +e- 2Σg a 3Σg 2sσ
10 10
+
H 1Σg 3sσ
+
c 3Π u 2p
+ +
c +e- 2Σg E,F 1Σg 2sσ + 2pσ2
8 H2 8

+
6 b 3Σu 2pσ 6

+ H(1s)+H(1s)
b +e- 2Σg
4
4
- 10
H2 -
X + e - 2Σ g
+ H(1s)+H
-
(1s2)

2 5 2
H2
+
0
X 1Σg 1sσ 0

0 0.1 0.2 0.3 0.4 0.5 0.6

nuclear distance R / nm

Fig. 3.41 Potentials for the most important states of the H2 molecule (for comparison also for the
anion H− +
2 and the cation H2 ) after S HARP (1971). The equilibrium distance of H2 in its X Σg
1 +

electronic ground state is R0 = 0.07416 nm, the dissociation energy D0 = 4.476 eV. The b 3 Σu+
state is repulsive. For brevity we omit the description 1sσ . Note that the ion pair p + H− (1s 2 ) is
formed at 17.5 eV

clei. The molecular potential for a state φγ is found by minimizing (3.95). In the
R ITZ variational method one may construct φγ as linear combination of MOs φg
and φu . The MOs needed for the ansatz (3.147) may e.g. be found by first solving
3.6 Construction of Total Angular Momentum States 209

17 17
V 1 Π u 4f
H(1s)
+H(3ℓ)

H2+ +
m 3Σu 4fσ
+
16 g 3Σg 3dσ 16
D1 Π u 3p
X 2Σ+ d 3Π u 3p
g
i 3Π g 3d
g 3d δ
J,j 1,3∆
I Π g 3d
D'' 1 Π u 5p D' 1 Π u 4p
+
15 h 3Σg 3sσ 15
B'' 1Σ + 4pσ
u + H(1s)
potential energy V(R ) / eV

B' 1Σu3pσ
+H(2ℓ)

14 +
+ 14
H 1Σg 3sσ

C 1 Π g 2p
3
a Σg 2sσ
13 e 3Σ+ 2pσ 13
u
+
E,F 1Σg 2sσ + 2pσ2

12
c 3Π u 2p
H2 12

+
B 1Σu 2pσ

+
b 3Σu 2pσ
11 11
0.05 0.1 0.15 0.2 0.25 0.3 0.35
nuclear distance R / nm

Fig. 3.42 Enlarged part of the potential energy diagram for H2 Fig. 3.41 after S HARP (1971).
Note the interesting states designated E, F and H with double minima, and the “predissociating”
states with double minimum potential marked I , i and h
210 3 Diatomic Molecules

the S CHRÖDINGER equation for H+

2

i φ(r i ) = W φ(r i ).
H

However, it is also possible and quite instructive to start with LCAO-MOs as a first
approximation, in the most simple case with (3.110). Explicitly for the ground state
(without normalization) this leads to:

φX(S) (1, 2) ∝ Φ1s (rA1 )Φ1s (rB2 ) + Φ1s (rA2 )Φ1s (rB1 ) (3.150)

+ Φ1s (rA1 )Φ1s (rA2 ) + Φ1s (rB1 )Φ1s (rB2 ) . (3.151)

The two parts (3.150) and (3.151) of this wave function may be identified with a
covalent and an ionic contribution, respectively. The covalent part represents a rel-
atively uniform distribution of the two electrons at both atoms. For large distances
between the atoms, the covalent wave function eventually merges with the descrip-
tion of two separated atoms H + H. In contrast, the ionic part localizes both electron
either at atom A, i.e. Φ1s (rA1 )Φ1s (rA2 ), or at atom B, i.e. Φ1s (rB1 )Φ1s (rB2 ). This
corresponds thus to the situation H− + H+ and H+ + H− , respectively. Asymptot-
ically the thus described state is found for H2 (in its vibrational ground state) at an
energy WI + D00 − WEA  17.32 eV, where WI is the ionization potential of the H
atom, with D00 being the dissociation energy of H2 in its vibrational ground state,
and WEA stands for electron affinity of the H atom. Looking at Fig. 3.41 tells us that
this is far above the H + H asymptote. Interestingly, the ionic state contributes to
the excited B 1 Σg+ state which is much weaker bound than the X 1 Σg+ ground state
of H2 .
To finally evaluate the energy according to (3.95) with the simple trial function
(3.150)–(3.151) one again introduces the coordinate transformation (3.97) in order
to compute the integrals which evolve. We refrain from presenting here the details
of the calculation – in particular because the results of this procedure is not really
convincing: on obtains with this ansatz R0 = 0.08 nm and De = 2.68 eV – which has
to be compared with the precisely calculated and experimentally determined values
R0 = 0.07414 nm and D00 + ω0 /2 = 4.747 eV (see Table 3.2).

3.6.8 Valence Bond Theory

Instead of starting from MOs, one may simply use a covalent charge distribution as
trial function. One chooses, e.g. for the H2 ground state (not normalized)
 
φX(S) (1, 2) = Φ1s (rA1 )Φ1s (rB2 ) + Φ1s (rA2 )Φ1s (rB1 ) × χ00 (1, 2) (3.152)

and computes the energy of the X 1 Σg+ state from

|φX(S) 
φX(S) |H
WX(S) = .
φX(S) |φX(S) 
3.6 Construction of Total Angular Momentum States 211

Correspondingly the ansatz

 
φb(T ) (1, 2) = Φ1s (rA1 )Φ1s (rB2 ) − Φ1s (rA2 )Φ1s (rB1 ) × χM
1
S
(1, 2)

leads to the repulsive potential of the non-bonding 3 Σu+ state.

This so called valence bond theory, first introduced by H EITLER and L ONDON,
is less complicated than the MO method and gives, in the case of H2 , even somewhat
better results that the simple MO ansatz just discussed.11
With today’s state-of-the-art computers and quantum chemical methods one may
obtain nearly ‘exact’ ab initio molecular potentials, wave functions and other prop-
erties – at least for small molecules, without recurring to the above discussed ap-
proximations. Today, such programmes use large, well tested MO or AO basis sets
and have already been discussed in Vol. 1 for atomic multi-electron systems. Sophis-
ticated H ARTREE -F OCK methods (HF-SCF) with configuration interaction (CI) and
various corrections are applied. We refrain from listing the different acronyms for a
large variety of elaborate approximations. Today, for larger molecules DFT is used
more and more, which ultimately is based on variational principles for the electron
density, similar to those sketched here for MOs. A range of very powerful comput-
ing programmes are available on a commercial basis for various application areas
(e.g. G AUSSIAN 2013; M OLPRO 2012; T URBOMOLE 2010; GAMESS 2010).

