Elmer E. Anderson - Modern Physics and Quantum Mechanics-W. B. Saunders Company (1971)

MODERN
PHYSIOS
AND
QUANTUM
MEOHANIOS:
ELMER E. ANDERSON, Ph.D.
Chairman, D"port*ent of Physics,
Clnrkson College of Technobgy
Potsdarn, New York
W. B. SAUNDERS COMPANY PHILADELPHIA . LONDON . TORONTO

1971
W. B. Saunders Company: West Washington Square
Philadelphia, Pa. 19105
l2 Dyott Street
London, WCIA lDB
1835 Yonge Street
Toronto 7. Ontario
Modern Physics and Quantum Mechanics sBN 0-7216-1220-2
e D7l by W. B. Saunders Company. Copyright under the International Copyright Union. All
rights reseived. This book is protected by copyright. No part of it may be reproduced, stored in a
relrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording. or otherwise, without written permission from the publisher. Made in the United States
of America. Press of W. B. Saunders Company. Library of Congress catalog card number
77-145554.
Print No.:
ffiffiffiffiffiffiffi
This book grew out of my experience in teaching a one-year course in

quantum physics given to our seniors in the physics curriculum. It is designed
to bridge the gap between the descriptive course at the sophomore level and a
graduate course in quantum mechanics in which formai operator methods are
used freely. I have written this text because I have been unable to find a
book which suits our needs, although there ate a number of excellent texts
which are satisfactory for certain parts of the course.
Since many diverse areas ofmodern physics have developed simultaneously,
a strictly chronological presentation would not necessarily be the most logical
to follow. Therefore, I have chosen to develop the ideas that I consider
necessary for an understanding of quantum physics without regard to historical
order and with no attempt at complete coverage of all of twentieth century
physics. The use of the word "Modern" in the title is not intended to suggest
the same list of topics as is included in the sophomore course bearing that name.
It is assumed, in fact, that the student has previously taken such a course at
the sophomore level, as well as previous or concurrent courses in classical
mechanics, electromagnetic waves, and thermodynamics. Much of the material
of the first four chapters should not be new to the student but is included for
the sake of logical completeness in the development of quantum mechanics.
Four great conceptual edifices form the principal themes for the first
four chapters. These are: (l ) the special theory of relativity and some ex-
amples of its role in contemporary physics ; (2) the quantum theory of radiation
and the concept of the photon as a wave packet which can display both particle
and wave properties; (3) the Bohr theory of the atom and the quantization
of angular momentum; and (4) the representation of a particle by a wave
packet and the development of a wave mechanics for the description of a
particle's dynamical states. In Chapter 5, Schrddinger's wave mechanics is
applied to one-dimensional problems. In that chapter the linear harmonic
oscillator is solved by both the series and the operator methods in order to
familiarize the student with both of these approaches. This is followed in
Chapter 6 by a summary of the postulates of quantum mechanics and the
development of matrix mechanics. The angular momentum operators and
their eigenfunctions are introduced in Chapter 7 where they are applied to
ilt
tv PREFACE
the hvdrogen atom. The complexities resulting from spin and its interactions
u-ith orbital angular momenta are treated in Chapter B. The phenomenon
of exchange degeneracy resulting from the symmetry properties of identical
particles is also discussed in this chapter. Chapter 9 is an important chapter.
Here some approximate methods are introduced which enable the reader to
solrre large classes of real problems by means of the solutions obtained for the
ideal systems in earlier chapters. In Chapter 10, approximate methods are
applied to some specific atomic problems, and Chapter 11 is an introduction
to quantum mechanical scattering theory.
The book contains over 250 problems in addition to the numerical ex-
amples which are worked at appropriate points in the text. The importance
of solving a large number of problems in order to achieve a mastery of the
material cannot be overemphasized.
It is a pleasure to acknowledge the help of the following people and to
thank them for their indispensable roles in the preparation of this book: my
students over the past few years for their enthusiasm and inspiration; Agatha
Hollister and Amelia Anderson for typing and preparing the manuscript;
Rita Arajs for making all of the drawings; my colleagues, Professors Fred
Otter and Sigurds Arajs, as well as Professors Henri Amar of Temple University,
Paul H. Cutler of the Pennsylvania State University, Thomas B. W. Kirk of
Harvard University, John Reading of Northeastern University, John G.
Teasdale of San Diego State College, and an anonymous reviewer, for their
most valuable critical comments; to John J. Hanley and others of the staff
of the W. B. Saunders Co. for their assistance and encouragement; last of all,
to my mother and to my family-Amy, Kenneth, Mark, Scott, Ruth, and
Carl-to whom I dedicate this book.
Potsdam, Ir{ew York E. E. A.
wffiffiffiffiffiffiffi
Sometimes the over-dramatization of the concepts of modern physics

tends to give the erroneous impression that success is achieved in the physical
sciences in proportion to one's rejection of what was previously regarded as
"intuitively obvious." The truth is, however, that the great natural philosophers
in the modern era have not rejected the obvious but have forced us to redefine
the obvious after taking a closer look at the evidence. The transition from
lgth to 20th century physics is replete with illustrations of redefinitions of such
concepts as space, time, causality, measurement, waves, particles, and so on.
Admittedly, there have been some bold postulates made by modern physicists,
but these generally arose out of the compelling urge either to incorporate new
experimental facts into known theory or to make existing theories consistent
with each other. It is hoped that the student of modern physics will learn to
appreciate its conceptual structure as well as its application to specific physical
problems.
CONTENTS
Chopter 2
APPENDICES A To C 65
Appendix A. Calculation of Jeans' Number b5

Appendix B. The Maxwell-Boltzmann Distribution Function 67
Appendix C. The Maxwell Velocity Distribution Function . . . . 70
Suggested References . . 72
Chopter 3
,
no
THE CONCEPT OF THE NUCIEAR ATOM lc
l. The Atomic Models of Thomson and Rutherford . tc
2. Classical Scattering Cross Sections . lc
3. Scattering Cross Sections in the C and L Frames .80
4. Bohr's Theory of Atomic Spectra ,83
5. The Franck-Hertz Experiment . . .88
6. X-ray Spectra and the Bohr Theory .89
7. Nuclear Structure and Spectroscopy o9
8. Fundamental Forces and Exchange Particles .99
9. Angular Momenta and Magnetic Vloments l0r
10. Tlre Larmor Theorem and the Normal Zeeman Effect . . . 104
I l. Spatial Quantization . . l. . . 107
ll0
ll0
Chopter 4
THE DEVETOPMENT OF WAVE ,\,IECHANICS llr

l. Introduction . lll
2. The Old Quantum Theory: Wilson-Sommerfeld Quantization
Rules tt2
3. Sommerfeld's Relativistic Theory of the Hydrogen Atom t14
4. The Wave Nature of Particles . . 120
5. The Difiraction of Particles . . . 12t
6. The Wave Function for an Electron . . t26
7. The Fourier Integral and the Delta Function 130
8. Particles as Wave Packets 135
9. The Schrcidinger Wave Equation t4r
10. The Conservation of Probability Density . t43
I l. Observables, Operators, and Expectation Values 146
12. Separation of Space and Time in the Schrodinger Equation:
Energy Eigenvalues 150
13. Dirac Bracket Notation r54
Summary r54
CHAPTER I
Wffiffi ffiffiffiffiffiffiffi WWffiffiffiffiW

ffiW ffiffifuffiWffiWffiWW
The special theory of relativity is much more than a conceptual revolution

of interest only to philosophers of science. The consequences of the postulates
of the special theory of relativity to dynamical systems and to the interactions
between matter and energy are so far-reaching that scarcely any area of
contemporary physics is free of them, and the relativistic equations now play
a significant role in the research activities of many scientists. Therefore, an
understanding of the special theory of relativity is extremely important for
the student of quantum physics, whether or not he ever encounters the formalism
of relativistic quantum mechanics.
I. CLASSICAL PRINCIPLE OF RELATIVITY:

GALILEAN TRANSFORMATION
The theory of relativity is concerned with the way in which observers who
are in a state of relative motion describe physical phenomena. The idea of
an absolute state of rest or motion has long been abandoned, since an observer
"at rest" in an earth-bound laboratory is sharing the motion of the earth
about its axis, the earth about the sun, the solar system through the Milky
Wuy, and so on. It is also common knowledge that one can perform simple
experiments with bouncing balls, oscillating springs, or swinging pendula in a
laboratory fixed on the earth or in a smoothly running truck moving at
constant velocity and obtain identical results in both sets of experiments.*
We describe this fact by saying that the laws of motion are coaatiamt, that is
they retain the same form when expressed in the coordinates of either frame of
reference.
Consider two frames having relative translat'ional velocity V in their
common r-directions. It is evident that the coordinates are related by
t(':x-Vt
)' :)
nl-n
/-
tt:t,
* This is true only for experiments that are insensitive to the rotation of the earth.
THE SPECIAL THEORY OF RELATIVITY
zz'
Figure l-l Two inertial frames of reference.
provided that t : t' -- 0 when the origins coincide. Differentiating with

respect to time we find,
uL:u*-V
u'o : uo
ttL : u,
where ua :
dxldt, and so forth. This again checks with our every day experience
with velocities. Differentiating once more we have,
aL:a*
a'u : a,
al" : a",
where 4, : du*ldt, and so forth. Then,

F'--md':md:F. (1.1)
Equation 1.1 says that Newton's laws have the same form in both coordinate
frames, that is, they are covariant. Note that the coordinates and velocities
are different in the two frames but each observer always knows how to obtain
the other's values. Frames in which Newton's laws are covariant are called
inertial frames. Inertial frames are equiualent in the sense that there is no
mechanical experiment which can distinguish whether either frame is at rest
or in uniform motion; hence, tltere is no preferredframe. This is known as the
Galilean or classical principle of relatiuiQ and the coordinate transformation given
above is called a Galilean transformation. Strictly speaking, the earth is not an*
inertial frame because of its rotation and its orbital motion, but it can often
be treated as an inertial frame without serious error.
As a consequence of the principle of relativity, observers in different
inertial frames would all discover the same set of mechanical laws, namely,
\ewton's laws of motion. By way of illustration, consider the following thought
experiment. An observer, riding on a flat car which is moving at constant
velocity past an observer on the ground, fires a ball vertically upward from a
small cannon and notes its maximum height as well as the total time of flight
fior it to return to the muzzle of the cannon. The observer on the ground also
ELECTROMAGNETIC THEORY, GALILEAN TRANSFORMATION
measures the maximum height and the time of flight. Although the two
observers will not agree on the shape of the trajectoiy of the ball, they wilt
agree on the height, the time, the calculated muzzle velocity, and the value
of g, the gravitational acceleration.
2. ELECTROMAGNETIC THEORY AND THE

GALILEAN TRANSFORMATION
Long after the Galilean principle of relativity was well established,

Maxwell formulated his electromagnetic field equations, out of which arose a
finite and constant velocity for light in vacuum. In spite of the great success
of Maxwell's equations in describing the behavior of electromagnetic phenom-
ena in space and time, it was quite disturbing to find that they are not co-
variant under a Galilean transformation. Thus, the classical transformation
which permits a covariant description of mechanical forces, does not hold for
electromagnetic forces.
PROBI.EM ,jliilii:ilia1jLii,:ii.
'-'
Show that the equation V .i : p is not covariant under
a Galilean transformation. That is, assume the covariance
of the Lorentz force, F : q(& + I o x h\ and show that
this leads to a contradiction. Hint: require p' : pi
otherwise, a simple measurement of charge would distinguish
one frame from another.
Another serious difficulty arose from electromagnetic theory, in connection

with a medium which was postulated to sustain the wave motion. This ethereal
substance was simply called "the ether," and was endowed with infinite
elasticity and inertia but zero mass. These contradictory properties were
required in order to sustain the transverse vibrations of lighf *irr.r and yet
prevent any longitudinal vibrations. Furthermore, the ether had to pervade
all of space without inhibiting the movements of celestial bodies, since it was
known that these bodies move in a non-viscous medium.
It was hoped that the ether might provide the necessary reference frame
for measuring absolute motion. However, these hopes were destined to be
unfulfilled, as the following summary will show:
a. As early as I 728 it was known that the observed position of a star varies
slightly throughout the year.r'2 This phenomenon, which is called the aberration
of starlight, is due to the fact that the observed direction of a light ray depends
upon the relative velocity of the light source and the observer. Since the earth
moves in its orbit at about 30 km/sec, a telescope should be tilted towards a
star at an angle 0 ,'.-, 301300,000 : I x 10-4 radian or 20 seconds of arc. Six
1J. Bradley, Phil. Trans.35,637 (1728).
2A. B. Stewart, "The Discovery of Stellar Aberrationr"
Sci.entifc American, p. 100 (March 1964).
True
Positiorlgp-
l7
I
t Apptrrent
Position
Figure l-2 The angular position of a star is

w shifted by the amou nt 0 because the telescope moves
with the earth during the time the light travels
down the tu be. ln the course of a year, the telescope
sweeps out a cone of abberation of angular diameter
20.
!u A,t
months later, the telescope should be tilted the same amount in the opposite
direction. See Figure 1-2. An important conclusion can be drawn from the
fact that aberration of the correct amount is experimentally observed, namely,
that the earth mlaes through the ether. The ether, then, was assumed fixed with
respect to the stars.
b. A number of experiments were devised to measure the earth's motion
through the ether, that is, to detect an ether wind with respect to the earth.
The most famous of these was the Michelson-Morley experiments which will
be described below. This experiment was repeated many times but always
with a null result, that is, no ether wind was detected. Another famous but
unsuccessful attempt to detect an ether wind was the experiment of Trouton
and Noble. a They sought to detect a torque on a charged capacitor due to its
motion through the ether.
c. All attempts to explain the absence of an ether wind while retaining
the ether concept were also refuted. It could not be argued that the earth
carries the ether along with it since this would contradict the observed aber-
ration of starlight. Some proposed that the velocity of light adds vectorially
to the velocity of the source. This theory (historically called the emission
theory) would easily account for the null result of the Michelson-Morley
experiment, but it contradicts a known fact about wave motion, namely,
that the velocity of a wave depends only upon the properties of the medium.
Perhaps the most convincing evidence for rejecting the emission theory comes
from the study of the periods of binary stars as proposed by De Sitter. s The
relative velocity (to an earth observer) of a distdnt star performing circular
motion will vary sinusoidally with time. Therefore, the measured Doppler
shift* of the light emitted by such a star would show the same sinusoidal
3 A. A. Michelson, Amer. Journ. of Science (3),22t 20 (lBBl) ; A. A. Michelson and E. W. Morley,
ibid.34r 333 (lBB7). These results were confirmed by R. J. Kennedy, Proc. Nat. Acad.l2r 62l (1926)
and K. K. Illingworth, Phys. Reu.3Or 692 (1927).
4 F. T. Trouton and H. R. Noble, Phil. Trans. Ro1. Soc., A2O2r l65 (1903) ; Proc. Ro2. Soc. (London),
72, t32 (1903).
5 W. De Sitter, Proc. Amsterdam Acad.161 395 (1913).
* The mathematical details of the Doppler shift, which will be discussed in section 11 of this
chapter, are not required for the present argument.
ELECTROMAGNETTC THEORY, GAL|LEAN TRANSFORMATTON
variation if the speed of light were constant. On the other hand, if the emission
theory were true, the speed of the emitted light would be greater as the star
approached and less as the star receded. In this event the measured Doppler
shifts would no longer show a sinusoidal variation but the curve would be
distorted as in Figure 1-3(a). An actual plot of the Doppler shifts measured
for the components of the binary star Castor C is shown in Figure l-3(b).d
The absence of distortion in the curves is strong evidence lor the validity of
the constancy of the speed of light. Many such binaries have been studied
and there is no indication of distortion other than that due to known eccentric-
ities in the orbits. More recent experiments which refute the emission theory
of light are measurements of the velocity of light in the form of gamma rays
emitted from positron annihilation? and from the decay of zo mesons.s
A third attempt to explain the null result of the Michelson-Morley
experiment was the Lorentz-FitzGerald contraction theory. It was postulated
that a length is contracted by a factor (l - r'lt\ l, where a is the component
of the velocity parallel to the length and c is the velocity of light. This con-
traction is just the right amount to explain the null Michelson-Morley result,
although it was not derived from first principles. The Lorentz-FitzGerald
.n
60
a.
+
r20
(,
c)
vE o
Prirnrrrv
tio
0.4 0.8
Time-days
b
Figure l-3 The measured Doppler shifts due to the orbital velocity of astar performing circular
motion would appear as in (a) if the speed of light were ro increase as the star approached and to de-
crease as the star receded from the observer. Shown in (b) are the relative velocities of the components
of the binary star Castor C as determined from the Doppler shifts of their spectra. The absence
of distortion is strong evidence for the constancy of the speed of light. (The data are from A. H. Joy
and R. F. Sanford, Astrophysical lournol 64,250 (1926).)
6 A. H. Joy and R. F. Sanford, Astrophls. J. 64, 250 (1926).

?D. Sadeh, Phys. Reu. Letters lO,271 (1963).
8 T. Alviiger, F.
J. M. Farley, J. Kjellman and I. Wallin, Ph2s. Letters 12r 260 (1964).
6 THE SPECIAL THEORY OF RELATIVITY
contraction was extremely difficutt to refute,e and its physical significance did
not become clear until after the work of Einstein'
3. THE MICHELSON-MORLEY EXPERIMENT
As mentioned, in the previous section, if the ether is assumed to be at rest

in the frame of the "fixe8," stars, then the earth's motion through the ether
should result in an ether wind for an earth observer. Michelsonl0 proposed
to detect this wind by looking at the interference pattern produced by two
coherent beams of light, wheie the optical paths of the two beams are re-
spectively parallel to"and perpendicuiar-to the earth's motion through the
ether. If the two beam, uri initially in phase and if their optical paths diFt
by zero or an integral number of wavelengths, they wilt still be in ph_us9 when
they are brought"together again. In general, the optical.paths of the two
beams will ditrir by ui arbitralry amount, and either destructive or constructive
interference can occur. The arrangement used by Michelson is shown in
Figure l-4. One of the mirrors of the interferometer is usually tilted slightly so
thit a pattern of alternately bright and dark lines, called fringes, is viewed
in the eyepiece. Each dark'fringe corresponds to a path difference of an odd
number ofhulf *uvelengths in the two arms of the instrument. Slightvariations
in either path prod.rce"cl while observing thlgugh the eyepie.ce will result in
an apparent motion of the fringe pattern. Shifts corresponding to a fraction
of a wavelength can be easily detected.
Suppose" the interferometer is se t uP so that one light Pu!h, say
/r, is
colineai with the earth's motion through the ether with velocity a. Tl.tt,
assuming a Galilean transformation, the time for one round trip along that
path is
t_
/r- * lz : ?tt, :21, (r t ,-,4c (I + pr),
,^ - r62)-1
b2*-
c _aTfia:-p -az - c
where 0 : ulc. Since the velocity of the earth in its orbit is about 30 kilometers
per second,'F 10-a and terms high.. than pz are neglected in the expansion.
^For -
the puit, that is perpendiculir to the relative motion the situation is
analogous to the problim-of rowing a boat across a river having a current u.
If the rower's rp.id is c, his path is actually of length

ct:{M,
where / is the time for one transit. Then, his velocity for a direct crossing
is
lrlt :1/rz rz, and the time for a round trip transverse to the current is
-
Kennedy and E. M. Thorndike, Plryts. 400 (1932)'

'R.J. \a'-!2t
1o Se! footnote 3, and R. S. Shankland, Amer. J. PhTs'32116 (1964)'
THE MICHELSON.MORLEY EXPERIMENT
1,
------+
_ ] Cornpensator
Half-silvered surface
Source
To eyepiece
Figure l-4 Michelson interferometer with arm f, parallel to the earth's motion.
The phase difference between the two beams as they enter the eyepiece is
Ad, : QrrlD.(pathdifference) : (2nl)').c(tr-tr):(2n11).12(/,
- fr) +
0'(2/, - {t)1. By rotating the interferometer through 90" the roles of the two
beams are interchanged and the phase difference for this case is
L6, :a 2n
l2(/,- /,) + p,(2/,- ()1.
Thus, the effect of rotating the instrument through 90'is to enable the observer
to detect a total phase difference of A{, -l L6r. This corresponds to a fringe
shift of
2"
t
(Ad' f Ad,) :#,t, * /r). (1.2)
For lt
- fz :10 meters and ,i ^,5000 A, Equation 1.2 predicts a shift of
about 0.4 fringe. Such a shift could be easily detected, as Michelson's instru-
ment could resolve 0.01 of a fringe at that wavelength.
The Michelson-Morley experiment was repeated many times, at all hours
of the day and in all portions of the earth's orbit, yet no fringe shift has ever
been observed. The null result requires us either to accept the non-physical
Lorentz-FitzGerald contraction or to reject the ether theory, since the ether
cannot be fixed with respect to the earth as well as the distant stars.
THE SPECIAL THEORY OF RELATIYITY
PROBLEM 1"2
Show that the Lorentz-FitzGerald contraction can account

for the null result of the Michelson-Morley experiment.
4. THE POSTULATES OF SPECIAL RELATIVffY

The failure of such efforts as the Michelson-Morley and the Trouton-
Noble experiments to discover a preferred frame for the electromagnetic
equations of Maxwell suggested that the latter must conform to a principle
of relativity. However, the fact that the Galilean principle of relativity-
which was known to be valid for classical mechanics-failed for Maxwell's
equations was a considerable source of frustration to physicists at the turn
of the century. After a critical examination of the concepts of space, time,
and simultaneity, Einsteinll discarded the Galilean principle of relativity and
postulated in its stead a principle of relativity for all physical laws. His
postulates may be stated as follows:
(l) There is no preferred or absolute inertial system. That is, all inertial
frames are equivalent for the description of all physical laws.
(2) The speed of light in vacuum is the same for all observers who are in
uniform, rectilinear, relative motion and is independent of the motion of the
source. Its free space value is the universal constant c given by Maxwell's
equations.
Since the first postulate says that Maxwell's equations as well as Newton's
laws are covariant, a new set of relativistic transformation equations must be
obtained to replace those of Section l. The second postulate follows from the
first, since, if an observer were to measure, for example, the value c - u for
the speed of light, he would then have a means of determining the relative
speed a, which would violate the principle of relativity.
Maxwell's equations enable one to express the universal constant c in
terms of the fundamental constants of electromagnetic theory. Thus, in
m.k.s. units, c : (rolto)-*, and in c.g.s. units, c is given by the ratio of the
electrostatic unit of charge (esu) to the absolute unit (emu). The value of a
is very nearly 3 X 108 meters per second. Frequent attempts have been made
to determine c to as many significant figures as possible. Some of the methods
employed have involved direct measurements of distance and transmission
timl; others have been concerned with precise measurements of A and v in
order to calcul ate c from their product, and still others have utilized inter-
ferometry. There is a large amount of literature dealing with this experimental
problem and a number of good summaries exist which are replete with
ieferences.12,13,14 At the present time the accepted value for the speed of light
is
c :299,792.5 + 0.4 km/sec.
11A. Einstein, Ann. Physik 17' 891 (1905).
t, F. Mulligan, Am-. J. Ph2s.20,165 (1952).
J.
ta j. F. Mulligan and D. F. McDonald, Am. J. Phys.25' 180 (1957).
trJ. H. Sandirs, Thc Velocity of Light, New York: Pergamon Press (1965).
THE LORENTZ TRANSFORMATION 9
This value was adopted by the International Union of Geodesy and Geophysics
in 1957.14
Before deriving the new relativistic transformation, let us look briefly at
the concept of time. According to classical physics, t : t', and it was tacitly
assumed that clocks in different inertial frames could be easily synchronized.
The concept of simultaneity was not questioned. Einstein, however, realized
that the process of synchronization requires the transmission of signals at a finite
velocity which can be at most the velocity of light in free space. Furthermore,
simultaneity is a meaningful concept only in a frame in which light sources,
clocks and observers are all at rest. By way of illustration, suppose observer O
has measured two equal distances tr1 and x, from his reference position and
installs identical flash lamps ,S, and ,S, as shown below:
OfO -'+v
Stlt- xr X2 -45,
o
He can fire these flash lamps simultaneously by means of matched cables or
radio operated relays, and he will assert that the flashes reach him simultane-
ously. However, to a second observer O', who is in a state of relative motion
parallel to the line joinins Sr and Sr, the flashes will not appear to be simultane-
ous because of the finite distance through which he moves during the transit
time of the light waves. Thus, if the flashes occur just as O'passes O, then O'
will claim that S, flashed earlier than Sr.
Since simultaneity cannot be defined independently of spatial coordinates
for observers who are in relative motion, their clocks are not synchronized and
their measured times are different. As a consequence of this, taken together
with the constancy of the observed speed of light in either frame, we conclude
that the statement t' : / is no longer valid.
5. THE LORENTZ TRANSFORMATION

Although the Galilean transformation is not the correct expression of the
principle of relativity, the new transformation should reduce to the Galilean
transformation for small relative velocities, since the latter gives correct results
fior Newton's laws in our macroscopic world. Also, the transformation must
be linear in order to make the intervals between events independent of the
choice of origin in the space-time coordinate system. The simplest trans-
formation that will satisfy both of these requirements will have the form
s'-C(x -ut))
)' :) l (1.3)
z'-7, )
where C --- I as u ' 0, and where u is parallel to the common direction of the
x and "r' axes. Since the transformed time must, in general, depend upon both
IO THE SPECIAL THEORY OF RELATIYITY
time and space coordinates, the simplest form to try is,

t':At*Bx, (1.4)
where A -* I and B -- 0 as a --+ 0.

Consider two obseryers in different inertial frames having relative velocity u
along their common r-directions. At the instant that their origins coincide
a light is flashed by the unprimed observer and a spherical wave emanates
from his origin, the equation of which is
cztz __ xz I )z I zz.
However, by Einstein's postulates, the primed observer also must see a spherical
wave emanating at speed c from iis origin as its center. If he were not to see a
spherical wave but, say, an ellipsoidal wavefront, he would have a means of
determining which frame were moving with velocity u. The equation of the
wave front in the primed system is,
C2t'2 - X'2 *)'2 * Z',
Then we have,
,c2 +y + 22 - c2t2: x'' *)'2 i z'2 - czt',L (1.5)
If the new transformation satisfies Equation 1.5, it will be consistent with

Einstein's postulates.
PROBTE/U l-3 :,1.;jii.ri;!liil,iffi;tir:'i;.i::,;.'i,1l,:,'i iii;ji.iiljirli#
By means of Equations 1.3, L.4, and 1.5, determine the

Lorentz transformation,
y':y(x-Bct)
)' :)
ql-a
L (1.6)
ct,:y(ct-px),
where
f :ulc and y:(l-p\-t.
PROBLEM
'1
Show that the wave equation,
Y'6(*,), z, t) ::r# 6@,), z, t),
is covariant under a Lorentz transformation.
6. RELATIVISTIC KINEMATICS.
We define an event as a physical occurrence that is localized in space and
time. That is, to an observer who is at rest in the frame of the event, the time
RELATIYISTIC KINEMATICS II
interval of the event can be measured by means of a single clock and the position
of the event can be determined by means of a meter stick. We call the time
interval in the rest frame the proper time interual, rr.
If the event consists of the measurement of the length of an object in its
rest frame, the coordinates of the two ends of the object can be measured,
simultaneously by an observer in that frame. We call the length of the object
measured in this fashion its proper length, Lo. To an observei in any ineitial
frame other than that of the event, a measurement of either a length or a time
interval will require the use of both a meter stick and at least oni clock. We
will now look in detail at the way lengths and time intervals transform under
a Lorentz transformation.
a. Preservation of lengths perpendicular to the direction of relative motion.
Consider the following thought experiment. The lengths of two meter sticks
are known to be exactly the same in a frame where they are both at rest.
If they are now placed in relative motion in a direction perpendicular to their
lengths, as in Figure 1-5, what do observers O and O' conclude about their
lengths ? O can install synchronized flash cameras at each end of his meter
stick which he can trigger just as O' passes him. O' will agree that the flash
Pictures were simultaneous since he is at all times equidistant from the two
flash lamps. Since both observers agree that the flashes were simultaneous,
!h.y must agree with the result shown on the photographs. That is, the res.rli
is an absolute one, not a relative one. Now the experiment can be repeated
with the flash cameras in the primed frame and once again both observers will
agree that the flashes were simultaneous. The same result must be obtained
in both sets of experiments. If, for example, the stick in the unprimed frame
were found to be shorter in the first set of pictures, then the stiik in the un-
primed frame must be found to be shorter in the second set of pictures. How-
ever, such a result would contradict the principle of relativity since in that
case motion to the right would stretch a stick and motion to the left would shrink
it. Therefore, we conclude that the lengths of the two sticks must remain
unchanged. This result is consistent with the symmetrical conditions of the
experiments.
b. Contraction of lengths parallel to the direction of relative motion. Let
the two meter sticks of Figure 1-5 now be placed in relative motion parallel
to their common .r-axes and the photographs repeated as before. The first set
Figure 1.5 Length measurements per-

pendicular to the line of motion.
l2 THE SPECIAL THEORY OF RELATIVITY
w: (u) (b)
Figure l-6 Length measurements parallel to the line of motion. ln (a) the photographs are
simultaneous to observer O in S; in (b) they are simultaneous to observer O'in S'.
of photographs, taken by O, would look like Figure 1-6(a), and the second,
taken by O', would look like Figure 1-6(b). In the first, O would say that the
"moving" stick was contracted. O'would explain the photographs by claiming
that the flashes fired by O were zol simultaneous. In fact, he would say that
the leading end of his meter stick was photographed before it reached the coin-
cidence position and the lagging end of his meter stick was photographed after
it had passed the position of coincidence. Thus, both observers accept the
photographs but since they do not agree that the flashes were simultaneous
they have different explanations for them. In the second experiment the roles
are reversed and now obsewer O' says that the meter stick in .S is contracted
by virtue of its "motion." We cannot argue, as we did in the previous thought
experiment, that the sticks should appear equal in length since there is n0 agree-
ment on simultaneit2 in the present case. We must conclude that the stick that
is moving with respect to the observer is contracted.
This contraction may be demonstrated analytically as follows. Suppose
observer O' has a meter stick situated so that the end points are at xl and x'2
for any time l' in his frame. By the Lorentz transformation, an observer in ,S
will obtain the result
xi,:y(xr-Bctr)
xL:y(xr-|ttr),
from which we obtain
x', - x'r: y(xz - rr) - FyrU, tr).
But xi - xl : Lo, the proper length, since the stick is at rest in S '. Further-
more, if r, and xL ate measured simultaneously in S, then tcz - x1 : Z, the
length of the stick as observed in .S. Thus, for l, : tz,
Ls:yL or L:41.-PzLo. (1 .i)
The same result can be obtained by using the inverse transformation, but
the algebra requires a few more steps. To illustrate this we write,
xL: y@l + fcti)

xz:y@L+ fctl)
x2 - xL: y@L - ni) + FvtQL - ti).
RELATIYISTIC KI N EMATICS
But the time transformations must be used here since the simultaneous measure-
ment of r, and x, in frame S will not appear simultaneous in frame S'. Then,
'^',i
bz : y(ctr
- fx')
from which
t
=YQh
- frr),
c(ti - ti) : yc(tz
- tr) - fy(r, - xr).
Putting \ : tz and elineinating (tL
- t{) from the above equations,
xz - xr: y(xl *'r)
- - fryr(r, - xt),
which reduces immediately to Equation 1.7.

This result is identical with the Lorentz-FitzGerald contraction, but it
now has a theoretical basis in the postulates of Einstein. Note that the proper
length is the maximum length that can be measured by any observer.
c. Time dilation. Suppose an observer in S measures a proper time
interval zo by a single clock and that the coordinates of the evenhhaiking the
start and the finish of the interval are (xo, /r) and (ro,tr). An observer in the
inertial frame S' will observe the coordinates of the events as follows:
cti:TQtr-Fro)
ct'r:y(ctz-fro)
c(ti - t) : yc(tz - tr) - fy(ro - ro).
Letting t; - ti: z ?nd t, tt : rot

-
r : Tro or ro: 1/L Bz (l.B)
".
As in the case of Equation I .7, the inverse transformations can be used to derive
Equation l.B, but additional algebraic steps would be required to eliminate ri
and x'2. Note that the proper time interval is the minimum time interval that
any observer can measure between two events. This phenomenon is called
time dilation and provides the basis for the statement that "moving clocks run
slow." If the two events are two "ticks" of a clock, Equation 1.8 says that the
minimum interval between ticks will occur for the clock in the rest frame of the
observer, that is, the proper clock will run fastest.
It is possible to perform thought experiments in order to compare measure-
ments of time intervals by clocks which are in relative motion. In the course
of such experiments it is assumed that any number of clocks in the same rest
frame as an observer can be synchronized to any desired degree of precision.
It is also assumed that clocks in different inertial frames can be photographed
or compared visually from either frame at the instant that the clocks pass the
same point in space. Consider the following example. Let observer O'send a
light flash a distance / to a mirror at rest in his frame, S', at the instant that
he arrives at the origin of the unprimed frame where observer O is stationed.
O' sees the light reflected directly back to his position and he records the
proper time interval for the round trip by means of a single clock. The proper
time interval is zo :2/lc. fn the unprimed frame the light beam is rehecled
t4 THE SPECIAL THEORY OF RELATIYITY
Figure 1.7 Comparison of time intervals measured in two frames.
from a point to the right of observ er Oras shown in Figure l-7, since the mirror
is moving to the right with velocity u. The time at which the beam returns is
measured by observer ,B on a clock which had previously been synchronized
with the one used by observer O. If the elapsed time in the unprimed frame
is z, then from the figure we have,
(!cr)z : /2+ (*r")'

-2-
a-
ffr * (fi')'
(1 - 0'),' : 2
T6
T : TTo,
as in Equation 1.8. Observer O thus concludes that the clock used by O' runs
slow. Had O' photographed the clocks at O and at.B as he passed those points,
his photos would confirm the time interval reported by O. However, O' would
claim that the clocks used by O and B were not synchronized but that the
clock used by B was fast by the amount p2z. Therefore, according to O', the
time interval in the unprimed frame should be
, - 0'r: (1 - p')" : (l - P')Tro: lI - P'"0,

and O'would also conclude that "moving clocks run slow."
Before concluding this section, mention should be made of the visual
appearance of objects traveling at relativistic speeds, since misconceptions are
prevalent in the literature. A visual image formed at the retina or a camera
film is produced by photons that have arriued simultaneously, but they did not
necessarily leaue the object simultaneously. Thus, photographs of high speed
objects must be corrected for the transit time of the photons to the various
points of the image in order to determine the effects of the Lorentz contraction.ls
15 See V. T. Weisskopf, Physics Todq 13124(1960); also, G. D. Scott and M. R. Viner, Amer. J.
Pftrys.33,534 (1965).
RELATIYISTIC DYNAMICS t5
PROEIEU 1.5
A meter stick lies at rest in the xy-plane at an angle of 45"

with the r-axis. What are the length and orientation of the
stick to an observer moving at relative velocity Fc in the
r-direction ?
PROBIE'H
'{
If the galaxy has a diameter of 106 light-years, what will the
diameter appear to be to a cosmic ray particle traveling at
the relative speed 0 : 0.99 ? How long will the trip take
as measured by a clock riding with the particle?
PROBLEM
'.f
What synchronization error will an observer moving at
relative speed Bc detect in two stationary clocks separated
by r distance Z along the path of the motion ?
7. RELATIVISTIC DYNAMICS
By means of the Lorentz transformation given in Equations 1.6 we can
readily obtain expressions for the velocity components which each observer
will measure in his own frame. Thus, first taking the differentials,
dx' : y(dx - pc dt)
d)' :d)
dz' -dz
dt' : nr(r,
/ \-- -l*),
we define,
ul
dx' ur- fc
dt' t Urf
I ---
c
: d)' : u'
u!, (l.e)
dt'
,(-ry)
uL: dz'
uz
dt'
,F-ry)
Note that p, the relative velocity of the frames, is in the common r-direction
in Equations 1.9. The numerators of these expressions are just the velocity
-a
transformations as given by the Galilean principle of relativity; these are the

transformations that are normally used in the everyday world. The denomina-
tors represent the correction due to the special theory of relativity; they all
approach unity as P gets small. These equations can be used to obtain the
relative velocity of one body with respect to the other when two bodies are
moving with respect to the laboratory frame. For example, suppose two
electrons are fired in opposite directions, each with speed V with respect to
the laboratory. Let thg S frame be the rest frame of the electron moving in
the negative r-direction, and the ,S' frame be the rest frame of the laboratory.
Then, in the S frame the laboratory is moving to the right with velocity pc : V;
in the ,S ' frame, uL : Z is the velocity of the other electron. Since we seek z'
we must use the inverse transformation,
ua:
uL+pc V+V
, 7v-''
r -T--"Lp r |
(1.10)
cc"
If -, -
V is large, such as 0.9c, then u,: (1.80/1.81)c, which is very close to, but
still less than, the velocity of light in free space. An observer in the laboratory
will assert that the electrons are separating at a relative velocity of 1.80c.
This is not in conflict with the theory of relativity, since neither electron is
moving at a speed greater than c in any frame of reference. If the particles
in this example were photons fired by two lasers, Equation l.l0 tells us that
relative to each photon, the other would have a velocity of
2c
Ur:T:t,
which is the result required by Einstein's second postulate.
In the rest frame of an electron, another electron approaches

at a speed u. What is the relative velocity of an observer
who measures equal and opposite velocities for the electrons ?
PROBLEM 1.9
A cosmic ray mu meson moves vertically through the atmos-

phere at a relative speed of 0 : 0.99. Its mean lifetime in
its own rest frame is 2.22 microseconds.
(u) What is the mean lifetime measured by an observer
on earth?
(b) What mean path length will be "seen" by the meson
and by the earth observer ?
PROBLEM I.IO
A stream of electrons has velocity components V* :0.9a,

Vu :0.3c, V":0, with respect to the rest frame of the
ellctron gun from which the electrons were ejected. Find
the velocity components and the total velocity of the
RELATIYISTIC DYNAMICS 17
electrons with respect to an observer approaching the

electron gun at relative velocity f : 0.8 along the r-axis.
PROBLEM T.IT
Consider three inertial frames such that S and ,S' have

relative speed Brc along their common r-directions and ,S'
and S " have relative speed B2c along their common x-direc-
tions. Find the expression for transforming velocities from S
to S". Express Bc, the relative speed of S" with respect to
^S,
in terms of Brc and Brc.
PROBLEM T-T2
(u) Derive the result expressed by Equation I .7 by allowing

one observer to measure how long it takes for a fixed
length in a moving frame to pass him.
(b) Show that Equation l.B can be obtained if an observer
in S measures the coordinates of two events which
occur at the same point ri in S'.
PROBLEM I.'3
A radioactive atom is moving with respect to the laboratory

at a speed of 0.3c in the "r-direction. If it emits an electron
having a speed of O.Bc in the rest frame of the atom, find the
velocity of the electron with respect to a laboratory observer
when:
(u) the electron is ejected in the r-direction, (b) it is
ejected in the negative r-direction, and (c) it is ejected
in theT-direction.
(Ans.: 0.89c; 0.659c; 0.Bl9c at 68.5".)
PROBLEM T.I1
An observer in S places a laser so that the beam makes an

angle 0 with his r-axis. What angle does an observer in ,S'
see if the relative velocity between ,S and S' is Bc in their
common r-directions ? (This is the relativistic equation for
the aberration of starlight.) What does the -S' obserrre, see
whencos0 <p?
The velocities defined by Equations 1.9 were obtained by differentiating

with respect to the local time in each observer's frame. It is convenient to
define the proper uelocitlt by differentiating with respect to the proper time,
that is,
dx dt
uor:Y:
dto dt dro
(1.r1)
where u,2 : uI + ui + zl is the square of the local velocity.

The other components follow in like fashion. The advantage of using the
proper velocity will be evident when we discuss four-vectors in the next section.
Now consider the following thought experiment. Two identical volley
balls, having rest mass mo when measured in the same rest frame, are given to
two observers in different inertial frames with relative velocity pc in their
common x-directions. Each observer is instructed to throw his volley ball so
that it will make a head-on collision with the other volley ball at the instant
that the line joining the two observers is perpendicular to their common tr-axes
(see Figure 1.8). Each observer claims that his own volley ball moves parallel
to his own j,-axis both before and after the collision and that it undergoes a
momentum change of 2moun. Moreover, each claims that the other volley ball
has an .r-component of velocity before and after the collision. Since Newton's
laws are covariant and momentum is conserved, either observer can write,
:
2moun (2mrur)' : 2*[r{4 u* Provided u, 4c.
Then, letting m[ : m, we may write
ffi : Tt|to, (1. 12)
which is the expression for the relativistic change of mass. That is, a body
of rest mass mo has its mass increased by the factor 7 when it moves at a
S'
lv'
I
I ------ B"
I
,L\
I
\ \
Bc <-
./
---..
(a) (b)
Figure l-8 The collision of two identical volley balls: (a) as observed in S, (b) as observed in S'.
The y-axes of S and S'are colinear at the instant of impact.
RELATIYISTIC DYNAMICS t9
velocity a relative to an observer. After squaring Equation 1.12, it can be

written as
mzcz _ mLuz : mtcr,
Differentiating with respect to both m and u, we obtain
or
2mcz dm
- 2muz dm - 2m2u du : 0,
c2 dm - t12 dm * mu du. (1.13)

Also,
;+ dF d , +\:* dn -dm
,:ll:n\mu) ati, dt,
and the change of kinetic energy is
dT : F.as : *fr.d6 + t : mu Qu I uz dm, (1.14)

frd.m
where the last expression holds for dd parallel to ri. From (1.13) and (1.14)
we have
dT : c2 dm,
which can be integrated as follows:
Jo:o
f d,T : f* ,, d*.
J*o
Then,
T: (m
- mo)c, : (y - l)mocL, ( l.l5)
which says that the kinetic energy of a body is equal to the product of its mass
change times the square of the velocity of tight. This resuli can be compared
with the classical expression for the kinetic energy by writing
T_ ,/4lfrocz
- ntocz : moc'l(l - p')-+
- 1]
: mocz(l * tp' +' ' . - 1) (for small B)

:mocz(tp'+...)
,-.,, $m6az.
Thus, if B is small enough to permit discarding all terms higher than Br,
Equation 1.15 reduces to the classical result.
Still more information can be gleaned from Equation 1.15 by writing it
in the form,
E-mc2:T+ffi0c2, (1.16)
where we now call E the total energy and mocz the rest mass energy. Equations
l.l5 and 1.16 express the equivalence of mass and energy and provide the
basis for restating the conseryation laws of classical mechinics. N.ith"r mass
nor the quantity ( Z + Z) need be conserved separately in relativity*; instead,

the quantity mcz is conserved.
The ordinary momentum is
: rffioi'
and its magnitude may u. *ri,l::'
. mczu
:-
Eu:- PE (l . l7)
F:niu
'czcZc
Using Equations 1.16 and I .L7, we find that
p') -- (*ot')'
E',(r -
"'[' - eY)
: (mocz)z
or
- (pt), I (mrc2)2.
E2 (1.18)
Further,
g +mocz)r: (pt)'+ (*ot\
and
(Pt)': Tz {2Tmrc,. (1.1e)
Equations 1.18 and l.19 provide useful relationships between the momentum
of a body and its energy.
It is worth pointing out that the kinetic energy of a body, according to
Equation 1.15, will increase rapidly as the velocity increases. In the limit as
u --+ ct 7 would become infinite and hence the kinetic energy would also be in-
finite. We conclude that c must represent a limiting velocity that cannot be
achieved by any particle that has a rest mass. Conversely, we also conclude
that photons and neutrinos must have zero rest mass since they are known
to travel with speed c. Another way of stating this is to say that only those
entities for which a rest frame exists can have a rest mass. Since an entity
traveling at the speed of light has no rest frame (its speed will be c to all ob-
servers), it can have no rest mass. A zero rest mass particle canrhowever, have
a momentum and kinetic energy. From Equation 1.18 we see that the
momentum and kinetic energy are p - Elc and T : Pc - E, respectively.
Find the velocity and the momentum of an electron whose

kinetic energy equals its rest mass energy.
PROBLEM I.'6
Through what potential difference must an electron "fa11"

in order to achieve a speed of 0.95c? What is the mass of the
* A potential energy V may be added to each side of Equation 1.15, and then Equation l.16
becomes E: tncz : T+ V * mocz.
FOUR.VECTOR NOTATION 2I
electron in the Z-frame ? What is its momentum ? How

could you verify experimentally that it actually has that
mass and speed ?
PROBLEM l-rl ili:r;;;''1i
Derive an expression for the radius of the orbit of a particle

of mass mo and charge q moving with kinetic energy T at a
relativistic velocity in a transverse magnetic field of intensity
B. Express your answer in terms of q, B, c, T, and mo.
8. FOUR.VECTOR NOTATION
In ordinary three-space the components of a vect or i in two coordinate
systems S and S' with a common origin are related by:
A| : r4r cos (r't, rr) * A, cos (.ri, xr) * ,43 cos (x't, xt)
AL : z{., cos (xL, xt) * Arcos (ri, *r) * r43 cos (rL, r") (1.20)
AL :,4, cos (xL, xt) * A, cos (.ri, xr) * ls cos (*'", r").
These equations may be summarized by Ai: i cnr,A* - cr,Ar,, where the

summation over the repeated index is understood-iT rfr. last expression. One
can readily show that for a simple rotation through an angle 0 in two dimensions
this reduces to the form,
- rcos 0 +) sin
tc' B
]' : -x sin 0 +) cos 0, (1.21)
which can be written in matrix notation as follows:
(:) : ( j*"" :1 ;) fi)

: ( in"?j::J,)
In general
(;):(;ll ;il0)
The cni are not all independent since",::
they are direction cosines and must
satisfy the relations
: f0, for i+f
citcit, 6or:
tr, for j : k
and (r.22)
det cri - l.
22 THE SPECIAL THEORY OF RELATIYITY
Show that the relations in Equations 1.22 must hold for a

Cartesian coordinate system.
In addition to their transformation properties, another important property

of vectors is the invariance of lengths. That is,
(Ar), - (Ar)2 + (Ay), + (Lz)r: (Ar')s + (Ll')' +(Lz')2 : constant.
We have seen that in special relativity we require the Lorentz invariance of

the quantity
(Ar), : (A,r), + (Ay) , + (Lz)z - 6z(Lt)2.
Therefore, if we can treat these quantities as the four components of a vector
in four-space and the Lorentz transforrhations as a rotation of coordinates in
four-spaie, then the desired invariance properties are automatically satisfied.
a. The position four-vector. The position four-vector is easily constructed
by adding to the three-vector components the fourth component, ict. Thus,
/x,\ /'\
Xu:l::l:{ll (1 23)
\i:/ \;,1
When the index is a Greek letter, a four dimensional space is implied; a
Roman letter will represent a three dimensional vector. With this convention
AX,, AXp : (As)z, as given above.
The matrix of the Lorentz transformation is given by
in terms of which the transformation

(1, l:'l)
becomes,
Q24,
X| : fuJn or X' : fX, (1.25)
in either tensor or matrix notation. We will show that Equation 1.25 is

equivalent to Equation 1.6 by writing the full matrices, that is:
6:(r,,l :',l)(3
FOUR-VECTOR NOTATION 23
By means of matrix multiplication we obtain the four equations:

Xi:T(Xr*ifXn)
Xi:X,
Xi:Xt
Xi:y(Xn-ifXr)
substituting the components from Equation 1.29, we obtain
tc'-y(x-pct)
)'.:)
p.7'- o
ct,:y(ct_px),
which was to be shown.
PROBLEM I'tg ::iirl:':.::
Show that the Lorentz transformation is equivalent to a

rotation in four-space, that is, that the transformation is
similar to Equation 1.21 with the circular functions replaced
by hyperbolic functions. Hint: I-Jse the x,
- xqplane as in
the accompanying figure and derive the expression, 0
arc ban (ip).
-
PROBLEM I-20
Show that the Lorentz transformation from the primed to

the unprimed system may be expressed by a matrix equation
of the form
x - f-1X"
where f-l is the matrix inverse of f given by
r-1 _
00
10 'r)
(,[, 01
00
b. The velocity four-vector. The components of the four-velocity are

obtained by differentiating the position four-vector components with respect
to the proper time. This was done for one component in Equation 1.11 and was
defined as the proper velocity. Thus, the first three components of the velocity
four-vector are the components of the proper velocity, that is to say, the local
velocity multiplied by
y:ffi,
,T
Vt -"
where z2 is the square of the local velocity. The fourth component is
rla: {,x,: .#,

#rfu'q
Then the four-velocity is
U,:(t (1.26)
ffi)
and its transformation is given by Equation 1.25 as UL: Tp,u".
c. The momenfum four-vector. The four-momentum is formed simply
by multiplying the four-velocity by the rest mass. Then,
P:(l,W) (r.27)
W:(T)
The first three components reduce to the three-momentum components and
the fourth is the total energy multiplied by ilc.
1r..
I-'l ::tuu::i .,.....,,'.. .r1:::: :" :.
PROBLEM I.2T .,.1
Find the scalar invariants associated with the four-velocity

and the four-momentum.
d. The force four-vector.* The components of the four-force are obtained

by differentiating the four-momentum with respect to the proper time. For
* This is frequently called the Minkowski force after the man who first formulated the four-vector
notation.
FOUR.VECTOR NOTATION E
examPle,
Ft: I ,dVrt:T
r, ,dP,
7t: T,FN'
and
FE: ,'#rrr: y':# : y'! F'a'
Thus, the first three components of the four-force are the three-vector force
components multiplied by y'.
F': (1.28)
(,1'!'')
e. The current density four-vector. If po is the charge density in a frame
where the charges are at rest, then a current density four-vector may be defined
as the product of po and the four-velocity. That is,
Ju: PN,r:
(l!';,h:(T;) (1.2e)
fi,)
Here we have defined p : y'po, illustrating the increase in charge density
due to the change in volume element. Then the product of local charge density
and local velocity gives the ordinary three-space current density, (i*, ju,i,).
f. The electromagnetic potential four-vector. Since a charge distribution
can be at rest in only one frame, it will be observed as a current accompanied
by.u magnetic field by all other observers. Therefore, we would not expect
the components of either electric or magnetic fields to become components
of the four-vector. Using the vector and scalar potentials defined by
d:i xA and e:-V6 -#,

we define the potential four-vector as
Ar:h) (1.30)
where the Ao are the components of the three-space vector potential.

g. The electromagnetic field four-tensor. It is often desirable to work with
the components of the electric and magnetic fields themselves rather than with
charges, currents, or potentials. However, in order to express the electromag-

netic fields in four-space we require a second-rank tensor rather than a
vector. The construction of this tensor is generally done in a course in
electromagnetic theory so it will merely be given here16:
0 g, 8*
-go C
0 g*
-9, -!c
a
6a
(l .31)
go
-g* 0
-,_ 8,
C
3mx
8. - 6ot 8"
c'c
PROBLEM 1.22
By means of the transformation TLu : I uuoJ oo, find the

Lorentz transformation equations for the six components of
the electric and magnetic fields. Show that these may be
expressed by the following four equations:
eir: gr +
: Et
8'u
g'L:y(Etao x6)'
CIt:
'(*':#)
where "parallel" and "perpendicular" refer to the line of
Q.g2)
relative motion.
PROBLEM I-23
A meter stick is at rest and at equilibrium under the action

of the forces F as shown.
(u) What angle, 0', will the meter stick aPpear to make
rG See, for example, W. K. H. Panofsky and M. Phillips, Classical Electricit2 and Magnetism' Addison-
rA-esley Publishing Co., Inc., Reading, Mass. 1955.
RELATIVISTIC COLLISIONS
with the x-axis in a frame moving in the positive x-

direction at relativistic speed Bc?
(b) What angle will the forces make with the meter stick
in the moving frame ?
(") Calculate the torque in the moving frame.
h. The proper time as a four-scalar. Using the symbol (Ar)2 for an interval
in three-space, we may express the length of an interval in four-space as
AXp AXp : (Ar)z : (Ar)z - 6z(Lt), - invariant.
Now, when two events are observed by the same clock, Lr :0 and A, : Azo,
the proper time interval. Therefore, in any inertial frame the square of the
interval is
(At)' - -cz (A"o)'.
Since this is a scalar quantity in all frames, Azo is called a four-scalar. We
can show that the proper time is the minimum time interval by rearranging
the above expression as follows,
(A?'.
(Ar)r: (Azs)z +
when (A"o)' ) 0, Azo is real and,,, ;; ur r;*rt)ure'e interual.In this case one can
always find an inertial frame traveling at a velocity less than c such that the
two events occur at the same point in space but at different times. Two events
separated by u time-like interval can never be regarded as simultaneous by
any observer. When (A"o)' { 0, Azo is imaginar2 and is called a space-like interual.
In such a case one can always find an inertial frame traveling at a velocity
less than c such that the two events occur simultaneously but at different
points in space. No observer can eliminate the space interval between the
events regardless of his state of relative motion. The imaginary proper time
interval corresponds to the transit time for a tight pulse between the two
spatial points.
At this point it is worthwhile to summ arize the principal features of the
four-vector formulation. First, all four-vectors are Lorentz covariant, that is,
they transform like Equation 1.25. Second, the "length" of a four-vector is a
scalar that is invariant in any inertial frame, as, indeed, is any scalar product
of two four-vectors. Third, the first three components of a four-vector form a
vector in three-space, but not all three-vector components are components of a
four-vector (for example, the three-velocity). Fourth, unlike a three-vector,
a four-vector can have zero ma.gnitude without all of its components being zero.
9. RELATIVISTIC COLLISIONS
In the majority of experiments involving relativistic velocities, the ob-
server is at rest in the laboratory and the events under study are taking place
between particles which, for the most part, are moving at high velocities with
respect to the laboratory. The physical nature of these events, however,
depends upon the amount of energy available to do work in the zero-momentum
frame of the particles. (The zero-momentum frame is the center-of-mass frame
of classical mechanics.) If two particles approach each other and exchange
momentum and energy we call the event a scattering event. If the particles
are not the same in number or in kind after the collision we say that a reaction
has occurred. In either case, it is often convenient to do the theoretical analysis
in the zero-momentum frame and then transform the results to the laboratory
frame in order to check against experiments.
a. The zero-momentum frame (C-frame). We will designate the total
energy, total kinetic energy and total momentum in the zero-momentum frame
by E*, T*,and,P*, respectively. Of course, F* - 0, by definition. Consider
two identical particles of rest mass mo, one particle having momentum F, and
kinetic energy To and the other being at rest in the laboratory (Z-frame).
If we call p* the relative velocity of the C-frame, the Lorentz transformation
may be written,
0 0 i|*y*
r0
(il:( _'i ,
0
01 0
00 y*
(il
where Pt: Po, Pz: Ps :0, and Pa : iElc : (ilt)(E" + E) : (ilt)(To *
2moc2).
PI:T*(Pr*if*Pn)
Pt:P,
Pt:P"
Pf :/*(Pn-i|*Pr).
Substituting for P, and Pnit follows that:
p*_ r*(r"-ry-o
and
E*:y*(E-f*tP").
From the invariance of the length of the momentum four-vector we have,
E*z-Ez-Plcr.
From the first of these three equations we see that the velocity of the C-frame
is given by
F* :ry : t' 33)

RELATIVISTIC COLLISIONS 29
Putting Poc : B*E into the second or third equation above,

E* : \/l - P*'z E. (1.34)
From the third equation we can easily obtain an expression for the kinetic
energy of each particle in the C-frame in terms of the laboratory kinetic energy
of the incident particle, f. Thus,
E*2-(8"+mocz)z-Plcz
: (E: - (*ot\
PStr) ! 2Eomoc2 +
: (mocz), + 2Eomocz * (*ot\
:2mocz(Eo * mocz)
:2moc2(7, + 2mocz). (1.35)
Since the energy is equally divided between the two particles in the C-frame,
the energy of either particle is
Etr : fI + mocz : iE*,

and the kinetic energy of each particle is
Tt:+E* -ffiocz.
EXAMPLE
In the rest frame of an electron, a positron is o\served coming directly

toward the electron with momentum given by lc : 4 MeV. Find the relative
velocity of the C-frame, the total energy in the C-frame, and the kinetic energy
of each particle in the C-frame.
Each particle has a rest mass energy of mocz:0.51 MeV. Then the
energies of the positron and the electron in the L-frame are:
Eo:\/@W - \/4, + |o.l)Mev :4.03N{ev

Eo : fttscz :0.51 MeV
and
E- Eo I Eu :4.54 MeV.
The relative velocity of the C-frame is given by Equation 1.33:
f* :H: o.BB,
Then the total energy in the C-frame is
E* : 1/L B*z E :2.14 MeV,

and
Tt : Tf :1.07 - 0.51 : 0.56 MeV.
Figure l-9 Symmetric collision in the l-frame.
d'
b. Elastic collision of two particles of equal mass. Consider two un-

charged particles of rest mass mo such that particle a is moving toward particle
D, which is at rest in the Z-frame. In order to keep the algebra simple let us
treat the symmetric collision, after which the momenta of the two masses make
equal angles with the direction of the incident particle in the Z-frame. Primes
will be used to denote values after the collision. . Using the conservation of
momentum, llsin 0 : p; sin 0, or lL : p;. Then, Po -- p'"cos 0 + P; cos 0 :
2p'" cos 0. It follows that
lE-(Jnrc"y
cos 0
-Po
2PL
- 2\/Ere' (1.36)
From conservation of energy we have, E" + mscz : EL + EL : 2E'o, since the

particles share equally the momentum and the energy. Then Equation 1.36
becomes
cosg: lW Eo I mocz
-VIE" + 3rnoc'z' (1.37)
V tf', * mocz)z
- (2msc2)2
In the classical limit the kinetic energy of the incident particle is small compared
to its rest mass, and Eo ffi0c2. Then cos 0 : {T12, and the angle between
-
the paths of the particles is 90'. This result holds even for asymmetric scattering
in the classical limit as long as the masses are equal. As the incident energy
increases so that Eo t ffi0c2, the scattering angle gets smaller and both particles
tend to go off in the forward direction.
c. Inelastic collisions. If kinetic energy is not conserved during a
collision we call that collision inelastic, and the kinetic energy that disappears
is converted to mass or potential energy. By way of illustration consider the
following example: A particle of rest mass mo moving with velocity ptc collides
inelastically with and sticks to a stationary particle of rest mass Mo. Find the
velocity of the composite particle. The momentum in the L-frame is
P :7tmo0{,
and the total energy in the Z-frame is
E:Tfltocz*Mocz.
RELAT|VISTIC COLLISIONS 3t
Then, from Equation 1.33, if we call B* the relative velocity of the C-frame,
D*,: pc Ttmo|{z
P' : : Ttmo|r
E r*ot, a 74t, r*, * 4'
and p*c is the velocity of the composite particle in the l,-frame. All of the
kinetic energy that existed in the C-frame before the collision has been converted
to mass, since there is no kinetic energy in the C-frame after the collision.
PROBLEM 1.24
What is the rest mass of the composite particle in the above

example after the collision ?
PROBLEM 1.25
A I GeV proton is moving toward an alpha particle which

is at rest in the L-frame.
(u) What is the velocity of the proton in the Z-frame ?
(b) What is the velocity of the C-frame (zero momentum
frame) relative to the L-frame?
(.) What is the total kinetic energy in the C-frame before
the collision ?
(Ans.: 0.875c; 0.298c; 740 MeV.)
d. The Q equation for inelastic scattering. The change of kinetic energy

in the C-frame is often designated by the symbol 8, that is,
Q:T'a -T*'
where the prime denotes a value after the collision. Since the change in kinetic
energy is numerically equal to the change in rest mass energy in the C-frame
we can write,
Q : r'* - T*: (; *,), - (; *o)', (1 .38)
For Q - 0, the collision is elastic; for I > 0, the rest mass decreases, the
kinetic energy in the C-frame increases, and the reaction is exoergic; for
Q < 0, the rest mass increases, the kinetic energy decreases and the reaction
is endoergic. Equation 1.38 is most useful when Q is expressed in terms of
the energy of the incident particle and its angular deflection by the collision.
Let us designate the incident particle as particle a with rest mass m and kinetic
energy To in the l-frame. The target particle D has a mass M and is initially
at rest in the Z-frame. Then from conservation of momentum we have
Po:Po +F;
P'o' : PZ + PL' - 2!'P'ocos o,
where 0 is the angle through which particle a is scattered in the Z-frame.

