Elementary Wave Mechanics - Heitler

SECOND
ll>ITIO:\
ELEMENTARY
WAVE
MECHANICS
WITH APPLICATIONS TO
QUANTUM CHEMISTRY
SECOND EDITION
W.
HEITLER
The
first
edition of this
hook was
published in 1945. This second

edition has been revised and reset,
and is about 50 per cent, longer

than the first edition. In particular
the sections dealing with the applications to chemistry have
been
enlarged, as the book was designed
In
for chemists.
words
'With
this respect the
applications
to
Quantum Chemistry' have been

added to the
title-page.
2l/NET
1
,N U.K
ON^
ELEMENTARY
WAVE MECHANICS
ELEMENTARY
WAVE MECHANICS
WITH APPLICATIONS TO
QUANTUM CHEMISTRY
BY
W.
HEITLER
PROFESSOR OF THEORETICAL PHYSICS IN THE
UNIVERSITY OF ZURICH
SECOND EDITION
OXFORD
AT THE CLARENDON PRESS
$r
Oxford University Press,
QLASflOW
Amen
House, London E.G.
NEW YORK TORONTO MELBOURNE WELLINGTON
PREFACE TO THE SECOND EDITION
BOMBAY CALCUTTA MADRAS KARACHI KUALA LUMPUR

CAPE TOWN IBADAN NAIROBI ACCRA
As
this little book was primarily designed for the use of

chemists and other non-mathematical readers I have added
a section on diatomic molecules and above all much ex-
tended the chapters on the chemical bond. These are in

the same elementary style as the rest of the book. I hope
that these additions (which seem compatible with the idea
of a small pocket book) will afford a good illustration of the
methods of wave mechanics as well as its usefulness for
chemical problems. To make the general theory more
complete a section on the time-dependent wave equation
has also been added.
W.H.
Zurich
January 1956
FIRST EDITION
945
REPRINTED I946 (TWICE), ALSO LITHOGRAPHICALLY

IN GREAT BRITAIN AT THE UNIVERSITY PRESS, OXFORD,
FROM SHEETS OF THE THIRD IMPRESSION 1947, I948, 1950
SECOND EDITION 1956
REPRINTED LITHOGRAPHICALLY I958, I961
CONTENTS
I.
EXPERIMENTAL BASIS OF QUANTUM

MECHANICS
1.
2.
3.
II.
and electron diffraction

wave and particle properties
Reconciliation of wave and particle views
states
1.
The
2.
Discrete
3.
free electron
17
quantum states
The Schrbdinger wave equation
The time -dependent wave equation
21
24
27
THE HYDROGEN ATOM

1.
The ground
2.
Excited states
32
37
3.
39
4.
Normalization and linear combination of wave
state
states
functions
IV.
DERIVATION OF THE WAVE EQUATION
4.
III.
Quantum
Relations between
43
ANGULAR MOMENTUM. ZEEMAN-EFFECT.

SPIN
2.
Sharp and unsharp quantities

Angular momentum of s and p states
3.
Zeeman-effeet
4.
5.
The
electron spin
6.
Two
electrons with spin
1.
V.
46
50
54
56
59
64
states, directional quantization
PROBLEM OF TWO ELECTRONS
5.
The wave equation for two electrons

Solution of the wave equation for two
Exchange degeneracy
The Pauli exclusion principle
The spin wave function
6.
General formulation of the Pauli principle
1.
2.
3.
4.
electrons
68
70
74
76
79
82
CONTENTS
ii
VI.
PERTURBATION THEORY
1.
2.
3.
VII,
General theory
atom, exchange energy
The orthogonality theorem
He
86
88
93
EXPERIMENTAL BASIS OF
QUANTUM MECHANICS
THE PERIODIC SYSTEM OF ELEMENTS

1.
2.
The electron configuration

The atomic states
97
100
Quantum states and electron diffraction

There are two main groups of experimental phenomena
1.
VIII.
DIATOMIC MOLECULES
1.
2.
3.
4.
IX.
The electronic states

The rotation of molecules
The vibration of molecules
Ortho- and para-molecules
107
which are inconsistent with
115
(i)
1.
The hydrogen molecule

The saturation properties of the chemical bond
(ii)
Spin valency
2.
Crossing of atomic interaction curves. Valency

of carbon
3.
Interaction in diatomic molecules
2.
3.
4.
5.
Interaction of several atoms with one valency

electron
Quantum
states
143
150
chiefly derived
from spectroscopy, shows that an atom cannot

states of continuously varying energy
but only in different discrete states of

energy, symbolized in Fig. 1, and referred to as 'discrete energy levels'.
These levels and the spacing between
exist in
;
s
hv
2
155
them
164
chemical elements, but are identical

Discrete energy
for all atoms of the same element. Fig.
levels.
The change from one level E2 to
Ej)
another Ex is associated with the emission {if E2
or absorption (if E2
Mj) of light, whose frequency v is
determined by the relation
Activation energy, non-localized bonds

Directed valencies
Interaction of atoms with several electrons
170
Binding energies of hydrocarbons
184
INDEX
does not do so
evenly or continuously but in 'quantum' steps;
the fact that a beam of electrons exhibits interference phenomena similar to those of a light wave.
A great wealth of experimental material,

139
XL POLYATOMIC MOLECULES
1.
changes, owing
it
123
VALENCY
1.
namely:
136
(i)
X.
when the internal energy of an atom

to emission or absorption of light,
117
THEORY OF HOMOPOLAR CHEMICAL

BOND
2.
classical physics,
112
179
191
are different for the various
*1
>
<
E E ~
2
hv.
(1)
EXPERIMENTAL BASIS OF QUANTUM MECHANICS

h is a universal constant known as
Planck's quantum of
action whose value is 6-6 X 10-" e rg
sec. The frequency
2
for the
same energy change,
e.g.
to
always the
same. Furthermore, a beam of light
with frequency v such
as may be emitted by a large
number of like atoms all carrying out quantum jumps from
E% to Ex will not have a
continuously varying energy but will
consist of a number
oi quanta' each having an
energy h>. The energy of the
2
lt
Interference of waves
known that light waves when reflected

or
refracted from a regularly spaced
grating interfere
and
form what
2.
whose spacing we can determine the wave-length of the

light using the
formula
nA
where d
crystal
is
Electron diffraction
called a 'diffraction' pattern.

This phenoused, for instance, to prove
the wave nature
oi X-rays. For this purpose
the regular spacing is provided
by the atoms of a crystal, as an
ordinary grating would
be too coarse. The crystal may
be used in the form of a
WaV6S fr
their source
(Kgpass
is
2d sin-,
the distance apart of the rows of atoms in the
and m the angle of
diffraction,
1, 2, 3,...
for
the various rings.

We can now carry out a similar experiment with a beam
of electrons, all having the same velocity, v, and it has been
found that the result is similar. To produce the beam of
may
be used. The electrons

V which
accelerates the electrons to a velocity, v, determined by the
\mv z The electrons then pass through the
relation eV
crystal powder B, etc., as above.
The diffraction pattern so found is very similar to that
produced by X-rays. This experiment shows that waves
are associated with a beam of electrons, formerly considered
as consisting of particles only. It is the purpose of wave
mechanics to show how the properties of an electron regarded as a wave can be reconciled with its properties as
a particle. At the same time it will be seen that the facts
of the discrete energy levels follow from the properly
developed theory of wave mechanics.
electrons a cathode-ray tube
It is well
Fig.
is
beam is thus either lk v or 2hv or Zhv... or

nhv, etc., or in
other words, consists of 1, 2,
3,..., n,... light quanta.
(ii)
pass through an electrical potential difference
menon was
l\u
the
fi?
through
O.
crystalline
powder
A
2)
B and arrive on the screen
Ihe diffraction pattern observed

on
series
G consists of a
of concentric rings
(Debye-Scherrer rings) from
2.
Relations between
wave and
particle properties
To develop the theory of wave mechanics we start from

the experiment described above, showing that a wave is
associated with a beam of electrons. The first question
which naturally arises is: What is the wave-length A found
to be
It can be
tion rings.
measured from the spacing of the diffracA depends on the
The measurements show that
velocity, v, of the electrons. This velocity can be varied

by
varying the potential used to accelerate the electrons.

The wave-length is then found to be inversely proportional to the velocity; A oc 1/v, the slower the electrons the
longer is the wave-length. The proportionality constant
can also be measured in this way. It turns out to be
equal to Planck's constant divided by the mass of the
electron
needed for either v or v^. Unfortunately both v and v$

cannot be determined experimentally. Instead, we shall
establish a further relation on theoretical grounds. It is
plausible to assume that there must be a connexion
between the particle velocity v and the velocity of propagation v^ of the wave. It would be wrong, however, to
m:
n.
n.
mv
or
(2)
or
where mv is called the momentum p. The relation (2) is

due to de Broglie (1924) and marks the starting-point of
wave mechanics.
The velocity, v (henceforward
city
),
is
Fig.
a concept relevant to the electron pictured as a
concept relevant to a wave. The two

concepts are connected by Planck's constant h. This constant was first discovered as giving the connexion between
the energy E and the frequency v of a light quantum,
particle, whilst A is a
E=
Whilst the frequency
is
3.
Group
velocity.
called the 'particle velo-
hv.
a concept relevant to a wave, a
quantum rather resembles the concept of a particle.

Now we have seen that the same constant h also connects
the wave picture with the particle picture for electrons.
A wave is not determined by its wave-length alone but
by the three quantities, wave-length A, frequency v, and
wave velocity v which are connected by the relation
assume straight away that v = v^. The reason is the

following: In our case A was seen to depend strongly on v
and therefore presumably also on v^. Our waves thus
show a strong dispersion, similar to, but much stronger
than, the dispersion of a light wave in a refracting medium,
glass or water, say. In this case one distinguishes between
two velocities of a wave. In addition to the phase velocity
v^ defined above, one can define a so-called group velocity
light
= Vj
In the case of a light wave v^ = c
Av
We
(3)
(velocity of light).
have already found a connexion between A and the

but in order to determine the properties
of the electron waves completely one further relation is
particle velocity v,
(4)
Only in the case where v^

length
is
independent of the wave-
vg , according to (3), identical with v^. Whilst

the phase velocity is the velocity of propagation of a monoA, is
chromatic, infinitely long wave, a closer examination

shows that the group velocity describes the velocity of
travelling of a short wave pulse (Fig. 3). Such a pulse,
often also called wave packet, does not have a well-defined
wave-length but is composed of waves with many different
wave-lengths, centred about the basic wave-length A just

as a pulse of sound (noise) is composed of many different
We
can use this to determine
since
v,
on integration
Each of these monochromatic partial waves of

which the wave packet is composed has a different phase
velocity. It can now be shown that this has the following
result: Supposing the wave packet has the shape of Fig. 3
tones.
at the time
0, then for a short time afterwards

retained and the whole wave packet travels
with the group velocity vg (X Q ), different from v^{X Q ). (In
(a)
or
(6)
the shape
is
course of time the wave packet will, however, alter its

shape.) For a simple example the relation (4) will be
derived and demonstrated in Chapter II, section 4. Of
course, in the case of light pulses
v^
and vg are almost
where dispersion
is
small,
equal. (In vacuo they are even exactly
Now
waves we have seen that

the particle velocity v depends on the wave-length very
strongly, vcc 1/A, showing a very big dispersion. Accordingly, v$ and v are very different. If we wish to equate
g
the particle velocity v to the velocity of propagation of the
in the case of electron
waves we have to decide whether to assume v

v
= vg
Since a particle resembles a small
= vj
=v
v =
dv
dv
h dv
d{ljX)
d{mv/h)
mdv'
to
K2
=E
E,
.
'
(7)
light
quantum with frequency
v,
we now
see that the
same
relation prevails between the energy of the electron considered as a particle and the frequency of the electron
hv is thus universal, and this
waves. The relation E
may
consistency
be regarded as a confirmation of our
choice v
= uf
We now have a complete connexion between the particle

wave
beam
of electrons (velocity, energy) and
its
properties (wave-length, frequency):
Wave
Particle view
Velocity =* v
Energy
(momentum p
mv)
concept
mv
11
B
mv* =
rr;
h
h2
.
Jmy*
We
This would be so at any rate, if we picture a 'particle'

as a wave packet of small extension, which, as will be
seen later, is correct. Our choice v
vg will be seen to
be confirmed by all experimental facts.
We thus assume that v
v or
g
v dv
it
(5)
which is the energy E of a particle of mass m and velocity

v. Remembering that hv = E was also the energy of a
wave packet
rather than an infinitely long monochromatic wave,

would seem more likely that
v
or
h J
hv as hnv
fmv*2
or
properties of a
equal.)
= to
(6)
can also express the phase velocity by the particle

vX
$v, but this relation is not of great
importance. In fact the phase velocity hardly has a physical meaning. Like the frequency with which it is connected
by vX
v$, it is not a measurable quantity (compare footnote). This is also illustrated by the following fact which
velocity v^
t In (7) an additive constant of integration has been omitted. This

would mean adding a constant to the energy; this has no physical
meaning. Accordingly, the frequency of the electron wave is only

determined to within an additive constant (in contrast to the wavelength A) and is not a measurable quantity.
we mention without proof. If we were to treat our electron

waves according to the principles of special relativity,
the phase velocity would turn out to be not v/2 but c 2jv.
It would thus be larger than the velocity of light c, but
no velocity larger than c can be measured. The phase
velocity has therefore a purely mathematical significance.
This, of course, is in contrast to light waves where both
the frequency v and the velocity c have a well-defined
meaning.
experiment. We carry out the same diffraction experiment

(Fig. 2) with single electrons, letting the individual electrons pass one after another through the crystal powder B
and record their arrival on a scintillation screen 0. If each
individual electron were in fact a complete wave we should
3.
wave and particle views

We have seen that a beam of electrons must be described
Reconciliation of
'partly' as consisting of a number of individual particles

and partly as a wave. As yet the two concepts stand quite
apart and seem irreconcilable. Speculations as to what the
'medium' of the wave is, have proved fruitless.
We must now seek a better and deeper understanding
of the mutual relationship between the wave and particle
concepts. For this purpose we consider the intensity of the
electron beam. It is well known that the intensity of a wave
at each point of space
wave amplitvde
>p
say.
is
proportional to the square of the
We may regard
ifs*
as a measure of
the intensity of the electron wave. $ will in general be a

function of space and time. On the other hand, the local
intensity of the beam is equal to the number of particles
per cm. 3 For an intense beam, as used above, we therefore

equate 2 to the number of electrons per cm. 3 in the beam.
The apparent
contradiction between the two pictures becomes now very striking if we consider a single electron.
What can the meaning of 2 be then? How can we
conceive a particle, such as exists in an atom, to be
associated with or described as a wave ? Is the electron
a particle and a wave at the same time !
The answer to these questions can be given by an
i/
expect that each time an electron arrives on the screen

the whole diffraction pattern with all the rings would
appear simultaneously which is hardly conceivable, as
the rings have macroscopic extension! This is not indeed
what happens. Instead, we observe that the individual
electrons arrive on the screen at individual points scattered
over the whole screen. However, when a number of electrons have arrived, we find that the distribution of the
points is not uniform, they lie preferably at such places
where in the former experiment using an intense beam
of electrons the maxima of the diffraction pattern would
lie. No electrons, or very few, are found to arrive at the
minima of the diffraction pattern (Fig. 4: the left-hand
diagram shows ^r 2 on the plane of the screen, the righthand graph shows ^ 2 for a radial cross-section through
the screen.) In course of time when a very large number
of electrons have arrived the rings are built up completely,
as in the former experiment with an intense beam.
It is evidently not possible to predict where any one
individual electron would appear on the screen (although
it is possible to say where an electron would not appear,
namely, between the rings). This leads us to interpret the
z
intensity
(on the screen, plotted in Fig. 4) as the
tfr
probability of a single electron hitting the screen at a certain

point. Of course, if a large number of electrons is used,
each will be at a certain point with a probability 2 and

are led back to the former interpretation of 2 as the
intensity or average density of particles. More generally,
function of space, will be the probability for the
i/r
we
âsa
t/r
10
electron to be present at this particular point of space (at
the time at which the
wave function
probability interpretation of
to Born (1926).
is
considered).
for single electrons is
The
due
by numerous experiments, of which the above
11
diffraction
only one example. The necessity of reconciling these two 'natures' leads directly to the probability
interpretation of 2 and to the fundamental principle of
experiment
is
t/f
the uncertainty of physical quantities.
This principle of uncertainty will also hold for other

physical quantities such as, for instance, the angular
momentum,
etc.
We
cannot always attach sharp values
to a given physical quantity; if a quantity does not
have
is given by the square of the wave

amplitude ^ 2 which, naturally, is a function of time and
a sharp value, we can only give the probability for its

having certain values. On the other hand, there are of
course physical quantities, which may have a sharp value
in certain cases. An example is the wave-length in the
hjp, it
diffraction experiment discussed above. Since A
follows that p (or v) also has a sharp value (it is measured
by the applied potential). In addition, the energy for such
a beam of electrons is a function of the velocity only, so
that it also has a sharp value. We shall see that it depends
on the external conditions, especially on the experiment
performed, which physical quantities have sharp values
and which have not. In the diffraction experiment the
space.
situation
Fio.
4.
Probability distribution.
now a very important and far-reaching conIn contrast to classical mechanics we cannot
predict exactly which way the electron will go. We can only
tell that the probability for finding the electron at a certain
place, at a certain time, is, say, 50 or 3 per cent., but we
cannot say where the electron will be found precisely.
This has
sequence.
This probability
,
It follows
from the above that no
definite value
is:
can be
attached to the position of the electron at any time. Its

position may be anywhere within the extension of the
wave by which it is described. We say in such a case that
the position of the electron is 'uncertain' or 'not sharp'
to distinguish it from something which has a 'sharp' value,
such as, for instance, its velocity in the above diffraction
experiment. The fact that not always can sliarp values be
attached to every physical quantity (in our case to the position) is a fundamental feature of quantum mechanics or wave
mechanics. It is, as we have seen, due to the double nature
of the electron, as a particle and as a wave, as it is exhibited
Unsharp
Sharp
p
v =
Er. m
z
p,
*
x, y, z.
2m
Hence there are physical quantities with sharp values

and others for which we can only give probability values.
Another example of a case where the energy is sharp and
the position is unsharp is provided by an ordinary atom
which shows discrete energy values Ex E2 E2 etc. We
,
>
are able to state the energy value of the atom in a particular instance, yet the position of the electron is not
sharp, although a 'probability distribution' for the various
12
positions of the electron can be given. It will be derived

in Chapter III.
A small wave
packet also has sharp and unsharp quanLet us contrast a plane monochromatic wave, which extends indefinitely and has a fixed
velocity, with a very small wave packet. Such a small
wave packet has a small but finite extension Ax, say, but
as we shall see, the velocity is not defined.
A wave packet can, in fact, be regarded as being built
up by a superposition of monochromatic waves in much
the same way as a short pulse of sound is built up by
a spectrum of monochromatic tones. Therefore a wave
packet consists of waves of several wave-lengths, A varying,
tities (Fig. 3 (a) or (6)).
between A +AA and A

AA. The momentum varies
accordingly between p +Ap and p Ap;p
A/A
Hence
the velocity or momentum cannot be sharp, in contrast to a
say,
monochromatic wave which has infinite extension but sharp

wave-length and momentum. For a wave packet the
energy E
p zj2m is also unsharp. We call A^> the uncertainty of the momentum. The smaller the spatial extension, Ax, of the wave packet, the wider is the range of
wave-lengths required to build up the wave packet, and
the larger, therefore, is the uncertainty of the momentum
Now Ax
a measure for the uncertainty of the

^ is different from zero only within
the extension of the wave packet. Aa; may be very small,
in which case the position is nearly sharp. A monochromatic wave has infinite extension Az
oo, but sharp
momentum Ap
0. Thus we find the following theorem.
The more accurately the position of a particle is defined
the less accurate is its momentum or velocity (small wave
Ap.f
position,
is
since
packet),
t
and the more accurately the momentum of the
The group velocity

Av = Apjm.
within
vg (p. 5)
is
the average of the various velocities

particle is defined the less accurate is the position
chromatic wave). This
13
(mono-
Heisenberg's uncertainty relation
is
(1927).
These considerations can be carried out quantitatively,

calculating the range of wave-lengths of
build
up a packet of
AxAu
size
Ax. It
or
is
AxAp
%irm
where h
is
Ap
required to
found that
(8)
77
again Planck's constant. The formula demon-
strates the uncertainty principle directly; the product of

is a constant proportional to h. A monochromatic wave is a limiting case Ap
oo, a very
0, Aa:
small wave packet ('particle' with fixed position) has
Ax
oo. In Chapter IV, section 1 we shall find a
0, Ap
mathematical criterion which will enable us to tell which
quantities have a sharp value in a certain instance.
Our next task will be to derive a general equation for the
propagation of the wave amplitude or wave function
which is to replace Newton's equation of motion for a
particle. Before doing so, some further observations about
the uncertainty principle must be made.
The right-hand side of (8) is also proportional to 1/m (if
v is used, not p). It follows, then, that for a very heavy
particle for which hjm is very small, the product of the
two uncertainties AxAu becomes very small too. In this
case both the position and the velocity are practically
sharp, f Especially for macroscopical bodies with
so
large that hjm 0, all uncertainties vanish, and all
quantities have sharp values. This is the limiting case of
classical mechanics. Classical mechanics holds for heavy
bodies; the 'uncertainties' are a peculiarity of quantum
the uncertainties
*jt
t The momentum, though, will still be unsharp, but for a heavy body
a moderately small momentum means vanishing velocity.
14
mechanics which applies to light
EXPERIMENTAL BASIS OP QUANTUM MECHANICS
particles, especially elec-
trons, protons, etc.
A further question which will present itself if the foregoing principles are to be logical and not self-contradictory
is the following: we have stated that in the diffraction
experiment the momentum has a sharp value (it was
determined by experiment) whilst the position was uncerBut what happens when the electron has arrived on
the screen and the scintillation is observed at a particular
tain.
point?
electron
Then we obviously do know the position of the

and we could hardly admit that the latter is still
uncertain.
We may
ask, for example,
what would happen
if
we
place a second scintillation screen immediately behind the

first (making the first screen so thin that the electron can
pass through) and observe the position once more on the

second screen. If there is any meaning at
all in saying 'the electron was found at a
*
certain point then surely
Fig.
5.
Two
subsequent observations of
position of
electron on two
scintillation
screens.
we must
find it
again at the corresponding point on the

second screen (Fig. 5), otherwise the first
statement of position would be valueless.
On the other hand, what has now become
of the wave function after the position is
observed ? z is the probability distribution
of the position and this extends over a
\ft
large area {Fig. 4). If 4s were to remain

unchanged there would be no reason whatsoever why the
electron should appear at the same point on the second
screen. It might jump to another point with an equally
large probability, perhaps a few centimetres away! But
not what happens.

The answer to this question
this is
reaching.
We
is
as simple as
it is far-
state the answer without being able to
15
develop the theory in this book up to a point where it

would appear as part of the general structure of quantum
mechanics.
The moment the electron has been observed at a particular point on the screen it has, indeed, a sharp position.
Then the electron is to be described by a small wave packet
which has a sharp position but an uncertain momentum.
Accordingly, the wave function
changes from a monochromatic wave, suddenly and decisively, into a small wave
packet. The change is effected by the measurement of the
position on the screen. By this measurement we force the
electron into a state with sharp position. (A subsequent
second measurement of position, following immediately
after the first, yields then the same result.) At the same
time we destroy the knowledge we had before of the momentum (gained by measurement!). If, afterwards, the
momentum were measured again we should find all sorts
of values for p, each with a certain probability, and these
tfi
values would be very different from the

electron
had when
it left
momentum
the
the cathode-ray tube. It depends
on the experiment performed which quantities have sharp
A quantity that is measured is forced to be sharp.

In atomic physics a measurement no longer leaves the
measured object uninfluenced. Two measurements of
the same quantity (p, say) carried out in succession will
not yield the same result if a different quantity (a:, say)
is measured in between them.
This is a very striking
feature of wave mechanics and contrary to classical physics.
The sudden change of the
function shows that the
wave field of an electron is something very different from
the familiar kind of field such as the electric or magnetic
field. The latter is directly observable and measurable, it
never changes suddenly, and a measurement has no influence on it. The wave field of an electron is nothing that
values.
tji
16
observable directly. Its physical meaning is a probability

2
(rather a 'probability amplitude', for the probability is
).
ingredient
essential
of
an
nevertheless
it
is
For that reason
is
tft
the nature of the electron and vital for its description. For
without the tp function no laws of nature could be formulated in the domain of atomic physics.
These considerations were given for the sake of logical
They
not be further needed in this

book. In atomic physics one is nearly always concerned
with cases where the energy is sharp. We shall develop
the theory of wave mechanics mainly for this case.
completeness.
The
philosophically
II

1.
The
free electron
will
A monochromatic
'
reality ' or
minded reader may decide for himself whether

field of an electron (^r function) as part of
only as a product of the human brain, useful
'
travelling in the direction of
the x-axis, with wave-length A and frequency

amplitude tjj given by the equation
he would consider the wave
an objective
wave
tfi
v,
has an
= ABm2irl-vt\
(1)
'
for predicting the results of experiments (it happens, though, that these
predictions always agree with the 'objective* facts). The author does
not wish to influence him in his belief on this point. He only suggests
that the question of what 'objective reality is, be cleared up first.
'
by the corresponding cosine expression, which describes

wave as (1) with a shift of phase. There exist
also standing waves, in which the maxima and -minima do
or
the same
not travel along:
tft
The
sin
= A sin -^ cos 2md.
and cos can here
also be replaced
(2)
by
cos
and
sin
respectively.
The standing waves are more important for describing

the motion of an electron in an atom. In a discrete level
we certainly are not dealing with a wave running to infinity

but it will be seen that the electron behaves as such a
standing wave. Hence we will confine ourselves to waves
depending on time like (2). Furthermore, we shall consider
the space part, A sin %ttxj\, and the time part, cos 27n>(,
separately. We shall see that for the type of problem considered below we need not be concerned with the latter.
5881

Hence we may use either of the two expressions:
18
Uf
the wave-length A
hjmv.
relation A
= A sin -^
A
by the particle velocity v using the

The energy is then
E = \mv* = im(i)
(3)
19
!
.
2-nx
= A cos -^
.
tfi
A
or,
more
Hence the expression

equation (7) becomes
generally,
4,
= Acos(^+h\
{2irjX}
occurring in the above
wave
(4)
SnhnE
where 8 is an arbitrary phase between and 2it.
In wave mechanics one often has to deal also with complex waves containing the imaginary unit i ^J(-l).
Instead of (3) we may, for instance, also write
iff
This
is
also
shall see
=A
= Ai
cos- -|-tsin
a wave with wave-length
(section 4)
a ways
e
_ 2nivt
^ 2m4 _
The waves
(5)
gin
(4)
or (5) twice
we find
QV =
dx 2
^ W
(9)
Every solution of (9) is to be multiplied by a factor

exp( 2-rrivt) (using the complex expression (6)); v is
E occurring in (9):
factor
time-dependent
the
In the following
Ejfi.
interest.
will not be of much
So far the wave equation (9) describes only waves
directly connected with the energy

2ttv
extending in one dimension, namely, in the ^-direction.
satisfy the equation
dx 2
A.
instead of sin or cos. Differentiating
thus obtain a wave equation in which the energy E

of the particle occurs. For convenience we introduce
H. Hence the wave equation becomes
instead of h, hj2ir
We
,2ir&r/A
In particular we
that the time-dependent factor is
(8)
er-
Naturally,
o.
(7)
is
the basis
We now wish to relate (7) to the motion of an electron

considered as a particle. In order to do this we express
will,
in general, depend on
all
three co-
ordinates x t y, z. If we had, for instance, a wave extending

in the ^-direction d 2 tfjjdx 2 would have to be replaced by
dtyjdy 2
This is a characteristic 'wave equation' and

of the whole theory.
tft
d 2ijtjdx 2
It
is
is
not
difficult to see
to be replaced
that in the general case
by
S^f
sty
8x*~t~
dy 2
aty
dz*'
20
For
this expression
we
use the short notation

equation is

Hence
8]
where
motion of a particle. Whilst Newton's equation of motion

allows one to calculate the orbit of a particle for any time
accurately (in any given circumstances), the wave equation only allows one to calculate t/>, i.e. the probability for
finding the particle at a certain position. This replacement
of accurate and predictable orbits by uncertain values and
The wave equation (10) describes monochromatic waves

only, with a particular given frequency v or energy E.
from classical physics to wave mechanics. It is due, as we

have seen, to the double nature of electrons, as particles
and waves.
the general
wave
^.
(10)
probability distributions
occurs in (10) directly and also in the time-dependent

(6). For all solutions of (10) the energy has therefore a sharp value. Naturally, there may also be waves
quantum
is
the chief step in the transition
factor
2.
which E is not sharp (for instance wave packets) and

whose wave function will not then be proportional to a
factor (6) with one given frequency v, but will depend
on time in a more complicated way, allowing many fre-
From the wave equation (10) we can draw a most important conclusion, namely, the existence of discrete energy
levels. We have stated in Chapter I that an atom can
only exist in a number of discrete energy states. We now
derive this important result for a very simple case. We
for
quencies,
i.e.
different energies, to occur.
In order to
form a wave packet, for instance, we must take several

different solutions of (10) with different E's, multiply
each by its appropriate time-factor, and take a linear

combination of them. (Similar to the representation of
by monochromatic waves.) Such a wave

no longer a solution of (10) but is a solution of a
a sound pulse
packet
is
more general wave equation, the so-called time-dependent

wave equation. This will be derived and discussed in
section 4. Otherwise, we shall throughout this book only
be interested in cases for which E has a sharp value, i.e.
monochromatic waves. In this case (10) is the wave
equation always to be used. The time-factor is then
always (6) and of no interest. It will henceforth be
in
omitted.
The wave equation

(17),
(10)
derived in section
and the more general equation
3,
replace Newton's equation of
Discrete
states
moving along the s-axis, between two

which are perfect mirrors for the electron.
Considered from the wave point of view this is a case
where we have to deal with standing waves. It is similar
to the vibration of a violin string fixed at two end-points.
Since the electron cannot pass through the walls the amplitude
must be zero outside the walls, and therefore at
consider an electron
reflecting walls
each wall itself. The amplitude, $, is therefore zero at

x
and L, say. These are new additional conditions
which we call 'boundary conditions'. The standing waves
must therefore have a node at each wall. Hence it follows
that not all wave-lengths are possible, but only those, as
2L or
we see from Fig. 6, which have wave-length A
2/2, 2L/3, 2L/4, etc. This, for instance, is the condition
2L gives the
determining the tone of a violin string. A
ground tone and the other wave-lengths A

give the different overtones.
=
= 2L/2, 2L/3,...
22
In mathematical form this

wave equation
is
The
two forms
derived as follows:
is
solution of the
either of the
Por the purpose of showing the discreteness of the energy

levels
we introduce instead of A the energy
= AA sin 2^ x
2
= AA cos
x
**
or
2 [mXj
A
(11)
Then, inserting for A the possible values of A n from
have
23
(13),
we
(15)
Therefore only certain values of E, namely, Eni are possible

and these are the only ones allowed. The lowest energy is
L
Fig. 6. Discrete
wave functions
Ex =
between two walk).
(electron
The
cosine solution does not satisfy the boundary condition ift

at x
0, and must therefore be discarded.
The
sine solution satisfies this condition.
boundary condition
Bin -r-L
is
for certain values of A,
and
0,
this
The second
is
only true
namely:
=
where n
is
fin,
(12)
a whole number.
The latter condition gives the value

!
An
where n
2L
>
\*
1, 2, 3, 4,...,
2L,
etc.
A2
There
of A straight
= -
is
an
infinite
away as
The amplitudes
= A n sm~ x.
A n may, of course,
(16)
1),
and the higher energies are 4, 9,... times tins value. There
are no energy values in between these values. The energy
values En are usually called eigenvalues and
n eigenifi
functions.
If these discrete energy levels are plotted (Fig. 1) it is
seen that the spacing increases as 2n-\-l. It is due to the
boundary conditions that the energy levels are discrete.
discrete
quantum
states
can
be understood from the wave-mechanieal viewpoint.

