QM LecNotes 01

LECTURES ON QUANTUM MECHANICS-I
By
A. K. Kapoor
School of Physics
University of Hyderabad
Hyderabad 500046, INDIA
Version 1.1
November 4, 2010
DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Contents
1 Waves vs Particles 5
§ 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
§ 2 Gedanken Experiments With Waves . . . . . . . . . . . . . . . . . . . 6
§ 3 Gedanken Experiments With Bullets . . . . . . . . . . . . . . . . . . 7
§ 4 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 9
§ 5 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Inadequacy of Classical Theories and Old Quantum Theory 11

§ 1 Inadequacy of Classical Theory . . . . . . . . . . . . . . . . . . . . . 11
§ 2 Emergence of New Concepts . . . . . . . . . . . . . . . . . . . . . . . 12
§ 3 Superposition, Indeterminacy and Appearance of Probability . . . . . 12
§ 4 Measurement of Position . . . . . . . . . . . . . . . . . . . . . . . . . 13
§ 5 Measurement of momentum . . . . . . . . . . . . . . . . . . . . . . . 15
3 States and Dynamical Variables 16

§ 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
§ 2 Structure of Classical theory . . . . . . . . . . . . . . . . . . . . . . . 16
§ 2.1 Newtonian formulation . . . . . . . . . . . . . . . . . . . . . . 18
§ 2.2 Lagrangian form of classical dynamics . . . . . . . . . . . . . 19
§ 2.3 Hamiltonian form of classical dynamics . . . . . . . . . . . . . 19
§ 3 Emergence of New Concepts . . . . . . . . . . . . . . . . . . . . . . . 20
§ 4 Postulates of Quantum Mechanics . . . . . . . . . . . . . . . . . . . . 21
4 Probability and Average Values 23

§ 1 The Third Postulate . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
§ 2 Computation of Probabilty and Average Value . . . . . . . . . . . . . 25
5 Canonical Quantisation 27
§ 1 Canonical Commutation Rules . . . . . . . . . . . . . . . . . . . . . . 27
§ 2 Canonical commutation relations . . . . . . . . . . . . . . . . . . . . 27
§ 3 Generalized Uncertainty Relation . . . . . . . . . . . . . . . . . . . . 28
§ 4 Hamonic Oscillator Energy Levels Using Commutators . . . . . . . . 30
§ 5 Angular Momentum Eigenvalues Using Commutators . . . . . . . . . 32
§ 6 Summary of important results . . . . . . . . . . . . . . . . . . . . . . 36

6 Time Evolution 38
§ 1 Superposition principle and time development . . . . . . . . . . . . . 38
§ 2 Time dependent Schrodinger equation . . . . . . . . . . . . . . . . . . 40
§ 3 Solution of the time dependent equation . . . . . . . . . . . . . . . . 42
7 Working with Representations 45

§ 1 Introduction to representations . . . . . . . . . . . . . . . . . . . . . 45
§ 2 The coordinate representation . . . . . . . . . . . . . . . . . . . . . . 46
§ 3 Momentum representation . . . . . . . . . . . . . . . . . . . . . . . . 49
§ 4 Change of representation . . . . . . . . . . . . . . . . . . . . . . . . . 49
§ 5 Wave function as probability density . . . . . . . . . . . . . . . . . . 53
8 Time Dependent Schrodinger Equation 54

§ 1 Conservation of probability . . . . . . . . . . . . . . . . . . . . . . . . 54
§ 2 Schrodinger equation for a charged particle . . . . . . . . . . . . . . . 56
§ 3 Time Reversal Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . 57
§ 4 Solution of Time Dependent Schrödnger Equation . . . . . . . . . . . 58
§ 5 Free particle solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
§ 6 Wave Packets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
§ 7 Propagator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
9 Energy Eigenvalues and Eigenfunctions in One Dimension 66

§ 1 Energy Eigenvalues and Eigenfunctions . . . . . . . . . . . . . . . . . 66
§ 2 Particle in a Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
§ 2.1 Rigid wall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
§ 2.2 Infinite Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
§ 3 Square Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
§ 3.1 Bound states solutions . . . . . . . . . . . . . . . . . . . . . . 70
§ 4 Delta Function Potential . . . . . . . . . . . . . . . . . . . . . . . . . 72
§ 4.1 δ- function as a limit of a square well . . . . . . . . . . . . . . 72
§ 4.2 Direct Integration of the Schrödinger Equation . . . . . . . . . 73
§ 4.3 Solution in Momentum Space . . . . . . . . . . . . . . . . . . 75
§ 5 Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
§ 6 Periodic Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
§ 7 General Properties of Motion in One Dimension . . . . . . . . . . . . 81
10 Reflection and Transmission in One Dimension 88

§ 1 Reflection and Transmission Through a Potential Well . . . . . . . . 88
§ 2 Square Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
§ 3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
11 Potential Problems in Two and Three Dimensions 93

§ 1 A Rigid Box in Three Dimensions . . . . . . . . . . . . . . . . . . . . 93
§ 2 Harmonic Oscillator in Three Dimensions . . . . . . . . . . . . . . . . 94

12 Spherically Symmetric Problems in Three Dimensions 96
§ 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
§ 2 Coordinate Representation and Angular Momentum . . . . . . . . . . 97
§ 2.1 Separation of Variables . . . . . . . . . . . . . . . . . . . . . . 98
§ 2.2 Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
§ 3 Spherically Symmetric Potentials . . . . . . . . . . . . . . . . . . . . 101
§ 3.1 Summary of Results on Spherically Symmetric Potentials . . . 102
§ 4 Solution of Radial Equation for a Constant Potential . . . . . . . . . 104
§ 4.1 CASE II : E − V0 < 0 . . . . . . . . . . . . . . . . . . . . . . 105
§ 4.2 Some Spherical Bessel functions . . . . . . . . . . . . . . . . . 106
§ 5 Radial Equation for a Free Particle . . . . . . . . . . . . . . . . . . . 106
§ 6 Particle in a Rigid Spherical Box . . . . . . . . . . . . . . . . . . . . 107
§ 7 Square Well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
§ 8 Hydrogen Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Version 1.1

Unit 1
Waves vs Particles
§1 Introduction
From the time of Newton, it has been debated whether light is like a beam of cor-
puscules (particles) or is it a wave? It was believed that only one of these properties
could be associated with a beam of light as the characteristic properties of classical
particles and waves appeared to be incompatible. We briefly recall the differences
between waves and particles.
A particle cannot be subdivided, whereas one can subdivide the waves indefinitely.
For a wave, the intensity can be reduced continuously without any limit. The waves
transfer energy, momentum in a continuous fashion. For a beam of particles, the
intensity cannot be reduced indefinitely, beyond having one particle at a time. An
exchange of energy and momentum from particles takes place in a discrete fashion.
A particle is localised in space whereas the waves are, generally, not localised. A
particle follows a well defined trajectory, no trajectory can be associated with waves.
A particle cannot bend round a corner, whereas waves exhibit the phenomenon of
diffraction.
The waves show phenomenon of interference. Thus, two waves of intensity I,
meeting at a point, can give rise to a resultant wave of zero intensity, or of intensity
4I. In fact, whereas the resultant wave can have intensity between 0 and 4I, clearly
a similar situation for particles is not possible; two particles coming from different
paths, and meeting at a point, cannot destroy each other to give no particle, nor
do they create particles to give four particles. A beam of particles does not show
phenomena of polarisation exhibited by (transverse) waves.
For a complete specification of the state of a system of particles we require a
finite number of coordinates and momenta. The waves are specified by infinite
number of “coordinates“, an amplitude at each space point. All waves, for example
sound waves, can be thought of as a collection of infinite number of oscillators. For
sound waves these oscillators are just the particles of the medium. Light can also be
thought of as a collection of infinite number of oscillators, but without any medium
being associated with it.
Some of the differences between the wave and particle nature, outlined above, will
be brought out with the help of some Gedanken experiments ( thought experiments)
in next few sections.

§2 Gedanken Experiments With Waves
Single Slit Experiment with Waves
The interference and diffraction are characteristic properties of waves. When a
wave falls on a slit, a diffraction pattern is observed as the waves can bend round
the corners of the slit. Consider the experimental set up of Fig.1.1 where a wave
travelling along the z- axis passes through a slit and hits a screen where the intensity
of waves is measured as a function of x. If a is the slit width and λ is the wave
length, the intensity as a function of x is given by
2
sin α
I = I0 , (1.1)
α
where
πa x
α= sin θ, sin θ ≈ tan θ = , (1.2)
λ L
and L is the distance of the screen from the slit. The waves not only reach the point
x = 0, but a significantly large part of the wave is received for large values of x.
If the intensity of the incident wave is reduced, one would continue to get the
same pattern, even for arbitrarily low values of the intensity of the incident waves.
Fig. 1.1.
Double Slit Experiment With Waves

In a double slit arrangement shown in figure above, a part of the incident wave travels
through each slit and on reaching a point P on the screen, the two parts interfere and
produce an interference pattern. This is because waves travelling through different
paths have a phase difference and the resultant amplitude φ at any point is given
by the sum of the amplitudes, φ1 , φ2 of the waves coming through the two slits:
φ1 (~r, t) = A1 exp(i~k · ~r), (1.3)

φ2 (~r, t) = A2 exp(i~k · ~r + iδ), (1.4)
φ = φ1 + φ2 . (1.5)
The resultant intensity is then given by
I12 = |φ1 + φ2 |2 = |φ1|2 + |φ2|2 + (φ∗1 φ2 + φ1 φ∗2 ). (1.6)
Here the first, the second and the third terms give the intensity due to the first
slit, the second slit and the interference term, respectively. When the intensity

of the incident waves is reduced, we continue to get the same intensity distribution
pattern on the screen. Assuming that the detectors placed on the screen can measure
arbitrarily low intensity by collecting waves received over a large period of time, the
intensity pattern would look the same irrespective of the intensity of the incident
beam.
Double slit Experiment With Only One Slit Open at A Time

Let us now consider a variation of the double slit interference experiment, keeping
only one slit open at a time. Let us assume that only one slit is kept open for time
0 < t < T and the second slit is closed. During the time T < t < 2T , the second
slit is kept open and the first slit is kept closed. In this case what intensity pattern
do you expect to see on the screen? Would one get the interference pattern
I12 = |φ1 + φ2 |2 ? (1.7)
or simply the sum of intensities due to the two slits:
I = I1 + I2 = |φ1 |2 + |φ2 |2 ? (1.8)
Interference can take place only between parts of a wave arriving simultaneously at
a point with a phase difference. This happens when different parts of a wave arrive
taking different paths. When only one slit is kept open at a time, we expect to
observe the intensity pattern I1 + I2 which is just the sum of intensities due to the
two individual slits. While there is no doubt about the conclusion, the issue has to
be settled only by doing an experiment.
§3 Gedanken Experiments With Bullets

Single Slit Experiment With Bullets
Waves can turn around the corners but not particles, certainly not the bullets.
Consider an experiment with bullets fired along the z- axis from a gun and hitting
a wall with a slit. Next to the wall there is a screen where all bullets are stopped
and are collected in containers placed at different positions. The number of bullets,
N(x), received in a given time t in a container placed at x is counted. We plot the
number of bullets as function of x.
Assuming an ideal parallel beam of bullets, travelling in the z- direction, only
the containers with positions in the range −d ≤ x ≤ d will receive the bullets, where
2d is the slit width. The container placed at far away positions will not receive any
bullet. We should expect to see a distribution as in Fig.1.2. The bullets cannot
bend round the corners of the slit and cannot reach containers at positions |x| > d.

Fig. 1.2.
Double Slit Experiment

If a double slit experiment is performed with two slits, one would get a count dis-
tribution pattern, N(x), which is just the sum of the patterns corresponding to the
two individual slits.
N(x) = N1 (x) + N2 (x). (1.9)

Note that each bullet must pass through one of the slits, unlike waves where parts of
the incident wave could pass through each slit and produce an interference pattern.
The result will remain unchanged if the experiment with the bullets is repeated
with each slit open only half the time.
A Probabilistic Description
The classical theory is deterministic in the sense that, knowing the initial position
and momentum of a bullet, one can predict where it will hit the screen. Now imagine
a gun firing the bullets randomly at different angles, with a maximum number of
bullets along the z- axis. In this case, the bullets would reach positions outside the
range (−d, d) and the number of bullets reaching a point x will be as shown above.
Assuming that the angle between the z- axis and the initial direction of a bullet
is random, and not known a priori, it will not be possible to predict where the
bullet would hit the screen. For a small number of bullets the distribution N(x) of
the bullets on the screen cannot be predicted. But, if the experiment is repeated
sufficiently many times we expect to see the same distribution N(x), irrespective of
the rate of firing the bullets.
How do we describe the outcome of an experiment with a single bullet? We
associate a probability distribution P (x) that a bullet will hit the screen at a position
x. If N0 is the total number of bullets in the experiment, then
N(x)
P (x) = . (1.10)
N0
In a double slit experiment we would expect to see a distribution P12 (x):
P12 (x) = P1 (x) + P2 (x), (1.11)
where,

P12 (x) = Probability that a bullet reaches x when both the slits are open,
P1 (x) = Probability that a bullet reaches x when only the first slit is open,
P2 (x) = Probability that a bullet reaches x when only the second slit is open.
The outcome of an experiment with waves can also be described in probabilistic
terms. We normalise the intensity of waves at a position x by the intensity of the
incident waves, I0 , and associate a probability P (x) = I(x)/I0 , which would give
the intensity pattern as a function of x.
Here it must be emphasised that a probabilistic description for waves is purely a
matter of taste or convenience, but for bullets we are forced to introduce the prob-
abilistic description because of incomplete information about the initial momentum
of the bullets. Even in this case, the classical theory is deterministic in the sense
that each bullet takes a well defined trajectory which can be predicted if the initial
momentum is known accurately. A knowledge of the probability distribution, of the
angle at which bullet is fired, will in principle permit us to compute P (x) and the
number of bullets reaching a container at position x.
§4 Summary and Discussion

In previous sections single and double slit experiments with waves and beams of
particles have been analysed. These experiments clearly bring out the fact that a
wave bend round the corners, whereas particles don’t. Interference phenomenon
seen in the double slit experiment emphasises that a part of a wave passes thorough
one slit and another part passes through another slit and an interference pattern is
produced when parts of waves from different paths meet on the screen. The particles
have a well defined trajectory and for a beam of particles, each particle must pass
through one of the slits and no interference pattern is produced.
While in interaction with other physical systems, the waves transfer energy and
momentum in a continuous fashion, in contrast, the particles transfer energy and
momentum in a discrete manner.
Another important difference between the waves and particles is that a wave can
be subdivided indefinitely by cutting down the intensity. For a beam of particles,
the intensity cannot be reduced beyond having a single particle at a time.
The particle and the wave aspects appear to be incompatible and it would seem
impossible that they could coexist for a physical system. However, the nature has
its own way and it is amazing that all particles show a wave nature and all waves
have a particle nature. First indications came from the photoelectric effect from
its explanation in 1905 by Einstein. Next came the matter wave hypothesis by de
Broglie and its confirmation by Davisson Germer experiments. It was established
that the electrons exhibit wave nature. It is now accepted that all matter has a
wave nature associated with it. and
§5 Questions
• What pattern you expect to see for a wave hitting a single slit in the limit
(a) Very wide slit (b) Very narrow slit ? Does your answer coincides with the

answer in (a) small wavelength, and (b)large wave length limit?
• Plot the intensity in the above limits and compare with the answer you guessed.
10

Unit 2
Inadequacy of Classical Theories

and Old Quantum Theory
§1 Inadequacy of Classical Theory

There were several experiments which pointed to the inadequacy of the classical the-
ories such as mechanics, electromagnetic theory of light, and statistical mechanics.
Here we list some of the important experimental facts which had no explanation
within the classical theories.
1. Black Body radiation
2. Photoelectric effect
3. Atomic spectra
4. Frank Hertz experiment
5. Compton effect
6. Stern Gerlach Experiment
7. Zeeman splitting of spectral lines in magnetic field
8. Stark effect
9. Wave nature of electrons
10. Alpha decay
11. Specific heat of solids and gases
While efforts to explain some of the above experiments played a crucial role in the
development of quantum theory, an explanation of some others had to wait for a
fully developed quantum mechanics.
11

§2 Emergence of New Concepts
Many of the classical concepts have to be reviewed in light of developments that
shook Physics in the early twentieth century. Several new concepts were brought in
by the revolution that took place during these years. A historical account of all these
developments is very fascinating and is well documented but the historical path is
not always an easy path for a beginner attempting to learn the subject. Therefore,
we will not present a historical development of the subject. Instead, we follow will
an approach which, in our opinion, serves the most important purpose of presenting
the subject in a logically coherent fashion.
We make an attempt to convince the reader about most important points of
departure from the classical theory. We shall focus on the new concepts and the
classical concepts that had to be revised, and this will be done by an analysis of
thought experiments in light of experimental and theoretical facts available. Very
briefly, our focus will be on superposition principle, unpredictability, probabilistic
nature, uncertainty in simultaneous measurement of dynamical variables. With our
present knowledge, and within the framework of commonly accepted interpretation,
these new features appear to be inherent in the nature of the physical systems and
can not be bypassed by construction of a better, more and more refined, equipment
or by modifications in the quantum theory.
We avoid making commitment to a particular interpretation of quantum me-
chanics. The details of a particular interpretation are best learnt by a beginner only
after he feels comfortable with the new concepts, mathematics and structure that
come along with quantum mechanics. Hence, the issues related to interpretations
of quantum mechanics will not be addressed here, except for occasional comments
in a few places.
§3 Superposition, Indeterminacy and Appear-

ance of Probability
Photon polarisation experiment
Consider a light beam incident on a polariser crystal which allows light vibrations
perpendicular to its optic axis to pass and absorbs all the vibrations which are
parallel to the optic axis. By rotating the crystal, we can get a beam which is
polarised along different directions. The light beam is now allowed to pass through a
similar second crystal. Let the second crystal be placed with its optic axis parallel to
the first, all the light coming out of the first one will have vibrations perpendicular
to the optic axis of the second crystal and will be fully transmitted through the
second crystal too. Next, if the first crystal is rotated by π/2, no light will pass
through the second crystal. In case the angle of rotation of the first crystal is α, the
light from the first crystal will be obliquely polarised and a fraction cos2 α of the
incident intensity will be transmitted.
Next, we consider the same arrangement with intensity of the incident beam so
low that at any time only one photon passes through the apparatus. It is obvious
that when the two optic axes are parallel, the photon will be transmitted through the
12

apparatus and that it will be absorbed, when the two optic axes are perpendicular.
What do we expect when the light incident on the second crystal has an oblique
polarisation? From what is known in wave theory, one cannot conclude anything
about the outcome of this experiment. Certainly, we cannot say that a part of photon
is transmitted and a part is absorbed; because the photon is indivisible! However,
when we wait for a long time, we will find that a fraction cos2 α, of the total number
photons incident on the second crystal, will pass through. Therefore, all that can be
said about the single photon is that sometimes it will pass through the second crystal
and sometimes it will get absorbed. What can we say about the states of photon in
the three cases of parallel, perpendicular and oblique polarisation? Thus in order
to explain the outcome of the experiment for a large number of photons, it must be
admitted that the photon in oblique polarisation is partly in the state with parallel
polarisation and partly in the state with perpendicular polarisation. Obviously,
every state of polarisation of photon must then be regarded as a superposition
of states of polarisation along two perpendicular directions. This superposition of
states comes with indeterminacy that no prediction is possible about the outcome of
an experiment involving a single photon, only the outcome of experiment repeated
several times can be predicted. For a single photon with oblique polarisation, one
is forced to assign a probability of getting transmitted or absorbed. One arrives at
similar conclusions by analysis of other thought experiments involving electrons.
§4 Measurement of Position
Measurement using a slit
Let us consider a beam of electrons incident on a thin slit as shown in the Fig.2.1.
The slit allows only a small width of the beam along the z- axis to pass through,
the rest of the beam is blocked. This could be regarded as an arrangement for
measurement of the z- component of the position of an electron in the beam. For an
electron passing through the slit, the z- component of the position is known upto an
accuracy equal to the width of the slit d. Let p be the momentum of the electrons
in the beam. Then the wave length associated with the electrons is λ = h/p. Due
to diffraction of the electron waves the electrons, after passing through the slit, will
come out diverging at an angle α and z- component of momentum will be uncertain
by an amount
∆pz ≈ p sin α (2.1)
where sin α ≈ λ/d. Thus
λ
∆z ∆pz ≈ (p sin α) × d = (p sin α) (2.2)
sin α
∴ ∆z ∆pz ≈ ~ (2.3)
13

z
p
α
Fig. 2.1.
Heisenberg microscope
Let us consider measurement of position of a point particle by means of a microscope.
Let the particle be at a point P and its position be measured by allowing light to
fall on the particle and collecting the light scattered from the particle.
The resolving power of the lens, λ/ sin α, sets the limit on the accuracy upto
which the x- component of position, (see Fig.2.2 ) can be measured. In the process
of measurement the photon gets scattered from the particle and, after the scattering,
the photon will enter the microscope if it is travelling in a cone of angle ǫ. Thus
the x- component of the photon momentum will be uncertain by an amount p sin α,
where p is the initial momentum of the photon. This introduces an uncertainty
h
∆px ≈ p sin α = sin α (2.4)
λ
in the momentum of the particle after the measurement has been made. Thus, we
have
λ h
∆x ≈ , ∆px ≈ sin α, (2.5)
sin α λ
∆x ∆px ≈ h. (2.6)
Fig. 2.2.
14

§5 Measurement of momentum
Measurement using Doppler shift
Let us consider an experiment involving measurement of momentum of an atom by
measuring the Doppler shift of the radiation emitted by the atom. Let ν be the
frequency of the radiation emitted by the atom at rest and ν ′ be the frequency when
it is moving with velocity v. Then the nonrelativistic formula for the Doppler shift
gives
v ν − ν′
≈ . (2.7)
c ν
The uncertainty in measurement of velocity depends on the accuracy of measurement
of ν ′ , which in turn depends on the time τ which is the duration of measurement of
the frequency. The error in the frequency measurement is given by ∆ν = 1/τ . The
time T , when the photon is emitted, can have any value between 0 and τ . When
the photon is emitted, the velocity changes abruptly by an amount δv = hν ′ /mc
and the position becomes uncertain by an amount δvT . Since this time T can have
any value between 0 and τ and it cannot be determined, the uncertainty in position
will be
hν ′ τ
∆x = δvT ≈ . (2.8)
mc
The error in measurement of the momentum, using Eq.(2.7), is given by
m∆ν ′ mc
∆p = m∆v = = . (2.9)
cν τν
Hence, taking ν ≈ ν ′ , we get
∆x∆p ≈ h. (2.10)
Some other thought experiments, that can be analysed for momentum mea-
surement of a charged particle, consist in using of motion in a magnetic field, and
measuring the change in wavelength in scattering of photon, Compton effect. In
all such experiments one is led to the Heisenberg uncertainty relation as a conse-
quence of wave particle duality. It must be noted that a measurement of position
(momentum) gives disturbs momentum (position) of the particle by an amount that
is uncontrollable fashion and can not be measured and thereby leading to the un-
certainty relation.
15

Unit 3
States and Dynamical Variables
§1 Introduction
The transition from classical mechanics to quantum mechanics not only meant that
a new theory of the physical processes was necessary to describe the physical phe-
nomena at the microscopic level, but many of the old concepts had to be revised
and were replaced by radically new concepts. The quantum theory required intro-
duction of radically new hypotheses and brought its own mathematical structure.
When predictions of the quantum theory were worked out, many surprising results
were encountered.
In the next few lectures, we are concerned with the structure of quantum theory
and showing how different it is from that of the classical theory. In particular, we
will discuss specification of states of a physical system in classical and quantum
mechanics.
§2 Structure of Classical theory

Some Physical Systems of Interest
We begin with examples of some physical systems of interest.
• System of point particles moving in a force field.
• A point particle constrained to move on a curve or a surface.

• One or more point particles. moving on a surface of a sphere.
• Rigid body.
• Vibrating string.
• Water waves and waves in other media.
• Electromagnetic waves.
• Interaction of charged particles and electromagentic fields.