3.6.9 Nitrogen and Oxygen Molecule

As examples for more complex, diatomic, homonuclear molecules – in comparison

to H2 in a 1s 2 configuration – the potentials for N2 and O2 are shown in Figs. 3.43
and 3.44, respectively.
According to Table 3.7, N2 is the most stable of the small molecules. K and L
shells are completely filled, which include all bonding MOs based on 2p atomic or-
bitals. In the X 1 Σg+ ground state 6 bonding 2p electrons participate in the bonding
and the spins are saturated (singlet state); N2 represents, so to say, a closed molec-
ular shell. Correspondingly, N2 is also chemically very inert. The first excited state,
A 3 Σu+ state lies with 6 eV rather high above the ground state, and the ionization
potential of 15.6 eV belongs to the highest ones among all molecules. In contrast,
the anion is not stable at all. Its ground state X 2 Πg is drawn in Fig. 3.43 as grey
line and lies 1.6 eV above the neutral N2 ground state. As (short-lived) resonance it
may, however, be observed very clearly in electron scattering. We shall come back
to this in Sect. 8.1.2.
Corresponding to the atomic configuration of the O atom in the ground state,
2s 2 2p 4 3 P, one expects for the O2 molecule a more complex electronic structure.
According to Table 3.7, two antibonding πg∗ electrons have to be added to the very
stable N2 core (1 Σg+ ) with its spatially symmetric wave functions. In Sect. 3.6.5 we

11 Whichisn’t really surprising here, as we have seen that the ionic component of the MOs leads to
completely wrong asymptotic states.
212 3 Diatomic Molecules

28
N(2Do) + N+(3P)
2
u
26
N(4So) + N+(1D)
2Σ –
C 2Σ + u
u
2∆
24 u 4
N(4So) + N+(3P)
4Σ –
u
u

22 B 2Σ u+ 4∆
u

4Σ + D2 g
u
20
+
N2 A 2 Πu
–
18 + +N
N

16 X 2 Σ g+
C' 3 N(2Do) + N(2Do)
u
14 +
b' 1Σu N(4So) + N (2P o)
+
potential energy V(R ) / eV

E 3Σ
g N(4So) + N(2Do)
12 3
C u
5 –
u +
7Σ N(4So) + N (3P)
10 a1 g 5Σ
+ u
w 1∆u g
1Σ –
N(4So) + N(4So)
a' u 3∆
– u
8 B' 3Σu

B 3Π g
6 +
A 3Σu
–
4 N2 N2
X2 g
2
X 1Σg+

0
0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36

nuclear distance R / nm

Fig. 3.43 Potentials for the nitrogen molecule and its ions according to G ILMORE (1965). The
potential of the unstable N−2 anion has been adjusted to fit the experimental observations from
electron scattering resonances (see Sect. 8.1.2)
3.6 Construction of Total Angular Momentum States 213

24

O(3P) + O+(2Do)
22
2Φ
2Σ
– u O(1D) + O+(4So)
g
2Δ 6Σ +, 6Σ + , 6 6
20 g g u u, g
4Σ
+
4
– g g O(3P) + O+(4So)
2Σ
b 4Σ g g

18 2Σ
+ +
A2 u g and 4Σu
+
16 O2
a4 u

X2 g
potential energy V(R ) / eV

14

1Σ
+
12 u O(1D) + O(1S)

10 O(3P) + O(1S)
1Δ
u
O2 O(1D) + O(1D)
8
O(3P) + O(1D)
3Σ
–
1 B u 5 5Σ + 5Δ 5 +
g u g g and Σg
6 3 3
1
u
– +
5Σ , 3Σ and 1Σ
+
g u u u g
+
A 3Σ u O(3P) + O(3P)
A' 3Δ
4 u -
c 1Σ u 2
– o
u
O(3P) + O (2P )
+
b 1Σ
g –
2
a 1Δg
O2
–
0 X 3Σ g X2 g

0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36

nuclear distance R / nm

Fig. 3.44 Potentials for the oxygen molecule and its ions according to G ILMORE (1965); minima
of the A, A and c states have been modified according to J ENOUVRIER et al. (1999)
214 3 Diatomic Molecules

have already discussed that the ground state X 3 Σg has highest multiplicity accord-
ing to H UND’s rules. The next higher states (built from the same atomic electron
configurations) are the a 1 g and b 1 Σg+ states.
As already mentioned, oxygen gas O2 in its ground state X 3 Σg is paramagnetic,
due to its two electrons with parallel spin, S = 1. We also note that a stable anion
exists, O− − −
2 in the ground state (in contrast to H2 and N2 ). Nevertheless, the anion
−
decays when vibrationally excited: O2 → O2 + e; conversely, it may also be gen-
erated in a vibrationally excited state, again as a resonance in electron scattering
by O2 .
The overviews on the potentials of N2 and O2 presented here (as well as that
on NO in Fig. 3.53) are based on the very systematic work of G ILMORE (1965).
Spectroscopic data have been evaluated there in the spirit of the RKR method (see
Sect. 3.4.6), but also quantum chemical computations have been included. Even
though since then many new measurements and calculations have been published,
we have added only a few modifications to this data compilation, since a comparable
critical overall evaluation of all available data has – up to now – not been reported.

Section summary
• The electronic states of diatomic molecules are characterized by projections
of total angular momenta (sums over all active electrons) onto the internuclear
axis.
• In the simplest cases (e.g. for H UND’s case a) the projection of the total orbital
angular momentum, Λ, and total spin, S, are good quantum numbers. States
with Λ = 0, 1, 2 etc. are designated by Σ, Π,  etc., while the spin quantum
numbers S = 0, 1/2, 1 etc. lead to singlets, doublets, triplets etc. In analogy
to atoms, typically the notation is 2S+1 Λg,u is used.
• The five H UND’s cases differ by the strength of the interaction between in-
ternuclear field, orbital, spin and rotational angular momentum. The ensuing
different angular momentum coupling schemes lead to spectroscopically rel-
evant differences in the energy spectra.
• In addition to angular momenta, molecular states are characterized by their
symmetries. For homonuclear, diatomic molecules, the relevant symmetry op-
erations are inversion in respect of the origin, and reflection at planes through
the molecular axis.
• For multi-electron systems (exemplified by H2 and O2 ) positive or negative
reflection symmetry lead to different energies for Σ states – and only for
these. For other values of Λ, states with ± reflection symmetry are degenerate
if the molecule does not rotate. However, high rotational levels show a small
splitting, the so called Λ-type doubling.
• Potential energy diagrams for H2 , O2 , and N2 show many interesting struc-
tures. In all three cases negative ions (anions) may be observed as short-lived
resonances in electron scattering, while only the O− 2 anion has a stable vibra-
tional ground state.
3.7 Heteronuclear Molecules 215

• N2 is the most strongly bound diatomic molecule (De  9.9 eV) and has also
a particularly high ionization potential. H2 and N2 have rather isolated elec-
tronic ground states, while O2 has three lowest states separated by less than
1.7 eV and ordered according to H UND’s rules: the X 3 Σg− ground state (para-
magnetic), the a 1 g and the b 1 Σg+ state.