In an atom, however, the position is not quite the same
as in our example, since in an atom there are no actual
boundaries. On the other hand, the electron is moving in
e*fr shown
the field of an electric potential given by V
(13)
number
of solutions, since for each value of n we obtain

a wavelength, say, A n and a corresponding solution
&i
(n
2
Our example shows how the
2tt
Li
8mi
(U)
be different for each
n.
in Fig. 7.
Now
if
the electron
is
bound to the
nucleus,
its
energy must be negative. On the other hand, its kinetic

energy is always positive. Therefore, if we consider the
motion of the electron classically, it can only move up to
a maximum value of r shown by the curves of Fig. 7; the
kinetic energy is zero when the electron touches the curves.
This is indeed a condition similar to the boundary conditions
considered above.
24
In wave mechanics the boundary condition is slightly

different because the curves of Fig. 7 are not really
impenetrable walls with an infinitely high potential as in the
example of Fig. 6. Nevertheless, if the electron has negative

energy and is therefore bound, it cannot leave the atom.
This means that
zero at large
>r
this condition is quite suffi-
cient
Electron in Coulomb
must be
distances and
t/t
to
lead
to
discrete
energy levels of the atom (see

Chapter HI, especially the
end of section 1) but, of
course, the spacing of the
levels is quite different
potential.
from
that in the above example.

The fact that, in classical theory, the electron could not
move beyond the curves of Fig. 7, finds its expression in
25
and the energy E
is not moving in an external field, V =
in the
wave equation
3.
The
Schrftdinger
wave equation
The wave equation derived in section 1 applies still to

electrons moving freely in space but not subject to
any
external forces. If an electron moves under the influence
of an electric field, say, it is deflected and accelerated
and
consequently we must expect that in the wave picture
also the waves are somehow deflected. We must
therefore
generalize our wave equation to take account of
such
external forces. Let us consider an electrical potential
and
try to obtain a wave equation which shall be
general for
such cases. The equation describing the free motion of an
electron makes no distinction between kinetic
energy,
= E
The energy levels would all be the same as for a free

also the same, whatever
electron and the wave function
ifj
might be. This certainly cannot be true.

We therefore assume that E in the wave equation for the
free electron means, in fact, the kinetic energy, i.e. EV,
if the electron moves in a field of potential V. The general
wave equation then becomes
the potential
vv+f^-F)</. =
total energy, E.
where
V is the potential
The
total energy is
energy.
If,
E'= T+V,
therefore, the electron
(17)
o.
wave equation found by SchrOdinger (1926). It describes quite generally the motion of an
electron under the influence of any electrical potential F.
Our procedure, namely the replacing of E by T E V
is also supported by the following purely formal considera(17) is
the celebrated
which will also be useful later on.

Let us rewrite the wave equation (10) for a
in the form
tions,
^=-^
V2*=
free particle
a2
2m \dx 2
and
)*'
(18)
let us compare this equation with the corresponding

expression for
in classical theory:
T,
and
energy and
wave mechanics
in that the ^-function decreases rapidly

to zero for values of r larger than the classical
maximum.
(10) is equal to the total
the kinetic energy at the same time. But if a potential V

is present, the question arises whether we should then
insert in the wave equation, instead of E the total energy
V. If we
E T+V, or the kinetic energy, T
should decide for the former alternative the wave equation
being then the total energy.
would remain unaltered,
m
E = j(v*x +vl+v%).

Introducing mvx px as the momentum, we have
26

the
wave equation
Eif,
We
see that a great similarity exists between the two

equations. In order to pass from the classical expression for
E, (19), to the wave equation, (18), we have to substitute
for
pi the 'operator' W^ftx 2 ), and let this operator act

on the wave function ^. Or, in other words, we have to
replace
by the
z itself
operator:
*7B-
(20)
This replacement seems mysterious at this stage, but is

part of the general theory of quantum mechanics, where
it is seen to have a deep meaning which
cannot be fully
explained in this book.
If we now consider the case of an electron moving in
a potential F, the classical equation for the energy is
T = E - v = ^*+vlti)'
If
we
again replace each
p x by
h 8
the operator ?
and let
% dx
operate on a wave function t/,, we obtain Tift
(E V)ifi,
which is just the wave equation (17). Ultimately, it is of
course experiment which has decided in favour of the wave
equation (17) and therefore also the formal replacement (20).

Thus the wave equation (17) bears a close relation to the
classical equation for the energy. If we denote the classical
energy function expressed as a function of the coordinates
E=
H[x
...
...,p x ,...),
px ,.)
also called
H = H ix
= JEty,
Hamiltonian\
= (px +pl+p*) + V(x,y,z),

{21 a)
hd
\
(216)
Again, all the solutions of the wave equation (17) are

those with sharp energy. This does not mean, and this is
in contrast to the case of a free electron, that the momentum is also sharp. For now the energy E is the sum of the
T and the potential energy V. The latter

coordinates of the electron which certainly
on
the
depends
kinetic energy
E 7*4-7 is to be sharp, T
Chapter
III). This is connected
or p
classical
theory
the momentum of
that
in
fact
the
with
the electron does not remain constant in course of time,
except for a free electron (V
0). This finds its expression in quantum theory (as we cannot explain here) in the
fact that when the energy is sharp only such quantities
can also be sharp (but need not necessarily be so) which
classically remain constant in course of time. An example
(angular momentum) is considered in Chapter IV. For
the various values of p a probability distribution, analogous to the a for the position, exists, but this will not be
needed in this book.
cannot be sharp
(see
/f
it
and momenta by H(x
is
are not sharp. Therefore, i
h 8
27
4.
The time -dependent wave equation
The wave equation (17) only applies to cases where the

energy is sharp. We have already stated that this is not
the most general case and that in fact a time-dependent
wave equation exists, which also describes running waves
and has solutions for which E is not sharp. Also, even if
E is sharp, the complete wave function is not merely the
solution of (17), tft(x,y, z), but this is to be multiplied by a
time-dependent factor like exp(iEtjh). Now we seek a
more general equation which does not contain the energy
E explicitly but which reduces to (17), multiplied by a
28
time-factor, as a special case for sharp E. In general, the

solution of such a wave equation will be a function
^(#, y,
z, t)
of the coordinates as well as of the time.
a wave equation which such a function
is
to satisfy
solution
we remember that a
multiplied by a time-factor
easily settled if
must be
(17)
we
shall
have a term
correct factors in (17)
In order to obtain the

we must then write the wave
equation as
vyvtIndeed,
2im
dtjt
,</r
field strength.
Previously
we had
interpreted
as the probability for
is
a certain point. Now when
complex it need hardly be proved that this probability
2
Most of the solutions occurring in this book,
will be
however, will be real and in this case the probability is 2
The solution of (22) which describes a running wave for
finding the electron at
I^!r|
iff
tf/
is
obviously
tr
= Ae
2lTixlX - iEilJi
with
d^jdt.
because
remarkable that in the wave equation

will there% occurs. f In general
fore be complex and this again underlines the fact that ^f
cannot be a measurable field quantity like the electrical
special
Since in (17)
and not 2 occurs we see that only the
first derivative can occur. Thus instead of the term
E,
(17)
It is
must
exp( iBtjh).
would not lead back to
(22) the imaginary unit
with respect to the time. The question is now whether

this is the first derivative dtff/dt or perhaps the second
derivative a^/Si 2 a s in the case of the coordinates (0 2^/&c 2
is
latter
derivative of cos (or sin) does not reproduce this
function.
external conditions under which the electron moves.

Thus it is clear that the term (2mEjfi, 2 )ifi which occurs
in (17) will be replaced by some derivative of ijj(x,...,t)
This
first
Now
only contain the derivatives of ^r with respect to the

coordinates as well as the time but must not contain any
constant like the energy or the wave-length. Such constants occur in special solutions only. On the other hand,
the potential V will occur because this describes the
etc.).
The
cos Et/fi.
the
29
= Ae
-
2viixlX - vC>
(23)
(Again it is not the real form cos 2tt{xjX vt).) (23) is a

wave travelling along the positive #-axis with wave-length
A and frequency v or phase velocity Xv.
A still more general solution is obtained by superposing
several running waves of the type (23) with different frequencies. As a simple example we superpose two waves
0.
we obtain a solution with sharp energy
(22)
E if we put
A +AA
with but slightly different wave-lengths A, say A
and A
A
AA where AA is small compared with A This
example will afford a derivation and illustration of the
group velocity v
The frequencies are vQ -\-Av and v Q Av.
Let the amplitudes of both waves be the same. Using then
the fact that
,
#(*, V,
z, t)
= ${x, y, s)e-^'/
ft
(22')
and then
tfi(x,y,z) satisfies the old equation (17). (22) is

the general time-dependent wave equation. When
solved, it tells us quite generally how the wave function
develops in course of time. For sharp energy
we see
AA
now
that the correct time-factor
is
the exponential and not
A
t
The
sign with which
(22') is irrelevant
used consistently.
+AA
i
occurs in (22) as well as in the time-factor

as long as the same sign is
and a mero convention
the solution
*a 4[e
ib
is
2'"W^+ÂM"'(ri-Av)a_j_
=
It
may
be
a solution of
v
30
= h/2mX
^ *-"^cos277(â;+AvA.
A
show that
to the reader to
left
(22).
and
2ê 2w
e 27rqx/(Ao-iA)--(i' D -4j')fl|
(24)
(24) is really
Now there is a relation between v and A,
therefore,
upon
differentiation,
The right-hand side is the particle velocity v (p. 4, eqn.

(2)) which we had identified with the 'group velocity' vg
(compare Fig. 3). The latter was defined as vg = dvjd(\jX)
and since v = Ejh = p 2/2m7i = h/2mX 2 we obtain
,
d(l/X 2
{hj2m)
= hjmX = v.
d(l/A)
Thus the superposition of two waves with slightly

example for the
group velocity and its identity with the particle velocity.
For the rest of this book the time-dependent wave equation
will not be used any more.
different wave-lengths provides a simple
= __Av,
h
AX
(25)
where we have inserted A for A. Thus the wave function

tft
The
first
ft=
2Ae 2
^x^~ ^coB Avt^ x~t\v
2tt
factor (the exponential) describes
is
(26)
an ordinary
running wave with wave-length A However, this wave is

now 'modulated' by the second factor (the cos). For a
.
\fV\[\MAl\jv
Fio.
8.
Modulated wave.
fixed time t 0, for example, this is a periodic factor and

has maxima at x
A/mA Av t etc. Since A*> is small,
0, x
this is a wave with a very long wave-length h/mX Av. The
whole modulated wave looks like Fig. 8. Now as a function
of t
and of x the modulating factor also describes a
wave travelling along the z-axis. The maxima of the
modulating factor, however, travel with a velocity, given
by the ratio of the factors of x and t in the cosine of
(26). We call this velocity v
g It is given by
% - mX. = v
(27)
31
THE HYDROGEN ATOM

and
differentiating again,
52
Ill
^_
dx 2
THE HYDROGEN ATOM

The ground
In the
atom we have one
electron with a charge
with charge +e. We

take the proton as the centre of our system of coordinates.
e 2/r, and substituting this in
The potential is then V
the wave equation II (17), we get
We
o.
tft
and
~dz~
&c
Consider
2x
=2
^J{
x 2 +y 2 -\-z 2 )
_x
" r"
If
tfs
c 24i
~dr
The simplest
dx
where a
is
or
df
dx
dr 2
r2,
we
and
re-
find
dr^rdr*
2 d$
2m
UMW-+3*-*
,
(2)
is
- e-~
(3)
a constant to be determined.
Substituting
(3)
into
(2),
we have
-<*-",
dr
ds*
dr 2
dr
~ drdx
dr2
differentials together
rdr^dr 2
A(r)
It will
diji
'
solution of (2)
depends only on r but not on the polar angles,

difi
3 2 <A
The wave equation then becomes
^ =(+$+>
^ ^
2
2
*/
r^dr^r 2
fit
dy
dift
rdr
rdr
now
[dx 2
r 3 dr
i/r
dr
dr
rdr
Adding these three second

membering that x 2 -\-y 2 -{-z 2
(i)
attempt to find all the solutions of this

which
would require some mathematical labour.
equation,
shall
give a few simple solutions.
Instead we
The potential e 2/r is spherically symmetrical. There will
which are also spherically symbe a class of solutions
for
which
depends
on r only. To solve (1) let
metrical, i.e.
2
2
=
that
r
remember
us
^/(a: +^ + 22 ) an <l therefore that
shall not
r 3 dr
I dtp
dy 2
V+f(^) =
-l2??X2
?f
For the remaining terms we simply replace x by y and
state
rotating round a central proton
ty
rdr
8 2 <p
1.
33
be noticed that e~ra occurs in each term of (2). There-
fore, dividing
x dif/
rdr'
by
this factor throughout,
*-*-+5?NH
5881
we
obtain
(4)
THE HYDROGEN ATOM
34
Now
THE HYDROGEN ATOM
collecting the terms
which are independent of r in

this equation, namely, the first and third terms,
and then
those proportional to 1/r, namely, the second
and fourth
terms, we see that both must be equal to
zero, because
(4) holds for aU values of r.
2mE
and
r2
of a shell with radius r, and the probability of the electron

2 z
Plotbeing in this shell is therefore proportional to r ij/
2
2ra against r (Fig.
9) we observe that
ting this function T e~
the probability of finding the electron at a distance r from
.
the nucleus
factor r 2 )
35
small at small distances (because of the
is
and at very large distances (because of the factor
(5)
= me*
h*
E=
and hence
If
aW
me*
2m
'W
'
(6)
E and a have the above values, the suggested exponen-
function ftr)
*-~ (3), is a solution. It also satisfies
the boundary conditions described in Chapter
II, section 2,
tial
Fro.
since (3)
becomes zero at large distances from the nucleus!

Actually, this is the solution with the
smallest value of E
(which cannot, however, be shown here), and
therefore
describes the ground state of the
atom.
First
is
we note
bound.
that
E is negative,
E is determined by m, e,
therefore the electron

and h entirely and has,
a fixed value, since these are universal constants!

It can be measured experimentally.
is the 'ionization
energy', that is, the energy required to remove
the electron
from the nucleus. This energy is known from spectroscopy
to be equal to 13-5 eV (electron volts). On
the other hand,
the values of the universal constants m,
e, h are also
therefore,
~E
known. Inserting these values into the expression

E = me4/2 we find also
= 13-5 eV which is in
perfect agreement with the experimental
value.
well
We
-#
now consider the solution 0(r) itself, ^dr gives the

probability of finding the electron in the
volume element
dr. If we introduce polar coordinates,
47rr 2 dr is the volume
9.
Probability distribution of hydrogen ground state.
e~ra ). It has a
the most
nucleus.
maximum
at r
1/a.
The
distance 1/a
The length
constants, m,
e, ft
1/a is also expressed
{equation
(5)),
and
is
by the
universal
called the 'Bohr
radius'. It first occurred in Bohr's theory of the
in which the electron
is
from the
probable distance of the electron
was assumed to move on a
H atom,
circle
with
radius 1/a round the nucleus. It is interesting to note that

e% 2E,
such an electron would have a potential energy
is the kinetic energy
by (6). The amount -fme 4 /2^ 2
giving a total energy E, Although Bohr's theory was the
= E
to give an account of the quantum phenomena in an

atom, it is now replaced by the more consistent and more
comprehensive theory of quantum mechanics (or wave
mechanics). The numerical value of the Bohr radius, as
0-53 X 1CH* cm.
found from (5), is 1/a
We know now that the electron does not quite move on
a circle, as its position is not, for any time, certain, but
first
'
THE HYDROGEN ATOM
36
THE HYDROGEN ATOM
we can say
that the probability for finding

the electron
at a certain distance from the
nucleus is biggest if that
distance is equal to the 'Bohr radius'.
lie probability distribution, Kg.
9, affords an illustration of the uncertainty principle.
While in classical theory
(and in Bohr s theory) the kinetic
energy of an electron
in a Coulomb field is T =
-E, this cannot be
wave mechanics because neither V nor
T is sharp
theless,
this
relation
will
true
Never-
hold
approximately, in the
following sense. From Fig. 9 it
appears that the position
of the electron is uncertain
within limits of the order
of
magnitude A*
1/a. Hence the momentum
is uncertain
to within A
U, The average absolute value of
P
the
momentum must be at least of the same
order of magnitude. In the ground state the
electron will not have more
kmetic energy or momentum than
it need have. Thus
Ha
will actually be the average
momentum. Consequently
tne average kinetic energy will
=
=
be
2m
This
is
2m
2fi
- j&, i.e. the same as in Bohr's theory

interesting feature of our solution
r" is the
just equal to
Another
following:
Although
decreases rapidly for large r
it is
not quite zero, however large r
may be. There is therefore
a small but finite probability for
finding the electron even
at large distances from the
nucleus. In classical theory
this could never happen.
For there the kinetic energy can
never be negative and the particle
can never reach a Joint
where the potential energy would be
larger than its own
total energy E. The
potential energy being V
e 2 /r
and the total energy E given by
(6), we see that r could
not be larger than 2h*Jme *
2/a. Such a
tfs
=-
a*
is
represented by
(see Fig. 7)
by a
particle, according to classical mechanics. This is
so in
wave mechanics. On account
potential barrier
cannot be penetrated
of
its
not
wave nature the
electron has a certain small chance of penetrating such

a potential barrier and of being found outside it (unless the
high as in the example (Chapter II) of

an electron moving between two walls). This is in fact one
of the most conspicuous consequences of wave mechanics.
At first sight this also seems to contradict the principle of
conservation of energy, for the electron does not have the
energy to move beyond the distance 2 fa. Our statement
If we wish to
is to be regarded in the following sense.
observe the position we have to carry out an experiment.
Now such an experiment has a decisive influence on the
barrier
is infinitely
changes its wave function (Chapter I)).

that the measurement also supplies
the necessary energy. If the electron is observed at a large
distance from the proton the energy required to move it
there is supplied by the measuring apparatus.
electron itself
It can
2.
37
(it
now be shown
Excited states
In addition to the state of lowest energy there are also

higher states with energies En
v if we now denote the
energy of the ground state given by (6) by v We shall
not derive all these solutions here but only give the results.
>E
The
energies of the various excited states are
me*
1 _,
(?)
or generally,
where n
If n
is
Fn
me *
2ft
n2
n*
Ev
(7')
1, 2, 3,..., oo.
very large, E is very small, and in the limit n -> oo

This is then the case of a, free electron.
En becomes zero.
THE HYDROGEN ATOM
38
The
THE HYDROGEN ATOM
energies (7) agree also with the experimental facts:
Balmer found that the frequency of the spectral lines

emitted by hydrogen can be represented by the formula
\m 2
-=
n 2J
const.
(8)
39
somewhat similar to those of an electron

moving between two walls (Fig. 6). All n 's decrease,
however, rapidly for large r and satisfy the boundary
conditions, namely, that $ should vanish for large r.
The average distance of the electron from the nucleus is
increasingly larger for the higher n 's than for v
States whose wave functions are spherically symmetrical
solutions is
i/r
i/r
According to Chapter I the frequencies of such spectral

lines are given by
hv
If
we
insert (7) for
= En -Em
are called s states.
There
are,
tf/
however, also solutions
functions are not spherically symmetrical.
En we just obtain
the Balmer formula
(8).
whose wave
These we shall study in section 3.
Apart from the states with negative energy which form
what is called a discrete spectrum En (7') there is also
an infinite variety of solutions with positive energy.
These describe a free electron moving in the field of the
nucleus and correspond to the hyperbolic orbits of classical
theory. The electron comes from and goes to infinity.
Such solutions form a continuous spectrum, every value
E>
3.
is
possible.
states
Although the potential V in the wave equation
Fig. 10.
Wave
The wave functions
functions of hydrogen.
n belonging to the energies

n
symmetrical, i.e. functions of r
only, or more precisely: To each energy value
(7) there is a
spherically symmetrical solution
^n (r). We leave it to the
reader to verify, for instance, that tp
e~m/2 (2 ra) is a
\js
depends on the distance r only, there are solutions which

depend on x, y, z separately. We can show that
(7) are also spherically
solution of (2) belonging to the energy value

2 (7). This
solution, together with
^, and ^3 belonging to the energies
are shown in Fig. 10 (multiplied by suitable
x and
3
m xf(r)
(9)
a solution, if f(r) is suitably determined. Inserting

(9) into the wave equation we have
is
factors).
dx
drdx
r Br'
As we may
'
well expect, the behaviour of these
dx*~ rdr
r &r
r 3 &i
'
r 2 dr 2
THE HYDROGEN ATOM
40
In a similar
way we
THE HYDROGEN ATOM
find
express this fact by saying that we have three energy

levels which happen to lie together. Wave functions of
ty^xydf
dy
the type (11) are called p wave functions, and the energy
state belonging to them a p state.
p state is always
r dr'
^ = zQf_wyfy,xyy8*f
rdrr
dy
41
r Br^~ r
threefold degenerate.
'
r dr 2
d2f/8z z arises from d 2f/dy 2 by changing into

y
the three second derivatives we obtain
z.
Adding up
It
The reason for the occurrence of this degeneracy is clear:

is the spherical symmetry of the potential V. Whenever
we have a solution which is not spherically symmetrical

we can at once obtain further solutions by rotating
the system of coordinates (change x^-y, etc.) as we have
itself,
r dr
r or
dr 2
r dr
done above. This applies to any spherically symmetrical

dr
2 ^~
r Br'
If we insert this expression into the

wave equation we
find that each term is proportional to x.
Thus, dividing
by x throughout,
sy 4 3/ 2m
{E ~ V ^=
8r^rdr + 'W
This
is
(see below).
$x
(10)
an equation for f{r) only, which can be solved

if we
insert for V(r) again the
xf(r)
Coulomb
potential
V = e /r
2
Thus we have found a solution of the type
which
is therefore not spherically

symmetrical
Since, however, there is no distinction
between x, y, z in
the
wave equation, we can say

^v
= yf{r)
are also solutions.
equation
In
and
at once that
^ = zf(r)
(11), f{r) also satisfies
(II)
the same
with the same value of E. Hence there

are three solutions,
$x ^, fc, for each energy value of
(10) with the same function/(r). Any of the three
functions
is a solution for the same
value of the energy.
A case where more than one independent solution exists
for the same energy value is called
degeneracy. In our case
we are dealing with a threefold degeneracy. We may
also
(10),
Quite generally, symmetry properties of the

A case of axial
symmetry will be discussed in Chapter VIII. Only if the
wave function is itself symmetrical, is the state not degenerate. This holds also the other way round: Degeneracies
do not as a rule occur unless they are caused by such
symmetries of the potential V. The only exception is the
potential.
potential give rise to such degeneracies,
case of the
Coulomb
field
/r,
for
which an additional
degeneracy occurs (see below), but this has special reasons

which we cannot discuss here.
So far we have not made use of the special form of the
potential V(r)
e 2 /r. For any spherically symmetrical
potential V(r) p solutions of the kind (11) exist, provided
f(r) satisfies (10). In the case of the Coulomb potential
a solution it is again the one with the lowest energy
value can easily be found. We put
e~ iar with the
same value of a as before, namely (5). By substituting
this
into (10), we easily find that this is a solution, if
tfs
(12)
is the same energy value as that of the second lowest
energy -level given by (7). Now the energy -levels (7) are
This
THE HYDROGEN ATOM
42
those with spherically symmetrical tp functions, i.e. of the

s states. We see now that the lowest p state has the same
energy as the second lowest s state. The level with energy
2 is
therefore fourfold degenerate (one s
functions).
This
is,
THE HYDROGEN ATOM

The p wave functions (9) and (II) may
somewhat as
vanish.
x is
tp
43
be pictured
For large values of r, x and tp v

zero on the t/z-plane and
on the xz-plane.
in Fig. 12.
tft
ift
ff
and three p wave
however, a peculiar property of the
ea
->x
-*-x
Fio. 12.
tf/
x is largest
p wave
functions.
along the positive and negative
a;-axis
and
ift
along the y-axis.
In a similar way
4.
ijt
would extend along the
z-axis.
Normalization and linear combination of wave

functions
From any
solution
other solutions
Fio. 11. Energy-levels of hydrogen.
Coulomb
potential. For any other potential V(r) the s and

have different energies. For all the other atoms,
for instance, the potential in which a particular electron
moves is also influenced by the other electrons and therefore not simply e 2/r. The s and p states then have different
states
energies.
There are also higher p states whose energies are also

equal to E3 i4 ,..., etc. Furthermore, there are also solutions which depend in a still more complicated way on
a;, y, z.
For instance, there are wave functions proportional to a; 2 xy, xz, etc. These are called d functions and
they are five-fold degenerate. Thus we obtain the level
scheme, Fig. 11.
^r
of the
by multiplying
wave equation we can get

by an arbitrary constant
c; aji is also a solution if ^r is a solution, as can be seen

immediately by inserting c\jj into the wave equation. What
fixes the value of this constant now ? Since *p z is proportional to the probability of finding the electron at a certain
point in space, we might fix the constant in such a way
that
i/r
is
actually the absolute probability of finding the
electron at a certain place. The total probability of finding

the electron somewhere must then be equal to unity. Thus
we have
j$*dr=l.
(13)
This normalization condition (13) fixes the constant c,

apart from the sign. A reversal of sign of ip has no physical
all. The normalized solution of the ground
hydrogen is, for instance, a i e -ra/V7r, as can easily
be found by integration.
significance at
state of
THE HYDROGEN ATOM
THE HYDROGEN ATOM
44
StiU
more freedom in the choice of the functions
we have two
prevails
more) solutions belonging to the same

energy. This is the caso of degeneracy. Let fa and fa be
two normalized or not normalizeddegenerate solutions,
assuming, of course, that tp z is not just a multiple of fa,
i.e. assuming that
fa and fa are really different functions
if
(or
In Chapter II
combination
,
of space.
We
can then get further solutions by taking the sum

or difference or an arbitrary linear combination of fa and
fa thus:
atfit+bfa
is also a solution
energy value
tp
if
fa
~ fa
and fa are solutions
for the
same
The first and third terms are equal to zero because fa is

a solution, and the same is true for the underlined terms
because fa is a solution. Thus
is a solution.
Finally,
can, of course, be normalized again by multiplying by
a suitable constant.
In the case of the three p functions we can take any
linear combination we like of xf(r), yf(r) zf(r). We shall
see in the next section that the choice
ifr
i}s
>
- {x+iy)f{r)
fai = (x-iy)f(r),
= V(-l))
fa = zf(r)
0+i
(14)
(i
is
a particularly important combination. The three func-
tions (14) are connected with the angular

the electron and with the Zeeman-effect.
momentum
Another example of a two -fold degeneracy

wave functions (sin and cos) for a free electron
Chapter II). Both belong to the same energy E
is
of
the two
((3), (4)
of
h /2mX
(5)
we have
This
wave function
=e
also belongs to the
reason for this degeneracy

travel in the
already formed the linear
2ttx
.
2ttx
== cos^f-fcsin-^-
is
45
9
rA
inixlA
.
same energy. The
here that the electron can
+x and x directions (symmetry with respect
to a reflection of the :r-axis).
ANGULAR MOMENTUM. ZEEMAN-EFFECT
47
the operator
IV
(2)
tdx
ANGULAR MOMENTUM. ZEEMAN-EFFECT.

SPIN
and
let the operator on the right side of (1) operate on

a wave function if/(x,y,z). The coordinates in V are not
replaced by any operators. Let us then, after this replace-
ment, denote the right side of (1) by ôpcrator The resulting wave equation Eifs
EOJtemtor will have certain
solutions ^fn belonging to the energy values E say. In
n
each case the energy has a sharp value En E
acting
0Jierat0T
on n produces here the same wave function ^n only
.
Sharp and unsharp quantities

In Chapter III we have studied different types of states
(s, p, d states) whose wave functions, if expressed in polar
coordinates, depend in a different way on the angles. The
1
question arises as to
what the physical
these different types of states

are distinguished
is.
We
shall see that
they
electron has in these states. Before, however, such a ques-
we must remember that
in
quantum
mechanics a physical quantity does not always have a

sharp value. For instance, as far as we have developed the
theory at present, the energy has a sharp value, whereas
the position of the electron is not sharp, there is only a
probability distribution for
it.
We now speak about another physical quantity, namely,

momentum of the electron, and we must ask
whether or in what circumstances this quantity has a sharp
value or whether only a probability distribution exists for
ijt
multiplied
It
is
by the
now
have seen that
tf^d+ri+jpi)+P(s
goes over into the
wave equation
if
y,a)
we
replace
(i)
p x by
constant
En
The
result is
En
if/
quite plausible to generalize this result in the
following way:
Suppose Q{x,y z,p Xi p y ,p z ) is a certain

physical quantity which in general can be expressed by
the coordinates and the momenta. If we now carry out
the
same replacement
(2)
of
p z by
px p
,
the operators
h 8
idx''"'
^ becomes an
6tC "*
operator
@ opemtor
say.
Now
Pera* e on the wave function ift. It may or

not then happen that <? operator reproduces the wave
function ^r with a certain factor q, i.e.
^
e*
ôperator
may
the angular
the various values of this quantity. We decide this question

by analogy with what we already know about the energy
and the way in which the wave equation is connected with
the classical expression for the energy. In Chapter II we
tjj
significance of
by the angular momentum which the
tion can be answered
ôperator
= #
(3)
where q is a constant number (not an operator). In the

case where (3) holds we are justified by the analogy to
the energy in saying that the physical quantity Q has a
sJiarp value,
namely, the value
q.
On
the other hand,
we
must expect in many cases that Q

will be a function
opcrfttor
f (#, y> 2) quite different from $ itself; in this case Q does
not have a sharp value and there will only be a probability
distribution for the various values of Q. This probability
distribution can in fact be calculated, but this requires
a further deepening of the theory which
goes beyond the
48

this book.
framework of
An example
of the latter case is

a function of the
the coordinate x itself. x^[x,y,z)

coordinates totally different from ^t{x y z) itself and not
just i/r multiplied by a constant factor. Therefore x does not
have a sharp value.
As a further example we consider waves with sharp
is
This solution, which also

the required property
is expected to have a sharp value for

wave
of a free electron, because such a
monochromatic
a
and there is a unique
wave-length
well-defined
has
a
wave
relationship between wave-length and velocity (or momentum). The wave functions for such a wave have been
given in Chapter II, equations (3) and (4). However, if
momentum p x p x
operator e 2nix[\
Fx
fi
we apply the
operator
px
=
to these wave functions,
ox
Pi
for instance to the sin solution

.
H B
2irx
2ttx
2ttX
2irti
ir
C08
px =
right-hand side,
= COS iam~ = &***,

2-77X
ip
j-
2irixlX
27J
./.
A~
the coefficient of ip on the

2tt#/A, and this is just our old relation
is
bination
= e~
_sin
cos
describes a particle travelling with
the
2irix/*
momentum
2-nHjX in
direction.
A wave packet consisting of a superposition of several

monochromatic waves with different wave-lengths cannot
have a sharp momentum. Our criterion verifies this. For
instance,
^ ae
8x
the wave function
therefore free to form linear combinations, for instance

the combination Chapter II, eqn. (5);
the wave equation, has
between wave-length and momentum. The linear com-
^r'
is not reproduced, contrary to what

is true for the cos solution. The
same
we expect. The
cos solutions are standing waves
sin
and
The
reason is this:
moving to and fro along the
electron
describe
an
and
instance between two walls).
(for
directions
both
x-axis in
Thus p x occurs with both signs and cannot be expected to
be sharp itself; only the absolute value or p% is sharp, and
this is easily seen to be the case. But we can also easily
construct a wave function for which p x itself is sharp. We
have only to remember that both solutions (sin and cos)
are degenerate and belong to the same energy. We are
?n*
_
&
The numerical value of p x
S
-
satisfies
49
and b may
(a
fcrfafl,_|_6 *rfa/A,
parts, say ]T
k
still
^jt
Al
Â
(4)
2 ),
A,
depend on time). The right-hand side
of the last equation

generally true:
If
is
not proportional to
a wave function
iff
t/.f
This
is
quite
consists of several
such that for each part the quantity
has
a sharp value qk but the qk 's are different, then Q cannot

be sharp for a state described by the wave function $. The
statement also holds the other way round: If a quantity Q
t The wave packet equation
apply the operator
does not even have a sharp energy
(4)
= ~
(4) is
not a solution of the wave
equation, (17), Chapter II, whose solutions always have a sharp energy.
(4) is, however, a solution of the time-dependent wave equation, Chapter
II, (22), provided a and b depend on time in a suitable way (compare,
for
example, Chapter
6881
II, (24)-(26)).
50
not sharp in a state with wave function ^r, then tp can be

considered as composed of several parts ^j with different
sharp values for each part. Examples are given in the
is
61
M has then only a 2-component
M = m{xv
z
yvx = mr
)
^|.
following sections.
The reader
is
will also easily find that the
momentum p x
not sharp for any of the hydrogen wave functions given
in Chapter III, in agreement with the considerations
pp. 27, 36.

It is seen that the
wave function of a
particle
on
with a
given sharp value of the momentum is necessarily complex.

