You are on page 1of 74

Introduction to Quantum Mechanics

Victor Hugo Ponce


Instituto Balseiro and Centro At omico Bariloche
Universidad Nacional de Cuyo and Comision Nacional de
Energa Atomica
October 9, 1996
2
Chapter 1
Origins of Quantum Mechanics
1.1 The Classical World Before December 14th,
1900
The physicists of late nineteenth century believed to live in an ordered world,
where (almost) everything was known about the laws obeyed by nature. The
universe was formed by two classes of elements:
1) Matter, aggregate of elementary pieces starting to be identied through
its chemical properties as atoms: particles that occupy a position in space as
a function of time. Matter is identied because when acted on by an external
inuence (force) changes its state of motion, i.e. its velocity.
2) Electromagnetic waves, that where known at the time in the form
of radiation of heat, light and X-rays. Its velocity of propagation already
assumed independent of external inuences.
This world was ruled by Newton

s laws, that related the motion of matter


with the sources of interactions, and by Maxwell

s equations that determined


the elds of force (electromagnetic elds) produced by a known distribution
of electric charges and magnetic dipoles. Added to this formalism, thermo-
dynamics and statistics provided the tools to understand complex systems
formed with many elementary parts.
It seemed that everything important about the behaviour of the universe
was already known, the scientists of those years being left with the task of
solving some problems of second order importance, left behind by the fast
progress of knowledge.
This world was really a little boring, but happily some observations in the
3
4 CHAPTER 1. ORIGINS OF QUANTUM MECHANICS
behaviour of nature could not be explained with the formalism of classical
physics.
In the following section we will describe the rst observation that classical
physics could not describe.
1.2 Black Body Radiation
If a chapter of classical physics passes the examination of daily experience
with highest qualications, that is of thermodynamics. Therefore, nineteenth
century physicists were shocked by their failure to describe the phenomenon
of thermal emission by hot bodies, a problem that could be solved in principle
by using the basic postulates of thermodynamics.
1.2.1 The Black Body and Classical Physics
We will use elementary concepts of thermodynamics to obtain some prop-
erties of the radiation emitted by a hot body. Matter radiates and absorbs
electromagnetic waves because it is formed by bound electric charges capable
of absorbing or giving away energy, the frequency of the radiation being that
of the oscillations of the charges in their bound states. Due to the multitude
of oscillations present in condensed matter, from low frequencies of collective
modes, through molecular bindings, atomic orbitals of electrons and nuclear
bound states of protons and neutrons, the frequency range of radiated energy
is quite wide, and extends to very high frequencies.
We will consider a system in thermodynamical equilibrium, so all its parts
have the same temperature: a cavity in bulk matter whose walls are kept at
a temperature T and the radiation trapped inside. We separate the internal
surface into two sectors that may dier in the chemical composition of the
material and the topography of the surface. There is a mutual interaction
among these regions through the emission and absorption of radiation (see
Figure 1). The emissive power of sector i is dened as the energy radiated at
a given frequency per unit area and unit time: E
i
(, T), while the absorbence
A
i
(, T) is the fraction absorbed of incident radiation of frequency . At
equilibrium the energy ux across each sector should cancel
E
i
(, T) = A
i
(, T)
1.2. BLACK BODY RADIATION 5
At equilibrium, the uxes of energy between the sectors should be equal,
so we conclude that
E(, T)/A(, T) (1.1)
is independent of the material and the form of the cavity. This relation
is called Kirchho

s law for radiation.


If we perform a small hole in the body so that radiation from the cavity
can be emitted, but the energy radiated out is negligible compared with that
contained in the cavity. Any radiation falling from outside on the (small) hole
has a negligible probability of coming out again due to the many reections
on the inside (large) surface of the cavity, so A(, T) = 1, that denes what
is called a black body.
The emittance E of the black body, according to (1.1) independent of
the nature of the cavity, can be related to the energy density u(, T) inside
the cavity, which is isotropic and homogeneous in order to be in thermody-
namical equilibrium (if it were not homogeneous, a temperature gradient will
develop between two absorbing particles placed at points of dierent energy
density; if it were anisotropic, a tempertarue gradient will appear between
two absorbing sheets placed back to back facing the direction of anisotropy).
The emitted ux is geometrically related to the density inside the cavity
through:
E(, T) =
c
4
u(, T)
with c the speed of light.
Let us calculate now the energy density u: according to Kirchho

s law
(1.1), this should be independent of the material of the surface and form of the
cavity. We are then free to assume that the walls behave as those of a perfect
metallic conductor, so the electromagnetic elds should be zero just at the
surface. Therefore, in thermodynamic equilibrium the eld inside the cavity
is formed by waves that have nodes at the walls. Invoquing the principle of
equipartition of energy, we assume that the degrees of freedom of the eld
inside the cavity are in thermodynamical equilibrium with the oscillators at
the cavity walls. In fact, since there is no interaction between a wave with
nodes at the surface and the charges of the medium, the radiation system may
not be in equilibrium with the matter system; to put them in equilibrium, we
introduce in the surface a grain of non-metallic impurity, that may absorb
6 CHAPTER 1. ORIGINS OF QUANTUM MECHANICS
energy from the eld and transfer to the metallic walls and viceversa. It will,
for suciently long times, establish equilibrium between eld and walls. The
hypothesis of stationary waves (with nodes at the surface) is not destroyed
by the impurity.
We will now count the number of degrees of freedom that the eld has in
the cavity. To this end, we assume a cubic cavity (Kirchho

s law assuring
us that the form does not matter for the black body energy density). An
electric eld that cancels at the walls has the general form:

E =

E
0
sin(k
x
x) sin(k
y
y) sin(k
z
z)
with
k
x
=
2n
x
L
, k
y
=
2n
y
L
, k
z
=
2n
z
L
where (n
x
, n
y
, n
z
) are integer numbers.

E
0
has two independent compo-
nents in the plane normal to the wave vector

k = (k
x
, k
y
, k
z
), so there are
two independent stationary waves for each normal mode n
x
, n
y
, n
z
of vibra-
tion of the cavity eld. The number of normal modes between wave vectors
k and k +dk is then:
dn(k)=2[volume between k and k + dk]/[volume belonging to a given
mode]
dn(k) = 2
4k
2
dk
[2/L]
3
=
L
3

2
k
2
dk
and the density in frequency of radiation modes:
() =
1

2
_
2
c
_
3

2
The spectral distribution of energy radiated by the black body becomes:
E(, T) =
1

2
_
2
c
_
3

2
T =
2
c
2

2
T (1.2)
this is the result provided by classical physics, known as Rayleigh

s law
of radiation.[?]
Integrating (1.2) we obtain the total energy irradiated by the cavity:
1.2. BLACK BODY RADIATION 7
E(T) =
_

max

min
dE(, T) =
_

max

min
d
2
2
T
c
2

3
max
as
max
0 (1.3)
We have a nonphysical result, since in actual materials
max
can be very
large so the radiated energy becomes extremely large.
1.2.2 Experimental Results of Black Body Radiation
The rst accurate measurements of black body radiation were performed by
Lummer and Pringsheim between 1897 and 1900[?]. for temperatures in the
range of 400 to 1600

K. Figure 2 show the spectral distribution for the energy


radiated from a black body at three representative values of T.
Max Planck solved this inconsistency in 1900, with the following reason-
ing: for a normal mode with frequency =
c

and a given polarization, the


probability of having an energy E is given by the Boltzmann distribution
P(E) =
e
E/T
T
if the energy E is a continuum variable, the average value E) becomes
T, in accord with the equipartition rule for the energy stored in the electric
and magnetic components of the mode. But if we assume that that there
is only a discrete set of values allowed for the energy, as for example:
E
n
= nh
we obtain
E()) =

n
E
n
P(E
n
) =

n
exp(nh/T)
1 exp(h/T)
nh
=
h
exp(h/T) 1
that exactly coincides with the measured spectral distribution for the
black body made by Lummer and Pringsheim in 1899.
8 CHAPTER 1. ORIGINS OF QUANTUM MECHANICS
The number of modes of vibration in a frequency interval d is obtained
from1.3, with k=
2
c
dn() =
4L
3
c
3

2
d
so, the spectral distribution of the energy density in the cavity results
u(, T) =
8
2
c
3
h
exp(h/T) 1
(1.4)
and the energy density is
U(T) =
_