Conservation of energy in the L-frame requires that
Eo-l-Eo:EL+E;
To * (M + m)cz : fL + m'cz * f{, + M'cz.
Then, Equation l.3B becomes:
Q : W +m - M' - m')cz : TL + T; -- T* (r.3e)
Since this equation gets quite complicated for relativistic velocities it will be
derived here for the classical case instead. It has been shownlT that the correct
relativistic expression can be obtained from the classical result by substituting
for each rest mass mo the value mr I (Tcl2C), where To is the value in the
Z-frame. Then, using the classical expression for the kinetic energy, Equation
1.39 becomes:
bL' _t- P;' PI , I

a-- :-2m' I
-P'":!L'
2M, 2m ;a ffi+ 2A,r,(PZ + PL' -
- 2\.P;cos o)
;(# . #) pL' +;(# -*) p'" -t#cos o
of,
e : r;(t . #) - r"(, - #) -z{@Wcos o (r 40)
Equation 1.40 is frequently used in nuclear physics to obtain the energy released
by u nuclear reaction or the threshold energy to induce a reaction, both of
which are given by the value of Q. Note that To, T',, and 0 are all measured
in the Z-frame.
IO. THE CREATION AND ANNIHILATION OF

PARTICLES
An important consequence of the equivalence of mass and energy is that

under certain circumstances particles can be created or destroyed, the mass
difference.being accounted for by the disappearance or appearance of energy.
Whether or not such a transformation occurs in a given physical situation is
governed by the requirements of conservation laws and the availability of
1?A. B. Brown, C. W. Snyder, W. A. Fowler, and C. C. Lauritsen,Phlts. Reu.821159 (1951).
THE CREATION AND ANNIHILATION OF PARTICLES 33
competing reactions or processes. The conservation laws with which we shall

be concerned here include the classical requirements on total linear momentum
and total angular momentum, as well as the relativistic statement of the con-
servation of mass f energy. The classical conset'vation laws for mass and
energy are no longer valid separately, but the quantity conserved is the total
relatiaistic energ) which includes the rest masses of all particles in the system
under consideration. In addition, we shall accept and make use of the con-
servation laws for total charge and total spin* for a system of particles.
These laws are very simple to apply. For example, if a system has,
initially, a total charge of zero and a total spin of zero, it must remain in a state
of no net charge and no net spin. Thus, if a charged particle is created or
destroyed, it must be accompanied by another particle of equal but opposite
charge. The same rule applies to spin, where we can use the terminology
"positive or negative" or "up or down" to designate the orientation of a
hypothetical vector which represents the spin.
In the case of particle creation there must be sufficient kinetic energy in
the C-frame to provide for the masses of the new particles. Consequently, in
the Z-frame the experimentalist must have a means of obtaining projectiles
of very high energy. In the case of heavy particle production it is sometimes
more practical to use two colliding beams of high energy projectiles in order to
optimize the energy available in the C-frame. For instance, if two particles
of equal mass and velocity could be directed toward each other in a head-on
collision, the C-frame as a whole would have no kinetic energy, and the available
energy in the C-frame would be a maximum.
By way of illustration, consider the production of a proton-antiproton
pair by bombarding protons with protons. Since the proton has a charge and
a spin angular momentum, the conservation laws require that its antiparticle
(having opposite charge and spin) must be created simultaneously. Multiplying
the mass of two protons by t'gives a rest energy of 1876 MeV. Therefore,
the threshold energy in the C-frame for this reaction is 1876 MeV. The question
we must answer is, what must be the energy of the bombarding proton in the
Z-frame in order to initiate the reaction ?
The reaction may be represented by the expression,
p+p-,p+p+p+p.
Here the bar over the particle symbol indicates an anti-particle. Note that
before the collision there are two particles, but after the collision there are
four particles of equal rest mass. In what follows we will assume that the target
proton was initially at rest in the Z-frame. After the collision, the minimum
total energy in the C-frame is
E* : 4Moc2,
assuming that all four particles are at rest in this frame. Before the collision,
the total energy in the C-frame was
2Moc2 + T*.
* Spin will be discussed in detail in section 9 of Chapter 3 and in Chapter B.
Therefore, in the C-frame the threshold value of T* is 2Moc2. Then

T* :2Moc2 :2(M - M)t' :2(y* -- l)Moc,
T* :2'
From Equation 1.34, the total energy in the Z-frame must be
E:y*E* :28* :BMocz.
Since E : T + 2Moc2, we find that the threshold kinetic energy in the Z-

frame for this reaction is Z : 6Moc2.
PROBLEM 1.26
A positive pion can be produced through the reaction

p+p --- p + n -l n+ by bombarding protons at rest with
high-energy protons. Find the minimum kinetic energy for
the incident protons (in the L-frame) to initiate this reaction.
Take the z+ rest mass energy to be about 140 MeV.
The previous examples have illustrated the creation of particles from

kinetic energy, but it is also possible to convert electromagnetic radiation
quantat to rest mass energy. A common example of this is the production of a
positron'electron pair by the annihilation of a gamma ray quantum, a process
frequently referred to as "pair-production." Since the electron and positron
each require 0.51 MeV of rest mass energy, it is evident that the threshold
photon energy for this process is slightly over I MeV. Pair production occurs
only in the presence of a nucleus or some other particle which can be given
some of the initial momentum of the photon.
PROBLEM l-27 ,:' ',:.' -:'' ':.'' 'i:ir "i!,' ',:r' 'l
Show that the conservation laws cannot be satisfied for the

conversion of an isolated photon into a positron-electron
pair. Let the energy of the photon be ltv and take its
momentum to be ltalc.
The inverse process, the annihilation of a particle-antiparticle pair, also

occurs with the emission of two or more energetic photons. At least two photons
are always required to conserve linear momentum, but we will see that the
f The quantum theory of radiation will be developed in Chapter 2. However, the concept of the
photon as a quantum of energy of amount hy will be used freely here, since it should be familiar to the
reader from an earlier course.
THE RELATIVISTIC DOPPLER SHIFT 35
number of photons emitted is determined by considering the spin conservation

law as rvell as the dynamical conservation laws.
Since electrons and positrons each have a spin of *, there are two possibil-
ities for the initial spin state. If the particle rpirrr are parallel, the total spin
of the system is unity; if they are antiparallel, the system has ,.to rpin.
Photons, on the other hand, each have a spin of l, so the production ofiwo
photons would necessarily result in either a spin of 0 or a spin of 2. In order to
obtain a state having a total spin of I and satisfying the linear momentum
conservation law, we must have at least three photons ! Therefore, for electron-
positron annihilation two photons will be emitted in the antiparallel spin case
and three photons will generally be emitted in the case of parallel spins.
Assuming that electrons and positrons having equal energies

collide head-on, what laboratory kinetic energy per particle
is required to produce the reaction,
e+ * e- ---> n+ | n-?
Use 139.6 MeV for the rest mass energies of the pions.
(Ans. : 139.1 MeV.)
II. THE RELATIVISTIC DOPPLER SHIFT
The four-vector formalism already derived may be used to obtain the

Doppler frequency shift in a straightforward manner. We need only incorporate
tlre Planck expression for the energy of a photon, E : lt!, and the relaiivistic
expression for the momentum, ! : hplc, where y is the frequency. Consider a
monochromatic source at rest in the unprimed frame as shown in Figure l-10.
An observer moving to the right at velocity Bc will detect a frequency which is
Bc
-->
Observer
E:h2,,,tr:$
/
Sotrlce
\ 1l_
Figure l.l0 Relative motion of observer and light source.
lower than the proper frequency. For the case of the moving observer we have
the transformation equation,
PL : rPu'
The equation for the fourth component becomes
(Pn)':T(Pn-iFPr),
ort
and
+:r(+ -iPT*'u)
v
^., _vo(l -Bcos0)
- '\/L p,
-
As the observer in the S' frame moves toward the source from the left,
he light emitted at the angle 0 : n and he measures the frequency,
sees
,ot/l 1
',
u
\/-r-P
B
(Approaching observer)
As he passes the source he sees light emitted at the angle 0 : nl2 and he
measures,
yo
y' : (Transverse Doppler effect)
ffir.
As he recedes to the right, the observer in S' measures light emitted at the
angle 0 : 0 and he obtains,
, : ,o{T-p
vt' (Receding observer)
ffi.
Now suppose we consider the inverse transformation in which the observer
is regarded as stationary in the S frame and the source is moving. In matrix
form this is (see Problem 1-20),
Pu : f -1P;.
The equation for the fourth component is
Pe:y\i+ipP)
OI,
*f cos0')
- \/t_
.,_r,o(1
p,
Again we have the three cases:
(u) Approaching source : 0' :0, and , : "!,L + |

\/t p
THE RELATIVISTIC DOPPLER SHIFT 37
(b) Receding source ; 0' :2, and , :'o1.1 - P

\/t + p
(.) Transverse'motion : 0' : a"d
|,
The above results are consistent with the principle of relativity in that it is
not possible to distinguish between motion of the source or motion of the
observer. A relative velocity of separation decreases the observed frequency
and a relative velocity of approach increases the observed frequency. Note
that the transverse effect is simply an example of the phenomenon of time
dilation, that is, that "moving" clocks run slowly. These results may be sum-
marized by the vector equation,
ph.'a)
v':
vo(L
-
\/r- p, '
where B is always positive, i is the unit wave vector, and A is the unit vector
in the direction of the velocity of the primed frame.
PROBLEM I-29
The light emitted by hydrogen atoms which are traveling at

a velocity of 0.1c is observed in the laboratory from the
forward and backward directions by means of a mirror.r8
Calculate the wavelength separation between the forward
and backward beams for the spectral line of wavelength
1o :4861 A.
(Ans. : 977 A.)
PROBT.EM
'.30
Assume that the stars of a binary pair revolve about their
center of mass with a linear velocity of 6000 km/sec. If one
of the stars emits a spectral line of wavelength 6000 A, what
will an observer measure for this line when the emitting star
is receding, approaching, and moving transverse to the line
of sight if (a) the observer is in the rest frame of the center
of mass, and (b) the observer is moving with speed 0.6c with
respect to the center of mass frame of the binary ?
(Ans.: (a) 6120 A, 5BB0 A, SSSS A; (b) 12,2+A A, 3060 A,
50e4 A.)
18This is the experiment of H. E. Ives and G. R. Stilwell, J. Opt. Soc. Am.28r2l5 (1938) and 31,
369 (1941), but with an exaggerated aromic velocity.
THE SPEC|AL THEORY OF RELATIVITY
PROBT E/r4
'-3'
In its own rest frame, a no meson decays into two photons
of equal energy and equal but opposite momenta. If a n"
has a velocity ao in the laboratory, what are the highest
and lowest photon frequencies that can be produced when
it decays ?
12. GEOMETRIC REPRESENTATIONS
It is possible to make two-dimensional plots of space-time events if we

confine our attention to only two components of X, say X, : x ?rtd Xn -- ct.
Such a diagram is shown in Figure 1-I 1. The line given by x : c/ is the "world
line" of a light wave, since it represents the progress of events moving in space-
time at velocity c. The world line of a particle can trace out a curve of any
shape provided the angle between the time axis and the tangent to the curve
<45". This restriction is obvious, since
Idx
tan?: cdt
':c <t.
*u,
Now consider a frame S' traveling at velocity pc relative to S in the x-
direction. The line r' - 0, which corresponds to ct : xl0 (from Equation 1.6),
becomes the time axis in the primed frame. The Iine ct' - 0, which corresponds
to ct : pxris the space axis in the primed frame. Hence, the angle oc between
Figure l-ll Space-time diagram.

The curved line represents the world
line of a particle. The angle through
which the primed axes are rotated is
given by tan o: f .
GEOMETRIC REPRESENTATIONS 39
Figure l'12 Calibration curves for the coordinate axes. The segments OP1,OP2, OQ, and OQ,
each represent unit interval along their respective axes.
the two space axes and between the two time axes is given by
tan a. : f,
and these axes can be drawn once B is known. Now it is necessary to obtain
the calibrations for the two sets of axes. For convenience define the new vari-
ables s : ct and s' : 6f,t . Then the Lorentz transformations are:
x'-y(x-fis) + fs')
x -- y(x'
I .41)
r'-y(s-Fx) s:y(s'+ 0x') (
Now draw the hyperbolae 12 s2 : * I as shown in Figure I-12. Event 0, it

-
the space-time point determined by the intersection of the line s : Px and the
right branch of the hyperbol z, x2 .r2 : 1. Then the coordinates of Q, are
x : T and s : y P in the unprimed- frame, but in the primed frame they are
1,' : I and .t' : 0. The latter are obtained from the relations in Equation l.+1.
Therefore, interval OQ, corresponds to unit length along the spatial axis in
the primed frame. Similarly, OQ, ,orresponds to unit time along the time
axis in the primed frame. Likewise, the vertices of the hyperbolal give unit
calibratipns of the unprimed axes.
The concept of simultaneity is readily visualized from space-time diagrams,
as a glance at Figure l-13 will show. Events P, and P, are simultaneous in
the primed frame, but occur at times srlc and srf c in the unprimed frame.
Figure l-13 Concept of simultaneity. Pt and P,

are simultaneous in the primed frame while Pr and Pn
P,, Pr are simultaneous in the unprimed frame.
t-;,
Conversely, events P, and Pn are simultaneous in the unprimed frame but

not in the primed frame. The Lorentz contraction and time dilation are shown
in Figure i-t+, where Io and r0 represent the proper lengths and proper times
in each case.
the form of space-time diagram that is
fn discussions of cosmology,
frequently used is the Minkowski or world-time diagram such as that shown
in !-igure 1-15. Note the similarity between this figure and Figure l-12,
with the calibration curves removed.
Figure l-14 Graphical illustration of time dilation and the Lorentz contraction. The scale used
here is for y : ;.96.
SUMMARY 4t
I
Absolute Future
Ar, is time-like
a
,/p
Present Present
Ar,, is space-like
Present
Absolute Past
Figure l-15 World-time diagram. An event P will appear simultaneous to an observer at O by

transforming to aframe x'. An event Q willappear to be localized at O by transforming to a frame s'.
This section has been merely a brief introduction to some of the basic
ideas of the geometrical approach to the solution of problems in special
relativity. To some the geometrical approach is fresh and appealing; others
prefer the analytical methods stressed in this book. Those readers who wish
to pursue the intricacies of the geometrical method are referred to the recent
book by Shadowitz.rg
SU'I,IMARY q
The two postulates of the special theory of relativity are that (1) all
inertial frames are equivalent for the description of all physical laws, and (2)
that the speed of light in vacuum is the same for all observers. Since the
Lorentz transformation follows directly from these postulates, any physical
law which retains the same mathematical form under a Lorentz transformation
automatically satisfies the principle of relativity. Such a law is said to be
Lorentz couariant. Two important kinematical effects which derive from the
theory of relativity are the contraction of lengths parallel to the relative motion
and the dilation of time. An important dynamical effect is the increase of
mass with velocity, which leads to the statement of the equivalence of mass and
re Albert Shadowitz, Special Relatiaitl. W. B. Saunders Company, Philadelphia, 1968.

energy, E : mcz. The conservation of total energy, which includes rest mass
energy, now replaces the separate conservation theorems for mass and energy
in classical physics. The concept of a four-vector is developed from the treat-
ment of time as a fourth dimension on an equal footing with the three
dimensions of Euclidean geometry. The invariance of the "length" of a
four-vector is shown to be a useful property when solving physical problems.
Applications are made to particle collisions, particle creation and annihilation,
a.td the Doppler shift. The use of the space-time diagram to solve problems
graphically is briefly introduced.
REFERENCES
p. G. Bergmann, Introduction to the Theor2 of Relatiaitl. Prentice-Hall, Engle-
wood Cliffs, N.J., 1962.
A. Einstein, The Meanirg of Relatiuit2. Princeton University Press, Princeton,
N.J., 1946.
A. P. Fiench, Special Relatiuity. W. W. Norton, New York, 1968'
J. D. Jackson, classical Electrodlnamics. John wiley and sons, New York,
1962, Chap. ll, 12.
R. B. Leighton, Principles of Modern Physics. McGraw-Hill, New York, 1959,
Chap. l.
C. MOller, The Theoryt of Relatiuit2. Oxford University Press, New York, 1952.
W. K. H. Panofsky and M. Phillips, Classical Electricit2 and Magnetism. Addison-
Wesley, Reading, Mass., 1955, Chap.14-17'
R. Resnick, Introduction to special Relatiuit\. John wiley and Sons, New York,
r968.
W. G. V. Rosser, An Introduction to the Theoryt of Relatiuity. Butterworths,
London, 1964.
J. H. Smith, Introduction to Special Relatiaity. W. A. Benjamin, New York, 1965.
f. W. Van Name, Jr., Mod.ern Ph-ysics. Prentice-Hall, Englewood Cliffs, N.J.,
1962, Chap. 3.
F. K. Richtmyer, E. H. Kennard, and J. N. Cooper, Introduction to Modern
Physics. McGraw-Hill, New York, 1969, 6th edition'
A. Shadowitz, special Relati.uiti,. w. B. Saunders, Philadelphia, 1968.
CHAPTER 2
Wffiffiffi ffiffiffiffiffiffiWffiffiffiffi ffiW Wffiffiffi
In this chapter we will review some of the significant events which led to
the realization that electromagnetic radiation is quantized, and further, that
these quanta-called photons-display both wave and particle characteristics.
What at first seems like a conflict between contradictory natures will be
resolved by adopting the point of view that a photon is a wave packet which
can participate in either wave-like or particle-like interactions.
I. BLACKBODY RADIATION
The quantum theory of radiation had its origin in the search for an
explanation of the spectral distribution of the radiant energy emitted by u
heated body. By the term spectral distribution, we refer to the relative amount
of energy associated with each wavelength interval of the emitted radiation.
It has been known for a long time that the color of a heated object changes
to a dull red at about I100"K and that the color of the visible light emitted
shifts toward the blue end of the spectrum as the temperature rises further.
Experimental curves of the distribution of energy with wavelength for a given
equilibrium temperature show the same general characteristics, regardless of
the material of the body. Hence, it is natural to define an ideal blackbodl,
which is a perfect absorber (and emitter) of radiation. Since it reflects no
light at all, it must appear perfectly black unless itis emittinglight in the visible
region of the spectrum. If a pulse of visible light, made up of a narrow band of
frequencies, were incident upon such a blackbody, all of the light would be
absorbed and would then be reradiated in all directions at greatly reduced
intensity and with a different spectral distribution. It turns out that a black-
body which is in thermal equilibrium with its surroundings has a constant
spectral distribution of radiated energy which is characteristic of all blackbodies
maintained at that same temperature. Moreover, the spectral distribution
curve for a real object can be predicted by multiplying the blackbody curve by
the absorptivity of the real body.
A study of the radiation from an ideal blackbody can be approximated
experimentally by observing the light emerging from a small hole in an iso-
thermal enclosure, such as a hollow block of carbon. If the hole is sufficiently
43
THE BEGINNINGS OF THE GIUANTUM THEORY
L
lr
E 100
Figure 2-l Spectral distribution
t< of blackbody radiation for several
a)
d ifferent temPeratu res.
d50
24
Wavelength, in microns
small,.the energy radiated through it will have a negligible effect upon the
equilibrium state in the cavity. For such a cavity, a typical plot of the spectral
distribution of energy for several absolute temperatures is shown in Figure 2-1.
The curves shown were obtained by Lummer and Pringsheiml in 1900,
although their general characteristics were known earlier. It should be noted
that (1) the short wavelength cutoff advance s toward the origin as the tempera-
ture increases, (2) raising the temperature increases the energy of all spectral
components, and (3) the peak of the curve shifts to shorter wavelengths as the
temperatuie increases. The shift of the peak of the curve was found to obey
the following empirical relationship, commonly called Wien's displacement law,z
).oT : constant, (2.1)
where the symb oI )., refers to the value of the wavelength corresponding to
the peak of the curve. A thermodynamic expression also exists which relates
the total power radiated per unit area of a blackbody to its absolute tempera-
ture. This is known as the Stefan-Boltzmann law and is expressed mathe-
matically as
E(T) - oTa, (2.2)
where o : 5.6699 x l0-8 watts m-2 deg-a. Thus the total energy radiated in
a given time by u heated object is proportional to the fourth power of its
absolute temperature. The monochromatic emissive power, E()"r Z), is the
power radiated per unit area at a given wavelength. This is related to the total
power radiated per unit area by the integral,
E(r) :I- E(1, T) dL.
1 O. Lummer and E. Pringsheim, Verhandlungen der Deutschen ph_rsikalischen Gesellschaft 1r 23, 215
(1899); 2, 163 (1900).
2 W.
Wien, Ann. Physik 58, 662 (1896).
THE RAYLEIGH.JEANS THEORY 45
Wien proposed an empirical form for the monochromatic emissive power,

E(1, T), by merely constructing a mathematical function to fit the experi-
mental blackbody curves. Although he did not completely define the functional
form, but wrote only E(A,T) :T-TQ,T), the following expression is
commonly known as Wien's law:
EQ',7):# (2.3)
The quantities a and b are not derived but are simply curve-fitting parameters.
Although Wien's radiation law is consistent with both the Stefan-Boltzmann
law and the displacement law, it is quite unsatisfactory as a theory since it is
not derivable from a physical model. That is, there was no attempt to relate
the emitted radiation to physical processes taking place within the radiating
body.
The peak of the radiation curve for a certain blackbody

occurs at a wavelength of 10,000 A. If the temperature is
raised so that the total radiated energy is increased 16-fold,
at what wavelength will the new intensity maximum be
found ?
(Ans.: 5000 A.)
PROBLEM 2.2
From Wien's law as given in Equation 2.3, derive the

displacement law and the Stefan-Boltzmann law. Express
the constants in Fquations 2.1 and2.2 in terms of a and b.
(e",. , o =#, :!.)
^,,
2. THE RAYLEIGH-JEANS THEORY

The simplest model of a radiating body is to regard it as a collection of a
Iarge number (on the order of 1023) of linear oscillators performing simple
harmonic motion. Since the particles undergoing the oscillations are, in
general, charged particles, they will radiate electromagnetic waves. In the case
of a cavity as discussed above, at thermal equilibrium the electromagnetic
energy density inside the cavity will equal the energy density of the atomic
oscillators situated in the cavity walls. When the walls are raised to a higher
temperature, the following events take place: more energy is put into existing
oscillator modes by increasing their amplitudes, new modes corresponding to
stiffer spring constants (higher frequencies) are excited, and the radiation
density in the cavity is increased until a new equilibrium point is reached.
# THE BEGINNINGS OF THE QUANTUM THEORY
According to the classical theory of the equipartition of energy, an

average energy of $kZis associated with each position coordinate or momentum
that appears as a quadratic term in the Hamiltonian. Since the Hamiltonian
for a linear oscillator may be written in the form
H:T+ ,r-PT
' - 2*- tkx',
ithas two such variables , p* and x (called degrees of freedom), and hence the
average total energy is simply fr 7. All that is necessary, then, to obtain the
spectral energy density in the cavity is to find n(r), the number of oscillators
per unit volume at each frequency a, and to multiply this number by kT. This
number is known as Jeans' number and is calculated in Appendix A of this
chapter. Its value is*
/\ : Bnyz /4\ : Bn
n\v) or n\/.) (2.+)
,"
^n
We then obtain the important Rayleigh-Jeans law,3 namely,
I(1, T) : B&T (2.5)

14'
where I(1, T) is the energy per unit volume at wavelength 2 at the equilibrium
temperature ZoK. The relationship between I()", Z) and E(1,7) is given in
Problem 2-5.
A difficulty with Equation 2.5 immediately appears when we consider very
small wavelengths. Although the Rayleigh-Jeans law describes the experi-
mental curve quite well in the long wavelength region, it diverges (in the
mathematical sense of having no limit) as the wavelengths approach zero.
This failure was such a crushing blow to classical physics that it is historically
referred to as "the ultraviolet catastrophe." We can gain some appreciation
of the difficulty physicists faced at that time if we realize that the Rayleigh-
Jeans theory utilized the important equipartition theorem of classical physics,
and further, that it had no adjustable parameters. Its failure implied that
there was something fundamentally wrong with either the equipartition
theorem or the theory of electromagnetic radiation.
3. PLANCK'S QUANTUM THEORY OF RADIATION
Being well aware of the shortcomings of both the Rayleigh-Jeans and

the Wien radiation laws, Planck examined the mathematical statistics of these
* Conversion between the parameters r., and ,i is easily done if one remembers lhat n(a) dv :
n(i.) d2.. -I\en
ldll c
n(u) : nQ,) ' : -;
lal "(i)
3 Lord Rayleigh, Phil. Mag.49, 539 (1900) and J. H. Jeans, Phil. Mag.l0' 9l (1905).
PLANCK'S OUANTUM THEORY OF RADIATION 47
theories to ascertain what changes, if any, might result in a reasonable descrip-

tion of the experimental radiation curve. As a result of this work, he was led to
certain conclusions about the nature of the electromagnetic oscillators which
are in equilibrium with the energy density within the blackbody cavity. These
postulates, which have become the foundation of the quantum theory of
radiation ) are as follows:a
(1) The amount of energy emitted or absorbed by an oscillator is pro-
portional to its frequency. Calling the constant of proportionality h, we then
write for the change in oscillator energy,
L,e : hy.
(2) An oscillator cannot have an arbitrary energy but must occupy one
of a discrete set of energy states given by
en : nh'u'
where z is an integer or zero. It was assumed that the ground state corre-
sponded to the zero energy state. The value Planck gave for the constant of
proportionality, lt, was 6.55 x 10-27 erg-sec. He obtained this value by
fitting curves such as those shown in Figure 2-1 with his radiation law (which
will be derived below [see Equation 2.8]). However, the best values for i is
now believed to be
h : 6.626196 x 10-27 erg-sec.
Planck's constant is a universal constant which plays an important role in all
quantum phenomena.
The previous picture of a continuum of oscillator states is now replaced
by u discrete set of "quantized" states. Furthermore, the amount of energy
emitted or absorbed is also quantized, since each quantum must correspond to
the energy difference between two states of a given oscillator. Each quantum
of electromagnetic energy is called a photon. The absorption of a photon of
frequency r will raise the energy of an oscillator of frequency ? by an amount
given by ha; it will have no effect on an oscillator of frequency v' t' a. Emission
of a photon occurs when the oscillator energy drops to the next lower energy;
the frequency of the emitted light will correspond to the oscillator frequency.
In what state is an oscillator most likely to be found ? If nothing excites
it, it is most likely to be found in its lowest energy state or ground state. Hence,
at absolute zero one would expect to find all oscillators in the zero energy state
according to the above model. This would be true in classical mechanics and
can be assumed here without affecting our answer. But we will see later that
quantum mechanics predicts a so-called "zero-point motion" at absolute zero
instead of the complete cessation of all vibration. At higher temperatures
thermal agitation excites some oscillators to higher states so that some sort of
distribution of oscillators over all possible states will exist for each temperature.
a M. Planck, Ann. Phltsik 4' 553 (1901).

5 The best value has been obtained from measurements of eifr using the effect in superconductors
predicted by B. D. Josephson in Physics Letters lr25l (1962). See B. N. Taylor, W. H. Parker, and D. N.
Langenberg, Reuiews of Modern Phl,sics 41r 375 (1969).
48 THE BEGINNINGS OF THE OUANTUM THEORY
The required distribution is the Maxwell-Boltzmann function derived in

Appendix B of this chapter, that is,
Ir{(r) - lt[os-€"tkr (2.6)
The higher energy states are thus less likely to be populated, and as the energy
increases indefinitely the number of such oscillators becomes vanishingly small.
It can now be seen that this is going to eliminate the problem of the
ultraviolet catastrophe, because the latter arose as a result of the assumption
that oscillators of all bnergies were excited with equal probability; hence,
all energies contributed equally to the emitted radiation. From the quantum
hypothesis we see that a high energy oscillator can contribute more radiated
energy than a low energy oscillator only if it is excited. But its probability of
being excited is so low that the energy which appears at the high-frequency
end of the spectrum is much less than the classical Rayleigh-Jeans theory
predicted.
Mathematically, this can be shown by u calculation of the average energy
per oscillator, €, based on Planck's quantum hypothesis.
@
I N(n)e"
a : *- , where the sums are taken over the energy levels, €0: €1r €2 ' '
.
I
n:0
/v(n)
-
2
n:o
lr:ot-"'nlkr nhu
0* hue-hvlkr a 27rt-znvlter + 3hye-BhvlkT a
I a s-nnt*, + t-2hvlteT * fsnvl*T *
I Not-"onlo'
n:0
: hyx (r
+ + 4xB +
2x + 3x2
where * - t-hvltcT
+r 1-' + -gr -L
I
),
, ' (l - x)-z: hvx Itv
: nYi
(1 -
"1-r
T=x: w_r (2.7)
PROBLEM 2.3
By treating the summation above as an integration over an

energy continuum, show that the average energy per
oscillator becomes the classical value, C : kT.
PROBLEM 2-1
Show that e may be written as a : -dldaln 7 r* :

-dld.uln (1 - x)-1, where a
: I lkT and, r : r-"Til Per-
form the differentiation and obtain Equation 2.7.
PLANCK'S OUANTUM THEORY OF RADIATION 49
Using Jeans' number, Equation 2.4, and the average energy per oscillator
obtained in Equation 2.7,we write the Planck radiation law as
I(1, T) : n(1) '€ - ry75\(/'ctt'tcr , )_,1

of, (2.8)
I(u, T) : n(r). €-
- ry (ehvttcr ,)-J
Planck's law in this form is in units of energy per unit volume per interval of
wavelength (or frequency). To express it in terms of radiated power per unit
area per interval of wavelength (or frequency), one need merely multiply by
the factor cl4.
PROBLEM 2.5
Assume that IQ., Z) results from y'{ beams, each of energy

density U (1, 7) and speed c. Derive the result that EQ,, T) :
t4
rQ,7), wher e EQ,, 7) is the radiated power per unit area
per wavelength interval.
PROBLEM 2.6
Using Planck's law as given in Equation 2.8,

(u) Show that it reduces to the Rayleigh-Jeans law in the
limit as 2 gets large.
(b) Show that it reduces to Wien's law
in the short wave-
length limit. Evaluate a and b.
(.) Evaluate the Stefan-Boltzmann constant.
(d) Evaluate the constant in the Wien displacement law.
(Ans.: (b) a : 2nhc2 :3.73 x 10-16 watt-m2'
b :+ : t.4z x to-2 m-oK

,\
(.) o : 2n5ka
5.67 x l0-8 watt/mz-ol(a
ffi:
Itc
(d) loT : : 2.898 x 10-3 *-"K)
4.96sk
As Problem 2-6 illustrates, the Planck theory of radiation incorporates

all that is valid from the older theories as special cases. It thus serves as an
excellent example of a conceptual advance which opened exciting new frontiers
while still preserving much of the older physics.
50 THE BEGINNINGS OF THE AUANTUM THEORY
PROBLEI{I 2.7
Einstein6 considered the thermal vibrations of a solid to be

equivalent to a large number of 3-dimensional oscillators, all
having the same frequency rr. If the average energy per
l-dimensional oscillator is E-,'.., then the total energy per
moleisU:3lfoE"".
(u) Using the value for En, given by the equipartition
theorem, show that the specific heat is given by the
Dulong-Petit value,
Cv- : 3R,
*:
where R : Âloft, the molar gas constant.

(b) Now use Planck's value for Eu' and obtain Cu. Show
that this reduces to 3,R at high temperatures.
PROBLEM 2.8
Show that the Stefan-Boltzmann law can be derived from

thermodynamics if one starts with the classical expression
P:tu;
where P is the pressure and u the energy density of isotropic
radiation. Assume that the ideal gas law holds in the cavity.
PROBLEM 2-9
Assume that the sun radiates as a blackbody. If A, :

5000 A for the solar spectrum, what is the surface tempera-
ture of the sun ?
4. EINSTEIN'S TRANSITION PROBABILITIES
Einstein? made use of Planck's quantum hypothesis to study the probability

of radiative transitions for a system in equilibrium with electromagnetic
radiation. The radiation field may be thought of as a photon gas of energy
density I (r) . Let the atomic system consist of i[ atoms, with i/, atoms occupying
energy level E, and l/, atoms excited to energy Er; that is, E, ) Er. Photons
6 A. Einstein, Ann. Ph2sik 22' lB0 (1907).

?
A. Einstein, Verhandl. deut. phlsik. Ges.18,3lB (1916); also Physik. Z.l8' l2l (1917).
EI NSTEI N'S TRANSITION PROBABILITIES 5I
with energy hv : Ez E, can be absorbed by atoms in the E, state, thus

-
raising them to the E, state i or, they can induce atoms in the E, state to emit
photons_of energy ltv and drop to the E, state. The photons radiated by induced
or stimulated emissiln-are in phase with the radiation field and are thus in phase
with each other; that is, they are coherent. Einstein also assumed that spon-
taneous transitions can occur from E, to Er, but not in the other direction. The
probabilities for these three types of transitions may be summarized, as follows:
Brrl(a) is the probability of absorption of a photon,

Brtl(u) is the probability of stimulated emission of a photon, and
A^ is the probability of spontaneous emission of a photon,
where A and B are constants. Then the transition rate for emission is
lArr*Brrl(u)l .Nr,
and the transition rate for absorption is
BrrIQ') . ifr.
At equilibrium these two rates must be equal, so we have
irt Ar, * BrrIQ,)

(2.e)
n BrrI(v)
Now tlre equilibrium populations, frr and lr{r, are given by the Boltzmann
distribution function (Equation 2.6) as:
lrt
_::: Er\
N2
exp (- kr) 'p e): exp (#),
where 7 is the temperature corresponding to the equilibrium state. Then,
exp (!; Brrl(a):Azr{Brrl(u)

and
A,,
I(r) :
,Bu exp (v) - 8,,
For extremely large energy densities it is reasonable to assume that ir/,

and that Brtl(u) ) Arr. Then from Equation 2.9 we get the result that - i/,
Bn : Brr,
and
(2.10)
It is evident that this is the Planck radiation law (Equation 2.8) and that the
ratio of the Einstein coefficients is
A4 Bthvg
'
Bu: cs
The.B coefficient is readily obtained* quantum mechanically by regarding the

atom as a dipole oscillator driven by the time-dependent electric field intensity
of the radiation field. Contributions from other charge configurations and
from interactions with the magnetic field may also be considered, but the most
pronounced effect of the radiation field is simply to give each atom an induced
electric dipole moment. Hence, the radiation resulting from the interactions
between atoms and electromagnetic radiation is normally called electric dipole
radiation. The calculation of the relative intensities of the spectral lines arising
from electric dipole transitions is one of the triumphs of modern physics.
The extension of Einstein's analysis to non-equilibrium cases has become
the basis for the amplification of signals in laser and maser devices. Note that
if the population of the upper level is increased or the population of the lower
level ii decreused such that Nrlnft > 1, more photons of energy hv will be
emitted than absorbed. Furthermore, the emitted photons are in phase with
the incident beam, so that amplification can result. In order to achieve the
non-equilibrium state, an additional process is required or at least one other
e.t"tgy level must be involved. For example, suppose a state E" t E, exists
such-that its lifetime is short and its most probable spontaneous decay involves
a transition to state E, rather than a return to E, (or any other lower state).
Suppose further that E, is a metastable state.t Then an external signal whose
freqiettcy corresponds to (8, - Er) f h canbe utilized to "pump" atoms from the
E, state to the .8, state. This has the effect of depopulating state E, and, if
(metastable)
Signal
E, (metastable)
Pump
hv Signal
Fast Decay
Figure 2-2 Three-level lasers.
* This is done in section 6 of Chapter 9.

t A state is said to be metastable if the probability for a spontaneou.r transition out of that state is
very small or zero. The physical basis for metastable states will become clear later when quantum
mechanical transitions are discussed.
EINSTEIN'S TRANSITION PROBABILITIES 53
the ,E --- Ez decay is fast, over-populating state Er. Alternatively, suppose

a state Eo exists such that the Et
- Êo transition
state. Then pumping can occur between
is fast and E, is u rn.tustuUt.
the Eo and ,8, levels. Th... are also
four-level systems which combine the proc.rr., shown in Figur e 2-2. In all
cases of amplification of a continuous signal by meanS of stimulated emission,
the pump must run continuously to maintain the non-equilibrium populations
required for successful operation.
PROBLEM 2-t0 ixi'*+x:',,1;llxr;t:'i, r,ii;lx;:nxji;il ,,E
using the Boltzmann distribution, Equation 2.6, find the

temperature that corresponds to equal populations for two
levels. what temperature is required in order for the popula-
tion of the upper level to be greater than that of thi lower
level ? How does this statistical use of the term "tempera-
ture" compare with its use in thermodynamics ?
PROBLEM 2.' I
Which of the systems shown in Figure 2-2 would operate

more efficiently if the laser material were cooled ? Show
this analytically.
PROBLEIA 2.12
A certain semiconductor has electron energy levels just

0.1 ev
below its conduction band. In order io control the
number of electrons in the conduction band, an experimenter
plans to cool the material and excite electrons to the con-
duction band by means of a weak source of photons of the
correct energy. He decides to use a carbon rod inside the
cryostat as a source of radiation.
(+) If the sample is kepr at 4"K, what is the probability
that an electron will be thermally excited to the
conduction band ?
(b) At what temperature must the carbon rod be kept so
that the peak of the blackbody radiation curve corre-
sponds to the energy gap of 0.1 eV ?
Much of the significance of laser (or maser) action lies in the fact that it
can provide an intense source of coherenl radiation. In fact, an incoherent
source acting as a pump can be converted to a coherent source at the pump
frequency if an appropriate pair of energy levels can be found in a physicat
system.
s4 THE BEGINNINGS OF THE QUANTUM THEORY
Although the term "laser" can be used for all such devices employing
photons, it is common practice to regard a device operating in the microwave
region as a maser and one operating at optical frequencies as a laser. However,
this distinction cannot be held too rigidly, since it is possible to pump at
microwave frequencies and to amplify at optical frequencies, or vice versa.
Since there are regions of the electromagnetiCspectrurn-wherein it is extremely
difficult either to obtain a monochromatic source or to amplify an existing
signal, it would be attractive to devise lasers which could operate at these
frequencies but which could be pumped at frequencies that are readily attain-
able (such as "white" light). This goal provides impetus for a great deal of
research on the energy levels of atomic, molecular and many-body systems.s
5. THE PARTICLE NATURE OF PHOTONS
The measurement of the charge of the electron by Millikane in 1909

established the fact that charge, as well as energy, is quantized. That is, any
accumulation of charge must consist of an integral multiple of the electronic
charge. Once the value of the electronic charge was known, it was possible to
determine the electron's mass from the value of ef m obtained by Thomsonl0 and
others. As a result of a great deal of research with cathode ray tubes, a number
of interesting demonstrations were devised to verify the particle nature of the
electron. The electron's mass is mo:9.109 x 10-28 gram and its charge is
e :4.803 x 10-10 statcoulomb : 1.602 x l0-1e coulomb.
It is easy to accept the particle nature of the electron because we can define
its mass, we can accelerate it, and we can make it behave as we think a particle
ought to behave. A photon, on the other hand, has no rest mass and cannot be
accelerated; however, we have seen that it has momentum associated with it
in both classical electromagnetic theory and in the special theory of relativity
(see Equation l.lB). We will now discuss two very important experimental
events which can be best explained by assuming that the electron interacts with
a single photon as if the photon were a localized particle rather than a wave
front. Thus, the quantum or particle nature of light dominates its wave
nature in these experiments. The first is the well-known photoelectric ffict and
the second is the Compton efect. ":
6. THE PHOTOELECTRIC EFFECT
Light incident upon a metal surface can, under some conditions, eject
electrons from the surface. These electrons are called photoelectrons, not that
they differ from other electrons, but merely to identify their source. The
following facts must be explained by u satisfactory theory of the photoelectric
8 B. Lengyel, Introduction to Laser Physics. John Wiley and Sons, Inc. New York, 1966. See also,
Lasers and Light, Readings.from Scientifc American. W. H. Freeman and Co., San Francisco, 1969.
0 R. A. Millikan, Phil. Mag.19, 209 (1910) ; Phys. Reu. 32,349 (l9l l).
toJ.J. Thomson, Phil. Mag. tAt293 (1897).
Light quanta
Figure 2-3 Apparatus for determining the photocurrent (switch in position. l) and the maximum
kinetic energy of the electrons produced by photoemission (switch in position 2).
t
I
b0
li
0)
0.)
!
!i
o
P
Saturation Current x
pi
2
0
Frequency --->
I u rlate voltage
Plate --
--+
-w
T.o*
Figure 2-4 Plot of photocurrent Figure 2-5 Maximum kinetic
for a given photon fre-
vs. plate voltage energy of photoelectrons vs. photon
quency. Tmax is determined from the frequency.
value of the reversed potential that cuts
off the photocurrent.
Figure 2-4. Figure 2-5 shows a plot of 7}a* ar. ? which permits the experi-
mental determination of both h and W in Einstein's equation. The first
experimental verification of this equation was made by Millikan in 1916.13
PROBLEM 2.13
If a retarding potential of 5 volts just stops the fastest

photoelectrons emitted from cesium, what is the wavelength
of the most energetic incident photons ? Use l.B eV for
the work function of cesium.
(Ans.: 1840 A.)
PROBLEM 2.14
If the photoelectric threshold wavelength of sodium is

5+20 A, calculate the maximum velocit-y of photoelectrons
ejected by photons of wavelength 4000 A.
(Ans.: 5.36 x 105 m/sec.)
ls R. A. Millikan, Ph2s. Reu.7' 18,355 (1916).

THE COMPTON EFFECT 57
PROBLEI{I 2.15
(u) Light of wavelength 4000 A is incident upon lithium.

If the work function for lithium is 2.13 eV, find the
kinetic energy of the fastest photoelectrons.
(b) What would be the maximum wavelength of photons
capable of ejecting photoelectrons from lithium at a
velocity of 0.95 c ?
(Ans. : 0.97 eV; 0.01 A.)
PROBLEM 2.16
If the photocurrent of a photocell is cut off by a retarding

potential of 0.92 volts for monochromatic radiation of
2500 A, what is the work function of the material ?
(Ans.: 4.08 eV.)
With the advent of lasers capable of emitting coherent radiation at high

power levels, two-photon photoemission has now been observed in sodium
metal.la Theory predicts that the double-quantum photocurrent should be
proportional to the square of the power of the incident radiation as opposed
to the nearly linear relationship that holds for the single-quantum photoeffect.ls
7. THE COMPTON EFFECT
It was generally known among the early workers with monochromatic

x-rays that a scattered beam always contained a longer wavelength component
in addition to the incident wavelength. Comptonl6 made a systematic study
of the scattering of x-rays from carbon and obtained the spectrum shown in
Figure 2-6. The surprising thing is that the wavelength shift is independent of
the wavelength of the source and the scattering material, although it varies with
the scattering angle. However, Compton succeeded in explaining this peculiar
effect by treating the x-ray photon as a particle which undergoes a collision
with a rest mass particle such as an electron or an atom as a whole. This is
shown schematically in Figure 2-7.
The mathematical analysis of the collision requires only the conservation
of energy and momentum. From momentum conservation we have
lo:p'cos0+pcos{
and
0:!'sin0-psin{,
lf M. C. Teich, M. Schroeer, and G. 6l I
15
J. J. Wolga, Ph1s. Reu. Letters 13, (1964).
R. L. Smith,
Ph1s. Reu.l'28r2225 (1962).
r'A. H. Compton, Ph2s. Reu.22r409 (1923).
58 THE BEGINNINGS OF THE AUANTUM THEORY
Figure 2-6 Scattering of molyb-

denum Ko x-radiation from graphite.
(a) Variation of the shifted line with
scattering angle. The peak at io is
due to the incident x-ray beam.
U
(b) Compton's data for the wave-
length shift for 90' scattering.
(From A. H. Compton, Phys. Rev.
22, 409 (1923), used with permis-
sion.)
tr,, I --) Io Wavelength --+
(a) (b)
from which we can eliminate d bV isolating the term in { on the left side of each
equation, squaring, and adding. Then,
!t, : p3 + p', - 2pop' cos 0. (2. 12)
The conservation of energy requires that
Eo-E': T,
or
T
Po-P': c
(2. 13)
Squaring this,
PZ + P'" - 2Po!' : (Tr (2.r+)
h
I,
Figure 2-7 Collision of a photon with a free, stationary mass.

THE COMPTON EFFECT 59
Subtracting Equation 2.14 from Equation 2.12,
p2-+:2!op'(l
c -coso). (2.15)
Since the velocity of the particle after the collision could well be relativistic,
we must use the relativistic momentum, Equation 1.19,
b,
lo
:7 (T' + 2moczT).
Then
Pz -# : 2moT : 2moc(!o - P), (2. r 6)
where Equation 2.13 has been used to obtain the last expression. Using
Equation 2.15 and Equation 2.16,
moc(Po
- p') : PoP' (1 - cos 0)
11 lrt cos 0)
Y - Po: n?oc' -
of,
).'-)": Ll 'Tltoc - cos 0). (2.r7)
Using the mass of the electron for ffi,, the quantity hf moc : 0.0243
angstrom and is now called the Compton wavelength. Notice that for 90"
scattering, Equation 2.17 predicts a new x-ray line just 0.0243 angstrom
longer than the primary line. Figure 2-6 shows this line as the large peak
measured by Compton. The presence of the primary peak at 90o might at
first seem surprising. Compton explained this by considering the scattering of
a photon from the atom as a whole. Thus, if one uses the mass of a whole
carbon atom instead of the electronic mass in Equation 2.17, the wavelength
shift will be reduced by a factor of 20,000 and amounts to roughly one millionth
of an angstrom. Therefore, the line scattered by an atom is for all practical
purposes unshifted.
Compton's work provided rather convincing evidence that a photon can
undergo particle-like collisions with both atoms and unbound electrons. Later
studies of the recoil electrons and their energies added further confirmation to
the predictions of the theory.u
PROBLEM 2.17
Show that a photon cannot transfer all of its energy to a

free electron. What is the maximum recoil kinetic energy
that can be given to an electron by u photon of energy .Eo ?
(See Problem 1-27.)
i:t' ii:i::ri: ;=:iiili:rii:r,,r+iirri;iii. iiiii iri4:':;. l+;r.rl i.+r=;1;1;,:.;i11;.,1,i;:riq.i.;i::r..; i;;ffi
C. T. R. Wilson, Proc. Royal Soc. (London) 104,

17
I (1923); W. Bothe, Z. Pfutsik 20,237 (1923);
A. A. Bless, Ph1ts. Reu.29, 9lB (1927).
60 THE BEGINNINGS OF THE EUANTUM THEORY
PROBLEM 2.18
What is the recoil kinetic energy of a free electron, initially

at rest, after a Compton scattering event in which a I MeV
gamma photon is scattered at 90" ? Find the recoil angle of
the electron.
(Ans. 0.662 MeV; 18.7".)
PROBLEM 2.'9
A 2 MeV gamma photon is scattered through an angle of

180' by an electron. What is the recoil kinetic energy of the
electron ?
(Ans. 1.77 MeV.)
PROBLEM 2.20
Photons of energy 0.1 MeV undergo Compton scattering.