The occurrence of complex wave functions is quite general
itself is not a
in wave mechanics. This shows that
physical quantity that could be measured directly in some
way or other, as is the case, for instance, for an electric
field. Its role is mathematical, but for that reason none
the less vitally important for the description of nature.
is complex) has a direct physical
Only i^ 2 or \ifi\ z (if
meaning, being the probability distribution for the positft
t/r
tion of the electron.

2.
Angular
In
momentum
classical
of s
and p states
mechanics the angular
momentum
the well-known expression mr 2 X angular velocity.

For a motion in three dimensions all three components
wiU be different from zero. They are:
z
x
v
This
is
M M M
M = yp zp
M = zp
M = xp yp
,
In
classical
~ rcos<,
=v
dx
dt
dy
dt'
dt
,
<fy
a particle moves in a field that

are
three components of
they remain constant in course
if
all
known, for instance from the

motion of a planet round the sun. Is this also true in
quantum mechanics ? Clearly, before such a question can
be raised, we must decide whether x My, z have sharp
values, for if this is not the case we cannot say that these
quantities have constant values.
In wave mechanics we again replace p x etc., by (2). The
is
angular
momenta then become
i^dz
operators:
-|2
Byy
X
dx
8z
(6)
&y
In order to find out whether the angular momenta have

sharp values and, if so, what their values are, we let the
operators (6) act on the wave function
In particular
t/f.
dr
64
-rsin^-jH-cos^-j-,
(5)
constants of the motion, i.e.

of time, a fact that is well
JfT
rsin<,
xp si
where v is the velocity and r the radius from the origin

of the coordinate system, r X v denotes the vector product. If the motion is in the rcy-plane we may introduce
mechanics,
x.
is spherically symmetrical,
p),
polar coordinates
vector given by
M = m(r x v) = (r x
y,
dr
Even if x is not sharp there is a certain meaning in saying that

a constant of the motion. What then remains unchanged in course
of time is the probability distribution for the various values of
x
In this sense all three components of
are constants of the motion,
t
it is
also in
wave mechanics.
52
we
M's which the

The wave functions
are interested in the values of the
electron has in
an
5, |i,
or d state.
of an s state are functions of r only, for those of a
we
^r
choose the wave functions, Chapter III
(14),
state
+1
t/>
This
an equation of the form
is
M
s
iff
A^r,
63
the multiplying
has
constant A being -{-h. Therefore in the state +v
z
the sharp value +#
As regards the other two wave functions, t/r.j and ^ we
easily find in the same way that
tf/
Let the operators (6) then act on $ and let us see

whether, for instance,
A^, where A is the multis ift
plying constant.
M =
Let
t/r
first
be an s wave function,
i.e.
MJ =
9
t/r(r),
Therefore, in a
momentum has
0,
p state the z-component of the angular

ft, if the electron
the three values +ft, 0,
has the wave functions ^ +1 fa, tfj_ x respectively.

If we now try to find the corresponding values of
x
we shall find that x $ is not equal to A^, for any
and
dr r
dr r
v>
0. Thus A =
In a similar way
x^ =
v
s state. We have therefore found that in an s
three
components of
the
tfi
angular
momentum
of the three
for
an
functions
ifi
^.j,
and
d_
dz
'SjjJ'/W
fyj
p wave functions,
show that we obtain three
be one of the three
(14).
different values for
We
shall
M for the three p functions

s
= x +iy)f(r),
0_! = (x-iy)f{r),
fa - s/(r).
0+i
Differentiating
+1
state, all
are sharp
have the value zero.
Let now
Chapter III
i/
As an example, take
is
if/
two p functions ^ +1 0_ r t
,
t If
^ +I we get
<f>_i
and
4C# = * *{if+(*+iy) \ I) -ly(f+(*+iy) \ %
-M-T y)f = H*+iy)f = H n
zf(r)
not the same function of x, y, z as ip
electron
is
the
sharp
value
if
a
does
not
have
Therefore
x
in the state
The same is true for u and for the other
yf(r)
we had choBen
then
>li
the wave functions
would not be sharp,
4i
îsa
and <p v inBtead of i(>+1 and

linear combination of ^+1
.;
Thia illustrates the statement made in section 1, namely, that if a quanconsists

tity does not have a sharp value for a wave function ifi, then
of parts with sharp values for each part. In this case ipx consists of
two parts with the values -f ft and ft for
64
We summarize
Table
our results in Table
MM
and p
states
an electron has then a magnetic dipole. The dipole moment

is, as is well known,
55
H, say. To understand what happens in this case consider

first the classical case of an electron moving on a circle in
the plane perpendicular to the direction of the field. Such
1.
Angular momentum in
State
v^Tc'
not sharp
-h
the radius of the circle, v the velocity of the electron,
and c the velocity of light. Now the angular momentum
of the electron is m(rx v), and if we call the direction of
mrv, or
the magnetic field the z-axis, we can write
z
if r is
One may wonder why

x and Jl^ since, after all,
cular direction
This
is
is
plays a role so different from

in the wave equation no parti-
distinguished from any other direction.
p wave
functions.
We could
equally well choose three different linear combinations,
(y iz)f(r), ccf(r) changing x^y,

y->z,z->x. For these three wave functions x has sharp
values, and they are also ~\-h, h, 0, whilst
z and My
are now unsharp. Thus we find in general: In a p state
only one component of the angular momentum can have a
for instance {y-\-iz)f(r),
M
M
sharp value, and then has one of the three values -{-ft, 0,
ft,
the other two components are then not sharp. The particular
direction in which the angular momentum is sharp can be
chosen arbitrarily.
also see that, as far as the angular
a sharp value at
momentum
all,
it is
3.
This expression holds quite generally, even if the electron

does not move on a circle or in a plane which is perpendibeing then, of course, the
cular to the magnetic field,
3
in the z direction.
momentum
component of the angular
be proved that
fact
in
and
can
It is very plausible
if
only for z the
theory,
quantum
(7) should also hold in
according to
has
electron
the
which
value is substituted
electron has
the
field
magnetic
the
quantum theory. In
states.
Zeeman-effect
an additional energy
*-*-&,**
This energy
(8)
around
which the angular momentum is sharp can be removed if
we consider the electron moving in a weak magnetic field,
arbitrariness as to the choice of the axis
is,
of course, in addition to the usual kinetic
and potential energies occurring in the wave equation. Now

the energy, being an observable quantity, has a sharp value.
It is therefore obvious that we must choose the direction
in
The
constant.
quantum
(7)
--SS5*
has
For h and zero are

surely constants that do not change in course of time.
Later we shall use the values of z to classify and label
the
M=
due to our particular choice of the linear combina-
tions of the three degenerate
We
which the angular
momentum is sharp to be the direction
H. The measurement of the energy in the

magnetic field can be regarded as a measurement of
z
value.
sharp
a
forcing
assume
to
z
of the field
56
an atom
placed in such a field, it will

p
have different energies according to the three different
values of z namely,
If
in a
state
is
component of the angular momentum

sharp values (as we shall not show in
M=
2
2mc
(See Table 2.)
hH
still
or
(9)
or
2mc
TiH
we
that the
plot the energy levels in a magnetic field
level splits
up
In a magnetic
2ft.
a d level would accordThere are also levels with
field
into five levels.
higher angular
h,
0,
momentum
splitting
up
into
It is often convenient to represent the angular
we
see
into three levels (Fig. 13) with
7, 9,...,
but these are of lesser importance, at any rate

for light atoms.
etc., levels,
tum by
If
up
ingly split
h,
28,
can have the
detail)
five
57
an absolute value
the
maximum
momentum
of the angular
value which M can assume;
momen-
We
the picture of directional quantization.
define
as being
i.e.
d states
<
Fig. 13. Zeeman-effect of p state.
a separation given by formula (9). This splitting up of

a single level into three in a magnetic field is called the
for each state the angular momentum by

a vector of length I. Relative to the z direction I can have
only a certain discrete set of directions, namely, such directions that the component of I in the z direction assumes
and represent
the values
'Zeeman-effect', after Zeeman, who was the first to discover that certain spectral lines emitted by atoms in a
strong magnetic field split up into several lines. The separation is proportional to the field strength H. Formula
(8) has been fully justified by the experiments, at any rate
for those levels which split up into three levels (compare,
however, the following sections).
MJh.
Table 2 Angular momentum

M,
t
not
sharp
not
sharp
of d state
I
+ 2K
+1
-2fi
\
4.
d states, directional quantization
In addition to
and p
states there are also states with
higher angular momentum. In Chapter IH, section 3, for

instance, d states have been mentioned. In a d state the
The
quantization* picture is shown in

The various components of the orbital angular
momentum are h, 0, % for a p state, and 2h n, 0, n,
and 2h for a d state respectively. In all there are 2/+1
'directional
Fig. 14,
58
different directions for
into
which the
This
Z.
is
also the
levels
original level splits in the Zeeman-effect
The values of z always differ by an

multiple of ft and are:#Z, ?i(l I),..., ft(l 1), M.
in a magnetic field.
integral
number of

2
vanish. Ml+Ml has the sharp value M Ml =
The formula
The
5.
(10)
has been confirmed by
all
59
hH.
experiments.
electron spin
The question now
whether all the above results

with
the experiments. The answer
are in perfect agreement
and
this
will
lead us to a further expanis only partly yes,
p-5tate
arises
sion of the theory.
The SchrOdinger equation always
gives perfect results
In the presence of a
true that cases exist where a level splits
up into three or five levels. There are, however, also cases
observed where a level splits up into only two levels. This is
if
there
is
magnetic
state
Fio. 14. Directional quantization.
The number I, defined as the maximum value of MJft

has also a slightly different significance. We ask what the
absolute value of the angular momentum
is, or its square
M Ml+Ml+Ml
The reader may easily verify that M has a sharp value for
2
and p
generally so for
namely
and 2ft 2 respectively. This is
an electron in a central field of force. The
general formula
is
states,
M =
2
ft
l{l+l).
(10)
We might have expected to find ftH
was the maximum

a maximum we would imagine
.
value of MJh and when z is

the other components to be zero. This is not so because
the other two components of
are not sharp and do not
field present.
field it is
indeed the case for all atoms with only one valency electron,
lithium, sodium, etc. A splitting up into two or four levels
can never be obtained from the wave equation, which
always leads to states with a 1, 3, 5,... -fold degeneracy.
The problem was solved by Goudsmit and Uhlenbeck
(1925) by an ingenious suggestion. These authors assumed
that the electron, in addition to its orbital motion, is
capable of carrying out itself a spinning motion. For the
moment we may picture
the electron as a small solid body,
rather than a point, carrying out a rotation about its own

axis similar to that of a gyroscope. Thus in addition to
orbital angular momentum I the electron has its own
angular momentum which we call the spin and which we
denote by s. The value of s must be chosen so that we can
account for the splitting into two levels in a magnetic
its
field.
The
%
no magnetic
orbital angular
momentum
a magnetic
The
levels in
field.
gives rise to 2l-\-l
spin angular
momentum
In order to make
must have the value .
will therefore give rise to 2s-\-l levels.
equal to two, 2s+l = 2, s

The component of the spin momentum
this
in the z direction
60
can then have the two values only:
61
There the spin appears no longer as an independent addiwave mechanics but is derived from the principles
of these two theories. The value (11) of the magnetic
moment follows then also. However, the picture of a
rotating ball is not quite substantiated in this theory but
the proper explanation goes beyond the framework of this
tion to
JC
-P,
the subscript s refers to the 'spin momentum'.

In the picture of directional quantization (Fig.
are only
two
there
5)
directions for the spin; namely, the direction
of the field itself or the opposite direction. In a magnetic

field the spin alone will split up into two levels with equal
book.
Since
momenta
angular
all
occurring in atomic physics
numbers multiplied by ft, it is convenient to

introduce these numbers themselves instead of
We
s
denote them by mz> etc:
are simple
separation from the original level.
=m
MJh
<
MJh = m
z,
etc.
3S ,
ms etc., and
momentum'. Such numbers as
I, m
mSZf etc., which can be used to label the quantum
z
states, are called quantum numbers.
For simplicity we
simply
them
call
shall
henceforward use the
'angular
Fig. 15. Spin: directional quantization and
Zeeman -effect.
From the very fact that the spin causes a splitting in

a magnetic field it follows that a magnetic moment is also
attached to the spin, as must be expected from the picture
of a charged rotating ball. From the magnitude of the
splitting the value of the magnetic moment can be obtained.
It is found to be the same as that of an electron with
angular momentum ifs = ft, namely (7),
"*
2W
One might have expected that
is
in
an
zero and the spin
is
the only angular
If the electron
(U)
only half that value

because the mechanical momentum of the spin is only ft/2.
But this is not the case. The explanation of this factor
cannot be given here. The spin and all its properties are
now understood as a necessary and natural consequence
of the theory of relativity combined with wave mechanics.
is
electron.
s state
momentum I
momentum of the
the orbital
In this case the level actually splits up into two

however, the electron has an orbital angular
levels only. If,
momentum
>
the result becomes more complicated,
0,
then the orbital momentum and spin momentum combine.

In order to see in what way this combination takes place
we consider the z components of the two momenta and
denote the orbital momentum now by
h The total
angular momentum is obviously the sum of the two
m = m^+m^.
= m can have
it is
For a p state
1,
one of the following
six values:
+*
m =
z
-r
1
2>
2>
2*
2>
62
\ occur twice, as they can be obtained
The values and
\\ or \
0+.
the % component of the total angular
must therefore be possible to consider
two ways: \
in
Now mz is
momen-
it as the
tum, and it
component of a new vector denoting the absolute value of
the total
angular
momentum corresponding
to the
magnetic
tum
field itself.
forcing
magnetic
orbital
it,
field
This interacts with the spin
63
momen-
as in Fig. 16, either in the direction of this

is the same as the direction of the
or in the opposite direction. The
cases is, of course, slightly different.
(which
momentum)
energy in the two
of the
momentum and
the s of the spin. This quantity

we denote by j. It represents the total angular momentum
arising from the combination of spin and orbital momenorbital
tum. The components of j,
i.e.
z,
must be
similar to
fe ,
Fig. 16. Vector addition of spin
m there must
= with mz = f
Now, with the above values of

different values for
andj =
Table
3.
with
j,
namely, j
m =
z
\,
be two
z,
(Table
Combination of spin and
\,
\,
3).
orbital
j-4
OP
momentum
and
orbital
momentum.
We estimate the order of magnitude of this splitting compared with the ordinary distances of the energy levels in
an atom. The magnetic action of the orbit is that of a
magnetic dipole placed at its centre with a moment
(7)) of the order of magnitude ehjmc. The spin
a magnetic moment of the same magnitude, but
is, of course, placed on the orbit. Two magnetic dipoles at
a distance r have a mutual potential energy of the order
3
fjfi/r
For r we may insert a distance of the order of
magnitude of the Bohr radius \\a h 2 jme 2 Thus the
energy-splitting is of the order of magnitude
(equation
also has
|
1
mz
-|
-1
-I
Thus an
electron in a
state can actually exist in
two
momentum
The
different states with total angular

first is
or |.
four-fold degenerate, the second two -fold; the degree
of degeneracy
is
2j+ 1
in each case. This can be represented
in the directional quantization picture, Fig. 16.
can put its spin either in
The electron
the direction of its orbital
momentum
or in the opposite direction.
Actually, the two states have slightly different energies

even in the absence of a magnetic field, for the following
reason:
an
The
electron rotating in
electric circular current
and
its
orbit
is
equivalent to
therefore produces a
weak
Here we have left one factor e 2a. For this is the potential
energy of the electron in its orbit and is the order of
magnitude of the atomic energies. In (12), then, the factor
e 2 jhc occurs.
This is a pure number without physical
dimensions, (e 2 has the dimension energy x length, h that
of energy x time, and c that of length/time.) Its numerical
value, calculated from the known values of e, h, c, is
l/137.t Thus, from (12) it follows that the splitting due

to the spin-orbit coupling is 137 2 times smaller than the
ordinary energies of an atom, which is very small. This
Consider first the simple case in which there is no

angular momentum, i.e. where l x and / 2 are both zero.
Both electrons are then in s states. The two spins then
combine to give a resultant spin for the total system. The
way in which they combine is exactly the same as in the
case of a spin-orbital momentum combination (section 5).
64
splitting is called the fine structure of the spectral lines,

e 2 jhc
and
the fine structure constant.

is applied each level j splits
If also an external field

into
2j+l
up
The two spin components
levels (Fig. 17).
J=*
field
m _ 44
= +*
V
With
(2)
/*
No
in the z direction are
"
m>S>
iff
j-l
field
Fig. 17. Splitting of p level with spin.
An electron in a d state will have

momentum j = 2+f = or j ~ 2 f
65
No
With
Field
field
a total angular
f.
The
Fio. 18. Zeeman-effect of singlets
first
and
triplets.
splits in a magnetic field into six, the second into four
levels
6.
The
Two
electrons with spin
electrons.
l2
the
is
=s =
2
sum
=l,
of these, giving the four pos-
0,
0,
-1.
field
(atom) each
and an orbital
momentum
or opposite directions
momenta
of two electrons are to be
combined. In a spherically symmetrical
and
In the same way as in section 5, we conclude that the

combination of the two spins gives rise to two different
1 and 3
0. In other words,
states with a total spin s
the spins can have the same direction (ff) giving s = 1
We consider now the question of how the spins
orbital angular
electron has spin s l

2j
So far we have considered systems with one electron

only, but of course most atoms and molecules have several
and
resultant
sible values
respectively.
very strange number indeed, remembering that it is built up of

t
universal constants only. Its explanation is one of the greatest unsolved
problems of modern physics.
(f J,)
giving 5
0.
will show no splitting,

In a magnetic field the state 8
but the state a = 1 will split up into three levels much in
the same way as a p state. The generally accepted notation for states with various total spins (irrespective of how
6881
66
many electrons the system
has)
is
ly
s=0,
l,
singlets,
doublets,
triplets.
=s =
s = 3J-1-S2 =
and s x
The two
orbital
States with different total spin alwajr s have very different

energies, and the energy difference of the two states arising
use capital letters:
(in a He atom, for

very large. The reason for this fact is due to
a fundamental quantum-mechanical phenomenon (the
'exchange phenomenon') which will be explained in
Chapter V.
We now consider the more general case where one or
both electrons (in different orbits) have also an orbital
angular momentum l lt l 2 respectively. The question arises
then as to the order in which we have to combine all these
momenta. As was mentioned above, states with different
total spin have very different energies, in other words, the
coupling between the spins is always large, whereas the
coupling between spin and orbital momentum is small,
giving rise to only a very small splitting. The coupling
between the two orbital momenta of the two electrons is
We
P if =
from the combination of two spins

is
also usually fairly large.
We therefore arrive at the follow-
First combine the electron spins
and obtain the total

spins and also combine the two orbital momenta and obtain the
total orbital momentum. Afterwards combine the total spin
ing rule
with the
total orbital
momentum
to
obtain the total angular
momentum. The energy

combination
tude (12)).
is
splitting arising from the latter

usually very small (of the order of magni-
The following example
will
make
the rule clear.
Let
and
spins
triplet).
momenta combined give a total orbital

the reader will now easily find to be
momentum which
To denote
The combination of the

(singlet
1,
One would think that the energy difference of the two

states (singlet and triplet) arising from the combination
is very small and due only to the small magnetic influence
of one spin upon the other. But this is not the case.
instance)
\.
67
2,
0.
1,
states with various total orbital
(I
0),
{I
1),
momenta we
(I
2), etc.,
d refer to one single electron.

speak, for instance, of a 'singlet P state' written as
while small letters like

l
as *S
if
therefore,
and s
= and
<s
x
1
orbital
0,
1, s t
1
*S,
S,
these six
All
p,
and of a 'triplet S state' written

1, etc. For our example we find,
that the following states arise out of a com-
bination of
Finally
s,
=5 =
P,
P,
W,
Z).
have rather different
states
we may combine the
momentum
\*.
to obtain the total angular
energies.
total spin with the total
momentum
P, *D the combination gives

nothing new because one of the two total momenta is zero.
The total angular momenta for these states are therefore
j.
For the states
S,
3
0, 1, 1, 2 respectively. For
j
3
three states j
(I
2, 1,0. For
three
=
states j
(/
D =
3, 2,
1.
2, s
The energy
1,
1)
1)
we
we
find
find also
differences of these
states with different j are usually very small and will play
no role in the following. In a magnetic field each state
with total
The
momentum j
for the
rule
will further split into 2j-\-l levels.
combination of angular momenta
explained above is called the vector addition rule (see also

chapter VII section 2.)
,

the two electrons repel each other, which
by the mutual interaction potential
The wave equation
for
two electrons
For
the motion of an electron in a potential

found the wave equation
V we have
Ci
J;
tfi(x,y,z). This equation, with tft depending on the

three coordinates of the electron, refers only to one single
electron.
molecules
In order to treat more complicated atoms or

we have to generalize it to describe two or more
To obtain guidance
electrons.
we
as to
how
this
can be done
consider again the classical expression for the energy,
E ^ pI+pI+pI
(2)
where r12 is the distance between the two electrons.

In order to translate equation (1) into wave mechanics
we use the old method of replacing p x by the operator
-
acting on a wave function $.
8x
the two electrons

tp
accounted for
V -i-

I.
is
we have
To
distinguish between
h 8
p Xi by
to replace
acting on ti. Obviously, now,

dx 2
of the coordinates of both electrons
by
iff
-r
If
we now
consider
two
t and p Xt
must be a function
*i>
<P{
x i>yi> z v x 2>y2>*2)-
Hence
a2
/e 2
2m
a2 \
a2
/a 2
e 2 \,
2m T/
A*"
V(x,y>z).
(3)
"2m
potential field
69
electrons
moving
same
in the
with
the energy will be
= V(x ,y v z )-\-V{x ,y
= v +v +v
1
l2
2i
z i )-{'Vl2
Using the abbreviations
2m
2m
+ y^vyi,z )+V(z
l
2r
y 2l z 2 )+Vn
and
>
e2
M M
+ te\
a2
a2
V|
for
t-s+t-2+^~2'
d2
dx\
a2
f
f r
(1)
where pXi pXt etc., are the momenta of the first and
second electrons respectively, and x x
v 2 y v y2 etc., their
coordinates. The potential V is the same function for both
electrons, but the two electrons may, of course, have
different positions. Vix^y^Zj) is the potential energy of
the first and V(z 2 y2 z ) of the second electron. In addition
2
,
a2
r2
V*
dyl
8zl
we
get
W+Vfr+|?C-W-0.
(4)
This is the wave equation for two electrons. The potential

function Ffo,...,^,...). ( 3 ) is a function of the coordinates
3
of both particles. The wave function

all the coordinates of both particles
i.e.
70
is
also a function of
^{^y^z^x^y^z^,
of six variables.
>
(x2 y 2 z 2 ) depends on
one particular kind of solution.
b
the second electron only. We insert (6) into the wave
(2) is a conequation (5). Vf acts on a (l) only, while
stant with respect to the differentiation Vf. The first term
of (5) is then ^ b (2)Vf a (l). On the other hand, Vf. acts on
only and the second term becomes a (l)Vl^6 (2).
b {2)
Substituting in (5) (always neglecting Vn ) we get then
electron at the position
x l ,yly z 1 and
at the
second electron at the position x

v y2 z2
,
Solution of the
</r
same time
the
wave equation
</r
(/>
for
fe
tjs
A6 (2)Vf,A a
2.
*i>
i/r
In analogy to the physical interpretation of $* for

one electron (Chapter I, section 3) we interpret now
^( x i* yi> z i. z 2 y% z 2 ) as the probability for finding the first
71
(l}+Wl)Vi^(2) +
2M
^-^ {l}^(2)a
two electrons
We have seen that for two electrons the Sehrddinger

equation becomes
1*+W+f?(jr-*5~r,-)* =
(5 )
The exact
solution of such an equation is always very

In order to solve it we must make approximaAs a first approximation we neglect the term e 2 /r
difficult .
tions.
12
now
the case where only the electron 1 is

present, moving, of course, in the field Vv Its wave function will then satisfy the equation for one electron
Consider
o.
In other words, we regard the electrons as moving independently of each other in an external potential field. We
should then expect that each electron can be in certain
energy-levels irrespective of the energy-level occupied by
the other electron. These energy-levels are the same
only one electron were present moving in the field
V,
namely, Ev E2 ,..., Ek ,..., say. The two electrons can then
occupy two of these levels, E and E for instance, and
+E
b,
the total energy will be E

Each electron can occupy
a
b
any of the levels k all distributions are possible. This
will
be exactly what
(5).
We find a solution by putting $ equal to the product of
the two functions,
shall find
by solving equation
- lMl)&<2),
(6)
depends on the coordinates of

the first electron only, the index a being used to
distinguish
*pa (l)
a (x ly y ltZl )
ift
(7)
t/r
(7).
Thus
ijj
a (l)
must
satisfy the equation
V?^(l) +
â-WaU) =
0,
(8)
then really satisfied. Similarly,

were alone present, would satisfy
exactly the same equation (7), only with the coordinates
and we
shall see that (6') is
the second electron, if it

1
and 2 interchanged:
V!<A(2)+~(-W(2)
e.g.
^
where
we
o.
Since the two electrons are moving independently, and

from what we have said above, it is clear that the a (l)
which occurs in (6) should be identical with some solution of
as if
v 2 0(i)+^(-W(i)
particular solution
is
E = Eb
<>.
(9)
<At>(2).
The complete set of solutions of (9) is identical with the

complete set of solutions of (7) since it can make no
72
difference whether
we
a single electron No. 1 or No. 2.

But for a certain special solution each electron may be
in a different energy state
or
In the same way as
call
Ea
Eb
above we assume that 6 (2) occurring in

with some solution of (9):
(6) is identical
2m
Vi^(2)+^(^-F M6 (2) =
2
Making use now of the fact that

$a and
equations (8) and (10) separately, we easily
0.
(10)
-fâW
and the second and
fifth
atom,
very similar to that of the
the only difference being that the nuclear charge is 2e
instead of e. (It can easily be seen from Chapter III
that the energy of each level is then just four times the
energy of the corresponding level of hydrogen.) A state
of the He atom is then characterized by a distribution
of the two electrons on the set of energy-levels of the
set of energy-levels is
satisfy the
see that (6')
provided that E is suitably chosen. The

fourth terms of (6') are together equal to
satisfied
first
is
e
6 (2)
together equal
-^2^(2)^(1).
cc
-y
Fig. 19. Distribution of two electrons on one-electron levels.

(6')
reduces,
f?lM2)*a (l),
on cancelling the common factor
E = Ea +Eb
(li)
a solution of the wave equation for two electrons with

no interaction between them, the energy of the whole system
is
just the
sum
Various such distributions are shown hi Fig. 19.

is in the ground state and
the other in the first excited state. In y both electrons
are in excited states, and in S both electrons are in the
ground state. The lowest state of the atom is obviously
He +
to
Thus the product of two wave functions for one electron each
is
and
e
Hence
73
of the energies of the two electrons, as
was to
be expected.
Let us consider, for instance, the helium atom. It has
a nucleus with charge 2e and two electrons, 1 and 2. Each
electron moves in a potential V = 2e 2 /r, and can exist
in a set of energy-levels for one electron <He+ ion).
This
ion.
In distribution a one electron
the distribution S.
For all these considerations we must not forget that we
have neglected the mutual interaction of the two electrons.
If the latter is taken into account the states of the atom
can no longer be pictured in such a simple way as the
on the levels of the He + ion.

for the two electrons is no longer a
distribution of electrons
The wave function tp

function in three -dimensional space. It is a function of
six coordinates x lt y v z v x 2> y 2 z 2 If it were at all to be
,
represented it would have to be a function in a sixdimensional space (or a space of Zn dimensions for n
electrons). That this must be so is quite clear from the
physical meaning of 0. There must be a probability for
finding one electron at one point and the other at another
exchanging the labels of the electrons, in other words, to the

complete symmetry of the wave equation in the electron
74
point,
and therefore tp must be a function of two

The fact again shows how different
coordinates.
a measurable
field in
sets of
is
from
ordinary space.
We
call it
We
and
of(6)and(12):
= 0.(.l)fc(2)+lM2)iMl),
4>- = *.(l)fc(2)-A(2)iMi).
wave equation
(6)
a (l)06 {2)
of the
(5) without the interaction e 2 /r

12 describes
a state of the system where one electron is in the state a,
the other hi the state 6, a and b being states of the oneelectron problem. This is now not the only solution de-
same
distribution of electrons.
The reason
is
this: Actually, (6) describes
labelled
No.
is
in level
the state where the electron

a and the electron No. 2 in b.
Physically, however, the
two electrons are indistinguishable

also clear from the fact that the
completely symmetrical in the two
from each other. This
wave equation
electrons 1 and
We therefore
(5) is
is
2.
expect that
= ^(2)^(1)
also a solution of (5), and this is easily verified.
solution (12) also belongs to the energy
is
E = Ea +Eb
(12)
The
It describes the state
where electron 2 is in a and electron

Hence there are two wave functions belonging
to the same energy. We have to deal with a case of
is
(13a)
^+
have seen that the solution
scribing the
exchange degeneracy.
(12) we can build up more wave functions

(6)
belonging to the same energy. We may take any linear
combination as was the case with the three p wave functions. For instance, we could choose the sum or difference
From
Exchange degeneracy
3.
labels.
7r>
in b.
degeneracy; similar to that found before for a p state. The

new kind of degeneracy is due to the possibility of
(13) is also
a solution of
(5).
(13 a)
and (136) are
13 *)
parti-
These linear combinadoes not change

tions have namely the property that
essentially if we interchange 1 and 2. Obviously + satisfies
this requirement as it is symmetrical in 1 and 2. 0_ changes
i}j_ is not really
its sign if 1 and 2 are interchanged, but
we
can multiply
a wave function different from 0_, since
without
changfactor
every wave function by an arbitrary
solutions
the
For
both
ing its physical meaning.
(13) there
is therefore no real distinction between the two electrons.
The use of the linear combinations (13) instead of the
mere products (6) and (12) is really compulsory. We must
accept it as a very fundamental principle that no distinction
can be made between two electrons. The probability of finding
electron No. 1 at x v ... and electron No. 2 at x 2 ,... must be
equal to the probability of finding electron No. 1 at x 2 ,.. and
No. 2 at x lt .... All that has a physical meaning is the
probability of finding one electron at x v ... and one electron
at x 2l ... irrespective of any labels. Hence it follows that
cularly important combinations.
tfj
^(1,2) =^(2,1),
and hence
0(1, 2)
= 0(2,1)
or
0(1,2)
=-0(2,1).
(13')
76
The two wave functions (13) are just the ones which
satisfy these two alternative conditions respectively.
The exchange degeneracy, of course, only exists so long
as the interaction between the two electrons is neglected.
Otherwise we shall see {Chapter VI) that (13 a) and (136)

triplet
states
and fj are
singlet states.
77
(Each
triplet
comprises 3 degenerate states.) In cases where the two

spins are parallel the atom has a magnetic moment due
to the spin (cases a and y). If the two spins are 'antiparallel' as in jS or S there is no magnetic moment, a and jS
belong to different energies. If the interaction between

is taken into account the solution will be
some wave function i/r(l, 2) that can no longer be split up
into products (6) or (12) or their combinations (13). On
the other hand the symmetry properties (13') must be
maintained. It is still true that no distinction can be made
between the two particles and that therefore
the two electrons
Fig. 20.
He
states, allowed
and forbidden.
Wi,*)ttM),
no matter what ^(1,2) is otherwise. So one of the symmetry properties (13') must hold for each solution but
symmetrical and antisymmetrical solutions will in general
have different energies.
4.
The Pauli exclusion
same energies.
Whereas for atoms with only one
principle
So far we have been dealing with the orbital motion of

the two electrons only and have not considered their spin.
For a full description we must take into account the
spin. As we have seen before, each electron has a spin
momentum whose z component can have the two values
m8z = -Hi or i- We symbolize these two spin-states by
arrows f and j respectively. The spins combine to a total
spin
and 1, which we also symbolize by |j and
ff
3=0
respectively.
States
ff
are three-fold degenerate.
Let us, then, consider the various states of two electrons,

for instance the distributions a
and S of Fig.
To describe
we must draw an arrow through each electron
showing the spin. It is obvious that we obtain then four
them
can be distinguished from each other experimentally, since

a would split in a magnetic field into three states (see
Chapter IV) whilst in /? the spins cancel. The same is true
for y and 5. Without a magnetic field a and /3 have the
19.
fully
times more cases, because each electron can have two spin
directions. They are shown in Fig. 20.
Cases |f are
electron it could be
stated that, after the inclusion of the spin, the results of
the theory are in perfect agreement with the experiments,
one further fundamental principle has now to be added to
the theory
when we
are dealing with
two or more
electrons.
We can compare our results with the experimental facts,

for instance for the He atom. The ground state of the
He atom could be either or j8; both have the same energy.
It is now an experimental fact that the ground state of
He does not split up in a magnetic field. Besides, He is
a.
known to be diamagnetic and never paramagnetic. This

means that the He atom in its ground state has no magnetic moment. We must therefore conclude that the state
(Fig. 20) does not occur in nature. On the other hand,
the spectroscopical evidence has shown that all the other
states of Fig. 20, , y, 8, do occur in nature. It appears
78
that
we have
79
W e may include the description of the spin also

T
to deal here with a
ciple according to
which not
all
new fundamental
prin-
the states derived from
electron.
in the description of its state
wave mechanics (generalized by the inclusion of the spin)

are possible. The principle is due to Pauli (1925) and is
electron different
called the exclusion principle.
There can never be two (or
The exclusion
5.
if
and
call
the state of an
either its spin or its orbit
is
different.
assumes the simple form:

more) electrons in the same state.
principle then
The spin wave
function
The formulation of the Pauli principle given in the preceding section rests on the neglect of the interaction
OOCr $ *
(ft
Fia. 21. Three-electron states
(Li).
between the electrons. In general it is not possible to

speak of a distribution of electrons amongst the oneelectron levels and the wave function ^(1,2) is not in
general a product ^ (l)i/r6 (2) or a linear combination of
such products. We must look, therefore, for a more general
formulation of the exclusion principle which can be applied
also if the interaction between the electrons is taken into
account. For this purpose it is useful to describe the spin
also by a 'wave function', similar to the wave function
tt
In order to find out which are the 'forbidden' and which

the 'allowed' states, we consider the more general case
of three electrons, for instance the Li atom. We can
obviously describe the states of the atom again by distributing the electrons in all possible ways over the oneelectron levels, i.e. the levels of the Li++ ion and drawing
an arrow through each electron to describe its spin. Some

of the distributions with low energy are given in Fig. 21.
Distributions where no arrows are drawn stand for all the
distributions with various spin directions. Now those distributions which are found not to occur in nature are
crossed out.
The
exclusion of the states
a, y, *,
7}
was
derived by Pauli from a careful study of the spectra of

more complicated atoms. It is easy to read off from Fig. 2
the general rule: All distributions with three electrons in the
same
level
are forbidden.
Two
electrons
may
be in the
same
tjs
a (x,y,z)
two
wave
Since there are
describing the orbital motion.
states of the spin of
functions a and
j8,
one electron, we
shall
momentum mS2 namely

The two spin functions a, j9
the three orbital wave functions of an
of the z component of the spin
-f and
If
we
respectively.
correspond to
electron in a
have two
say, corresponding to the values
state.
picture the electron as a sphere rotating about
the axis of the spin direction either from west to east or

the other way roundf the azimuthal angle of this rotation
<f>
f This picture of the spinning electron as a rotating ball must not

literally. No physical reality whatsoever can be attached to
the 'structure of the electron'. The only observable fact is that tho
electron has an inner degree of freedom, the spin, and we use this
but must then have opposite spin directions. For electrons in different levels there is no restriction of the spin
be taken
directions.
picture merely to illustrate this fact. But no further conclusions should

be derived from this picture and questions of what the radius of such
a ball would be, etc., are void of any physical meaning (compare p. 61).
level,
We
can also formulate this exclusion principle in a

simpler way if we change what we call a 'state' of a single
'
80
may
will
be the variable, upon which the spin functions a,

depend. We need not know what functions of
a((f>)
and
In
a,
(<) are.
One may
/?
and
/?
with
m =
sz
of the two electrons
The
is
total spin in the z direction
then, for our four spin functions,
<f>
m = m&'+mW
also use different variables.
8Z
our following considerations the variable on which

depend will not occur; all we need is the fact that two
all
f$
*p a (x,y,z)p.
(14)
These describe an electron in the same orbital state a but

with different spin directions. If we have two electrons,
there will be two spin functions a^), a(< 2 ), etc., or more
shortly a(l)
and
a(2), etc.
The
total spin
(16 a).
(166)
a(l)j8(2)+a(2)jS(l)
-1
jB(l)j8(2)
(16c)
(16d)
(l)j8(2)-(2)j3(l)
tff
a (x y,z)a,
+1
(1)(2)
such spin functions a, jS exist.