• Electromagnetic fields in presence of a dielectric medium.
16

• Gas molecules enclosed in a volume.
It is seen that the classical systems of interest can be broadly classified as consisting
of (a) point particles, (b) waves, and (c) particles and waves.
Classical theory of a physical system consists of components listed in table below.
SN Components
1 Description of physical system of interest

2 States of the physical system; ‘Co-ordinates’
3 Dynamical Variables
4 Laws of Motion
5 Forces, Interactions
Physical states: By state of a physical system we mean method of specifying

complete information about the system.
Dynamical variables: The dynamical variables of a classical system are functions

of the state of the system and can be computed when the state has been specified.
Laws of motion: Not only are we interested in knowing about a system at a given
time, we also want to know how the system changes with time. In order to describe
behaviour of a system under time evolution, one needs to know the laws of motion.
Classical systems are governed by the Newton’s laws.
Interactions: A classical description is completed by specifying the forces, or the

interactions of the system. It should be remarked that while the laws of motion are
general and are applicable to a wide variety of physical systems, the nature of forces
or the explicit form of interactions are specific to a given system and will differ from
system to system. The interactions are specified by giving explicit expressions for
forces acting on particles.
There are several different formulations of classical mechanics.
1. Newtonian Formulation
2. Lagrangian Formulation
3. Hamiltonian and Poisson Brackets
4. Hamilton Jacobi Formulation
17

§ 2.1 Newtonian formulation
You are all familiar with the newtonian mechanics from your school days. It starts
with the three laws of motion. A complete specification of state of a particle re-
quires three position coordinates and three velocities. The observable quantities
such as energy, momentum and angular momentum are functions of coordinates
and velocities.
Point particles: The equations of motion are given by the second law. For a
single point particle the equation of motion is
d2~r
m = F~α (3.1)
dt2
Note that the EOMare a set of second order differential equations in time. There-
fore, one needs to know the values of position and velocities at a time t0 in order to
be able to predict trajectories at a later time. One also needs to have information
about all the forces acting on the particle. The Newton’s laws require that the equa-
tions of motion be set up using the cartesian coordinates to describe the particle. For
a system consisting of several particles one needs to know all the forces, including
the forces of constraint. In order to set up equations of motion in a non-cartesian
system of coordinates one has to start from the cartesian system and change the
variables to to the desired set of coordinates.
Waves: The state of a vibrating string is described completely by specifying the

displacement and velocity of the string at each point. The vibrations are also gov-
erned by the Newton’s Laws which can be used to derive the wave equation giving
the propagation of waves in a medium.
Electromagnetic Field: The systems consisting of charged particles interacting

with electromagnetic fields are very important. These are governed by the Maxwell’s
equations and the Lorentz force equation. The state is described by specifying
position and momenta of the charged particles and the electromagnetic fields, or the
scalar and vector potentials, at all points in the space.
Thermodynamics, Statistical Mechanics: For systems consisting of a large

number of particles, such as gases, the classical mechanics is not very useful, one
needs statistical methods. While thermodynamics and statistical mechanics were
successful in describing the behaviour of a large number of systems very closely,
there were some notable disagreements with experiments.
The Newtonian formulation has limitations which make it unsuitable for descrip-
tion of several physical systems. Many different formalisms exist which generalise
the Newtonian formalism. We mention a few of these here which are useful for
systems with a few degrees of freedom.
18

§ 2.2 Lagrangian form of classical dynamics
In the Lagrangian approach, one needs a set of generalised coordinates and velocities.
The generalised coordinates are not restricted to be cartesian. They are a set of
independent variables needed to specify the system completely. The knowledge of
generalised coordinates, and their time derivatives, for a system allows us to compute
all dynamical variables of the system. The dynamical laws or the equations of motion
are given in terms of a single function, L(q, q̇, t), of generalised coordinates and
momenta called Lagrangian of the system. Knowing the Lagrangian, the equations
of motion are given by

d ∂L ∂L
− = 0, k = 1, . . . , n (3.2)
dt ∂ q̇k ∂qk
The Lagrangian formalism offers distinct advantages over the Newtonian formalism.
§ 2.3 Hamiltonian form of classical dynamics

In the Hamiltonian approach to the classical mechanics, the state of a system at time
t is described by giving the coordinate and canonical momenta qk , pk , (k = 1, · · · , n)
at that time. The canonical momentum pk is defined as derivatives of the Lagrangian
of the system w.r.t. the generalised velocity q̇k :
∂L
pk = , k = 1, 2, · · · , n. (3.3)
∂ q̇k
The interaction is specified by giving Hamiltonian H(q, p) which determines the

EOM. The EOMin the Hamiltonian approach take the form
∂H ∂H
q̇k = , ṗk = − k = 1, . . . , n (3.4)
∂pk ∂qk
For two functions F (q, p), G(q, p), the Poisson bracket [f, g]PB is defined as
X ∂F ∂G ∂F ∂G

[F, G]PB = − (3.5)
∂qk ∂pk ∂pk ∂qk
k
The Hamilton’s equations, 3.4, written in terms of Poisson brackets assume the form
q̇k = [q, H]PB , ṗk = [p, H]PB (3.6)
In general, the time evolution of any dynamical variable f (q, p) is given by
Ḟk = [F, H]PB (3.7)
• The Hamiltonian form of mechanics turns out to be most the convenient and
suitable for making a transition to quantum mechanics.
19

§3 Emergence of New Concepts
The inadequacy of classical theory and efforts to explain the observed physical phe-
nomena led to a major revision of classical concepts and of the mathematical struc-
ture. Some of the earliest points of departure from the classical theory were discrete
nature of physical processes, as in Planck’s hypothesis and quantisation of dynam-
ical variables, for example angular momentum in Bohr’s theory. The wave particle
duality had far reaching consequences and has changed the way we understand and
do physics. Here we will briefly highlight the important points that are best seen
to emerge from the implications of wave particle duality on thought experiments.
These are the indeterminacy, introduction of probabilities, uncertainty in simulta-
neous measurements of physical quantities.
The principle of superposition has been known for the superposition of amplitude
of waves and is responsible for the phenomenon of interference and polarisation. In
order to correctly describe the particle nature of light, i.e., the behaviour of a single
photon, in interference and polarisation experiments, one is led to the principle of
superposition of states. This superposition principle is not quite the same as in the
classical theory of light. For example, superposition of a quantum state with itself
does not produce a new state.
In a double slit interference experiment for electrons, we would expect to see an
interference pattern as is the case for any wave. However, when the intensity of the
incident beam is reduced and we seek the result of the experiment performed with a
single electron, we are inclined to conclude that one would see a spot on the screen,
and not an interference pattern, due to indivisibility of the particle nature. Still an
interference pattern appears when the experiment is repeated with a large number
of times. Within the commonly accepted interpretations of quantum mechanics, this
is explained by associating a probability amplitude of electron reaching a point on
the screen and accepting that a prediction about the exact location of the spot in
the experiment with a single electron is impossible and indeterminacy has entered
in an essential way.
It has to be emphasised that the indeterminacy is of fundamental nature of
physical systems and not a limitation of the quantum theory which is in complete
agreement with the experiments. However, one may still ask whether the quantum
theory is complete in its present form and if the unusual features appear due to the
restrictions on the domain of application such as nonrelativistic situations. We shall
not dwell upon these and other similar questions here, as a meaningful discussion is
beyond the scope of these lecture notes.
An analysis of several thought experiments to measure position and momentum
of particles leads to the Heisenberg uncertainty principle which asserts that the
position and momentum of a particle cannot be measured simultaneously. This
happens due to an uncontrollable disturbance, usually in momentum (position),
when a measurement of position (momentum) is made.
We are then naturally led to ask how are the states and dynamical variables
are to be described in quantum theory? How do we compute the allowed values of
dynamical variables which may be quantised? What is the method to compute the
probabilities of different possible outcomes of an experiment? The answers to all
such questions can, in principle, be reduced to an application of the postulates of
20

quantum mechanics.
§4 Postulates of Quantum Mechanics

1. The state of a quantum mechanical system is represented by a vector in a
complex vector space with inner product (Hilbert Space). The null vector does
not represent any state. Two non-null vectors represent the same state if and
only if they are linearly dependent. Thus two vectors |ψ1 i and |ψ2 i represent
the same state if there exists a complex number c such that
|ψ2 i = c|ψ1 i (3.8)
The vector that represents state of a physical system will be called state vec-
tor.The quantum mechanical states represented by vectors in Hilbert space are
called pure states.
There are other possible states which are called mixed states. These are rep-
resented by a density matrix ρ. The density matrix is an operator having
properties that it is hermitian and that its eigenvalues are between 0 and 1.
A density matrix ρ corresponds to a pure state if and only if ρ2 = ρ.
2. The dynamical variables of a physical system are represented by linear opera-

tors in the vector space. A linear operator representing a dynamical variable
must have real eigenvalues and their eigenvectors must form a complete set.
These properties are satisfied by self adjoint operators (hermitian operators).
So we demand that the dynamical variables be represented by self adjoint op-
erators in Hilbert space. An operator representing a dynamical variable will
also be called an observable.
3. A result of any measurement of a dynamical variable is one of the eigenvalues

of the corresponding operator. Conversely, every eigenvalue of an observable
representing a dynamical variable is a possible result of a measurement of the
dynamical variable.
As an example, let |u1i, |u2i, · · · , |un i represent the eigenvectors of an ob-
servable Â. If the state vector of a physical system, |ui, is an eigenvector
of an operator Â representing a dynamical variable A, a measurement of the
dynamical variable gives the corresponding value α with probability 1.
Conversely, if the measurement of A gives the value α with probability one, the
state of a system will be represented by a vector which will be an eigenvector
of the operator Â corresponding to the eigenvalue α.
Let the state of a system be given by |ψi, in general, a measurement of the
variable A results in values α1 , α2 , · · · , αn with probabilities |c1 |2 , |c2 |2 , ·, |cn|2
where c1 , c2 , ·, cn are the coefficients in the expansion of the state vector |ψi in
terms of eigenvectors of Â.
X
|ψi = ck |uk i (3.9)
k
21

Here |ψi and |uk i are assumed to be normalised.
hψ|ψi = 1, huk |uk i = 1 (3.10)
4. The operators corresponding to the generalised coordinates and momenta

qk , pk of a classical system satisfy
qî qˆj − qˆj qî = 0 (3.11)

p̂i pˆj − pˆj p̂i = 0 (3.12)
qî pˆj − pˆj qî = i~δij (3.13)
The above relations are called canonical commutation relations.
5. The time development of a system is governed by the Schrodinger equation

∂
i~ |ψti = Ĥ|ψti (3.14)
∂t
where |ψti is the state vector of the system at time t and Ĥ is the operator
representing the Hamiltonian of the system.
6. For a system of identical particles, the state of the system remains unchanged
under exchange of a pair of particles, it should be either symmetric or anti-
symmetric under an exchange of all the variables of the two identical particles.
22

Unit 4
Probability and Average Values
§1 The Third Postulate

The third postulate has several parts. In this section we use the notation A to
denote a dynamical variable, Â will denote the corresponding hermitian operator.
The eigenvalues will be real and it is assumed that the eigenvectors have been chosen
to be orthonormal. If we denote the eigenvalues of Â by α1 , α2 , . . . , αk , . . . and the
corresponding orthonormal eigenvectors by |u1i, |u2i, . . . , |uk i, . . . then we have
A|uk i = αk |uk i, huk |um i = δkm . (4.1)
1. The first part of the third postulate says that the only outcome of a measure-
ment of a dynamical variable A is one of the eigenvalues of the corresponding
operator Â. Thus, if an experiment to measure a dynamical variable A is
performed, the result must be only one of the eigenvalues αk . In particular,
an answer different, from every eigenvalue, cannot be the outcome of measure-
ment of A.
2. The next part of the postulate tells that if a system is represented by one of
the eigenvectors |un i, a measurement of the dynamical variable A will give the
corresponding eigenvalue αn .
3. The third postulate also tells us about the outcome of measurement of A

when the state vector |ψi is not an eigenvector of Â. In this case, the result
is some times one eigenvalue and sometimes another. We cannot predict the
result of a single measurement fully. When repeated measurements are made,
different eigenvalues αn will be obtained with different probabilities which can
be predicted. To compute the probabilities, we first expand the state vector |ψi
in terms of the eigenvectors of the operator Â corresponding to the dynamical
variable A which is being measured and write
X
|ψi = ck |uk i. (4.2)
k
Then the probability of getting value αk is given by |ck |2 . We continue to

assume that the state vector |ψi and the eigenvectors |uk i are orthogonal i.e.
23

,
hψ|ψi = 1, huk |uk i = 1 (4.3)
4. How do we compute the coefficients ck in Eq.(4.3)? The eigenvectors of a

hermitian operator are orthogonal and this helps in computing the coefficients.
Taking scalar product of Eq.(4.2) with |un i gives
X X
hun |ψi = ck hun |uk i = ck δkn = cn (4.4)
k k
Note that, in the right hand side of Eq.(4.4), only the term with k = n survives,
all other terms where k 6= n will vanish due to orthogonality property of the
eigenvectors.
∴ cn = hun |ψi (4.5)
5. The Parseval relation X

hψ|ψi = |ck |2 (4.6)
k
implies that X
|ck |2 = 1. (4.7)
k
if the state vector |ψi is normalised, hψ|ψi = 1. This suggests that the inter-
pretation of expressions |ck |2 as probabilities is consistent with the requirement
that the sum of all probabilities be equal to unity. We shall call the coefficient
ck as the probability amplitude for obtaining a value αk for the dynamical
variable A when the system is in state |ψi.
As a consequence of the the postulates, we have indeterminacy in the theoretical

predictions. The origin of this indeterminacy can be traced to the superposition
principle which in turn is needed to incorporate the wave nature of matter. In
classical mechanics the result of measurement of position, momenta, and every other
dynamical variables, can be fully predicted. This is no longer true in quantum theory.
Here is a summary.
Remarks
• A single measurement of A does not lead to a definite answer when the state
vector is not an eigenvector Â.
• In general, a result of a measurement of A must be one of the eigenvalues. The

outcome of a single experiment is indeterminate, and the quantum theory is
probabilistic by its nature in contrast to the classical theory which is determin-
istic. When measurement is repeated several times, we will sometimes get an
eigenvalue αj sometimes some other eigenvalue αk , and only the probabilities
of each outcome can be predicted.
• A simple consequence of the above discussion is that a measurement of a

dynamical variable A will give a value αm with probability 1 if and only if the
state is represented by corresponding eigenvector |um i.
24

• It must be remembered that some obvious changes, described later at the end
of the next section, will be needed when the eigenvalues of Â are continuous.
• Finally, we leave it as an exercise for you to convince yourself that the assump-
tion about the proabilities, as stated above, is correctly contained in the the
following statement. Given that the system is in a state described by the state
vector |ψi, the probability that it will be found in the state given by the vector
|φi is equal to |hφ|ψi |2.
§2 Computation of Probabilty and Average

Value
Let A denote a dynamical variable and Â the corresponding hermitian operator
representing A in quantum mechanics.Let |un i be normalised eigenvector of Â with
eigenvalue αn .
Â|un i = αn |un i (4.8)
These eigenvectors, being eigenvectors of a hermitian operator, will satisfy the or-
thogonality relation
hum |un i = δmn. (4.9)
If several repeated measurements are made on a system with state vector |ψi, one
would get αk with probability pk = |ck |2 where
ck = huk |ψi . (4.10)
The average of results of measurements of A in the state |ψi, to be denoted by hAiψ ,

will then be given by X X
hAiψ = pk αk = αk |ck |2 . (4.11)
k k
We will now show that the above expression coincides with hψ|Â|ψi . Without loss
of generality, we may assume that the state vector |ψi and the eigenvectors |uk i are
normalised. To prove this result we recall that ck are the expansion coefficients
X
|ψi = ck |uk i (4.12)
k
and compute hψ|Â|ψi .
hψ|Â|ψi = (ψ, Âψ) (4.13)

X X
= (ψ, Â ck u k ) = ck (ψ, Â uk ) (4.14)
k k
X X
= ck (ψ, αk uk ) = ck αk (ψ, uk ) (4.15)
k k
X X
= ck αk c̄k = αk |ck |2 . (4.16)
k k
25

which is seen to be equal to the average hAiψ from Eq.(4.11). When the state vector
|ψi is not normalised, the average value will be given by
hψ|Â|ψi
hAiψ = . (4.17)
hψ|ψi
Case of continuous eigenvalues: So far our discussion has been restricted to the
case when the eigenvalues of Â are discrete. Now consider the case when the eigen-
values α of Â are continuous and the corresponding eigenvectors |αi are normalised
to Dirac delta function
hα|α′i = δ(α − α′ ). (4.18)
In this case the probability that a measurement of A will give a value in a small
range α and α + dα is equal to |hα|ψi |2dα. For probability of finding the result in
Rb
between a and b, we would have the answer a |hα|ψi |2dα.
26

Unit 5
Canonical Quantisation
§1 Canonical Commutation Rules

The first three postulates of quantum mechanics give a general framework applicable
to any physical system. The first and the second postulates are about the mathe-
matical structure of the quantum theory. These two postulates furnish a description
of the states and dynamical variables of quantum systems. The third postulate
makes contact with experiments and talks about the possible outcomes of result of
a measurement. Recall that while the second postulate says that the dynamical
variables of a theory are represented by hermitian operators, no clue is provided in
the first three postulates about properties of these operators. For example, the po-
sition and momentum variables in quantum theory are replaced by operators, which
we denote by q̂ and p̂. However, the rules to manipulate these operators must be
formulated. In general, the product of two operators depends on the order in which
they are multiplied, so information about their commutator will be useful. The
canonical quantisation rule are important assumptions about the commutation rela-
tions of operators that represent position and momentum, and more generally, about
the commutation relations obeyed by a pair of operators representing generalised
coordinate and canonical conjugate momentum.
§2 Canonical commutation relations

Let q1 , q2 , . . . , qN be generalised coordinates of a classical system having N− de-
grees of freedom and let the corresponding canonical conjugate momenta be denoted
by p1 , p2 , . . . , pN . The canonical quantisation procedure consists in assuming that
the corresponding operators satisfy the following canonical commutation relations
(CCR).
[q̂i , q̂j ] = 0; [p̂i , p̂j ] = 0 (5.1)

[q̂i , p̂j ] = i~δij (5.2)
It should be noted that ~ has the correct dimension of product qk · pk and that the
commutators in Eq.(5.1), Eq.(5.2) are i~ times the corresponding Poisson brackets
in the classical theory.
27

The canonical commutation rule for a particle in one dimension the operators
q̂, p̂, corresponding to the position q and momentum p, takes the form
[q̂, p̂] = i~. (5.3)
For a particle in three dimensions the position and momenta are x, y, z, px , py , pz

and the nonzero commutators are
[x̂, p̂x ] = [ŷ, p̂y ] = [ẑ, p̂z ] = i~, (5.4)
and the commutators of all other pairs of operators are zero.

For general dynamical variables, other than coordinates and momenta, one has
the following correspondence between the commutators and Poisson brackets in the
limit ~ → 0..
1
lim [F̂ , Ĝ] = {F, G}P.B. (5.5)
~→0 i~
§3 Generalized Uncertainty Relation

Using the canonical commutation relation
ˆ
[x̂, p̂] = i~I, (5.6)
where Iˆ is the identity operator, we will derive the Heisenberg uncertainty relation
~
∆x∆p ≥ . (5.7)
2
The position momentum uncertainty relation is a special case of generalised relation
1
(∆A)ψ (∆B)ψ ≥ hψ|C|ψi (5.8)
2
for two hermitian operators A, B having the commutator [A, B] = iC.
Definition and properties of uncertainty ∆X

The uncertainty of a physical quantity X will be denoted as (∆X)ψ and is defined
as
(∆X)2ψ = hX 2 iψ − hXi2ψ , (5.9)

2
= h(X − X̄) iψ . (5.10)
where X̄ψ denotes the average in state |ψi. We assume the state vector is normalised,
X̄ = hX̂iψ = hψ|X̂|ψi . (5.11)
The operator X̂ must be hermitian and the average value will be real for every |ψi.
The two expressions for the uncertainty take the form
(∆X)2ψ = hψ|X̂ 2 |ψi − hψ|X̂|ψi 2 = hψ|(X̂ − X̄)2 |ψi . (5.12)
28

Since X̂ is hermitian, X̄ is real, the uncertainty in X is equal to the square of the
length of the vector (X̂ − X̄)|ψi:
k(X̂ − X̄)|ψik2 = hψ|(X̂ − X̄)† (X̂ − X̄)|ψi , (5.13)

= hψ|(X̂ − X̄)2 |ψi (5.14)
= (∆X)2ψ . (5.15)
It, therefore, follows that the uncertainty of a dynamical variable X in a state |ψi
is zero if and only if the state is an eigenstate of the operator X̂ and X̄ as the
eigenvalue.
We now assume that commutator of two hermitain operators A, B is
[A, B] = iC (5.16)
and we will show that the uncertainties in A, B satisfy the inequality in Eq.(5.8). To
simplify the notation, we drop the suffix expression of uncertainties and averages.
We will be using Cauchy Schwarz inequality
kf k2 kgk2 ≥ |hf |gi |2 (5.17)
with |f i = (A − Ā)|ψi and |gi = (B − B̄)|ψi. So we have
hf | = hψ|(A† − Ā∗ ) = hψ|(A − Ā) (5.18)

hg| = hψ|(B † − B̄ ∗ ) = hψ|(B − B̄) (5.19)
Therefore,
kf k2 = hf |f i = hψ|(A − Ā)2 |ψi (5.20)

kgk2 = hg|gi = hψ|(B − B̄)2 |ψi (5.21)
hf |gi = hψ|(A − Ā)(B − B̄)|ψi (5.22)
Using these expressions in the Cauchy Schwarz inequality we get
(∆A)2 (∆B)2 ≥ |hψ|(A − Ā)(B − B̄)|ψi |2 = |hψ|X|ψi |2 (5.23)
where we have used the notation X ≡ (A − Ā)(B − B̄). We write the operator X
as a sum of anticommutator and a commutator of operator (A − Ā)
(A − Ā)(B − B̄) = [(A − Ā), (B − B̄)]+ + [(A − Ā), (B − B̄)] (5.24)

C
= [(A − Ā), (B − B̄)]+ + i (5.25)
2
This splitting allows us to separate real and imaginary parts of hψ|X|ψi and to
compute the square of absolute value needed in Eq.(5.23).
1
hψ|X|ψi = hψ|[(A − Ā), (B − B̄)]+ |ψi +i hψ|C|ψi (5.26)
| {z } 2 | {z }
29

Both the expressions marked with braces, being averages of hermitian operators, are
real. Therefore, the inequality Eq.(5.23) becomes
(∆A)2 (∆B)2 ≥ |hψ|X|ψi |2 (5.27)

1
= |hψ|[(A − Ā), (B − B̄)]+ |ψi |2 + |hψ|C|ψi |2, (5.28)
4
1
∴ (∆A)2 (∆B)2 ≥ |hψ|C|ψi | 2
(5.29)
4
This gives us the desired uncertainty relation
1
(∆A)ψ (∆B)ψ ≥ |hψ|C|ψi |. (5.30)
2
When do we have a state for which the uncertainty product assumes the minimum
value? Going back to the first step and analysing the proof, it is easy to see that
the relation (5.30) becomes an equality if and only if |f i = λ|gi, for some complex
number λ, and if the average of the anticommutator in Eq.(5.28) is zero. Thus the
conditions for state |ψi to be state with minimum uncertainty becomes
(A − Ā)|ψi = λ(B − B̄)|ψi (5.31)

[(A − Ā), (B − B̄)]+ |ψi = 0. (5.32)
§4 Hamonic Oscillator Energy Levels Using

Commutators
Operator algebra for harmonic oscillator
The classical Hamiltonian for harmonic oscillator in one dimension is
p2 1
H= + mω 2 x2 (5.33)
2m 2
The corresponding operator Ĥ is obtained by replacing the position and momentum,
x, p by operators x̂, p̂ satisfying canonical commutation relations [x̂, p̂] = i~. We
introduce operators N, a, a† by
1
a = √ (p − imωx), (5.34)
2mω~
1
a† = √ (p + imωx), (5.35)
2mω~
N = a† a. (5.36)
It is easy to see that these operators satisfy the following identities
[a, a† ] = 1, [N, a† ] = a† , [N, a] = −a, (5.37)

p2 1
Ĥ = + mω 2 x2 = (N + 1/2)~ω. (5.38)
2m 2
30

Eigenvalues of N are positive
Let ν is an eigenvalue and |ψi be the corresponding normalised eigenvector:
N̂|ψi = ν|ψi. (5.39)
Then ν = hψ|N|ψi and the eigenvalue ν must be positive because
hψ|N|ψi = hψ|a† a|ψi = (aψ, aψ) = kaψk2 ≥ 0. (5.40)
The operator a lowers the eigenvalues of N

Let ν be an eigenvalue and |ψi corresponding eigenvector of N as in Eq.(5.39). The
operator a acts as a lowering operator for the eigenvalues of N. Consider the state
|φ1 i = a|ψi. Consider N|φ1 i
N|φ1 i = Na|ψi = (aN − a)|ψi (5.41)

= (aν − a)|ψi = (ν − 1)a|ψi (5.42)
∴ N|φ1 i = (ν − 1)|φ1i. (5.43)
and hence |φ1 i is an eigenvector of N with eigenvalue ν−1. Note that the result,an |ψi
, of applying n powers of the operator a on |ψi would give eigenvector with eigenvalue
ν − n. There must exist an integer n such that
an |ψi =
6 0 and an+1 |ψi = 0 (5.44)
Thus for the vector |φi ≡ an |ψi we have
a|φi = 0 ⇒ N|φi = a† a|φi = 0 = 0|φi (5.45)
and |φi is an eigenvector of N with zero as eigenvalue.
The eigenvalues of N are all nonnegative integers

Let normalised eigenvector of N with zero eigenvalue be denoted by |0i. Applying
a† on eigenvector |0i we get a sequence of vectors
a† |0i, (a† )2 |0i, , · · · (a† )m |mi, · · · , (5.46)
which are eigenvectors of N with eigenvalues 1, 2, · · · , m, · · · . We shall denote the

corresponding normalised vectors by
|0i, |1i, |2i, · · · , |mi, · · · . (5.47)
These are also eigenvectors of the Hamiltonian Ĥ, the ket |ni corresponds to the
eigenvalue (n + 1/2)~ω, because
Ĥ|ni = (N + 1/2)~ω|ni = ~ω(N|ni + 1/2|ni) = (n + 1/2)~ω|ni (5.48)
Hence the desired energy eigenvalues of the harmonic oscillator are En = (n+1/2)~ω.
31

Properties of eigenvectors
Let |ni be a normalised eigenvector of N with eigenvalue n. Action of a† on |ni
gives a vector proportional to |n + 1i and we write
a† |ni = cn |n + 1i. (5.49)
The constant cn can be found by taking the inner product of Eq.(5.49) with itself.
This leads to
hn|aa† |ni = |cn |2 hn + 1|n + 1i , (5.50)

(5.51)
and the left hand side of Eq.(5.50) can be shown to be equal to (n+1) by computing
aa† |ni = (a† a + 1)|ni = (N + 1)|ni = (n + 1)|ni. (5.52)
Therefore, we get (n + 1) = |cn |2 and hence the result

√
a† |ni = n + 1|n + 1i. (5.53)
In a similar fashion, one can get

√
a|ni = n|n − 1i. (5.54)
All the states |ni can be obtained by applying powers of a† on |0i:

1 1
|ni = √ a† |n − 1i = · · · = √ (a† )n |0i. (5.55)
n n!
§5 Angular Momentum Eigenvalues Using Com-

mutators
Angular momentum algebra
In this section we derive eigenvalues of angular momentum operators using commu-
tation relations.
[Jx , Jy ] = i~Jz ;
[Jy , Jz ] = i~Jx ;
[Jz , Jx ] = i~Jy ;
It is easy to see that J~2 = Jx2 + Jy2 + Jz2 commutes with each of the three components
Jx , Jy , Jz
[J 2 , Jx ] = 0 , [J 2 , Jy ] = 0 , [J 2 Jz ] = 0 . (5.56)
Note that J~2 and Jz form a commuting set of hermtian operators, we cannot add
any other component of J to this set. Therefore, there exists a complete set of
32

simultaneous, orthonormal, eigenvectors of these operators. Let us assume that |ψi
is a simultaneous eigenvector of J~2 and Jz with eigenvalues λ~2 and µ~ respectively.
J~2 |ψi = λ~2 |ψi, Jz |ψi = µ~|ψi. (5.57)
Note that ~ has dimensions of angular momentum, hence λ and µ will be dimen-
sionless numbers. Our aim is to get restrictions and find permissible values of λ and
µ . The results will apply to any set of operators obeying the above commutation
relations. Two most important techniques in the derivation of the eigenvalues will
be use of positivity of every operator of the form X † X, where X is any operator,
and use of ladder operators constructed out of the angular momentum components.
We define J± = Jx ± iJy and summarise some other important commutation

relations which are easily derived from the basic commutation relations given above.
[J+ , J− ] = 2~Jz (5.58)
[Jz , J+ ] = ~J+ (5.59)
[Jz , J− ] = −~J− (5.60)
J+ J− = J 2 − Jz2 + ~Jz (5.61)
J− J+ = J 2 − Jz2 − ~Jz (5.62)
1
J~2 = (J+ J− + J− J+ ) + Jz2 . (5.63)
2
Use positivity of J+ J− and J− J+ to get a bound on µ

The relation J+ = J−† implies that
hψ|J+ J− |ψi = (ψ, J+ J− ψ) = (ψ, (J− )† J− ψ) (5.64)
= (J− ψ, J− ψ) = ||J− ψ||2. (5.65)
It therefore follows that the operators
J+ J− = J 2 − Jz2 + ~Jz , (5.66)
J− J+ = J 2 − Jz2 − ~Jz , (5.67)
are positive operators in the sense that theiry have positive expectation values
hψ|J+ J− |ψi ≥ 0. (5.68)
for an arbitrary ket vector |ψi. Using J+ J− = J 2 − Jz2 + ~Jz , we get
hψ|J~2 − Jz2 + ~Jz |ψi ≥ 0 .
Since ψ is an eigenvector of J 2 and Jz we get
(ψ, (λ~2 − µ2 ~2 + µ~2 )ψ) ≥ 0 ⇒ λ − µ2 + µ ≥ 0
Similarly, starting with (ψ, J− J+ ψ), we can prove that
λ − µ2 − µ ≥ 0 .
These two relations give µ2 < λ, thus absolute value of eigenvalue of Jz is bounded.
This result corresponds to the fact that the length of projection of a vector is less
than the length of the vector.)
33

J+ is a raising operator for Jz .
Action of J+ on |ψ > gives a vector |φ1 i
|φ1 i = J+ |ψi
which is also a simultaneous eigenvector of J 2 and Jz with eigenvalues λ~2 and
(µ + 1)~, respectively, i.e.
J 2 |φ1 i = λ~2 |φ1 i (5.69)
and Jz |φ1 i = (µ + 1)~|φ1 i. (5.70)
Proof:
J 2 |φ1 i = J 2 (J+ |ψi) (5.71)
= (J 2 J+ )|ψi ( because 2
J J+ = J+ J ) 2
(5.72)
= J+ (J 2 |ψi) (5.73)
= J+ (λ~2 |ψi). (5.74)
Hence
J 2 |φ1 i = λ~2 J+ |ψi = λ~2 |φ1 i. (5.75)
Also
Jz |φ1i = Jz J+ |ψi use Jz J+ − J+ Jz = ~J+ (5.76)
= (J+ Jz + ~J+ )|ψi (5.77)
= (J+ µ~ + ~J+ )|ψi (5.78)
= (µ + 1)~J+ |ψi = (µ + 1)~|φ1i (5.79)
Jz |φ1i = (µ + 1)~|φ1i. (5.80)
Similarly, the states |φ2 i = J+ |φ1 i, |φ3i = J+ |φ2 i · · · , obtained by repeated action
of J+ on |ψi, are eigenvectors of J 2 and Jz as given below.
States Eigenvalue of J 2 Eigenvalue of Jz
|φ1 i = J+ |ψi λ~2 (µ + 1)~
|φ2 i = J+ |φ1 i = J+2 |ψi λ~2 (µ + 2)~
|φ3 i = J+ |φ2 i = J+3 |ψi λ~2 (µ + 3)~
··· ··· ···
J− is a lowering operator for Jz .