0
du(, T) =
8
5

4
15h
3
c
3
T
4
(1.5)
which is the well known Stefan-Boltzmann law.
Results 1.4 and 1.5 are veried by experiments. The appeal of the
study of black body radiation comes from the extremely simple
and basic principles in which the classical reasoning is based: ther-
modynamic equilibrium that makes results independent of the ma-
terial in which the cavity is made, equipartition of energy to de-
termine the energy of each mode of the radiation eld, and the
existence of simple experimental results such as the spectral dis-
tribution of the radiation and the Stefan-Boltzmann law that are
not veried by the classical description.
We conclude that an electromagnetic wave with frequency has a
quantied value of energy, where only values nh are allowed. The
quantum h is so small for frequencies in the range of visible light that it
is impossible to observe the loss of continuity of the energy in macroscopic
measurements (h = 6.629 10
27
ergssecond)
The quantication of the energy carried by an electromagnetic wave, needed
to explain the behaviour of nature, had to be added to the principles of clas-
sical physics since December 14, 1900 when Mak Plack presented his papeer
On the Theory of the Energy Distribution Law of the Normal Spectrum
before the members of the German Physical Society in Berlin. Needless to
say, the physics community was very uncomfortable wuth this prescription
assumed to be, even by his author, as a mathematical artice that brings a
theoretical result to coincide with measurements, but an unknown deeper ex-
planation based on the principles of classical physics waited to be discovered.
1.3. PHOTOELECTRIC EFFECT 9
1.3 Photoelectric Eect
When light falls on a metalic surface, it may happen that electrons are emit-
ted. This occurs when the frequency of the radiation is higher than a thresh-
old value
0
characteristic of the metal. The ux of emitted electrons depends
on the radiation intensity, while the fact that the electrons be emitted or not
is independent of it.
Einstein in 1905 explained the fotoelectric eect using the experience
gained in the interpretation of the black body radiation. He generalized the
limitation on the amount of energy carried by a light wave of frequency ,
that was assumed to be restricted to discrete values nh, with n=0,1,2,3,...;
his hypothesis was that light was constituted by particles, or photons, whose
energy was h for the monochromatic wave of frequency . The photoelectric
eect is understood assuming that the electrons can absorb an integer number
of photons, and really at most one photon with any appreciable probability.
When this happens, the electron gains the energy h of the photon, and is
able go out of the metal if this energy is larger than the potential barrie
of the surface. Furthermore, the energy of the emitted electrons is a linear
function of the frequency of the light incident on the surface. See Figure 2.
1.4 Compton eect
The picture of electromagnetic radiation as composed of particles is rein-
forced with the Compton eect, observed when X rays (radiation of frequency
beyond visible and ultraviolet light) falls on a thin foil: the radiation is scat-
tered, and for any given direction of observation beyond the foil radiation
appears with two specic wavelengths.
Assuming that radiation is composed by particles, and since from clas-
sical electrodynamics we know that the energy and momentum of radiation
are related by P=E/c, we assign to each photon the momentum p=h/c. We
now consider the collision of the particle photon, with energy E=h and mo-
mentum p=h/c, with a free electron. Studying the energy and momentum
of a photon dispersed in the direction , conservation of energy and momen-
tum gives the energy (say the wavelength) of the photon as a function of .
Therefore, at a given direction of observation we see radiation with a denite
wavelength = c/, which is greater than the wavelength of the radiation
incident on the foil because the photon gives energy to the recoiling electron.
10 CHAPTER 1. ORIGINS OF QUANTUM MECHANICS
Finally, at any direction of observation we will see radiation with wavelength
equal to the incident beam, produced by the scattering of photons on the
nuclei of the medium, since they are much more massive than the electrons,
their recoil energy is negligible.
Now that we are convinced on the corpuscular nature of radiation, we
may ask ourselves how are we going to reconcile this fact with the wave
properties of this same radiation.
Chapter 2
Wave nature of matter
If electromagnetic waves have properties of particles, shouldnt particles to
show wave properties?. We found that for a photon the linear momentum is
directly related to the wavelength: p=h/c = h/. The french physicist De
Broglie proposed in 1923 that material bodies have an associated wavelength,
dened by the same relation as for photons:
= h/p
The wavelength associated to macroscopic bodies is so small that its wave
properties are unobservable. For example, for a 1 gram particle with a speed
v of 1cm/sec. is: = h = 6.6 10
27
cm.
To verify the De Broglie hypothesis a beam of electrons can be directed
toward the surface of a crystaline solid; see Figure 3.
The wave associated to any given electron should be dispersed by the
atoms of the crystal planes. The waves emerging from each atom in a plane
superpose and propagate with wavefronts that represent constructive inter-
ference along well dened directions; these are the directions where the op-
tical paths coming from the scattering atoms dier in a whole number of
wavelengths. The angles of constructive interference are given by Braggs
law:
sin = n/2a
Electron difraction was observed by Davisson and Germer in 1926, and
the De Broglie assumption veried without any doubt.
11
12 CHAPTER 2. WAVE NATURE OF MATTER
Electron difraction is observed even when the intensity of the electron
beam is so weak that only one electron at a time is interacting with the
crystal. See Figure 4. In this case, a screen that absorbs the electrons
registers each arrival, which appears for example as a point in a photographic
plate used as screen. This means that the electron behaves as a particle when
interacting with the screen. But as more electrons arrive, their distribution
is exactly the same as that predicted by the diraction of the De Broglie
wave.
The presence of wave properties tells us that we should abandon the
notion of a well dened trajectory of particles, because if the electron had a
precise trajectory we would knew from which crystal atom it scattered, and
the result of that scattering should not depend on the presence of atoms.in
the neighborhood.
Chapter 3
The Wave Function
3.1 Wave Packet
Wave phenomena tied to the motion and interaction of particles as is the
case of electron diraction, require the introduction of an associated wave.
This is the De Broglie wave where the wavelength is related to the particle
momentum. By analogy to the case of light particles where the energy is
related to the frequency of the wave, we generalize this connection:
= h/p
= E/h
We build a one-dimensional plane wave with these denitions:
A cos(2x/ 2t) = A cos[(px Et)/ h]
where h = h/2.
If we are going to describe particles with these plane waves, the resulting
wave function should manifest in the space region where it is possible to nd
the particle; so it is necessary to superpose plane waves in order to localize
the particle:
(x, t) =
_
dpA(p) cos[(px Et)/ h] (3.1)
This form can be generalized to three dimensions.
13
14 CHAPTER 3. THE WAVE FUNCTION
The wave function 3.1 will become localized according to the form selected
for A(p). For example, if we take
A(p) = 1forp
0
a < p < p
0
+a
A(p) = 0otherwise
This selection implies that p
0
is the average value of the momentum, and
p = 2a (3.2)
is the dispersion of values. The wave function at time t=0 results
(x, 0) =
_
p
0
+a
poa
dp cos(px/ h) =
2 h
x
sin(ax/ h) cos(p
0
x/ h)
If we assume that p
0
a, the momentum of the packet is well dened.
The form of (x, 0) is given in Figure 5.
The wave packet, centered at the origin of coordinates, is concentrated
roughly between the rst two zeros of the modulating function
h
x
sin(ax/ h);
its space dispersion is then of the order of:
x =
2 h
a
(3.3)
Relations 3.2 and 3.3 satisfy a well known condition for classical waves:
the more localized in space is a wave packet, the larger is the dispersion
in its wavelengths. Translated to momentum and position of a particle we
see that a particle with well dened position has necessarily a poorly dened
momentum, and viceversa:
p.x 2 h
It can be proven that, no matter the form of the wave packet, the product
p.x satises the relation:
p.x > h
This is the famous Heisenberg uncertainty relation. Due to the small
value of Planck constant, the unavoidable dispersion in the values of position
and momentum is negligible for macroscopic bodies.
3.1. WAVE PACKET 15
Let us consider now the wave function at later times t0. Since the energy
of a free particle is E=p
2
/2m , we obtain from 3.1:
(x, 0) =
_
a
a
dq cos[(p
0
+q)x/ h (p
0
+q)
2
t/2m h] (3.4)
where we have used p=p
0
+q. We consider times such that q
2
t/2m h 1,
then 3.4 gives:
(x, t) =
_
a
a
dq cos[(p
0
x p
2
0
t/2m)/ h +q(x p
0
t/m)/ h] (3.5)
=
2 h
x

sin[
ax

h
] cos[(p
0
x

+p
2
0
t/2m)/ h]
where
x

= x p
0
t/m (3.6)
This wave function is exactly the same we show in Figure 5, but now the
maximum of the packet is shifted to the position
x
0
= x p
0
t/m (3.7)
while the term p
2
0
t/2m appearing in the cosine argument has no relevance,
it only changes the phase of the oscillations represented in Figure 5. From
Equation 3.7 we nd that the velocity of propagation of the wave packet in
space is
v =
dx
0
dt
=
p
0
m
(3.8)
the same relation as for a free classical particle.
For long times such that q
2
t/2m h is not negligible, the form of the
wave packet is not conserved, it widens and disperses. This is the
eect also observed for classical waves moving in a dispersive medium, where
the phase velocity of the waves v
p
= v/k is dierent form the group velocity
v
g
= d/dk.
16 CHAPTER 3. THE WAVE FUNCTION
Chapter 4
Wave Equation
We have already dened waves associated to material particles that allow a
space localization and in the case of free particles the packet moves with the
velocity of a classical particle. We need now to determine the shape of these
waves as a function of the forces and constraints that act on the particle.
We start with the case of the free particle, where we know the relation that
connects the wavelength with the frequency:
E = p
2
/2m
with
E = h, p =
h

The wave equation is a dierential operator acting on the wave packet;


it is formed by space and time derivatives. If we think of waves describing a
free particle we have two candidates:
cos[(px Et)/ h], sin[(px Et)/ h]
We propose a linear combination as wave function of the particle:
(x, t) = cos[(px Et)/h]+B. sin[(px Et)/ h] (4.1)
As for the dierential equation, we start suggesting the simplest form.
Since

2
x
2
operating on the trigonometric functions produces a factor p
2
, and

t
a factor E, we propose a wave equation of the form:
[a

2
x
2
+b

t
](x, t) = 0 (4.2)
17
18 CHAPTER 4. WAVE EQUATION
Replacing 4.1 we get:
a

2
x
2
(x, t) =
ap
2
h
2
cos[(px Et)/ h]+B. sin[(px Et)/ h]
b

t
(x, t) =
bE
h
Bcos[(px Et)/ h]sin[(px Et)/ h] (4.3)
These equations are satised when:
B =
1
B
,
ap
2
h
= bEB (4.4)
that gives:
B = i,
a
b
=
i h
2m
(4.5)
We choose b=-i h, so that:
[
h
2
2m

2
x
2
i h

t
](x, t) = 0 (4.6)
and the wave function 4.1 describing the free particle results:
(x, t) = A.e
i(pxEt)/h
(4.7)
Equation 4.6 is the wave equation or Schr odinger equation, who presented
it in 1926 starting the era of Quantum Mechanics. It presents complex coef-
cients, so the solution is a complex function.
We need to nd the physical meaning of the wave function 4.7. The
properties of a particle, such as its mass and charge, energy, position are all
real quantities, while a complex function is formed by two real functions that
we may express as:
(x, t) = (x, t)e
iS(x,t)
(4.8)
We will separate the Schrodinger equation in its real and imaginary parts;
we write:

h
2
2m

2
x
2
(x, t) = i h

t
(x, t) (4.9)
and the complex conjugate:
19

h
2
2m

2
x
2
(x, t) = i h

t
(x, t) (4.10)
Multiplying 4.9 by

and 4.10 by , and sustracting them:

h
2
2m


2
x
2


2
x
2

= i h

t

(4.11)
that we may write as:

h
2m

x
[
2
(x, t)

x
S(x, t)] =

t

2
(x, t) (4.12)
This equation has the same form as the continuity relation for a uid of
density and velocity eld S. If we integrate it between points x
0
and x
1
we
nd:
h
2m

2
(x
0
, t)

x
S(x
0
, t)
h
2m

2
(x
1
, t)

x
S
1
(x, t) =
_
x
1
x
0
dx

2
(x, t)
t
(4.13)
The square of the modulus of the wave function clearly represents a sort
of density attached to the particle. Equation 4.13 is the continuity equation
of this density: the left member gives the balance of the amount that gets
into the region (x
0
, x
1
) per unit time, and the right member says that it is
equal to the net change of the quantity inside the interval.
20 CHAPTER 4. WAVE EQUATION
Chapter 5
Probabilistic Interpretation
The wave packet describing a particle is dierent from zero in the region
where we expect to nd the particle; then if we make x
0
, x
1
in
Equation 4.13,
2
(x
0
, t) =
2
(x
1
, t) = 0 for a particle localized in some nite
region of space, so we get:

t
_

dx
2
(x, t) = 0 (5.1)
this says that the integral over all space of
2
is independent of time.
The physical property whose density is represented by
2
(x, t) is something
independent of time. What kind of density
2
represents?. It can not be a
mass density, because we said before that a wave packet disperses as times go
by, so the space extension of a particle would depend on its age: an electron
born today would have dierent properties from one existing for millions of
years. As we know that this is not the case, we have to think in something
else. The hypothesis put forward by Max Born is that
2
(x, t) represents
the probability density of nding the particle at the position x at
time t. Quantum mechanics will then be a probabilistic theory, where we
know the probabilty of nding a particle as a function of the position. Being
a probability density [ [
2
should satisfy a normalization condition:
_
[ (x, t) [
2
dx = 1 (5.2)
Given a generic wave function:
(x, t) =
_
dpA(p)e
i(pxEt)/h
(5.3)
21
22 CHAPTER 5. PROBABILISTIC INTERPRETATION
the probability of nding the particle between x and x+dx is:
dP(x, t) =[ (x, t) [
2
dx (5.4)
From the point of view of the measurement of the position, this result
says that if we prepare the initial state of the particle at the initial time, let it
evolve up to t and measure the position, we will nd a value x. Repeting this
procedure many times, the distribution of measured x values will be given by
the function
2
(x, t) =[ (x, t) [
2
. The average value of the measured values
is:
x) =
_
dx.x [ (x, t) [
2
(5.5)
We will also assign a physical meaning to the Fourier transform
(k, t) =
1

2
_
dx(x, t)e
ikx
(5.6)
There is a complete symmetry between and since
(x, t) =
1

2
_
dk(k, t)e
ikx
(5.7)
From Equations 5.3 and 5.7 we nd that the variable of the Fourier space
is related to the momentum:
k = p/ h (5.8)
Due to the equivalence of knowing the wave function (x,t) or its Fourier
transform (p/ h, t), it is assumed that represents the wave function of the
particle in momentum space, having in this space the same properties as
in position space; then:
P(p) =[ (p/ h, t) [
2
is the probability density of measuring the momentum p. The average
value of the measurements of momentum is thus:
p) =
_
dp.p [ (p/ h, t) [
2
(5.9)
We will transform this result to position space; from
23
p) =
1
2
_
dk
_
dx