The complete wave function of an electron is the product of the orbital wave function a {%, y, 2) multiplied by
one of the spin functions
*fr
81
These four spin functions are associated in a unique way

with the total spin s of the two electrons. (See Chapter
IV, section 6.) It will be remembered that two electrons
with spins \ combine to give a resultant spin
5
and
0,
wave function
the product of two such spin functions (the spins move

practically independently). In all we have, for two elec-
is
with the components
m^ =
1, 0,
1,
and
0.
trons, four spin functions:

It is clear that the spin functions (16 a)
electron
and (16 c) belong
s=l, because only 5=1 can have a comm^ = +1 or 1. One of the two spin functions
to a total spin
electron 2
ponent
o(l)a(2)
(166) or (16c) belongs also to s
jS(l)jS(2)
(15)
< 1)0(2)
and the other to
0.
seems plausible that (166) should belong to s = 1 and

because (166) is symmetrical and the other
(Ida*) to 5
spin functions (16 a) and (16 c) which we already know to
It
0<1)<*(2)
we can form any linear combination of the four

functions (15). To comply with the fact that the two electrons are indistinguishable from each other we shall again
Again,
choose, just as in section 3, the symmetrical and antisymmetrical combinations of (15). a(I)a(2) and /3(l)jS(2)
are already symmetrical. Instead of a(l)j9(2) and a(2)/J(l)
we choose a(l)0(2)+a(2)j3(l) and a(l)0(2) a(2)0(l), of
which the first is symmetrical, the second antisymmetricaL
Now each a describes a spin with a 2 component K = +|
belong to s
are also symmetrical.
(16d)
is
the only
antisym metrical spin function. The three spin functions

1 belong to one and the same state (a triplet state),
of s
at any rate as long as there is no magnetic field present,
and it is certainly plausible that they should all have the
same symmetry properties.

That this must really be so follows from the following
consideration: The three spin functions for 5 = 1 correspond closely to the three wave functions of a p state
5681
82
(Chapter III, section 3). They aTe degenerate and we are

allowed to choose any linear combinations of them we like,
whilst the spin functions for 5
and s
1 belong to
(1)(2)
)
[Wl)iM2)-* a (2)fc{I)]
However, the spin functions, or any

linear combinations of them, must still he perfectly either
symmetrical or antisymmetrical in the two electrons.
Suppose now the spin function (IQd) to belong to a total
spin s 1. We should then be entitled to form linear
combinations of {IQd) and (16a) since both would belong
to s = 1. Such a combination would no longer be either
symmetrical or antisymmetrical as it consists of two parts
with different symmetry properties. Only combinations
between (16 a, 6, c) remain always symmetrical. It follows
that (16 d) must stand apart and must be the wave func-
83
(17 c)
fll)j8(2)
a(l)j9(2)
a(2)j8(l)
different energies.
tion belonging to s
We
0.
functions:
Symmetry of spin functions and
Symmetrical
Antisymmetrical
6.
have written out these eight wave functions so that

symmetry properties are evident. (17 a) and (17 d)
are symmetrical in the two electrons ((17 c?) changes its
sign twice if 1 and 2 are commuted). (17 6) and (17 c) are
their
antisymmetrical.
=
=
f f
t
total
spin
Triplet state ( 1 6 a-c)

Singlet state (16 d)
Now
according to the Pauli exclusion

wave functions can be
principle not all of these eight
allowed.
To
the case where the two
a =
b,
(17 c)
automatically.
and
We know from the exclusion principle that

two
electrons
two electrons are
in different orbits,
excluded) but by (17
(since for
also antisymmetrical in the
the electrons. We multiply the two parts of the wave

functions for two electrons, the orbital wave functions
(13), and the spin wave functions (16). We obtain altofunctions
(1)(2)
]
(17a)
JS(1)JS(2)
I
(l)jB(2)+(2)jS(l)
Wa(lft(2)+ô (2)0 6 (l)]x[a(l)jff(2)-a(2)i8(l)]
one
triplet
state exists in nature in addition to the singlet state (17 6).

It is very probable that this is not described by (17 a)
now be easy to formulate the Pauli principle in

that does not depend on the interaction between
bMl)iM2)+ik(2)^(l)]
orbit,
(17 d) vanish in this special case
way
consider
same
must have opposite spin and

the three wave funcTherefore
the state is a singlet state.
the one describing
and
is
tions (17 a) are forbidden,
(176)
is antisymmetrical.
function
wave
the state occurring. This
If the
General formulation of the Pauli principle
wave
we
functions
electrons are in the
It will
gether eight different
wave
find out the forbidden
in this case the

4.
(17I)
We
i.e.
thus arrive at the following scheme for the spin
Table
[0 tt (l)^(2)-^(2)^(l)]x[a(l)A2)-a(2)Al)].
(17 6)
6 (17 a)
is
two
electrons.
c).
(17 c) is
Furthermore,
if the two electrons

has already been seen to be
allowed, and we conclude therefore that (17 d) must also
be forbidden. The allowed wave functions are therefore
(17 6) and (17 c), and both are antisymmetrical. We are
therefore led to the following principle: The total wave
only one singlet state
are in different orbits.
known even
is
(17 6)
function of two electrons is always antisymmetrical in the

electrons. If the wave function is split up into an orbital
part and a spin part the orbital part can be symmetrical
84
or antisymmetrieaL The spin part

or symmetrical respectively.
is
then antisymmetrieal
The following table gives the allowed total wave functwo electrons with no interaction either in two
different orbits a =$ b or in the same orbit a = b.
tions for
Table
5.
Symmetry of orbital wave junctions and spin

Orbited
Spin
* B U)&(2) + * (2)&(1)
a (l)j3(2)-(2)fll)
Total s
the reader to show that then the wave function vanishes

automatically for all the forbidden states, namely a, y, c, r)
of Fig. 21.
This general formulation of the exclusion principle was
given by Dirac (1926). It has been found to be in perfect
all relevant facts.
spin functions was to show how the
the
The role of
wave functions is connected with
orbital
the
symmetry of
at a proper formulation of the
arrive
to
and
the total spin
will only occur in this
functions
spin
The
principle.
Pauli
agreement with
(singlet)
f
*a (l)*
<2)
connexion.
1)0(2)
U(l)0(2) + a (2)0(I)J
*(1)M2)-* B <2)&<1)
(1)(2)
(triplet)
0)fl2)-(2)fll)
(singlet)
The new formulation of the exclusion principle is not

confined to the case where the interaction of the electrons
is neglected, nor is it confined to two electrons. For the
symmetry of a wave function is a property that persists

if the wave function is not merely a combination of products a (l) 6 (2) but is some more general function 0(1, 2).
Also for three or more electrons we can demand that the
wave function be antisymmetrieal in all electrons. To
satisfy the exclusion principle we have to postulate that
every leave function of two or more electrons is, if the spin
variables are included, antisymmetrieal in all these electrons,
changes sign if the labels of any two electrons are

exchanged. Thus for three or more electrons 0(1,2,3,...)
i.e.
0(2,1,3,....)
-0(1,3,2,...), etc.
trons, the interaction is neglected, the
built
up of products such
as
If,
for three elec-
wave function
is
a (l)0 (2)0c (3), a(l)a(2)0(3)

and the antisymmetrieal combination obtained by permuting the variables 1, 2, 3 is to be taken.
leave it to
fc
We
85
PERTURBATION THEORY
modification of the orbits will not be very large either.

At each position of the particles they have an interaction
energy V
e 2 /r 12 If we wish to know the average change
whole s>stem due to V we have to average
the
of energy of
VI
PERTURBATION THEORY
General theory
Now
which
also
to
If the interaction of two electrons is neglected, we have

seen that the energy is the sum of the individual energies
of the two electrons in states a and b
over both orbits.
themselves,
1.
is
small,
it is
If
V12 is inserted
into the
the perturbation of the orbits
due
clear that,
f
first
of the orbits themselves.

approximation to
The question now arises as to how the total energy changes

if we take into account the mutual interaction energy,
if
approximation, we
can form the average value of
V by neglecting the change
for
first
Vn= +
87
The
the
change of energy is therefore

obtained by assuming that the
particles still move on their
orbits, as if no
Fio. 22. Unperturbed and permutual interaction existed. turbed orbits of two particles.
The change of energy of the
whole system, due to the fact that they have_an inter-
unperturbed
wave equation, the
far too complicated to be solved exactly.
latter
We
becomes
can only
look for suitable approximations. Hence we shall use a
rather crude approximate method which, however, is sufficient for most problems and gives us a proper insight into
what really happens. Better methods would have to be
used if we wished to calculate atomic energy-levels with
great accuracy, say of 1 or 2 per cent. But we shall not
be concerned with such precision work here. The method
we shall use can best be understood if we consider the
problem in classical mechanics (and was actually developed
first for astronomical purposes).
Consider two particles
moving on two different orbits (Fig. 22). In the absence
of the interaction V12 the two particles move on the
'unperturbed orbits' (full curves). The interaction V12
causes the two orbits to be somewhat modified (dotted
curves), but if the perturbation V12 is not very large the
action energy V, is given by the average value V, averaged

over the unperturbed orbits.
The method can easily be applied to wave mechanics.
Instead of well-defined orbits we have to deal with probaposition of
bility functions giving the probability for the
the two particles. What correspond to the unperturbed

wave
orbits of classical theory are now the unperturbed
functions of two electrons.
The two electrons in, say, the He atom have an interaction
e2
F=+-.
perturbation energy we have to mulat
tiply F12 by the probability of finding the two particles
2
probathe
two specified positions, i.e. by (l, 2), which is
To form the average
t/r
bility of finding the electrons at positions
and 2. For 0(1,
2)
PERTURBATION THEORY
PERTURBATION THEORY
approximation, use the unperturbed wave

those calculated as if there were no mutual
perturbation. The average value is obtained by integrating
over all possible positions of the two electrons. Thus the
The perturbation e 2 /r12 is symmetrical in the two electrons 1 and 2. When integrated over the coordinates
of both electrons the first two terms of (3) give equal
contributions. Hence, cancelling the factors 2, we obtain
88
we may,
as a
functions,
first
i.e.
89
average change of energy becomes
A =
Wl 2)V dr^*
(1)
f
where dr x dr 2 are the volume elements of the two electrons.
In (1) it has, of course, been assumed that ^(1,2) is
t
A =
square does not represent the absolute probability. If $ is not normalized yet
we have to write, instead of (1),
f
its
AE gives the first approximation of the energy change

of the whole system due to the perturbation V. (2) is quite
general and can be used for any kind of perturbation.
He atom, exchange energy
the change of energy due to the perturbation F

e 2/r 12
12
we take the square of the wave function and insert it in
equation (2). If the two electrons are in different unperb,
the
wave function
C
'12
finding the electron that is in the

tfsl is the probability of
point, or the average charge
certain
a
at
a
quantum state
density of the state a. Similarly, tf>l is the average

charge density of the electron in the state b. The two
charge distributions repel each other according to their
Coulomb
As an application of our general theory we consider the

case of the two electrons in the He atom. To obtain
is
given
by
symmetrical or antisymmetrical combinations are used, and

not the product wave functions
^ a (l)^r6 (2). Hence
(13).
call it
writing dr for drx dr%.
turbed states a and

Chapter V, equation
if/
(2)
j^dr
2.
We shall not work out &E numerically (which can be

done by inserting the wave functions tp a and b in (4)).
Instead we confine ourselves to a qualitative discussion.
Consider the first integral of equation (4) in the numerator
and
F^ 2 dr
(4)
{^(l)^(2)/ra (l)^(l) /r u (2)^{2)} dr.dr.

1
normalized to unity, otherwise
W(l)(2)^ (lM*(Iâ (2)^ t (2)}-- dr,d
interaction.
The
mutual
which has thus a simple
potential energy of this
interaction is just the integral C,
natural meaning.
This Coulomb interaction of the two charge clouds C

however, not the only contribution to the change of
energy. The second term of the numerator of (4), which
is,
we denote by A,
is
It is essential that the
-/* (1)^(2)^(1)^(2)-^
(5)
12
cannot be so simply interpreted. It arises obviously

from the fact that the two electrons are identical and can
be exchanged. The electron 1 is 'partly' in the state a and
PERTURBATION THEORY
90
PERTURBATION THEORY
partly in the state 6 (0 ( 1)^(1)). If we wished we could

a
A also as the Coulomb interaction of two charge
interpret
densities, or rather of different parts of
one charge density

This charge density would then be
Û 1 )&,(!) (5) represents the Coulomb interaction of each
part of the charge density
a ( 1)^,(1} with every other part
â( 2 HW 2 )- We call it the exchange charge density. A will
be called the exchange energy.
It must be remembered though that the interaction of
with each other.
exchange charge with itself, namely A, occurs with

both signs -\-A and A, and this fact cannot be explained
by such a semi-classical picture of interacting charge
clouds. The very occurrence of the exchange charge and
the exchange energy A is a typical result of quantum
medmnics. It is due to the fact that two electrons are
this
For a discussion of
section
is
but for the excited state

negative values (cf. Chapter III, Fig.
positive,
ifj
all
the
has positive and

always
2
10). e /r 12 is
The
largest contributions to the integral arise

values
of r 12 i.e. from two parts of the charge
from small
coordinates of the two electrons not
with
the
density t/> a ^ 6
(2)</r (2) will then nearly
other.
each
from
different
very
a
6
positive.
t/r
always have the same sign as </r(l)i/r6 (l), and whatever the
sign is, the integrand will be mostly positive. We therefore understand that the whole integral A is positive. The
change of energy AE is now
AE = CA,
with
C and A
(6)
both positive quantities.
occurs with two different signs, and we obtain therefore two different values for the perturbation energy. We
remember from Chapter V that the two signs of the
unperturbed wave function
the understanding of chemistry.
and more still for

The homopotar chemical
bond between two atoms rests on such an exchange energy

(Chapter IX).
As to the denominator of (4) we may assume that ip
are normalized,
shall furthermore
show
,/r
of great importance indeed
for the understanding of atomic spectra,
further remark that
1.
The 'exchange energy'
and
we
factors occurring in C are positive. Therefore C is positive.

The same is true for A, although this cannot be shown so
easily. If a is the ground state, it is true that ip a is always
indistinguishable
from each other, and besides, of course,

are described by wave functions. We have to accept the
existence of such an exchange energy as a rather fundamental consequence of wave mechanics which cannot be
explained on grounds of a classical picture. We shall
discuss it in more detail in Chapter IX at the end of
(4)
01
J ^(1) d7l
2
J"
(2)
dr 2
the perturbation energy
AE = C+A
(s
Q)
(5
1).
(7 a)
1.
We
and the
triplet the
energy
A = C-A
0.
Therefore the second integral in the denominator

to 0, and the denominator is equal to 1.
(l)^{2)^(2)^(l)
are attributed to the different values of the total spin

3
and 1 respectively. Therefore the singlet state has
in section 3 that
1^(1)^(1)^
in which one electron

electron
in an excited
one
is in the ground state a and
triplet and
both
for
is
energy
state 6, the unperturbed
If
is
equal
He atom
(7 6)
we have,
for instance, a
singlet
tion
PERTURBATION THEORY
PERTURBATION THEORY
92
Ea -{-Eb
AE, the
If
we add the energy due
to the perturba-
have a lower energy than the

Thus the unperturbed state splits up into
triplet state will
singlet state.
two
states with different energies, the energy difference

being 2A. The e-xcJiange degeneracy is now removed. A is
by no means a small quantity. It depends on the Coulomb
interaction e 2/r12 of the two electrons which is on the
average, though, smaller than the potential energy of each
electron but of the
same order of magnitude. The
More we cannot expect from such a crude perturbation

method.
Especially,
excited state of
Fig. 23
He
is
it
shows to be the
has been found that the

triplet,
Unperturbed
the spins are tied to the symmetry of the orbital wave

function.
If both electrons are in the
is
same unperturbed level a,

^ a (l)â (2) and the change of
energy simply
&e= ^mm-d^dr,
-I
2E
Triplet
Singlet
Fig. 23. Perturbation energy of
He
levels.
With the help of more refined methods the He states

can be calculated with much greater precision. In particular the ground state has been worked out by Hylleraas
(and others) with an error of less than 0-0001 per cent.
The perfect agreement of this result with the experimental
value is a very valuable test of the theory and shows that
the many-body problem is also described correctly by
wave mechanics.
3.
(8)
'12
first
what
Perturbed
singlet-
distance of the atomic levels

also in the following way: The coupling of two electron spins
is accompanied by a large coupling energy, depending on
wave function
this is precisely
Singlet
magnitude of the
themselves. We can express this
say, the
and
case.
triplet splitting is therefore of the order

of
whether the two spins are parallel or antiparallel. This

coupling energy is of the order of magnitude of the electrical
forces and by no means, as one might think, of the very
small order of magnitude of the magnetic interaction of
the spins. The reason is that, through the Pauli principle,
93
The orthogonality theorem

In the preceding section
we have used
the fact that
AE
has only one value in accordance with the fact that

only a singlet state exists, by the Pauli exclusion principle.
The change of energy due to the perturbation of the
two electrons in the He atom is shown qualitatively in
Fig. 23, for the ground state and the first excited state.
The values of the integrals have been worked out
numerically, and the energy-levels have been found to
agree within 20 or 30 per cent, with the observed ones.
1^(1)^(1)^
(9) is
0.
(9)
quite a general theorem, called the orthogonality
relation,
quantum
different
moves.
which we have now to prove. It refers to the

states of a single electron and holds for any two
states and in whatever potential field the electron
PERTURBATION THEORY
PERTURBATION THEORY
94
We write down the wave equation for the two

states
a and
quantum
over y) and d*jdz
We therefore
2m
v v+(^- f m.
h
= .
2m
+ ^(E ~V)fa =
b
(Ea
If
(106)
We
multiply (10 a) by fa and (106) by fa, integrate over

the whole of space, and take the difference. The terms
containing V then cancel and we obtain
now a and
Ea ^ Eb
95
Hence the left-hand side of
find that
(10a)
0.
-E )jfafadr =
b
6 are states with
two
0.
^)dr^~{E ~E
J<fcvv-* a
b)
J*'
dr.
(11)
Consider the left-hand
The
side.
in-
tegration over oV includes an integration over z from

oo
to +oo.
take the first term d 2 jbx z and integrate over
x. Integrating by parts,
vanishes.
different energies
the integral of (13) must vanish, which proves
our statement (9).

It may be that ip a and fa are two different wave functions belonging to the same energy. In this case we are
dealing with a degenerate state. The relation (9) need then
not necessarily hold. On the other hand, if u ip b are two
degenerate wave functions, we are free to choose any two
linear combinations of ift a and fa, and this fact can be used
to make the two new wave functions orthogonal (i.e. to
choose such combinations that (9) holds). Suppose, for
instance, that the integral (9) has the value c. We then
choose, instead of fa
We
fia
("
We
was normalized
If fa
00
(13)
tfj
1 1
always
and
fa,
= fa Câ*
0* oV = l), we find
J
4>a>
fafa dr
fib
cc
$* dr
= 0.
J*
can therefore assume that (9) holds for any two

wave functions, whether they belong to two dif-
different
ir'to'"''
"dx
n)
_m
DO
l&e 8*
a* ax)"*'
same energy. The reader will

p wave functions fa,
(Chapter
The same is true for
are
orthogonal.
III)
z
fa,
the combinations ip +l fa, if/_ v They are, of course, also
orthogonal to the wave function of the ground state of
ferent energies or to the
easily satisfy himself that the three
(12)
The integral on the right is obviously zero. The first term

on the right has to be taken at an infinitely large distance
+oo or -co, but here the wave function must vanish.
This is indeed the 'boundary condition' which we had to
impose on all wave functions (Chapter II). (12) therefore
vanishes. The same is true for the parts d 2 jdy 2 (integrating
ifr
hydrogen.
In Chapter V, section 5, we had also introduced spin
wave functions a and j8 which describe the spin of one
electron in the two states with spin component in the z
PERTURBATION THEORY
96
m^ i- Also these two spin functions must

be orthogonal to each other because they belong to two
different states which have different energies in a magnetic
direction
VII
field:
| o0
The
integration
is
0.
to be carried out over
variable (or variables) on which a
Chapter V, section
5,
(14)
and
all
/?
values of the
depend.
(In
we had provisionally introduced some
angle of rotation for this variable but this is irrelevant.)

Also a and /3 may be assumed to be normalized
1.
The
The
electron configuration
different chemical elements consist of
atoms of the
ze, surrounded by z electrons, where z

an integer ranging from 1 (hydrogen) to 92 (uranium).
If we neglect the interaction between the electrons each
electron moves independently in a potential V = ze 2 jr.
This is the same potential as in a H atom, only that e 2 is
replaced by se 2 The arrangement of the energy-levels for
the individual electrons is therefore the same for all atoms
and especially the same as for hydrogen. Only the scale of
energy is different. We see that at once from formula (7),
Chapter III, giving the energies of the hydrogen levels.
a nucleus of charge
is
/*-/*-
(15)
We
obtain the levels for the individual electrons of any

atom by replacing e 4 by z 2 e 4 Apart from the scale
the arrangement of the one -electron levels is therefore
always that of Fig. 11. To find the levels for the whole
other
atom we
distribute
all
the
available electrons
amongst
ways.
In doing so we must remember, however, that not all
distributions are possible. The exclusion principle (Chapter V) demands that in each one-electron level not more
than two electrons can be placed (and they must have,
the energy-levels of Fig.
11, in all possible
then, anti parallel spin).
The
different one-electron levels are classified
angular
$,
I,
p,
momentum,
d,.,.
e.g.
0, 1, 2,..., etc.
levels respectively. If the angular
then the level
is
by the
We call them
momentum
is
(2Z+l)-fold degenerate and actually
98
consists of 2J-f
coinciding levels. In such a level there-
can be placed.
arranged according to increasing
2p, 3s, 3p, 3d, 4s, etc., where Is, 2s,...
fore 2(2Z-f-l) electrons
The order of the

energies, is Is, 2s,
levels,
denote the lowest, the second,
etc., s levels.
99
of the nucleus depending on the distance and ranging

from z to s-2.
The effect is, that the 2s and 2p levels are more or less
separated. The same is true, of course, for the 3s, 3p, 3d
levels, the separation is bigger for the higher elements
where
s is large.
Actually, the separation
is
such that for
which is the lowest

one, the p d levels lying higher, which cannot, however,
be shown here in detail. The separation of the 2s and 2p
levels is, though quite appreciable, by no means so large
as that of the Is and 2s levels or between 2s and 3s.
We are chiefly interested in the lowest state which is
the normal state of an atom. We obtain it by filling up
gradually all the lowest one-electron levels with the maxi-
light elements it is
always the
s level
Fig. 24. Charge cloud of two I electrons.
mum
electrons permitted by the exclusion

Take, for instance, the case of the Li atom with
three electrons. The first two electrons can be placed in
the Is level. The third electron is then in the next higher
level, which is 2s. Next comes beryllium with four electrons, of which two are placed in Is and two in 2s. For
the heavier elements the 2p level is gradually filled up. It
6 electrons. This is the case for
is completed by 2.3
number of
principle.
In the
is
H atom
the 2s and 2p levels coincide, the same

3s, 3p, 3d levels. This is rather an
true for the higher
and is only true for a Coulomb potential

In fact the s and p levels are somewhat separated
except for hydrogen itself, for the following reason: Consider the case of three electrons with two electrons in the
Is level and one in the 2s or 2p level. It would now be
a very poor approximation if we were to assume that the
electron in the higher level moves in the pure field of
the nucleus ze 2 jr. The two electrons in the lower level,
which are, according to Fig. 10, p. 38, much nearer to the
nucleus than the 2s electron, represent a negative charge
accidental fact
ze 2Jr.
shown in Fig. 24. This charge cloud

Coulomb field to some extent. At a very large
distance the potential is only (z2)e 2 /r. At medium
distances the potential is between ze 2 jr and (s 2)e 2 /r.
cloud, roughly, as
screens the
Actually, therefore, the electron in the higher level does

not move in an exact Coulomb field, the 'effective charge'
Ne, which has ten electrons, two in Is, two in 2s, and six
in 2p. When a level is filled completely we call it a 'closed
shell'. Hence the first closed shell is completed for helium.
Since the separation of the 2s and 2p levels is not very
large, one usually considers them together and speaks of
the next closed shell only when both are completed, which
is first the case for Ne. In this way we find the electron
distribution for all the elements. For the first two periods
of the periodic system it is shown in Table 6.
The chemical behaviour of an atom depends on those
electrons which are not parts of a closed shell. The
chemical valency is in fact either equal to the number of
THE PERIODIC SYSTEM OP ELEMENTS
100
101
free electrons not contained in
a shell or to the number

of electrons missing to complete a shell, whichever is the
smaller of the two numbers. Elements which differ only
by a closed shell and have otherwise similar electron configurations behave chemically very much alike. From
Table 6 it is clear that the chemical behaviour must be
Table
6.
He
Li
Be
Electron distribution in periodic table
Ne
Is
2
2
2s
2p
Na
Ca
Al
Si
11
12
13
14
15
16
1*
17
IS
2
5
2*
2
6
2
6
2p
2
6
3s
Zp
CI
10
For instance, H, Li, Na have, apart from closed

one free electron. F, CI,... have one electron missing
to complete a shell. The elements which have only closed
shells, namely, He, Ne, A,... are the rare gases.
In this way quantum theory accounts for the wellknown periodic system of chemical elements.
periodic.
Shells,
extra-shell electrons in the lowest state of the
is the lowest state. The various states arising from

one electron configuration can be found by the vector
addition rules of Chapter IV.
We discuss now all the elements of the first period The
H atom has one electron in the Is level. The spin
2
is , therefore the lowest level is a S state. Helium has
two electrons in the Is level. They must have antiparallel
spin (Pauli principle), therefore the ground state is l S. In
them
the Li atom the third electron is in the first excited

2s level with spin
\. Since the two electrons of the Is
do not contribute to the spin the ground state is 2 S,

All the alkalis have a 2 S as ground state since the closed
shells do not contribute to the spin. For beryllium the
l
electron configuration is similar to that of helium with a S
the
2p
as ground state. Boron has one electron more in
shell
The atomic states

We shall now deal in more
detail with the first period

of the periodic table. The chemical behaviour of an atom
depends not only on the electron configuration but also
on further details such as the spin directions of the free
electrons, etc.
It is especially the spin directions of the
In addition, the
1
The atomic
p electron has an angular momentum
2
H,
He, Li, Be,
elements
five
state is therefore P. For the
the lowest
from
state
arising
B there is only one atomic
level.
The
total spin
is
therefore s
\,
'
electron configuration.
2.
atom which
are of importance. We shall see that, in general, out of

each electronic configuration several atomic states arise
which are distinguished by the total spin and angular
momentum. Owing to the interaction between the electrons (which has bo far been neglected) these states will
have different energies and we have to find out which of
Their ground states are repre-
sented in Fig. 25.
From carbon onwards

plicated.
to the two
the atomic states are more com-
electrons in the 2p shell in addition

denote such a configuration
electrons in 2s.
C has two
We
The p level really consists of three levels with

mz = +1, 0, 1. The two electrons can be distributed in
various ways amongst these three levels. In addition, the
by
s 2p 2 .
102
two
electrons can have parallel or antiparallel spin directions. Only, of course, if they are both in the same level,
the spins must be antiparallel because of the exclusion

principle.
He
L!
States
Ground
ft
-1
tt-
I,
1,
0,
D,
mz
+1
S.
t
-1
mz
44
spectroscopical facts.
+ +
The next element is nitrogen, which has three electrons

2 3
in the p shell (configuration s ^ ). If we are only interested
consider those configuraonly
in the ground state we need
z<* m iz "2
= 2,
Chapter VI, that the energy is lowest

3/2
s
when the spins are parallel. Therefore
Ground
Fio.
27.
we conclude that the ground state of
state of N.
l
l
the carbon atom is 3 P, the D and S
being excited states. This is also in agreement with the
ng
4 4
we have m s
three states have different energies. We

can easily find the ground state. We
remember from the perturbation theory,
The various distributions of the two p electrons are

shown in Fig. 26. Altogether there are nine different configurations. Each configuration has a total spin a = 1 or
0, according to whether the two spins are parallel or anti-
ft
If the interaction between the

electrons is taken into account, all these
Btates of H,..., B.
0,
configurations therefore
gether we have three atomic states arising