In a manner similar to the step II above, the commutation relation
Jz J− − J− Jz = −~J−
or
Jz J− = J− (Jz − ~)
can be used to show that the states |χ1 i, |χ2 i, · · · , obtained by repeated application
of J− on |ψi are also eigenvectors of J 2 and Jz with eigenvalues as given below.
34

States Eigenvalue of J 2 Eigenvalue of Jz
|χ1 i = J− |ψi λ~2 (µ − 1)~
|χ2 i = J− |χ1 i = (J− )2 |ψi λ~2 (µ − 2)~
|χ3 i = J− |χ2 i = (J− )3 |χ3 i λ~2 (µ − 3)~
··· ··· ···
Thus, if µ is an eigenvalues of jz , (µ±1)~, (µ±2)~, (µ±3)~ · · · are all eigenvalues

of Jz . This statement holds as long as an application of J+ (or J− ) gives a non-zero
vector.
The difference between the maximum and minimum values of µ is an

integer
We have already seen that the eigenvalues λ and µ satisfy
µ2 ≤ λ.
This relation shows that for a fixed λ, the eigenvalue µ of Jz cannot increase or
decrease indefinitely. Thus, for a given value of λ, there is a maximum eigenvalue
and there is a minimum eigenvalue of Jz .
Let µ1 be the minimum eigenvalue of Jz for given λ and let |χi be corresponding
eigenvector
J 2 |χi = λ~2 |χi Jz |χi = µ1 ~|χi.
Then action of J− on |χi must give null vector, otherwise J− |χi will be eigenvector
of Jz with eigenvalue (µ1 − 1)~ and this would contradict the assumption that µ1 is
the minimum eigenvalue of Jz (for fixed λ). Hence
J− |χi = 0 .
Taking norm of J− |χi, we get
hχ|J+ J− |χi = 0 ⇒ hχ|J 2 − Jz2 + ~Jz |χi = 0 ⇒ λ~2 − µ21 ~2 + µ1 ~2 = 0. (5.81)
Hence, we have the relation

λ − µ21 + µ1 = 0. (5.82)
Repeated application of J+ on |χi, will generate eigenvectors of Jz by with
eigenvalues (µ1 + 1)~, (µ1 + 2)~, · · · . Let µ2 ~ be the maximum allowed eigenvalue,
the mu2 = µ1 + N for some N. Let |φi be the corresponding eigenvector of J~2 and
Jz . Then J+ |φi = 0. Taking norm of J+ |φi we get
hφ|J− J+ |φi = 0 or hφ|J 2 − Jz2 − ~Jz |φi = 0.

or
λ~2 − µ22 ~2 − µ2 ~2 = 0,
Therefore, we have
λ − µ22 − µ2 = 0.
35

The above equation and corresponding equation for µ1
λ − µ21 + µ1 = 0
imply that
µ22 + µ2 − µ21 + µ1 = 0, (5.83)

or (µ2 − µ1 )(µ2 + µ1 ) + (µ2 + µ1 ) = 0, (5.84)
or (µ2 + µ1 )(µ2 − µ1 + 1) = 0, (5.85)
µ2 = −µ1 or µ2 = µ1 − 1. (5.86)
Since µ2 = µ1 − 1 =⇒ µ2 < µ1 which contradicts our earlier assumptions that

µ2 is the maximum and µ1 is the minimum allowed eigenvalue of Jz , hence we must
have
µ2 = −µ1 .
This equation, when used with µ2 − µ1 = N, implies that
N
µ2 = −µ1 = , (5.87)
2
N N
and λ = µ2 (µ2 + 1) = +1 . (5.88)
2 2
Next we change notation and call N2 = j, and state our results as follows. The
eigenvalues of J~2 are j(j+1)~2 with allowed values j = 0, 12 , 1, 32 , · · · . For a fixed value
of j, the eigenvalues of Jz are between j~ and −j~. The simultaneous, normalised,
eigenvectors of J~2 and Jz will be denoted by |j, mi. The m~ is the eigenvalue of Jz
where m can take (2j + 1) values −j, −j + 1, · · · , j.
§6 Summary of important results

1. It is customary to denote the normalised, simultaneous,eigenvectors of J~2 and
Jz by |j, mi so that
J~2 |j, mi = j(j + 1)~2 |j, mi, (5.89)

Jz |j, mi = m~|j, mi, (5.90)
and m = −j, −j + 1, · · · , j.
2. The operators J± acting on |j, mi give a ket vector proportional to |j, m ± 1i.
The proportionality coefficient can be worked out using the relation Eq.(5.61)
and Eq.(5.62). Writing
J+ |j, mi = C|j, m + 1i, (5.91)
and taking scalar product of this equation with itself we get
hj, m|J− J+ |j, mi = |C|2 hj, m + 1|j, m + 1i . (5.92)
36

Using the relation J− J+ = J 2 −Jz2 −~Jz , and the fact that |j, mi and |j, m+1i
are normalised, we get
p
|C|2 = j(j + 1)~2 − m2 ~2 − m~2 ⇒ C = j(j + 1) − m(m + 1)~. (5.93)
This result and similar steps for J− |ψi give

p
J+ |j, mi = j(j + 1) − m(m + 1) ~ |j, m + 1i, (5.94)
p
J− |j, mi = j(j + 1) − m(m − 1) ~ |j, m − 1i. (5.95)
3. The operators J+ and J− annihilate the states with the highest and lowest Jz
values, respectively, because the Jz value cannot be increased beyond j nor
can it be decreased beyond −j.
J+ |j, ji = 0, J− |j, −ji = 0. (5.96)
4. In the above discussion we have worked with J~2 and Jz , however the results
remain true for component of J~ along any direction. In fact one can find
simultaneous eigenvectors of J~2 and any one component, Jn = n̂ · J, ~ along a
~
fixed direction given by the unit vector n̂. The eigenvalues of J are j(j + 1)~2
2
where j can be integer of half integer. The eigenvalues of a component of Jn

take values from −j to j in step of 1.
5. The above results are applicable to operators satisfying angular momentum

commutation relations except that the half integral values are ruled out for
the orbital angular momentum, because of additional requirement of single
valuedness of the wave function.
6. A crucial techniques in the derivation of the eigenvalues of angular momentum

has been use of positivity for operators of the form X † X and raising and
lowering action of J± . These techniques are very general and appear in many
other applications involving the use of commutator algebra.
37

Unit 6
Time Evolution
Introduction
So far we have been concerned with description of a quantum mechanical system at
one time. Now we take up the time evolution of a quantum system. The guiding
principle for time evolution operator will be superposition principle and preservation
of norm. This restricts time evolution operators to a unitary operator. Correspon-
dence with the classical equations of motion in the Poisson bracket form will be
established. Stationary property of energy eigenstates will be seen as a simple con-
sequence of the Schrodinger equation.
§1 Superposition principle and time develop-

ment
Let |ψt0 i represent the state of system at time t0 and |ψti represent the state at
time t. We assume that |ψt0 i at time t0 determines the state at time t completely.
The principle of superposition should apply between these two times t0 and t. If we
have a relation at time t0
|ψt0 i = α|χt0 i + β|φt0i, (6.1)
between three possible states, |ψi, |χi, |φi, the same must hold at all time t when
the system is left undisturbed :
|ψ(t)i = α|χti + β|φti. (6.2)
Thus, if we write
|ψti = U(t, t0 )|ψt0 i etc.. (6.3)
Then U(t, t0 ) must be a linear operator independent of ψ. Obviously U must reduce
to the identity operator at time t = t0
U(t0 , t0 ) = I . (6.4)
Next we demand that the norm of vector |ψti should not change with time and
hence
hψt|ψti = hψt0 |ψt0 i for all t. (6.5)
38

The above requirements (2) and (5), respectively, imply that the operator U must
be a linear operator and that it must be unitary.
UU † = U † U = I. (6.6)
We shall now derive a differential equation satisfied by the state vector at time t.
We, therefore, compute
d |ψ, t + ∆ti − |ψti
|ψti = lim (6.7)
dt ∆t→0 ∆t
(U(t + ∆t, t) − I)
= lim |ψti (6.8)
∆t→0 ∆t
= X̂|ψti (6.9)
(6.10)
where the operator X̂ is given by
U(t + ∆t, t) − I
X̂(t) = lim = (6.11)
∆t→0 ∆t
U(t + ∆t, t) − U(t, t)
= lim (6.12)
∆t→0 ∆t
U(t + ∆t, t0 ) − U(t, t0 )
= lim |t0 =t (6.13)
∆t→0 ∆t
d
= U(t, t0 )|t0 =t (6.14)
dt
In the last two steps, the arugment has been changed to t0 so as to write the operator
X̂(t) as a time derivative. The operator X̂ can be shown to be anti-hermitian,
therefore we write Eq.(6.8) as
d
i~ |ψti = Ĥ(t)|ψti, (6.15)
dt
where
1 ∂
Ĥ(t) = U(t, t0 )|t0 =t . (6.16)
i~ ∂t
We shall now show that H(t) must be a hermitian operator. Consider
U † (t, t0 )U(t, t0 ) = I. (6.17)
Differentiating w.r.t. t we get

∂ † † ∂
U (t, t0 ) U(t, t0 ) + U U(t, t0 ) = 0. (6.18)
∂t ∂t
Setting t0 = t and using U(t, t) = I we have
d † d
U (t, t0 )|t0 =t + U(t, t0 )|t0 =t = 0 (6.19)
dt dt
or (+i~Ĥ)† + (i~Ĥ) = 0 (6.20)
or − iĤ † + iĤ = 0 (6.21)
or Ĥ † = Ĥ (6.22)
39

Thus, the time evolution of a quantum system is governed by the equation
∂
i~ |ψti = Ĥ(t)|ψti (6.23)
∂t
Using Eq.(6.3) in Eq.(6.23) we get
∂
i~ U(t, t0 )|ψt0 i = Ĥ(t)U(t, t0 )|ψt0 i (6.24)
∂t
This equation must hold for all vectors |ψt0 i. Hence the time evolution operator U
must satisfy the differential equation
∂
i~ U(t, t0 ) = Ĥ(t)U(t, t0 ) . (6.25)
∂t
We shall follow Dirac and using correspondence with classical mechanics to show
that the operator Ĥ represents the energy, i.e. the Hamiltonian, of the system.
§2 Time dependent Schrodinger equation

Hamiltonian as generator of time evolution
We shall now show that the operator Ĥ appearing in the time evolution equation
d
i~ |ψti = Ĥ|ψti (6.26)
dt
represents the Hamiltonian of the system. For this purpose we obtain an equation
for time evolution of average quantities and use correspondence with the classical
equation for the time development of dynamical variables.
Let F (q, p, t) be an dynamical variable of the system and let F̂ denote the cor-
responding operator. We are interested in finding out how the average value
hF̂ i ≡ hψt|F̂ |ψti (6.27)
changes with time. The time dependence of the average value comes from depen-
dence of the three objects, the operator F̂ , the bra vector hψt|, and the ket vector
|ψti, present in 6.27. The equation conjugate to the Schrodinger equation
d
i~ |ψti = Ĥ|ψti (6.28)
dt
is given by
d
−i~ hψt| = hψt|Ĥ † . (6.29)
dt
Since the operator Ĥ is hermitian, the above equation takes the form
d
−i~ hψt| = hψt|Ĥ (6.30)
dt
40

Therefore we get

d d dF̂ d
hF̂ i = hψt| F̂ |ψti + hψt| |ψti + hψt|F̂ |ψti . (6.31)
dt dt dt dt
Using Eq.(6.29) and Eq.(6.30) in Eq.(6.31) we get
d 1 dF̂ 1
hF̂ i = − hψt|Ĥ F̂ |ψti + hψt| |ψti + hψt|F̂ Ĥ|ψti. (6.32)
dt i~ dt i~
The above equation is rearranged to give the final form
d ∂ 1
hF̂ i = hF̂ i + h [F̂ , Ĥ] i. (6.33)
dt ∂t i~
Comparing the above equation with the corresponding equation in classical me-
chanics for time evolution of dynamical variables
dF ∂F
= + {F, H}P B (6.34)
dt ∂t
and remembering that the commutator divided by i~ corresponds to the Poisson
bracket in the limit ~ → 0, we see that Ĥ must be identified as the operator
corresponding to the Hamiltonian H of the system. Unless mentioned otherwise, we
shall always assume that H is independent of time.
Constants of motion
If an operator F̂ commutes with the Hamiltonian operator Ĥ we have
[F̂ , Ĥ] = 0 (6.35)

∂F
and if F̂ does not contain explicit time dependence, ∂t
= 0, Eq.(6.33) shows that
d
hψt|F̂ |ψti = 0 (6.36)
dt
Therefore, in an arbitrary state, the average value of F̂ does not change with time.
Such a dynamical variable will be called a constant of motion.
Ehrenfest theorem
Consider a particle moving in a potential V (x), with Hamiltonian given by
p2
H= + V (x), (6.37)
2m
and the Hamiltonian operator is obtained by replacing x, p by the corresponding
operators x̂, p̂. Using the commutators
p̂ \
dV (x)
[b
x, Ĥ] = i~ , [b
p, Ĥ] = −i~ , (6.38)
m dx
41

we get the equation for time development of averages as, see Eq.(6.35),
d p̂
hx̂i = h i (6.39)
dt m* +
d \
dV (x)
hp̂i = − . (6.40)
dt dx
Similarity of the above equations with the classical equations is obvious.
§3 Solution of the time dependent equation

We now wish to solve the time dependent Schrodinger equation
d
i~ |ψti = Ĥ|ψti, (6.41)
dt
and derive an expression for the state vector at time t in terms of the state vector |ψ0 i
at time t = 0. For present purposes we assume that the Hamiltonian of the system is
independent of time. Since H is hermitian, its eigenvectors from a complete set we
can expand |ψti in terms of the eigenvectors. Denoting the normalised eigenvectors
by |En i, we write X
|ψti = cn (t)|En i. (6.42)
n
Substituting the expansion 6.42 in Eq.(6.41) we get

d X X
i~ cn (t)|En i = Ĥ cn (t)|En i (6.43)
dt n n
X dcn (t) X
⇒ i~ |En i = cn (t)Ĥ|En i (6.44)
n
dt n
Taking scalar product with |Em i and using orthonormal property of the eigenvectors
|En i, we get
dcm (t)
i~ = Em cm (t) (6.45)
dt
which is easily solved to give
cm (t) = cm (0) exp(−iEm t/~) (6.46)
Therefore, the state vector at time t can be written as

X
|ψti = cn (0) exp(−iEn t/~)|En i. (6.47)
n
The coefficients cn can be determined if the state vector is known at initial time,
say, t = 0. Setting t = 0 in Eq.(6.47), denoting the state vector at time t = 0 by
|ψ 0i, we get
X
|ψ 0i = cn (0)|En i. (6.48)
n
42

The unknown coefficients cn (0) can now be computed, taking the scalar product of
Eq.(6.47) with |Em i one gets
cm (0) = hEm |ψ 0i . (6.49)
The Eq.(6.47), 6.49 give the solution of the time dependent Schrodinger equation.
Recalling the definition of function of an operator, the Eq.(6.47) can be cast as
X
|ψti = exp(−iĤt/~) cn (0)|En i (6.50)
n
. = exp(−iĤt/~)|ψ, 0i. (6.51)
In general, if the state vector is known at time t0 , the time evolution solution
Eq.(6.51) takes the form
X
|ψti = exp(−iĤ(t − t0 )/~) cn (t0 )|En i (6.52)
n
. = exp(−iĤ(t − t0 )/~)|ψ, t0i. (6.53)
Comparing with
|ψti = U(t, t0 )|ψt0 i, (6.54)
we see that the time evolution operator is given by
U(t, t0 ) = exp(−iĤ(t − t0 )/~). (6.55)
Stationary states
Let us now consider time evolution of a system if it has a definite value of energy at
an initial time t0 . The value of the energy then has to be one of the eigenvalues and
the state vector will be the corresponding vector. So |ψt0 i = |Em i, then at time t the
system will be in the state given by |ψt0 i = exp(−iEm (t−t0 )/~)|Em i. Note that the
state vector at different times is equal to the initial state vector times a numerical
phase factor (exp(−iEm (t − t0 )/~)). Therefore, the vector at time t represents the
same state at all times. Such states are called stationary states because the state
does not change with time. Every eigenvector of energy is a possible stationary
state of a system. In such a state the average value of a dynamical variable, X̂,
not having time explicitly, is independent of time even if X̂ does not commute with
Hamiltonian. In fact the probabilities of finding a value on a measurement of the
dynamical variable are independent of time.
43

Summary
• Given the state of the system at a time t0 , the state vector at any other time
is related to it by a unitary transformation U(t, t0 ).
|ψti = U(t, t0 ) |ψt0 i.
• The equation of motion of quantum system is the Schrodinger equation

d
i~ |ψti = Ĥ|ψti.
dt
where Ĥ is the Hamiltonian operator of the system.
• The time evolution operator satisfies the equation

∂
i~ U(t, t0 ) = Ĥ(t)U(t, t0 ).
∂t
• If the Hamiltonian does not depend on time, the evolution operator is
U(t, t0 ) = exp[−iĤ(t − t0 )/~].
• The average value of a dynamical variable,F̂ , satisfies
d ∂ F̂ 1
hF̂ i = h i + h [F̂ , Ĥ] i.
dt ∂t i~
• A dynamical variable is a constant of motion if it commutes with the Hamil-

tonian.
• The energy eigenstates of a system are stationary; they do not change with
time. The state vector of a stationary state at any time is equal to the initial
state vector multiplied by a numerical phase factor.
• The average value of a constant of motion G is independent of time in every

possible state of the system including nonstationary states.
• The average value of every dynamical variable is independent of time in sta-

tionary states.
44

Unit 7
Working with Representations
§1 Introduction to representations
In the following sections we will introduce the coordinate and momentum repre-
sentations and discuss the relationship between the two representations. Here we
briefly recall main results for representations in a finite dimensional vector spaces.
Let {|en i, n = 1, · · · , N} be an o.n. basis in a finite dimensional vector space. The
orthogonality property is
hem |en i = δmn . (7.1)
The completness property of the basis states that the sum of all projection operators
|en ihen | is the identity operator.
N
X
ˆ
|en ihen | = I.. (7.2)
n=1
A representation with respect to a chosen basis is constructed by constructing a n-

component column vector f for every vector |f i in the vector space and an n × n
matrix T for every operator T . The rules for writing these representatives are
 
he1 |f i
 he2 |f i 
|f i → f = 
 ..  ,
 (7.3)
hen |f i
 
he1 |T̂ |e1 i he1 |T̂ |e2 i · · · he1 |T̂ |en i
 he2 |T̂ |e1 i he2 |T̂ |e2 i · · · he2 |T̂ |en i 
T̂ → T = 
 ···
. (7.4)
··· ··· ··· 
hen |T̂ |e1 i hen |T̂ |e2 i · · · hen |T̂ |en i
Having introduced a representation, all equations involving vectors and operators

can be written as matrix equations. This is most conveniently achieved by making
45

use of the completeness relation Eq.(7.2), for example
|vi = T |ui ⇒ hm|vi = hm|T |ui (7.5)

X
⇒ hm|vi = hm|T |nihn| |ui (7.6)
n
X
⇒ hm|vi = hm|T |nihn|ui . (7.7)
n
The last equation is just the matrix equation, v = T u, written in terms of the
representatives of the abstract vectors |ui, |vi. In quantum mechanics one has to
work with infinite dimensional vector space. The number of components of f becomes
infinite and the matrices T have infinite rows and infinite columns. Two of the most
useful choices representations are the coordinate representation and the momentum
representation, in which the index n is replaced by a a continuous index, such as
x, or p, which can take all real values instead of positive integral values, as is the
case for a finite dimensional vector space. In these situations one we can not display
corresponding column vectors or the matrices, but all rules of matrix manipulation
will apply; the summation over an index n is replaced with integration over x.
Keeping this in mind, we come to the most commonly used representation called
the coordinate representation.
§2 The coordinate representation

The CCR [x̂, p̂] = i~ shows that if x0 is an eigenvalue of x̂, x0 +a is also an eigenvalue.
This is most easily seen by making use of the relation
x̂ exp(−iap̂/~) = exp(−iap̂/~) (x̂ + a) (7.8)
here a is a real number. To see this we note that U(a) ≡ exp(−iap̂/~) is a unitary
operator. If |x0 i is an eigenvector with eigenvalue x0 , then U(a)|x0 i is an eigenvector
of x̂ with eigenvalue x0 + a. Using Eq.(7.8) we get

x̂ U(a)|x0 i = exp(−iap̂/~) (x̂ + a)|x0 i (7.9)

= (x0 + a) U(a)|x0 i . (7.10)
This show that (x0 + a) is an eigenvalue with U(a)|x0 i as the eigenvector. As a

is any real number, all real values are allowed for the eigenvalues of x̂. a similar
argument shows that all real values are allowed as eigenvalues of p̂.
We shall at first consider a particle in one dimension. To set up the coordinate
representation we use the eigenvectors of the position operator x̂ as o.n. basis. We
assume that the eigenvalues of x̂ are all real values in the range (−∞, ∞). Let x be
one such eigenvalue and |xi be the corresponding eigenvector, i.e.
x̂|xi = x|xi. (7.11)
The orthogonality property hm|ni = δmn now assumes the form
hx′′ |x′ i = δ(x′′ − x′ ) (7.12)
46

P
and the completeness relation n |nihn| = Iˆ takes the form
Z ∞
ˆ
|xihx|dx = I, (7.13)
−∞
P
where Iˆ denotes
R the identity operator. Everywhere the sum n over all eigenvalues
is replaced by dx as the eigenvalues are now continuous. This choice of basis leads
to the coordinate representation or the position representation also known as
the Schrodinger representation.
Thus an expansion of abstract vector |ψi in the basis |xi becomes
Z
|ψi = dx|xihx|ψi . (7.14)
and the abstract vector |ψi is represented by numbers hx|ψi for x having a value
from real numbers. Instead of arranging all the components of |ψi in form of a
column vector, we write them as a function of x:
hx|ψi → ψ(x). (7.15)
Due to the fact that the eigenvalues of x are continuous, it is not meaningful to ask
for the probability of getting a single value x0 ; instead we must ask for probability
that position has a value in a specified range, such as x and x + dx. The third
postulate tells us that the probability of finding the position between x and x + dx
is given by
|hx|ψi |2dx = |ψ(x)|2 dx (7.16)
Thus the function ψ(x), to be called the wave function, gives the probability
density of position. The Parseval relation
Z ∞
hψ|ψi = |ψ(x)|2 dx. (7.17)
−∞
states that the total probability will be unity for normalised vectors |ψi.
In the coordinate representation, the states are, therefore, represented by square
integrable functions of x, i.e. functions ψ(x) such that
Z ∞
|ψ(x)|2 dx = hψ|ψi ≤ ∞. (7.18)
−∞
While the square integrable wave functions will be appropriate for describing the
bound states, a class of other functions, plane waves, will also be acceptable and
describe physical situations.
The operator x̂ will be represented by an infinite dimensional matrix, ‘diagonal
matrix’, labeled by continuous indices x′′ , x′ and the matrix elements of x̂ are
hx′ |x̂|x′′ i = x′ δ(x′ − x′′ ). (7.19)
I will give the matrix representation of the momentum operator without proof.
d
hx′ |p̂|x′′ i = −i~ ′
δ(x′ − x′′ ). (7.20)
dx
47