(x, t)e
ikx
_
dx

(x

, t)e
ikx

. hk (5.10)
=
h
2
_
dk
_
dx

(x, t)e
ikx
_
dx

(x

, t)

x

e
ikx

(5.11)
integrating by parts in x

and considering that (, t) = 0 :


p) =
ih
2
_
dk
_
dx

(x, t)
_
dx

(x

, t)e
ik(xx

)
the k-integral is immediate, giving the Dirac delta function:
_
dk.e
ik(xx

)
= 2(x x

)
so we get
p) =
_
dx

(x, t)(i h

x
)(x, t) (5.12)
This result shows the most distinctive features of quantum mechanics:
the probabilistic nature of the results, and the denition of the dynamical
variables, such as the momentum, by dierential operators instead of being
functions of time as in classical mechanics.
The distribution of measured values of momentum is determined if we
know all moments p
n
) of the distribution P(p); they can be calculated as
we did for p) in 5.12:
p
n
) =
_
dx

(x, t)(i h

x
)
n
(x, t) (5.13)
We conclude that if we know the wave function ,the operatori h

x
acting on it determines the distribution of measured values of the momentum
of the particle.
The relation 5.5 can be written in a form similar to 5.12:
x) =
_
dx

(x, t)x(x, t) (5.14)


The operator corresponding to measuring the position x is then the same
variable x.
24 CHAPTER 5. PROBABILISTIC INTERPRETATION
Chapter 6
Schrodinger Equation. Particle
in a One-Dimensional Potential
6.1 Mathematical Description of Physical Mea-
surements
We have found that the dynamical variables of a particle that can be mea-
sured in an experiment, are represented by operators acting on the wave
function. These variables are called observables of the system. The opera-
tor that represents the momentum in the x direction is:
p = ih

x
(6.1)
while the operator representing the position x is:
x = x (6.2)
The distribution in the measurements of these observables is found through
the knowledge of the moments p
n
), x
n
) dened as:
p
n
) =
_
dx

(x, t)( p)
n
(x, t)
x
n
) =
_
dx

(x, t)x
n
(x, t) (6.3)
25
26CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


For a particle moving along one direction the only observable apart from
x and p is the energy. For a free particle this is:

E= p
2
/2m, that as a function
of p has the following form of dierential operator:

H(p) =
h
2
2m

2
x
2
(6.4)
We will call Hamiltonian to the operator that represents the energy of the
system. The Schr odinger equation 4.6 that determines the time evolution of
a free particle can be written in terms of the Hamiltonian of the particle:
[

H(p) i h

t
](x, t) = 0 (6.5)
When a force is acting on the particle, there is a term of potential energy
in the Hamiltonian. We will assume that the function V(x,t) that denes
the potential energy in quantum mechanics is the same that denes the force

F =
V
x
in classical mechanics. As the position operator x is directly the
coordinate x, the Hamiltonian (representing the total energy) becomes:

H(p, x) =
h
2
2m

2
x
2
+V (x, t) (6.6)
The generalization of the Schr odinger equation when a potential acts on
the particle results from keeping the form 6.5 of the equation, with the use
of the Hamiltonian 6.6:
[
h
2
2m

2
x
2
+V (x, t) i h

t
](x, t) = 0 (6.7)
In the case of a particle moving in three dimensions the Schr odinger
equation generalizes to:
[
h
2
2m

2
r
+V (

r , t) i h

t
](

r , t) = 0 (6.8)
where
2
r
=

2
x
2
+

2
y
2
+

2
z
2
is the Laplacian operator.
6.2 Stationary States
We now consider the case where the force acting on the particle is conserva-
tive, so the potential energy is independent of time. In classical mechanics
this leads to the conservation of the total energy of the particle.
6.2. STATIONARY STATES 27
The solutions of the Schr odinger equation
[
h
2
2m

2
x
2
+V (x) i h

t
](x, t) = 0 (6.9)
can be obtained by the method of separation of variables:
(x, t) = (x).(t)
giving:
1
(x)
[
h
2
2m

2
x
2
+V (x)](x) =
i h
(t)

t
(t) = 0 (6.10)
For this equation to be valid it is necessary that each member be equal
to a constant that we name E:
[
h
2
2m

2
x
2
+V (x)](x) = E(x) (6.11)

t
(t) =
iE
h
(t) (6.12)
Equation 6.12 is easily solved:
(t) = e
iEt/h
(6.13)
while Equation 6.11, the stationary Schr odinger equation, encloses
all the features of the forces acting on the particle, and gives the distributions
of measured momentum and position. We need to know which values can be
assigned to the separation constant E; the answer is any value that produces
a well dened wave function for the particle. For example, from:
_
dx [ (x, t) [
2
=
_
dx
2
(x) = 1 (6.14)
we see that
2
should be integrable. (We assume that is real because the
equation 6.11 dening it has real terms). There may be many values of E that
produce functions satisfying 6.14, we call them eigenfunctions corresponding
to the eigenvalue E:

H(p, x)
E
(x) = E
E
(x) (6.15)
The equation 6.9 that we write as:
28CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


[

H(p, x) i h

t
](x, t) = 0 (6.16)
describes the time evolution of the wave function. If we prepare the
particle in such a way that at time t
0
is described by a wave function
(x, t
0
) = f(x)
equation 6.16 will give us the wave function (x, t) at any later time tt
0
.
The initial condition should be a normalized function
_
dxf
2
(x) = 1 in order
to describe a density of probability. We could then have as initial conditions
any of the eigenfunctions 6.15
(x, t
0
) =
E
(x) (6.17)
so the wave function at later times becomes
(x, t) =
E
(x)e
iEt/h
(6.18)
The density of probability of nding the particle becomes:
[ (x, t) [
2
=
2
E
(x) (6.19)
that is independent of t. For that reason, the states 6.18 are called sta-
tionary states. If we measure the energy for a stationary state we get, using
6.15
H) =
_
dx

(x, t)[
h
2
2m

2
x
2
+V (x)](x, t) (6.20)
= E
Furthermore, the average value of the square of the energy results H
2
) =
E that togeter with H) = E says that the distribution of measured energy
values has zero dispersion:H
2
)H)
2
= 0. See Figure 6. This means that we
have a delta distribution of energy values and there is certainty of obtaining
the value E when measuring the energy of the particle. The wave functions

E
(x)e
iEt/h
describe states with a precise value of energy for a particle acted
on by a conservative force, in the same way as the plane waves did for a free
particle.
6.2. STATIONARY STATES 29
The most general wave function for the motion will be a linear combina-
tion of eigenstates:
(x, t) =

{E}
a
E

E
(x)e
iEt/h
(6.21)
the coecients a
E
are only subject to the condition of normalization

{E}
[ a
E
[
2
= 1. It can be proved that the functions
E
(x) form a complete
basis set, this means that any function can be written as a linear combi-
nation of them. Furthermore, we prove now that the eigenfunctions
E
(x)
are orthogonal: we study the matrix element
_
dx

(x)

H(p, x)(x) between


two generic wave functions , and treat by integration by parts the kinetic
energy term:
_
dx

(x)
2
(x)/x
2
=

(x).(x)/x [


_
dx

(x)/x(x)/x
the rst term on the right cancels, and integrating again by parts:

_
dx

(x)/x(x)/x =

(x)/x(x) [

+
_
dx
2

(x)/x
2
(x)
so that we obtain the important result that: in a matrix element
[

H [ ) the action of the Hamiltonian on the wave function at
the right of it is equivalent to the action of the Hamiltonian on the
wave function at the left:
_
dx

(x)

H(p, x)(x) =
_
dx

H(p, x)

(x)(x) (6.22)
Operators that have this property are called Hermitian:
This result applied to any pair of eigenfunctions
E
(x),
E
(x)
_
dx

E
(x)

H(p, x)
E
(x) = E
_
dx

E
(x)
E
(x)
_
dx

H(p, x)

(x)(x) = E

_
dx

E
(x)
E
(x)
indicates that the right members are identical:
(E E

)
_
dx

E
(x)
E
(x) = 0 (6.23)
so that they are orthogonal for E,= E

.
30CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


The energy levels that have more than one eigenfunction are
called degenerate. For them Eq.6.23 does not determine the orthogonality
of degenerate eigenfunctions, but since any combination of them

a
j

j
E
(x)
is also an eigenfunction with eigenvalue E, we are able to orthogonalize the
set
j
E
(x), and conclude that the set of eigenfunctions of

H can always be
assumed orthogonal.
With this result, we can write the average value of the energy for a general
wave function of the form 6.21
H) =

{E}
[ a
E
[
2
E (6.24)
6.3 Continuity of the Wave Function for a
One-Dimensional Potential Well.
We consider a potential function of the form:
V (x) = 0for a<x<a (6.25)
= V
0
for [ x [> a
The potential energy V(x) that appears in the Hamiltonian of the quan-
tum formalism is assumed to be the same that gives the force in the classical
version of the problem:
F(x) = dV (x)/dx (6.26)
then the functional form of V(x) is obtained by looking at the force acting
on a classical particle placed in the environment that denes our quantum
system. Most of the forces have already well known potentials, specially
the electromagnetic forces that act on the atomic, molecular and solid state
systems. The potential energy of particles of charge q
1
, q
2
is then of the
form q
1
q
2
/ [

r
1

r
2
[, where

r
1
,

r
2
are the positions of the particles. The
potential energy of a nucleon (either proton or neutron) when it belongs to
an atomic nucleus can be approximated by the form (6.25); this fact has been
extracted from many observations of the properties of nuclei, because in this
6.3. CONTINUITYOF THE WAVE FUNCTIONFORAONE-DIMENSIONAL POTENTIAL WELL.31
case we do not have an equivalent classical system to provide us with the
space dependence of the force.
Since the force is the derivative of the potential, this must be a continuous
derivable function. Our potential well should then be of the of the form
presented in Figure 7. The stationary Schrodinger equation that gives the
allowed values E for the energy is:
[
h
2
2m

2
x
2
+V (x)](x) = E(x) (6.27)
with V(x) given by Eq.6.25, except in the immediate neighborhood of the
points x=a, as seen in Figure 7. If these regions are very small, we will
determine the wave function only outside them, and connect the solutions
outside the interval [ x [< a with the solution inside it. Let us assume
we know the wave function and its derivative at the point x=a-: (a
),

(a ). To determine (a + ),

(a + ) we integrate

2

x
2
between a-
and x (a- < x < a +) using equation 6.27:

(x)

(a ) =
2m
h
_
x
a
dx[V (x) E](x) (6.28)
We get the following relation:

(x)

(a ) <
2m
h
E. max[V (x) E](x).(x a +) (6.29)
where max[V (x) E] is the maximum value that function reaches in
the interval a-, a +. For x=a+ this result gives.