3
from the two p electrons, namely P,
2P
Fio. 25.
1,
+1. These three
Be
'5
For
I, i.e. & *P state.
+1, +2, and one more 0.
They belong to two atomic states with I =* 2 and 1 = 0.
We therefore obtain D and 1 S. Alto-
41 ft
4* -ft 44
Element
represent a state, with
-f
is
I, Z+l,., +i. We
with components m z
together
those configurations
to
group
therefore have
the same value of I.
to
belonging
values
with different
z
1 we have
Collecting first the configurations with s
turn
wia
2P
is
103
0+2-1
O+1-i
three electrons have parallel spin. There

must then be one electron in each of the three levels. There
Here
0, therefore
is only one such configuration.
s
tions
+1
angular momentum the z component of which is simply

the sum of the contributions of the two electrons. It is
given in the bottom line. To find the atomic states we
an
orbital
momen-
(Fig. 27). Since $

|, the state is a *S state. This
I
of
state
N. In addition, there are also
ground
the
fact
in
is
excited states arising from the same electron configuration.
They are in fact 2 2) and 2 which we leave to the reader
parallel, and belongs therefore to a triplet or singlet state

respectively. In addition, each configuration has an orbital
state has
all
Fio, 26. Electron distribution in C.
remember that each atomic
where
to show.
The remaining three elements, O,
F,
Ne
are easily
104
state with as
The
We confine ourselves to the ground state, i.e. the
discussed,
many electrons with parallel spin
behaviour of carbon, as it is responsible for the four

valencies of C. This question will be dealt with in Chap-
as possible.
electron configurations for the three elements in this

shown in Fig. 28. The ground states are there-
ter
case are
fore:
*P
for
(s p*),
^ for F
(s
p%
and *
+r
r
-1
m2
fj
ft
Ne
(s p*).
"',
Ground
7.
He
l
*S
We may
2p
-i
Ground
states of the first
Li
Be
*S
1s
states of O, F, Ne.
IP
ap
*S
ip
%p
Fig. 29. *S state of C.
the level arises;
(ii)
the screening of the inner electrons
which causes the s~p splitting; (iii) the interaction of the

electrons which causes a splitting of the various terms of
the same electron configuration (neglecting the very small
fine structure splitting). Although these effects are all of
the same order of magnitude they become, oh the whole,
decreasingly smaller in the order they were mentioned.
To illustrate this we give in Table 8 the excitation energies
of some of the levels of carbon as they are known from
period of elements
Ne
spectroscopy.
also consider further excited states
sidering higher electron configurations.

instance,
2s
There are also excited states of 0, arising out of the same

electron distribution. They are *D and 1 S. F and Ne have
no excited states from the lowest electron configuration.
Finally we give a table of the ground states of all the
elements of the first period (Table 7);
Fig. 28.
Table
X.
The energies of the atomic levels are mainly determined

by three factors; (i) the electron configuration from which
Ne
for
105
we can
by con-
For carbon,
Table
for
one of the 2s electrons into the 2p

level. There are then three electrons in 2p and one in
2s
(configuration sp a ). All these four electrons can have
parallel spin {Fig. 29). This will be a quintet state (s
2)
and since I
(the three p electrons have a total orbital
momentum zero as in the case of N) the state is 5 . This
6
S state of C is of great importance for the chemical
8.
Exci tation energies of carbon
levels
raise
Levels
Number
of electrons
Atomic term
Excitation energy (eV)
2a
2
3jp
1-3
2p
2a
2p
1*
2;i
2-7
4-2
3.s-
Sp
ip
7-4
7-7
The terms of the lowest configuration are only separated

by 2 or 3 electron volts. Raising one 2s electron into the
106
2p level requires at least 4 volts. There are many more

terms arising from this configuration, apart from the & S
(not given in the table), which lie much higher. The raising
of a 2p electron into the next higher level 3* requires much
more energy, 7-8 eV. The energy required to remove one
2p electron altogether, i.e. the ionization energy, is 1 l-2eV
Of all the terms arising from the same electron configuration it is always the term with the highest multiplicity
which is the lowest, in agreement with what we found in
Chapter VI.
The situation is very similar for all the light elements.
VIII
DIATOMIC MOLECULES
1.
The
electronic states
this chapter we consider the quantum states of a

diatomic molecule consisting of two nuclei and a number
of electrons. In general all these particles are in motion
and interact with each other. At first sight the problem
In
quantum states looks rather formidable, but we

more concerned with a general survey than with an
exact calculation of levels and this is obtained quite easily.
The chief clue is the fact that the nuclei are very much
heavier than the electrons. They move, therefore, much
more slowly. If they were at rest and in fixed positions,
at a distance R say from each other, the problem would
of finding
are
be reduced to finding the energy levels of the electrons

moving in a potential created by the two centres of force,
but the motion of the nuclei would not enter into the
problem. Now when the nuclei move but slowly this cannot
create a drastically new situation. For each instant of
time t we may, approximately, regard the nuclei as if they
were at rest and determine the energy of the electrons for
the particular distance R(t) which the nuclei have at the
time. In this way we obtain electronic energy levels En {R)
which all depend on a parameter R, the distance between
the nuclei. The problem is thus split up into two parts,
(i) the determination of the electronic levels, and (ii) the
motion of the nuclei themselves. The latter problem will be
considered in sections 2 and 3. It is true that this splitting
up of the problem is not exact (because the motion of the
nuclei influences that of the electrons and vice versa) but it
is a good approximation and quite sufficient for our purpose.
DIATOMIC MOLECULES
DIATOMIC MOLECULES
108
and would
at once dissociate. Such repulsive states are

important when we consider the interaction between atoms
in connexion with the chemical bond but need not be con-
We can easily obtain a qualitative idea of the
^-dependence of at least the low energy levels. When the molecule

is in equilibrium, i.e. in a state of minimum energy, the
nuclei are at a finite distance R
The lowest electronic
level Et (R) has therefore a minimum energy at R
For
larger and smaller R, Ex will increase. It is convenient to
include in E(R) also the Coulomb interaction between the
two nuclei which is always positive (repulsion) because the
nuclei are both positively charged. Thus at small distances
(and this also applies to all excited states) E^R) increases
rapidly and becomes positive. On the other hand, for
R i? E1 also increases. When R is very large, the
energy is equal to that of the two
separated atoms. The lowest electronic molecular state will naturally
go over asymptotically into two atoms
in their ground state. The ^-dependence of the excited molecular states
En (R) will be similar. It happens
frequently that different molecular
states go over asymptotically into
sidered in this chapter.
We
of an
>
and
the same states
atoms
two atoms in
or, in
of the
other words,
is
according to their total spins

Accordingly, we
I.
momentum
states. A similar
let
definite states interact
E{oo)E(R
classified
with each other and

let their distance decrease, several types of molecular
states may arise (see below). Thus we obtain the type of
molecular states or 'atomic interaction curves' shown in
Fig. 30.
in the preceding chapters that the states
their total angular

t
separated
when we
have seen
atom can be
speak of 2 S 3 P, etc.,
classification can
also be carried out for molecules. Consider in the first
place the orbital angular momentum. If a particle moves
in a field of central symmetry (atom), then according to
classical mechanics all three components of the angular
momentum (and therefore the absolute value of the
angular momentum) remain constant in course of time.
In quantum theory this finds its expression in the fact
that two quantum numbers, the total angular momentum I
and its z component mz exist which remain constant in
course of time and can therefore be used to characterize
the atomic state, whilst the other two components m m3
y
are not sharp. This is not quite so for a diatomic molecule.
The electric field of such a molecule no longer has central
symmetry but is only symmetrical with regard to rotations
about the axis of the molecule. A particle, or several
particles, moving in such a field of axial symmetry have,
according to classical mechanics, but one component of
angular momentum which remains constant in course of
time, namely the angular momentum about the axis of the
molecule. In quantum theory this must mean that a
quantum number p, independent of R, exists, which denotes the angular momentum about the molecular axis, and
that fx like ms can have only integral values, fi can also
be negative because the particles in a molecule can rotate
about the axis clockwise as well as anticlockwise. Now
the energy of the molecule cannot depend on the sense of
Fig. 30. Atomic inter

action curves.
109
the dissociation energy.
In addition to the above type of states which lead to

stable molecules there are states where the electronic
energy is always larger than that of the separated atoms.
(Two such curves are also shown in Fig. 30.) An example
is derived in Chapter IX, p. 133. These are not proper
molecular states because the atoms then repel each other
DIATOMIC MOLECULES
110
rotation.
States with angular
DIATOMIC MOLECULES
momentum
ft
and
/x
wiU
therefore have the same energy. The molecular states are

0. We
thus always two-fold degenerate, except when fi
can, in analogy to the atomic l define the absolute value of
the angular momentum A
^ and classify the molecular
/x.
are
states according to the value A. Only when A
the states not degenerate. We call, in analogy to the
S, P, D,... states of an atom, the states with A
0, 1,...
, II, A,... states respectively. Thus we have the scheme
= =
=
S
A
=
=
Next consider the
A
2,-2
The
spin.
able to retain
its free rotation,
spin has practically no

is
no matter in what sort of
same properties and

as in an atom. There
addition rules
follows the
will
same
be molecular
states with spin 0, |, 1,... or singlet, doublet, triplet states,

etc. Molecular states will therefore have notations such
J
n, HI, A.., etc.

What states of a molecule arise, then, if two atoms,
specified by their s and I values, are brought in contact ?
The spin values are obtained from the ordinary vector
addition rules and require no further explanation. Let l t
and l 2 be angular momenta of the two atoms and m 1 and
m 2 the z components. m 1 = llt ..., -\-l 1 ,m 2 =l2 ,,,,, -\-l2
Let also the axis of the molecule be the z-axis. The angular
momentum about the molecular axis p is obviously the
sum m x -\-m 2 Thus a variety of ^-values arise by adding
up the various m t and m 2 Those /^-values which differ
only by their sign belong to the same A and are part of a
two -fold degenerate state. Different A's belong to different
as
Z,
3i>-f 3P
*n,
s,
s,
S,
n,
2j U>
s,
Uj 1 A)
S>
S>
S) 3Ut
3rlj 3 A>
n, n, a.
have different energies when the atoms

of them are states of a stable molecule.
Some, in fact the majority, correspond to repulsion of the
All these states
the particles move. The spin of a molecule has there-
fore exactly the
\S+ 2 PîS,
interact.
...
interaction with the orbital motion of the particles. It
field
...
1,-1
111
molecular states. Thus an atom in an S state and an atom

in a P state give rise to two molecular states with A
and 1 i.e. to a and a IT state. Both occur with all possible
values of the spin obtained by the vector rule. We give
two examples; a pair of atoms in Z S and 2 P states respectively, and a pair of atoms both in 3 P states
Not
all
two atoms. In the case of two H atoms both in 2 S the two

states arising are 2 S and 3 H; the *S is the molecule and
the 3 E will be seen to be a state where the two atoms repel
each other (Chapter IX).
We comprise the characteristics of a molecular state,
namely spin and angular momentum, by one word, 'race*.
We say that two states belong to the same race if they
have the same spin and the same angular momentum, but
if one or both of them are different.
Two X S states belong to the same race, but a 1 and *!!
to a different race
to different races. |
Thus when two atoms in specified states are brought

modes of interaction are possible,
according to which of the above states are formed. In
some cases a stable molecule is formed (in one of its electronic states) and in some cases the atoms will repel each
into contact several
other.
f Actually there are still more characteristics which define a molecular state, namely certain reflection properties of the wave function.
We shall
not use them in the
books on molecules about
follo\ving,
this point.
and
refer the reader to special
2.
DIATOMIC MOLECULES
DIATOMIC MOLECULES
112
The
rotation of molecules
the total energy
move in space.
Now
If their distance
the nuclei are allowed
R changes the electronic
As is well known the centre of gravity could only

move with uniform velocity and this is not an interesting
motion. lfm m 2 are the masses, r 1( r 2 the position-vectors
rest.
=-m
we use this relation and Vi

of the nuclei becomes
If
R = i-j r2
If
and
we
v
X v2 )
1 -\-m 2
The square
is
(4)
-(Rv) 2 ).
(5)
given by
m2 (R2v
Furthermore, the change with time of the absolute value

of the distance R is given by
(2)
dR
introduce the distance
= Vj v
= dR
= - (7n +m )(T X vj
= m(Rxv).
/Tin
(l)
is the reduced mass.

between the nuclei
ti
r2
is
it is
M = m^Ti X v )+m (r
= drjdt the kinetic energy
This
and
gravity
of the two nuclei with the centre of gravity as origin, then

r1
(3)
then vary between two extreme values, R x and R 2 say,

given by the two branches of the potential curve Fig. 30.
In addition the direction of R can change freely in space.
If R were constant, this would be a free rotation. Thus we
are dealing with a vibration (change of R between two
extreme values), as well as a rotation round the centre of
gravity. To see this more clearly we put the kinetic energy
into a different form. For this purpose we introduce the
angular momentum of the two nuclei round the centre of
the centre of gravity of the two nuclei f as the origin of our

frame of reference and we may well assume that this is at
{R).
energy changes according to the function E^R). This

change must now govern the motion of the nuclei. In
other words EX (R) will play the role of a potential energy
under the influence of which the nuclei will move.
Let us first consider the motion of the nuclei, according
to classical mechanics. It will be convenient to introduce
We
Let us assume that the electrons are, for example, in their

lowest electronic state, with energy EX {R) represented by
to
\m\*+E
the energy function for the motion of the nuclei

seen that it depends only on one variable, namely
R, and its time derivative.
is, of course, constant,
II
E. Consider now an energy value E higher than the
minimum of the 'potential' Elf i.e. higher than E^Rq),
but lower than the electronic energy of the separated
atoms (2^(00)). The absolute value of the distance R can
nuclei to be in fixed positions not necessarily at the equilibrium distance.

now consider the motion of the nuclei.
the lowest curve of Fig. 30.
is
H=
In the preceding section we have seen that the electronic

levels of a molecule can be calculated by assuming the
113
dt
g dtd VR - (Rv)
R
2
Thus with the help of (5) and

2
of
(dRjdt)* smd R:
t Since the electrons are very light compared with the nuclei, the
centre of gravity of the moleeule is practically identical with the centre
of gravity of the two nuclei.
(6)
we can express v 2 in terms
= MR 2+ [dRy
\dt)
m
2
V2
5SW1
(6)
'
DIATOMIC MOLECULES
DIATOMIC MOLECULES
114
and the energy of the
nuclei
becomes
H=
(7)
Now in many cases the extreme values R
and R 2 between
very much. This is espe-
can vary, do not differ

energy is not much higher than the minimum
E^Bq). (In the minimum itself R equals R and the two
extreme values coincide.) In such cases we may well
replace R in the first term by a constant, for example by
the average (R^R^j^.f The energy then splits up into
two contributions,
/fra t+-#vib> a rotational energy
which
cially so, if the
H=
116
simultaneously with the total energy. The case of a rotating

molecule could only differ by the mass of the particle, but
this enters in the expression for the energy only, and not
2
Thus (9) must hold for a molecule also.
in the value of
has a
Thus vot itself (and therefore also vlb
Tot )
H = EH
sharp value but this fact depends on the approximation

J?
~ const.
K,
rot
//.
rot
= ^+i;
^Kl+i)-
(10)
We obtain a series of rotational energy levels for =

/
0, 1,
These energy values and their spacing are, of course,

much smaller than for electrons because the (large) mass
moleof the nuclei enters in the denominator. For the
2
2,...
rot
" 2m.fi
(8 a)
cule, at the equilibrium distance
(86)
3,
ri+m 2 ri
(by
(1)) is
the
moment
of inertia
of the molecule.
Now
also the
motion of the nuclei has to be treated
quantum mechanics. This is quite simple if

the above separation into rotation and vibration is a feasible
according to
approximation. Consider first the rotational energy (8a).

As we have put R
const, here, this depends on the square
of the angular momentum only. We have considered the
angular momentum of an electron (in a central field of
2
force) in detail in Chapter IV and we have seen that
has a sharp value
M = M(Z+1),
2
The
vibration of molecules
The vibrational energy of a diatomic molecule is given

by (86). If we treat the vibration according to wave
mechanics we have to replace the velocity by the momentum P = mdR/dt and then replace P by the operator
iftdjBR.
It
(9)
E^B)
is
a one-dimensional motion with the disThe wave equation
tance, R, between the atoms as variable.

is
thus
,, i
d-ifi
It
is
o^,
2m
/v
mt*\\J.-
(11)
quite easy to obtain a qualitative idea of the energy

Looking at Fig. 30 we can say at once that all
levels.
discrete levels
0,1,2,...
t This, of course, cannot be done in the potential energy

because this varies very much between t and % .
OS A.U.), we obtain
B^ = fr(^+E {B).
2
7-3 X 10- 3 eV whereas the electronic energies are

h2 l2A
of the order of a few eV. In the next section we consider
the vibrational energy.
and a vibrational energy
mR = A = m
(R
must
En
must he between two energy
limits:
En
certainly be larger than the minimum of the potential
E^Rq), and
it
must be smaller than the asymptotic value
Ei(oo) of the atomic interaction curve.
For
if
is
larger
DIATOMIC MOLECULES
116
than E^co) the atoms can move to infinity and we have

a continuous spectrum just as in the case of the Coulomb
potential. In fact the curves of Fig. 30
have this qualitative feature in common
with the Coulomb potential that both tend
to a finite value at large distances (compare Fig. 7, p. 24).
It might even be
Fig. 31. Vibra
conjectured that the energy levels crowd
tional energylevels.
as E approaches the limit Eôo). This
is
in fact the case for
molecules.
most electronic states of diatomic

Thus we obtain the qualitative level scheme
shown in Fig. 31.

For the low levels a feasible approximation can be made.
We may replace the curve Fig.

the
mniimum by a
31 in the neighbourhood of
-piR-R^+EÊ,);
(12)
/ is determined by the curvature of the curve near the

minimum. Introducing then R~ R = x as variable the
wave equation becomes that of a harmonic oscillator
3+^-w-^
Mr
(13)
= EE^Rq).
solutions of (13) are not difficult to find. The reader

easily verify that the three lowest solutions are
all have equal spacing. This is, of course,

only true as long as the interaction curve can be approximated by a parabola.
The lowest level is higher than the minimum of the
potential Ej{S 9 ) by an amount \ftw. This 'zero point
energy' is due to the uncertainty principle. To make the
energy small we have to make the kinetic energy small.

Now in order to minimize the kinetic energy (and hence
the momentum) a certain extension in space is required,
but at the minimum of the potential the position would

be fixed at R Q Therefore a certain possibility for vibration
of energy.
is needed which in turn means a certain increase
much
usually
levels
is
vibrational
the
low
The spacing of
For
the
levels.
rotational
the
low
2
larger than that of
molecule, for example, 1ia> is about \ eV. Thus each
vibrational level is followed by a band of narrowly spaced
The experimental
analysis of the vibra-
tional levels is an important tool for the determination of

the actual atomic interaction curves.
4.
Ortho- and para-molecules
If a diatomic molecule consists of two like atoms, for

etc., special considerations are needed
example
2 ,...,
H N
(not normalized)
2,
quantum states.
H nucleus, or
The
Consider
2
protons are
two
and
particle
elementary
an
proton, is
From
identical in the same sense as two electrons.
numerous experiments it is known that the proton also
has a spin s = \ and obeys the Pauli exclusion principle.
We can therefore conclude that the wave function of two
protons must be antisymmetrical in the two particles if
in order to obtain the correct selection of
first
"
= 1*1 2>
with the corresponding energies
W = &w,
t
E'%
is
The energy levels
rotational levels.
The
may
In fact the general formula for the energy levels
117
parabola
E,(R)
DIATOMIC MOLECULES
jiSoi,
E'z
Ihw,
=f/1
the simplest case
DIATOMIC MOLECULES
DIATOMIC MOLECULES
the spin wave function is included. The latter

the same as for two electrons (Table 5, p. 84):
antisymmetrical:
a(l)j8(2) (2)j3(l),
exactly
is
0,
(f],)
((0(2)
]
symmetrical:
Ja(l)j3{2)+a(2)jS(l)
lj8(l)]8(2)
s is the total spin of
(16)
1,
(f |)
both protons.
for the motion of the nuclei
The energy function

by (7) and (8),
are concerned with ^rot only.

For an electron moving in a central field of force we have
obtained the wave functions, as far as they depend on the
direction of the radius vector r, in Chapters III and IV.
the angular
functions
is
given
fcbi
R is regarded as constant in
#vib depends only on the absolute value |RJ
of the distance and involves a differentiation with
respect
(provided that the distance
=R
corresponding to
(xi!/)f(r)>
<Ao
R=
const.,
differentiations with respect to the angles,
6,
$, say,
#rot &ot
#vlb ^vib
is
(tfrot +#vib)*W^,
^vib
2).
to
do
is
to replace
for
(18a)
0,
X const.,
for
(186)
1.
depend on the direction of R only, not on R

r x r2
itself.)
If we interchange the two nuclei, R
changes sign. Hence x, y, z also change sign whereas R
remains the same.
Whenever the energy function is the Bum of two parte,

each of
which depends on different variables, the wave
function is a product
of two functions depending on the two
sets of variables separately
Ihe case of two electrons without interaction
is another example
(Chapter V, section
we have
(x/R, etc.,
tfvlb-
= ^rot&ot^vib + ^vib^rot^vib "" ^rot^vlb

then indeed satisfied for E = #
-h#vib .t R = |r r
t
all
xiy
rot
+1, namely
0,
1,
(17)
The wave equation
H* =
(x+iy\
formed
*t(M)*Tib(*)
^rot 0rot
*/(").
= const.,
by the vector R (with respect to a chosen frame of reference).

The wave function consists then of a product
ifj{r)
we obtain
^pot
In 7/rot we have regarded R as constant. //

rot
therefore only depends on the direction of R and
involves
is indeis zero,
=
i=lwe have three wave
the position vector r, by the vectorial distance of the two

nuclei R. If we denote the Cartesian components of R
again by x, y, z, and furthermore take into account that
only.
and
For a rotating molecule
-tfrot)-
to
momentum
When
pendent of the direction.
and we have seen that both parts have a sharp value
evidently symmetrical in the two particles and therefore ^ Tib (-R) is symmetrical. Our symmetry considerations
is
When
119
^rot
~~
We
therefore see that
Asymmetrical
for
lantisymmetrical
for
=
=
1.
For I > 1 the symmetry properties are similar. We had

mentioned on p. 42 that for d states {1=2) the wave
2
functions behave like x 2jR 2 , xyjR etc. They are evidently
,
DIATOMIC MOLECULES
120
DIATOMIC MOLECULES
again symmetrical. In the general case the symmetry
is
as follows
,
(symmetrical,
(antisymmetrical,
When we combine
with
(19)
principle into account,
(16)
= even
= odd.
even:
odd:
=
=
'
and take the Pauli
0,
(20)
1.
spin.
The
spin directions of nuclei are usually hard to change.

of collisions (or adsorption at a solid sur-
A large, number
face) is required before
a change of sign
When A is even, the total spin of the nucleus can

only be s
0, 1, 2,..., etc., whereas if A is odd the possible
values are s
, ,..., etc. The actual value cannot be
foretold but can be taken from experiment. We consider
two examples: (i) oxygen has A
16, and the spin of
14 and 5=1.
the nucleus is s
0; (ii) nitrogen has A
In both cases the total wave function of the two nuclei
must be symmetrical. Now the total nuclear spin of 2 is
zero. Thus the orbital wave function of
2 must be sym-
rule.
even and s = 1 (and also with I odd, 5

0)
would have a symmetrical total wave function and are
forbidden. The rotational levels with even I have only
the (21+ l)-fold orbital degeneracy, whereas the levels with
odd I have in addition a three-fold degeneracy due to the
Levels with
two neutrons the wave function changes sign, we have

A times. Thus, the wave function of the
two nuclei will be symmetrieal/antisymmetrical in the two
nuclei if A is even/odd. We see that for even A the wave
function of the two particles is symmetrical, in contrast to the
cases considered so far. The spin of a nucleus is composed of
the spins of the protons and neutrons, by the vector addition
or
find that the rotational levels

are connected with the total spin of the nuclei,
we interchange in reality A pairs of

Since for each interchange of either two protons
atomic weight A,
particles.
we
121
a change of spin direction takes
The two types of states behave almost as if they

belonged to two different kinds of molecules. They are
called para-hydrogen [s > 0) and ortho-hydrogen (s =
1).
metrical.
Hence we
find
place.
Para-hydrogen has only rotational levels with even l

t
ortho-hydrogen those with odd L In ordinary hydrogen
the two kinds of molecules occur in the ratio 1 3 because
ortho-hydrogen has three times as many spin levels (on
the average the ^-degeneracy is the same).
Finally we consider briefly molecules consisting of higher
atoms. A nucleus is known to consist of a number of
:
protons and. neutrons. The neutron has the same weight

as a proton and has also spin and obeys the Pauli principle. The number of protons plus neutrons is the
atomic
weight A. Also two such composite nuclei are 'identical
particles
When we interchange two like nuclei with
1
2:
only even
allowed.
(21)
Half of the rotational levels are missing altogether. For

nitrogen the total nuclear spin is obtained by combining
the two spins 5=1. The total spin can be 5

2, 1,
(compare the addition rules in Chapter IV, sections 5 and
Without actually proving the fact the reader will
6).
readily believe (from what has been shown so far) that the
spin wave function of the two nuclei is symmetrical when
the total spin is
2 or 0, and antisymmetrical when 5
1.
Thus the rotational levels are connected with the total
nuclear spin as follows:
N2
even,
or 2 (ortho -nitrogen)
(22)
odd,
5=1
(para-nitrogen).
DIATOMIC MOLECULES
122
The spin degeneracy

are 5
+ =
1
is
25+1. Thus the
levels
= even
6-fold degenerate {apart from the ^-degeneracy)
IX
the levels I
odd are three-fold. Ortho- and para-nitrogen
occur therefore in the ratio 2:1.
THEORY OF HOMOPOLAR
CHEMICAL BOND
The experimental facts underlying the above analysis

of the rotational levels of nitrogen have played an important historical role in establishing the fact that nuclei are
composed of protons and neutrons (and not of protons and
was believed before) and that the neutron has
spin \ and follows the Pauli principle.
electrons, as
The hydrogen molecule
One of the great puzzles confronting physicists and

chemists in the days before wave mechanics was found,
was the problem of the chemical bond. Roughly speaking,
there are three types of chemical forces which lead to an
attraction between atoms and molecules (although the
distinction of these three cases is not always quite sharp),
namely :f
(i)
(ii)
(iii)
As to
the hetero polar or ionic bond;

the homopolar or covalent bond;
the van der Waals forces.

(i)
there
is
no
understanding
difficulty in
ions with opposite charges, for instance
why two
Na + and
CI",
The van der Waals forces are usually

very weak compared with the true chemical forces; they
attract each other.
do not so much lead to the formation of molecules, but

are rather responsible for keeping the molecules of a liquid
together.
The
We shall not consider them here.
real
problem
lies
in
(ii).
On
the basis of classical
physics it is quite impossible to understand why two

atoms, form a molecule.
neutral atoms, for instance two
has
the striking feature of
bond
the
covalent
Moreover,
showing saturation expressed in the concept of valency.
f The so-called metallic bond is more complicated than the covalent

bond and leads to the formation of crystals rather than molecules. It
is also explicable by wave mechanics.
124
THEORY OF HOMOPOLAR CHEMICAL BOND
atom can bind one other H atom, but not two or

three. A C atom can bind four H atoms but no more.
Even if in classical physics sufficiently strong attractive
between neutral particles were known, it would be
quite impossible to understand why a third atom should
forces
not also be attracted by the two atoms already bound.

The feature of saturation is entirely foreign to classical
125
The formulae for E(R) will be very similar to

the formulae for the perturbation energy of the levels of
electrons.
He atom.
According to the general principles of the perturbation
theory we have first to find the unperturbed wave function
the
The
of the system.
electron of the
first
atom whose
physics.
The problem has been solved on the basis of wave

mechanics (Heitler and London, 1927). Wc shall see that
wave mechanics not only explains the attraction between
neutral atoms but also leads to a full understanding of the
saturation properties.
We start with the simplest case, the H molecule. The

most natural way to come to an understanding of the
attraction between two H atoms will be the following:
We consider the two atoms with their nuclei placed at
two points a and 6, say, at a distance R apart. Let the
two atoms be both in their ground states. The total energy
of the system will then differ from the sum of the internal
energies of the two atoms (i.e. twice the energy of the
ground state of H) by a certain interaction energy, E(R),
which will depend on the distance R. Only if R is very
large will this interaction energy be zero. If now the
atoms are really attracting each other E{R) must decrease
when R = oo to negative values when R

At small distances R it is to be expected that
E(R) increases again and becomes positive, for otherwise
no equilibrium distance would exist. Thus E(R) will be a
curve of the type shown in Fig. 30 or 3 1
We can calculate E{R) most conveniently by using the
from the value
is finite.
perturbation theory of Chapter VI. In fact the problem is

very similar to the calculation of the energy-levels of the
He atom, for, as in the He case, we have to do with two
Fig. 32. Charge clouds of two
atoms.
nucleus is at a, say, is described by the wave function a (l)

given by equation (3) of Chapter III, p. 33, namely
^r (l)
e -an apart from the normalizing factor. | rla is
a
here the distance of the electron from the nucleus which
t/
is
placed at a.
We have labelled this electron as number
1.
atom
the electron of the second
(whose nucleus is at b) number 2, then the electron of the
second atom is described by the wave function $b (2). ^ b is
the same function as U only it depends on the distance r2b
of the second electron from the nucleus at b. The charge
Similarly, if
we
call
cloud of the second electron is just shifted by the distance

R to the point b as its centre. Fig. 32 shows the two charge
clouds schematically. Except when R is very large the two
charge clouds partly overlap. For the wave function of the
t The a occurring in the exponential is the reciprocal of the Bohr
radius and has, of course, nothing to do with the o labelling the nucleus
of the first
atom.
126
whole system we have then to take, as before, the

product
0O)lM2).
The
is now that, just as in the case of

the
obtain a second wave function by interchanging the roles of the two electrons. The second
wave
function is ^(2)^(1 ); it describes exactly the
same system,
namely two
atoms at a distance 8, only the electron at
a is labelled number 2 and the electron at b number 1.
We
have seen in Chapter V that in this case we are
forced to
take the symmetrical and antisymmetrical
combinations
of these two wave functions in order to conform
with the
fundamental fact that the two electrons are indistinguishable and cannot be affixed with names.
Thus we have the
following two undisturbed wave functions:
crucial point
He atom, we
0+
^W
(2)+&,<2)^(1),
(la)
Now
(2) gives the Coulomb interaction between the two

atoms provided that the electron at the nucleus a is
number
The wave function
however, conis at a and

number 1 at b. Clearly in this case the interaction between
the two atoms is not (2) (e 2 / r 2a * s * nen *^ e internal potential
energy of the atom a), but an expression obtained from
(2) by interchanging 1 and 2:
labelled
tains also parts
1.
(1),
where the electron number 2
y_
a_j^_1
"
_,
r l2
(2')
r la
r 2b
Thus 7 is to be used as interaction for the first part of

the wave function i/r(lty/ 6 (2) and 7' for tp a (2)tfjb (l).
The perturbation energy of the two atoms is now given
by Chapter VI, equation (2), p. 88,
(16)
jv^
In order to find the perturbation energy E(R)

we have
to take the square of our wave functions
(1), multiply
by the interaction function 7, and integrate over

all
the coordinates of the two electrons. Now
what is this
interaction function 7? In the He case
7 was simply the
Coulomb interaction of the two electrons e 2 /r
In our
12
127
E(R)
(3)
s
Or
'
J"
it
case the unperturbed potential energy

is, however, only
the interaction between the electron 1 and
the
nucleus a
and between the electron 2 and the nucleus
6, provided
the electron 1 is attached to a and
2 to b. But, clearly,
the electron 2 also interacts with the
nucleus a and also
1 with b, and finally the two nuclei also
interact with each
other.
All these interactions are part of the

perturbation
in our case is
between the two atoms. Thus 7
ft
r,.,
r 1h
a
where the integration is over the coordinates of both

and the '7' to be inserted is either V equation
electrons
equation (2') for the two parts of the wave