The action of the position and momentum operators, x̂ and p̂, on the wave function
ψ(x) is seen to be
d
x̂ψ(x) = xψ(x), p̂ψ(x) = −i~ ψ(x). (7.21)
dx
d
Thus in the position representation, we have p̂ → −i~ dx . This is seen to be consistent
with the canonical commutation relation This representation of p̂ is unique upto a
change of basis. An operator corresponding to a dynamical variable can be obtained
by making the replacement
x → x̂; p → p̂. (7.22)
The most imporatnt dynamical variable of a system is energy or the Hamiltonian:
p2
H= + V (x) (7.23)
2m
(7.24)
The corresponding operator is

p̂2 ~2 d 2
Ĥ = + V (x̂) = − + V (x̂) (7.25)
2m 2m dx2
Several degrees of freedom

Generalisation to a particle in three dimension is straight forward. The basis vectors,
|~ri, in this case are simultaneous eigenvector of position operators x̂, ŷ, ẑ:
x̂|~ri = x|~ri, ŷ|~ri = y|~ri, ẑ|~ri = z|~ri. (7.26)
The orthogonality and completeness relations take forms:

Z
′′ ′ ′′ ′
h~r |~r i = δ(~r − ~r ), ˆ
d3 r|~r ih~r | = I. (7.27)
Expansion of an arbitrary |ψi in the basis |~r i assumes the form

Z
|ψi = |~r ih~r |ψi d3r. (7.28)
We call the function h~r |ψi the wave function and also denote it by ψ(~r). The
absolute square |ψ(~r)|2 dV is the probability of particle being in a small volume dV .
The corresponding probability for the particle to be in a finite volume V is given by
the volume integral
y y
|h~r |ψi |2d3 r = | V psi(~r)|2 d3 r. (7.29)
V
The action of the position operators ~rˆ is to multiply the wave function by ~r and
that of the momentum operators p~ˆ is −i~ times differentiation as given below.
~rˆψ(~r) = ~rψ(~r), ~pˆψ(~r) = −i~∇ψ(~r) (7.30)
48

§3 Momentum representation
In many problems it is useful to work with the momentum representation. In
the momentum representation the basis is eigenvectors of the momentum operators.
The following table summarises the coordinate and momentum representations
Comparison of coordinate and momentum representations
Choice of basis vectors |~r i are simultaneous eigen- |~p i are simultaneous eigen-
vectors of position operators vectors of momentum
operators
Orthogonality h~r ′′ |~r ′ i = δ(~r ′′ − ~r ′ ) h~p ′′ |~p ′ i = δ(~p ′′ − p~ ′ )
R 3 R 3
Completeness formula d r|~r ih~r | = Iˆ d p|~p ih~p | = Iˆ
R R
Expansion in the basis |ψi = d3 r|~r ih~r |ψi |ψi = d3 p|~p ih~p |ψi
State representative Coordinate space wave Momentum space wave
function ψ(~r ) function ψ̃(~p )
R R
Length squared hψ|ψi hψ|ψi = |ψ(~r )|2 d3 x hψ|ψi = |ψ̃(~p )|2 d3 p
R ∗ R ∗
Scalar product hψ|φi ψ (~r)φ(~r )d3 r ψe (p)φ(~
e p )d3 p
e p ) = +i~ ∂ e
Action of the position oper- x̂ψ(~r) = xψ(~r ) x̂φ(~ φ(~p )
∂px
ators
∂ e p ) = px φ(~
e p)
Action of the momentum p̂x ψ(~r ) = −i~ ψ(~r ) p̂x φ(~
∂x
operators
∂ ∂
Operator for F̂ (qj , pk ) F̂ q̂j , −i~ F̂ + i~ , pk
∂qk ∂pk
It must be noted that two different operators can be written for the same classical
dynamical variable. Classically pq and qp are the same functions but operators p̂q̂
and q̂ p̂ are different. So which operator should correspond to pq(= qp)? This is
known as operator ordering problem. The only guideline avaliable is the operator
must be hermitian. So we must select
1
pq → p̂q̂ + q̂p̂ (7.31)
2
For more complicated expressions the requirement of hermiticity is not sufficient to
get a unique answer.
§4 Change of representation
We shall now discuss the connection between the two representations introduced in
the previous sections. The relationship between two representations is most conve-
niently displayed using the completeness formula.
Z Z
ˆ
|xihx| dx = I |pihp| dp = Iˆ (7.32)
49

So given a ket vector |f i how do we relate the two representatives in the coordinate
and momentum representations? This process is illustrated below. We begin with
the wave function f (x) :
ˆ i
f (x) = hx|f i = hx|I|f (7.33)
Z
= hx| |pihp| dp|f i (7.34)
Z
= dphx|pi hp|f i (7.35)
Z
= ˜
dphx|pi f(p) (7.36)
This gives the desired relation between the coordinate space wave function f (x)
and the momentum space wave function f(p). ˜ However, we still need to answer:
what are the transformation functions hx|pi ? This is easy to answer. For every
vector |f i, the inner product hx|f i is the wavefunction for the state represented by
|f i. So hx|pi is the wave function for the state |pi, i.e., the wave function for state
with definite momentum value p. This is easily computed by solving the eigenvalue
problem for the momentum operator.
Eigenfunctions of momentum operator

Delta function normalisation: Let up (x) denote the wave function for the
state with momentum p. Then
dup (x)
p̂up x = pup (x) ⇒ −i~ = pup (x). (7.37)
dx
The solution of the last differential equation is easy to obtain and is given by
up (x) = C exp(ipx/~). (7.38)
The normalisation constant C is fixed by demanding the δ- function normalisation
which follows from the orthonormality property of the states |pi:
δ(p′′ − p′ ) = hp′′ |p′ i . (7.39)
Z
= hp | dx|xihx||p′i
′′
(7.40)
Z
= dx hp′′ |xi hx|p′i (7.41)
Z
= dx u∗p′′ (x)up′ (x) (7.42)
It has been shown that the function up (x) must satisfy the δ- function normalisation
condition Z
dx u∗p′′ (x)up′ (x) = δ(p′′ − p′ ) (7.43)
Using this condition we get
Z
dx C ∗ exp(−ip′′ x/~)C exp(ip′ x/~) = δ(p′′ − p′ ) (7.44)
|C|2 2π~ = 1 (7.45)
50

where in writing the last step we have used the identity
Z
dx exp(i(p′ − p′′ )x/~) = 2π~δ(p′′ − p′ ). (7.46)
√
Thus we see that C = 1/ 2π~ and the normalised momentum eigenfunctions of
momentum are given by
1
up (x) = √ exp(ipx/~). (7.47)
2π~
The relation between the coordinate and momentum space wave functions becomes
Z
1
f (x) = √ exp(ipx/~)f˜(p)dp. (7.48)
2π~
The inverse relationship is easily written down.
Z
˜ 1
f (p) = √ exp(−ipx/~)f (x)dx. (7.49)
2π~
All the equations have been written down only in one dimension, the generalisation
to three dimensions should be obvious.
Box normalisation: The operators x̂ and p̂ are unbounded operators and we

must deal with infinite dimensional vector spaces. These require a deeper level of
mathematics than what has been introduced so far. We shall take the time honoured
approach of working out everything a finite space with range of x restricted to a
’box’ of size L and taking L to infinity. We shall then give a set of working rules
sufficient for later purposes.
To begin we , therefore, start with Hilbert space of all square integrable functions
over the interval (−L/2, L/2). Further it is required that the functions in this space
satisfy periodic boundary conditions
u(x + L) = u(x) (7.50)

This ensures that the operators x̂, p̂ defined by
∂u(x)
x̂u(x) = xu(x) p̂u(x) = −i~ (7.51)
∂x
are hermitian. Also the position and momentum operators satisfy the required CCR.
We wish to have self adjoint operators x̂ and p̂ satisfying canonical commutation
rule [x̂, p̂] = i~ . It is easy to see that the operator p̂ defined by
d
p̂u(x) = −i~ u(x) (7.52)
dx
satisfies the canonical commutation relation.
Next we seek the eigenvalues and the eigenvectors of momentum operator p̂. For
this purpose we set up and solve the eigenvalue equation p̂u(x) = pu(x) in the x−
representation:
d
−i~ u(x) = pu(x) (7.53)
dx
51

The solutions are given by
up (x) = N exp(ipx/~) (7.54)
Imposing the periodic boundary conditions one get eigenvalues of p̂ as
p = 2πn(~/L), n = 0 ± 1, ±2, (7.55)
The normalisation of eigenvectors is chosen to be

Z L/2
|up (x)|2 dx = 1 (7.56)
−L/2
giving
1
N=√ .
2π~L
and the normalised eigenvectors of p̂ with an eigenvalue pn are therefore
1
upn (x) = √ exp(ipn x/~) (7.57)
2π~L
The eigenvectors are seen to satisfy the completeness relation
X
u∗pn (x)upn (y) = δ(x − y) (7.58)
n
The change of representation formula, relating the coordinate and momentum space
wave functions is Z L/2
ψ̃(pn ) = u∗pn (x)ψ(x)dx (7.59)
−L/2
The inverse relation is

X
ψ(x) = u∗pn (x)ψ̃(pn ). (7.60)
n
Eigenvectors of position
We now seek eigenvectors of position operator x̂. It is not difficult to see that no
eigenvector exists in the conventional sense because if f (x) is an eigenvector of x̂
with eigenvalue x0 , then we must have
x̂f (x) = x0 f (x) (7.61)

(x − x0 )f (x) = 0 (7.62)
This equation tells us that f (x) = 0 for all x 6= x0 . Thus f (x) = 0. The equation
Eq.(7.62) has a formal solution f (x) = δ(x − x0 ) which we shall call ’generalised
eigenvector’ of x̂. All these ’difficulties’ are intimately connected to the fact that x
is an operator having continuous eigenvalues.
52

§5 Wave function as probability density
Let |ψi denote the state vector of a particle and ψ(x) ≡ hx|ψi be the corresponding
wave function. We now come to physical interpretation of the coordinate space wave
function ψ(x). Note that ψ(x) is the coefficient of |xi in the expansion of the state
vector |ψi Z Z
|ψi = |xihx|ψi dx = ψ(x)|xi dx (7.63)
The third postulate tells us that the absolute square of the wave function gives
the probability density for position. Thus the probability of finding
Rb the position
of particle in the range (x, x + dx) is |ψ(x)|2 dx. The integral a |ψ(x)|2 dx is the
probability that position of the particle will be found in the range (a, b).A similar
statement holds for the momentum space wave function ψ(p) e ≡ hp|ψi .
53

Unit 8
Time Dependent Schrodinger

Equation
§1 Conservation of probability
Consider motion of a particle in a potential well V (~x). The time dependent
Schrödinger equation is
∂ψ(~x, t) ~2 ~ 2
i~ =− ∇ ψ + V (~x)ψ(~x, t) (8.1)
∂t 2m
Taking complex conjugate we get (assuming the potential to be real V (~x)∗ = V (~x))
∂ψ ∗ (~x, t) ~2 ~ 2 ∗
−i~ =− ∇ ψ (~x, t) + V (~x)ψ ∗ (~x, t) (8.2)
∂t 2m
If we multiply Eq.(8.1) by ψ ∗ (~x, t) and Eq.(8.2) by ψ(~x, t) and subtract, the potential
terms cancel, and we get
∂ ∂
i~ψ ∗ (~x, t) (~x, t) + i~ψ(~x, t) ψ ∗ (~x, t)
∂t ∂t
2
~ ~ 2 ψ(~x, t) − ψ(~x, t)∇
~ 2 ψ ∗ (~x, t)]
=− [ψ ∗ (~x, t)∇ (8.3)
2m2
Eq.(8.3) can be written as
∂ ~2 ~ ∗ ~ x, t) − ψ(~x, t)∇ψ
~ ∗ (~x, t)]
i~ {ψ ∗ (~x, t)ψ(~x, t)} = − ∇[ψ (~x, t)∇ψ(~ (8.4)
∂t 2m
The right hand side of Eq.(8.4), when expanded gives four terms, two of which cancel
giving the right hand side of Eq.(8.3). We define
ρ = ψ ∗ (~x, t)ψ(~x, t) = |ψ(~x, t)|2 (8.5)

~j = − i~ ψ ∗ (~x, t)∇ψ(~
~ x, t) − ψ(~x, t)∇ψ
~ ∗ (~x, t) (8.6)
2m
With ρ and ~j defined as in Eq.(8.5)-(8.6), Eq.(8.4) can be written as
∂ρ ~ ~j
= −∇. (8.7)
∂t
54

or
∂ρ ~ ~
+ ∇.j = 0 (8.8)
∂t
This equation is called the equation of continuity. In electrodynamics a similar re-
lation holds between charge density (ρ) and current density (~j) and it represents
conservation of total charge. Here, in QM, ρ = |ψ|2 represents probability density
and therefore ~j is called probability
R ∞ current density. The Eq.(8.8) has the conse-
quence that the total probability −∞ |ψ|2 d~x is independent of time. To see this we
integrate Eq.(8.8) over a volume V enclosed by a surface to get
∂ y 3 y
~ ~j dV
ρd x = − ∇.
∂t
V V
x
= − (~j.n̂) dS (8.9)
S
where n̂ is the unit vector perpendicular to the surface. If V represents volume of a

sphere of radius R, S will be the surface of the sphere. The surface area increases
as R2 when R becomes very large. If ~j decreases faster than 1/R2 , as R → ∞, the
right hand side of (9) becomes zero when R → ∞. The left side becomes integral
over all space and we get
∂ y∞
ρ d3 x = 0 (8.10)
∂t −∞
Therefore the norm
y∞
ψ ∗ (x, t)ψ(x, t) d3 x = (ψ, ψ) = kψk2 (8.11)
−∞
is independent of time and Eq.(8.10) represents conservation of total “norm” and is a

consequence of the fact Ĥ is a hermitian operator. While Eq. Eq.(8.10) representing
the conservation of total probability follows from Eq.(8.8); Eq.(8.8) is, in fact, a
stronger equation representing a local conservation law.
∂ρ ~ ~
+ ∇.j = 0
∂t
Electrodynamics Quantum Mechanics

ρ = charge density Probability density
~j = (charge) current density Probability current density
R
V
ρd~x = charge in volume V Probability of finding
particle in volume V .
R
~j.n̂dS current through S; Probability that the particle
S
charge flowing out of S crosses the surface in
per unit time unit time.
The physical interpretation of the probability current density can be seen by

evaluating it for plane waves
ψ(~x) = N exp(i~p · ~r/~)
2
ρ = |N| , ~j = |N|2 ~p/m = |N̄|2~v = ρ~v (8.12)
55

The expression (8.12) for ~j is similar to that for the current density (~j = ρ~v ) in the
electromagnetic theory. This expression for ~j can be understood in classical terms
if we interpret ρ as number density of particles, then the flux of the beam can be
expressed in terms of ~j. To see this let us consider a beam of particles incident on a
small surface of area ∆S, see Fig.4 below. The number of particles crossing ∆S in
time ∆t equals the number of particles in a cylinder of height |~v|∆t, base ∆S, and
having its axis parallel to the velocity ~v of the particles. The volume of the cylinder
is given by ∆S · cos θ(|~v|∆t), where θ is the angle between the velocity ~v and the
normal to the surface ∆S. If the number density of the particles is ρ = |N|2 ,the
number of particles in cylinder = |N|2 ∆S(cos θ)|~v|∆t = ~j · ~n∆S∆t. This is the
number of particles crossing the surface ∆S in time ∆t. Thus the flux of particles,
defined as the number of particles of the incident beam crossing a surface per unit
area per unit time can be written as ~j · ~n
q
DS v
vDt
Fig. 4
In the above we interpreted ρ as the number density of particles in the beam and
concluded that ~j.~n gives the number of particles crossing a unit area in unit time.
For a single particle, ρ has the interpretation of being probability density, and hence
~j will called the probability current density.
§2 Schrodinger equation for a charged particle

The Lagrangian for a particle, having charge q, and moving in electric and magnetic
~ r , t) and scalar potential φ(~r, t) is given by
field described by vector potential A(~
1 q ~ r, t) − qφ(~r, t).
L = m~v 2 + ~v · A(~ (8.13)
2 c
where ~v = d~
r
dt
is the velocity of the particle. You must verify that this is the correct
Lagrangian by showing that it gives the correct equations of motion. The classical
Hamiltonian for a charged particle in a electromagnetic field is easily obtained from
the Lagrangian by first computing the canonical momentum p~ defined by
dL q~ 1h q~ i
p~ = = m~v + A(~ r, t) ⇒ ~v = p~ − A(~r, t) . (8.14)
d~v c m c
The Hamiltonian is then seen to be
d~r
H = ~p · −L (8.15)
dt
1 q~ 2
= p~ − A(~r, t) + qφ(~r, t) (8.16)
2m c
56

p
~ 2
Comparing with the free particle Hamiltonian 2m , we see that the Hamiltonian in
presence of electromanetic field is obtained from the free particle Hamiltonian by
making replacements
q~
~p → ~p − A, H → H + qφ (8.17)
c
The Hamiltonian operator is obtained from (8.17) using the quantisation rule ~p →
−i~∇. This gives the time dependent Schrödinger equation
dψ 1 q ~ 2
i~ = −i~∇ − A ψ + qφ(~r, t)ψ. (8.18)
dt 2m c
Gauge invariance
The electric and magnetic fields remain unchanged under gauge transformation of
the potentials
~→A
A ~′ = A
~ + ∇Λ(~r, t), (8.19)
1 ∂Λ(~r, t)
φ → φ′ = φ − . (8.20)
c ∂t
Under the gauge transformations the change in Lagrangian (8.13) is a total time
derivative and hence the equations of motion remain unchanged. In quantum me-
chanics the Schrodinger equation does not remain invariant under a gauge transfor-
mation. The observable quantities in quantum mechanics remain unchanged if the
wave function transforms as
iq
ψ(~r, t) →= ψ ′ (~r, t) = exp Λ(~r, t) ψ(~r, t). (8.21)
~c
Thus, for example, the averages computed using the potentials A,~ φ and the wave
~ ′ , φ′ and
function ψ will be the same as computed using the transformed potentials A
transformed wave function ψ ′ .
§3 Time Reversal Symmetry

Consider classical motion of a particle under influence of a force field. If at some
instant t0 the direction of the velocity of the particle is reversed, the particle will
retrace its path. For a charged particle in magnetic field, the path will be retraced
if the direction of the magnetic field is also reversed.
To see the time reversal symmetry of the classical equations in another way
consider thought experiment of motion of a particle thrown up in a gravitational
field. Let the motion of the particle be captured on a film while going and on a
second film while coming down. Now let one of the films be run backwards, what
you see cannot not be distinguished from what you see in the second film running
forward. Neglecting effects of air friction, no measurement on the motions seen in
the two films, one running backwards and the other film running forward, will be
able to distinguish between them.
We say that the classical Newton’s laws retain their form under time reversal
t → t′ = −t. This means that one cannot distinguish the motion of a particle
57

in a force field with time reversed motion. What about quantum mechanics? In
quantum mechanics the time dependence is described by the Schrödinger equation
dψ(~r, t) ~2 2
i~ =− ∇ ψ(~r, t) + V (~r)ψ(~r, t). (8.22)
dt 2m
It is easy to see that the form of Schrodinger equation does not change, if we make a
replacement t → t′ and also take the complex conjugate of the Schrödinger equation.
dψ ∗ (~r, t′ ) ~2 2 ∗
i~ ′
= − ∇ ψ (~r, t′ ) + V (~r)ψ ∗ (~r, t′ ). (8.23)
dt 2m
The wave function ψ ∗ (~r, t) describes the time reversed motion. The Schrödinger
equation for a charged particle in presence of magnetic field has time reversal sym-
metry, if the sign of the magnetic field is also reversed. As a simple example, exp(ikx)
represents a particle moving to the right on the x- axis and the complex conjugate
wave function, exp(−ikx), represents a particle moving to the left.
§4 Solution of Time Dependent Schrödnger

Equation
For time independent Hamiltonian the Schrödinger equation can be solved by sep-
aration of variables. We show this for a system of single particle in one dimension.
The Schrödinger equation is
∂ψ(x, t) ~2 d2 ψ(x, t)
i~ =− + V (x)ψ(x, t). (8.24)
∂t 2m dx2
We seek solution for the wave function ψ(x, t) subject to the initial condition that
the wave function at time t = 0 coincides with function φ(x). To achieve separation
of variables we substitute ψ(x, t) = T (t)u(x) in the Schrödinger equation to get

dT (t) ~2 d2 u(x)
.i~u(x) = T (t) − + V (x)u(x) (8.25)
dt 2m dx2

1 dT (t) 1 ~2 d2 u(x)
⇒ i~ = − + V (x)u(x) . (8.26)
T (t) dt u(x) 2m dx2
The left hand side in Eq.(8.26) depends on t only, and the right hand side only on
x. Hence each side must be a constant, say, E. Thus we get two equations
~2 d2 u(x) 1 dT (t)
− + V (x)u(x) = Eu(x), i~ =E (8.27)
2m dx2 T (t) dt
The time equation has the solution
T (t) = Const × exp(−iEt/~), (8.28)
The equation for u(x) is eigenvalue equation for the Hamiltonian. Denoting the
eigenvalues and eigenfunctions by En and un (x), we get the following solutions of
the time dependent Schrödinger equation.
ψn (x, t) = Ce−iEn t/~un (x), . (8.29)
58

where C is a constant. The time dependent Schrödinger equation is a linear equation
and the most general solution is obtained by taking linear combinations of all ψn (x, t)
in Eq.(8.29). Thus we get
X
ψ(x, t) = Cn e−iEn t/~un (x). (8.30)
n
Thus the equation for u(x) is the eigenvalue equation for the Hamiltonian. Denoting
and the functions u(x) are eigenfunctions of the Hamiltonian Now we require the
initial condition that the wave function at time t = 0 reduces to φ(x). Setting t = 0
in Eq.(8.30) gives X
φ(x) = Cn un (x), (8.31)
n
and the constants cn can be determined by making use of the orthogonality of the
energy eigenfunctions of un (x). Assuming un (x) to be normalised, we can find the
coefficients to be Z ∞
Cn = (un , φ) = u∗n (x)φ(x) dx, (8.32)
−∞
Thus the scheme of obtaining the time development of the wave function is as
follows. First we obtain the energy eigenfunctions. Next write the wave function at
time t = 0, φ(x), as a superposition of energy eigenfunctions as in Eq.(8.31). The
coefficients Cn are determined using orthogonality of the energy eigenfunctions as in
Eq.(8.32). The time development is obtained by supplying a factor exp(−iEn t/~)
in term with energy En as in Eq.(8.30).
The eigenfunctions of the Hamiltonian describe stationary states having definite
energy and it is seen from the above discussion that they are also needed to obtain
the time development of the wave function.
The simplest situation of free particle will now be taken up. We shall obtain free
particle energy eigenfunctions and use them to obtain motion of free particle wave
packets.
§5 Free particle solution

The classical Hamiltonian for a free particle is
p2
Hcl = . (8.33)
2m
The corresponding operator Ĥ, in the coordinate representation, is given by
~2 d 2
Ĥ = − (8.34)
2m dx2
end the energy eigenvalue equation reads
~2 d 2 ψ
− = Eψ(x). (8.35)
2m dx2
59

For E > 0 we write the eigenvalue problem as
r
d2 ψ 2mE
+ k 2 ψ(x) = 0, k= (8.36)
dx2 ~2
and the most general solution is given by
ψ(x) = Aeikx + Be−ikx (8.37)
There is no restriction on k hence all positive energies are allowed. For each value
of energy E there are two solutions
ψ(x) = eikx , e−ikx (8.38)
which correspond to momentum eigenvalues ±~k and represent the particle moving
to the right and left respectively.
It must be noted that two linearly independent free particle solutions can also
be written in alternate form as
ψ(x) = sin kx, cos kx (8.39)
While solutions given in Eq.(8.38) are also eigenfunctions of the momentum opera-
tors, the solutions Eq.(8.39) are not.
No solution exists for E < 0 We shall now show that the energy eigenvalue
problem has no solution for E < 0. For E < 0 we define α2 = 2m|E|
~2
and rewrite the
differential equation 8.35 in the form
d2 ψ
− α2 ψ(x) = 0, (8.40)
dx2
which has the most general solution
ψ(x) = Ceαx + De−αx (8.41)
In order that ψ remains finite as x → ∞ we must have A = 0, and similarly,

demanding hat the solution remains finite as x → −∞ gives B = 0. The two
constaints A = 0, B = 0 imply that the wave function must vanish everywhere. Such
a solution does not represent a physical state. Thus we arrive at the conclusion that
E < 0 is not possible. For energy E > 0 the eigenfunctions are not normalisable
and one must use delta function normalisation
Z ∞
ψE1 (x)ψE2 (x) = δ(E1 − E2 ) (8.42)
−∞
if there is a need to fix the normalisation.
60

Free particle in two and three dimensions
The free particle solutions corresponding to energy eigenvalue E are given by
uE (x, y, z) = N exp(ik1 x + ik2 y + ik3 z) (8.43)
where
~2 k 2
E= , k 2 = k12 + k22 + k32 (8.44)
2m
and k1 , k2 , k3 are otherwise arbitrary. If we write ~k = kn̂, where n̂ is a unit vector,
the solution uE (~r) can be written as
uE (x, y, z) = N exp i(~k · ~r) = N exp(ikn̂ · ~r) (8.45)
These energy eigenfunctions are also eigenfunctions of momentum operator with

eigenvalue ~~k. For a fixed energy there are infinite number of solutions, one corre-
sponding to each direction of momentum.
Similar results hold for a free particle in two dimensions. One can impose a delta
function normalisation or a periodic boundary conditions with box normalisation on
the free particle eigenfunction.
§6 Wave Packets
Wave packet
A classical particle has a precise value of position and can have a very precise value
of momentum.On the other hand a wave is not localised in space, it can have a very
precise value of momentum. A quantum particle, due to its dual nature, cannot have
100% very precise values of position and momentum simultaneously.Any attempt to
make position well defined results in large uncertainties in momentum and vice versa.
How are then particle to be represented in quantum mechanics? Consider a particle
with a definite value of momentum p. These are represented by plane waves:
ψ(x) = eipx/~ (8.46)
and the probability density is |ψ(x)|2 = 1 for all positions. Such a quantum particle
has equal probability of being anywhere in space and is, therefore, not localised.
However, if we form a superposition of plane waves over a range of momenta values
and consider the state given by the wave function
Z ∞
Ψ(x) = φ(p)eipx/~dp (8.47)
−∞
it is possible to describe a localised particle. A suitable choice of the function φ(p) is

one which has a maximum around a momentum value p0 and a is small over outside
a range ∆p around p0 . One such function is shown in the Fig.8.1. The Fourier
transform ψ(x) function will be peaked about some point x0 and will be appreciable
for values within a range ∆x around x0 , schematically shown in Fig.8.2. Note that
|ψ(x)|2 , |φ(p)|2 give the probability densities for position and momenta of the quan-
tum particle. Such a superpostion represents a quantum particle localised around
61

position x0 and having an uncertainty in position of equal to ∆x. The value p0 gives
approximately the momentum of the particle. The spreads in position and momenta
will be constrained by the uncertainty relation ∆x∆p ≥ ~/2. Such a superposition
of plane waves is called a wave packet. A wave packet is the closest possible rep-
resentation of a classical particle having definite position and momentum.
Fig. 8.1.
Fig. 8.2.
Several questions need to be answered.
How can we locate the peak x0 ? What happens as time passes? Does the time evo-
lution of wave packet in any way resemble the motion of a classical particle? To
answer these question we take up a very special case of wave packet known free
gaussian wave packet.
Free gaussian wave packet

. A gaussian wave packet is described by the following wave function.