(a +)

(a ) <
2m
h
max[V (x) E](x).2 (6.30)
so that when the interval 2 where the potential varies form V
0
to 0 is
made innitesimal we get

(a +) =

(a ) (6.31)
If we integrate again equation 6.29 we get
_
a+
a
dx

(x) <
_
a+
a
dx

(a)+
2m
h
max[V (x)E](x)
_
a+
a
dx(xa+)
32CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


=

(a )2 +
2m
h
max[V (x) E](x)2
2
(6.32)
so for small we have
(a +) = (a ) (6.33)
Results 6.31,6.33 say that the wave function and its derivative are
continuous at a point where the potential presents a nite discon-
tinuity (which is what happens in our Figure 7 when 0). Therefore,
the stationary Schrodinger equation 6.27 for the one-dimensional square well
in the limit of almost vertical walls ( 0) results:
[
h
2
2m

2
x
2
+E]
i
(x) = 0for [ x [< a (6.34)
and
[
h
2
2m

2
x
2
+ (E V
0
)]
e
(x) = 0for [ x [> a (6.35)
with the following conditions of continuity at the boundary x=a:

i
(a) =
e
(a)

i
(a) =

e
(a) (6.36)
6.4 Symmetry of the Potential and Parity of
the Eigenfunctions.
The Hamiltonian for the one-dimensional motion of a particle is
H(x) =
h
2
2m

2
x
2
+V (x) (6.37)
We notice that choosing the origin of coordinates at the center of the
potential well, the Hamiltonian is independent of the sign of the coordinate
x:
H(x) = H(x) (6.38)
6.4. SYMMETRYOF THE POTENTIAL ANDPARITYOF THE EIGENFUNCTIONS.33
This apparently trivial condition leads to a simplication in the task of
nding the eigenfunctions of the problem. It is also the rst and simplest
example of what we may call the constants of the motion of a quantum sys-
tem. In classical mechanics there are dynamical variables whose value does
not depend on time, such as the energy and the three components of the an-
gular momentum for the planetary motion. The equivalent cases in quantum
mechanics are those magnitudes whose value is known with certainty and
remains independent of time.
We have found the condition for the energy to be a constant of the motion
in Section6.2: the Hamiltonian operator should be independent of time. In
classical mechanics (see H.Goldstein, Classical Mechanics, Addison Wesley
Pub.Co., Reading, Mass., 1959) a variable not appearing in the Hamiltonian
is called a cyclic variable, and this absence determines that the canonical
moment P
q
conjugate to the variable q is a constant of the motion as seen
from the Hamilton equation
dP
q
/dt = H(P
q
, q; t)/q
The spatial coordinate x has as conjugate moment the linear momentum
p
x
, while time has as conjugate moment the energy: P
t
= E.
Returning to our example: which should be the magnitude that is con-
served due to the absence of the sign of x in the functional dependence of
the Hamiltonian?. To get the answer in quantum mechanics we dene the
operation of changing the sign of x for a function like the Hamiltonian or the
wave function:

P(x)f(x) = f(x) (6.39)


so that:

P(x)

H(x)(x) =

H(x)

P(x)(x) (6.40)
This relation says that the Hamiltonian is transparent to the parity
operation, and it is valid for any function , so that

P(x)

H(x)

H(x)

P(x) = 0
meaning that the operators

P and

H can be commuted, so that the
commutator is zero:
[

P(x),

H(x)]

P(x)

H(x)

H(x)

P(x) = 0 (6.41)
34CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


Let us see the eect of operating with

P on both members of the station-
ary Schr odinger equation:

P(x)

H(x)(x) =

P(x)E(x)
Using Eq.6.41 to commute operators on the left member, we get:

H(x)

P(x)(x) = E

P(x)(x) (6.42)
This results says that

P(x)(x) = (x) is also an eigenfunction of

H
with the same eigenvalue E as the original state (x). If the level E is non-
degenerate we must have

P(x)(x) to be identical to (x) (except at most
for a constant factor):

P(x)(x) = p(x) (6.43)


The possible constant factor p can be determined applying

P twice on
(x)

P(x)

P(x)(x) =

P(x)p(x) = p
2
(x)
using directly the denition Eq.6.39:

P(x)

P(x)(x) =

P(x)(x) = (x)
we get:
p = 1
Therefore:

P(x)(x) = (x) = (x)


If the level E is degenerate

P(x)(x) is not required to be proportional to
(x); but if both are eigenfunctions of energy E, so do are the combinations:

(x) = (x)

P(x)(x) = (x) (x)
which are even and odd functions of x:

P(x)

(x) =

(x)
This concludes the demonstration that when the Hamiltonian is indepen-
dent of the sign of the coordinate x the parity operator commutes with

H,
and the energy eigenfunctions can always be chosen as eigenfunctions of the
parity.
6.5. BOUND STATES IN A SQUARE POTENTIAL WELL 35
6.5 Bound States in a Square Potential Well
The stationary states for a particle in a square well are described by the
eigenfunctions
[
h
2
2m

2
x
2
+V (x)]
E
(x) = E
E
(x) (6.44)
with
V (x) = 0for a<x<a
= V
0
for [ x [> a
They have a well dened parity, so we start studying the set of even
eigenfunctions.
For [ x [< a Eq.6.44 results:

h
2
2m

2
x
2

E
(x) = E
E
(x), [ x [< a
whose solution of even parity is:

E
(x) = A. cos(kx) (6.45)
where
k =
_
2mE/ h
2
For [ x [> a Eq.6.44 gives:
[
h
2
2m

2
x
2
+V
0
]
E
(x) = E
E
(x)
We consider rst the case where we look for solutions with EV
0
.
Then, the normalizable solution of even parity results:

E
(x) = B.e
x
, forx>a (6.46)

E
(x) = B.e
x
, forx<a
where:
36CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


=
_
2m(V
0
E)/ h
2
Wave functions of this type are shown in Figure 8. The constants A,B
are determined through normalization and connection of the wave function
at the points x=a. Continuity of
E
(x) and its derivative at both points
give:
A. cos(ka) = B.e
a
A.k. sin(ka) = B..e
a
at x=a, and:
A. cos(ka) = B.e
a
A.k. sin(ka) = B..e
a
at x=-a. These relations bring two independent equations, that together
with the normalization
_
dx [
E
(x) [
2
= 1
determine the constants A,B and the eigenvalue E so that the wave func-
tion is continuous, derivable and normalizable. The three conditions can be
expressed as:
Normalization:
A
2
_
a
a
dx. cos
2
(kx) + 2B
2
_

a
dxe
2x
= 1 (6.47)
Continuity of wave function:
A. cos(ka) = B.e
a
(6.48)
Continuity of logarithmic derivative

E
(a)/
E
(a) :
= k. tan(ka) (6.49)
6.5. BOUND STATES IN A SQUARE POTENTIAL WELL 37
Eqs.6.47 and 6.48 determine the factors A,B, while Eq.6.49 xes the phys-
ical value for the energy eigenvalue E; expressed in terms of the wave number
k:

_

2
(ka)
2
1 = tan(ka) (6.50)
where:

2
= 2mV
0
a
2
/ h
2
, E = h
2
k
2
/2m (6.51)
Equation 6.50 presents always at least one solution, because the left mem-
ber decreases from innite to zero for k growing from zero to , and the right
member goes from zero to innite for 0 ka /2. It is possible to have sev-
eral solutions when > , they are shown as the intersections of the curves
representing both members of Eq.6.50 in Figure 9. Each solution E
n
denes
a stationary state
n
(x). We have found a distinctive feature of quantum
mechanics: the energy of bound systems appears only in discrete
values.
The same procedure can be applied to nd the odd eigenfunctions, whose
form is:

E
(x) = A. sin(kx), for [ x [< a (6.52)

E
(x) = B.e
x
, forx>a

E
(x) = B.e
x
, forx<a
and are presented in Figure 10. The continuity and normalization condi-
tions now give:
A
2
_
a
a
dx. sin
2
(kx) + 2B
2
_

a
dxe
2x
= 1 (6.53)
A. sin(ka) = B.e
a
= k. cot(ka)
The last equation gives the energy eigenvalues for the odd stationary
states:
38CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL

_

2
(ka)
2
1 = cot(ka) (6.54)
That does not have solutions for ka/2. Figure 11 shows the graphical
solutions of this equation.
6.6 Continuum States. Normalization Con-
dition
We have found the stationary states of a particle in a potential well, cor-
responding to eigenvalues of the energy smaller than the depth of the well.
The solutions produce discrete values of E and normalized and continuous
wave functions.
We look now for solutions with EV
0
; for even parity they are:

E
(x) = A. cos(k
1
x), for [ x [< a

E
(x) = B. cos(k
2
x), for [ x [> a
with:
k
1
=
_
2mE/ h
2
k
2
=
_
2m(E V
0
)/ h
2
This function can not be normalizaed, since for A,B,= 0 :
A
2
_
a
a
dx. cos
2
(k
1
x) + 2B
2
_

a
dx cos
2
(k
2
x) =
We see that the mean probability of nding the particle far away from the
potential well is nite:
1
2
B
2
. It is not possible to obtain normalized stationary
states with EV
0
of a particle in the potential, but we may think of a situation
where there are many particles lling the space close and away from the region
where V(x),= 0. The physical situation is that of a stationary ux of particles
coming to the region of the potential well, being scattered by it and receding
from it. The symmetry of parity is lost, because the boundary condition
6.6. CONTINUUM STATES. NORMALIZATION CONDITION 39
is already unsymmetric, as we see in Figure 12. Furthermore, the square
modulus of the wave function does not represent anymore the probability
density of nding a particle, now it describes the number of particles found
per unit volume.
The particles in the incoming beam should be far away one from another
so their mutual interactions can be neglected, and each one of them obeys
the Schr odinger equation 6.27. The incoming beam is represented by a plane
wave moving, for example, from left to right:
A.e
ikx
The boundary condition determines the amplitude A, it is related to the
ux of particles through Eq.4.11, of the form:
d
dx
Flux =
d
dt
Density (6.55)
where
Flux =
i h
2m

d
dx

d

dx
(6.56)
For a plane wave = e
ikx
:
Flux =[ A [
2
hk
m
(6.57)
The ux is the number of particles coming in per unit area and unit time.
We will consider in detail a simple situation of particles in unbound states
of a potential, such as the step potential: (see Figure 13a)
V (x) = 0, forx<0 (6.58)
V (x) = V
0
forx>0
The general solution for x0 of
[
h
2
2m

2
x
2
]
E
(x) = E
E
(x)
is the combination of plane waves with momenta k :

E
(x) = A.e
ikx
+B.e
ikx
(6.59)
40CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


and
k =
_
2mE/ h
2
The factor A is determined by the boundary condition that xes the ux
through Eq. 6.57. For x0 the solution of
[
h
2
2m

2
x
2
+V
0
]
E
(x) = E
E
(x) (6.60)
presents two dierent forms, depending on the sign of E-V
0
:
For EV
0
the general solution is:

E
(x) = C.e
x
+D.e
x
(6.61)
(see Figure 13b).The condition that the wave function should be nite
for x requires that D=0. The undetermined factors B,C are xed by
the continuity of the wave function at x=0:
A +B = C
and the continuity of its derivative:
ik(A B) = C
that produces:
B = A
k i
k +i
(6.62)
C = A
2k
k +i
(6.63)
We notice that there is no condition on the value chosen for E, so E can
have any real value. We get a continuum of levels for unbound states.
The component
B.e
ikx
represents the particles reected by the step; its ux is -[ B [
2 hk
m
, since
[ B [=[ A [ it cancels the incoming ux, so there is no build up or depletion
of particle density in the space around the step.
For EV
0
the general solution is:
6.6. CONTINUUM STATES. NORMALIZATION CONDITION 41

E
(x) = C.e
ik

x
+D.e
ik

x
(6.64)
(see Figure 13c), with:
k

=
_
2m(E V
0
)/ h
2
(6.65)
Since there is no incoming ux from the right, we must set D=0. The
conditions of continuity now read:
A +B = C
ik(A B) = ik

C
that produce:
B = A
k k

k +k

(6.66)
C = A
2k
k +k

(6.67)
The ux reected by the step is smaller that the incoming ux:
R =[ B [
2
hk
m
= [ A [
2
hk
m
(
k k

k +k

)
2
(6.68)
The ux transmitted over the step is:
T =[ C [
2
hk

m
=[ A [
2
hk

m
(
2k
k +k

)
2
(6.69)
We see that the sum R+T of reected and transmitted uxes equals the
incident ux [ A [
2 hk
m
.
42CHAPTER 6. SCHR

ODINGER EQUATION. PARTICLE INAONE-DIMENSIONAL POTENTIAL


Chapter 7
Schrodinger Equation in More
than One Dimension
7.1 Symmetries and Constants of the Motion
We studied the symmetry of parity for the Hamiltonian of a one-dimensional
system. We saw there that the invariance of the Hamiltonian under the
inversion of the coordinate x meant that the energy eigenfunctions were also
eigenfunctions of the parity operator. This is a simple case of a powerful
feature of the quantum formalism, that helps in nding the stationary states
by looking at the symmetry operations that commute with the Hamiltonian.
We consider the less trivial case of a two dimensional system, where the
potential depends on only one coordinate:
[
h
2
2m
(

2
x
2
+

2
y
2
) +V (x)]
E
(x, y) = E
E
(x, y) (7.1)
with:
V (x) = V
0
, for [ x [< a (7.2)
= 0, for [ x [> a
(See Figure 14).
The coordinate y is missing in the dependence of the Hamiltonian, in
classical mechanics this means that the conjugate moment, the linear mo-
mentum p
y
, is a constant of the motion. In quantum mechanics this signies
that the operator p
y
commutes with

H ; the commutator
43
44CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
[ p
y
,

H] = [ih

y
,

H] = 0 (7.3)
is zero because the Hamiltonian does not depend on y. The operators p
y
and

H have a set of simultaneous eigenfunctions, as can be seen from the
equation 7.1, that is separable in x and y variables; writing.

