This book provides a first course on quantum mechanics and describes
simple applications to physical phenomena that are of immediate and
everyday interest.

The first five chapters introduce the fundamentals of quantum mech-
anics and are followed by a revision quiz with which readers may test
their understanding. The remaining chapters describe applications,
including the theory of spin and its application to magnetic resonance
imaging, physics of lasers, molecular binding, simple properties of
crystalline solids arising from their band structure, and the operation of
junction transistors.

Ideal either as a course text or a self-study text, the book contains
nearly 100 exercises and hints to their solution. It will be valuable for
undergraduate students at 2nd or 3rd year in physics, electronic
engineering, mathematics, chemistry.

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:09:15 GMT 2014.
http://ebooks.cambridge.org/ebook.jsf?bid=CBO9781139171243
Cambridge Books Online © Cambridge University Press, 2014

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:09:15 GMT 2014.
http://ebooks.cambridge.org/ebook.jsf?bid=CBO9781139171243
Cambridge Books Online © Cambridge University Press, 2014

Essential Quantum Physics

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:09:15 GMT 2014.
http://ebooks.cambridge.org/ebook.jsf?bid=CBO9781139171243
Cambridge Books Online © Cambridge University Press, 2014

Downloaded from Cambridge Books Online by IP 150.org/ebook. http://ebooks. 2014 .cambridge.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press.244.208 on Thu Feb 13 17:09:15 GMT 2014.221.

org/ebook.244.221.Essential Quantum Physics Peter Landshoff University of Cambridge Allen Metherell University of Central Florida Gareth Rees University of Cambridge CAMBRIDGE UNIVERSITY PRESS Downloaded from Cambridge Books Online by IP 150. http://ebooks.208 on Thu Feb 13 17:09:15 GMT 2014.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press.cambridge. 2014 .

Madrid. New York. QC174. .org/ebook. Landshoff. / Peter Landshoff. Metherell and W. Tokyo Cambridge University Press The Edinburgh Building. accurate or appropriate. org Information on this title: www. Sao Paulo. Cape Town. Subject to statutory exception and to the provisions of relevant collective licensing agreements.cambridge. Cambridge CB2 8RU. travel timetables and other factual information given in this work are correct at the time of first printing but Cambridge University Press does not guarantee the accuracy of such information thereafter.208 on Thu Feb 13 17:09:15 GMT 2014.. Rees 1997 This publication is in copyright.L37 1998 530. Melbourne. no reproduction of any part may take place without the written permission of Cambridge University Press. Information regarding prices. .. Gareth Rees. http://ebooks.org/9780521629935 © P. CAMBRIDGE UNIVERSITY PRESS Cambridge. I. Rees.12. 1937- Simple quantum physics. Singapore. p.cambridge. 2014 . UK Published in the United States of America by Cambridge University Press. Gareth. III. V. Allen Metherell. II. and does not guarantee that any content on such websites is. Allen. First published 1997 Reprinted 1999.2001 A catalogue recordfor this publication is available from the British Library Library of Congress Cataloguing in Publication data Landshoff.12-dc21 97-25151 CIP ISBN 978-0-521-62011-6 Hardback ISBN 978-0-521-62993-5 Paperback Transferred to digital printing 2010 Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. cm. Title.ISBN 0 521 62993 4 (pb) 1. Cambridge.244. Metherell. 1959. 1937. Includes index. A.2nd ed. ISBN 0 521 62011 2 (hb). Delhi. New York www. G. Dubai. Quantum theory. Downloaded from Cambridge Books Online by IP 150. Peter.221. or will remain.

Wave packets. The one-dimensional square well. Problems 4 The superposition principle 36 Linear operators.244. 7 Introduction to perturbation theory 70 Time-independent perturbation theory. Commutators. Continuity conditions. Spher- ically symmetric potentials.208 on Thu Feb 13 17:09:26 GMT 2014. Contents Constants of quantum physics x 1 Preliminaries 1 Atoms. Probability interpretation and normalisation. Operators and observables. The lin- ear harmonic oscillator. Problems 2 The Schrodinger equation 8 Wave functions and operators. 2014 . The WKB approximation. The delta-function potential. The deuteron. Problems 3 Special solutions 19 Particle in a box. Many- electron atoms. Beams of particles. Problems vn Downloaded from Cambridge Books Online by IP 150. Wave nature of matter. Problems. Her- mitian operators.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. Photons.cambridge. Alpha decay. Transition probability. Time-dependent per- turbation theory. Other molecules. Example: the one-dimensional potential well. Two-body systems. Problems Revision quiz 62 6 The hydrogen molecule 63 The ionised hydrogen molecule.221. The tunnel effect. Energy uncertainty principle. http://ebooks. Example: decay of tritium. The hydrogen atom.org/ebook. Ehrenfest's theorem. Orbital angular momentum. Sudden change in the Hamiltonian. Problems 5 The hydrogen atom 49 Good quantum numbers.

Population inversion. Spin | . Band structure. Pair excitation in intrinsic semiconductors. The ammonia maser. Motion in an external electric field. Simple consequences of band structure.Vlll 8 Spin 82 Two kinds of angular momentum.244. Problems 11 Electron motion in crystals 126 Electron velocity.208 on Thu Feb 13 17:09:26 GMT 2014. Spin precession. Electric current. n. Resonant absorption and stimulated emission.221. The laser. The junc- tion transistor. The diode. Problems 10 Band structure of crystals 112 Electrons in crystals.cambridge. Two simple circuits. Problems 12 Transistors 141 Impurities. http://ebooks. Electric dipole transitions. The Zeeman effect. Semiconductor junction. 2014 .org/ebook.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. Number of levels in a band. The ammonia molecule. Impurities and crystal colour. The electro- magnetic interaction. Problems Appendices A Power-series solutions 158 B The delta function and Fourier transforms 162 C Orbital-angular-momentum operators 173 D Electrodynamics 175 E Bloch waves 178 Hints for the problems 180 Index 201 Downloaded from Cambridge Books Online by IP 150. Thermal excitation.and p-type semiconductors. Problems 9 Masers and lasers 93 Radiative transitions. Effective mass and holes. Band overlap. Holog- raphy.

the physics of masers and lasers. the main change is the addition of a chapter on the theory of spin. Preface This book is intended as a first course on quantum mechanics and its applications. In this new edition.doi. The remaining chapters concentrate on applications. 2014 . the first application is to scattering problems. as they form an integral part of the course. and have made a number of other minor changes. These include molecular binding.244. Peter Landshoff Cambridge Allen Metherell January 1997 Gareth Rees IX Downloaded from Cambridge Books Online by IP 150. In most courses on quan- tum mechanics. and are fol- lowed by a revision quiz to test the reader's understanding of them.208 on Thu Feb 13 17:09:38 GMT 2014. A few problems are included at the end of each chapter. and the operation of junction transistors. and its application to magnetic resonance imaging. It is designed to be a first course rather than a complete one. While recognising the importance of scattering theory. http://dx.001 Cambridge Books Online © Cambridge University Press. A previous edition of this book was published under the title Simple Quantum Physics in 1979. The book should be suitable also for engineering students. simple properties of crystalline solids aris- ing from their electronic band structure. and it is based on lectures given to mathematics and physics students in Cambridge. We have also described the WKB approximation and its application to a-decay. The first five chapters deal with basic quantum mechanics. we have chosen rather to describe the application of quantum mechanics to physical phenomena that are of more everyday interest. Christ's College.org/10. Some hints on their solution may be found at the end of the book. We urge the student to work through all of these.221.1017/CBO9781139171243.

05 x 10" 3 4 J s Charge of electron -e = -1. 2014 .67 x 10" 2 7 kg = 938 MeV/c 2 Electron volt 1 eV = 1.org/10.00 x 108 m s" 1 Mass of electron rae = 9.60 x 10" 19 C Fine-structure constant e2IA-KSQTIC = 1/137 Speed of light c = 3.11 x 10" 3 1 kg = 0. http://dx.244.38 x 10" 2 3 J K" 1 Downloaded from Cambridge Books Online by IP 150.51 MeV/c 2 Mass of proton rap = 1. Constants of quantum physics Dirac's constant h = h/2iT = 1.60 x 10" 1 9 J Boltzmann's constant kB = 1.002 Cambridge Books Online © Cambridge University Press.221.1017/CBO9781139171243.208 on Thu Feb 13 17:09:48 GMT 2014.doi.

the diameter of the whole atom is about 10~10 m.003 Cambridge Books Online © Cambridge University Press. 1 Preliminaries Atoms An atom consists of a positively charged nucleus.1. and neutrons. The mass of the proton or neutron is some 2000 times that of the electron. so that it keeps the neutrons and protons very close together. It is natural to think of the electrons as being in orbit round the nucleus. together with a number of negatively charged electrons. Downloaded from Cambridge Books Online by IP 150.221. http://dx. there is a serious difficulty. each of which carries positive charge e. it has been ionised. The nuclear force does not affect electrons. The nuclear force is much stronger than the electrical force. so that when the atom has Z electrons it is electrically neutral. the diameter of a nucleus is of the order of 10~15 m. The electrons are held in the atom by the electrostatic attraction between each electron and the nucleus. According to classical electrodynamics. when a charged particle is accelerated it inevitably radiates energy (this is the basic principle of radio transmission). figure 1. The electrostatic force that keeps the electrons in their orbits is an inverse-square-law force. which have no charge. The protons and neutrons are held together in the nucleus by a different type of force. The charge on each electron is —e. the atomic number. just as is the gravitational force that keeps the planets in orbit. just as the planets are in orbit round the sun.208 on Thu Feb 13 17:10:02 GMT 2014. By contrast.org/10. If some of the electrons are stripped off. the nuclear force. is the number of protons.244. where Z. So the charge on the nucleus is Ze. so that the two systems would seem to obey precisely similar equations. but this is about 10~40 times less strong. so that for many purposes one can think of the nucleus as a point charge. so nearly all the mass of the atom is in the nucleus.doi. So according to classical physics the electron would continuously lose energy and its orbit would form a spiral which would gradually collapse into the nucleus. 2014 . It is a very short-range force. There is also an attraction because of the gravitational force. When a particle moves in a curved orbit its velocity vector is continuously changing: the particle is being accelerated towards the centre of its orbit. and its attraction more than counteracts the electrostatic repulsion between pairs of protons. Inside the nucleus there are protons. the atom then has net positive charge. and so may be neglected. However.1017/CBO9781139171243.

1 Preliminaries Figure 1.org/10. Atoms can be bound together to form molecules (see chapter 6). but hydrogen has a stable isotope. while 'heavy' water D2O has deuterium nuclei instead of the ordinary hydrogen. In quantum mechanics. The reason that this does not happen is that very small systems. To describe an atom one has to use quantum mechanics. such as atoms. the separation between these levels is so small that this is not a very real restriction. Classical picture of negatively charged electrons in orbit round the posi- tively charged nucleus of an atom.208 on Thu Feb 13 17:10:02 GMT 2014. The chemical properties of heavy Downloaded from Cambridge Books Online by IP 150.003 Cambridge Books Online © Cambridge University Press. 2014 . Z = 1. they are said to be isotopes of the same element. For example. it cannot radiate any more energy. Thus atoms whose nuclei differ only in the number of neutrons that they contain have similar chemical properties. One can also use quantum mechanics to describe the solar system.1. http://dx. called deuterium.doi. its energy is allowed to change only to that of one of the other allowed discrete levels.1017/CBO9781139171243. and classical mechanics is perfectly adequate to describe the system. If a planet in orbit is given an impulse. as opposed to classical mechanics. whose nucleus consists of one proton and one neutron. However. one cannot arbitrarily choose a value for the energy of the orbiting particle and then find an orbit corresponding to that energy. only certain discrete values of the energy are allowed. and different isotopes of the same element do this in the same way. The effects of quantum mechanics are generally only important for submicroscopic systems.244. the allowed energy levels of the planets are discrete.221. do not obey classical mechanics. and so the total collapse of the atom is not possible. the atom of the common form of hydrogen contains just a single proton. that is. Just as for the electrons. When the electron is in its lowest allowed energy level. Ordinary water H2O consists of molecules containing two hydrogen nuclei and one oxygen nucleus. The chemistry of an atom is determined by the charge on its nucleus.

Photons water are exactly the same as those of ordinary water. (1. called the work function of the metal. and W is the energy that must be given to the electron to enable it to overcome the electrostatic attraction of the metal.1) Here h is Planck's constant. The outermost orbital electrons of the atoms are almost free. Downloaded from Cambridge Books Online by IP 150.221. but the energy with which they escape does not depend on the beam intensity.) These results are explained by the hypothesis that a beam of light can be thought of as a collection of particles. Rather it depends on the colour or frequency v of the light.208 on Thu Feb 13 17:10:02 GMT 2014. 2014 . there is a definite minimum value Wo. (The value of W varies. The number of electrons that escape in a given time rises with the intensity of the beam of light.1017/CBO9781139171243. called photons. (1. The kinetic energy T with which the electrons escape is found to obey the equation hu = T + W. some of the electrons can actually escape from the surface of the metal and can be detected as an electric current.doi. http://dx. The number of electrons ejected rises as the intensity of the light is in- creased because there are then more incident photons. For a given metal. it is denser because of the extra neutrons. the atoms are effectively anchored to fixed sites by the electro- static forces due to all the other atoms.2) The electron is ejected from the metal when one of the photons collides with it and is absorbed by it. but there are some differences in its physical properties. The number of photons is proportional to the intensity of the light. If the metal is bombarded with light. so giving up all its energy to the electron. In particular.003 Cambridge Books Online © Cambridge University Press. Photons In a metal. and move through the metal when an electric field is applied (see chapter 11).org/10. and the energy E of each photon is proportional to the frequency. This is the photoelectric ef- fect.626 x 10" 34 J s. E = hv. h = 6.244. and so there is a greater chance of a photon being absorbed. according to the state within the metal from which the electron is ejected.

and the conductor must experience and equal and opposite force. http://dx.003 Cambridge Books Online © Cambridge University Press. they cannot be absorbed because it can be shown (see problem 1. Equations (1. sending the electron from the lower to the higher level. that is.208 on Thu Feb 13 17:10:02 GMT 2014. The kinematics of the process can be worked out using (1. The absorption of the photon then excites the atom. an atom in an excited state can decay by emitting a photon. a free particle can scatter the photon.244. Equation (1. and the results are found to agree with experiment.doi. that is. When photons collide with free electrons or protons. This is a realisation of the clas- sical idea of radiation pressure. (1.1017/CBO9781139171243.221. and only if.2) also helps to explain atomic spectra. not bound into a solid.4 1 Preliminaries Photons move with the speed of light.2).2) that this would violate conservation of energy and momentum. (The number of photons that can be absorbed in this way of course depends on how many atoms happen to be in the lower of the two states to start with.4).36) so that for a photon E = cp. according to the results of quantum mechanics. Downloaded from Cambridge Books Online by IP 150. their energy is equal to the difference between the energies of two electron levels. (1. In terms of the momentum p of the particle. W. so their kinematics must be de- scribed by the laws of special relativity. the allowed energy levels of the electrons in atoms are discrete. so changing its energy and therefore its frequency.org/10.3a) reads E = c(m2c2+p2)1'2.) However. the momentum of each photon is reversed. (1. the frequency of the photon depends on the difference between the initial and final energies: hv = E2 -Ei. We have said that. If a beam of light is shone on a collection of atoms.) Thus only photons with certain discrete frequencies are absorbed.4) and the relativistic energy-momentum conservation laws (see problem 1. 2014 . the photons can be absorbed by the atoms if.2) and (1. (In the photoelectric effect some of the energy. is absorbed by the other particles in the metal.4) If a beam of light is shone normally on a perfect conductor it is reflected. (1. The energy of a particle whose speed is v and whose rest mass is m is E = mc2/{l-v2/c2)ll\ (1. at the same time the particle recoils.4) are tested in the Compton effect.3a) so that when v = c the energy can be finite only if m = 0. Conversely. photons have zero mass. This must occur by some sort of force being exerted on the photons.

The spacing of these fringes is calculated from the wavelength A of the light.003 Cambridge Books Online © Cambridge University Press. it also has wave-like properties. http://dx.doi. If a beam of electrons is passed through a crystal. a coherent beam of light is diffracted when it is shone through a pair of closely separated slits: if a screen is placed at large distance behind the slits. 2014 . The regularly spaced atoms in the crystal cause the diffraction. the two components cancel. Wave nature of matter Slits Screen Figure 1. where n is an integer. Dark fringes appear at points on the screen such that their distances from the two slits differ by (n + |)A. So quantum mechanics gives light a dual nature.208 on Thu Feb 13 17:10:02 GMT 2014. in others like a wave.2. so that the spectrum of frequencies absorbed and emitted provides a useful way of identifying substances. Then the light received from the two slits is exactly out of phase.2. In some respects it behaves like a collection of particles.org/10. The same is true for electrons and other particles. it is diffracted. If a fluorescent screen is set up behind the crystal. There is darkness at points on the screen such that the path difference between rays that pass through the two slits is (n -f ^ The energy levels of an atom (or molecule) depend on what element it is. a diffraction pattern appears on the screen. Wave nature of matter Although light can be thought of as a collection of photons. The pattern can be explained by associating with the elec- trons a wave of wavelength A. For example.244. which changes with the momentum of the Downloaded from Cambridge Books Online by IP 150. a pattern of light and dark fringes appears on it. See figure 1. The double-slit experiment. The necessity for this is illustrated by the phenomenon of electron diffrac- tion.1017/CBO9781139171243.221. quantum mechanics associates waves with every kind of particle.

1).org/10.6 1 Preliminaries electrons according to de Broglie's relation A . if the experiment is repeated many times. the frequency v is not directly measurable and there is some arbitrariness in its definition.5). (1. Because the classical waves have speed c.h/p. The waves of quantum mechanics may describe either a collection of particles or a single particle.5) applies also to photons. Classical electromagnetic waves are associated with a very large number of photons (see problem 1. or with any other kind of particle. However. It is important to understand that quantum-mechanical waves are more abstract than classical waves.4) gives (1.2) and (1. In this case we cannot predict with certainty what will be the angle 9 through which the photon or electron is diffracted.208 on Thu Feb 13 17:10:02 GMT 2014. or one can calculate the average value of an observable quantity if the experiment that measures it is repeated many times. we find a probability distribution for the angle 9 that has the same shape as the variation of intensity with 9 in an experiment where there is a continuous beam of photons or electrons. which may be chosen so as to be related to the electron energy E by the relation E = hv as for photons (see (1. http://dx. 2014 .003 Cambridge Books Online © Cambridge University Press.244.6) and combining this relation with (1. or where electrons are diffracted through a crystal. this implies that \v = c (1. Downloaded from Cambridge Books Online by IP 150. is somehow statistical. However.2)). As will become clear. Consider an experiment where light is diffracted through a pair of slits. according to quantum theory one can never predict with certainty what will be the result of a particular experiment: the best that can be done is to calculate the probability that the experiment will have a given result. in the case of electrons.5) The waves may be assigned a characteristic frequency v. The de Broglie relation (1.1017/CBO9781139171243. Suppose that only one photon or electron is allowed to come into the experiment. The part played in the theory by the electron frequency will become apparent in the next chapter. This follows if we make the assumption that the quantum-mechanical waves that describe photons have the same frequency v and wavelength A as the corresponding classical electromagnetic waves. We explore this in the following chapters.221.doi. for example. This suggests that the association of a quantum-mechanical wave with a photon. E may or may not include the rest-mass energy of the electron.

5 Associated with the electron there is an antiparticle.003 Cambridge Books Online © Cambridge University Press. charge. which has equal mass and equal. How many photons does it emit per second? 1.3 A particle has mass 1 kg.244. show that an electron that is not in a bound state cannot absorb a photon. Show that the sum of the wavelengths of the two photons is Ao(l — cos#). http://dx. where 6 is the angle between their directions of motion and XQ is the Compton wavelength of the electron. and the quantity h/mc is the Compton wavelength of the electron. deduce from the conservation of energy and momentum that as the result of the scattering the wavelength of the photon changes by (h/mc)(l — cos#).208 on Thu Feb 13 17:10:02 GMT 2014. Wave nature of matter Problems 1. Using relativistic kinemat- ics.org/10.2 Using energy-momentum conservation.1017/CBO9781139171243.doi. but opposite. A positron impinges on an electron which is at rest. where 9 is the angle through which it scatters and m is its rest mass. (1. (This scattering process is known as Compton scattering.4 A photon of momentum p. 1.1 A radio transmitter operates on a wavelength of 100 m at a power of 1 kW. They annihilate into two photons. the positron. is comparable with the wavelength of visible light? What is the corresponding answer if the particle is an electron? 1. Downloaded from Cambridge Books Online by IP 150.5).221. How long does it take to move through a distance of 1 m if its de Broglie wavelength. and therefore of wavelength /i/p.) 1. scatters on an electron that is initially at rest. 2014 .

2
The Schrodinger equation

The basic equation of quantum mechanics is known as the Schrodinger
equation. It is not possible to prove that the equation must be true, just
as we cannot prove that Newton's laws, the basis of classical mechanics,
must be true. All that can be done is to work out the consequences
of the equation in different physical contexts, and to compare them as
exhaustively as possible with experiment. In this chapter we begin by
showing how the experimental facts that we have already described make
the truth of the equation plausible.
The Schrodinger equation describes non-relativistic particles, whose en-
ergy E and momentum p are related by

E=p2/2m. (2.1)

Non-relativistic kinematics can be used so long as the energy E is not
comparable with, or larger than, the rest-mass energy me2. Relativistic
quantum mechanics is much more difficult than the non-relativistic theory,
and will not concern us in this book. In particular, we shall not describe
the quantum mechanics of the photon; for this particle m = 0, and so
relativistic mechanics must always be used.

Wave functions and operators
We explained in chapter 1 that with a particle of energy E and momentum
p we somehow associate a wave of frequency v — E/h and wavelength
A = h/p. Instead of using v and A, it is convenient to introduce the
angular frequency

and the wave vector fc, whose direction is in the direction of wave propa-
gation and whose magnitude is

k = 2TT/A.

We also work with
h = h/2n

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:10:12 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.004
Cambridge Books Online © Cambridge University Press, 2014

Wave functions and operators

(pronounced '/i-cross') instead of h. Then we have

p = hk
E = hu. (2.2)

For a free particle, which is not interacting with any other particle or with
a potential, p and E are constant. Hence we expect such a particle to be
described by a wave for which k and u are constant:

tt(r,<)= tfei(*'r-wt). (2.3)

Here r denotes the position vector and TV is a constant. In many physical
applications, we write a wave in the form of a complex exponential, as is
done here, and understand that only the real part is physically meaningful.
In quantum mechanics, however, it will turn out that both the real and
the imaginary parts of \I> are needed. The plane wave (2.3) is the simplest
example of a wave function: it describes a free particle. Instead of (2.3),
we could use JVe""1^"1""^; the choice of sign is a matter of convention.
In order to guess how to arrive at wave functions that describe particles
that are not free, we perform a simple manipulation on (2.3). Differentia-
tion with respect to a component Xj (j = 1, 2 or 3) of r simply multiplies
\I/ by i times the corresponding component kj of k. So, with (2.2), we
have
)V=PjV (j = 1,2,3) (2.4a)

or, in more concise vector notation,

(-iftV)tf = p * . (2.46)

Equation (2.4a) says that if we apply the differential operator —ihd/dxj
to #, the result is to multiply \I> by the number pj. We say that # is
an eigenfunction of the operator (—ihd/dxj) with eigenvalue pj. (More
properly, since the differential equations (2.4) do not determine ^ until we
have imposed suitable boundary and continuity conditions on \I>, we should
at this stage specify these conditions. However, we defer the discussion
of this until the end of this chapter.) In this way, an experimentally
observable quantity, the momentum p of a particle, is associated with
a differential operator — iftV. We generalise this to other experimentally
observable quantities by introducing three basic assumptions:

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:10:12 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.004
Cambridge Books Online © Cambridge University Press, 2014

10 2 The Schrodinger equation

(a) States of a system are described by wave functions # .
(b) Observable quantities are associated with operators.
(c) When the value of an observable Q is known to be g, the system is
in a state whose wave function is an eigenfunction of the operator
Q corresponding to Q, with eigenvalue q. That is,

qV. (2.5)

We shall elaborate on these assumptions in chapter 4.
Consider the particular example of the observable that is the energy of
a particle. If it is a free particle, this is just p2/2m. So one would
expect the corresponding differential operator to be obtained by replacing
p in this expression by its corresponding operator —i^V, and so we have
(-iftV) 2 /2ra, that is, — Ti2V2 /2m. If the particle moves in a potential
V(r), this operator becomesf

H = (-h2/2m)V2 + V(r) (2.6)

corresponding to the classical expression p2/2m + V(r). H is called the
Hamiltonian operator. To find the possible energy levels E of the particle,
we must find the eigenvalues E of H, that is, we must solve the equation

(2.7a)

(subject to appropriate boundary conditions). This equation

[(-h2/2m) V2 + 7(r)]tt(r, t) = £ * ( r , t) (2.7b)

is the time-independent Schrodinger equation.
Notice that if the particle is not a free particle, so that 7 ^ 0 , the plane
wave (2.3) is not a solution of the Schrodinger equation; the wave function
ty will be more complicated.
The time-independent Schrodinger equation applies when the particle is in
a state such that its energy takes a definite value E. In a general situation
this may not be the case, for example when V depends explicitly on the
time t as well as on the position r of the particle. In order to guess what

f Although H is an operator, we shall follow the usual convention and not write the
operator symbol " over it.

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:10:12 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.004
Cambridge Books Online © Cambridge University Press, 2014

Example: the one-dimensional potential well 11

is the generalisation of the Schrodinger equation that describes this, we
return again to the plane-wave solution (2.3) appropriate to a free particle.
Making use now also of the second relation in (2.2), we find that the plane
wave satisfies

The obvious generalisation of this to the case where V ^ 0 is

[(-h2/2m) V2 + V]*(r, t) = \h 8V/dt (2.8a)

or
HV = ihdV/dt. (2.8b)
This is the time-dependent Schrodinger equation.
This is the equation that is valid in all circumstances whether or not the
particle is free and whether or not it is known to be in a state of definite
energy. In the particular case where the particle is known to be in a state
of definite energy E, the time-independent equation (2.7) is valid also.
When both equations hold,

This means that the time dependence of \P(r,4) is then trival:

y(r,t) = $(r)e-iEt/n. (2.9)

The function ^ ( r ) is called the time-independent wave function and it
satisfies the time-independent Schrodinger equation

Hip = E\j). (2.7c)

We repeat that a solution of the form (2.9) is applicable only when the
energy takes a definite value. For reasons that will be explained below, a
solution of this type is known as a stationary-state solution. Comparing
(2.9) with (2.3), we see that in the special case of a free particle

if>(r)=Neik'r (2.10a)

where, from (2.7),
E = h2k2/2m. (2.106)

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:10:12 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.004
Cambridge Books Online © Cambridge University Press, 2014

doi. in particular. for example an elec- tron bound in an atom. Suppose that the wave function ip depends on only one coordinate. If E is finite. ip(x) has to be continuous.. we deduce that in the region outside 0 < x < a. where E — Ti2k2/2m. we call this a one-dimensional model. Downloaded from Cambridge Books Online by IP 150. 2014 . The simplest mathematical model of a bound-state situation is as follows. Hence. As only one of the three coordinates is explicitly involved. the energy levels of a bound state are quantised. where V is infinite. To find the states of definite energy E we solve the time-independent Schrodinger equation.004 Cambridge Books Online © Cambridge University Press.221. which here reduces to [{-h2/2m)(d2/dx2) + V]t/>{x) = Et/){x).. the values of E that are allowed by the time- independent Schrodinger equation are discrete. x say. The condition at x = 0 picks out the sin&x solution for -0.1017/CBO9781139171243.\ As we shall see at the end of this chapter. since it vanishes for x < 0 and for x > a. we must impose the conditions tp(O) = ip(a) — 0. Otherwise the term VI/J in the Schrodinger equation (2. f Throughout we shall use primes to denote differentiation with respect to a space vari- able and dots to denote differentiation with respect to a time variable..208 on Thu Feb 13 17:10:12 GMT 2014. where V = 0.2. which is not possible because neither of the other two terms in the equation would be similarly infinite. That is. and so the equation would not be satisfied.org/10. http://dx. In the region 0 < x < a. 12 2 The Schrodinger equation Example: the one-dimensional potential well It turns out that for a particle in a bound state.12) would be infinite in this region.. the wave function ip(x) must vanish. it has to be continuous at x = 0 and x — a. and that at x — a imposes a restriction on the allowed values of k: k = nn/a n = 1.12) This equation is to be valid for all x. the Schrodinger equation reduces to IL / £ lib] (1/ I Ju / ""~~ 1 j (// I %XJ I and its general solution for ^ is a linear combination of smkx and coskx. (2. and let the potential be the infinite square well 0 0<x<a ( 2 n ) oo otherwise.244.

14) corresponds to saying that there is unit probability that a measurement of the position of the particle finds that the particle is actually somewhere.004 Cambridge Books Online © Cambridge University Press. The corresponding allowed energy values are En = n2h2n2/2ma2 n = 1.org/10. An exception is the special case of the plane wave (2.doi. but this would not give any additional wave functions.£)| 2 d 3 r is the probability that a measurement of the position of the particle de- scribed by \I> will give a result in the infinitesimal volume element d 3 r in the neighbourhood of the point r.. In practice. That is. |\&(r. Indeed. Downloaded from Cambridge Books Online by IP 150. In order for this interpretation to be consistent. and the integration is supposed to extend over all space. It has |\I>|2 = \N\2.14) diverges for all values of the multiplying constant N. If we have found a solution # to the Schrodinger equation that is not correctly normalised so as to satisfy (2. \I/ must be arranged to satisfy the normalisation condition (2.. the allowed energy values for the bound system are discrete. so that it gives equal probability of finding the particle anywhere throughout space.14) / • Here we have used the symbol / d 3 r as a shorthand for JJJ dx dy dz. Probability interpretation and normalisation 13 We could also allow negative integers n.. it was not until after it had been discovered that the Schrodinger equation gives the right energy levels of a system that the correct interpretation of \I> was suggested. http://dx. we can usually get one that is by simply multiplying by a suitable constant.3). So (2. it is not like the waves of classical physics. for which the normalising integral (2.£)| 2 is a probability density.2. (2. and the basic as- sumption is that |\I/(r. 2014 .136) That is. Probability interpretation and normalisation As we have said in chapter 1. We explained that the wave function has a statistical interpretation.208 on Thu Feb 13 17:10:12 GMT 2014.14). v J where the Cn are normalising constants (we explain below how these are chosen).. independent of r.1017/CBO9781139171243.221. The reason for this is that strictly the plane wave does not correspond to a physical situation.244. the wave function \& associated with a parti- cle in quantum mechanics does not describe a concrete wave. the allowed wave functions are Mx) = {Cn sm(nnx/a) 0 < x< a } ^ nV ' \ 0 elsewhere.

if we integrate over any volume fi.1017/CBO9781139171243. and by dividing ^ by a suitable finite constant we can make it converge to 1.org/10. Then if \l/(r.14 2 The Schrodinger equation one knows that the particle is confined within some volume £)o> f° r example within a given building. 2014 .156) Downloaded from Cambridge Books Online by IP 150.221. of a physically allowed solution. (2. d * Jn Jn = Jn[ d r q3 3 (*** + * * * ) . which can in fact be circumvented by methods that we shall not discuss here. it remains so.* V * * ) . we use the Schrodinger equation and its complex conju- gate: [(-h2/2m)V2 [(-h2/2m)V2 + Thus — at 2miJa ' = -77—.3) is best thought of as a mathematical limit. For.244. Suppose that the particle is confined to a finite volume £2o> so that its wave function * vanishes outside that volume.No problems arise for finite f2o> &s then the normalisation integral converges. and if the volume QQ is large (for example when fi0 represents the volume of a typical experimental apparatus.15a) To evaluate this. the numerical results are changed very little by taking the limit QQ —> oo.14). where QQ —>• oo. which is almost always very large when measured on a quantum-mechanical scale). and that therefore the wave function vanishes outside fio. Apart from the normalisation problem. the mathematics of the f2o —> oo situation is rather simpler than that of the proper wave functions.doi. The plane-wave solution (2. http://dx. t) is correctly normalised at a given time i. \ d3r V • O T V * .004 Cambridge Books Online © Cambridge University Press. (2.208 on Thu Feb 13 17:10:12 GMT 2014. as is required in (2.

and its direction is that of the outward normal to S. H Jeffreys and B S Jeffreys. since we are assuming that \I> vanishes outside some volume f2O5 it is sufficient to integrate over any volume that contains fi0. See.* V * * ) . and so the integral (2. (2. [ [dS-j Js where j = (ft/2rai)(\I>*V\I/ .org/10. f The divergence theorem states that for any suitably well-behaved vector U = f dS u. http://dx.004 Cambridge Books Online © Cambridge University Press. For the normalisation integral (2.15).1017/CBO9781139171243. and therefore j .doi. vanishes all over it.14). for example. When the wave function corresponds to a state of definite energy. we have to integrate over all space.15b) vanishes.1): =. What we have said so far about normalisation applies to any wave function \I/(r.208 on Thu Feb 13 17:10:12 GMT 2014.Because the surface S is then outside the volume $~20? ^ . However. (Cambridge University Press) Downloaded from Cambridge Books Online by IP 150. 2014 .1. Js Here the vector dS is a vector whose magnitude is equal to an element dS of the area of the surface S bounding the volume Q.221. Probability interpretation and normalisation 15 Figure 2. The divergence theoremf can now be used to convert this to an integral over the surface S that encloses the volume of integration (see figure 2.15c) This relation is valid for any volume Q. The integration volume for (2. and its bounding surface S.244. t). Methods of Mathematical Physics.

This average value is called the expectation value of the particle density. In the case of an electron beam. for example a beam of electrons from an accelerator fired at an atom. i)\2 is the total number of particles at time t.221. A positive value for this integral would correspond to a depletion of the total number of particles within S: (2. (2. Exper- iments often involve beams of particles. the particle density is not too high. This is the reason that we call a state of definite energy a stationary state.t).1017/CBO9781139171243.doi. This is such that.9). that is. In this case . This means that we may describe the problem in terms of the same Schrodinger equation as describes a single electron moving in the potential produced by the atom. we have considered a single particle.i) is the average density that would be measured by setting up the same state many times and repeatedly measuring the particle density. 2014 . t). Beams of particles So far. as in (2. each electron interacts with the atom independently of the presence of the other electrons.004 Cambridge Books Online © Cambridge University Press. Then |#(r.14) appropriate for a single-particle problem. Suppose that when there is just one electron the potential produced by the atom results in a wave function \&(r. the same wave function describes the density of particles at the point r at time t. if we take any closed surface 5.244. If the beam is not too intense.*)| 2 = h/>(r)|2 so that the probability density at each point r is independent of time.16) Here p(r.208 on Thu Feb 13 17:10:12 GMT 2014. we may define j . it is convenient then to normalise \I> not by the integral (2. and interactions between the electrons are negligible. http://dx. However.16 2 The Schrodinger equation its time dependence factorises out trivially. the expectation value of the particle flux. A q • _ the expectation value of the number of particles per unit time that cross the surfaces'.org/10. such as an electron.£)| 2 = p(r. In addition to the expectation value of the particle density. p. but instead such that the integral Jd 3 r|\I/(r.17) Downloaded from Cambridge Books Online by IP 150.

221. estimate the difference between the speeds of the particle when it is in the first excited state and when it is in the lowest-energy state: (i) an electron confined by an infinite square-well potential whose width is roughly equal to the radius of an atom (about 2Q-io m j . (This means that.15c). V is often defined so as to vanish at infinity.3 Determine the constant Cn in the wave function (2.) 2. the point at which V vanishes may be chosen ar- bitrarily. as in classical mechanics.doi.t) and j(r. The infinity would have to be balanced by another infinite term and there is no such term unless V has an infinity. Show Downloaded from Cambridge Books Online by IP 150. if at all. The physical consequences of this are not surprising: both p(r. on the atomic timescale.208 on Thu Feb 13 17:10:12 GMT 2014. http://dx. 2014 . Hence there is effectively a stationary-state situation.18) 2rai In most cases.004 Cambridge Books Online © Cambridge University Press. That is. This means that if the potential V is well behaved. no particles are being created or destroyed at any point r.9) and p and j independent of t. so as to satisfy the normalisation condition (2. ^ a t e n n i s b a n confined by an infinite square-well potential whose width is equal to the width of a tennis court.8) is supposed to be valid at all points r. the wave function and its space derivative must both be continuous functions of r. this is just the relation (2.org/10. the potential and the rate at which the beam delivers par- ticles to the target vary very slowly.14). Continuity conditions 17 With # now normalised according to (2. and this cannot be because there cannot be just one term in the Schodinger equation that has an infinity. the appropriate expression for j is j A ( .1 How are the stationary-state solutions of the Schrodinger equation changed if a constant is added to the potential V(r)? Show that this change has no observable consequences.1017/CBO9781139171243. Problems 2. Continuity conditions The Schrodinger equation (2. By convention.t) have no discontinuous changes as r varies. with the wave func- tion factorising as in (2. (2.13a) for the infinite square well. That is.16). the second derivative V 2 ^ would be infinite at the points of discontinuity. For if this were not so.244. What do your answers tell you about the mechanics of microscopic and macroscopic systems? 2.2 For each of the following.

244. r2) = )V.004 Cambridge Books Online © Cambridge University Press. Show that as n —> oo these average values approach the values that are obtained from classical mechanics. In the general case.5 A system consists of two particles.18)) is independent of both t and x. and not with each other? Show that in this case the stationary-state wave functions have the structure ^(ri.2(^2).18 2 The Schrodinger equation that the probability interpretation of the wave function leads to the average values (x) = \a. the potential will be a function F(ri. Does this result have a simple generalisation to the three-dimensional case? 2.208 on Thu Feb 13 17:10:12 GMT 2014. r2).221.1017/CBO9781139171243. http://dx.4 Show that for a one-dimensional system in a stationary state the par- ticle flux (see (2.What equations do the functions ^i and V>2 obey? Downloaded from Cambridge Books Online by IP 150. What is the structure of the function V in the special case where the particles interact only with an external field of force.(x))2) = (a 2 /12)(l . What is the obvious probability interpretation of ip? Write down the time-independent Schrodinger equation. 2014 . ((x .org/10. 2.6/n27r2).doi. so that its stationary-state wave functions VKri? r2) are functions of the coordinates of each.

(3. Within the box. qn rn sir ki = — .3) must vanish on each of the six planes x = 0.244. z = 0.1017/CBO9781139171243. 3 Special solutions Particle in a box Consider the stationary states of a particle confined within the rectangular box-shaped volume 0 <x < a. http://dx. because the potential is infinite at all points outside the box. k2 = — .221. (3. it must also vanish on the faces of the box. the particle moves freely and the time-independent Schrodinger equation is 2ra (ITS ~t~ dy2 +dzin \dx2 + ITS 2 Elementary solutions to this equation are sin sin sin W\r) = kix k2y k3z cos cos cos where E = (h2/2m)(k21+kl + k23). 2014 . To make ij)(r) continuous. x = a. rather than the cos functions. a b c 19 Downloaded from Cambridge Books Online by IP 150. 0 < y < 6.doi.208 on Thu Feb 13 17:10:23 GMT 2014. The first three conditions pick out the sin functions. the wave function rp(r) must vanish for r outside the box. 0 < z < c. Hence (3.3) As we explained for the example of the one-dimensional infinite square- well potential in chapter 2.005 Cambridge Books Online © Cambridge University Press.org/10. &2 and k3: .1) Suppose that within the box the potential is zero and outside it is infi- nite. y = 6. y = 0. The last three conditions impose restrictions on the allowed values of &i. z = c. k3 = —.

r and 5 are integers. The significant feature of these results is that the allowed values Eqrs of the energy are discrete. over all space.6).r. These are the stationary-state solutions.7) q. or equivalently. which can be verified to satisfy the time-dependent Schrodinger equation (2. Thus the allowed wave functions are 1/2 sm Vqrs ( r ) = — . since i\) — 0 outside the box.9). (3.4) v } \abcj a b c and the allowed values of the energy are hn (q2 r2 s2 ^ 2m \ a2 b2 cz Notice that all these allowed values of Eqrs are positive. from (2. r and s are integers.221.1017/CBO9781139171243.org/10. a change that has no physical significance. (3.244.s=l where the cqrs are constants.4) we have included the normalisation factor (8/abc)1/2.8) and also have built into them the correct boundary condition that they vanish on the faces of the box. By superposing the stationary-state solutions (3. http://dx. so as to satisfy the normalisation condition the integration being over the volume of the box.— sin —^. Notice that we may as well confine them to being non-negative integers.005 Cambridge Books Online © Cambridge University Press. *qrs(r.sin . The corresponding time-vary ing wave functions are.t) = ^grs(r)e-iE<>r*t/h. 2014 . because of the condition that 9.6) In general.20 3 Special solutions where q. In (3. c qrs®qrs{rj).doi. we obtain more general wave functions \I>.208 on Thu Feb 13 17:10:23 GMT 2014. if we imagine that we somehow introduce the particle into the box at time t = 0. since changing the sign of any one of them merely changes the sign of the wave function. (3. Imagine we know that we have introduced the particle into the box in such a way that its wave function Downloaded from Cambridge Books Online by IP 150. it is unlikely that we will succeed in putting it into a stationary state.

8).7) becomes {-h2/2m)(d2ip/dx2) = Ex/} \x\ (-h2/2m)(d2ip/dx2) -Uip = \b. http://dx.1. where both the potential and the wave function are functions of one of the three coordinates only. (3. is thus determined in the form of an infinite series. = (— 1 V sin sin —-— sin \abcj ^ This is just a triple Fourier series for / ( r ) . (a) The one-dimensional square-well potential. due to the nucleus (see figure 3.0) = f(r).244. We investigate the bound states corresponding to the square-well potential 0 \x\ > ±6.1017/CBO9781139171243. the time-independent Schrodinger equation (2. The one-dimensional square well We now consider a one-dimensional problem. the complete wave function (3. (3 g) V = -U <h with U positive.9) Downloaded from Cambridge Books Online by IP 150. say x. which can be inverted by stan- dard methods to give the constants cqrs.7). The one-dimensional square well 21 b U (a) (b) Figure 3. for all t > 0. 2014 . (In practice. With the potential (3. This potential represents a very crude one-dimensional model of the potential experienced by an electron in an atom. at t — 0 is *(r. it would be very hard to determine /(r). If f(r) is known.) Then x 1/2 oo .org/10.doi.208 on Thu Feb 13 17:10:23 GMT 2014. qirx . (6) The Coulomb potential produced by a charged atomic nucleus.221. my .005 Cambridge Books Online © Cambridge University Press.1).