(2) or
function (1) respectively. We write out the numerator
explicitly
Yp dr = f 7^(1)^(2)^^ + / Pftmfti)*^
2 J 7^(1)^(1)^(2)^(2)^^.
In the
last
term of (4)
it
(4)
may be doubtful whether to insert
or 7' because here the product of the
two parts of the

wave function occurs and the electron 1 is 'partly' at the
nucleus a and partly at b. This, however, makes no dif-
(2)
ference, for the last integral of (4) does not
change
if
we
128
instead of F. To see this we only remark that

a change of notation of the variables over which we
integrate does not change the value of the integral. If we
insert
then interchange the notation of the two electrons, 1 -> 2

and 2
1 the product
n ( 1 )0 (2)06 ( 1 )<6 (2) does not change
other hand,
smaller than
therefore
fl
and V goes over
and therefore the integral does

not change if we replace F by V. In the same way we
see that the first two integrals of (4) are equal.
Thus, using the same abbreviations as in Chapter VI,
into F',
is,
1.
It
O =
6,
except for very small distances,

would be equal to 1 only if R
thus
= j #(1) dr
02(2)
a\ 2
(R
129
much
and
0).
The 'overlap integral' 8 therefore only plays the part of

a more or less minor correction (the denominator does not
differ much from unity), and we can ignore it for a qualitative discussion.
We finally
obtain for our interaction energy of the two
H atoms the two formulae

p JL
E+ (R) = O1+8
(8a)
*-CT
(86)
(6)
We may
the numerator of (3) is 2C2,A.

the wave functions of the two
assume that
atoms are normalized
/ 05(1)
î
J 02(2) dr z
1.
The denominator of (3) then
becomes 2 28, where
= / 0a(l)0 (l)0a(2)0
Unlike the case of the
He
atom,
6 (2)
dr.dr,.
where
Jâ (l)06 (l)
and
dr x
(7)
does not vanish here.
The orthogonality theorem (Chapter VI,

says that
section 3) which
applies only to the case
are two different
wave functions of the same

two different stationary states, but in
our case ip a and ^6 are wave functions of two different atoms
put at a distance R apart. Two such wave functions are
not orthogonal. Looking at the charge clouds as shown
ijj
tf>
atom belonging
to
we also see that the integral of the product of

the two wave functions does not vanish unless the two
in Fig. 32
atoms are so
overlap at
far apart that the
all.
where G, A,
S are the integrals
integrals depend, of course,
two charge clouds do not

E(R) vanishes.) On the
(In this case also
given by (5)-(7). All these
on R.
We see that, as in the case of the He atom, we obtain two

We know from Chapter V
different interaction energies.
that the wave function of two electrons is always symmetrical in the coordinates of the two electrons if the two
electrons have antiparallel spins, and it is antisymmetrical
if the two electrons have parallel spins (see Table 5, p. 84).
Now the interaction + {R) belongs to the symmetrical
combination (la), whilst E_(R) holds for the antisymmetrical wave function (16). Thus, if the two electrons
have antiparallel spins (^ state) their interaction energy
is
E+ (R).
If the spins are parallel
tion energy
atom, we
is
E_{B).1[
3
(
state) the interac-
Again, as in the case of the
He
obtain two widely different interactions, their
atoms are in S states it is evident that only E states

f Since both
can be formed for the s system (see Chapter VIII, section ]).
6881
130
difference
is roughly 2A. The reason

for this difference
again the exchange phenomenon.
We
is
now discuss our result: The integral

(5) can
easily be interpreted. 0(I) is the
charge density sur-
rounding the nucleus a and

$?(2) that surrounding the
nucleus b. C therefore represents the Coulomb

interaction
of the charge cloud round a with the
nucleus b (term
e2A*ifc)> that of the charge cloud
round b with the nucleus
a, the interaction between the
two charge clouds (e 2/r12 ),
and finally the interaction between the two
nuclei. If
we are at all to think of a classical picture that might
give
us an idea of the interaction of two
neutral atoms, we
should think that C would be the interaction
energy. But
we shall see that not C but
is the chief contribution
and that the formation of a molecule rests essentially
on A
A does not permit such a simple interpretation as C.
As we have seen, it is due to the fact that the two
electrons
are indistinguishable and can be exchanged.
However, if
we wish, we can, as in Chapter VI, section
2,
define
an
exchange charge density p

2
a (l)fc(l). The term e /r
of
A represents, then, the Coulomb interaction of 12this
exchange charge density with
itself (e 2
gtl)p(2)
^^\
and the other terms may be interpreted as

the'interaction
of the exchange charge with the nuclei,
multiplied, though,
by the total value of this exchange charge
'fp(2)**ra f*L\fr
is
is
minor importance), and in parThis is not quite

so easy as in the case of the He atom. In C as well as in A
both negative and positive contributions occur. The sign
of C and A depends on whether the attraction of the charge
clouds by the nuclei outweighs the repulsion between the
two nuclei and between the
y
and S
(the latter is of
ticular to decide
what
their signs are.
charge clouds themselves. The

chief features can, however,
easily be seen from a look at
Figs. 32
and
33.
In the first place, consider

very large distances R. The
exchange charge density p is
then zero, because a and b
Fig. 33. Exchange charge.
do not overlap and a {l) is very
small everywhere where 6 (l) is large and vice versa. Also
Consequently both C and A
all r's are then very large.
ifi
ifj
tfr
</r
must be expected.
Next consider medium distances, say of the order of
magnitude of the Bohr radius I fa. At such distances the
two charge clouds (Fig. 32) just overlap nicely because
vanish, as
I fa is also the distance from the nucleus at which the

charge cloud is largest. First look at Fig. 33. Owing to
the overlapping the exchange charge density is then quite
large. We also see that certain parts of this exchange
charge density are quite close to the nuclei and are there-
by them, giving rise to a large

negative contribution to A. On the other hand, the
average distance of the different parts of the exchange
charge density is quite large (and so is the distance between
the nuclei). Thus the positive contributions to A due to
the repulsion of different parts of the exchange charge
fore strongly attracted
the contribution from the term e 2 /r

16
In Fig. 33 the distribution of the
exchange charge
density
131
shown schematically.
In order to decide now whether the two

atoms attract
or repel each other we have to
evaluate the integrals C,
with each other (and also of the two nuclei) will be
132
relatively smaller
than the negative contributions. We

thus conclude: For medium distances A is negative. Looking
at Fig. 32 we see that for the same
the charge cloud 2,
for example, is on the average farther away
from the
nucleus a than is the exchange charge from one
of the
nuclei at the nearest point, and also the two charge
clouds
are relatively far away from each other. We expect
THEORY OF HOMOPOLAR CHEMICAL BOND 133

plotted separately in Fig. 34. We see that E^(R) is always
and therefore no stable molecule is formed if the
E+(R) is indeed negative and far
two
larger than C (the difference between E + and C is approximately A). E+(R) has a minimum at a distance Ra 1-6
positive
spins are parallel.
there-
fore that
distances.
and
C is numerically much smaller than A

The
sign
positive for
is
attraction,
medium
of E(R) is therefore determined by A
E_ and negative for E+ E+ means
for
E_ repulsion of the two H atoms.
Finally, for very
small distances R the two nuclei are very close

to each
other and repel each other strongly. The two
charge
clouds overlap completely, but are, of course,
always a
distance I/a
tribution to
away from both
Far the largest conC and A arises, then, from the Coulomb
repulsion of the two nuclei. The factor e 2jR can
be taken
out of the integrals and these cancel in the
numerator or
denominator of (8). Thus for small R, E (R) = e 2 /R,
in
+
both cases, the atoms repel each other strongly.
To sum up: Coming from large distances the two H atoms
nuclei.
attract (repel) each other if the spins

of their electrons are anti-
parallel (parallel).
energy
E + (R)
has a
For
Fig. 34. Interaction of two
antiparallel spins the interaction
minimum
at
the order
of magnitude of the Bohr radius. For smaller distances the
attraction goes over into repulsion.
These qualitative considerations are fully borne out

by
the calculations. The evaluation of the integrals
O, A, S
presents no difficulty but is rather cumbersome.
We give
the results in form of a graph. In Fig. 34
we have plotted
the energies E b (R) and E_(R) as a function
of the distance
R using as units the Bohr radius 1/a. As units for the energy
we use electron volts (for comparison remember that
the
ionization energy of the
atom is 13*5 eV).
is also
atoms.
some distance of
and for smaller R, E+ becomes positive. If, therefore, the

two spins are antiparallel a stable combination of the two H
atoms is formed, and this means the formation of a H 2
molecule.
The minimum of
E+ (R)
gives the equilibrium
and the binding energy (or dissociation energy) of

the H molecule. For comparison we have also plotted
(dotted curve) the interaction curve of two H atoms
as it is known from various experiments. We see that
distance
the theoretical curve fits the experimental one rather

well. Table 9 gives the numerical data for the equilibrium
134
distance, the binding energy,
and the frequency of vibranear the minimum, (See Chapter VIII, section
3.)
The agreement is as good as can be expected. We must
not forget, of course, that we have used a perturbation
method which only gives approximate results. There are
more powerful and more complicated methods to calculate
tion
a>
Table
9.
molecule {theoretical and experimental)
135
As we have seen, the covalent bond rests on the quantummechanical exchange phenomenon. In view of the importance of this effect and the difficulty it represents for an
understanding, we discuss now the question of what precisely is the physical meaning of this exchange. We have
emphasized in Chapter VI that it cannot be fully understood by the picture of interacting charge clouds (which
otherwise
is
useful) because
occurs with
two
different
a typical quantumsigns. Above all, the exchange

finding a 'classical
at
attempts
mechanical effect and all
from the large
apart
Nevertheless,
analogue' must fail.
aspect of the
another
is
there
contribution to the energy,
is
Dissociation energy,
eV
Equilibrium distance,
Ra
Frequency of vibration
the binding energy of the
But
Exp.
Theor,
4-5
3-2
1-4
1-6
8-1
90xI0 u sec. _1
molecule with great accuracy.
was not our purpose. We set out to understand

why two H atoms attract each other and form a molecule.
This we have achieved:
The reason for the formation of a molecule is the quantumthis
mechanical exchange phenomenon, and the bulk of the

binding energy is the exchange energy A.
When two
atoms meet, a molecule is not always
formed. The two atoms repel each other if the spins of
the two electrons are parallel. Indeed, if two
collide,
and if we do not know
the chance
atoms
their relative spin directions,
only 1:4 that they will attract each other

repel each other. This is so because
(Chapter V) there are three spin wave functions for the
triplet state but only one for the singlet state. The
existence
of a repulsive interaction between two
atoms has also
been found experimentally.
is
and 3:4 that they
The 3 molecule has, according to our theory, a spin zero.

This also is in agreement with the facts, for
2 is known to
be diamagnetic and to have no magnetic moment.
exchange phenomenon that might be helpful for its understanding. In Chapter I we have seen that to each energy
E a frequency Ejh is attributed. We may ask what is the
meaning of the frequency attached to the exchange energy,
freviz. Ajht Is there anything that vibrates with this
the
that
think
to
tempted
be
might
one
first
quency ? At
frequency
a
with
places
exchanging
really
are
two electrons
this
Ajh, since A is due to the exchange of electrons. But
phenomenon
exchange
the
of
basis
very
The
correct.
is not
indistinguishable and
is the fact that the electrons are
an exchange of electrons is unobservable in prinAjh can, though, be interpreted as the exchange
therefore
ciple.
frequency of spin directions. We can only describe tins

qualitatively, without proof. We may imagine that by
means of some experiment we have obtained a knowledge
that
of the individual spin directions of the two atoms, and
the
of
spin
the
we found that at a certain instant of time
electron at a
was
(i.e.
m =
sz
+\) and that
at 6
j.
Once
these individual spin directions are determined, this does

not mean that they will remain the same in course of
time. They will in fact exchange. We may perform the
same experiment a
little later
and
find,
with a certain
136
probability, that the spin at a is

J and that at b f The
probability is largest after a time A/4h. The change
is
periodical and the frequency of exchange is just Ajh.
The
electron of
frequency Ajh
is
He which
has the same spin direction
137
{Fig. 35).
Now we have seen in the case of two atoms that the

exchange of two electrons with parallel spins leads to
thus the frequency of exchange of spin
directions.
2.
The saturation properties
r
of the chemical
bond
it
In the preceding section we have seen that neutral atoms

can attract each other to form a molecule. It remains
now
Fig. 35. Interaction of
repulsion.
proof will follow in Chapter XI.

We consider first a simple example: the interaction of a
He atom with a atom. He in its ground state has two
electrons, both in the lowest level. According to
the Pauli
principle the two electrons must have antiparallel
spins
and are in a l S state. If now the
atom is brought near so
that interaction is taking place, the exchange phenomenon
will
We conclude therefore: A He atom and a H atom

We expect that the interaction energy
be given by a formula of the type E_(R) (86),
B{R)
This consideration can at once be generalized to all rare

For the electron configuration of all rare gases consists of closed shells only, with all electrons arranged in
pairs with antiparallel spin. Therefore a rare gas atom
repels any other atom. This is in agreement with the
chemical behaviour of the rare gases, which are known

not to enter into chemical compounds with any other atom.
atom
As a next step we consider the interaction of a
with a 2 molecule which we assume to be already formed.
As long as the
atom is at a distance much larger than
the equilibrium distance of
itself we may consider the
2
the fact that it consists
molecule
as
a
whole
and
ignore
2
atoms. The electron state of 2 is now, according
of two
electron could be exchanged with the electron of

He
which has a spin J, then a state of the He atom would arise
where both electrons have the same spin directions, namely

ff, but this is forbidden by the Pauli principle. It follows
(9)
is''
gases.
that the
= C-A
and the actual calculations show that this is the case,

where, of course, C, A, S are not quite the same as in the
case of two H atoms but are of similar type and have
the same qualitative properties.
come into play. In contrast to the case of two

atoms, however, there can be only one mode of interaction.
For the vector addition rule for the spins shows at once
that there is only one value for the total spin of the system,
namely. , giving a 2 2 state for the He-H system. Is this
mode of interaction, then, attraction or repulsion ? We
can decide the question in the following way. Let
the
spin of the
electron be, for example,
At first one
f
might think that this electron could be exchanged with
both the electrons of He. But this is not the case. If the
Ho and H.
repel each other.
will
He
show that these chemical forces have the saturation

properties known from the chemical facts. In this section
we show this in a qualitative way, a more quantitative
to
electron can only be exchanged with the one
138
to the preceding section, a singlet state with the two electrons forming a pair with antiparallel spin. It resembles
therefore the electron configuration of He (at least as far
as the spin is concerned). The electron of
can only be
VALENCY
exchanged with the one electron of H 2 which has the same

spin direction. It follows, then, that the H atom is repelled
by the
molecule.
This, then,
1.
repelled
by
an account of the valency and chemical binding

The first and most important step in this direction will be quite easy and straightto give
properties of the higher atoms.
it.
This consideration applies only as long as the

atom
is comparatively far away from the
If the
2 molecule.
distance between
and 2 is comparable with the mole-
forward.
H
of H
As we have seen
in the two preceding sections, a moleonly formed by two atoms if each atom can contribute an electron with a free spin so that a pair with
antiparallel spins can be formed between the two atoms.
If all the electrons of one atom are arranged in pairs within
the atom (rare gases) so that the total spin of the atom is
zero, no other atom can be bound. Moreover, as we have
seen from the case of three
atoms, an atom with one
free electron is capable of binding one other atom with a
free electron, but no more. We therefore conclude that the
valency properties of an atom must depend on the number
of electrons with free spins (i.e. not used up in pairs).
In what circumstances may we then expect that an
atom A, say, could bind two
atoms? Obviously, if the
atom A has two free electrons which have or can have
parallel spin, i.e. a total spin 1. The two
atoms can
then be bound if their electrons arrange their spins in the
direction opposite to that of A So just one pair between
A and each atom is formed. Generally, we expect that
an atom with n free spins can bind n
atoms. Fig. 36 (a)
illustrates this for n 3, for which nitrogen is an example.
Much the same considerations apply if an atom with
cule
cular distance
itself, the situation is more compli2
cated and will be considered in Chapter XI, section 2,
where the above conclusions will be verified by explicit
calculation.
The repulsion of closed shells is also of great importance

The ions Na + and CI- both have the
same electron configuration as a rare gas; the only difference is that they are not neutral. They attract each other
at large distances on account of the Coulomb forces which
for the ionic bond.
is
is
have a very large range, much larger than the exchange

forces. At close distances they repel each other on account
of the exchange forces of the closed shells. What is commonly called the 'size' of an ion is due to this repulsion.
The exchange forces become strong very rapidly once the
appropriate distance
Spin valency
We must now generalize our considerations of Chapter IX
the explanation of the saturation properties.

A 2 molecule, once it is formed, is no longer capable of
binding a third
atom, but any further
atom is
is
reached.
VALENCY
VALENCY
valency n forms a molecule with another atom with

valency m. Then n or
pairs can be formed between the
two atoms according to whether n or m is smaller. A
saturated molecule is formed if n
m. Fig. 36(6) shows
this for the
molecule
with
n
3.
Every saturated
2
molecule is expected to have spin
and to be therefore
diamagnetic. (For exceptions see the following section.)
which can arrange their spins in parallel. f In Chapter VII,

we have seen that the lowest state of each configuration is usually that where the spins are, as far as
possible, parallel. Thus the valency of the ground state
(defined by (a)) will be identical with the, valency of the
lowest configuration (defined by (6)). The states i S, 2D, 2 P,
of nitrogen have valencies 3, 1, 1 respectively, and the
140
section 2,
valency of the
D
>
cr
t)
1)
ct
4)
(t
1)
(31
We thus see: What in

an atom
NH
chemistry
is
and N.
called the valency of
is the
called the valency against hydrogen.
What
is
called
bond between two atoms is a pair of electrons with antiA double bond means two pairs formed, etc.
The term 'have or can have parallel spin' requires a little
more precision. If an atom has several electrons in the
outermost shell, various atomic states arise from this configuration. For example, nitrogen has three
p -electrons,
giving rise to the states *S, 2Z>, and 2 P. We can then
define (a) either a valency of each specific atomic state,
parallel Bpin. This
directly connected with the total spin of this state

(s
n/2) or the multiplicity ((+ l)-tuplet). Or else, (b), we
is
can consider the whole configuration and define the valency

of the configuration as the
maximum number
is 3.
what extent excited
states (or
we can
read off the valency of an atom directly from the spin of the
ground state. In Table 7, p. 104, we have given the ground
states for the atoms of the first period of the periodic table.
In Table 10 we give the valency of these elements, as derived
from the ground states by the above rule, n should then
be the maximum number of
atoms that can be bound.
In the last row we give the molecules of the type AH with
the maximum number of
atoms as they are known to
exist in chemistry.
Table
parallel spin.
namely the number of electrons n with
configuration
discuss to
selves, to start with, to the lowest configuration,
number of free sjiins of the atom or the total

spin multiplied by 2. More precisely, this number is what
is
3,
excited configurations) can be essential for the formation

of molecules even for their ground state. Confining our-
(b)
Fig. 36. Spin valency:
We must still
&
141
of electrons
10.
H
Total spin
Valency n\
from
ground
state
Exp.
Valency of elements of
the first
period
1/2
1/2
1/2
3/2
1/2
BH
[CH 4 ]
He
Li
Be
Ne
AH
LiH [BeH]
NH OH
3
FH
We see that the observed valencies agree in general

with the expected ones, with two exceptions (indicated
We
shall see below (section 3) that for the treatment of the intert
action of such atoms also two alternative starting points are possible,
corresponding to the two views (a), (6).
VALENCY
142
by square
VALENCY
namely Be and, most important, C.

Carbon is very well known to have valency 4 (the whole
of organic chemistry rests on this fact), whereas its ground
state is 3 P with only two free electrons. In order to explain
a valency 4 we should expect a state for the C atom with
four free spins, i.e. a quintet state. Now we have seen in
brackets),
Chapter VII, section 2, that such a state really exists. It is

a *S state. But this quintet state is not the ground state
of C, it arises from the excited electron configuration sp 3
It will be seen in the following section that this excited
143
Crossing of atomic interaction curves. Valency

carbon
2.
of
The valency 4 of carbon can only be understood as due

b S state. One might think that
it would be
to the excited
first to spend the energy of excitation (about

eV, according to Table 8, p. 105) before the four
necessary
4-2
state
We
really responsible for the four valencies of C.

are thus led to consider not only the ground states
is
of the atoms as responsible for their chemical properties

but also their excited states. This will be of particular
importance if an excited state has a higher valency than

the ground state. In the following sections we investigate
the influence of the excited states, and the case of carbon
be treated in detail as an example. (The case of Be
very similar.)
will
is
In addition to the
an atom has also in

underneath. These would always repel
free electrons
general closed shells

the other atoms. Since they are much closer to the nucleus
their influence will only be noticeable at small distances.
At the equilibrium
distance they will at most be able to

diminish the binding energy of the molecule somewhat
but not hinder the formation of a molecule.
Long
before
wave mechanics was known Lewis put
forward a semi-empirical theory according to which the

covalent bond between atoms was effected by the formation of pairs of electrons shared by each pair of atoms.
We see now that wave mechanics affords a full justification
of this picture, and, moreover, gives a precise meaning to
these Lewis's electron pairs: they are pairs of electrons
with antiparallel spin.
Fiq. 37. Atomic interaction curves.
atoms can be bound. But this will be seen not to be the

case. In order to understand the valency of carbon properly
we
investigate in this section the influence of the excited
states quite generally.
Consider two atoms A and B both in specified atomic

When these atoms are brought into contact they
will interact in various ways, according to which of the
molecular electronic states is formed. The variety of states
which can be formed has been derived in Chapter VIII,
section 1 Suppose we calculate the interaction energy by
means of perturbation theory, much in the same way as
we have done for two
atoms. Some of the modes of
interaction will be attractive, some repulsive, according to
how many electron pairs with antiparallel spin are formed.
It may now happen that an attraction curve starting
from excited atomic states crosses a repulsion curve (or one
states.
with weak attraction) starting from the atoms in their

ground states. In Pig. 37 such a case is indicated and the
VALENCY
VALENCY
144
crossing region
for instance,
dotted. In particular this will happen if,

atom in an excited state (A*, say) has a
is
an
larger valency than in its
ground state and is able to form

more pairs with its partner.
At the crossing-point anew degeneracy would arise. There
would be two different states which at the particular distance have the same energy.
Now, in our previous exposition we have seen that
whenever a degeneracy occurs there is a special reason for
it.
An
atomic
electron
moves
state
is
with the spacing of the atomic energy-levels. Only if the

atomic interaction curves are far away from each other
can we derive their qualitative behaviour from pure valency
considerations. In a crossing region a new effect is taking
place which, in
with different energies. In our case there
is
for the degeneracy at the crossing-point.
We
no such reason
expect therefore that in reality no such crossing-point can exist (with the
exception of the case mentioned below). In other words,
the degeneracy in the would-be crossing-point
removed and the two states split up.
we
are led to such a crossing
is
de facto
by valency arguments
of the kind used above or by calculating atomic interactions with the help of the perturbation theory of Chapter
VI this only shows that in the neighbourhood of the
crossing region this form of perturbation theory is in-
adequate. This latter only holds if the perturbation is small,

in particular only if the state whose perturbation we wish to
far away from all the other states. More precisely: the perturbation energy must be small compared
calculate
is
cases, modifies the results of section
1.
three-fold degenerate because the
IE
field
must also exist and they must belong to the same energy.
The same applies to the two-fold degeneracy of a molecular
IT state, the reason being the axial symmetry of the molecular field. Whenever this symmetry is removed, for
instance by an external magnetic field, the degeneracy is
also removed and the state splits up into several states
If
some
II
of spherical symmetry. The wave

function in this case (Chapter III) is xf(r), but on account
of the spherical symmetry wave functions yf[r) and zf(r)
in
145
(a)
Fig, 38. 'Crossing' of atomic interaction curves.
We
can describe qualitatively what must happen in the

would-be crossing region thus: To avoid actual crossing the
curves must give place to each other and deviate in such a
way that they do not come very close. They virtually
'repel' each other. Beyond the crossing region their mutual
influence diminishes again and the curves join the original
branches. However, the first curve must then join what
was originally part of the second curve and vice versa.
This is shown in Fig. 38 (a): the dotted curves would cross,
but the real curves as they actually are in nature, are the
full-drawn ones.
A further point concerns the
wave functions. Approxiproducts of atomic wave functions.

Suppose the branches I and II of Fig. 38 (a) belong to
mately, these are
still
where
A*
functions ^/fe and t^tftg

state of atom A, for example. If we ignore
excited
is an
the mutual influence of the two curves, branch IV is a
continuation of I, and III of II. In IV, therefore, the
not A $B The true curves start
wave function is
wave
respectively,
^./fe,
from I
5831
(or II), pass
tfr
through an intermediate region but

L
VALENCY
VALENCY
146
join the branches III (or IV) again.
Thus
B must go
over gradually into $1*^, and tpA *^ into v^^r This
B
means that in the intermediate region the wave functions
must be linear combinations
& i'B+on A 4>u

A &. 4>b+c'ii & fe
ci
*f*
(for
i-in)
(for
n-ivj j
ifr
1
*
(1)
The
coefficients c I} ci, etc., change with the distance, as we

pass along the true curves. Evidently, for large distances
f
for the upper curve, and c\
1, cn
0, c n
1,
&l
for the lower curve. At small distances, beyond the crossing
region,
portion
Cj
IV
0,
is
cn =
n=
1, and c^
0. Although the
1, c'
reached from atoms in their ground states
(A and B) directly the wave function is here ^, $g. This,

is only true if we have moved sufficiently far
away from the crossing region.
of course,
All these conclusions can, of course, be verified
by
calcu-
a simple improvement of the perturbation

theory of Chapter VI suffices to treat the above effect
quantitatively, but this we need not carry out.
There is one exception to the above conclusions and
this is if the two interaction curves in question belong to
different 'races' (see p. 111). Then they can really cross.
This can be seen in the following way. Supposing we start
from the branch I of Fig. 38 (a). The spin and angular
momentum are constant numbers which cannot change
suddenly if we slowly change the distance of the atoms.
Coming to the crossing region we meet a state with different
spin or angular momentum. If we were to deviate on the
full-drawn line to branch III, we would have to change
either the spin or the angular momentum because branch
III is the continuation of the curve II which belongs to a
different race. Therefore the above effect can only happen
if the two curves belong to the same race.
Curves of
lation; in fact
different races are, in a
may
147
way, foreign to each other; they
accidentally cross, then an 'accidental degeneracy*
by symmetry reasons) occurs.

The above effect even takes place if the two curves would
only come near to each other without actually crossing.
(not caused
Their distance is enlarged (Fig. 38(6)). Thus, molecular

attraction curves are further deepened (giving still stronger
attraction) if, from some excited atomic state, a curve
starts belonging to the same race. We now see the reason
why we obtained too small a binding energy of the hydrogen

molecule in Table 9, p. 134. This is due to the influence of
the higher atomic states.
It is now easy to understand the four valencies of carbon.
atoms have all
For simplicity we assume that the four
the same distance R from C and are brought close to C
simultaneously. The total energy of the system is then
only a function of one distance R. The ground state of C
3 >
2 z
is f (s p configuration). In this state the valency of C is
2 and the four hydrogen atoms cannot all be attracted.
In fact, if they are all brought in contact with the C atom
simultaneously they are attracted only very weakly (as a
calculation shows).
We
consider in particular the state
where the total spin is zero. Thus from the ground state
of C an almost horizontal curve starts. Of all the excited
&
5
states we need only consider the S. The S state has four
atoms
valencies and is therefore able to attract all four
strongly. This attraction curve, which naturally also has
3
spin 0, is now bound to 'cross' the curve starting from P.
The
situation
is
similar to Fig. 38
(a).
Instead of crossing,
the curve starting from the ground state is pushed down

and leads to a stable molecule, whilst the curve actually
from B S remains high.

Although it is the decisive influence of the excited 5 $
atoms, the
state that makes it possible to bind four
starting
VALENCY
148
VALENCY
atoms when removed from the carbon will leave the C atom
in its ground state. And vice versa: When four H atoms
collide with a C atom, either simultaneously (which in
practice does not occur) or one after another, it is by no
means necessary first to excite the C atom to the 5 8 state.

They can all be bound one after the other.
Similar to C is the N + ion. It has the same electron configuration as C and therefore four homopolar valencies. In
can bind a negative ion in a heteropolar bond.
In this way molecules like (N+H 4 )C1- can be formed.
Other examples for the mutual influence of such molecular interaction curves are the molecules C and
2
2
From section 1 we should expect that all saturated molecules are in singlet states. Whilst this is in general found
to be the case, these two molecules are exceptions. C is
2
in a 3 IT and
in a 3 2 state. In fact, oxygen is well known
2
to be paramagnetic, which means that it has a permanent
magnetic dipole moment and must therefore have a spin
addition
1*9
becomes the ground state of C 2 More detailed considera3

II {and not
tions are needed to show that it is actually a
.
or 3 A) state.
should be mentioned that ionic states may

energy, even in a
also contribute to increase the binding
energy is
ionization
the
Although
molecule.
homopolar
Finally
it
3P+5S
it
3p+3p
different from zero. We consider briefly as an example

the case of C 2 If both C atoms are in their ground states
3
the strongest attraction curves should be the singlets,
while the triplets with only one pair formed should lie
higher but should also be slightly attractive. The low
.
excited states of C are l D, 1 S and & S. All these when

interacting with the second C atom in 3
give rise only to
triplets (or higher multiplicities).
In particular, from
i
jS+ 3 P a triplet curve with strong attraction should arise

because two pairs are formed. This triplet curve will cross,
or at any rate come near to, the triplet with weaker attraction starting from 3 P-\-*P. The result is that the latter
curve is strongly pushed down even below the singlet curve
3
from 3
P. This is shown qualitatively in Fig. 39 (only
the essential curves are drawn; many more curves start in
fact from the atomic states in question). A triplet then
5
P+
Fig. 39. Molecular states of Cj.
by
then usually high this is to some extent compensated
come
may
curve
ionic
the
and
ions
the
the attraction of
down deep enough to strengthen the bond of the homopolar curve.
above modifies
It is seen that the crossing effect discussed
in
considerations
valency
pure
from
drawn
the conclusions
order
the
circumstances
special
in
changes
It
some
cases.
to increase
of the molecular terms and always contributes
in the
valency
a
replaces
never
it
But
the binding energies.
posmolecule
a
of
formation
the
make
sense that it would
particular,
In
binding.
electron-pair
proper
sible without
moderately low
the crossing effect makes it possible that a
its full part
play
can
valency
higher
with
excited state
In broad
spent.
being
first
energy
excitation
without an
an atom are
outline, therefore, the chemical properties of
the
few lowest
valency
spin
of
maximum
the
determined by
In this way the
agreement with the facts.
states of the atom.
in full
results of the theory are
150
3.
VALENCY
VALENCY
We call this the
Interaction in diatomic molecules
It will be useful to corroborate the

considerations of the
preceding sections, especially of section
1, by explicit calculation. Before we do this we
must first make some remarks
about the approximations to be used.
For atoms with

more than one electron there are essentially
two different
approaches: The most direct way to treat
the interaction
is to consider the two
atoms in their ground state (or in
another specified atomic state) and
calculate their interaction energy. We may use the
exact wave functions of
each atom. We call this the procedure
(a). We have seen
with the exception of carbon, the

ground state of the
general the correct valency, and we
expect to
understand the formation of the molecules
observed
that,
atom has
in
chemistry. However, in this way

we neglect the influence
of the excited states which
(section 2) is negligible only
when the curves starting from these states
are sufficiently
far away from the molecular curve
considered. For large
atomic distances where the interaction
energy is still small
this is always the case.
On the other hand, the low excited states
(again with
the exception of carbon) all arise
from the same electron
configuration and their energy
difference is due to the
interaction of the electrons within
the atom. If we neglect
the latter (to start with) several
atomic states will coincide
and the electrons in the outermost
shell will be free to
arrange themselves and their spins
(as far as the Pauli
pnncipie permits) in this shell. Let now
an atom in such
a configuration interact with other
atoms. In the lowest
molecular state the electrons will then
arrange themselves
so as to give an optimum
binding rather than form
specific
account
atomic
all
In this way we evidently take into

the atomic states of the same
configuration
states.
151
procedure (6). It is a good approximation

he very close so that the interaction
states
atomic
the
if
ground state and the excited states
the
from
starting
curves
of the electrons within each
interaction
The
close.
are also
atom can be taken into account as a perturbation but the
unperturbed wave functions of each atom are used (i.e.

products of single electron wave functions).
A great advantage of {b) is that this method easily
accounts for the directional properties of the valencies.
These are, at least partly, due to the influence of the excited
states. We shall treat them in Chapter XI, section 3.
The truth usually lies somewhere between (a) and (6).
Better than both (a) and (6), but also more complicated,
is
the procedure described in section
2:
consider the curves
starting from specified atomic states and take their mutual

influence as well as the correct excitation energy into
account But this we shall not carry through quantitatively.

For many atoms, for example N, (a) is a sufficiently good
approximation if we do not insist on very accurate results
and are mainly interested in the binding properties, although the excitation energies are of the same order of
magnitude as the binding energies of the molecule. The
reason is that the excited states have smaller valency and
give rise to repulsion. Thus the attraction curve starting
from the ground state and the repulsion curves starting
from the excited states are far away from each other in the
.
neighbourhood of the equilibrium distance.