E
(x, y) = (x)(y)
1
(x)
[
h
2
2m

2
x
2
+V (x) E](x) =
1
(y)
h
2
2m

2
x
2
(y) (7.4)
the solutions are:
(y) = e
iqy
and (x) is a solution of the type we found before, (see equations 6.44,
6.45, 6.46, 6.52) :
1
(x)
[
h
2
2m

2
x
2
+V (x)](x) = (E
h
2
q
2
2m
)(x) (7.5)
The wave function

E
(x, y) = e
iqy/h
(x) (7.6)
is obviously an eigenfunction of both

H and p
y
.
We analyze the symmetry operation that the momentum p
y
. The trans-
lation along the coordinate y: for a function F(y) in an innitesimal distance
dy is:
F(y +dy) = F(y) +
dF(y)
dy
dy
and can be expressed in terms of the momentum operator p
y
= ihd/dy :
F(y +dy) = F(y) +
idy
h
p
y
F(y)
For a nite translation b the function F(y+b) can be expressed in terms
of the original F(y) by using the Taylor expansion:
7.2. SIMULTANEOUS EIGENFUNCTIONS OF COMMUTINGOPERATORS45
F(y +b) =

n=0
b
n
n!

n
y
n
F(y)
that can be written as:
F(y +b) =

n=0
(ib p
y
/h)
n
n!
F(y)
where we recognize the expansion of the exponential:
F(y +b) = e
ib py
F(y)
The operator of linear momentum p
y
is then the generator of translations
along the direction y. The Hamiltonian of our example is invariant under
those translations because it does not depend on y.
7.2 Simultaneous Eigenfunctions of Commut-
ing Operators
A fundamental property of the eigenfunctions of an operator

A is that
they can be taken to be simultaneous eigenfunctions of all the
operators that commute among themselves and with

A. We prove
this property for two commuting operators, and the procedure can be easily
generalized to an arbitrary number of them. If:

Af(x) = af(x) (7.7)


and
[

A,

B] = 0 (7.8)
then:

A.

Bf(x) =

B.

Af(x) =

Baf(x) = a

Bf(x) (7.9)
so

Bf(x) is also an eigenfunction of

A. There are two possibilities:
a)that the eigenvalue a is non-degenerate so there is only one eigenfunc-
tion that satises 7.7, then

Bf(x) can dier from f(x) at most by a constant
factor:
46CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION

Bf(x) = bf(x) (7.10)


but this means that f(x) is also an eigenfunction of

B, which proves our
thesis.
b) if the eigenvalues is degenerate there is more than one eigenfunction
that satises 7.7; let us consider the case of two degenerate eigenfucntions:

Af
1
(x) = af
1
(x)

Af
2
(x) = af
2
(x) (7.11)
In fact, any linear combination is an eigenfunction:

Ad
1
f
1
(x) +d
2
f
2
(x) = ad
1
f
1
(x) + d
2
f
2
(x) (7.12)
Since

B commutes with

A from 7.9

Bf(x) is an eigenfunction of

A , so
that:

Bf
1
(x) = c
11
f
1
(x) +c
12
f
2
(x)

Bf
2
(x) = c
21
f
1
(x) +c
22
f
2
(x)
that can be written in matrix form:

B
_
f
1
f
2
_
=
_
c
11
c
12
c
21
c
22
_

_
f
1
f
2
_
or in shorthand notation:

B

f = C

f (7.13)
The operator

B acts on the coordinates of the functions f
1,2
, so it com-
mutes with a matrix of constant numbers. We can use this to multiply 7.13
on the left by a matrix M and get

BM

f = MCM
1
.M

f (7.14)
since M is arbitrary, we may choose it to diagonalize the matrix C
MC = dM
7.3. DEGENERATE LEVELS 47
producing:

BM

f = d.M

f (7.15)
There are two matrices M with two eigenvalues d that diagonalize 7.14(there
are n in a space of n dimensions), so we get combinations M

f that are eigen-
functions of

A (Eq.7.12) and also of

B (Eq.7.15). This completes the proof.
7.3 Degenerate Levels
The example of the two-dimensional motion of a particle in a potential that
depends on only one of the coordinates also serves to see the conditions
required to have degenerate energy levels. We notice that not only the linear
momentum in the y direction commutes with the Hamiltonian, but also the
parity operator of inversion of this coordinate:

P(y)f(y) = f(y)
[

H,

P(y),

P(y)

H] = 0
The eigenfunctions can not be chosen as simultaneous eigenfunctions of

H, p
y
and

P(y) because the last two do not commute:
[ p
y
,

P(y)]f(y) = ih
d
dy

P(y)f(y) + ih

P(y)
d
dy
f(y)
= ih
d
dy
f(y) + ih

P(y)
d
dy
f(y)
the derivative is a function of y: f(y)=
d
dy
f(y), so
d
dy
f(y) =
d
d(y)
f(y) =
f

(y); so the commutator acting on any function f(y) results:


[ p
y
,

P(y)]f(y) = 2ihf

(y)
and we arrive at:
[ p
y
,

P(y)] = 2ih

P(y)
d
dy
(7.16)
48CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
Going back to our example, we may build eigenfunctions of

H and either

P(y) or p
y
; if the set
E,q
is
(

H E)
E,q
= 0
( p
y
q)
E,q
= 0
since [

H,

P(y)]=0, the state

P(y)
E,q
is also an energy egienfunction
with eigenvalues E, but it can not be simply proportional to
E,q
because
[ p
y
,

P(y)] ,= 0. The conclusion is that when in the set of symmetry op-
erations that leave the Hamiltonian invariant, there are some that
do not commute among themselves, the energy levels are degener-
ate. In our case, accompanying the
eigenfunction given by Eq.7.6

E
(x, y) = e
iqy/h
(x)
there is the degenerate state

P(y)
E
(x, y) = e
iqy/h
(x) (7.17)
that describes a particle moving freely in the opposite sense along the y
direction.
7.4 Particle in a Three Dimensional Central
Potential
We will study the evolution of a particle in a eld of central forces. In classical
mechanics the angular momentum is a constant of the motion for this system.
The denition of angular momentum in quantum mechanics uses the same
function dependence on coordinate and momentum as the classical case:

L =

p (7.18)
so:
L
x
= y.p
z
z.p
y
7.4. PARTICLE INATHREE DIMENSIONAL CENTRAL POTENTIAL49
L
y
= z.p
x
x.p
z
L
z
= x.p
y
y.p
x
(7.19)
but now p
x
, p
y
, p
z
are dierential operators:
p
x
= ih

x
, .....
The commutation relations between the components of the angular mo-
mentum are:
[

L
x
,

L
y
] =

L
x

L
y


L
y

L
x
(7.20)
= h
2
(y.p
x
x.p
y
)
= ih

L
z
and
[

L
y
,

L
z
] = ih

L
x
(7.21)
[

L
z
,

L
x
] = ih

L
y
(7.22)
Furthermore, the square modulus

L
2
commutes with any of its compo-
nents:
[

L
2
,

L
i
] = 0 (7.23)
The angular momentum components are the generators of rotations: if
we rotate a function f(

r ) through an innitesimal angle and around an axis


dened by

, the function transforms into f(

r ), and this can be


expressed as:
f(

r

r ) f(

r )

r .

r
f(

r ) (7.24)
= f(

r )

r

r
f(

r ).

= f(

r ) +

ih
.

Lf(

r )
50CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
Finite rotations are then produced by the operator:

R = e
i


L/h
(7.25)
We express the angular momentum operator in spherical coordinates,
which is the natural system for central forces:

L
x
= y. p
z
z. p
y
(7.26)
= ih(sin

+ cot cos

L
y
= z. p
x
x. p
z
(7.27)
= ih(cos

+ cot sin

L
z
= x. p
y
y. p
x
(7.28)
= ih

The Hamiltonian for a particle of mass in a central force eld is:

H(

r ,

p ) =
h
2
2

r
+V (r) (7.29)
The Laplacian term the represents the kinetic energy can be expressed in
spherical coordinates:

h
2
2

r
=
h
2
2

2
r
2

h
2

1
r

r
+
h
2
2
1
r
2

L
2
(7.30)
Therefore, we nd that

H,

L
2
and one of the components

L
i
commute
among themselves; these mutually commuting operators dene a common
set of eigenfunctions; the form of

L
2
in spherical coordinates is:

L
2
= h
2
(

2

2
+ cot

+
1
sin
2

2
(7.31)
The eigenfunctions of this dierential operator:
7.4. PARTICLE INATHREE DIMENSIONAL CENTRAL POTENTIAL51

L
2
Y

(, ) = Y

(, ) (7.32)
are the well known spherical harmonics
Y
m
l
(, ) = N.P
m
l
().e
im
(7.33)
where N is normalization factor, and P
m
l
the associated Legendre func-
tions:
(

2

2
+ cot


m
2
sin
2

+

h
2
)P
m
l
() = 0 (7.34)
To have a continuos function in the coordinate dened in the range
0 2:
e
im
[
=0
= e
im
[
=2
we must restrict m to be an integer. Furthermore, to have a nite function
in the interval of denitions 0 , the allowed values of should be
restricted to:
= h
2
l(l + 1) (7.35)
and those of m are further restricted to be
l m l (7.36)
The basic condition of the quantum formalism that the wave
function should be nite and continuous restricts the allowed values
of the observables of the system. We have just found that the square
modulus of the angular momentum should only present discrete values h
2
l(l+
1) with l=0,1,2..., and any of its projections should be hm, with -l m l.
The wave function that describes the evolution of a particle in three di-
mensions can be expanded in the complete basis of the spherical harmonics.
When measuring the square modulus of L we will nd a distribution of values
h
2
l(l +1) , while the measurement of a projection will produce a distribution
of results quantized in units of the constant h. We notice that the measure-
ment of a projection never reaches the value of the modulus:
hm <
_
hl(l + 1)
52CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
The reason is that if L
2
z
) = L
2
), then since: L
2
) = L
2
x
+ L
2
y
+ L
2
z
) it
will result L
2
x
) = L
2
y
) = 0, and all three components would be known with
certainty, which is not possible because the operators of the components of

L do not commute among themselves. Only when l=m=0 is the values of


the three components are known with certainty.
7.5 The Hydrogen Atom
7.5.1 Center of Mass and Relative Coordinates
We consider the two-particle problem of an electron of mas m
e
and charge -e
and a point nucleus of mass M charge Ze. The interaction is approximated
by the electrostatic potential V(r)=
Ze
2
r
, where r=[

r
e

R [, with

r
e
,

R the
positions of electron and nucleus.
The stationary Schr odinger equation for the two-particle system is:
(
h
2
2M

R

h
2
2m
e

2

r
e

Ze
2
r
)(

r
e
,

R) = E(

r
e
,

R) (7.37)
Since the potential energy depends only on the relative distance r, it is
convenient to use center of mass and relatives coordinates

R
CM
=
M

R +m
e

r
e
M +m
e

r =

r
e

R (7.38)
This gives


R
=
M
M +m
e

R
CM

r
e
=
m
e
M +m
e

R
CM
+

r
so the kinetic energy results:
7.5. THE HYDROGEN ATOM 53

h
2
2M

2

R

h
2
2m
e

2

r
e
=
h
2
2M
(
M
M +m
e
)
2

R
CM
+
2

r
2
M
M +m
e

R
CM
.

r
)

h
2
2m
e
(
m
e
M +m
e
)
2

R
CM
+
2

r
+ 2
m
e
M +m
e

r
.