We can manipulate the equations as they stand. and we find that a = /?tan|/?6.116) where C and D are constants. which are symmetric about the origin. We now impose the boundary conditions that i/> and d^/dx are continuous at x — ±§&. (3. where E = h2k2/2m. and so we should impose the boundary condition \ij)\ -> 0 as x -+ ±oo. E + U = h2p2/2m. = Cpcos |/36 | By eliminating A.221.244. Hence we reject the case E > 0.11).9) are where A and B are constants. from (3. E = -h2a2/2m.9) is i/>(x) = Csinfix + Dcospx. with a and (3 > 0. For these.' Ae~*ab = Csin \(3b + Dcos |/36. (We have rejected a term eax for x > | 6 . (3. B.005 Cambridge Books Online © Cambridge University Press.Dp sin |/36. we obtain a relation between P and a. since when k is real the solutions do not satisfy this requirement. C. J? from these equations.208 on Thu Feb 13 17:10:23 GMT 2014. http://dx. 2014 .12a) Downloaded from Cambridge Books Online by IP 150. A = B and C = 0. we must have E < 0.10) the required solutions to the first equation in (3. for x < —16.org/10. (3.7): (a) Even-parity wave functions. for a bound state the particle should be more or less confined to values of x in the vicinity of the potential. However. The solution to the second equation in (3.doi.22 3 Special solutions The solution to the first equation is a linear combination of elkx and e~lkx.1017/CBO9781139171243. ip(x) = ip{-x). To satisfy the boundary condition. and a term e~aa. but the algebra is somewhat reduced if we first realise that the possible wave functions for the system fall into two classes (see problem 3. = Cficos \$b . so as to make |^| —> 0 as x —> ±oo). If we write.

we obtain a discrete set of allowed values for the energy E.13) We have sketched the two sides of these equations. The linear harmonic oscillator We now consider another one-dimensional problem.org/10.005 Cambridge Books Online © Cambridge University Press.12) we have to take care to accept only those points of intersection for which tan ^fib or — cot \fib is positive. or cot |/36 = . A = -B and D = 0.221.12a) or (3. The equations (3. For this reason the simple harmonic oscillator plays a f Because we have squared equations (3. This is not unexpected: the potential is then repulsive instead of attractive. One can show that if U < 0. In each case the two curves intersect in just a finite number of pointsf. at least if the amplitude of oscillation is sufficiently small.doi.1 + 2mU/h2/32 2 (for odd-parity solutions). If we combine either (3. with the simple harmonic solution x = smut or cosut.126) The reason for the existence of these two classes of wave function is that the potential is symmetric about x = 0. The number of such allowed values depends on the magnitudes of U and 6. which are antisymmetric about the origin.10). so resulting in a finite set of allowed values for E. in figure 3.244. For these. (3. and a = -0 cot \(3b.12). and so has no bound states.126) with (3. The linear harmonic oscillator 23 (6) Odd-parity wave functions.2. for which the potential is V(x) = \mu2x2.14) In classical dynamics. this would correspond to an attractive force whose magnitude is proportional to the distance from the origin. the bound states correspond to energies lying within the well. that is. 2014 . (3. against /?. In classical physics most oscillat- ing systems are simple harmonic. but we can get some information about the solutions by graphical methods. http://dx. Each of these lies in the range 0 > E > . il>(x) = ~^{-x).208 on Thu Feb 13 17:10:23 GMT 2014. there are no bound-state solutions.1017/CBO9781139171243. Downloaded from Cambridge Books Online by IP 150. (3.[ / .10) and (3.12) can only be solved by numerical methods.10) and (3. If we eliminate a between (3. we find that either tan 2 |/?6 = — 1 + 2mU/Ti201 (for even-parity solutions).

.14) is {-h2/2m){d2^/dx2) + (\mu2x2 . = 0. just as we shall find that in quantum mechanics the energy of a simple linear oscillator with the potential (3.2. These quanta are what we know as photons. so also the energy in an electromagnetic field appears in discrete quanta. Thus.org/10. Downloaded from Cambridge Books Online by IP 150.. So the mathematics that follows. and the same is true in quantum physics.005 Cambridge Books Online © Cambridge University Press. Solution to (3.244. (3. n = 0.13).14) can only take certain discrete values.208 on Thu Feb 13 17:10:23 GMT 2014. since accord- ing to classical electrodynamics the electromagnetic field in a light wave oscillates harmonically at each point of space. The time-independent Schrodinger equation for the potential (3. (3. This has the consequence that E must take one of the values En = (n+ \)hu.1017/CBO9781139171243.1.doi.221. as in the previous problem.24 3 Special solutions Figure 3. 2014 .16) For other values of E the equation has no solutions obeying the boundary condition. with suitable adaptations.. central role in classical physics. http://dx. applies also. for example.. to the quantisation of the electromagnetic field.2.15) To this is applied the boundary condition that ip —> 0 as x —> ±oo.

17) where Hn(x) is a polynomial of degree n in x (it is known as the Hermite polynomial). Hn(x) contains only even powers of a. In classical mechanics the lowest possible energy is. http://dx.doi. so that i[. We show there that the corresponding wave functions are tl>n(x)=e-m»x2'MHn(x). by superposing stationary-state solutions: x. the potential is symmetric about x = 0. Hn(x) contains only odd powers.org/10.221.. according to (3. it will not be in a stationary state. 2014 . so that <0n(-x) = —ij)n^x). zero.t). satisfying the appropriate boundary conditions. As in the case of the square well. and so we again find that the solutions have definite parity. This wave function satisfies only the time-dependent Schrodinger equation.208 on Thu Feb 13 17:10:23 GMT 2014. Downloaded from Cambridge Books Online by IP 150. the probability density at each point x oscillates in time with period 2TT/U. The time-dependent wave function corresponding to (3. We may write general solutions to the time-dependent Schrodinger equation. (3. it does not correspond to a stationary state.244. if the oscillator is somehow set in motion.005 Cambridge Books Online © Cambridge University Press. If n is even. For arbitrary choices of these constants.17) is In general.18) n=0 where the constants cn may take any values. (3. of course. Notice that. but in quantum mechanics this is not allowed by the uncertainty principle. the lowest possible energy of the oscillator is EQ = \Tiu. This is known as zero-point energy.1017/CBO9781139171243.16). if n is odd. which we explain in the next chapter. Notice that although each \I/n is a stationary-state wave function and so satisfies the time-independent Schrodinger equation. this is not true of the wave function \I> that we get by superposing them on each other. It corresponds to the particle being at rest at the origin.n(—x) = ipn(x). The tunnel effect 25 We give an outline proof of this in appendix A.. though the oscillation is not simple harmonic.

In the region of the barrier. For x < 0 the Schrodinger equation is (-h2/2m)(d2^/dx2) = Elf). 2014 . Suppose that the incoming beam of particles has energy E and has the time-independent wave function &n = Aeikx. as we discussed in chapter 2. 0 < x < a. so that if E < [/. http://dx. for simplicity. in quantum mechanics a beam of particles will be partially transmitted through this barrier even if the energy of each particle is less than [/. with U positive.1017/CBO9781139171243. Applying these boundary conditions gives us four equations. However. this happens also when E > U. In fact. (3. which is not allowed by the laws of classical physics and which is the basis of some of the ap- plications in the later chapters of this book. n is real. is the tunnel effect. the constant A depends on the density of particles in the beam. (3.221.org/10. Consider. In order to Downloaded from Cambridge Books Online by IP 150. but so far we only have two unknowns. In classical mechanics a particle approaching the origin from x < 0 would see this as a barrier which it could surmount only if its energy were greater than U.208 on Thu Feb 13 17:10:23 GMT 2014. some of the particles pass on to x = +oo. (3. that is.005 Cambridge Books Online © Cambridge University Press.doi. similar to the square well but opposite in sign: T/ _ // 0 x < 0 and x > a. the Schrodinger equation is (-h2/2m)(d2ilj/dx2) + U^ = E%l>.20) where. a one-dimensional potential.23) where h2K2/2m = U — E. There are two boundary conditions at each edge of the barrier: ip and dil)/dx are continuous at x — 0 and at x = a.21) so that if V>in is to satisfy it.26 3 Special solutions The tunnel effect An important phenomenon of quantum mechanics. B and C.244. while others are reflected back towards x = — oo. k and K being determined once we have chosen E. V ~\U 0<z<a.22) so that there the wave function is BeKX + Ce~"x (3. h2k2/2m = E.

where T and R are constants. (3.208 on Thu Feb 13 17:10:23 GMT 2014.26) We may solve these equations. To see this. and none is lost. we find that T = ^hA ( 3 27) eika[e«a(k + IK)2 + e~^a(n + ik)2)' K } so that there is indeed a non-zero transmitted wave.28) This equation is the condition that every particle is either reflected or transmitted. T. we use the particle flux (2. http://dx.1017/CBO9781139171243. If one solves also for i?. one finds that |i?|2 + \T\2 = \A\2. in this one-dimensional problem it is Downloaded from Cambridge Books Online by IP 150. 2014 .18).org/10. (3.005 Cambridge Books Online © Cambridge University Press.24) in x > a.221. J5 and C in terms of the incoming amplitude A. Applying the boundary conditions at x = 0 and x — a now gives us A + R = 5 + C.doi. (3. where only certain discrete values of E turn out to be allowed. a = «(Be Ka . The tunnel effect 27 be able to satisfy the boundary conditions we must introduce two more unknowns. The complete wave function is then (3.24) and a reflected wave in x < 0. so we need a transmitted wave in x > a.Ce~Ka).244. this is to be contrasted with bound-state problems.23) in 0 < x < a and *P(x) = Aeikx + Re~ikx (3.25) in x < 0. In particular. Notice that there is a solution for any positive value of E. to determine i?. . (3.

in such a way that all remains fixed and equal to A.32) Downloaded from Cambridge Books Online by IP 150. we retrieve (3.4.24) gives j = (hk/m)\T\2.2)). as we expect. where the particles do not have a single speed. (3.306) equal to each other.org/10. so that (3. (3. so it is often simpler to consider a mathematical limit that is easier to work with. we have to solve the time-independent Schrodinger equation (3. The factor hk/m in (3.19) and let a -> 0.1017/CBO9781139171243. or indeed in three.24). so that no particles disappear. is more complicated because some of the particles are moving in one direction.) To find the sta- tionary states.25) gives j = (hk/m)(\A\2-\R\2).30a). where the particle density does not vary with time. this must be independent of rr. was in terms of the simplest possible potential. the form of j does not have such a simple interpretation. in x > a the particle density is \ip\2 = \T\2.doi. see appendix B.31) (For a description of the delta function.28). potentials in one. 2014 . more realistic. In many respects the solutions for other. (3.244. and of bound states in one dimen- sion. the rest in the other.221. with at the same time U -> oo. (3. Consider the square-barrier potential (3. say. dimensions have similar proper- ties.005 Cambridge Books Online © Cambridge University Press. Notice that. applicable in x < 0.306) Putting (3.306) says that.208 on Thu Feb 13 17:10:23 GMT 2014.) In x < 0. the particle flux is equal to the particle density times the speed. The resulting potential is just V(x) = XS(x).306) is the speed of the particles (see the relation between wave number and momentum in (2. see problem 2. http://dx. In a more general situation.28 3 Special solutions In a stationary state. the square barrier or square well. The delta-function potential Our discussion of the tunnel effect. (This particular simplicity of the consequences of the particle-conservation re- quirement is peculiar to one-dimensional problems. according to (3. the solutions for even these simple potentials involve a fair amount of algebra. However. The form (3.30a) and (3.30a) and in x > a (3. These form prototypes for repulsive or attractive potentials in one dimension.

35) and. because such a singularity in \j)" is not balanced by a similar singularity in any other term of the Schrodinger equation (3.(x) is continuous even at x = 0. Then V'O^) would contain a delta-function part (or a worse singularity if the discontinuity of i\) is infinite) and so ip" would contain the derivative of S(x) (or worse). just as in (3.34) f To obtain the correct discontinuity of ip (x) at x = 0.33a) Let the transmitted wave be as in (3. (3. http://dx.2eEi/>(0) -> 0. Hence (-h2/2m) disc V' + A^(0) = 0 (3.mX). we conclude that i/. integrate the Schrodinger equation (3. 2014 .org/10.32). ip'(x) must be discontinuous at x = 0. the value of 1 ip'(x) at x just greater than zero minus its value at x just less than zero.ik(A . The various terms of the Schrodinger equation give A / J -£-e p£ E I dxip(x) . J —£ where disc ip denotes the discontinuity of tpf at x = 0.doi. that is. With (3. However.m\).32) from x — — e to x = +e and then let e —> 0. Downloaded from Cambridge Books Online by IP 150.34).24): i/j(x)=Teikx x>0.336) In order that the wave function be continuous at x — 0. Take the wave function in x < 0 to be an incoming plane wave together with a reflected wave. from the wave function (3.005 Cambridge Books Online © Cambridge University Press.1017/CBO9781139171243. (3.221.25): %l>(x) = Aeikx + Re~ikx.33).36) It is straightforward to check that the particle-conservation condition (3. {-n2/2m)[ikT . A + R = T. (3. we thus find that R = m\A/(ih2k .R)] + XT = 0. For suppose that ij)(x) were discontinuous at x = 0. This cannot be.28) is satisfied. so that ij)"{x) contains a delta-function singularity that balances that of the second term of the Schrodinger equation.208 on Thu Feb 13 17:10:23 GMT 2014. Prom it. T = ih2kA/{ih2k . (3.244. The WKB approximation 29 This equation is to be valid for all x.

which was derived more rigorously by Wentzel. In the re- gion xi < x < x2 we may use (3.2 [2m(V(x) . each of infinitesimal width Ax. where as before K2 = 2m{U-E)/h2. if we follow the first potential barrier by another one for which the corresponding parameters are K! and a'.3.30 3 Special solutions The WKB approximation We have seen how to calculate the tunnelling probability for a square barrier. A smooth.244. Suppose that the width a of the barrier is so small that na » 1. Kramers and Brillouin.396) This is the WKB formula.221. then ln|T/A| 2 will become modified by the addition to it of —2n'a'.208 on Thu Feb 13 17:10:23 GMT 2014.005 Cambridge Books Online © Cambridge University Press.37) and (3. 2014 .386) Because our approximations have led to a result for ln|T/A| 2 that is lin- ear in Ka.38a) and. ln|T/A| 2 « -2na (3. we will not be able to find an exact solution and will have to rely on approximate methods.37) Then the intensity transmission coefficient \T/A\2 (the ratio of the trans- mited intensity to the incident intensity) is approximately (Jf^J>2"a (3. http://dx.org/10. Consider first the transition amplitude (3. if Ka is large enough.1017/CBO9781139171243.27) for a particle of energy E incident on a square-barrier potential of height U.E)/n2] 2 Ax (3. Downloaded from Cambridge Books Online by IP 150. as shown in figure 3. (3.39a) and the transmission coefficient for the whole barrier is approximately dx[2m(V(x)-E)/h2]^\ (3. arbitrarily shaped potential barrier V(x) can be treated as a series of adjacent rectangular barriers.doi.38b) to write the contribution to ln|T/^4|2 from one of these infinitesimal barriers: . How can we adapt this analysis to irregularly-shaped barriers? In general.

The a- particle is held within the nucleus by a strong attractive force. A stationary-state wave function corresponding to positive energy E that satisfies the Schrodinger equation for this potential in the well 0 < x < R and vanishes at x = 0 and x = R.1017/CBO9781139171243.3. and simplify further by pretending that the potential is one-dimensional and that the a-particle moves only in the ^-direction.e~iKX) = 2L4 sin KX (3.40a) oo otherwise. we include the effect of the Coulomb force later.221. Alpha decay 31 V(x) \ \ \ \ Figure 3. So the nuclear-force potential is ' -U 0 < x < i?. http://dx. it is subjected outside the nucleus only to the weaker Coulomb force. is i>{x) = A(eiKX . This force has finite range. A potential barrier of any shape approximated by a series of rectangular barriers. Alpha decay We use the WKB approximation to derive a very simple model of the a-decay of a radioactive nucleus. In this decay.005 Cambridge Books Online © Cambridge University Press. which is composed of two neutrons and two protons. so that if the a-particle escapes from it by tunnelling. 2014 .208 on Thu Feb 13 17:10:23 GMT 2014. the nucleus emits an a- particle. Model the potential corresponding to the nuclear force by a square well.244. V(x) = (3. so that its mass M is about 4 times that of the proton and its charge is 2e.org/10.doi. Consider first the wave function of an a-particle bound within the nucleus by the nuclear force. the nuclear force.41a) with E + U = h2K2/2M such that sin«i? = (3.416) Downloaded from Cambridge Books Online by IP 150. as is required when the potential is infinite outside the well.

as in (3. and so the wave function (3.doi. so it is a reasonable approximation to keep the form (3.005 Cambridge Books Online © Cambridge University Press. the a-particle cannot escape if its energy E is less than Vb.4. Because the potential is no longer infinite for x > i?. With classical mechanics. The current j then Downloaded from Cambridge Books Online by IP 150.41). 2014 .221. and they will even become time-dependent so that we should use the time- dependent Schrodinger equation.24). For very large positive x the potential is extremely small.406) where Ze is the charge of the 'daughter' nucleus that remains after the a-particle has escaped. However. We leave the potential unaltered for x < i?. that is.1017/CBO9781139171243. we can assume that A varies very slowly with t. but in quantum mechanics it can tunnel through the barrier.32 3 Special solutions V(x) E 0 -U Figure 3.4. for energies E much less than Vb the tunnelling probability is small and the wave function in 0 < x < R is changed only a little by the addition of the Coulomb tail to the potential.244.208 on Thu Feb 13 17:10:23 GMT 2014.org/10. We now include the Coulomb force. http://dx. we no longer require the wave function to vanish at x — R. The coefficients of e1KX and e~1KX will no longer have equal magnitude. so that the wave function there is to a good approximation Telkx with h2k2/2M = E. the maximum height V(R) of the barrier. so that it is as shown in figure 3.41) will become modified. so that for x > R the potential be- comes 2Ze2 V(x) = (3. One-dimensional model potential for a-decay.

221. A varies very slowly with t\ in fact P obeys the continuity condition (2.1017/CBO9781139171243.P2 / M\W (3.456) Downloaded from Cambridge Books Online by IP 150. the current at x = R is equal to that at very large x.39b) gives where r = 2Ze2/4neoE is the distance at which the Coulomb potential is equal to E. So we find that the current at large x is If we normalise the wave function suitably. As we have explained. Jo where S is the cross-sectional area of the nucleus in our one-dimensional model. because all the particles that leave the nucleus escape to infinity.17): where here j is the current at the surface of the nucleus.43) =s . http://dx. the probability that the a- particle is within the nucleus is rR dx\iJ>(x)\ 2 = 2S\A\2R.30b).208 on Thu Feb 13 17:10:23 GMT 2014. If the a-particle energy E is small compared with the height of the barrier. Alpha decay 33 is (hk/M)\T\2. The integration can easily be done by making the change of variable x — r 1 / 2 sin#.43) and (3.doi. So from (3.44) we have dP P -* .42). (3.005 Cambridge Books Online © Cambridge University Press. 2014 .org/10.244. as in (3. The WKB approximation (3. that is at x = R (because the potential is taken to be infinite for x < 0 there is no current at x — 0). (3."7 < with /M\2 i 7. the integral is approximately ^Trr1/2. so that r > R. As we have said.

Problems 3. http://dx. What are the new wave functions? [Hint: replace x by a new variable. To obtain the half-life of the nucleus itself. it well describes the observed strong dependence of the half-life of a-particle emitters on the a-particle energy.34 3 Special solutions Integration of (3. It is found that the average distance of the particle from the centre of the well can be rather greater than the width of the well.005 Cambridge Books Online © Cambridge University Press. (ii) a narrow barrier (ka < 1) whose height is equal to twice the kinetic Downloaded from Cambridge Books Online by IP 150. in terms of your sketches? 3. How is this possible.244. Show that the condition that there is just one bound state is that 3. we must divide by the probability that two neutrons and two protons within the nucleus come together to form an a-particle. calculate the n = 0 and n = 2 wave func- tions of the linear harmonic oscillator (leave them unnormalised). Although this model is quite crude. where Po is the value of t at time t — 0.8).org/10. An exponential decay law such as this is typical for unstable systems in quantum mechanics. 2014 . Show that each energy level is reduced by e2£2 /2mu2. 3.] 3. Verify that they are orthogonal. After time r approximately half of the a-particles have escaped from the nucleus. we call r the half-life for an a-particle to remain in the nucleus once it has been formed. so that the potential becomes V(x) = \mw2x2 — e£x.2 A particle of mass m is bound by a one-dimensional potential well of depth U and width b.45a) gives P = Poe~t/T.1017/CBO9781139171243.221.3 A one-dimensional harmonic oscillator carries charge e and is placed in a uniform electric field £.208 on Thu Feb 13 17:10:23 GMT 2014. £ dxipo(x)ijj2{x) = 0.1 Sketch the wave functions for the one-dimensional square-well potential (3. that is.5 A beam of particles is incident on (i) a very deep square-well potential.doi.4 By consulting appendix A.

3. 2014 .doi. what is the half-life of jf Downloaded from Cambridge Books Online by IP 150.org/10.7 A potential V(x) is such that V(x) = V(—x) and each bound-state energy level corresponds to only one independent wave function. Verify that this agrees with that for the square-well potential (3. The a-particle energies are 7.8) in the limit U -» oo. Notice how different these results are from those which obtain in the corresponding situations in classical mechanics. Show that each of these wave functions has either even parity or odd parity. with the same value of E. with Ub fixed and equal to JJL.8 The nuclei IQQFm and 928U have approximately the same radius and both are a-particle emitters. while in case (ii) they are nearly all transmitted. so also is ij){—#). respectively.6 Find the energy level of the bound state for the potential V(x) = —fiS(x).] 3. http://dx. Alpha decay 35 energy of each particle.208 on Thu Feb 13 17:10:23 GMT 2014.005 Cambridge Books Online © Cambridge University Press.1017/CBO9781139171243. Show that in case (i) the particles are nearly all reflected.221. The half-life of ^oFm is \ hour.2 MeV. assuming that the probability of the formation of an a-particle is the same.4 MeV and 4. 3.244. [Hint: the first step in the proof is to show that if ip(x) is a solution of the Schrodinger equation.

http://dx.1017/CBO9781139171243. we get another wave function that describes a possible state.7) and (3. or at infinity. which we normally require to vanish outside some region of space.org/10. By a linear operator Q.1) An example of an operator that is not linear is the operator S that squares the wave function: In chapter 2 we said that one of the basic assumptions of quantum me- chanics is that observables correspond to operators. we mean that (4. the boundary and continuity conditions on the wave function ^ are linear.doi.208 on Thu Feb 13 17:10:33 GMT 2014. Notice that the operators we have met so far. By linear boundary or continuity conditions. as in our examples of (3. This assumption is usually supplemented by the further requirement that the operators be linear. then so does the function (ci^i + c2\I/2). This is evidently the case for wave functions. which appear in the Schrodinger equation. 4 The superposition principle Linear operators In the examples of chapter 3. It is often useful. The reason that this works is that ihd/dt and the Hamiltonian operator if. The consequence of linearity is that if we superpose wave functions that describe possible states of a system.244.18). and in this chapter we explore some of the consequences of this. are linear. where C\ and c2 are any complex numbers.7) and (3.18). we took linear combinations of solutions to the Schrodinger equation and so constructed further solutions: see (3. and. to superpose stationary-state solutions.006 Cambridge Books Online © Cambridge University Press.221. — iftV corresponding to momentum and H corresponding to energy. 2014 . A basic mathematical property that can be proved is that these stationary-state wave functions 36 Downloaded from Cambridge Books Online by IP 150. are linear. we mean that if ^ i and \I/2 are two functions that satisfy them. furthermore.

7)).4) expresses the wave function as a superposition of components having different values of momentum p and energy p2/2m. This means that any allowable # ( r . Wave packets 37 are complete. (4. For finite f2 the superposition (4.3) describes a mathematical limit where the volume Q of the system has been allowed to become infinite. The resulting wave function is not a stationary state. as we explained in chapter 2. 2014 . if we wish to localise the particle in space. the plane wave (4. (4. i ) . can be written as a sum of stationary-state wave functions Wave packets As an example. Setting t = 0. N(p) is a constant. any wave function that satisfies the time-dependent Schrodinger equation and the appropriate boundary conditions for the system. we must take a superposition of the stationary states (4. which is here a continuous variable.208 on Thu Feb 13 17:10:33 GMT 2014. This is because. that is. consider a free particle moving in the ^-direction.221.1017/CBO9781139171243. Rather. there is equal probability of finding the particle at each point x. [dpN{p)eipx/h-ip2t/2mh. As the squared modulus of this wave function is independent of re. The stationary states are the plane waves corresponding to a definite value p of the momentum and a definite value p2/2m of the energy. As Q. we have = [dpN(p)eipx'h.doi.244. To construct a wave function corresponding to a particle that is localised in space.org/10.3). http://dx. it does not have definite momentum or energy.4) Here N may take a different value for each value of p: it is now a function of p. These are labelled by the momentum p. —> oo this sum becomes an integral: .2) would be a sum over discrete values of p (see (3.006 Cambridge Books Online © Cambridge University Press. (4.5) Downloaded from Cambridge Books Online by IP 150.

N(p) is the Fourier transform of $(x.6).0)| 2 is localised around the point x = x$. Inserting the result into (4.4).244.38 4 The superposition principle That is.0)|2 = 1. http://dx.1017/CBO9781139171243.doi.7) is called a wave packet: the probability density |^(x. (4.8) we find that. With (4. suppose that we take a Gaussian: *(ar. 2014 . 0) = No e-(*-*o)2/a2+iPox/hm (4. As a simple example.0) (see appendix B).7) Here NQ is a normalisation constant. By inverting the transform. chosen so that Jda.0)\2=x0. at t = 0. A measure of the width of their distribution is the quantity Ax.221. which requires that N0 = (ia a *).1 / 4 - The wave function (4. we can calculate N(p).006 Cambridge Books Online © Cambridge University Press.7) into (4.208 on Thu Feb 13 17:10:33 GMT 2014.9) Ax is called the width of the wave packet.8) The results of the separate measurements will be scattered around this average value.|\I>(a. we can then find the wave function # at other times t. (4. Ax = \a. If we prepare the state a large number of times and each time measure the position of the particle at t = 0.0): N(p) = ^jdx^{x^)e-'^lh.org/10.0)\2. we obtain an integral whose evaluation is discussed in appendix B. defined by (Ax)2= [dx(x-(x))2\y(x.. we may express N(p) in terms of \I>(:r.6) Hence if we know the wave function \fr at t = 0. because |^| 2 is the probability density the average value will be (x) = Jdxx\V(x. (4. It gives 2 2 2 N(p) = Downloaded from Cambridge Books Online by IP 150. If we insert (4.

If we prepare the state many times and measure the momentum each time.doi. in this case the integral (4. in this sense a Gaussian is the best possible shape in that it gives Ax Ap = \h rather than greater than \7i. Hence at t = 0 AxAp= \U. where d>(p) = [l/(2nh)1/2]N(p). defined by (4. EhrenfesVs theorem 39 This has the same functional form as #(£. The interpretation is that.12) With (4. the probability of obtaining a result in the interval (p.9) diverges and Ax = 00. since |\I>|2 has the same value for all x. The allowed degree of accuracy depends on the shape of the wave packet.244. Downloaded from Cambridge Books Online by IP 150. we find that Ap = h/a. http://dx.006 Cambridge Books Online © Cambridge University Press. p + dp) is just \(j){p)\2dp. There is a general principle. a property that is peculiar to the Gaussian.221. (4.1017/CBO9781139171243. the average value so obtained will be (p) = Jdpp\ct)(p)\2=po.208 on Thu Feb 13 17:10:33 GMT 2014.0).11) and the distribution of values obtained will have a width Ap defined by (Ap)2 = Jdp(p-(p))2\ct>(p)\2. (4.13) Ax and Ap are the uncertainties in the position and momentum associated with the wave packet. the momentum takes a definite value. just as |\I>|2 is the probability density in position space.7). satisfy AxAp> \Tt. so Ap = 0. One can check that J dp\(/)(p)\2 = 1. Ax and Ap.12). However. That is. so \(f)\2 is the probability density in momentum space. This means that we cannot know the position and momentum together to better than to a certain degree of accuracy.org/10. (4. which says that for any wave function. the Heisenberg uncertainty principle. 2014 . if we measure the momentum.9) and (4. Notice that in the case of a plane wave.

that states that whatever the shape of the wave packet.14a) and (d/dt)(p) = <-W>. For a sufficiently narrow wave packet. t) for t / 0 (see problem 4.4). and the width Aa: of the wave packet increases with increasing time.org/10.doi.221.40 4 The superposition principle Ehrenfest's theorem Having obtained N(p) in (4.t) remains Gaussian in shape at subsequent times t.244. 2014 . Ap of the wave packet in position and momentum are small.1017/CBO9781139171243. (4.146) where = /*d3rW|«f (see problem 4. This does not mean that it describes a classical situation because. (d/dt)(r) = (p)/m (4. it is necessary that if a classical description is to be a good approximation. The result is that ^(x. the quantum- mechanical description is very close to the classical description if the widths Ar. we may now use (4. as we have seen. The reason that the wave packet spreads out as t increases is that the different momentum components of the wave function correspond to different velocities p/m. L and P will be parameters some- how typical of the system under investigation.2). Crudely speaking. The average position (x) of the particle moves with velocity po/m.4) to calculate *(x. This statement is not precise because we have not specified a length L and a momentum P with which we must compare Ar and Ap in order to decide whether they are indeed small.14) show that the wave packet as a whole moves like a classical particle. Because we know from the uncertainty principle that ArAp > | / i .208 on Thu Feb 13 17:10:33 GMT 2014. http://dx.006 Cambridge Books Online © Cambridge University Press. ( W ) is almost equal to the value of W at r = (r). and for a general potential V.10). the momentum and position of the particle cannot be simultane- ously known with absolute certainty. Then the two equations (4. There is a theorem. due to Ehrenfest. Downloaded from Cambridge Books Online by IP 150.

Hermitian operators 41 Hermitian operators We have said that it is assumed in quantum mechanics that to each ob- servable Q there corresponds a linear operator Q. Let ^ i and ^2 be any two of the allowed wave functions for the system under discussion. as usual. so that Just as in (2.org/10. we may transform this to an integral over the surface enclosing the volume of integration and conclude that for a closed system.166) As an example. (4-17) where Downloaded from Cambridge Books Online by IP 150. as we now show. An important property of a Hermitian operator is that its eigenvalues are real. and we shall not discuss it.221.doi. We now introduce an additional assumption about this operator: that it is Hermitian.15) In this definition the integration is. where the wave functions vanish outside some volume £2o5 the integral van- ishes. We write (4. To show this. for all choices of \&i and #25 *.006 Cambridge Books Online © Cambridge University Press. since the potential V is real. the Hamiltonian operator H is Hermitian.208 on Thu Feb 13 17:10:33 GMT 2014. is a little more subtle. then (\I/. (4. The operator Q is said to be Hermitian if. if fd3r^l(Q^2) = /d3r(Q#i)*#2.1017/CBO9781139171243. over all space. 2014 . For if Qty = g\I>.16a) that is. corresponding to the mathematical limit fi0 ->• 00. As we explained in chapter 2. Qty) = q{*&.244. http://dx. where q is one of the eigenvalues. an unbounded system. (4. we use the fact that V = V*.15). \fr).

By this we mean that any allowable wave function \I> of the system can be expanded as a linear combination of the eigenfunctions \&r of Q: #r. we cannot conclude that \I>i and \&2 are orthogonal. However. which is not altogether straightforward to check in practice. Because.42 Jf.18) where 8rs = 1 if r = s and 8rs = 0 if r ^ s. we write the chain of equations q2 ( Here we have used (4.221. the eigenvalues of the operator Q corresponding to an observable Q are possible values obtained in measurements of Q.19) Downloaded from Cambridge Books Online by IP 150.006 Cambridge Books Online © Cambridge University Press. we then have an orthonormal set: (4.16a) for the case \&i = Hence.244. \P) > 0. So if q\ ^ q2-> we have the required result. q = q*. A basic mathematical assumption. so that it satisfies ( ^ . the result is one of the eigenvalues of Q.org/10. http://dx. in this case any linear combination (ci^i +02^2) is also an eigenfunction of Q.208 on Thu Feb 13 17:10:33 GMT 2014. it is natural to impose the restriction that Q be Hermitian so as to ensure that these values are real. as we explained in chapter 2. since by definition Q^i = qi^i and Q\&2 = #2^2. and we can choose two linear combinations that are orthogonal (see problem 4. since (\P. ^ ) = 1.7). with the same eigenvalue. If we normalise each eigenfunction.17) and use (4. is that the set is complete. We shall argue below that whenever we measure <2. The superposition principle Take the complex conjugate of (4. Another important property of a Hermitian operator is that two eigen- functions \&i and ^ corresponding to different eigenvalues qi and #2.doi. are orthogonal.1017/CBO9781139171243.16a) and the fact that qi is real. If qi = q2. By this we mean that To show this. (4. 2014 .

* ) | 2 . However. (4-20) To verify this.21) Operators and observables We explained in chapter 2 that it is assumed that when the value of an observable Q is known to be g. .*) = c . and that the natural time development of the system has not resulted in a change in the value of Q. http://dx. We say that the system is then in an eigenstate of Q. the first measurement must have left the system in an eigenstate of Q and the result of the measurement is the corresponding eigenvalue q. Operators and observables 43 If we want \I> to be correctly normalised. use the orthonormality condition (4.19) of \I> in terms of the eigenfunctions \I>r.1017/CBO9781139171243.doi. From this condi- tion we find also that (*. which eigenstate the system will jump into because of our act of measurement. we know what will be the result of the second measurement and so. is P r = |c r | 2 = | ( * r . This assumption implies that any measurement of Q will leave the system in an eigenstate of Q. we can calculate the probability of obtaining a particular eigenvalue. So the first measurement must have disturbed the system. we expect to get different answers. so that (#. that is. (4. 2014 . The result of measuring Q is one of the eigenvalues of Q but.22) Downloaded from Cambridge Books Online by IP 150. unless ^ happens to be an eigenstate of Q.org/10. the system will not be in an eigenstate of Q before the first measurement. *) = 1. The two measurements must be close enough in time so that we can be sure that no outside influence has disturbed the system. before any measurement of Q is made.18). One would guess that this probability is related to the coefficients cr in the expansion (4. we cannot predict which eigenvalue it will be. Suppose that.208 on Thu Feb 13 17:10:33 GMT 2014. the system is in a state whose wave function is \I>.221.. This follows if we assume that quantum physics is not so capricious that. with eigenvalue q. the coefficients cr will satisfy X>| 2 = 1. according to the assumption. In general.006 Cambridge Books Online © Cambridge University Press. The most natural assumption is that the probability Pr of throwing the system into the eigenstate * r by making the measurement. in rapid succession. if with suitable care we make the same measurement of Q twice. and so obtaining the eigenvalue qr as the value of Q. After the first measurement. (4.244. the system is in a state whose wave function is an eigenfunction of the corresponding operator Q. so as to throw it into an eigenstate of Q.

(4-23) That is. with eigenvalue q. simply multiplies any wave function on which it acts by r . because of the normalisation condition (4.006 Cambridge Books Online © Cambridge University Press.25) The operator f.25) for (p) is equivalent to the definition (4. It is often written (Q). \£)| 2 . the expectation values of position and momentum: (r)= /d3r**(r#)= /d3rr|*|2.44 4 The superposition principle Then. the probability that a measurement of Q will give the eigenvalue qr is Pr — |(\t r . (b) If the value of Q is known to be g. http://dx.18). When the system is in the state \I/.19) of the c r . (c) The result of measuring Q can be any of the eigenvalues of Q. (p) = /d 3 r#*(p#) = f d3ry*(-ihVV). the probability of getting some result for the measurement is 1. 2014 .221.doi.11) in terms of (f)(p).1017/CBO9781139171243. and the orthonormality condition (4. the system is in an eigenstate of Q. Let us now recapitulate our basic assumptions about observables: (a) To each observable Q there corresponds a linear.208 on Thu Feb 13 17:10:33 GMT 2014. If we somehow prepare the same state ^ many times. and on each occasion measure Q at the same relative instant in time.org/10. Downloaded from Cambridge Books Online by IP 150.244. corresponding to position. Earlier in this chapter we have already encountered two particular examples. Hermitian operator Q. the average answer will be If we use the definition (4.20). (4.24) This is called the expectation value of Q in the state \I/. we find that J] < (4. It may be verified that the expression (4.

This is another manifestation of the uncertainty principle.org/10.13) for Ax Apx to an arbitrary pair of observables can be shown to be (see problem 4. However. The first measurement will throw the system into an eigenstate of Q. some pairs of observables may simultaneously be measured ex- actly. one cannot know the exact values of two observables Q and R simultaneously. 2014 .208 on Thu Feb 13 17:10:33 GMT 2014. We then say that the operators Q and R commute.R] = QR-RQ (4. if we operate on any wave function first with Q and then with R. Thus. which we have already encountered for the particular case of the observables r and p.27) On the right of this inequality there appears the modulus of the expectation value of the commutator of the corresponding operators.doi. in the state for which the measurement is being made. of course. To check this. it turns out that the operator is a simple number: [xi. http://dx. their commutator [Q. itself an operator. the result is the same as when we apply the operators in the reverse order. we lose the information we have about Q. For the particular case of components of position and momentum.244. because generally their eigenstates are not the same. introduce an arbitrary wave function \I> and let the commutator operate on it: Downloaded from Cambridge Books Online by IP 150.28) where i and j label the different components. and only if. Commutators 45 Commutators Suppose that we measure an observable Q and then another observable R.pj] = ih6ij (4. So the condition that we may simultaneously know the values of two observables is that the corresponding operators commute.1017/CBO9781139171243.26) vanishes. That is. The generalisation of the uncertainty principle (4. In general. this will not be an eigenstate of R.006 Cambridge Books Online © Cambridge University Press.15) (4. and <Jy = 1 if i — j and 0 if i 7^ j . so that the second measurement will disturb the system again. It can be shown that two operators Q and R have a complete set of common eigenstates if. by measuring i?. In general.221. The commutator is.

27) then agrees with (4.8 Show that the momentum operator — i W is Hermitian for a bounded system.7) at t = 0 by taking expectation values of the appropriate operators.] 4.13): we cannot measure the exact values together.46 4 The superposition principle Thus the two sides of the relation (4. and even controversial. these produce second-order effects on the accuracy. Downloaded from Cambridge Books Online by IP 150. subject.6 A dart of mass 1 kg is dropped from a height of 1 m.28) is ift.208 on Thu Feb 13 17:10:33 GMT 2014. 4.244. we can measure simultaneously the exact values of a given component of r and a different component of p. 4. Find a normalised linear combination of them that is orthogonal to ipi. show that this is only of the order of a tenth of a nuclear diameter.5 Show that both (x) and (p) are zero for a one-dimensional harmonic os- cillator and use the uncertainty principle to deduce that the minimum energy must be greater than. equations (4. and so the general uncertainty relation (4. When we take the same components of r and p. 2 4. 4. Problems 4. the values of (p) and (p2) calculated from the momentum-space probability density |^(p)|2 are equal to the values obtained from the expectation values of the cor- responding operators using the coordinate-space wave function tp(x).28) vanishes when i ^ j . Calculate 3/(z.221.4 Prove Ehrenfest's theorem. the intention being to hit a given point on the ground below.1017/CBO9781139171243. which is the same thing as saying that (4. 4. for the one-dimensional case where the wave function depends on x and t only. Because the commutator (4. the commutator (4. our discussion has necessarily been brief.4) far enough to verify that (x) — x0 + pot/m.org/10. The uncertainty principle imposes a limitation on the accuracy that can be achieved.3 Show that. (Ax) = \a2 + h2t2/m2a2. http://dx. [Neglect the uncertainties in position and momentum in the vertical direction. The theory of measurement in quantum mechanics is a deep. and satisfy (/0ij'02) = c.£) from (4.10) and (4.14).28) are equal when applied to any wave function #.doi. or equal to.28) is a true identity.2 A Gaussian wave packet is given at t = 0 by (4. 4. 2014 .006 Cambridge Books Online © Cambridge University Press. \hw. for any one-dimensional wave function.1 Calculate (p) and Ap for the Gaussian wave packet (4.7).7 The wave functions ipi(x) and -02(x) are normalised.

14 Show that for a free particle _d 2 _ (pxx + xpx) d2 2 _ 2(p|) Downloaded from Cambridge Books Online by IP 150. Find a complete orthonormal set of eigenfunctions of Q.10 Q is a Hermitian operator and the wave function * is such that Qfy = q\&^ where q is a number. Show that for any solution \& of the time-dependent Schrodinger equation 4.13 The operator Q does not depend explicitly on t. 2014 . 4.doi.org/10.208 on Thu Feb 13 17:10:33 GMT 2014. 4. [i.9 Use the divergence theorem (page 15 footnote) to show that in a bounded system the expectation value of the one-particle kinetic- energy operator is positive.11 Show that. Show that 4. The system is left undisturbed and after a time t the observable is measured again. The observable is measured and is found to have the value +1. B]] = 0. $ is any other wave function.1017/CBO9781139171243. with different eigenvalues En.221. C]] + [S. The operator Q corresponds to an observable and is such that Wi=fe Q^2 = ^u Q*Pn=0 forn>3. (The second relation is called the Jacobi identity. Commutators 47 4. A]] + [C. [C. deduce that for a particle moving in a potential well of arbitrary shape the lowest bound state has an energy level that is always higher than the bottom of the well. Calculate the probability that the value +1 is found again. [A.244. http://dx.) 4. [B.12 A quantum system has a complete orthonormal set of energy eigen- states V>n. for any linear operators A.006 Cambridge Books Online © Cambridge University Press.B and (7. By considering the expectation value of the Hamiltonian.