3.7 Heteronuclear Molecules

3.7.1 Energy Terms

We generalize now the considerations on molecular orbitals outlined in Sect. 3.5.

For diatomic molecules with two different atomic nuclei inversion symmetry is no
longer relevant, and the molecules can no longer be characterized as even or odd.
As a rule, the covalent bonding energy is weaker as for homonuclear molecules, and
HAA = HBB . We recall the simplest approximation for orbital energies (3.107) and
(3.108) in the case of homonuclear molecules

"g,u = HAA ± HAB .

For the sake of simplicity we have set the overlap integral S = 0 – which will
change the overall trends only when both atoms come very close. For heteronuclear
molecules the secular equation (3.106) thus is

(HAA − ")(HBB − ") − HAB HBA = 0.

With HAB HBA = |HAB |2 the energy eigenvalues become now


HAA + HBB (HAA − HBB )2
"± = ± + |HAB |2
2 4

HAA + HBB HAA − HBB 4|HAB |2
= ± 1+ .
2 2 (HAA − HBB )2

For many heteronuclear molecules |HAA − HBB | |HAB | holds (in particular for
those with very different nuclear charges). Then the two eigenvalues are approxi-
mately

|HAB |2 |HAB |2
"+  HAA + and "−  HBB − . (3.153)
HAA − HBB HAA − HBB
Figure 3.45 illustrates the raising and lowering of the atomic orbital energies. We
have assumed here that HAA > HBB .
Instead of the even and odd functions (3.110) and (3.111), respectively, we obtain
now for the orbitals
(±) (±)
φ± = cA ΦA + cB ΦB
216 3 Diatomic Molecules

Fig. 3.45 Schematic of

|HAB |2
raising and lowering of ε+ ___________
atomic orbital energies for HAA − HBB
HAA
heteronuclear, diatomic
molecules due to the |HAB |2 HBB
resonance integral HAB
___________ ε-
(3.105) HAA − HBB B A

Fig. 3.46 Orbital energies 6σ*

for heteronuclear diatomic
molecules 2 * 2p
2p 5σ
1

4σ *
2s
2s 3σ

2σ *
1s
1s 1σ

A B

(+)
cA HAB HAA − HBB
with  = 1 and
(+)
cB "+ − HAA HAB
(−)
cA HAB HAB
− − = 1.
cB(−) "− − HAA HAA − HBB

The molecular orbitals are thus rather similar to the AOs, i.e. φ+  ΦA and φ− 
−ΦB . The lower level correlates as expected with ΦB .

3.7.2 Filling the Orbitals with Electrons

A direct consequence of the asymmetric molecular orbitals is an unequal distribution

of charge in the molecule. Consequently, heteronuclear diatomic molecules have a
permanent dipole moment. In comparison to the homonuclear molecules Fig. 3.37
the building-up of MOs is somewhat more specific.
For molecules made of atoms with similar nuclear charges (e.g. CO) one obtains
MOs from LCAO as schematically indicated in Fig. 3.46. Since there is no longer a
difference between g and u, the designation is done simply by consecutively num-
bering the levels for each λ. The respective correlation diagrams are based on this
scheme, and in the limit of separated atoms two different energies appear for the
atomic orbitals 1s, 2s, 2p etc. Apart from this the same non-crossing rules are valid.
A typical correlation diagram is shown in Fig. 3.47 – to be compared to Fig. 3.36.
As for the homonuclear molecules the states with higher λ but equal  lead to higher
3.7 Heteronuclear Molecules 217

3d δ 3π
3d 3d π 7σ* σ*3sA
3dσ* 3sA
6σ* π 2pB
3p 3p π 2π σ2pB 2pB
3pσ 5σ* π 2pA 2pA
3s σ2pA
3sσ* 4σ* σ*2sB 2sB
2p π 1π 2sA
2p 3σ σ2sA
2s 2pσ
2sσ* 2σ* σ*1sB
1sB
1sA
σ1sA
1s 1sσ 1σ
united atom molecule separated atoms ZA > ZB
(R = 0) R (R → ∞)

Fig. 3.47 Correlation diagram for heteronuclear diatomic molecules

Table 3.9 MOs for heteronuclear diatomic molecules

Molecule Expected electronic configuration Bond-order 2S+1 Λ Observed
R0 nm De eV
BeO (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 2 1Σ + 0.13309 4.69
BeF (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )1 2 12 2Σ + 0.13610 ≈6.0
BNa (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)3 (5σ )1 2 3Π 0.13291 3.99
BO (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )1 2 12 2Σ + 0.12045 8.40
BF (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 3 1Σ + 0.12625 7.89
CN+ (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 2 1Σ 0.11729 4.93
CN (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )1 2 12 2Σ + 0.11718 7.83
CN− (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 3 0.114 ≈10
CO+ (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )1 2 12 2Σ + 0.11151 8.47
CO (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 3 1Σ + 0.11283 11.11
CF (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 (2π)1 2 12 2Π 0.12718 5.75
NO+ (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 3 1Σ + 0.10632 11.00
NO (1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 (2π)1 2 12 2Π 0.11508 6.61
R0 and De as far as not otherwise mentioned according to H UBER and H ERZBERG (1979)
a LI and PALDUS (2006)

energies. We have discussed this in the context of the S TARK effect in Sect. 8.2.7,
Vol. 1. For large internuclear distances the energetic ordering is thus σ < π < δ.
However, due to the necessary correlation with the united atom scheme this trend
may be inverted in the MO region. In Fig. 3.47 this is particularly evident for the
1π , 5σ , 2π and 6σ states.
In Table 3.9 the electron configurations, equilibrium distances R0 and the bond
energies De of the ground state for a number of heteronuclear diatomic molecules
are summarized.
218 3 Diatomic Molecules

It should be noted that the scheme communicated here gives only a first overview
of potential MOs. In practice one has to investigate the energetics and each atomic
configuration in detail to find a suitable MO basis set for describing a particular
molecule. As already mentioned for homonuclear molecules, today a number of
sophisticated methods and efficient computer programmes are available for this task.
To get somewhat more specific, we shall in the following discuss a few interesting
examples in more detail.