Find the energy of a photon scattered at 60o, the recoil
angle of the electron, and the recoil kinetic energy of the
electron.
(Ans. 0.91 MeV; 0.009 MeV; 55.4'.)
PROBLEM 2.21
Gamma rays of energy 1.02 MeV are scattered from

electrons which are initially at rest. Find the angle for
symmetric scattering at this energy (that is, the angles 0 and
S are equal). What is the e nergy of the scattered photon for
this case ?
(Ans. 41.5"; 0.68 MeV.)
8. THE DUAL NATURE OF THE PHOTON
The particle nature of light, as illustrated by the photoelectric effect and

the Compton effect, is no longer viewed as irreconcilable with the over-
whelming evidence for its wave-like behavior. Such phenomena as interference
and diffraction are peculiar to a wave description wherein the region of
interaction is extended over a large portion of the wavefront, in contrast with
the localized interactions of particles. Regarding the photon as a wa'ae packet
consisting of a superposition of many waves imparts to it some of the properties
of both waves and material particles. Thus, the photon exhibits its wave
nature when it interacts with an object such as a grating, where the details of
the instantaneous phase of each of the constituent waves are important. On
the other hand, it manifests its particle nature when energy and momentum
THE DUAL NATURE OF THE PHOTON 6I
of the packet as a whole are transferred to another particle. In the latter case
the details of the phases of the constituent waves are unimportant. That both
the wave and particle aspects of photons are required for the description of
light is known historically as Bohr's principle of complementarit2, That is, the
wave and particle aspects complement each other. In order to acquire a better
understanding of the nature of a wave packet, let us first consider the super-
position of two plane waves.
In classical physics we frequently represent a plane wave traveling in
the positive x-direction by either the real or the imaginary part of one of the
following equivalent expressions
V - Aei(ku-ortl - /ri'2tht1-tlT) - trri<oncala-t) (2.18)
Here z4 is the wave amplitude, k :2nl), is the propagation constant, cr.r :2rv
is the angular frequency, and ?tr is the period of the harmonic oscillation. The
velocity of propagation of the wave front is the phase velocity, u : alk - Lu.
Suppose that two waves having slightly different frequencies and wave-
lengths are propagating together through a medium. For simplicity let us take
their amplitudes and initial phases to be equal. Then we may represent these
two waves by the expressions,
Yr : ,4 sin (kx - ot)

and
Yz : .4 sin [(f + dk)x - (,, + dco)tf,
where dk and da are infinitesimal quantities. Making use of the trigouometric

identity,
sin a * sin 0 :2 cos -|(a - P) 'sin {(a + P),
we may express the displacement resulting from the superposition of these two
waves by,
Y:Y.*Ys:2Acos +(n sln a)
(*-)
)
Note thatk':k+dkf2,-,k, and 111':u1-dal2- @t since dk<k and
da K a. Then the'factor containing the sine is essentially the same function
as Y, and may be thought of as a "carrier wave" of phase velocity u : a'lk' -
,lk. The cosine factor has the effect of modulating the amplitude of the
carrier wave, and the modulation envelope moves at the so-called group
velocity given by , : daldk. The transmission of energy (that is, a signal)
must o...rr at the group velocity and not with the phase velocity. The reason
for this will be clearer after a discussion of the Fourier integral theorem in
Chapter 4, but it may be noted here that the transmission of a signal alwa_ys
involves modulation of one kind or another. An infinitely long wave train of a
single frequency can never be used to transmit information at its phase velocity.
Signaling always involves chopping (keying), amplitude modulation, frequency
z"o.'g(*
)
sin (k'x - t,l't) - sin (kx - <ot)
Figure 2-8 One half-wavelength of the modulation envelope formed from the superposition
of two plane waves of nearly equal frequencies and equal amplitude. This pattern is repeated con-
tinuously throughout space.
modulation or phase modulation. These processes all result in the super-

position of many plane waves of different frequencies grouped around some
center frequency, that is, the formation of a waae packet. Since the envelope of
the packet must have a finite spatial extent, the sum of the amplitudes of all
of its plane wave components must be zero everywhere except where the packet
is localized. The energy carried by the packet at its group velocity is thus
analogous to the kinetic energy of a free particle of the same velocity. The
packet shown in Figure 2-8, which is made up of only two components, is not
zero elsewhere, but its envelope is repeated throughout space. In principle,
however, the amplitudes elsewhere can be cancelled if enough component
frequencies are used. How this is done analytically will be shown in Chapter
+, where the Fourier integral theorem will be used to construct localized
packets to represent material particles.
If all of the wavelengths of a packet travel through a medium at the same
phase velocity, there is said to be no dispersion. A physical consequence of this
case'is that the packet retains its shape as it propagates, regardless of the
number of frequencies which comprise it. When dispersion occurs, the packet
changes shape as it propagates. T'he dependence of frequency upon the
propagation constant is called the dispersion relation and may be written,
@ - u(k) 'k.
If the phase velocityu is a constant for all wavelengths, there is no dispersion

and the group velocity and the phase velocity are equal. That is,
du
g. THE HEISENBERG UNCERTAINTY PRINCIPLE
In the previous section it was argued that a localized packet of electro-

magnetic energy must be composed of a mixture of a large number of plane
waves of differing frequencies. Paradoxically, the more nearly monochromatic
the packet is, the broader it becomes. Thus, if the frequency spread is nearly
THE HEISENBERG UNCERTAINTY PRINCIPLE 63
zero, then the packet is so broad (its wave train is so long) that its emission
time is extremely large. Light radiated by atoms during electronic transitions
has a finite wave train which corresponds roughly to the lifetime of the state.
In the time domain these bursts, which are of the order of 10-8 seconds, result
in a spread of frequencies rather than in a single emitted frequency, and
produce what is called the natural linewidth of the spectral line. The longer
the lifetime, the fewer extraneous frequencies in the spectral line.
Denoting the uncertainty in the frequency (the frequency spread) by Ar,r
and the uncertainty in emission time by At, we can write,
^
A@ ,\, 2n r
N.*
On the other hand, if the degree of monochromaticity is described in terms of

the spread in ft-values (where k : .lt), the smaller the value of Aft, the greater
the extent of the packet in coordinate space. That is, the uncertainty in the
propagation constant and the uncertainty in position are related by
.r
L'k* '--'
t*.*
Making use of the Planck expression for the energy of the photon and the
relativistic value for the photon momentum, the above relations suggest that
LE . Lt ,-,h
and (2.1e)
Lp*' L,x ,-, lt.
This is an intrinsic limitation to the precision with which certain pairs of

physical variables can be simultaneously measured under ideal circumstances
in which there are no experimental or instrumental errors. Although these
limitations are of great importance in quantum systems, they are completely
unnoticeable in macroscopic systems because of the smallness of the Planck
constant fr. Thus, for a momentum which is known to within l0-7 g-cm/sec,
the uncertainty in r is n,lQ-zo cm, which is certainly not detectable. The
relations in Equation 2.19 are two examples of the well-known uncertaint2
in L927 .LB The same constraints apply to any
principle of Heisenberg, enunciated
generalized coordinate and the generalized momentum associated with that
coordinate. Such variables are known as canonicallt conjugate variables in
classical mechanics. Thus, if q is a generalized coordinate, then po is the
generalized momentum conjugate to q. Other examples of pairs of conjugate
variables are (2,!o), (r,!,), (0,p), and so on. In formal language, the un-
certainty principle may be stated as follows: The product of the uncertainties
in the measurement of two canonically conjugate variables must be greater
* These relationships will be derived in section B of Chapter 4.
18
W. Heisenberg, Z. Ph2sik 43r 172 (1927).
than a quantity of the order of fi. That is,
L,q. L,po > h. (2.20)
The uncertainty principle will be discussed quantitatively in sections B

and 1I of Chapter 4 in connection with the behavior of wave packets and the
concept of expectation value. Numerous examples of how it can be applied to
physical problems will occur throughout the subsequent chapters of this book.
The reader who is interested in pursuing this topic further at this point is
referred to Heisenberg's own discussion of the subject.le
SU/I4IT4ARY
The concept of the quantization of electromagnetic energy waS the key to

Planck's explanation of the spectral distribution of the energy radiated from a
thermally excited blackbody. As a consequence of the Planck theory we now
regard the process of absorption or emission of electromagnetic energy as the
exchange of a photon-which is a quantum of energy and momsnfurn*-
between matter and the electromagnetic field. Although the photon can
interfere with itself to produce diffraction effects, it behaves like a zero rest
mass particle in such processes as the photoelectric effect and Compton
scattering. These classically contradictory properties are. reconciled by
regarding the photon as a wave packet, that is, a superposition of a large
number of waves representing a continuum of frequencies whose center
frequency is given by v : Efh, that is, the packet energy divided by Planck's
constant. The packet momentum is p : hf A, which is required for consistency
with the special theory of relativity. The particle and wave aspects of a photon
are thus regarded as complementary properties which are manifestations of
the particular way in which the photon interacts with a specific detector. Thus
the photon exhibits its wave nature when it interacts with an object such as a
grating, where the details of the instantaneous phase of each of the component
waves are important. On the other hand, it displays its particle nature when
it transfers energy and momentum to a material particle.
Also discussed in this chapter are the Einstein probabilities for spontaneous
and induced transitions. These are shown to lead to a radiation energy
density that is consistent with the Planck law. They are then used to illustrate
the essential ideas behind the operation of a laser.
10 See Chapter II of W. Heisenberg, The Ph2sical Principles of the Quantum Theorlt. Dover Publi-
cations, Inc., New York.
t The photon also has unit spin, which is ignored in the present discussion.
CHAPTER 2 APPENDICES A to C
APPENDIX A. CALCULATION OF JEANS' NUMBER

Consider the one-dimensional wave equation,
02u | 02u
0x2 7 at,'
where z is the displacement and c is the velocity of the wave. Since the variables
are independent, we assume a product solution of the form u : T(t) .
6@.
Substituting this into the wave equation we obtain the separated .q,ruiior,,' '
_ i(t)
'"ffi:fi)-
^16"@) -@2' (A1)
where -@2 is simply the separation constant. Solutions of the two equations
(Al) are
T:Csinicut+Dcosrcot
and (A2)
6:ysinkxfdcos,tr,
where k : alc :2r11.
An infinite medium can have any values of ro and ,tr, but a finite medium
such as a string fastened at each end has only certain allowed values. These
allowed values are called eigenualues, and the modes are known as standing
waves. If we impose the constraint thatu:0 for x:0 and x: L, then wi
find that d :0 and k : nnf L, where n is an integer. The general solution to
(Al ) becomes
u(x, t) : 7 sin
T t, sin a-l/ + D cos a-rt),
which may be rewritten in the form,
u(x, t) : Fnsin (k,x) cos (a;,t) + G*sin (fr,r) sin (a,r,t). (A3)
Although we will leave (A3) in its present form, one can show that it can be
written as the superposition of waves traveling to the right and to the left.
Also, the most general solution is one including a summation over alt possible
a)n. If we consider one frequency, orn, we note that
- nTrC
@n:knc:i:2run,
65
55 APPENDICES A TO C
of,
2L
fl--Tn'
C
If the frequency modes are closely spaced we can express the density of modes as
d,n :" ,r,

C
where dn is the number of modes in the frequency interval between r and

y*dy.
An analogous procedure can be used for three dimensions where we now
write the wave equation as
: | 02u
Yzu
F-atr,
and constrain u(x,_y, z,t) to be zero for.r :0, L; I :0, L; z:0, L. Since
all of the variables are independent, the solution is a product function, the
first term of which is
F,sin (T).,r"(T) .,i., (T).cosc,,,/, (A4)
provided that (nclL),(r', + n? + nZ) : @cnlL)'z. That is, n is now

@?, :
defined in terms of the three integers ft*, flu, and nr. Rearranging the last
expression,
n2: (+l : (T1,, (A5)
Each set of integers (n*, flu, n") determines a point in lattice space and each such
point occupies unit volume of the lattice. (Each unit cube contains eight
points, but each lattice point is contained in eight such cubes.) Since there is a
one-to-one correspondence between the volume of the space, the number of
lattice points contained in that volume, and the number of allowed modes of
a given frequency, we can count modes by merely calculating the volume in
the lattice space. Therefore, the number of modes of frequency pn in the first
octant of a sphere of radius z is
nr-1.!n^,
iV:B'Trr":6
n
(?",r (,{6)
Then the number of modes in the frequency interval between an and vn I dt is
dr,{ :T t"r, dr,

APPENDICES A TO C 57
and the number of modes per unit volume in the same frequency interval is
dN 4no,
L":-u'"da'
In the case of transverse waves there are actually two times this number of
modes, one for each sense of polari zation. Finally, we obtain
Jeans' number,
which is
Bn
n(u) : Y2. (A7)
-c"
APPENDIX B. THE MAXWELL-BOLTZMANN

DISTRIBUTION FUNCTION
Stppose that a given volume is divided into z cells and that n distinguish-
able but otherwise identical particles are to be distributed among the z cells.
If each particle has the same a priori probability of occupying any unit volume
of the space, then intuition tells us that in the final distribution the number of
particles occupying each cell should be proportional to the volume of the cell.
Thus, a uniform distribution is the most probable distribution, and if the cell
volumes are the same each cell would be expected to containnf zparticles. We
will prove this conclusion formally before deriving the Max*ill-Bolt zmann
distribution function.
Let 9e be the fractional volume and no the number of particles occupying
the ith cell such that
2 go:l
i:l
and 2 ,o:
i:L
n.
Then the probability of a given distribution with n, particles in gt, ltz ir gr,
and so on, is
n!
W: nrlnrl ...n,!
(g')"(gr)"'''' (g")'" (81)
The first factor is the expression from combinatorial algebra for the number
of configurations corresponding to a given distribution (that is, the number
of ways it can be obtained) when neither the order in which particles enter the
cells nor the designation of specific particles is important- The remaining
factors give the a priori probability of obtaining a single configuration. Foi
example, consider the case of distributing 9 particles over 3 equal cells such
that n, : 2, ltz :3, and fts : 4. The 2 particles in the first cell can be put
there in 2!equivalent ways, ft2:3 can be obtained in 3!ways, and nr:g
can be obtained in 4 ! ways. Thus, the number of different configurations
corresponding to the distribution (2, 3, +) is given by
9!
ztgt+t
But the a priori probability of putting any particle in a cell is l; for two
particles in the same cell it it (*)'; and so on for no particles. Therefore, the
probability of the distribution is
w:#^(l),(+),(+-)n,
which agrees with (Bl). Before maximizing (Bl) it is con'u'enient to take its
logarithm as follows:
r" (w) :'Ir''1"

*li l':'';.'*l?' - ; ':':"'x""
This can be simplified by means of Stirling's formula,
ln (n!) - nln (n) - n)
where the approximation is valid for large n. Then,
In (W) : n Ln (n) * n,'" (fi) * nz'" (fi) + . * n"^ (f) (B2)
In order to find the most probable distribution we must maximize (82).

For this we will use the method of Lagrange's multipliers, which is as follows.
Suppose we wish to find the extremum, given by (ror)or ro), of a function
.f (r,), z) :0, subject to the constraintt dr(r,), z)
:0 and Qr(r,), z) :0.
We introduce two new parameters, ir and ).r, and write,
F :f * 1'4, * 116r:0.
The conditions for the extremum are given by the five equations
AFAFAFÂFÂF
l:0. l:0. -- -- --
il-'' o7 -o
or-"' -o
aLr.-"' -n
olr-"'
wlrich permit us to obtain the five parameters xu)b zo, 1r, and )'r. The method
of Lagrange's multipliers provides a direct and symmetric approach to all such
problems involving large numbers of variables subject to any number_ of
ionstraints. To apply the method to the present problem we note that f :
ln (W), and since there is only one constraint we need only one multiplier, 2.
The constraint is
6:n flr :0'
Then, for the ith equation we have,
3
dnn'tt" w)l+ ^y-0,
oftt
APPENDICESATOC 69
and we must solve the set of
z equations obtained by letting i run from I to z.
From the equation obtained by differentiating with t.rp.Jt to n, we obtain:
h/e\
\ntl
:).+I
or
3r : flro^*' (83)
Similar expressions are obtained from aII z equations, which, when added,
yield
(g' + 3z * ''' * g,) : (ry * nz + ''' + ,,)t^,''
I -r*'
-:u
n
Then (B3) becomes
ni
8t:;t
which says that the number of particles per cell is proportional to the size of the
cell. That is, the most probable distribution is the uniform distribution.
Now we will consider the same problem with an additional constraint
on the energy of the system. The cells may now be regarded as a discrete set
of energy states, er, Such that the total energy is
e : lnoe..
The i,D Lagrange equation is
,.06,, ^ a6,
*,rt^ (w)l * h6 + ^,#,: 0,
where $r.: n - lnr :0 and $z : e - ) nueo -- 0. We then obtain the

set of equations,
ftL:gre-(litl'e-tz't
ttz:grg-(|ftt),e-xzez
Adding,
ft : s-()'r*t'(grt-^r', * gze-tzez * * g"e-xz€'1
_ ,-(),iL) 7,
where we have defined the partitionfunction, Z,by

Z :2g6d1rru : nexrll (84)
Then,
,.: 9tft ,-i"'. (85)
rLi-
Z"
To obtain the Maxwell-Boltzmann function we define )uz: Ilkf. Then,
fli : floe-€tllcT,
Examine the function, dt, fot extrema subject to the

dt
constraint, t|-y - c,
"fS(-t)z
where-2(0) : 0 and -t(0) : 0.
(Ans.: j : -112.)
APPENDIX C. THE MAXWELL VELOCITY

DISTRIBUTION FUNCTION
A useful application of the Maxwell-Boltzmann function can be made

to a monatomit- gas to derive the Maxwell distribution law for velocities.
Consider a volume element in velocity space,
dr : du, duo du".
Since the cells are infinitesimal, we can replace the summation over g, by
an integration over dr. That is,
r-
2 so - J d, : III
du, dun dt," : !* +nr' dr,
where u is the radius of a sph;. ," velocity'space. The total number of
molecules may be expressed using Equation B5 as
,: Zfti
-ii
: 2 Agor-^z'i : A l, go..O ( - ry)
-r,.*o
n '^Jo- ---r\(- ry\
: 4nAf 2 I
or. (cl)
Similarly, the total energy is
E - 2'n'o
E:2,A*t*aaexp ?n n. (c2)
APPENDICESATOC 7I
The integrals occurring above appear frequently enough in theoretical

physics to justify th e following listing:
Io : r-o" dx :;J: (Gauss' probability integrat)

[o*
rr : Io-
rr-',' or :
*.
r,:l**rr-or'
' Jo dx: -dlo
da -I 4Va3
17
fs :
J*
r"r-",' dx :- d*4!]
: *,
Ir : j* rnr-,* dx :
# :3r/;
rs:[o*xue-'r'dx:#:*
r,n: (-1)"
f)t,
Izn+t: (- ,)" rr.
f*
Then,
N:AffiT (c3)
and
E:*J(#f : nr' 3n
(c4)
Using the result that the energy of an ide al gas is ,!nkT, we have
gnkr'
or #:
1r: #
which is consistent with the definition of the previous section. Then,
A:'(#J
V p V Vrnr.
Figure 2-9 Maxwell velocity distribution function.
From (Ct) we then find that the number of molecules having speeds between
aandu*duis:
fto:+"r(Huzexp ?m, (cs)
which is the Maxwell velocity distribution.
PROBLEM 2.23
Using the Maxwell velocity distribution and the integrals

evaluated above, find the average molecular speed, the rms
speed, and the most probable speed for the molecules of an
ideal gas of temperature T. Show that these speeds are
proportional to l.L2:1.22:1, respectively.
SUGGESTED
REFERENCES
A. Beiser, Perspectiues of Modern Ph2sics. McGraw-Hill Book Co., Inc., New
York, 1969.
C. H. Blanchard, C. R. Burnett, R. G. Stoner, and R. L. l\reber, Introduction to
Modern Ph2sics, 2nd ed. Prentice-Hall, Inc., Englewood Cliffs, New Jersey,
1969.
M. Born, Atomic Ph1,,sics, 6th ed. Hafner Publishing Co., New York, 1959.
R. M. Eisberg, Fundamentals of Modern Physics. John Wiley and Sons, Inc.,
New York, 1961.
Richard P. Feynman, Robert B. Leighton, and Matthew Sands, The Feltnman
Lectures on Physics. Addison-Wesley Publishing Co., Inc., Reading, Mass.,
1963.
R. B. Leighton, Principles of Modern Physics. McGraw-Hill Book Co., fnc., New
York, 1959.
Ph2sics, 6th ed. McGraw-Hill Book Co., New York, 1.969.
Henry Semat, Introduction to Atomic and Nuclear Ph2sics, 4th ed. Holt, Rinehart
and Winston, New York, 1969.
Paul A. Tipler, Foundations of Modern Pfutsics. Worth Publishers, Inc., New
York, 1969.
F. W. Van Name, Jr., Modern Ph2sics, 2nd ed. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1962.
Hugh D. Young, Fundamentals of Optics and Modern Phltsics. McGraw-Hill Book
Co., New York, 1968.
CHAPTER 3
YWffi trtr#Sffigpr ffiff YFnffi

ffiH,#ffi&ffi&ffi ewffiM
This chapter addresses the question of the structure of the atom, its
relationship to the more familiar elementary particles, the origin of electro-
magnetic spectra, and the source of the atomic magnetic moment. The Bohr
model, when combined with the Planck theory, leads to a simple but unified
quantum theory of matter and radiation.
I. THE ATOMIC MODELS OF THOMSON

AND RUTHERFORD
The assumption that electrons are constituents of all atoms was a reasonable
inference from such experiments as the photoelectric effect in metals, the
ionization of gases in discharge tubes, x-ray bombardment, and so forth. A
difficulty arose, however, in attempting to design a stable atom which could
contain both electrons and the positive charges necessary to make the atom
electrically neutral. It was supposed that these positive charges were many
times heavier than electrons, and this was confirmed by the elm measurements
of positive ions by Thomson.l The hydrogen ion turned out to be 1836 times
as heavy as the electron if one assumed the same magnitude of charge for each.
Thomson proposed an atomic model in which electrons were embedded in a
massive matrix of positive charge filling a volume of roughly one atomic
diameter (which was known to be about I angstrom). This model has been
called the "plum-pudding" model and the "jellium" model, the former term
referring to the role of the electrons analogous to the raisins in a pudding, the
latter deriving from the fact that the electrons were permitted to vibrate in
order to account for the radiation spectrum discussed in the previous chapter.
After the identification of alpha particles by Rutherford and his co-
workers,z'3 a series of experiments was performed by Geiger and Marsdena-G in
'J.J. Thomson, Phil. Mag. 13, 561 (1907).

2 E. Rutherford and H. Geiger, Proc. Roy. Soc. (London) 81, l4l (1908).
s E. Rutherford and T. Royds, Phil. Mag.lTrZBl (1909).
a H. Geiger and E. Marsden, Proc. Ro1t. Soc. (London) 82r 495 (1909).
5 H. Geiger, Proc. Rolt. Soc. (London) 83r 492 (1910).
6 H. Geiger and E. Marsden, Phil. Mag.25, 605 (1913).
7t
74 THE CONCEPT OF THE NUCLEAR ATOM
which alpha particles of known energies were scattered from gold foils. The
deflection of an alpha particle caused by u Thomson atom can be estimated
by the following argument. Neglecting the binding energy of the electron to
the atom, the maximum momentum that can be transferred to an electron of
mass mo by an alpha particle of mass M and velocity u is 2mou. Then the
maximum angle of deflection of the alpha particle is approximately
Ad*u* :W : ffi- lo-a radian,
since the momentum given to the electron represents the change in momentum
of the alpha particle. Similarly, one can estimate the deflection due to the
continuous charge within the atom by calculating the total impulse given the
alpha particle by the Coulomb force during its transit through the positive
jellium. Thus,
2ft' (lF)*,*
2R fz'2e dq 4Zez
muz Jo Rz MRuz'
-t
where R is the atomic radius (-10-t cm) and u'-'2 x lOe cm/sec. Substitut-
ing numerical values results in a A{,,ax - 10-a or 10-5 radian. Since the
probability of scattering from more than one electron within a single atom is
quite small, we will assume that A{-ax - l0-a radian per atom. Multiple
scattering from a layer of such atoms is completely random and obeys the laws
L+: \/N. Ad*u* (3. 1)

and
1/(d) : i/(0)e-tdttdt2, (3.2)
where i/ is the number of atoms contributing to the scattering and A{ is the

average total deflection of the alpha particle. GeigerT found that the average
deflection of alpha particles passing through a 0.5 micron gold foil is about Io.
This seems reasonable in the light of the above discussion, since a one micron
film would be about lOa atoms thick and the average deflection for random
scattering would be
n6 : f-tOn. 10-4 rad -, 10-2 rad ^, lo.
However, for angles much larger than Io, Equation 3.2 predicts an infinitesimal
probability. In particular, the fraction of particles scattered through 90"
would fs The earlier work of Geiger and Marsden8 showed that about
...,,s-s021
one alpha particle in 10a was scattered 90o or more. Though this might seem
like a small number, it is many orders of magnitude greater than any prediction
based on the Thomson model of the atom.
7 See footnote 5.
8 See footnote 4.
CLASSICAL SCATTERING CROSS SECTIONS 75
In order to account for the unexpected large-angle scattering of alpha

particles, Rutherforde proposed the model of the nuclear atom in which
-oit of
the atomic mass and all of the positive charge is concentrated in a nucleus of
dimension much less than the atomic dimension. By considering only the
Coulomb interaction between the incident alpha particle and the target
nucleus, Rutherford calculated the scattering cross section and obtained
results which agreed remarkably with the experimental results. His calculation
will be shown in the next section after a brief discussion of classical scattering
theory.
2. CLASSICAL SCATTERING CROSS SECTIONS
There are two principal methods of studying the forces between elementary
particles, namely, the study of the bound states of the particle system and a
statistical study of the scattering of a beam of particles of one type by u target
containing the other particles. The study of bound states is more limited
because we must be content with the bound systems provided by nature and
we must adapt our detectors to the energies required by those states. In the
case of the neutron-proton system (the deuteron) there is only one bound
state, so that there is not a great deal that one can learn. However, in a
scattering experiment a wide range of particle energies and scattering angles
may be studied, and the resolving power of the scattering apparatus is poten-
tially very great, because the de Broglie wavelength * of high-energy particles can
become quite small in comparison with the range of the forces being studied.
Furthermore, such quantum mechanical effects as spin-dependent forces can
be studied in scattering experiments by polarizing the particles of the incident
beam or target, or both.
In the present section we will discuss only classical scattering and will
consider the situation illustrated in Figure 3-1. The incident beam consists of
monoenergetic particles such that I particles pass through unit area normal to
Bettur
<lf
intensitY \ -
I -
-
Detector
Figure 3-l Schematic diagram of a scattering experiment.
e E. Rutherford, Phil. Mag.2l,

669 (1911).
* The de Broglie wavelength varies inversely with the momentum of the particle. This subject
will be discussed in detail in section 4 of Chapter 4.
THE CONCEPT OF THE NUCLEAR ATOM
the beam per second. That is, the beam intensity is .I particles per unit area per
second. Consider a ray of the beam that would miss the scattering center O
by the distance s if it were undeflected. We call s the impact parameter. Since
there is cylindrical symmetry about the scattering axis, we will call { the
azimuthal angle about that axis, measured from some arbitrary reference.
Then for a detector located at (0, il which detects dI\f (0, il particles per
second, we define the scattering cross section as
dN(o' 6)
d.o : (3.3)
I
The scattering cross section must have the dimensions of area since the product
I do gives the number of particles per second arriving at the detector.
Because of the cylindrical symmetry, the total number of particles
scattered into the conical wedge bounded by 0 and 0 + d0 is equal to the
number of particles incident on the washer of area2ns ds. That is,
dN(0) :2nls ds
oft
dI'{(O, 6) : Is ds d$, (3.4)
since the integral of d$ is 2n. Combining (3.3) and (3.4) *. have
do:sdsd$ (3.5)
as the classical scattering cross section. In order to arrive at a cross section

that is independent of the physical size of the detector, we define a dffirential
scattering cross section obtained by dividing do by the solid angle subtended by
the detector. That is,
do
:-: sdsd$ sds (3.6)
da -sin 0d0d6 -sin 0d0'
For a specific interaction between particles one first obtains the relationship
between s and 0;
then dsld0 can be found and the differential cross section can
be calculated immediately. Experimentally, the number of particles counted
per second per unit solid angle at the detector is
dN(0, 6) _ ,ao
T -'ddr'
do I dN(O, +)
_: (3.7)
dolda
A comparison of theory and experiment can be made by noting the results
obtained for Equations 3.6 and 3.7.
CLASSICAL SCATTERING CROSS SECTIONS
The total cross section, d, rrray be obtained by integrating the differential

cross section over all solid angles, that is,
o: In#on. (3.8)
PROBLEM 3.1
consider the elastic scattering of a hard sphere of radius a

from a stationary sphere of radius D. The incident sphere
will rebound from the tangent plane at an angle equal to its
angle of incidence as shown in Figure 3-2. Show that
do ldA : *@ + b)2 and find o.
Tangent plane
Figure 3-2 Classical scattering of hard

-"I
spheres.
Now we will discuss the case of Coulomb or Rutherford scattering.

Consider a projectile of mass m and charge ze incid,ent upon a stationar"y
scattering center of charge Ze. For z and Z both positive, a Coulomb repulsion
occurs, and the scattered particle will have a hyperbolic trajectory as shown
in Figure 3-3. The angular momentum about the target nucleus is constant
for any central force. Hence,
flu)sS : mf20 : COnStant.
Figure 3J Coulomb scattering.

Then,
r(o), : "+.
,,2
"z
(3.e)
Using the radial part of the equation of motion,
"z!' : m(i
"o* - rdr). (3.10)
12 -
Letting r : llu,
0: SUottzt
drdu,
:h;;0: du
f -'rofr, (3.11)
and
i: -su, *ry, o
: -szfiu, #.
Substituting these expressions into Equation 3.10,
zZez _ _,, _dzu

&
mszuf, d02'
Define D, the collision diameter, as the closest distance of approach in the case
of a head-on collision. That is, D is the distance from the target at which all
of the kinetic energy of the projectile is converted to electrostatic potential
energy. Then,
2zZez
D_@,
and the equation of the trajectory becomes
d2u D
dfrr+u:-*' ' (3'12)
A solution ofEquation 3.12 is u : Acos 0 *,8 sin 0 - Dl2sz. Now as 0 goes

to zero, r becomes infinite and u goes to zero. Then A : Dl2s2. Also, as 0
goes to zero, f becomes -00. From Equation 3.11 we have
(g :L: B.
\dole:o r
Therefore our solution becomes
1D
u::sin0 +W(cos0-1). (3.13)
CLASSICAL SCATTERING CROSS SECTIONS 79
The scattering angle @ corresponds to the situation when 0 : O and u : 0,

since r becomes infinite here also. Then, from Equation 3.19,
,:2/L:g'ot D o D o
(3'14)
2\ ,iro-/:Ttanr:Zcotr,
where o plays the role of 0 in Equation 3.3 and those which follow.
This provides the relationship which is needed to calculate the differential
scattering cross section in the center of mass system from Equation 3.6. Since
dsDO cscz
de: - ,,
then
do D cot O.. D O : D2/ .O\-1
(sin Q)-'4 csc2
de: , 2- , * \sina 7/ . (3.15)
This is the now famous Rutherford result. To find the fraction of particles
scattered at a given angle @ we use Equation 3.7 and write
dN@, D2 I
+) :t6
--f
-@do
stn4
U
per scattering center. Then for C scattering centers per unit area,
4{gD
I
:o:
16 +.2zsin
,O
@da
rrDzC sin (D dO
(3.16)
sln- sirrn
t f,
For gold, the constant C : plr{otlA:5.92t x 1022 atoms/cm2, where p is the
density in g/cm3, I is the thickness of the film in cm, A is the atomic weight, and
/[o ir Avogadro's number. For alpha particles having energies of the order of 5
MeV, D is about 2 x 10-11 cm. Thus the fraction of particles scatterdd into
the cone bounded by ,D and (D + dO, where
O : 90",
1S
ry91 :++dQ,_-,to-.dQ,
for a foil one micron thick.
PROBLEM 3-Z i;:l;i:i:ii:ii,rr:'i;irr:i:ii
What fraction of incident alpha particles having 2.5 MeV of

kinetic energy will be scattered through an angle of 60o or
more by u gold foil one micron thick ?
THE CONCEPT OF THE NUCLEAR ATOM
Impact parameter
Figure U Relationship of impact

parameter to scattered angle for Coulomb
scattering.
Scattering center
For alpha particles of energy 7.68 MeV scattered from a

platinum foil two microns thick, find: (u) the collision
diameter, (b) the differential scattering cross section, and
(.) the fractional number of particles scattered through
angles of 90' or more.
One of the important results of the scattering experiments of Rutherford

and his collaborators was that an upper limit was obtained for the size of the
nucleus. This figure is of the order of l0-12 centimeters. For impact parameters
of the order of 5 x 10-1s cm, the scattering of alpha particles from light
elements showed anomalies which were attributed to a force other than the
pure Coulomb force.lo This nuclear force was assumed to be an attractive,
short-range force which dominates the Coulomb force at distances which are
of the order of 10-13 cm. (This distance is now called the fermi. That is,
1F:1 x 10-1t.-.)
3. SCATTERING CROSS SECTIONS IN THE

C AND L FRAMES
Since scattering experiments are performed in the laboratory frame,

while the theory is conveniently worked out in the center of mass frame, it is
necessary to be able to transform scattering results from one frame to the other.
(See also section 9 of Chapter l.) Consider a non-relativistic particle of mass
mrand velocity u, moving toward a particle of mass mrand velocity uz :0 in
the laboratory frame (Z-frame), as shown in Figure 3-5. The velocity of the
zero momentum frame (C-frame) relative to the laboratory is a, : (mLal) I
(*, + *r) : Falmz, where p is defined as the reduced mass. The total
kinetic energy in the L-frame is Ta : $mrul and the kinetic energy in the C-
frame is Tc : tpr?. Since the kinetic energy of the center of mass of the
10E. Rutherford, Phil. Mag.37,537 (1919).
SCATTERING CROSS SECTIONS IN THE C AND L FRAMES 8t
(a)
v(.
Figure 3-5 Scattering of mass m. from mass m2 as viewed in the laboratory frame (a) and in
t,he C-frame (b).
system with respect to the L-frame is $(m, + mz)a?, the reader should show that
for elastic scattering this energy is equal to T" - Tc. We then have the useful
relationship,
TcP (3.17)
mt
rL mr
for elastic scattering. Note that in the C-system each mass is scattered at
the same angle, 0r; thus, conservation of kinetic energy requires conservation
of velocities and each mass has the same kinetic energy after the collision that
it had before the collision. It is convenient to draw the following vector
triangle:
Then the following relations are obtained from the triangle:
(rr.
- ar) sin 0, : u'tsin 0a
a, * (r, - a,) cos 0c : uicos 0a

and
sin 0, sin 0, _ sin 0,
tan 0a
ac mL, n rf cos0,
, (3.18)
ut-4, f
cos o,
-+cosoc
llt2
where r is the ratio of the masses, mrlmr.

In order to relate the scattering cross sections in the two frames it is
convenient to draw an auxiliary triangle with its sides labeled in accordance

with the quantities in Equation 3.18.
r*cosd"
Using the fact that the total cross section is the same in each frame, we
write,
g, doLd+L:IIffi)",t' oc doc d+c,
ilffi),sin
,"1(fr)""in g1 d0L : r"l(f")"sin 0s d0g.
Then,
(do\ . sin 0" .d0"
(#)" _
- \aal" sin g, dot (3.1e)
Differentiating Equation 3.18,

| *r cos0" aa,A
secz 0t' do" (r $6;*'c'
or
--------------:-0,
d|c: (r * cos 0c)2 'sec2
. (3.20)
dTt- L *rcos0"
From the triangle above,
secz0t _ (l + r' + 2r cos ldi- (3.2 r )

tit 0, (t * cos gc,)2 sin 0"
Combining Equations 3.19, 3.20, and3.2l, we obtain our final result,
(d"\ : (do\ (1 + r' + 2rcos gc)B (3.22)

\dolL \dolc r*rcosot
An isotropic scattering cross section in the C-frame will thus be peaked in the
forward direction in the laboratory by the factor (1 1r).
PROBLEM 3.4 :i, ii:l : I rn: ii.: ri : ri .: 1:
Sketch the differential scattering cross-section as a function of

angle for the laboratory and zero momentum frames for
the scattering of alpha particles by u carbon film.
BOHR'S TI'{EORY OF ATOMIC SPECTRA 83
4. BOHR'S THEORY OF ATOMIC SPECTRA
T'he nuclear atom proposed by Rutherford settled the probiem associated

with the scattering of alpha particles, but did not explain the stability of the
atom. Since it is impossible to have a stable configuration of charges subject to
electrostatic forces only, a dynamical system was proposed, analogous to a
planetary system. Such a system could account for the fact that the nucleus
is only of the order of 1-x 10-12 cm while the atom as a whole has an effective
diameter of the order of 1 x l0-8 cm. However, a serious problem arose in
connection with electromagnetic theory, namely, that a charge undergoing
continuous centripetal acceleration should radiate continuously. If this were
the case, the energy of the dynamical system would decrease continuously
and the planetary charge would spiral into the nucleus after a nominal lifetime
of about l0-8 second. That this does not occur is borne out by the infinite
lifetimes of most elementary atoms and by the nature of their radiation spectra.
Atoms do not radiate unless excited, and when radiation does occur its spectrum
consists of discrete frequencies rather than the continuum of frequencies
required by the classical theory of radiation.
In 1913, Bohrll proposed a theory which was successful in explaining the
radiation spectra of one-electron atoms, although it is in direct disagreement
with the classical theory of radiation. Bohr's postulates may be summarized
as follows:
(l ) The Coulomb force on a planetary electron provides the centripetal

acceleration required for a dynamically stable circular orbit.
(2) The only permissible orbits are those in the discrete set for which the
angular momentum of the electron equals an integer times fi, where
h : ltl2n.
(3) An electron moving in one of these stable orbits does not radiate.
(4) Emission or absorption of radiation occurs only when an electron
makes a transition from one orbit to another.
From the second postulate note that we now have angular momentum (as
well as charge and energy) quantized in atomic systems. The third postulate
rejects the troublesome claim that an accelerated charge must radiate in
atomic systems, in spite of its validity in the macroscopic world. The fourth
postulate provides the link with Planck's theory of radiation, since the frequency
of the photon emitted or absorbed is given by the energy difference of the two
states divided by lt.
In order to appreciate the implications of the Bohr theory, let us consider
a one-electron atom of nuclear mass M, nuclear charge Ze, electronic mass m0
and electronic charge e. We will use the reduced mass pc given by
p: MmoM (3.23)
+mo'
11N. Bohr, Phil. Mag.26, I (1913).
in order to neglect any motion of the nucleus. Then, from the first postulate,
puz:ry, el+)
rr"
and from the second,
p,aT : nh. (3.25)
By combining Equations 3.24 and 3.25 we obtain the expression for the radius
of the nth orbit,
n2h2 nzao
Tn: (3.26)
@F-';.
The quantity,
h2
" r-0.53A,
ao: ffiooz
is the radius of the first orbit of hydrogen calculated for a fixed nucleus, that
is, for F : ltto. It is often simply called tlte Bohr radius. Note that Equation
3.26 specifies that the radii of the allowed orbits are proportional to the squares
of the integers.
Using Equation 3.24 we may write the kinetic energy of the electron as
'' :Z-!'
T: !uu' 2r
Since the potential energy of the electron in the Coulomb field of the nucleus is
,, -
-zt'r ,
then the total energy is
E:T+V: Zez p (Ze')' woZ'

(3,27)
- ----i-
/-rJ
2r -
2 n2h2 no
r
where -
wo : To=to
^, I3.6 ev
2h2
is the magnitude of the ground state energy of hydrogen for a fixed nucleus.
Equation 3.27 indicates that the quantization of energy has arisen as a result
of assuming that the angular momentum is quantized. The frequency of a
photon emitted by u transition from the nth level to the kth level, where n ) k,
is given by
hy : .(:) : hcn - En - Ek : try #, - *)
Expressed in wave numbers, this becomes
- 2n2 p,Zzea
" hsc $-:,):R)(h-*), (3.28)
BOHR'S THEORY OF ATOMIC SPECTRA
where R7 is the Rydberg constant calculated from the reduced mass. For
hydrogen R) : 10968l cm-l, whereas the experimental value from optical
spectroscopy is R) : 109677.576 + 0.012 cm-r. The Rydberg constant
calculated for a fixed nucleus (that is, for p, : mo) is
R)- : 109737.309 cm-l.

Furthermore, the experimentally established spectral series for hydrogen are
a1l explained by Equation 3.28 by setting thi value of & and lettini n run
through the integers such that n > k. Foi example, k: l, n:2, zl +, ...
corresponds to the Lyman seriesi k:2rn:3,4,5,... corresponds_to the
Balmer series; k : 3, +, and 5 correspond to the paschen, B.uckeit, and
Pfund series, respectively. (See Figu.. 5-O;. Equation 3.28 can also be used
to calculate the ionization potential of hydrogen by letting n ---> e. Then we
obtain
v:ffi-*^, 13.6volts.
The excellent agreement between the spectroscopic l,alues for the Rydberg
constant, the ionization potential, and the emission ipectrum of hydrogln wa"s
Energy
(eV)
Continuum of positive energy

ionization states
././ ,/,/ ./,/ ,/ ,/ ,/ ,/ ,/ ,/ ,/ ,/ / ,/ ./ ,/ ,/ ,/ ,/ ,/
Paschen Brackett
series series
Lyman -r3.53
series
4a,, l6q,
Radius, r
Figure 3-5 Bohr radii and allowed energies of hydrogen. The first few lines of several of the
spectral series of hydrogen are shown at the left.
a great triumph for the Bohr theory. However, it was immediately apparent
that the Bol-rr theory was good only for a one-electron atom, since it failed
badly in the cases of helium and lithium, the next simplest neutral atoms. But
for thd spectra ol singly ionized helium and doubly ionized lithium the Bohr
theory is again quite good. We call such atoms "hydrogen-like" or "hydro-
genic" since they are one-electron atoms but their nuclear masses and charges
differ from that of l-rydrogen. The spectra of the alkali metals can also be
calculated as if their single valence electron moves in an effective field of
reduced Z due to the noble gas core of electrons. Although the gross fcatures
of these spectra are quite similar to that of hydrogen, there are some serious
discrepancies due to the non-Coulomb nature of the electrostatic field and due
to spin-orbit coupling. *
Interesting applications of the Bohr theory can be made to the sliort lived
mesic atoms and to positronium. Mesic atoms are ordinary atoms which have
captured an orbital negative mu meson in lieu of one of the orbital electrons.
Alihough tl-rese atoms have short lifetimes, t the muon completes many millions
of orbits bcfore it decays or is captured by the nucleus. During its short lifetime,
it can make transitions to or from excited states by absorbing or emitting quanta
of radiation. The study of these transitions provides additional information
about tlre nucleus.12 Since the muon is about 207 times the mass of the electron,
the radius of a Bohr orbit for a bound muon is about 7616 the radius of the
corresponding orbit for an electron. Taking into account the difference
in nuclear charge, a muon bound to a heavy element will have orbital radii
"-,10*a times the hydrogen radii, or ^,10-12
centimeter. According to Wheelerl3
a muon in its ground state orbit spends nearly half of its lifetime within the
captor nucleus. Accordingly, we conclude that the nucleus is nearly transparent
to the muon, which is to say that the muon interacts weakly with nucleons.
(u) Use the Bohr theory to calculate the allowed energies

and circular orbits for the two hydrogenic atoms
positronium and muonium. Positronium consists of a
positron and an electron, each of mass mo; muonium
consists of a proton of mass I836 mo and a negative mu
meson of mass 207 mo. Each particle has a charge of
magnitude e.
(b) Discuss the possible decay schemes of orthopositronium
(parallel spins) and parapositronium (antiparallel
spins) in the context of the discussion in section 10 of
Chapter 1.
* This will be treated in section 2 of Chapter B.

t The lifetime of the mesic atom is limited by the lifetime of the muon which decays into an
electron and a neutrino. See problem l-9.
12
V. L. Fitch and J. Rainwater, Phys. Reu.92.789 ( I e53).
tt
J. A. Wheeler, Plqts. Rea. 92, Bl2 (1953).
BOHR'S THEORY OF ATOMIC SPECTRA 87
PROEI-E/U 3-6
(u) Using the Bohr model find the energy of the photon
emitted when a mu meson makes a transition from the
first excited state to the ground state of mu-mesic
Pb2o8.
(b) The 2Pg-lS+ transition energy for mu-mesic Pb208 is
known to be 5.963 MeV. Suggest several reasons for
the discrepancy between this value and your answer
in (a).
PROBLEM 3.7
Determine the effect of nuclear mass on the first line of the

Balmer series of hydrogen by calculating its wavelength
shift for deuterium and tritium.
(Ans.: 1.79 A; 2.38 A.)
PROBTE/I4 3.8
Using the simple Bohr theory for circular orbits, calculate

the first ionization potential of helium.
(Ans. : 54.4 volts.)
PROBLEM 3-9
If helium gas in a discharge tube is excited by u potential

difference of I I volts, which, if any, spectral lines corre-
sponding to the Balmer series of hydrogen will be excited ?
(Use the simple Bohr theory and neglect spin.)
PROBLEM 3.IO
An interesting application of the Bohr theory of the hydro-

genic atom occurs in solid state physics. A phosphorus-doped
silicon crystal may be regarded as a collection of hydrogenic
atoms of unit net charge at each phosphorus site. The
effect of the periodic potential, due to all of the silicon
nuclei, on the motion of the orbital electron is taken into
account by giving the electron an effective mass appreciably
different from the electronic rest mass. In addition, the
Coulomb force is diminished by the dielectric constant of
silicon since the orbital radius is equivalent to many lattice
spacings.
Calculate the energy and the radius of the first Bohr

orbit if the dielectric constant of silicon is 1l'7 and the
effective electron mass is 0.2 mo.
(Ans.: -0.02eV; 30 A.)
5. THE FRANCK.HERTZ EXPERIMENT

The experiment of Franck and Hertz provided additional confirmation
of the discrete energy states predicted by the Bohr theory.la In this experiment
mercury vapor was bombarded with electrons of known kinetic energy. When
this kinetic energy is less than the energy of the first excited state of atomic
mercury, the only energy which the colliding electron loses is the small amount
of kinetic energy (about one part in 105) that it can transfer to the massive
mercury atom by an elastic collision. However, when the kinetic energy of the
incident electron just exceeds the energy of the first excited state of mercury,
then the electron gives up virtually all of its kinetic energy to the mercury
atom in an inelastic collision. Thus, by comparing the kinetic energy of the
electrons before and after the collision one can determine how much energy is
transferred to the target atoms.
Franck and Hertz used a simple apparatus consisting of a mercury tube
containing a cathode, a plate, and an accelerating grid which was located
physically n.ur the plate (see Figure 3-7). Both the plate and the grid were
maintained at positive potentials with respect to the cathode, but the plate
potential was slightly lower than the grid potential. This small retarding
potential on the plate had the effect of preventing contributions to the plate
current from electrons having negligible kinetic energy. Hence, if electrons
were to lose most of their kinetic energy in inelastic collisions with mercury
atoms, the retarding potential would prevent them from reaching the plate and
a drop in plate current would result. Such a drop in plate current was found
to ocCur at 4.9 volts, as shown in Figure 3-8. Furthermore, Franck and Hertz
noted that at this voltage the 2536 A spectral line of mercury appeared in the
emission spectrum of the vapor. A simple calculation shows that the photon
energy of [he 25g6 A line corresponds to 4.86 electron-volts! At slightly higher
voltages a large drop in plate current occurs and new lines appear in the
Gas inlet
Figure 3-7 Apparatus for Franck-Hertz experiment.
ra
J. Franck and G. Hertz, Verhandl. deut' ph2sik. Ges.161 457 and 512 (1914)'
X.RAY SPECTRA AND THE BOHR THEORY 89
? 3oo
L
,i
': 200
L
Figure 3-8 Current vs. voltage in the Franck- i3
Hertz experiment. (After J. Franck and G. Hertz,

Verhondl. deut. physik Ges. t6, 457 (1914).) U
L
t<
o loo
A')
510
Plrrte volttrge, \/olts
emission spectrum of the mercury vapor. This behavior is repeated at multiples

of 4.9 volts as shown in the figure. We must conclude frorn this that the Bohr
concept of discrete energy states is qualitatively correct.
6. X-RAY SPECTRA AND THE BOHR THEORY
Jvhgn high-energy electrons are used to bombard a target made of a heavy

metal, the electrons are decelerated and they will radiite. The resulting
radiation produces a continuous spectrum called Bremsstrahlung which i;
characterized by
? turp cutoff at the short wavelength end. A typical
spectrum is shown in Figure 3-9. This continuous spectrum can be accounted
for classically by the theory of electromagnetic radiati,onls or quantum mechani-
cally by invoking an inverse photoelectric effect. That is, the change in kinetic
energy provides an upper limit on the photon frequency, whereas in the
K"
a
5 25 kV
-4
!
L
ti
Figure 3-9 X-ray spectrum of molybdenum.
Kp The continuous spectrum is the bremsstrahlung
X 15 kv radiation.
hz
a)
t2
Wavelength, angstroms
15
F. K. Richtmyer, E. H. Kennard, and T. Lauritsen, op. cit.rp. 349.
photoelectric effect the photon frequency provides an upper limit on the change
in kinetic energy. Since the initial kinetic energy of an electron is determined
by the voltage on the x-ray tube, the short wavelength cutoff is given by
ltc 12,+00
^
Ac : (3.2e)
,V: V,
where Zis in volts and A is in angstroms. Thus, a 12 kilovolt x-ray tube has a
cut-off wavelength of about I angstrom. Not many photons having wave-
length )", are emitted, because the majority of electrons perform mechanical
work on the target when they stop and a large fraction of their kinetic energy
is converted into heat. Equation 3.29 provides an accurate methodr6 for
determining the ratio ef h, since a plot of r, vs. V is a straight line whose slope
is elh.
Superimposed upon the continuous spectrum are a few discrete peaks
whose wavelengths are characteristic of the target material. A few of these
peaks are also shown in the figure . The characteristic x-ray spectra of many
elements were studied by Mosel.y,tt who showed that the frequency of a given
line varies from element to element as the square of the atomic number. A
plot of some of Moseley's data is shown in Figure 3-10. This is precisely the
dependence predicted by the Bohr theory in Equation 3-28. Furthermore,
the Bohr theory accounts for the discreteness and the order of magnitude of
the photon energies of the characteristic spectrum by assuming that these
photons are emitted as a result of transitions involving inner electron shells.
If for example, a high energy electron knocks out an electron from the first
shell of a target atom, the vacancy may be filled by a transition from the second
shell, or the third, or higher shells. Since the electrons of the first shell have
been traditionally called K-electrons, the photon emitted by a transition from
the second shell to the .K shell is called Ko radiation, that due to a transition
from the third shell to the K shell is called Kuradiation, and so on. The second
9zo
X
--
N
16
Figure 3-10 Moseley's results for the effect of
irz atomic number on the frequency of a given x-ray line.
9s
f-
216202428323640
Atomic number
16J. DuMond and V. L. Bollman, Ph1ts. Reu. 51, 400 (1937).

17
H. G..f. Moseley, Phil. Mag. 26, 1024 (l91 3) ; 27, 703 (1914).
X.RAY SPECTRA AND THE BOHR THEORY 9I
thgll is designated as the L shell, the third M, and so forth, so that Lo, LB,. . . ,
Mo, M*. . . radiation is defined in a similar fashion.
Since Equation 3.28 shows that the photon energy is proportional to Zz,
a transition from the second to the first shell for coppe, (Z : 29) would result
in the emission of a photon having an energy about 84l times the energy of the
longest line of the Lyman series of hydrogen. This approximate calculation
gives l.++ Afor the K* line of copper, wherlas the correci value is 1.54 A. The
agreement is not nearly so good for transitions involving shells of higher order
because of the shortcomings of the Bohr theory and the importanf screening
effect of the nuclear charge by electrons in the innermost shells.
A process which generally competes with x-ray emission is the internal
photoelectric effect or Auger efrtt. Here, the x-ray photon does not actually
appear, but an equivalent amount of kinetic energy is given to an outer
electron, which is in turn ejected from the atom.
PROBLEM 3.1I ,,i 4;l. riir,;j1t !$1; ii{r#
Calculate the wavelengths of the ,K" and Ku lines of the

characteristic spectrum of vanadium using the elementary
Bohr theory. Compare these values with the accepted
values.
PROBLEM 3.12 ia :; tj iir al :; :: : li ; ; i" i: :: :; :: ::r i:;r :i

-:
i ;; ;; i. i ir :: ::': ;
i;r ir r: rl; :l i; ii ;,. a ai; lr iria:: l; ; i,:r X :1i:;;1 i:' :'i.:r i. a: iir':ii?
=r
What is the kinetic energy of an Auger electron resulting

from a radiationless transition from the L to the K shell of
chromium ? Use the elementary Bohr theory.
PROBLEM 3.13
(u) If the Bohr atom is interpreted as a classical oscillator

whose frequency is rro : ul2nr, show that the classical
oscillator frequency can be expressed as
2*o
lo : t
rrr"
where wn is given in Equation 3-27.
(b) Show that in the limit as n gets very large, the Bohr
frequency relation reduces to the expression in (u).
This is an example of the correspondence principle, which
states that a quantum theory result should agree with
the equivalent classical solution in the limit of large
quantum numbers.
7. NUCLEAR STRUCTURE AND SPECTROSCOPY

The nucleus of the Rutherford atom was at first thought to be composed
of a number of protons equal to its atomic mass number A, and a number of
electrons equal to (A - Z), that is, the difference between the mass number
and the atomic number. Although this composition accounted for the existence
of the electrons that appeared in the case of p emitters, it was early rejected by
Rutherford, who first proposed the existence of the neutron. The subsequent
discovery of the neutron by Chadwick established it as a nuclear constituent.
We now refer to both neutrons and protons as nucleons and call .4 either the
mass number or nucleon number. The quantity A - Z is often called the
neutron number, l[. Figure 3- l I shows a plot of 1[ versus Z for some nuclei.
There are now about 900 known nuclei, of which fewer than one-third are
stable.
The nuclear radius was determined to be of the order of l0-12 centimeters
by Rutherford. Subsequent experiments have led to the empirical'relationship,
r: roA+, (3.30)
which expresses the radius of any nucleus in terms of its nucleon number and
t44
z
L
j
.J
/
Eao
=
L
264
CJ
z
^:.
48 =r/
32 7
l6
l6
Proton number, Z
Figure 3-l I Plot of N vs. Z for stable nuclei.
NUCLEAR STRUCTURE AND SPECTROSCOPY 93
the constantro. The accepted value of ru is 1.2 x l0-13 cm (or 1.2 F), which was
obtained from measurements of the nuclear charge distribution by means of
high-energy electron scattering.18 Other methods of estimating nuclear radii
are in good agreement with this 1g5uh.1s,2o Equation 3.30 leads to two
interesting conclusions. First, the nuclear force whicl-r holds the nucleus together
must be charge independent, that is, the n-n, p-p, n-p interactions must be
essentially the same since it is .z{, and not Z or //, which determines the nuclear
radius. Second, the nuclear density is apparently the same for all nuclei, since
it is given by the expression
A A
V 4rt
T,,U
Thus, nuclear matter is roughly (oolro)s:1012 times as dense as ordinary
matter.
The binding energ) of a nucleus can be expressed in terms of the difference
in rest mass energy between the constituent nucleons and the composite nucleus.
Thus, the binding energy is the mass defecl multiplied by c2. For example, the
binding energies of the deuteron and the alpha particle are as follows:
Ea : (M, + Mn - :2.2 MeV

Mo)t,
Eo -- (2M, + 2M,
- M")t, : 28.3 MeV.
Figure 3-12 shows a plot of the binding energy per nucleon (that is, Ef A) as a
function of A. Note that the binding energy per nucleon is roughly 7 or B MeV
z,
j
O
O
L
q)
b0
(')
C.)
b0
7l
(_) 50 100 t50 200

Nucleon number, A
Figure 3-12 Binding enerty per nucleon for stable nuclei.
ll!*a for example, Robert Hofstadter, Annual Reuiews o.f Nuclear Science 7r231 (1957).
le L. R. B. Elton, Introductor2 Nuclear Theor2. W. B. Saunders Co., Philadetphia, 1966.
20 R. D. Evans,
The Atomic Nucleus. McGraw-Hill Book Co., New york, 1955.
for most nuclei except the lighter nuclei. The sharp increase in binding energy
for A : 4 accounts for the unusual stability of the alpha particle. The gradual
fall-off of binding energy per nucleon for the heavy nuclei is due to the effect of
the Coulomb repulsion, which is always present although it is dominated by
the nuclear potential. fhiq increase in Coulomb repulsion is partially com-
pensated by an increased nuclear attraction resulting from an increase in
the I,{IZ ratio as Z increases (see Figure 3-tl). The nearly constant nuclear
energy per particle contrasts sharply with the dependence of electronic binding
energy per electron on Z (and hence A) in atomic potentials.
The study of positron emission from mirror nuclei provides an estimate of the
Coulomb contribution to the binding energy as well as an approximate value
of the nuclear radius. Mirror nuclei are pairs of nuclei having odd ,4 such that
Z and i[ differ by one. For example, Oru-(Z : B, ff : 7) and N15(Z : 7,
1/ : B) are mirror nuclei, each containing 15 nucleons. Thus, the binding
energies of these nuclei should be the same except for the correction due to the
difference in the Coulomb interactions. Using the classical expression for
the Coulomb energy of a uniformly charged ball of charge Ze and radius r,
the difference in Coulomb energy between nuclei of atomic numbers Z and
Z-lis,
L,E :Yrt, - (z - t),1 : Y* {rt - t). (3.31)
PROBLEM 3-14
Find the mass defects of O15 and Nr5. Compare this differ-
ence in rest mass energy with the Coulomb energy calculated
from Equation 3.31.
By measuring the energy of p+ decay from element Z to element Z - 1, r can be

calculated. Studies of a number of such reactions confirm Equation 3.30 with
a value of ro ^, 1.5 F. The fact that the binding energy of a nucleon is inde-
pendent of its charge, except for the Coulomb correction, adds further con-
firmation to the validity of the assumption that the nuclear force itself is
independent of the charge of the nucleon. The nucleon force is known to be an
extremely short-range force, since the Coulomb law adequately describes the
scattering of alpha particles for impact parameters greater than about I x
10-12 cm. A commonly used form for the nucleon potential is the co-called
Yukawa potential,zl
o--r/b
l/ ,--," (3.32)
, ,
where & is the range of the nuclear force, ^-,lQ-ra cm : I F. This expression
is not valid for r < 0.4 F, where a repulsive term dominates. Forr: D : I F,
21H. Yukawa, Ph2s. Math. Soc. Japan 17' 48 (1935).