Downloaded from Cambridge Books Online by IP 150. *) = ( The operators Q and . Show that: i (i) if F is Hermitian. Show that the expectation value of (Q + i\R){Q-i\R) in any state * is real and > 0.org/10.doi. then FT = F.221.48 4 The superposition principle 4. 2014 . (iii) (FG)t = GtFt. Hence prove that and so deduce the general uncertainty principle (4.R correspond to observables and A is any real number. (ii) (F + iG)t =Ft-i(?t.F#i)*.15 We define the Hermitian conjugate FT of an operator F such that. F t * 2 ) = (*2.244.1017/CBO9781139171243. (iv) if $ = F * then ($.27).208 on Thu Feb 13 17:10:33 GMT 2014. for any pair of allowed wave functions of the quantum system under discussion. http://dx.006 Cambridge Books Online © Cambridge University Press. ( * i .

This is because the system will have undergone a natural time development.221. If this state is also an eigen- state of H. if we measure an observable Q at two different times. (In addition. we shall obtain different eigenvalues of Q.H] = 0. the system may have been disturbed from outside.) Orbital angular momentum An observable that is often useful for labelling stationary states is the orbital angular momentum. we throw the system into an eigenstate of the corresponding operator Q.007 Cambridge Books Online © Cambridge University Press. If we measure the observable Q. 5 The hydrogen atom Good quantum numbers In general.13.9). The stationary states are the eigenstates of the Hamiltonian operator H. (See also problem 4. because it does not change with time.1) In these circumstances we say that the observable Q is a good quantum number. 2014 .244. http://dx. The calculation of 49 Downloaded from Cambridge Books Online by IP 150.208 on Thu Feb 13 17:10:44 GMT 2014. we must have the condition [Q. In quantum mechanics the corresponding operator is usually called TiL.2) An important property of the operator L is that its square is closely related to the operator V2 which appears in the Hamiltonian. in order to ensure that the eigenstates of Q and H are the same. for which the particle density |\fr|2 is constant in time at each point r. Its value can be used to label a stationary state.1017/CBO9781139171243.doi.org/10. for example by a measurement of some other observable. governed by the time-dependent Schrodinger equation. its time development is trivial and a subsequent measurement of Q will give the same result as the first measurement. According to the discussion given at the end of the last chapter. In classical mechanics the orbital angular momentum of a particle about the origin of coordinates is r A p. (5. and it is given by (5.) The only states whose natural time development is trivial are the station- ary states (2.

If we now define the polar axis 9 = 0 of the spherical polar coordinate system to lie along the ^-direction. Downloaded from Cambridge Books Online by IP 150.221.5) f For the calculation of V in spherical polar coordinates.doi. I? is described in appendix C and the result is that 2 d i d (5. (Cambridge University Press). 2014 . In spherical polar coordinates. (5.1). for example H Jeffreys and B S Jeffreys. Spherical polar coordinates.! r. see any textbook on vector calculus. 50 5 The hydrogen atom Figure 5.3) sin 0 where 6 and (j) are polar and azimuthal angles of spherical polar coordi- nates (figure 5. http://dx.2 V 1 (5.007 Cambridge Books Online © Cambridge University Press.2) of L that its z-component takes the simple form Lz = —i d/d4>. it can be shown from the definition (5.1017/CBO9781139171243. Methods of Mathematical Physics.1.4) =- so that the part of V 2 that involves the angles 9 and </> is just —L2/r2.208 on Thu Feb 13 17:10:44 GMT 2014.244.org/10.

and also to ensure that any linear combination of the wave functions Yim gives a probability density that is single valued (the probability density should not change when <p is increased by 2?r.8) is free from infinities.3). tells us that Y/m varies with <j> simply as eim<f>. a given eigenvalue 1(1 + 1) for L2 corresponds to the square of the angular momentum being equal to U2l(l + 1).LZ] = 0.3). That is. .007 Cambridge Books Online © Cambridge University Press.org/10. Using the second equation of (5.6). for a reason that will become ap- parent shortly (see (5.9) below).6) That is. That is. we find that 1 d sin0d0 The solutions of this equation. since values of <f> differing by 2TT correspond to the same point r ) . This means that L2 commutes with Lz. where m = — /.208 on Thu Feb 13 17:10:44 GMT 2014. m is the eigenvalue of Lz but. m for L z .244. However. and so they do not then yield a physically accept- able wave function. .1017/CBO9781139171243. where P[n(9) is some function of 9 only. I — 1. together with (5. 2014 . Orbital angular momentum 51 According to (5. . only d/d(/). that is. The first equation in (5. which is known as Legendre's equation. the eigenvalue of L2 is written as 1(1 + 1). [L2. . . — I + 1 . so that the two operators have simultaneous eigenfunctions. For most values of / and m the solutions are infinite for 0 = 0 or TT.doi.6).9) and then one of the two solutions of the differential equation (5. http://dx. where the labels I and m are defined in terms of the eigenvalues of L2 and Lz as follows: L2Ylm=l(l + l)Ylm. we have the eigenvalues 1(1 + 1) for i 2 . have been well studied. It is usual to denote these by Y/m(0.221. . L2 does not involve <j> explicitly. where I = 0 .5). we must choose / to be a non-negative integer and restrict m to integers from —I to /. the usual terminology is to call the value assigned to / the orbital angular momentum Downloaded from Cambridge Books Online by IP 150.</>). together with (5. . . It is found that in order to avoid infinities. (5. Strictly. 1 . 2 . or / (5.

doi. the value of / will remain unchanged but the system will have been disturbed in such a way that the information about Lz is lost. . In quantum theory even a point particle may have spin. we can predict the probability of obtaining any of the possible values ra'. Instead of the simultaneous eigenfunctions Yim of Lz and Z 2 .221.org/10. Then it is easy to show that La commutes with L2. http://dx. but we shall not describe this calculation here. one can work with the analogous simultaneous eigenfunctions of La and Z 2 . we will have determined a value m' for this new component. apart from orbital angular momentum. There is in quantum mechanics another type of angular momentum.52 5 The hydrogen atom of the state. where m' = —/. it may also be half-integral.007 Cambridge Books Online © Cambridge University Press. and the possible eigenvalues of La are m'. not associated with the motion of a particle. for the case of electrons Downloaded from Cambridge Books Online by IP 150. that is. 2014 .244.1017/CBO9781139171243. The value of m is called the magnetic quantum number. Instead. This is an intrinsic angular momentum. . (The only exception is when there is no orbital angular momentum. However. It turns out that in the case of spin angular momentum the restriction on the quantum number / is a little less severe: in addition to possibly being integral. so that this new direction could equally well have been called the z-direction and the result follows because we already know that Lz commutes with L2. with m an integer between —/ and +1. . .208 on Thu Feb 13 17:10:44 GMT 2014. It is not possible. to have certain knowledge simultaneously of more than one component of the angular momentum. In particular.—/ + 1 . and we cannot simultaneously assign definite values to the corresponding observables. / — 1 or I This means that it is possible to know simultaneously the exact values of both the orbital angular momentum I and of any one of its components. and it is given the name spin. there is nothing special about the z-direction.) Suppose that a measurement of the square of the orbital angular momentum gives the value h2l(l + 1). If we know the value of I and the original value of m. Notice that [L^Lj] ^ 0 for i =fi j . Then all components of the orbital angular momentum are zero. Let La be the component of L in any other direction. Hence two different components of L do not have simultaneous eigenfunctions. however. where m! also is some integer between —/ and +/. This can be done by direct calculation or by realising that any three orthogonal directions can be chosen for the coordinate axes. If we now measure some other component. and that a measurement of the ^-component gives the value mh. the reason for this is that effects resulting from the quantisation of the z- component of the angular momentum are observed when atoms radiate photons in a magnetic field which points in the ^-direction. / = 0.

1017/CBO9781139171243. The corresponding possible values for the magnetic quantum number associated with the spin are m = ± | .007 Cambridge Books Online © Cambridge University Press. to avoid confusion with the magnetic quantum number m. and we can use these values to label the stationary states. we have to solve the time-independent Schrodinger equation & {l& h) (5.) Hence for the case of a spherically symmetric potential [H.) In addition to their spin. Similarly.3)).org/10. and not on the angles 6 and fa so that V = V(r). Making use of (5. That is.Lz] = 0 (5.11) We look for solutions i/j(r) that are simultaneous eigenfunctions of L2 and Lz. [V. Spherically symmetric potentials 53 and protons the spin angular momentum has / = | . Then [V.208 on Thu Feb 13 17:10:44 GMT 2014. L2 and Lz. [H. and that depends on the direction of r as well as on its magnitude.4). but not d/dr (see (5. We discuss spin further in chapter 8.221. and this is always true for all electrons and all protons.L2] = 0. because this operator is also spherically symmet- ric: p2/2M — (p2 + py +p2z)/2M involves each of the three coordinate directions equally. we say that these particles have spin | . on the other hand.doi. as we have already described. we can find stationary states where L2 and Lz take definite values.12) Downloaded from Cambridge Books Online by IP 150. (We have written the particle mass as M. so that their angular dependence is 1>(r) = R{r)Ylm(d. Spherically symmetric potentials The properties of the orbital angular momentum are of particular im- portance in the analysis of the motion of a particle in any potential V that is spherically symmetric. it depends only on the spherical polar coordinate r. (5.10) and we can find simultaneous eigenfunctions of H. This result is not true in the case of a potential V(r) that is not spherically symmetric. http://dx.it>). (Photons. the particles may carry orbital angular momentum. The operators L2 and Lz also commute with the kinetic-energy operator. If V is spherically symmetric about the origin. because L2 involves only the derivatives d/d6 and d/d(j). have spin 1. L2] = 0. 2014 .244. Lz] = 0.

Thus it has the effect of an additional potential.13) and look for a solution for the electron wave function Downloaded from Cambridge Books Online by IP 150. In the differential equation (5. so that the repulsive centrifugal potential is then absent. and replacing L2 by its eigenvalue /(/ + 1).17) that this is indeed the case for the hydrogen atom. moving in the Coulomb field of the fixed proton.007 Cambridge Books Online © Cambridge University Press. though for some choices of the potential V(r) the ground state can correspond to a larger value of /.13) for R(r).doi. (5.org/10. http://dx. the term involving 1(1 + 1) is simply added to V(r). we find that R(r) satisfies the differential equation Because m does not appear in this equation.9) all have the same energy. for a spherically symmetric potential.54 5 The hydrogen atom where Y\m is defined by (5. which is repulsive. Inserting this structure (5. It is known as the centrifugal potential. We insert this potential into (5. For a rotationally symmetric Hamiltonian. This field corresponds to the potential V(r) = -e2/47T£0r.1017/CBO9781139171243. with the tightest binding) corresponds to / = 0. for each value of / the states with the (21 + 1) possible different values of m given in (5.6) and R(r) is some function that remains to be found.11). by analogy with classical mechanics. the electron.14) where we have taken the proton to be at rest at the origin.12) into (5. Since the proton is nearly 2000 times heavier than the electron. That is.244.221.208 on Thu Feb 13 17:10:44 GMT 2014. the state of lowest energy. The reason for this degeneracy is that. the proton is then almost at rest and we may as a good approximation treat it as if it were. Usually we expect to find that the ground state of a system (that is. The approximation has reduced the problem to that of a single body. no direction in space is physically different from any other. Suppose that the centre of mass of the atom is at rest. We shall see below in (5. We say that these states are (2/ + 1)- fold degenerate. The hydrogen atom The hydrogen atom consists of an electron of charge — e bound to a proton of charge +e. 2014 . the eigenvalues E obtained from the equation will depend only on / and not on m. the energy cannot depend on the magnitude of the angular-momentum component in any particular direction. We explain how to relax this approximation later.

. Bohr's theory turned out to be too simple and restrictive in its applications. It is interesting that the result (5.l.15) Then (5. Schrodinger and Dirac. (5. just as in the case of the one-dimensional square well (chapter 3).. The ground state corresponds to n = 1 and therefore I = 0. and from (5.2. Z = 0. using what is now known as 'old-fashioned' quantum theory. When / = 0..14) is zero at infinity.17). The hydrogen atom 55 that corresponds to a bound state. we therefore expect that the energy E will be negative.13) becomes We show in appendix A that this has solutions such that R(r) —> 0 at infinity and the behaviour at r = 0 is suitable if. So we write E = -ftV/2Af. http://dx. His work appeared some twelve years before the foundations of modern quantum theory were laid by Heisenberg. We show in appendix A that the ground-state wave function is Downloaded from Cambridge Books Online by IP 150.208 on Thu Feb 13 17:10:44 GMT 2014. Me2/AKSQKTI is a positive integer n..1017/CBO9781139171243. (5.2.7) and (5.17) for E was first obtained by Bohr.3. and only if. we must have m = 0. In a bound state. but it does happen to give the correct result for the bound-state energies of the hydrogen atom. the electron does not have enough energy to escape to infinity. The ground state of the atom corresponds to n = 1 and I = 0 in (5. For a bound-state solution.doi.. Bohr postulated that the electron moves round the proton in a classical circular orbit. so that the electron is localised somewhere near the proton.17) The integer n is known as the principal quantum number.244. That is..(n-l). 2014 . Me 4 1 E & ^ " = 1..007 Cambridge Books Online © Cambridge University Press. and imposed a quantisation condition which requires that the orbital angular momentum be an integral multiple of h. we impose the condition that R(r) -> 0 as r —> 00. Because the potential (5.. and / is restricted to integers from 0 to (n — 1).8) one can see that Foo(#50) is a constant (remember that Ybo is the solution that is everywhere finite).221.org/10.

the Pauli exclusion principle. we find in (5. (For an atom with very many electrons. As we have explained. and so obtain the configuration of the ground state of the atom. In the proper quantum theory the expectation value of the radius of the atom in its ground state i S o / err |\I/f r = 4n drr6 3— = | a 0 . until we have allocated all the electrons. Each electron interacts with the nucleus through a potential —Ze2/A-KE^r so that if we neglect the interactions between the electrons the possible electron energy levels are obtained from (5.17) an additional degeneracy. the electron has spin | . Many-electron a t o m s Consider now an atom of atomic number Z > 1.244. the discussion is more compli- cated. states that no two identical particles of spin | may occupy the same quantum-mechanical state. so that the charge on the nucleus is Ze and there are Z electrons when the atom is electrically neutral.208 on Thu Feb 13 17:10:44 GMT 2014.221. </>) there are two different electron states. We fill in the levels in this way.17) by making the replacement e2 —> Ze2.17).1017/CBO9781139171243. because then m can take any of the three values . n = 2. 0 . http://dx. and six with / = 1. This degeneracy is peculiar to the Coulomb potential. (5.) Downloaded from Cambridge Books Online by IP 150. which is found for any spherically symmetric potential. 2014 . and so it is known as accidental degeneracy. In the next level. because the ^-component of the electron spin can be ei- ther + | or — | .007 Cambridge Books Online © Cambridge University Press. Jo ™o Notice that in addition to the (21 + l)-fold degeneracy with respect to ra. + 1 . Hence in the ground state of the atom. that is.56 5 T/ie hydrogen atom where . it is known as the Bohr radius.18) It happens that ao is the radius of the ground-state orbit of the electron in Bohr's model. starting from the lowest. n = 1. Cor- responding to each wave function R(r)Yim(6'. there can be two electrons with / = 0. The neglect of the electron-electron interaction turns out to be a fair approximation if Z is not too large.1 . because the energy of the interactions between pairs of electrons cannot then be neglected. two electrons occupy the lowest-energy level (5. For each value of n.org/10. A basic principle of quan- tum mechanics. Z = 0. we have the same value of E for the n different values of /.doi.

244.007 Cambridge Books Online © Cambridge University Press. Because V does not depend on R. as one would do in solving a two-body problem in classical mechanics: so that r is the relative separation of the two particles and JR is the position of their centre of mass. we must introduce a wave function ip{ri. we assumed that it is a good approxima- tion to regard the proton as being at rest.1017/CBO9781139171243. Hence the Hamiltonian operator takes the form H = -(ft 2 /2Mi)V? _ (h2/2M2)V22 + V(rx . 2014 .r2) that depends on the coordi- nate ri of the electron and also on the coordinate r 2 of the proton (see problem 2.208 on Thu Feb 13 17:10:44 GMT 2014. http://dx. (5. (5.org/10.19) may be reduced with a little algebra to the form H = -{h2/2M)V2R . the time-independent Schrodinger equa- tion H<ip{rur2) =Et has solutions of the form where Downloaded from Cambridge Books Online by IP 150.19) where V\ operates only on the coordinate r\ and V 2 operates only on the coordinate r 2 . We then wrote the Schrodinger equation for the wave function ip{r) for the electron. If we do not make this approximation. for a system of two interacting particles the potential will be a function of the separation between the particles. In terms of these variables.5).(h2/2»)V2r + V(r).21) where M = M1+M2. n = M1M2/{M1 + M2).doi.19) the first two terms are the separate kinetic-energy operators for the two particles. which reduces the problem to the single-body case. (5. Generally.221. It is convenient to change to centre-of- mass variables. In (5. (5.22) The quantity \x is known as the reduced mass of the system.r 2 ). Two-body systems 57 Two-body systems For the hydrogen atom at rest.

such as the electron. Downloaded from Cambridge Books Online by IP 150. (5. This result is precisely that obtained in classical mechanics. there is good agreement. As we explained in chapter 1. Hence in expression (5. The orbital motion of the electron generates a magnetic field. http://dx.17). in rather crude terms this is because a charged particle moving in a closed orbit is equivalent to a current loop. If the system is at rest. because the mass of the electron is much smaller (by a factor of about 2000) than that of the proton. The most important of these terms arises because the electron has both orbital and spin an- gular momentum.17) for the energy levels of the hydrogen atom. 2014 . it leads to a small correction to the energy levels (5. When the result (5. The energy of interaction between this magnetic field and the electron's dipole moment must be included in the Hamiltonian.221. is compared with experi- ment. and when all corrections have been made the calculations agree with experiment to better than one part in 106. E1 = 0. However. As we explain in chapter 8.doi.244.23) says that the centre of mass R of the system has a motion equivalent to that of a single particle of mass M moving freely. the com- parison is made by examining the emission or absorption spectrum of the atom.17). and the magnitude of this dipole moment can be calculated with high accuracy.org/10.208 on Thu Feb 13 17:10:44 GMT 2014.23) The second equation in (5.007 Cambridge Books Online © Cambridge University Press. necessarily carries an intrinsic magnetic dipole moment. /z « M. modified in this way. in zero potential. There are other effects of a relativistic origin.1017/CBO9781139171243. the mass M of the electron should be replaced by the reduced mass \i of the system. The main discrepancy between theory and experiment may be ac- counted for by including in the Hamiltonian magnetic interaction terms in addition to the electrostatic Coulomb potential. For a given value of n.58 5 The hydrogen atom E + E' = Etot. The first equation states that the relative motion of the two particles is the same as the motion of a single particle of mass \i moving in the potential V. It can be shown from relativistic quantum mechanics that a charged particle of spin | . the states with different values of / are then not ex- actly degenerate.

so that we indeed have a bound state. J" r >a (5. so that the repulsive centrifugal potential is absent.007 Cambridge Books Online © Cambridge University Press. as in (5. According to (5.221. and treat this as in (5.208 on Thu Feb 13 17:10:44 GMT 2014. the relative motion of the proton and the neutron is described by the Schrodinger equation \-V]ip{r) =E^{r).24) Suppose that V depends only on \r\ — r. (5. As a simple model for the potential.1017/CBO9781139171243.26) v v . we want both R(r) and Rf(r) to be continuous. v } where E = -h2k2/2^ E + U = h2K2/2fi. (5.11). http://dx. This gives the equation k = -K cot Ka (5. At r = a.13). satisfy the Schrodinger equation at the point r = 0. We explained that we expect to find that the ground state of the system. It is a bound state of a neutron and a proton. the reduced mass of the system is /i « | M . The deuteron 59 The deuteron The nucleus of the isotope of hydrogen known as deuterium is called the deuteron. As their masses M are almost equal. it would not.244.23).27) Hence we have the solution \ie:k" f Vr ? >a (5.29) Downloaded from Cambridge Books Online by IP 150. where now L is associated with the relative orbital angular momentum of the two particles.doi. take the spherical well —U for r < a ^0 for r > a. corresponds to / = 0. We have also required that R(r) does not diverge at the origin: if it possessed a singularity of the form r " 1 . in fact. with the tightest binding. Then for / = 0 (d2/dr2)ri?(r) = ( ^ l ) .org/10. ' where A and B are constants. ' ' [ -KzrR(r) for r < a. 2014 . Replace L2 by its eigenvalue.28) v for r < a. Here we have imposed the constraint that R(r) —> 0 as r -» oo.

208 on Thu Feb 13 17:10:44 GMT 2014. The wave function goes to zero very slowly as r —> oo. http://dx.org/10. It turns out that when realistic values are taken for U and a.3 Calculate the energy levels of the hydrogen atom according to Bohr's theory as follows.18) of the hydrogen atom. Calculate also the Bohr radius.6 Derive an equation that determines the energies of the bound states of orbital angular momentum / = 0 of a particle moving in the attractive potential V{r) = (-Ti2U/2m) S(r . 2014 . Verify that it is dimensionless. together with (5. and of the system nevertheless being a bound state. 5.(f)) = Cre~Kr ^ By explicitly applying to it the operators Lz.007 Cambridge Books Online © Cambridge University Press.18). Calculate the velocity of the electron in the lowest-energy Bohr orbit and determine whether the use of non-relativistic kinematics is justi- fied.</>) change if the coordinate axes are rotated through an angle a about the z-direction? 5. the deuteron is very weakly bound and the ground state is the only bound state. where U and a are constant. 5. it is approximately equal to 1/137. given in (5. The possibility of this being so. Then require L to be an integer.] Downloaded from Cambridge Books Online by IP 150.1 How does the function Y/m (#. most of the time the neutron and proton are separated by a distance greater than the range a of the potential. Assume that the electron is in circular orbit and cal- culate its orbital angular momentum TiL in classical mechanics. 5.O. L2 and H.244. corresponds to an equation for E with discrete solutions.a).doi.221. is a peculiarity of quantum mechanics. determine the values of the quantum numbers /.ra and n and express n in terms of the Bohr radius a$.5 A hydrogen atom has electron wave function iP{r. 5.2 The quantity e2/Airsohc is called the fine-structure constant. That is. Problems 5.27). [See the discussion of delta-function potentials in chapter 3.4 Find the most probable distance from the nucleus of an electron in the ground state (5.1017/CBO9781139171243. and (r) « 2a.60 5 The hydrogen atom which. Verify that this leads to the same energy levels as the correct procedure based on Schrodinger's equation.

The deuteron 61 5. 5.9 The binding energy of the deuteron is approximately 2 MeV and its radius 2 x 10~ 15 m.8 In a one-dimensional problem. assuming that it is much greater than the binding energy.7 The alkali atoms have an electronic structure that is approximately hydrogen-like.244. two particles.208 on Thu Feb 13 17:10:44 GMT 2014. http://dx. in that their chemical properties and spectral lines are determined essentially by a single electron. 2014 . Following the procedure described in appendix A. where 5. Downloaded from Cambridge Books Online by IP 150. Find the energy levels of the system when its centre of mass is at rest.007 Cambridge Books Online © Cambridge University Press.221.1017/CBO9781139171243. interact through the potential ^muj2(xi — x2)2. As a model for the poten- tial in which this electron moves. Estimate the depth U of the potential in the spherical-well model. where x\ and x2 are their position coordinates.org/10. show that the energy levels are Enl = (-e2/87re0a0)[n-D(l)}-2. each of mass m.doi. take V(r) = (-e2/47TSOr){l + b/r).

org/10. Revision quiz This quiz is designed to check that you have followed and understood the basic features of quantum theory.doi.244. http://dx.221. and how are they derived? Why are the conditions different for the problem of a particle in a box and for a particle moving in a delta-function potential? What are the conditions in these cases? 9 What is the tunnel effect? 10 How does one use stationary-state solutions to express general solutions of the time-dependent Schrodinger equation? What mathematical as- sumption is implied? 11 What is the Heisenberg uncertainty principle? Under what conditions can one simultaneously know the values of two observables? 12 What is meant by a Hermitian operator? What properties do its eigen- values and eigenfunctions have r 13 What are the basic assumptions about observables? 14 What is meant by the expectation value of an observable. 1 What are the diameters of an atom and of its nucleus? 2 What do we learn from the photoelectric effect? 3 What is the wavelength associated with an electron. 2014 . What are their de- generacies? 17 What is the Pauli exclusion principle? 18 How does one deal with a two-body system? 62 Downloaded from Cambridge Books Online by IP 150.1017/CBO9781139171243. and how is it calculated? 15 Why is the orbital-angular-momentum operator useful in problems where the potential is spherically symmetric? 16 Describe the energy levels of the hydrogen atom. and how do we know this? 4 What are the operators corresponding to the momentum and energy of a particle? 5 Write down the time-dependent Schrodinger equation.021 Cambridge Books Online © Cambridge University Press. How and under what circumstances can one derive the time-independent Schrodinger equation from it? 6 What is the probability interpretation of the wave function? 7 What is the expression for the particle flux j in terms of the wave function? 8 What are the continuity conditions that are usually satisfied by the wave function.208 on Thu Feb 13 17:11:11 GMT 2014.

As the protons are much heavier than the electron. which consists of two protons and one electron. we investigate how the presence of the electron overcomes this repulsion and holds the molecule together.244. considered as a function of the proton separation i?.1) 63 Downloaded from Cambridge Books Online by IP 150. the tunnel effect.doi.208 on Thu Feb 13 17:11:40 GMT 2014. http://dx. where EQ is the energy of the ground state of the hydrogen atom. We show that the expectation value of the energy. we may neglect their motion compared with that of the electron. We shall confine our discussion to the type known as covalent binding. the lowest-energy level of the system is EQ. The ionised hydrogen molecule The simplest molecule is the ionised hydrogen molecule. which we have already encountered in chapter 3. The Coulomb force between the two protons tends to push them apart. Suppose first that R is so large that in the vicinity of each of the protons the Coulomb field of the other is completely negligible. There are two different states corresponding to this energy. We now give a rather abbreviated account of how atoms bind together to form molecules. In particular.18) of the hydrogen atom: ' (6.org/10. 6 The hydrogen molecule In chapter 5 we saw how in quantum mechanics electrons are bound to nuclei so as to form atoms.008 Cambridge Books Online © Cambridge University Press. where r\ is the coordinate vector of one of the protons and <j> is the ground-state wave function (5. has a minimum for a certain value of i?. 2014 .221. In one of them the electron is described by the wave function c/)(r — ri). Then there are two sets of stationary states: the electron is bound to one or other of the two protons with a wave function corresponding to the ordinary hydrogen atom. so that there is a stable equilibrium configuration.1017/CBO9781139171243. but we can discuss the general features of the bonding by making suitable approximations. There is more than one type of molecular binding. The possibility of this type of binding relies on an effect that is peculiar to quantum mechanics. An exact calculation is difficult. and so regard them as fixed.

http://dx.n) 1 = 1. |c 2 | 2 is the probability offindingthat it is in orbit round the sec- ond proton. The system eventually settles down into a rather different sta- tionary state. The stationary-state wave functions are the eigenfunctions of the Hamil- tonian 2 P2 P2 P2 Lv 2m r v 4TT£O \r — 2 h (6 2) where m is the mass of the electron.244. The wave function ip(r) has to satisfy the time-independent Schrodinger equation H^ — EI/J.2. Similarly. in orbit round the first proton. Notice that these wave functions are real. This means that </>(r — r\) and <fi(r — r2) are no longer stationary-state wave functions.008 Cambridge Books Online © Cambridge University Press. Now reduce R to a value such that the Coulomb field of one proton is beginning to be felt in the vicinity of the other.221. because it would re- quire energy from outside to tear the electron from one proton and attach it to the other. that is. The expansion (6.36) Then |ci| 2 is the probability of finding that the electron is in the con- figuration with wave function </>i. which we now discuss.1017/CBO9781139171243.org/10. Here.3a) where fa = 0 ( r . Insert in this equation the expansion (6. But in quantum mechanics the tunnel effect does allow this transition to occur without any external energy being supplied. 2014 .3) for %j). and Downloaded from Cambridge Books Online by IP 150.208 on Thu Feb 13 17:11:40 GMT 2014. Then pre-multiply the equation by <f>i and integrate over all space. We write the stationary-state wave function for the ground state of the molecule as a superposition of <f>(r — r±) and (p(r. where r2 is the coordinate vector of the other proton. these terms are impor- tant only when R is taken to be so small that the extra Coulomb energy of repulsion between the protons is readily able to excite the electron into one of these states.r 2 ): (6.doi.3) represents an approximation. the first term is the kinetic- energy operator for the electron — remember that we are neglecting the kinetic energies of the protons.64 6 The hydrogen molecule In the other state. it should contain also terms corresponding to configurations in which the electron is bound to one of the protons with a wave function associated with one of the excited states of the hydrogen atom. (6. In classical physics the configuration of the system would be little altered. However. the wave function is <f>(r — r 2 ).

and we have used the normalisation (<f>i. http://dx. there is equal probability of finding the electron in orbit round either proton. (ii) E = E. E) where Hij = (faHfa) ij = 1 or 2. We must now attempt a rough calculation of Hu and Hi2. (6.4) gives / TIT J?}2 ( J{ fT TT \ 2 which leads to the two solutions n 12 (i) E = E+= . So if R » a0.org/10. with d = -c2.This means that eliminating c\ and c2 from the equations (6.1) that </>(r') is very small when \r'\ » a0.6) 1 — K\2 In both these solutions. Crudely speaking. we may write the integral in (6.7) A v } —~ ' \r . then \r'\ < R Downloaded from Cambridge Books Online by IP 150.008 Cambridge Books Online © Cambridge University Press. Hu — H22 and H\2 — H2i.244. (6. / d6r—r-±— + -—-.1017/CBO9781139171243. This result shows how dramatic are the consequences of the tunnel effect. |ci| 2 = |C212. the situation is very different from that in classical mechanics. so = Eo .3) of 0i. that is.221. Because of the symmetry of the problem.= Hn J*12. The first two terms of the Hamiltonian H in (6.2) are nothing but the hydrogen-atom Hamiltonian of which cf>i is the ground-state eigenfunction. 2014 . withci=c2.doi. The ionised hydrogen molecule 65 repeat this with 4>2: n-E) = c2(K12E-H12).1).r2\ 4ire0R Because of the definition in (6.<f>i) = (</>25</>2) = 1 implied by (6. the electron continually tunnels back and forth from the potential well surrounding one proton to that surrounding the other. We see from (6.7) as where we have written r' = r — r\ and R = r2 — r\.208 on Thu Feb 13 17:11:40 GMT 2014.

and so dominates for small R.221. (6. Nev- ertheless.7) diverges as R -> 0. when R —> 0 the integral (6.org/10.106) We consider H\2 similarly: H12 = Ai2^o .8) evidently remains finite. where (6.1017/CBO9781139171243. and so when R is large Hence for large R we may neglect the last term in (6.008 Cambridge Books Online © Cambridge University Press. 2014 . Hence we may use the expansion l/[( r ' _ fi)2]l/2 = [r/2 _ 2f . relatively small values of r' are most important in the integral.~A / d r.The second term. . . because the volume element d 3 r ' contains a factor r'2 when it is written in spherical polar coordinates.12) This integral does not diverge at r' = 0.13a) Downloaded from Cambridge Books Online by IP 150. the correction is of order OQ On the other hand.9) When we insert this expansion into the integral. where the factor CLQ appears here because (pi is appreciable only in the region where r' « oo. Hence for large i?.That is.11) compared with the others. the first term integrates to give —e2/AneoR. r+ 4T£ J \ \r-n r \ The integral here is —V(R). Hence for small R we have (6. integrates to give zero. The last term in the Hamiltonian (6./ .R + fl2]-l/2 = (1/i?) [1 + ( r ' • i?/i? 2 ) + . (6. http://dx.66 6 The hydrogen molecule throughout that part of the integration from which the integral receives most of its contribution.7) reduces to i f n « E o.208 on Thu Feb 13 17:11:40 GMT 2014.] . being odd in each of the three coordinates of r'. and (6.doi. (6.244.10a) This approximation is correct up to and including terms of order ao/i? 2 .

Hence the plots of E± against R must have the shapes drawn in figure 6. By inspection of (6.1017/CBO9781139171243. The corresponding value of the energy E+ is less than EQ.13) together. so that the configuration is a bound state. the covalent bond can be strong only between similar atoms: Downloaded from Cambridge Books Online by IP 150.14). there is a configuration of stable equilibrium. If we use this. Therefore E+ must have a minimum at some value of R\ that is. and vice versa.1. E± (6.008 Cambridge Books Online © Cambridge University Press. because when (pi is large 02 is nearly zero.are both very large and positive for small R.) When only one electron is involved.org/10.12) we see that V(R) is positive. http://dx. and put (6.10) and (6. and it decreases in magnitude as R is increased. the energies E± in (6.14) R small. (6. from its definition in (6.doi. 2014 .1.244. most inorganic molecules are bound mainly by a different mechanism. and in some inorganic molecules.5). becoming small for large R. #12 is small for large i?.11) is dominant and #12 « #12 e2/4<jre0R. known as ionic binding. E± plotted against R. E+ and E.221. At large i?.6) become £0 =F V(R) R large. which is the minimum energy of a separate proton and hydrogen atom. Other molecules A similar type of covalent binding provides most of the binding in the neu- tral hydrogen molecule. (However. so that for small R the last term in (6. in most organic molecules.12) is evidently finite at R = 0.208 on Thu Feb 13 17:11:40 GMT 2014. E+ approaches its asymptotic value Eo from below. Because of (6. The integral (6. Other molecules 67 Figure 6.136) Finally.

[You may find it helpful to refer to the discussion of delta-function potentials in chapter 3.org/10. in (6. but with <j>i now the wave function for the state where the first electron is attached to atom A and the second to atom 5 .208 on Thu Feb 13 17:11:40 GMT 2014. (See problem 6.244.Let the energies of the ground states of the two separate atoms be respectively EQI and JE 02 .1017/CBO9781139171243. http://dx. Problems 6.2 The muon is a particle whose properties are closely similar to those of the electron. When a muon binds two deuterons together the distance between the deuterons is sufficiently small for there to be a substantial overlap between their wave func- tions. but quadratic. with EOi < #02.4). v (A{eKX±e~KX) 0<x<R and find equations that give the bound-state energy as a function of R.] 6. 2014 .68 6 The hydrogen molecule the symmetry properties Hn = #22 and Hu = #21 were of crucial im- portance in our analysis.. with a lifetime of about 10~6 s).14). two electrons can form a strong covalent bond between two atoms A and B even when these are different.1 As a one-dimensional model for the ionised hydrogen molecule.doi. (The consequence is that a single electron does not give strong covalent binding between unlike atoms. The theory is similar.) However.221. as in (6. sup- pose that the potential near an atom is a delta function.3.3 Suppose that. Show that the wave function of the electron has the form . how does the size of the resulting molecule compare with that of the electron-bound molecule? [We explained at the end of chapter 5 that the radius of the deuteron. If it is used to bind together two nuclei. Hu ^ H22 and Hi2 ^ H2i. It is evident that in this case the equalities Hn = H22 and H12 = #21 do hold. and so there is a good chance that the two nuclei will fuse and form a helium nucleus before the muon decays.008 Cambridge Books Online © Cambridge University Press. and 02 the wave function for the corresponding state where the electrons are interchanged.Show that for large R the lowest-energy eigenvalue E differs from EOi by an amount that is not linear in small quantities. is quite large. so that for the molecule V(x) = (-h2U/2m)[8(x -R) + 5(x + R)].) Downloaded from Cambridge Books Online by IP 150. the nucleus of deuterium. except that it is some 200 times heavier (and it is un- stable.] 6.

doi. Other molecules 69 6.221. it would have wave function <^. http://dx.6) for the energy levels.244. 2014 .008 Cambridge Books Online © Cambridge University Press.4 Three atomic nuclei are fixed at the corners of an equilateral triangle and an electron is introduced into the system. Downloaded from Cambridge Books Online by IP 150.208 on Thu Feb 13 17:11:40 GMT 2014.org/10.1017/CBO9781139171243. Obtain expressions analogous to (6. If this electron were bound to the atom i in isolation.

and time-dependent perturbation theory.25) is an approximation to it. (7.221. time-independent perturbation theory.<l>k) = 6jk. can we now calculate. Time-independent perturbation theory In chapter 5 we described a model of the deuteron in which the proton and neutron are bound together by the simple spherical-well potential given in (5. . (7. correspond- ing to eigenvalues Ej.244. 2. 7 Introduction to perturbation theory In this chapter we describe the approximation method for problem solving known as perturbation theory. used to calculate certain quantities called transition probabilities.org/10.doi.25). and so the simple form (5.2) and that they form a complete set.1017/CBO9781139171243. This is certainly not the true potential.208 on Thu Feb 13 17:12:03 GMT 2014. 2014 . There are two types of perturbation theory. that the Ej are all distinct. as discussed in chapter 4. j — 1. Thus the question arises: suppose that we are able to find the eigenvalues of a Hamiltonian Ho. Assume that there is no degeneracy. be the various eigenfunctions of HO. The true potential will also be an attractive potential.1) that is not too different from Ho? Let 0j(r). . .009 Cambridge Books Online © Cambridge University Press. the eigenvalues of another Hamiltonian H = H0 + H1 (7. 3 . that is. http://dx.3) 3 with the Cij constant. This allows the eigenfunctions ^ of H to be expanded as linear combinations of the _ y&. We shall also suppose that the (j)j are orthonormal. meaning that they meet the condition (<f>j. used to find the stationary states of a system. at least approximately. 70 Downloaded from Cambridge Books Online by IP 150. but it is the model that most easily leads to a solution of the time-independent Schrodinger equation.

7) That is. we obtain (Ho + Hx) Y. we expect that our approximations will be good when AEi is small compared with the difference between E{ and the next eigenvalue of #o. is just the expectation value of Hi calculated from the original wave function fa. neglect terms that are quadratic in small quantities. are such that |cy | < 1. fa) — 1.6) Now pre-multiply by <^(r) (remember that Hi may contain differential operators).208 on Thu Feb 13 17:12:03 GMT 2014.244. (7. use the definition HQ(/)J = Ejfy of the eigen- functions (f)j. except for c#. However. with k ^ i.8) Downloaded from Cambridge Books Online by IP 150.6) by </>£(r). if we wish also to calculate the perturbed wave function ^ . (7.5) To simplify this equation. Ci^i(r) = (Ei + AEi) £ ^ ^ ( r ) . (7. Time-independent perturbation theory 71 We consider the case where Hi is small. omit those terms where either Hi or AEi is multiplied by cy with i ^ j . Then + X > . Usually this is as far as we need go in the calculation: the level shift is the quantity of most interest. and again integrate over all space.009 Cambridge Books Online © Cambridge University Press. On inserting the expansion (7.1017/CBO9781139171243.221..3) into the time-independent Schrodinger equation Hfa = (Ei + AEi) fa. Then if we normalise fa so that (fa. 2014 . and replace ca by 1. This gives = (cf>uHi<j)i). the shift AEi in the level Ei resulting from the addition of the perturbation Hi to the original Hamiltonian HQ. Also. integrate over all space and use the orthonormality condition (7.org/10. http://dx. which implies that = 1 ' (7-4) cu will differ from 1 by a quadratically small quantity. and that corre- sponds to an eigenvalue Ei + AEi that is not too different from E{. We then find that cik = (<j>k. (7. By this we mean that corre- sponding to a given eigenfunction fa of HQ with eigenvalue E^ there is an eigenfunction ipi of H that is not too different from </>. . and that all the coefficients cy. we instead pre-multiply (7.2).Edarfjir) = AJE^(r). that is.Hifa)/(Ei -Ek) k± i.doi. More precisely.

and that Hi is small in the sense that we have described. 4>qrs(r) = -T.sin .244.org/10.208 on Thu Feb 13 17:12:03 GMT 2014.— s i n — . an obvious choice is to identify Hi with V. (7. (7.8) vanishes. / 8 \ . http://dx. so that Ho is the kinetic-energy operator whose eigenfunctions are given in (3. the only requirements are that we can find the eigenstates of HQ. together with ca = 1. Take the field to be parallel to the rr-axis.009 Cambridge Books Online © Cambridge University Press.I s i n . Example If we want to apply perturbation theory to calculate the eigenvalues of a Hamiltonian H..doi. / . For two degenerate states i and k such that the numerator in (7. the first step is to write it in the form (7. It is possible to extend the calculation to give expressions for the higher-order terms. my .3). we have an expression for ipi(r). as described at the beginning of chapter 3. We shall not pursue the case where the numerator does not vanish.10) \abcj a b c Downloaded from Cambridge Books Online by IP 150. As a simple example. Notice the importance of our original assumption that the level E{ is non-degenerate.221. but we shall not do this here. so that this potential is weak.1017/CBO9781139171243. Our calculations are correct to first order in small quantities.72 7 Introduction to perturbation theory By inserting this result into (7.8) that we have obtained for Cik would be singular for some A. but rather that it carries a charge e and is acted on by a uniform electric field £. 2014 . our calculation is valid so long as that level is non-degenerate. it does not matter if some of the other levels happen to be degenerate. If we are interested in the perturbation AE{ to a particular level E{. the way in which we choose to divide H will have an obvious physical significance. This division of H into two parts Ho and Hi may be a purely mathematical procedure. however. namely . otherwise the expression (7. It is possible also to deal with the case where the level of interest is degenerate. snz . Usually._ . qnx . Suppose that the particle does not now move freely within the box.9) If either e or £ is small. and the previous analysis applies without modification.1). so that the potential takes the form V = -e£x. and our assumption that all the c^ with k ^ i are small would be inconsistent.4). there is no problem. consider again the problem of a particle in a rect- angular box.

we must evidently confine ourselves to the case where the lengths a. Hi(t) -4 0.13) We suppose that. at t = — oo.7).5). (7.org/10. Because we are interested in the time dependence of \I>.£) of the system.t) = <j>j(r)e-iEJt'\ (7. http://dx. 2 2 2 \ In order to be able to apply the perturbation theory developed above.12) •/box In this simple problem.208 on Thu Feb 13 17:12:03 GMT 2014. so that initially H coincides with Ho. s integers. Time-dependent perturbation theory We now investigate the behaviour of a system under the influence of a perturbation that varies with time: H = H0 + H1(t).3). (7. for which the shift AEqrs is being calculated. r. as before.*). We shall suppose that it is initially in the stationary state i and we shall denote by ^i the wave function that coincides with <3>.244.14) where the <j>j(r) are the same eigenfunctions as in (7.15) Suppose that as t -> -oo.doi. where AEqrs = = -e£ / dxdydzxc/)2qrs = -\e£a. Because H varies with t. from (3. the perturbation has turned out to be the same for every level. 6. #0 does not vary with t.221. The corresponding unperturbed energy levels are. the energy levels in the presence of the electric field are Eqrs + AEqrs. From (7.1017/CBO9781139171243. we must now include in the stationary-state wave functions of Ho their time variation: *j(r. it does not have stationary-state wave functions. (7. is non-degenerate. 2014 . we must solve the time-dependent Schrodinger equation H^(r.009 Cambridge Books Online © Cambridge University Press. Downloaded from Cambridge Books Online by IP 150. Thus initially the system may be in a stationary state of Ho. c of the sides of the box are unequal. We again use the stationary-state wave functions of Ho to make an expansion of the wave function \t(r. so as to ensure that the level Eqrs. Time-dependent perturbation theory 73 with q.t) = i/i*(r.

because H\ may contain the derivative operator d/dr. (7. and all the other a^ are zero. However.18) we have had to be careful about the order in which we have written the wave functions $ relative to Hi.t) = ih$j which defines $ j . We shall suppose that the value of t is such that au(t) is still close to 1. ihdik = Notice that in (7.H1tf)<j>i)eitE'>-B*'"i.org/10. will start to become different from $.20) where the 5ik is the integration constant that takes account of the initial conditions at t — — oo.1017/CBO9781139171243. 2014 .16) Notice that the coefficients a^ are functions of t. We use the expansion r. we have aik {t)=8ik-1. and integrating over all space. which means that on the right-hand side we keep only the term in the sum for which j — i and in it we replace an by 1: ihaik(t) = ($*. we find that y^i^-. Downloaded from Cambridge Books Online by IP 150.#!$.) = (&.208 on Thu Feb 13 17:12:03 GMT 2014. This means that we expect the coefficient functions a^-(t) to depart rather little from their original values.18) only those terms that are first order in small quantities. So far. At t — — oo.244.221.18) is exactly equivalent to the time-dependent Schrodinger equation.009 Cambridge Books Online © Cambridge University Press.17) and (7. \I>. http://dx.15).19) Integrating with respect to t. we obtain. We then retain in (7. We now suppose that Hi is small.doi. these simply multiply H. an = 1. (7.74 7 Introduction to perturbation theory When the perturbation Hi (t) begins to switch on. (7. with the help of the orthonormality conditions ($&. at least if the time t is not too large.17) On pre-multiplying this equation by $£. we do not allow Hi to contain the operator d/dt. f dt'(4>k.tfi&) e ' ^ . we have made no approximations: (7. because the $j are not stationary-state wave functions for the perturbed system. and that aij(t) for i ^ j is still small.W * . so that H does not operate on the coefficient functions a^(t). and using the equation Ho$j(r. (7.t). Inserting this expansion into the Schrodinger equation (7.. $j) = 5k j .