3.7.3 Lithiumhydrid

LiH is quite a remarkable molecule – albeit not available in a gas bottle. A number
of specialities may be studied with it. In recent years it has even attracted interest
in the context of B OSE -E INSTEIN condensation of ultracold atoms (see e.g. C ÔTE
et al. 2000; J UARROS et al. 2006), not least because of its large dipole moment
which makes one expect specific long range effects. One may e.g. apply two-photon
processes to generate such molecules by photo-association from ultracold atoms.
The constituents of the LiH molecule have the configuration 1s 2 2s (Li) and
1s (H), whereas the closed 1s 2 does not participate in the bonding. These elec-
trons fill the 1σ MO localized at the Li nucleus. A comparison of the first excited
atomic states 1s 2 2p (Li) and 2s, p (H) with 1.848 eV and 10.2 eV excitation energy,
respectively, shows that a correlation diagram would not lead us very far in this case.
According to the above considerations, the active MOs will be determined by the
easy to excite AOs of Li. The ground state is essentially given by the 2s electron
of Li which interacts with the 1s electron of the hydrogen atom: the atomic orbitals
2s (Li) and 1s (H) form a 2σ MO. In Fig. 3.48(a) the potentials for some states of
LiH are collected from several sources.
Most clear is the situation for the triplet states 3 Σ, where the PAULI principle
provokes a spatial repulsion of the σ MOs from n − 1s. A closer inspection shows
(not recognizable in Fig. 3.48) that here too long range dispersion interaction leads
to a weak attraction. For the a 3 Σ + state this is of importance in a distance of 1 −
2 nm and is of significant relevance for the interaction among ultracold atoms.
More complicated are the conditions for the singlet states for which in addition
to the symmetric, neutral AO configuration Li0 H0 also the ion pair Li+ H− must be
accounted for. The hypothetical energy of the ion pair is indicated in Fig. 3.48(a)
by the dashed red line. As seen, the asymptotic energy of this ion pair lies below
the ionization energy of Li (5.39 eV) by just the electron affinity of the H atoms,
WEA = 0.756 eV. A suitable MO which accounts for this, will be characterized by a
significantly asymmetric charge distribution.
In the most simple approach one realizes this by a linear combination of a 2s and
a 2pz AO of lithium (a |σ 2s and a |σ 2p AO, in respect of the internuclear axis z).
This is sketched in Fig. 3.49.12 Such a combination of atomic orbitals with different

12 We remind the reader that such schematic polar plots of orbitals just display the magnitude of
the wave function under a given polar angle θ at fixed radial distance. They do not give a realistic
image of the spatial extension of the orbital.
3.7 Heteronuclear Molecules 219

V R) / eV
V( Li++H
dipole moment / ea0
–
5 (a) Li++ H WEA (b)
4 well depth R 4
ca. 34.5 meV A state
3 Σ+ X state
3 3

2 B 1Π +
2
Li(2p 2P) + H(1s 2S)
A 1Σ +
1 1
Li+H– Li(2s 2S) + H(1s 2S)
0 a3Σ+ 0
B state
−1 -1
X 1Σ +
−2 -2
0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0
R / nm

Fig. 3.48 Potentials (a) and dipole moments (b) of the LiH molecule. Ground state and excited
state A 1 Σ + are strongly influenced by the ionic Li+ H− configuration. Triplet states and B state
are essentially covalent (adapted from PARTRIDGE and L ANGHOFF 1981; Y IANNOPOULOU et al.
1999; C ÔTE et al. 2000)

Fig. 3.49 Formation of an

s-p hybrid orbital + z - + z

1
_ |σ 2s〉 √3
√
|σ 2p〉
+

2 2
hybrid: = |2spσ 〉

+ z
–

angular momenta is called a hybrid orbital. According to PAULING (1931), who first
introduced this important concept of hybridization, the bond will be strongest for a
wave function which has the largest value in the direction of the bond. In the√present
case one easily verifies that this implies superposition coefficients 1/2 and 3/2 for
the |σ 2s and |σ 2p orbital, respectively, obviously resulting in a quite asymmetric
shape of the orbital. We recall that such type of orbital was already encountered in
Sect. 8.2.8, Vol. 1 in the context of the linear S TARK effect for the 2p, 2s states of
atomic hydrogen. When a molecule is formed, the very strong electric field of the
two atomic nuclei acts correspondingly. The phenomenon always occurs if ns and
np states have similar energies. A more detailed treatment of hybridization will be
given in Sect. 4.4.2.
220 3 Diatomic Molecules

For LiH one consequence is a significant displacement of the charge distribution,

away from the Li atom towards the H atom. This in turn leads to a large dipole mo-
ment for the X state as documented in Fig. 3.48(b). It is remarkable that it reaches
its maximum at (0.2 to 0.3) nm, i.e. just there where the hypothetical potential curve
for Li+ H− (red dashed) is particularly close to the true (computed) potential of the
X 1 Σ + ground state. Obviously the dominantly ionic character of the orbitals in-
volved leads to the bonding. Hence, the singlet state lies here (contrary to the usually
valid H UND’s rules) lower than the triplet state which has no ionic character.
To avoid misunderstandings: the dashed potential for the ionic Li+ H− bonding
does not represent a real state at small distances: the corresponding MOs are, so to
say, an integral part of the X 1 Σ + ground state, but to some extend also of the ex-
cited A 1 Σ + state.13 In the region of the “avoided crossing” at about (0.4 to 0.6) nm
the A state has even a larger dipole moment as seen in Fig. 3.48(b): one reads a
maximum of 4.2ea0 at 0.5 nm, corresponding to a charge delocalization of more
than 40 %. Obviously the sign of this charge displacement reverses for the A state
at small internuclear distances, i.e. when the H atom dives fully into the excited 2p
orbital of the Li atom.
The potential of the A state has another remarkable property: it has a negative
value of ωe xe , i.e. in contrast to the usually found behaviour (as described e.g. by a
M ORSE potential) it is flat at the bottom and becomes steeper at higher energies –
just another consequence of the locally very ionic character of the molecular bond.
Interesting is also the B 1 Π state, which correlates with the excited 2px or 2py
AOs. The overlap with the 1s H orbital is minimal. Due to the π character of the or-
bitals involved there is, however, no avoided crossing with the hypothetical ionic po-
tential curve (the ground state of Li− has the configuration 1s 2 2s 2 ). This gives rise
to the very flat shape of the B state shown in Fig. 3.48(a). The well depth is, how-
ever, sufficient to sustain 3 stable vibrational states (PARTRIDGE and L ANGHOFF
1981), which could possibly be used as intermediate states for the proposed molec-
ular B OSE -E INSTEIN in this molecule (J UARROS et al. 2006; D ULIEU and G AB -
BANINI 2009).
Also a number of higher excited singlet and triplet states are known, some with
interesting shapes of their potential minima. The computation requires extended
basis sets up to f orbitals (Y IANNOPOULOU et al. 1999).