ψ(x) = (2π∆2 )−1/4 exp − (x − x0 )2 /4∆2 + ip0 x/~ . (8.48)
62

For such a wave packet we have
Z ∞
hxi = ψ ∗ (x)xψ(x) dx = x0 , (8.49)
−∞
Z ∞
d
hpi = ψ ∗ (x)(−i~ )ψ(x) dx = p0 , (8.50)
−∞ dx
Z ∞
(∆x)2 = hx2 i − x20 = ψ ∗ (x)x2 ψ(x) dx − x20 = ∆2 , (8.51)
−∞
Z ∞
2 2 2 2 d2
(∆p) = hp i − p0 = (−~ ) ψ ∗ (x) 2 ψ(x) dx − p20 = ~2 /∆2 . (8.52)
−∞ dx
The above wave packet represents a particle localised around x0 and having momen-
tum average p0 . We will now obtain time evolution of the gaussian wave packet and
demonstrate explicitly that the time development of a wave packet closely follows
the motion of a classical particle.
Motion of free particle wave packet

Given the wave packet Eq.(8.48) as wave function at time t=0, and taking the
Hamiltonian as the free particle Hamiltonian, how does the wave packet move?
This can be answered by finding the wave function at time t following the steps
given below.
1. Find the eigenfunctions of the Hamiltonian and for a free particle these are
given by plane waves
ψp (x) = eipx/~. (8.53)
2. Write the wave function at t = 0 as superposition of the energy eigenfunctions:

Z ∞
ψ(x) = φ(p)eipx/~ dp (8.54)
−∞
where the superposition coefficients are given by

Z ∞
1
φ(p) = ψ(x)e−ipx/~ dx (8.55)
2π~ −∞
3. The wave function at time t, Ψ(x, t), is obtained by supplying a factor

exp(−iEp t/~) with the plane wave inside the integral in Eq.(8.54).
Z ∞
Ψ(x, t) = φ(p)eipx/~ e−iEp t/~ dp (8.56)
−∞
In the present case of free particle, Ep = p2 /2m.
4. Computing φ(p) from Eq.(8.55) and substituting in Eq.(8.56) and doing the p
integral gives the wave function at time t.
h n oi
2 −1/4 i~t 1 x2 ip0 x ip20 t
Ψ(x, t) = (2π∆ ) 1+ exp − 1+i~t/2∆2 m
− 4∆2
+ ~
− 2m~
(8.57)
.
2∆2 m
63

The position probability density is given by
" #
~ 2 2
t x2
|Ψ(x, t)|2 = (2π∆2 )−1/4 1 + exp − ~2 t2
. (8.58)
4m2 ∆2 4∆2 (1 + 4m2 ∆2 )
The average position at time t, and the uncertainty in position at time t, (∆x)t are
given by
Z ∞
hxit = Ψ∗ (x, t)xΨ(x, t) dx, (8.59)
−∞
p0 t
= x0 + = x0 + vt. (8.60)
m
1/2
~2 t2
(∆x)t = (∆x)t0 1 + , (8.61)
4∆2 m2
The average position changes with time as expected according to the classical free
particle motion. Also, remembering that the spread in position at time t = 0 is
∆, we see that the uncertainty in position increases with time. Thus, there is a
spreading of the wave packet with time. The spreading takes place because the
wave packet is a superposition of plane waves and different waves have different
velocities.
§7 Propagator
In this section we introduce the propagator for the time dependent Schrödinger
equation. The propagator, to be denoted by K(x, t; x0 , t0 ) is a solution of the time
dependent Schrödinger equation
d
i~ K(x, t; x0 , t0 ) = HK(x, t; x0 , t0 ). (8.62)
dt
and satisfies the initial condition
lim K(x, t; x0 , t0 ) = δ(x − x0 ). (8.63)

t→t0
Given the wave function at time t0 to be φ(x), the wave function at time t, ψ(x, t),
can be written in terms of the propagator as
Z
ψ(x, t) = K(x, t; x0 , t0 )φ(x0 )dx0 . (8.64)
That ψ(x, t) satisfies initial condition
lim ψ(x, t) = φ(x) (8.65)

t→t0
is obvious from Eq.(8.63) and ψ(x, t) is a solution of the time dependent Schrodinger
equation
d
i~ ψ(x, t) = Hψ(x, t). (8.66)
dt
64

is easily seen to follow from Eq.(8.62). To describe in words, the propagator
K(x, t; x0 , t0 ) is seen to coincide with the wave function of a particle having precise
position x0 at time t0 . To obtain an expression for the propagator K(x, t; x0 , t0 ), we
recall that the most general solution of the time dependent Schrödinger equation is
given by X
Cn e−iEn t/~un (x). (8.67)
n
terms of the energy eigenvalues En and eigenfunctions, un (x). Writing the propaga-
tor K(x, t; x0 , t0 ) equal to the expression 8.68 and setting t = t0 we get
X
δ(x − x0 ) = Cn un (x). (8.68)
n
Multiplying by u∗k (x), integrating over x and using orthonormality property of the
energy eigenfunctions we get
Ck = u∗k (x0 ) exp(iEk t0 /~). (8.69)
Substituting Cn in Eq.(8.67), the propagator takes the form
X X
K(x, t; x0 , t0 ) = Cn e−iEn t/~un (x) = e−iEn (t−t0 )/~u∗n (x0 )un (x). (8.70)
n n
Free particle propagator

The free particle propagator, being solution of the time dependent free particle
Schrödinger equation, has the form given by superposition of free particle solutions.
Thus we can write
Z ∞
1
K(x, t; x0 t0 ) = √ C(p)eipx/~e−iEp t/~ dp. (8.71)
2π~ −∞
p2
where Ep = 2m
. The functions C(p) are determined by setting t = t0
Z ∞
1
δ(x − x0 ) = √ C(p)eipx/~e−iEp t0 /~ dp. (8.72)
2π~ −∞
and taking inverse Fourier transform of the above equation. Thus we have
Z ∞
−1/2
C(p) = (2π~) δ(x − x0 )eipx/~e−iEp t0 /~ dx
−∞
−1/2 −ipx0 /~ −iEp t0 /~
= (2π~) e e (8.73)
Therefore, the the expression for the propagator becomes
Z ∞
1
K(x, t; x0 , t0 ) = e−ip(x−x0 )/~e−iEp (t−t0 )/~ dp (8.74)
2π~ −∞
Substituting Ep = p2 /2m and doing a Gaussian integral gives the final answer for
the free particle propagator
1/2
m m(x − x0 )2
K(x, t; x0 , t) = exp − . (8.75)
2πi~(t − t0 ) 2i~(t − t0 )
65

Unit 9
Energy Eigenvalues and

Eigenfunctions in One Dimension
§1 Energy Eigenvalues and Eigenfunctions

Solving the eigenvalue problem for the Hamiltonian gives the allowed values of en-
ergy of a system. Working within the Schrödinger representation, the eigenvalues
and eigenfunctions for free particle, particle in a box, square well and delta func-
tion potentials will be obtained. In the Schrödinger representation the Hamiltonian
operator, for a particle in one dimension, is
~2 d 2
H=− + V (x), (9.1)
2m dx2
and the eigenvalues and the eigenfunctions are obtained by solving the Schrodinger
equation
~2 d2 u(x)
Hu(x) = − + V (x)u(x). (9.2)
2m dx2
Let us talk about the requirements that must be imposed so that the solution may
be acceptable. When the potential is a continuous function of x, the differential
equation Eq.(9.2) demands that u(x) and its first derivative u′(x) must be continu-
ous. The same requirements holds for the potentials, such as square well potential,
which are piecewise continuous with a finite jump at the discontinuity. For more
singular potentials, correct requirements on the solution and its derivative will be
found either from the Schrodinger equation directly, or by regarding the potential as
a limiting case of a suitable potential. I will go through this exercise for the infinite
well and the Dirac delta function potential.
Additional physical requirements on acceptable wave functions follow from the
fact that absolute square, |u(x)|2 , has the interpretation of being the probability
density. For motion in one dimension, in order that the position probability may be
finite the solution must be less singular than |x − a|−1/2 for all x. Also for the bound
state solutions, the probability density, |ψ(x)|2 must approach zero as x → ∞. This
means that the solution u(x) must go to zero faster that x−1/2 for large distances.
An exception will be made for continuous solution which describe scattering from
a potential. A standard discussion of the scattering problems requires use of wave
packets. For those scattering problems when the potential has a short range I will
66

follow a common, though not so rigorous, practice to accept and use solutions which
behave like plane waves at large distances. This is consistent with the physical
picture for scattering that the particle behaves like a free particle at large distances.
We summarise a few important points about the nature of solutions of the
Schrodinger equation in one dimension. Let Vm be the absolute minimum value
of the potential, i.e.,
Vm ≤ V (x), ∀x.
Denoting the asymptotic values of the potential V (x) as x → ±∞ by V± , we will
see that
1. Energy must be greater than the minimum value Vm
2. If the bound states exist, their energies are quantised and must be less than
both V± , in addition to being greater than Vm .
3. For energy greater than V+ , or greater that V− , acceptable solutions can be

found and hence there is no quantisation of these energies.
A detailed discussion of nature and properties of solutions will be taken up in a later

lecture.
To be Included
• Improper integrals and examples on square integrable functions.
• Explanation of why bound state energy is quantised
• What happens for potentials not bounded from below?
§2 Particle in a Box
§ 2.1 Rigid wall
We will first derive boundary condition at a rigid wall. A rigid wall is a boundary
such that the potential on one side of the wall is infinite. We consider a rigid wall
as a limiting case of a potential step given by
(
0, x < 0,
V (x) = (9.3)
V0 x > 0.
when V0 → ∞. Considering energy 0 < E < V0 , the solution of the Schrodinger

equation
~2 d2 u(x)
− + (V (x) − E)u(x) = 0, (9.4)
2m dx2
in the region x < 0 is
2mE
uI (x) = A cos kx + B sin kx, k2 = . (9.5)
~2
67

and for x > 0 is
2m(V0 − E)
uII (x) = C exp(−αx) + D exp(αx), α2 = . (9.6)
~2
We note that the solution must remain finite at infinity, this gives D = 0. Demanding
continuity of the solution u(x) and its derivative at x = 0 gives
uI (0) = uII (0) ⇒ A = C + D, (9.7)
uI (0) = uII (0) ⇒ kB = −α(C − D). (9.8)
Using D = 0 gives the two conditions become
A = C, kB = −αC. (9.9)
When V0 tends to infinity, α → ∞, in order that B is finite, C must go to zero,
giving A = 0 but kB, which is the value of the derivative of uI at x = 0, becomes
an indeterminate product, 0 × ∞.
To summarise, we have shown that at the solution must vanish for x > 0 and that
it must be continuous at x = 0 and that no condition is forced on the derivative of
the solution at x = 0. We leave it for the reader to show that the same conclusions
must hold if the potential is infinite over a finite region of x, instead of extending
to infinity as was the case here (x > 0). The boundary condition was derived here
assuming a constant potential, equal to 0, on one side of the wall. However, the same
boundary conditions apply when the potential assumes ’any’ nonconstant function
of x next to the rigid wall.
§ 2.2 Infinite Well

The energy levels of the infinite well
(
0 0≤x≤L
V (x) =
∞ outside
can be found by taking the solution for square well potential problem and letting
Vo → ∞. This gives boundary condition
ψ(x)|x=o = ψ(x)|x=L = 0
There is no corresponding condition on the derivative of the wave function. The
solution for 0 ≤ x ≤ L is
ψ(x) = Aeikx + Be−ikx
and ψ(x) must vanish outside the region 0 ≤ x ≤ L. Imposing the boundary
conditions one gets
A+B =0 ⇒ B = −A
and AeikL + Be−ikL = 0 . These two conditions imply
eikL = 1 or 2kL = 2nπ (9.10)
nπ
k = (9.11)
L
68

and
~2 k 2 ~2 n2 π 2
E= =
2m 2mL2
and the wave function for the n-th state is
r nπx
2
ψn (x) = sin 0≤x≤L
L L
where n can take all positive integral values. Note that n = 0 is unacceptable
because it implies ψ(x) = 0 for all x.
To Be Written
Discussion of wave function and solution; Correspondence principle Classical limit,
Draw Figures showing first few wave functions.
§3 Square Well
We shall discuss the energy spectrum for a square well potential shown below.
Within the range of the well, it is an attractive, and constant, potential. With
suitable reference for potential energy, the potential can be chosen to be zero inside
the well. It is again a constant outside the range of the potential well. We have
chosen V0 > 0 to denote the value of the potential outside the well. In Fig. 1 we
show a square barrier potential. While the square well has at least one bound state,
the barrier does not admit any bound state. However both the square well, as well
as the square barrier problems have continuous energy solutions which are doubly
degenerate. We shall now discuss the solution for energy levels of a square well
potential in one dimension. The square well potential is given by
V (x)
6
V0
(
0 0≤x≤L
V (x) =
V0 outside I II III
- x
Fig. 1
Since the potential has different expressions for different values of x, the Schrödinger
equation is solved in the three regions (i) x < 0 (ii) 0 ≤ x ≤ L and (iii) x > L
separately. Also the two ranges of energy 0 < E < V0 and E > V0 will be considered
separately.
69

§ 3.1 Bound states solutions
The bound states correspond to 0 < E < V0 . For the bound states one must
insist that ψ(x) → 0 at large distances, because |ψ|2 dx represents probability of
the particle being found between x and x + dx. Thus in the limit x → ±∞, we
must have lim ψ(x) → 0. The solutions will be obtained in the three regions I,II,
and III separately. Besides vanishing of the solution at infinity, we shall impose the
requirement of continuity on the solution for the eigenfunctions and their derivatives.
Region I, x < 0: For x < 0, V (x) = V0 and the Schrödinger equation takes the
form
~2 d 2 ψ
− + (V0 − E)ψ = 0
2m dx2
or
d2 ψ 2m
− 2 (V0 − E)ψ = 0
dx2 ~
2m 2
Denoting ~2 (V0 − E) = α , where α is real, the most general solution for x < 0 will
have the form.
ψI = F eαx + Ge−αx
Region II, 0 < x < L: The Schrödinger equation is
~2 d 2 ψ
− = Eψ
2m dx2
or
d2 ψ
2
+ k2 ψ = 0
dx
2 2
where k = 2mE/~ and most general solution is
ψII (x) = A sin kx + B cos kx
Region III, x > L: When x > L, V (x) = V0 and the Schrödinger equation takes
the form
~2 d 2 ψ
− + (V0 − E)ψ = 0
2m dx2
or
d2 ψ 2m
− 2 (V0 − E)ψ = 0
dx2 ~
The most general solution for x < 0 is
ψIII = Ceαx + De−αx .
Boundary conditions at infinity

For the bound states the wave function must vanish for large distances.
(i) We want that ψI (x) should → 0 as x → −∞.

∴G=0
70

(ii) Also ψIII (x) should → 0 as x→∞
∴ C=0
Continuity Conditions Next we require that the wave function and its derivative
be continuous at x = 0 and x = L.
(i) Continuity conditions for the solution and its derivative at x = 0 give
ψI (x)|x=0 = ψII (x)|x=0 (9.12)
ψI′ (x)|x=0 = ψII
′
(x)|x=0 (9.13)
writing out these and using G = 0 gives
F = B (9.14)
αF = kA (9.15)
which implies
B = kA/α (9.16)
(ii) Continuity conditions for the derivative at x = L give

ψII (x)|x=L = ψIII (x)|x=L (9.17)
′ ′
ψII (x)|x=L = ψIII (x)|x=L (9.18)
These equations imply
A sin kL + B cos kL = De−αL (9.19)
Ak cos kL − kB sin kL = −Dαe−αL (9.20)
Eliminating B in favour of A, using Eq.(9.16), we get two equations for A and

D. These two equations can be written in form of a matrix

sin kL + αk cos kL −e−αL A
2 =0
k cos kL − kα sin kL −αe−αL D
These equations have a nontrivial solution only when the determinant of the
matrix on left hand side is zero. This requirement gives a condition on the allowed
values of energy and can be cast in the forms
k k2
α(sin kL + cos kL) + (k cos kL − sin kL) = 0 (9.21)
α α
(k 2 − α2 ) sin kL − 2kα cos kL = 0 (9.22)
2kα
i .e. tan kL = (9.23)
k2 − α2
Defining tan θ = α/k, it is easy to see that bound state energy eigenvalue must
satisfy
k tan kL/2 = α or k cot kL/2 = −α (9.24)
Energy E appears in the above quantisation condition through k and α and can
be determined graphically.
71

To Be Written
• Graphical solution for bound states
• Discussion of wave functions
• Continuous energy solutions
§4 Delta Function Potential

We shall use the following three methods for obtaining the solutions to the energy
eigenvalue problem for the Dirac δ- function potential
V (x) = −gδ(x), (9.25)
where g is a constant. The three methods are
1. Dirac delta function potential as a limit of square well potential.
2. Solution of the eigenvalue problem by direct integration of Schödinger equa-
tion.
3. Solution of the eigenvalue problem in momentum space.
For E > 0 the eigenvalues will be seen to be continuous and doubly degenerate. The
bound states exist only for E < 0. It will be shown that there is only one bound
state with energy E = −|E| and
mg 2
|E| = . (9.26)
2~2
§ 4.1 δ- function as a limit of a square well

To solve for energy eigenvalues for the δ- function potential problem, we consider it
to be a limiting case of a square well potential of the form
(
0, |x| > a,
V (x) = (9.27)
−V0 , |x| < a.
If the strength of the potential V0 → ∞ and the range a → 0 in such a way that the
area under the potential energy curve remains constant (= −g),
Z
V (x)dx = −2V0 a, (9.28)
we would get
lim V (x) = −gδ(x). (9.29)
The solution of the square well potential problem is known and we have


uI (x) = A sin kx + B cos kx, |x| < a,
u(x) = uII (x) = C exp(−αx), x > a, (9.30)


uIII (x) = D exp(αx), x < a.
72

where
2m(V0 − |E|) 2 2m|E|
k2 = ,α = . (9.31)
~2 ~2
The boundary conditions that the wave function and it derivative must be continuous
at x = ±a give the result that the energy eigenvalue must satisfy one of the following
two conditions.
k tan ka = α, or k cot ka = −α. (9.32)
In order to take the limit V0 → ∞, a → 0 with g = 2V0 a fixed, we substitute
V0 = g/2a and consider the limit a → 0. In this limit, assuming |E| to remain finite,
we can substitute
2m(V0 − |E|) 2mV0 mg
k2 = 2
∼ 2
∼ 2 (9.33)
r ~ ~ ~a
mga
∴ ka ∼ (9.34)
~2
The condition k tan ka = α becomes
r r r
mg mga 2m|E|
2
tan 2
= (9.35)
~a ~ ~2
r r r
mg 1 mg √ 2m|E|
2
√ tan 2
a = . (9.36)
~ a ~a ~2
The result limx→0 tan αx/x = 1/α in Eq.(9.36) gives

r r
1 mg √ mg
lim √ tan 2
a = . (9.37)
a→0 a ~ ~2
Thus we get the bound state energy condition

r
mg 2m|E|
2
= . (9.38)
~ ~2
Hence we have
mg 2
E = −|E| = − . (9.39)
2~2
as the final answer for the energy eigenvalues. It can be shown that there are no
other solutions for the energy.
§ 4.2 Direct Integration of the Schrödinger Equation

We now solve the Schrödinger equation to get the boundary condition on the deriva-
tive of the eigenfunction. We rewrite the Schrödinger equation for the δ- function
potental
~2 d 2 u
− − gδ(x)u(x) = Eu(x) (9.40)
2m dx2
in the form
d2 u 2mg 2mE
2
+ 2 δ(x)u(x) = − 2 u(x). (9.41)
dx ~ ~
73

We want to solve for bound state energy, hence E is negative and we set E = −|E|.
and write it as
d2 u 2mg
+ 2 δ(x)u(x) − α2 u(x) = 0. (9.42)
dx2 ~
r
2m|E|
where α = . Since δ(x) = 0 for x 6= 0, the Eq.(9.42) takes the following
~2
form for x < 0 and x > 0.
d2 u
2
− α2 u(x) = 0 (9.43)
dx
which has two solutions eαx and e−αx . We write the most general solution as
(
u1 (x) = A exp(αx) + B exp(−αx), x < 0
u(x) = (9.44)
u2 (x) = C exp(αx) + D exp(−αx) x > 0.
Taking the boundary condition at ±∞ into account, we get B = 0, C = 0 and the

solution for the eigenfunction becomes
(
u1 (x) = A exp(αx), x<0
u(x) = (9.45)
u2 (x) = D exp(−αx), x > 0.
Demanding that the wave function be continous at x = 0 gives
u1 (0) = u2 (0) =⇒ D = A. (9.46)
Thus, (
u1 (x) = A exp(αx), x<0
u(x) = (9.47)
u2 (x) = A exp(−αx), x > 0
We integrate the Schrödinger equation Eq.(9.42) from −ǫ to ǫ and take the limit
ǫ → 0.
Z ǫ 2 Z Z ǫ
d u 2mg ǫ 2
dx + 2 δ(x)u(x) dx − α u(x) dx = 0 (9.48)
−ǫ dx
2 ~ −ǫ −ǫ
ǫ Z ǫ
du 2mg 2
or + 2 u(0) − α u(x) dx = 0. (9.49)
dx −ǫ ~ −ǫ
The solution u(x) is continuous at x = 0, therefore the last term vanishes in the
limit ǫ → 0. Therefore, we have
du2 du1 2mg
− + 2 u(0) = 0. (9.50)
dx x=ǫ dx −x=ǫ ~
Now using the explicit solution, we have
u(0) = u1 (0) = u2 (0) = A (9.51)

du
|x=−ǫ = Aαe−αǫ (9.52)
dx
du
|x=ǫ = −Aαe−αǫ (9.53)
dx
74

Hence Eq.(9.49) becomes
2mg
−Aαe−αǫ − Aαe−αǫ + A = 0. (9.54)
~2
Taking the limit ǫ → 0 and assuming A 6= 0, we get
2mg m2 g 2 2m|E|
= 2α =⇒ = α2 = (9.55)
~ 2 ~ 4 ~2
mg 2
or |E| = (9.56)
2~2
mg 2
=⇒ E = − 2 . (9.57)
2~
The final form of the energy eigenfunction
u(x) = A exp(−α|x|), (9.58)
is obtained by fixing the constant A using the normalisation condition

Z ∞ Z ∞
2
|u(x)| dx = 1 ⇒ 2 |A|2 exp(−2αx) dx = 1. (9.59)
−∞ 0
Hence r
−1/2 2m|E|
uE (x) = α exp(−α|x|), α= . (9.60)
~2
and the corresponding bound state energy eigenvalue is given by Eq.(9.57).
§ 4.3 Solution in Momentum Space

The Schrödinger equation for the δ- function potential is
~2 d 2 u
− − gδ(x)u(x) = Eu(x). (9.61)
2m dx2
The momentum space eigenfunction ũ(p) is related to the coordinate space eigen-
function by Z ∞
1
ũ(p) = √ e−ipx/~u(x) dx. (9.62)
2π~ −∞
Therefore, we take the Fourier transform of the Schrödinger equation to convert it
into an equation for the momentum space wave function:
Z ∞ Z ∞ Z ∞
~2 2
−ipx/~ d u −ipx/~
− e dx − g e δ(x)u(x) dx = E e−ipx/~u(x) dx. (9.63)
2m −∞ dx2 −∞ −∞
In the first term we integrate by parts twice and use the boundary conditon that
u(x), du
dx
→ 0 as x → ±∞. This gives
p2 √ √
2π~ũ(p) − gu(0) = 2π~E ũ(p). (9.64)
2m
75

Solving for the momentum space eigenfunction ũ(p) gives
2mgu(0)
ũ(p) = (2π~)−1/2 . (9.65)
p2+ 2m|E|
The coordinate space solution can now be computed
Z ∞
1
u(x) = √ eipx/~ũ(p) dp, (9.66)
2π~ −∞
Z
2mgu(0) ∞ exp(ipx/~)
= 2
dp. (9.67)
2π~ −∞ p + 2m|E|
mgu(0)
∴ u(x) = exp(−α|x|). (9.68)
α~2
where use has been made of the result
Z ∞
exp(ipx/~) π 2m|E|
dp = exp(−α|x|)), α2 = . (9.69)
2
−∞ p + 2m|E| ~α ~2
This result can be proved easily by the method of contour integration. Setting x = 0
in Eq.(9.68) and remembering that u(0) cannot vanish, u(0) 6= 0, we get
mg mg 2
= 1 =⇒ |E| = , (9.70)
α~2 2~2
and
u(x) = u(0) exp(−αx) (9.71)
and u(0) is fixed by normalising the eigenfunction to unity.
Z ∞
|u(x)|2 dx = 1. (9.72)
−∞
The final expression for the bound state energy and the eigenfunction turns out to
be
mg 2
E=− 2, uE (x) = α−1/2 exp(−α|x|) (9.73)
2~
with α given in Eq.(9.69).
To Be included
A discussion of wave functions; Uncertainty in x is not always the range but deter-
mined by the bound state energy; Wave function is continuous but not the derivative;
Continuous energy solutions;
§5 Harmonic Oscillator
We shall now outline the steps for deriving energy levels and wave functions for
harmonic oscillator in the coordinate representation. The eigenvalue equation
Hψ = Eψ (9.74)
76

for the harmonic oscillator becomes the following differential equation in coordinate
representation 2 2
−~ d 1 2 2
+ mω q ψ(q) = Eψ(q) (9.75)
2m dq 2 2
The main steps in solution of the eigenvalue problem in coordinate representation
are as follows.
1. In terms of dimensionless variables ξ = αq, λ = 2E/~ω, where α2 = mω/~,

the Schrödinger equation (9.75) becomes
d2 ψ
+ (λ − ξ 2 )ψ = 0. (9.76)
dξ 2
2. It can be seen that for large ξ solutions to the differential equation behave as
2
a polynomial times e±ξ .
3. Define H(ξ) by means of the equation