Frequently one can show without detailed calculation that this term vanishes for all but a few final states fa. and so throws the system into the stationary state <!>£. Provided that this probability is small. usually (though not always) their corresponding transition probabilities are very small. is.org/10. according to the discussion given in chapter 4 and to (7. is the probability that our measurement finds that the per- turbation Hi has induced a transition from the state $. Those final states fa for which (7. so that our approximations are valid. Transition probability 75 Transition probability We have supposed that initially. the system will be thrown into a stationary state of Ho and will remain in this state.Suppose that after some time T the perturbation Hi(t) vanishes. regardless of the value of t'. and that the system is then in the stationary state $i of Ho. we may calculate it from (7.009 Cambridge Books Online © Cambridge University Press.244. consider again the problem of a charged particle in a box. and let the perturbation again be caused by a uniform electric field in the ^-direction. then. The probability that the measurement results in the value Ek.21) This. to the state $fc. at t = —00. the Hamiltonian is just i?o. so that the Hamiltonian is again Ho.221. Notice that a^ involves the term (fa.9).20): m 2 \«ik(t)\2 = To (7. but with a transition probability that is non-zero only at second or higher order in the perturbation H\. - Jo (7.16). just |a^| 2 . and then switched off again at time T.1017/CBO9781139171243.208 on Thu Feb 13 17:12:03 GMT 2014. Hi fa).doi. without detailed calcula- tion one can find those final states fa that are readily accessible from the initial state fa. That is. If we now measure the energy of the system. The probability that a measurement at some time t > T will find that there has been a transition to the stationary state 4>QRS is 2 2 (1/h ) I dt'(<f>QRS.21) vanishes may still be possible final states. We assume that for t < 0 the system is in the stationary state (f)qrs. Suppose that this perturbation is switched on at time t = 0. Such transitions are known as forbidden transitions. http://dx.22) Downloaded from Cambridge Books Online by IP 150. as given in (7. 2014 . Example As an example.

)| 2 = (4/UJ2) sin 2 \uT. (7. in fact. this principle also governs the accuracy with which it is possible to measure the energy of a system: it can be shown that if a limited time T is available for making the measurement. and is a manifestation of the energy uncertainty principle.24) So first-order perturbation theory allows only those transitions for which R = r . The shorter the duration T of the perturbation.221. http://dx.244. Energy uncertainty principle Prom (7. 2014 . The height of the central peak is T 2 .doi.10): {(/>QRS.23) The term ((f)QRs.009 Cambridge Books Online © Cambridge University Press. and.org/10. (7. This somewhat surprising result is peculiar to quantum mechanics. then the energy cannot be Downloaded from Cambridge Books Online by IP 150. Alternatively.x<f>qr8)= / dxdydz(j)QRSx(j)qrs Jbox 0 Q .1017/CBO9781139171243. We explained in chapter 4 that making a measurement on a system is liable to induce a perturbation.25) A plot of this function against UJ is shown in figure 7.x<j>qrs) is readily evaluated.q even (Q ^ q) X | 8Q/[2(Q2 2)2] Q _ q (7. the system may remain in its initial state. the greater will be the spread in energy of this band of preferred final levels.76 7 Introduction to perturbation theory where I(UJ) = / dt'eiuJt' Jo u = (EQRS-Eqrs)/h.208 on Thu Feb 13 17:12:03 GMT 2014.1.23) we find that |J(o. while that of the peaks next to it on either side is some twenty times less. This peak is centred on UJ — 0 and it has half-width 2TT/T. SO the levels to which transitions are most likely are those whose energy EQRS differs from the original energy Eqrs by an amount huj of order h/T. Hence the transitions that occur most readily are those to levels for which UJ lies within the central peak. using the explicit form of the wave functions given in (7. 5 = 5 and Q differs from # by an odd integer.

http://dx.2. Plot of |/(CJ)|2 against w (see (7. We suppose that we are able to solve the time-independent Schrodinger equation for both Hamiltonians. but then at time t = t0 the Hamiltonian instantaneously changes... The energy uncertainty principle is important also for understanding how the tunnel effect is possible: it allows a particle momentarily to acquire enough energy to surmount a potential barrier. we have considered perturbations H\[t) that vanish for very large t.org/10. determined better than to within an amount of order h/T. and that we know that the eigenvalues of Ho are Ej. as before..doi... no measurement can determine its energy to within an accuracy better than about Tt/r.009 Cambridge Books Online © Cambridge University Press.208 on Thu Feb 13 17:12:03 GMT 2014. Sudden change in the Hamiltonian Up to now. so that the Hamiltonian ultimately returns to its original form HQ.221. and those of Hf0 are E'^ j = 1. 2014 .1017/CBO9781139171243. if a system is unstable and so has a finite lifetime r.2. In particular. j = 1.1.244. Sudden change in the Hamiltonian 77 -2TT/T 2TT/T Figure 7.. so that for all t > to it is equal to the time-independent operator H'Q.25)). Consider now the case where the Hamiltonian is initially equal to the time-independent operator i? 0 . Let the system initially be in a stationary state Downloaded from Cambridge Books Online by IP 150.

*)| 2 . t ) = < Ebj&jfat) t>t0.28) would not then be satisfied at t = t0 since H merely has a finite discontinuity there. the wave function \I/(r.t) = *. In the previous work we calculated the first term in a power-series expansion for the probability.14). equal to $*. http://dx. For t > to we expand it as a linear combination of the <fr ^: ($i{r. (7.009 Cambridge Books Online © Cambridge University Press. and therefore the coefficients bj are constant. and so throw the system into the corresponding stationary state of HfQ. For if it were not.1017/CBO9781139171243. t 0 ).31) Downloaded from Cambridge Books Online by IP 150.are stationary-state wave functions of the Hamiltonian. t o ) = *t(r.29) 3 Multiplying this equation by $'fc* (r.208 on Thu Feb 13 17:12:03 GMT 2014.«o).doi.org/10. (7. The wave function (7. we obtain (7.78 7 Introduction to perturbation theory of Ho.26) where Note that here the prime does not denote differentiation.t) of the system is.(r)e-"3«/\ (7. by assumption.26).27) I 3 For t > to the <J>^. 2014 .221. (7.27) will automatically satisfy the Schrodinger equation HV = ihd$/dt (7. with energy E{. t) as in (7. The continuity condition at t — t0 gives ^ ( r . d^jdt would have a singularity proportional to S(t — to) at t = to (see appendix B).t) t<t0} * ( r . At t = t 0? * must be continuous. We do not need to assume that the difference between HQ and H'o is small.28) for both t < to and t > t0.(r. and the functions $'• may be completely different from the $j. Let those of H'o have the orthonormal wave functions ». We calculate the probability that a measurement of the energy at some time t later than t0 will result in the value J5J-. For t < to. Let the stationary states of Ho have the orthonormal wave functions <frj(r.244.30) Hence the probability that a measurement of the energy at some time t > to yields the value E'k is M 2 = l(&. and the Schrodinger equation (7. here we are able to give an exact expression. integrating over all space and using the orthonormality conditions in (7.

the deuteron. this break- up is not allowed by energy conservation: the binding of the deuteron is strong enough to ensure that the rest mass of the deuteron is less than the sum of the rest masses of two protons and an electron. The probability that we do find this to be so is calculated from (7. http://dx.32) (The neutrino is a particle that has zero charge and also zero mass. the nucleus would have to break up. before its decay.31) by using Downloaded from Cambridge Books Online by IP 150. Suppose now that. 2014 . If we measure the energy of the orbiting electron after the decay of the nucleus. the decay may not be possible. If its neutron were to decay. so that quite a lot of kinetic energy is available in the final state.244.221. so that the initial- state energy mnc2 is sufficient both to create the final-state particles and to give them some kinetic energy.009 Cambridge Books Online © Cambridge University Press. the charge on the nu- cleus suddenly changes from e to 2e. The reason that the decay is possible is that the rest mass mn of the neutron is greater than the sum of the rest masses of the proton and the electron.32). Consider. namely the isotope 3 He of helium. so that the tritium atom was neutral. However. The decay is tritium —>> 3 He + electron + neutrino.) Consider this decay in the frame of reference in which the neutron is at rest. so that like the photon it moves with velocity c.33) occurs. Because the 3 He has only one orbiting electron. (7. the nucleus 3 He is bound rather more tightly than the original tritium nucleus. When a neutron is not bound into a nucleus it is unstable: it undergoes the decay neutron ->• proton + electron + neutrino.33) Further. This is because there is a bound state of two protons and a neutron. when the neutron is bound in a nucleus. so that the stationary atomic states correspond to new electron wave functions.doi. However. it is not neutral but positively ionised.org/10.208 on Thu Feb 13 17:12:03 GMT 2014. When the decay (7. The electron is so energetic that it rapidly escapes. energy conservation does allow one of its neutrons to decay as in (7. (7. we do not necessarily find that the orbiting electron is in the new ground state.1017/CBO9781139171243. for example. because there is no bound state of two protons. the tritium nucleus was accompanied by an orbiting electron in the atomic ground state. rather than being trapped in an orbit corresponding to one of the atomic states of the helium atom. In the case of tritium. Example: decay of tritium 79 Example: decay of tritium Tritium is the isotope of hydrogen whose nucleus consists of one proton and two neutrons.

34) where the Bohr radius an is calculated in terms of the appropriate reduced mass and is approximately the same in both wave functions.208 on Thu Feb 13 17:12:03 GMT 2014.18) and the subsequent work in chapter 5. Thus. ^ ( ) a o ) . 2 = (8/^og) 47rr 2 dre.1017/CBO9781139171243.doi.2 r / a 0 ) (7.80 7 Introduction to perturbation theory the original ground-state wave function for fc and the new ground-state wave function for (f)'k. 2014 .org/10. Initially the particle is in the stationary state of energy Downloaded from Cambridge Books Online by IP 150.3r/ao = 512/729.1 A hydrogen atom is placed in a weak electric field 5.009 Cambridge Books Online © Cambridge University Press.3 A quantum system is capable of existing in only two independent states 0i and </>2.35) So the probability that the helium ion is found to be in its ground state is a little less than 3/4. In this case it will subsequently decay back into the ground state. and if the electron has not escaped from the atom. (ii) in lowest- order perturbation theory. (7.244. The system is modified by the addition of a term H± to the Hamiltonian. Problems 7. http://dx. Hence tritium gas provides a useful source of illumination that needs no external supply of power.3 / 2 e .4 A particle of mass m and charge e is contained within a cubical box of side a. 7. According to (5. respectively. 7. Show that to first order in £ the ground-state energy is unchanged.2 Calculate the first-order perturbation to the ground-state energy of a linear harmonic oscillator due to an additional potential ex2. Find expressions for the new eigenvalues E[ and E'2\ (i) by exact calculation. If the helium atom is not in its ground state. emitting one or more photons in the process (see chapter 9). 7. which are eigenstates of the Hamiltonian Ho with eigenvalues E\ and E2. Verify that your result agrees with the exact answer to first order in e. Verify that the perturbation-theory result is an approximation to the exact result.221. it will be in an excited state.

assuming that the electric field varies with time as £ e~f /T . 2014 .org/10. Show that the probability that the elec- tron is excited from the ground state n = 0 to the n — 1 state is 7r\2h6/2vsm8uj6.221.6 A particle is in the ground state in a one-dimensional box (given by the infinite square-well potential (2. 7.5 A charged particle is scattered by an electron bound to a hydro- gen atom.208 on Thu Feb 13 17:12:03 GMT 2014. Example: decay of tritium 81 3h2/K2/2ma2.doi. Obtain the expression in lowest-order perturbation theory for the probability of eventually finding the particle in a state of energy 3/i 2 7r 2 /raa 2 .1017/CBO9781139171243. At time t = — oc a uniform electric field is switched on parallel to one of the edges of the cube. The problem is considered in terms of a crude one- dimensional model in which the electron is described by a harmonic- oscillator wave function and the charged particle induces the pertur- bation Hi — X8(x — vi).244.009 Cambridge Books Online © Cambridge University Press.11)). What is the probability that it remains in the ground state when the wall separation is suddenly doubled? Downloaded from Cambridge Books Online by IP 150. http://dx. 7.

with also spin angular momentum because it spins about an axis.221. however. However. we outline how quantum theory allows even a point particle to have spin. [LZ. the eigenvalues of L2 and Lz are given in (5. (8.1) These allowed values for / and m may be deduced by explicitly solving the differential equation (5. quantum theory introduces an orbital angular momentum vector TiL — r x p asso- ciated with the motion of a particle. 8 Spin Two kinds of angular momentum In chapter 5 we explained that.010 Cambridge Books Online © Cambridge University Press.1. . We explained that there is another type of angular momentum. not associated with the motion of a particle. where m = —/. just as in classical mechanics. .Lz] = \LX. and vanishes in the limit of zero radius. which instead uses what is called the algebra of angular momentum. We explain in this chapter some of the many other experimental reasons why we know that electrons.8) for the simultaneous eigenfunctions of L2 and Lz.28) between the components of the position and momen- tum operators.Ly] = iLz. which is given the name spin. For the case of orbital angular momentum.244.9): 1(1 + 1) for X2. But in quantum theory even a point particle may have spin angular momentum. These two types of angular momentum oc- cur too in classical mechanics. there is another method. 2014 . . the earth has orbital angular momentum because of its motion round the sun. .LX] = iLy. — I + 1 . First.2) 82 Downloaded from Cambridge Books Online by IP 150.208 on Thu Feb 13 17:12:37 GMT 2014. and derives from them commutation relations between the components of L: [Lx. for example. (8. photons and other particles have spin angular momentum. I — 1.1017/CBO9781139171243. where / = 0.doi. . . We have already mentioned in chapter 5 that the fact that electrons have spin influences how the Pauli exclusion principle controls the energy levels of multi-electron atoms. or /. .org/10. m for L z . In classical mechanics the spin angular momentum depends on the radius of the spinning body. http://dx. One starts with the commuta- tion relations (4. . [Ly. protons. 2 .

then the allowed values for j are \l . and use instead US when we are referring to spin. Downloaded from Cambridge Books Online by IP 150. (8. for example 7r-mesons (these are produced when protons are smashed together at high energy). Its total angular momentum TiJ is the vector sum of the two.244. one may show from these commutation relations that. We denote the eigenvalues of S2 by 5(5 + 1). ..1). § .. the rules (which follow from the algebra) for calculating the allowed values j(j + l) for the eigenvalues of J2 are more complicated. where the minimum magnitude of the sum of two vectors is again the difference between their separate magnitudes. Photons have spin 5 = 1. This requires then that — I and / are separated by a non-negative integer. Thus. and we write J = L + S.221.208 on Thu Feb 13 17:12:37 GMT 2014. 2014 . or (I + 5). a particle will have both orbital angular momentum and spin. . . §. . If the eigenvalue of L2 is 1(1 + 1) and that of S2 is 5(5 + 1). . It is convenient to keep the symbol UL for orbital angular momentum. without solving any differential equa- tion. we call it spin. rather it is intrinsic to it. (8. then. (/ + 5 . but it also allows instead / = § .1).1017/CBO9781139171243. . for given /. and the maximum is again their sum. Some particles have zero spin. Spin \ 83 By simple algebraic manipulation. The values of / given in the first line of (8. http://dx..5) This is something like classical mechanics. the allowed values of m are just those shown in the second line of (8. |/ . As it is not associated with the motion of the particle. However.5|. .8) that is associ- ated with orbital angular momentum.1) satisfy this condition. § . so we identify them with a different kind of angular momentum. (8.010 Cambridge Books Online © Cambridge University Press. but in quantum me- chanics the details are more complicated and we shall not elaborate on them. | . so that 21 is a non-negative integer. 1 . | .3a) These new solutions for / are consistent with the algebra of angular mo- mentum but not with the explicit differential equation (5.doi. as do protons and neutrons. 2 .s\ + 1. .org/10. The angular-momentum algebra requires 25 to be a non-negative integer: 5 = 0. (8. Generally.4) The rule for calculating the eigenvalue of Jz for a simultaneous eigenstate ij) of Lz and Sz is simple: if Lztp = mip and Sztp = sz%l) then Jzip = (m + sz)ij). electrons all have spin 5 = | .36) Experiment finds that a given type of fundamental particles always has the same spin 5.

so obtaining axay + ayax = 0 . (8.2iay. which we obtain from (8.86) to set <J\ — \ and then add these relations. multiplying each term first on the left and then on the right: (Jza2x — crxazax = 2iayax.2) by replacing L with \a\ [crx. (8. (8. ± | . whenever we use it the operator 5 2 takes the value 3.84 8 Spin Spin \ The special case of spin \ is particularly important. [<Jy.8a) There is nothing special about the ^-direction: we may choose the z-axis to be in any direction without changing the physics. First. [(Jz.<rz] = 2iax. we now use the algebra satisfied by the com- ponents of a. For spin 5. 2014 . because when we apply it to a spin-| state.208 on Thu Feb 13 17:12:37 GMT 2014.86) To derive further relations.010 Cambridge Books Online © Cambridge University Press.ay] = 2iaz.org/10. a2y = 1. we represent the algebra in a very simple particular concrete form. (8. Since we are describing the case where S2 has eigenvalue 5(5 + 1). (8. and only for spin | .1017/CBO9781139171243.6) The operator a is known as the Pauli spin matrix. Sz has two possible eigenvalues. it is useful to write for this case 5 = \a.crx] . a2 has the single eigenvalue 3. in particular a\ = 1.9) Multiply the last of these relations by crx.244. proton and neutron. Since we only use the operator a when it is applied to states of spin | . (8. oz has possible eigenvalues ±1 and o2z therefore has only one possible eigenvalue: effectively a2z = 1. and so effectively a2 = 3.221.7) Further. http://dx.106) Downloaded from Cambridge Books Online by IP 150.10a) We use (8. because it applies to the principal constituents of matter: the electron. (8.doi. So the square of the component of a in any direction must be 1.

(S!)(?)~(! so that u^ is the eigenstate of GZ with eigenvalue +1. GZGX = \Gy. We can do this by associating a 2 x 2 matrix with each component of a: /on x { 8 U a ) ' This set of matrices is called a "representation" of the components of a.1 .010 Cambridge Books Online © Cambridge University Press. Although we have written = signs in (8. The representation (8. Each matrix is a Hermitian matrix.208 on Thu Feb 13 17:12:37 GMT 2014. This means that its eigenstates have a particularly simple representation.221. more properly we should say "is represented by". this gives axay = \az. 2014 . (8. We usually label these eigenstates by t and | . because the operator it represents is a Hermitian operator. Downloaded from Cambridge Books Online by IP 150.11a) is not unique: we could have used larger matrices.11a). It has the feature that the matrix that represents GZ is chosen to be diagonal.14) The obvious interpretation of \ipi(r)\2 is that it is the probability density that a measurement of GZ will give the result +1.1017/CBO9781139171243. They are represented by It is easy to see that !)(i)-(i).org/10.2(^)|2 is the probability density that it will give the result . The coefficients that multiply them will usually be functions of r. It is not even unique if we restrict ourselves to 2 x 2 matrices. (8. To obtain a general spin-| stationary state.244.doi. it is often useful to cast them into a more concrete form. We choose to normalise it so that Jd3ruUr)u(r) = Jd3r (h/>i(r)|2 + |^(r)| 2 ) = 1. http://dx.8c) In a similar way we may show also that GyGz = \GX. The electromagnetic interaction 85 With the first commutation relation of (8. and |V. So the state has a representation of the form We call u a "two-component spinor".8d) While everything involving spin \ can be calculated by using the six simple relations (8. we take a linear combination of u^ and u^. and u± is its eigen- state with eigenvalue —1. (8.9).8).

B = VAA. (8. and that we switch on an external electromagnetic field.org/10. One of the direct pieces of evidence that the electron has spin \ comes from the structure of this degeneracy breaking. so that its electric and magnetic vectors are given by A. to- gether with (8. now give Mr = -VV -e£ -er AB. It is found that the correct way to do this is to follow the same procedure as in classical physics. where the Hamiltonian is V(r). In classical mechanics. we must first add a term to the Hamiltonian to take account of the interaction of the electron with the external electromagnetic field.17). have energies that are independent of the eigenvalue m of the ^-component of the orbital angular momentum.t) and a vector potential j l ( r . As we shall show. mr = — VV.dp' d* " dr' These equations are known as Hamilton's equations.doi.244. Suppose that the particle is an electron.18) In order to verify this.15).) The Hamiltonian then changes from Ho to H = V + (p + eA)2/2M .16) Hamilton's equations give r = p/M.1017/CBO9781139171243. one shows that Hamilton's equations (8. for a particle moving in a potential V(r). (8. a Hamiltonian H is used to give the equations of motion of a system by eliminating p in the equations _ dp__8H (8 5) dt .VV.208 on Thu Feb 13 17:12:37 GMT 2014. http://dx. 2014 . so that it has charge —e. To study this. such as the Coulomb potential which binds the electron in the hydrogen atom.221. correctly. (8. and together these equations give. this degeneracy is broken when the atom is placed in a magnetic field.19) Downloaded from Cambridge Books Online by IP 150.86 8 Spin The electromagnetic interaction We saw in chapter 5 that the bound states of a spherically-symmetric potential. p = . (8.010 Cambridge Books Online © Cambridge University Press.17) (See appendix D. For example.ecj). This field may be described by a scalar potential </>(r. i ) .

21a) If there is no external electric field.18) becomes an operator. We can therefore write (p + eA)2 = p2 + 2ep • A + e2A2.21c) Downloaded from Cambridge Books Online by IP 150. the external electromagnetic field is small. (8. In most applications.17) give H = HQ + i?i.19). In quantum mechanics. which can therefore be neglected.20) The Zeeman effect Suppose now that a hydrogen atom is placed in a uniform external mag- netic field B pointing in the z-direction. Then the term ep • A is rather larger than e 2 A 2 . 2014 . A vector potential that satisfies the gauge condition V • A = 0 and B — V A A is A = \rf\B. ihA • V * = -ih(V • A)V = 0.208 on Thu Feb 13 17:12:37 GMT 2014.1017/CBO9781139171243.16). and so (8.20) gives # i = (e/2M)r A B. and it is convenient to impose on it the further gauge condition V • A = 0 (see appendix D).216) This commutes with the original spherically-symmetric hydrogen-atom Hamiltonian Ho. (8.A-e<l> = {e/M)A • p . be- cause p must be replaced by the operator p = —i/iV.244. Hi = (e/M)p .12) of Ho which we found in chapter 5 corresponded to definite values m for the operator L z .org/10. http://dx. so in these states Hi takes the value AE = (eh/2M)Bm. For if \I> is any wave function. the last two terms are just the correct Lorentz force for the action of the electromagnetic field on the charged electron.(r A p) = {eh/2M)BLz. So then (8. The vector potential A is not fully determined from the definitions (8. in the sense that it is much smaller than the internal Coulomb field that binds the electron in the atom.17). The eigenstates (5. 0 = 0. (8. when we multiply out the term (p + eA)2 in (8.ecf).doi. the order in which we write p and A in their dot product does not matter. The Zeeman effect 87 We do this in appendix D.16) and (8.221.010 Cambridge Books Online © Cambridge University Press. the Hamiltonian (8. (8. In (8.p = (e/2M)B. This has the consequence that.

The discrepancy is explained by the fact that the electron has spin. 2014 . that is they could be multiplied by any spinor whose components are independent of r. So we expect that the n = 2 to n = 1 transition should give three lines in the spectrum whose frequencies are close to one another. but which are eigenfunctions of oz. so placing the atom in a magnetic field removes some of the degeneracy. There are two of these. it is not affected by the external magnetic field.208 on Thu Feb 13 17:12:37 GMT 2014.0 or — 1. This affects the spectrum of photon frequencies seen when the atom makes transitions between different levels. there can be an intrinsic magnetic dipole moment pointing in the direction of S. n = 1. http://dx. for which m can take any of the values 1.1017/CBO9781139171243. n = 2. because the Hamil- tonian was independent of a. so that the n — 2 level is split into three levels that are close together (because in practice B is "small" in the sense we have already explained). for a particle that has spin vector 5. When we add in the term (8.88 8 Spin This has to be added to the original energy levels (5.doi. corresponding to eigenvalues ±1 for az. For example. The m = ±1 states have their energies shifted by ±(e/2M)B. which again is unaffected. Here n is the principal quantum number: see (5.244. So every energy level we had before is split into two. This is not what is observed experimentally.221.org/10. However. we obtain eigenfunctions of the new Hamiltonian by multiplying VWi(^) by a two- component spinor u whose components do not depend on r.17).22) in Hi. (8. in particular to the n = 1 ground state. because there is no direction intrinsic to the particle along which it can point. depending on whether Sz has eigenvalue ±1. The electron's intrinsic magnetic dipole moment vector is found to be — (eh/M)S. to its ground state. with additional energy ±(e/i/2M). and therefore it can have an intrinsic magnetic dipole moment.116). This applies even to the / = 0 states. so that in the presence of the external magnetic field B pointing in the z- direction there is an additional term {eh/M)S • B = (eh/2M)Baz. the stationary-state wave functions 4>nim{r) could be in any spin state. u^ and u± given in (8. Before we included this term.17). but it can also be / = 1. but in the case of the electron it may also be calculated from a relativistic wave equation for the electron which was derived by Dirac. which were independent of m. A spinless particle cannot have an intrinsic magnetic dipole moment. The ground state is necessarily / = 0 and so m = 0.010 Cambridge Books Online © Cambridge University Press. consider a transition from the first excited state of the hydrogen. The n = 2 level can be I = 0. Its magnitude may be found from experiment.22). Downloaded from Cambridge Books Online by IP 150.

Hu(r.\wox. 2014 . http://dx. Spin precession 89 There should be another term in Hi.UJSX = Ho . for example it might contain a term involving az as in (8.010 Cambridge Books Online © Cambridge University Press. Suppose that its normalised time-dependent wave function is ${r. its precise form may be calculated from Dirac's relativistic equation.22) is the interaction of the intrinsic magnetic dipole moment with the field. where a charged particle in orbit behaves effectively like a loop of electric current. so we may think of (8. So we expect Hi to include a term that represents the interaction between the two magnetic moments. though it is relatively unimportant unless the external field B is very weak. we must use the time-dependent Schrodinger equation and allow the spinor to depend on t as well as r.216) as the interaction with the field of a magnetic dipole moment —(eh/2M)L.1017/CBO9781139171243. so when B is large enough this extra term is less important than the ones we have described. (8.22).208 on Thu Feb 13 17:12:37 GMT 2014. The proton in the nucleus also has a magnetic dipole moment. If we are interested instead in a state that is not stationary. t) = ih—u(r. The Hamiltonian then becomes H = Ho . so that it is described by the spinor ( (8. which interacts with that of the electron.221. (8.246) Suppose that at time t = 0 a. It satisfies the time-independent Schrodinger equation Hu(r) = Eu{r).13) we introduced a two-component spinor u(r) to describe a station- ary state of a spin-1 particle. magnetic field B is switched on pointing in the rc-direction.23a) The Hamiltonian H is represented by a 2 x 2 matrix.t) and that its spin points in the ^-direction.24a) with H0V(r.org/10. This is familiar from classical electromagnetism. This extra term is called a spin- orbit interaction.25a) Downloaded from Cambridge Books Online by IP 150. Just as (8. Spin precession In (8. and a loop of electric current has an associated magnetic dipole moment. (8.t) = ih—V.236) A simple example is a proton which. (8. t). Ev- idently it does not depend on B.doi.244. before any magnetic field is switched on is in a bound state corresponding to some spin-independent Hamilto- nian Ho. again this is relatively unimportant if B is not too weak.

org/10. (8. «(r.0 = • ( . We look for a solution to the time-dependent Schrodinger equation (8.26) Substitute this into (8.28c) We have said that the spin state at t = 0 is such that t*i(0) = 1 and « 2 (0) . (8.244.11a) for ax: 0 -±7i( fa 0 JU2(t)J -""dt\u2{t) or = ik diUl^ ih^-u2(t).90 8 Spin where u) = 2hvB.221. (8.256) with /i the magnetic moment of the proton. 2014 . So. so that Uo = Vo = 1.28a) and (8. i^-r~ = §ti>V.276) dt If we introduce = U1(t)+U2(t). from (8. = Ul(t)-u2(t). (8. .doi.246) and the representation (8.286) Their solution is = Uoe-*iwt.0.010 Cambridge Books Online © Cambridge University Press. The effect of adding a spin- dependent term to the Hamiltonian is to make the spin state of the proton change with t. V(t) = V0e5'Mt. Downloaded from Cambridge Books Online by IP 150. http://dx.208 on Thu Feb 13 17:12:37 GMT 2014.28c).236) of the form (8.276) become simple: ift-rr = ~^0JU.236) and use (8.26).28a) the equations (8. (8.1017/CBO9781139171243. (8.

Because the magnetic moment of the proton points along its spin vector.306) What we have found is that.221. http://dx. Problems 91 Because the original wave function \&(r. this causes the emission of electro- magnetic radiation with angular frequency u. show that the spinors given in (8. the spin vector steadily rotates in the yz plane with angular frequency u: this is called spin precession.11a) find representations for its corresponding eigenstates. A specimen is subjected to a magnetic field. the spinor u(r.org/10.30a) Using the representation (8. Conversely. (8. The expectation values of the components of the spin vector 5 vary as if.1017/CBO9781139171243.2 What are the eigenvalues of ax? With the representation (8. Just as in classical physics. 8. Evaluate also S-U± and S Downloaded from Cambridge Books Online by IP 150. at later times there is non-zero probability that a measurement will find that it points in the — z-direction.208 on Thu Feb 13 17:12:37 GMT 2014. (8.t) is normalised to 1 also: it satisfies the normalisation condition (8.t).1 Derive the commutation relations (8.2).cosu.£. if initially we know that the spin points in the +z direction.14) for all t. Problems 8.244. 8. The expectation values of the components of S at time t are (5) = fd3rJ(r. The probability that a measurement of Sz gives + | at time t is cos2 \ut.116) obey S-U^ = u± and S+u± = u^.t) \au(r. Electromagnetic radiation is shone on the specimen and measuring the frequencies that are absorbed by dif- ferent parts of it gives information about the magnetic moments of the nuclei concerned and hence about its structure. so that the spins of the protons within it precess. and because cos2 ^cut + sin2 \ut = 1.3 With the definitions S± = (Sx ± iSy). t) was supposed to be normalised to 1.010 Cambridge Books Online © Cambridge University Press. 2014 . if electromag- netic radiation of angular frequency u is shone on the precessing proton.11) for the components of a we find (5) = (0. An important application is to nuclear magnetic resonance imaging. in classical terms.£). used in medicine and chemistry. when its spin precesses the direction of its magnetic moment vector also oscillates. there is a relatively high probability that it is absorbed.-sinu.doi.\ is sin2 \wt. and the probability that it gives .

1017/CBO9781139171243.5) give as the eigenvalues of S2 ? It is clear that the two-particle spinors u^ and u^ are both eigen- spinors of Sz with eigenvalue 0.] 8.208 on Thu Feb 13 17:12:37 GMT 2014. such that the angle between the directions of B and B' is 45°. Show that the superpositions x^ = ^(uti ^ ^It) a r e e ig e n " 2 spinors of S . Use perturbation theory to calculate the new stationary-state energy correct to order B' and compare with the exact answer. verify that both u t and satisfy 2 8. [Hint: use the results of the previous question. A small additional magnetic field B' is switched on. http://dx.5 A particle of spin | and magnetic moment | TIJAG is at rest with its spin pointing along a magnetic field B.org/10.92 8 Spin By writing S2 = S2 + §(S+ S_ + 5-54.doi. 2014 . What can be said about a further measurent of the y-component of its spin? Downloaded from Cambridge Books Online by IP 150.). 8.4 The operators S^1) and S^ are the spin vectors of two electrons and the components of S^ commute with those of S^2\ The total spin vector is the operator S — S^ + S(2\ What are the eigenvalues of Szl What does (8.6 An electron is in a state in which the y-component of its spin is known to be + | .010 Cambridge Books Online © Cambridge University Press. however. calculate the probabilities of obtaining each of the possible results. they are not eigenspinors of S2.244. The x-component of its spin is now measured.221. and find the eigenvalues.

Let the stationary-state wave functions of Ho be $^ = (j)k^~iEkt^h. because the higher states can spontaneously emit ra- diation and so change their configuration. If the system is initially in a state other than the ground state. Resonant absorption and stimulated emission We consider the interaction between an electron bound to an atom and an electromagnetic wave. On the other hand. corresponding to energies E^. By taking all the levels E^ to be stationary states. in which case a spontaneous emission is said to occur.011 Cambridge Books Online © Cambridge University Press. 9 Masers and lasers Radiative transitions When a system changes its energy as the result of the emission or the absorption of a photon. Our calculation will consider only the stimulated emission (and absorption) induced by the external 93 Downloaded from Cambridge Books Online by IP 150. it will allow us to understand the main features of the proper. In the absence of the electromagnetic field. more exact treatment. which we describe in this chapter. Although the model is very simple.244. it can emit a photon and so shed energy. In this case the emission process is known as stimu- lated emission. the electron wave function satisfies the Schrodinger equation with Hamiltonian HQ. it is said to undergo radiative transition. only the ground state is exactly a stationary state.208 on Thu Feb 13 17:12:49 GMT 2014. In the presence of an electromag- netic field the system can absorb a photon. Stimulated emission is the basis of maser and laser action.doi. so that its energy is raised to a higher level. the two types of radiative transition where there is an external electromagnetic field.org/10. 2014 . This can happen with- out the presence of any external electromagnetic field. http://dx. In reality. We shall deal mainly with absorption and stimulated emission. the probability that it will emit a photon can be greatly enhanced.221. First.1017/CBO9781139171243. we are making an approximation in which the possi- bility of spontaneous emission is neglected. if the excited system is placed in an electromagnetic field that varies in time with the appropriate frequency. we construct a simple model that allows us to investigate how the internal structure of a quantum system changes as the result of a radia- tive transition. There are three kinds of radiative transition.

we must impose the constraints. parallel to the wave vector K.5) JO where ufi = (Ef .21)) is the squared modulus of 10 }.6) Downloaded from Cambridge Books Online by IP 150.20) and (7. Imagine that the field is switched on at time t = 0 and is switched off again at some time T. after the field has been switched off. u = cK. (9.1017/CBO9781139171243. (9.2) In order to satisfy Maxwell's equations (which are listed in appendix D). We shall take this field to be the plane wave given by 0 = 0. K'A0 = 0. (9. the electric and magnetic vectors are S = UJAQ COS(K • r — ut) B = K A Ao cos{K • r .Ei)/h.1) Then. (9. We suppose that the electron was initially in the eigenstate <!>i of Ho. This transition probability (see (7. will find that a transition has occurred to some other eigenstate $ / of Ho.ut).94 9 Masers and lasers electromagnetic field.4) We suppose that the field is weak enough to allow us to use the time- dependent perturbation theory developed in the last chapter.20) and (9. 2014 . http://dx. A = Ao sin(K • r . is in the ^-direction.cut).011 Cambridge Books Online © Cambridge University Press.221.org/10. Hi — (eAo/m)px sin(if z — u)t) = (eA0/2mi)px[e[iKz-^ .244.1). from (8. (9. and such that Ao is in the ^-direction. We shall choose the directions of the coordinate axes such that the direction of propagation.17).e-K**-"*)].208 on Thu Feb 13 17:12:49 GMT 2014. according to (8.doi. Then. and calculate the probability that a measurement.3) The second of these equations also makes A satisfy the gauge condition V • A = 0. (9.

where Ef > E^ so that cj/i > 0. but the result is perhaps surprising for the case of emission: the presence of the field stimulates the electron to emit a photon. Unless u is within a distance of about 2TT/T from a. this estimate ignores the fact that excited levels of atoms have a finite lifetime and so.doi.) Let us consider what happens when u. which may be neglected in comparison with it.7) leads to an analogous expression. In either case. ^ ^ . The integral (9.244. is equal to CJ/J. their energy is not precisely determined. the second term in (9. the first term becomes dominant and the transition probability is K / l h2 {ufi-u)* • (9 8) ' In the case of emission. in fact.208 on Thu Feb 13 17:12:49 GMT 2014. the frequency corresponding to the energy Downloaded from Cambridge Books Online by IP 150. so that when T is equal to one second. we see that the transition probability is proportional to the square of the intensity of the field. . The term with the small denominator then dominates over the other term. In practice. and the more complete treatment in which the field is quantised shows that.8): see (7./*.221..25) and figure 7. it is easy to understand intuitively that the probability should increase with the field intensity.011 Cambridge Books Online © Cambridge University Press. In the case of absorption.org/10. the transition probability is very small. Then the denominator of one of the two terms in (9.7) a Ufi — UJ ujfi + UJ Suppose that the angular frequency u of the external field is chosen such that hu is close to the difference between the energies E{ and Ef of the initial and final states. the electron has absorbed a single photon from the field. where Ufi < 0. because of the energy uncertainty principle which we described in chapter 7. (However. In the case of absorption. Resonant absorption and stimulated emission 95 If Ef > Ei. (9. Ufi is often in the optical- frequency range. If Ef < E^ the electron has emitted a single photon. and so u) has been tuned to give a resonance in the transition probability. We have already encountered a function very similar to (9.5) gives aif = .1017/CBO9781139171243.1. the frequency of the incident radiation.7) is small. the transition occurs readily only if u differs from Ufi by less than one part in 1014. http://dx. Allowing for this typically increases the frequency spread to about one part in 105. the transition must have occurred through the absorption of energy from the electromagnetic field. 2014 .