Special considerations are needed for carbon. The excited
5
S state is all-important but the other states arising from
the configuration sp* he very high and (6) would be a poor
approximation. For molecules which require only two (or
one) of the four valencies of carbon (like CH, CH 2 and C 2 )
,
b
the co-operation of 3 P and S is essential. On the other
hand, for molecules in which three or four valencies are
VALENCY
VALENCY
152
needed the procedure

cannot be too accurate,
(a)
(applied to
S) t although
it
may well be used, provided we are

only interested in the stable molecule. We have seen in
section 2 that beyond the region where the & S curve
would cross the curves starting from 3 P, i.e. near the
potential minimum of the molecule, the wave function of
the carbon atom is close to that of the 5 S state. Thus the
part of the interaction curve describing the molecule can
be calculated by taking into account only the & S state, but
it must be borne in mind that this is not permissible if we
study the asymptotic behaviour for large distances.
Let us then use in this section method (a). Let a, b be
two atoms with n,
valency electrons with parallel spin.
The wave functions of a and b are then

â (l,...,n) and
|M4- If .,*+**)> where we have numbered the electrons
of a 1 to n those of b n+1,..., n+m. Since the spins in
each atom are parallel their spin wave functions are entirely symmetric and thus
â and b must be antisymmetric in all their electrons. We may assume n > m, and
then between a and b a number of pairs, say
p pairs can
s
be formed, p ranges from

spin of both atoms is
to m. Accordingly, the total
(2)
interaction energy will be expressed by certain

which are a generalization of the Coulomb and
integrals
exchange integrals
The Coulomb
where
C and A
integral
is
=*
is
also
an exchange
xVf
fl
in
of Chapter IX, section
f tt</iV
dr,
1.
which the electrons numbered
+m)
+2
n+m ).
(4)
and n+ 1 are exchanged.
exchanged.
In addition integrals occur in which, for example, two

pairs of electrons are exchanged, 1 -> + 1 2 *-> n+ 2. In
fact all kinds of permutations of the n-\-m electrons occur.
All these integrals are numerically much smaller than A,
,
We
shall neglect them in

as a closer examination shows.
the following. Also the 'overlap integral' S (p. 129) may
be neglected.
Without very excessive labour
it will
not be possible to
calculate the values of these integrals numerically. As,

however, the structure of the integrals is similar to that for
hydrogen, we are justified in assuming that their behaviour,
as a function of the distance,
is
similar.
Thus we can
C and A
assume that for large and medium distances both
A is numerically considerably larger
than C, The formation of a molecule, and the strength of

the bond, therefore depends on the sign and the factor with
which A occurs.
How does the interaction energy then depend on C and
This is most easily derived by the method explained
the result
in Chapter XI, sections 1 and 4. We anticipate
((5) is derived on p. 183):
A?
E = C+A\$(np){mp)\.
is
dr,
The reader will easily convince himself that the integral has
the same value if any other two electrons of a and b are
(3)
the interaction of all the electrons of a with

those of 6 and the nucleus of b, etc. The integration is
over the coordinates of all electrons (dr
dT dT ,..dT
integral
(ft+l,2,...,ftWI
are negative and that
n-\-m
The
There
153
When the maximum number of pairs is formed, p

E C+pA.
have
(5)
we
(6)
VALENCY
154
The exchange integral occurs with a factory corresponding

to the fact that a#-fbld bond between the two atoms exists.
For smaller p the factor of A decreases and then becomes
negative (repulsion). Consider, for example, two
atoms,
=m=
states
XI
POLYATOMIC MOLECULES
Both are in 4 states and therefore only 2

are formed. The energies are
3.
= 3, 8 = 0,
= 2, 8=\,
p=l, s = 2,
8 = 3,
p=
C+3A, p
G+A, p
B = C 3A,
C-9A,
0,
1.
2,
*,
(?)
Z.
Apart from the lowest state with a triple-bond

(p = 3)
is an excited, but still stable, triplet state
(p
2).
Although two bonds still exist, the energy is only C+A.
Excited triplet states of

one bond is formed (p
N are indeed known. When only

= 1) the energy is CZA which
certainly means repulsion
This is rather surprising because

formed, between the two atoms. It is a general
feature of the theory that the binding energy between
two
atoms is very much diminished, or even overeompensated,
if both atoms have either free
unsaturated valencies
one pair
understand the chemical behaviour of a system of

atoms in polyatomic molecules as well as their reactions,
we calculate the interaction energy. This will depend on
the distances between each pair of atoms. We shall see
that it can be expressed in terms of Coulomb and exchange
is
integrals.
To begin with, we consider a number of atoms a, 6,..., etc.,

placed somehow in space and with one valency electron
each. It is convenient to assume an even number of atoms.
The case of an odd number can be reduced to this by
assuming one further atom and putting the interaction
integrals of this last atom with all other atoms equal to
zero.
or are
bound
jointly to a third atom.
with the
facts,
ofN 2 is known.
Our
result
is
in
atoms with one valency
To
*2,
there
Interaction of several
electron
agreement
though, because no molecular quintet state
Atom a
has then a wave function
according to
function is then
its
spin direction.
i/r
a (l)(l)
or
a ( 1)0(1)
tfi
typical total
wave
0(l)^(2MB (3)...a(l)j5(2)(3)....
Since
we are most interested in molecule formation,
(1)
let
the
be equal (total spin zero).

number of a-factors and
with (1), are obdegenerate
Further wave functions, all
attributed
electrons
of
the
tained by a different numbering
into
a-factors
of
the
some
to a, 6,..., as well as by changing
j8-factors
and vice versa.

As a first step, we can now construct spin wave functions
such that between specified pairs of atoms a bond, or an
j8
15(3
is:
we must also attribute an arrow to

Then the three spin functions (2) and the
spin functions matters,

electron pair with anti-parallel spin,
established.
For
we can have a bond between a and
example,
between
(1) this
is
and d. If the
spin wave function
c
b and also
numbered as in
electrons are
is
each dash.
relation (3) can be represented as follows:
+b
(omitting factors relating to
the remaining atoms)

*i
Xt
i[(l)j8(2)-(2)i9(l)][a(3)j8(4)-flt(4)j3(3)].
(2a)
0m
Alternatively, bonds can be established between
a and c
and between b and a*, and finally we could have a-d and
b~c bonds. The corresponding spin functions are
""
(3')
1
c
*Wl)j8(3)-(3)]8(l)][a(2)i8(4)-a(4)i8(2)]
(26)
-> d
0n
We can choose two of the three spin functions as an 'Ladeand n We call a wave
pendent basis', for example
<
< z
0n =
0m -
function with fixed bonds, like
<J> t
<
or
<f>
a valency
However, these three spin functions are not independent.
four atoms, more valency

structures exist, and for those relating to any four atoms
a linear relation (3) or (3') exists, but there are no further
By
linear relations.
iWl)i3(4)-(4)j8(l)][ a (2)iS(3)-(3)^(2)].
structure. If
(2c)
multiplying out the brackets one sees that
0m 0ii 0i-
(3)
Thus one of the three spin functions is a linear combination

of the other two. The factors have been added in order
that the <'s are normalized to unity
J-jA = /*i=l,
when use
is
made
of
aj8
0,
a2
/J
(4)
2
1 for
each
electron (see p. 96). It is important to realize, however,

that ^j, ^ n <
are not orthogonal to each other. Multiplying (3) by ^j, < n <
in succession and using (4), we
easily obtain
,
We
0i0ii
0ii0m
<Wiii
=h
5)
can represent the three spin functions graphically in

a very simple manner, indicating already their connexion
with chemistry. We connect the atoms between which a
pair is formed by a dash. Since, however, the sign of the
we have more than
basis of independent structures is conveniently obtained if we imagine the atoms all arranged
on a circle (on paper, this has nothing to do with the true
arrangement in space). Then a structure where two
valency dashes cross (as in lllt (3')) can be expressed by
<f>
those where no dashes cross. An independent basis is

obtained by constructing all possible valency structures
in which no two dashes cross. Complete wave functions
are
now
h = A#ttMB>~fc
0n
iMi)&(2)...*ii. etc
6)
These wave functions are not yet antisymmetric. This is

easily remedied. We add to D for example, all those wave
functions arising by a permutation of the numbering of the
electrons with the proper sign. If we change the numbering
a, b
1 2 -> 2, 1 of the two electrons located in (6) at atoms
added
we obtain ^ a (2)/f (l)...^ I (2, 1,...) and this has to be
</f
ft
to
(6)
two electrons
at any two atoms can be exchanged and
with a minus
located in (6)
sign.
Similarly, the
a corresponding term to be added with a minus

sign. For brevity we write these terms T tp
ab a b .. .Tab j> 1} etc.
where T^ has the function of exchanging the numbering
of the electrons at a and 6. This exchange is to be carried
out in the orbital functions ^fa ^b ... as well as in the spin
function v There are also more complicated permutations
in which three or more electrons are involved. For example,
the numbering 1, 2, 3 can be permuted into 2, 3, 1, or
1, 2, 3, 4 -> 2, 1, 4, 3, etc. All these occur in the completely
antisymmetric wave function with proper signs but we
need not write them down explicitly. Thus from (6) we
obtain an antisymmetric wave function, again denoted
the original ip l7 tp n ,.. The coefficients 6j,.., must be determined from the problem itself, and we shall see that they
even depend on the arrangement of the atoms in space
158
there
is
<ff
<j>
159
(their distances, etc.).
We
proceed as foEows: supposing ^r is the exact wave

the complete energy function comprising the kinetic energy of all electrons and the
whole potential energy, including the interaction between
the atoms, then the wave equation is satisfied E*f/
Eift
consists of the
(see p. 27). Now the energy function
internal energy of the atoms
say, and the perturbation
V, which in our case is the interaction between the atoms,
-\-V. Similarly E is the value of the internal energy
of all atoms EQt say, plus the perturbation energy Ai?
which we wish to calculate. If we denote the unperturbed
wave function (8) by ^f then E
Hqi/sq. Now the exact
but if we
wave function
differs, of course, from ^
confine ourselves to an approximate determination of AE
in the sense of perturbation theory, we can identify
with tp (see Chapter VI). Then we can subtract E from
E and from and we find that the perturbation energy
&E and the interaction function must also satisfy a wave
equation approximately
function of the whole system,
H=H
h = ^(i)^(2)...#i(i,2...)-gr
fl6
^^...ra6 ^+....
(?)
The sum a, b
is to be extended over all pairs of atoms. For

each valency structure we have then one unperturbed antisymmetric wave function of the type (7). If
such independent valency structures exist, we have
degenerate
wave functions and they all belong to a total spin s
0,
because all valencies are saturated.
Our next task is to calculate the interaction energy. For
this purpose we must now apply a different method, for
the following reason: From the
degenerate wave func-
N
N
*j/
i/
if/
tions
tpi,...,ipx,
tions like
we
,
"P
In previous
+-"+< ;vAv;
(8)
we
could find the 'correct' linear combination by mere considerations of symmetry, but here no
such argument presents itself. Yet, when the atoms interact, the degeneracy is surely removed, and for each energy
state the
cases,
A^ - 7^
can, of course, also form linear combina-
= cm+Cn&
wave function must
be, approximately,
Below we write
E again for AE.
(9)
is an unperturbed wave
with as yet unknown coefficients c It ...
For lt etc., (7) is to be substituted.
V is the interaction between each pair of atoms
function of type
/r
(8)
tft
= vab +vao +vbe +....
(10)
a definite
combination of the type (8), with fixed coefficients and we

have no reason to think that these wave functions are just
To make use of (9) we multiply this equation by tp a

and integrate over the coordinates of
1 )^r (2)...
6
all electrons.
Since
160
dr x
1, etc., we
J $1(1)
equal to unity multiplied
in
still
obtain on the left a term

by the spin function
i+ c ii ii+which two electrons are exchanged give

ci
The terms
first
<
rise
to integrals of the type
^ = J sMl)^W
n (2)^(l)
dr x dr 2
Thus the left-hand
side of (9) reduces to
(c I fo+...+eiV fv ).
On the
right-hand side
we encounter
taining V. First the term without
(11)
similar integrals con-
bc
where use
made
of the fact that
Vab
i-
ab
terms in
In addition further contributions arise from
An
interchanged.
which three or more electrons are
example
is
^(1)^(2)^(3)^(2)^(3)^(1)
z +...
integral also 4
(12)
With the above
simplifications
N N
Next consider the term containing
.
Tab
It
is
given by
A* = j ^(l6 (2)4(3)...Fâ (2)A6 (l)^(3)...rfr
= f ^(1)^(2)^^(2)^(1) dr dr +
+ f B (l)fc(l) dr f &(2)0 (2)F (23)#(3) dr
x
te
(9)
ab
'
(15)
Our next task will be to find out what is the effect of the
be very
exchange T on the spin functions &,... This will
.
ab
<f>
dr 3 +...
(13)
4*
expressed directly in terms of

t Since the interaction energy is
A&.
these terms, they may be included
etc., there is no need to neglect
all the distances R^,
on
depends
ac 11^
that
(14)
noteworthy
is
It
the chemical bond there are
at the same time. Thus in the theory of
three-body forces, etc.
also
but
atoms,
only forces between two
obtain from
itself a linear
easy indeed and we shall see that T^fa is
between
equation
combination of the <'s. Thus (15) is an
the
independent,
are
the <'s. Since these spin functions
up
splits
then
must vanish. (15)
coefficients of each
<f>
we
is
factor c1 <f> l -\- . .
(14)
dr.
electrons, any
Since each term of V depends only on two
factor and is
as
such integral contains an overlap integral
this type of
again comparatively small. We shall neglect
independent of
electron 3 and therefore
dr
1.
(12) is nothing
$?(3)
3
J"
but what we have called the Coulomb energy in Chapter IX.
It represents the sum of the Coulomb interactions betweeen
the charge clouds of the atoms. (12) occurs also with the
is
exactly the exchange integral

It refers to the interaction of a
is
than the first. f Thus we expect A^ to be

Chapter IX
not very different from the exchange integral in
A occurs on the right-hand side of (9) with the factor
>
integral of (13)
of Chapter IX, section 1.

second
and b only and depends on the distance a-6. The
interthe
as
described
be
can
integral is of a new type. It
'exchange
the
with
of
c
action of the charge cloud $(3)
the overlap
charge* ^{2)^(2 ), but this is multiplied by
"
of
term
(13) k therG
The second
l dr
Tab yields
= (^+^ +...)(l)^(2^(3)dfr dr dra+...

f
- j V* (1W*(2) dr dr + J Wi(2)#(3) &r dr
- Cab +C +...,
x
first
integral j if> a (l)M

fore rather smaller
These are the 'overlap integrals' already encountered in

Chapter IX. They are fairly small and do not play an
important role. For simplicity we shall neglect them in
the following.
The
161
not
6881
162
into
linear equations
From
the
for
unknown
coefficients
and
in a similar
and the energy E can be

determined. If, however, we only want to know the energy
we can put the problem even more simply. Instead of the
coefficients Cj,... we regard the ^'s themselves as unknowns.
cr,..., cN
(15) is
these both the c's
then evidently satisfied
if
we
put, for each
Ta Ai =
163
way
Tbdti
-fan
*i-*ii
(17i)
<f>,
We now
a,b
{E-Cttn+yAaTafaÔ
With the abbreviations
insert (17) into (16).
= A^+A^
= A ac +A bd
w = Ate+Aaa
(16)
(18a)
we
obtain
{EC u+v#i+(w )^n =

Since these are
linear
homogeneous equations for the
N unknowns $ the determinant

<f>'s
must vanish. This
What
of the coefficients of the

an equation for E.
will provide
now
without calculation. Each valency dash (with

direction) represents a pair of electrons formed between
specified atoms. All we have to do to obtain J^ ^ is to
exchange the two end-points a, b of the two valency dashes
unknowns
< r ,
from
ending in) a and 6, retaining the directions

and keeping, of course, the other ends of these valency
dashes (atoms c, d,.) fixed. If, in <, a valency dash connects a and 6 then T& just reverses its direction and
T^^
If ^ is a structure such that a is connected
with c and b with d, then
is a structure where 6 is
connected with c and a with d.
Consider as an example four atoms with the two spin
functions
(or
<f>.
<f> It
T^
<f>
n>
(3).
We have
m <f>j = Ttffa
Tabfal = Tab \ \
J\ ~
<f>
n must
vanish:
ECu+v
u+w
diately,
starting
(186)
oJ
It follows that the determinant of the coefficients of the
the effect of the exchange ab on each j> ?

This can be read off from the valency structures (3') immeis
- u +w)<l> + {E C+uv)tf> n
wv
-0,
EC+uv
(19)
or, explicitly,
(E Of (u-v) 2 +{w )( w)
0.
Hence we obtain two energy values
E=
C^j{u z -{-v 2 -\-w 2 uvuwvw}.
(20)
The twofold degeneracy, due to the two possible valency

structures, is removed: we have two different energies.
formula for the interaction of four
shall use it in sections 2 and 3 for two
(20) is the general
electrons.
We
different purposes.
of some interest to determine also the coefficients

cIf c n in (8). Clearly only the ratio is determined. We
1)
give the result without derivation (putting Cj
It
is
Tai><f>i
4>iii
= <f>n<f>i ~
^cdÎH
(11a)
= Ciî+cn n ==(/'i+
,
E C u+v
uw
fa.
(21)
There are two wave functions

corresponding to the two
energy values E. c^c-^ depends on the exchange integrals
A and hence on the distances between the atoms.
atom then still be repelled ? We find the answer

Will the
left-hand
to this question if we now remove gradually the
very
removed
is
atom
the
D. When
atom:
either
now
can
D.
D
far we are left with a system
a triplet or in a singlet state, in which cases the
164
if/
The generalization of this method (Heitler, Rumer, Weyl,

1931) for atoms with more than one electron will be given
in section 4.
2.
Activation energy, non- localized bonds
consider the interaction of three hydrogen

atoms in more detail. This will give us a new insight into
the mechanism of the covalent bond and we shall see also
that new concepts arise. We begin with some purely
qualitative considerations which will be corroborated
afterwards by the theory.
In order to distinguish the
atoms it is convenient to
assume that one of the three atoms is in fact an atom of
heavy hydrogen
(or deuterium).
Such a
atom behaves
chemically exactly like H, its electron has the same wave
functions and energy-levels as H, it differs from H only in
that the mass of the nucleus is twice that of H. For simplicity we assume that the three atoms are always placed
on a straight line and we only change the distances. Other
cases can, of course, be treated equally well.
In the first place we assume that the two
atoms have
formed a molecule and that the D atom approaches the
H D. We have
2 molecule from a large distance:
seen in Chapter IX, section 2, that then the D atom is
repelled because there is no possibility of forming a pair
with one of the
atoms. This is true as long as the dis-
HH
tance
D is much larger than the distance
of
the molecule. But what happens when the
atom has
come near to the
Supposing we have
2 molecule?
brought the
atom to just about the same distance from
the right-hand
atom as the equilibrium distance
H.
H-H
be in
atoms repel or attract each other respectively. At first,
one might think that the former must be the case (repul-
when the
sion), since
We now
165
by the two
D atom was far away it was repelled
atoms. This, however,
We
is
not true, as can
started from a state
HH-D
be seen as follows:
which has an energy equal to the binding energy of H 2
and this is the lowest energy which the three atoms can
whole
have in their present arrangement. The spin of the
the
change
we
if
Now
change.
not
does
and
system is
,
distances no interaction curve with the
can
cross,
and
therefore,
same total spinf

however we change the distances,
given
the energy must remain the lowest possible for each
the
bring
we
If
spin
(and
atoms
\).
arrangement of the
atom,
the
remove
afterwards
atom near and
HH-D
H-HD,
we always remain in the lowest energy state, at each stage.

state is
Finally, we have an HD system and its lowest
hi the end
the singlet (attraction). We are therefore left
the
changing
While
words:
molecule. In other
with an
a
into
over
goes
distances as above, the bond between
HD
HH
D.
bond between
atoms
This changing over of a bond from one pair of
There
suddenly.
happen
course,
to another pair cannot, of
half
speak,
to
so
which,
in
must be intermediate states
bond
bea
half
and
a bond is established between
HH
tween
HD:
f
--H--H
The angular
momentum
D
is
.here
always zero.
H
D
166
By symmetry reasons
HH
this will
be especially the case when
HD
the
and
distances are equal. We are dealing
here with a case where the chemical bond is not localized
between pairs of atoms. The number of electron pairs
formed between
composed of
i.e.
all directions.
wave function
parts, each with a sharp

with localized bonds. In our case
there are two parts, one with a pair localized between
and the other with a pair between
D. This
intermediate state is called a transition state.
How now does the energy change while we change the
distances ? From what was said above it is clear that at
first the energy increases when the D atom approaches
the
Later, when the intermediate stage
D is
2
number of
is
pairs,
H H
HH
reached, the increase of energy will be less rapid as a

between
and is being built up (at the cost of the
bond
bond, though). It may even be that in the stage

H
the energy has a small relative minimum. At any rate, the
energy will decrease rapidly again if we remove the
atom, because we shall in the end be left with an
molecule. Thus, if we change the distances as above, the
energy follows a curve of the type of Fig. 40, where it is
plotted qualitatively as a function of i?
H D/îm- Ifc flas
a maximum.
H
HD
Our conclusions have an important bearing on the theory

of chemical reactions. Supposing
reaction
amount of energy, the so-called

available. For before the D atom
made
activation energy, is
place unless a certain
can be attracted, it must be brought across the maximum

of the curve of Fig. 40. The activation energy required
is equal to the height of this
maximum (in the figure called
F). Naturally, F depends on
the direction from which the L>
atom approaches, but qualitaRhd/Rhh
tively the curve is the same for
and
is neither zero nor one (it is
one when the D atom is far away, zero when
is removed); in fact, this number has in the intermediate state
H--D not a 'sharp value' (in our previous language). In Chapter IV, p. 53, we have seen that then the
17
H +D = HD+H.
2
we study the exchange

Although the energies before
and after the reaction are exactly the same (namely, the
binding energy of
2 ) and the reaction therefore isothermal, we see from Fig. 40 that the reaction cannot take
F is lowest when
Fig. 40. Activation energy.
the
the three atoms are on a straight line and then has
experimental value 0-35 eV.
The same considerations as above can be applied to
reactions between molecules, for instance,
H +D =
2
2HD,
only the activation energy will be much larger because

the repulsion between two saturated molecules is stronger
than that between a molecule and a free atom.

This explanation of the activation energy and the more
quantitative theory below was first given by London (1928).
These more intuitive considerations are fully borne out
by the theory. The above conclusions are all contained in
for the interaction energy and (21) for the
formula
(20)
only three atoms, let the

interaction integrals of one of the four atoms (c) vanish.
The lowest energy is that where the square root occurs
with negative sign. We identify the atoms a, b, d then
with H, H, D, in the order of their arrangement. Call the
wave
function.
Since
we have
between the neighbours CH h> ^hh ail d ^hd> ^hd

respectively and those between the non-neighbours C, A'.
C',A' are always rather small. Then
integrals
E = Ohh+^hd4-^'
-V{Îh+Îd+^' 2 -^hh^hd-(^hh+^hdM'}.
(22)
Consider first the case where i? HD is large and -H H n near

the equilibrium distance. Then
HD is also small and by
expansion of the square root we find
We can represent E as a function of J?hh an(i ^hd

(always in linear arrangement) by an energy surface. This
is shown qualitatively in Fig. 41, where the lines of equal
168
E = CH h+Chd + C'-M hh
-^ hd -M')]
= CH H+iiHH+CHD + C'-Hii H D+^').

(23)
(The A's are negative.) Chh +.4hh = ^h,
the energy
of the H molecule.
H d I^hd and C'\A' are cer-
169
energy are plotted. The 'cheapest' way in which an

can be effected is
-> H-f
exchange reaction
2
along the dotted line. It leads from the valley on the
H +D
HD
is
tainly positive, which
Similarly
if
RKn is
means that the

large but
RHJ}
atom
is
repelled.
near the equilibrium
value,
An
HD
The two
molecule
is
formed and the third
cases go over into each other
is
by a mere
repelled.
adiabatic
12
change of the distances because they are both special cases

of the same energy state represented by (22). When both
distances are equal,
A BU = A ni>
= MHD = E
for i? HH
Q,
(24)
The energy differs from that of a single H 2 molecule by

Qhh + C'-A'. A glance at Fig. 34, p. 1 33, shows that C'A'
> 0) at the distance 2^ is much larger than HH at i?

Thus the energy is certainly higher than EHt On the other
hand, it is difficult to decide at exactly what distance i? HD
(
maximum of the energy lies. It may be that the energy

when R RS = R is kept fixed and J?HD
is allowed to decrease down to the value E
Or it may be
Q
that the maximum lies at a slightly larger value of J?
H d.
in which case the position .Bhd = ^hh = Rq would reprethe
increases steadily
right (i?HH
minimum.
We
Rhd/Oo
HHD system.
across the saddle-point S, to the valley
tion energy
quantitatively, because
have neglected too
is
the small
H atoms
and depends sensitively on the
values of the exchange integrals and the simplifications
made. More than an agreement in order of magnitude
cannot be expected, and this the theory achieves.
Finally consider the wave function. This is given by
and binding energy of
(21), section 1.
i/r
2)
corresponds to the valency structure

D
atoms, i.e. to
D. In our present notation $ is
with a bond between the

tfi
is
the structure
H H
HH
= to E C -^hh+^hd
toiA
^HH
much
(overlap integral, three-body forces, influence of

higher states, etc.) to be able to predict such minor details.
it
difference of two large quantities (energy of three
sent a relative energy
on the top (R-zn = )- The saddle-point may (or may

not) have a little dip.| It is hard to calculate the activa-
say, (22) reduces to
E - Chh +,4 hh +(Chh+<7'-4')

= EHt +(CBU +C -A'),
Fio. 41. Energy surface of the
t Actually S lies at a distance -Rhh

equilibrium distance Ii of U a
.
Rw>
slightly larger
than the
170
We
obtain in the three cases (23)-{24),t (omitting A')

I
valencies would extend in any direction and the attraction

atoms would be indepenand the two
forces between
2
dent of the angle between them. The form of the
A1
= 0i + 2A
"T ^ 0ii
(^hd
HH
arg e >
^hd small),
(25 a)
= -IA5S ^x +^n
2^
(i?
large,
HH
i HH small),
(256)
(^HH
In (25 a) the factor of
tft
is
^HDr -^HH
small and
= HH D
= ^Hd)'
25c )
essentially the
is
if/
HH
D is thus a superposition of
transition state
two valency structures. The bonds are not localized and the
number of pairs formed between a specified pair of atoms
does not have a sharp value.
We
shall see that the
characteristic for almost
atoms
behaviour of three
aU polyatomic molecules.
is
Directed valencies
3.
The valencies of many atoms with more than one valency

definite angles. The tetrahedral struc-
have a bias to form
ture of the four valencies of carbon
is
well
known. Also
the two valencies of oxygen extend in definite angles, the

water molecule has the form of a triangle with a basic angle
of 105.
3 is a pyramid. These directional properties
of the valencies of 0, N, or C must be due to the structure of these atoms themselves. Supposing, for instance,
that the
atom was spherically symmetrical, the two
NH
We
have multiplied (25
in order to
make
6)
by
A^jÂ^-Â^) ~
the factor of ^ n equal to unity. It

the ratio of the two coefficients that matters.
is,
atoms therefore repel each other. It would

most stable form of H 2 is a straight
H. This is not found to be the case.
line: H
In our previous considerations we have chiefly concentrated our attention on the question whether or not
a bond between two atoms can be established, and we
have seen that this depends on the formation of pairs.
We have not, so far, inquired to what extent the forces
between two atoms, due to the formation of a pair, depend
on the quantum states and wave functions of the two
electrons concerned. Now there is an essential difference
in the valencies due to electrons in 5 and p states. In
The two
follow that the
Tor J?hd = this is

L
exactly so. When -R Hh is large, the wave function is
essentially Îr =
H H D. In the case i2H H = ^hd
both wave functions occur with the same factor. The
wave function
molecule would then depend on the forces between the

in its ground state has two electrons
atoms alone. Now
atoms must both have
with parallel spin and the two
the same spin direction, namely opposite to that of 0.
II
= Al+0H
171
^hb/2^hd
of course, only
particular
shall see that valencies of
we
electrons
have
just the desired directional properties.

As an example we consider the simplest case, the water
begin again with some qualitative conmolecule.
We
atom, in its ground state < 3 P) has,

according to Chapter VII, section 2, four electrons in the
2p shell. For the wave functions of the three degenerate
p levels we choose the functions given in (9) and (11) of
Chapter III, namely (neglecting the interaction between
siderations.
The
the electrons),
*,
*/(r),
0,
y/(r),
fc
= */(r).
(26)
They are pictured in Fig. 12, p. 43. Of the four electrons two
must be placed with antiparallel spin in one of these three
states, say, for instance, in 2 The other two electrons can
t//
172
then be placed in ipx and tp respectively, and can have

y
parallel spin. In the ground state 3 P they have parallel
spin (see Fig. 42; only x and ^ are shown; ip would
B
extend in the direction perpendicular to the plane of
a direction between x and
drawing).
first
i//
tf
173
y, it is also attracted,
but the
attraction will be strongest from the z or y direction

itself.
let the second H atom be brought near. If the

atom has already formed a pair with the electron
in x then the direction from which the second H atom
is most strongly attracted is the ^-direction. The second
H atom can form a pair with the electron in y It follows
that the most stable configuration of the water molecule
Now
ift
H
,
tfj
H
is
a triangle with basic angle of 90:
This will be
0
by the fact that the two H atoms repel
The angle will then be slightly larger than
slightly modified
each other.
90, as is actually
By permitting
Fig. 42.
Water
Now let a hydrogen atom be brought in contact with

the oxygen atom. In Chapter IX it has been found that
the interaction energy of two atoms (the integrals C and
A) depends on the fact that the wave functions of the two
atoms overlap. If they overlap very little the interaction
energy
is
very small.
interaction
We
see at once that in our case the
must depend on the
direction
from which the
atom approaches the O atom. If the H atom comes

from the 2-direction, its wave function (which is spherically
symmetrical) will overlap a great deal with that of the pair
placed in s but overlaps very little with the two electrons
>jt
must be that the H atom is

H atom approaches either
from the x or y direction. Then a pah* can be formed
between H and, say, the free electron in tfc. The effect of
the pair in f will then be small. If the H atom comes from
in
^r
and
t}i
y.
The
result
repelled. Attraction arises if the
to form a pair with the

with
the
second
atom, we
y
actually permit these two electrons to arrange their
spins independently. They cannot always have parallel
spin (as in the 3 P state) because an exchange of spin
direction with the hydrogen partner takes place. This
means that, in addition to 3 P, we also take the singlet states
^D, 1 S) into consideration. In fact we apply
of
what in Chapter X, section 3 was called the procedure (6).
It is in this approximation where the directional properties
of the valencies are exhibited most directly.
In a very similar way the structure of
3 can be
explained. The three p electrons of N can just be placed
in the three states ipx
The three
atoms can all
yy ^2
be attracted if they come from the x, y, z directions respectively.
3 should therefore have the form of a pyramid
N
with an angle
of 90. Again this pyramid will
be flattened by the repulsion of the
atoms. The actual
first
molecule.
the case.
the electron in
H, and that
in
/r
NH
NH
ift
H H
angle
is 109.
174
This explanation of the directional properties of valency
was given by Pauling and Slater
(1931).
These qualitative considerations can easily be expressed

framework of our general theory.
Applying, thus, approximation (b) we consider the configuration of oxygen where two electrons are in
z and
one electron each in x and i/^.f We make no restriction
in formulae within the
ifs
C2v etc., those between the two H atoms by Oh, A&.