3.7.4 Alkali Halides: Ionic Bonding

The alkali halides (i.e. LiF, NaCl, NaI, KBr etc.) are prototypical examples for ionic
molecular bonding. As crystal solids some of them are also of considerable practical
importance. Their preparation as free molecules for spectroscopic purposes requires

13 Inprinciple, one may let a Li+ cation collide with an H− anion in a scattering experiment. For
large distances, the interaction potential of such a system is indeed given by the red dashed line
(∝ −1/r) and remains – for sufficiently high kinetic energies – a good first approximation also at
smaller distances. Transitions at the (avoided) crossings may indeed be induced in such collision
processes.
3.7 Heteronuclear Molecules 221

Fig. 3.50 Several 1 Σ + 4

potential curves for NaCl as a NaCl
3 1Σ +
typical example for ion Na(3p) + Cl(3p)5
bonding: the red dashed –
2 Na+ + Cl
curved shows the potential Vγ (∞) =1.522 eV

potential energy / eV
for a hypothetical, purely 1
A1Σ+
ionic system Na+ + Cl− 0
which converges for large R
– Na(3s) + Cl(3p)5
towards 1.522 eV. The -1 Na+Cl
ground state is dominated by
-2
the ionic bonding. At about
X1 Σ +
1 nm one sees the classical -3
avoided crossing with the RH
excited A 1 Σ + state which D e=
becomes covalent for small 4.23eV 0.2 0.4 0.6 0.8 1.0 1.2 1.4
internuclear distances R0 = 0.236079 nm R / nm

some efforts (e.g. evaporation at high temperatures). Nonetheless a wealth of pub-

lications exists about these interesting molecules. They have played a key role in
understanding elementary chemical reactions in molecular beam studies, for which
H ERSCHBACH, L EE and P OLANYI (1986) have been honoured with the N OBEL
prize in chemistry. The key theme is reactive processes of the type

A + BC → A+ + (BC)− → A+ B− + C, (3.154)

where A is typically an alkali atom and BC a halogen containing molecule (e.g.

Br2 , CCl4 etc.). The rather low ionization potential WI of the alkali atoms and the
high electron affinity WEA of the halogens lead to a positive energy balance for the
electron jump indicated as intermediate in the reaction (3.154). For this to happen
efficiently, the reaction partners have to approach each other close enough, say to a
distance RH .
Once the ion pair is formed, the C OULOMB attraction of the ions leads to intense
interaction. The whole process is called harpooning: the electron of atom A is, so
to say, fired with a harpoon at the molecule BC to catch it. In real scattering exper-
iments one measures the angular distribution of reactants and/or reaction products
after the interaction process. From the harpooning radius RH one obtains an esti-
mate for the reaction cross section σr  πRH 2 . One may illustrate the electron jump

in principle already by a diatomic molecule, as shown by Fig. 3.50 for the example
NaCl.
The configuration of AOs in this system is 3s (Na) and 3s 2 3p 5 (Cl). The 3s
valence electron of the Na atom is only weakly bound and is given off easily to
complete the 3p shell in Cl. The ionization potential of Na is WI = 5.1391 eV, the
electron affinity of Cl is WEA = 3.617 eV. At infinite distance one thus needs an
energy

Vγ (∞) = WI − WEA = 1.522 eV

222 3 Diatomic Molecules

to form the ion pair Na+ and Cl− . However, at shorter distances energy is gained by
the C OULOMB attraction and the ionic potential is expected to be

e2
Vγ (R) = Vγ (∞) − ,
4πε0 R

which is plotted as red dashed line in Fig. 3.50. At distances R < RH  1 nm

C OULOMB interaction and bond energy for Na+ + Cl− are more than compen-
sated and the electron may ‘jump’ from Na to the Cl atom. At internuclear dis-
tances R < RH one thus expects spontaneous formation of the ion pair Na+ Cl− ,
and consequently of a bound molecule. At very small distances the electron densi-
ties of the closed shells of Na+ and Cl− begin to overlap – what finally leads to a
strongly repulsive potential just as for rare gases. The overall result is the potential14
for a rather well bound molecule in the X 1 Σ + ground state (full black line) with
De = 4.2303 eV at R0 = 0.236 079 nm. Comparison of the hypothetical ionic poten-
tial with the true potential (semi-empirically computed on the basis of spectroscopic
data) shown in Fig. 3.50 illustrates impressively that the ionic bonding model repro-
duces the situation very well between R0 and RH . Consequently, the dipole moment
of NaCl is large: DX = 30 × 10−30 C m (see Table 3.2). At equilibrium distance R0
this corresponds an effective charge displacement of about 0.8e!
At RH we see the aforementioned avoided crossing with the excited A 1 Σ + state.
Obviously, the latter is dominated by covalent orbitals up to RH – beyond which the
ionic character takes over. The potential thus forms a small well in which several
vibrational states can be bound. For energetic reasons, the next higher state, corre-
lating with the excited 3p electron of the Na, can not cross the ionic state.
The relations found in the case of NaCl are rather typical for all alkali halides –
except for spin-orbit interaction which naturally increases with the atomic number
and can no longer be neglected for large Z. For atomic iodine it amounts already
to 0.9078 eV, much larger that the respective exchange interaction. We discuss now
the consequences and some more details for NaI on the basis of relatively recent ab
initio computations by A LEKSEYEV et al. (2000). The electronic configuration of
valence electrons in this calculation included 9 + 7 electrons in the atomic orbitals:
2s 2 2p 6 3s (Na) and 5s 2 5p 5 (I).
For the highest of these outer shell LCAO-MOs this is illustrated schematically
in Fig. 3.51. The strongly bound X 1 Σ + ground state is dominated by the . . . σ 2 π 4
ionic configuration. The lowest excited states are 1 Π and 3 Π with the covalent MO
configuration . . . σ 2 π 3 σ ∗ . They differ only little in energy, since exchange interac-
tion between the σ ∗ and π orbitals is very weak. As indicated in Fig. 3.51, these

14 For small distances and around the potential minimum we have used the potential derived an-

alytically from spectroscopic data reported by R AM et al. (1997). For larger distances the semi-
empirical valence bond calculations of C OOPER et al. (1987) appeared more plausible: experimen-
tal data were available up to v = 8 only (ca. 350 meV). It does not seem reasonable to assume
that the extrapolation of such data to about 4 eV excitation energy gives reliable information in the
vicinity of RH .
3.7 Heteronuclear Molecules 223

σ*
3s σ* antibonding
π π 4 "lone pairs"
5p
σ bonding
σ
Na NaI I

Fig. 3.51 MOs from AOs for the alkali halogenide molecules. Due to their very different orbital
energies the two “lone pairs” of π electrons are localized at the I atom and do not participate in
bonding