R
CM

that reduces to:

h
2
2M

2

R

h
2
2m
e

r
e
(7.39)
=
h
2
2M
T

R
CM

h
2
2

r
where
M
T
= M +m
e
=
Mm
e
M +m
e
The Schr odinger equation expressed in center of mass and relative coor-
dinates is:
(
h
2
2M
T

R
CM

h
2
2

r

Ze
2
r
)(

r ,

R
CM
) = E(

r ,

R
CM
) (7.40)
We notice that

R
CM
is a cyclic coordinate (the Hamiltonian does not
depend on it), so the canonical moment is a constant of the motion. This is
the total linear momentum of the system:

P
CM
= ih

R
CM
that commutes with the Hamiltonian:
[

P
CM
,

H] = 0 (7.41)
54CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
The stationary states are separable: (

r ,

R
CM
) = (

R
CM
)(

r ) and are
simultaneous eigenfunctions of

P
CM
and

H :

h
2
2M
T

R
CM

K
(

R
CM
) = E
K

K
(

R
CM
) (7.42)
(
h
2
2

2

r

Ze
2
r
)

r ) =

r ) (7.43)
with:
E = E
K
+ (7.44)
and is the free wave solution:

K
(

R
CM
) = e
i

R
CM
(7.45)
The non trivial part of the hydrogen atom system is the equation 7.43,
that is equivalent to the Schr odinger equation for a particle of mas moving
in a xed central eld of forces. This is the one-particle equivalent problem,
that already appeared in the classical mechanics study of motion of two par-
ticles interacting through a force dependent on their distance of separation..
7.5.2 Eigenfunctions and Eigenvalues of the Hydrogen
Atom
We will solve the wave equation for the one-particle equivalent problem:
(
h
2
2

2

r

Ze
2
r
)

r ) =

r )
where =
m
e
M
m
e
+M
is the reduced mass of the electron-nucleus system, with
m
e
the mass and -e the charge of the electron. The

r ) are simultaneous
eigenfunctions of

H,

L
2
and

L
z
. They are separable in radial and angular
coordinates:
(
h
2
2

2
r
2

h
2

1
r

r
+
h
2
2
1
r
2

L
2
(, )
Ze
2
r
)

r ) = 0 (7.46)
7.5. THE HYDROGEN ATOM 55

(

r ) = R
l
(r)Y
m
l
(, ) (7.47)
where:

L
2
(, )Y
m
l
(, ) = h
2
l(l + 1)Y
m
l
(, ) (7.48)
(
h
2
2

2
r
2

h
2

1
r

r
+
h
2
2
l(l + 1)
r
2

Ze
2
r
)R
l
(r) = 0 (7.49)
It can be found (see e.g. A.Galindo and P.Pascual, Mecanica Cuantica,Eudema
Universidad, Madrid,1989; L.Schi,Quantum Mechanics, 3rd.Edition,McGraw
Hill,1968, page 91) that in order to have solutions R
l
nite and normalizable,
the eigenvalues only admit the values

n
=
Z
2
e
4
2h
2
(n
r
+l + 1)
2
(7.50)
with n
r
0 and integer is called the radial quantum number.
The stationary states of the hydrogen atom are also eigenstates of the
square modulus of angular momentum and of one of its components. Since
the Hamiltonian commutes with the three components of the angular mo-
mentum, we are in the case of degenerate energy levels, as seen in the example
of 7.3. We in fact see that the energy does not depend on the quantum
number m, so each energy level of quantum numbers n
r
, l is (2l+1)-fold de-
generate. Furthermore, since the energy depends only on the sum n
r
+l+1=n
, all combinations of states with n
r
+l that produce the same value of n have
the same energy. Each energy level of value

n
=
Z
2
e
4
2h
2
n
2
has a degeneracy:
n1

l=0
(2l + 1) = n
2
(7.51)
In 8.3 we will introduce an internal degree of freedom to the electron, its
spin, that has two distinguishable states, both with the same energy in an
56CHAPTER 7. SCHRODINGEREQUATIONINMORE THANONE DIMENSION
electrostatic eld. The degeneracy 7.51 of each energy level has then to be
multiplied by a factor 2.
The l-degeneracy of the hydrogen atom should be tied to another invari-
ance of the Hamiltonian apart from

L
2
and

L
z
. Our example of 7.3 shows
that level degeneracy appears when the set of physical operators commuting
with

H has some operators that do not commute among themselves. For the
hydrogen atom this happens to be the Runge-Lenz vector, that in classical
mechanics denes the orientation of the major axis of the orbit of planetary
motion and we know is a constant of the motion:

A =

L
m
+
e
2
r
r
(See e.g.: Arno Bohm, Quantum Mechanics, Springer-Verlag,1993, page
207).The corresponding quantum operator commutes with

H but not with

L.
A scheme of the energies, angular momentum values and degeneracy of
the levels of the hydrogen atom is given in Figure 15. We introduce the
spectroscopic notation to name the levels, originated in the classication of
emission and absorption lines by atoms: s (sharp) for l=0 states, p (principal)
for l=1, d(diuse) for l=2, f for l=3, etc...
Chapter 8
Systems of More than One
Particle
8.1 Constants of the Motion
For a system of several particles we may dene collective observables such
as the total energy, linear and angular momentum, that are the sum of the
values corresponding to each particle. In fact, for identical particles collective
magnitudes are the only dynamical variables that can be dened: particles
are undistinguishable, and even a magnitude tied to an single particle, when
measured represents the average of that magnitude over all the individuals
of the system.
We consider the case of a multielectron atom , where the external eld
is central and the interation between particles depends only on their relative
distance:

H =
h
2
2m
N

i=1

2
r
i

i=1
Z
r
i
+

i>j
1
r
ij
(8.1)
The total orbital angular momentum, specied for a two-particle system,
is:

L =

L
1
+

L
2
(8.2)
Even though the eld acting on each particle is non central, we will show
that

L commutes with the Hamiltonian; the only non trivial term is the
57
58 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
commutator with the interparticle potential f(r
12
):

L
1
f(r
12
) = ih

r
1

r
1
f(r
12
)
= ihf

(r
12
)

r
1

r
12
/r
12
and

L
2
f(r
12
) = ih

r
2

r
2
f(r
12
)
= ihf

(r
12
)

r
2

r
21
/r
21
since:

r
21
=

r
12
we get:
(

L
1
+

L
2
)f(r
12
) = 0
Then
[

L, f(r
12
)] =

Lf(r
12
) f(r
12
)

L (8.3)
= (

Lf(r
12
)) = 0
This result can be generalized to any number of particles, so
[

L,

H] = [

L,

i>j
1
r
ij
] = 0 (8.4)
The operator

L has the same rules of commutation among components
as for the case of one particle, Eqs.7.20-7.22:
[

L
x
,

L
y
] = ih

L
z
, etc.
Furthermore:
[

L
2
,

L
2
i
] = [

L
2
i
+

i,j

L
i

L
j
,

L
2
i
]
and since

L
i
commutes with itself, with any of its components and with

L
j
for j,= i :
8.2. COUPLING OF ANGULAR MOMENTA 59
[

L
2
,

L
2
i
] = 0 (8.5)
but it does not commute with

L
iz
because

L
2
has the terms

L
i

L
j
=

L
ix

L
jx
+

L
iy

L
jy
+

L
iz

L
jz
, and [

L
iz
,

L
ix
],= 0. The set of commuting operators
is thus formed by

H,

L
2
,

L
2
i
,

L
z
.
8.2 Coupling of Angular Momenta
We will study the way to obtain the set of simultaneous eigenfunctions

{}
(
1
,
2
,
1
,
2
) of

L
2
,

L
2
1
,

L
2
2
,

L
z
and their eigenvalues.
We consider the subset of states with eigenvalues h
2
l
1
(l
1
+1) for

L
2
1
, and
h
2
l
2
(l
2
+ 1) for

L
2
2
. This subset is given by the (2l
1
+ 1)(2l
2
+ 1) products
Y
m
1
l
1
(
1
,
1
) Y
m
2
l
2
(
2
,
2
), l
1
m
1
l
1
, l
2
m
2
l
2

that we will abbreviate as


Y
m
1
l
1
(
1
,
1
) Y
m
2
l
2
(
2
,
2
) (
1
m
1
,
2
m
2
)
Therefore, the states that are also eigenfunctions of

L
2
,

L
z
should be linear
combinations of those product pairs.
The operator

L
z
= ih(

1
+

2
) has as eigenfunctions each of the
products:
[

L
z
h(m
1
+m
2
)](
1
m
1
,
2
m
2
) = 0 (8.6)
Since there is only one term with eigenvalue h(l
1
+ l
2
): (
1

1
,
2

2
)
this is also necessarily eigenfunction of

L
2
. The orbital quantum num-
ber of

L
2
h
2
( + 1)(
1
m
1
,
2
m
2
) = 0 can not be less than the magnetic
quantum number m = l
1
+l
2
of

L
z
hm(
1
m
1
,
2
m
2
), because it is a prop-
erty of the angular momentum components to be smaller that the modulus
of the vector. It can not be greater either, because if l > m = l
1
+ l
2
, then
eigenfunctions corresponding to eigenvalues of m = l
1
+l
2
should be present
in the subset, and they are not. We conclude that by coupling two angular
momentum operators

L
1
+

L
2
=

L, the largest value of the orbital quantum
number for

L
2
is l = l
1
+l
2
.
60 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
There are two products with eigenvalue h(l
1
+l
2
1) of

L
z
: Y
l
1
l
1
Y
l
2
1
l
2
and Y
l
1
1
l
1
Y
l
2
l
2
. A given linear combination of them should represent the
state with l = l
1
+ l
2
and m = l
1
+ l
2
1. There is another independent
linear combination also with the same m, whose only possibility is to be an
eigenfunction of

L
2
with l = l
1
+l
2
1.
Going on with this reasoning, we nd three products with m = l
1
+l
2
2,
and so on until we reach the number m =[ l
1
l
2
[= l
1
l
2
if we assume
l
1
> l
2
. Here we nd 2l
2
+ 1 products :
Y
l
1
l
1
Y
l
2
l
2
, Y
l
1
1
l
1
Y
l
2
+1
l
2
, ...Y
l
1
l
2
l
1
Y
l
2
l
2
which determine the eigenfunctions with l = l
1
+ l
2
, l
1
+ l
2
1, ...l
1
l
2
.
The next group of eigenfunctions of