6) for uffi by unity: (9. as we have already remarked. that is the number of states in the range (Ef. and take the origin at the nucleus of the atom.. observed to be constant. but rather to a narrow range of states.1).244. Hence it is a good approximation to replace the exponential in the expression (9. http://dx. we obtain the transition rate I\_+/ = — luji \2g(Ei + fiu)).doi.1017/CBO9781139171243.221.Ef + AEf) is given by Here g(Ef) is the density of states at energy Ef.8) becomes \aif\2 = lu^^T2/^2.12a) Downloaded from Cambridge Books Online by IP 150.9) is over the energy interval AEf.9) can then be rewritten as (9. The integration in (9. However. Then if ^ and c/>f are electron wave functions corresponding to low-lying atomic levels.011 Cambridge Books Online © Cambridge University Press. (9. 2014 .208 on Thu Feb 13 17:12:49 GMT 2014. Then (9. Making the change of variable a = |(cj/i — o. We can reconcile these facts if we consider not a transition to a single final state / . (9-11) This is known as Fermi's golden rule. they are small unless Kr «: 1 (see problem 9.8). The probability P of a transition from state i into any state whose energy is in the range {Ef. however. Recall that \aif\2 is the probability that a transi- tion has taken place by time T. the transition rate is. that is \a>if\2 is linear in T.8) is extremely small unless Ufi is almost equal to a. In macroscopic systems. it is necessary to calculate u^K Consider a hydrogen atom for definiteness. for large T the transition probability (9.10) Dividing by T.org/10.96 9 Masers and lasers difference between the initial and final states.Ef + AEf) is g(Ef)AEf. so that it is an excellent approximation to replace g(Ef) and | ^ | 2 with their values at E = (Ei + TIUJ) and to extend the integration to ±oo.)T. Electric dipole transitions In order to evaluate the transition probability (9.

the electron in orbit round the nucleus is equivalent to an electric current loop.— exfc) vanishes: compare the result (7.6) by unity.13) give Since Ho = (p2x + p* +p2z)/2m + V(r). In consequence. (9.14).221.org/10. Inserting this relation into (9. and using the fact that fc and </>/ are eigenfunctions of #o. http://dx. and also terms corresponding to magnetic dipole and multipole moments. If we had not replaced the exponential in (9. this gives two relations which when added together yield [x.pz] = 0. The magnetic terms may be understood from the fact that. [x. are known as electric dipole transitions. according to the commutation relations (4. we obtain u% = [Ao(Ei .1017/CBO9781139171243.011 Cambridge Books Online © Cambridge University Press. Downloaded from Cambridge Books Online by IP 150. This leads to certain selection rules: radiative transitions readily occur only between certain pairs of atomic levels. for many pairs of levels E{ and Ej the quantity (</>/.H0] = {m/ih){xH0 .Hox). 2014 . the transitions corresponding to the approximation we have made. However. The ammonia molecule 97 Now. the other two relations in (9.px] = 2ihpx.py] = [x.28). we thus have px = (m/ih)[x. (9. -excfn).244.24) for the case of a particle in a box.px]=ih. that e±lKz « 1. crudely speaking.12a). the corresponding lines in the emission or absorption spectrum of the atom are much more pronounced than those that are associated with transitions that can occur only via a magnetic or higher electric multipole mechanism. we should have obtained additional terms corresponding to electric quadrupole and higher multipole transitions.126) Now. —ex is the rr-component of the electric dipole moment of the atom. The transition probabilities are greatest for the electric dipole transitions. [x. which gives the atom a magnetic moment that interacts with the electromagnetic field. and V(r) trivially commutes with x. For this reason. (9.Ef)/2h]{(j>f.doi. Similarly.13) Pre-multiply and post-multiply the first of these relations in turn by px. It may be shown that spontaneoues emission is subject to exactly the same selection rules.208 on Thu Feb 13 17:12:49 GMT 2014.

Thus it is a very complicated system: there are four nuclei and.org/10.221. the spin vector 5 is normal to the plane defined by the three hydrogen nuclei (that is. when the molecule has no net charge.1 (a) simply by turning the molecule over. namely the total angu- lar momentum. However. and so would not correspond to a different internal state. It turns out that in the ground state the total angular momentum of the molecule is non-zero: it has non-zero spin arising from the sum of the orbital-angular-momentum and spin vectors of its separate constituent particles. http://dx. correspond to differ- ent quantum-mechanical states.1). S is the spin vector and /X is the electric dipole moment. The direction in which S points distinguishes the two sides of this plane.1(6) would be obtained from figure 9.1017/CBO9781139171243. so that the two configurations shown in figure 9.98 9 Masers and lasers (a) (b) Figure 9. In the ground state. ten electrons.1. figure 9.208 on Thu Feb 13 17:12:49 GMT 2014. the expectation values of the spin components in directions parallel to the plane are zero). only one of these need concern us here. If the molecule had no spin. a large number of parameters would be needed.244. Consider the ammonia molecule in a coordinate frame in which its centre of mass is at rest. To specify fully the states of motion of the four- teen constituent particles. A full calculation of the energy levels of this fourteen-body system is much too difficult to undertake. 2014 . but the system has an important property which we can understand by making an analysis similar to that given in chapter 6.doi. in which the nitrogen nucleus is positioned on opposite sides of the plane. Two configurations of the ammonia molecule (neither is a stationary state). where we described the ionised hydrogen molecule.1. The ammonia molecule The ammonia molecule is a bound state of an atom of nitrogen together with three hydrogen atoms (figure 9. Downloaded from Cambridge Books Online by IP 150.011 Cambridge Books Online © Cambridge University Press.

208 on Thu Feb 13 17:12:49 GMT 2014. Be- cause of the symmetry of the system.17) corresponds to a frequency of 24000 MHz. Be- cause of the tunnel effect. in terms of </>i and (j)2 only. (9. These wave functions differ only in that they make the expectation value of the coordi- nate of the nitrogen nucleus appear on different sides of the plane defined by the expectation values of the coordinates of the hydrogen nuclei. respectively. unlike in chapter 6. the molecule is unlikely to be found in one of these states unless the tem- perature is so high that there are frequent and violent collisions between molecules. the wave func- tions fa and <f>2 are functions of many coordinates. for example. This level separation is very much less than the energy needed to excite the state of next-lowest energy. the l / \ / 2 is just a normalisation factor).1017/CBO9781139171243.15) applies.1(6). rotational excitations of the whole molecule or excitations of its separate atoms. If this were not the case. 12.£ + = hu0 (9.221.3) and (6. 2014 . an analysis based on an expansion of the wave function similar to (6. The corresponding wavelength. and as in chapter 6 the stationary states are V>± = ( 0 i ± f c W 2 (9. It is found experimentally that the level separation E_ .16) (see (6. #11 = # 2 2 and H12 = H2i. http://dx.011 Cambridge Books Online © Cambridge University Press. although the molecule has very many other stationary states. Without making a detailed calculation. would not have been a good approximation. The cor- responding energies E± are to be calculated from a very complicated four teen-particle Hamiltonian H and. This is the frequency of the photons that are emitted or absorbed in a transition between the two lev- els.6). The calculation is not possible in practice.doi. The ammonia maser 99 Let the wave functions fa and $2 describe configurations of the molecule corresponding to figure 9.org/10. the wave func- tions (j>i and (fi2 do not describe stationary states.1 (a) and figure 9.244. is in the microwave range. it may however be shown that the ground state of the molecule corresponds to the wave function ?/>+. transitions occur between these states.15) where H^ = ( As in the discussion of the hydrogen molecule in chapter 6. Downloaded from Cambridge Books Online by IP 150. Thus.5 mm. corresponding to.3). Their significance is merely that they represent configurations for which (9.

be tuned.) We now consider in some detail the two stages in the operation of the maser. at least partially. the electrons of the molecule tend to be found closer to the nitrogen nucleus than to the plane containing the hydrogen nuclei.100 9 Masers and lasers The ammonia maser The word maser is an acronym for microwave amplification by stimulated emission of radiation.doi. those molecules in the lower level E+ from those in the higher level E-.org/10. using ammonia gas. Hence the maser acts as an amplifier of any external radiation of frequency OJQ that is fed into the cavity. At normal temperatures. so that the two states fa and fa have equal and opposite dipole moment. as shown in figure 9. The component enriched with molecules in the higher level is then squirted through a resonant cavity. The emission process is stimulated emission. (9.o. The first maser was constructed in 1954. to some extent. tuned to angular frequency u. for example by placing the maser in a magnetic field.221. Thus. The first stage in the operation of the maser is to separate. 2014 . so that some of the molecules emit photons of energy TILOQ and emerge from the cavity in the lower level E+.208 on Thu Feb 13 17:12:49 GMT 2014. it would need a very strong external field £ for this not to be the case. This dipole moment exists because the molecule is not spherically sym- metrical. so that the number of photons emitted into the cavity by the gas increases if the intensity of the radiation already present in the cavity is increased. http://dx. H22 = Eo + vo'£.1017/CBO9781139171243. that is. the frequency LOQ at which it operates can be varied. The separation The separation is achieved by passing the gas through a static electric field S.1. This means that small changes in the geometry of the molecule Downloaded from Cambridge Books Online by IP 150. if /i 0 is ^ e expectation value of the dipole moment in the state fa. which we described in the last chapter. normal to the plane of the hydrogen nuclei. To achieve this.244. This field interacts with the electric dipole moment /JL of the molecule.011 Cambridge Books Online © Cambridge University Press.18) where Eo is the value of Hn or H22 in the absence of the electric field. We shall suppose that £ is small compared with the internal field of the molecule. (The maser can. In fact.—£?+). very nearly equal numbers of molecules of the gas are found in each of the two lowest levels E+ and E-. one has to vary the energy gap (E.

is made to pass into the resonant cavity. the energy of the upper level E.17).18) for Hn and #22 in equations (6. so that in the expressions (6. http://dx.6) for E± we may neglect Ku compared with 1.208 on Thu Feb 13 17:12:49 GMT 2014. the two components are deflected in opposite directions (figure 9. depend mainly on the overlap if 12 = (</>i.doi. the two components of the gas will experience forces -VE'± = ±(l/huo)V(tio • Sf (9. and it is sufficient to account for it classically. Hence not only the dipole moment /io can be taken to be unchanged by £.has been increased a little by the presence of the electric field. and the energy of the lower level E+ has been decreased.1017/CBO9781139171243.221. ^2) between the wave functions <f>i and fa. K\2 < 1. The component that contains the molecules with wave function ?/>-. According to (9. which. as in (6.21) in opposite directions.20) Thus. £ will still be constant on the atomic scale of length. In practice.to be huo.16). The ammonia maser 101 due to the applied field £ can be neglected. The stationary-state wave functions also will be changed a little from the ij)± of (9.org/10. 2014 . we define the difference between E+ and E. elim- inating c\ and C2 from them we find that in the presence of the electric field £ the energies E± are changed to (9.6) gives H12 = -±huj0.13). however. Downloaded from Cambridge Books Online by IP 150. in the upper of the two energy levels. That is. so that (6.19) If we now use the values (9.4).011 Cambridge Books Online © Cambridge University Press. since in practice \£\ is small on the atomic scale. In practice.244.2). but also H12 and #21. So if the gas is squirted through a region where £ varies across the jet. whatever the direction of £ relative to that of /LIQ. Suppose now that the magnitude of £ varies with position. this change is small. (9. so that we need not take account of its variation in the quantum-mechanical Hamiltonian.

(9.23) and the relation UJ = cK. However. (9.doi. B = (0.23) The need to satisfy the boundary conditions at z — 0 and z — a has led us to consider a standing wave (9.2).208 on Thu Feb 13 17:12:49 GMT 2014. The presence of an electromagnetic field (9. Suppose that a jet of ammonia gas enters the cavity with each of the molecules initially in the ^_ state. the analysis of the interaction of the radiation (9.org/10.1017/CBO9781139171243. with empty space between. 0.22) where Because the planes z = 0 and z = a are conductors. n integer. z = 0 and z = a. Schematic diagram of the ammonia maser. 0). Stimulated emission in a radiation cavity Consider two parallel conducting planes. Because of (9.244. However. (£0/c) cos Kz smut.22) with the ammonia molecule can be carried through in a way closely similar to that described earlier.102 9 Masers and lasers Output Input Ammonia jet Separator Collimator Resonant cavity Figure 9.221. unless u is tuned so as to be very close to U)Q. we must have £x — £y — 0 o n them. This requires that K = riir/a.22) will stimulate the molecules to emit photons and so make transitions downwards in energy into the i/>+ state. 0).011 Cambridge Books Online © Cambridge University Press. http://dx.2. A solution to Maxwell's equations (listed in appendix D) in the space between the planes is £ = (£Q sin Kz cos art. the probability of such transitions occurring will be very small.22). 2014 . instead of a travelling wave as in (9. this tuning may be achieved by suitable choice of the distance a between Downloaded from Cambridge Books Online by IP 150.

The momen- tum vector of each photon is normal to the planes. The standing wave (9. so that the field builds up rapidly. as the other photons that are already present in the cavity. This is an important factor in the design of the maser.22) corresponds to a collection of photons that bounce back and forth between the two conducting planes. and the photon that is emitted adds to the intensity of the electromagnetic field (9.208 on Thu Feb 13 17:12:49 GMT 2014. The molecules continually collide with one another and so exchange energy.1017/CBO9781139171243. the original electromagnetic field (9. 2014 . Even a small quantity of gas contains a very large number of molecules — about 1025 kg" 1 . the net effect is zero: there are as many upward transitions that absorb photons as there are downward transitions that emit them.244. and therefore from (2. That is. and when a photon hits one of the planes its momentum is reversed. equal mixture of the two com- ponents ?/>+ and ip. http://dx. What our simple discussion does not show is that each extra photon that results from the stimulated emission actually has the same wave vector. If this is allowed to occur. When u is tuned in this way. until it has left the cavity it is now liable to absorb radiation and so make a transition back to the state V>-.22). so that the probability of further molecules undergoing stimulated emission is then enhanced.2) the same momentum. Population inversion the conducting walls of the cavity. The reason for this is that once a molecule has undergone stimulated emission and so made a downward transition to the ground state ?/>+.011 Cambridge Books Online © Cambridge University Press. It therefore exactly reinforces the existing field. Population inversion Because a molecule in the ground state -0+ is liable to make a transition that absorbs energy from the field instead of adding to its intensity. If an unseparated. This problem is avoided by ensuring that the molecules which have dropped into the ground state escape through the exit hole before an appreciable probability of reabsorption occurs. the emission occurs fairly readily. Each transition further reinforces the field and so increases the probability of yet more transitions occurring.22) rapidly becomes amplified.221. It is completely impracticable to try to follow the Downloaded from Cambridge Books Online by IP 150.org/10. one must consider in more detail the equivalence between electromagnetic fields and systems of photons. the initial separation of the gas into two components is a vital stage in the operation of the maser. To show this.enters the cavity. which is beyond the scope of this book. it decreases the intensity of the field in the cavity instead of amplifying it. The speed with which the ammonia gas passes through the cavity deter- mines the time that the molecules spend in the resonant system.doi.

In certain circumstances it is important to take account of quantum- mechanical effects. i — 1. or Kelvin. Then according to classical statistical mechanics. Let these have energies £?». A remarkable result of statistical physics is that it predicts the average distribution of the states of the molecules without knowing anything in detail about what happens in the collisions.24) adequately describes the distribution of states of the molecules in the gas. the average number U{ of par- ticles expected to be in the ith level at any given instant is given by the Maxwell-Boltzmann distribution: m = Cgie~*i/KBl.24) Here T is the absolute temperature and ks is a universal constant. one needs to know how many molecules at any instant may be expected to be in a given quantum-mechanical state. The collisions become more frequent and more violent as the tem- perature of the gas increases.org/10. since by definition the average kinetic energy of the molecules is proportional to the temperature (measured on the ab- solute. Instead.221. scale).25) Evidently C varies with the temperature T.208 on Thu Feb 13 17:12:49 GMT 2014. here we just quote the results. with corresponding degeneracies gi (so that the level E{ is associated with gi independent states). Suppose that each particle may be found in any one of a number of different states. Consider any system of particles that is in dynamical equilibrium. The demonstration of this may be found in any book on statistical physics. but at room temperatures (9.. The Maxwell-Boltzmann distribution is a formula of classical mechan- ics. one resorts to statistical methods.2. The constant C may be calculated in terms of the total number N = ]T^ n^ of particles: C =N V>e-*'/*Br . There is little interest in determining at any instant the state of a given molecule.doi. (9.104 9 Masers and lasers effect of the collisions on the state of each individual molecule. so that in particular no heat flows into or out of the system. known as Boltzmann's constant. and it turns out that this is quite sufficient to give an accurate description of the macroscopic properties of the gas.. (9. That is. what matters is the configuration of the system as a whole.011 Cambridge Books Online © Cambridge University Press.. Because the energy Downloaded from Cambridge Books Online by IP 150.. http://dx. 2014 . rather than which particular molecules are in that state.244.1017/CBO9781139171243. The molecules continually collide with each other and so change their states.

The laser 105 difference between the levels E+ and £L of ammonia is small. whose operation was announced in 1960. the material that emits the photons is kept stationary within the radiation cavity.208 on Thu Feb 13 17:12:49 GMT 2014. light with a wide range of frequencies. If it is left to return to equilibrium. instead of being squirted through the cavity.doi. the populations of the higher-energy levels are rather smaller than those of the lowest-lying levels E+ and E-. and the gas is not in dynamical equilibrium. The end faces of the ruby cylinder were rendered accurately parallel to each other and then both were silvered.011 Cambridge Books Online © Cambridge University Press. When the flashtube is fired. and nowadays also in masers. the population of the lower level E+ being only slightly larger than that of the higher level E-. The flashtube emits white light. This laser had as its active medium a cylinder of synthetic ruby. one fully and the other partially. A resonant cavity of this form was used in the first successful laser.24) we may see also the truth of our earlier assertion that. If the amplifying material is solid. In this component the majority of the molecules are in the higher level. unless the temperature T is high. The active ingredient of the ruby is the chromium. Population inversion was achieved by a process known as optical pumping.org/10. intermolecular collisions ensure a rapid return to the Maxwell-Boltzmann distribution.244. whose (simplified) energy-level diagram is shown in figure 9. The separation process achieves a population inver- sion in the component of the gas that is injected into the cavity. which is an aluminium oxide crystal containing a small percentage of chromium. As we have explained.24) that in a quantity of gas that is in dynamical equilibrium the numbers of molecules in these two levels are very nearly equal. the relative probabilities of transitions between different levels of a sys- tem are governed by selection rules. http://dx. that is. 2014 .1017/CBO9781139171243. In lasers. The laser The principles underlying the operation of the laser are much the same as those of the maser. most of which dissipate their en- ergy as heat. From (9. which uses a helical flashtube wrapped around the ruby cylinder.221. the ruby crystal is flooded with photons. the basic difference being that the photons produced correspond to the optical range of frequencies. we deduce from (9. some in the blue and green ranges of the spectrum are absorbed by the chromium and cause transitions from the ground state to one of the two energy bands shown in the figure.3. However. In the case of the energy bands it Downloaded from Cambridge Books Online by IP 150. the parallel mirrors that form the radiation cavity may be the end faces of the material itself.

a standing-wave electromagnetic field of the form (9.org/10. but rather it appears in the form of phonons. First.doi. a large number of chromium ions are driven into the metastable state: a large population inversion occurs.3. In this way. for example the semiconductor lasers described in chapter 12. The partially silvered end of the crystal allows the leakage of an intense pulse of laser light.) So if the flashtube is fired with high power. The metastable state decays to the ground state by spontaneous emission of photons. turns out that the preferred decay is to a metastable state. http://dx.1017/CBO9781139171243. it can be very intense: the power per unit area of cross-section in a laser beam can be at least 109 times that which may be obtained from a conventional Downloaded from Cambridge Books Online by IP 150.221. The light emitted by a laser has two important properties. (Its lifetime is rather long on the atomic scale: a few millisec- onds. but just a few move in the right direction to be reflected back and forth between the two silvered end faces.) Most of the photons emitted by sponta- neous emission pass out of the crystal and are lost. compared with the 10~8 s typical of spontaneous transitions when dipole transitions are allowed.208 on Thu Feb 13 17:12:49 GMT 2014. 2014 . Energy levels in a ruby crystal.244. (This decay is a non-radiative transition: the energy is not given up by the emission of a photon.22) begins to be set up. Laser technology has developed very rapidly since 1960. also indicated in figure 9. so that the intensity of the field (9. This is the output of the pulsed ruby laser.106 9 Masers and lasers Energy Non-radiative transitions Resonant Metastable state absorption of green and Stimulated emission blue light Ground state Figure 9.011 Cambridge Books Online © Cambridge University Press.22) builds up very rapidly as the metastable state depopulates.3. which are quanta of vibration of the other atoms in the crystal lattice about their mean positions. and there are now many types of laser. This field stimulates the emission of further photons.

an inter- ference pattern is established: if the two components of the light reaching P are exactly out of phase. but these are extremely small. The two sources S and O will be mutually coherent. http://dx. Light reaching a point P directly from a coherent source S and reflected from a point object O. light source. In a laser beam.011 Cambridge Books Online © Cambridge University Press. the phase difference between the light at points along the beam separated by as much as 100 km is likely to be unchanging with time. mainly caused by fluctuations in the distance between the mirrors resulting from thermal and mechanical disturbances. There are some variations in phase of the field as a whole. the stimulated emission is precisely in phase with the electromagnetic field that causes it. then P is a point of maximum brightness. Holography 107 Figure 9.208 on Thu Feb 13 17:12:49 GMT 2014. An important application of the high degree of coherence of laser light is to holography.4). Secondly. 2014 . such as non-linear optics.doi. Holography The high intensity of laser light has obvious applications. monochromatic point source of light S at a dis- tance R from a point object O (figure 9. then there is darkness at P. that is. and hence O behaves effectively as a second point source. In a conventional light source.1017/CBO9781139171243. Light that comes from S in the direction of O will be scattered by O. In a laser. such as to cutting and welding in engineering or medicine. laser light is not only highly monochromatic.org/10. It also makes possible the study of fundamental phenomena. Imagine first a coherent. The condition that Downloaded from Cambridge Books Online by IP 150.4. Hence.221. the phase difference between the light from them will be constant in time at any point P in the space surrounding them. but also highly coherent. We digress briefly from our main subject of quantum physics in order to describe this technique of three-dimensional photography. the atoms decay in an uncorrelated fashion and so the phases of the electromagnetic wave packets associated with the different photons are random. while if they are in phase. arising from the changes in the properties of materials when they are subjected to very strong electromagnetic fields.244.

org/10.5. and a is a possible phase change induced by the reflection of the light from S on the object O. darkness.26a) where = 0.4. 2014 .221. (9.108 9 Masers and lasers (a) Photographic plate (b) T Figure 9.208 on Thu Feb 13 17:12:49 GMT 2014. http://dx.doi.244. there be destructive interference at P. (a) Interference fringes corresponding to figure 9.1017/CBO9781139171243. Similarly. v\ and r 2 are the distances of P from S and O as in figure 9. (6) Reconstruction of a virtual image O1 and a real image O". using a light source Sf.±2. with a photographic plate on which a hologram is recorded.011 Cambridge Books Online © Cambridge University Press..4. the condition that there be constructive interference Downloaded from Cambridge Books Online by IP 150. that is. is K{rx -R-r2) + a = (2m + 1)TT.±l. Here K is the wave number of the light (so that its wavelength is 2TT/K).

6. To view the holographic image. This plate records a hologram.56). (9. which is a thin section of the interference pattern.1017/CBO9781139171243.266) where The points P that satisfy either of the two conditions (9.221.244. Diffraction occurs in two directions.5(a). Set-up used to produce a hologram of a three-dimensional object. the source S' does not have to be coherent.r2) + a = 2ra'7r. To view the Downloaded from Cambridge Books Online by IP 150. The light from S' that falls on the hologram is diffracted by the light and dark fringes recorded on it. Holography 109 Mirror Photographic Laser plate Mirror Figure 9. with a high density at points P correspond- ing to constructive interference.doi. at P is K(rx . 2014 . Unlike 5. having the same position relative to it as the original source S and emit- ting light of the same colour. one illuminates the hologram with a source S". as is shown in figure 9. and the other such that the rays converge on the point O" that is the reflection of O' in the photographic plate (see figure 9.org/10. Thus there is a virtual image at O' and a real image at O". http://dx. one such that the rays emerging on the other side of the hologram appear to have originated from the point O' corresponding to the position of the original object O. Development of the plate causes silver grains to be deposited in the emulsion.R . In this figure we show also the cross-section of a pho- tographic plate that has been placed so as to intersect the interference pattern.011 Cambridge Books Online © Cambridge University Press. and low density in between at points of destructive interference.208 on Thu Feb 13 17:12:49 GMT 2014.26) lie on a sys- tem of hyperboloids of revolution having O and S as foci.

What would be the temperature of the black body that emitted photons at the same rate. The real image is made apparent by placing a screen behind the plate. By ap- plying the uncertainty principle to estimate the transverse momentum of the photons. In practice. A helium-neon laser operating at a wavelength of 633 nm with an output power of 1 mW has a linewidth of 1 kHz and produces a beam of cross-sectional area 10~6 m2.3 For a black body. Verify that the ground-state wave function (5. the power emitted per unit area in the wavelength interval from A to A + dA is (2nhc2/\s)[d\/(ehc'Xk*T .011 Cambridge Books Online © Cambridge University Press.org/10. passing through the point O". 9.244.221.1)]. one looks at the plate from behind but focuses the eye on O' because this is the point from which the rays appear to have diverged. 2014 . show that a helium-neon laser beam which initially has diameter 1 mm can be collimated to within an angle of about 3 x 10~3 degrees.18) is small for all values of r that do not satisfy Kr <C 1.110 9 Masers and lasers virtual image. and the other is reflected onto the plate by the object.6.4 The beam from a laser is usually extremely well collimated. The beam is spread out by a lens and split into two parts.17)).doi. show that the stationary states ip± have zero dipole moment. because a suitably coherent point source of light S is not. http://dx. this is why holography produces a three-dimensional effect. The interference fringes now have a much more complicated geometry. The set-up needed to produce a hologram of an extended three-dimensional object is shown in figure 9. 9. or by placing the eye beyond the point O" and focussing it on that point (see figure 9. a laser beam is used to produce the hologram. Problems 9.5(6)).208 on Thu Feb 13 17:12:49 GMT 2014. in fact.2 Assuming that the states fa and fa of the ammonia molecule are eigen- states of the electric-dipole-moment operator. in the same frequency interval and from the same area? 9. but the principles involved are the same as for a point object.1 The wave number K corresponds to the radiation associated with transitions between the two lowest energy levels of the hydrogen atom (see (5. available. How small would the filament of an ordinary household torch have to be to achieve the same collimation? Downloaded from Cambridge Books Online by IP 150. One part is reflected directly onto the photographic plate.1017/CBO9781139171243.

5 A gas of non-interacting particles of mass m is contained in a cube of side L and obeys the Maxwell-Boltzmann distri- bution.221.6 In the previous question.org/10. the Maxwell-Boltzmann distribution is replaced by the Fermi-Dirac dis- tribution where the Fermi energy ( is independent of E{.208 on Thu Feb 13 17:12:49 GMT 2014. It is permeable to all particles of the gas whose normal component of velocity is greater than u$.011 Cambridge Books Online © Cambridge University Press.1017/CBO9781139171243.doi. What form does this distribution take at T = 0. Show that as L —> oo the probability that the ^-component of the velocity of a given particle lies between u and u + du is 9.244. Write down the formula for the probability that a given particle is in the state corresponding to wave function (8/L 3 ) 1 / 2 sin(q7Tx/L) sin(r7ry/L) s'm(s7r z/L).7 In quantum statistical mechanics for a system of electrons. a membrane of area A is embedded in one wall of the cube. http://dx. Show that the rate at which particles escape from the box is (NA/L3){kBT/2<irm)1/2 e- m 9. Holography 111 9. Downloaded from Cambridge Books Online by IP 150. 2014 . and why? Notice that when E{ » £ the two distributions are approximately the same.

112 Downloaded from Cambridge Books Online by IP 150. We shall show in this chapter that. called its site in the lattice.208 on Thu Feb 13 17:12:58 GMT 2014.1(6)). 10 Band structure of crystals Electrons in crystals A crystal consists of a collection of atoms arranged in a regular array. If L is large enough. but rather to the crystal as a whole.221. These bands are separated by energy gaps that are forbidden to the electrons. We shall not find it necessary here to discuss the details of how this comes about. Suppose that the spacing L is now reduced (figure 10. each original discrete atomic energy level spreads out into a band of closely spaced levels (fig- ure 10. Each atom is more or less anchored to one point. To a reasonable approximation. In figure 10. Imagine that we can assemble a crystal of identical atoms whose spacing L can be altered at will.1017/CBO9781139171243. The band structure is all-important in determining the properties of electrons in crystals. the spacing between atoms being of the same order of magnitude as the dimensions of the atoms. In a crystal. nor shall we consider the various patterns in which the atoms can be arranged. with its electrons in discrete bound states. 2014 . the motion of a given individual electron in a crystal can be treated as a single-particle problem. as L is reduced. however.org/10.doi. particularly those that are closest.012 Cambridge Books Online © Cambridge University Press. Each atom then behaves as if it were isolated. We have seen in chapter 5 that the energy of an electron bound to an atom is restricted to certain discrete values.2). an electron is not bound to any one particular atom.244. the motion of an electron in one of the atoms will be affected to a negligible extent by the electrons and nuclei of the other atoms. The potential V(r) in the neighbourhood of a given atom is now affected by the presence of the nuclei and electrons of the other atoms. by the electrostatic forces produced by all the other atoms. with the potential determined by the electric field produced by all the other particles in the crystal. An exact calculation of this situation is a complicated many-body problem and is not tractable. http://dx.1 (a) we have drawn a schematic diagram of the potential V(r) in which an electron moves in this situation. If it is not to escape from the crystal. It will be sufficient to remember the essential feature that the structure of the crystal is periodic in space.

(b) when they are closer together.012 Cambridge Books Online © Cambridge University Press. its energy must be negative (where the zero of energy is defined to be the potential of the free space outside the crystal). we shall confine the discussion to the electrons that remain bound in the crystal.244. To explore the band structure. An electron in one of the low-lying bands. electrons in these bands are readily transmitted over the potential barriers.1(6)). Downloaded from Cambridge Books Online by IP 150. however. we consider mainly a one-dimensional model. for example. require a discussion of electron motion in more than one dimension.1. Schematic diagram of the potential in a crystal: (a) when the atoms are widely separated. this is called thermionic emission. in order to under- stand why divalent metals such as calcium and magnesium are reasonably good electrical conductors. We shall find that there are also energy bands between Vmax and 0. an electron can gain sufficient energy by collisions with the vibrating nuclei to escape from the crystal. http://dx.208 on Thu Feb 13 17:12:58 GMT 2014. If the crystal temperature is high enough.doi. Band structure 113 V(r) (a) V(r) (b) Figure 10. such that its energy is less than F max (see figure 10. 2014 . this is necessary. passes from one atom to another by tunnelling through the potential barriers between the nuclei. This provides a good guide to electron behaviour in real crys- tals. Certain aspects.1017/CBO9781139171243. However.org/10.221.

208 on Thu Feb 13 17:12:58 GMT 2014. 2014 . between the wells. The Schrodinger equation is in the wells. we consider an electron in a one-dimensional periodic square-wave potential (figure 10. we define the zero of the potential to coincide with the top of the wells.012 Cambridge Books Online © Cambridge University Press. Band structure We do not need to know in detail how the potential varies with position in the crystal. The most important feature is its periodicity.1) Downloaded from Cambridge Books Online by IP 150.org/10. This model is sufficiently realistic for many purposes.3).doi. Consider first an electron with energy E such that —U<E<0.244. For convenience.2. To have a readily solvable problem.1017/CBO9781139171243. rather than with the value of the potential in free space outside the crystal. (10. The band structure that develops when the atomic separation L is de- creased. http://dx.114 10 Band structure of crystals E A Figure 10.221.

(10. One looks for solutions of the form (10.2) consists of simple exponentials. with the same period L as the period of the crystal lattice. (10.3.L < x < 0.5) •{ Ce(a-ik)x 0 < x < a.4) \CeQX+£»e-QI 0 < x < a. (32 = (2m/hz){E + U). at least in the limit where the total length of the crystal is taken to be infinite.221.org/10.L < x < 0. 2014 .208 on Thu Feb 13 17:12:58 GMT 2014.244. The electron belongs not to one atom. which is derived in appendix E.3) Aei({3-k)x a .3) represents a wave of wavelength 2?r/A. The procedure for solving (10. Downloaded from Cambridge Books Online by IP 150. http://dx. but has equal chance of being found in the vicinity of any of them. Bloch's theorem states that solutions of the type (10.doi. We write a . (10.2) is suggested by a theorem due to Bloch. but is the same from one atom to the next..3) form a complete set.1017/CBO9781139171243.3) where k is real and Uk{x) is periodic. Then from (10. where a2 = -2mE/hz. whose amplitude Uk{x) varies across each atomic site.2) in the wells. Band structure 115 V(x) o -U Figure 10. The solution to each of the two differential equations (10. Evidently (10.012 Cambridge Books Online © Cambridge University Press. or „ _ r a2i> between the wells. One-dimensional periodic square-well potential.

11) According to (10.k)A . http://dx.2).6). and so also il>(x). when a. When this 4 x 4 determinant is expanded it results in the condition [(a 2 .9) Equations (10. the quantities a (or a1) and p are functions of E./32)/2a/3]smhaa sin/3b + cosh aa cos pb = cos kL. A.5) determines uk{x).5) we have A + B =C + D. then a 2 .ik)Ce^-[k)a -(a + ik)De^a+ik)a. then they are satisfied for every well. of course.org/10.(a + ik)D.8) are four homogeneous equations in four un- knowns. Imposing the conditions at x = 0 on the two functions in (10. (10. i(P .i(/3 + k)B = {a . it is valid also when E > 0.i(p + k)B el = (a .221.11) Downloaded from Cambridge Books Online by IP 150. or equiv- alently u and u' to be continuous. In particular. In order that they have a non-trivial solution the determinant of their coefficients must vanish.208 on Thu Feb 13 17:12:58 GMT 2014.116 10 Band structure of crystals and because we require uk(x) to be periodic.012 Cambridge Books Online © Cambridge University Press. (10.ik)C . (10.5) and to (10. and so the left-hand sides of (10. B.1017/CBO9781139171243.244. (10.7) and (10.L and U are fixed. i{p .k)A e-W~k)b .10) and (10. This.doi.3). (10. The periodicity of Uk{x) guarantees that if these conditions are satisfied at the edges of just one well. defined in (10. for a < x < L we replace x by (x — L) in the first equation of (10.10) Although we have derived this equation for the case where E lies in the range -U < E < 0.2).7) Prom the conditions applied at x = a to the second function in (10. is the reason for looking for a solution of the Bloch type (10. 2014 . is negative and in (10. (10. everywhere.5): At both edges of each well we require if) and rj)1 to be continuous.10) we write a = \al and so obtain [-{a12 + p2)/2a'P]s'ma'a sin pb + cos a'a cos pb = coskL. C and D.8) where b = L-a.

We write these equations together as f{E) = cos kL (10.L and U. above the top of the potential wells.244.1017/CBO9781139171243. Notice also that the band structure persists even for positive values of E. correspond to a complicated function of E. the bands corresponding to the lowest-lying levels are narrowest: the levels associated with the electrons that are most tightly bound to the atoms are affected least by the presence of the other atoms. but they can still be of appreciable width even when E is so large that the electron nearly has enough energy to escape from the crystal into the surrounding free space. Because k is real.4. Plot of the function f(E) defined in (10.I/| < 1. that is. Thus those values of E for which \f(E)\ > 1 are not accessible. 2014 . an alternative possibility is that neighbouring energy bands overlap.12).4. Number of levels in a band 117 Gap Gap Figure 10. when the level structure is calculated in more than one dimension. |cosA. As we shall describe below. with forbidden energy gaps between the bands. The energy gaps decrease in width with increasing E.208 on Thu Feb 13 17:12:58 GMT 2014.012 Cambridge Books Online © Cambridge University Press. that is.221. and this is found to be the case in certain types of crystal.4 we plot the function f(E) against E for a typical set of values of a. the allowed values of E fall into bands determined by the condition |/(i£)| < 1. Number of levels in a band Except for the smallest microcrystals.12) and in figure 10. the properties of the crystal and the disposition of the allowed energy levels do not depend critically on the Downloaded from Cambridge Books Online by IP 150.doi. Notice that.org/10. according to figure 10. http://dx.

2014 . but they give very slightly different energies to corresponding levels.5). .6). (10. we find from (10. The wave function ip is required to take the same value at corresponding points on opposite faces of the crystal. for example -n/L <k< n/L. without losing any of the states. Thus we may. .012 Cambridge Books Online © Cambridge University Press. http://dx. However. . a standing-wave solution is often unsatisfactory because transport of energy or charge is far more conveniently expressed in terms of travelling waves. so generating another travelling wave moving in the negative-a. (10. ± 2 . (10. (10. Thus each band has just N allowed levels. or in the linear model ip = 0 at each end of the chain of atoms.3).3) we see that this requires that eikNL = 1.1017/CBO9781139171243. one must instead take combinations of Bloch waves to achieve a standing wave (a travelling wave of the type (10. Downloaded from Cambridge Books Online by IP 150. this has two disadvantages. Both types of boundary condition result in the same number of levels within each band.14) As k takes each of the allowed values (10. (10.3) moving in the positive-:*. confine k to any chosen interval of length 2n/L. In the linear model the periodic boundary condition is equivalent to bend- ing the chain of atoms round to form a closed loop. that is ip = 0 on the faces. direction). Choosing the coordi- nates such that the ends of the chain are x = 0 and x = JVX.13) Furthermore. n = 0. the boundary conditions can then not be satisfied by the simple Bloch wave functions (10. First. so that an electron is not reflected at the faces but leaves the crystal and simultaneously reenters it at the corresponding point on the opposite face. This is achieved with periodic boundary conditions.13) that fall within this interval. so that the possible values of k are discrete: k = 2nn/NL. ± 1 .3). From (10.221. The simplest choice of bound- ary condition is to require that the electrons be strictly confined within the crystal.244. we require ^(0) = ip(NL).10) and (10. direction is reflected at the crystal face.doi. For all but the smallest microcrystals the levels within a band are so closely spaced that they almost form a continuum and so this difference has no consequence.118 10 Band structure of crystals boundary conditions applied at its surfaces. Since the crystal surface has little effect on the bulk properties of the crys- tal. where N is the total number of atoms in the crystal.org/10. . Secondly.11) that both if) and the corresponding energy E are unchanged if k is either in- creased or decreased by an integral multiple of 2n/L. one may as well choose boundary conditions that more or less eliminate it. we pass through every one of the allowed levels in each band.208 on Thu Feb 13 17:12:58 GMT 2014.

the single level corresponds to N different states. which relate k and i£. According to figure 10.4.11) involve k only through cos kL the solutions for E are symmetric about the origin in k. are called dispersion relations. because the electron can be attached to any one of the N different atoms.10). It is not possible to obtain an explicit expression for this solution. The model of figure 10. When the atoms are very far apart each band consists of just a single level. As the atoms are brought together.208 on Thu Feb 13 17:12:58 GMT 2014. for E > 0.012 Cambridge Books Online © Cambridge University Press. The number of values for which E < 0. but. However.10) and (10. We discuss the general features of the solution.14) this corresponds to either k = 0 or k = ±n/L. A given value of k corresponds to many values of E. Prom these considerations. Because (10.10) and (10. This is a generalisation of the two-atom problem that we discussed in chapter 6.221. in figure 10.5.5 we show a one-dimensional potential that can be used as a model for a crystal in which the unit cell consists of two atoms. We consider now the solution to these equations. the number of levels in each energy band is equal to the number of unit cells in the crystal. In many cases the crystal lattice consists not of a regularly spaced array of single atoms. we deduce that the dispersion curves have the form drawn in figure 10. (10.org/10.6. but rather of an array of unit cells that each consist of more than one atom. the edges of the allowed energy bands occur when cos kL = ± 1 . Generally.doi. it is necessary to use either graphical or numerical methods. Band overlap Equations such as (10.11) always gives an infinite number of E values for a given k. the energy level of the solitary atom. corresponding to (10. Band overlap 119 Figure 10.3 represents a crystal in which the unit cell consists of a single atom. the AT-fold degenerate levels spread out into bands. Downloaded from Cambridge Books Online by IP 150. One-dimensional model for a crystal having two atoms per unit cell. The property that the number of levels in each band is equal to the number of atoms in the crystal is special to the simplest crystal structure. http://dx. and if the range of definition of k is chosen to be (10.11).1017/CBO9781139171243. 2014 . depends on the depth and width of the atomic potential wells. which expresses E as a function of A. This result is not unexpected..244.

1017/CBO9781139171243. The dispersion curve becomes a three-dimensional surface in four-dimensional (£7.120 10 Band structure of crystals k -n/L n/L Figure 10. Dispersion curves corresponding to figure 10. re- gardless of how high up in energy the bands are. so that if we draw sections of this surface corresponding to different directions for k the resulting curves need not be the same.7).208 on Thu Feb 13 17:12:58 GMT 2014. is a vector. In three dimensions the Bloch wave function (10.6. http://dx.doi. the bands overlap and there is no gap between them: it is possible to get from one to the other by continuously varying k.244. The properties of the crystal may be different in different directions.3) becomes tl>(r)=eik'ruk(r). 2014 .6. k) space.012 Cambridge Books Online © Cambridge University Press.org/10. but the positions of these gaps and their widths may vary with the direction chosen for k. If this happens. it may be that the lowest position of the top of a given gap between two bands is lower in energy than the highest position of the bottom of that gap (figure 10. in the real world of three dimensions the bands may overlap. and has important consequences.221. In the one-dimensional model. Downloaded from Cambridge Books Online by IP 150. However.4. As this direction is varied. This occurs in divalent metals and semimetals. (10. gaps always occur between the bands. with gaps between the bands.15) so that A. Each such section will have an appearance similar to figure 10.

Although most elements crystallise into forms containing one atom per unit cell. let the number of atoms per unit cell be m. We Downloaded from Cambridge Books Online by IP 150. the N levels of a band can accommodate at most 27V electrons.doi. Since the electron has two independent spin configurations (the ^-component of its spin has eigenvalues ± | ) . it has ZmN electrons when it is electrically neutral.208 on Thu Feb 13 17:12:58 GMT 2014. It is conventional to classify a solid as an insulator.org/10. the ratio of the conductivity of the best conductor to that of the best insula- tor is about 1028. At the absolute zero of temperature the electrons will be found in states of as low an energy as possible. each having charge Ze on its nucleus. http://dx. the levels are filled up. a semiconductor. starting from the lowest level of the lowest band. Simple consequences of band structure 121 Figure 10. partly according to how good a conductor it is.012 Cambridge Books Online © Cambridge University Press. a semimetal or a metal.1017/CBO9781139171243. Then.244. Band overlap. that is. We explained in chapter 5 that the Pauli exclusion principle allows at most one electron to be in each quantum-mechanical state. with two electrons in each. in particular on its electrical conductivity and its optical prop- erties. Suppose that the crystal consists of one type of atom only. The electrical conductivity takes a wider range of values for different ma- terials than does any other physical property: at room temperature. if the crystal contains N unit cells.7. In order to keep the discussion general. a few of interest to us here form crystals with two atoms per unit cell. 2014 .221. Simple consequences of band structure The band structure of a material has a profound effect on its physical properties.