N UCLEAR STRUCTU RE AND SPECTROSCOPY 95
Repulsive core
Coulomb repulsion
l
Figure 3-13 Schematic d iagram of the
nucleon- nucleon potential. Since the exact shape
of the well is not known, it is approximated by a
square well. For a neutron there is no Coulomb
repulsion. Yukawa potential
- -40 MeV
i t,i
V ,-.,, I to l0 MeV for a nucleon;- for r : ljb : l0 F, V ,-' l0 to 100 eV. A
schematic diagram of the nucleon-nucleon potential is shown in Figure 3-13.
Another property of the nuclear force, which is quite a departure from the
behavior of Coulomb and gravitational forces, is that of saturation That is,
a nucleon seems to interact with only a limited number of other nucleons,
analogous to nearest-neighbor interactions in solids or the saturation of
chemical bonds in ligand theory. It is this characteristic that is evoked to
explain the unusual stability of the alpha particle. There appears to be little
or no attraction between an alpha and another nucleon and hence there is no
stable nucleus of A : 5.
As a result of the discovery of natural and induced radioactivity, as well
as the research leading to the identification of the "mysterious emanations"
called q, f , and y rays, it became evident that the nucleus itself must have an
inner structure. From careful measurements of the energies of emitted particles
and y rays, schematic diagrams of the energy levels and decay schemes of many
nuclei'have been produced (see Figure 3-14). One striking difference between
Ground strrte
Grotrnd state
Z, N Z-2, N-2 Z,N Z+I,N

(b)
(a)
Figure 3-14 Schematic energy level structure for hypothetical nuclei. An a-emitter (a) or a
p-emitter (b) may decay directly to the ground state or to an excited state of the daughter nucleus.
optical and nuclear spectroscopy is associated with the energies of the photons
in each case. Recall that the energy of a photon in the visible region is of the
order of a few eV, while an x-ray photon is generally in the keV regime.
Since nuclear energies are of the order of MeV, 7 photons are frequently in
the MeV energy regime. However, there are no sharp demarcations between
these energy regimes and accordingly, "hard" x-rays may equally well be
called "soft" y rays. The atom emitting a photon must suffer a recoil in order
to conserve momentum as well as energy. For visible photons this recoil
energy is often negligible, but it can become significant for energetic gammas.
PROBIE/I4 3-15
(u) Show that the energy shift of an emitted photon due to

the recoil of the emitting atom may be expressed as
E2
LE - wIr" (3.3 3)
where E is the transition energy and M is the mass of

the atom.
(b) Calculate the recoil shift for the emission of the most
energetic Lyman line from a free atom of hydrogen.
(.) Calculate the recoil shift for the emission of a 14.4 keV
y ray from a free atorn of Fe57.
The effect of such recoil on the light emitted from a source consisting of many
atoms is to broaden the spectral line associated with the transition. * Of
course, every spectral line has a natural width (there is no perfectly mono-
chromatic source), but the recoil broadening can be many times greater than
the natural line width. An important exception to this occurs in some crystals
when the emitting atom is bound to the lattice with sufficient energy to prevent
its recoil. In such cases the emitted line is very sharp since the mass in the
denominator of Equation 3.33 becomes the mass of the whole crystal and
AE ^,0. When recoilless emission occurs the linewidth is essentially its
natural width, and the photon may be absorbed by an unexcited nucleus of
the same species by the process of resonance absorption. This phenomenon is
known as the Miissbauer ffict after its discoverer.22 It has become an important
tool in many areas of physics, but particularly in solid state and nuclear
physics.2s
Brief mention should be made of the gamma emission process when recoil
does occur. Here we may distinguish three separate cases. (1) If the free-atom
* See Chapter 2, section 9.

2sR. M<issbauer, Z. Pfusik|.5lr l24 (1958) i Naturwiss.45,538 (1958).
23 See, for example, G. k. Wertheim, Mdssbaver Efect: Principles and Applications. Academic Press,
\ew York, 1964.

NUCLEAR STRUCTURE AND SPECTROSCOPY 97
Zero
p'i Phonon
Line
i^
Figure 3-15 A hypothetical gamma ray spec-
trum for an atom in a solid at low temperature.
The narrow zero-phonon line is the one normally
used for Miissbauer spectrometry. z
Plionon Wing
O
- 0.1 - 0.05 0
Energy Sliift - eV
recoil energy, Equation 3.33, is greater than the binding energy of the atom
in the solid (15 to 30 eV), the recoiling atom is dislodged from its lattice site.
(2) If the free-atom recoil energy is less than the binding energy but large
compared to the phonon* energy (-0.1 eV), the gamma-emitting atom will
remain in its site and dissipate its recoil energy to the lattice as heat. (3) If the
free-atom recoil energy is less than the phonon energy, then it is possible to
observe either recoil-free gamma emission or gamma emission accompanied
by the excitation of a discrete phonon of energy ha ^, 10-2 eV. This third
case is the effect discovered by Mijssbauer, and the gamma spectrum consists of
a sharp peak corresponding to recoil-free gamma emission (the zero phonon
peak) as well as a continuous spectrum on the low energy side (the phonon
wing) corresponding to gamma emission accompanied by phonon excitation
(see Figure 3-15). The reader should note that recoil-free gamma emission is
analogous to the elastic scattering of photons as in the production of the
unmodified line in the Compton effect (see section 7 of Chapter 2).
A frequently used source for Mossbauer studies is ,rCo57, which captures a
-K-electron and decays to an excited state of ,.Fe57. When Fe5? returns to its
ground state it emits a recoilless photon having a linewidth of about l0-1r
times its energy of 14.4 keV. Such a sharp line can easily excite an Fe5? atom
in its ground state by the process of resonance absorption. Thus, a sample of
Co57 provides a source of radiation of unbelievable spectral purity and un-
excited Fe57 provides an equally sharp detector. To get some appreciation fior
the sharpness of this line, note that a relative velocity of only a few hundredths
of a millimeter per second between source and detector will produce a Doppler
shift great enough to prevent the resonance absorption.
(u) Calculate the Doppler shift due to a relative velocity of

3 cm/sec.
(b) Find LEIE, where E : 14.4 keY and AE is the Doppler
shift obtained in (a).
* Phonons are the quantized lattice vibrations referred to in Problem 2-7.

PROBLEM 3.17
(u) What is the line broadening due to recoil of a Ni60

nucleus upon emission of a 1.33 MeV gamma ray?
(b) What is the ratio LEIE for this case?
(Ans.: 15.8 eV; ^*,10-5.)
Several nuclear models have been proposed to account for the observed
properties of nuclei. Unfortunately, no single model satisfactorily explains all
of tt experimental facts, so that more than one model must be invoked unless
one is" concerned only with a specific property of the nucleus.
Perhaps the most popular model is the shell model. In this model each
nucleon is regarded as moving in an orbital state under the influence of a
nuclear field *tri.h represents the average effect of all the other nucleons. The
allowed quantum states are found by using the j-j coupling scheme of combining
angular momenta, which will be discussed in section 5 of Chapter B. The
noieworthy successes of the shell model are its ability to account for nuclear
spins, the stability of the alpha particle and certain other nuclei (those associated
with the so-called "magic numbers"), and for many features of nuclear spectra.
The shell model not only ignores such problems as the saturation of the
nuclear force, the nearly constant nuclear density, and the binding energy per
nucleon, but it gives the wrong results for the scattering of neutrons from
nuclei. It turns out that neutrons interact so strongly with nuclei for certain
discrete energies (sometimes spaced only a few hundred eV apart) that they
are often trapped for a while before being ejected by the target nucleus. Such
trapping events are called "resonances."
Th; model proposed to account for these resonances is the liquid-drop
model. Here we regard the collective behavior of the nucleons as a many-body
system analogous to the lattice vibrations of a solid. The energy states of the
system are now the excitation energies of the collective system (analogous to
the normal modes), which one would expect to be a set of discrete but closely
spaced levels. Hence, a neutron having one of these energies will, in a collision
with the nucleus, immediately share its energy with the whole system. A
considerable time interval will elapse (compared to the transit time through a
distance equal to the nuclear diameter) before a neutron accumulates enough
energy to emerge as the "scattered" particle.
In addition to explaining the resonances, the liquid-drop model is useful
in visualizing the satuiation of nuclear forces and the phenomenon of fission.
It has also been used to develop a phenomenological formula known as the
semi-empirical mass formula.za
Another useful model of the nucleus is called the optical model. Here the
nucleus may be regarded as a cloudy crystal ball, since it is essentially opaque
to short wavelength neutrons (high energies) and it gets increasingly transparent
as the neutron wivelength increases (lower energies). Its cloudiness is a measure
2a See, for example, R. D. Evans, op. cit., p. 366, or L. R. B. Elton, op. cit., p. l1B.
FUNDAMENTAL FORCES AND EXCHANGE PARTICLES 99
of the absorption probability and its index of refraction is related to the deflec-
tion of scattered neutrons. The phenomena of absorption and resonances can
be accounted for by means of canceling and reinforcing traveling waves,
respectively.
8. FUNDAMENTAL FORCES AND EXCHANGE

PARTICLES
The historical problem of accounting for "action at a distance" was

avoided in classical physics by postulating fields of force which pervaded all of
space. Thus, the effect of body A on a remote body B can be explained-by the
fact that body .B moves in the force field of A, and vice versa. The field has an
energy density associated with it and it possesses momentum which can be
transferred to and from massive particles. We have already seen in the case
of the electromagnetic field that this energy density may be regarded as arising
from a collection of quantized oscillators, each having a frequency corre-
sponding to one of the allowed modes. The entity which possesses the quantum
of energy (and the corresponding momentum) is the zero mass wavepacket
which we know as the photon. Thus we may regard an electromagnetic field
as a pltot\n gas. To proceed a step further, the electromagnetic interaction
between two charges is propagated by means of the exchange of uirtual photons
between the two charges. These photons are called virtual becauie their
emission and reabsorption would constitute a violation of the conservation of
energy were it not for the fact that they cannot be detected.
The uncertainty principle* permits a discrepancy in the energy of an
amount AÊ provided that this non-conservation lasts for a time interval no
greater than something of the order of hlL,E Thus, a charged particle can
continually emit photons of energy A.E and reabsorb them without violating
the conservation of energy since the whole process is undetectable . By means o-f
the same argument, a photon (not a virtual one) may be regarded as continually
creating and annihilating virtual electron-positron pairs. These two processes
are indicated schematically in Fig. 3-16.
When a virtual photon is emitted by one particle and absorbed by u
different particle, momentum is transferred between the particles. The time
rate of change of such a transfer of momentum is the force. There is a further
complication, however, since the exchange of photons must account for both
Virtual photon Virtual Ptrirs

_
e/\e .--.. Y"y
--G e*
(a) (b)
Figure 3.l6 (a) Emission and reabsorption of a virtual photon by an electron. (b) Creation
and annihilation of an electron-positron pair by a photon.
* See Section 9 of Chapter 2.

IOO THE CONCEPT OF THE NUCLEAR ATOM
\
\
\
){
(a) (b) (c)
Figure 3-17 Feynman diagrams depicting (a) the interaction of two electrons, (b) the Compton
effecr, and (c) electron-positron annihilation. A dashed line represents a photon and a vertex indicates
the absorption or emission of a photon.
attractive and repulsive forces. The interaction of two electrons by means

of the exchange of virtual photons is symbolized in the Feynman diagranr
shown in Fig. 3- I 7 (a). Neither the angles nor the lengths of the lines have any
significance, but the intersections (or vertices) represent the physical processes.
The Compton effect and (real) electron-positron annihilation are also shown
in Fig. 3-17.
In a similar manner, the gravitational force may be regarded as being
propagated by the exchange of virtual particles between gravitational masses.
Since this force (the weakest of known forces) has an infinite range, we expect
the exchange particle-which is called the grauiton-to have zero rest mass.
Although it is being hotly pursued at the present time,2s the graviton is still
undetected.
At the other extreme, the strongest force known is that between nucleons,
the so-called nuclear force which was mentioned in the previous section.
Unlike the weak electromagnetic interaction and the infinitesimally weak
gravitational interaction, the nuclear force requires an exchange particle having
a rest mass in order to account for its extremely short range. Yukawa2u pto-
posed a mass intermediate between that of the electron and the nucleon; hence,
the name mesln (intermediate) was given to this hypothetical exchange particle.
Subsequently, a variety of mesons was discovered, but it is now generally
believed that the pi meson or pion is the Yukawa particle. The charged pion
has a rest mass that is 273 times the electron rest mass. It is illuminating to
note how Yukawa was able to predict the mass of the nucleon exchange
particle. We saw in the last section that the range of the nuclear force is
b ,-, I.4 x l0-1s cm. If we take this to be the distance that the exchanged
meson must move, then the transit time is at least Lt ,-'blc. Since the emission
of a pion in the time At witl produce an uncertainty in the energy of the nucleon
of AE ,--, M,cZ, then from the uncertainty principle we have
LE , L,t ,-' Mobc ,-' h,
25J. Weber, Phys. Today2l,34 (1968) ; Ph2s. Reu. Letters 2l' 395 (1968).
26 H. Yukawa, Ph2s. Math. Soc. Japan 17r 48 (1935).
ANGULAR MOMENTA AND MAGNETIC MOMENTS l0t
Mo-*-+*, ,-'270 mo,
where 2, is the Compton wavelength for the electron. Note the reciprocal
relationship between the mass of the exchange particle and the range of the
force. This explains why particles of zero rest mass are requireJ for the
inverse square forces which extend to infinity. Conversely, u fo.ce with a
finite range requires an exchange particle with a non-zero rest mass.
There is at least one additional force which is involved in the so-called
"weak interactions" of particle physics. This is the force that is responsible for
nuclear p'decay, muon decay, and other lepton decay schemes. The exchange
particle involved in this force has not yet been discovered.
9. ANGULAR MOMENTA AND MAGNETIC MOMENTS
In elementary electromagnetism one learns that a current loop has

associatedwith it a magnetic moment whose magriitude is equal to the pioduct
of the current and the area enclosed by the loop, and whoie direction is per-
pendicular to the plane of the loop. Then, for a circular Bohr orbit of radius r,
we have
it:i xiA,
where 7 it
i., absolute units of current and the caret signifies a unit vector. But
A: nrz andi: -ril2ncr, where e is in esu. Then
e: e;
Ft: - :-fvL?:---l
2c 2mc
xF:-2*,el (3.34)
According to Bohr's postulate, the angular momentum t is quuntized and

must be equal to an integer times h, or 7 : *rhi. In particular, the angular
morhentum of the first Bohr orbit is given by *, l, o, 7 : h?. Then
-
+ehÂ
Ft: - 2-rf :- Haf. (3.3 5)
Here m is the mass of the electron. The unit vector I is included to show that
the magnetic moment is directed antiparallel to the orbital angular momentum
because of the negative charge of the electron. The quantity ltn is called the
Boltr magneton, and its value is approximately
eh
Fn : : 9.27 x l0-2r erg/gauss.
%nc
102 THE CONCEPT OF THE N UCLEAR ATOM
The ratio of the magnetic moment to the orbital angular momentum is called
the classical gyromagnetic ratio (o. the magnetomechanical ratio) and is
expressed as
rt : li'l 2mc
0 :Fr
(3.36)
l7l: h
In addition to its orbital momentum the electron also has an intrinsic

spin angular momentum which cannot be explained classically. Its existence
was postulated by Uhlenbeck and Goudsmit2T in order to explain the fine
structure of the hydrogen spectrum. It was later found that many of the details
of the multiplet structure of many-electron atoms immediately became clear
when spin was taken into account. Spin arises in a natural way in Dirac's
relativistic quantum mechanics,28 but it must be treated phenomenologically
in the nonrelativistic Schrodinger formulation which will be used in this book.
Therefore, we will postulate its existence and use it at will. Spin has a gyro-
magnetic ratio that is approximately twice the classical valud for orbital
moments. That is,
Ts: /'t e
(3.3 i )
SI mc
This value is called anomalous because it has no classical explanation.

Equations 3.36 and 3.37 are often combined by writing y : gef2mc. The
quantity g is called the spectroscopic splitting factor because of its effect on the
splitting of spectral lines under the influence of a magnetic field. In the case of
spin only, the g-factor may be taken as approximately 2, although its experi-
mental value is 2.0024. Since the g-factor is unity for orbital angular momenta,
one might expect it to have non-integral values for states which are mixtures of
orbital and spin angular momenta. Such cases will be discussed in section 4 of
Chapter B.
We have seen that spin is twice as effective as the orbital angular momen-
tum in producing a magnetic moment. Another peculiarity of spin is that it
can have hallintegral units of angular momentum, in contrast to the integral
multiples of fi required for the component of orbital angular momentum along a
given axis. Thus the spin of an electron is !h. Protons and neutrons also have
intrinsic spins of !h, while photons have one unit of spin, namely, ft. The
magnetic moment due to the spin of the electron is, however, also equal to one
Bohr magneton because of the cancelling factors of 2. Thus
lp'l : z" lsl :*r: Fa'
From the foregoing we note that the smallest unit of magnetic moment for the
electron is the Bohr magneton, whether one considers orbital or spin angular
momentum.
27 G. E. Uhlenbeck and S. A. Goudsmit, Naturwiss.13r 593 (1925).
28 P. A. M. Dirac, Proc. Rol,tal Sociefi,t (London) ll7, 610 (1928) and ll8' 351 (1928).
ANGULAR MOMENTA AND MAGNETIC MOMENTS t03
The intrinsic magnetic moment of the proton, which is also a spin- ] particle,
and the deuteron, which is a spin- l particle, were determined by Stern and
his co-workers from beam deflection experiments on hydrogen and deuterium
using the molecular species Hr, D., and HD.2e Later, more accurate values
were obtained for these nuclei by means of beam resonance and nuclear
magnetic resonance techniques. These methods have now been applied to
nearly all stable nuclei. The neutron magnetic moment was determined from
neutron scattering experiments in magnetized iron.s0
By analogy with the Bohr magnetorl we define the nuclear magneton as
L
en
FN : 2M"r:
5.050951 x l0-24 erg/gauss,
where the mass of the proton has replaced the electronic mass. The current
values of the proton, neutron and deuteron magnetic moments are:
Itrp : 2.792782p.'
Itrn: -1.9135Pr1-
Itra :0'8576Pr.''
Note that the deuteron magnetic moment is not the algebraic sum ofthe moment
due to the neutron and proton. This discrepancy is believed to be evidence for
a non-central contribution to the interaction between nucleons.
(u) By means of a classical calculation, estimate the average

flux density seen by an electron in the first Bohr orbit of
hydrogen due to the proton revolving about it.
(b) Assuming that the spin moment of the electron experi-
ences the magnetic field calculated in (u), find the
energy required to flip the spin direction.
(.) What is the frequency of the photon having this e nergy ?
PROBLEM 3.19
What is the magnetic field seen by an electron in the first

Bohr orbit of hydrogen due to the intrinsic spin moment of
the proton ?
2e For references and an excellent summary of the beam experiments, see N. F. Ramsey, Molecular
Beams, Oxford University Press, N.Y. (1956), p. 102 cf.
s0 L. W. Alvarez and F. Bloch, Phys. Reu.57,
lll (1940).
t04 THE CONCEPT OF THE NUCLEAR ATOM
The intrinsic spin angular momenta and associated magnetic moments

of the elementary particles have provided the strongest arguments against the
existence of electrons in the nucleus. The electron, having a magnetic moment
roughly 1000 times greater than any known nuclear moment, could hardly
remain undetected within a nucleus. Furthermore, it is known that nuclei
with even A and odd Z and i[ have integral spin, whereas if such a nucleus
contained ,4 protons and y'/ electrons its spin would have to be half-integral.
On the other hand, the'nuclear shell model accounts for all known nuclear
spins by regarding the nucleus to be comprised only of nucleons.
IO. THE LARMOR THEOR.EM AND THE

NORMAL ZEEMAN EFFECT
Let us now consider the effect of a weak magnetic field on an electron

performing circular motion in a planar orbit such as that discussed in the pre-
vious section. In the classical treatment there is, of course, no electron spin,
so our model simply consists of the interaction of a classical magnetic dipole
y', (arising from the orbital angular momen tum V) with the applied magnetic
field, A, measured in gauss. Assume that the orbital angular momentum
vector is oriented at the angle 6 with respect to the z-axis along which the
magnetic field is applied (see Figure 3-18). The torque on V is given by
i, x 9-, which is directed into the plane of the page in the {-direction. Since,
from elementary mechanics, the torque also equals the rate of change of the
angular momentum, we have
df
dt
== i,xA:y,/xA.
ButldTl: / sin 0 d.4, so we may write the scalar equation
I sin 6.4!
dt
: ytl # sin 0.
Defining the precessional velocity by r'ta: d$ldt, we obtain the result
0)a -- ytfi :t# (3.38)
The angular velocity a" is often called the Larmorfrequenqt.

An important consequence of this, which is known as Larmor's theorem,sl
may be stated as follows: the effect of a weak magnetic field upon an atom can
be removed (to first order in 9) by a mathematical transformation to a rotating
ttJ.H.Van Vleck, The Theory d Electric and Magnetic Susceptibilities, Oxford University Press,
London, 1932, p. 22.
LARMOR THEOREM AND THE NORMAL ZEEMAN EFFECT t05
Figure 3-18 The precession of the angular I

momentum vector as a result of the torque produced
by the action of a magnetic field on its associated I
magnetic moment.
coordinate system. Thus, if the energy levels of an atom are known (with no
applied field) and then a magnetic field is turned on, a transformation to the
coordinate system rotating at the Larmor frequency will restore the original
field-free problem. It follows that the solution of a problem involving a static
magnetic field can be written approximately as the superposition of the no-
field solution and a rotation at the Larmor frequency. The exact expression
which shows the quadratic term in the field is considered in Problem 3-20.
Using the Planck relation, the energy associated with the Larmor frequency is
L,E: Iconh:'-2mc
*tlg -- +I Iua%, (3.3e)
where the signs refer to the sense of the rotation. This energy difference
should be recognized as the potential energy of a magnetic dipole whose
moment is one Rohr magneton, since the dipolar energy is given by
L,E : _i, .d
Hence, the plus sign (higher energy) in Equation 3.39 corresponds to anti-
parallel alignment (as in Figure 3-18), while the minus sign (lower energy)
indicates parallel alignment. If the energy of an electron having a moment
p-a is Eo with no applied magnetic field, then it can take on one of the energies
Eof FaCI in the magnetic field #,provided that one can neglect quadratic
terms in 0. It was found experimentally that in a collection of identical atomic
systems of the type we have described above, a magnetic field actually produces
a triplet of levels (called a Lorentz triplet) whose energies are{Eo, Eo * Ita%\.
The existence of these levels can be confirmed by studying the optical tran-
sitions to and from these states. This phenomenon is known as the normal
Zeeman tfttt.,, The reason for the appearance of three levels will become
apparent in the next section when we discuss spatial quantization' There we
will see that the three energies correspond to the three allowed projections of the
sz P. Zeerr,an, Phil. Mag. 5, 43, 226 (1897).

angular momentum. However, the classical explanatio-n invoked the Larmor

fre.lrrercy to account for the two shifted levels, and the unshifted level was
attributed to the component of the motion parallel to the magnetic field,
for which there would be no interaction.
The Zeeman effect is really more complicated than the above classical
model would lead one to believe. We now know that electrons have spin and
that this spin also has a magnetic moment associated with it. Thus, when a
magnetic held is applied, both the spin and the orbital angular moments will
find themselves actid upon by precessional torques. The resulting energy level
splittings, howeve., .un.rot bo interpreted from the classical theory because
oi the inomalous half-integral spin (in units of h) of the electron and the spin
g-value which is twice the classical value . As a consequence of this inexplicable
fiehavior, the more general Zeeman effect, including spin,, was historically
misnamed the "anoriralous Zeeman effect" in contrast with the "normal"
effect, which is the less common variety. This topic will be discussed in con-
siderable detail in section 4 of Chapter B.
Before leaving this classical model the reader should note that the torque
produced by the magnetic field acting on the magnetic dipole does not align
ifr. aipole with the field. Instead, the torque causes the dipole to precess in
such i *uy as to maintain its initial polar angle 0 (see Figure 3-tB). In an
ideal system (having no dissipative forces) this precession would continue as
long as the neia *.ri to remain. Hence, it would be impossible to magnetize a
pieJe of iron in such an ideal system, however gre.at the applied field; the
greater the field, the more rapid would be the precession. In the real world,
f,o*.rr.r, some sort of "viscouJ" damping always exists. As energy is dissip?t:d
to the lattice, the polar angle of the precession decreases at the expense of the
dipolar energy until complele alignment of the dipote occY.rs.. The time interval
required for-co-plete align-..rf is known as the longitudinal relaxation time, or,
in ihe case of a dipole moment due to spin, it is called the spin'lattice relaxation
time. These relaxation times play an important role in magnetic resonance
phenomena as well as in other areas of contemporary physics.ss
PROBLEM 3.20 -t:;ilr':l:ji+;::;r;,:.il;l::rr :rl:riirflri.;r-;;.;l:;5r':l:r::i,rl :rrii$::iri:rr:
(u) For an electron moving in a circular orbit about a

nucleus of charge e in a magnetic field d, set up the
equation of motion using the Lorentz force,
F:e@+L;"h1. c
(b) Slrow that o is given approximately by- t I l2mc for

small magnetic fields by estimating the relative magni-
tudes of the neglected terms.
3s See, for example,C. P. Slichter, Principles of Magnetic Resonance, Harper and Row, New York,
Ig63; also, M. Sparls, Ferromagnetic-Relaxation Theory, McGraw-Hill, New York, 1964.
SPATIAL OUANTIZATION 107
I t. SPAT|AL QUANTTZATTON
The classical model of the previous section is often useful for describing
certain aspects of resonance and relaxation experiments, but at the same timi
it can be misleading if taken too seriously. For example, the picture of a pre-
cessing angular momentum vector relaxing through a continuum of pôlut
angles conflicts with a fundamental concept of quantum mechanics which is
referred to as spatial quantization. Stated briefly, spatial quantization means
that the projection of an angular momentum vector along any single axis in
space can be only one of a discrete set of allowed values. At a given time there
can be only one axis of quantization; if it is altered, szy, by changing the
direction of an applied uniaxial magnetic field, quantized states ."irt-o.rly
along the new field direction. The close connection between space and angular
momentum will be elaborated further in section 3 of Chapter 7.
The experiment proposed by Stern and performed by Gerlach and Sternaa
provided the first conclusive evidence of spatial quantization in 1922. Silver
atoms were evaporated in an oven and collimated into a narrow beam which
was passed through an inhomogeneous magnetic field as shown in Figures
3-19 and 3'20. Although a uniform field produces no net force on a magnetic
dipole (only torque), a non-uniform field can deflect a dipole with I net
translational force. This can readily be seen by considering a magnetic fieId, g
which acts in the positive z-direction, and which is also designed to have a
gradient in the positive z-direction (FiS. 3-19). Assuming that all other
derivatives of the magnetic field are zero, we find that the translational force
on a dipole, oriented at an angle 0 with the z-axis, is
i; : -iE : ifi. si : i(pscos o)
as ^
-l.rcos0-;-k,
oz
Fz: /rcos0 as
^ .
oz
Thus, a dipole aligned with the field is acted upon by an upward force of
magnitude p'(1CI102), while a dipole aligned antiparallel to the field is acted
uPon by u downward force of the same magnitude. For arbitrary orientations
the force could have any value between
-pQola) and + p,egaloz).
In the absence of spatial quantization, the beam of particles would contain
a continuous distribution of angular orientations of the precessing dipoles, and
the action of the non-uniform magnetic field would spread the narrow beam
into a band at the detector screen. However, instead of a continuous band,
Gerlach and Stern obtained two distinct lines whose breadth was due to the
spread in particle velocities rather than to a continuum of dipole orientations.
This can only be explained if the dipoles are permitted just two orienta-
tional states, one state having a component in the { z-direction and the other
3a O. Stern, Z. Pltsik 7 r 249 ( l92l ) ; W. Gerlach and O. Stern, Z. Ph2sik B, I l0 and 9, gag 0922) ;
Ann. Ph2sik 74,673 (1924).
Figure 3-l9 An inhomogeneous magnetic field having a large

gradient in the field direction.
W, N
deflected toward the

having a component in the-z-direction. The former aredeflected
most intense region of the field while the latter are toward the
weakest region.
The quantization condition underlying the Stern-Gerlach experiment
may be staled as.follows: The allowed orientations of the angular momentum are such
dfft, successiuely b2 one unit of h. Thus,
thai tlwir projections on the axis of measurement
if the *L*i*rrrn projection is lft, along the z-axis, there are three allowed
orientations corresponding to z-components of 1,0, and -lh. In general, if
the maximum component is fh, there are 2l * I allowed orientations of the
angular momentum and its associated magnetic moment. Note that if the
*u*i-rr- component of the angular momentum is an integer there will
always be an odd number of allowed orientations, and hence there should be an
odd number of lines on the screen. However, fot silver only two symmetric
lines were observed in spite of the fact that the magnetic moment was found to
be one Bohr magneton. Tn 1927, Phipps and Taylor repeated the experiment
with hydrogen and again found just two lines.35 The explanation, of course,
is that the moments observed in these two cases were due to spin and not the
orbital angular momentum. Because of its anomalous g-value, a spin projection
of $h has a magnetic moment of one Bohr magneton. However, in order to
satisfy the quantization condition, atoms with half-integral spin cannot have
zero proj.ciiotrs, so that the only allowed states correspond to +hh. Hence,
2(+) + | : 2, which accounts for the two lines in the original experiments.
Thns, the experiment which first demonstrated spatial quantization also showed
the exist.r.i of intrinsic spin angular momentum, although it was several
years before the latter was recognized.
If the above experiment is repeated with particles having one unit of
orbital or spin angulir momentum, the beam is split into three distinct beams
as required by spatial quantization. It is natural to inquire what would
Collimator
( -t I Screen
Figure 3-20 Particle trajectories in the experiment of Gerlach and Stern. The deflections of
the beams are greatly exaggerated.
36 T. E. Phipps and J. B. Taylor, Phys. Reu.29, 309 (1927)'

SPATIAL OUANTIZATION t09
Detecting
screen
Beam of
particles
of
angular
momentum
)h
Screen
,h:tl:,,
Apparatus for verifying the quantum states of particles having one unit of angular
,.J","t:;.t-t,
happen if each of these three beams were, in turn, subjected to another appar-
atus identical with the first. In Figure 3-21 three such beams are shown
entering a second magnet with the same field orientation as the first. If only
one beam at a time is permitted to enter the second magnet, just one beam will
appear at the detecting screen. Hence, no further splitting will occur, regard-
Iess of which of the three beams is allowed to enter the second apparatus.
The second magnet produces an additional deflection of the upper and lower
beams from which one can verify that the magnetic moments of the particles
which comprise them correspond to + 1t,u and - Fa, respectively. The center
beam is undeflected and thus corresponds to zero projection of the magnetic
moment along the vertical axis. From this result we draw the following con-
clusions. The first magnet establishes an axis of quantization and sorts the
initial beam into three quantum states. The second magnet does not alter
these quantum states but serves to confirm that the sorting of the beam into
pure states has been preserved during the transit of the particles from one
magnet to the other. On the other hand, if one were to alter Figure 3-21 by
rotating the second magnet through an angle about the beam axis, the situ-
ation would be quite different. Each beam from the first magnet, which was
a pure state in the first magnetic field, now appears as a mixture of the three
quantum states in the new field orientation. Accordingly, if one beam at a
time from the first magnet is allowed to enter the second, each will split into
three components in the second apparatus.
PROBLEM 3-21
(u) A beam of silver atoms is obtained by heating silver to

a temperature of 2300"K. Assuming a Maxwell distri-
bution of velocities, what is the most probable velocity
of a silver atom ?
(b) Calculate the angular deflection of this beam in a
Gerlach-Stern apparatus for a path length through the
inhomogeneous magnetic field of 4 cm and an average
field gradient of 5 x 104 gauss per cm.
il0 THE CONCEPT OF THE NUCLEAR ATOM
su/ylMARY .*,iiii:r
The Rutherford-Bohr atom is a dynamical atom in which nearly all of the

atomic mass is concentrated in a nucleus having a diameter of about lOF
(^-,10-4 A). Charge neutrality is achieved by placing as many electrons in
orbits about the nucleus as there are protons in the nucleus. The orbital radii
and velocities of these electrons are such that the Coulomb attraction provides
the appropriate centripetal acceleration to achieve dynamical stability. Bohr
postulated that the orbital angular momentum associated with an electron is
quantized in units of fi : hl2z', where ft is Planck's quantum constant. This
restriction on the possible values of the angular momentum results in an in-
finite but discrete set of allowed orbits and energies for each electron. Since the
diameter of an allowed orbit is of the order of I to l0 A, the "size" of an atom is
roughly 105 times the size of the nucleus.
The Bohr theory predicts the energy levels of hydrogen and hydrogen-like
atoms reasonably well, although serious discrepancies arise for non-hydrogenic
atoms. An additional postulate of Bohr, namely, that the absorption or emission
of a photon by an atom occurs when an electron makes a transition between two
of the atomic energy levels, was an important link betwe e n the quantum theory
of radiation and the quantum theory of matter. The magnetic moments
associated with orbital angular momentum and with intrinsic particle spin are
derived and the normal Zeeman effect is discussed. The phenomenon of spatial
quantization, that is, that the angular momeutum calt have only discrete pro-
jections along a given axis, is illustrated by means of the Stern-Gerlach experi-
ment.
REFERENCES
American Institute of Physics, Nuclear Structure, Selected Reprints, 1965.
A. Beiser, Perspectiues of Modern Ph-ysics. McGrarv-Hill Book Co., Inc', New
York, 1969.
C. H. Blanchard, C. R. Burnett, R' G. Stoner, and R. L. Weber, Introduction
to Modern Ph\sics,2nd ed.Prentice-Hall, Inc. Englewood Cliffs, N.J., 1969.
M. Born, Atomic Ph2sics,6th ed. Hafner Publishing Co', New York, 1959.
R. M. Eisberg, Fundantentals of Modern Physics. John. Wiley and Sons, Inc.,
New York, 1961.
L. R. B. Elton, Introductory Nuclear Theor2. W. B. Saunders Co., Philadelphia,
1966.
Robley D. Evans, The Atomic Nucleus. McGraw-Hill Book co., Inc., New York,
1966.
Richard P. Feynman, Robert B. Leighton, and Matthew Sands, The Feynman
Lectures on Ph2sics. Addison-wesley Publishing co., Inc., Reading, Mass.,
ton, principles of Modern ph2sics. McGraw-Hilt Book co., Inc., New

". "]l??it
York, 1959.
Phltsics,6th ed. McGraw-Hill Book Co., Inc., New York, 1969'
Henry Semat, Introduction to Atomic and Nuclear Ph2sics,4th ed. Holt, Rinehart
and Winston, New York, 1969.
Paul A. Tipler, Foundations of Modern Ph2sics. Worth Publishers, trnc., New
York, 1969.
F. W. Van Name, Jr., Modern Ph4sics,2nd ed. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1962.
Hugh D. Young, Fundamentals d Optics and Modern Physics. McGraw-Hill Book
Company, New York, 1968.
GHAPTER 4
Wffiffi ffiffiWffiffiffiWWffiffiW ffiW
I. INTRODUCTION
In this and the ensuing chapters we shall develop and use the formalism
of non-relativistic quantum mechanics as it is understood today. We begin
this chapter with a brief aside on the "old" quantum theory and its application
to the hydrogen atom. Although this material may be omitted without
hampering the reader's understanding of the current theory, it is included here
for the following reasons. It represented a forward leap in that it attempted
to incorporate quantum concepts into current theories. It correctly predicted
a large body of experimental results from a few simple rules. It is of consider-
able historical importance because it occupied many of the greatest minds, and
like the ether theory of pre-relativity days, it thus set the stage for the appear-
I
ance of the modern theory.
The new quantum mechanics appeared in two forms which were later
shown to be equivalent. One form is known as waae mecltanics and is generally
regarded as Schrcidinger's formulation,l although it is based heavily on the
work of de Broglie.z The other approach is called matrix mecltanics and it is
chiefly credited to Heisenberg, although Born and Jordan shared in its develop-
ment.8 Since wave mechanics is more easily grasped intuitively we shall begin
our study with the development of the de Broglie theory of particle waves and
the evolution of the Schrcidinger equation. All of Chapter 5 is devoted to
applications of the Schrodinger method to one-dimensional systems. The
essentials of matrix mechanics will be introduced in Chapter 6. In the re-
maining chapters the wave and the matrix formulations will be used freely
and interchangeably.
1E. Schrcidinger, Ann. Ph2sik (4)79,361 (1925); 79,489 (1925) ; 80,437 (1926); 8l' 109 (1926).
sL.de Broglie,Naturell2r540(1923); Thesis,Paris(1924); Ann.Ph2sique (10)2(1925).
3W. Heisenberg, Z. Physik33,B79 (1925); M. Born and P. Jordan, Z. Physik 34' B5B (1925);
M. Born, W. Heisenberg and P. Jordan, Z. Ph2sik 35r 557 (1926).
ilt
il2 THE DEVELOPMENT OF WAVE MECHANICS
2. THE OLD QUANTUM THEORY:

WILSON-SOMMERFELD QUANTIZATION RULES
Wilsona and Sommerfelds independently discovered a method ofquantizing

the action integrals of classical mechanics, and this method was subsequently
applied to a number of physical systems. A necessary condition for the appli-
.iiion of this method is that each generalized coordinate q* and its conjugate
momentu m p* must be periodic functions of time. Then the action integral
taken over one cycle of the motion is quantized; that is,
r- oor : n*h. (4. 1)

$
To illustrate the method, consider a one-dimensional simple harmonic oscillator

whose equation of motion is
mx +kx:0,
or
i +q'x -0,
where
k
692:-
m
Then,
x: xo sin cull
and
!* : mi -- mlJxo cos a-rl'
Equation 4.1 becomes
nh: or : *,'rfr cosz at dt
fn hq-
Ii
I .t
: maxil cos'odo
Jo
marxfi.
Therefore,
ua-
oflh
'rO- ,
@m7T
or the amplitudes are quantized. The energy states are
E:T+y:gmxzalkxz
: lmazxfi,
of,
En: nath : nltY., (+.2)
a W. Wilson, Phil. Ma* 29r 795 (1915).

5 A. Sommerfeld, Ann. Ph2s,5lr l (1916).
THE OLD EUANTUM THEORY lt3
In both the classical theory and the old quantum theory, the ground
state energy of an oscillator is incorrectly given as zero. Howcver, th; level
spacings are correct in th.ese older theories. From the oscillator energy levels
obtained in Equation 4.2, there is no information about which transitions are
most likely to occur, or in fact, whether any are forbidden. Information of this
kind goes under the general heading of selection rules and is readily obtained in
the new quantum mechanics. However, in the old theory, selection rules were
inferred by comparing the system with the behavior of a classical system;
that is, by employing what is called Bohr's correspondence principle. Thus, since
a classical oscillator will emit only one frequency (and no harmonics) , if a
quantum mechanical oscillator is to correspond to the classical result in the
limit of large z, then we must have the selection rule Az : f l. We had
already assumed transitions between adjacent levels in our discussion of
Planck's oscillators in section 3 of Chapter 2.
If we treat a two-dimensional harmonic oscillator as two independent
one-dimensional oscillators in the x- and 2-directions, the energy levels are
En,,nu : h(nro, * nnan).
If the oscillator is isotropic (k, : kr), then @a : @a : a,,, and
En : nhat,
where n:7t,*nn. But z: I now corresponds to the two states (n,:0,

fla: l) and (nr: lrftn:0), which have the same energy. The two states
are said to be degenerate. In general, the level of energy En is (, + l)-fold
degenerate. Similarly, the energy of the nth level of a three-dimensional iso-
tropic oscillator is also given by Equation 4.2 with ft : fl, I no * n,. The
degeneracy in this case is $(z + 1)(n + 2).
Duane and Comptond applied the Wilson-sommerfeld method to a
corpuscular model for the diffraction of ,r-rays by . crystal. If the z-direction
is normal to a set of identical atomic planes of separation d, then the quantum
condition becomes
:fo"o, :lud:nh, (4.3)

$o"o'
in the absence of forces. A photon of momentum ltl), incident at the angle g,
as shown in Figure 4-1, will be reflected at the same angle and will transfer an
b
Figure Cl Bragg reflection of a photon I
by a crystal.
Atomic planes
6 W. Duane, Proc. Nat. Acad. Sci. 9n l58 (1923); A. H. Compton, ibid.91 359 (1923).
il4 THE DEVELOPMENT OF WAVE MECHANICS
amourrt of momentum in the z-direction equal to (2hl),) sin 0. But the momen-
tum must satisfy the quantum conditions given by Equation 4.3, so we have
nA:2dsin0, (+.+)
tlte well-known Bragg equation.

One can obtain the equivalent Bragg expression for electrons by letting
:
!" 2mu sin 0 in Equation 4.3 to obtain
n!:2dsino.
ma
(4.5)
By comparing Equation 4.4 and Equation 4.5, it is apparent that the de

Broglie wavelength, ). : ltf mu, could also have been predicted by this theory.
PROBLEM 4-l ;i': ';l:1:',J
Use the Wilson-Sommerfeld method to obtain the energy

levels of a rigid rotator of angular momentum, Po : Ia,
where ,I is its moment of inertia about the rotation axis.
(Ans.i En:n2h.2121.)
PROBLEM 4.2
Use the Wilson-Sommerfeld method to obtain the energy

states of a perfectly elastic particle in a cubic box of edge a
and with perfectly rigid walls.
(Ans. : En : nzhzl\maz, where nz : n? + ,l + n2.)
PROBLE/I4 4.3
Find the energy states of a perfectly elastic ball bouncing

in the gravitational field by applying the Wilson-Sommerfeld
quantization condition.
(Ans. i En : (9gzlt2mn'zl32)1.)
:ii;r,,,,,1;;:.:lt: ,;;r;ii .;.,, r;".;;;r.,*.r';:,:'; .i.r,li1:+"i:,;i+,i:iii:,: ,:i:rr,i:i;tifl.;l;at+,:
3. SOMMERFELD'S RELATIVISTIC THEORY OF THE

HYDROGEN ATOM
Although the Bohr theory was quite successful in predicting the spectrum
of hydrogen, there remained an unexplained fine structure or splitting of the
lines. This splitting amounts to about one part in 104 and cannot be seen in
spectrometers of low resolving power. Sommerfeld proposed that if elliptical
SOMMERFELD'S RELATIYISTIC THEORY il5
Figure 4-2 Relationship between

the planar and spherical coordinates.
The angle p is the azimuthal angle in
the x'-y'-plane, / is the azimuthal angle
about the z-axis, and 0 is the polar
angle measured from the z-axis.
as well as circular orbits were allowed, the electron's velocity in an orbit of

large eccentricity could become relativistic. He showed that the energy
correction resulting from the relativistic treatment was of the proper order of
magnitude to account for the fine structure splitting of hydrogen. Unfortu-
nately, Sommerfeld's theory is not the correct e*planition for the fine structure
of atomic spectra. We now know that the fine structure is due to the fact that
the electron has an intrinsic spin angular momentum which produces the so-
called spin-orbit interaction. This will be treated in detail in the discussion
of the quantum mechanical theory of radiation. In spite of the shortcomings
of the Sommerfeld theory, it is of such great historical importance that it is
worth sketching briefly here.
Consider an electron of reduced mass pt to be revolving about a fixed
nucleus. Since the motion in a given orbit is confined to a plane, we will
describe the position of the electron by the planar coordinates (r, B) measured
in the plane of the orbit. In Figure 4-2, the x',2'-plane is the plane of the orbit
and the unprimed system is used as the frame for a set of spherical polar
coordinates. The relationship between the planar angle and the spherical
coordinates is obtained by considering an infinitesimal angular displacement
dP. Then the arc r dp is the hypotenuse of an infinitesimal spherical triangle
whose legs are r d0 and r sin 0 d$. Then we have
(, dF)' : (r d?)z + (r sin 0 d.+)', (4.6)
which we will need later in our discussion of the quantization rules.

The equation of the motion in the x'r1t'-plane is
Zez
lti : pr p2
-,,
I16 THE DEVELOPMENT OF WAVE MECHANICS
Using the fact thatpO: przP : /: constant, we obtain
.. {2 Zez
pr:Tr"- ,r.
Multiplying by i and integrating,
u!,ar:l(#-#) *,
or
iprr:-L*tt'-rF
2pr' ' r
-r r'' (4'7)
The constant of integration E is the total energy of the system. In order

to get the path equation we eliminate the time by means of the substitution
: /'L'
':!/f dt dP'
dP prz
Substituting this into Equation 4.7 and multiplying by 2pl/'results in
(ldrY_ 1
,2p,Ze2 ,2pE
\*oBt 7- r', - f''
Introducing z : 1lr,
du Idr ,I o 2p,Ze2
u-,2pE
dP:-FaB:-V-u'r ,"'z /'
and
du
rd0: l2tL + 2t Zt' tt
l-i -t, - uz
lntegrating,
u :!r : rz:'
dz +.\ry!
'! f4 +ry.sin
' lz (B - oi, (4.8)
for negative total energy. The equation of an ellipse of semiaxes 4 and b,

eccentricity e and orientation po is
I 1*esin(p-pr) : a {a'z-62-.,n
It:1:ffi *+ n, sin(B - 0o),
where bfa: \/l - e'z. Comparing this with Equation 4.8 we immediately
SOMMERFELD'S RELATIVISTIC THEORY lt7
obtain the following results:

Zez
a- -28
{
b: (4.e)
{ -zpn
2Ef2
I -e2
P'Zzst'
PROBLEM 1-1
Show that the time averages of 7" and Z satisfy the same
relation as that satisfied by circular Bohr orbits, namely,
E:+V:-f.
Applying the quantization condition given in Equation 4.1 to the planar

angle and the three spherical coordinates, we have
n oo : ktt (a.loa)
$
f
Q 1,, d4 : mh (4.10b)
J
Q pu d0 : neh (a.10c)
J
r
a, : n,lt
! f,
(4.10d)
Equation 4.10a can be integrated immediately since Pp: /: corlstant, and

the axis about which / is measured is fixed in space. Then,
!p: /: kh. (a.11a)
Similarly, lo it a constant and the z-axis about which it is measured is fixed, so
Po : mh. (4. 1 Ib)
The quantum number m is called the magnetic quantum number because of the
role it plays in distinguishing the energy levels of the atom in the presence of a
magnetic field. Since the axis about which P, is measured is neither unique
nor stationary, we must transform Equation 4.10c before it can be integrated.
If Equation 4.6 is multiplied by pldt, we obtain
prrp dp : Ltrz sinz 0 ' E d+ | pr20 d0,
": ped;:podi+pedo.
Substituting this into Equation 4.10c,
: : (k -
f tn dP - lod6)
ltoh m)h,
of,
k -- ,u * m. (a.11c)
The integer ,t is called the azimutltal quantum number. It can take on the
values lr213r. . ., with zero excluded.
Sommerfeld's integration of Equation 4.10d will not be repeated here but
we will merely state his result, namely,
kh(-L-
\1/l e2
r)/ :n,h,
-
or
! :Y! k :: k'. (4.11d)
b- k
The quantity z in the last expression is called the total quantum number, since it
is defined as the sum of the radial and azimuthal quantum numbers, which
is to say that
tL:nr*ne*m.
Combining Equations +.9, +.11a, and 4.11d, we obtain the equalities
a nh Zez l- t,
b:7:- /lE'
from which the quantized orbits and energies are given by:
nzao
O:2,
.
h-_
ka knao
u--t
NL
and (4.r2)
LooZ'
un-
H
)
nZ
where the constants ao and w, were defined in section 4 of Chapter 3. Note

that the semimajor axis, a, is the counterpart of the radius of a circular orbit
SOMMERFELD'S RELATIVISTIC THEORY il9
n:2
k:1
b : 2a,,
n:3
k: I
b: 3a,,
n:3
k:3
Figure 4-3 Sommerfeld's orbits for n : l, 2, and 3.
in the Bohr theory. Furthermore, the energy of an orbit is independent of

the value of the semiminor axis. This means that all of the orbits of different
eccentricities associated with the same n value are degenerate. A few of these
orbits are depicted in Figure 4-3.
It was at this point that Sommerfeld introduced a relativistic correction
for the mass of the electron. For an orbit of large eccentricity the electron
would pass close to the nucleus at a very high velocity such that the relativistic
increase of mass would become noticeable. With this refinement to the theory
the revised energies become
woZ'f . oz/z ll
En---;l'--\Z- *)l (4.13)
The quantity
e2
h.c
: 7.297 x l0-3 ^,Ir37
is called the fne structure constant since the term in Equation 4.13 in which a
appears, correctly accounted for the fine structure splitting of the lines of the
hydrogen spectrum.
Although the old quantum theory achieved many successes in atomic and
molecular spectroscopy, it was an incomplete theory in the sense that none
of its recipes for quantization were derived from first principles. Since it
THE DEVELOPMENT OF WAVE MECHANICS
could not be applied to aperiodic systems, most collision and scattering

problems were beyond its pale. Furthermore, it contained errors, contradictions,
and ambiguities.? It did have some virtues, however. It predicted a large
body of experimental results from a few simple rules, and it set the stage for the
new quantum mechanics which soon replaced it. We will now proceed to
discuss the wave mechanics of de Broglie and Schrodinger.
4. THE WAVE NATURE OF PARTICLES
The idea of associating both a wave and particle nature with the electron
was first proposed by de Broglie in his doctoral thesis in 1925.8 His-work was
motivatea ny the mystery of the Bohr orbits, which he attempted to explain
by fitting a standing wave around the circumference of each orbit. Thus,
de Broglie required that n). : 2rr, where 2 is the wavelength associated with
the nth orbit and r is its radius. Combining this with Equation 3.25 we immedi-
ately obtain the result that
hh
"1-_-_
- ml, P'
Assuming the existence of a natural symmetry in the properties of matter and
energy, he proposed that a material particle of total energy E and momentum I
must be accompanied by u phase wave, analogous to that ascribed to the
photon, whose wavelength is given by I : hlp and whose frequency is given
by the Planck formula, y : Elh. The Planck and de Broglie relations may be
expressed in the useful forms,
E:ha
and (+.r4)
!:hk'
where h. : hl2r, k :2rf )', and a-l :2nv.
The physical nature of such a particle wave was not clearly described by
de Broglie. Unlike a classical wave, the energy E of the particle wave is not
thought of as spread out over the extent of the wave, but is regarded as localized
with the particle. However, the. accompanying wave is essential in order to
account for the phenomena of interference and diffraction.
The concept of the de Broglie wavelength is one of the cornerstones of
modern quantum theory, and the simple relationship
1-- ph
holds for photons as well as for both relativistic and non-relativistic material
particles, provided that the appropriate expression for p is used. On the other
?
Albert Messiah, Quantum Mechanics, North-Holland Publishing Co., Amsterdam (1958),
Chapter l.
E
L. de Broglie, Ann. Ph2s. (Paris) 3,22 (1925).
THE DIFFRACTION OF PARTICLES t2t
hand, the de Broglie frequency has not been a very useftrl concept and it
comes into play only in the calculation of the phase velocity. Here a dis-
tinction appears between the relativistic and non-relativistic cases. If the rest
mass energy is included in the total energy E (as in de Broglie 's treatment),
then the phase velocity of the wave becomes
or'
t*) : cz. (4.15)
In obtaining Equation 4.15 we have expressed the relativistic momentum as
P : ymou, where , : Ft is the particle velocity and y : (l - 1lz\-tz. From
our knowledge of waves* we identify the particle velocity u with the group
velocity of the wave packet. Since special relativity requires that a be less
than c, we note that Equation 4.15 calls for phase velocities greater than c.
However, as no energy (that is, no signal or information) is transmitted at the
phase velocity, the fact that u > r constitutes no violation of the postulates of
special relativity.
In non-relativistic quantum mechanics the rest mass term is neglected
and the total energy E is merely the sum of the kinetic and potential energies.
Accordingly, the phase velocity of the wave associated with a non-relativistic
free particle (Z : 0) is,
*- @Ephka
k p 2m 2m-2'
that is, one-half of the particle velocity. Thus the phase velocity is not of any
physical significance. The group velocity of a particle wave, however, is given
by
da dE 1.
"-dk-kdk
5. THE DIFFRACTION OF PARTICLES
Whether or not particles of a given momentum will exhibit their wave

characteristicswill be determined by the relative magnitude of their de Broglie
wavelength in comparison with the physical dimensions of the environment
in which they are found. For wavelengths that are much smaller than the
dimensions of apertures and obstacles, diffraction and other wave effects are
not ordinarily observed. In such cases we can assume rectilinear propagation
and problems can be treated by means of ray diagrams (for example, visible
light in our everyday world). However, for wavelengths which approximate
or exceed the dimensions of objects, diffraction effects become quite important
and ray diagrams become meaningless (for example, audible sound in our
everyday world). In order to get some insight into the kinds of behavior to
* See Chapter 2, section B.
t22 THE DEVELOPMENT OF WAVE MECHANICS
TABLE 4-l de Broglie Wavelengths

Associated with Certain Particles.
t44 volt electron IA

I volt electron 12 A
I volt proton 0.29 A
I volt alpha particle 0.15 A
expect from particles, the de Broglie wavelengths for a few special cases are
given in Table 4-1. Massive particles such as a bullet fired from a 0.22 rifle
or a fast baseball have de Broglie wavelengths of the order of l0-23 or 10-24
angstrom.
Since the wavelengths associated with 10 to 100 volt electrons correspond
to the atomic spacings of most crystalline solids, it was conjectured that
electrons of these energies ought to be diffracted by crystals in the same manner
as x-rays. This was confirmed experimentally in 1927 by the work of Davisson
and Germer,e in which they scattered low energy electrons from a nickel
crystal. Although electrons were scattered in all directions from the nickel
crystal, a distinct peak in intensity occurred at an angle which corresponded
to the first Bragg reinforcement for a wavelength of 1.65 A. The size of this
peak varied with the energy of the incident electrons and was found to reach
a maximum of intensity for 54-volt electrons whose de Broglie wavelength is
1.67 A! (See Figure 4-+.)
The Bragg equation, which gives the condition for intensity maxima in
the reflected beams, may be expressed as
n). : 2d sin $ : 2d cos 0, (4.16)
where z is the order of the diffraction, d is the distance between the reflecting
planes, and the angles 0 and $ arc shown in Figure 4-5. For a mono-energetic
beam of particles at normal incidence, the first order maximum will be found
at an angle from the normal given by
20 :2 cos-'h.
The higher ordered reflection maxirna occur at progressively smaller angles

from the incident beam.
44 eY 48 eV 5'1 eV 64 eV 68 eV
0o r: A
t.AS 0o r: t.ZZ A 0o I: 1.67 A 0' r: t.ss A 0' l,: t.+S A
Target
Figure4{ Angular plots of scattered intensity for low-energy electrons incident upon a nickel
single crystal. The energies and de Broglie wavelengths of the electrons are given for each plot.
(From C. Davisson and L. H. Germer,Phys. Rev.30,705 (1927). Used with permission.)
e C. Davisson and L. H. Germer, Ph2s. Reu.3Or 705 (1927).