L
z
has m = l
1
l
2
1, and there are
also 2l
2
+ 1 products:
Y
l
1
1
l
1
Y
l
2
l
2
, Y
l
1
2
l
1
Y
l
2
+1
l
2
, ...Y
l
1
l
2
1
l
1
Y
l
2
l
2
Therefore, they are only enough to produce the eigenfunctions
with l such that l
1
+ l
2
l l
1
l
2
. See Figure 16 for a graphical
explanation of the preceeding discussion.
We see that the number of eigenfunctions of

L
2
,

L
2
1
,

L
2
2
,

L
z
found is
l
1
+l
2

l
1
l
2
(2l + 1) = (2l
1
+ 1)(2l
2
+ 1)
which coincides with the number of original states, indicating that none
of these was left outside the build up of simultaneous eigenfunctions.
The conclusion is that for two particles in eigenstates of their in-
dividual angular momenta

L
2
1
,

L
2
2
with orbital quantum numbers
l
1
, l
2
, the possible eigenvalues for the total angular momentum

L
correspond to the range
[ l
1
l
2
[ l l
1
+l
2
(8.7)
of the orbital quantum number for

L
2
.
Finally, the stationary states for a system of particles moving in an ex-
ternal central eld and subject to mutual interactions that depend on their
separation are eigenstates of the square modulus of each individual angular
8.3. INTRINSIC ANGULAR MOMENTUM (SPIN) 61
momentum

L
2
i
, of the square and one projection of the total angular momen-
tum

L, and of the energy. For the case of two particles the eigenvalues are
h
2
l
1
(l
1
+1),h
2
l
2
(l
2
+1) for

L
2
1
,

L
2
2
where l
1
, l
2
can have any value 0,1,2....; the
eigenvalues of

L
2
are h
2
l(l + 1) with [ l
1
l
2
[ l l
1
+ l
2
, and those of

L
z
are hm with [ m [ l. The eigenvalues E of the energy depend on the form
of the external potential, and on l but not on m. These conclusions can be
generalized to systems of more than two particles, the coupling of angular
momenta proceeds by coupling rst two of them:

L
1
+

L
2
=

L(1, 2), then
a third particle with the momentum of the rst pair:

L
3
+

L(12) =

L(123)
and so on. It can be proved that the resulting set of eigenfunctions does not
depend on the order followed to perform the coupling.
8.3 Intrinsic Angular Momentum (Spin)
Up to now we assumed that particles have only external degrees of freedom

r associated with their position in space. It was found that to explain


atomic spectra it was necessary to assign an internal magnetic moment to
the electron. In fact, all particles may present this internal structure which
manifests in a magnetic moment

, so the energy levels of bound systems
of particles such as atoms and nuclei are split in the presence of an external
magnetic eld

H, due to the added interactions

H.

.
The classical description of a magnetic moment is that of a loop of electric
current; for material particles, the loop of moving charges is tied to a motion
of mass that produces an angular momentum. There is a linear connection
between

and

l through a fator g called gyromagnetic ratio. In quantum
atomic physics there will be magnetic momenta related to the orbital motion
of the electrons:

o
= g
o

l (8.8)
and the intrinsic magnetic moment will be related to an intrisic angular
momentum (spin)

s :

i
= g
i

s (8.9)
The gyromagnetic ratios are found from experiments or can be determined
from a relativistic treatment of quantum mechanics, where the spin appears
62 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
as a necessary property emerging from the covariance of the theory. Their
values are: g
o
= e/2mc, and g
i
= 2.00232 e/2mc, with c the speed of light.
The wave function of a particle with internal states will be an ordered set
of functions of position

r and time t:

1
(

r , t),
2
(

r , t), ....
n
(

r , t) (8.10)
whre [
i
(

r , t) [
2
gives the probability of nding the particle in

r at
time t and in the internal state i. This set can be compacted in one function
(

r , , t) with adopting the discrete values 1,2,...n.


Apart from the magnetic moment, there are no other manifestations of
the internal structure of the particles, so it is unjustied to try to build a
model of an extended distribution of charge and mass for the elementary
particles such as the electrons. The intrinsic angular momentum is the only
observable, through the magnetic moment, of the internal structure of the
particle.
Electrons, protons and neutrons have two internal states of spin, that
correspond to two possible projections of the intrinsic angular momentum;
since there are only two projections and for a quantum number l there are
2l+1, this indicates that l=
1
2
. These two projection values
h
2
are observed
when measuring the magnetic moment of these particles (Eq.8.9). The fact
that we have half-integer values of the magnetic quantum number m, already
indicates that this angular momentum is not related to a spatial rotation and
internal angular coordinates.
Photons and mesons have an odd number of internal states, and corre-
spond to projections hm of spin with integer values of its intrinsic m and
l quantum numbers.
The space and internal state of an electron is described by the two-term
wave function:

1
(

r , t),
2
(

r , t) (8.11)
It is convenient to dene a two dimensional vector space, where the two
dimensions describe the internal states, and the amplitude of each one gives
the space and time dependence of the state:
_

1
(

r , t)

2
(

r , t)
_
(8.12)
The eigenstates of projections
h
2
are:
8.3. INTRINSIC ANGULAR MOMENTUM (SPIN) 63

+
=
_
1
0
_
,

=
_
0
1
_
(8.13)
each one can be multiplied by a general function of

r , t that gives the
space part of the quantum state of the particle.
The operators in the internal space will be matrices of 22 dimensions;
the projection of spin in a direction z is expressed as
z
=
h
2

S
z
, so the

S
z
is
dened through:

S
z

+
=
+

S
z

(8.14)
that determines the matrix:

S
z
=
_
1 0
0 -1
_
(8.15)
The rest of the components of the operator

are determined consider-
ing that if this is an angular momentum operator, it should obey the same
commutation relations we have obtained for the orbital angular momentum.
From Eqs.7.20-7.23:
[

S
2
,

S
i
] = 0
[

S
x
,

S
y
] = 2

S
z
[

S
y
,

S
z
] = 2

S
x
[

S
z
,

S
x
] = 2

S
y
(8.16)
the forms of the rest of matrices is determined as:

S
2
=
_
1 0
0 1
_
64 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE

S
x
=
_
0 1
1 0
_

S
y
=
_
0 -i
i 0
_
(8.17)
A general state of the spin of an electron will be
_
a
b
_
where [ a [
2
+ [
b [
2
= 1, being [ a [
2
([ b [
2
) the probability of measuring spin up(down).
In the same way as we coupled two orbital angular momenta, we can
couple the spin to the orbital angular momentum of a particle, or to the
other particles to get the total angular momentum of the system. In the case
of an electron with orbital quantum number l, the total angular momentum
operator

J =

L +

S (8.18)
has two values of the quantum number j if l,= 0:
j
1
= l +
1
2
, j
2
= l
1
2
The 2(2l+1) products of the type Y
m
l
(, )
+
, Y
m
l
(, )

generate ac-
cording to the analysis of 8.2:
the 2l+2 states
m
1
j
1
with magnetic quantum numbers [m
1
[< j
1
and
the 2l states
m
2
j
2
with magnetic quantum numbers [m
2
[< j
2
.
8.4 Systems of Identical Particles
Schrodinger equation for a system of identical particles becomes:
[
h
2
2m
N

i=1

2
r
i
+V (

r
1
,

r
2
, ...

r
N
, t) ih

t
](

r
1
,

r
2
, ...

r
N
, t) = 0 (8.19)
We consider the simplest case of two identical particles, two electrons.
The wave function has as independent variables the degrees of freedom of
both particles, now the position and internal coordinates, the we call

r ,
8.4. SYSTEMS OF IDENTICAL PARTICLES 65
for an electron, and

s , for the other. The wave function assigns a com-
plex number = (

r , ;

s , ) to each set of values

r , ;

s , . Since the
particles are undistinguishable, we can not identify one of them with the
coordinates

r , and the other with

s , . For this reason, the function
(

r , ;

s , ) and the function (

s , ;

r , ) represent the same quantum
state; in particular, the probability to detect a particle in

r , and the
other in

s , must be the same independent of the function considered:
[ (

r , ;

s , ) [
2
=[ (

s , ;

r , ) [
2
. The relation between the functions
should then be:
(

r , ;

s , ) = e
i
(

s , ;

r , )
Repeating the exchange

r , =

s , on the right member:
(

r , ;

s , ) = e
i2
(

r , ;

s , )
that gives
e
i2
= 1
We have two possible solutions:
e
i
= 1
Observations on systems of identical particles show that some of them
are described by symmetric functions
(

r , ;

s , ) = (

s , ;

r , ) (8.20)
these are called bosons, and correspond to particles with integer or zero
values of spin. The particles that are described by an antisymmetric function
(

r , ;

s , ) = (

s , ;

r , ) (8.21)
are called fermions and correspond to particles with half-integer spin.
In the general case of n identical particles, the wave function has the
property of symmetry or antisymmetry under exchange of the space and
spin coordinates of any pair of particles.
66 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
8.5 Systems with two electrons.
8.5.1 Angular momentum coupling for two electrons
Consider two particles in eigenstates of the square modulus of their orbital
angular momenta
Y
m
1

1
(
1
,
1
)m
1
[
1
[ ,Y
m
2

2
(
2
,
2
)m
2
[
2
[
We have seen that the operators representing the square modulus of each
orbital angular momentum

L
2
1
,

L
2
2
, the total orbital angular momentum

L
2
and its projection

L
z
commute among themselves, therefore they have a
common set of eigenfunctions. The (2
1
+ 1) (2
2
+ 1) products
Y
m
1

1
(
1
,
1
) Y
m
2

2
(
2
,
2
)
are all the possible eigenfunctions of

L
2
1
and

L
2
2
with eigenvalues h
2

1
(
1
+
1) and h
2

2
(
2
+1) respectively. They are also eigenfunctions of

L
z
, each one
with eigenvalue h(m
1
+ m
2
). Specic linear combinations of these products
should then be the eigenfunctions of

L
2
, and then come to be the set of
common eigenfunctions of the four operators, all of them with the same
eigenvalues h
2

1
(
1
+ 1) , h
2

2
(
2
+ 1) for

L
2
1
and

L
2
2
.
We have seen that the eigenvalues for

L
2
are h
2
L(L+1) where [
1

2
[
L
1
+
2
. The eigenfunction corresponding to the orbital quantum numbers

1
,
2
, L =
1
+
2
, M =
1
+
2
is the product

L,M=
1
+
2
= Y

1
(
1
,
1
)Y

2
(
2
,
2
)
while the one correspondig to L =
1
+
2
, M =
1
+
2
is

L,M=
1
+
2
= aY

1
1

1
(
1
,
1
)Y

2
(
2
,
2
)
+bY

1
(
1
,
1
)Y

2
1

2
(
2
,
2
)
we are going to determine the coecients a, b that
8.5. SYSTEMS WITH TWO ELECTRONS. 67
8.5.2 Spin coupling for two electrons
Each electron has two possible spin projections with eigenstates identied as:

+
(),

()
Then a complete basis for the spin state of two electrons is:

+
(
1
)
+
(
2
)

+
(
1
)

(
2
)

(
1
)
+
(
2
)

(
1
)

(
2
)
To build in this space eigenfunctions of the total spin

S
2
and its projection

S
z
notice that each one of these products is already an eigenfucntion of

S
z
:
8.5.3 The two-electron atom
We consider two electrons moving in the Coulomb potential of a xed nucleus
and subject to their mutual electrostatic repulsion:
[
h
2
2m