2 /2
ψ(ξ) = H(ξ)e−ξ (9.77)
then H(ξ) satisfies equation.
H ′′ − 2ξH ′ + (λ − 1)H = 0. (9.78)
4. The above equation is well known Hermite equation and can be solved by the
method of series solution. To solve the Hermite equation we write a series
expansion
H(ξ) = ξ c (ao + a1 ξ + a2 ξ 2 + · · · ). (9.79)
The series Eq.(9.79) is substituted in Eq.(9.78), and coefficient of each power
of ξ coming from the L.H.S. of Eq.(9.78) must be set equal to zero. This gives
value of c
c(c − 1) = 0 ⇒ c = 0, 1, (9.80)
and recurrence relations for the coefficients are
2n + 2c + 1 − λ
an+2 = an . (9.81)
(n + c + 1)(n + c + 2)
5. For c = 0, all the even coefficients are determined in terms of ao and all the
odd coefficients are proportional to a1 , and ao and a1 are arbitrary. Thus one
gets
H(ξ) = a1 y1 (ξ) + a2 y2 (ξ) (9.82)
For c = 1, the solution for H(ξ) is proportional to y2 (ξ) and is already con-
tained in Eq.(9.82). Hence this case, c = 1, need not be considered separately
and we will set c = 0.
Note the Eq.(9.78) is a second order differential equation and the most general
solution is a linear combination of two independent solutions y1 (ξ) and y2 (ξ).
77

6. Next we must explore large ξ behaviour of solutions of Eq.(9.78). The relation
(9.81) for large n takes the form
an+2 2
∼ . (9.83)
an n
which coincides with the ratio of the expansion coefficients, in the series for
exp(ξ 2 )
X ξ 2n
exp(ξ 2) = . (9.84)
n!
Thus the two solutions y1 (ξ) and y2 (ξ) behave like exp(ξ 2 ) for large ξ and
2 /2
ψ(ξ) = H(ξ)eξ (9.85)
ξ2 −ξ 2 /2 ξ 2 /2
ξ→∞ ∼ e ×e =e (9.86)
This behaviour of ψ(ξ) for large ξ makes the solution unacceptable because
ψ(ξ) would not be square integrable.
7. The only way one can get a square integrable solution for ψ(q) is that the
solution H(ξ) must reduce to a polynomial. If H(ξ) is to contain a maximum
power n then we must demand the following conditions.
(i) an+2 = 0
⇒ 2n + 2c + 1 − λ = 0
λ = 2n + 1
and
(ii ) a1 = 0 if n = even
a0 = 0 if n = odd.
8. The condition λ = (2n + 1) is equivalent to the energy quantisation

1
E = (n + )~ω .
2
The wave functions are obtained by using conditions, as in (i) and (ii) above,
and the recurrence relations to solve for the coefficients an . The resulting
solutions for Hn are Hermite polynomials and the normalised eigenfunctions
are given by
1/2
α
ψn (q) = √ n Hn (αq) exp(−α2 q 2 /2)
π 2 n!
These coincide with the wave functions obtained from operator methods
1
n!
(a† )n φ0 (q).
9. First few Hermite polynomials are
H0 (ξ) = 1, H1 (ξ) = 2ξ H2 (ξ) = 4ξ 2 − 2 (9.87)
78

To Be Included
Discussion of properties of wave functions; Uncertainty principle; Correspondence
with classical theory
§6 Periodic Potential
We will discuss properties of periodic potentials. Let V (x) be periodic with period
a:
V (x + a) = V (x). (9.88)
Let u1 (x) and u2 (x) be two linearly independent solutions of the Schrödinger equa-
tion with the periodic potential V (x):
~2 d 2 u
− + V (x)u(x) = Eu(x). (9.89)
2m dx2
Every other solution of the Schrödinger equation will be a linear combination of
the two solutions.u1 (x), u2 (x). Now it is easy to verify that u1 (x + a), u2 (x + a)
are also solutions of the Schrödinger equation. Then u1 (x), u2 (x) must be linear
combinations of u1 (x), u2 (x). Therefore, we write
u1 (x + a) = c11 u1 (x) + c12 u2 (x), (9.90)

u2 (x + a) = c21 u1 (x) + c22 u2 (x). (9.91)
Floquet’s Theorem: There are solutions with the property
ψ(x + a) = λu(x), (9.92)
where λ is a constant. Of course we have
ψ(x + na) = λn u(x). (9.93)
Proof: Let ψ(x) be any solution then we can write it as
ψ(x) = Au1 (x) + Bu2 (x). (9.94)
The requirement ψ(x + a) = λψ(x) leads to
Au1 (x + a) + Bu2 (x + a) = λ(Au1 (x) + Bu2 (x)), (9.95)

or (Ac11 + Bc21 )u1 (x) + (Ac12 + Bc22 )u2 (x) = λ(Au1 (x) + Bu2 (x)) (9.96)
(9.97)
Because the solutions u1 (x), u2 (x) are assumed to be linearly independent, the above
equation holds if and only if coefficients of u1 (x) and u2 (x) are separately equal. This
gives
Ac11 + Bc21 = λA, (9.98)

Ac12 + Bc22 = λB. (9.99)
79

A nontrivial solution for A, B will exist if and only if

c11 − λ c
21 = 0. (9.100)
c12 c22 − λ
This gives a quadratic equation in λ:
(c11 − λ)(c22 − λ) − c12 c21 = 0, (9.101)

2
λ − λ(c11 + c22 ) + c11 c22 − c12 c21 = 0. (9.102)
Let the two roots of this equation be denoted by λ1 , λ2 and corresponding solutions
for A, B will define two functions ψ1 (x), ψ2 (x) from Eq.(9.94). The Wronskian of
the two solutions ψ1 (x), ψ2 (x) defined as
W (x) = ψ1 (x)ψ2′ (x) − ψ1′ (x)ψ2 (x) (9.103)
is known to be independent of x. Hence λ1 λ2 = 1. Now |λ1 | = |λ2 | = 1. It can be

seen by checking that the operator T̂ defined by T̂ ψ(x) = ψ(x + a) is unitary. Hence
its eigenvalue λ obeys |λ| = 1.
We write λ1 = exp(iKa) and λ2 = exp(−iKa) where K is real number such that
π π
− ≤K≤ (9.104)
a a
Thus for all solutions we have
ψ(x + a) = eiKa ψ(x), , (9.105)

ψ(x + na) = eiKna ψ(x). (9.106)
If we define uK (x) by
ψ(x) = eiKa uK (x), (9.107)
we have
uK (x + a) = uK (x). (9.108)
This result is known as Bloch’s theorem. Bloch’s theorem gives the solution for all
x if it is known in one period. Let u1 (x), u2 (x) be any two linearly independent
solutions. Define
ψ(x) = Au1 (x) + Bu2 (x), 0 ≤ x ≤ a. (9.109)
which satisfies
ψ(x + a) = eiKa ψ(x). (9.110)
Define functions
ψ1 (x) = ψ(x), . 0≤x≤a (9.111)

ψ2 (x) = ψ(x), . a ≤ x ≤ 2a (9.112)
iKa
= e ψ(x − a), . a ≤ x ≤ 2a (9.113)
= eiKa {Au1 (x − a) + Bu2 (x − a)} (9.114)
80

which represent the solutions in regions (0, a) and (a, 2a). Demanding continuity of
wave function and its derivative at
ψ1 (x)|x=a = ψ2 (x)|x=a , (9.115)
ψ1′ (x)|x=a = ψ2′ (x)|x=a . (9.116)
These equations give
Au1 (a) + Bu2 (a) = eiKa {Au1 (0) + Bu2 (0)}, (9.117)
Au′1 (a) + Bu′2 (a) = eiKa {Au′1 (0) + Bu′2 (0)}. (9.118)
These equations have a nontrivial solution for A, B if the determinnat of the coeffi-
cients A, B vanishes. This gives

u1 (a) − eiKa u1 (0) u2 (a) − eiKa u2 (0)
′ (9.119)
u1 (a) − eiKa u′1 (0) u′2 (a) − eiKa u′2 (0)
which simplifies to
[u1 (0)u′2(a) + u1 (a)u′2 (0)] − [u2 (0)u′1 (a) + u2 (a)u′1 (0)]
cos Ka = . (9.120)
2(u1 u′2 − u2 u′1 )
This equation gives the condition for existence of solutions for the periodic potential
and can be satisfied only if the right hand side is between −1 and 1. This gives rise
to bands of allowed energies whose limits are obtained by setting the right hand side
to be ±1.
To Be Typed
§7 General Properties of Motion in One Dimen-

sion
1. Energy values are bounded from below:
If the absolute minimum of the potential energy V (x) of a particle be V0 , for
all x V (x) ≥ V0 . The classical particle must have energy greater than V0 and
the motion is possible only in the region where E > V (x). This is seen by
noting that the total energy is a sum of kinetic energy and potential energy
and that the kinetic energy is always positive.
p2
E= + V (x) ≥ V (x) ≥ V0 . (9.121)
2m
The solutions of the Schrödinger equation Hu(x) = Eu(x) exist only for E > 0.
This follows from the facts that the average value of the kinetic energy must
be positive and the average value of the potential energy must be greater than
V0 in every state φ(x). To see this, write
hp̂2 iφ = (φ, p̂2 φ) = (p̂φ, p̂φ) = kp̂φk2 ≥ 0, (9.122)
Z Z
hP.E.iφ = φ∗ (x)V (x)φ(x) dx ≥ V0 φ∗ (x)φ(x) dx = V0 .(9.123)
p2
∴ hEi = hK.E.i + hP.E.i = h i + hV (x)i ≥ V0 . (9.124)
2m
81

2. Minimum bound state energy and the uncertainty principle:
If the potential function has a minimum at xo with a value Vmin . In classical
mechanics, a state with zero momentum, p = 0, and x = xo can exist and
the energy will be Vmin . In QM x and p cannot have sharp values simulta-
neously, and for the lowest bound state the energy will, in general, be greater
than Vmin . The ground state energy can be estimated using the uncertainty
principle. We shall illustrate this by means of the harmonic oscillator.
1
V (x) = mω 2 x2
2
Vmin = 0 classically x = 0, p = 0, E = 0 is a possible state. Quantum
mechanically, the values of x and p will have some uncertainties ∆x and ∆p
which are subject to the uncertainty relation ∆p∆x ≃ ~. Taking the averages
of x2 and p2 of the order of (∆x)2 and (∆p)2 , respectively, and using ∆p ≈ ∆x
~
,
we have
(∆p)2 ~2
< KE > ≈ = (9.125)
2m 2m(∆x)2
1
< V (x) > ≈ mω 2 (∆x)2 (9.126)
2
2
~2 1 1
E ≈ + mω 2 (∆x)2 (9.127)
2m ∆x 2
Minimising E w.r.t. ∆x we get

~2 −2 1
3
+ mω 2 2(∆x) = 0, (9.128)
2m (∆x) 2
2
~ 2 ~
(∆x)4 = × 2 2 ⇒ (∆x)2 = (9.129)
2m m ω mω
2
~ mω 1 ~
E≈ + mω 2 = ~ω (9.130)
2m ~ 2 mω
If we had used ∆p∆x ≥ ~/2, we would have obtained
~ω
Emin =
2
which matches with the exact ground state energy of the harmonic oscillator.
In general this argument can be used to get a quick estimate of the ground
state energy for a given potential.
3. Wave function at large distances:
Let us assume that the potential energy becomes a constant V± as x → ±∞.
Then the energy eigenfunction decays to zero exponentially if the energy is
less than the value of the potential energy at infinity. The solution becomes
oscillatory, like plane waves, if the energy is greater than the value of the po-
tential energy at infinity. This can be easily seen by looking at the Schrodinger
equation at large distances. For example, consider the Schrodinger equation
~2 d 2 u
− + V (x)u(x) = Eu(x) (9.131)
2m dx2
82

For x → −∞ we replace the potential V (x) by its limiting value V− and the
resulting equation
~2 d 2 u
− + V− u(x) = Eu(x), (9.132)
2m dx2
has the solution
2m(V− − E)
u(x) ∼ Aeαx + Be−αx , α2 = (9.133)
~2
for x → −∞. The coefficient B must be set equal to zero, otherwise u(x)
diverges to ∞ as x → −∞. Thus we get
u(x) ∼ Aeαx for x → −∞ (9.134)
In the other case, when E > V− , the asymptotic form of the solution is oscilla-
tory and can be written in several ways. For example, any one of the following
forms
A1 eikx + A2 e−ikx ; B1 cos kx + B2 sin kx; C sin(kx + δ) (9.135)
where k 2 = 2mE/~2 gives the behaviour at infinity. Similar statements obvi-

ously hold for x → +∞.
4. Nature of Solutions:
The most general solution of the Schrodinger equation is a linear combination

of two independent solutions u1 (x), u2 (x) and contains two arbitrary (complex)
constants‘:
u(x) = Au1 (x) + Bu2 (x). (9.136)
of these only a ratio of A, B can be fixed by the boundary conditions.
(a) For the bound states the wave function must go to zero as x → ±∞.
Therefore the corresponding energy must be greater than the minimum
value E > V0 and less than the potential energy at infinity on both sides,
E < V− and E < V+ .
For the bound states the two requirements that the solution must vanish
at ±∞, fix the ratio and the energy value. Thus the energy cannot take
continuous values, assumes only discreet values and is quantised. Later
we will show that in one dimension bound state energy is nondegenerate.
(b) In the situations V− 6= V+ , the solutions for energy in between the two
asymptotic values V± , the solution of the Schrodinger equation will be
oscillatory on one side and exponentially decaying on the other side. The
wave function going to zero at infinity on one side gives a relation be-
tween the two constants A, B and determines their ratio. The solution is
fixed upto an overall constant to be fixed by normalisation. In this case
no restriction is imposed on the energy eigenvalues. Thus the energy
eigenvalues will be continuous and nondegenerate.
83

(c) For the energy values greater than both V− , V+ , there will be no restriction
coming from behaviour at infinity and we will have to linearly indepen-
dent solutions with no condition on energy. Thus energy eigenvalues will
be continuous and doubly degenerate.
You may check validity of the above rules for the potential problems for which
you have seen exact solutions such as square well, harmonic oscillator and
other potentials
5. Non degeneracy of bound state eigenvalues one dimension:

Proof: We shall show that if for a given bound state energy eigenvalue E there
are two eigenfunctions ψ1 and ψ2 , the two solutions must be proportional. Thus
we have
~2 d2 ψ1
− + V ψ1 = Eψ1 (9.137)
2m dx2
~2 d2 ψ2
− + V ψ2 = Eψ2 . (9.138)
2m dx2
Multiply (1) by ψ2 and (2) by ψ1 and subtract to get

~2 d2 ψ1 d2 ψ2
− ψ2 2 − ψ1 2 = 0, (9.139)
2m dx dx

d dψ1 dψ2
or ψ2 − ψ1 = 0. (9.140)
dx dx dx
Integrating we get

dψ1 dψ2
ψ2 − ψ1 = const. , C
dx dx
The constant C can be fixed by evaluating the left hand side at x = ∞. As

x → ±∞, ψ1 → 0, ψ2 → 0 for bound states
... C = 0
Thus we get
dψ1 dψ2
ψ2 − ψ1 = 0 (9.141)
dx dx
1 dψ1 1 dψ2
or − = 0 (9.142)
ψ1 dx ψ2 dx
Integrating we get
ln ψ1 − ln ψ2 = const., K (9.143)
or ln(ψ2 /ψ1 ) = ln K (9.144)
or ψ2 = Kψ1 (9.145)
... ψ1 and ψ2 are linearly dependent. Hence the bound state eigenvalues in one
dimension are non degenerate.
84

It is obvious that the above proof also applies when the solution decays to zero
on one side at large distances. Thus for energies between the two asymptotic
values V− , V+ of the potential we have only one linearly independent solution
and there is no degeneracy.
An exception to this result is particle in twin, (or more) boxes described by
the potential


0 0 ≤ x ≤ L,
V (x) = 0 2L ≤ x ≤ 3L ,


∞, otherwise
For this potential each energy eigenvalue has two linearly independent solu-
tions.
Another exception is provided by von Neumann Wigner type potentials which
have bound states for energy in the continuum. For these potentials the bound
states solutions exist inside the range of energies where we expect only con-
tinuous energy solutions.
6. Parity :
If the potential is an even function of x, i.e., V (−x) = V (x), the parity
operator commutes with the Hamiltonian.
P̂ Ĥ − Ĥ P̂ = 0 (9.146)
If uE (x) is an eigenfunction of energy with eigenvalue E, v(x) = P u(x) =

u(−x) is also an eigenfunction of Hamiltonian with the same eigenvalue E.
This is easily seen by applying Ĥ on v(x).
Ĥv(x) = Ĥ P̂ u(x) = P̂ Ĥu(x) (9.147)

= E P̂ u(x) = Ev(x) (9.148)
Now there are two possibilities.
(a) When the eigenvalue is non-degenerate there is only one linearly inde-
pendent eigenfunction and u(x) and v(x) must be proportional. There
must exist a constant c such that
u(x) = cv(x) (9.149)
Noting the relation v(x) = u(−x), we have
u(x) = cu(−x) (9.150)
Making a replacement x → −x in this equation implies
u(−x) = cu(x). (9.151)
Now Eq.(9.150) and Eq.(9.151) imply that c2 = 1 and hence c = ±1.

This gives u(x) = ±u(−x) and u(x) must be an eigenfunction of parity.
85

(b) In the first case when u(x) and v(x) are linearly independent, v(x) is a
new solution of the eigenvalue problem. This happens if and only if the
energy is degenerate. This is the case for example for a symmetric square
well for positive energies. If form the combinations w1 (x), w2 (x) defined
by
w1 (x) = u(x) + v(x) = u(x) + u(−x) (9.152)

w2 (x) = u(x) − v(x) = u(x) − u(−x) (9.153)
and these will be eigenfunctions of parity.

Similar comments, though differing in details, will apply for any operator
which commutes with Hamiltonian of the system.
7. Tunnelling through a barrier

Consider an example of a particle is initially confined to a box whose walls
can be represented by a potential barrier of finite height V0 . For example,
considering a one dimensional box having walls represented by a potential of
height V0 . Let the potential inside and outside box be zero.
If the energy of the particle is less than barrier height V0 , classically the particle
will always remain confined to the box. Similarly, for the two potentials shown
in Fig. 2 and Fig. 3 bounded motion is possible for a classical particle for
energies between V1 and V2 if the particle is on the left of the maximum at
x = b, it cannot cross the barrier at x=b when E < V2 .
However, in quantum mechanics, particle with energy less than the barrier
height can tunnel through the barrier if the energy is more than the potential
energy on the other side of the barrier and bound states do not exist if the
particle can escape to infinity by tunnelling. Exactly in a similar fashion, a
classical particle, with E < V2 , incident on the barrier from the right (x > b)
cannot reach the region x < b, whereas a quantum particle can. Expressed
differently, a particle can tunnel from one classical region to another even
though they may be separated by a finite potential barrier, giving rise to a
non-classical region.
For the potential of Fig. 2, acceptable solutions exist only for E > V0 , i.e.,
energies greater than absolute minimum value of potential. Even though a
bounded motion is possible classically, if the energy is between V1 and V2 ,
quantum mechanical particle can tunnel through through the barrier and es-
cape to ∞. Considering potential of Fig. 3, we see that solutions of the
Schrodinger equation exist only for E > V1 and tunnelling is not possible if
the energy is less than the value of the potential energy on the right of the
barrier, i.e. E < V0 . Thus if the bound states exist, the energy will be in the
range V0 < E < 0. For E > 0, the solution will be oscillatory as x → ∞ and
energies will be continuous.
This phenomenon is known as barrier penetration or tunnelling. earliest know
example of tunnelling phenomenon is α decay.
86

V2
V2
a b a b
V1 V1
V0
Fig 2 Fig 3
8. Periodic potentials, Energy bands Let V (x) be a periodic potential with period
L
V (x + L) = V (x) .
The energy eigenvalues has bands of allowed energies and forbidden energies
and the energy level diagram is schematically shown in Fig. 4.
E
Allowed
values
Forbidden Band
Allowed
values
Forbidden Band
Allowed
values
Fig 4 Bands in Energy Level Diagram of a Periodic Potential
To Be Written
Oscillation theorem
87

Unit 10
Reflection and Transmission in

One Dimension
§1 Reflection and Transmission Through a Po-

tential Well
The transmission and reflection coefficients for a particle through a potential will
be defined in terms of large distance asymptotic properties of the solutions of the
Schrodinger equation. We shall, therefore, consider the motion of a particle in one
dimensional potential V (x). We assume that the potential is such that it approaches
a constant value V1 as x → −∞ and a constant value V2 as x → +∞.
lim V (x) = V1 , lim V (x) = V2 . (10.1)

x→−∞ x→+∞
If the energy of the particle E is greater than V1 and V2
E > V1 , E > V2 , (10.2)
the motion of the particle is unbounded from −∞ to +∞. The Schrödinger equation
is
~2 d 2
− ψ(x) + V (x)ψ(x) = Eψ(x) . (10.3)
2m dx2
To find solutions for x → −∞, we replace V (x) with V1 in Eq.(10.3) and solve the
resulting equation
~2 d 2 ψ
− + V1 ψ(x) = Eψ(x) . (10.4)
2m dx2
or
d2 ψ 2m(E − V1 )
+ ψ(x) = 0 . (10.5)
dx2 ~2
Defining k1 by r
2m(E − V1 )
k1 = (10.6)
~2
for x → ∞, we get
d2 ψ
2
+ k12 ψ(x) = 0, (10.7)
dx
88

Therefore, we have
ψ(x) → Aeik1 x + Be−ik1 x . (10.8)
for x → −∞. Similarly, as x → ∞ the wave function behaves as
ψ(x) → Ceik2 x + De−ik2 x , (10.9)
where r
2m(E − V2 )
k2 = .
~2
When E > V1 and E > V2 the solution ψ(x) has asymptotic behaviour given in
Eq.(10.8) and (10.9). The energy eigenvalues are doubly degenerate and all the
parameters A, B, C, and D in Eq.(10.8), (10.8) will not be determined completely.
The solution (10.8) is a linear superposition of plane waves travelling in positive and
negative x directions. Similarly (10.9) is also a superposition of waves travelling in
the positive and negative directions.
Now we consider a beam of particles of definite energy incident on a target whose
effect may be represented by a potential V (x) satisfying conditions (1). A part of
the beam will get reflected and part of the beam will be transmitted. If the beam
is incident from the left on the target, we shall have transmitted beam to the right
and part of the beam will be reflected and will travel to the left.
Incident Reflected Transmitted

Beam Beam Beam
Fig. 5
Thus to the left of the target, we have superposition of beams having momenta
in both positive and negative x directions. However to the right of the target, we
have a beam travelling in the positive x direction only. This experimental situation
will be described by a wave function which satisfies the boundary conditions
ψ(x) x→−∞
−→
A exp(ik1 x) + B exp(−ik1 x), (10.10)
ψ(x) x→+∞
−→
C exp(ik2 x), (10.11)
i.e. we should look for solutions of the Schrödinger satisfying Eq.(10.10) and
Eq.(10.11), ( i.e., with D = 0 in Eq.(10.9)) . The physical interpretation of the
coefficients A, B, C in wave function is obtained by computing the current density

~j = ~ ∗ d d ∗
ψ ψ−ψ ψ (10.12)
2im dx dx
for large distances. We have
as x → −∞, j → (|A|2 − |B|2 )~k1/m, (10.13)
as x → +∞, j → |C|2 ~k2 /m. (10.14)
89

In the region x → −∞, we have a superposition of waves travelling to the right
~k1 ~k1
and to the left with fluxes |A|2 and |B|2 respectively. In the region x → ∞
m m
we have a wave travelling to the right with flux |C|2 ~k m
2
. Hence we are led to define
the transmission and reflection coefficients as
Transmitted Flux |C|2 k2
T = = , (10.15)
Incident Flux |A|2 k1
Flux for the reflected beam |B|2 k1 |B|2
R = = = . (10.16)
Incident Flux |A|2 k1 |A|2
Conservation of probability holds for a real, local potential. We have
∂ρ ~ ~
+ ∇.j = 0 (10.17)
∂t
~
2 ~ ∗~ ~ ∗
ρ = |ψ(x, t)| , j = ψ ∇ψ − ψ ∇ψ . (10.18)
2mi
For one dimensional problems, ~j has only one component and

~ ∗ d d ∗
j= ψ ψ−ψ ψ . (10.19)
2im dx dx
As ψ(x, t) corresponds to definite energy E, then it represents a stationary state
and its time dependence is given by
ψ(x, t) = e−iEt/~ψ(x, 0) (10.20)
and hence
ρ = |ψ(x, t)|2 = |ψ(x, 0)|2. (10.21)
∂ρ
Therefore, ρ(x, t) is independent of time and hence = 0. Thus, for a stationary
∂t
state, in one dimension, we have
d
j(x) = 0 → j(x) = const, (10.22)
dx
Thus, j(x) is a constant independent of x, and hence
j(+∞) = j(−∞) (10.23)
Using Eq.(10.13) and Eq.(10.14) we get
~k1 ~k2
|A|2 − |B|2 = (|C|2 ) (10.24)
m m
Divide by ~k m
1
|A|2 to get
2
2 2 k2 C
1 − |B| /|A| = → 1 − R = T. (10.25)
k1 A
Thus, as a consequnce of probability conservation, we get the result that the sum
of reflection and transmission coefficients is unity.
T +R =1 . (10.26)
Thus, as a consequence of conservation of probability, we get the result that the
transmission and reflection coefficients must add to unity.
We shall solve some problems exactly and compute the reflection and transmis-
sion coefficients in the tutorial session.
90

§2 Square Barrier
We wish to compute the reflection and transmission coefficient for a beam with
energy E incident on a on a target represented by a square barrier. The potential
is assumed to be of the form


0, x < 0
V (x) = V0 , 0 ≤ x ≤ L (10.27)


0, x > L
Solve eigenvalue problem

We assume E < V0 and write the solutions in the three regions I, II, III.
ψI (x) = A exp(ikx) + B exp(−ikx) (10.28)