THE DIFFRACTION OF PARTICLES t23
Incident beam
Reflected beam
Figure
s
G5 Bragg reflections of electron waves or x-rays for normal incidence on a crystal.
In addition to the,Bragg reflection, the electron waves should undergo

refraction at the crystal surface because of the abrupt change in potential
energy at the surface. The index of refraction may be express.d as,
F :7, L
where the prime refers to the wavelength inside the crystal. For electron
energies that are small with respect to the rest-mass energy, the total energy
and the momentum are related as follows:
E-4*v. 2m
Then,
p:{2m(E-V)}+,
and the index of refraction becomes,
p:*:x:{t;n')', (4.r7)
where it has been assumed that V : 0 outside of the metal. If we take the
potential well of nickel to be V' ,-,
100 eV total energy we find that
-5 volts, then for incident electrons of
{iff}-: {r.05}} :
tt ,\-, r.02.
424 THE DEVELOPMENT OF WAVE MECHANICS
Figure 46 Diffraction of 50 kV
electrons from a disordered film of
CurAu alloy. The alloy film was
400 A thick. (Photograph kindly
furnished by Dr. L. H. Germer.)
Figure 4-7 Diffraction of 300 volt electrons

from a clean (l l0) surface of a tungsten single
crystal. (Photograph kindly furnished by Dr.
L. H. Germer. After L. H. Germer and J. W.
May, Surfoce Science 4,452 (1966). Used with
permission of North-Holland Publishing Co.,
Amsterdam.)
THE DIFFRACTION OF PARTICLES t25
Figure 48 Neutron Laue photograph of NaCl. (Photograph kindly furnished by Dr. E. O. Wollan.)
what is the wavelength of a photon whose energy equals

the rest energy of an electron ? what is the significanie of
this length ? How does it compare with the claisical radius
of the electron ?
PROBLEM 1.6
using 2.15 A for the lattice constant of nickel, calculate
the angular positions of the first and second order maxima
for 100 eV electrons incident normal to the surface.
t26 THE DEVELOPMENT OF WAYE MECHANICS
Having established the fact that electrons have a wave nature, Davisson
and Germer continued their work by looking at higher order Bragg reflections
-tb
1"9 by studying grazing angles of incidence. They were able assign an
index of reflection slightly greater than one to nickel for 100-volt electrons.
_ Pressing the analogy with x-rays a step further, Thomson proposed that
electrons passing through a crystal should produce diffraction patteins similar
to Laue patterns. He subsequently observed these from foils of gold, aluminum
and other metals.lo Later, diffraction patterns were observed for other ele-
mentary particles (see Figure 4-B) as well as for atoms and molecules. *il
6, THE WAVE FUNCTION FOR AN ELECTRON

As was seen in the previous sections, a beam of electrons behaves very much
like a beam of photons for the proper choice of energies and physical di-
mensions. As a further illustration, consider the following particle analog of
Young's double slit experiment in optics.* Let a mono-energetic beam of
electrons be incident on two slits whose dimensions and spacing are chosen
to be of the same order of magnitude as the de Broglie wavelength of the electrons $I
so that diffraction effects can be detected. The detector may be regarded as
an afiay of microscopic counters or as a fluorescent screen which can be
photographed. If either slit is blocked off, the pattern observed on the screen
would look like that shown in Figure 4-9(b). However, if both slits are open,
the pattern resembles Figure 4-9(c), where the interference effects are striki"sly
evident at the points shown. If the electrons behaved like classical particles
there would be no interference effects and the pattern on the screen would
approximate that shown in Figure 4-9(d). By repeating the experiment with
progressively lower beam intensities, the same pattern is observed after a
sufficient time. Surprisingly enough, even if only one electron at a time were
fired at the slits, the interference pattern of Figure a-9 (c) would be produced
after a sufficiently large number of electrons were fired. (See Figure 4-10).
This forces us to conclude that each electron interacts with both slits at nnce even
though our classically trained intuition tells us that each electron can pass
through only one of the slits. Thus, from the point of view of quanium
mechanics, it is meaningless to ask which slit the electron goes through. Ary
attempt to determine experimentally which slit an electron goes through will
destroy the interference pattern just as effectively as if the other slit had been
blocked off!
It follows from the above discussion that it is impossible to predict at
which point a given electron will strike the detecting screen. However, we
can relate the relative height of the particle distribution curve at position x ffi
to the relative probabilit2 that an electron will strike the screen at position .r.
Drawing again upon the optical analogy, if monochromatic photons had been
incident upon the slits of Figure a-9(a), the curve shown in (c) would be a plot
10
G. P. Thomson, Proc. Roy. Soc. A. ll7,600 (1927); 1l9p 651 (1928); 125,352 (1929); 133, I
(le3l).
* For an elementary review of this subject see Hugh D. Young, Fundamentals of Optics and Modern
Phltsics, McGraw-Hill Book Co., New York, 1968, Chapter 3.
Detecting
screen
_=_5
-'* ._1
--_->
--=+ 3_1,^
____-+
Incident
particles I
A
(a) (lr)
BA
|-t|l
(c) (d)
G9 (a) Double slit experiment with particles. (b) Distribution of particles recorded
Figure
on the screen due to diffraction from either slit A or B. (c) Distribution of particles recorded on the
screen due to diffraction with both slits A and B open. (d) Hypotheticil distribution of particles
recorded on the screen if wave effects are neglected.
(a) After 28 electrons
. Figure 4-10 (a), (b), and (c). Computer-simulated

growth of a two-slit interference pattern for electrons.
(d) An actual photograph of a two-slit pattern produced
by electrons. (Parts ("), (b), and (c) from E. R. Huggins, (b) After 1000 electrons
Physics ,, W. A. Benjamin, lnc., New York, 1969. Part (d)
is from C. Jtjnsson, Zeitschrift f Ur Physik, 16,, 454 ( l96l).
Used with permission.)
(c) After 10,000 electrons
refifiil
(d) Two slit electron Pattern
127
'
of relative light intehsity versus position on the screen. The light intensity
at each point is interpreted as the square of a wave amplitude vector which
can be represented by the real or imaginary part of a function of the form,
q : )re(w-ail.
Therefore, in quantum mechanics we define a wave amplitude function, or

simply awauefunction, for an electron, Y(r, l), such that its modulus squared is
proportional to the probability of finding the electron at position x at time t.
This wave function is a complex function which we write as,
Y(r, t) : Y osi0t'-'t) : Y ooibo,-Et)lh,
where the de Broglie relation has been incorporated in the last expression in
order to relate the frequency and wave number of the wave to the energy and
momentum of the particle. Although this complex wave function is not
directly observable (that is, measurable) its physical significance rests on the
assumption that the quantity
lY(rr, t1)12 dx : Y* (rr, tr) .Y(rr, t1) dx
is proportional to the probability of finding the electron in the elem ent dx

centered at x, at time /r. Then the total probability for finding-the electron
argtwltere in the space in question is proportional to the integral of lY(r, t)12
?
over all of the space. Thus, the total probability, P, is
-n
,-l** lY(", t)12 dx,
for the one-dimensional space along the r-axis. This integral must be finite
in order to represent a real particle. It is convenient to define the probabilit2
densit2 for the particle as
'
P(x,t) :
lY(x,t)12
{
(4.18)
t*t", t)t dr'
f F
I
f
I
where we then have I
i
dx : l. (4.1e)
I-:rr,q I
i
I
For physically acceptable wave functions it is'always possible to introduce an I
r
{.
appropriate factor in the wave function such that I*
i
i{!
J- t*{r,
t)lz dx :J]J.,', /)Y(r, t) dx - 1. (4.20)
I
I*
g
i-
,
f
I
THE WAVE FUNCTION FOR AN ELECTRON
'29
When Equation 4.20 is true, the wave function Y is said to be normalized. It is
evident from Equation 4.18 that when a wave function isnormalized the
probability density is simply the square of its absolute amplitude.
Furthermore, in order to account for interference effects we assume the
validity of the principle of superposition. When superposition is valid in optics
(that is, in the case of coherent light) we add the amplitudes at a point vectori-
ally and square the resultant amplitude to obtain the intensity at the point.
Thus,
I - (i, -l ir), : AL + A? + 2iL. i, - I, * I, I 2ir. ir,
where ,I represents the time average of the intensity over a full cycle. On the
other ha4d, when light is incoherent (the relative phases of the different
sources are washed out), the resultant intensity just goes as the sum
I-L|-Iz,
and the interference term 2i, .,4-, vanishes. Applying the principle of super-
position to the experiment of Figure 4-9, we write :
p(x, t) ^.,, lY(r, t)1, -- lYo@, t) + Ys@, t)12
- lYzl2 * lYal' * YjYa * YsYz
- Pz * pa * lYjYrl t lYsYel.
Here, the last two terms are the interference terms which depend upon the
relative phases of the two waves. The plus and minus signs correspond to
constructive and destructive interference. The phase factors in the wave
functions Y, and Vr play a role analogous to that of vector addition in the
above example from optics, thus indicating the importance of choosing complex
wave functions.
The device of representing an electron by u complex wave function can
be extended, of course, to other kinds of particles and even to atoms and mole-
cules. In order to be physically admissible, however, a wave function which
represents a particle must be finite, single-valued, and continuous. Furthermore,
it must vanish suitably as r --+ @ so that it can be normalized, as indicated
by Equation 4.20. The latter requirement is necessary for the validity of the
probabilistic interpretation of the modulus squared. The great significance of
the principle of superposition is that a superposition of physically acceptable
wave functions is itself acceptable for the representation of a real particle.
We will now study the Fourier integral theorem. Its use will enable us to
extend the mathematics of superposition from the simple case treated in section
8 of Chapter 2 to the formation of packets consisting of a continuum of fre-
quencies.
I3O THE DEYELOPMENT OF WAVE MECHANICS
PROBLEI,I 4-7 Fi;ii.i;:;!r;:::r
What are the phase and group velocities of an electron whose

de Broglie wavelength is 0.01 A? What is the kinetic energy
of the electron ?
(Ans. i I)o :0.925c; uptt :1.08c; 0.836 MeV.)
-i
PROBLEM 1.8 il:r! i e

t
A particle is defined in the space, -o ( x 1@, by the

wave function,
Y(r, t) : 7;aztifta-att.
(u) Find the normahzation constant ,4.
(b) What is the probability of finding the particle in the
interval bounded by r and x I dx at time t?
(.) What is the total probability of finding the particle
somewhere between
-o and *o? t
:
PROBLEM 4.9
A particle may be represented in the space, -a 1x I'

*
by either of the wave functions I

(u) V(t) : z4 cos nxl2a or (b) Y(r) : .8 sin nxla.
Find the normalization constants A and B. t
h
il
PROBLEM 4.'O $r
{t
A particle may be described in the space, -a < x < a, by
a superposition of the two wave functions of the preceding
problem, namely, ffi
Y(") :lcosfi+Bsin!.
(u) Normalize this new function.
(b) Sketch the probability density as a function of x.
7. THE FOURIER INTEGRAL AND THE

DELTA FUNCTION
Any periodic function, such as f (t + T) : "f (t), can be expanded in
terms of sines and cosines provid ed thatf (t) is piecewise continuous and differ-
entiable throughout the interval of the expansion.ll If the interval of expansion
11R. Courant and D. Hilbert, Methods of Mathematical Physics. Interscience Publishers, Inc.,
New
York, 1953, Chapter 2.
THE FOURIER INTEGRAL AND THE DELTA FUNCTION I3I
includes the end points of the period, then we add the assumption that the
value off (t) at each end pointis the arithmetic mean of its values at the right
and left end points. Then, in the interval, -(Tl2) < t < Tl2,
:
-f (t)
f + 2 o* cos o-nt + : bn sin ,.nt, (4.21)
where
an : + [:; ,,f
(t) cos ant' dt' ,
bn : + I:; , 'f (t') sin ant' dt',

2nn
@n:7.
This may be written in a more convenient form* by means of the Euler identity:
(t) : .|-rno'^' and cn : (t')r-n'nt' d,t', (q.22)

-f + f; , -f
where cm : t@^ -;;;, cI : c-, and,@m : -,)-n.
PROBLEM 1-ll rt-?iiilt:rrii.,j liii:iiiiliiiiiil#
Verify that Equations 4.21 and 4.22 are equivalent.
i:;i1l;;,i;riliiii i :',, ,' rl:".1., .:'..,'.':: ";.1-"t -.1
It is desirable to extend the interval of the expansion from T to @ so

that non-periodic functions can also be represented by expansions in sines
and cosines. To do this we must add the requirement that the integral
"f1. "f @t dl exists. We rewrite Equation 4.22 in the form
I
f @ : -2-+ I:; , .f
(t'),n'n(t-t't ,,'l, (4.2s)
and note that we must eliminate the factor 1/7" before going to the limit.
Since z is restricted to integral values,
Aar:@n*L-0)n _ 2r(n * l) - 2nn

:- 2n
T,
* Although the complex representation is introduced here as a mathematical device to represent
real functions, its general utility will become evident when it is used to represent complex wave
functions.
I Aar
(4.24)
-:-
T 2rr
Substituting Equation 4.24 into 4.23,
"f
(t) ---*,,:." o' [1i,,-f
(t'1sn'*tt-t't 46"
In the limit as T + @r the frequencies are distributed continuously instead of

discretely and
.f (t) :* I__^ Il-t rrr'i@G-'I'' dt' ,
if the integrals are absolutely convergent. The Fourier integrals are written
in a more nearly symmetric form by defining a function g(r) as follows:
(t) :
and
-f
#['*^ s(*),n"
(4.25)
g(r) :
#f_r, tr6'),-,,'i'
The functions/(t) and g(ar) are called Fourier transforms of one another.
Any reasonably well-behaved function /(l) can be represented by a super-
position of harmonic functions with continuously varying frequency and
weighting function g(ar). Conversely, g(r) can be represented by a super-
position of harmonic functions in time, each function multiplied by the
weighting factor ;f (t') .
The above expansion is in the time-frequenqt domain. It is evident that an
analogous expansion may be obtained in the position=wave-vector domain, or
what is conveniently called the coiirdinate-momentum domain. Physically, this
implies that our starting point was a spatially periodic function
,t@ + L) : rp(x),
whose Fourier expansion is

@
,p(x) :
n:-@
where
cn :L I-ryrx')e-it'na' dx' and'kn :ry

Following the same procedure as before, we eliminate the factor lll before
taking the limit as I -' oo by the device that Lkn,--'2rll and l lL - L^k*12r.
THE FOURIER INTEGRAL AND THE DELTA FUNCTION r33
The results are:
',t,@):#l-- dkg(k)eck
I . (+.26)
6&):#[:_ dx'y(x')e-'*')
Equations 4.26 are readily generalized to any numbe r of dimensions. For

example,
,t e) : ef y! norl)eifrr
l (4.27)
dG) : (*ldj di',p(i')e-,n,' ,)

where i has comp_onents (x,1t, r), E has components (k,,k,k,), di = dx d1 dz
and dE : d'k* dko-dk,. It is aisopossible to incorporate the time a.to, by wriiing
V(i, t) : y(i)e-r,t,
but the time will be omitted until it is specifically required.

The need for the Dirac delta function arises naturally from the Fourier
integrals. Suppose we substitute 6&) into rp(.r) in Equations 4.26 as follows:
,p (x) : * I__on f_or, fx,) eik(u-u' t .
Interchanging the order of integration,
,r,@) :* I__rr',t,@') !**onro^n-a').

Now define the delta function,
d(* - x') :*l__onrik@-al. (4.28)
The delta function has the following properties:
6(* - x') :0, if x' * x.

_ x,) ar, :
tur- {oi, li :arrl,f .- 1a
Since the delta function is zero everywhere but at the singular point, the only
contribution to an integral oecurs at that point. Then, using Equation 4.28
,p(x) : If@ dx'y(x')6(x - x') - rp(x).

J-o
Thus the delta function may be thought of as a spike function having unit
area but a non-zero amplitude at only one point, the point of singularity, where
the amplitude becomeJ infinite. When integrated over all of space its effect
is to yiita the remaining factors of the integrand evaluated at the singular
point.
It is often convenient to place the origin at the singular point, in which
case the delta function may be written as
d(r) : Lul:- lftsitca
An alternative definition of the delta function may be obtained by inte-

grating as follows:
(:-4: sinax
d(") :
:g# t;*'dk: 1'-
2zr o-* \ ir /
lim
on* rrx
, (4.2e)
where a is positive and real. Let us examine the behavior of this function for
both small and large x. First, consider the limit as .r goes to zero:
sin ax a ,. sin ar :-. 4

llm-:-lllfl
a+o TltC 7T a*o AX Tl
Thus, d(0) : l:L*aln ---> oo, or the amplitude becomes infinite at the singularity.
For large lxl, sin axlrx oscillates with period 2rla, and it1 amplitude falls off
as |/lxl. But in the limit as a ---> o, the period gets infinitesimally narrolu so_
that the function approaches zero everywhere except for the infinite spike of
infinitesimal width ai the singularity. It now remains to show that the integral
of Equation 4.29 over all space is unity:
sinax 2 f*sinax,dx: 2ir

l-]* rrx
dx:hm
e+@ -l
n Jo X
-.--l
iT2
This establishes the validity of Equation 4.29 as a representation of the delta

function.
PARTICLES AS WAVE PACKETS I35
PROBLEM 4-12 tiiit;i:fi
Show that the following two functions are valid represen-

tations of the delta function, where e is positive and real:
(u) d(r) :]ri* Lr-"r

\/fie+oYe
(b) d(") :*.r*

=+n
Verify the following properties of the delta function:
(u) d(") : d(-r).
(b) x6(x) - 0.
(.) d'(-") : -d'(")
(d) x6'(x) : -d(t).
(.) c6(cx):d(r),c>0.
8. PARTICLES AS WAVE PACKETS
The assumption that a particle can be represented in space and time by u

complex wave packet is one of the fundamental premises of wave mechanics.
We have seen how the concept of a complex wave function has resolved the
classical wave-particle dualism through the de Broglie relations which connect
the energy and momentum of the particle with the frequency and propagation
constant of the wave. Also, in section B of Chapter 2, it was pointed out that
in the classical limit the particle velocity a is associated with the group velocity
of the packet.* Hence,
b ldE
r-',,,_tt-
dE
*- " - hdk -
dE :la,
and
E-L**r, (4.30)
* This is another example of the correspondence principle which states that quantum mechanics must
give the same result as classiial mechanics in the appropriate classical limit. For particles, the classical
limit occurs for very short de Broglie wavelengths.
where V rrray be regarded here as the integration constant. Experiments which

depend upon particle-like properties can be regarded as observations of the
behavior of the packet or wave-group as a whole, whereas experiments which
depend upon interference and diffraction effects are observations of individual
wavelength components of the packet. With the help of the Fourier integral
theorem we are now able to construct and study some elementary packets
composed of waves so closely spaced that they form a continuum of frequencies
or fr-values. This has the decided advantage of producing a truly localized
packet having essentially zero amplitude except in one specific region of space.
By way of contrast, the reader should recall that although a finite sum of
waves whose frequencies differ by finite amounts can produce a modulation
envelope that resembles a packet, this kind of packet repeats itself regularly
throughout all of space. It is evident, then, that a packet which is used to
represent a particle in free space is best constructed by means of the Fourier
integral theorem.
As our first example of the use of this theorem, let us look at the effects
of chopping a pure, infinite sine wave in both the time and the coordinate
domains. Consider a sine wave of frequency co' which is chopped to a lifetime
of 2T seconds. That is,
f(t\:ei'rt-
J\/)-
for -T <t <7.
-f (t) :0, for l/l > T.
The chopping (or keying) process will introduce many new frequencies in
varying amounts, given by the Fourier transform,
g(,) :+f* J-* 0,11,)e-t't

! 2n
tPT
: o,
fnJ_;n,.,_,,,
sin (aro - ro) ?"
: li . ,.
-
V; (ro-,)T
This function is plotted schematically in Figure 4-11. The attempt to chop in
the time domain results in the mixing of an infinite number of frequencies,
albeit in small amounts except for those frequencies close to a,l'. The principal
contribution is still at the frequency al'. Note the similarity with the amplitude
pattern for single slit diffraction or for a shutter, which is its analog in the space
domain (the shutter chops an infinite wave front and produces a spread in
propagation constants). Nature provides an example of this kind of chopping
in the emission of photons during electronic and nuclear transitions in atoms.*
Notice that the central peak in Figure a-l1(a) has a breadth given by
. :7.2rr or T' Lv:
Or 1.
r See section 9 of Chapter 2.

PARTICLES AS WAVE PACKETS t37
g(r) d(k)
I zr o I.
a;'r-1 ','*T
(a) (b)
Figure 4ll The spectral distribution resulting from chopping a pure, infinite sine wave in the
time domain (a) and in the spatial domain (b).
Multiplying by h and replacing the time interval T by At, we have the state-
ment that the minimum uncertainty product for this type of packet is
L,t.L,E:h.
Therefore, we may state Heisenberg's uncertainty principle for simultaneous
measurements of energy and time for such a packet as
L,t.L,E>h. (4.31)
The spatial counterpart of the above keying process can be illustrated by

chopping an infinite, plane wave front with a shutter such that the length of
the packet is 2a. (Here 2a :2cT, where 2T is the time interval thit the
shutter is open.) Then,
ys(x):stlcor, for -a <x <a,

,p(x) : 0, for lxl > a.
Then,
+(k) : #,ll-rol,-ikn dx
:+ f" dr
! 2n J-" 'n'o'-o'*
l, sin (n'o
- k)o
v; fto - k)a
This function is also shown in Figure 4-lI, but here it is the wave vector (or
the momentum) that takes on a spread of values around fr,.* The breadth of
* Likewise, in single slit diffraction, we generally regard ks as a constant and allow its direction
to
take on a spread of values. This is equivalent to treating the slit as a source 9f Huygens wavelets.
the peak is Afr :2nla, from which we obtain the result

a' L,p, : 7,
which enables us to write the uncertainty principle for simultaneous measure-

ments of position and momentum as,
L,x.Lp,>h. (+.32)
For the packet shape shown in Figure 4-11, the minimum value of the
uncertainty product is fr at t :0. The packet shape that achieves the absolute
minimum value of hl2 is the Gaussian packet, which will be discussed later in
this section. (See Problem 4-2I.) However, as time progresses, any packet
will broaden in coordinate space so that for a given L,p, the uncertainty product
increases with time. (See Problem 4-16.) If we denote the time during which
the packet retains its form as To, then
mm
ro-E (Ar)',
where Ax is its initial breadth in coordinate space. The more massive the
particle, the more slowly its packet spreads with time. This is what we want
in order to satisfy the correspondence principle. Thus, for an electron which
is localized to I angstroffi, To n,,|Q-ro second, whereas for a .22 caliber bullet
localized to 0.1 mm, To
- l02a seconds. It is evident that the concept of a
classical trajectory, though meaningless for the electron, is quite appropriate for
a macroscopic particle.
Now suppose that we wish to represent a free particle as a superposition
of a continuum of plane wave states, namely,
Y(x, /) :Joiol* *@, t) dk. (4.33)
Each value of $(k) serves as a weighting factor for the wave function it multi-
plies. That is, it tells how much of that particular wave function contributes
lo the packet at position x and time /. A plot of the values of $(k) verzus *
is known as the qpectral distribution of states, or simply as the spectrum of the
states. The plane wave states of Equation 4.33 may be represented by
Y*(', t) : Aen&v-@t\,
where ,4 is the normalization factor. Since infinite plane waves are not well-
behaved functions in the sense that they do not vanish at x : * @, special
proced.ures must be used in order to normalize them. One such procedure is
to use delta-function normalization; thus,
:
t- Yf,Yo dx : VPI*:'rue-ik'e dx IAPI:-enr&-k'l dx.
PARTICLES AS WAVE PACKETS I39
Comparing this expression with Equation 4.28, written for the delta function
in k,
6(k k')
1r- eie(tc-k')
- - 2n J--
I dX,
we see that by choosing lAl : I | {G we can normalize a plane wave function

to the delta function, so that the integral of its modulus sqlared over all space
is'unity.
It is now evident that the integration over a continuum of states is identical
with the Fourier transform of Equation 4.26 with the time factor included.
The function Y(x,t) is said to be the wave function of the particle in the
coordinate representation and O(,t) is the wave function in the momentum represen-
tation. The two wave functions are Fourier transforms of one another. Consider
a packet whose spectrum of &-values is a delta function in momentum space,
that is,
6&) : 6(k _ ko).
Then,
,p(x) : - ko)ro** : rf,r*',,

hl*** 6(k
which is the normalized wave function for a packet containing a single mo-
mentum component hko. Note that the infinite plane wave ind thi delta
function are Fourier transforms of each other.
As a second example, consider a packet whose spectrum of ,t-values is the
Gaussian function,
f(k) : 4r-tcztzoz,
where o is the standard deviation of the distribution and is a measure of the

spread of the packet. (See Figure 4-12) Then,
,p(x) : + f* dkr-r',zo2sika.
! 2rr J--
I I -)
t* zr/i- oj
r.
k
Figure 4-12 Gaussian packets. Each packet is the Fourier transform of the other. The breadth
of the packet is arbitrarily taken to be its standard deviation, as shown.
I4O THE DEVELOPMENT OF WAVE MECHANICS
In order to complete the square we multiply by exp [(x2o'12) - (x2o2l2)l : I

and obtain:12
:
,t,@)
f" I**on.,.o - # * [
ikx +
ry - ry)
=
#r-.,",,,f_dk
expl-
*&, - 2ikxoz - r,qf
: ir4')
fn'-."''' f*dk l- *, & +
exP
=#r_.,",,rf*e-u, du
- Aoe-'zo'lz'
Note that if o(,t) is normalized by settin g A : (#)",' th.r, v(x) is also a
normalized Gaussian function; that is, the Fourier transform of a Gaussian

function is also a Gaussian. The spread of the packet in coordinate space is
given by I lo and illustrates the reciprocal nature of Aft and Ax as required
by the uncertainty principle.
.ROBLEM 4-r4 riljii:it:,ii.
Show that (dldt)to(e)l : 0.
PROBLEM 4.15
Show thatif the coordinate wave function is normalized at

t :0, then both the coordinate and momentum wave
functions remain normalized for all time.
PROBLEM 4.16 irliil;1ii#ri::"ltli i;:ili
For a free particle show that a wave packet in coordinate

space broadens with time (corresponding to increasing
uncertainty in the position of the particle), while its rep-
resentation in momentum space retains its shape and
changes only in phase (since momentum is conserved).
12
See J. L. Powell and B. Crasemann, Quantum Mechanics, Addison-Wesley Publ. Co., Inc.,
Reading, Mass. (1961), p.475.
THE scHnciorNGER wAvE EeuATtoN t4t
PROBLEM 4.'7
What must be the frequency bandwidth of the detecting

and amplifying stages of a radar system operating at pulse
widths of 0.1 psec ? If the radar is used for ranging (dis-
tance measurements), what is the uncertainty in the range?
PROBLEM 1.18
(u) Find the normalization constant // for the Gaussian

wave packet,
,P(x) : 1Yt-k-agtzl2Kz.
(b) Obtain the Fourier transform of y(x) and verify that

it is normalized.
PROBLEM 4.'9
(u) use the uncertainty principle to obtain the uncertainty

in the momentum of a particle of mass rn constrained
to the volume of a cubical box of side a.
(b) What is the corresponding uncertainty in the kinetic
energy of the particle ?
(") How do you interpret the answer to (b) ? (See Problem
4-2.)
9. THE SCHRODINGER WAVE EQUATION
In section 6 we defined a complex wave function, V(", l), which is assumed

to completely describe the dynamical state of a particle in the Schrcidinger
representation. Although the wave function Y(x, t) cannot be measured di-
rectly, we interpret its "intensity," that is, its modulus squared, to correspond
to the relative probability of detecting the particle at position x at time t. How-
ever, a knowledge of the state of the particle at one point in space and time is
not enough, since we would also like to know how the particle is behaving
at some later time. In order to describe the progress of the particle in space
and time it is necessary to find a wave equation, preferably a differential
equation, whose solutions will correspond to the motion of the particle. This
equation cannot be derived, but must be postulated and then tested against
experimental results. The equation should be linear in Y(x, /) so that if
Yt(t, /) and Yr(r, t) are solutions, vty linear combination of V, and Y, will
also be a solution. This requirement insures the validity of the superposition

principle. Furthermore, the differential equation must be of the first order
with respect to time so that ifY(x, /) is known at, say, to, then it will be uniquely
specified at alI later times. Our last requirement is that the wave equation
must be consistent with de Broglie's hypothesis and the correspondence principle.
For a free particle the total energy is
"o --Pi
2*' (4.3+)
which becomes in the de Broglie formulation, Equation 4.1+,
,h :
h2k2
,2* (4.35)
We have seen that such a free particle can be represented by an infinite plane
wave of the form
Y(r, t) : I o'r"-"' (4'36)
iZn
Note that a single differentiation of Equation 4.36 with respect to time gives
AV
(4.37)
;: -irouY,
whereas differentiating twice with respect to x results in
a:Y : (4.38)
0x2 -k2Y.
From Equations 4.35, +.37, and 4.38 we see that we can write,
..
t6-
o'e
av h2 azY
(4.3e)
0t 2m 0x2'
which is Schrodinger's wave equationls for a free particle in one dimension.

In three dimensions the wave function for a free particle is
l-
Y(i,) t)t : ;rrt'i-'tt (+.40)
(2n1t
and the corresponding Schrodinger equation is
., a ,-^,_ t)
., h, n, (+.41)
ih t),
i,Y(r, fiYzY(r,
,
ls E. Schrddinger, Ann. Ph1s.79r 361 (1926).
THE CONSERVATION OF PROBABILITY DENSTTY I43
where V2 is the differential operator,
a2 a2 a2
Trr*W--Azz'
If
the particle is not free but is acted upon by u force which can be ex-
pressed as the derivative of a scalar potential function, we merely add this
potential to the right hand side of Equation 4.3+. Thus,
E-f**''
and
ah.:H * r.
Equation 4.39 now becomes
ih!Y(x. t\ : (
-Y I * v\v1r, t1 : trv(x, t).
dt^\^r"/-\ 2mOx2,'l'
The operator, ,ff : *
V, is called the Hamiltonian operator
-(h2l2m)(02l0xz)
since, analogous to the classical Hamiltonian function, it is the sum of the
kinetic and potential energy operators. In three dimensions the time-dependent
Schrodinger equation is,
: "/rY (i, t),

:tY(i, 4
th (4.+2)
where
.r -- - nv2 + v(i, t). (4.+3)
Equation 4.42 is a far more useful form than Equation 4.39 since all of the
important physical problems do contain forces. In fact, one can almost go so
far as to say that all of the interesting physics is contained in the potential term.
IO. THE CONSERVATION OF PROBABILITY DENSITY

As in section 6 of this chapter, we define the probability of finding a
particle in the volume element dr : dx d1, dz as,
p dr : pe, t) dx d2 dz : Y* (r, ,)Y(i, t) dx d) dt, (4.+4)
where Y(f, r) is the normalizedwave function associated with the particle. Irlow
suppose that as time progresses, p decreases for one volume element in space.
In order to be consistent with the fact that the total probability density over
all of space must be unity, we must then have p increasing in some other volume
144 THE DEYELOPMENT OF WAVE MECHANICS
element.* This shift or flow of probability density may be thought of as a

"probability density current." We shall now derive the formal statement of
the conservation law for probability density.
The time rate of change of probability density in any convenient volume
is given by:
+, Lp
dr :JF ],v. + Y* *,*) o"
Now, Y satisfies Equation 4.34,
trY : ih.!,v,
while V* satisfies the conjugate wave equation,
ffV* : _ih.
,Lrv..
The reason for the fact that the conjugate wave functions require a different
wave equation may be argued as follows. To claim that Y and V* satisfy the
same equation is to say that V : Y*.
To go a step further, if Y represents momentum transfer to the right, then
Y* transfers momentum to the left; Y:r€ - Y implies no momentum, that is,
no dynamical state. Then,
ry:
dt
!**: hffiv2+r)":*wY-in*,
th,
+:+.trY* :+(#vz + n)*. :#vs** +'rw*,

and
+Y +Y-# --#[(v,v*)Y -Y*(v,Y)].

Implicit in the last expression is the assumption that I/Y*V : Y* ZY. In the
next section we will see that this is equivalent to assuming that the operator
for the potential commutes with the wave function.
Now, by Green's theorem we have the vector identity
v ."f(vg) : (V/) . (ve) +"f(V,g)

v 'g(v/) : (vs) . (v/) + s\'f),
* This discussion assumes that matter is conserved. Therefore, these results are not valid for
relativistic reactions where particles are created and annihilated.
THE CONSERVATION OF PROBABILITY DENSITY I45
where the latter merely interchanges the roles of the functions f and g. Sub-
tracting,
v ' ["f(vg) - e(v/)] : "f (v'd - s(vy).
Letting/-Yandg -Y*,
v.[Y(vY*) - Y*(VY)l : Y(V2Y*) - \rt (VrV).
Then,
:
* [,0 a, I, - *v . ;v(vY*) - Y*(vY)] d".
Now we define the probabilit2 densitlt curcent,
^f(i, 4 : *[Y(vY*) - Y*(vY)]. (4.+5)
Then
* l,p dr :1, -Y ' s dr,

or
ff*v'.f:0. (4.+6)
Equation 4.46 expresses the conseruation of probability densit2. Its similarity to

classical conservation laws in the absence of sources or sinks should be immedi-
ately recognized.
The probability density current may be written in a more convenient
form by noting that Y*(VY) is the complex conjugate of V(VY*). Thus the
bracketed quantity in Equation 4.45 is equal to 2i times the imaginary part
of Y(VY*), and Equation 4.45 becomes:
î(a r) : +rm [Y(vY*)]. (+.+7)
By way of illustration, consider a wave packet representing a free particle

of momentum hko : mu, in the r-direction,
Y(r, t):
' +ri&sa-a)tt.
{2,
Then the probability density current ill the x-direction associated with the
motion of this particle is
s,: #t^;v(-rlrJr/*)l : hkom IVY'|'l : p(x)ro.
Hence, the probability density moves with the particle velocity.

PROBLEM 4.20
If the Gaussian packet of Problem 4-18 represents a free

particle, calculate the probability density current.
I I. OBSERVABLES, OPERATORS, AND EXPECTATION VALUES

In the preceding sections we have seen how a particle can be represented
by u complex wave function and we have obtained the wave equation which
describes its progress in space and time. In addition, the probability density
and its conservation law have been discussed. Yet the reader may justifiably
wonder how all of this relates to the dynamical variables which can be measured
in the laboratory. After all, this wave function, which contains all of the
available information about the particle in the system under study, cannot
even be directly observed !
In quantum mechanics the connecting link is that each dynamical
observable is represented by a mathematical operator through the corre-
spondence principle. When this operator operates on the wave function, a
number is obtained which corresponds to a possible result of a physical measure-
ment of that quantity. The operators we use in quantum mechanics to represent
physical observables are linear operators. Thus, if Pis a linear operator, then
P (orp, * brpr) : aPVt I bPVr,
where a and b are constants. The sum and the product of two linear operators
also form linear operators. Although the sum of two operators is commutative,
the product is, in general, not commutative.
We say that two quantities a and D commute under multiplication if
ab : ba; that is, if the order of multiplication is immaterial. This property
is taken for granted in ordinary algebra. However, you. have already seen in
vector algebia that the vector product Z x E + E x i, although the scalar
product n . E : E .i. Thus, the dot product is commutative, whereas the
cross product is not.
In operator algebra we must not assume commutivity unless it is proven
specifically for the operators in question. Consider the product of the operators
x and dldx operating on a function of x:
-
fr tra : xf '(x).
But,
!* rt f-l : -f (r) + xf ' (x).
Subtracting,
?fr- *lrrrt : -f (r).

OBSERYABLES, OPERATORS, AND EXPECTATION VALUES 147
Therefore, the equivalent value for the bracketed quantity is - l. This bracket
is called the "commutator bracket" of r and df dx and is usually written in the
abbreviated form,
: -1.
[",
Evidently,
ld ,r] : *1.
ldr
Therefore, two operators commute if tlteir commutator bracket is zero.
Let us now consider a measurement of a position coordinate of a particle.
The probability of finding the particle in the volume element dr is, from
Equation +-lB,
z' t)Y-(r'''v' z' t)
!"-
p(x,), z, t) dr - j,V*
-Y!-|.*-'l'
(r,), z, t)Y(x,1t, z, t) dr'
.
The fact that p dr is called a probabilit2 implies that one would not necessarily
expect to obtain the same result for two successive measurements on the system.
We can define an aaerage ualue, or expectation ualue, of the position variable
by considering either a large number of measurements on the same system, or a
single measurement on each of a large number of identical systems. Thus, the
expectation value of, say, the variable x, is the weighted average,
(4.48)
The expectation value of a component of the momentum may be determined

directly from the momentum wave functions in like manner. Thus:
(f,) : li- $*!"+ dk. (+.+e)

.[:- d* 6 dk,
It is also possible to obtain the expectation value (r) by means of the momentum
wave functions, provided that the appropriate operator is used. In order to
find the correct operator expression, let us insert Equation 4-26, the Fourier
transforms of Y* and Y, into Equation 4-+8, where we assume that Y is
normalized. Then,
(x) :
*fidxdkdk' 6* 1k)e-tk, xeik'u 6&')
xeik'' 6&).
+, fu I d.k 6* &),-o*.
Idk'
Performing a partial integration over k', the last integral becomes
l-i6@')eno''12* + t| ,nr'' L o&') dk"

I48 TH E DEYELOPMENT OF WAYE MECHANICS l
I
where the first term is zero since the wave function vanishes at both limits.
i
Then,
(x) : *IIlr*i(k'-k)x dk dk' 6*(k) t# 6(k')
:
IIor
dk,6*&) t
* +&,) 6(k - k'),
where Equation 4-28 has been used. Performing the ,t' integration,.
(r) : oo r. &) i *,6&) : 6* ( p) ih or n

I Idp, S,
Thus, the expectation value (r) can be obtained from the momentum wave
functions by means of the operator, ih(010p,). In like manner, it can be
shown that the momentum operator in the coordinate representation is
-ih.(7l0x). That is,
(!.) :f v- (-n *)" o- (4.50)
By comparing Equations 4-34 and 4-39 of section 9, it is evident that the

requirement of the correspondence principle, namely,
(p?)*-_ hz*,
otc"
is consistent with the last equation. Applying the correspondence principle

to the left members of Equations 4-34 and 4-39, we see that the total energy
operator may be represented by
E<-+ih!.
ot
A few of the commonly used operators are summarized in Table 4.2 for
convenience. il
Before closing this section, it is worth noting that the concept of the
expectation value enables us to make a more precise statement of the uncertainty
principle. The difference between a given measurement ofx and the expectation
value of r is called the deuiation of x. We define the uncertainty of r as the root-
mean-square deviation of r. Thus,
Ar: {GW
: \/ (x, - 2l4tc> + xrD
: \/ <tP> \x><,c> + (xY
: \/ <,cz) - <xy. (4.51)
OBSERVABLES, OPERATORS, AND EXPECTATION YALUES t49
TABLE 4-2 Commonly Used Operators
Coordinate Momentum
Variable Representation Representation
a
x ,n
,n
q t
-o' un
a
P, - ift --dx P*
az
pi -ftz pi
-ox'
a o
E th= oh
dt u,
By means of Equation 4.51, we are now able to define the uncertainty product
given in Equation 4.32 asla
L,x . Lp, : f (xr> 1xy' {W- A.Y

h
-2
Equation +.31, however, cannot be expressed in this manner since the time I is
not a dynamical variable but is a parameter in the wave function. The meaning
of At in Equation 4.31 is that of a time interval during which the system makes
a transition from one state to another.rs
PROBLEM 4.2'
(u) For the Gaussian packet of Problem 4-lB, calculate the

expectation values (x), (x'), (P), and (p*z) using the
coordinate representation.
(b) Find the same expectation values using the momentum
representation.
(.) Find the minimum uncertainty product Lx ' L'p*. The
Gaussian packet has the unique property of providing
the theoretical minimum of uncertainty.
(d) Find the expectation value of the kinetic energy, (7').
la See A. Messiah, Quantum Mechanics. North-Holland Publishing Co., Amsterdam, 1958, Vol. l,
p. 133.
rd lbid., p. 135.
I5O THE DEYELOPMENT OF WAYE MECHANICS
PROBLEM 4-22
Work out the following commutators:
lr' P*], l*, !of , lr, !'1, I !o !,], I P,, P'f '
Do the results just obtained depend upon whether the
coordinate or the momentum wave functions are used in the
calculation ?
12. SEPARATION OF SPACE AND TIME IN THE

SCHNOOINGER EQUATION:
ENERGY EIGENVALUES
There are many problems of physical interest in which the potential

energy of the system is independent of the time. In such cases it is possible to
simplify the Schrodinger equation (+.+2) by regarding the wave function
V(i f) as a product of two functions, one containing only spatial coordinates
and the other containing only the time. Expressed mathematically, we assume
that if V : V(i), then
:
Y(i, t) 'lte) ' x,@'
Substituting this form of the wave function into Equation +.+2, we obtain
A
ihy(i) : x(t) ' trrP(i),
*x@
since .tr is independent of the time. This may be written in the separated
form,
ih a-
,ttl :
*y!!)
- E, (4.52)
m' }'x\t) 'Pe)
where the symb ol E has been used for the separation constant, since we will
soon see that it is readily identified as the total energy of the system. The
equation involving the time can be immediately integrated, and its solution is
x@ : Ce-QthtEt'. (4.53)
The spatial equation deriving from Equation 4.52 may be written in the
form
"/frpe) : Erp(f) (+.5+)
where ,t : -(hrl2m)V2 + V(f). Since each solution of Equation 4.54 will

correspond to a definite value of the energy, let us write the ntk solution as
SEPARATION OF SPACE AND TIME l5t
,pn-Q). Thus, a particular solution of Equation 4.42 may be written as
Y t_(i, t) : Vn,(i) . e-(itilEnt (4.55)
Equation 4.54 is often called the time-independent Schrtidinger equation. It

belongs to the class of equations which is known collectively una t.iaitionally
as the eigenualue problem. Before continuing with the application of the
Schrcidinger equation to physical problems, let us briefly discuss some of the
properties of eigenequations.
Consider an operator Q which performs a mathematical operation on a
functionf This is indicated by the produ ct Qf. Suppose now thaf the operation
Q;f results in reproducing the same functionf multiplied by a constant. That is,
Q"f : +tf, (4.56)

where c is a positive constant.
When this occurs,,f is called an eigerfunction of the operator Q, and *c
is called the eigenualue associated with I For example, for the function yf :
sin 2x and the operator Q : (dzldxz),
q-f : fi btn z*) - -+sin 2r : -+Jt.
In this case the eigenvalue is -4 and the eigenfunction is sin 2x. If there is
a whole set of functions fn, there is an eigenvalue of Q associated with each
function. Thus Qf* : cn-fn. In the above example
-fn : sin nx
and
Cn: -n2.
Not all functions are suitable as eigenfunctions; for some functions there is
no opetator such that an eigenvalue exists. We say that an eigenfunction must
be 'iwell-behaved"; in particular, it must be single-valued ind the integral
of its modulus squared must be finite. In general, functions which are contiru-
ous, single-valued, and which are finite or vanish at infinity are suitable. It is
evident that the suitability of a function depends somewhat on the nature of
the operator, but generally the restrictions on eigenfunctions are the same as
those that we require for wave functions which describe particles.
One of the main activities of quantum mechanics is that of solving an
eigenvalue problem when only Q is known. That is, we want solutions of
Q.f : tcl where the problem is to find both c and f. As an example, consider
again the operator dzf dxz:
dT
_-LfT
,f
dxz - -L"J'
For positive c this has the solution-f : Aeri" 1 3s-{",. This function diverges
-> * oo, so it is not a suitable eigenfunction. However, if A : 0 and JtL :
as ,{
&s-\/i*, the function;{, is suitable for x > 0 and has the eigenvalue c. For x 1 0,
t52 THE DEYELOPMENT OF WAVE MECHANICS
the function f, : Asl'* with B - 0 is suitable for the negative half-space. It

also has the eigenvalue c.
To obtain well-behaved eigenfunctions for all-space, we must consider the
eigenvalue -c. Then,
a:
dxz
_rfr
-f :Dsint/ix f Ecos lir.

This function is well-behaved over all of space, and an eigenfunction exists for
Q fot all eigenvalues -c. In this example there is a continuum of possible
eigenvalues, whereas in many physical problems (such as standing waves on a
string) the eigenvalues form a discrete set of numbers.
Returning now to the discussion of the energy eigenequation (the time'
independent Schrodinger equation), which we will write as
,trrpu^?) : Enla^(i), (+.s7)
we wish to show that the energy eigenvalue E, (which is a particular value of

the separation constant for Equation 4.52) is a real number for the boundary
conditions that we impose on the wave functions. From Equations 4.55 and
4.18, we have
P : Y* (i l)Y(1, r) : ''ph,(i)'pE,(i)e-uth)(En-En*It
'
for yr^(i) normalized. Now, making use of the conservation of probability,

Equation 4.+6,
+ V . ^S(i t) :
?
-h (8, - E|)rpfr^(i)rpr^(f) 0.
Integrating over all of space,
(8" - nh Lrph-Q)rp'^(i) dr : I, 1,, Sd, :lo[,t do,
where the latter is a surface integral. For a sufficiently large bounding surface,
the probability flow through it must vanish, so we find that En : El , or En
is real. This important result tells us that when space and time are separable,
the time factor, Equation +.53, is a harmonic function and not an exponentially
rising or decaying function. Y(i, t) and yp*(i) differ, then, only by u time-
dependent phase factor. An immediate consequence of the reality of En is
that the probability density is independent of time, since
p : Y* (i, /)Y(i, t) : ytk^(i)tPn-e).

SEPARATION OF SPACE AND TIME ts3
- Following the pattern of Equation 4.48, for Y normalized, the expectation

value of the energy is:
(E ) :,[*- Q, t)ih
*,*rr, t) dr
: i n)
e( t
E't
ih (- E -) v r,Q1 r-( t t h) E
^t dt r
!,vh ^tr) ;
: E,l,rh,(i)yp^(i) dr
: E*t,p d'
-tr
-
un.
The meaning of the above results may be summarized as follows. When

the Schrodinger equation is separable in space and time, each of its solutions
corresponds to a definite energy which remains constant in time. The solutions
that correspond to the spectrum of these energy eigenvalues are called stationary
states or eigenstates. The probability density for an eigenstate is constant in
time, and the expectation value for the energy of an eigenstate is precisely the
energy of that state for all time. The reader can easily show that the expectation
value of any operator which does not explicitly depend upon time is also
independent of time for a stationary state.
We will devote all of Chapter 5 to the application of the Schrodinger
eigenequation for the energy, Equation 4.57,
.trrpr,(i) : Enrpn^(f),
to a number of one-dimensional systems. For unbound systems we will find

that a continuum of eigenvalues and eigenfunctions exists, while for bound
systems only discrete values of En exist. The state with the lowest value of En
is taken as the ground state, and the states corresponding to higher values oi
En are called excited states. Since excited states generally have short lifetimes,
one would suspect that the ground state is the only state which deserves to be
called a "stationary state." In spite of this, however, we do approximate the
states of a physical system (such as an atom) with the spectrum of absolute
stationary states obtained by solving the energy eigenequation. To get a
quantitative feeling for the validity of this approximation, recall thai the
nominal lifetime of an atomic state is l0-8 second (section 9 of Chapter 2) and
that the period of an electron in a Bohrorbit is second, sinc. optical
frequencies are ,--,10r5 sec-l. Hence, the electron -10-15
makes about l0? revolutions
in the state before leaving it. Since this is comparable to l0 million years in a
solar orbit, we must regard the electron's orbit in an excited state as relatively
stable !
13. DIRAC BRACKET NOTATION
It is convenient to introduce Dirac's abbreviated notation for integrals of

the type that are required for the normalization of wave functions and the
calcuiition of expectalion values. If ,p, represents some definite state function,
then the integral of its probability density over all of space is
d,.
t,vlv.
We will now also represent this integral by the symbol
(p,lg,"), or simPlY bY (a I a).
Note that there is no explicit designation of the complex conjugate or the

variables of integration. lt is understood that the complex conjugate of th.e
function to the left of the vertical bar will be used when the integration is
actually carried out. There is rarely any difficulty caused by omitting reference_
to the representation being used, but in case of ambiguity, the variables of
integration may be easily included in the brackets as follows:
(Po(r, t) | Y,(x, t)).
When finding the expectation value of an operator Q, the integral may be

written as
@,1Qtp") or (rp,lQl rt,,), (4.58)
where the form on the right implies that the operator is Hermitian. Since the
Hermitian property will not be discussed until Chapter 6, _we will restrict
ourselves to the form on the left for the time being. By way of illustration, the
expectation values given by Equations 4.48 and 4.49 now take the form:
(y rY)
(r):ffi and (!,):ffi
I (d I
p,+>
Although we will use Dirac's notation freely in the next chapter, its deeper
significince will not become evident until we take up Chapter 6.
SU/YIMARY :iri+:ii;::i':
The old quantum theory is briefly described and illustrated by the Sommer-
feld treatment of the hydrogen atom. Its demise was brought about by
Schrijdinger's wave mechanics, which is developed in -the rernainder of the
-Aft.t
chapter. a discussion of the de Broglie wavelength of a particle andparticle
didaction, a complex wave function is defined. Although the wave function
SUMMARY I55
is not directly observable, its modulus squared is interpreted as the relative

probability of finding the particle at a given point in space and time.
One of the fundamental assumptions of quantum mechanics is that a
wave packet composed of a superposition of complex wave functions can be
used to represent a localized particle. As required by the uncertainty principle,
the more narrowly localized the packet in coordinate space, the greater the
number of *-values required to form the packet. The Fourier integral theorem
provides the mathematical tools necessary for constructing packets. A com-
pletely unlocalized particle may be represented by u single plane wave having
one frequency and one *-value; that is to say, the Fourier transform of an
infinite plane wave is a delta function in ,t-space.
The concept of the wave packet resolves the classical wave-particle
dualism through the de Broglie relations, which connect the energy and
momentum of the particle with the frequency and propagation constant of the
wave. By means of this correspondence between packets and particles, a wave
equation is constructed which enables us to describe the progress of a particle
in space and time. Physical measurements are represented by linear mathe-
matical operators which operate on the wave function. The average value of
a large number of identical measurements is called the expectation value of the
operator.
The principal activity of non-relativistic quantum mechanics is the solution
of the Schrcidinger equation for systems of fundamental particles or atoms.
When the total energy of such a system is independent of the time, the solution
of the Schrirdinger equation reduces to the solution of the energy eigenequation.
The total solution is then simply the energy eigenfunction gn^multiplied by the
time factor, exp (-(ilh)E"t). The energy eigenstates are also referred to as
stationary states, since the probability density for such a state is constant in
time. Furthermore, the expectation value of any operator that is itself not a
function of time is constant for a stationary state.
CHAPTER 4 APPENDIX A
APPENDIX A. NATURAL UNITS

The dimensionless constant a, which was named the fine structure con-
stant, has a significance which goes beyond its role in Equation +.13. Its
magnitude may be regarded as a measure of the strength of the coupling of
pu.ii.l.r of charge le to electromagnetic fields. Indeed, one may call e2 the
:'strength" of ttre interaction between particles of charge te and electro-
-ug.t.ti. fields, but it is more "natural" to use the dimensionless quantity
a : erlh.c for this purpose. Since a is independent of mass it can be used to
describe the electromagnetic interactions of all particles of charge f e.
An appropriate set of natural units for describing atomic phenomena is
as follows:
Variable Natural Unit Description
mass mo electron rest mass
length I, : h Tfloc
Compton wavelength divide d, by 2rr
h
time -+
ffioc' reciprocal of the de Broglie frequency
associated with electron rest mass
energy
energy ffiocz electron rest mass energy

velocity r speed of light
angular momentum h Planck's constant divided bY 2n
In terms of the natural units and the coupling constant a, the relationships
that were previously found for the hydrogen atom may be expressed as follows:
ao:
hz 18,
First Bohr orbit:
ffloaz q.
-:-
ntoo4
Ground state energy: -Ws
:-- luzmocz
2h2
Rydberg constant: R)*:#r:#,

Orbital velocity: A:&C
t56
APPENDIX A t57
Thus, when expressed in natural units, we may regard the hydrogen atom
as being rather weakly bound, which is to say that thi electromag.r.ii. inter-
action is a weak interaction in the world of elementary particles. For instance,
suppose that the coupling constant o( were nearly unity- Then the Bohr radius
would be nearly the Compton wavelength, and the ground state of hydrogen
would be of the same order of magnituJe as the ,.ri-us energy of the elec-
tron. The density of matter then would be greater by the factor (137)3.
It is evident from the form of the orbital velocity that uz is the appropriate
correction for relativistic effects, which are always proportional to (iic1z.'
Some form of simplified units is always used in advanced courses,'although
it may not be the system described here. A frequent practice in quantulm
electrodynamics is to define h, : c : tnt: 1, in ordir to simplify all equations.
When the answer is obtained it is then multiplied by the-appropriite com-
bination of these universal constants which gives it the .otr.cf dimensions.
In a non-relativistic treatment of atomic systems, the "natural" units
turn out to be the so-called atomic units wherein the unit of length is the
Bohr radius ao, the unit of mass is mo, and energies are expressed in teims of ws.
SUGGESTED
REFERENCES
sidney Borowitz, Fundamentals of Quantum Mechanics. w. A. Benjamin, Inc.