Fig. 3.52 The most 3

important potential curves for
0+ NaI –
Na+ + I
NaI in Ω symmetry 2
according to A LEKSEYEV et Na(2P 1/2) + I(2P
3/2)
B0+
potential energy / eV

al. (2000). The red dashed 1

curve gives again the A0+ Na(2S1/2) + I(2P1/2)
potential for the hypothetical 0
ionic system Na+ + I− , – Na(2S1/2) + I(2P3/2)
which for large R converges Na+I
-1
to 5.105 eV. In the present
case this leads to two avoided -2
crossings at RH1 = 7.2 and
X 0+ RH1 RH2
RH2 = 12.5a0 . These in turn
D e=
create two quasi bound, 0.4 0.6 0.8 1.0 1.2 1.4
excited states A0+ and B0+ 3.026eV
R0 = 0.2836Å R / nm

orbitals are localized in great distance from each other, at the Na and I atom, respec-
tively. The next states, A1 Σ + and 3 Σ + states, which are formed with the . . . σ π 4 σ ∗
MOs, lie somewhat higher, as the excitation of the σ orbital requires more energy
than exciting a “lone pair” of electrons out of the π orbitals. At this stage of the com-
putation one obtains a picture very similar to that in the case of NaCl in Fig. 3.50:
with pronounced curve crossings as a consequence of the ionic influence.
However, one now has account for spin-orbit interaction and diagonalize the
whole Hamiltonian again with it included. This leads now to a range of Ω ± states
which arise from Λ + Σ as we have discussed formally already in Sect. 3.6.3. Fig-
ure 3.52 shows only three of these states, which can be accessed by optical excitation
from the ground state.
According to A LEKSEYEV et al. (2000), the above mentioned Λ states with
S = 0 and 1 are responsible for the X0+ ground state and five more Ω states (2(I ),
1(I ), 1(II), 0− (I ) and A0+ ) which correlate asymptotically with Na(3 2 S1/2 ) + I
(5 2 P3/2 ). For symmetry reasons, of those only the A0+ state interacts with the
ionic configuration and leads to an avoided crossing. In addition, there are the states
(1(III), 0− (II) and B0+ ) which correlate for R → ∞ with Na(3 2 S1/2 ) + I(5 2 P1/2 ).
Of those, again only the B0+ state interacts with the ionic configuration and avoids
the crossing. Those two avoided crossings give rise to two excited, quasi-bound
states, A0+ and B0+ , as clearly seen in Fig. 3.52. For completeness we also have
224 3 Diatomic Molecules

drawn another state, 0+ (IV) according to C OOPER et al. (1987), which correlates
with Na(3 2 P1/2 ) + I(5 2 P3/2 ) which is, however, not bonding since the ionic poten-
tial does not cross this state. The diagonalization of the spin-orbit interaction hardly
influences the ground state. It is, however, responsible for a lowering or raising
of the asymptotic potentials by −1/3 and +2/3 of the atomic spin-orbit splitting
(0.9426 eV). Thus, the overall bonding energy of the X0+ ground state becomes
De = 3.02631 eV. The bond-length is R0 = 0.2836 nm.
NaI has played a key role in the development of femto-chemistry. With ultrashort
laser pulses one may generate wave-packets on the repulsive part of the excited A0+
state potential (just above the minimum of the X0+ at R0 , see Fig. 3.52). These
wave-packets may oscillate between the left and right border of the A0+ potential,
just as a classical oscillator. However, these vibrational states are not completely
stable. Speaking figuratively: at the avoided crossing the potential well has a leak,
through which during each oscillations a certain fraction of the wave-packet ‘flows
out’. With femtosecond pump-probe techniques one may follow this process. For his
seminal work studying such transition states in chemical reactions with femtosecond
spectroscopy Ahmed Z EWAIL (1999) received the N OBEL prize in chemistry.

3.7.5 Nitrogen Monoxide, NO

Before ending this discussion of diatomic molecules we briefly mention the NO

molecule. It is of great importance for several disciplines of science and technology,
e.g. in biology and physiology, in atmospheric chemistry, in technical combustion
processes and many more. NO is a very reactive radical but may nevertheless be
stored and transported in bottles. It has a dipole moment (i.e. it is infrared active) and
has a well known electronic structure. In molecular physics it is thus a very popular
reference object for which countless spectroscopic studies have been carried out.
In the context of multiphoton ionization processes in molecules it is valued almost
as a kind of ‘Drosophila’ of molecular physics (similar to Na in atomic physics).
Figure 3.53 gives an overview of the most important potentials for the states of
neutral NO and the lowest states of the NO+ cation. The data presented here are
again adapted from G ILMORE (1965).
Corresponding to the AO configuration . . . 2s 2 2p 3 (N) and . . . 2s 2 2p 4 (O) of its
constituents, the ground state of NO is characterized by a

(1σ )2 (2σ )2 (3σ )2 (4σ )2 (1π)4 (5σ )2 (2π)1

MO configuration. Obviously, NO has one unpaired 2π ∗ electron in its valence

shell. This is the cause of its high reactivity as a radical. NO is paramagnetic and its
ground state must be a X 2 Π state.
What can be illustrated very clearly by this example is the difference between
the so called molecular RYDBERG states and valence states. If only the outer, anti-
bonding 2π ∗ valence orbital15 is excited, then the excited molecule will be stronger

15 See Fig. 3.46 for the ordering of orbitals.

3.7 Heteronuclear Molecules 225

1Σ+ and 3Σ+

22 7Σ +
N+(3P)+O(3P)

1Δ 5 Σ+
20
1Σ - N(4So)+O+(4So)

A1
18 3 Σ-

w 3Δ

16 b3

NO+
14
potential energy V(R ) / eV

a 3Σ+

12
X 1Σ + 4 Σ- o
N(4S )+O(1S)
o
N(2P )+O(3P)
10 S 2Σ+, M 2Σ+, K 2 (from top to bottom)
F 2Δ H' 2 and H 2Σ o
2 Σ+
N(2D )+O(3P)
E 2Σ +
2Φ o
8 i 4 4Δ N(4S )+O(1D)
D 2Σ + G 2Σ
-
i
4Σ+ and 6Σ+
6
C2 B' 2Δ i o
N(4S )+O(3P)
6 -
5 and 5 Σ
A 2Σ + 3
B2 r
- o - o
b 4Σ N(4S )+O (2P )
4 a4 i

NO
2 NO-
-
X 3Σ
0

X2 r (1 )2 (2 )2 (3 )2 (4 )2 (1 )4 (5 )2 (2 )1
-2
0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36
nuclear distance R / nm

Fig. 3.53 Potentials for nitrogen monoxide and its ions according to G ILMORE (1965)
226 3 Diatomic Molecules

bonded than the ground state (smaller R0 , more rigid potential). In this manner a
whole series of potentials is generated, A 2 Σ + , C 2 Π , D 2 Σ + , E 2 Σ + , with nearly
identical shape of their potentials. As seen in Fig. 3.53, these states converge to-
wards the ground state X 1 Π of the NO+ cations and reflect already its potential:
the outer electron is far away from the ion core of the molecule, and does not par-
ticipate in the bonding. It behaves like an atomic RYDBERG electron, hence one
speaks about molecular RYDBERG states. In contrast, the states a 4 Π , B 2 Π , b 4 Σ −
show neither a great similarity to the neutral nor to the ionic ground state. They are
significantly weaker bound than either of those: they are generated by exciting elec-
trons from lower lying, bonding MOs with subsequent rearrangement of the whole
electron configuration and a softening of the bonding. One speaks of valence states.
As states of equal symmetry must not cross, a whole series of avoided crossings
arises, both for RYDBERG and valence states. These are clearly seen in Fig. 3.53 at
energies between 7 eV and 9 eV.
As also indicated in Fig. 3.53, here too an anion exists. NO− is, however, not very
stable (electron affinity WEA = 24 meV). It may be observed readily as resonance in
low energy electron scattering. We shall come back to this briefly in Sect. 7.2.6.