A is small enough for there to be a significant electron concentration in the conduction band. distinguished by whether the number Zm of electrons per unit cell is even or odd. T = 0. Therefore the field does not cause a net electric current to flow. some of the electrons in the valence band may be excited into the conduction band.1017/CBO9781139171243.012 Cambridge Books Online © Cambridge University Press. Hence at the absolute zero of temperature. and therefore also the electric current. Insulators and semiconductors In these crystals Zm is even and no band overlap occurs. there are three basically different ways in which the bands can be populated. At temperature T > 0 some of the valence-band electrons are thermally excited into the conduction band. A is so large that the electron concentration in the conduction band. the quantum state of the crystal cannot change when such a field is applied.122 10 Band structure of crystals have said that the maximum number of electrons in any one band is 2N. http://dx. In the next chapter we show that the number of electrons that appear in the conduction band varies rapidly with the width A of the energy gap and with the temperature T. is called the conduction band. As all the valence-band states are already filled.doi. In the next chapter we explain how this enables the semiconductor to carry an electric current. and the next band above it. the electrons in the conduction band can fairly easily change their states by going to nearby unfilled levels in the conduction band. However.244. which can change their states by occupying the holes in the valence band. which is empty. If the crystal is placed in a strong electric field. Hence the semiconductor carries a current.8).208 on Thu Feb 13 17:12:58 GMT 2014. In chap- ter 12 we shall explain that the conductivity of an insulator or an intrinsic Downloaded from Cambridge Books Online by IP 150. and if it is even by whether or not there is band overlap. The topmost filled band is called the valence band. At T = 0 a number of bands are completely filled with electrons and the remainder are empty. are negligibly small even up to the melting temperature of the solid. In a semiconductor. Another contribution to this net drift comes from some of the valence-band electrons.221. this increases as T increases.org/10. with an electric field of ordinary magnitude the electrons do not acquire enough energy to jump the energy gap between the bands. If an electric field is applied. In an insulator. the quantum state of the crystal changes in such a way that there is a net drift of electrons. We consider these cases in turn. 2014 . Such a semiconductor is known as an intrinsic semiconductor. The electrons that are thermally excited into the conduction band leave unfilled states or holes in the levels they pre- viously occupied in the valence band (see figure 10.

244.) and of the alkali metals (sodium. These have an even number of electrons per atom and crystallise into a structure with one atom per unit cell. This means that even at T = 0 there are some electrons in the lowest levels of the conduction band.012 Cambridge Books Online © Cambridge University Press.221. but crystallise into a structure having two atoms per unit cell.1017/CBO9781139171243. leaving unfilled those states in the valence band that have higher energy. the material is then said to be a doped semiconductor. silver etc. Metals The atoms of the noble metals (copper. These crystals are therefore fairly good conductors. semiconductor may be increased by the presence in the crystal of a small concentration of certain impurity atoms. so that again each has an even number of electrons per unit cell. potassium etc. A similar situation occurs in the divalent metals. diamond.org/10.208 on Thu Feb 13 17:12:58 GMT 2014.7).doi. Since the total number of electrons per unit cell is even. The high concentration of conduction-band electrons ensures that metals are good electrical conductors. they would be insulators were it not for the fact that their conduction and valence bands overlap to a small extent (we recall figure 10. which Downloaded from Cambridge Books Online by IP 150. http://dx. Hence their valence band is completely filled and their conduction band half full. they are known as semimetals. 2014 .8. Excitation of electrons from the top of the valence band of a semiconductor into the bottom of the conduction band. Simple consequences of band structure 123 Conduction band Valence band Figure 10. For insulators and semiconductors. of which calcium and strontium are examples.) have an odd number of electrons and one atom per unit cell. antimony and bismuth each have an odd number of electrons per atom. the magnitude A of the energy gap be- tween the valence and conduction bands has an important influence also on the colour of the pure crystal. Semimetals and divalent metals Arsenic. Consider. for example.

012 Cambridge Books Online © Cambridge University Press.124 10 Band structure of crystals is one of the crystal forms of carbon.4 The tight-binding model for a crystal is as follows. These remarks apply to pure crystals.208 on Thu Feb 13 17:12:58 GMT 2014. so that an electron in the valence band cannot absorb such a pho- ton. it has energy Eo and wave function <f>. independent of position in the crystal. take the potential as Investigate the level structure.3 (i) As a model for a one-dimensional crystal.org/10. 10. 2014 .1 In the free-electron model of a metal the potential is taken to be constant. less than A. on the other hand. Downloaded from Cambridge Books Online by IP 150. Problems 10.doi. has A = 2.14) generalise to this three- dimensional case? How many allowed fc-vectors are there per unit volume of fe-space? If there are Z electrons per atom. analogous to those of figure 10. How does (10. so all visible light passes through the crystal.4 eV and be knocked into the conduction band. Sulphur.4 eV.2 A crystal cube has N atoms arrayed in a simple cubic lattice and its volume V is large. A photon in the visible frequency range has energy in the range 1. (A more realistic model is the nearly-free-electron model. (ii) Explain how to repeat the calculation for the potential which is a model for a crystal having two types of atom. There are not enough conduction-band electrons to give appreciable absorption. which starts from the free-electron model and then uses perturbation theory to introduce a small additional potential around each atomic site.244.221. the introduction of impurities can have an important effect.1017/CBO9781139171243. http://dx. and the crystal is yellow in appearance.6 eV. This has A = 5. using a Bloch wave function. so that its valence-band electrons can absorb photons of frequency u) such that TIUJ > 2. Thus the blue end of the spectrum is absorbed.6.5 eV. When the electron is bound to a solitary atom.) 10.7 to 3. Diamond is therefore colourless. What form do the Bloch waves take in the one-dimensional version of the model? Sketch the dispersion curves. what is the largest value of \k\ for the states occupied at zero temperature? 10. as we indicate in chapter 12.

Downloaded from Cambridge Books Online by IP 150. 2014 .doi. By considering the expec- tation value of the Hamiltonian if. or instead the electrons are strictly confined within the crystal.org/10.221. http://dx.5 Show that the number of states in a band is unaffected by whether periodic boundary conditions are applied. 10. In the crystal.012 Cambridge Books Online © Cambridge University Press. Simple consequences of band structure 125 and the binding is tight.10) and one can show that it approximately describes a stationary state. as given. the atoms are not very close together. where VQ(X) is the potential that would correspond to a solitary atom at the origin.244. (Notice that ip(x). show that £?« Eo+ 2cos kl / dx</>*(x)Vo(x)<l)(x-L). is not normalised.1017/CBO9781139171243.208 on Thu Feb 13 17:12:58 GMT 2014. The wave function clearly satisfies the Bloch condition (E.) Deduce that each atomic level Eo gives a single band of energy levels in the crystal.

2014 . From (11.26) The Bloch wave function (11.1017/CBO9781139171243. In terms of uk. (11.3) with where V is the crystal potential.. As we know from the discussion in chapter 4. 11 Electron motion in crystals Electron velocity In the previous chapter we showed that in a perfectly regular crystal the stationary-state electron wave functions are the Bloch waves *k(r.1) and (11. apart from questions of degeneracy the wave functions corresponding to different values of k are orthogonal.244. except that we have added the factor e~lEkt/h so as to write the time-dependent wave function. these equations read h(k)uk = Ekuk.org/10.013 Cambridge Books Online © Cambridge University Press.t)=uk(r)Sk-r-E>t'h\ (11.208 on Thu Feb 13 17:13:17 GMT 2014.2a) Here. and we may choose the normalisation = 6kk. but if the crystal is large enough these discrete values are so close together that for many purposes k may be treated as a continuous variable. (11.1) This expression is the same as (10. (11. (11. The values allowed for the components of the vector k are discrete. in which case it is equal to 1. the integration is over the volume of the crystal and dkk> is zero unless the vectors k and k' are the same.1) satisfies the Schrodinger equation if*fc = ih4>k.15).4) 126 Downloaded from Cambridge Books Online by IP 150. = <W.2a).221. http://dx.doi.

1) is not an eigenfunction of the electron mo- mentum operator p = —i/iV.9a) or v = (l/h)(dEk/dk). Sk may be chosen to be essentially in any direction and also arbitrarily small in magnitude.5) In order to evaluate this.8) give (p) = (m/h)(dEk/dk) (11.6) similarly gives zero contribution. dh(k)/dk = (h/m)(p + hk).h(k)uk>) = Ek(uk.4) gives h(k)uk> + 6k • (dh(k)/dk)uk = Ekuk> + 6k • (dEk/dk)uk. (11.221.26). The first term on the right-hand side of (11. 2014 . The other terms in (11.8) Thus (11. However.7) remains true if we cancel the factor 5k on each side.4) with k replaced by some other allowed vector k' = k + 6k.013 Cambridge Books Online © Cambridge University Press.6) We now pre-multiply this equation by u*k and apply the integration / d3r.6) give where again (11. (11.uk>). (11. If we approach the continuum limit. The first term on the left-hand side becomes (uk.1017/CBO9781139171243. we first write (11. (11.4). http://dx.7) and (11. Because the direction of 6k is arbitrary.5). (p + hk)uh). (11. (11. -ihv*k) = (I*. This vanishes because of (11. (11.26) has been used to simplify the right-hand side.208 on Thu Feb 13 17:13:17 GMT 2014.96) where v = (p)/m is the expectation value of the velocity of the electron.org/10. If we neglect quadratic and higher powers of Sk.244. Prom the definition of the operator h(k) in (11. where we have again used (11. Downloaded from Cambridge Books Online by IP 150. we may calculate the expectation value of the momentum: (P) = (#*.4).doi. Electron velocity 127 where The Bloch wave function (11.

Plots of Ek.6. because of collisions between the electron and the atoms of the crystal. 2014 . for the three lowest bands. In the one-dimensional model described in the previous chapter.221. We reproduce these in figure 11.96) shows that in quantum mechanics v is constant in both direction and magnitude.013 Cambridge Books Online © Cambridge University Press. the direction and magnitude of the electron velocity would continually be changing.1.208 on Thu Feb 13 17:13:17 GMT 2014. together with the graphs of dEk/dk and d2Ek/dk2. corresponding to no variation with time of the probability density at each point in the crystal. the graphs of Ek plotted against k are as drawn in figure 10.1) describe stationary states. This result is valid only for a perfectly regular crystal lattice. If we wish to consider the motion of an electron in the presence Downloaded from Cambridge Books Online by IP 150. The result (11. In classical mechanics. This is because the width of the bands increases with increasing band number.1017/CBO9781139171243.6.244. Notice that the maximum value of v = h~1dEk/dk increases as we go up to higher bands.14)) remains fixed.128 11 Electron motion in crystals Ek against k dEjdk against k d2Ek/dk2against k Figure 11. while the range of definition of k (given in (10. http://dx.doi. dEk/dk and d2Ek/dk2 against k for the three lowest bands of figure 10.1. Motion in an external electric field The Bloch wave functions (11.org/10.

and /d 3 fc means integration over each of the components of the vector k.10).org/10.221. How- ever. that a wave packet moves like a classical particle. 2014 . chosen a wave packet whose spread in r was less than the dimension of a unit cell of the crystal.244. (11.4 we proved Ehrenfest's theorem (4. Because each Bloch wave (11. the expectation value of the electron velocity would have varied as the wave packet moved across each unit cell.1). This means that each component of the wave packet corresponds to ap- proximately the same value v = dEk/dk of the expectation value of the electron velocity. If we had. (11. dEk/dt = k • dEk/dk = Tik • v. We explained in chapter 4 that a narrow spread in k corresponds to a wide spread in r.1) satisfies the time-dependent Schrodinger equation (11. for a wave packet that spans many unit cells.3). (11. However.11a) However. This is a superposition of stationary-state wave functions: . Hence in an electric field £ the vector k about which N(k) is peaked changes with time in such a way that the rate of change of the electron energy is equal to the rate of work done by the field: dEk/dt = -eSv.013 Cambridge Books Online © Cambridge University Press.1017/CBO9781139171243. and so k = -(e/h)£.208 on Thu Feb 13 17:13:17 GMT 2014. In prob- lem 4. instead.14). it moves in this direction until it reaches the limit of the range of definition Downloaded from Cambridge Books Online by IP 150. We shall suppose that N(k) is sharply peaked at some value of k and that it falls rapidly to zero outside the immediate vicinity of this value.10) is not a stationary state. At whatever point of the dispersion curve k starts. That is the situation in the absence of an external electric field. the variation of Uk{r) is averaged out and the time variation of v is negligible.t). When £ is constant. This variation may be traced back to the variation over the unit cell of the factor uk(r) in the Bloch wave function (11. Motion in an external electric field 129 of an external electric field. http://dx.doi.10) where the function N(k) determines the shape and spread of the wave packet. it is more appropriate to use a localised wave packet.116) This result holds provided that the necessary empty electron states exist so as to allow k to change. (11. k increases uniformly with time in the direction opposite to 5. so also will any superposition (11.

2.2.208 on Thu Feb 13 17:13:17 GMT 2014. all the allowed representative points along the correspond- ing dispersion curve are occupied and the exclusion principle therefore forbids the distribution of these points to change when a field is applied. k) curve like a chain of beads. The representative points for the electrons. this is the point G in figure 11.1. of the vector k.1 we see that this means that the electron velocity v = h~1dEj£/dk oscillates back and forth as the cycle is repeated. as in figure 11. In the one-dimensional model. For each electron with velocity v there is another with velocity — v. Prom figure 11. it reappears at Gf. Because the states with k = n/L and k = — n/L coincide. corresponding to k = n/L.2 and then it again moves across towards G. and so the total current due to all the electrons in a full band vanishes. if. When the point reaches G.244. When there is band overlap. the discussion must be modified accordingly. This means that in real space the electron oscillates back and forth also.130 11 Electron motion in crystals -n/L n/L Figure 11. the representative point now reappears at the point G' in figure 11. even a weak field may induce the electron to make a transition from one band to another.013 Cambridge Books Online © Cambridge University Press. We have assumed that no band overlap occurs. as is true in the one- dimensional model. will be strung out along the (v. so that the total current Downloaded from Cambridge Books Online by IP 150. Let the dispersion curve for the conduction band have the shape drawn in figure 11. When there is no electric field the points are distributed symmetrically about k = 0.doi. Motion of the representative point in the presence of an electric field £.221.org/10.3(a). http://dx.2. which are initially in the lower half of the band. instead. But suppose that the conduction band is half full and is separated from the band above it by a gap. it is the other way up like the middle band in figure 11. 2014 . Electric current In a full band. and then the motion is more complicated.1017/CBO9781139171243.

the crystal has no electrical resistance. The vector sum of the velocities of the collection of electrons is now non-zero. and so on.1017/CBO9781139171243.3(c). In either case. This rather surprising result. is zero. the representative points begin to move to the right. does not apply if the conduction band overlaps the next higher band. In practice this is Downloaded from Cambridge Books Online by IP 150.3. the steady state described by Ohm's law is not found.org/10.013 Cambridge Books Online © Cambridge University Press.doi. figure 11.3(6). (6). This is because the discussion has assumed that the crystal lattice is perfectly regular.244. and so a net electric current flows in the direction of the electric field (remember that the electrons have negative charge). An electric current nowflowsin the reverse direction. until in figure 11. The velocity of conduction band electrons (a) when there is no electric field.221. and if all the higher bands similarly overlap each other. http://dx. Electric current 131 \v .3(rf) the sum of the velocities has changed sign.208 on Thu Feb 13 17:13:17 GMT 2014. Subsequently. Then the electric current increases steadily. (c) and (d) with an electric field 8. that the direction of the electric current oscillates back and forth with time. When the electrons reach the right-most edge of the range of definition of k they reappear at the left. it changes direction again. as shown in figure 11. 2014 . at successive times. remaining in the direction of the applied field.r\ I (a) (b) i 1 1 r\ 1 1 j 1 k 1 % 1 1 1 Figure 11. When a field is applied towards the left.

as do photons. Imperfections and impurities apart. these travel through the crystal with speeds of the order of the speed of sound. this gives m*dv/dt = -eS (11. the quantum of energy is called a phonon. their vibrations are correlated. the main source of resistance in a solid is the thermal vibration of the atoms about their lattice sites. When an electric field is applied to the crys- tal. We now show that.org/10. but oscillate with zero- point energy (see the discussion of the harmonic oscillator in chapter 3). (1L12) If we now use (11. A steady state is rapidly reached: the distribution of representative points becomes static. so is the energy of a lattice wave. the representative points illustrated in figure 11. in nearly all crystals.132 11 Electron motion in crystals never exactly true.013 Cambridge Books Online © Cambridge University Press. In addition. the electron may sometimes move in the direction opposite to the force. As the electrons move through the crystal.116). Phonons behave in many respects like particles. Even at the absolute zero of temperature the nuclei are not completely anchored at their lattice sites.96). the exact periodicity of the lattice is disturbed. the collisions become more frequent and more violent. http://dx.208 on Thu Feb 13 17:13:17 GMT 2014. As the temperature T increases.doi.3(6). Just as the energy of an electromagnetic wave is quantised. Prom (11. This effective mass m* varies with the energy Ek. but this motion is opposed by the phonon scatter- ing. The scattering with phonons tends to prevent the electrons from increasing their energy and so from rising into the uppermost parts of the conduction band. as a result of the presence of the crystal lattice. by the presence of imperfections such as geometrical dislocations and concentrations of impurity atoms. Because the displaced atoms exert forces on one another. and in a part of each band it is negative. consider the one-dimensional case. they repeatedly collide with phonons.1017/CBO9781139171243.244. dt dt [h dk ) ~h Ok* k.221. 2014 .3(a) begin to move according to (11. For simplicity. This correlation allows an analysis in terms of lattice waves. A steady electric current then flows. displaced asymmetrically from its original position as in figure 11.13) Downloaded from Cambridge Books Online by IP 150. the electron moves as if it had a mass m* different from its real mass m. Effective mass and holes We have seen that when a force is applied to an electron in a perfectly regular crystal.11).

m* is negative near the top of the band. We have plotted d2Ek/dk2 for the one-dimensional model in figure 11.221. This occurs in a semiconductor. The holes correspond to an absence of these electrons and so they have positive charge and positive mass. If the levels were filled with electrons.1017/CBO9781139171243.15a) where the summation extends over all the states i except for the unfilled state j .15a) may also be written as J = evj.org/10. For each of the bands. these would have negative charge and negative effective mass.244. Effective mass and holes 133 with 1 1 d2Ek (11. called holes. with a few levels near the top of the band unfilled.14) m* h2 dk2 The relation (11. because for each state corresponding to velocity v there is another corre- sponding to velocity — v. We illustrate this by considering the case in which a single electron is missing from an otherwise full band.13) is just Newton's law for the motion of the electron. Hence 0= - where the summation extends over the whole band. and so (11. the effective mass m* is proportional to (d2Ek/dk2)~1. when the band is full there is no net electric current.156) This is just the current that would be contributed by a solitary positively charged carrier with velocity Vj equal to that which an electron occupying Downloaded from Cambridge Books Online by IP 150. The dynamics of the charge carriers in the nearly filled band may then be described either in terms of the large number of electrons in the band or in terms of the small number of holes. However. http://dx. 2014 . but with an effective mass m*.1.208 on Thu Feb 13 17:13:17 GMT 2014. It is convenient to regard the unfilled states as fictitious particles. This means that the acceleration of an electron near the top of a band is opposite to the applied force. Suppose that a particular band is nearly full.013 Cambridge Books Online © Cambridge University Press. but at non-zero temperature a few of the electrons from the top of the valence band are thermally excited into the conduction band.doi. When an electric field is applied. the total electric current contributed by the band is u (11. (11. where at zero temperature the valence band is filled and the conduction band is empty.

the rate of change of momentum of all the electrons in the band is dP x—s *dv{ /i 1 1^? \ (1L16a) X>'dT However. 2014 . the distribution tails off smoothly beyond energy (Q. If the level E{ has degeneracy Qi (with Downloaded from Cambridge Books Online by IP 150. there are no electrons in the conduction band and no holes in the valence band. Because the effects of the exclusion principle are important.244.16a) may also be written as dP/dt = (-m^dvj/dt (11. At T > 0 some electrons are excited into the conduction band and leave behind holes in the valence band. In a semiconductor or insulator at T = 0. all states in the conduction band with energies up to some maximum value (o are filled. In a solid in thermal equilibrium. To study the effects of temperature it is rarely necessary to consider the details of the collision processes that cause electron excitation. the classical Maxwell-Boltzmann distribution (9. the appropriate distribution is the Fermi-Dirac distribution.24) is not applicable.1017/CBO9781139171243. and assigning to the unfilled states j fictitious particles of effective mass (—m*) and velocity Vj. so that then the rate of change of its total momentum vanishes: Hence (11. the exclusion principle forbids its overall quantum state from changing.As a result of thermal excitation. http://dx. Although the electrons continually change levels as a result of collisions. instead.doi.221. At temperatures T > 0 the distribution of occupied states does not end abruptly at £o.013 Cambridge Books Online © Cambridge University Press. the electrons are distributed in the available en- ergy levels in accordance with the predictions of statistical mechanics.134 11 Electron motion in crystals the level j would have had. Similarly. when the band is filled.208 on Thu Feb 13 17:13:17 GMT 2014.166) So the rate of change of momentum also may be calculated by ignoring the electrons. statistical mechanics gives their expected average distribution among the levels independently of the details of the collisions. Thermal excitation In a metal or semimetal at zero temperature. electrons are found with energies greater than (o.org/10. and the rest are empty.

p(E)dE i. but it varies with the tem- perature T. in this figure we have supposed that all the degeneracies gi have the same value g. of electrons in each level is then very small. The Fermi-Dirac distribution (11.17) is derived in any textbook on sta- tistical mechanics. It has the property that at zero temperature 9i Ei < Co. SO that it varies rather slowly with T.20) where C = e^^kBT. It is independent of energy. the parameter ( is called the Fermi energy or. 2014 . more correctly. The ex- pected average number of electrons in this range is then. but for a large system they are usually so close together that it is an excellent approximation to treat Ei as a continuous variable. So for energies for above the Fermi energy we again have the classical Maxwell-Boltzmann distribution (9.24). In practice. the exponential in the denominator of (11. so that the exclusion principle has little effect.244.17). the average number of electrons expected in that level is Ui = (^ Here.17) dominates and E / k T (11. Let p(E) &E be the number of states in the energy range E to E + dE. where £0 is the value of £ at T = 0. For large values of E{.doi.21) Downloaded from Cambridge Books Online by IP 150. Thermal excitation 135 a minimum value gi = 2 when spin is taken into account).org/10.1017/CBO9781139171243.208 on Thu Feb 13 17:13:17 GMT 2014. http://dx.221. (11. In any bound quantum system of finite size the allowed values of Ei are discrete.013 Cambridge Books Online © Cambridge University Press. This is because the average number n. such that (E{ — () » k^T.4. as the continuum limit of (11. For temperatures T > 0 the distribu- tion has the shape drawn in figure 11. the variation between T = 0 and room temperature is almost negligible. It is determined in terms of the total number N of electrons by the implicit equation This usually has the effect that £ varies as log &BT. the chemical potential.

Unless the temperature T is very high.18): p(E) dE N = J (11. for the case in which all the levels have the same degeneracy g.1017/CBO9781139171243.org/10. p(E) is a function that depends on the nature of the crystal. there are just enough electrons to fill all the levels up to the top of the valence band: N -L dEp(E). /•Co JV= / dEp(E). The Fermi-Dirac distribution at temperatures T > 0. Downloaded from Cambridge Books Online by IP 150.doi.24) to be consistent. (11. whereas in the gaps p(E) = 0. at T = 0 all the levels up to the Fermi energy ( 0 are filled. while for an insulator or a semiconductor it lies somewhere in the energy gap between the valence band and the conduction band. In an insulator or an intrinsic semiconductor. and all those above it are empty. However.is the analogue of (11. http://dx.24) J — OO That is.23) (11.136 11 Electron motion in crystals g Figure 11.221.208 on Thu Feb 13 17:13:17 GMT 2014. where Ev is the energy at the top of the valence band.4 (with g set equal to 1). 2014 . In the energy bands.21) is the product of p(E) and a factor that has the shape drawn in figure 11.23) and (11. We demonstrate this last result.22) e(E-O/kBT The average number density n(E) given by (11.22) evaluated at T = 0.4.244. the Fermi level £ for a metal or semimetal lies in the conduction band.013 Cambridge Books Online © Cambridge University Press. from (11. The equation that determines C. For (11.

21) gives the total number of electron-hole pairs as On the other hand. The number of pairs so excited can be calculated from the Fermi-Dirac distribution (11. Pair excitation in intrinsic semiconductors 137 p(E) must vanish for values of E between Ev and (o? so that ( 0 must be in the energy gap. most of these excited electrons will actually be near the bottom of the conduction band. The first way is to bombard the crystal with photons whose frequency u is such that Tiuj > A.208 on Thu Feb 13 17:13:17 GMT 2014. The rate of pair production increases with the intensity of the light. so producing an electron-hole pair of current carriers. when T > 0 pair excitation occurs because of collisions between electrons and phonons. This is the principle of the photoelectric cell. Pair excitation in intrinsic semiconductors There are two obvious ways in which an electron in an intrinsic semicon- ductor can be knocked from the valence band into the conduction band.doi. but remains in the energy gap until T becomes much greater than room temperature. from what we have said earlier in this chapter. the Fermi-Dirac distribution (11. If again Ev is the energy at the top of the valence band.21). http://dx. be Ne. Thus Downloaded from Cambridge Books Online by IP 150. Let the number of electrons excited into the conduction band. Secondly.244. both the excited electrons and the holes will nearly all have positive effective mass ra*. where the total number N of electrons is given by (11. and A is the energy gap. and most of the holes that they leave behind will be near the top of the valence band. and so when a steady electric field is applied the electric current also increases with the light intensity.1017/CBO9781139171243. or into even higher bands.ATe). where A is the magnitude of the energy gap between the valence band and the conduction band. the total number of electrons in the valence band and in lower bands is This must be equal to (N . Hence.23).221. Be- cause the exponential in the denominator of the Fermi-Dirac distribution grows rapidly as E increases. 2014 .org/10.013 Cambridge Books Online © Cambridge University Press. As T is increased ( changes slowly.

and most of the con- tribution to the integral on the right-hand side comes from the top end.29) So in this special case £ is independent of T and lies exactly in the middle of the energy gap.org/10.27) comes from near the bottom end of the integration.27) This is the equation that determines For purposes of illustration.5). Downloaded from Cambridge Books Online by IP 150. -L dE- p(E) (11. 2014 .doi. http://dx.208 on Thu Feb 13 17:13:17 GMT 2014.27) is satisfied by = Ev (11. consider the special case where the level den- sity p(E) has the same shape near the bottom of the conduction band as at the top of the valence band (figure 11.013 Cambridge Books Online © Cambridge University Press.221.1017/CBO9781139171243. so we do not need to know about p(E) elsewhere. this is true in silicon. Model in which the state-density function p(E) has the same shape at the top of the valence band as at the bottom of the conduction band. One may readily check that (11. To a reasonable approximation.28) Most of the contribution to the integral on the left-hand side of (11. germanium and many other semiconductors. Then near the energy gap p(E)=p(Ev (11.244.138 11 Electron motion in crystals P(E) Figure 11.5.

2 An atomic monolayer is formed by evaporating a metal onto a flat substrate. (ii) In the case of a simple square lattice of area A. (i) In the case of a one-dimensional lattice.25) and evaluating it for the case k&T < A.29) into (11. and is infinite otherwise.244. the exponen- tial decreases rapidly as T decreases. show that p(E) = 11. so that the exponential in the denominator dominates. suppose that the potential V(r) vanishes for 0 < z < a.1 Consider the free-electron model for a crystal (problem 10. so that the exponential is multiplied by a very large number.221. and at room temperature the conductivity is fairly good. where a is the atomic diameter. However. Electronic devices that contain semiconductors frequently incorporate feedback loops in their circuits in order to compensate for changes of temperature. we see that approximately Ne oc e" A / 2fcBT .31) At room temperature. Problems 11. However.208 on Thu Feb 13 17:13:17 GMT 2014. and at around 0°C the conductivity of silicon or germanium is only a seventh of its value at room tempera- ture (remember that T = 0 corresponds to — 273° C). Pair excitation in intrinsic semiconductors 139 Inserting (11. Hence the electrons Downloaded from Cambridge Books Online by IP 150. http://dx.org/10. For silicon or germanium.1017/CBO9781139171243.30) Since most of the contribution to this integral arises from values of E near the lower end of the integration. we have -L Ev+A (11. show that p(E) = {NL/Kh)(m/2E)ll2. defining the constant potential to be zero inside the crystal. k^T « ^j e V. (11.013 Cambridge Books Online © Cambridge University Press. 2014 .1). As a model for the behaviour of the electrons in the metal film.doi. show that p(E) = (Am/nh2). so that the exponential is very small indeed at room temperature. the total number N of electrons is very large. (iii) In the case of a simple cubic lattice of volume V. the en- ergy gap A is about 1 eV.

it will collect all the electrons that are emitted.) The conduction band is a continuum of states of energy E > 0.221.1017/CBO9781139171243.244. (That is. Downloaded from Cambridge Books Online by IP 150. the valence band has zero width. If a collecting electrode is placed near the metal surface and is maintained at a sufficiently positive potential relative to the metal. How does the density of states p(E) vary with El 11. the electrons behave as a set of independent particles obeying the Fermi-Dirac distribution. with density given by p(E) = with A a constant. Show that the number of electrons in the conduction band at temperature T is where.208 on Thu Feb 13 17:13:17 GMT 2014. suppose that there are N electron states all having the same energy —A. if T is so small that e c//eBT < 1 and that n < TV.4 As a model of a semiconductor.140 11 Electron motion in crystals travel freely in the film.doi.org/10. 2014 .3 In the free-electron model of a metal. and that these are all filled at zero temperature.013 Cambridge Books Online © Cambridge University Press. this is called thermionic emission. An electron having momentum component normal to the metal surface greater than a critical value (2mVb)1/2 can escape from the surface. In practice Vb ^> ( + k&T. Show that the current per unit area emkBT c(C 11. http://dx.

244. but with one of the wells having a different depth. These have the property that when the required continuity conditions on the wave function are satisfied at the two edges of one of the potential wells. In order to study the effect of doping.25)). they are automatically sat- isfied also at the edges of all the other wells. Observable effects occur with impurity concentrations as low as a few parts in 108.221. Ui say (figure 12. Hence the total wave function to the left of the impurity atom is eikxuk{x) +Re~ikxu-k{x). The semiconductor is said to be doped with impurity atoms. we might guess that in the case of the potential of figure 12. We recall that for the perfectly regular crystal. (12.3). the stationary state solu- tions are Bloch waves of the form (10. are drastically af- fected by the addition of small traces of impurity atoms. If such a wave is incident on the impurity. 2014 . represents a wave travelling from left to right. The Bloch wave elkxuk(x). unlike metals or semimetals. As a one-dimensional model of this situation. and increasing the impurity concentration to one part in 105 can increase the conductivity by as much as a factor of 103 at room temperature and 1012 at liquid-helium temperatures. with k real. in which the crystal lattice is perfectly regular.1017/CBO9781139171243. but that it has a different form in the neighbourhood of the impurity. However. and the remaining part is transmitted across the impurity (com- pare (3. part of it is re- flected. For this reason.1).014 Cambridge Books Online © Cambridge University Press.24) and (3. the electrical prop- erties of semiconductors. we take the same infinite chain of square wells as in chapter 10. except that just one of the atoms has been replaced by an atom of a different type.208 on Thu Feb 13 17:13:29 GMT 2014.doi.1. 12 Transistors Impurities We have so far described the properties of intrinsic semiconductors. we first consider a crystal consisting of a periodic array of one type of atom.1) 141 Downloaded from Cambridge Books Online by IP 150. the stationary-state wave function is again composed of Bloch waves in the parts of the crystal where the lattice is perfectly regular.org/10. http://dx.

at the edges of the wells other than the impurity.244. and to the right of the impurity atom it is Teikxuk(x).10). determine E in terms of k. B and C are determined by the four conditions that tj) and ij)1 be continuous at both edges of the impurity well. 2014 . Because of the exponential factors.221.o) where The boundary conditions at the edges of the wells corresponding to the host atoms.4) and to the right of the impurity it is (12.1. except near to the impurity. in which the electron is more or less localised in the vicinity of the impurity atom.208 on Thu Feb 13 17:13:29 GMT 2014. Take the origin x — 0 to be in the middle of the impurity well.doi. In this solution. To the left of the impurity the wave function is eKXvi(x) K > 0. the wave function still satisfies the Bloch condition (E.1017/CBO9781139171243.014 Cambridge Books Online © Cambridge University Press.142 12 Transistors *— L u 1 Figure 12.org/10. The relationship is exactly as in chapter 10: there is the same band structure as if the impurity were not there. T. the solution to the time-independent Schrodinger equation is JLJ e "T u e yiz. so the electron is trapped in the neighbourhood of the impurity. but k is pure imaginary. One-dimensional model for a crystal with a single impurity atom. that is. (12. http://dx.5) where the functions vi(x) and V2(x) are both periodic. (12.2) Here. There is also a different type of solution. the wave function goes to zero as x -» ±oo. R and T are constants. Inside the potential well that represents the impurity atom. The constants i?. Downloaded from Cambridge Books Online by IP 150.

for example (ip^ip) = 1. if their concentration becomes greater than about one part in 103. in the same way as the levels of the host atoms spread into bands when a large number of host atoms are brought together to form a crystal.244. http://dx. there are discrete trapped-electron levels resulting from the impu- rity. we again ensure that when we satisfy the required continuity conditions at the edges of any single well on either side of the impurity. Imposing a normalisation condition on the overall wave function. For these v\{x) = +V2(—x). we expect a discrete set of solutions. One class has positive parity and the other negative parity.and p-type semiconductors 143 By making vi(x) and v2(x) periodic. Inside the impurity well. A say. they are automatically satisfied at the edges of the well corresponding to all the host atoms. the functions v\ and V2 are determined in a way precisely similar to the determination of Uk(x) in chapter 10. Since there are only two unknowns. Thus. n. provided that their concentration remains so low that they are far enough apart for there to be little overlap of the trapped- electron wave functions. but phosphorus has a valence of Downloaded from Cambridge Books Online by IP 150.(—x). This result continues to apply when further impurity atoms are sub- stituted for host atoms.doi. the positive-parity solution is B(ehx + e''nx). so that the pair of functions v\ and v2 now contains one multiplicative constant that remains to be fixed. (12. Silicon has a valence of 4. the solutions fall into two classes. For definiteness. 2014 .6) where 7 is defined in (12. For a given K. in addition to the usual band structure associated with the host atoms. n.1017/CBO9781139171243.3). There are two more continuity conditions to be satisfied: that \j) and ip' are continuous at one edge of the impurity well (the cor- responding conditions at the other edge are then automatically satisfied. because of the even-parity condition ip(x) = ifj(-x)). the impurity levels begin to spread out into bands. However.and p-type semiconductors Suppose that a crystal of silicon is doped with a few parts per million of phosphorus.014 Cambridge Books Online © Cambridge University Press. A and E (which determines 7 and «).221.208 on Thu Feb 13 17:13:29 GMT 2014. determines the constant B in terms of A. Since the complete potential is symmetric about x = 0. Each impurity level has degeneracy equal to twice the number of impurity atoms (the factor of 2 corresponding to the two possible electron spin states). ip(x) = ±if. Each contains a multiplicative constant that remains to be fixed and that determines its normalisation. consider the positive-parity solutions (the negative-parity ones may be discussed similarly).org/10.

2014 . It turns out that the lowest impurity level lies just below the bottom of the conduction band (figure 12.208 on Thu Feb 13 17:13:29 GMT 2014. So now the crystal appears red.221.2(6)). the crystal is transparent. such as boron atoms. it is said to be doped with donor atoms. They leave behind holes in the valence band.doi.014 Cambridge Books Online © Cambridge University Press. A material doped in this way is called an n-type semiconductor. in which it is trapped near an impurity atom.244. than in the impurity level. If instead titanium is added. in pure aluminium oxide the energy gap between the valence band and the conduction band is so large that visible light is not absorbed. The effect of doping silicon with phosphorus is therefore to provide an excess of electrons in the conduction band. The impurity atoms each bring with them one electron less than is needed to saturate the nearest- neighbour covalent bonds at the sites that they occupy. the position Downloaded from Cambridge Books Online by IP 150. At temperatures T > 0 electrons are thermally excited from the valence band into the degenerate impurity level. being trapped near the impurity atoms and so providing the missing bonds. in such a position that photons corresponding to green and yellow light are absorbed. say | % . the silicon crystal may be doped with atoms that have a valence of 3. Impurities and crystal colour The presence of iitipurity atoms can have an important effect on the colour of a crystal. For example. it is said to be doped with acceptor atoms. The extra electron either occupies one of the degenerate discrete impurity levels.1017/CBO9781139171243. In terms of the Fermi-Dirac distribution. The material is known as a p-type semiconductor. So at each lattice site where a phosphorus atom replaces a silicon atom there are five valence electrons. where there are very many levels close together. whereas only four are needed to form the covalent bonds that anchor the atom to its nearest-neighbour host atoms (figure 12.2(a)). http://dx. this is because the density of states p(E) is much greater in the conduction band. it is ruby. These are situated in the energy gap. Alternatively. when T > 0 the probability is high that the electron is thermally excited into the conduction band. However. So at zero temperature the fifth valence electron is in this level.org/10.144 12 Transistors 5. there are two relatively narrow bands of impurity levels. If a small quantity of chromium is added. The lowest impurity level is then found to be just above the top of the valence band. that is an excess of negative carriers which can contribute to an electric current. or it is found in the conduction band. because it is the lowest level available to it. These are mobile and are positive carriers which can contribute to an electric current. in fact.

http://dx. Suppose that each piece has a face that is accurately plane. (o) A silicon crystal doped with the donor impurity phosphorus. Semiconductor junction Consider two pieces of a given semiconductor. This forms a pn-junction.2. of the impurity levels is such that red light is absorbed.org/10. Semiconductor junction 145 Lattice site carries charge Si (a) Fifth valence electron Conduction band Impurity level (b) Valence band Figure 12. rather than Downloaded from Cambridge Books Online by IP 150.1017/CBO9781139171243.014 Cambridge Books Online © Cambridge University Press. it is sapphire. The crystal now appears blue. and imagine bringing the two pieces into contact at their plane faces.208 on Thu Feb 13 17:13:29 GMT 2014. one doped with donor atoms and the other with acceptor atoms.244.221.doi. with (b) the impurity level just below the conduction band. (In practice. 2014 .

compared with the p-type material. The electric field — V</> drives the electrons towards the n-type material and the holes towards the p-type material. we imagine that the transition is abrupt. the junction is manufac- tured from a single piece of host crystal by varying the doping in different parts of it as the crystal is grown. electrons and holes recombine in the junction region. the n-type material has an excess of electrons in the conduction band.146 12 Transistors pressing together two separate pieces of crystal. which is the basis of the solar cell. compared with the n-type material. the device operates as a semiconductor laser. are polished flat and made parallel to each other. This produces a transition region be- tween the p-part and n-part that is typically about 1 fim in width.1017/CBO9781139171243. so that when dynamical equilibrium is reached the band structure becomes as drawn in figure 12. This is the light-emitting diode. perpendicular to the junction plane. so producing photons.244.3(6). 2014 . After the contact is made. In our discussion. In an exactly similar way. See figure 12. The shape of this electrostatic potential 4> is drawn in figure 12. it is energetically favourable for some of the excess electrons in the conduction band of the n-type material to cross to the p-type material and annihilate some of the holes there.3(a).org/10. and this eventually stops the flow of further electrons across the junction. so that the absorbed photons create electron-hole pairs. So if the two sides of the crystal are joined to an external circuit. If two faces of the crystal. Similarly. the effect of the heat is to drive a current through the crystal from the n-type side to the p-type side. This is the photovoltaic effect. and round the circuit. It causes an additional energy — e<j> to be superimposed on the level struc- ture.221. http://dx. Its operation may be reversed: if. a current is generated if light is shone on the junction. current is driven through the junction by an external battery. and a net positive charge in the n-type material. so that there is an appreciable electric field — V</>. Thus an electrostatic potential is set up. As they do so. so that extra electron-hole pairs are created there by thermal excitation. Typically. The diode We now explain how the pn-junction acts as a diode rectifier: when a bat- tery is connected such that the p-side is at a higher electrostatic potential Downloaded from Cambridge Books Online by IP 150.014 Cambridge Books Online © Cambridge University Press. a net negative charge is built up in the p-type material. Suppose that heat is applied to the junction. the region where the elec- trostatic potential <f> varies appreciably.3(c).doi. instead. is of width 10 fim.208 on Thu Feb 13 17:13:29 GMT 2014.) At the instant when the two pieces are brought together the p-type mate- rial has an excess of holes in the valence band.

but only a limited current flows when the polarity of the battery is reversed. current flows readily. The difference between (a) and (6) is caused by the electrostatic potential <j> drawn in (c). Its position is shifted when the crystal is doped.244. than the n-side. http://dx. The diode 147 Conduction band (a) Valence o o o o band ° °o°o P-type n-type p \ \ (b) \ \ <t>0 (c) p-type n-type Figure 12.1017/CBO9781139171243.3. p Downloaded from Cambridge Books Online by IP 150.014 Cambridge Books Online © Cambridge University Press.org/10.doi.221. 2014 . Suppose first that no external electrostatic potential difference is applied. In chapter 11 we explained that in an intrinsic semiconductor the Fermi energy ( lies somewhere in the energy gap. The pn-junction: (a) the band structure at the instant the two pieces are brought together and (b) in dynamical equilibrium.208 on Thu Feb 13 17:13:29 GMT 2014. and the Fermi energies of each of the two pieces.