ψII (x) = C exp(αx) + D exp(−αx) (10.29)
ψIII (x) = F exp(ikx) + G exp(−ikx) (10.30)
(10.31)
where r r
2mE 2m(V0 − E)
k= , α = (10.32)
~2 ~2
Now we ask what boundary conditions and matching conditions are to be imposed to
fix the unknown constants. The wave function and its derivative must be continuous
at x = 0 and x = L. Apart from these general requirements, a boundary condition
specific to this problem of finding reflection and transmission coefficients is to be
imposed. Let us assume that a beam of particles is incident from the left. In general,
some particles will be reflected and some particles will be transmitted. Thus we
should expect particles travelling to both left and right in region I, but in region II
only transmitted particles, travelling to the right, are expected in region III. The
solution in region III, Eq.(10.30), should have only exp(ikx) term and hence G = 0.
Setting G = 0 and requiring continuity of the wave function and its derivative at
x = 0 and x = L gives
x=0 A + B = C + D, (10.33)
ik(A − B) = α(C − D) (10.34)
x=L C exp(αL) + D exp(−αL) = F exp(ikL) (10.35)
Cα exp(αL) − Dα exp(−αL) = F ik exp(ikL) (10.36)
Solving these equations for F/A one gets the transmission amplitude given by
F 2iαk
S(E) = = (10.37)
A 2iαk cosh αL + (k 2 − L2 ) sinh αL
and the transmission coefficient is
−1
2 sinh2 αL
T (E) = |S(E)| = 1 + (10.38)
4(E/V0 )(1 − E/V0 )
91

A classical particle with energy E < V0 cannot cross the barrier. However quantum
mechanically there is a nonzero probability that the particle will be transmitted and
we say that particle ‘tunnels through the barrier’. For E ≈ V0 , α ≈ 0 and one has
−1
mV0 L2
T (E) → 1 + (10.39)
2~2
For an opaque barrier αL >> 1 the transmission coefficient is given by
16E(V0 − E)
T (E) ≈ exp(−2αL) (10.40)
V02
§3 Discussion
For a discussion of resonances, relation with bound states and tunnelling see class
notes.
92

Unit 11
Potential Problems in Two and

Three Dimensions
§1 A Rigid Box in Three Dimensions

A rigid rectangular box is represented by zero potential inside the box and infinite
potential outside the box. Taking one of the corners of the box as the origin, a rigid
box in three dimensions, of sides a, b, c, will correspond to the potential
(
0, if 0 < x < a, 0 < y < b, 0 < z < c,
V (x, y, z) = . (11.1)
∞. otherwise.
Since the potential is infinity outside the box, the wave function must be taken to
be zero. The boundary conditions to be satisfied are
ψ(0, y, z) = ψ(a, y, z) = 0 for all y, z, (11.2)
ψ(x, 0, z) = ψ(x, b, z) = 0 for all x, z, (11.3)
ψ(x, y, 0) = ψ(x, y, c) = 0 for all x, y. (11.4)
Recall that there is no condition on the derivative of the wave function, when the
potential jumps by an infinite amount across a boundary. Inside the box, the
Schrödinger equation is free particle equation. Thus, for 0 < x < a, 0 < y <
b, 0 < z < c,
~2 ∂ 2 ψ(~r) ∂ 2 ψ(~r) ∂ 2 ψ(~r)
− + + = Eψ(~r). (11.5)
2m ∂x2 ∂y 2 ∂z 2
The solution must vanish everywhere on the walls of the box. This equation is
separable in several coordinate systems, but the geometry of the problem demands
that we must use the cartesian coordinates. Thus, writing ψ(~r) as a product of three
functions X(x), Y (y) and Z(z) leads to separate, free particle, differential equation
for each of the three functions.
~2 d2 X(x)
− = E1 X(x), (11.6)
2m dx2
~2 d2 Y (y)
− = E2 Y (y), (11.7)
2m dy 2
~2 d2 Z(z)
− = E3 Z(z). (11.8)
2m dz 2
93

where E = E1 + E2 + E3 . The boundary condition (11.2) gives
X(0)Y (y)Z(z) = X(a)Y (y)Z(z) = 0
for all y, z. This condition can be satisfied for all y, z by a nontrivial solution, if and
only if X(0) = X(a) = 0. Similarly, we get Y (0) = Y (b) = 0 and Z(0) = Z(c) = 0.
Thus, the answers for X(x), Y (y) and Z(z) can be written down from the known
solutions of particle in a box problem. A little thinking shows that we must have
r
2 n1 π ~2 k12 ~2 n21 π 2
X(x) = sin k1 x, k1 = , E1 = = ; (11.9)
a a 2m 2ma2
r
2 n2 π ~2 k22 ~2 n22 π 2
Y (y) = sin k2 y, k2 = , E2 = = ; (11.10)
b b 2m 2mb2
r
2 n3 π ~2 k32 ~2 n23 π 2
Z(z) = sin k3 z, k3 = , E3 = = . (11.11)
c c 2m 2mc2
where n1 , n2 , n3 are nonzero positive integers. Hence, the normalised wave function
is given by r
8
ψ(x, y, z) = sin(k1 x) sin(k2 y) sin(k3 z), (11.12)
abc
and the corresponding energy is

~2 π 2 n21 n22 n23
E= + 2 + 2 . (11.13)
2m a2 b c
For a cubical box of side L, setting a = b = c = L, we get
~2 π 2 2
E= (n + n22 + n23 ). (11.14)
2mL2 1
The ground state corresponds to n1 = n2 = n3 = 1 and is nondegenerate. The next
energy level corresponds to (n1 , n2 , n3 ) = (1, 1, 2), (1, 2, 1), (2, 1, 1) and has energy
equal to 4(~2 π 2 /2mL2 ). There are six permutations of choice 1,2,3 for (n1 , n2 , n3 ),
which correspond to the same energy level 14(~2 π 2 /2mL2 ).
§2 Harmonic Oscillator in Three Dimensions

The harmonic oscillator in three dimensions is described by the potential
1 1 1
V (~r) = mω12 x2 + mω22 y 2 + mω32 z 2 . (11.15)
2 2 2
The Schrödinger equation for the harmonic oscillator in three dimensions

~2 ∂ 2 ψ(~r) ∂ 2 ψ(~r) ∂ 2 ψ(~r) 1 2 2 1 2 2 1 2 2
− + + + mω 1 x + mω 2 y + mω 3 z ψ(~r) = Eψ(~r),
2m ∂x2 ∂y 2 ∂z 2 2 2 2
(11.16)
94

can be solved by separating the variables in cartesian coordinates. Substituting
ψ(~r) = X(x)Y (y)Z(z) gives separate harmonic oscillator equations for the three
functions X(x), Y (y), Z(z).
~2 d 2 1
2 2
− + mω1 x X(x) = E1 X(x), (11.17)
2m dx2 2
~2 d 2 1
2 2
− + mω 2 y Y (y) = E2 Y (y), (11.18)
2m dy 2 2
~2 d 2 1
2 2
− + mω 3 z Z(z) = E3 Z(z). (11.19)
2m dz 2 2
Using the known solutions from harmonic oscillator problem, we get the energy as
3
X 1
E= nk + ~ωk . (11.20)
k=1
2
In special case of ω1 = ω2 = ω3 ≡ ω, known as the isotropic oscillator, we get the

wave functions as products of corresponding solutions for x, y, z variables:
ψ(~r) = NHn1 (αx)Hn2 (αy)Hn3 (αz) exp(−α2 (x2 + y 2 + z 2 )/2), (11.21)
where α2 = mω/~ and n1 , n2 , n3 are nonnegative integers. The normalisation con-

stant N is given by
α 1/2 α 1/2 α 1/2
N= √ √ √ . (11.22)
2n1 πn1 ! 2n2 πn2 ! 2n3 πn3 !
The energy eigenvalues are

3
E = n+ ~ω, where n = n1 + n2 + n3 (11.23)
2
For a fixed n, several values of n1 , n2 , n3 give the same energy. For a given n, n3 is
determined in terms of n1 , n2 and equals n − n1 − n2 . The degeneracy of the level
En = (n + 3/2)~ω is, therefore, given by
n
X n
X n
X
1= n2 = (n + 1)(n + 2)/2. (11.24)
n1 =0 n2 =n1 n1 =0
For the isotropic oscillator in three dimensions the potential is 12 mω 2 r 2 and depends
on the variable r alone. This problem can, therefore, be solved by separating the
variables in r, θ, φ coordinates.
95

Unit 12
Spherically Symmetric Problems

in Three Dimensions
§1 Introduction
Features of Classical Motion For a particle in a spherically symmetric potential
• The angular momentum is conserved
• The radial motion is equivalent to motion in one dimension with an effective

potential given by
L2
Veff = V (r) + ,
2µr 2
where L is the angular momentum and the second terms is well known cen-
trifugal term.
• The angular momentum is conserved
• As a consequence of angular momentum conservation, the motion is confined

to a plane.
We will see that for a quantum mechanical system, again the effective potential
appears in the equation for radial motion and that the angular momentum is con-
served. The orbits are not well defined in quantum mechanics, so we ask if we
have a correspondence with the classical theory? Also what are new features in the
quantum motion?
We begin with a derivation of eigenvalues and eigenvectors of the orbital angular
momentum operators L ~ 2 , Lz . The Schrödinger equation for a spherically symmetric
potential problems can be solved by separation of variables spherical polar coor-
dinates. The differential equation for the angular variables θ, φ turns out to be
precisely the eigenvalue problem for the operator L ~ 2 with solution given by spheri-
cal harmonics.
The solution of the radial equation for a free particle, square well, infinite well and
hydrogen atom will be presented and properties of the solutions will be discussed.
96

§2 Coordinate Representation and Angular Mo-
mentum
The orbital angular momentum of a particle is given by L~ = ~r×~p and the components
of the angular momentum operator, in coordinate representation, are

∂ ∂
L̂x = −i~ ŷ − ẑ , (12.1)
∂z ∂y

∂ ∂
L̂y = −i~ ẑ − x̂ , (12.2)
∂x ∂z

∂ ∂
L̂z = −i~ x̂ − ŷ . (12.3)
∂y ∂x
In terms of spherical polar coordinates these expressions take the form

∂ ∂
L̂x = i~ sin φ + cot θ cos φ . (12.4)
∂θ ∂φ

∂ ∂
L̂y = i~ − cos φ + cot θ sin φ , (12.5)
∂θ ∂φ
∂
L̂z = −i~ . (12.6)
∂φ
Also

±iφ ∂ ∂
L± = i~e ∓i + cot θ (12.7)
∂θ ∂φ
~ 2 , given by
The operator L
~ 2 = L̂2 + L̂2 + L̂2 ,
L (12.8)
x y z
takes the form

~ 2 = −~2 1 ∂ ∂ 1 ∂2
L sin θ + . (12.9)
sin θ ∂θ ∂θ sin2 θ ∂φ2
The components of orbital angular momentum satisfy the same commutation rela-
tions as angular momentum,
[Lx , Ly ] = i~Lz ; [Ly , Lz ] = i~Lx ; [Lz , Lx ] = i~Ly .
These commutation relations imply that L ~ 2 commutes with ~n · L̂ for all numerical
~ 2 and a component of
vectors n̂. Hence we can find simultaneous eigenfunctions of L
~ along any direction ~n. Taking n̂ to be along z− axis, the eigenvalue equations
L
~ 2 Y (θ, φ) = λ~2 Y (θ, φ),
L (12.10)
and
Lz Y (θ, φ) = µ~Y (θ, φ), (12.11)
become differential equations

1 ∂ ∂ 1 ∂2
sin θ + Y (θ, φ) + λY (θ, φ) = 0, (12.12)
sin θ ∂θ ∂θ sin2 θ ∂φ2
97

and
∂
−i Y (θ, φ) = µY (θ, φ). (12.13)
∂φ
~ 2 , Lz , so that
Factors of ~2 and ~ have been written explicitly in the eigenvalues of L
λ, µ are dimensionless.
In the next subsection, we shall explain why acceptable solutions exist only for
λ = ℓ(ℓ + 1), µ = m, (12.14)
where ℓ can take only positive integral values 0, 1, 2, · · · and m must satisfy
m = ℓ, ℓ − 1, · · · , −ℓ + 1, −ℓ, (−ℓ < m < ℓ). (12.15)
There are (2ℓ + 1) eigenvalues of Lz for a fixed L ~ 2 and the spherical harmonics
Yℓm (θ, φ) will be seen to be the corresponding eigenfunctions. These results on
~ 2 and Lz will be proved by solving the differential
eigenvalues and eigenfunctions of L
equations by the method of separation of variables.
§ 2.1 Separation of Variables

To solve the differential equations, Eq.(12.12)-(12.13), we substitute
Y (θ, φ) = Q(θ)E(φ) (12.16)
in Eq.(12.12)-(12.13) and divide by Y (θ, φ) = Q(θ)E(φ). This gives
d2 E(φ)
2
+ µ2 E(φ) = 0 (12.17)
dφ
and

1 1 d d 1 1 d2 E(φ)
sin θ Q(θ) + + λ = 0. (12.18)
Q(θ) sin θ dθ dθ sin2 θ E(φ) dφ2
Rearranging this equation we get

2 1 1 d d 2 1 d2 E(φ)
sin θ sin θ Q(θ) + λ sin θ = − . (12.19)
Q(θ) sin θ dθ dθ E(φ) dφ2
While the left hand side of the above equation is a function of θ, the right hand side
is a function of φ alone. Hence each side must be a constant, and the constant, from
Eq.(12.17), is µ. Thus we get

1 d d µ2
sin θ Q(θ) + λQ(θ) − Q(θ) = 0. (12.20)
sin θ dθ dθ sin2 θ
98

Solution of the φ equation
The solution of Eq.(12.17) is
E(φ) = A exp(iµφ) + B exp(−iµφ). (12.21)
A wave function must be single valued function. For a fixed r, θ, φ the values of φ
and φ + 2π correspond to the same point. Hence the solution should have the same
value for φ and φ + 2π. Thus we demand that E(φ) must satisfy
E(φ + 2π) = E(φ), (12.22)
for all φ. This implies that we must have
A exp(iµ(φ + 2π)) + B exp(−iµ(φ + 2π)) = A exp(iµφ) + B exp(−iµφ), (12.23)
or
A exp(iµφ) exp(2πiµ) + B exp(−iµφ) exp(−2πiµ)) = A exp(iµφ) + B exp(−iµφ).

(12.24)
For µ 6= 0, the linear independence of the exp(±iµφ) imples that the corresponding
coefficients must be equal separately giving
exp(±2πiµ) = 1 ⇒ µ = m, (12.25)
where m is an integer and two linearly independent solutions of the φ equation are
E(φ) = exp(imφ), m = ±1, ±2, · · · . (12.26)
If µ is zero, the most general solution of the φ equation is
E(φ) = A + Bφ, (12.27)
and the requirement of single valuedness gives B = 0. Thus, the for the eigenvectors
of Lz can be written as
E(φ) = exp(imφ), m = 0, ±1, ±2, · · · . (12.28)
Solution of the θ equation

If we substitute w = cos θ in Eq.(12.20), it takes the form

d 2 dP (w) m2
(1 − w ) + λ− P (w) = 0, (12.29)
dw dw 1 − w2
where we have introduced P (w) ≡ Q(cos θ) and have used
dP (w) dP (w) dw dP (w)

= · = − sin θ . (12.30)
dθ dw dθ dw
This Eq.(12.29), known as associate Legendre equation, can be solved by the method
of series solution. Since Eq.(12.29) is a second order differential equation, there are
99

two linearly independent solutions of this equation. For general values of λ both
the solutions become infinite at w = ±1 corresponding to θ = 0, π. Therefore these
solutions are unacceptable. For special values λ = ℓ(ℓ + 1), where ℓ is a positive
integer, and with ℓ ≥ |m|, one solution remains finite, but not the other solution.
Thus we choose
λ = ℓ(ℓ + 1), ℓ ≥ |m|. (12.31)
For the above choice, the nonsingular solution for P (w) is known as the associated
Legendre function and has the form
d|m|
Pmℓ (w) = (1 − w 2 )|m|/2 Pℓ (w), (12.32)
dw |m|
~ 2 and
where Pℓ (w) is Legendre polynomial of degree ℓ. Thus the eigenfunctions of L
Lz are the
Yℓm (θ, φ) = NPmℓ (θ)eimφ , m = ℓ, ℓ − 1, · · · , ℓ. (12.33)
The normalisation is fixed by demanding

Z 2π Z π
∗
dφ Yℓm (θ, φ)Yℓm(θ, φ) sin θ dθ = 1. (12.34)
0 0
The functions Yℓm (θ, φ), in Eq.(12.33), are known as spherical harmonics.
§ 2.2 Remarks
~ obeying angular momentum algebra can be obtained
The eigenvalues of operators, J,
by making use of commutation relations alone. (See Lecture Unit on Canonical
Quantisation). This gives rise to integral as well as half integral values for the total
angular momentum. Here the half integral values for the orbital angular momentum
case are not allowed due to an extra requirement of single valuedness of the wave
functions.
The action of operators J± relates the states |jmi with |j, (m ± 1)i by
p
J± |jmi = j(j + 1) − m(m ± 1)|j, (m ± 1)i. (12.35)
The corresponding relation for the orbital angular momentum assumes the form
p
L± Yℓm (θ, φ) = ℓ(ℓ + 1) − m(m ± 1)Yℓ(m±1) (θ, φ). (12.36)
Similarly, L± acting on the wave function Yℓ,±ℓ (θ, φ) corresponding to the highest
(lowest) m value, m = ±ℓ, will result in zero:
L+ Yℓ,ℓ (θ, φ) = 0, L− Yℓ,−ℓ (θ, φ) = 0.. (12.37)
These equations, written as differential equations, can be solved to get Yℓ,±ℓ (θ, φ)
and the raising and lowering action of the differential operators L± , as in Eq.(12.37),
can be used obtain expressions for Yℓm (θ, φ) for other values of m.
100

§3 Spherically Symmetric Potentials
We shall now discuss energy eigenvalue problem in three dimensions for a spherically
symmetric potentials. A spherically symmetric potential depends only on r and does
not depend on θ and φ. The Hamiltonian for such a system is
~p2
H= + V (r) (12.38)
2m
For a spherically symmetric potential the Hamiltonian commutes with the angular
momentum operators L ~ = ~r × p~ and the angular momentum components Lx , Ly , Lz
are constants of motion and therefore H, L ~ 2 , Lz form a commuting set of operators.
It is seen that the parity operators P commutes with all these operators and that
the set of operators
~ 2 , Lz and P~
Ĥ, L
is a complete set of commuting operators. This means that L ~ 2 , Lz , P are constants
of motion and that the energy eigenfunctions can be selected to have definite values
~ 2 , Lz , P also. We shall see these features in the following specific examples to
of L
be discussed later.
• Free Particle, V (r) = constant.

e2
• Hydrogen atom, v(r) = −
r
• Square well and other similar potentials.
Schrodinger Equation for Spherically Symmetric Potentials

The Schrodinger equation for a spherically symmetric potential is

~2 2 ~ = Eu(~r).
− ∇ + V (r) ψ(r) (12.39)
2m
The Laplacian ∇2 in spherical polar coordinates is given by

2 1 ∂ 2 ∂ 1 ∂ ∂ 1 ∂2
∇ = 2 r + 2 sin θ + 2 2 . (12.40)
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2
Therefore, Eq.(12.39) takes the form

1 ∂ 2∂ 1 ∂ ∂ 1 ∂2 2m
2
r + 2 sin θ + 2 2 2
ψ(r, θ, φ)+ 2 (E−V (r))u(r, θ, φ) = 0.
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ ~
(12.41)
Separation of Variables
Substitute
ψ(r, θ, φ) = R(r)Y (θ, φ) (12.42)
101

in Eq.(12.41) and divide by R(r)Y (θ, φ) to get

1 1 d 2 ∂R 1 1 d dY 1 1 d2 Y 2m
r + sin θ + + (E−V (r)) = 0.
R(r) r 2 dr ∂r Y r 2 sin θ dθ dθ Y r 2 sin2 θ dφ2 ~2
(12.43)
2
Multiply by r and rearrange to get

1 d 2 dR 2m 2 1 1 d dY 1 d2 Y
r + 2 (E−V (r))r = − sin θ + = 0.
R(r) dr dr ~ Y sin θ dθ dθ sin2 θ dφ2
(12.44)
The left hand side of the above equation is a function of r alone and the right hand
side is a function of θ and φ only. This is possible only when each side is a constant,
say λ. Thus we get two ordinary differential equations

1 d 2 dR 2m
r + 2 (E − V (r))r 2 = λ, (12.45)
R(r) dr dr ~
and
1 1 d dY 1 d2 Y
sin θ + = −λ. (12.46)
Y sin θ dθ dθ sin2 θ dφ2
On rearranging Eq.(12.45), we get the radial Schrodinger equation

1 d 2 dR 2m λ~2
r + 2 E − V (r) − R(r) = 0. (12.47)
r 2 dr dr ~ 2mr 2
Eq.(12.44), rewritten as

1 d dY 1 d2 Y
− sin θ + = λY (θ, φ), (12.48)
sin θ dθ dθ sin2 θ dφ2
is seen to be just the eigenvalue problem for square of the angular momentum
~ 2.
operator, L
The variables θ and φ can be separated in Eq.(12.48) by writing
Y (θ, φ) = Q(θ)E(φ)
the resulting partial differential equation

1 1 d dQ 1 1 d2 E
sin θ + =λ (12.49)
Q sin θ dθ dθ E sin2 θ dφ2
separates into two ordinary differential equations, one of which is just the eigenvalue
equation for Lz . These equations have already been discussed analysed earlier and
physically acceptable solutions are obtained when λ = ℓ(ℓ+1), m = ℓ, ℓ−1, · · · , −ℓ−
1, −ℓ. The solutions for Y are the spherical harmonics Yℓm .
§ 3.1 Summary of Results on Spherically Symmetric Poten-

tials
1. The solutions of the Schrodinger equation

~2 2
− ∇ + V (r) ψ = Eψ (12.50)
2m
102

for a spherically symmetric potential V (r) are of the form
ψ(r, θ, φ) = Rℓ (r)Yℓm(θ, φ) (12.51)
2. Noting that |Yℓℓ (θ, φ)|2 = | sin θ|2ℓ , we see that for m = ℓ and for large ℓ, the
probability of particle being outside the plane θ = π/2 becomes very small.
This result corresponds to the classical result that the orbit lies in a plane for
spherically symmetric potentials.
3. The angular part of the wave function, Yℓm (θ, φ), is simultaneous eigenfunction
~ 2 , Lz and with eigenvalues ℓ(ℓ + 1)~2 and m~, respectively. Note that only
of L
ℓ appears in the radial equation and that it does not contain m. Hence the
energy eigenvalues are independent of m; there are (2ℓ+1) linearly independent
solutions, ψEℓm (r, θ, φ), for each fixed ℓ, all having the same energy. Thus they
are (2ℓ + 1) fold degenerate.
4. The energy eigenfunctions are also eigenfunctions of parity operator. Under

parity ~r → −~r. In terms of polar variables this implies that under parity we
have
r → r, θ → π − θ, φ → π + φ. (12.52)
Thus the action of parity on the wave function (12.51) can be seen to be given
P Yℓm (θ, φ) = (−1)ℓ Yℓm (θ, φ). (12.53)
Therefore, the energy eigenfunctions have a definite parity (−1)ℓ .
5. The radial part of the wave function Rℓ (r) is called the radial wave function
and satisfies the radial Schrodinger equation

1 d 2 dR 2m ℓ(ℓ + 1)~2
r + 2 E − V (r) − R(r) = 0 (12.54)
r 2 dr dr ~ 2mr 2
1
6. If we substitute R(r) = χ(r), the radial equation takes the form of one
r
dimensional Schrodinger equation. Using
dR(r) 1 1 dχ(r)
= − 2 χ(r) + (12.55)
dr r r dr
dR(r) dχ(r)
r2 = −χ(r) + r (12.56)
dr dr
1 d 2 dR 1 dχ d2χ dχ
r = 2 − +r 2 + (12.57)
r 2 dr dr r dr dr dr
2
1d χ
= (12.58)
r dr 2
Eq.(12.54) takes the form

~2 d 2 χ ℓ(ℓ + 1)~2
− + V (r) + −E χ=0 (12.59)
2m dr 2 2mr 2
103

This equation looks like one dimensional Schrodinger equation with potential
V (r) replaced with
ℓ(ℓ + 1)~2
V (r) + ≡ Veff (r) (12.60)
2mr 2
The second term in Veff (r) is just the centrifugal potential term which also
appears in the classical equation for the radial motion. The radial Schrodinger
equation for χ, Eq.(12.59), can be analysed in the same manner as one dimen-
sional problems. However, there is one difference that we must demand
χ(r) → 0 as r → 0, (12.61)
χ(r)
so that the radial wave function R(r) = does not become singular at r =
r
0. In addition to above boundary condition on the solutions, another difference
between Eq.(12.59) and a one dimensional problem is that the variable r takes
values in the interval (0, ∞) instead of (−∞, ∞).
§4 Solution of Radial Equation for a Constant

Potential
The solutions of the radial equation for a constant potential are known in terms of
Bessel functions. We shall list these solutions and discuss their properties before
taking specific examples such as free particle, square well potential. Let us assume
V (r) = V0 for some range of values of r. Then for this range of values the radial
equation takes the form

1 d 2 dR 2m ℓ(ℓ + 1)~2
r + 2 E − V0 − R(r) = 0. (12.62)
r 2 dr dr ~ 2mr 2
We shall consider the cases E − V0 > 0 and E − V0 < 0 separately.
§ 4.0.1 CASE I: E − V0 > 0

We define k 2 = 2m(E−V
~2
0)
and the two linearly independent solutions of the resulting
radial equation, given by jℓ (kr) and nℓ (kr), are known as spherical Bessel functions.
These are related to the Bessel functions Jν (kr) as follows
π 1/2
jℓ (kr) = Jℓ+ 1 (kr), (12.63)
2kr 2
π 1/2
nℓ (kr) = (−1)ℓ+1 J−ℓ− 1 (kr). (12.64)
2kr 2
The most general solution of the radial equation is a linear combination of the above
solutions:
R(r) = Ajℓ (kr) + Bnℓ (kr). (12.65)
We need to know the behaviour of the solutions for r ≈ 0 and for r → ∞.
104

Small r : The solution jℓ (ρ) goes to zero, but nℓ (ρ) is singular for ρ ≈ 0. In fact
as ρ → 0, we have
ρℓ
jℓ (ρ) → , (12.66)
(2ℓ + 1)!!
nℓ (ρ) → (2ℓ − 1)!!ρ−ℓ−1 . (12.67)
Here (2ℓ + 1)!! means (2ℓ + 1)(2ℓ − 1) · · · 1.
Large r : For large ρ both jℓ (ρ) and nℓ (ρ) are oscillatory. As ρ → ∞
1
jℓ (ρ) → cos(ρ − (ℓ + 1)π/2), (12.68)
ρ
1
nℓ (ρ) → sin(ρ − (ℓ + 1)π/2). (12.69)
ρ
Thus for E > V0 , both jℓ (ρ) and nℓ (ρ) are acceptable solutions as ρ → ∞.
§ 4.1 CASE II : E − V0 < 0