For simplicity we neglect the interaction between the
electrons within the
atom (which is by no means necessary).
Formula (20) with the abbreviations (18 a) of
section 1 becomes
Cjte,...,
E=
C'H-C^+C^ A u A^
*j/
as to their spin directions.
*J{u
We consider then two H atoms placed either (i) both

on the -axis on both sides of the O atom (angle 180) or
(ii) one on the #-axis and one on the /-axis (angle 90) and
we shall show that the binding energy is greater in the
where
latter case.
distances of the
the 180 position
To
calculate the interaction energy
= A lx +A
l
now
Consider
we can
use the
we have
to do
is
states enters in the integrals
to replace
states
tp
3
w=
2x>
(compare
(9)
atom with
of Chapter IX, section
and
90,
it
Co2x
Lr,
'0i
jCX i
->
Lr
A-^
xx
i>
;=
xL. 0'
= A^ = A
180:
E=
2C -\-4C'+Ca -2A'-\A
+A'-A n
(Aq-\-A'~- j4 h
between the
is
(27 a)
certainly negative because the distance
H atoms
rather large
is
and
An
small).
From the above overlapping considerations it follows

that \A Q is rather large and \A'\ considerably smaller.
We denote
= 2C +4C?'+Ck+^ +^'-^ H
2).
second
A,
and let the

be equal. For
follows from symmetry reasons that
positions 180
becomes
x) ip v .
are the interaction integrals of the
an electron in
and
tp
= A lv +A
A lv = A iy =
two of the 'atoms'
Only one addition is required,

namely the interaction of the two H atoms with the
electron pair occupying ips It is very easy to see that this
leads merely to an addition CA for each H atom where
C A
2y>
the two
Gi~
la
1.
The nature of these
by the two
Although this formula was meant

to apply to four atoms with one electron each it is valid
for four electrons occupying any four different one-electron
All
+v 2j(-w 2 uvuwvw}
H atoms from the oxygen
derived in section
only.
O = Clx -j-C2x -\-C y-\-C2y -\-CSt
general formula for the interaction of four electrons
states.
175
the interaction integrals between the
H atom with the electrons in
ip
x , etc.,
first
or
atom is
Consider next the 90 position. The second
on the y-axis. From symmetry reasons it follows that
by C^, A-^,
t The electrons of oxygen can, of course, be arranged in various ways

in the p- shell and all these configurations have in general to be taken
into account (for example two electrons in <px and two in t ). In the
two cases considered below where the
atoms are placed either at
180 or at 90 it can be shown from symmetry considerations that none
of the other configurations enter the problem.
C2y
C^ = C
A^x
\ft
f
rule,
An
A = Aix = A Q
Zil
" A ly = A}2 = A^ = A.
explicit calculation
now
now
shows that A' has, contrary to the general
a positive sign, whereas
is,
as usual, negative.
176
where
and A' are the same
integrals as in (27 a).
We
obtain
90:
E=
2C
+4C'+CH -2^'-
-V{448+44'-44 il'+4fi-24 H 4
(
+^')}.
(27 6)
Let us now compare (27 a) with (276). The

has a lower energy if
4Jg+44' 2 -4^
or
Now
^'+^l-2^ H (^ + 4')
> (A +A'-A s y,
J
{A^-A'Y
if
90 position
where r 1( r 2 are the position vectors of the two electrons.

The exact wave function differs, of course, from (28) but
the symmetry properties are the same. (28) has its largest
value when the two vectors r^ r 2 are perpendicular to each
other. This fact, however, does not so far give rise to
directional properties. The theory for the interaction of
atoms with several electrons in antisymmetrical states will
be developed in the next section. A very simple calculation
shows that the energy is, apart from Coulomb terms and
the contribution from the pair in z just one exchange
tf/
integral
> 0,
on account of the larger overA Q Thus the 90 position is more
for each
A =
this is just the case
lapping in the integral

stable than the 180 position.
.
compared with A Q
the two H atoms repel each
certainly
Since
AK
is
rather small,
seen from (27) that

(The
term
other.
it is also
It is instructive to consider the same problem also when

we take only the 3 P state into account (procedure (a)).
The wave function of the two O electrons in x and v is
then antisymmetric. If we were to neglect the interaction
between the two electrons we should have
tft
-0(2,
With the help of the
1)
tf/
- iwi^(2)-W2^(l)}.
explicit expressions (26) this
can also
be written
0(1,2)
= ^(*
sl
-* 2 y 1 )/(r1 )f(r2
H atom, where
dT
j 0(1,2)M3)F0(1,3)0H (2)
<
29 >
atom. The integral

of the
atom
lies on the x or
the
whether
has the same value
direction from
the
of
independent
it
is
the y axis.f In fact
of H 2 is the
energy
The
approaches.
atom
which the
This is
atom.
each
for
integrals
(29)
sum of two such
neglects
one
fact,
if
H.
In
independent of the angle
the interaction of the O electrons, one obtains just the
expression (27 a) we found before for the 180 position.
The lowering of the energy in the 90 position is due to the
fact that we have allowed the two O electrons to arrange
their spins freely, which implies that singlet states of O
are also taken into account. It is the influence of these
(in reality excited) states which causes the directional
preference and it is in accordance with our considerations
in Chapter X, p. 147, that the energy is lowered.
and
i/f
is * he
wave function
in (276) diminishes the binding energy and A& is smaller

than this term.) The angle with the lowest energy is
therefore slightly larger than 90, as is in fact the case.
#M) =
177
= ^ (r, x r,),/^)/^.),
(28)
HO
In addition to the influence of the excited states there

is a second effect which contributes to the directional
properties. We have always confined ourselves to the
exchange of only two electrons. It was mentioned in
f 0(1, 2) merely changes sign if a;
and y are interchanged and
twice in (29).
5881
<Ji
occurs
178
section
that higher exchanges also occur. For example,
numbered 1, 2, those of the

H atoms 3, 4, we can have an exchange 1 *> 3 simultaneously with 2 -e> 4. Such exchanges give rise to smaller
contributions to the energy which, however, depend on
the position of the three atoms simultaneously and not
only on the distance between two atoms. If these are
if
the two
electrons are
taken into account even the

tial directions
state alone gives preferen-
The wave function
(28) is largest
when
the
two
electrons are at right angles and the value of such a

double exchange integral is also largest when the two
atoms form a right angle with O. (Artmann, 1946.)
Thus there are two factors contributing to the directional

valencies of
H 0:
2
(i)
D,
S,
figuration with three jo-electrons give rise to a pyramidal
S state, although it is spherically

symmetrical as a whole, has a wave function which is
largest when the three electrons are at right angles. Double
exchange leads to the same structure.
The case of carbon is a little more complicated. It can
be shown that the co-operation of 3 P and 5 also leads to
preferential directions, namely the directions from the
centre of a tetrahedron towards its corners, although this
cannot be shown by such intuitive considerations as for
or N. Furthermore, the wave function of 5 S is largest
when the four position vectors r^, r 2 r3 r 4 extend in the
directions of a tetrahedron. This is in fact quite easy to
see. Thus the four valencies of carbon have a bias to form
a tetrahedral structure.
Although the directional properties of the valencies are
very important in that they determine the geometrical
structure but even the
179
impor-
tant if we are interested in the binding energies. For

pyramid
example, the energy required to press the
3
flat is only 0-26 eV, a small fraction of the binding energy
where the energy difference between
(in contrast to
2
the 90 and the 180 positions is large). Also for carbon
the energy required to bend the valency angles is rather
NH
small compared with the binding energy. Both effects

which give rise to the directional bias involve comparatively small energies. In the following sections, where the
emphasis is on the binding energies, we shall ignore the
directional properties (we can use
section
3,
was
called the
method
what in Chapter X,
(a)).
the influence of the excited states
and (ii) double exchange, which is effective even

in the ground state 3 P. For molecules like NH 3 the situation
is very similar. The excited states arising from the con1
structure of a molecule, they are in many cases less
4.
Interaction of
atoms with several electrons
We now generalize the theory of section
to atoms with
several electrons. Let a, b, c,... be atoms with na nb , nc> ...

valency electrons. For simplicity we assume that no
,
degeneracy other than that due to the spin exists. Further

that the n n electrons of atom a have parallel
n
spins so that atom a is in an +1 S state, and we do not take
into account electron pairs within the atom. The orbital
wave function of a is then antisymmetrical in the n a
electrons, the spin wave function is symmetrical. The
simplifications are made solely for the purpose of laying
we assume
most essential and interesting in

the bond mechanism. It would be very easy indeed to
take the degeneracy of P states and electron pairs, etc.,
into account. Thus we apply what in Chapter X, section 2
was called the procedure (a) and we have pointed out that
this can even be applied to carbon (*S state) for the neighbourhood of the energy minimum.
The modification of section 1 for the present case requires
hardly any calculation. The first step is again the
the emphasis on what
5881
is
N2
180
construction of the valency structures. This is much as in

I, the only difference is that now more than one
between atoms a and
section
electron pair can be formed between
Suppose, for example, n a
atoms a and
> n^
then evidently up to nb
atom can form
bonds with several atoms. For each such structure a spin
wave function
can be constructed. For example, if we
have two C atoms and two
atoms the following three
valency structures exist:
pairs can be formed.
A ab =
6.
Also, of course, an
<f>
j
a
181
b, is
na )W I +l,..,n +)FX
W1.2
fl
(+1.2,...,n a )^(l,o +2,..., o +6)dT.
(33)
of
have numbered the electrons of a 1, 2,...,na those
exchange
the
involves
and
(33)
b +l,.,n +n 6
We
integral has the
The
c^c
c=tc
>H
&
Again there
is
c=tc
X
H H
hi
(30)
It
is
we
shall neglect
The
them.
interaction energy
is
again determined from the set
N independent valency structures)
(31)
(34)
There are no further linear relations. Again an 'independent

basis' of spin wave functions is found by the non-crossing
The structures
electrons
exchange
belonging to a and b are exchanged. There are also
than
more
integrals in winch higher permutations involving
two
for
occur
two electrons occur. These can now even
against
a
of
atoms (exchange, for example, two electrons
exchanges
electrons of 6), and there are, of course, also
of equations (for
*m &i &
rule.
any other two
these are conin which three or more atoms take part. All
integral and
siderably smaller than the ordinary exchange
a linear relation between the structures
any four bonds involving four atoms.

exactly the same as in section 1, for example,
if
two
h. ii
relating to
same value
(fl-Ctov
+ ^rfZWw-O.
(30) are, of course, constructed without
any prejudice as to the formation of a molecule C H

2
2
The interaction integrals are now defined in the following
manner: The Coulomb integral is, as before,
.
C = ...F dr
J
= Crt +C +C
fle
te +...
(32)
The integration is to be extended over all valency electrons

of all atoms and V is a sum of terms each of which describes
the interaction between two atoms. The exchange integral,
The next question is: what is the effect of the exchange

Tab on $ lt etc. ? Evidently, Tab is the sum of interchanges
dashes ending
of the end-points a and 6 of any two valency
a with b
connect
bonds
in (or starting from) a and 6. If p (sb
p ub^' If a is
the corresponding contribution is Tub
then
d
hjp
bd bonds,
connected with c by p ac bonds, b with
is
result
the
p^Pm $',
P Pbd exchanges can take place and
ac
bonds
where <f>' is the structure which has p M 1
d.
It is
and
a
b
c
l bonds 6 d, and one bond each
<f>
p bd
We
better to express this in formulae.
a
a
182
by
6,
[ab]
p^ bonds
[6a].
<
The
Then the
c, etc., is
symbolize a bond
structure with
Then T^^
is
The
-8^ch
E G 2-4QC 2^4 OH
quadratic equation for E
solution of this
written in the form (with
by
= 0.
can be
= CCc+2Cch)
E Ccc+^cc+2(Cch+-4ch) -^ cc/3 (2^)'
[ac][bd\-[ab][cd].
given by
(35)
cd
The sum
EC4A o0 +SA 0a
Aqc
linear relation of type (31) is expressed

[ad][bc]
It follows that
p^ bonds
denoted by
[aby<^[ac]p 'e [bd] 3) 'x'...
183
lacjlbaj
A GO
*m 2A
b-44/K- t
,,+
(37)
CH
d other than
We shall use and discuss this formula in detail in section 5.
c and d are the same atom (if

both a and 6 are connected with the same third atom c).
The structure in the second term of (35) has one additional
bond ad and b c in place of a bond a c and 6 d.
We give two examples. Consider the structures fa fa
(30) which will describe the C 2 2 molecule.
For the
exchange Tco between two C atoms we have
second, very simple, example consider two atoms

valency electrons respectively. Assume
with n and
and consider also the case where only p pairs are
a, b.
It
is
to be extended over
may
also
atoms
all
c,
be that
^ccî
= 4î
As a
n^-m
formed,
< m,
free valencies.
atom a has np, atom 6

To reduce this case to the above we
so that
mp
intro-
duce a third atom, L say, which has no interaction with

A aL CbL A hL 0) but which can
a and 6 (i.e. CaL
absorb the free valencies of a and 6, There is then only
one valency structure, namely
H are two atoms connected by a bond in fa then

Teak = ~*fai -4*n+4fc, To* fa = -fa. (366)
We can now calculate the energy of the C H system. Let
If C,
[ab]p[aL]*-*>[bL]m -*>.
(38)
the atoms connected by a bond in

^ n be neighbours (^ n is
then the chemical formula of the C
2
2 molecule), and let
us neglect the interaction of non-neighbours. The C
interaction occurs twice, so CC h and C m have to be multiplied by a factor 2. Inserting (36) into (34) we obtain
(E-G}fa+(-*Aca+&AcBMi-ZA(mfal
~ A cci>i+(^-0)faM-^co-2A
cn )fa
0,
0.
The
application of (35) gives at once
Tab<l> =
-p<f>+{n-P)( m -~P)4>-
The exchange of the end-points a, b of two valency dashes

connecting a with L and 6 with L leads back to the same
structure. Thus the energy is
= C+[> (np)(mp)]A.
This is the formula anticipated and discussed in Chapter X,
from our formulae the excitation energy of *S, viz. ve,

where v is the number of C atoms in the molecule, and e
184
section
3.
185
the excitation energy of 5 S.

5.
Binding energies of hydrocarbons

The theory of the preceding
section allows one to calcu-
The interaction integrals which occur are all negative

and so is the binding energy. It is more convenient to deal
with positive quantities. So
of the Coulomb and exchange integrals. There is, of course,

little hope that we could calculate these integrals theoretically without excessive labour, as could be done for hydrogen. Instead, a semi-empirical method suggests itself. If
we have a number of different molecules consisting of the
same kind of atoms so that in each molecule the same
interaction integrals occur we can use the experimental
data of some of the molecules to determine the integrals
and then check the theory by calculating the energies of
the remaining molecules and compare them with the
experimental data. A particularly favourable case is
tive dissociation energy into
available in the simple organic molecules consisting of

C and atoms only. Thus we shall consider the molecules
state (calculated in
CH4 C2 2 C2H4 C2H6 and the radicals obtained by

removing an
atom CH3 C 2 H, C 2 H 5
It has been explained in Chapter X, section 2, that for
the treatment of molecules containing carbon it is a feasible
approximation to consider the B S state of carbon with its
,
four valencies alone, provided that we confine ourselves to

the energy minimum and leave out of consideration molecules in which only one or two valencies operate (CH, CH ).
2
The binding energy thus obtained is then, however, the
energy required to separate the atoms such that each C

atom is left in the & S state. In reality the molecule dissociates into atoms in their ground state ( 3 P for carbon)
and this dissociation energy is also what is measured. To
obtain this quantity we have to subtract from the positive
dissociation energy (= minus the binding energy) obtained
now be the posiD=

atoms in their ground states.
In the molecules CH 4 C 2 6) and C 2 H 4 the interaction
between neighbouring H atoms (in the same CH 3 and CH 2
group) must be taken into account. To simplify writing
late the binding energy of a polyatomic molecule in terms
we introduce the
following notation:
Cbc = c
is
let
>
the interaction energy of two
H atoms in the repulsive

h > 0.
Chapter IX, section
CH 4
and
1).
CH 3
are easily calculated.

The
identical with the
structure
valency
one
only
There is
are all parallel.
atoms
the
of
spins
The
formula.
chemical
occurs
repulsion
The
other.
each
repel
atoms
Thus the
energies of
six times in
CH 4 and
three times in
CH4 D=
CH3 D -
CH3 We
.
obtain
-+4(6+j8)-6A,
(39 a)
-e+3(6+|B)-3fc.
(396)
The energy of C 2H 2 has already been
calculated in the
preceding section. We have to subtract the excitation

energy of & S twice because there are two C atoms.
C2 H
jp
-2+2(6+jB)+c+3 y -y/3 (),
(40)
25
f^-rrtj
(40')
&
2j3
186
(the minus sign of the square root gives the lowest state of
the molecule). The energy of C
is obtained by replacing
2
26
by
action
and
2jS
by
j8
because there
is
only one
separate the two
value /j
3.
CH inter-
D = -&+&+ /S+c-f 3y-y/ <2).

(41)
The molecules C H and C H are more complicated. One
can safely assume that the H atoms in each CH or CH
3
fi
group have parallel spin and can be treated like an atom

with two or three valencies respectively (it can be shown
that this incurs only a negligible error). The reader will
easily convince himself that then three structures exist for
C 2 4 and four for C 2 6 The resulting equations of third
and fourth order can readily be solved numerically, because
the energy depends, apart from certain additive constants,
on the ratio
y/2 only. The result is:
4:
6:
CH
2?ch is the energy of the

3 radical. The exchange
integral between the C atoms occurs with a factor 2.
where
Although each CH 3 group has a free valency the two

radicals repel each other at large distances. Thus a (probably very small) activation energy has to be overcome to
form the C 2 H 6 molecule from the radicals. On the other
hand two CH 2 or CH groups attract each other strongly.
(42)
2 (|),
0-2
(43)
C 2H5
If
it
3+1 HH neighbours.
D = -2+5(6+j5)-4*+c+y-y/ (gf).
1
(44)
were not for the functions / the formulae would have
a natural and simple
significance.
Each bond of the
formula contributes just one corresponding
exchange integral and, of course, the Coulomb integral.
In addition two neighbouring
atoms contribute a
chemical
A. But the functions /

and the result is more complicated.
repulsion
are
One observation about C 2 H6 may be
08
,
and 6h by 4A. (There are
0-6
The functions flt /a / 3 occurring in

the energy of C 2 C a 4 C S S
are plotted in Fig. 43 for the range

of values of y/2j8 which come into question. The energy
of the radical C 2 5 is obtained by replacing 66 and 6/J by
,
5^3
0-4
FlG. 43.
The functions flt f2 f3
56 and
2DGHg +c-2y,
=
C H D = -2e+4(6+jS)-2fc+c+2y-y/
C H D = -2c+6(6+^)-6A+c+y- y / (f).
2
groups (c, y small) fx tends to the

is then
The energy
C 2 H:
CH3
187
use our results numerically we first remark that

atoms
The interaction between
known both from theory and empirically. For the
To
e
is
= 4-16 eV is known.
known H H distance h = 0-5 eV.

four unknown constants 6, c, jS, y.
We
try to adjust these to
fit
We
are left with the
four of the experimental
It turns out, however, that the theoretical results

depend in practice on three parameters only. So only
data.
CH
by no means zero
three data can be fitted, for which we choose

4 C2
2
and C 2 6 The remaining four dissociation energies are
we
then determined practically whilst for the constants themselves a certain range of values is still possible. The test
of interest. If
188
of the theory
is
therefore rather severe.
In Table 11 the
theoretical dissociation energies together with their experi-
and the repulsion of the
189
H atoms h, no additivity prevails
because of the presence of the functions fv f2 fz which

depend on the ratio y/2-3 and are far from zero. The theory
also accounts well for the differences of the energy required
to remove an
atom in the various molecules. The fact
that these are not equal again shows that the binding
,
mental valuesf are given for the following choice of the

constants (in eV) c = 1-9, y 3-8, b
2-75.
3-35, /J
Table
Energy
Exp.
Tbeor.
11
Dissociation energies of organic molecules
energies are not additive.
CH,
C S H,
("=11,
C,H,
Cri^-H^Crij
CgH^>-Cjli
Cjii^-^CaHg
17-2
170
23-5
29-1
4-45
5-25
4-25
(17-2J
(17-0)
23-7
(29-1)
4-55
5-55
4-20
The non-additivity of the binding energies is closely

connected with the valency structures. As in the case of
three
atoms
exception of
seen that the agreement
very good indeed, and

much better than might have been expected from the
numerous simplifications we have made. This can only
be understood in the sense that many of the neglected
It
is
effects are
embodied
is
tures
<j,
reasonable.
formula
shows that the theory
is well able to
account
must be
however, that the binding energies of

these molecules are not additively composed of contributions due to each bond (with the exception of CH 4 and CH 3 ).
Even if we take into account the excitation of the B S state e
It
(30).
realized,
t The experimental values depend on the heat of evaporation of

diamond which is still not finally known. The figures in the table refer
to a value of 7*4 eV (= 170 kcal./mol.)- An alternative value which
cannot be entirely excluded is 6-1 eV. If this should be the true value,
the experimental values of the first four columns would be smaller by
1-3 eV per C atom. In this case a similar adjustment of the constants
can be made and the agreement is about equally good but no agreement
could be reached for a substantially different value.
The bonds
$
The
result
</>
(with the
C 2 H 2l these are the
struc-
are therefore not localized
between pairs of atoms (compare section 2). For the above

values of the energy and the constants one finds that
presumably the reason why the

Coulomb integrals turn out to be comparatively large.
Otherwise the values of the interaction integrals are very
The
is
a superposition of several
structures. In the simplest case,
is
for the so-called bond-energies of the single, double, and triple

bond, which occur in the energies of C 2 3 C 2 4 and C 2 6
CH3
and
in the empirical values of the inter-
action constants. This
wave function
{section 2) the
CH4
coefficient of
factory feature:
<j>
is
^x
is
^ n + 0-165^.
rather small. This
is
a very
satis-
just the structure of the chemical
HC= C H. Quite generally the structure of the
chemical formula is always predominant amongst the

various structures which occur in the wave function. This
fact provides the physical justification for the formulae which
have always been used in chemistry
(especially in organic
chemistry) with such outstanding success. Nevertheless,

the other structures also play an important part and contribute very considerably to the energy. They are also
necessary to obtain the correct asymptotic behaviour for
large distances (compare section 2).
There are molecules where the bonds are not even nearly
The benzene ring with the two equivalent for-
localized.
mulae [
chemistry.
and
has always been a problem of organic

(J
In wave mechanics
it
is
obvious that, for
190
symmetry
reasons, both structures must occur in equal

proportions in the wave function (as was first noticed by
Pauling). Thus the double bond has no localized position
INDEX
in the ring.
Reviewing the contents of the last three chapters it can

be said that wave mechanics is the tool for a complete
understanding, on a physical basis, of all the fundamental
facts of chemistry.
Activation energy, 167 ff., 187.

Additivity of binding energies,
of molecule,
147; see also Ex-
change degeneracy.
Deuterium, 164.
188.
Angular momentum, 50
Degeneracy, 40, 44, 74, 95, 144.
accidental,
ff.
orbital, 61.
Diffraction, 2.
spin, 59.
Directional properties of valencies,

171 ff.
quantization, 57, 60.
Discrete quantum states, 1, 21.
Dispersion, 6,
Dissociation energy, of
133.
2
total, 62.
109, 113.
Antisymmetrical wave function,

75, 81, 83, 118, 126, 157.
Atomic interaction curves,
108,
143.
of hydrocarbons,
Double bond (C aH 4 ),
Balmer formula,
188.
188.
Doublets, 66, 110.
38.
Benzene, 189.
Binding energy, see Dissociation
d, D-states, 42, 57, 67.
A -states,
110.
energy.
Bohr radius, 35, 131.

Boundary conditions,
C atom,
22, 34.
excited states, 105.
Estate,
104, 142.
valency, 142, 147.

molecule, 148.
Charge density, 89, 98, 125.
Closed shell, 99, 137, 142,
,
C3
Complex wave
functions, 18, 29,
44, 50.
Constants of the motion, 27, 51,

54, 109.
Continuous spectrum,
Coulomb
interaction,
integral,
89, 130, 152, 160, .180, 185.
two
CH 4 CH3 CjH
,
carbons.
2,
149.
Equilibrium
distance,
of
2
molecule, 133.
Exchange charge density, 90, 130.
degeneracy, 75, 126.
energy, integral, 89, 128, 153,
H atom,
influence on valency,
on directional properties,
Hydro-
ff.
149, 151.
ff.,
interaction
etc., see
37
of C, 105.
Exclusion principle, 76
spins, 66, 92.
Crossing of atomic
curves, 143 ff,
Excited states, of
Coupling, spin-orbit, 63, 66.
Eigenvalues, 23,
of hydrogen, 34, 37, 42.
Electron configuration, distribution, 73, 77, 100 ff.
pairs (chemical bond), 133, 140,
161, 181, 185.
39.
field, 24.
energy,
Effective charge of the nucleus, 88.

Eigenfunctions, 23.
Fine structure, 64.

Forbidden states, 78.
Free electron, 17, 29, 37.
177.
82, 117.
INDEX
INDEX
192
Frequency,
Molecule, electronic states, 107
I, 4, 7.
of vibration of H
diatomic molecules,
of spin exchange,
134.
2,
in
116.
135.
rotation, 112
energy,
Hydrogen atom, 32
molecule, 124
dissociation
188.
ff.
of two
H-
atoms, 133.
of two
91
diatomic molecules,
of several H-atoms,
inH O,
polyatomic molecules, 180
electrons,
in
2,
NH
ff.
152, 183.
ff.,
186.
Operator, 26, 47.

Ortho-molecules, 120.
Orthogonality relation, 93, 96.
121, 148.
Para-molecules, 120.
ciple.
Phase
velocity, 4.
II -states, 110.
Quantum numbers,
Lewis's electron pairs, 142.

Light quanta, 2.
Linear combinations of
of
wave
functions, 43, 95, 146.

Lithium, electron states, 78.
Localized bonds, see Non-localized

bonds.
55, 60.
action,
38, 61.
Race, 111, 146.
Rare
80, 96.
Standing waves,
17, 48.
gases, 100, 137.
Repulsive interaction, 109,
132,
185.
Rotational levels (molecule), 114,

120.
Saturation, 123, 138.

Schrodinger equation,
equation.
Screening, 98.
s,
81, 118, 126.

^-states, 39, 52, 67.
S-states, 110.
der Waals forces, 123.

1
Vector addition
rule, 63, 67.
Vibrational levels (molecule), 116.
sm Wave
Water molecule, 171
ff.
Wave
amplitude, 8.
equation, 18 ff., 24 ff., 28.
function of hydrogen, 33, 38,
40.
of oscillator, 116.
of s, p, d states, 40, 42, 52,
119, 171.
Three-body forces, 161.

Time-dependent wave equation,
p; P-states, 39, 43, 52, 67, 171.

Ionic bond, 123.
Ionization energy, 34, 106.
Measurement, 15.
Modulated wave, 30.
60.
Symmetrical wave functions, 75,
Potential barrier, 36.

Probability, 9ff., 16, 35, 70.
Interference, 2.
moment,
magnetic moment,
of molecule, 110.
valency, 139
wave function,
173.
Periodic system, 99 ff., 141.

Perturbation energy, 88, 159.
175.
in
Magnetic dipole, 55.
structure, 157, 169, 180.
Van
ff.
Superposition of waves, 29, 49.

of valency structures, 170, 189.
121, 140, 154.

3,
13, 36, 117,
ff 149.
Nuclei, 120.
Uncertainty relation,
Valency of an atom, 141, 149.
ion, 138.
Particle velocity, 4, 6, 31.

Pauli-principle, see Exclusion prin-
163.
12, 17.
Non-localized bonds, 166, 170, 189.

Normalization, 43, 90, 96.
O s molecule,
energy,
an
Spin, 59
4, 12.
Overlap integral, 129, 160,
ortho, para, 120.
Interaction
Size of
Oscillator, 116.
ff.
Singlets, 66, 82.
119.
ff.
Neutron, 120.
ff.,
vibration, 115
30.
Hamiltonian, 26.
Helium atom, 72, 88, 93.
interaction with hydrogen, 137.
Heterpolar bond, 123.
Homopolar bond, 123 ff.
Hydrocarbons, 184 ff.
10,
13, 47, 54, 166.
Monochromatic waves,
104, 141.
Sharp and unsharp quantities,
154.
Momentum,
Ground states, 34,
Group velocity, 5,
ff. f
193
28 ff., 49.
Transition state, 166.
Triple bond (C 2 S ), 188.
Triplets, 66, 82, 110.
length,
packet,
Zeeman
5, 49.
effect, 55, 60.
Zero-point
117.
3, 7, 29.
energy
(vibration),
PRINTED IN CHEAT BRITAIN

AT THE UNIVERSITY PRESS, OXFORD
BY VIVIAN RIDLER
PRINTER TO THE CNIVERSITY
OXFORD
UNIVERSITY PRESS
Amen
London
Glasgow v
New York
58
House, E.C. 4
Buchanan
Street
417 Fifth Avenue, N.Y. 16
480 University Avenue

Toronto
Melbourne
Bowen Crescent
Wellington Empire Building, Willis
Street
Bombay
Apollo Bunder
Lai Bazar
Calcutta
Madras
Karachi
Dacca
Lahore
Kuala Lumpur
Hong Kong
Cape Town
Ibadan
Accra
Nairobi
Salisbury
26
Mount Road
McLeod Road
Jinnah Avenue
Bank Square
Holland Road
The University
Thibatdt House,
Thibault Square
Iddo Gate
Kwame Nkrumah Avenue

Government Road
Baker Avenue
WAVE
By
the
same Author
THE QUANTUM THEORY

OF RADIATION
Third edition, 1954.
MECHANICS
HE1TLER
Pp. 444, with 37 text-figures

50s. net
{International Series of Monographs on Physics')
The first edition of this important hook, on which so much of the

modern developments in cosmic rays are based, appeared in
1936, and the second edition in 1945. This third edition has been
entirely revised and brought up to date. In particular some of
the most recent developments of the theory are dealt with
thoroughly, including the line shift of hydrogen and the anomalous magnetic
moment
SECOM)
EDITION
of the electron.
The Quantum Theory
of Radiation is a field of great and

increasing importance. It is the basis of the proper theory of the
shapes of spectral lines, which are of leading interest to astrophysicists today. It is the basis also for the study of cosmic rays
as shown by the phenomena they exhibit in their passage through
matter. It is the basis of the theory of the creation and annihilation of positive electrons. Finally, it is the field in which limitations of the present form of the Quantum Theory have already
been suspected and in which they must be further investigated.
As such it is a subject that few physicists can afford to neglect.
This book presents the first systematized account of the theory
in English (or indeed in any other language), and includes also
some account of the theory of the passage through matter of
charged particles of high energy as well as radiation.
'The appearance of a
new
edition of this standard work is an exciting event

working in the held of high energy radiation
experimentalists as well ... it is in every way worthy of
for all theoretical physicists
many
physics and for

iis
Proceedings of the Physical Society.
predecessors.'
(All prices are operative in the United
only and are subject
to alteration
Kingdom
without notice)
OXFORD UNIVERSITY PRESS
Stf,
[8/61]
OXFORD

Elementary Wave Mechanics - Heitler

Uploaded by

Copyright:

Available Formats

Elementary Wave Mechanics - Heitler

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Elementary Wave Mechanics - Heitler

Uploaded by

Copyright:

Available Formats

SECOND

published in 1945. This second

and is about 50 per cent, longer

enlarged, as the book was designed

this respect the

Quantum Chemistry' have been

PROFESSOR OF THEORETICAL PHYSICS IN THE

House, London E.G.

NEW YORK TORONTO MELBOURNE WELLINGTON

PREFACE TO THE SECOND EDITION

BOMBAY CALCUTTA MADRAS KARACHI KUALA LUMPUR

this little book was primarily designed for the use of

tended the chapters on the chemical bond. These are in

REPRINTED I946 (TWICE), ALSO LITHOGRAPHICALLY

EXPERIMENTAL BASIS OF QUANTUM

and electron diffraction

THE HYDROGEN ATOM

Normalization and linear combination of wave

DERIVATION OF THE WAVE EQUATION

ANGULAR MOMENTUM. ZEEMAN-EFFECT.

Sharp and unsharp quantities

electrons with spin

states, directional quantization

PROBLEM OF TWO ELECTRONS

The wave equation for two electrons

General formulation of the Pauli principle

THE PERIODIC SYSTEM OF ELEMENTS

The electron configuration

Quantum states and electron diffraction

The electronic states

which are inconsistent with

The hydrogen molecule

Crossing of atomic interaction curves. Valency

Interaction in diatomic molecules

Interaction of several atoms with one valency

from spectroscopy, shows that an atom cannot

but only in different discrete states of

chemical elements, but are identical

Activation energy, non-localized bonds

Binding energies of hydrocarbons

A great wealth of experimental material,

when the internal energy of an atom

THEORY OF HOMOPOLAR CHEMICAL

are different for the various

EXPERIMENTAL BASIS OF QUANTUM MECHANICS

same energy change,

known that light waves when reflected

whose spacing we can determine the wave-length of the

called a 'diffraction' pattern.

the distance apart of the rows of atoms in the

and m the angle of

the various rings.

be used. The electrons

electrons a cathode-ray tube

beam is thus either lk v or 2hv or Zhv... or

EXPERIMENTAL BASIS OF QUANTUM MECHANICS

pass through an electrical potential difference

Ihe diffraction pattern observed

To develop the theory of wave mechanics we start from

measured from the spacing of the diffracA depends on the