148 12 Transistors

and n, are further shifted when the pieces are pressed together to form the
junction. However, statistical mechanics tells us that any system of elec-
trons in dynamical equilibrium has a unique Fermi energy. So dynamical
equilibrium for the pn-j unction is reached when the Fermi energies ( of
the p-part and n-part coincide. We do not need to know the precise final
value of the common Fermi energy, but in practice it lies somewhere in
the overlap between the energy gaps on the two sides of the junction, as
drawn in figure 12.3(6).
According to the Fermi-Dirac distribution (11.21), the number of electrons
on the n-side that are above the bottom E% of the conduction band on the
p-side is

where pn(E) is the density of states on the n-side. Some of these electrons
will be moving in the right direction to cross over to the p-side, and some
of these will actually reach the p-side conduction band before collisions
reduce their energy to below E%. Hence there is a diffusion current of
electrons from the n-side to the p-side (that is, an electric current from
the p-side to the n-side):
/•OO

Jo oc / dEfr(E)e-(E-Mk»T. (12.8)
JEI

Similarly, some of the electrons in the conduction band on the p-side dif-
fuse into the n-side. In order that there be dynamical equilibrium, the
corresponding current must be equal and opposite to Jo- Exactly similar
considerations apply to the hole carriers.
Suppose now that an external electrostatic potential difference V is ap-
plied across the junction, by connecting the two sides of the crystal to a
battery. We assume that all this potential difference appears across the
junction region; if this is not a good approximation, the Ohmic potential
drop IR in the rest of the crystal may be allowed for once the current I
is known. With the extra driving potential V, the crystal is no longer in
dynamical equilibrium; there is no longer a single Fermi energy through-
out the crystal. However, on each side far from the junction there is a
situation close to equilibrium, not much affected by the additional current
flow caused by V. This is provided that V is not too large, so that the
drift velocity superimposed on the electron motion, corresponding to the
additional current, is small. Then outside the junction region each piece

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:13:29 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.014
Cambridge Books Online © Cambridge University Press, 2014

The diode 149

EP+eV-

EP+eV
n (a)

I I
(b)

p n 1-
V

Figure 12.4. (a) The band structure of figure 12.3(6) when there is an additional
potential difference V whose sign is indicated in (b).

of crystal may, to a good approximation, be described by a Fermi-Dirac
distribution, with the Fermi energy in the same place relative to the bands
as it was before V was applied. (See figure 12.4, and compare it with fig-
ure 12.3(6) — we have chosen to define the origin E = 0 of the energy
scale such that its position relative to the n-type bands has not changed,
and have drawn the figure for the case where the sign of V is such that
the p-type bands move upwards in energy.) In the narrow transition re-
gion at the junction, the Fermi level is not defined: the additional electric
field corresponding to V is not small, and the situation there is far from
dynamical equilibrium.
Because the p-type side has the same Fermi-Dirac distribution as before
(in spite of the shift of origin of £7), the number of electrons in the conduc-
tion band on that side is the same as before. Thus the diffusion current
arising from electrons passing from the p-side to the n-side is still Jo given
by (12.8). However, the diffusion current in the other direction is changed.
Now in the n-type region only those electrons above the level E =

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:13:29 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.014
Cambridge Books Online © Cambridge University Press, 2014

150 12 Transistors

I

-*» V

Figure 12.5. Plot of current / against potential difference V for a semiconductor diode.
The signs of V and / are defined in figure 12.4(6).

have sufficient energy to pass over to the p-type conduction band, and so
they carry a diffusion current proportional to

dEpn (12.9)

Assuming that the density of states pn(E) varies only slowly with E, this
is approximately

Hence the total electric current that passes through the crystal from the
n-side to the p-side is

(12.10)

This is the contribution from electron carriers; the hole carriers contribute
similarly.
/ is plotted against V in figure 12.5. When V is positive, / < /Q, but for
negative V, the current increases exponentially with the magnitude of V
(until the approximations made in the derivation are no longer valid).
The semiconductor diode can be made to operate as a laser (see chap-
ter 9). The first semiconductor laser was made in 1962, using stimulated
emissions associated with transitions from the conduction to the valence
band as the source of laser radiation. The pn-junction is forward-biased
(the p-type side is at a higher electrostatic potential than the n-type side),
providing the necessary pumping, and the end-faces of the diode, perpen-
dicular to the plane of the junction, are polished to give high reflectivity.

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:13:29 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.014
Cambridge Books Online © Cambridge University Press, 2014

The diode 151

Conduction
band

(a)

Valence
band
n-type p-type n-type

Emitter Base Collector
I I I

R
(b)

4-'
Emitter Base Collector

Conduction
band

(c)

Valence
band

n-type p-type n-type

Figure 12.6. The npn-junction transistor (a) with no external potential, (c) with
external potentials applied as in (6).

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:13:29 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.014
Cambridge Books Online © Cambridge University Press, 2014

152 12 Transistors Semiconductor lasers can be made very small — the size of a pin head — and have wide application.014 Cambridge Books Online © Cambridge University Press. so that almost all the electric current is carried by electrons rather than holes. the disposition of the energy bands at each of the two junctions is an in figure 12. (12.221.3(6). So almost all of them reach the collector terminal.org/10.6(a). before they are stopped. The potential applied to the collector lowers the level of the bottom of the conduction band in the collector region by a large amount.6(6). their energy is high enough for them to be able to suffer many collisions. Alternatively it may be a pnp-sandwich. for example in compact disc players and in fibre-optic communications. The donor impurity concentration in the n-type material is made much greater than the acceptor impurity concentration in the p-type material. This can be much greater than the input. Hence when the electrons arrive in the collector region from the base- region conduction band. so that most of these electrons diffuse across it before collisions stop or reverse their motion. as in the case of the diode. The p-type slice (called the base) is at slightly lower electrostatic potential than the left-hand n-type piece (the emitter). The junction transistor The junction triode transistor consists of a thin slice of p-type crystal sandwiched between two pieces of n-type crystal. as for the diode.11) The emitter current / e is controlled by the magnitude of the potential dif- ference between the emitter and the base. That is. The output of the circuit is the potential difference ICR across the resistor shown in figure 12.1017/CBO9781139171243. and cascade down the levels in the conduction band. almost all the current of electrons that enters the emitter emerges from the collector: /c~/e.208 on Thu Feb 13 17:13:29 GMT 2014. A Downloaded from Cambridge Books Online by IP 150. for then the collector potential would be lowered so much that the device would not operate. for the case where the two pieces of n-type material are doped with equal impurity concentrations. http://dx. We draw this in figure 12. The base is made very thin.6(6). 2014 . if R is large — though R must not be too large. The potential across the emitter/base circuit causes a flow of electrons through the crystal from the emitter to the base. and the right-hand n-type piece (the collector) is at rather higher potential.244. This potential difference may be regarded as the input to the circuit. When no external potentials are applied. Suppose now that the device is connected to a pair of batteries as in figure 12. it may vary with time. less than 5 /im across.doi. but for definiteness we consider the npn case.

Largely for constructional reasons. but as these cannot be manufactured so conveniently circuits are designed in such a way as to replace them with transistors as far as possible. A field-effect transistor.1017/CBO9781139171243. such an integrated circuit may also contain resistors and capac- itors (though not inductors). under these conditions electrons cannot readily cross the p-type region. these are called the source and the drain. This causes little flow of current since. These can also have the advantage of much lower power consumption. We have seen that the essential physics of the transistor is in the junction region.244. A wafer of insulator is sandwiched between a layer of metal. as we have seen. called the gate. Hence a very large number of tran- sistors can be accommodated in a single integrated circuit. and a piece of doped semiconductor. integrated circuits also commonly use a different type of transistor. This system replaces the base of a junction transistor. The junction transistor 153 Gate—^ Insulator — n c ' ^ Drain Source Figure 12.org/10. http://dx.doi. However. the remainder of the material has no active function. shown in figure 12.208 on Thu Feb 13 17:13:29 GMT 2014. Consequently.014 Cambridge Books Online © Cambridge University Press. this causes an electric field to be set up near the upper surface of the p-type material. 2014 . By doping a single piece of host crystal. With suitable doping tech- niques. the field-effect transistor. Although the gate is electrically insulated from the p-type material. power gain of 104 is typical. and holes are the principal carriers. the conduction and valence bands are lowered in energy in this region (fig- Downloaded from Cambridge Books Online by IP 150.7. it is placed between two pieces of n-type crystal. There are various types offield-effecttransistor. The pnp-transistor works similarly: the signs of all the potentials and currents are reversed. which is very thin. If the semiconductor is p-type.221. many junction transistors can be placed together on a single chip of crystal. several million on a crystal chip the size of a finger nail.7. The drain is raised to a positive potential relative to the source. the situation changes if the gate is raised to a positive potential. an example is the metal-oxide-silicon field- effect transistor (MOSFET).

current flows through the diode and the resistor when the input to the diode is positive. which is one of the first stages of a radio receiver.org/10. This is shown in figure 12. the arrow indicates the direction in which current can pass freely. That is. If the resistor is now replaced by a capaci- tor (figure 12.9(6)).doi. The larger the gate voltage. This inversion layer at the surface allows electrons to flow from the source across to the drain. The output is now Downloaded from Cambridge Books Online by IP 150. ure 12. modulated in amplitude by the lower-frequency audio signal. even though in the bulk of the p-type material holes are the majority carriers. 2014 . According to the Fermi-Dirac distribution. Lowering of conduction and valence bands near the surface of the p-type material when a positive potential is applied to the gate. there is then a substantial population of electrons in the conduction band in this region.208 on Thu Feb 13 17:13:29 GMT 2014.8. Two simple circuits The simplest application of the diode is in the demodulator circuit. is ap- plied between the p-type terminal of the diode and earth. in the circuit symbol that denotes the diode. the greater the electron concentration in the inversion layer and the greater the current.9(a). The radio-frequency carrier wave. If the other side of the diode is connected to earth through a resistor.1017/CBO9781139171243. and the Fermi level £ appears in the conduction band.8).221. http://dx.014 Cambridge Books Online © Cambridge University Press. since the impedence of the capacitor is low for high frequencies. the radio-frequency component is removed. the output potential across the resistor corresponds to the input with its lower half chopped off.244.154 12 Transistors Gate /Insulator Conduction band Depth below surface of p-type material Figure 12.

If this function is represented as a Fourier integral.208 on Thu Feb 13 17:13:29 GMT 2014. cosptcosut = 5 [cos (a.org/10. Two simple circuits 155 Diode t -M- o t Input Output o L Earth (a) Diode T -H. it is found to have a low-frequency component corresponding to the audio signal. G(t) is non-negative. but a function that has much the same shape as F(t) is G(t) = cos2 \pt cos2 \ut. + p)t + COS(CJ — p)t]. suppose that the input is the very simple amplitude-modulated wave cos pt cos ut where u is the radio frequency and p is the audio frequency.1017/CBO9781139171243. To see this. just the audio-frequency signal. (12.doi. Downloaded from Cambridge Books Online by IP 150. amplitude modulated with an audio- frequency signal. where 9{x) = 1 for x > 0 and 0(x) = 0 for x < 0. applied to a diode circuit (a) chops off the negative components of the input.221. and so the capacitor would short circuit the whole signal if the rectifier were not there. Notice that the rectifier is needed.9. http://dx. the input is the sum of two high-frequency components. Like F(t). A radio-frequency carrier wave.12) Hence. which is amplified in further stages of the radio receiver.244. The signal passed by the rectifier is F(t) = cosptcosut6{cosptcosut). It is not easy to calculate the Fourier transform. However.014 Cambridge Books Online © Cambridge University Press.T t Input Output J T Earth (b) Figure 12. 2014 . However. and both F(t) and G(t) oscillate in a similar fashion. (b) also removes the high-frequency component.

In a digital circuit. 2014 . for example car- rying a significant current and carrying little or no current.244. which is the audio signal superimposed on a constant potential. We have already described the use of the junction transistor as an am- plifier.13) As we have seen in (12.12).10.208 on Thu Feb 13 17:13:29 GMT 2014. http://dx. these are short- circuited by the capacitor. (12.221. a given circuit element operates in only two recognisable states. G(t) = \(COSpt+1)(COS Ut+ 1) = ^(cosptcosut + cosut + cospt + 1). rather make use of digital circuits. in particular calculators.doi. The flip-flop circuit.156 12 Transistors Figure 12.014 Cambridge Books Online © Cambridge University Press.org/10. So the output from the circuit is ^(cospt + 1). This is an example of an analogue circuit: the output varies continuously with the input. Many electronic instruments.1017/CBO9781139171243. as does the second term. the first term in this expansion consists of high-frequency components only. An example Downloaded from Cambridge Books Online by IP 150.

In the n-type material the Fermi level lies in the conduction band. The transition region at the junction is made so narrow that tunnelling readily takes place across the interface. This is as it should be if the transistor 1 is conducting. (6) when the n-type material is connected to a small negative potential V relative to the n-type. Sketch the energy bands near the interface (a) when no external po- tential is applied. http://dx.doi. Then the point A is at rather lower potential than the battery terminal E. or transistor 1 is switched on by a positive signal applied to its base. This is drawn in fig- ure 12. Then the potential of point A rises. Thus transistor 2 is not conducting.10. Investigate the energy-level structure.and n-type materials which are both very heavily doped. and in the p-type material it lies in the valence band.2 The tunnel diode is a junction between p. and therefore also that of point C and of the base of transistor 2. (The correct curve has a region of negative 'resistance' dl/dV. and thus the base of transistor 1.244.org/10. and so also the base of transistor 2. so the situation is stable. the circuit returns to the first stable configuration. If either transistor 2 is switched off by a negative signal applied to its base.014 Cambridge Books Online © Cambridge University Press. and (c) when it is connected to a fairly large negative potential.1017/CBO9781139171243. is at a comparatively low potential. Suppose that transistor 1 is conducting.208 on Thu Feb 13 17:13:29 GMT 2014. Thus transistor 2 now conducts. so that the impurity levels are spread out into bands.1 In a one-dimensional crystal an impurity atom replaces the atom at the origin. Hence the point C. taking as a model for the potential 12. Hence also the point D. and so the point B is at a potential somewhat higher than point A. so that it stops conducting. Problems 12. is at a comparatively high potential. Suppose now that a negative signal is applied to the base of transistor 1. this is the other stable situation.) Downloaded from Cambridge Books Online by IP 150. 2014 . Sketch the graph of the current I through the junction against the potential V. Two simple circuits 157 of such a circuit is the bistable circuit or flip-flop. where two npn-junction transistors are used.221.

221.I) We are interested in solutions having the property I/J —>• 0 as x -> dboo.244. This procedure is used for finding the stationary-state wave functions of the linear harmonic oscillator (chapter 3) and of the hydrogen atom (chapter 5). but if we used this instead. (A. For any given eigenvalue £?.(2mujx/h)Hf + [2m(E . Appendix A Power-series solutions In this appendix we describe briefly the procedure for solving differential equations by means of power-series expansions.doi. 2014 . so we begin by examining the differential equation (A.2) because the exponential factor has the property that for large x The exponential e+mujx l2h has the same property.asXs. when x is sufficiently large E is much less than ^muj2x2 and so the term — Ety in the equation is then much less important than the other two terms.208 on Thu Feb 13 17:13:39 GMT 2014.E)%l> = 0. The linear harmonic oscillator The Schrodinger equation to be solved is given in (3.1017/CBO9781139171243.4) s=0 158 Downloaded from Cambridge Books Online by IP 150. (A.2) into (A. (A.I) at very large values of x.I): H" . This suggests that it may be useful to write tl>(x)=e~mu'x2'2hH(x). H(x) would have to obey very much more stringent conditions at x — ±oo in order to satisfy the required boundary conditions for -0- Substitute (A. http://dx.3) Look for a power-series solution: oo Y.015 Cambridge Books Online © Cambridge University Press.15): {-h2/2m)(d2tf>/dx2) + (\mu2x2 .\hu)/h2]H = 0. (A.org/10.

16) and (3. http://dx. E— \TIUJ = 2nTujo.±hu = (2n + l)hw.org/10.221. these two possibilities correspond to the results quoted in (3.6) for large s. which can be justified. because the differential equation (A. so that H(x) is a polynomial. Its coefficients also have the property (A. Consider now the series expansion of the function AemuJX /h xB. E . so that the series terminates at s = 2n + 1. The hydrogen atom 159 We substitute this into the left-hand side of the differential equation (A.17). and for some odd value s = 2n + 1. we do not have the desired behaviour for ip at infinity.015 Cambridge Books Online © Cambridge University Press. If we make the plausible assumption.1017/CBO9781139171243. we see that this will happen if and only if: either a\ = 0. There are two parameters ao and a\ that can be chosen arbitrarily. so that all odd powers of x are absent.doi.4) tends to zero as s -> 00.6) (22 fi S This means that.244. we see that. The only way to avoid this is to arrange that the series expansion of H(x) terminates at some value of s. (A. When s is large. this relation determines as for all s > 1. where A and B are constants. or ao = 0. so that the series terminates at 5 = 2n. Downloaded from Cambridge Books Online by IP 150. From (A.5). the ratio of successive terms in the series (A. and therefore the series converges. we conclude that for large x Inserting this behaviour into expression (A. for any fixed value of #. that the form of H(x) at large x is controlled by the high powers in the series. 2014 . whatever may be the values of A and J5.5) gives 2mcj (A.2) for -0. so that all even powers of x are absent.208 on Thu Feb 13 17:13:39 GMT 2014. Together.3) is second-order. and for some even value 5 = 2n.3) and equate to zero the coefficient of xs: 2m \ ( ) / [ \ For any choice of ao and ai.

It tells us that if we want to choose the constant a such that a0 ^ 0. so that the expansion (A. the procedure is just the same. If we neglect the other two terms.015 Cambridge Books Online © Cambridge University Press. the differential equation to be solved is given in (5.244. The lowest resulting power of r that we find is ra~2. we find Now make the modified series expansion 5=0 and substitute this expansion into the left-hand side of the differential equation (A. we must take a slightly more general form of series expansion. By analogy with (A. (A.208 on Thu Feb 13 17:13:39 GMT 2014. http://dx.16): We again want a solution R(r) that goes to zero as r -> oo.4).160 Power-series solutions The hydrogen atom For the hydrogen atom.org/10. the resulting differential equation for rR(r) has solutions e ± K r .8) Substituting this into the differential equation (A. Downloaded from Cambridge Books Online by IP 150. then either a = — I or a — I + 1.1017/CBO9781139171243.2). and on equating the coefficient of this to zero we find ao[a(a-l)~ 1(1 + 1)] = 0. otherwise. in the previous example we knew that we had obtained both the independent solutions of the second-order differential equation (A.7). 2014 . How- ever.7) . 10) begins with the power r°". The procedure for obtaining the solution is similar to that above.doi.3) by making the expansion (A. In order to reveal both of the two independent solutions. the term At2 dominates over the other two terms in the square bracket in (A. (A. because the two parameters ao and a\ could be chosen freely.221. we therefore write rR(r) = e-Krf(r). If we proceed here in exactly the same way. When r is large.9).ll) This is known as the indicial equation. we obtain only one solution.

doi. Downloaded from Cambridge Books Online by IP 150.18). which is equivalent to the result (5. we see that [(Me2/A-KSQKTI2) — a] must be a non-negative integer JV.org/10. (A.015 Cambridge Books Online © Cambridge University Press. 12). even though it does satisfy the equation at all other points r . From (A.) For the ground-state solution. This means that then ip(r) does not satisfy the Schrodinger equation at r = 0. we must reject this solution. we conclude that for large r f(r)~Ae2KrrB. with N — n — I — 1. from the analysis of orbital angular mo- mentum in chapter 5.221.17). as+i/as ~ 2K/S. We therefore set a = / + 1. we have / = N = 0 and f(r) = ao. this determines all the other coefficients as. with the multiplicative constant chosen to normalise it correctly.244.So the wave function is just (5. 2014 .1017/CBO9781139171243. Because the Schrodinger equation is to be satisfied at every point r . When s is large. that / must be a non-negative integer. which is not an acceptable behaviour.208 on Thu Feb 13 17:13:39 GMT 2014. arguing as before.13) Hence. (Recall that we already know. http://dx. So again the series must terminate. the resulting wave function ip(r) diverges like r~l~l at the origin. We now equate to zero the coefficient of For any choice of ao. The hydrogen atom If we make the choice a = — /.

Our presentation will be sufficient for practical purposes. (B.I.org/10. 1958). Define the function A£(x) = (l/eV5r) e~x /£ (e > 0).016 Cambridge Books Online © Cambridge University Press. the function becomes extremely small. Outside the peak.208 on Thu Feb 13 17:13:52 GMT 2014.2) When plotted against #. (B. (B.1017/CBO9781139171243.f The delta function The delta function is an example of a generalised function or distribu- tion. the function A£(x) has a peak at the origin. The simplest way to understand its properties is to use a limiting procedure.I) For each value of the constant parameter e. This means that if we define an integral poo Af] = / dxA£(x)f(x).3) J -c then for a wide class of functions f(x) the value of the integral when e is very small depends almost entirely on the values that f(x) takes very M J Lighthill. Appendix B The delta function and Fourier transforms In this appendix we present a heuristic discussion of the properties of the delta function 5(x). so that if e is allowed to become very small the peak becomes very tall and very narrow.244. and we show how by using the delta function one may readily and simply understand the properties of Fourier integrals. 2014 .doi. This peak has height 1/ey/n and width of order e (exactly how we define the width does not matter).221. a mathemati- cally rigorous treatment may be found in pages 15 to 29 of the book by Lighthill. as in figure B. this function satisfies a:) = l. http://dx. 162 Downloaded from Cambridge Books Online by IP 150. Introduction to Fourier Analysis and Generalised Functions (Cambridge University Press.

and ultimately lim/ e [/] = y dx Ae(x)/(0) = HO).4): dxS(x)f(x)=f(0) (B. close to the origin.org/10. By making a simple change of variable in (B.016 Cambridge Books Online © Cambridge University Press.I.4) Here we have used (B.208 on Thu Feb 13 17:13:52 GMT 2014.221.5) The mathematically rigorous definition is arrived at via (B. The delta function 163 Figure B.6). {B. (B.a)f(x) — f(a). Different types of generalised function or distribution are defined by using different classes of test function.6) for a suitable class of 'test functions' f(x).2). 2014 .1017/CBO9781139171243. for example tempered distributions are defined using functions f(x) that. are bounded by some fixed power of x at x = ±00.I) Downloaded from Cambridge Books Online by IP 150. The test functions are ordi- nary functions.doi. http://dx. (B. Plot of the function A£(x) against x. one obtains dx 6(x . A crude definition of the delta function is S(x) = lim A£(x).3) and (B. together with all their derivatives.244.

221. and the two peaks each approach very close to the origin. When e is small.doi. and higher derivatives. in such a way that for any positive a and b. one positive and one negative. At the origin.1017/CBO9781139171243. Now.org/10. the derivative of A£(x) has two peaks close to the origin. as drawn in figure B. Plot of A'€(x) against x. One may define also the derivative 5'(x) of 8(x).164 The delta function and Fourier transforms Figure B./ dxA£(x)f(x). From the way in which we have arrived at 8(x)1 it follows immediately that S(x) = 0 except at x = 0. 2014 . (B. each of these peaks becomes very narrow and very tall. (B.8) J —< This is why it is not a function in the ordinary sense. an integration by parts gives / ]!!'0O. This may be compared with the property of the symbol Hence 6(x — a) is a generalisation of 8{j to the case where its discrete suffixes become continuous variables.244. As £ —> 0.2. http://dx.016 Cambridge Books Online © Cambridge University Press.9) J— J— Downloaded from Cambridge Books Online by IP 150. 5(x) is positive infinite.208 on Thu Feb 13 17:13:52 GMT 2014. dxS(x) — 1.2.

208 on Thu Feb 13 17:13:52 GMT 2014.244.I) of Ae(x). S£(x)= f dyA£(y). So.10) J-co J-co We have used (B. the step in this function becomes progressively steeper. Higher derivatives of S(x) may be defined similarly.1017/CBO9781139171243.ll) J — oo This is plotted against x in figure B.doi. successive derivatives of 5(x) are more and more singular at the origin.016 Cambridge Books Online © Cambridge University Press. Downloaded from Cambridge Books Online by IP 150.221. dxS(x)f'(x) = -f'(0). (B. In fact dy6(y)=6(x). Because of the explicit definition (B. http://dx. the first term on the right-hand side vanishes unless f(x) explodes violently at infinity. until finally the function changes abruptly from 0 to 1 at the origin.6) for the case where f(x) is replaced by f'(x).3. Because the rth derivative of A£(x) changes sign r times very near the origin when e is very small.org/10. Consider now the indefinite integral of Ae(x). by letting e —> 0. 2014 . The delta function 165 1— Figure B. (B. Plot of @£(x) against x.3.12) where 6(x) = 0 x<0. As e -> 0. we arrive at the definition of 8'{x)\ pCO pOO I dx8'(x)f(x) = . (B.

12) with respect to x.208 on Thu Feb 13 17:13:52 GMT 2014.244. any finite value will serve.a). and equal to the ordinary differentiable function g(x) for x > a (see figure B. defined by the Gaussian function (B.f'(x)]6(x .221. Fourier transforms In our discussion of the delta function. we used a particular function Ae(ai). we obtain initially [g(x) — f(x)]5(x — a). This means. (B. 2014 .6) is unchanged if we change the value of its integrand by a finite amount at not more than a finite number of points.a). but we may replace x by a inside the square bracket because the delta function vanishes unless x — a. in particular.13) Thus distribution theory allows us to define the derivative of a function that has a finite discontinuity.4). that it does not matter what value we assign to 6{x) at x = 0. almost any function having this property will be just as good.f(a)}S(x . defined to be equal to the ordinary differentiable function f(x) for x < a.2) for each value of e. but 0(0) = \ is perhaps the most natural definition. There is nothing special about this particular choice of A£(x).doi. we obtain d6/dx = S(x).166 The delta function and Fourier transforms Notice that the value of an integral such as (B. http://dx. Another example is the function Consider now the integral Downloaded from Cambridge Books Online by IP 150.I). (B. provided it is chosen so as to satisfy the integral condition (B.14) and so. applying the usual rule for differentiation of a product of functions.a) + [g(a) .f(x)]0(x .15) In the last term.016 Cambridge Books Online © Cambridge University Press. which has a peak that becomes progressively taller and narrower as e —> 0. If we differentiate (B.1017/CBO9781139171243. (B.org/10. Consider the function F(x). F'(x) = f'(x) + [g'(x) . Hence two generalised functions that differ by a finite amount at a finite number of points are equal in the sense of distribution theory. We have F(x) = /Or) + [g(x) .

doi. http://dx. It is understood that. a damping factor e~£^ is to be included in the inte- grand so as to make the integral converge. However. as in (B.016 Cambridge Books Online © Cambridge University Press. Downloaded from Cambridge Books Online by IP 150.4. (B. integrate from —oo to oo with respect to ±.5). and after the integral has been calculated e is allowed to tend to zero.208 on Thu Feb 13 17:13:52 GMT 2014.18). This is simple to evaluate by elementary means: it is just / d k e ikx+ek + J- 1 IX + £ IX — 6 = 2TTA£(X).244. Fourier transforms 167 f(x) Figure B.17) To proceed further in a rigorous fashion. we write £W lim = 2TT6(X). Plot of F(x) aganist x. (B. in the same informal sense that we understand the delta function to be defined by (B.221.org/10.18) It is usual even to write (B.1017/CBO9781139171243. we should multiply both sides by a test function f(x). and then let e -> 0.19) The integral here does not converge in the usual sense. 2014 .

20) and (B. Otherwise it is necessary to include in the integrand a damping factor e~£^ and let e —> 0 after the integral has been calculated.doi.168 The delta function and Fourier transforms Consider now the function $(fc). We insert the representation (B.221. a damping factor e~e\h\ may be needed in order to make this integral converge. Most of the textbooks consider only classical Fourier transforms.22) lkx (f)(x) = -4= I dke-' $(k). this integral converges in the usual sense. Putting (B.19) and then (B. in this case it may be shown that both $(k) and (p(x) are ordinary func- tions.244. If <j>(y) is bounded by some power of y as y -> ±00.1017/CBO9781139171243. so does the other. It may be that both the integrals in (B.org/10.208 on Thu Feb 13 17:13:52 GMT 2014. We say that the functions $ and <j> are Fourier transforms of each other.20) to x. defined by the integral [ (B. This is the case of classical Fourier trans- forms.21) together.7). http://dx. This is the Fourier inversion theorem: if one of the relations holds.016 Cambridge Books Online © Cambridge University Press. The possibility of including damping factors allows the very pow- erful extension of Fourier transform theory to include the possibility of $ Downloaded from Cambridge Books Online by IP 150. and changing the dummy variable y of integration in (B. We now calculate 1 4 Again.22) converge without needing any additional damping factors. this procedure will give a convergent result.20) for $(k) and obtain -£= (°° dke-'lkx$(k) = ~ f 27r V2TT J-00 7- where we have used (B. we have dxeikx(f)(x) (B. If (j)(y) -> 0 sufficiently rapidly as y -> ±00. 2014 .20) V27T J- for some suitable function <j>(y).

Often. because of (B. It is possible to show that if either $ or 0. 1 f° \/2W-oo V2^ The inverse relation is obtained from (B.23a) For.6). Further.and y-integration. (B. together with all its derivatives.21) we have implicitly interchanged the order of the k. is bounded by some power at infinity. by replacing the dummy variable k of integration by —A. However.1017/CBO9781139171243.236) v2 This is an example of a classical Fourier transform: no damping factors are needed.22). (B. (ii) e-*'/a» <_> « e-'*V4. Simple examples of Fourier transforms (i) S(x) <—> -=. then so does the other.22).016 Cambridge Books Online © Cambridge University Press. then no factor appears in front of the other. Notice the factor 1 / ^ T T in front of each of the integrals in (B. we shall use the definition (B.221.19). 2014 .244.doi. then these interchanges are justified. We have 2 o o-a fc2/4 r ^2^ J_ Downloaded from Cambridge Books Online by IP 150. http://dx. it may be shown that if either one of the functions $ or <fi has this property at infinity.. In the manip- ulation in (B. Fourier transforms or <j> being a distribution rather than an ordinary function. one chooses to define the Fourier transforms instead by writing 1/2TT in front of one of the two integrals.208 on Thu Feb 13 17:13:52 GMT 2014.org/10. and when damping factors are needed we have interchanged doing the integration with taking the e -¥ 0 limit.

http://dx.org/10.221. (B. As e —> 0.208 on Thu Feb 13 17:13:52 GMT 2014. by differentiating the second relation of (B. except for a pole at k = -\e. putting in the necessary damping factor.) (iii) For any pair of Fourier transforms </>(#).12). we show how to verify the inverse relation. i / 1 \ lim dfce . after the change of integration variable the y-integration is no longer along the real axis in the complex y plane. For.23c) This is proved immediately. the integra- tion may be restored to the real axis by using Cauchy's contour-integral theorem.L I dxeikx~£X..\\ka2) — y and use the standard integral dye-y la = a v ^ (Strictly.1017/CBO9781139171243.. make the change of variable (x .24 ) The integrand is analytic throughout the complex k plane.doi. and this integration may be performed by elementary means. it is important to retain the ie in the answer.244. Notice that because the result is singular at k = 0.170 The delta function and Fourier transforms To perform the integration here. (iv) 0(x) <—> lim -4= ( . (B. To illustrate this. this pole approaches the origin.22) repeatedly with respect to x.016 Cambridge Books Online © Cambridge University Press.23d) where 6(x) is the step function defined in (B. ). 2014 . but the term Downloaded from Cambridge Books Online by IP 150. --L f dxeikx9(x) = lim . We have to evaluate 1 f°° . However. -/*= / ^ = TT7Z ' ( B .

h .208 on Thu Feb 13 17:13:52 GMT 2014. ie tells us that it approaches the real axis.k2) Downloaded from Cambridge Books Online by IP 150.23e) This is known as the convolution theorem. along which lies the contour of integration. When x > 0.221. the integral vanishes by Cauchy's theorem. Then the exponential factor in the integral (B.org/10. without changing the value of the integral. because the pole is in the lower half-plane. (v) Let </>i(x) and fafe) be the Fourier transforms of and respectively. Suppose now that x < 0.016 Cambridge Books Online © Cambridge University Press. See figure B. we instead add on to the original integration an integration over an infinite semicircle in the lower half-plane.244. Fourier transforms 171 / \ / \ / \ / \ I / \ ^Pole / \ / \ Figure B. 2014 . Hence we may add this contour to the original integration contour along the real axis.5.doi.1017/CBO9781139171243. The proof is as follows: J_ dk2 5(k .24).24) vanishes as k —> oo in the upper half of the complex k plane sufficiently rapidly that the integral taken over an infinite semicircle in the upper half-plane would be zero. which now lies within the contour. The integral is then evaluated by taking the residue at the pole. http://dx. from below rather than from above. Then 4= f dk> (B. The integration contour is now closed and.5. Integration contours for the integral (B.

we prove Parseval's theorem. Finally. This is that.016 Cambridge Books Online © Cambridge University Press.1017/CBO9781139171243.221. 2014 . for any pair of trans- forms (j)(x) and $(&).org/10. POO /*OO / dx\</)(x)\2= / dk\${k)\2.23e).doi.208 on Thu Feb 13 17:13:52 GMT 2014. http://dx.172 The delta function and Fourier transforms and this integral is equivalent to the integral in (B.244.25) J— oo J — oo For the left-hand integral is /»OO / dx<f)*(x)<t>{x) J — oo /•o = / J— Downloaded from Cambridge Books Online by IP 150. (B.

http://dx.zd/dy).2) In spherical polar coordinates (see figure 5. y = psin</>.xd/dz). Prom (C. Lz = -i(xd/dy-yd/dx).5) Solving these three simultaneous equations for dp. (C. d<j> and d0 we have dp = cos (/> dx + sin (/> dy.3) where p = r sin#. (C.2): hL = r A(-ifiV).1017/CBO9781139171243.1) X = p COS (/>. and d/dcj) is calculated with p and # constant. we need ^ + + dx dx dp dx 89 dx dcf) and the similar expression for d/dy. Ly = —\(zd/8x .3). dy = sin <f> dp + p cos <f> d</>.244.208 on Thu Feb 13 17:14:04 GMT 2014. Here the partial derivative 8/dp is calculated keeping 9 and (j) constant.I) so that Lx = -i(yd/dz . each partial derivative with respect to x is calculated with y and z constant.p cosec2(9 d0. (C. 2014 . To calculate Lz in spherical polar coordinates. pdcj) — — sin 0 dx + cos 0 dy.6) 173 Downloaded from Cambridge Books Online by IP 150.doi. Appendix C Orbital-angular-momentum operators The orbital-angular-momentum operator is defined in (5.221.017 Cambridge Books Online © Cambridge University Press. On the other hand. pd9 = sin 0 cos 0(cos 0 dx + sin 0 dy) — sin2 0 dz. dx = cos (f)dp — ps'mcf) d(/>. dz = cot 9 dp . 8/86 is calculated with p and <£ constant.org/10. z = pcot9. (C. (C.

The result is given in (5. The other two components of L may be calculated in the same way. with x and y from (C. Ly — —i(cos (j> d/dO — cot 6 sin <j> d/dcj)). Lz = -id/d<f>. we set dy = dz = 0.4). 7— = . — = —sine/).sin6cos0cos </>. 2014 .208 on Thu Feb 13 17:14:04 GMT 2014. Downloaded from Cambridge Books Online by IP 150.244. and calculate d/dy similarly.174 Orbital-angular-momentum operators To calculate the partial derivatives with respect to x.1017/CBO9781139171243.5).3). http://dx. Using these expres- sions. (C.doi.221. we calculate Lz from (C.3).017 Cambridge Books Online © Cambridge University Press. and so have — = cos</>. (C.7) ox ox p ox p We insert these in (C.org/10.8) It is then straightforward to evaluate L = LXLX + LyLy + LZLZ) with the result given in (5.2). so that altogether Lx = i(sin (j)d/de + cot 9 cos (j)d/d(j)).

They are V • £ = 0. In the presence of electric charges or of a magnetically or electrically polarisable medium. V • B = 0. we find that V A (£ + A) = 0. denote partial derivatives with respect to t. We have written the equations for fields propagating in empty space. The constants So and no are related to the speed of light c by c2 = 1/eoMo. B may be expressed in terms of a vector potential A: B = VAA. http://dx. it is straightforward to verify that the second equa- tion of (D. (D. (D. Maxwell wrote down these equations before either special relativity theory or quantum mechanics were known.3) By calculating V • B . V A£ = -B. If we insert (D.I) is indeed satisfied if B is expressed in this way.1017/CBO9781139171243.3) into the third of Maxwell's equations.244. Appendix D Electrodynamics The electric field £ and the magnetic field B obey four equations known as Maxwell's equations.org/10.208 on Thu Feb 13 17:14:13 GMT 2014.5) 175 Downloaded from Cambridge Books Online by IP 150.221. but neither of these two developments led to any need to change the equations. but the converse result is also valid. where the dots.018 Cambridge Books Online © Cambridge University Press. (D-2) A general theorem of vector calculus allows one to deduce from the second of Maxwell's equation that. 2014 .4) Another general theorem of vector calculus then allows us to deduce that £ + A can be written as the gradient of a scalar potential (f>: (D. at least locally.doi. as usual. the equations must be modified.

which is the condition introduced in chapter 8. e(r • V)A. (D.18) we wrote the Hamiltonian for an electron of charge — e in inter- action with an electromagnetic field: H = V + (p + eA)2/2m .7) In classical physics.ecp.221. and d/dp denotes the corresponding differentiation with respect to p. remain unaltered. d e at VV m {[(p + eA) • V)A + (p + eA) A (V A A)} + eV</>. and one example of a possible gauge condition is V • A = 0. http://dx. In (8.6). A -> A + V / .018 Cambridge Books Online © Cambridge University Press.3) and (D. Hence the transformation (D. and again the converse is also true. given by (D.244. (D. which is called a gauge transformation. The fields B and S are physically measurable quantities.5).t) and change A and <j> together in the following way.9) gives (d/dt)p = mr-e(d/dt)A = mr -eA.4). If we take some arbitrary scalar function f(r. This freedom allows one to impose on (p and A certain constraints known as gauge conditions. If we apply Hamilton's equations to the Hamiltonian (D. 2014 . The first equation in (D.7).10) Downloaded from Cambridge Books Online by IP 150. then the fields B and £. we have used the vector identity Vu2 = 2(u • V)u + 2u A (V A u).1017/CBO9781139171243. the Hamiltonian leads to equations of motion through the elimination of p from Hamilton's equations (8. can have no physical consequences. The differentiation d/dr is carried out with p kept constant. the operator d/dr is usually written as V.15).doi. and r is kept constant for d/dp. we obtain r = (p + eA)/m. but the functions A and (j) are not directly measurable.9) To write the second of these equations.176 Electrodynamics It is again straightforward to verify that if S is written in this way then it automatically satisfies (D. which are dr_dH dp_ dH Here. (D.208 on Thu Feb 13 17:14:13 GMT 2014.org/10.

and so obtain the equation of motion mr = .5). (D. Electrodynamics 177 Here.221.244.1017/CBO9781139171243.208 on Thu Feb 13 17:14:13 GMT 2014. so that its total derivative dA/dt is not simply equal to dA/dt = A. but rather dA/dt — A + (r • V)A.018 Cambridge Books Online © Cambridge University Press.er A B.9) using the first equation of (D. we have used the fact that A generally varies with both t and r.W . 2014 .3) and (D.ll) Downloaded from Cambridge Books Online by IP 150.doi. We reduce the right-hand side of the second equation of (D.eS .org/10. http://dx.9) and also (D.

(E. [D. Let %j) be any one of these.019 Cambridge Books Online © Cambridge University Press.5) where c is the eigenvalue of D and is a constant. http://dx. 2014 .org/10.3) If an operator gives zero when it is applied to any function /(#). Because the operator D does not correspond to an observable. that is. it need not be Hermitian.doi.2). and so we have DH(x)f{x) = H(x + L)f(x + L)= H(x)Df(x). so that Dt/){x) = &l)(x).221. this is equivalent to saying that the operator vanishes. (E.244. Thus (E. Because of the definition (E.4) As we explained at the end of chapter 4. and c can be complex. where we have used (E.5) gives ).H] = 0. (E.2) This definition applies only to functions / . Hence.H(x)D)f(x) = 0. However. this implies that the operators D and H have a complete set of common eigenstates. not to operators such as H.I) and (E. Hf is a function. (E. (E.I) We define the translation operator D such that for any function f(x) Df(x) = f(x + L).6) 178 Downloaded from Cambridge Books Online by IP 150.1017/CBO9781139171243.208 on Thu Feb 13 17:14:32 GMT 2014. (E. the stationary states can be chosen to be eigenstates of D. H{x)]f(x) = (DH(x) .3) gives [D. Appendix E Bloch waves The Hamiltonian that describes the motion of an electron in a one- dimensional model of a crystal is unchanged when the electron coordinate is displaced through a distance L: H(x + L) = H(x). for any function f(x). (E.2) of JD.