In this case we define
2m(E − V0 )
2
= −α2 , α = real. (12.70)
~
In this case two linearly independent solutions are jℓ (iαr) and nℓ (iαr) and the most
general solution is
R(r) = Ajℓ (iαr) + Bnℓ (iαr). (12.71)
Again, nℓ (iαr) has unacceptable singular behaviour at r = 0. To discuss large r
behaviour, introduce Hankel functions of first and second kinds by
(1)
hℓ (ρ) = jℓ (ρ) + inℓ (ρ), (12.72)
(2)
hℓ (ρ) = jℓ (ρ) − inℓ (ρ). (12.73)
Then, as ρ → ∞, we have
(1) 1
hℓ (iρ) → − exp(−ρ) (12.74)
ρ
(2)
and hℓ (iρ) blows up, becomes infinite, as ρ → ∞ and is unacceptable. It may be
remarked that the solutions jℓ (ρ) and nℓ (ρ) are linear combinations of cos ρ, sin ρ
(1,2)
multiplied by powers of ρ. Similarly, hℓ (ρ) are exponentials multiplied by terms
containing powers of ρ.
Table : Form of Acceptable Solutions of Radial equation
Near r = 0 For large r r in finite interval (a, b)
E − V0 > 0 jℓ (kr) Ajℓ (kr) + Bnℓ (kr) Ajℓ (kr) + Bnℓ (kr)
E − V0 < 0 jℓ (iαr) jℓ (iαr) + inℓ (iαr) Ajℓ (iαr) + Bnℓ (iαr)
105

§ 4.2 Some Spherical Bessel functions
We shall now tabulate first few spherical Bessel functions. Many books use a different
(1,2)
convention for nℓ and Hankel functions hℓ ρ.
sin ρ
j0 (ρ) = (12.75)
ρ
sin ρ cos ρ
j1 (ρ) = − (12.76)
ρ2 ρ

3 1 3
j2 (ρ) = 3
− sin ρ − 2 cos ρ (12.77)
ρ ρ ρ
cos ρ
n0 (ρ) = − (12.78)
ρ
cos ρ sin ρ
n1 (ρ) = − 2 − (12.79)
ρ ρ

3 1 3
n2 (ρ) = − − cos ρ − sin ρ (12.80)
ρ3 ρ ρ2
(1) 1
h0 (iρ) = − exp(−ρ) (12.81)
ρ

(1) 1 1
h1 (iρ) = i + exp(−ρ) (12.82)
ρ ρ2

(1) 1 3 3
h2 (iρ) = + + exp(−ρ) (12.83)
ρ ρ2 ρ3
(2)
The functions hℓ (ρ) have exp(ρ) as a factor which is bad for large ρ; only h(1) (iρ) is
useful when one needs a solution valid for large ρ. When one needs a solution valid
for intermediate values of ρ, one can take a linear combination of jℓ (ρ) and nℓ (ρ).
§5 Radial Equation for a Free Particle

The radial equation for a free particle, V (r) = 0, for all r is

1 d 2 dR 2 ℓ(ℓ + 1)
r + k − R=0 (12.84)
r 2 dr dr r2
2mE
where k 2 = . The solution of the radial equation has the most general form
~2
R(r) = Ajℓ (kr) + Bnℓ (kr). (12.85)
Here we must set B = 0, because nℓ (kr) → ∞ as r → 0. Hence we get
Rℓ (r) = Ajℓ (kr), (12.86)
and the full free particle wave function is
Ψ(r, θ, φ) = Njℓ (kr)Yℓm(θ, φ). (12.87)
The above solution is a simultaneous eigenfunction of the free particle Hamiltonian,
~ 2 , Lz and parity operators. There is no restriction on the energy E except that it
L
must be positive.
106

Degeneracy: For a given value of energy E, the radial equation will have solutions
for all ℓ all values 0, 1, 2, . . . and for each ℓ, m has (2ℓ + 1) values from −ℓ to ℓ.
Therefore, for every energy value E > 0, there are infinite number of solutions
and the each energy eigenvalue is infinitely degenerate. These different solutions
correspond to spherical waves with definite angular momentum. It is apparent that
this result will hold for continuous energy solutions of every spherically symmetric
potential, not just free particle.
If we take linear combinations of solutions with fixed energy E we get most
general form of the solution for a given energy as
∞ X
X ℓ
Φ(~r) = Cℓm jℓ (kr)Yℓm (θ, φ) (12.88)
ℓ=0 m=−ℓ
In cartesain coordinates the free particle solutions for energy E are plane waves
exp(i~k ·~r). Thus, it is possible to write each of these two type of solutions as a linear
combination of the other type. In particular we have
X
exp(i~k · ~r) = Cℓm jℓ (kr)Yℓm(θ, φ). (12.89)
In a particular case of this equation, when k is along the z- axis, ~k · ~r = kz and we

have the expansion of plane waves
∞
X
exp(ikz) = (2ℓ + 1)iℓ jℓ (kr)Pℓ (cos θ) (12.90)
0
in terms of spherical waves. Note that only m = 0 terms contribute in the above
equation.
§6 Particle in a Rigid Spherical Box

The potential for a rigid spherical box can be written as
(
0, 0 < r < a,
V (r) = (12.91)
∞, r > a.
The full wave function will have the form
ψ(r, θ, φ) = R(r)Yℓm(θ, φ) (12.92)
We need to consider the solutions of the radial equation only. No solution can be
found for E < 0, therefore we consider E > 0. For r > a, the potential is infinite
and hence the radial wave function must be zero. Next we consider r < a, where
the potential is zero. Hence a general solution of the radial equation for r < a is
2mE
Rℓ (r) = Ajℓ (kr) + Bnℓ (kr), k2 = (12.93)
~2
Recall that near r = 0, nℓ (ρ) ∼ ρ−ℓ−1 and blows up as ρ → 0. Hence, in order that
the solution remains finite at r = 0, one must set B = 0. Thus we get
107

(
Ajℓ (kr), 0 < r < a,
Rℓ (r) = (12.94)
0, r > a.
Next, we must demand that the radial wave function be continuous at r = a. Recall
that there is no corresponding requirement on the derivative for this case of infinite
jump in the potential at r = a. The continuity requirement on Rℓ (r) becomes
jℓ (ka) = 0. (12.95)
The solutions of the above equation determine allowed values of k and hence the
2 2
allowed bound state energies E = ~2mk .
Energy levels and degeneracy

To get all the solutions, one proceeds as follows. First set ℓ = 0 and locate the
roots of j0 (ka) = 0, if we call them ρ0n , n = 01, 2, . . . etc. the corresponding energies
~2 ρ2
are given by E = 2ma0n2 . Here n denotes the nodes of the radial wave function for
ℓ = 0. Next, set ℓ = 1 and again find the roots of j1 (ka) = 0, calling the roots as
~2 ρ2
ρ1n , n = 01, 2, . . . etc., we get the ℓ = 1 energy levels E = 2ma1n2 . This process is to
be repeated for all values of the angular momentum ℓ. For each ℓ the number of
bound states turns out to be infinite for the present case. So the energy levels can be
characterised the radial quantum number n and azimuthal quantum number ℓ. For
each energy level Enℓ there are (2ℓ + 1) wave functions. Thus the degeneracy of the
levels is (2ℓ + 1). The energy increases with increase in ℓ and also with increasing
n. Thus the energy level diagram would appear as follows.
—————
—————
—————
n′ = 3
′
n =3
′
n = 3 —————
n′ = 2
′
n =2
′
n = 2 —————
n′ = 1
n′ = 1
n′ = 1 —————
n′ = 0
n′ = 0
′
n = 0 —————
l=0 l=1 l=2
nondegenerate m = −1, 0, 1 m = −2, −1, 0, 1, 2
3 fold degenerate 5 fold degenerate
108

§7 Square Well
The potential for a three dimensional spherically symmetric square well is
(
−V0 , r < a,
V (r) = (12.96)
0, r > a.
Bound states
The bound states will be found for E < 0. The radial Schrödinger equation becomes

1 d 2 dR 2 ℓ(ℓ + 1)
r + q − R = 0, r < a. (12.97)
r 2 dr dr r2

1 d 2 dR 2 ℓ(ℓ + 1)
r + −α − R = 0, r > a. (12.98)
r 2 dr dr r2
where
2m(E + V0 ) 2m|E|
q2 = 2
, α2 = . (12.99)
~ ~2
A most general form of the solution for the radial equation is given by
(
Ajℓ (qr) + Bnℓ (qr), r<a
R(r) = (1) (2) (12.100)
Chℓ (iαr) + Dhℓ (iαr), r > a
(2)
We note that nℓ (qr) diverges as r → 0, hence we must set B = 0. Also hℓ (iαr)
blows up at ∞ so we must have D = 0. This gives The radial wave function to be
(
Ajℓ (qr), r<a
R(r) = (1) . (12.101)
Chℓ (iαr), r > a
Next the requirement of continuity of the solution and its derivative at r = a gives
the following two constraints on the unknown constants A, C.
(1)
Ajℓ (qa) = Chℓ (iαa) (12.102)

dhℓ (iαr)
(1)
djℓ (qr)
A = C (12.103)
dr r=a dr
r=a
Eliminating A and C from the above two equations we get the bound state condition
to be
dhℓ (iαr)
(1)
1 djℓ (qr) 1
= (1) . (12.104)
jℓ (qa) dr r=a h (iαa)
ℓ
dr
r=a
This equation can be solved numerically to get the energy eigenvalues. In the case
of square well only a finite number of bound states exist for each ℓ. In general, will
be no bound state solution for ℓ greater than a certain value.
The lowest energy state corresponds to ℓ = 0. For ℓ = 0 this equation assumes
a simpler form, using j0 (qr) = sinrqr h′0 (iαr) = exp(−αr)
r
, we get
q cot qa = −α. (12.105)
109

The equation Eq.(12.103) has already appeared for the one dimensional square well
problem and has no bound state solution unless the potential is sufficiently attrac-
tive, i.e., V0 must be greater than a minimum value. This is a consequence of
the requirement that radial wave function must be finite for r = 0. This should
be contrasted with the result in one dimension, where one had another condition,
q tan qa = α, for bound states and which always had at least one solution.
Continuous energy states

The energy levels for E > 0 are continuous. We shall write corresponding solutions
which are important for discussion of scattering form a square well. When E > 0,
we define
2m(E + V0 ) 2mE
q2 = 2
, k2 = . (12.106)
~ ~2
A most general form of the solution for the radial equation is given by
(
Ajℓ (qr) + Bnℓ (qr), r < a,
Rℓ (r) = (12.107)
Cjℓ (kr) + Dnℓ (kr), r > a.
In order that the radial wave function be finite for r = 0, we must set B = 0. This
gives (
Ajℓ (qr) r<a
Rℓ (r) = (12.108)
Cjℓ (kr) + Dnℓ (kr), r > a
Next, we demand the continuity of the radial wave function and its derivative at
r = a and get
Ajℓ (qa) = Cjℓ (ka) + Dnℓ (ka) (12.109)

djℓ (qr) djℓ (kr) dnℓ (kr)
A = C +D (12.110)
dr r=a dr r=a dr r=a
While the overall constant corresponds to normalisation of the solution, the above
equations determine the two ratios of the constants A, B, C and gives a solution for
the radial wave function. For every energy E, (E > 0), the radial equation will have
a solution for every pair of ℓ, m values giving rise to infinite degeneracy. The con-
tinuous energy solutions will have physical applications in scattering problems.
§8 Hydrogen Atom
The classical Hamiltonian for an electron and a nucleus of charge Z is
~p21 p~2 Ze2

H= + 2 − (12.111)
2m1 2m2 |~r1 − ~r2 |
The case Z = 1 corresponds to H- atom and Z = 2 is the case of a singly ionised

He atom. Here ~r1 = (x1 , y1 , z1 ), ~r2 = (x2 , y2 , z2 ), denote the position vectors of
110

the nucleus and the electron respectively, and m1 and m2 are their masses. The
Schrodinger equation for the an electron and a nucleus of charge Z is
2 2
~2 d d2 d2 ~2 d d2 d2 Ze2
− + + Ψ − + + Ψ − Ψ = EΨ
2m1 dx21 dy12 dz12 2m2 dx22 dy22 dz22 |~r1 − ~r2 |
(12.112)
Since the potential depends on the relative position, in classical mechanics, the two
body problem can be reduced to a one body problem by changing to the centre
of mass frame. Introducing the centre of mass coordinate, R ~ = (X, Y, Z), and the
relative coordinate, ~r = (x, y, z) defined by
~ = m1~r1 + m2~r2 ,
R ~r = ~r2 − r~1 . (12.113)
m1 + m2
The centre of mass will move like a free particle, and the relative motion reduces
to motion of a particle of reduced mass. Therefore, it is not surprising that the
separation of variables for the Schrodinger equation Eq.(12.112), can be achieved
by changing to the centre of mass and relative coordinates. In terms of the new
~ ~r, the Schrodinger equation takes the form
variables R,
2 2
~2 ∂ ∂2 ∂2 ~ ~r)− ~ 2
∂ ∂ 2
∂ 2 2
~ ~r)− e Ψ(R,~ ~r) = EΨ(R,
~ ~r).
− + + Ψ( R, + + Ψ(R,
2M ∂X 2 ∂Y 2 ∂Z 2 2µ ∂x2 ∂y 2 ∂z 2 r
(12.114)
~
If we now write the full wave function Ψ(R, ~r) as
~ ~r) = U(R)u(~
Ψ(R, ~ r), (12.115)
~ ~r are separated and we would get
and substitute it in Eq.(12.114), the variables R,
~ u(~r) given by
the differential equations for U(R),
2
~2 ∂ ∂2 ∂2 ~ = Ecm U(R), ~
− + + U(R) (12.116)
2M ∂X 2 ∂Y 2 ∂Z 2

~2 ∂ 2 ∂2 ∂2 e2
− + + u(~r) − u(~r) = Eu(~r). (12.117)
2µ ∂x2 ∂y 2 ∂z 2 r
where M = m1 +m2 is the total mass, µ is the reduced mass µ = (m1 m2 )/(m1 +m2 ),
and Ecm , E are constants appearing from the process of separation of variables so
that E = Ecm + E. The equation for U(R), ~ Eq.(12.116), is a free particle equation
and the centre of mass moves like a free particle. The equation (12.117) refers to the
relative motion of the electron and the nucleus. The Schrödinger equation (12.117)
can be solved by separation of variables in polar coordinates r, θ, φ. The angular part
of the wave function is given by a spherical harmonic Yℓm (θ, φ) and we, therefore,
write
u(~r) = R(r)Yℓm (θ, φ). (12.118)
The radial equation for R(r) now takes the form

1 d 2 dR 2m e2 ℓ(ℓ + 1)~2
r + 2 E+ − R(r) = 0, (12.119)
r 2 dr dr ~ r 2mr 2

d2 R(r) 2 dR(r) 2m e2 ℓ(ℓ + 1)~2
+ + 2 E+ − R(r) = 0. (12.120)
dr 2 r dr ~ r 2mr 2
111

2
The radial equation involves an effective potential Vef f = − er + ℓ(ℓ+1)~
2
2mr 2
. Remember-
2 2
ing that ℓ(ℓ + 1)~ is eigenvalue of L , the second term here is the centrifugal barrier
which appears for the radial motion in classical mechanics. The effective potential
goes to zero for large ρ. Hence for E > 0 energy eigenvalues are continuous and
bound states will exist for energy E < 0. The continuous energy will be of interest
for scattering problems. Here we shall discuss the bound state solutions only, E is
negative so we write E = −|E|. It is convenient to work with dimensionless variables
ρ and λ where
8µ|E|
ρ = αr, α2 = , (12.121)
~2
and r
2µZe2 Ze2 µ
λ= = . (12.122)
α~ 2 ~ 2|E|
The equation for the radial wave function written in terms of ρ takes the form

d2 R(r) 2 dR λ 1 ℓ(ℓ + 1)
+ + − − R = 0. (12.123)
dρ2 ρ dρ ρ 4 ρ2
Large ρ behaviour
The above equation can transformed into a form similar to one dimensional equation
by introducing χ(ρ) = ρR(ρ),

d2 χ λ ℓ(ℓ + 1) 1
+ − − χ(ρ) = 0. (12.124)
dρ2 ρ ρ2 4
The behaviour of the radial wave function can be found easily by taking large ρ
limit Eq.(12.124). For large ρ the terms λ/ρ and ℓ(ℓ+1)
ρ2
can be neglected compared
to the constant term 1/4. The radial wave function for large ρ obeys
d2 χ(ρ) 1
− χ(ρ) = 0. (12.125)
dρ2 4
showing that χ(ρ) ≈ e±ρ/2 . The wave function must be bounded everywhere in-
cluding at infinity, so we must have χ(ρ) ≈ e−ρ/2 . This suggests that we write
R(ρ) = e−ρ/2 F (ρ), and solve for F (ρ). The equation for F (ρ) turns out to be

d2 F (ρ) 2 dF (ρ) (λ − 1) ℓ(ℓ + 1)
+ −1 + − F (ρ) = 0. (12.126)
dρ2 ρ dρ ρ ρ2
Solution by Frobenius Method

We now find the solutions of this equation by the method of series solution. Assum-
ing X
F (ρ) = ar ρc+m , (12.127)
m=0
substituting it in Eq.(12.126) and equating coefficients of powers of ρ to zero gives
c(c + 1) − ℓ(ℓ + 1) = 0 ⇒ c = −ℓ − 1, ℓ (12.128)
112

Since ℓ ≥ 0, the value c = −ℓ − 1 is discarded as it gives rise to radial wave function
diverging to ∞ at ρ = 0.
Taking c = ℓ, the recurrence relation for the coefficients ar turns out to be
(m + ℓ + 1 − λ)
am+1 = am . (12.129)
(m + 1)(m + 2ℓ + 2)
This ratio of coefficients, for large m, is
am+1 1
∼ . (12.130)
am m
which coincides with the corresponding ratios for the series ρk eρ . Hence if the series
does not terminate, the solution F (ρ) behaves like ρk eρ for large ρ, giving the radial
wave function diverging like ρk eρ/2 for large ρ. This is unacceptable and hence the
series must terminate. If the series terminates after n′ terms, an′ + 1 = 0 and from
Eq.(12.129) we must have
λ = n′ + ℓ + 1 ≡ n. (12.131)
The energy is then given by
Z 2 e4 µ
En = −|En | = − , (12.132)
2~2 n2
where n is called the principal quantum number and n′ is the radial quantum number
which is equal to the number of nodes of the radial wave function. The number ℓ
is the azimuthal quantum number giving the angular momentum. The principal
quantum number n takes values 1, 2, 3, · · · . Thus wee that there are infinite number
of energy levels for the Coulomb problem. For a fixed n, ℓ takes values 0, 1, 2, ..., (n−
1). It should be noted that the eigenvalues of constants of motion, L~ 2 , Lz and parity,
′
along with the radial quantum number n are the quantum numbers which label the
states of the H- atom.
Properties of H- atom wave functions

The final expression of the wave functions for hydrogen like problems is given by
unℓm (r, θ, φ) = Rnℓ (r)Yℓm (θ, φ), (12.133)

Rnℓ (r) = Nnℓ ρℓ L2ℓ+1
n+ℓ (ρ) exp(−ρ/2), (12.134)
v
u
u 2Z 3 (n − ℓ − 1)!
Nnℓ = t 3 , (12.135)
na0 2n (n + ℓ)!
2Z ~2
with ρ= r, a0 = . (12.136)
na0 µe2
Here Nnℓ is normalisation constant, Lpq (ρ) are associated Laguerre polynomials and
a0 is the radius of the first Bohr orbit for hydrogen (Z=1). The energy levels are
given by
Z 2 e4 µ
En = − 2 2 . (12.137)
2~ n
113

The first few radial wave functions are given by
3
R10 = (Z/a0 ) 2 2 exp(−Zr/a0 ) (12.138)
3
R20 (r) = (Z/2a0) (2 − Zr/a0 ) exp(−Zr/2a0 )
2 (12.139)
3 √
R21 (r) = (Z/2a0) 2 (Zr/ 3a0 ) exp(−Zr/2a0 ) (12.140)
For plots and properties of the wave functions, see for example
Schiff,“Quantum Mechanics”
Bransden and Joachain,“ Quantum Mechanics”.
Accidental Degeneracy of the Coulomb Problem

For hydrogen atom the energy levels are given by
Z 2 e4 µ
E=− 2 ′
2~ (n + l + 1)2
The energy does not depend on n′ and l separately but only on the combination
n = (n′ + l + 1). For a fixed n, l can have values 0, 1, · · · , n − 1 ( because n′ ≥ 0 )
and all these solutions correspond to the same energy eigenvalue. The energy level
diagram of H-atom, therefore, appears as shown below.
l=0 l=1 l=2 l=3 l=4
non-degenerate 3 fold degenerate 5 fold degenerate 7 fold degenerate 9 fold degenerate
n=7
n=6
n=5
n=4 n=4 n=4 n=4 n=4 n=4
7 fold 9 fold
degenerate degenerate
n=3 n=3 n=3 n=3
5 fold
degenerate
n=2 n=2 n=2
3 fold
degenerate
n=1 n=1
non-degenerate
Putting all the levels which have the same energy together we get the following
schematic representation of energy levels of H atom. This table also shows that the
allowed values of l for each n, and number of m values for each level. The number
of total m values, with the same energy, is n2 and the degeneracy, after taking spin
into account, becomes 2n2 .
114

l values number of m values degeneracy
0, 1, ..n − 1 n2 2n2
P
n=4 32 l = 0, 1, 2, 3 (2l
P + 1) = 1 + 3 + 5 + 7 = 16 2 × 16 = 32
n=3 18 l = 0, 1, 2 P(2l + 1) = 1 + 3 + 5 = 9 2 × 9 = 18
n=2 8 l = 0, 1 (2l + 1) = 1 + 3 = 4 2×4 = 8
P
n=1 degen=2 l=0 (2l + 1) = 1 2×1 =2
Comparing the hydrogen atom levels with those of a general spherically sym-
metric potential, we find that energies for states with several different values of
l (= 0, 1, 2 . . . n − 1) are the same. For a general spherically symmetric potential
different combinations of n, l values correspond to different bound state energies,
and are (2l + 1) fold degenerate. Thus, an extra degeneracy is present for H atom,
beyond the expected (2l + 1) fold degeneracy for every spherically symmetric prob-
lems. This phenomenon present in the case of hydrogen atom is known as accidental
degeneracy. Another well known case of accidental degeneracy is that of isotropic
harmonic oscillator ( V (r) = 21 kr 2 ) in three dimensions.
Remarks
1. It must be emphasised that the accidental degeneracy is due to the special

symmetry, of rotations in four dimensions, of the Coulomb problem. The hy-
drogen atom problem can be related to harmonic oscillator in four dimensions.
1
2. Any slight deviation of the potential from r
will result in splitting of energy
levels with different values of l.
3. It is known that the accidental degeneracy is present whenever the Schrödinger

equation Hψ = Eψ can be separated into ordinary differential equations in
more than one set of coordinate system.
H atom — Separation of variables for the Coulomb problem is possible in
(i) spherical polar coordinates r, θ, φ

(ii) parabolic coordinates ξ, η, φ
ξ = r − z = r(1 − cos θ) (12.141)

η = r + z = r(1 + cos θ) (Schiff) (12.142)
φ = φ (12.143)
Isotropic Harmonic Oscillator — Separation of variables for V (r) = 12 kr 2 is

possible in
(i) Cartesian coordinates.

(ii) Spherical polar coordinates.
115

4. Finally, I wish to remind you that the non-relativistic result −R/n2 for the
energy levels of H-atom is not the end of story for H-atom levels. Precision
experiments show that each level is not a single level. To understand the
experimental facts we must take into account of relativistic effects using Dirac
theory of electron
Dirac Theory ր Spin orbit coupling ց Fine Structure

ց Relativistic variation of mass ր
Also a hyperfine structure, seen in the energy levels, requires a treatment of

the spin-spin interaction of electron with the nucleus and an explanation of a
tiny ‘Lamb shift’ requires use of quantum field theory.
Hyperfine structure → Effect of Nuclear Spin

Lamb shift → Quantum field Theory, Vacuum Polarisation Effect
116

QM LecNotes 01

Uploaded by

Document Information

Original Description:

Original Title

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

QM LecNotes 01

Uploaded by

Copyright:

Available Formats

LECTURES ON QUANTUM MECHANICS-I

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

2 Inadequacy of Classical Theories and Old Quantum Theory 11

3 States and Dynamical Variables 16

4 Probability and Average Values 23

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

7 Working with Representations 45

8 Time Dependent Schrodinger Equation 54

9 Energy Eigenvalues and Eigenfunctions in One Dimension 66

10 Reflection and Transmission in One Dimension 88

11 Potential Problems in Two and Three Dimensions 93

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Double Slit Experiment With Waves

φ1 (~r, t) = A1 exp(i~k · ~r), (1.3)

The resultant intensity is then given by

I12 = |φ1 + φ2 |2 = |φ1|2 + |φ2|2 + (φ∗1 φ2 + φ1 φ∗2 ). (1.6)

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Double slit Experiment With Only One Slit Open at A Time

I12 = |φ1 + φ2 |2 ? (1.7)

or simply the sum of intensities due to the two slits:

I = I1 + I2 = |φ1 |2 + |φ2 |2 ? (1.8)

§3 Gedanken Experiments With Bullets

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Double Slit Experiment

N(x) = N1 (x) + N2 (x). (1.9)

P12 (x) = P1 (x) + P2 (x), (1.11)

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

§4 Summary and Discussion

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Inadequacy of Classical Theories

§1 Inadequacy of Classical Theory

1. Black Body radiation

4. Frank Hertz experiment

6. Stern Gerlach Experiment

7. Zeeman splitting of spectral lines in magnetic field

9. Wave nature of electrons

10. Alpha decay

11. Specific heat of solids and gases

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

§3 Superposition, Indeterminacy and Appear-

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

States and Dynamical Variables

§2 Structure of Classical theory

• A point particle constrained to move on a curve or a surface.

• Interaction of charged particles and electromagentic fields.

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

1 Description of physical system of interest

Physical states: By state of a physical system we mean method of specifying

Dynamical variables: The dynamical variables of a classical system are functions

Interactions: A classical description is completed by specifying the forces, or the

3. Hamiltonian and Poisson Brackets

4. Hamilton Jacobi Formulation

DRAFT -- DRAFT -- DRAFT -- DRAFT -- DRAFT -

Waves: The state of a vibrating string is described completely by specifying the

Electromagnetic Field: The systems consisting of charged particles interacting

Thermodynamics, Statistical Mechanics: For systems consisting of a large