Section summary
• MOs for heteronuclear molecules are constructed essentially by the same prin-
ciples as those for homonuclear molecules. Of course, the even (g) vs. odd (u)
symmetry is no longer relevant and correlation diagrams become somewhat
more complicated.
• The scheme for filling the orbitals, shown in Table 3.9, is consequently less
transparent and less rigorous than for homonuclear molecules.
• In principle, for all heteronuclear diatomic molecules ionic bonding is impor-
tant – the strength of its influence depending on the specific system. Caused
by this ionic character of the bonding, all have a permanent dipole moment,
i.e. they are microwave and infrared active.
• We have looked into the potential energy diagrams of several specific exam-
ples of heteronuclear, diatomic molecules, LiH, the alkali halides, NaCl and
NaI and finally NO, the ‘Drosophila’ of molecular physics. A number of inter-
esting properties have been discussed. We recall specifically the “harpooning”
mechanism in molecular reactions involving an intermediate ionic complex.

Acronyms and Terminology

AO: ‘Atomic orbital’, single electron wave function in an atom; typically the basis
for a rigorous structure calculation.
a.u.: ‘atomic units’, see Sect. 2.6.2 in Vol. 1.
BO: ‘B ORN O PPENHEIMER’, approximation, the basis when solving the S CHRÖ -
DINGER equation for molecules (see Sect. 3.2).
CI: ‘Configuration interaction’, mixing of states with different electronic configu-
rations in atomic and molecular structure calculations, using linear superpositon
of S LATER determinants (see Sect. 10.2.3 in Vol. 1).
Acronyms and Terminology 227

DFT: ‘Density functional theory’, today one of the standard methods for computing
atomic and molecular electron densities and energies (see Sect. 10.3 in Vol. 1).
E1: ‘Electric dipole’, transitions induced by the interaction of an electric dipole
with the electric field component of electromagnetic radiation.
EPR: ‘Electron paramagnetic resonance’, spectroscopy, also called electron spin
resonance ESR (see Sect. 9.5.2 in Vol. 1).
FIR: ‘Far infrared’, spectral range of electromagnetic radiation. Wavelengths be-
tween 3 µm and 1 mm according to ISO 21348 (2007).
FS: ‘Fine structure’, splitting of atomic and molecular energy levels due to spin
orbit interaction and other relativistic effects (Chap. 6 in Vol. 1).
FTIR: ‘F OURIER transform infrared spectroscopy’, (see Chap. 5, p. 298ff.).
FWHM: ‘Full width at half maximum’.
good quantum number: ‘Quantum number for eigenvalues of such observables that
may be measured simultaneously with the H AMILTON operator (see Sect. 2.6.5
in Vol. 1)’.
HF: ‘H ARTREE -F OCK’, method (approximation) for solving a multi-electron
S CHRÖDINGER equation, including exchange interaction.
HITRAN: ‘High-resolution transmission molecular absorption database’, http://
[Link]/hitran (ROTHMAN et al. 2009).
HOMO: ‘Highest occupied molecular orbital’.
IR: ‘Infrared’, spectral range of electromagnetic radiation. Wavelengths between
760 nm and 1 mm according to ISO 21348 (2007).
isotopologue: ‘Molecules that differ only in their isotopic composition’, [Link]
[Link]/wiki/Isotopologue.
LCAO: ‘Linear combination of atomic orbitals’, linear superposition of atomic, sin-
gle electron wave functions to form a molecular orbital (MO).
LUMO: ‘Lowest unoccupied molecular orbital’.
MO: ‘Molecular orbital’, single electron wave function in a molecule; typically the
basis for a rigorous molecular structure calculation.
NIR: ‘Near infrared’, spectral range of electromagnetic radiation. Wavelengths be-
tween 760 nm and 1.4 µm according to ISO 21348 (2007).
NIST: ‘National institute of standards and technology’, located at Gaithersburg
(MD) and Boulder (CO), USA. [Link]
NMR: ‘Nuclear magnetic resonance’, spectroscopy, a rather universal spectro-
scopic method for identifying molecules (see Sect. 9.5.3 in Vol. 1).
RKR: ‘RYDBERG -K LEIN -R EES’, precise method for determining molecular po-
tentials from spectroscopic data.
SCF: ‘Self-consistent field’, method for solving coupled integro-differential equa-
tions iteratively.
SI: ‘Système international d’unités’, international system of units (m, kg, s, A, K,
mol, cd), for details see the website of the Bureau International des Poids et Mé-
sure [Link] or NIST [Link]
html.
vdW: ‘VAN DER WAALS’, interaction between atoms or molecules, ∝ R −6 .
228 3 Diatomic Molecules

VMI: ‘Velocity map imaging’, experimental method for registration (and visual-
ization) of particle velocities as a function of their angular distribution (see Ap-
pendix B).