The wave function (E. we needed (E. so that c = eia with a real.I) and (E. the wave function is negligibly small over most of its interior.8) give |c| 2 = 1.019 Cambridge Books Online © Cambridge University Press. and it decreases exponentially with increasing distance from that end.10) This relation is known as the Block condition. namely xl>(x)=eikxuk(x).221. each time that we displace the coordinate x through distance L the wave function is multiplied by elkL.6) are valid for all pairs of points (x.ll) then contains a factor e~k2X. For this reason. Without loss of generality.208 on Thu Feb 13 17:14:32 GMT 2014. (E. so that Uk(x) has the periodicity L of the lattice. So for any positive or negative integer m <iP{x + ml) = eikrnLip(x). Then there are also solutions with k complex. This is not so when the crystal is taken to be of finite length.8) J — oo «/— oo Together. (E.7).3).7) and (E.1017/CBO9781139171243.9) where k is real. Suppose that the crystal extends from x = —oo to x = +00.ll) it is straightforward to see that the Bloch condition requires that Uk(x + mL) = Uk{x). (E. the solution is called a surface state.)= d^|V(^)|2 = 0 2 / (E. write k = k\ + i&2.6) and (E. 2014 .6) to be true for all values of x.x + L) that lie within the crystal. Depending on the sign of &2. and we have ignored them in this book. (E. (E. http://dx. so that (E. According to (E.244. In order to derive the result that k is real.6) implies that /•OO 1 /*O .doi. Block waves 179 The properties (E.tl. Unless the crystal is very small. this is largest at one or other of the two ends of the crystal.7) J-00 \c\ J- By making a change of integration variable from x to (x + L) we also have the identity /•OO /«OO / dx\ip(x)\2= dx\ip{x + L)\2. It turns out that surface states make a negligible contribution to the bulk properties of all but the very smallest microcrystals. we may set a = kL: c = eifcL. with ki and &2 real. If we choose to write the wave function in the form (10. Downloaded from Cambridge Books Online by IP 150.org/10. In the case of such a solution.

this is about 3 x 106 m s" 1 . Applying (1.36) to the electron after the scattering gives me2 + cp. so for a mass of 1 kg the answer is about 3 x 1020 years. so that its energy is me2. 1. We need the difference between the values of v for n = 2 and n = 1.1 According to (2.org/10.cp1 = c[m2c2 + (p . according to (2.doi.36).5 The calculation is similar to problem 1. 2014 . Chapter 2 2. Hints for the problems Chapter 1 1. according to (1. and for an electron about 10"3 s. the electron afterwards has momentum p and energy E = me2 + cp. 1. This means that E2 > m2c4 + c2p2. 1. the stationary-state solutions become multiplied by eiVot/h.9).136) so the allowed values of v are nhn/ma. adding a constant VQ to V(r) is equivalent to instead subtracting the same constant from E. its energy then is cp.2 The allowed values of \mv2 are given by (2.1 5 x 1029 photons per second.76). For the electron.4. http://dx.221. the answer is negligibly small: the velocity is effectively a continuous variable and quantum mechanics need not be used to describe the motion. Then according to (1.4) its energy has changed from cp to cp1. Let the momentum of the photon be p.4 Suppose that after the scattering the photon has momentum p'.208 on Thu Feb 13 17:14:46 GMT 2014. using p • p' = pp' cos 0. So if the photon is absorbed. Squaring both sides. Hence.4). in violation of (1.2 Work in the frame in which the electron is initially at rest. 2. 1. and dividing through by 2mc3ppf/h gives the answer.244. In the case of the tennis ball. This leaves unchanged the probability density |\I/|2 for all r and all t.3 The wavelength of visible light is of the order of 500 nm. p')2]1'2.020 Cambridge Books Online © Cambridge University Press. 180 Downloaded from Cambridge Books Online by IP 150.1017/CBO9781139171243. and for the tennis ball it is about 10" 33 m s" 1 .

calculate dj/dx and use the time-independent Schrodinger equation to show that it vanishes. &(x. Then the integrand is the probability of finding the first particle in the volume element d 3 ri and the second in d3r2. The Schrodinger equation is where V? is the V2 operator constructed from r» (i = 1. 2. This is manifestly independent of t.244. Downloaded from Cambridge Books Online by IP 150. and therefore where if)1 = di/j/dx. (x2) = [x2\^n(x)\2dx.208 on Thu Feb 13 17:14:46 GMT 2014. (x)= I" x\ipn(x)\2 dx Jo ({x-(x))2) = (x2)-(x)2-.221. To show that it is independent of x. x2dx = \a2 a a Jo Jo and so ((x — (x))2) = j^a2.2).org/10. where the integra- tions over both r\ and V2 extend over all space. the particle is equally likely to be found anywhere within the well and so 1 fa 1 fa (x) = . This does not imply that j is independent of r. "" ' ' 2mi V dx dx In a stationary state. Applying the divergence theorem to (2. Jo Classically.doi. In three dimensions both p and j are independent of t. Hints for the problems i 2.17) gives V • j = 0 . http://dx.t) = ip(x)e~lEt/h. 2.4 In one dimension. 2014 . {x2) = .1017/CBO9781139171243.3 Cn = (2/a) 1 / 2 (times an arbitrary complex number of modulus 1).020 Cambridge Books Online © Cambridge University Press.5 Normalise ij) such that Jd 3 rid 3 r2 |^( r i 5 r2)\2 = 1./ xdx = | a .

13) — see figure 3.10). Prom appendix A. and then there is a good chance that the particle will be found outside the well. the exponential fall-off of the wave function beyond the edges of the well is very slow. that is when it is at (3 = TT/6.116).11).3 With the changes of variable x' = x — eE/muP1 and E' = E + e 2 £ 2 /2raa. 3. V = Ui(ri) + Then there are separable solutions i/. from (3. So the new wave functions are just ipn(xf).4 Prom (3.org/10. the new Schrodinger equation becomes just the same as the old one (3.020 Cambridge Books Online © Cambridge University Press. the dashed curve moves downwards and the odd-parity solution disappears when its intersection with the cot 2 |/36 curve is where the latter touches the (3 axis.1 According to (3. where ^n is the function in (3. As U decreases.221. while outside the well it goes exponentially to zero.244. From (3. At the two edges of the well. If the values of U and b are such that in a given state the value of a is very small.2. within the well i/> oscillates sinusoidally. and the new energy levels are 2 2 2 \ 3. both ij) and dip/dx are continuous. the lowest levels correspond to the smallest allowed values for ft and so. HQ(X) is a constant Downloaded from Cambridge Books Online by IP 150. their wave functions have the smallest number of oscillations.doi.15). 3.208 on Thu Feb 13 17:14:46 GMT 2014.182 Hints for the problems If the particles do not interact with each other. the wave functions are Hermite polynomials Hn(x) multi- plied by an exponential factor.17).1017/CBO9781139171243. Chapter 3 3. 2 . 2014 .17).(ru r2) = ^ i ( r i ) ( ) 2rrii and E = Ex + E2.2 There must always be at least one even-parity solution of (3. http://dx.

and so E = —m/j. This gives a = m/i/h2.35).456) gives for nuclei of the same radius r oc E~ 1 / 2 exp(4.020 Cambridge Books Online © Cambridge University Press. 2014 . 3. Replacing x by —x in this equation. the Schrodinger equation is i/>" = a2tp. Hints for the problems 183 and (see (A. 3. as in (i).2mux2/h) with a 0 a constant.5 Use (3. use / ^ dxe~Xx = (TT/A) 1 / 2 and j°° dxx2e~Xx2 =-± (^ dxe~Xx\ 3.1017/CBO9781139171243. 3.10) gives /? -> oo such that [/ .mn/h2f3. This requirement also picks out the case E < 0. where E = -h2a2/2m.8 The mass of the a-particle is about 4 times that of the proton.12a) the right-hand side becomes mfi/h2. We choose the solutions Ae~ax in x > 0 and Beax in # < 0. then n is imaginary. Notice that (3. Because each value of E is supposed to correspond to only one independent wave function. (3. http://dx. so that ip —> 0 as x —> ±oo as required for a bound state. ft V/2ra = U-E to show that in case (i) \T\2 « : \A\2. Hence if C/6 = /i. so that a is real. where the a-particle E is in MeV. To verify the orthogonality.5 x 109 years.126) it diverges and gives no finite value for a. Combining these two equations gives C2 = 1.6 For x 7^ 0. This gives r\j/rFm w 5 x 1013 and hence T\J W 3 x 109 years. At x = 0.27) is valid also when U is negative. The other condition at x — 0 is (3. so the simple model gives a surprisingly accurate result.0Z J E~ 1 / 2 ).7 The first step is achieved simply by replacing x by —x throughout the Schrodinger equation (always valid for any equation!) and using V(x) = V(—x). I/J must be continuous. \fib .doi. Thus in (3.5)) H2(x) = a o (l . with A replaced by —/i. When [/ —>> oo. rather than E > 0. and in case (ii) \T\2 « \A\2. so C = ±l.org/10.h2f32/2m. but in (3.27) and the definitions h2k2/2m = E.244. it must be that ^{—x) = Cfi/j(x)) for some constant C.2/2h2.221. and e2/he0 = 4TTC/137. SO (3.208 on Thu Feb 13 17:14:46 GMT 2014. Downloaded from Cambridge Books Online by IP 150. we see that ip(x) = Cip(-x). so that A — B. The experimental value is 4. that is M « 3750 MeV/c 2 .

is a phase factor that varies with both x and t.(P))2) = (p2) .2 #(z.) in terms of the Fourier transform oi pcf>(p). Hence show that (px} = J dp \(f)(p)\2. where JVi = (a2 + 2iM/m)'1/2{ir/2a2)'1/A and e1.3 The operator corresponding to px is —ihd/dx.13). 4.1 /»OO = / dx \&*(—ihfy1).t) = Nie-(x-<x))2MAx)2eL~. this is f Integrating the first term by parts twice. from the uncertainty relation (4. A similar approach may be used for the expecta- tion value of any power of px.5 (x) and (p) are zero because of the symmetry about the origin. Write tp(x) as the Fourier transform of 0(p) and differentiate it to write ^'(o.1017/CBO9781139171243. 2014 .184 Hints for the problems Chapter 4 4. J—oo = UP .020 Cambridge Books Online © Cambridge University Press.221.244.(p)2 /•OO r I O II / v J — oo 4. so (px) = — ih J dx ip*if)'.208 on Thu Feb 13 17:14:46 GMT 2014. (E) > h2/8m(Ax)2 + Downloaded from Cambridge Books Online by IP 150. at at y J From the time-dependent Schrodinger equation.org/10. and using the fact that for a wave packet the wave function ^ vanishes at x = ±00. The second part is done similarly. 4.4 — (x) = — dx ty*xV = / dx(4/*x^ + **x*). 4. Because (p) and (x) are zero. this is (Ap)2/2m+±muj2(Ax)2. gives the result. http://dx.doi. So.

12 By taking linear combinations of the given wave functions.doi. It follows that in each bound state (V) = Jd3r V\*\2 > Jd3r where Vo is the minimum value of V.1017/CBO9781139171243.6 The maximum accuracy is achieved by attempting to drop the dart from directly above the target point. The second integral is (h2/2m) J d 3 r | V#| 2 . which is positive.16) is proved by integration by parts. 4.org/10. the dart hits the ground at distance R = Ax + (2l/g)1/2Ap/m > Ax + (2l/g)1/2h/2mAx. ft2 f 3 h2 f = -— d rV-(rV$) + . 4. The minimum occurs for Ax = (lh2/2m2g)llA. 4. because Q is Hermitian.221. giving (E) > \fiuj. Hints for the problems 185 The minimum value of the right-hand side is when (Ax)2 = h/2mu. http://dx. one finds that the complete set of eigenfunctions is fa = (fa - Downloaded from Cambridge Books Online by IP 150./ 2m J 2m J The divergence theorem transforms the first integral to an integral over a surface that encloses the system so that \I> vanishes over this surface. but the uncertainty principle implies that in practice the point of release is displaced through a horizontal distance Ax and the dart has a horizontal component of momentum Ap (in the same direction) at the time of release.244.. and ap- plying Q to them.7 4. 4.8 The identity (4.10 This follows immediately from (4. this integral therefore vanishes. soR> (8lh2/m2g)1/A.208 on Thu Feb 13 17:14:46 GMT 2014. where AxAp > \Tt.020 Cambridge Books Online © Cambridge University Press. 2014 . Since the time to drop a height I is (21/g)1/2.16) and the fact that q must be real.

org/10.27) is obtained from this inequality by making the replacements Q Q ( Q ) R R ( ) Chapter 5 5. 2014 . The expectation value in the state * is (*. 0-^6 and <f> -» <j> — a.244.1 Under the rotation. so ihd(x2)/dt = ([# 2 .208 on Thu Feb 13 17:14:46 GMT 2014.186 Hints for the problems The first of these has eigenvalue +1.F^FV) = ( $ .221. This expression has its minimum when A = — |i( [Q. whence the required inequality follows. so the system is left in this state after the first measurement. The uncertainty principle (4.doi. t ) | 2 is real and > 0. so dim (h) = ML 2 T" 1 .7). YJm is multiplied by e~ imQ .11.$) = (Q2) -i\([Q.X2(R2). 4. Downloaded from Cambridge Books Online by IP 150. because Q and R correspond to observables and therefore are Hermitian. From (iv). F t = Q + iXR. Multiplying out the terms in pip. (V. H = (p2x + p2 + pi)/2m. 5. $ ) . The only part of this which does not commute with x2 is the first term. which from its definition / d 3 r | $ ( r .16). The commutator may be simplified as in problem 4. http://dx. 4. F^F^). (ii). according to (5. At a later time t the wave function is lElt/h e~lElt/h = (</>!e~' elE2t/h)/V2. So. ($.R}) . (iii) and (iv) follow immediately from the definition of the Tsymbol given in the question. I/J2 2 e-'lE2t/h ++ I/J The required probability is Cos[(E1-E2)t/h}}. evaluate the commutator of the result for d(x2)/dt with H/ih.R] )/(JR2). a0 = (47rsohc/e2)(h/mc) « 5 x 10~ n m.13 This follows immediately from the time-dependent Schrodinger equa- tion. 4.020 Cambridge Books Online © Cambridge University Press. To differentiate again with respect to i. so dim (e2/47re0) = ML 3 T~ 2 . where F = Q-iXR and so. The momentum operator is (-iftV).p 2 ] )/2m.15 (i) is just the definition (4.1017/CBO9781139171243.2 e2/47re0r2 is a force.14 For a free particle.

where R -> 0 as r —> oo. so that according to (5. and since. so that v <ti c and non-relativistic kinematics are justified. Kr -eKr) r<a. the solutions to this equation are e±KT where n — (2mE/h)1^2. 2014 . Eliminating A and B gives the equation that determines the energies of the bound states: /^(coth/^a — 1) = U.doi. The corresponding energy is \mv2 .18). Also (5. The most probable distance is where P(r) is largest.e2/4ne0a = -e2/8n£0a = me4/32TT2 as in (5. 5.R must be continuous. we take «r r>a. http://dx.020 Cambridge Books Online © Cambridge University Press.6) shows that m = .4 When the wave function is spherically symmetric.221.1 .5 Applying L2 in (5.6) 1 = 1. which is at r = 2CLQ. For r 7^ a. 5. we require R to be non-infinite at r = 0.1017/CBO9781139171243.14) yield \Mr 2M Aneor-EJ so that /^ = l/2ao. Since we want a bound-state solution. 5. According to (5. disc [rR(r)]f = —U[rR(r)] r=a (compare with (3. and £7 is as given in (5.244.org/10. = rih say.3) to ip multiplies it by 2. Atr = a. Hints for the problems 187 5. where ao is given in (5. it satisfies [rR(r)]" + ^[rR(r)] = -U6(r . the velocity is e2 /A-KSQU = c/137. the wave function depends on r only.17).17) with n = 2. so that Ae~Ka = B(e~Ka—eK'a).208 on Thu Feb 13 17:14:46 GMT 2014.35)).13) and (5. (5. for the reason explained in the text. When n = 1. from the differential equation. Also.13). So -KAe~Ka + KB(e-Ka + e«a) = -UAe~Ka.3 The force towards the centre of the orbit is mv2/a = e2/4neoa2. Downloaded from Cambridge Books Online by IP 150.a) [rR(r)]. So TiL = mva = (mae2/A-KSQ)1!2.6 When / = 0. with P(r) = Anr2 \ip(r)\2. the probability that the distance is between r and r + dr is P(r) dr.

nA{eKR T e~KR) = -UB e~KR.9). write r = x\ — X2 and R = \(x\ +£2). Hence Be~«R = A(eKR ± e~KR) and KR .h2 — a must be an integer N = n — I — 1. 5. when the centre of mass is at rest we have m dr2 and ip2(R) = 0 (see (5. where ao is the Bohr radius given in (5. the Schrodinger equation is if)" = K2ip. for the linear harmonic oscillator by making the changes m —> \m and u —> y/2cu. ip is continuous and disc if>' = —Uip. 2014 . http://dx.n = K tanh KR or K coth KR. the first of these equations is obtained from equation.15). Chapter 6 6. 11) acquires the extra term —me2b/2ireofi2 = — 26/ao. the substitution (A.8 By analogy with (5. This gives U « 26 MeV.1 For x 7^ ±i?. The required root of the indicial equation is a = I + 1 — D(l). imposing the appropriate continuity and discontinuity conditions at x = ±R guarantees that they are satisfied also at x = —R.org/10.7 An extra term —me2b/27reoh2r2 appears in the square brackets in (A. However. The denominator of the right-hand side of (A. but the numerator is still the same and so again me2/47reoK.doi.7) and (A. With such wave functions.27) and (5.1017/CBO9781139171243.20).18). The square bracket in the indicial equation (A. 5.8) is the same.244.188 Hints for the problems 5.208 on Thu Feb 13 17:14:46 GMT 2014. Downloaded from Cambridge Books Online by IP 150. At x = ±i?. (3. The potential is symmetric about x = 0 and so (see problem 3. 12) has an extra term.9 Equations (5.23)).020 Cambridge Books Online © Cambridge University Press.29) give 1 cot Ka = — — = — K which « 0 if \E\ < U and so Ka w TT/2.221. where E — — TI2K2/2m. with /ip{xi^X2) — / 0i(r) ^ ( i ? ) . So the energy eigenvalues are E = (n + \)THJO\/2. Then. Eliminating A and B gives U .6) the wave functions have definite parity: either tp(-x) = ij)(x) or tl>(—x) = -ip(x).

Elimi- nate c\ and c2 between the two equations (6. 7. so there is no first-order energy shift. In (7. where J2 = u2+2e/m. so the required value of R is proportional to ao.244. for large R we have Hn « EOi and H22 ~ E02. Jo J-i The cos# integration gives zero. The exact value of the new ground-state energy is \Tiu' = \hw + eTi/2mu) + higher-order terms. 2014 . Hints for the problems 6.4).org/10. write E = Ei + S where S is assumed small.18). H0(x) is a constant.208 on Thu Feb 13 17:14:46 GMT 2014.4 With the obvious notation.221. 6. Downloaded from Cambridge Books Online by IP 150.3 (i) Let V be an eigenfunction of the Hamiltonian H0 + Hi. According to (6. The resulting equation for E has two de- generate solutions E = (Hn — Hi2)/(l — Ki2) and a third solution E = (Hn+2H12)/(l + 2K12).1). the muon-bound molecule is therefore some 200 times smaller. The resulting equation for E has one solution near E\ and the other near E2.doi.020 Cambridge Books Online © Cambridge University Press. http://dx.1). the energy shift (7. The per- turbed Hamiltonian \sp2/2m+\mJ2x2. take Ho as the Hamil- tonian before the field is introduced and Hi = e£z. 7.2 The equilibrium value of R corresponds to the minimum of the integral V(R) given in (6. the symmetry of the problem results in •Hii — H22 = i?33 and H\2 = H2\ = H2% = H$2 = H\$ = H%\. Eliminate c\.2 r / a o / d(cos0) eSr cos 9.1 Take the z-axis in the direction of £. From (5.7) is AE=(na30)-1 f = (2/a30) r ° d r r 2 e .4). and insert this in the equation.10a). Write (j) — ci<f>\ + c2cf)2 + c^fo and obtain the equations correspond- ing to (6.12). Prom the Schrodinger equation. With the ground- state wave function (6. so the normalised unperturbed ground-state wave function is Vo(#) = (nh/mu)'1^e~mujx2l2h.3 As in (6. The energy shift to first order is AJ5 = J dx ex2ifj% = eTi/2mu).c2 and C3 by making the determinant of their coefficients vanish.2 Prom appendix A. Chapter 7 7.1). To examine the solution near E\. 6. This gives that (E2 — E{)5 is quadratic in small quantities.1017/CBO9781139171243. the only length scale in the problem is that set by ao. and write it as the linear combination cifa +c2cf)2.

190 Hints for the problems

- Ei)fa
To this equation, apply in turn the operations Jd3r<f>l and /d 3 r(/> 2 ,
using the orthonormality of fa and fa (see (4.18)). Eliminating c\ and
c2 then gives

(fa, Hifa)+ E (fa, Hi fa)
((/>2, Hi fa) (fa^Hifa) + E2 — E
This is a quadratic equation for E, with two solutions. Expanding the
square root in the solutions gives, to first order, the same result as
perturbation theory.
7.4 According to (3.4) and (3.5), the state of energy 3h27r2/2ma2 has wave
function

fan — /a) sin(7n//a) si

The perturbation —e£(t)x can induce a transition to the state whose
wave function is

^211 = (8/a3)1/2sin(27rx/a) sin(7ry/a) sin(7r^/a),

but not to either of the other two states of energy 3h27r2/ma2. The
transition probability is calculated from (7.22), with

t'-t'2/T2
-/-,

(see (B.236)).
7.5 The normalised wave functions are ipo(x) = (a/7r)1/4e~2aa;2 and
•0i (a;) = (27r) 1 /2( a / 7r )3/4 ;re -iax 2 ) w h e r e a _ muJ/fl T h e required
probability is


J —c

f
2 2 2
= (2a2v2\2/h2) dtte -a v t +iut
J —a

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

Hints for the problems

The integral is

(see (B.236)).
7.6 The probability is given by (7.31), with fa = (2/a)1/2 sm{nx/a) and
0'fc = (l/a) 1 / 2 sin(7rx/2a). The integration is from 0 to a, and the
answer is 32/9TT2.

Chapter 8
8.1
[Lx,Ly] = -^[VPz ~ ZPy, ZPx ~ Wz]
a
Z]px+X[z,pz]py}

8.2 Solve
-o C
C
1 0) \u2
giving U2 = cui and u\ = CU2- So the eigenvalues of ax are c = ±1, the
same as those of a z . The corresponding eigenstates are represented
by the normalised spinors

0
.3 With 5 = | a , the matrix representations (8.11a) give

s+ s
-{o oj' --{i o
and when these matrices multiply the spinors given in (8.11b) the
results given for S-u^ and S+u± follow. Also S-u^ = S+u^- = 0.
.4 The possible results of measuring Sz and Sz are t t , J 4 ? t i a n d
It, for which Sz has eigenvalues +1,-1,0 and 0 respectively. If the
eigenvalues of S2 are s(s + 1), then according to (8.5) the allowed
values of s axe | | — | | and ( | + | ) .

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

192 Hints for the problems

Write S2 = S2Z + \{S+S- + 5-5+) with 5± = 5 ^ + S^. Use

= 0, etc

to derive 5 2 x+ = 2x + (so that 5 = 1) and S2x~ = 0 (so that 5 = 0).
8.5 Choose the z-axis to be along £?, and the a:-axis so that B' is in the
xz-plane. Then the particle is initially described by the spinor

••(!

and the addition to the Hamiltonian caused by switching on B' is

HUB'

The energy shift is u^Hiu = —/jihB'/2y/2. The initial energy was
| , while the exact final energy is

l
2 2 2
-\\& \B + B'\ = -±nhB ((1 + B'/BV2) + (B'I By/2) ) ,

and the square root may be expanded in powers of B' jB by Taylor's
theorem.
8.6 Initially the spin state is represented by a spinor that satisfies ayu = u.
With the representation for cry

where the factor l/\/2 normalises the spinor to unity. The possible
results of the measurement are the eigenvalues of |cr x , namely ± | .
The corresponding spinors u± obey ayu± =± u± and are

The two probabilities are calculated from (4.22) and are each equal
to | . A further measurement of the y-component of the spin will sim-
ilarly give one of the two results ± | , each with probability | .

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

Hints for the problems 193

(Of course, usually the probabilities will not be equal; we have ob-
tained this result because we have chosen directions that are orthogo-
nal to each other.)

Chapter 9
9.1 Equation (5.17) gives Tiu = 3me4/327r2h3SQ. For a photon, K = u/c.
So, with (5.18), Ka0 = 3e2/8ne0hc = (3/2)(l/137). Hence if it is not
the case that Kr <^ 1, then r/ao » 1 and the exponential in the wave
function (5.18) is very small.
9.2 The stationary states ip± are not eigenstates of the electric-dipole-
moment operator /i, but we may calculate the expectation of JJL in
these states. Because the eigenstates </>i and </>2 correspond to different
eigenvalues ±/i 0 , they are orthogonal. Hence, using (9.16),

|(0i>£<£i) + |(02, £<fo) = 0,
and similarly for the -0- -state.
9.3 The answer is 4.6 x 109 K.
9.4 The uncertainty relation (4.13) implies that when the width of the
beam is known to be A, the photons can have a transverse momentum
component of at least h/2A. The total momentum of each photon
is calculated from the de Broglie relation (1.5), and is 2-KTI/X. SO
6 > (h/2A)(27rh/\) = A/4?rA radians.
In the case of the torch, 0 = \wj'/, where w is the width of the filament
and / is the focal length of the reflecting mirror. With / = 10 mm,
w has to be 10~6 m to achieve the same value of 9. It would not be
possible to construct a filament of this size with a reasonable light
output.

9.5 The average number of particles in the given state is

nqrs = Ce-E«rs/kBT,

where C = N(EQtR9se-E**s'k»T)-1 and from (3.5),

Eqrs = (h27r2/2mL2)(q2 + r2 + s2).

(The fact that the states in this problem are degenerate makes no
difference here; the degeneracy enters only if we calculate the number
of particles in a level rather than in a state, as in (9.24).) The average

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

194 Hints for the problems

fraction of particles that are in the state is nqrs/N, which is also the
probability that a given particle is in the state.
The velocity of the particle is v = p/m = hk/m = (hir/mL)(q,r,s).
As L -> oo, the allowed values of EQRS become arbitrarily close, and

r r i 2 i"1

C ~ TV UmL/hn)3 / d3ve"mv lk»T\

= N UmL/h7r)3(27rkBT/m)3/2j .
In this limit, the number of allowed sets of values of (g, r, s) correspond-
ing to the volume element d 3 i; of velocity space ~ {mL/fnr)3d3v. If
d3v is infinitesimal, so that all the corresponding states have essen-
tially the same energy, the average number of particles in these states
is C(mL/h7r)3d3ve~^mv2/kBT. Integrate over vy and vz to obtain the
number having vx in the desired interval.
Notice that the final answer is in fact valid for a large box of any
shape.
9.6 Take the x-direction normal to the membrane, and use the result of the
previous problem. A particle whose rr-component of velocity is u will
reach the wall containing the membrane during the time interval At if
it is within a distance uAt of the wall, and a fraction A/L2 of such
particles will hit the membrane region of the wall. Hence the number
of particles in time At that pass through the membrane is

N f du(m/2nkBT)^2e'mu^2kBT(uAt/L)(A/L2).
JuO

9.7 At T = 0, ni = 0 for E{ > ( and rc* = g{ for E{ < (. This is because
at T = 0 the energy of the system is minimised, but the exclusion
principle does not allow more than one electron per state.

Chapter 10
10.1 The Bloch waves here have Uk(x) in (10.3) equal to a multiple of
e2lrnx/L, where r is any integer, and V + E = h2(k + 2rn/L)2/2m.
The dispersion curves are shown in the figure below. In this model,
there are no gaps between the bands, but these appear when the
model is modified so as to become the nearly-free-electron model.

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

Hints for the problems 195

r=-3

r=-2

10.2 The three components of k satisfy k{ — 2n;7r/F 1 / 3 , where the
Ui are integers. Hence there are V/(2TT)3 allowed vectors per unit
volume of fc-space. Taking account of the two spin states of each
electron, the number of electron states having |fc| < kF is therefore
(|7rA;p)Vy47r3. At T = 0, the NZ electrons occupy all the states up
to kF = (3TT 2 JVZ/TO 1 / 3 -

10.3 (i) For 0 < x < L, the potential is zero and so I/J and Uk have the
forms shown in the second equations of (10.4) and (10.5), respectively.
For — L < x < 0, Uk is obtained from the second equation of (10.5)
by replacing x by (x + L). Impose the conditions that, at x = 0, if) is
continuous and disc xp' = —Uip (0) - see (3.35). This gives

cosh ah — (U/2a) sinhaL = cos kL.

Downloaded from Cambridge Books Online by IP 150.244.221.208 on Thu Feb 13 17:14:46 GMT 2014.
http://dx.doi.org/10.1017/CBO9781139171243.020
Cambridge Books Online © Cambridge University Press, 2014

the number of atoms.5 In the case of periodic boundary conditions. take rp and uk as before. 2014 .208 on Thu Feb 13 17:14:46 GMT 2014. the main contribution to the denominator comes from those terms having m — n. the allowed values of k are given by (10.nL)0(a.mL) m. so does uk(x). H<t>(x — mL) = Eo(f>(x — mL) + V^ Vo(x — rL)(j)(x — m l ) . so that the denominator is approximately equal to TV.13). However.4 where and where n)L f dx 0*(:r .doi.org/10. m. There are N pairs of such terms and together they give the desired result. the main contribution to the numerator comes from those terms having r = n = m ± 1.020 Cambridge Books Online © Cambridge University Press.nL) <f>{x . and are used to form Bloch waves ipk(x) = Downloaded from Cambridge Books Online by IP 150. Likewise. http://dx.244. 10. Because V(x) now has period 2L. and the second con- tributes [dx(/)*(x-nL)V0(x-rL)<f)(x-mL) J2 eik(m-n)L f dx(f)*(x . -H"^) contributes £Jo to £".n J Because the binding is tight and the atoms are not very close together.1017/CBO9781139171243. .ml). and in L < x < 2L take similar forms but with different coefficients.196 Hints for the problems (ii) In 0 < x < L. Impose the conditions on ip and disc rp' at x = L. 10. Eliminate the four coefficients from the four conditions.221. from the definition of <^. The first term in (?/>. Use this to obtain uk(x) in -L < x < 0 from uk(x) in L < x < 2L.n n)L fdx<l>*(x-nL)H<l>(x-mL). and impose the conditions on ip and disc ipf at x = 0.

The required wave functions are then ij)k{x) — ?/>_*. In the free-electron model.1017/CBO9781139171243. and it will also vanish at x = NL if k = nn/NL with n = 1. so that there are Aj2ix2 states per unit area. Taking into account the two spin states. there is one allowed wave vector in the interval 2ir/NL of fc-space. In the other case. this combination is chosen so as automatically to vanish at x = 0.org/10.221. apart from a sign. 2014 . and so p(E) = (NL/7r)/{h2k/m) = (NL/nh){m/2E)1/2. the number of allowed values of k is thus the same as in (10. Hints for the problems 197 eikxuk{x). because of their periodicity they are then also equal to 1 at x — NL. because the corresponding negative values would again give the same wave functions. So p(E) = (Ak/n)/(h2k/m). there are therefore NL/n states per unit interval of &-space. so here they will be combinations of ipk a n d ^-fc- Define the normalisation of Uk and u-k such that they are equal to 1 at x = 0. (ii) This time there is one allowed wave vector in the area 4TT2 /A of fc-space. Chapter 11 11. So p(E) = (Vk2/7r2)/(h2k/m). For |fc| between k and (k + dk) there are therefore (yl/27r2)27r/jdfc states having all the same energy. 11. In the interval (10. take the ends of the crystal to be x = 0 and x = NL. (iii) There are V/4TT3 states per unit volume.2 p(E)/(Am/nJi2) J I I II E/(nznz/2ma2) Downloaded from Cambridge Books Online by IP 150.1 (i) Prom (10. (#).14).2.020 Cambridge Books Online © Cambridge University Press.208 on Thu Feb 13 17:14:46 GMT 2014. E = h2k2/2m and dE/dk = h2k/m. http://dx..244..doi.. and so do not correspond to additional quantum-mechanical states.3. For |fc| between k and (k+dk) there are (V/A7r3)A7tk2dk states of the same energy.13).. We here restrict k to positive values.13). Just as for the free particle in a box the wave functions are sine functions.

3 From part (iii) of problem 11. 2014 . Downloaded from Cambridge Books Online by IP 150. so the number of electrons per unit volume of p-space is obtained by dividing by Anp2.208 on Thu Feb 13 17:14:46 GMT 2014. http://dx. The current is therefore / dpx / dpy / dpz J(2mVo)1/2 J-oo J-oo e e</B 2 2 2 !2± e-p x/2mkBT e-p y/2mkBT e-p z/2mkBT 3 3 m 4 ^ 11. and for these p2/2m — ( » &BT. SO that the exponential in the denominator dominates.3. But also AVEdE e-(E-O/kBT Equating the two expressions for n gives e^/fciaT + e ( A + 2 < ^/ f c B T = (2iV/^l)(A.Br). Since E = p2/2m.3 / 2 7r. the first term is neg- ligible and the result follows. The Fermi-Dirac distribution (11.244. this begins at E = 4h2n2/2ma2. while q = 1. Then J5 > h2n2/2ma2 and the contribution to p(i?) is Am/nTi2 as in part (ii) above. with E = h2(k2 + k2)/2m + h2q27T2/2ma2. ad- ditional contribution.21) and integrating around E = — A: " --N-U. For q = 2.^ f c B T <C 1. The number of electrons in the valence band is obtained by inserting this into the Fermi-Dirac distribution (11.21) gives the number of electrons n(E) in the unit volume per unit E. When e .1. The allowed values of k\ and k2 are as in part (ii) of the previous problem. p(E) = NS(E + A).1017/CBO9781139171243. e-(A+C)/kBT This results in the given expression for n. the number per unit p is obtained by multiplying this by dE/dp = p/m and is {p2 dp/n2h3) ^e(p2/2m-O/kBT + ^ -1 The momentum distribution is spherically symmetric in the free- electron model.4 Near the valence band.221. 11.1 / 2 .doi. there is a similar.org/10.198 Hints for the problems The stationary-state wave functions are e^klX+k2y>) s'mq(7rx/a).2. Only electrons having pi/2m > Vo can escape. for unit volume of the metal p(E) = (2m3E/n4h6)1/2.020 Cambridge Books Online © Cambridge University Press. — Take first q = 1. And so on.

020 Cambridge Books Online © Cambridge University Press. http://dx.1 The function Uk{x) determined in problem 10.2). With the even-parity solution. because of the Downloaded from Cambridge Books Online by IP 150. gives on eliminating C. and disc if)' = —U21P there. and making Vi(x) periodic. D and ekL (C/1C/2 + \Ul ~ 2a 2 ) t a n h a L = 2aUx.244. But the tunnelling of electrons in the other direction is greatly increased.^~ ^ . we find that in L < x < 2L we have i/>{x) = Cea^x~L^KX a L KX +De.doi. There is no negative-parity solution.1) and (12. so that its energy levels are shifted upwards relative to those in the p-type region. we have if)(x) — Ceax + De~ax. the numbers of elec- trons passing across the interface in each direction are equal. the flow of electrons from p to n is reduced. Writing rj> as in (12. When (b) the n-type material is connected to a negative potential. together with the two previous equations. Because the valence band of the p-type material overlaps the conduction band of the n-type region.221. the remainder is due to electrons passing directly from one conduction band to the other.3.org/10.3 may be used to con- struct the solution given in (12.208 on Thu Feb 13 17:14:46 GMT 2014. that ip is continuous and disc ipf = —C/iV>.2 Conduction band Valence band p-type n-type p-type n-type (a) (b) In the case (a) where there is no external potential. 12. one of the two corresponding conditions is satisfied automatically. gives two equations. as in the normal junction diode.1017/CBO9781139171243.4). The same level structure obtains as in problem 10. where E = —h2a2/2m. Consider now the trapped-electron levels. The other. Hints for the problems 199 Chapter 12 12. where the potential vanishes. The constants R and T are determined by making I/J continuous at x = 0. some of the electron flow is the result of tunnelling between the two. 2014 . In 0 < x < L. Imposing the conditions at x — L.

http://dx.org/10. 2014 . in which the tunnelling does not occur. The current is then greater than for a normal diode. the tunnelling ceases and the situation is the same as for the normal junction diode (see figure 12.200 Hints for the problems large number of empty electron states in the valence band of the n- region.5).221.208 on Thu Feb 13 17:14:46 GMT 2014. When the potential difference is in- creased to such a value that overlap of the p-type valence band with the n-type conduction band no longer occurs. Normal diode *V Tunnel diode Downloaded from Cambridge Books Online by IP 150.244.1017/CBO9781139171243.020 Cambridge Books Online © Cambridge University Press.doi.

139.org/ebook. 123 201 Downloaded from Cambridge Books Online by IP 150.cambridge. 137 alpha decay 31 of molecules 99 angular momentum colour of crystal 123. http://ebooks. 173 Compton effect 4. 115. 115 orbital 49. 16 of pn junction 146. 82. 79 black-body radiation 110 diffraction 5 Bloch's theorem 115.59 acceptor and donor 144 for delta-function potential 29 alkali 61 convolution theorem 171 diameter of 1 crystal hydrogen 54 boundary conditions 118. 141 antiparticle 7 continuity conditions 9. 75 magnetic 58. 26. 97 transitions 96 chemical potential 135 Dirac equation 88 coherence 107 dispersion relation 119-20. 144 spin 52. 144 atomic number 1 lattice 112. 119 atomic spectra 4 resistance 131 band overlap 117. 19. 149 deuterium 2 beam of particles 16. 129-30 collisions divalent metals 113.221. 88-9. 123. 82 conductivity 121.208 on Thu Feb 13 17:14:56 GMT 2014. 144 addition 83 commutator 45 algebra 82 completeness 37. 42. Index acceptor atom 144 of electrons with crystal lattice 113. 55. 26 deuteron 59. 2014 . 7 orbital 52-3 conduction band 122.244. 45. 68. 120. 114 accidental 56 consequences 121 delta function 162 number of levels 118-9 demodulator 154 of npn-junction 152 density of particles 13. 9 131 degeneracy 54 band structure 112. alkali 123 127. 17. 21. 150. 28. de Broglie's relation 6. 119-20. 132. 155 Bohr atom 55 light-emitting 146 Bohr radius 56 tunnel 157 bound state 12. 130. 59 dipole box potential 19 electric 96 with electric field 72. 179 diode 146.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. atom 36. 125 many-electron 56 colour 123. 22.

41.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. 157 collisions with crystal lattice 127. 152.cambridge. http://ebooks. 159 eigenvalue 9. 143-5 Hamiltonian 10. 136 132. 27 hydrogen 63 forbidden transitions 75 momentum operator 9. 42-3. 121. 133. 46. Maxwell-Boltzmann distribution 104. 144. semiconductor 146. 141. 43 Hermitian operator 41. 141. 2014 .202 Index divergence theorem 15 Hamilton's equations 86. 77 field-effect transistor 153 metals 120. holes 122. 46 Fourier transform 38 MOSFET 153 free-electron model 124. 146-8. 135 in diode 147-48. motion of electron in 128 hydrogen atom 54-5. 88 Jacobi identity 47 trapped 142 junction transistor 152 velocity in crystal 127. 176 doping 123. 83-4 golden rule 96 normalisation condition 13 good quantum number 49 npn transistor 152 Downloaded from Cambridge Books Online by IP 150. 139 neutrino 79 gauge condition 87. 140. 82.221. 134 isotope 2 motion in electric field 128 spin 83. 150. 176 neutron 1. 123-4. 136 fine-structure constant 60 molecule 67 flip-flop 157 ammonia 98 flux of particles 16. 122-23 laser exclusion principle 56. 150 134 Legendre's equation 51 expectation value 16. 134. 117.244. 44 electric current 130. 130. 137 integrated circuit 153 diffraction of 5 inversion layer 154 momentum in crystal 127. 44 light-emitting diode 146 Lorentz force 87 Fermi energy 111. 134. 79. 137 152. 154 holography 107 electric field. 137. insulator 121-3. 53 collision with crystal lattice 113 impurities 132. 2. 59. 154 magnetic quantum number 52 in transistor 153 maser 100 Fermi-Dirac distribution 134. 144 135 Fermi-Dirac distribution 111 measurements 38.org/ebook. 46. 176 molecule 63 electron 1. 45-6. 160 electromagnetic field 175 in magnetic field 87 electromagnetic interaction 86. 129 sudden change 77 eigenfunction 9 harmonic oscillator 23.208 on Thu Feb 13 17:14:56 GMT 2014. 41 classical 86 effective mass 132 for two-body system 57 Ehrenfest's theorem 40. semiconductor 145 energy gap 112. 144. 86. 129-30 junction. 158 eigenstate 43-4 heavy water 2 common 45 Hermite polynomial 25.

52 quantum number orthogonality 42 good 49 orthonormal set 42 magnetic 52 orbital-angular-momentum 52 pair excitation 137. 13. resonant absorption 91. 2014 . 28. 137 selection rules 97. 84 periodic 114 Pauli matrix 84 spherical well 59. 35 radiation pressure 4 Parseval's theorem 172 reduced mass 57 particle density 13 resistance of crystal 131 Pauli resonance imaging 91 exclusion principle 56.208 on Thu Feb 13 17:14:56 GMT 2014. Index 203 nuclear force 1 spherically-syymetric 53 nucleus 1 power-series solutions for differential equations 158 observables: basic assumptions 10. 99. classical limit 2 82. 93. 52. 144 matrix 84 perturbation theory Schrodinger equation time-dependent 73 time-dependent 11 time-independent 70 time-independent 10 phonon 106. 52-3. 95. 41 principal quantum number 55 operator probability energy 10 of result of measurement 43 Hermitian 41 of transition 75 Hermitian conjugate 48 probability density 13 linear 36 in momentum space 39 momentum 9 in plane wave 13 orbital-angular-momentum 173 in stationary state 16 translation 178 proton 1. 102. 134 population inversion 105-6 special relativity 4 positron 7 spectrum 88 potential spherical polar coordinates 50 barrier 26. 136. 89 optical properties 121 optical pumping 105 quantum mechanics op tics.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press. 95 134 ruby 105. 37 semiconductor junction 145 pnp transistor 152 semimetals 120-1. 30.221. 132. 106 photoelectric cell 137 semiconductor 122-3. http://ebooks. 36. photoelectric effect 3 140-1 photon 3. 98 delta-function 60 z-component 56. 41-2 orbital angular momentum 49.non-linear 107 basic assumptions 10. 82. 24. 91 Downloaded from Cambridge Books Online by IP 150. 141 105 intrinsic 122. doped 123. 56. 121. 82. 147 photovoltaic effect 146 n-type and p-type 144 plane wave 11. 83. 146 parity 22. 53. 173 statistical nature 6 ^-component 50. 25. 77 spin 52. 136-7. 61 precession 89. 6. 137. 83. 130.244.org/ebook. 123. 133-4.cambridge.

2014 . 17. 102 unstable system 77 superposition 20 of stationary states 36 velocity of electron in crystal 127. 140 wave function tight-binding model 124 transistor Bloch 116.129. 31. 134 generalised 45.118. 132 trapped electron 142 triode 152 Downloaded from Cambridge Books Online by IP 150.124. 34 infinite 12. 130 thermionic emission 113.126.jsf?bid=CBO9781139171243 Cambridge Books Online © Cambridge University Press.128-9.204 Index spin-orbit interaction 58-9 tritium. 93 Zeeman effect 87 transition probability 75 zero-point energy 25.org/ebook. 30.cambridge. 28. http://ebooks. 40. 65. 30 99. 95. 49 for energy 76 statistical mechanics 104. field effect 153 179 npn 152 wave packet 37. 39. 106 tunnel effect 26. 100. 45 stationary state 11. 129 pnp 152 wave vector 8 transition WKB formula 30 dipole 96 work function 3 radiative 88. 113 square well 21.221. decay of 79 spinor 85 tunnel diode 157 spontaneous emission 93. 63. 48 stimulated emission 93. 19 uncertainty principle 25.244.120. 111. 77. 32.208 on Thu Feb 13 17:14:56 GMT 2014. square barrier 26. 38.

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.