New York, 1967-.
R. H. Dicke andJ. P. Wittke, Introductionto Qunntum Mechanics. Addison-Wesley
Publishing Co., Inc., Reading, Mass., 1960.
E. Merzbacher, Quantum Mechanics. John Wiley and Sons,.Inc., New york,
1961.
Albert Messiah, Quantum Mechanics. North-Holland publishing co.,
Amsterdam, 1958.
D. Park, Introduction to the Quantum Theory. McGraw-Hill Book co., Inc.,
New York, 1964.
L. Pauling and E. B. wilson, Introduction to Quantum Mechanics. McGraw-Hill
Book Co., Inc., New York, 1935.
J. L. Powell and B. crasemann, Quantum Mechanics. Addison-wesley publishing
Co., Inc., Reading, Mass., 1961.
F. K. Richtmyer, E. H. Kennard and J. N. cooper, Introduction to Modern
Phltsics (6th ed.). McGraw-Hill Book Co., New York, 1969.
V. Rojansky, Introductor2 Quantum Mechanics. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1938.
D. S. Saxon, Elementar2 Quantum Mechanics. Holden-Day, Inc., San Francisco,
1968.
Leonard I. Schifl Quantum Mechanics,3rd ed. McGraw-Hill Book Co., New
York, 1969.
Robert L. White, Basi.c Quantum Mechanics. McGraw-Hill Book Co., New
York, 1966.
CHAPTER 5
ffiffiffiWWffiffiffiffi ffiW ffiffiWffi ffiffiffi

ffiffiffiffiffiffiffiffiffieffi ffiwffiwffiMffi
In this chapter the Schrodinger wave equation will be applied to a few

one dimensional problems which will serve to illustrate the uppto*h ofquantum
mechanics as well as some of the interesting predictions of the theory. Although
the systems used here are idealized and are, consequently, somewhat artificiil,
this departure from reality serves to simplify the mathematics without destroy-
ing the essential physical features of the problems.
I. STEP POTENTIALS
We will begin with potentials that are constant in time and that are also
constant throughout prescribed regions of space. The simplest of these is the
step potential shown in Figure 5-1, in which the one dimensional space is
divided into two regions such thatV:0 for x 10,andV - Vofor x > 0.
Let us first consider the case when the total energy E of a particle of mass
m is greater than Zo. In the region where x I 0, which we will designate as
region I, the potential is zero and the time-independent Schrodinger equation
becomes
fftvt Yry:
2m dxz
Evr.
(a) (b)
t
Drift tubes II
x: 0 x --->
Figure 5-1. (a) A plot of energy vs position for a step potential. The dotted lines indicate
possible particle energies. (b) A physical approximation of a step potential. The dotted line indicates
a particle trajectory.
t58
STEP POTENTIALS I59
For a free particle in region I, the total energy of the particle is equal to its
kinetic energy, that is,
!? _,
2m-"'
or
!?:hzk?:2mE'
Thus, the Schrodinger equation for region I may be written as
drrl,,
-k?rp',
-:dxz
which has the general solution,
gr:Aeikte *BQ-o*rr. (5.1)
Following the same procedure for region II, we obtain

p? -- hzk? :2m(E - Vo)
d'rp' : ,
k'rrltrr,
dxz -
and
grt:Qsih2n lDe-ikr, (5.2)
Thus far we have said nothing about the direction of the incident particle. It
turns out that the continuous states for free particles are doubQ degenerate, that is,
there are two solutions for each value of the total energy. These solutions
correspond physically to the fact that a particle of energy E may be traveling
in either direction, *x. For convenience we will adopt the initial condition
that the particle arrives from the left. Therefore, the coefficient A in Equation
5.1 represents the amplitude of the wave associated with the incident particle,
.B is the amplitude of a reflected wave at the potential discontinuity at x : 0,
C is the amplitude of the transmitted wave in region II, and D is the amplitude
of an incident wave from the right. Our initial condition requires that D : 0.
Therefore,
Vtt : Qtihsa (5.3)
Equations 5.1 and 5.3 will together constitute a satisfactory solution of the
problem only if they fulfill all of the requirements of good behavior which
quantum mechanics imposes upon allowed wave functions. It was shown in an
earlier section that plane wave solutions of this type can be normalized so as to
satisfy the integral-square criterion. Although we will not normalize Equation
5.1 it is important to know that it canbe normalized. Two other requirements
are the continuity of the wave function itself and the existence of the first
derivative everywhere. Both of these conditions will be met if we require
91 and ?rr to be equal in magnitude and to have the same slope where they
160 SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
join at x :0. That is, we impose the boundary conditions:
-î:ll,:;;13,":,) (5 4)
From the first condition we have
A+B:C,
and from the second we obtain
c:\(A-
kr'
B\.
t
After solving these equations we find:

B _ft,
- ftr\
A-ml
c 2k, I (5'5)
V:G4)
Using the expression for the probability current in the previous chapter, we
have
sro: bvr
m
S"": -bWt
m
(5.6)
hk'
sr. : m
pp
Then the transmission and reflection coefficients are:
(5.i)
Note that T * ^R - 1, as it must if probability is to be conserved. Figure 5-2

t
E
Figure 5-2. Possible eigenfunctions for the

potential step. The amplitude of the wave is
larger in the region where the velocity of the
particle is lower, since the particle spends more
time in that region. The wavelength is shorter in
the region of higher kinetic energy.
x:0
STEP POTENTIALS 16I
illustrates schematically the behavior of solutions for the step potential. There
will always be a reflected wave unless the discontinuity in the potential vanishes.
Furthermore, Equation 5.7 indicates that the reflection coefficient is independent
of whether the potential discontinuity is a step up or a step down. Here the
behavior is quite different from that of a classical particle.
PROBTE/I4 5-,
Verify the probability currents given in Equation 5.6.
Now consider the case for a total particle energy less than Zo. Classically,
the particle would rebound from the step, but we will find that in quantum
mechanics there is a non-zero probability of finding the particle in region II.
The analysis of region I is identical with the first case and results in the solution
given by Equation 5.1. However, in region II we are now faced with the fact
that the momentum is imaginary since the total energy is less than the potential
energy. That is,
p? : hzKZ -- -2*(E - Vo) -- 2m(Vo - E).
This is equivalent to defining the propagation constant as
Kz : ikz
in Equation 5.2. The general solution to Schrodinger's equation in region II

then becomes
Vrt:Ce-Kz* lDeKrr.
Instead of the harmonic solutions that we obtained previously, we now have two
exponential terms, one increasing and the other decaying. Since the increasing
exponential is not well-behaved for positive tr we require D to be zero. There-
fore, the solution in region II is
K"
Vl: Ce (5.8)
Application of the boundary conditions leads to
B kL - iK}
v k, a iK,
(5.e)
C 2kt
:-
v k, I iKr'
and the result that
R:1 and T:0. (5. ro)
Here the reflection coefficient of unity, which agrees with the classical result,
arises from the assumption that no absorption occurs in region II. Thus, the
continuous damping results in continuous reflection until all of the incident

energy is ultirnately returned to region I.
PROBLEM 5.2
Verify the results given in Equations 5.9 and 5.10.
It is of interest to point out that for a time-dependent solution the boundary

conditions cannot be satisfied unless the same value for o is used on both sides
of the step. From our knowledge of the behavior of elastic and optical waves
we know that when a wave passes from one medium to another, the frequency
remains the same but the velocity and wavelength are both altered so that
())
- ku : constant. This, of course, assures the conservation of energy since
:
E ah. For a material particle this means that ar must be associated with the
total energy of the particle including its rest mass, which is consistent with the
original de Broglie hypothesis of Equation +.1+.*
2. THE FINITE POTENTIAL BARRIER
Consider the barrier shown in Figure 5-3, having height Vo and width
2a. Here there are three regions to be considered, but since the potential is
zero for x <
-o and for x ) a, these two regions have identical solutions.
For the case of E > Zo, following the same procedure as in the previous
section, we obtain the solutions:
gr : itsik* t Be-it'p, for x < -a

Vtt : Ceikze * De-ikzr, fot -a < x < a
Vttt:Feik", fotxla,
where hk.: t/Z*n and hkz: \/Wn.
t
"l
E<V.,
Figure 5-3. A finite potential barrier.
* This is confirmed experimentally by the coherent regeneration of short lifetime, neutral K mesons
inabsorbing material. See for example, Gunnar Kiillen, Elementar2 Particle Physics, Addison-Wesley
Publishing Co., Reading, Mass., 1964, p. 438.
THE FINITE POTENTIAL BARRIER t63
rl/
T
l
I
Figure 5-f. Transmission coefficient for a finite

square barrier. For E ) Vo the curve is given by Eq.
5.ll and for E < Vo it is given by Eq. 5.12.
Applying the boundary conditions (Equation 5.4) to ,p, and rp11 at )c -- -a and
to tp' and rpr' at x : q., orte can calculate all of the coefficients in terms of the
coefficient A. Then, from the probability currents the transmission and
reflection coefficients are determined to be
T:
l*l:l,.l'
l?:
l*t:l#l'
For the case of E > Vo the transmission coefficient is an oscillating function
as shown in Figure 5-4. An eigenfunction for this case is shown in Figure 5-5.
Its functional form is given by
I
T:
cosz (2k,a) * i e,.Elrr"' (2k,a)
' - L +dLTJsi*(2k,a)
T-
r-T (5.11)
It is evident that the transmission coefficient is unity whenever an integral

number of half wavelengths is contained inside the barrier, that is, when
2kra : nn, where z is an integer. This oscillation is the analog of such phenom-
ena as the optical transmission through a non-reflecting thin film, or the use of
Figure 5-5. Real part of barrier

eigenfunction. The function in region I
is a superposition of the incident and
reflected waves.
IU SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
matching stubs in wave guides, or the Ramsauer-Townsend effect* in the

scattering of low-energy electrons from noble gas atoms.
When E < Vr, corresponding to the classical case of no transmission, the
momentum within the barrier is imaginary as in the potential step discussed
previously. We will again find an exponentially damped wave within the
barrier, but if the barrier is sufficiently narrow an appreciable transmitted
wave can leak out into region III.
The functions 91 and ?ps1 vtt- the same as for E > Ze, but function g11
now becomes
lprr:gt*Kra *De-Kz',
where
hKr-ihkz:tt6.
'Ihe coefficients are obtained by satis$ring the boundary conditions as before.
The transmission coefficient for this case is:
T_
cosh2 (2Kra) *i(bn,- ft)'i'r" (2Kra)
of,
T- (5.12)
r- W
4E(vo Eisinh2
(2K'a)
-
It is of interest to evaluate the transmission coefficient for the case of f
j
E ,--' Zo when sinh (2Kra) * (2Kra); then, Equation 5.I2 becomes I
*l
T
I t'i
T:
D-7s ;1,kLay'
2m(vo-E)
'*#ulT
The same result is obtained by letting E ---> Vo in Equation 5.11.
Nature provides many examples of barrier penetration, although the
barriers in the real world are not rectangular in shape and they are generally
three dimensional. In spite of this, however, the idealized treatment given here
forms the basis for understanding all of the more sophisticated tunneling proc-
esses and frequently provides a reasonably good order-of-magnitude answer.
For example, alpha particle emission from heavy nuclei was first explainedl by
regarding it as a tunneling process through the nuclear barrier. Figure 5-6
shows a crude representation of a nuclear well which is about 40 MeV deep
for a heavy nucleus. The Coulomb barrier is roughly 16 MeV high in this
example and its breadth is of the order of 1 x 10-12 cm. A typical value for the
energy of an emitted alpha particle is 6 MeV, which is considerably below the
* At bombarding energies of about 0.7 eV, rare gas atoms are almost transparent to electrons.
This effect will be discussed in Chapter 10.
1 G. Gamow. Z. Playsik5lr204 (1928); R. W. Gurney and E. U. Condon, Ph1ts. Rea.33,127 (1929).
THE FINITE POTENTIAL BARRIER t55
E
(Mev) Coulomb repulsion
l6
Figure $'5. Approximation of the nuclear

Coulomb barrier by a rectangular barrier.
-40
summit of the barrier. Using Equation 5.12,
T--
1+ 4E(vovi Ei
sinh2 (2K'a)
-
: [1 + +*3 sinh2 (2Kra)]-1.
For 2Kra ) l, we may write,
T ,--, [sinh (2Kra)l-' ,-, lte'K'of- : l,s-4K2a
Now for the above example,
Kza-fi'vz;V;=-+H 2 x6.7 x 10-20(16 -6) x I.6x 10-6

:14.6
and
T '--'' 4e-58'4 a,' lQ-Zs'
This means that an alpha particle incident on the barrier has about one chance
in 1025 of tunneling through the barrier. Although this is a very small proba-
bility, we will see that a 6 MeV alpha particle trapped in the above nucleus
bangs against the barrier about 1021 times per second (see Problem 5-15).
Thus a tunneling event can be expected to occur with a probability of 10-25 x
1021 : t6-e per second, or the decay time is about 3 hours. To illustrate how
t66 SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
Figure 5-7. An optical example of barrier

penetration. A light ray is totally reflected in the
prism at the left. Howeyer, there is some penetration
of theair barrier and when this barrier is made narrow
enough a transmitted ray (dotted line) appears in the
prism at the right.
critically the mean decay time depends upon the strength of the barrier, suppose
that the barrier width were increased by only 20 percent, all other consiants
remaining the same. Then, T ,-' 10-30, and the mean decay time would be
,-,10e seconds or 100 years! Therefore, in spite of the crudeness of this model
it is able to accountfor tlte experimentalfact that the meun lifetirnes of atpha emitters aar)
-fro* 1 microsecond to l0ro Tears!
An elementary example of barrier penetration arises in optics in the case
of total internal reflection from a glass-air interface (see Figure 5-7). If a
second piece of glass is brought very near the reflecting surface, a weak trans-
mitted wave will appear in it. The intensity of the transmitted wave is strongly
dependent upon the thickness of the air barrier, and no transmission takei
place for barriers more than one wavelength thick.
There are numerous examples of the tunneling of electrons and quasi-
particles in solid state2 and molecular physics. However, there is little io be
gained by taking a detailed look at another example until we have developed
the mathematical techniques for treating more realistic potentials than the
rectangular potential. This will be done in Chapter 9.
PROBLEI4 5-3
Derive Equation 5. I l.
PROBLEM 5.4 i::ii:i:j,ir,:i i.l,:i:ir,i:l ii:ai:;:i::iill:i il::::,r,.r.:: ii.]l;ii; ril,r

=,!i:,,!;1,,1f,f:
Derive Equation 5.I2.
PROBTE/I4 5-5
A beam of 5-volt electrons impinges on a square potential

barrier of height 25 volts and width ao :0.52 A. What
fraction of the incident beam gets through the barrier ?
2 C. B. Duke, ed,, Tunneling in Solids: Solid State Physics Supplement 10, Academic Press, New York,
1969.
THE SOUARE WELL IN ONE DIMENSION
PROBLEM 5.5
For a 5-volt rectangular potential barrier what is the

minimum barrier thickness for 100 percent transmission of
l0-volt electrons ?
3. THE SQUARE WELL lN ONE DIMENSION

A square well potential is shown in Figure 5-8. For particle energies
greater than Votheeigenfunctions are similar to those of the barrier for E ) Vo,
ind the transmission coefficients are again given by Equation 5.11. For
energies less than Vowe will find that bound states are possible. These bound
statei exist for only certain discrete values of E, in contrast to the continuum
of states associated with all energies greater than Vo. In order to see how these
discrete bound states arise we will briefly outline the mathematical steps.
The solutions for the three regions are :
9r : AeK" , forx<-&
lprt : Ceikr' -f De-ikr*, fot -a < x < a
llrrt : Fe-K* , forxla,

where
hKr : ihh : t/Zrr4V, - 81, and hkz : t/Z*n'
Imposing the boundary conditions, Equation 5.4, aI x : -a one obtains:
c :4 r,ik'a (5'I3)
D- '
At r: at the boundary conditions-k;;11'o

give the following result:
c _ k, * i\ "_roo,o (5.r4)
D k2 - iKr-
-:-E
1
E
Figure 5-8. A square well potential.

ttl
E>V,
AV
V\, Figure 5-9. Eigenfunctions for the square
E1
well potential.
/,\
E,]
\/iv
E1
,-l\..
E:O x:0
I
The product of Equations 5.13 and 5.14 is CzlDz : 1, or
C : X.D.
For C : D, the boundary conditions require A : F and rp'r1 becomes a cosine

function; the eigensolution is then symmetric about the center of the well as
illustrated by Erin Figure 5-9. For C -
-D, A: -F, Vn becomes a sine
function, and the eigensolution is antisymmetric about the axis of the well,
such as Ea in Figure 5-9. Note that the penetration of the walls increases
with E.
In order to obtain specific information on the number of allowed solutions
and the energy associated with each, it is convenient to transform the above
equations in the following manner.
Let a : kza and B : Kta. Then, setting CID : I in either Equation 5.13
or 5.14, there results:
kr(tn" - t-n"): iKt(eio + e-n')

&, sin a : Kr cos oc
tanu:JK,B
k2 -'- d.
: f , for symmetric solutions.

a tan o (5.15)
In a similar fashion the condition CID : -l yields the equation

a cot n : -F, for antisymmetric solutions. (5.16)
Note that
a.2 + f, : or(k31 ^rcrr)

: #l2mE * 2m(vo - E)l
THE SOUARE WELL IN ONE DIMENSION t69
ofr
q.2 +F':ry (5.17)
Thus, the symmetric solutions can be obtained by solving Equations 5.15 and
5.I7 graphically, and the antisymmetric solutions are obtained by solving
Equations 5.16 and 5.17. These solutions are sketched in Figure 5-10. Note
that the number of solutions depends upon the radius of the circle, Equation
5.17, which in turn depends upon the mass of the particle, the depth of the well
and the width of the well. The prodtct a% is a measure of the "strength" of
the well in terms of binding a particle. That is, the greater the depth and
breadth of a well, the greater the number of bound states and the greater the
probability of retaining a particle in the well. In particular, if
(rn),.ry<@*r),n,,
then there are n f I symmetric bound states. Similarly, if
(n_ t),n,.ry <@*.t),n,,
then there are n antisymmetric bound states in the well. The total number
of bound solutions exceeds by one the largest integer containe din (alnh)\/rmv,
Note that there is always at least one bound symmetric state, regardless of how
shallow the well is.
In the limiting case of an infinitely deep well the energy eigenvalues are
readily obtained without performing a graphical solution. As Vo - oo, the
radius of the circle, Equation 5.17, becomes infinite and will intersect Equations
il
Figure 5-10. Graphical solu-

tions for the square well. Note that ot+B'
Ior az * 0':4, there is one anti-
symmetric solution and there are
two sym metric solutions.
Irr2
2
I7O SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
.c io
'd-o\
uo
I-/ -::+-_ L_= t-
'tr
o
o $
o @ sfn
@
e(
!rt
o
(ltL
=- -t'- 5G'
IJ !L
crc
:i
_+
1
o o -o
o o @
+,
I o
*oo :h
'0€!
t VFt
_) t-
OJ
o
{
sl o
o
N
SvE
-.c
sO
(, fit
:o
et
o
cs
GtI
.e$
L
>\o
o o o bo -o
to !r
o
$
o
ts -,/ co
{rlJ
tu
,{
tr<
t--
9T
r( d
-o 6E
o{
o o
€ -c-
F
o o
@
1
cr=
'L- -
L>\
G'L
,d
-Od
o!
dlt
o' fit
6c
o an '-
o
@
c0
LF
o.g
&r,
bo c)
et
t-
q, (,
r_ 9+,
H9
s<
Hd
et
Q)6
o o JL
$ 9 (J(u
N si fit lt
o-g
O=
at1
__ r_ )
GI
:_=€_ ci
.; .90
24)
j
o
N
o
N
o
@
o
er!- b.E
'- t^
-dc
g.
_ _f - ?.e
L e-c
ioe':
o o
L i;
o @ o
N rl
THE SGIUARE WELL IN ONE DIMENSION t7t
,i
o-c
.9.c
_l
''
gT
!
-f
o o iri
q,
@l
ol
N
I io
T' O\
,!
En;
tr\
-1
I
cl-
*'l .tr tri
*
u __l +, tyi
o
a6
(ol
o
o
! Ef
cr. o-
9E
-t
--- €<
o
1 I
€i
ELr
-cl
o<===-
?.-/
o
N
F
rt
o T.:
>u
Qv,
frl
oc
cct
O1
a0)
--l -=
_-l
ll'-c
-t4 O
o
n o o
o
o o
o oo z-
O'J
c+
o
opo
ll ::_l -4- _l
F"i
{, T'
o o
o
o
o
38
G
o o o F<
=o
og
__J ...:I
---t '<
o
j Es
o
.r-- -=\
o
o
rf
o
o
N
df
a\
EL
o!n
r-LL .*,
@lr
CL
'tr Gt
I Oa
+, .=
..1
__==n
+,
8s
(,h
-1 6E
JO
q-c
o
rf
ol
{l o
N el @
(o f,+,
og
>o
Gt-
;6T'
c.
6l
--r .er
a3
cth
o o o (,r
o
N
(,
n
\i
@ z
1\l
.A t-
j E€?
3- o
-l .g0T't
E -.E
. --- o
@
o
N
o
€t
oL
5.15 and 5.16 at their asymptotic limits. Then we have:
For symmetric solutions t & :;,T rT, .. .
For antisymmetric solutions, d, : trr 2nr 3r, . . .
fn general, a : nnl2, where the lowest state is symmetric, the next antisym-
metric, and so on, alternately. Since u : k20.,
k2a:;,(ilh=7,
E^: nzrPhz
Bmaz
(5. rB)
This will be recognized as the answer to the problem of a particle in a box with
rigid walls as solved by the old quantum theory (Chapter 4, Problem 2).
Note that n : 0 does not correspond to a physical state since it would require
that Y : 0 everYwhere in the box.
The eigenfunctions for the infinite square well do not leak out of the walls
of the well as they do in the finite potential well (Fig. 5-9). Since each solution
must have zero amplitude at both walls, the eigenfunctions then become sine
or cosine functions.
Before closing this section it is important to point out that a rcal particle is
represented by a wave packet consisting of many * values instead of just a
single value of ,t as we have assumed in the foregoing analysis. Furthermore,
since the time dependence of each component differs from the others, a com-
plete analysis of the interaction of a packet with a barrier or a well is quite
complicated even in one dimension. Goldberg et al\ have obtained computer
solutions and computer-generated motion pictures of Gaussian packets
scattering from one dimensional barriers and wells. Two of their scattering
events are illustrated in Figures 5-11 and 5-12.
PROBLEM 5.7
Assuming an infinite square well of radius 2.8 x 10-13 cm,

find the normalized wave functions and the energies of the
four lowest states for a nucleon.
PROBI.EM 5.8
Show that the energy of the lowest state in the infinite square
well is consistent with the uncertainty in momentum as
required by the uncertainty principle. Would the state
E : 0 satisfy the uncertainty principle ?
3 A. Goldberg, E[. M. Schey and J. L. Schwartz, Amer. J. Ph1ts.35, 177 (1967).

THE SEUARE WELL IN ONE DIMENSION t73
PROBLEM 5.9
(u) For the square well shown in Figure 5-8, find the
function tp11 defined in the region
-0. < x < a for the
caseofE>Vo.
(b) For what values of E > [ is the total probability with-
in the well a maximum ? A minimum ?
(") Is the maximum value of tL, p dx found in (b) greater
than what it would be in the absence of the well ?
Physically, this means that particles are "trapped" for
a short while in the well. Such events are called
resonances.
PROBLEM 5.IO
Calculate the minimum uncertainty product, Ar Lp, for

the ground state and the first excited state of the infinite
square well.
PROBIE/I4 5-r' F,li'ii,$?li:.1
Sketch the wave function ,p"(x) and the probability density

p"(x) for each of the three lowest eigenfunctions for the
infinite square well.
PROBLEM 5.12
Find the probability current density for the nth state of the
infinite square well.
PROBI.EM 5-'3
Consider a state function which is a superposition of the two

lowest eigenfunctions of the infinite square well, that is,
lnx 2.nx
1p: + srn
-a
{a
-COS- 2a \/a
-,-
(u) Normalize this wave function in the sPace -a < x I a.
(b) Sketch the probability density function, p(x).
(.) Include the appropriate time factor in each term of the
wave function and find the probability density function,
p(x, t).
(d) Using the normalizedwave function in (a), calculate the
expectation value of the kinetic energy.
(Ans. : (d) 17 Tznzl4\maz)
PROBLEM S.I4
Use the uncertainty principle to give an argument against

the existence of an electron in the nucleus of an atom. Take
the nuclear diameter to be ^,10-12 cm.
PROBIE/I4 5-r5
Approximately how many collisions per second would an

alpha particle have with the walls of a nuclear well of radius
I x 10-12 cm ?
4. MULTIPLE SQUARE WELLS
If two identical square wells are isolated from each other and each has a
particle in its ground state, the energy of the system is just the sum of the two
ground state energies. Now, if the two wells are separated by only avery narrow
barrier they will interact and the total wave function is a linear combination
of the two single-well ground state functions as shown schematically in Figure
5-13. In (a) the total wave function is symmetric in the two wells, and in (b)
it is antisymmetric. The energies of these two states are not the same, and the
energy splitting increases as the width of the barrier decreases. This may be
understood qualitatively by means of the following argument. In the limit as
the barrier vanishes, the symmetric wave function corresponds to the ground
state of a well having twice the width of the original wells; hence the energy
decreases. As the wells coalesce, the antisymmetric function, on the other hand,
corresponds to the first excited state of the well of double width. Thus, the
energy decrease due to the increased width of the well is compensated by
excitation of the particle to the first state above the ground state. The energy
of the double-well system for these two cases is shown in Figure 5-I4, where r is
the width of the barrier.
The most important feature of this model is that the two wells, which have
the same energy levels at large separations, interact more strongly as they are
brought closer together, and the interaction splits each original level into two
separate energy levels.
An important generalization of this model was made by Kronig and
Pennet' in treating the electron energies in solids. Here the number of
(a) (b)
Figure 5-13. Total wave functions for two interacting square wetls, (a) symmetric case, (b)
antisymmetric case.
{ R. de L. Kronig and W. G. Penney, Proc. Ro1. Soc. (London) A l3O, 499 (1931).
MULTIPLE SOUARE WELLS t7s
t
E
Figure 5-14. Energy versus barrier width r for two Antisymmetric

identical square wells.
Symmetric
interacting wells, l/, is extremely large so that each of the single-well levels is
split into y'/ Ievels spaced so close together that they form nearly continuous en-
ergy bands. Figure 5-15 shows a schematic square-well representation of the
potential seen by an electron due to a linear lattice of positive ions. Since the
potential function has the period of the lattice, that is, V(r + d) : V(x), we
expect our solution to have the same periodicity, provided that we avoid the
ends of the chain. This difficulty can be eliminated by imagining that the chain
is joined into a ring containing a large number of atoms (of the order of
Avogadro's number).
It was shown by Blochs that the solution for such a periodic lattice may be
written in the form of the product of a plane wave and a function having the
periodicity of the lattice. Thus,
,p(x) : eil'*u(x),
where u(x + d) : u(x) and k :2'rnxlNd. Thus,
,P(x + d) : eo*rP(x).
The last result enables us to obtain the solution anywhere in the lattice once
it is known for a single well. If we denote the width of the well by a, the width
of the barrier by D and the height of the barrier by Vo, then we may define the
propagation constants,
hkL : {2rrE and hKz : ihkz : \/MV, - E),
as before. Although the algebra is laborious, the procedure is straightforward

and one obtains:
For E I Vo,
cos kra. cosh Krb - (W) ,i., ft1a . sinh Kzb : ,orff ,
Figure 5-15. A sguare-well repre-

sentation of the potential seen by an
electron due to a linear array of positive
ions.
6 F. Bloch, Z. Physik 52' 555 (1928). This was first proven by Floquet in 1883.
andforElVo,
+ 'tit srn
' h'a'
- (? ' stn k'b : 2nn
cos kra. cos krb cos
\ffi) ..n/ '
Sincg the right hand side of each of these equations is restricted to the range
of values between I and - l, solutions exist only for those energies for which the
magnitude of the left hand side does not exceed unity. As a consequence of
this restriction, the energieswhich correspond to "physically" realizableiolutions
form continuous bands which are separated by forbidden bands of energies for
which no solutions exist.
Although the details of the energy band structure of a solid require a
much more refined model than that used here, it is instructive to see that the
general quantum mechanical behavior of the electrons can be predicted by
just using the elementary concepts of wells and barriers devel,oped in the
previous sections.
5. THE HARMONIC OSCILLATOR:

POLYNOMIAL SOLUTIONo
The harmonic oscillator is one of the most important problems in modern

physics. The overriding reason for its dominant role is that complex systems
can be reduced, by means of Fourier analysis, to the solution of a colleciion of
harmonic oscillators. Thus, the harmonic oscillator approximation has
become one of the principal methods for studying the colleiii,re excitations of
many-body systems such as collections of phonons and photons. We will use
two different methods to solve the harmonic oscillator. In this section the
polynomial method of Sommerfeld will be described, and later we will use the
more elegant operator techniques which are based on the factorization method
of Schr<idinger and Dirac.
The classical Hamiltonian for a linear harmonic oscillator is given by the
expression,
ff:!" { $azmxz,
2m
and by means of the correspondence principle the Hamiltonian operator

equivalent is
h2 d2
.ff: * lt fmxz
2m dxz
Then the energy eigenstates of the oscillator will be given by the solutions of the
time-independent Schrodinger equation,
#*r _ orxr)rl, _ 0, (5.1e)
6 A. Sommerfeld, Waae Mechandcs. Dutton, New York, 1929, p. ll.

THE HARMONIC OSCILLATOR: POLYNOMIAL SOLUTION 177
where the parameters or and p are given by:
n:ff and p:ry.

It is convenient to transform Equation 5.19 to the dimensionless variable
q, where
q: lix'
Substituting into Equation 5.19 the derivatives,
dy _dVdq _
dr-dqn- ^/:dV
"*dq
drrp: d 1&p1dqn4,
drrp
dr, aq\dr)n:
there results
:0,
# . (e - q')Y' (5'20)
where we have defined e as the dimensionless eigenvalue
, :;:
p2E
,h.
In order to find acceptable solutions of Equation 5.20, we will first con-
sider the form of the asymptotic solutions as q becomes infinite. Then it will be
assumed that a satisfactory solution of Equation 5.20 can be expressed as the
product of the asymptotic solution and a finite polynomial in q. That is,
V : Vo' H(q),
where H(q) is a power series in q which will later be terminated to the ap-
propriate polynomial. In order to arrive at the asymptotic solution, we note
that as 4 gets very large, Equation 5.20 may be approximated by
ry
all' - Qzqso
:0, (5.21)
which has the solution,

lpa:/sazlz*Be-q'lz.
We note that A must be zero in order to insure that yo is a well-behaved

function. If ,l,o is substituted into Equation 5.21 it is immediately obvious that
the latter is not satisfied exactly. That is,
d'rlto :
-
o
gzlso : t\o, where !)o frs-azlz'
#
I78 SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
However, since go vanishes as Q * @, rpo satisfies both Equations 5.20 and 5.21
in the asymptotic limit. Our trial solution of Equation 5.20 may now be
written as
g- e-q'lrH(q),
where the coefficient 3 has been absorbed in the polynomial. To determine

H(q) we now substitute rp into Equation 5.20 and obtain
H" (q)
- 2qH' (q) * (. - L)H(q) : 0, (5.22)
where the primes indicate differentiation with respect to q. Equation 5.22 is

known as the Hermite equation. It can be solved by assuming a power series
of the form
H(q) : : a*Ik : ao * atr * arr' * asQs I'''
,::
H'(q) : - at * 2a2q * 3arq, + ...
Ano*Otc-l
and
H"(q): : k& - L)aoqt-z :2az + 2.3arq * 3 .4anqz + .. .
Substituting ,rr.rl:i,rrr.rrorrs into Equation s.22 and,setting the coefficients of

each term of the resulting power series separately to zero, we obtain the
following relations:
2or+(e-l)ao:g
2'3at * (. - I - 2)at:g
3.4ar * (. - I - 2.2)ar:g
fn general,
(k + l)(r + 2)or*r* (. - I - 2k)a*: 0,
or
e - I -2k
ulc+z (5.23)
- - (ka116a27ax
Equation 5.23, which is called the recursion relation, shows how all of the co-
efficients can be determined from aoand a1. Thus, aoand arare the two arbitrary
constants required for the general solution of the second order differential
equation. The solution of Equation 5.22 may be written as the sum of two
series, one containing all odd powers of q and the other all even powers:
H(q) -,,(, *;qz +; ;.fir + ) .

;qn
+2
\
* a,(o * fir" +?,.fir' +; ; 7,{ * . . .)
(5.24)
'
THE HARMONIC OSCILLATOR: POLYNOMIAL SOLUTION 179
If the series does not terminate, the trial solution,
V - e-o't'H(q),
will diverge as q becomes infinite and
hence will be an unsatisfactory wave
function. This can be seen by examining the behavior of H(q) as & gets large.
From Equation 5.23,
at+z 2
o**k
Note that eq' : I + qz + *qn +. . . + 2kqzklk! + . . ., and the ratio of two
consecutive terms as ft gets large is
b**, 2
b* - k'
Thus, H(q) goes as e0', except for a constant factor, and
'tP ,- g-azlz ' gaz ,-,t ga2l2, WhiCh diVergeS.?
From this we conclude that the series solution of H(q) must be terminated to a pofu-
nomial in order to obtain a plrysicalfu acceptable wauefunction. The recursion relaiion,
Equation 5.23, tells us that whe n k : z, such that e : 2n * 1, one of the series
will terminate with a,,, since an*, and all higher coefficients will be zero. The
other series must be eliminated by setting ao equal to zero if z is odd, ot a1
equal to zero if z is even. As a consequence of the termination of the series we
obtain the energy eigenvalues from the condition
2E
::rh:2n * l' (5.25)
'
that is,
E : ah(n * il.
We may now label the energy states and the wave functions by means of the
index z which indicates the degree of the polynomial appearing in the solution.
That is,
,p"(q) : e-aztz17n7o1 and En : afi(n + t).
Note that the energy levels are equally spaced at intervals of o,h, as in the old
quantum theory. (See Figure 5-16.) However, the ground state is not zero, as
in the older theory, but is
Eo : $ah.
This energy is the so-called zero-point energy, which accounts for the fact that
7 P. M. Morse and H. Feshbach, Methods of Theoretical Physics. McGraw-Hill Book Co., Inc., New
York, 1953, Part II, p. 1640.
v: j''mx'
E,:1dh
Figure 5-16. Energy levels of
*--:--r,.:f;,,h the harmonic oscillator.
Yr:|rn
+----
+rA
"r:
elementary excitations can be present in a statistical system even when it is in

thermodynamic equilibrium at absolute zero.
As in the case of the bound states of the square well, the harmonic oscil-
lator provides another example of how the requirement of well-behaved wave
functions results in energy quantization.
5. METHODS OF GENERATING THE

HERMITE POLYNOMIALS
To illustrate how the Hermite polynomials can be obtained from the

Hr(q). From Equation 5.25 one obtains
recursion relation, consider
: 2(3) +|: 7'
Using e : 7:in Equati"" 5.2;
As 7 -t -2 2
d1 2.3
Then, from Equation 5.24,
Hr(q) : at(Q - &q").
The value of a, is arbitrary and will ultimately be absorbed in the normali-

zation factor of the wave function. However, a convenient convention is to
choose a1 (or ao) such that the coefficient of the term of highest order is 2".
In this example a1 would then be -12.
A few of the Hermite polynomials normalized in this fashion are as
follows:
Hok) : L
H,(q) - 2q
Hr(q):4qz -2
Hr(q)-Bq'-l2q
Hn(il:16q4 -4$qz +12.
Note that H*(q) is odd or even as z is odd or even. In quantum mechanics the
term parit2 is used to describe "even-ness" or "odd-ness." That is, ^F/"(g) has
METHODS OF GENERATING THE HERMITE POLYNOMIALS I8I
even parity if z is even, and odd parity if z is odd. Since e-qztz is an even
function, we note that y*(q) has the same parity as H*(q) and n.
In order to obtain two additional prescriptions for constructing the
Hermite polynomials, let us first define the following function of two variables,
F(s, q) - exp lqn - (s - q)rf : exp [s(2q - r)] :|n,rq) .t;. (5.26)
The last expression is merely aTaylor expansion ofI (t, q) about s : 0. Thus,
the A"(q) are given by
A,(q) : #F(r, r)]s:o ," #(r-,,-0,')],:0.

But 010(s - q) : 010s, since 4 is held constant here. Then,
A*(q)-eo,#(e_t,_ol,;]s:0(-1)",",#-(,_.,_o,,)],:o.
Ot,
A*(q): (-t)nraz{Ur". (5.27)
To show that the A"(q) are actually the Hermite polynomials, H,(q), we
proceed as follows: Taking the partial derivatives of Equation 5.26,
AF
(5.28)
*:2(q-s)F
!
dq
:2,, (5.2e)
H:2F
oq os
+2r{. os
(5.30)
Adding Equations 5.28 and 5.29,
a! _0F _r,n
a, - Tq -'v'',
and differentiating this sum with respect to q,
a2F
+2q{q (5.31)
@.Yr':2F
Eliminating the mixed derivatives from Equations 5.30 and 5.31,
AzF AF AF
(5.32)
@-'ouo*2s ar:u'
Now, using the definition of F in Equation 5.26, we find lFl0s:
#:2,o-r, # (5.33)
Substituting into Equation 5.32 we have
? P'A"
aq' - ,o* * 2n A-) ''i : r.
"?u\
In order for this to hold for all J we must have
Att _
1Ln 2qA**2nAn:0,
where the primes signify differentiation with respect to the argument q. But
the latter expression is identical with the Hermite equation, Equation 5.22,
with e replaced by its proper value of 2n + 1. We therefore conclude that the
An are the Hermite polynomials. Replacin g A*with Hn, Equations 5.26 and 5.27
then provide us with the following two new definitions of the Hermite poly-
nomials:
F(r, q): exp lq, - (s

- q)rf : exp [s(2g - r)] .t;, (5.34)
2n"fq)
and =
H*(q) : (- 9" s-o'
' oqn"
L)nsaz (s.35)
The first of these, Equation 5.3+, says that if the generating function is
expanded in a power series in s, the coefficient of s" is the Hermite polynomial
of order z divided by nl. The second, Equation 5.35, enables the calculation of
the polynomial of any order by means of a sequence of differentiations.
PROBLEM 5.T6
Calculate Hr(q) and Hn(q) bV means of the three prescrip-

tions given by Equations 5.24, 5.3+, and 5.35.
Some useful identities may be easily derived which can provide a ready
means of obtaining Hermite polynomials of other orders if H"(q) is known. By
equating Equations 5.28 and 5.33 it follows that:
Of,
@
sn-' @ cn*L
2 H"(q) . (z 1)! -g
m:o
-
1 2 H"(q)
' n:o ;l -2q 2r"al ''n :0.
THE HARMONIC OSCILLATOR WAYE FUNCTIONS I83
Collecting terms in like powers of s,
hlr"*k) *, .
_ rqH"Jd;lql r, :o
.
ffi
fn order for this to hold for arbitrary s, we must have
H"*r(q) * 2nH^_r(q) - 2qH"(q) - 0. (5.36)
In like manner, by equating the expressions for 0Fl0q from Equations 5.26 and
5.29, it follows that:
}r;ra .;: z,ion*1q) .';.,

Ul@ :2nlI*-r(q). (5.37)
The latter may be extended to higher derivatives as follows:
Ui@ :2'2n(n - l)H"-r(q) :4n(n - T)H"-r(q)

uii(il : Bn(n - 1) (n - 2)H,-r(q), and so on.
PROBLEM S-r7 i+1iii1{-i{ii,trtril
(u) Derive the following additional identities:
H"*'(q) :2qH^(q) - ni,k)

H"(-q) : (-l)"H"(q).
(b) Verify the four identities by substituting the polynomials
and their derivatives into the above expressions
(for n - +).
iiiiif iiailiiij,
7. NORMALIZATION OF THE HARMONIC

OSCILLATOR WAVE FUNCTIONS
Letting ff" b. the normalization coefficient to be determined, we write

the wave function for the nth state of a linear harmonic oscillator as
,P"(q) : Nne-o'/'H,(q).
Then we require that
t*-lr,{o)l'dq
: t.
At this point we will use the Dirac notation for such integrals over all of space
(see section 13 of Chapter 4) :
(,p,(q) f*
l,p-(q)): J-a ,p\(q),p^(q) dq,
where it is understood that the complex conjugate of the quantity to the left of
the vertical bar is used in performing the integration. Then we may write the
normalization integral as
(.p*lrp*) : lNnNnlf*r-*u-(q)H*,(q) dq : lNnNn,l @*(q) I H*, (q)>,
where it should be noted that the factor e-q' is understood in the last integral.
The condition n' : n may be imposed at alater time. The integral may be
evaluated in two ways. First, Equation 5.35 may be integrated by parts z
times, assuming n' > n, to obtain
(H*(q) I H*(q)) : (
- t)n'-n2n'n, ,l**ro(q) d.q: 0,
ffi e-o'
since e-q' and all of its derivatives vanishs for infinite q. For n' : flt the integral
becomes
(H,(q) I u-(q)> : 2nnt f* uo(o)e-a'z d,q : 2nnI\/;.

J-q
Therefore, we note that the Hermite polynomials are ortltogonal with respect
to the weighting factor r-c', since the integral vanishes for n' * n. For n'-: n
the normalization factor is determined by the result,
lN,l'2"nlt/n : l.
The normalized wave functions for the linear harmonic oscillator are then
,p*(q): (;;'-o'''Hn(q). (5.s8)
Normalization may also be accomplished by means of the definition in

Equation 5.34. Define two functions F and G such that:
F (q,
') in!
G(q, r')
# n'!
8 R. Courant and D. Hilbert, Methods of Mathematical Phlsics. fnterscience Publishers, fnc.,
New
York, 1953, Vol. I, p.92.
THE HARMONIC OSCILLATOR WAVE FUNCTIONS I85
Then,
f@ - fco Fl (n\FI .(n\ f@

FGe-o' dq :
IJ__ e-o' d.q
- ,r*'I:_s-k-s-s'tz 4r.
,Z,rnr'n'!-*u"(t)rur'(q)
Defining u : g- r- s', the integral on the right becomes
f*
,-u'du : t/n.
J--
Thus,
2@Jq) lH*@>;.':,, : 1/]sx,'
_r
:t/rlt*2ss'-T- , (2ss')" I
,(2ss,)r,
- ol -r"'l'
:G1ry
It is immediately obvious that there can be no solution unless n' - n, since the
two series expansions can be equal only when this condition is met. For n' : n
we have
; t(*l @*k) t H,(q)> -q(,,')' : o
Therefore,
(H^(q) I U"(q)) : 2"t/n n!, (5.3e)
and for n' + n, the integrand is zero as before.

It is sometimes convenient to use the normalization over r rather than the
dimensionless parameter 4. In this case the normalized wave function becomes
,t*@) : (5f r,,,rzH*({i4 . (5.40)
-Before closing this section it will be instructive to evaluate an integral of

the type that is used to calculate expectation values and transition probabilities
for the harmonic oscillator. Such integrals are called matrix elements in the
matrix formulation of quantum mechanics. Consider the following matrix
element of the operator r:
(,p*(x) I *pn(r)) :l_ ,pI',)x,p*,(x) dx : dq

+rll_r:fqlqv*,@)
:
*(,p*(q)lqy,*(q)),
where the wave functions in Equation 5.40 are used for the integral on the left
and those in Equation 5.38 are used for the integral on the right. Using the
latter and the definitions in Equation 5.26, *. *tit.:
:
Z,;. '# @-(q) I qH-,(q))
(,p,(q) I qy,*,(q))
- ,zsst f'r-,0-r-r'r' (q-

J--
\r r - s,)dq* (, f ,,)t ,rrr'l-
' t ' \ ' J--
s-k-s-s')z 4,t
: t/n(t I s')ezu'.
Then,
A,lrr"(q)lqH-,k)>;
';l: rn[('. 2szs' *T + "'
*ry + ) * (,, *2ss,z +T +.. .*r"'li'" . )]
By comparing series we see that the first series of the right member can be
equated to the left member only when nt - n
- I; the second series on the
right can be equated to the left only when n' - n -f l; for n' : n the integral
must vanish. The results are summarized below:
<H*l IHn+) :2n(n + :

\lf; (!,*l rrp,*',
(H"lqH*):0
J#'l
(V*lxy;*):0 ) 1S.+fy
<H*l IH*-r) - 2n-rnl\/; (v*l rrp*-, :

J*, )
The following example will illustrate further the method of evaluating expec-
tation values by means of the orthogonality properties of the Hermite poly-
nomials.
EXAMPLE
Calculate the expectation values (x), (!,), <T), <V) and (E) for the linear
harmonic oscillator. Utilizing the identities 5.36 and 5.37, we have
H**r(q) - 2qH"(q) + 2rH"-'(q) : 0. (5.42)
Multiplying by Hn(q)t-o' and integrating over all of space,

(H"l Hn+t) - 2(H*l qH"> I 2n(H,l H,-r> - 0.
From the orthogonality property of the Hermite polynomials, it follows that

THE HARMONIC OSCILLATOR WAVE FUNCTIONS I87
(H"l H^*r) : (Hnl H*_r> - 0. Therefore,

(H,lqHn):o (r):
*rr-lqH^)-o.
Now multiply Equation 5.a2 by Hngr-o'and integrate:
<H"l IHn+r) - 2(H,l qrH,) * 2n(H*l qH,_r) - 0.
Substituting the values for (HnlqH**r) and, (HnlqH*-) from Equation 5.41,
we obtain
<H,l q'H*> : ln-r(n + 1) I\/; I n, 2"-rnl\/; : W_t !) . 2"-'!I\/;.

Then
(xr) : lrr-l q,,p*) : q,H*> :'" *t
#@,1 .
To find the expectation value of the mornentum, we use the expression
(!,) : -th'/;(r"l +> : +ih\/i(,p*l Qe*) - ih\/ 2n\/;(y,,1 vn-),

where the identity given by Equation 5.37 was used in the last expression. Since
the first integral has been already shown to be zero, we have
(!,) - <H"l Hn-) : o,
from the orthogonality relation. Now, to find 1p2*), recall that
!r,: -h, #: -ah.z #

Utilizing Equation 5.20 we note that
.,*:.\ n'
: q2>. L-')
Q7> ah}(e -
Utilizing our results above,
<p',> : nh, (2n + I - ry) : aft.z

(ry)
Then'
(T) : fiuz, :g*(n *+) : ] a+ +),
(v) :ff :# <r,> @ +*) : j a+ +),

and
(E) : (f) + (V) : ath,(n + +).
The correspondence between the quantum mechanical expectation values

and their classical counterparts is again illustrated by the above example.
A few of the normalized harmonic oscillator wave functions are:
,po(q): (+)+r-oz1z
wnl
,p,(q) : (#Joe-qz,
,p,(q): (#f rrr -r)e-azrz
,p"(q) : (#f rro' - 3q)e-aztz
,pnh) :l(Hr$on - t2qz -t 3)e a'rz
,pu(q) :tr(#J(4q' - 2oq, + !Sq)e-a'tz
Some of these functions are sketched in Figure 5-I7 .
Figure $17. The first four harmonic

oscillator wave functions. The dashed lines
indicate the classical limits of motion.
APPLICATIONS TO MOLECULAR VIBRATIONS 189
8. APPLICATIONS TO MOLECULAR VIBRATIONS
An obvious application of the harmonic oscillator solutions is to the

vibrational spectra of linear diatomic molecules. A schematic diagram of the
potential energy versus internuclear distance for such a molecule is shown in
Figure 5-18. Note that this potential may be approximated by the parabolic
harmonic oscillator potential, at least for small oscillations about the equilib-
rium position, ro. Therefore, to the extent that this approximation is valid,
the vibrational energy states are given by
En:ath,(n++),
where 6sz - klp,, the force constant divided by the reduced mass of the two
atoms. A better fit to the real potential is obtained by using the so-called
Morse potential,
V(r) : VolL - r-a(o-totlz - Vo.
Using this potential, the energy levels becomee
12
En:afi(n++)
+ +) -
-14 tzfn@ * +)J
The effect of the second term is to depress the levels as z increases. Hence, the
interval between successive levels diminishes instead of remaining constant as
in the harmonic oscillator. This is shown schematically in Figure 5-19.
Another model that sometimes proves useful in molecular problems is the
double oscillator with a barrier between the two oscillator wells (see Figure
5-20). Aparticle having energy E <Vocan tunnel from one oscillator to the
Figure 5-18. Potential energy versus

interatomic distance for a diatomic molecule.
o See L. Pauling and E. B. Wilson, Introductionto Qwntum Meclwnics, McGraw-Hill Book Co., 1935,
p.271.
1
E
Figure 5-19. The vibrational

levels for the Morse potential.
other at a rate determined by the height and breadth of the barrier and the
total energy of the particle. This model can be used to describe the oscillation
of the nitrogen atom between two stable configurations of the ammonia mole-
cule. The geometrical arrangement of this molecrrle is that of a tetrahedron
whose basal plane contains the three hydrogen atoms at the vertices of an
equilateral triangle. The nitrogen atom is at the fourth vertex, either above or
below the basal plane (see Figure 5-21). The oscillation of the nitrogen atom
between these two equilibrium positions is a tunneling process whose qualitative
features are similar to those of the double square well of section 4. A rigorous
solution of the double oscillator shows that the energy splitting of the two
lowest levels, for example, isro
L,E : +r, lL g-vol€o ,
V zeo
Figure 5-20. The double os-

cillator. Each energ), level of a
single oscillator is split into two
levels. For E < Votunneling occurs.
10 Merzbacher, Quanturn Meclunics, John Wiley and Sons, New York, 1961, p. 64 cf.
THE HARMONIC OSCILLATOR: OPERATOR METHoD r9t
Figure 5-21. The ammonia molecule. The nitro-

gen atom oscillates between two stable configurations
by means of tunneling. ln each configuration it, ex-
periences a potential that is approximately harmonic.
where Vo ) €0 : $ah, the ground state energy of either single oscillator. This
corresponds to a tunneling frequency of LElh. The ratio of the tunneling time
z to the period of either oscillator, ?", it
'r -! i7". ovot,o

2ln
It is evident that the tunneling process is greatly affected by the barrier height
in the exponential factor.
(u) Calculate the vibrational energy states for the CO

molecule in the harmonic oscillator approximation.
Take the force constant to be k: I.B7 x l05dynes/
cm. Will any of these states be excited at room tem-
perature ?
(b) Assuming a rigid bond length of 1.13 A, calculate the
moment of inertia of the molecule about an axis per-
pendicular to its symmetry axis. Using the result for
the quantization of the rigid rotator, find the energies
of the first few states for this rotational mode. Will
these states be excited at room temperature ?
(") Estimate the energies of the first few rotational states
about the symmetry axis of the molecule. Will any
of these states be excited at room temperature ?
(Ans.: (u) Et :0.13 eV; (b) E ,-.'4 x 10-a eV;
(c) E ,'-,4 eV.)
9. THE HARMONIC OSCILLATOR:

OPERATOR METHOD
In this section we will solve the oscillator equation (Eq. 5.20), by the method
of Dirac.ll The method is a powerful one, frequently used for algebraic
11P. A. M. Dirac, Clarendon Press, Oxford, 1958, 4th edition, section 34.
Quanturn Mechnnics,
equations involving non-commuting operators, and is commonly called the

factorization method. Since the harmonic oscillator has already been studied
in great detail in the previous sections, it provides an excellent subject for
illustrating this important operator method.
For convenience let us define the dimensionless position and momentum
operators,
q: {nx'
p:+^:
h{ a'
* _+! : _+1jv: _io
{i a* t/i ar - aq Oq'
!': -y'
oq''
In terms of these operators, Equation 5.20 becomes
(P' + q')yt : ,rt,. (5.43)
Since p and q are operators associated with r and y', (see Problem 4-22), we do
not expect them to commute. In particular,
lp, ql :iro,,xf : -i . (b.4+)
If the commutator in 5.44 were zero, Equation 5.43 could be factored into
(q + ip)(q
- ip)tp : €v. However, because of the non-commutivity of p and,
q we can write:
(q+iP)(q-ip):q2+p2+r
(q - ip)(q + ip) : {rz + p, _ t.
Adding,
p' + q2 : tl? +;p)(q - ip) + (q - ip)(t +;p11.
If we define two new operators,
':+,(r +;p1
':aQ -i!),
a'1
then
P'+q':aaIaaIa,
and the Schrodinger equation, Equation 5.20, may be written as
(oot + oIo)yt : ey). (5.45)

THE HARMONIC OSCILLATOR: OPERATOR METHOD t93
The operators a and at satisfy the following commutation relation:
lo, oIl : aoj - ata: l.