2
r
1

h
2
2m

2
r
2

Ze
2
r
1

Ze
2
r
1
+
e
2
[

r
1

r
2
[
]
E
(

r
1

1
,

r
2

2
) = E
E
(

r
1

1
,

r
2

2
) (8.22)
where the solution should be antisymmetric under the exchange of
space and spin coordinates:

E
(

r
2

2
,

r
1

1
) =
E
(

r
1

1
,

r
2

2
)
Furthermore, since the Hamiltonian is independent of spin coordinates it
is possible to assume a separation of variables in

E
(

r
1

1
,

r
2

2
) =
E
(

r
1
,

r
2
)
E
(
1
,
2
)
which will be a valid description of the electronic state if it satises the
antisymmetry condition. This requires that either
68 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE

E
(

r
2
,

r
1
) =
E
(

r
1
,

r
2
),
E
(
2
,
1
) =
E
(
1
,
2
)
or

E
(

r
2
,

r
1
) =
E
(

r
1
,

r
2
),
E
(
2
,
1
) =
E
(
1
,
2
)
We already know the total spin space of two electron states. These are
four states that separate in a triplet and a singlet with denite symmetry
under
1


2
exchange.
We have already seen that due to the central character of the acting forces
the total angular momentum operator commutes with the Hamiltonian.
The total spin operator also commutes with the Hamiltonian, since this
does not depend on the spin coordinates
1
,
2
. Then
E
(

r
1

1
,

r
2

2
) can be
written as simultaneous eigenfunction of

H,

L
2
,

L
z
,

S
2
,

S
z
.
Furthermore, the independence on spin coordinates of

H allows for the
separation of variables

E
(

r
1

1
,

r
2

2
) =
E
(

r
1
,

r
2
)
E
(
1
,
2
)
with
[
h
2
2m

2
r
1

h
2
2m

2
r
2

Ze
2
r
1

Ze
2
r
1
+
e
2
[

r
1

r
2
[
]
E
(

r
1
,

r
2
) = E
E
(

r
1
,

r
2
) (8.23)
The three-body problem does not have a closed solution in the frame of
classical mechanics, this means that the time dependence of the six degrees
of freedom in the center of mass system can not be expressed in terms of
known functions (polynomials, trigonometric or special functions). Of course,
numerical solutions can be obtained with any degree of accuracy. The same
situation is found for the Schrodinger equation (8.22): it is not possible to
solve it and nd
E
(

r
1

1
,

r
2

2
) as a nite sum of known functions.
It is very convenient to have some kind of approximate solution to the
general problem of an N-electron atom, that could not only produce more
or less accurate values for the energy E and eigenfucntion
E
, but also that
8.6. INDEPENDENT PARTICLE APPROXIMATION FOR ATOMS 69
allows a physical interpretation of the physics ruling the atomic structure and
dynamics. We will consider in detail in the next section this approximation,
that replaces the eld perceived by each electron by an eective potential
that depends only on the electron distance from the nucleus. The equation
(8.22) simplies to
[
h
2
2m

2
r
1

h
2
2m

2
r
2
+v
ef
(r
1
) +v
ef
(r
2
)]
E
(

r
1
,

r
2
) (8.24)
= E
E
(

r
1
,

r
2
) (8.25)
This is the independent particle approximation, since the spatial wave
function can be separated in independent orbitals

E
(

r
1
,

r
2
) =
p
1
(

r
1
)
p
2
(

r
2
)
that satisfy:
[
h
2
2m

2
r
i
+v
ef
(r
i
)]
p
(

r
i
) =
p

p
(

r
i
) (8.26)
But the space-spin wave function should be antisymmetric under the
exchange of coordinates

r
1
,
1

r
2
,
2
, so (

r
1
,

r
2
) should be symmetric
under the exchange

r
1

r
2
when (
1
,
2
) is antisymmetric under
1

2
and viceversa.
8.6 Independent Particle Approximation for
Atoms
We consider the stationary Schr odinger equation for the case of an atom of
Z electrons:
[
Z

i=1
(
h
2
2m

2
r
i
+
Ze
2
r
i
) +

i>j
1
[

r
i


r
j
[
](

r
1

1
,

r
2

2
, ...

r
Z

Z
)
= E(

r
1

1
,

r
2

2
, ...

r
Z

Z
) (8.27)
The term of mutual repulsion between the electrons makes very dicult
to nd a solution of this equation, even an approximate one. But specially
70 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
if Z is large, we may approximate the potential felt by each electron by an
average of the interelectronic potential over the positions of the electrons:
_

i>j
1
[

r
i

r
j
[
_
r
j
= v(r
i
) (8.28)
This average over the positions of the Z-1 electrons is weakly dependent
of the electron i selected, so we may assume v(r) to be the same potential
for all electrons.
The resulting Schr odinger equation
[
Z

i=1
(
h
2
2m

2
r
i

Ze
2
r
i
+v(r
i
)) E](

r
1

1
, ...

r
Z

Z
) = 0 (8.29)
can be solved by the method of separation of variables. The solution
can be proposed as a product of one-particle functions
(

r
1

1
, ...

r
Z

Z
) =
1
(

r
1

1
)
2
(

r
2

2
)...
Z
(

r
Z

Z
) (8.30)
But physical solutions to a system of Z electrons should be antisymmetric
under exchange of any pair of space and spin coordinates. This can be
accomplished by a combination of antisymmetrized products of one-particle
states like that of Eq.8.30, which can be symbolized by a determinant of the
form:

1

2
...
Z
=
_

1
(

r
1

1
)
1
(

r
2

2
) ...
1
(

r
Z

Z
)

2
(

r
1

1
)
2
(

r
2

2
) ...
2
(

r
Z

Z
)
... ... ... ...

Z
(

r
1

1
)
Z
(

r
2

2
) ...
Z
(

r
Z

Z
)
_

_
(8.31)
that is called a Slater determinant. In fact, the average in Eq.8.28 can
be performed using the function 8.31, so the resulting one-particle equations
8.33 become coupled, and are named Hartree-Fock equations.
An interesting property of the Slater determinant is that it cancels when
two one-particle states are equal (a determinant is zero if two rows, or two
columns, are identical): this is the so called Pauli exclusion principle,
that says that electrons should occupy dierent one-particle states.
The determinant has two identical columns when two space-spin coordinates
are identical; this says that two electrons can not be found at the
8.6. INDEPENDENT PARTICLE APPROXIMATION FOR ATOMS 71
same point of space with the same spin. Finally, the states in the
Slater determinant have meaning if they are mutually orthogonal, since any
component c
j
on the i-th row:
i
+ c
j
gives a zero contribution to the
determinant

2
...,
i
+c
j,
...
j
...
Z

=
1

2
...
i
...
j
...
Z
+
1

2
...c
j
...
j
...
Z

=
1

2
...
i
...
j
...
Z

because:

2
...c
j
...
j
...
Z
= 0
A halfway model between ignoring antisymmetry (product of one-particle
functions of Eq. 8.30 to ) and full account of it (the Hartree-Fock method
with given by Eq.8.31), is to impose on the product of one-particle states
that these should be orthogonal, so only one electron is in any given state.
This is the Hartree model that obeys only the Pauli exclusion principle. The
equations satised by the Hartree proposal
(

r
1

1
, ...

r
Z

Z
) =
1
(

r
1

1
)
2
(

r
2

2
)...
Z
(

r
Z

Z
) (8.32)
with

i
[
j
) =
ij
(8.33)
are:

h
2
2m

2
r

Ze
2
r
+v(r))
p
]
p
(

r ) = 0 (8.34)
where:
v(r) =
Z 1
Z
Z

i=1
2

=1
_
d
3
r

[
i
(

r ) [
2
[

[
(8.35)
This is the average over the probability density of all the electrons of the
electrostatic potential produced by them on a point

r . We assume this is the


potential felt by an electron at

r , corrected by the factor


Z1
Z
because in the
average we included the interaction of the electron with its own probability
72 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
distribution. (This factor reduces the integrated value Z of the electronic
charge to the correct value Z-1).
The potential in Eq.8.34 is central, so the solutions are of the form:

p
(

r ) = R
nl
(r)Y
m
l
(, )

() (8.36)
When solving the Schr odinger equation we determine the radial eigen-
functions R
nl
and their eigenvalues E
nl
:
(
h
2
2

2
r
2

h
2

1
r

r
+
h
2
2
l(l + 1)
r
2

Ze
2
r
+v(r)
n
)R
nl
(r) = 0 (8.37)
The problem is similar to that of the hydrogen atom, but now the central
potential is not pure Coulombian. The potential is of the form -
1
r
on the
outskirts of the atom, coming close to the nucleus it gets stronger. Therefore,
states with small values of angular momentum (meaning small values of the
orbital quantum number l) that are able to reach the inner regions of the
atoms will feel a stronger eld of force than states with large values of l, that
are kept away by the centrifugal potential
h
2
2
l(l + 1)
r
2
. The correction to the degenerate energies of hydrogen for dierent l is
that for a given radial function R
nl
, the largest binding is for l=0 states, the
following energy level will be for l=1 states, and so on. The scheme of energy
levels and their degeneracy is presented in Figure 17.
8.7 Chemical properties of Atoms. Mendeleevs
Table
The structure of levels and their degeneracy in the Hartree model allows us
to build the electronic structure of all the elements, by successive additions
of electrons, each in one dierent state

p
(

r ) = R
nl
(r)Y
m
l
(, )

() (8.38)
We start the walk through the Atomic Table of the Elements by adding
one electron at a time, (and increasing each time by one the nuclear charge).
8.7. CHEMICAL PROPERTIES OF ATOMS. MENDELEEVS TABLE73
First comes hydrogen with one electron in the 1s orbital (orbital means the
space part of a one-particle state); the following electron (forming helium)
can be placed in the same spatial orbital and with opposite spin, the state is
symbolized by (1s)
2
. the third electron can not be placed in the 1s orbital,
then we go to the following solution of the radial equation 8.37; the cong-
uration of lithium is the (1s)
2
(2s). We continue in this way generating the
following scheme of lling of one-particle states :
Atom
H 1s
He 1s
2
Li 1s
2
2s
Be 1s
2
2s
2
B 1s
2
2s
2
2p
C 1s
2
2s
2
2p
2
N 1s
2
2s
2
2p
3
O 1s
2
2s
2
2p
4
F 1s
2
2s
2
2p
5
Ne 1s
2
2s
2
2p
6
Na 1s
2
2s
2
2p
6
3s
... ... ... ... ...
The resulting atomic structure with the systematic lling of orbitals de-
nes with total delity the chemical properties of the elements:
Noble gases He, Ne, Ar, Kr, Xe, Rn: their electrons have completed the
states of an energy level (shell) well separated from the next, as 1s, 2p, 3d,
...They have almost no chemical activity, since their electrons are strongly
bound and closely packed.
Alkalis Li, Na, K, Rb, Cs, Fr: They have a lone electron outside a closed
shell, it is weakly bound and can be easily taken away by neighbor atoms.
This produces the strong binding of ionic salts like NaCl.
Halogens F, Cl, Br, I, At: Their electrons ll a shell leaving only one
place empty. They are eager to accept electrons, specially from atoms with
weakly bound electrons like alkalis.
Transition Metals Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn: there are
some anomalies in the lling of the shells. For example the binding of 3d
electrons is higher but close to that of the 4s orbital. Then, after Ca that lls
the 4s
2
there appear the transition metals that may have one or two holes in
the 4s because the electrons prefer to go to the 3d orbitals. This mixing of
74 CHAPTER 8. SYSTEMS OF MORE THAN ONE PARTICLE
3d and 4s states with similar binding energies is the generator of the special
characteristics of the transition metals.

You might also like