References
A LEKSEYEV , A. B., H. P. L IEBERMANN, R. J. B UENKER, N. BALAKRISHNAN, H. R. S ADEGH -
POUR , S. T. C ORNETT and M. J. C AVAGNERO : 2000. ‘Spin-orbit effects in photodissociation
of sodium iodide’. J. Chem. Phys., 113, 1514–1523.
B ORN , M. and R. O PPENHEIMER: 1927. ‘Zur Quantentheorie der Molekeln’. Ann. Phys. Berlin,
84, 0457–0484.
C OOPER , D. L., S. B IENSTOCK and A. DALGARNO: 1987. ‘Mutual neutralization and chemiion-
ization in collisions of alkali-metal and halogen atoms’. J. Chem. Phys., 86, 3845–3851.
C ÔTE , R., M. J. JAMIESON, Z. C. YAN, N. G EUM, G. H. J EUNG and A. DALGARNO: 2000.
‘Enhanced cooling of hydrogen atoms by lithium atoms’. Phys. Rev. Lett., 84, 2806–2809.
D ULIEU , O. and C. G ABBANINI: 2009. ‘The formation and interactions of cold and ultracold
molecules: new challenges for interdisciplinary physics’. Rep. Prog. Phys., 72, 086401.
D UNHAM , J. L.: 1932. ‘The energy levels of a rotating vibrator’. Phys. Rev., 41, 721–731.
F LEMING , H. E. and K. N. R AO: 1972. ‘A simple numerical evaluation of Rydberg-Klein-Rees
integrals – application to X1 Σ + state of 12 C16 O’. J. Mol. Spectrosc., 44, 189–193.
GAMESS: 2010. ‘General atomic and molecular electronic structure system’, Gordon research
group at Iowa State University, USA. [Link] accessed:
9 Jan 2014.
G AUSSIAN: 2013. ‘Gaussian 09 rev. D’, Gaussian, Inc., Wallingford, CT, USA. [Link]
[Link]/, accessed: 9 Jan 2014.
G ILMORE , F. R.: 1965. ‘Potential energy curves for N2 , NO, O2 and corresponding ions’. J. Quant.
Spectrosc. Radiat. Transf., 5, 369–389.
G RISENTI , R. E., W. S CHÖLLKOPF, J. P. T OENNIES, G. C. H EGERFELDT, T. KÖHLER and
M. S TOLL: 2000. ‘Determination of the bond length and binding energy of the helium dimer
by diffraction from a transmission grating’. Phys. Rev. Lett., 85, 2284–2287.
H ERSCHBACH , D. R., Y. T. L EE and J. C. P OLANYI: 1986. ‘The N OBEL prize in chemistry:
for their contributions concerning the dynamics of chemical elementary processes’, Stockholm.
[Link]
H ERZBERG , G.: 1989. Molecular Spectra and Molecular Structure, vol. I. Diatomic Molecules.
Malabar, Florida: Krieger Publishing Company, 660 pages.
H OUGEN , J. T.: 2001. ‘The calculation of rotational energy levels and rotational line intensities
in diatomic molecules (version 1.1)’, NIST. [Link] ac-
cessed: 9 Jan 2014.
H UBER , K.-P. and G. H ERZBERG: 1979. Constants of Diatomic Molecules. New York: Van Nos-
trand Reinhold.
ISO 21348: 2007. ‘Space environment (natural and artificial) – Process for determining solar irra-
diances’. International Organization for Standardization, Geneva, Switzerland.
J ENOUVRIER , A., M. F. M ERIENNE, B. C OQUART, M. C ARLEER, S. FALLY, A. C. VANDAELE,
C. H ERMANS and R. C OLIN: 1999. ‘Fourier transform spectroscopy of the O2 Herzberg
bands – I. Rotational analysis’. J. Mol. Spectrosc., 198, 136–162.
J UARROS , E., K. K IRBY and R. C ÔTE: 2006. ‘Laser-assisted ultracold lithium-hydride molecule
formation: stimulated versus spontaneous emission’. J. Phys. B, At. Mol. Phys., 39, S965–S979.
K LAUS , T., S. P. B ELOV and G. W INNEWISSER: 1998. ‘Precise measurement of the pure rota-
tional submillimeter-wave spectrum of HCl and DCl in their υ = 0, 1 states’. J. Mol. Spectrosc.,
187, 109–117.
K LING , M. F. et al.: 2006. ‘Sub-femtosecond control of electron localization in molecular disso-
ciation’. Science, 312, 246–248.
References 229

K REMER , M. et al.: 2009. ‘Electron localization in molecular fragmentation of H2 by carrier-

envelope phase stabilized laser pulses’. Phys. Rev. Lett., 103, 213003.
L E ROY , R. J.: 1970. ‘Molecular constants and internuclear potential of ground-state molecular
iodine’. J. Chem. Phys., 52, 2683–2689.
L I , X. Z. and J. PALDUS: 2006. ‘Singlet-triplet separation in BN and C2 : Simple yet exceptional
systems for advanced correlated methods’. Chem. Phys. Lett., 431, 179–184.
L OCKWOOD , G. J. and E. E VERHART: 1962. ‘Resonant electron capture in violent proton-
hydrogen atom collisions’. Phys. Rev., 125, 567–572.
L OVAS , F. J., E. T IEMANN, J. S. C OURSEY, S. A. KOTOCHIGOVA, J. C HANG, K. O LSEN and R.
A. D RAGOSET: 2005. ‘Diatomic spectral database (version 2.1)’, NIST. [Link]
Diatomic, accessed: 9 Jan 2014.
M ANTZ , A. W., J. K. G. WATSON, K. N. R AO, D. L. A LBRITTON, A. L. S CHMELTEKOPF and
R. N. Z ARE: 1971. ‘Rydberg-Klein-Rees potential for the X 1 Σ + state of the CO molecule’. J.
Mol. Spectrosc., 39, 180–184.
M OLPRO: 2012. ‘Molpro quantum chemistry package’, H.-J. Werner, Universität Stuttgart, Ger-
many, and P. J. Knowles, Cardiff University, UK. [Link] accessed: 9 Jan 2014.
PARTRIDGE , H. and S. R. L ANGHOFF: 1981. ‘Theoretical treatment of the X 1 Σ + , A 1 Σ + , and
B 1 Π states of LiH’. J. Chem. Phys., 74, 2361–2371.
PAULING , L.: 1931. ‘The nature of the chemical bond. . . ’ J. Am. Chem. Soc., 53, 1367–1400.
R AM , R. S., M. D ULICK, B. G UO, K. Q. Z HANG and P. F. B ERNATH: 1997. ‘Fourier transform
infrared emission spectroscopy of NaCl and KCl’. J. Mol. Spectrosc., 183, 360–373.
ROTHMAN , L. S. et al.: 2009. ‘The HITRAN 2008 molecular spectroscopic database’. J. Quant.
Spectrosc. Radiat. Transf., 110, 533–572.
S HARP , T. E.: 1971. ‘Potential-energy curves for molecular hydrogen and its ions’. At. Data, 2,
119–169.
S TANTON , J. F.: 1999. ‘A refined estimate of the bond length of methane’. Mol. Phys., 97, 841–
845.
TANG , K. T., J. P. T OENNIES and C. L. Y IU: 1995. ‘Accurate analytical He-He van-der-Waals
potential based on perturbation-theory’. Phys. Rev. Lett., 74, 1546–1549.
T URBOMOLE: 2010. ‘Quantum chemistry (QC) program package’, COSMOlogic GmbH & Co.
KG, Leverkusen, Germany. [Link]
html, accessed: 9 Jan 2014.
Y IANNOPOULOU , A., G. H. J EUNG, S. J. PARK, H. S. L EE and Y. S. L EE: 1999. ‘Undulations
of the potential-energy curves for highly excited electronic states in diatomic molecules related
to the atomic orbital undulations’. Phys. Rev. A, 59, 1178–1186.
Z EWAIL , A. H.: 1999. ‘The N OBEL prize in chemistry: for his studies of the transition states of
chemical reactions using femtosecond spectroscopy’, Stockholm. [Link]
prizes/chemistry/laureates/1999/.