Using the latter result, the Schrodinger equation (5.45) can be written in
the following two additional but equivalent forms:
aatYt : (; *;), (5.46)
aray,: (; -;). (s.+7)
Before actually solving the Schrodinger equation, we will show that we can
construct an infinite set of eigenfunctions if we can somehow find one solution
of any of Equations 5.45, 5.46, or 5.47. Operating on Equation 5.46 with at
from the left, we have
aIaaly : (;* tr)^, (5.48)
Using the commutation relation, aIa : A0,I

- l,
(aot - l)oIrp :(; *i)u,
aaI (aly) :(; *tr)o, (5.4e)
That is, if rp is an eigenfunction of aaI corcesponding to the energy eigenvalue

el2 + $, then atyis also an eigenfunction of aaI, butithas the new eigenvalue
el2 +2. Since atyis also an eigenfunction of ata, from Equation 5.45 we see
that
(ooI + aI a)(aIy) : (. + 2)(oIrp),
where we have used Equations 5.48 and 5.49.

Similarly, operating on Equation 5.49 with at, we obtain
ot(ad)(otrp) : (ala)(oI),rp : (aaI - 1) (oI),rp : (;_';) (.\',p

Then,
aaI (al),,p : (; . Arfr,
|)
and (aaI + da)(d),rp: (. + 4)(oI),'p.
This process may be continued indefinitely, since the harmonic oscillator
has an infinite set of states. Since each successive operation with at raises the
oscillator to the next higher energy state, at is called the raising operator. In like
manner we will show that a is a lowering operator for the same set of functions
such that the following relations will hold:
alrpn : C+rpn+t and a!)n : C-rpn-r, (5.50)
where C* and C- arc constants of proportionality which will be determined

later.
Let a operate from the left on Equation 5.47 :
aatarp: (l I ata)ay : (; -i)*

or
ata(av): (; _i)*.
That is, if g is an eigenfunction of atawith eigenvalue (el2 - +), then ag is also
an eigenfunction of aIa butwith the new eigenvalue (el2
- 9). The lowering
process can also be repeated by successive application of the lowering operator
a as shown in Table 5-1. However, in contrast with the raising operation, the
rABLE s-r Eigenfuhct'ffi:,.ift,5i.?L}:ans of the Raising

The lowering operator is subject to the constraint, e ) 2n' + I
Eigenvalue Eigenvalue Eigenvalue

Oscillator of of of
Wave function oot oto pz + qz
el el
-r-
2'2 ,- 2
€
t e3 el
aty) -r- -r- e*2
2'2 2'2
e5 e3
(oI)'rp
2'2
-_L- -r-
2'2 ei4
e 2n*l e 2n-l
(ot)"rp
2'
-_L-
2 2'
-l_-
2
et2n
el e3
e -2
,-, ,-,
e3 e5
e -4
a2yt
,-, ,-,
€ 2n'-l € 2n'+l
a''ll
,- ,- , -2n'
THE HARMONIC OSCILLATOR: OPERATOR METHOD t9S
lowering operator can be applied only a finite number of times because the
eigenvalue, el2
- (2o' + \12, cannot become negative. The physical reason
for this is that the energy states of the harmonic oscillator must always corre-
spond to positive energies. This, of course, was a well-known classical result
and could be accepted in quantum mechanics on the grounds of the corre-
spondence principle. However, we can readily show its validity in quantum
mechanics by examining the expectation value of the energy. Thus,
,n :l- v*ert dq :ll-r.(p' + q,),p dq: -,L:,p. dqry,* *l:-,p*qr,p dq.

Integrating the first integral by parts,
(.): -[r. *l:-tvt,q,dq.

H]:_+J- %y,*
The first term vanishes since g must be well behaved. Then,
(.):[( Hl'* r tvl)dq2o,

since the integrand is positive definite.
If we designate the ground state wave function of the oscillator ds !)0, then
we must have
ago :0, (5.51)
since there can be no lower state. Then, from Equation 5.47, we have
olorpo:0:(;-!r)r,
E-r- ,
2Eo
rh
or
Eo : trh,
the ground state energy of the oscillator.
We will now see how easily the ground state wave function is obtained by
this method. Inserting the expiicit foim of the lowering operator,
":#'(o* &)'
into Equation 5.51, we obtain the first order differential equation,
W
* rro:0.
This may be integrated immediately to obtain the ground state wave function,
111o: Coe-a'lz,
where the constant of integration, Co, is chosen so as to normalize yo. Then

each higher state can be obtained by successive operations with the raising
operator, at. Thus,
lpr :* orro
L/o
1P
z : r, : ?,
(or)r,t, o
?o,
vn:%^{or)"r0. (5.52)
It will be shown below that the normalization constant C,,, when divided by
(2")L, becomes the same factor as that indicated by I{, in the discussion pre-
ceding Equation 5.38. It should be pointed out that Equation 5.52 provides
still another method of obtaining the Hermite polynomials, since
: q'tzHn(q) : - +t/2" (O r-",'.

C*(oI)nr-az1z
\' - lY
rpn Nne
0q/
That is,
H*(q) - saz/z(,
- &l
r*,,. (5.53)
Let us now evaluate the normalization coefficients above. We require

first that
(yo I ?o) : I: lcoPI-*ro' dr : lcol'z\/;.
Then, Co : Q 1t/ r1t.
To determine the ratio C"lCo we use Equation 5.52 and form the integral
(,p* l,p*) : * *vo

l 1ot1n,t,o).
l',rrtl
l
Recall from the definitions of a and aI that a* : aI; that is, the operators
are complex conjugates. They are, in fact, Hermitian conjugates or adjoints of
one another. The full meaning of this property will not be evident until one
uses matrix representations, but its usefulness here lies in the fact that it per-
mits us "to move an operator past the vertical bar" as follows:12
(otrp I oftt) : Q-Uo,.p> : (WI'p | 0.
12 The equivalence of these expressions may be established by performing partial integrations.

THE HARMONIC OSCILLATOR: OPERATOR METHOD lgt
Using the property of the adjoint,
(vn I ,,.n) : @ot)n,po).

'^l'rr,l
This expression can be greatly simplified by using the following identity:
o(al)" : (aI)"a * n(aI)"-t (5.54)
Its validity can be readily deduced from the commutation relation, that is,
'(")=tr",{-M"i;f
;'.a'I6(6r)n-'l
: n(ot)"-l + (oI)"o .
The result of applying Equation 5.54 to go is:
o (or)n!t o : n (oI)n-'rl, o,
since ago :0. Then,
a'(aI)"rl,o: na(aI)"-'rpo :
- l) (ot)
n(n n-rrpo
as(atlnrpo : n(n - l)a(at1"-rrpo : n(n - l) (" - 2) (oI)n-trpo.
a"(at)"rlto : nlyo. (5.55)
Using Equation 5.55,
kt*l !,*) :
*f"!(?o I vo) : *f^Pol't/n'
Therefore,
cn:(#J
and
Nn:(#J
The general form for the wave function for the nth state is then
't^k)
: (;1 e-a'zrz17n7o1
SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
I-4
V. ,/'
-5 -4 -3 -2 -l 0 I 2 3 4 5
(t ___+ (t: 1/ax)
Figure 5-22. Probability densities for a few harmonic oscillator states. The dotted curves show
the cf assical probabilities corresponding to thesame energies. (From Chalmers W. Sherwin, lntro-
duction to Quontum Mechanics, Holt, Rinehart and Winston, lnc., 1959. Used with permission.)
or
(5.56)
where
H^(q) : (- t)nsaz r-"'

and
f,
@m \/6
nn
PROBLEM 5.19 li,Tili.t*.i1l1::.itf
(u) Consider a particle of mass m performing simple

harmonic motion along the r-axis. Show that the
classical probability of finding the particle in the
THE HARMONIC OSCILLATOR: OPERATOR METHOD t99
element dx is
p(x) dx :
,1/ trz - *z'
where ,4 is the amplitude of the motion.
(b) Show that the classical amplitudes corresponding to
the quantized oscillator energies are given by
A'n:(2n * r)
,
(.) $::fi init\(;at the

"ru,,ilur
rmit on the motion or
the particle in the state tp,(q) is q : 1/2" -t 1.
PROBLEM S-20 iflt;tiiii{,ii,J-.if,iif,i
Using the normalized, linear harmonic oscillator wave

functions perform the necessary integrations to verify the
following:
(u) that yo and tpl are normalized.
(b) that yo and yl are orthogonal.
(.) that (rpol q'ltpo) : +.
(d) that (911 q' lrpr) : (Ttl !'lrpr) _3
-2.
(.) that (yol q ltpt) : (rptl q lqo) : \/-2
2
PROBLEM 5.21
Carry out the necessary differentiations and inte-

grations to find the following expectation values for
the first excited state of the harmonic oscillator: (x),
(x2), (!,), and (p!).
(b) Using the results of (a), find <T), (V), and (E) for the
same state.
(") Calculate the minimum uncertainty product Ar ' L!,
for this state.
PROBLEM 5.22 iff,f.1i.,ili"i,F,,'l",I-,i,ii
A harmonic oscillator has a time-dependent wave function

which is a superposition of its ground state and first excited
state eigenfunctions, that is
Y(*, t) : 1
t) I Vr(r, t)1.
,rlYo@,
Find the expectation value of the energy.
li11!li,iii.Ii!ii-{
2OO SOLUTIONS OF SOME ONE DIMENSIONAL SYSTEMS
PROBLEM S-23 i9;i.*'{,$.r+i,::x'ri
Where is the maximum of the probability ofgo(r), the ground

state harmonic oscillator wave function ? How does this
compare with the classical value ?
IO. EXPECTATION VALUES IN THE

OPERATOR FORMALISM
It is useful to evaluate the proportionality constants in Equation 5.50

before obtaining the matrix elements and expectation values of section 7 by the
operator method. Using Equation 5.52,
Ipn+L :T orr_,
or
aI,pn:
ff*,*,: (g+!'f ,-*,: \/n + r sn*I. (5.57)
Also, operating on Equation 5.52 with the lowering operator,
aen:?oo(o')"r0.
Using the identity (Eq. 5.54),
avn :prrarlna r n(oI)n-rlrt,o
cn
: nfr(ol)"-"1'o' (since avo : o)'
: (from Eq. 5.52),

"?, *!)n-r>
:n(? - l)!\+
\ il )v"-''
Thus,
alpn : {irp*-r. (5.5S)
EXPECTATION VALUES IN THE OPERATOR FORMALISM 2OI
It is convenient to transform p and q by means of the definitions of a and

aI. Thus,
q:+r@+,ll
(5.5e)
e:hr,-^.J
Now,
(a) : (V*l atp,) : t/i(rp*l y,*-r) : 0
(oI) : (rp^l aIrp,, : 1, a l9,*l rp**r) : 0
(aaI) : (,pnl oot,t,*) : (7 + (,p^l v*) : n * !

;)
(aIa) : (rt,nl oro,t'^> : (7 - il (rp,l y,*) : n.
From this it readily follows that:
* (ot)) : o
:72(@) 1
(q)
(!):m((at)-(a)) :o
(q') : *((aa) + (aaI) * (ata) -t @IaI))
: g(2n + l)
:n*t
(!2) :f f (aIaI: - @ta) - (aaI) * (aa))
:n*t'
Therefore,
(x): I<q>:o
!x
(!n):nt/i1py:O
(xz) : !r,, :la + +)
(p',): nh'<p'> : ah3(n + +)
: Ir :2ah(n +
(T) +)
2*<pi>
<v> :T o,, : gafi(n + +)
ah ah
(E):Vk):i(pr> +(qr)) :toh.(n ++).
The reader should compare these results with those obtained in section 7 using
the algebra of the Hermite polynomials.
PROBLEM 5.25
Calculate the matrix elements (!,*l ry,n*1) and (!,*l rrpr_r)

by means of the operators introduced in this section.
The formulation of the harmonic oscillator problem in terms of the

dimensionless operators a and aI has the important consequence that the
results are independent of the specific dynamical variables of the original
problem. In other words, the results are independent of the representation. Thus,
whenever the Hamiltonian of a dynamical system can be written in the form
ff:q2+f':aaIaata,
then the algebraic results of this section hold, regardless of the exact nature of
these variables. An important application of this method is that of the creation-
annihilation operator representation of field theory. In that formulation aI
creates a particle state, a annihilates a state, and aI a counts or numbers the
states. The latter is called the number operator and its function is readily seen
as follows: otorlrn: arlirp*-r,: {n. \/;!)n: frIpn. Therefore, the eigen-
value of the operator at a is the integer which numbers the state of the eigen-
function upon which it operates. In physical problems involving the
interactions of several different kinds of particles, it is customary to define a set of
creation-annihilation operators for the states of each class of particle. Thus, in
solids the many-body properties are conveniently treated as the collective
normal-mode excitations of systems of harmonic oscillators. The quanta of
excitation are called phonons for the set of oscillators associated with lattice
vibrations. In a magnetic system the normal modes are called spinwaves
and the quantized excitations are known as magnlns. In a physical problem
involving the interaction of electromagnetic waves with a magnetic solid, one
might require pairs of creation-annihilation operators for photons, phonons,
and magnons,
SUGGESTED
REFERENCES
S. Borowitz, Fundarnentals of Quanturn Mechanics. W. A. Benjamin, Inc., 1967,

particularly Chapter 10.
R. H. Dicke, and J. P. Wittke, Introduction to Qunnturn Mechani.cs. Addison-
Wesley Publishing Co., Inc., Reading, Mass., 1960.
R. M. Eisberg, Fundamentals of Modern Phltsics, John Wiley and Sons, fnc., New
York, 1961.
R. B. Leighton, Principles of Modcrn Phltsics. McGraw-Hill Book Co., Inc., New
York, 1959.
P. T. Matthews, Introduction to Quantum Mechanicsr 2nd ed. McGraw-Hill Book
Co., New York, 1968.
SUGGESTED REFERENCES 203
E. Merzbacher, Quanturn Mechanics. John Wiley and Sons, fnc., New York,
1961.
D. Park, Introduction to the Quantum Theory. McGraw-Hill Book Co., Inc., New
York, 1964.
L. Pauling, and E. B. Wilson, Introduction to Qtnntum Meclwni.cs. McGraw-Hill
Book Co., fnc., New York, 1935.
J. L. Powell, and B. Crasemann, Quantum Mechanics. Addison-Wesley Pub-
lishing Co., fnc., Reading, Mass., 1961.
Ph2sics,6th ed. McGraw-Hill Book Co., New York, 1969.
D. S. Saxon, Elementaryt Quantum Meclanics. Holden-Day, fnc., San Francisco,
1968.
L. I. Schifl Quanturn Meclanicsr 3rd ed. McGraw-Hill
Book Co. Inc., New
York, 1969.
R. L. White, Basic Qwnturn Mechanics. McGraw-Hill Book Co., fnc., New
York, 1966.
K. Ziock, Basic Quntum Mechanics. John Wiley and Sons, Inc., New York,
1969.
GHAPTER 6
Wffiffi ffiffiffiffiffiffi
wweffiwwffi ffiffiffiffi&ffiffiffiffi
Many of the concepts and some of the mathematical intricacies of quantum

mechanics have been introduced in earlier chapters. Up to this point the
emphasis has been upon making the development plausible to the reader.
In the present chapter we will alter this approach and, instead, focus our
attention on the logical structure of quantum mechanics and the formulation
of matrix mechanics. The attractiveness of this formal approach is that once
the postulates are accepted, the theory has a logical cohesiveness which has
great intellectual appeal, perhaps more so than the heuristic arguments used
earlier. Regardless of the plausibility or the logic of its postulates, the over-
riding justification for quantum mechanics is its tremendous success in
accounting for the observed behavior of atomic, nuclear, elementary particle,
and quasi-particle systems.
I. THE POSTULATES OF QUANTUM MECHANICS

Although there is general agreement as to what ideas constitute the.basic
postulates of quantum mechanics, there is great diversity in the way these pos-
tulates are presented in different textbooks. Even the numbering of the
postulates is quite arbitrary, since two or more postulates are sometimes com-
bined into a single statement. In what follows, numerous theorems and defini-
tions have been included in the discussions of the postulates, which are here
presented as five in number. Wherever the term "particle" is used, the phrase
;'quantum mechanical system" could replace it, since "particle" is being used
here in the sense of typifying the simplest quantum system.
Postulate I
The d2namical states of a particle can be described b2 a waae function which

contains all that can be known about tlte particle.
THE POSTULATES OF OUANTUM MECHANICS 205
A complex wave function can be used to represent a particle (see section 6

of Chapter 4), this wave function being the fundamental mathematical entity
from which all observable quantities can be computed. The wave function
may be denoted by V(i, t) in the coordinate representation or by {(/i) in the
momentum representation. Y and /, which are related by the Fourier trans-
forms, Equation +.27, are equivalent representations of the particle. Not all
functions make suitable wave functions. In order to be physically admissible,
both the function and its derivative must be finite, continuous, and single-
valued everywhere. In addition, the integral of the square of its modulus over
all of space must be finite so that the wave function can be normalized. Suppose,
for example, that
t)12 dx d1t dz - A,
,[,*,.
where A is a finite number. Then a wave function which is normalized to
unity may be constructed by writing the new function as Y' : 0l\/7)YQ, t).
(See Problems 4-8, 4-9, 4-10, and 4-15.) A quantum mechanical state is
defined by a specific set of values for the independent variables, such as
V(ir, tr) or +(il. Although neither Y(ir, t1) nor +(Er) is directly observable,
the square of its amplitude for a particular state is proportional to the probability
that the particle will be observed in that state. We defined the probability
density in Equation 4.18 as
p(it, tr) : lY(i1, rl)12
t*{0,, t1)12 dx d1t dz

J
For Y normalized, the probability of finding the particle in the volume element
dr centered around i, is
d.P : p(ir, t) dr : lY(ir, t)12 dt. (6. 1)
It follows that the probability for finding the particle somewltere in space, that
is, the total probability, is unity. Thus,
,:T,,,:T, lY(i, t)12 dr - 1,
when Y is properly normalized.
Definition
Two non-zero wave functions are said to be orthogonal if their scalar
product is zero. Thus, the orthogonality condition is,
YIV, dr : 0.
I,
THE FORMAL STRUCTURE OF EUANTUM MECHANICS
Wave functions that are both normalized and orthogonal satisfy the combined
orthonormality condition,
f
I YIY, dr : 6ii, (6.2)
Jr
where d' is unity for i : j, and zero for i + j.
Postulate 2
The superposition principle is ualid for functions representing plrysicall2 admissible
states,
Wave functions can be superposed to form a new wave function which

is, itself, a physically valid representation of a possible state of the particle.
This is the fundamental assumption underlying the representation of a particle
by a wave packet, where the Fourier integral provides the means for superposing
a continuum of states. (See sections 7 and B of Chapter 4.) In the case of dis-
crete states, the superposition postulate says that any linear combination of
the functions describing these states is also a possible physical state. In general,
then, such a state may be expressed as,
\r : 2tnYn, (6.3)
i
where the co are the expansion coefficients and the Y, are the known discrete
states.
Postulate 3
Tlte Schrtidinger equation describes the beltauior of a waue function in space and time.
If a wave function Y(i, t) is known at some particular time, then its

development in time is given by the Schrodinger wave equation (see section
9 of Chapter 4),
"ryve, t) : th
*vQ, t), (6.4)
where ,tr is the Hamiltonian operator for the system. In the particular case
of a time-independent potential, Equation 6.4 is separable in space and time,
and Y(i t) may be written as the product of a spatial function and a time
function. The spatial part is an energy eigenfunction; that is, it is one of the
solutions of the energy eigenequation, Equation +.57,
#rpr_(i) : Enl,r*(i), (6.5)
where the eigenvalue En is the energy of t}re nth state. Since the energy E^
remains constant in time, the solutions are known as stationary states. The
THE POSTULATES OF QUANTUM MECHANICS 2O7
nth particular solution of Equation 6.4 is
Y"(i, /) : Tn^(i)e4n"un. (6.6)
Utilizing the superposition principle, a general solution of Equation 6.4 may

then be expanded in terms of the solutions just obtained, namely,
YQ, t) (6.i)
The sufficient condition for Equation 6.7 to be a valid expansion for a general
wave function Y is that the functions yp. form a complete, orthonormal set of
functions. This matter will be discussed further in section 3 of this chapter.
Postulate 4
Each d2namical uariable q can be directl2 associated witlt a linear, Hermitian
operator Q. Tlte onQ possible result of a measurement of tlte obseruable q is one of tlte
eigenualues of the operator Q.
Definition
The operator Q is linear if it commutes with constants and if it obeys the
distributive law. Specifically, if
Qk,D : tQ(v),
and if
Q(p, + lpr) : Q(rt) + Q(rtr),
then Q is a linear operator.
Definition
An operator is Hermitian if it satisfies the following equality,
fr
IJr QrD*rp d, : JrI ,p* (QrD d",
In the Dirac notation of Equation 4.58, the Hermitian property is
<Qrplv): <QrplV)* : @lQrp). (6.8)
A more general statement is obtained by letting V : aVt -f brlr. Then Equation

6.8 becomes
(Q@rpt * brpr) I orpt + brpr) : (ay,, * brprl Q@rp, + btpr)).

208 THE FORMAL STRUCTURE OF OUANTUM MECHANICS
Using the properties of linear operators this becomes
a* a(Qrp'l y,t) * a*b(Qrprlrpr> + b*o(Qytrl V') + b*b(Qrprlrpr)

: a*a(tptl Qtp') I a*b(rprl Qrpr) * b*a(rprl Qrp') a bxb(rprl Qrpr).
Then,
a*a((Qrprlrpr) - (rprl Qrpr)) + n*U((Qrprlrpr) - @rl Qvr))

* a*b((Q'prlrpr> - (rprlQrpr)) + b*a((Qrprl y') - (rprlQrp')) :0 (6.9)
Since a and b are arbitrary, and their relative phase is arbitrary, the only
way Equation 6.9 can be satisfied for all possible values of a and D is for each
term to be identically zero. This immediately gives Equation 6.8 and the
more general statement of the Hermitian property,
(6.10)
@r l,Qrp,) -- (Q'p' 1,P,>.
A Hermitian operator is frequently called a self-adjoint operator. This terminol-
ogy arises from the fact that an equation similar to Equation 6.10 can be
written for any operator Q, provided that the Hermitian adjoint* of Q, denoted
by Qt, is used when it operates on the other function. Thus,
@rl Q'Pr) : (QI't'rl'Pr>' (6.rl)
When Q is Hermitian, QI : Q, which reduces Equation 6.1 I to the result

above. In the notation of Dirac, integrals involving Hermitian operators are
frequently written as
(rpl Q lrp),
to indicate that the operator, being Hermitian, can operate equally well on
the function to the left or to the right.
Now let us consider a measurement of the dynamical variable g which is
represented by a linear, Hermitian operator A. The fourth postulate says
that the measured value must correspond to one of the eigenvalues of Q, but
it does not tell us which one. In order to see what is implied here, let us suppose
that we wish to measure the energy of a hypothetical state which can be
represented by Equation 6.7, where the only non-zero expansion coefficients
are cL and cr, which are real numbers. That is,
Y(i, t) : crgsr(i)e-iu'tln * crrl,n"(i)e ia*ln: rrYr * cJtr, (6.12)
where normalization requires that cl * c? : l. A measurement of the energy

must yield a result which can be expressed mathematically by operating on Y
with the Hamiltonian operator. Thus,
.trY : ctffYt * csffYs : ctB#t * cgEJI's. (6.13)
* A precise definition of the Hermitian adjoint will be given in section 7.

rHE POSTULATES OF QUANTUM MECHANICS 209
The same result can be obtained by operating on Y with the energy operator
ih(7l0t) :
th :
*ty(r, r) ;nprvE,e) fir-ou,,,u * c{ps"@
*,r-ou*,1
: c1E1Yy * csEsYs. (6.14)
Equations 6.13 and 6.14 are evidently not eigenequations, since we cannot
gxpless the right hand side as the product of a scalar times the original function.
In this example the only accessible eigenvalues of ,ff are E. and4, ro we con-
clude that a measurement of the energy will yield either the value E, or E",
and not some intermediate or average value. The probability of measuring
the value E, is til^Q? c!) (or just cl if Y is normalized) and the probability o?
measurin1 EriscilQi^++ cil. The act of measurement, then, forces the sysiem
into one of its eigenstates, where it will remain unless it is disturbed. Once a
system is known to be in an eigenstate, the result of a repeated measurement
can be predicted with certainty.*
_ The previous discussion assumed, of course, that no time-dependent
fo-t::| act_ upon the system. Should the Hamiltonian be time-depetdent, an
additional c_omplication arises, since the Schrodinger equation is thin no longer
separable. We will assume that in such a case the wave function can still-be
expanded as in Equation 6.7 if we but add the assumption that the expansion
coefficients, co, are,also time-dependent (this will be dbne in Chapter 9j. The
1ct 9f measuring the energy will again yield one of the energy eiginvalues, say
Er, brlt in contrast with the previous case, the system will remain in
the jtk state. The time dependence in the coefficients will "o'to"g.r
cause some coeffi-
cients to grow at the expense of others, or even to oscillate. Therefore, successive
measurements of the energy will not, in general, yield the same eigenvalue.
Theorem
Tlte eigenualues of a Hermitian operator are real.
Pnoor. Given, Qytn : Qnvn; and Q : QI , For convenience let (tpn l rp,) :
l. 'Atro,
.(rtt*l Qrp"> : (pnl q*rp*) : gn(lpnlrp,) :' 4,.
The expectation value is
lQrprlr*):.(qnpnlrp,) : ql@*lrp^): qi.
g"i the lefi rru"a rid.r ur..q,rui
by the Hermitian property, so we have
8L : 8n,
ot In is real.
Theorem
. Tlte eigenfunctions of a Hermitian operator are orthogonal lf they correspond to

distinct eigenualues.
* See the discussion of the Stern-Gerlach experiment in section l l of Chapter 3.

2IO THE FORMAL STRUCTURE OF EUANTUM MECHANICS
Pnoor'. Given, Qrpo : qoyu and Qyto : qyp,, where gr # qi and Q : QI

Then, writing the integral,
@nl Qrpr) : (tsrl qoy,,) : r/%lrp,).

Also,
Qrpol Tr) : Q{pnlrp) : rr9ttl tpo),
where the fact that the eigenvalue is real for a Hermitian operator has been
used. The left hand integrals of each line are equal by the Hermitian property,
so by subtraction we obtain
o:(q,-er)@nlrp,).
Since Qr * gr, we must have (rpnlrp) : 0.
Therefore the functions Vt and Vt are orthogonal (see the definition

following Postulate 1). In section 4 it will be shown that if there are degenerate
eigenfunctions (that is, two or more eigenfunctions have the same eigenvalue),
these functions can be orthogonalized.
Postulate 5
The expectation ualue of a measurement of the uariable q is giuen mathematical[t as
(Y I 0Y)
(q) : (6.15)
If Y is an eigenfunction of Q such that QV : QY, then Y is said to be a pure

state, and the expectation value of I is simply the eigenvalue g each time the
measurement is made. Suppose, however, that Y is a mixed state such as that
given in Equation 6.12, namely, the superposition of two orthonormal eigen-
states of Q. Then,
Y-trYr*crYr,
QYr : Qfft
QYs : ItYs
and
(Yn lV,) : d,,'.
The expression for the expectation value of g becomes:
(q) :
(cifl, * crY, I Q(r'Yt * crYr)) :7a*
tlqt, * c?q"
'
Note that this number turns out to be a weighted average of the accessible
eigenvalues of Q, although a single measurement must always correspond to
rHE POSTULATES OF OUANTUM MECHANICS 2I I
just one of the eigenvalues of Q. Consequently, we interpret the expectation

value as either the average value of many measurements of ( on the same
system, or the average value of the same measurement on many identical
systems.
If we generalize these results to a mixed state that is a superposition of
all of the eigenstates of Q, the wave function is given by Equation 6.3, and the
expectation value becomes
(q):fu
2 lttl' qt
(6.16)
The probability that a single measurement will yield the eigenvalue qo is
lcnl'
2v)''
It has already been established that eigenvalues must be real-since they
correspond to physical measurements-and we have seen that the use of
Hermitian operators guarantees that this requirement is met. It is evident
also from Equation 6.16 that the expectation value must be real, but this can
be shown formally as follows.
Corollary
Tlte necessar.lt and suficient conditionfor a real expectation ualue is that tlte d2namical
uariable be represented b2 a Hermitian operator.
Pnoor. Given, (8) : (8)*, that is, the expectation value of I is real.
Then, (rp I Qil : (t, I QrD* : (Qrp l rD, which demonstrates that a is
Hermitian.
PROBLEM 6.1
An operator is said to be skew Hermitian or anti-Hermitian if

<Qv I V) : - (tl I Qil. Show that the eigenvalues of a skew
Hermitian operator are pure imaginary.
PROBLEM 6.2
Prove that the following operators are Hermitian for well-

behaved wave functions:
(u) ! : -ihY
E- th*t
lbl
\c) tc
::i .i4i:.,i, \,,'14-.'..1::,i1,1,:+,r..::r j+:r"r;'t;'4l,itl

212 THE FORMAL STRUCTURE OF EUANTUM MECHANICS
If ,4 is a non-Hermitian operator, show that (A f .41) and

i(A - Ar) are both Hermitian. This is a useful result since
the operator A can then be written as the following linear
combination of two Hermitian operators:
I
A : +(A + At) + %li(A - AI)1.
PROBLEM 6.1 -i,#bF-l*E{.zliii.S$,F,ff,S-ii,}
If A, B, and C are Hermitian operators, transform the

following expressions so as to eliminate the daggers:
(u) (ABC)i : (d) (AB - B!)' :
(ul (A"1t - (ilA,Bl)l (.) :
(.j (AB + BA)I : (f) (A + B)t :
Which of the above are Hermitian ?
PROBLEM 6.5 #F,i'**rc1*-$iF"+;gii.4*v;*f
Show that the variance, defined by (Q') - (8)', is zero only

when g is an eigenfunction of 8. (Let g
i
ga satisfy Qrl,u : g#)
,fjfl**.d.lFJ':*,s-eFLi{F-,;"S.t"tiit'-. .8'l .Petrs.€}i#-#fl'i,,H$;
2. MEASUREMENTS OF COMPATIBLE OBSERVABLES:

COMMUTING OPERATORS
Consider two operators P and Q associated with the observables p and q,
respectively. We wish to know what to expect if a measurement ofp is followed
by u measurement of q and, conversely, if a measurement of 4 is followed by u
measurement ofp. These two expectation values are expressed mathematically
by Equation 6.15 as
<QP> : (p I QP,p> and (PQ) : (t I PQrp>,

provided that y is normalized. Now, if rp is simultaneously an eigenfunction
of each operator then we have the relations
Pv : ltrl
Qrp : r{P,
and the two expectation values become:
<QP> : fi(st Qrp) : ltqt

(PQ> : It(rp Prl,) : (hh : frQr
MEASUREMENTS OF COMPATIBLE OBSERVABLES 2I3
Since p, and qL are simply numbers, they commute and the order of the
measurements is immaterial. Physically, this means that a system that is in
simultaneous eigenstates of two operators will be undisturbed by any sequence
of measurements of the observibles associated with these two operators.
Mathematically, we see that for this to occur the operators must commute;
that is,
lQ,Pl-QP-PQ-0. (6.17)
The bracket on the left is defined as the commutator bracket or simply the com-
of I and P. (See section l I of Chapter 4.)
mutator
Corollary
. .If tyo operators P and Q commute and either P or Q has non-degenerate eigenualues,
its eigerfunctions are also eigerlfunctions of the other operator.
Pnoor. Given [P, 0] : 0 and Py,n : !{p* where all po are distinct.
Then,
QPrpn : Qf rtpr : fn(Qrpr).
Using the commuting property,
QPrpn:PQVI:P(Q1D.
Equating the right members,
P(Qrpr) : pn(Qrp).
Thit says that Qrpo is an eigenfunction of P with !0, which will

eigenvalu e
lead to a contradiction unless Qyo differs from Vrby ahultiplicitive constant.
That is, we must have
QVr : Q{tr
Therefore, Vi is an eigenfunction of Q.
Wewill now examine the time dependence of an expectation value. From

our previous definitions it is evident that an expectation value calculated for an
eigenstate is a constant of the motion for the dynamical system described by
that eigenstate. In the language of operators, a necessary condition for a
dynamical variable to be a constant of the motion is that the operator associated
with that variable commute with the Hamiltonian for the system. This may
be shbwn as follows:
: #@l e,r) :
!, <o>
<#lr,).(, ln"+*(, lr{)
:i@,plQ,p
-1 ,. (#) + !h(,pt etr,p),
2/4 THE FORMAL STRUCTURE OF OUANTUM MECHANICS
where Equation 6.4 has been used. Then,
L,r, : *<,t,1Qt -"tru,p) * (#),

where the Hermitian property has been used. Therefore,
d n\ I _ *\di
lp ,\. (6.18)
dt<a):;6<lQ,.rl)
This tells us that the expectation value will be a constant of the motion if the
operator has no explicit time-dependence, and if it commutes with the
Hamiltonian.
PROBLEM 6-6 u.iiii1ii,i,-'i:,."i'i iirrr.llirr::rrlll,i':':,l;'
Use Equation 6.18 to verify the following expressions for the

linear harmonic oscillator:
(u) o!,<n.>
<#>
(b) *#,a>: (!,).
The statement that the variables in the classical equations of
motion can be replaced by quantum mechanical expectation
values is known as Ehrenfest's theorem.l
PROBLEM 5.7
- Evaluate the following commutators:
(u) lP, r'f (.) lP, P'l

(b) lp', r'f (d) lpr, V(r)l
(u) Using mathematical induction, show that

lx", Pf : ih'nx"-r lr, P"l : iftnPn-t'
and
(b) Ifl is a polynomial fuirction, show that
l-f @), ll : thy- and lr,.f ( P)l : ffi#.
1P. Ehrenfest, Z. Pfursik 45r 455 (1927).

LINEAR VECTOR SPACES 2r5
PROBLEM 6-9
For arbitrary operators A, B, and C, show that
lA, BCI : lA, BIC + BlA, Cl.
3. LINEAR VECTOR SPACES
We will now construct an abstract, linear vector space in which each

vector in the space is a wave function of the kind we have been using to represent
a possible physical state of a quantum system (in either coordinate or mo-
mentum space). A linear function space of this type is called a Hilbert space.
Any complete set of orthonormal eigenfunctions can serve as the basisfunctions-
that is, the unit vectors-for such a space. The concept of the superposition
of states is easily visualized by regarding the expansion coefficients, Equation
6.7, as the projections along the basis vectors made by the vector representing
the state in question. When the state vector lies along one of the basis vectors,
the system is in the eigenstate characterized by that basis vector. By way of
illustration, the complete set of orthonormal harmonic oscillator energy
eigenfunctions given in section 7 of Chapter 5 forms a basis for a Hilbert tpui.
of infinite dimension. Any superposition of harmonic oscillator states may be
regarded as a vector in that space.
Now let us examine the properties of a linear vector space. (l) There is a
null vector. (2) The associative and commutative laws hold for addition. Thus,
ifwe regard the g's as vectors, ?r -l- Vz * lps : rh * 0p, + rp") : (rp, + Vr) *
?q. Also, % * ,1,? : V? I ,t,,.. (3) There is a scalar product defined by (p, I tpr>,
which is linear (distributive) but not, in general, commutative. Thus, the
scalar or inner product of ,ltr. and g, is:
!,vIrrdr
: (v,.lv) : (ltzl vr)* (6.1e)
In a real vector space (or Euclidean space) the scalar product is real and
symmetric in the variables. In a complex vector space (or unitary space) the
scalar product is complex and possesses symmetry in the variables under
complex conjugation. The latter is known as Hermitian s2mmetr2.
To illustrate the linearity of the scalar product, let us form the product of
rp1 and (rp, * ,pr):
(rptl rp, * vs) : (Ttl rpr> + (rprl rpt).
Definition
Two vectors are orthogonal rf their scalar product is zero, provided that
neither vector is the null vector.
Definition
A set of functions is linearl2 independent if the linear equation,
2 tort,o
- 0,
is satisfie d onl2 when all the cu's are zerl. For example, if the Euclidean vector
at+bj+c[-is 0 only when a:b:c:0t then t, j, and ft are linearly
independent. fn other words, no one of these quantities can be written as a
Iinear combination of the others.
A set of i[ Iinearly independent, orthonormal vectors, {,$}, can provide a

basis for an /f-dimensional space, since any vector in the space rr'ay then be
written as a linear combination of the -ly' vectors. The set of vectors is said to
be complete if it spans the space.* That is to say, when the space is fully spanned
by the basis, no conceivable vector can have a projection which would require
an additional basis vector. Another way of describing completeness is as follows.
A set of functions is said to be complete if the addition of any other function
makes the set linearly dependent.
Theorem
The representation of oo arbitrary uector on a giuen basis is unique.
Pnoor. Given, Y: Lrn|*and suppose that Y is also represented by

N i:L
Y: where the same basis is used in both cases. Subtracting these two
Zt;,Q',
expressions yields
o:g ko-r|)Qu
Since the basis veetors {tn arerr"."r;;:ndependent, this result requires that cu :
c'0. Therefore, the representation is unique.
Theorem
The inner product of a uector with itself is positiue definite.
/N
(Y lY) : 2^/\
t,,0,) : > cfci(0,r1,0,> : l.cfc, 601
\4'n{'n i / d,i i,i
: >lcilz.
Therefore, (V lY) > 0, where the equal sign holds only for the null vector.
* The subject of completeness is discussed in R. Courant and D. Hilbert, op. cit., Vol. I, p. 369.
LINEAR VECTOR SPACES 217
Definition
The norm of a vector is defined by
normY-/(YTY):{ffir.
A vector with unit norm is said to be normaliz.O. ;", non-zero vector can be
normalized and the result is unique except for an arbitrary phase factor.
Theorem
Two uectoru V, andYo satisflt the Schwartz inequalit2,
(Y" lV")(Yu lY,) ) l(% lYr)lr. (6.20)
Pnoor. Let Y : Yo * bYo. Then, from the previous theorem, the inner
product of V with itself is positive definite. That is,
0 < (Y lv) : (Y" + 6Yu lY, + bYu)
: (Y, lv,> + b(Y"l Yu) + b*(yo l%) + lll, (y, lyu).

Although the choice of b is arbitrary, it is convenient to give it the value that
will minimize the right-hand sideoftheinequality. Setting elAq(y
we have
ly) : 0,
b - -g!-H'J*=-1f'Q
(Yo '-lYu) 1*,trlt>
Then,
0 < (Y, lY")(Vu lYu) - l(% lYr)|,,
which proves the theorem. If either Y, ot Vu is the null vector the equality is
satisfied trivially. Otherwise, the equality is satisfiqd only if Y, u.rd y, ur.
parallel, that is, if Yb : LYo, where ,t is a scalar.
In ordinary 3-space the Schwartz inequality simply reduces to the state-
ment
(i. nF.rt > G.h,,
and
txl . trt >x.i,
cos(,P, n:#r.
By analogy we may define the "angle" between Yo and Y6 as
cos(Yo,Yu):ffi
2t8 THE FORMAL STRUCTURE OF OUANTUM MECHANICS
4. THE SCHMIDT ORTHOGONALIZATION

PROCEDU RE
In section I (under Postul ate 4) it

was shown that non-degenerate eigen-
functions of a Hermitian operator are automatically orthogonal, but the case
of degeneracy was not discussed at that time. We will now define what is
meant by degenerate eigenfunctions and will show how to construct an orthog-
onalized set from a set of degenerate functions.
Definition
An eigenvalue q is n-fotd degenerate if there are n linearly independent
eigenfuncti,ons correqponding to this eigenvalue. The n functions are called
degenerate eigenfunctions.
Theorem
Any linear combination of degenerate, linearly independent eigerfunctions is also an
eigenfunction ltauing the same eigenaalue.
Pnoor. Given the set of z functions y,rall having the same eigenvalue q
associated with the operator Q. That is, Qrpo: QlPrfor i : L,2,...,n.nThe
most general linear combination of these z degenerate functions is V : Ztntpn.
Then,
QV : : : I : QY,
|r,Orn \rnort |r,rn
which proves the theorem.
Corollary
In the case of n-fold degenerac2,linear combinations of the n degenerate eigenfunctions
i
ma2 be found so as ti form linearll-independent, orthogonal eigenfunctions corresponding
to the same eigenualue.
A prescription for obtaining an orthogonal set of linearly independent

functions is known as the Schmidt orthogonalization method. It will be
illustrated in what follows. Let a degenerate set of eigenfunctions be given
by ,1,r, !)2, . . . , Vw. Normalize g, and let it be the first function of the new set,
0t. Define the scalar product,
@rl'P,) : arz'
Choose
Qr:Vz-anQt'
Since we require that (,0r1,0r) - 0, let us see if our choice for ri, satisfies this
THE SCHMIDT ORTHOGONALIZATION PROCEDURE 2t9
condition. Thus,
@rl'0): (hlrpr> - an(ftl'0t>: (tn

- atz:0, as required.
Therefore, 0t and ,$, are orthogonal. f', is normalized by settinS @rl,0r) : t.

Now define the scalar products,
@rl'Pt) : (hB
and
@rl rPr) -- azs-
Now take rits : y)s
- aBQr - ezs$, and require that
(0r 0) :0,
@, 0) :0,
and
@, 'Q) : l.
Let us test the orthogonality of ,0" with ,$, and $r:
QLrl'0'): (0t lp') - an(Qtl{'r> - ar"p['l'0r)

:aLg-aB:0.
(0r1,0"): (Qr l v') - an(fizl{"> - ar"('\rl'0r)
:azs-azs-0'
Therefore, Qr, Qr, and ri, form an orthonormal set of functions (vectors). This
process may be continued for the entire set of degenerate functions. When the
Vi are ordinary vectors instead of functions in a function space, the inner
product (tpulV,) reduces to the familiar dot product.
PROBLEM 6-10 $ftl.,l,"rr.Y'd,iJnat$ .:.al#.'i.,{ire.fi;$}jt
Suppose that an energy level Eo is three-fold degenerate. If

the orthonormal, degenerate eigenfunctions corresponding
to this energy level ?te !)11 rp2, and tpr, show that
I
*: r/r(tPt+2tPr*2tP")
is also an eigenfunction of H corresponding to the eigenvalue
Eo.
PROBLEM 6-l I i..?..j,fi;riirftfui.f'a+tir.4ir.Ji.ir$:i+1q-1it+li.1ii$i,#jJ. fi'i:iqjr'i4:3ililffi;tqtr*r'ir.{?f,i,f.*iEti#
Construct a set of orthonormal vectors which are linear

combinations of the vectors whose components are (1,0, 2,2),
THE FORMAL STRUCTURE OF OUANTUM MECHANICS
(1, 1,0, l), and (1, 1, 0, 0), respectively.
(e"r.: Taking Qt : ++(l ,0,2,2), then
Q, : +I: (2,3, -2,r) and Q" : +!* (2,r,2, -3).)
PROBLEM 6.'2 tr;,lf$''f,.i#;9,s..#.Fl
Consider a two-dimensional isotropic oscillator such that

e) : @a : @u. Construct normalized, linearly independent
wave functions for the ground state and the first two excited
levels. (Hint: the wave function for a degenerate level is a
linear combination of all of the degenerate functions.)
5. LINEAR TRANSFORMATIONS
Once we have constructed a function space for representing any physical

state of our quantum system, a question arises with regard to the interpretation
of the role of the operators that we have been using. The operator equation,
Ya : QYo,
may now be thought of as a linear transformation in the vector space, such that
the vector Yo is transformed into the vector Y, by the action of the linear
operator a. The eigenequation, Equation 4.56, thus appears as a special
case in which the direction ofYo is not changed under the action of the operator
and V6 is simply Yo multiplied by u scalar, that is,
qX": QYo.
Two important mathematical operators which must be incorporated in the
formalism are the null operator and the identit2 operator. Their properties are
immediately evident from the following:
OYo : 0,
lYo : 5rr.
The product of two operators is defined in terms of the effect produced

by performing successively the transformations associated with these operators.
Thus, if Y, : PYa andYu : QYo, then Yc : P(QV,) : PQY,: RYo. This
suggests that
R:PQ.
We define the product of two operators in this sense, that ,R operating on Yo
DIRAC BRA-KET NOTATION 22t
N-dimensional space spanned lrv {0'}
Figure Gl Schematic diagram of

two successive transformations.
R has the same effect upon the
vector Yo as the successive
operations by Q and P.
with Q and then with P. This is shown schematic-

is equivalent to operating first
ally in Figure 6-1.
If for every Yo there is a unique vector Ya given by Y, : QYo, and if
for every Yu there is a unique vector Yo, given by Yo : PYa, then it follows
that
Yo : PYb : PqYr: IYr.
Here we have the statement that

PQ : l, (6.21)
which means that if an operation by

Q is followed by an operation with P,
the net result is an operation with the identity operator. Equation 6.21 permits
us to define the inverses of P and Q, provided that these operators are nln-
singular in the following sense:
Definition
An operator Q is non-singular if the equation 0Y : 0 is satisfied only by
the null vector.
If P and Q are both non-singular, it follows from Equation 6.21 that
P:Q-t and Q:P-',

where P is the left inverse of Q and Q is the right inverse of P. If one can show
that the operators P and Q commute, then the inverses defined above are
valid for multiplication from either the left or the right.
6. DIRAC BRA.KET NOTATION

Dirac called the vectors of function space ket uectors or simply kets. In
his notation the vector Y, could be written as lYo) or as la), where the label a
generally denotes one or more eigenvalues or quantum numbers. The basis

functions are called eigenkels since they are the eigenfunctions of the older
terminology. An arbitrary ket can be written as a linear combination of basis
kets, such as
lrp) : c, ll) + czl2) + ... : ltolr>.
A linear operator acts on a ket from the left. Thus, the eigenequation takes
the form,
Q la) : Qola). (6.22)
The Hermitian conjugate of each ket is a bra uector which is written as (al.
The bras comprise a space having the same dimensionality as the ket space
and which is dual to it. Since there is a bra for every ket, and vice versa, there
is a one-to-one correspondence between these two spaces. Operators act upon
bra vectors from the right, so that the equivalent form for the eigenbra equation
is,
(al QI : Io (al, (6.23)
where the Hermitian conjugate of Q is used. Note that multiplication from

the left by the scalar go is a meaningful operation, whereas the statements
Q @l or 8t @l are not defined.
The scalar product of a ket with itself is defined as the product of the
ket with its corresponding bra. That is, the scalar product of la) and la) is
(ol o).
In general, the scalar product of lD) by la) is
@lb)'
which is not the same as the product of la) by lD). In agreement with Equation
6. r9,
(olb) : (b lr)*.
The expectation value of the operator Q then becomes, using Equation
6.22,
(allQla) : Io@la): qo,
for la) normalized. From Equation 6.23 we have
(alQt lla) : 8o @lo) : q".
It is evident that if Q is Hermitian these results can be combined in the form
(al Q'la) : Io,
which is in agreement with the discussion in section 13 of Chapter 4. Note

DIRAC BRA-KET NOTATION 223
that in the Dirac notation a closed pair of brackets ( is a scalar

(a number), while a half bracket,
) quantity
I ) or ( l, indicates a vector.
A new linear operator appears in this formalism, namely, the quantity
la) (al' Because of its structure it cun operate on either'abraor
a ket and in so
doing it produces a transformed bra tr ket, respectively. For
exampii r.i
la) (al operate on the bra (bl from the right:
(b la) (al : a (al, where q : (b I a).
We see that the vector (Dl is transformed to the direction of (al and
that it is
multiplied by a scalar factor which is numerically equal to the projectio.,
(bl along (al. Now let this same operator act upon thl ket oi
lc;:
la) (al c) : f la), where f : (al c).
Here lc) is transformed to the la) direction and is multiplied by the

scalar
projection of lc) upon la). As a of these properties the operator la) (al is
l.t_"lt
called the projection operator, 9o. JVlultiple upplications of this operator
reduce
to the equivalent of a single applicati".r, ,ir*,
92:9o9":la) (ala) (al: la)l (al: la) (al: go, (6.24)
and similarly for higher powers.

Suppose now that we have a- complete, orthonormal set
. of basis kets,
li) (equivalent to the 'h of section 3). O.n". the projection operator go, for
projecting an arbitrary vector
9trt9 the ket li). iet us conside, th. g.r.;ui
vector Y' which, when written in the earlier notation, has the e*pansiir,
nr
V@ : 2 t,,Lo.
i:L
operating on Y, with the projection operator yields the result
gy,: li) (il a) : crli),

since, by definition, the coefficient c, is the projection of y, on titu Now, if
we sum over all possible projections we obtain,
2**,:,i ri> (ir a) : 2,,,u : ra).
This gives the important result,
.nr
> lt) (il

i:1
: 1, (6.25)
which says that the projection operator summed laer arut complete basis is the identit2
operator.
PROBLEM 6.13 {" .*.+,f..{a,*:r,f;liul#,iF,# lj
Using the Dirac notation, show that (a) the eigenvalues of a

Hermitian operator are real, and (b) eigenkets belonging
to different eigenvalues of an operator are orthogonal.
r+i*.#ta}li#*1,t-F, j;!F.
7. MATRIX REPRESENTATIONS OF LINEAR

OPERATORS
Let {,i ,} be an orthonormal basis for the .lfl-dimensional vector space

such that
(0n1,0):Qli):60,,
for all i,i : 1, . . .ri Itr
Define Yo : 2 o,Q,, where the a1 are given by
i:L
&t:(\olY")-Ulo).
Then the transformation equation
Yo : QVo
becomes
NNN
Yo : 2 f*,Oo : Q : 2",Q,$,.
Pr"tLr
Taking the inner product with yto,
lrN
r) : 2", Q Q li> : 2 Qo$t
Z^P-(ul
Using the orthonormality property, we obtain

nr
fr : i:L
2 Qunt where Qu : <il Q lj>. (6.26)
Thus we see that the components of the transformed vector can be characterized
by the effect of the operator acting on the old components, where both sets
of components are defined on the same basis.
The coefficients Qo, can be written in a square array which we call the
"matrix of Q" or the matrix (Q). Then we can interpret the equation
MATRIX REPRESENTATIONS OF LINEAR OPERATORS 225
Ya : a% as a matrix equation in which Y, and Yo are column vectors with

Iy' corrrponents and Q is an /[ x 1/ matrix. Thus:
r:
fl(: :) 0
g
As an example of the correspondence between matrix multiplication and

operator multiplication, consider the two successive transformations dis-
cussed above,
Yu : QY,
and (6.28)
Y, : PYb : PQY": .RY".
.n/
Let Y. yofo, where Tx: &lr). Then, analogous to Equation 6.26,
lc
Tr, : 2i P*n7n , and Equation 6.28 becomes:
: : ProZQtiei : r*0,,) o, :
4 (4
Tr, 2 Profn
1 4 ô,o,.
Therefore, the operator equation R : PQ is equivalent to the matrix
equation,
(Rr,): (; PooQoo),
which will be recognized as the prescription for multiplying the matrix Q by

the matrix P.
In order to further establish the equivalence between the algebra of linear
operators and matrix algebra, let us consider a few more important properties.
The matrix equation V, : QYo can be inverted to
Yo : Q-tVu,
only if the inverse matrix Q-' is defined and is non-singular.
Definition
A matrix is non-singular if its determinant,
lQoi * 0.
Definition
The determinant lQn,l muy be written as an expansion in its cofactors,
A : > (-I)n**Qr*lqnrl,
lQnrl: ior&
where the summation is over only one index. Here, lfrol is the minor of the
element Qn*. It is a scalar whose value is given by the (nf
- l) x (lf - 1)
determinant remaining after striking out the ith row and the ,ttr column of the
original determinant. The cofactor of the element Qn, ir (-l)o*, lqnl.
Definition
The cofactor matrix is the matrix of the cofactors of the elements of the
original matrix.
Cof (Q,,) : ((- l)r+r' lquil.
Definition
The transpose of a matrix is the matrix formed by interchanging the rows
and columns.
The transpose of (Qor) : (Q) : (Qoo).
Definition
The adjoint of a matrix is the transpose of its cofactor matrix.
Adj (Q,,) : Cof (Q): Cof (Q,,): ((-l)t+i lqnl).
Definition
The inuerse of the matrix Q is determined by dividing the adjoint of Q
by the determinant of Q.
e-,: (eno)-l :Ad'u@ : I t(-r;,*r lqrJ).

Theorem
A matrix commutes with its inuerse; that is,
ee_t:e_te_1.
MATRIX REPRESENTATIONS OF LINEAR OPERATORS T27
Pnoor.
Theorem
Tlte inuerse of a product of non-singular matrices is equal to the product of the
inuerses in the reaerse order.
Pnoor'.
Let
(Anc;-' -o
Then,
(ABC)(ABC)-l : ABCD
1: ABCD
A-1 :A-IABCD:BCD
B-1A-1 :B-IBCD _CD
c_1B_1A._1 _C_ICD:D
Therefore,
(ABC)-I : C-rB-U.-I.
Following is a summary of useful definitions and properties of matrices.
Definition
Determinant: L: lQoi
Transpose: (Qr) : (Qon)
Cofactor: Cof (Q) : ((-l)t+t 1roU,
Adjoint: Adj (Qn,) : Cof (Qo,)
fnverse: Q-l :i oo, O
Complex Conjugate: (Qno)* : (Qil)
Hermitian Adjoint: (Qn,)t -- (Qn,)* : (O,[)
A matrix is symmetric if Q:0_

skew-s2mmetric if Q:-Q
real if Q:Q*
pure imaginarlt if Q:-Q*
Hermitian if Q:QT
shew Hermitian if Q:-Qt
ortltogonal if Q-t:0
unitarlt if Q-l : Qt
Properties
tfil ::3tat
(AB)t
Ba
(AB){€ : A*B*
(A'n;-t - B-1A-1
Properties of determinants
lAl : lAl
:
lA*l lAl*
lAtl :la*l :lAl*
We will now show that a Hermitian operator A is represented by u
Hermitian matrix. Using the notation of Chapter 5,
Q;t: $lqi>,
Qrj: (ajli): <ilail: Qr,
where the Hermitian property of Q has been used here. Then,
(Qn,) : Gil) : (Q)t,

and the matrix is Hermitian. The Dirac notation of section 6 is even more
transparent, as the following will show:
Qu : A qU>,
and
Qfi:<jlQli):Qt*
Therefore, (Q) : (Q)I .
It is evident that the elements along the principal diagonal of a Hermitian
matrix must all be real, that is,
<ilQ lf): (4Qli>*.
The matrix equivalent of this Hermitian form is
(F|Q,D : (gfQ,pn)*,
MATRIX REPRESENTATIONS OF LINEAR OPERATORS 229
where ytnis a column vector and its transpose, Fnris a row vector. This expression
may be readily verified by starting with the right hand side:
@iQrpr)* : (g,Q *rpil : (fitQrp,l) : ((6,)'pi) : (flQ,p,).
PROBLEM 6-11 Fstut"-ry,f1-if"s+,r9'J
Show that
cof (Qn,) : c6f Q").
PR'BLEM 6 ts
.':'T:j,#T,:-:"":T.'
fi\2 i2 sll)
(u) (b)
(; ::\
\-12
'l
' d
Ans.:
fi1) fi:l)
ft
l;_i -) U _i l)
d('l
PROBLEM EI6
'(_1
ffiffid,#P,
_i I 'l l)
Evaluate the detorminants of the four matrices in Problem
6-15.
(Ans. : 6, 41 5, 30.)
s#H
PROBLEMs.T
;;:T:-
Ans.:
:H:ffiHffiilffi :Tff
(u) l& -+ +\ (b) l+ o -+\
l-;i il
(.) I -? -+\
l;-l )
+ (d) l+ o o\
('-; (: ;:)
-; )
PROBLEM 6.18
By means of matrix multiplication show that the product of

each matrix in Problem 6-15 with its inverse in Problem 6-17
equals the identity matrix.
. :::::r...1._ " .-llllllllllrrrrr.::::ii;;;;:::i::ii;;;tri !a a::::::1 ::rr1111l:
::r I rrrr .'"r::i.,. .
8. THE MATRIX FORM OF THE

EIGENVALUE PROBLEM
Consider the eigenequation
Q la) : )"la),
which is equivalent to writing,
.^/.^/
2Qn,n1\r: A2";Q,
of,
.^r
- 16J,Qr:0.
4",(Qno
The condition for non-trivial solutions of this set of linear equations is that
the determinant of the coefficients is zero. That is,
Qtt-1 Qt, Qt
Q^ Qrr-1 Q*
lQu-16n,1 : Q", Q* Qr"-I'l:o. (6.29)
THE MATRIX FORM OF THE EIGENVALUE PROBLEM 23I
This equation, which is called the secular equation or the characteristic equation,
is a polynomial of order Nin )". Its i/ roots are the eigenvalues of the matrix
a.
It is evident that the problem of obtaining these eigenvalues would be
greatly simplified if the matrix (Qn,) were itself diagonal. In that case,
Equation 6.29 would become
(Q" - A)(Q,, - 1)(Q" - 1) '" (Qrr - 1) :0,
and the y'/ eigenvalues would be the diagonal elements Qrr.,Qrr,..., Qrr.
We will show in the next section that there is a transformation, called a unitar2
transformation, such that anlt Hermitian matrix ma2 be written in diagonal form. Such
transformations are analogous to the principal axes theorems of classical
mechanics and geometry, wherein cross products involving two different co-
ordinates can be eliminated by choosing an appropriate transformation of
coordinates. The off-diagonal terms of a matrix correspond to the cross terms
in the geometrical problem. For example, a surface in 3-space with its principal
axes along the coordinate axes may be described by the expression
d(i) : 2Ann*l: constant.
But the vector i is in a principal direction when it is parallel to the normal to

the surface; thatis, when e. Oi : i. Butthe direction cosines of the normal
are given by 0$l1xo, so the component equations for the principal axes are
a6
dxt -xn
of,
Auxr : Lxo, where ,?t is a constant.
This is immediately rocognized as the eigenequation in matrix form.
PROBLEM 5.19
Show that:
(u) The eigenvalues of a Hermitian matrix are real.
(b) The eigenvectors of a Hermitian matrix corresponding
to different eigenvalues are orthogonal.
We state without proof that Hermitian matrices have all of the properties
of Hermitian operators; that is, the operators and matrices are isomorphic
representations of the same abstract algebra.
THE FORMAL STRUCTURE OF EUANTUM MECHANICS
g. CHANGE OF BASIS: UNITARY TRANSFORMATIONS

It was shown in the previous section that the operator equation, Ya : QYo,
can be interpreted as a linear transformation of the vector Yo into the vector
Yo by the linear operator Q in the .lf-dimensional vector space defined by the
basis {rin}. Now we ask, is it posssible to describe this same transformation
from a different coordinate system, that is, using a different set of basis functions ?
We first define a new set of basis functions in terms of the old basis. Thus,
'0'n
: Lui,'4,,
i:L
where the basis {fr} ir a complete, orthonormal set of .l/ functions. Since we
require the new basis to be orthonormal, we have,
: 6,
@i1,0')
4,rr) : 6,i
/N
(> uh,O0
\ft
1,0,)
: 6nt
Zu,rrf,@r
un*u'f* : 6nt
7
(6.30)
Tuora|, -'6,r
.
Therefore, we conclude that

UUt:I
is the necessary and sufficient condition that the new basis be orthonormal.
We conclude that: A complete orthonormal basis for an N-dimensional space ma2 be
obtained b1 means of a unitaryt transformation on the old basis.
Next, we wish to show that the transformations given by Equation 6.28
go into
Yi : Q,VL
Yi : P'Y;:P'Q'YL:R'YL,
under the unitary transformation which takes the basis {rir} into the new
basis {rii}. It is desirable that Equation 6.28 preserve its form. Furthermore,
*. r..d to know if the operatorS P', Q', and.R' differ from P, Q, and ^R.
Let us write Yu : QY" in terms of the old and new bases. First, consider
Y":
n/nrN
Yro : 2 oo,$n
i:Lii
CHANGE OF BASIS: UNITARY TRANSFORMATIONS 233
Taking the scalar product with ti:o,
&k : 4 "iuf,6,*
: 4ilt"i.
(6.31)
In matrix form this is Vo : UtYl or
fl:uf)
To obtain the inverse transformation multiply Equation 6.31 by u,* and sum
over *:
: 6rai
Fu,*n*: ; \u;*fi,fia.; >
where Equation 6.30 has been used.
', d.'i :2uftq*.

That is,
Yi : Uv,
In like manner we have
Va : UtY; and Yi : U%.
Then, for Equation 6.28 we have:
Yr: a%
utY; : QUtY, (6.32)
Yi : UQU-lY;, (6.33)
since Ut : U-1. Thus we can preserve the form of Equation 6.28 if we define
Q' : UQg-t (6.34)

or, conversely,
Q : U-tQ'U
We conclude that: A giuen linear transformation can be expressed as a matrix equation

in an1 representations obtained from the original representation b1 means of a unitary
transformation.
This transformation may be interpreted as
(1) a transformation of the basis vectors, Equation 6.32, or
(2) a transformation of the operator, Equation 6.33.
To complete our verification of the validity of the matrix formalism in the
new representation, consider the equation R : PQ. Perform a unitary
2T4 THE FORMAL STRUCTURE OF EUANTUM MECHANICS
N-dimensional space spanned Uv { 0,} N-dimensional space spanned by { 0" }

Figure G2 Schematic diagram of a unitary t'ransformation.
transformation on each member.

URU- I o"-'
: $&U;1:o J--",t
::
R' $%Y-1)(uQU-r)
"'
We conclude that there is a one-to-one corre$pondence (an isomorphism)
between operators in the vector space and their matrix representations. This
is shown shematically in Figure 6-2:
Let us now examine what quantities are left invariant by u unitary
transformation. First, consider the transformation of Equation 6.29:
lQ' - 1'11 : IUQU-' - U,llu-tl : IUQU-I - ]Lll : lu(a -,i1)U-11,

since U and U-l commute with both i and 1. But the determinant of the product
of several matrices equals the product of their determinants. Therefore,
lu(a - ,er)u-'l : lul . la - ,111 '1u-t1

: lul . lu-'l . lQ - rtl
: luu-rl . la - nl
.'. lQ', - L',l| : lQ - 11|1.
Also, the trace of a matrix is invariant under a unitary transformation. The
trace is defined as the sum of the elements on the principal diagonal. The
proof of this is trivial since Q*o is a number and thus it commutes with U:
4 A;, : nUo*Q**U*' : nQn*Uo*Uot : 2 Q*'

We now can list the following properties of a unitary transformation:
1. The secular equation of the matrix is inuariant.
2. It follows that the eigenualues are inuariant.
3. The invariant.
determinant-which is the product of the eigenualues-is
4. The trace-which is the sum of the eigenualues-is invariant.
The following properties of the trace are stated without proof:
1. The trace of a product is not changed by u cyclic permutation of the
factors.
TI(ABCD) : Tr(BCDA) :T1(CDAB), and so on.

DIAGONALIZATION OF MATRICES 235
2. The trace of a commutator is always zero.
Tr(AB - BA) :Tr(AB) -Tr(AB) - 0
3. The trace of a sum is equal to the sum of the traces.
Tr(A +B) :TrA+TrB.
IO. DIAGONALIZATION OF MATRICES

A unitary transformation is just one type of transformation among a
larger class known as similaritlt transformations. If U-1 - tI, then the similarity
transformation U-IQU is called an ortltogonal transformation. A unitary trans-
formation, where I-J-l : U* : Ut, is an extension of the orthogonal trans-
formation to complex matrices.
Although not all matrices can be diagonalized, any square matrix can be
transformed by a similarity transformation such that its eigenvalues appear
on the principal diagonal (although the ofiF-diagonal elements are not nec-
essarily all zero). More important for our purposes, the matrix representations
of physical operators are usually Hermitian and these can be diagonalized.
The importance of a unitary transformation now becomes clear. For,
if there is a transformation which will result in a diagonal representation of
the operator Q, then the solutions to the eigenvalue problem are immediately
obtained by the process of diagonalization Defining Q as the diagonal matrix,
A : 1l (that is, A,, : 1$u), and Q'as its non-diagonal form (now called Q),
we have from Equation 6.3+,
QU : UA.
Writing out the matrices explicitly, this is
Qrt Qt, Qr" ul
Q^ Q* Qr" uZ
Qtt Q* Q* T u!
:) f
ul ul
ul u", 120
u!
(l u! 0 ).s
f
If we consider the columns of the U matrix as separate vectors, note that the
above result is equivalent to the separate eigenequations:
Q%, =t).fl/r\ (q - trJ)qlL - o\

Q%, =tlzQlzl (a - )"J)q/z: o
", ta- l, (6.35)
Q%"=tlsqlef LJ)qls:01
'/ "/
where
%,:(r) %z:(r) ands..n
We now have the prescription for obtaining the matrix of the transfor-
mation U:
First find the eigenvalues of the operator matri* Q.
For each eigenvalue 2,, obtain the eigenvector Qlo from Equation 6.35.
U is comprised of the Ql o as columns.
EXAMPLE
Given,
l1 0-2\
Q:
(; : i
Find the eige nvalues and eigenv ectors of Q. Construct U and U-l Show that
U-leU : A.
lr - ),
;'l : (-i)t(r -
0
I 0 - 4l :0.
I
-), 1)(4 - D
l-2
I
0 +-l
Then the eigenvalues ar e)":0, 0, 5. Taking lt : 0, we have from Equation
6.35, Q%r: 0, or
o -2\ /ui\
(i :;)l;;)-o
which yields the results, u!. :2t
2u!.
- Since az| is completely arbitrary, let it be zero. Then the simplest choice
for a normaliz ed Q/, is:
%'l:*(i)
Taking 1z :0, we have as above , uI : 2u3. Since e/, must be orthogonal
to Qly choose u3, : I and u?. : u3 : 0. Then,
%z:0 . 1
Taking ls -- the eigenequation itS - 51) %s:0, which becomes:

,5,
(-:;
\-, l) ffi:0
Then,
uI : -2u? r
uz : o'
and we obtain
%s:#( ;)
\ -zl
The transformation matrix is constructed by placing the %oalong the columns:
lz- o -r\
(
::, 'j) :+(i
'l i)
\/s {5/
lul : -1, Cofu - Adj u

,l'o tt's
: -75lo
\rI i)
and
IJ-l : U.
u.'QU:-(l
'l il( i I il(l 4 i)
:.( ,l
ilfi ; ,l)
:( ,l
)fi I il:fi Iil-^
Find the eigenvalues and eigenvectors for each of the
matrices given in Problem 6-15.
(Ans.: (a) I, 2, 3; (b) I, 2, 2; (.) 5, l, 1; (d) 2, 3, 5.
The eigenvectors are the columns of the matrices given in
Problem 6-21.)
PROBLEM 6.2'
The matrix U of the transformation that will diagonalize

each of the matrices of Problem 6-15 is given below.*
I
-\/2
-:
I
t/z
':,j)
0
* It should be pointed out that we use the symbol U for the transformation matrix even for those
cases where Q is not Hermitian and U is not unitary. In general, the eigenvectors of Q will not be
orthogonal unless Q is Hermitian. Also, lJ-L + Ut unless U is unitary.
(b) Cannot be diagonalized.
(.) ll
t/s \/,
I
--0
\/3
11
{)
t/g \/,
(d) lr 00\
(': o)
0' t/
Find the inverse U-r of each of these.
PROBLEM 6.22
carry out the multiplication to show that the matrices of

Problem 6-21 do indeed diagon alize the matrices given in
Problem 6-15.
PROBLEM 6.23
ir'l;i1::il::iltii,t iLJ
For each of the matrices of problem 6-15, verify that the

trace is the sum of its eigenvalues and that its determinant
is the product of its eigenvalues.
PROBLEM 5.24
Given the matrix,

/r i
A- Il_-o o
\1 0 i)
(u) Is A Hermitian ?
(b) Find the eigenvalues of A.
(.) Find the matrix U for the transformation that will
diagonalize A.
(d) Show that the eigenvectors comprising U are ortho-
normal.
240 THE FORMAL STRUCTURE OF QUANTUM MECHANICS
(.) Show that U is unitary by obtaining Ut and verifying

that UUt : 1 by means of matrix multiplibation.
(f) Carry out the matrix multiplication to verify that
U-1AU is diagonal. (Remember that LJ-l : Ut for a
unitary matrix.)
Ans.: (b) 0, -1,2; (c)
il
PROBLEM 5.25
Given the Hermitian matrix,
/v oilv\
r:l tt o I o l.
\-ify o y I
(u) Find the eigenvalues of f.
(b) Find the unitary matrix U for the transformation that
will diagonalize I.
(.) Verify by matrix multiplication that UUt : l.
(d) Verify that U-1IU is diagonal.
Ans.: (a) l, y(l + B).
"(L
:)
ffi.q.e.wP
II. APPLICATION OF MATRIX MECHANICS TO THE

HARMONIC OSCILLATOR
In the previous sections itwas shown that in a linear vector space an
arbitrary vector rrray be denoted by giving its components along a complete,
orthonormal set of basis vectors which span the space. Furthermore, each
linear operator, to which these vectors are subject, can be represented by u
APPLICATION OF MATRIX MECHANICS 241
matrix defined on the same basis vectors. The effect of applying a given
operator to a vector, then, can be determined by multiplying the matrix
representation of the operator times the column matrix of the vector components.
We have seen that the energy eigenfunctions of the harmonic oscillator
form a complete, orthonormal set, the only complication being that the set is
denumerably infinite. Thus, if we regard these eigenfunctions as our basis
vectors we will be using a space of infinite dimension. For example, let us
consider the energy matrix for the harmonic oscillator. This is, of course, the
matrix of the Hamiltonian operator using the oscillator wave functions as the
basis functions. Each matrix element is given by
Eu : A tr lj) : Eo(lj> : E,6n,.
Since both (il and 17) are eigenvectors of ,f , the effect of the operator is
merely to generate the eigenvalue Eti which, being a number, can be moved
outside of the integral. The integral is zero for all off-diagonal matrix elements
because of the orthogonality of the oscillator wave functions. If we label the
rows and columns in the order, Vo, Vt, 1pz, . . . , the matrix appears as follows:
(6.36)
Equation 6.36 illustrates the important fact that an operator has a diagonal
representation when its own eigenfunctions constitut6 the basis.
As a further example let us write the matrix of the operator x, whose
matrix elements were given in Equation 5.41. It is as follows:
0 IO 000
I o \/2 000
0 \/, 0 \/300
/\ I
o \/5
\/ t/20
(6.37)
0 002 o \/5
0 000
Notice that the matrix of x is not diagonal. This was to be expected since
the basis functions are not eigenfunctions of x.
For problems involving pure states, the value of the matrix formulation
is not evident. A somewhat better example of its utility is that of Problem
5-22, in which an oscillator is in a superposition of two states. The time inde-
pendent wave function as well as the function which includes the time de-
pendence are given below:
I I o-ouorn
t/z \/r"
I l'-
---= r-'u,'tu
f2 {z
,p(x) : 0 Y(r, t) : 0
0 0
The expectation value of the energy in the time varying state is:
ah I
T 00 o-o,o',n
\/r"
\.- I
{2
,ouorro. +
'{2 ,iEltrh g
ryo \/r"
o-'ur,lu
,u o2 5ah, 0
l: ah o-inotlh
{2
rouoln r+
'{2 riErtrh, o > o, 2\/r"
3ah o-iELtrFL
2\/r"
0
:- ah ,+3ah :
I- a)h.
+
SUMMARY 243
Although the last-example is still quite elementary, one can easily visualize
extending the method to more complicated situations. The real power of the
matrix method will appear when we discuss perturbation theory in Chapter 9.
PROBLEM 6-26
(u) Using Equations 5.56 and 5.57, obtain the matrices

for the raising and lowering operators, aI and a, with
the harmonic oscillator wave functions as the basis.
(b) With the help of Equation 5.58, obtain the matrix
representations of the operators p and q.
(.) In the same manner obtain the matrix representations
for x, P,, x2, p3, T, and V for the harmonic oscillator.
(d) Which of the above matrices are Hermitian ?
PROBLEM 5.27
Use the matrix method to calculate the expectation values

(x), <p*> and <E) for the following mixed state of the
harmonic oscillator:
: +yo(r,
{ro v\ / ' .hY,(x,r) + hY,@,t)
Y(", t) /)
'
++Y"(x.
{to
o\'7 t).
SUMMARV
A quantum system may be represented by u wave function which contains

all of the information that can be known about the system. The wave function
satisfies the Schrodinger wave equation, which provides a description of the
development of the system as time progresses. The superposition principle is
valid for wave functions. This means that any linear combination of *urr.
functions which represent physical states of the system is itself a possible physical
state. The connection between the mathematical wave function and physical
measurements is made by associating a linear, Hermitian operator with each
dynamical variable. A measurement of a particular variible can result in
onfit one of the eigenvalues of the operator corresponding to that variable. If
the system is known to be in an eigenstate of the operator, the result of a
measurement of the corresponding variable can be predicted, with certainty,
!o be that eigenvalue. Otherwise, only a probabilistic prediction can be madl.
The average value of a large number of measurements on the same or identical
systems is called the expectation value. This is the weighted mean of all of
2U THE FORMAL STRUCTURE OF OUANTUM MECHANICS
the accessible eigenvalues, where the weighting factor for each eigenvalue is
interpreted as the probability of obtaining that eigenvalue. After a measure-
ment of the variable q, the system is known to be in an eigenstate of the operator
a. If this is followed by a measurement of !, the act of measurement forces
the system into an eigenstate of P. If the operators P and Q commute, the
measurem ent of p will not disturb the eigenstate of Q, and, in fact, any sequence
of measurements of p and q will repeatedly yield the same two eigenvalues of
the operators P and Q. However, if P and Q do not commute, a measurement
ofp will destroy all previous knowledge concerning the state of Q, and vice
versa.
A quantum mechanical state, which was just regarded as simply a_wave
function, can be visualized as a vector in an abstract function space (Hilbert
space). Such a vector may be represented by u columnar matrix array of
numbers which characterize its projections on a given set of basis vectors.
In like manner, an operator Q may be represented by a matrix array composed
of the elements <,(tnl Q 1,,$,) calculated from a particular set of basis functions,
{{t,}. Just as a vector has an existence that is independent of the set of unit
vectors used to define its components, an operator transcends the particular
set of basis functions used to express its representation. The diagonal elements
of the matrix (i : j) are the expectation values of the operator Q, and the
ofldiagonal elements (i t' j) will later be identified as the transition probabilities.
Although ,Qo and {t, are orthogonal, the operator itself might connect the two
states. Physically, this means that the act of measurement produces an inter-
action which has a non-zero probability of causing a transition from one state
to the other.
A coordinate system in 3-space can be rotated so that an arbitrary vector
1 lies parallel to one basis vector, say fc. Then the vector has a particularly
simple form, namely,
t:
0
In like manner, a "rotation" in function space can be effected by applying a
unitary transformation to the matrix a . There is one particular transfor-
mation that will enable Q to be written in diagonal form such that its elements
are the eigenvalues of the oper ator Q and its basis functions are the eigen-
functions of q. An equally valid procedure would be to construct the new
basis by formittg upptbpriate linear combinations of the old basis functions
so as to make I diigo"ut. In general, however, it is easier to obtain the
eigenvalues directly by solving the secular equation._ -If !!. eigenfunctions
ari treeded, they can then be obtained easily by means of the diagonal matrix A.
SUGGESTED
REFERENCES
David Bohm, Qunnturn Theory. Prentice-Hall, Inc., N.Y., l95l'
R. H. Dicke and J. P. Wittke, Introduction to Quantum Mechanics. Addison-
P. A. M. Dirac, The Principles of Quantum Mechanics, Srd ed. Oxford University

Press, London, 1947.
R. B. Leighton,Principles of Modern Ph2sics. McGtaw-Hill Book Co., Inc.,
New York, 1959.
P. T. Matthews, Introduction to Qunntum Mechanics. McGraw-Hill Book Co.,
New York, 1968.
E. Merzbacher, Quantum Mechanics. John Wiley and Sons, Inc., New York,
1961.
Albert Messiah, Quantum Mechanics. North-Holland Publishing Company,
Amsterdam, 1958.
J. L. Powell and B. Crasemann, Quantum Mechanics. Addison-Wesley Pub-
Iishing Co., Inc., Reading, Mass., 1961.
V. Rojansky, Introductoqv Quantum Mechanics. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1938.
Leonard I. Schiff, Quantum Mechanics, 3rd ed. McGraw-Hill Book Co., New
York, 1969.
Robert L. White, Basic Quantum Mechanics. McGraw-Hill Book Co., New
York, 1966.
Klaus Ziock, Basic Qunntum Mechanics. John Wiley & Sons, fnc., New York,
r969.
CHAPTER 7
Wffiffi WffiWffi ffiffiIffi#&Wffiffiffi ffiffi
Wffiffiffiffi ffiffiffiffiffiffiffiffiffiffi
We will now apply quantum mechanics to three-dimensional systems. The

reader has at his disposal the formal tools of both wave mechanics and matrix
mechanics, as well as his experience in treating one-dimensional systems in
Chapter 5. After a brief discussion of the separation of variables in rectangular
coordinates we will turn to the two-body problem. This will be solved in
spherical coordinates and will be applied to the hydrogen atom. The important
role of the angular momentum operators and their eigenfunctions will be
emphasized.
I. RECTANGULAR COORDINATES IN THREE

DIMENSIONS
The time-independent Schrodinger equation in rectangular coordinates is
h2
:
- 2*Y'rpt(r,), z) + V(r,), z)rln@,), z) Eyu@,1t, z).
In the special case of the free particle (V :0), the solutions are plane waves
of the form
'ttn@, ), z) : (#'$i - ({J,'r" ,ikuu . ,ikrz (7. 1)
Whenever the coordinates are independent, that is, they are not connected
by interactions, the Schrodinger equation is separable and its solutions may
be written as a product of one-dimensional solutions as in Equation 7 .1.
Furthermore, the total energy eigenvalue is equal to the sum of the eigenvalues
of the one-dimensional problems, that is,
E :ry2m: L
2m
fo:, + ki + kZ) - E* * Eo + E,.
246
RECTANGULAR COORDINATES IN THREE DIMENSIONS 247
Even when a non-zero potential exists, these same principles apply,

provided that the potential can be expressed as
V(r,), z) : V"(r) i VuU) + V,(z).

In such cases, the Schrodinger equation may be separated as follows:
l- #*r,* v"@)f,pn *l- ##,. v,rt)fvz *l- #$,* v,@f'p'

: Erlrt,
Expressing the solution as a product function,
Tn : ,pn,@) ' vnuU) ' ,pn"Q), '
we obtain for the x-equation,
f_h'a'
L 2m ox2 + v"(r)
r
)rt,n,@)
: E*rttn,(x).
Similar equations hold for the other coordinates with the result that
E-8,+Er+8,.
Another example is that of a particle in a box. For simplicity, let the box
be a cube of side I. Although there is no force acting upon the particle and
its total energy is positive, it does not have a continuum of available energy
states like a free particle. The presence of the rigid walls (infinite potential
barriers) forces the particle to occupy one of a discrete set of energy states.
We found these states for the one-dimensional case of an infinite square well
in section 3 of Chapter 5. From Equation 5.18,
n2rn2h,2
En, :
2mL2 '
and
,tt@) : t . nanx
Zsrn L
Then,
rn,,nn,,": (?)- sinff ' snW' snff , Q.2)
and
D
n2nzfP
Lnln,n, , (7.3)
- 2*L,
where
y72:nl+n?,+n2.
THE WAYE. EOUATION IN THREE DIMENSIONS
PROBLEM 7.1
(u) Using Equation 7.3,whatis the ground state energy of a

particle in a box ?
(b) How do the energies of the first and second excited
levels compare with the ground state energy ?
(.) What are the degeneracies of the three lowest levels ?
(Ans.: (b) 2Eo,3Eo; (c) 1, 3, 3.)
PROBLEM 7-2
Calculate the density of states (number of states per energy

interval) for the particle in a box. (Hint: see the calculation
of Jeans' number in Appendix A of Chapter 2.)
'
(n",., #frm)881.)
PROBLEM 7.3 'b-:-:i:1'srr;:i*rt}lt.i;i;llliii.i,,jlfillf:;i.:ffi;ig-Tr*ii;-?q:+i+iLf
(u) Find the ground state energy of a three-dimensional

harmonic oscillator by treating it as three one-
dimensional oscillators having the same frequency.
(b) What are the energies and degeneracies of the three
lowest levels ?
(.) Show that the degeneracy of the nth level is given by
t@*r)(n+2).
(Ans.: (b) trh, **h, Lrh; 1, 3, 6.)
2. SPHERICALLY SYMMETRIC POTENTIALS

A two-body problem can be reduced easily to a one-body problem in a
conservative system in which the only force is one that is proportional to the
distance between the two bodies. In such a case, the problem has spherical
symmetry in a relative coordinate system whose origin is fixed at one of the
bodies.
Consider, for example, a proton of mass M with position vector /, and an
electron of mass m and position vector i, (see Figure 7 -1). Assume that the
only interaction between the particles is that given by the Coulomb potential,
o
-2
r/ -o
' - li, irl
-
SPHERICALLY SYMMETRIC POTENTIALS 249
t-:)
rc.- -:lrr
Figure 7-l Coordinates for the two-

body problem.
where ris the distance between the particles. The time-independent Schrcidinger
equation, Equation 4.54, becomes
l- h2 -,, hz 1
: EYu,
l- Trro? - frv3lv, - v(r)Ys (7.4)
where Vf operates only on the coordinates of the proton and Vr2 operates only
on the coordinates of the electron. A difficulty arises with this expression,
however, since the potential V(r) is a function of the coordinates of both particles.
This means that Equation 7.4 is not separable in the two sets of coordinates.
For this reason it is convenient to transform the equation to center-of-mass and
relative coordinates,
The position vector ,R for the center of mass of the two particles is given
by the relation,
Mlf, -i{t :mlR_.irl,
from which,
fi : M3! mrz. (7.5)
M +m
The kinetic energy of the center of mass is given by
T - +(M * m)R?,
while the kinetic energy of the two particles about the center of mass is
T' : +M i, - Fl' * im lR - 7,1r.
Using Equation 7.5, we can write
li'-Fl : #*lir- irl:ffin,

.MM
lR - Frl :
mlit - frl : M ffilil'
THE WAVE EQUATION IN THREE DIMENSIONS
From the time derivatives of these expressions we readily obtain
Mm
T' :'
M+m
Therefore, the Hamiltonian for the system becomes simply
#: T + T' + V : +(M + m)R' * LrFi' + v(r)

:ffi*e;+v?):-ffiq,.
+ + v?)f : ff'* * ffa"'
[- fr""
where
F : (M + m)R: -ih.i,^,
F : pi : -ih,i na,
and is the reduced mass defined in Equation 3.23. Since the Schrodinger
p,
equation is now separable in these variables, we may write its solution in the
form of a product function,
Yz : IPr*' garr
For a conservative system the solution Vcm represents the continuum of free
particle states for the system as a whole, and the function yor, represents one
of the discrete eigenstates of the bound system. Thus the total energy eigen-
equation rr'ay be written as
ffYt: (ffr* * ffn"r)!)r*'VR"r: (Er* * E^nr)tltr*'lparr: EYn.

We will now focus our attention on the bound system and will disregard
the energy of the system as a whole. Specifically, we will be concerned only
with solutions of the problem
# nrr,\)nrt : E R"r,l|art.
Hereafter we will drop the label "Rel" and will write the equation to be solved
as
l- #,vz + v?)f!,n : E,p'. (7.6)
In Cartesian coordinates, Equation 7.6 would not be separable because V(r)

would involve the radical (x, *), + z\t. Therefore, it is simpler to trans.
form V2 to polar coordinates and to make use of the fact that the radial and
angular puitr of Equation 7.6 can be separated. It is convenient to denote
the angular part of the Laplacian by 92 and the radial part by fr. That is,
SPHERICALLY SYMMETRIC POTENTIALS 25t
we let
l-
Y2:9 + ro g',
where
gE: (#.?*) (7,7)
and
g)z ra2
- ##(""'*,) + ,i"'0-
PROBLEM 7.4
Show that the Laplacian operator in spherical coordinates

is given by
v2 - (# .? *) . ; [# * (""' *) + # #,]
.'.
Since the potential term is independent of the angle variables, it is con-

venient to use a product wave function of the form
Vn(r,0, +) : ft(r) 'Y(0, +).
Then Equation 7.6 becomes:
l- #,^R?))YQ, il - #tszY(o,d)l 'n(,)

+ V(r).R(') .Y(0,6) : E' n(') ' Y(0, +).
Y(0, il is written outside of the first bracket to indicate that it is not affected
by the radial operator. Likewise, ^R(r) is not affected by the angular operator
in the second bracket. Multiplying each term by
2pr' I
hz R(r) Y(0, 4)'
.
we have the separated equations
r29R(r) 6) grY(0,
R(r)
+ryw -v(,)): - -wT:r\' a
(7.8)
Setting each member of Equation 7.8 equal to the separation constant, A,

252 THE WAVE EOUATION IN THREE DIMENSIONS
we have the two equations,
eR(r) +#W - v(r)lR(r) :#^r, (7.e)
and
gzY(,, o) : -.LY(', o). (7.10)
Since we know the differential operators 92 and 9, Equations 7.9 and 7.10
can be solved by the standard techniques for ordinary differential equations
once the exact form of the potential is put into Equation 7 .9. However,
the reader should recognize Equation 7.10 as an eigenequation of the type that
we have encountered before. Therefore, we prefer to solve Equation 7.10 by
means of operator algebra in a manner similar to that used for the harmonic
oscillator. But first we must digress to discuss angular momentum and its
operators in three dimensions.
3. THE ANGULAR MOMENTUM OPERATORS

The classical definition of the angular momentum vectorl is
E:i x P,
having components given by
L, :fiu
Ln : Z!, tT;l (7.1 l )
L, : xpo
We use the same definitions for the components of the angular momentum
operators, where it is understood that po is the momentum operatot,,-ih(7ldxo).
As we discovered earlier (Problem 4-22), a coordinate operator and its
conjugate momentum operator do not commute. Hence we should not expect
the components of the angular momentum operator to commute with each
other.
PROBLEM 7.5
Verify the following commutation relations:

(u) lrn, !,f : ih6il)
(b) lxo, x'f : 0 ). (7 .t2)
(.) l\rf,l:o )
1 Although
t :; x p'b.hu'u.s like an ordinary vector (polar vector) under addition, it is actually
a pseudovector (axial vector). Its different behavior appears under a reflection ofall coordinates through
the origin: i u.rd 1urc reversed, but Z is not.
THE ANGULAR MOMENTUM OPERATORS 2!'3
In particular, consider the commutator lL,, L;. The multiplication may

be carried out explicitly, or it may be indicated as follows:
lL,, Lrf : l(fi, - z!), (rp" - xp")l
00
:Ifi", z!,f - llp.
/v
x!,1 lzfo, zp,f + lzQ,
- x!"1
:il,lf u, zf * xgolz, !,f
-- -ihlP, ! ihxP, : ih(x!, - )P.) : ihL,.
In the second line above, the two indicated commutators are zero because all
of the factors in each bracket commute. In the third line, all commuting
factors have been taken outside of the commutator brackets. The commutation
relations for the other angular momentum components may be obtained easily
from cyclic permutations of x,1, z in the expression
Thus,
lL, L,f : (i.13)
and
lL,, L;
Asin the case of vectors, the square of the total angular momentum operator
may be defined by
L2:L'r*L'r+L'r.
PROBLEM 7.5
Verify that the commutation relations for the components

of the angular momentum operator may be summarized by
the mnemonic expression:
Ext:ihE.
If were an ordinary vector, E x
^d i would be identically
zero. Why is this not true here ?
PROBLEM 1.7
Prove the following commutation relations:

(u) lL', L,f : lL', L) : lL', L,l - 0.
(b) lL" L"l : lL" L\l : lL" LTl - o.
#,,*i;4.:,*.iit 9"s.rP;i.d:tr{i"1$.F"^F,riid;Ei+'#-..q,r o-.1"{i*;,6,rr,+;ii.F"i9fil}lF

'4.3;)ijBff
The commutation relations derived in Problems 7-6 and 7-7 have im-
portant consequences with regard to measurements of the total angular
THE WAVE EOUATION IN THREE DIMENSIONS
momentum and its components in a physical system. The fact that the operators
L, Ln, and L, do not commute means that a measurement of any one com-
ponent of the angular momentum introduces an uncertainty of the order of fi
in o.r. knowledge of any other component of angular momentum. Thus, in a
quantum system, a measurement of one component of the angular momentum
essentially nullifies what might have been learned previously about some other
component. In other words, a measurement of L, forces the system into an
eigenstate of Lr. This state cannot be an eigenstate of either Ln ot Lr. However,
the two operators Z, and 12 commute, so these operators can have simultaneous
eigenstates, and the knowledge of these states will be preserved throughout
any sequence of measurements of L* and L2. An attempt to measure L, or L"
will immediately destroy the eigenstate of L, but will not disturb the eigenstate
of L2.
PROBLEM 7.8
Show that the components of the angular momentum in

spherical coordinates are:
L* :;n(si' O
* * cot ocos rk)
/
Lo.: -ifr(cos d &_ cotosin r&)
_:
Lu
._a
-'O a6
PROBLEM 7.9
Using the expressions in Problem 7-B show that L2 -- -h,292,

where 92 is defined in Equation 7 .7.
We see from the result of Problem 7 -9 that the angular part of the
Laplacian operator is identical with the square of the total angular momentum
operatorexceptfor the factor (-h}). Furthermore, if we rewrite Equation 7.10 as
LrY(o, 6) : Lh2Y(0, 6), (7.1+)
it is evident that the eigenfunctions of the spatial operator 92 and the eigen-
functions of the angular momentum operator are the same set of functions.
These will be shown later to be the spherical harmonics. This close relationship
between geometry and momentum is real, and indeed, is summed up by the
statements that the conseruation of linear momentum in tlte absence of external forces
follows fio* the homogeneity of space and the conseruation of angular momentum in tlte
absence of external torques follows from tlte isotropy d space.
THE ANGULAR MOMENTUM OPERATORS 255
In order to illustrate these remarkq consider an infinitesimal displacement

of a particle from position i to (i + di). This may be thought of as a trans-
formation.Ta which transforms ?(i) into 9(i + di); that is,
Torp|) : ve + di) N rt'e) + di.Yrp(f), (7.15)
where the last expression is obtained from the first two terms of the Taylor
expansionof y(i + di) about ,t,e). Ifwe make useof the expression F : -ih.Y,
then we obtain
TyVF) : rp(F) + hur.p,t@: (t + iu'' F)ve) (7.16)
Therefore, the infinitesimal translation operator has the following explicit

dependence on the linear momentum:
Ta: t *1 6i' F. (7.r7)
A finite translation through a distance .i can be regarded as the repeated

application of the infinitesimal translation operator. Thus,
ri :ll: (t +iar.'l'"
: +;\;'
co 1 /; \n
F) (i.lB)
: ,Qlni*6 .
To show formally that translational invariance implies conservation of linear

momentum, let us consider a system for which the total energy is independent
of the choice of the origin of the coordinate system. That is, the Hamiltonian
of the system possesses translational invariance. Mathematically, this implies
that the Hamiltonian commutes with the translation operator, as the following
will show. Since the Hamiltonian operator and the energy are unchanged
by the translation we have
"trryt : EYsn
and
'trTs'Pn : ETsgn' (i. re)
Operating on the first of these equations with Z.r, we obtain
TSffrln : TSETP, : ETSltz, (7.20)
Subtracting Equations 7.19 and 7.20,
trTs-Ts,*-0,
or (7 .21)
l.f , Tsl - 0,
Figure 7-2 The relationship between an

--->
6,
infinitesimal displacement and an infinitesimal
rotation. d6' is perpendicular to the plane of /
and dy'.
which is the required result. Since the operator Ts depends explicitly on the
linear momentu^ F, the latter must also commute with the Hamiltonian and is
thus a conserved quantity.
Now let us consider an infinitesimal rotation through an angle dd by
regarding the displacement di to be given by di x i. Then the counterpart
of Equation 7.15 is:
Rurp(;) : rp(i) + dd x i' Vv(r)

and
fta:1+ddxi.V:1+d6'r*V
i-+
-1+ =66'L.
n
(7 .22)
As before, we can express a finite rotation about tlte same axis as the repeated
application of the infinitesimal rotation operator:
Ro :I," (t
\
* hu|
. tl'"
d--o
: i
'tt'Plf
Therefore, when the Hamiltonian has rotational symmetry about an
axis, the component of angular momentum along that axis commutes with the
Hamiltonian and is conserved. Further, if the Hamiltonian has spherical
symmetry the total angular momentum and its component along any single
axis are conserved.
PROBLEM 7.'O
Using the approach of Equation 7 .19 to 7 .21, show that

ur,g.rlut momentum is conserved when ,tr is invariant under
any rotation.
One important point should be emphasized before closing this discussion,

namely, thai fnite rolafions about dffirent axes do not commu(. .It is now easy to
understand why this'is so, since we know that rotations are intimately connected
EIGENVALUES OF THE ANGULAR MOMENTUM OPERATORS 257
with angular momenta and that the components of the angular momentum
do not commute with one another. Suppose we now consider two infinitesimal
rotations about the x- andlt-axes to see what effect the order of the operations
will have on the result. This is given by the commutator,
I
lRo,, Ruol
h2
lL,, Lul 6{" 66u (7.23)
ry0,
prouided the second-order infinitesimal 66" 66u is negligibly small. To the
extent that the latter is true, we can construct any finite rotation from a large
number of infinitesimal rotations. This difficulty does not occur in the case
of translations along different axes, since the coordinate operators commute
for both finite and infinitesimal translations.
4. EIGENVALUES OF THE ANGULAR MOMENTUM

OPERATORS
As in the case of the harmonic oscillator, the algebraic method for obtaining
che eigenvalues of the angular momentum operators requires the use of raising
and lowering operators which we will denote by the symbols L* and L_,
respectively. They are defined as follows:
L+:L,*iLu: ,'* (* * icot ,&)

\ (7.2+)
L-:Lr-iln: -he-io (*-icot',j,:r))

Let us first ascertain the commutation properties of these new operators. Thus,
L*L_ : (L, + iL,) (L, - iL,) : Li + Li - ilL,, L,f

L_L* : (L, - iL,)(L, + iLo) : L", + Li + ilL,, Lul.
Therefore,
lL*, L-l : L+L- - L-L* : -2ill,, Lrf :2hLu (7 .25)
PROBLEM 7.1I
Verify the following commutation relations:
(u) lL,, L*f : h,L+

l
(b) lL,, L_) : -hL_ I
(7.26)
(.) lL', L*f : lL', L-l : 0)
258 THE WAVE EQUATION IN THREE DIMENSIONS
It is also useful to note that

L*L- i L-L+ :2(L?. + Lil.
This permits us to write the operator for the square of the total angular
momentum in any of the four following forms:
L2:L?+Li+rZ
: t(L*L_ + L_L*) * ,r) (7 .27)
: L+L- - hLu + Lz
: L-L+ + h.L, + L2
The advantages of using L* and L- and the expressions derived above will
become clear shortly.
The problem we have before us is that of solving Equation 7.14, which is
derived from Equation 7.10, for both A and Ir That is, for each allowed value
of A (eigenvalue) there is a function I (eigenfunction) that satisfies Equation
7.14. Our goal is to obtain the whole set of these functions. We also have
another important bit of information at our disposal: the operator Z2 commutes
with the operator for lne component of angular momentum, which we will
elect to be L,. This means that the functions Y arc also eigenfunctions of L,
(see section 2 of Chapter 6). It was stated in section I I of Chapter 31hat the
z component of the orbital angular momentum is an integer times h, say mh.
This will be derived in what follows below. Let us then write:
LzY* __ 1\f7zf nt (7.28)

and
L"Y* -- mhY*, (7.2s)
where we have labeled a particular eigenfunction with its eigenvalue of L"

(without the fi).
Now let us operate on Equation 7.29 with the operator Z*:
L*L,Y* : mh.L+Y* (7.30)
But we may use the commutation relation (Equation 7,26) to rewrite the left
member of Equation 7.30 as:
L*L,Y* : (LrL* - hL*)Y* : L,L*Y* - hL*Y*. (7.3 r )
Equating the right members of Equations 7.30 and7.3l, we obtain:
L,(L*Y *) : (* + l)h(L+Y *). (7.32)
Equation 7.32 states that the quantity L+Y^ may be regarded as a new eigen-
function of Lu having a new eigenvalue (m + l)h. Now let us see if this new
EIGENVAIUES OF THE ANGULAR MOMENTUM OPERATORS 259
function satisfies Equation 7 .28:
L,(L*Y*) : L*(LzYm) : L+(Lhzf nt') : Lb,(L*Y,,,). (7 .33)
Thus, we see that L* is a raising operator. If Y"'is an eigenfunction of Lu and

L2, operation on Y* with I* will generate a new eigenfunction associated with
the same eigenvalu e of Lz but with an eigenvalue of L" which is greater by
one unit of fr.
Operating now on Equation 7 .32 with L* we have
' L*L"(L*Y*) : (m + l)h(LiY*), (7.34)
and from the commutation relation the left member becomes
L*L"(L*Y"') : L,L'*irm - hLz*Y"'. (7.35)
Equating the right members of Equation 7.34 and 7.35, we have
L,(LZ*Y*): (* +2)h(LiY^).
Also,
Lz(L\Y^) : L,*LrYm - fftz(L,"Y^).
Thus, once again we have obtained a new simultaneous eigenfunction of L"
and L2, where the eigenvalue of Lu has been raised another unit of fi but the
eigenvalue of 12 is unchanged. Generalizing these results, one may write
L,(I:*Y*) : (* {r)h(I\Y*)
Lz(L'*Y *) : Lh.' (LnY *).
PROBLEM 7.12
If LrY"' : mhY'o andLzY* : A.hzY.*, show thatL"LIY* :

(m - 3)hL3-Y* and LzL?-Y* : lvhz/;lY*.
We conclude from the above that a whole set of simultaneous eigenfunctions

of L, and Lz can be generated by operating with L* or L-, and it is quite natural
to ask what limitations there are on the number of these functions. Recall
that in the treatment of the harmonic oscillator the raising operator can be
applied any number of times, because there is no limit to the number of higher
states. However, there is a finite limit to the number of lowering operations
in the harmonic oscillator, because eventually the ground state will be reached.
fn fact, it is this limit which allows us to obtain the ground state eigenfunction
(see Equation 5.51). In order to determine the limitations in this present case
of the angular momentum eigenfunctions, we write:
(ti + LI)Y* : (L' - L\)Y* : (A - *z1hzY*

THE WAVE EQUATION IN THREE DIMENSIONS
Since the angular momentum operators L,, L.r, L,, and L2 are Hermitian
operators with real expectation values, the squares of these expectation values
must be positiue numbers. Thus the operator (L2, + Li) must have a positive
eigenvalue and /\ > m2. This means that there is a limit on the number of
times that either the lowering or the raising operator may be applied to a
given function Y*, and thus the set of eigenfunctions must terminate with a
minimum and a maximum eigenvalu e of Lr. For convenience let It be the '
maximum value of m and -lrbe the minimum value . It follows that
L*Y/' : 0
and (7.36)
L-Y -t, : 0
since, by definition, Y/' is the function associated with the top of the ladder
of states and. Y-t, is the bottom function. Letting z be the integer representing
the number of steps betweenl t' and Y-t', we have
n : lt - (- /r) : f, * fr. (7 .37)
Using Equations 7 .27 , 7 .28, 7 .29, and 7.36, we obtain:

Lzyt, : (L_L+ + hL, + L)Y' : (h, f, + h, 4)Y/' : lyhzYt,,
Lry -r, : (L+L_ _ h.L" + L|,)y -t, - (h, l, + h} lily -{' - lyhzy-t'
or
A: ft(ft + I) : /r(/, + 1). (7.38)
This implies either that f, - lr or that l, (1, * l). The latter choice is
rejected, however, since it contradicts the assumption that both /, and /, are
positive (Equation 7.36). From Equations 7.37 and 7.38 we find that:
oft
/:/r -tZ:2 (i.3e)
and
A: t(/+r). (7.+o)
From Equation 7.39 we learn that the maximum value of m is I and the mini-
mum ,uui,r. of m is - f,, where / can be an integer 0r a half integer. For the
classical orbital angular momentum which we are considering here, only
integral values of f-are physically meaningful, so the integer z must be even.
Hoiever, the reader can probably guess that the existence of the half-integral
states for odd z will permit this same formalism to be used for treating half-
integral spin angular momentum. From Equation 7.40 we see that the eigen-
valu-e of the tqrrut. of the total angular momentum is /2 + /, where / is the
maximum obsirvable component of the total angular momentum. Thus, one
can never measure a component of angular momentum equal in magnitude
to the total angular -o*.ntn-. A measurement of, say, the z-component will
THE ANGULAR MOMENTUM EIGENFUNCTIONS 251
nl
(:3
FigureT-3 The allowed projections of the angular momentum for the cases of /: 1,2, and 3.
always yield one ofthe ( + I allowed rn values given by /, I - l, I

2
- 2,
0, -1,. ., -{.
. This is illustrated few f-values
for a Figure in 7-3.
5. THE ANGULAR MOMENTUM EIGENFUNCTIONS

Using the results of the last section, we will now denote the angular
momentum eigenfunctions by Yf , where m, actually mh, is the eigenvalue of
L" and I is called the orbital quantum number. The latter is related to the
eigenvalue of L2 by the expression /(/ + l)h'z. If any one eigenfunction is
known, then the other 2l functions corresponding to the same f value can
be obtained by repeated applications of the raising or lowering operators.
In particular, if the known function is either Yj or YVl, corcesponding to the
top or bottom of the ladder, then repeated applications of the same operator
will generate the whole set of functions.
The clue to obtaining either of these functions is contained in Equation
7.36. Using Equation 7.2+, we may write the following equationfor Yl:
:orr!.(* +i cot o :0,

h)", r, d)
L*YI
(7.+t)
(*. icotA
fi)wro,6):
0.
Since this is a partial differential equation in two variables, we will employ

the same separation procedure as before.
Writing
Yt(o, d) : P(o). o(d),
262 THE WAVE EAUATION IN THREE DIMENSIONS
Equation 7.41 becomes
I 0 ,, icot? A
Pe) ao^ "o)

+
ffi aOa(01 - o'
Ifwe let the new separation constant be k, we obtain the two ordinary differ-
ential equations
f8:ikd6 (7.+2)
dP(0\
(7.43)
m:kcot1d?,
whose solutions may be readily written in the form
o(d) - rttcl
and
P(0) : (sin 0)k,
where the constants of integration have been neglected for the time being.
Then the product functron Y( is:
Yt -- (Cl) (sin o)t'stt'0,

where C( is the constant of integration, which will be determined later from
the normalization of the eigenfunctions.
Use the eigenequations
L'Yt' : /hY'
and
L'Y, -- /(/ + t)h,v,
to show that the separation constant k : /, and that the
product function is
YI : Q/s)$in o)teuo. (7.++)
The eigenfunctions corresponding to the other 2/ m-values can be obtained

by operating on Equation 7.44 with L-. It simplifies the bookkeeping if the
old normalization constant is replaced by the new one each time L* or L-
THE ANGULAR MOMENTUM EIGENFUNCTIONS 263
operates. Thus,
y,,-, :#, L_yt (7.+5)
: -ct-,"" (3*-i cot , gin o)tei/+

&)
: -cr-'ei(t-tto (# . / cot o) {ri' o)'.
But the operator
drd
ffi + f cot g - (ri'ly rt 6in o)t. (7 .+6)
Verify the identity given in Equation 7.46 by operating on a

function of 0 with each member.
Using Equation 7 .46 we may then write:
yrr_, :, _c/r_trtrt_r,, u)rr.

#)_, fi {rin
For the stD application of L_,
(7.+7)
: ( _ L)scr-s eiv--,d
$n (# fil Gr^ u),,.
Now let us make the following substitutions:
cos9: rl
dd 0n.
d0: -srn
Then,
yr_,(q):s/_sriu_s),(+J,-',,, (f,), r,, _ rt )/,

and if we set { - s : m,
yr(q) -crs'lma(*Y'(kl-- u
-,1\'. (7.+B)
Equation 7.48 may be simplified by using the definition of the associated

Legendre functions, namely,2
PT?i: (- \-(*f''(fr)/-m . ?r - L)/. (7.+s)
Then,
YT?i : (- \*Cf PT?i'u*6. (7.50)
For convenience, a few of the associated Legendre functions are tabulated

below. The argument here is 4, and the functions have not been normalized.
It will next be necessary to determine the coefficient Cf in Equation 7.50.
D0
ro I
-
Pi -- 2 cos 0
Dr
al sin 0
-
P8 :4(3 cos2 0 - l)
P\ :4 sin 0 cos 0
r2-
D2 sin2
r, \
-- sinz 0
0
P3 : 2+(5 coss 0 - 3 cos 0)

D1 _
rs_ 6sin0(5cos20-1)
D2-6sinz0cos0
rg
-
D3
tg- - sins 0
(u) Show that the associated Legendre functions given by

Equation 7.49 are orthogonal.
(b) Evaluate the integral
/._ rl _ m)l
(Pr (d I Pr (ri > : Ql f t)',
.
fx a 1 vTW.
:ll':.;; ":;i l:; l;i;:;;; " " "": l.il::1 1:: ";;;;;;;;: ll:: i ;: l; i: i r;:l;;;;;;;;;;n4:: li:"i;i:
5. NORMALIZATION OF THE ANGULAR MOMENTUM

EIGENFUNCTIONS
The normalization procedure which will now be used to determine the

constants C/ is straightforward but tedious. First it will be necessary to
obtain the normalization factor C! and then we will obtain the dependence
2 E. Merzbacher, Quantum Mechanics. John Wiley and Sons, Inc., New York, 196l' p. 179.
ANGULAR MOMENTUM EIGENFUNCTIONS 255
upon m which enters when we operate with I-. Our starting point is the re-
quirement that
Then,
<YilYh - 1.
f
L : lctlrJ,,liri" 0)teit67* . [(sin 0)/ea6f de
fn
:2n'lcll'Jo (sit o)2/sin o do.
Using the substitution "{ : coS 0, this becomes
t :2n lc^,[' e - r),(t + x)t dx. (7.51)
This integral may be evaluated by means of the properties of beta and gamma
functions:
B (m, n) : t*-'{t - )),-, d) : t (m) t (n)

t) l(m f- n)'
for m and n positive. In order to transform our integral to that of the beta
function, we next make the substitution t : +(l + .r). Then we have
1+x:2t
L - x :2(1 _ l)
dx :2 dt,
and Equation 7.51 becomes:
| :2r pir . 22/+t - 4t dt

[r'ur:
: 2n lcrl, .22t+rB(/ + t, / + L)
: I)
2z/+tf.(( l- -l)l(/-'+
2r lcr' ' 1
't TQl+2)
:2r il 1
lcltz.2z/+r L .
Qf +t)t'
since l(/ + l) : ll Therefore
(7.52)
and
yt,::
'' 2/ll\lW4n gin,l)teit6. (7.53)
266 THE WAVE EQUATION IN THREE DIMENSIONS
In order to introduce the m-dependence which arises when the lowering

operator is used, we write Equation 7 .45 in the form
yf : r cf t--r
r vm*L
/ (7.54)
h' C7*t'
The square of the norm of Yf is:
<Yr I Yr> : hlffi,l',r- Yr*' I r-vr*'>

I
h2 ffil' ,rr-' I LnL-Yv+11,
where we use the fact that L* and L- are Hermitian conjugates. Applying the
commutation relations, the ket on the right becomes (see Equation 7 .27):
W+L_Yry*') : l&, + hL, - LZ)Y\*'>

: h:lt(/ + r) + (* + 1) - (m + I),l Vf*'>
:h3(/ -*)(t *m *\Vf*'>.
Then
<yr ^t /
lyr) :l%l'. V - m)(/ + m * r)(Yr*'lvr*'>.
lcr*'l
Using the same procedure, one readily obtains:
<yf*, l ytr*,,
r / : %l'. V,- - m - r)(/
l -/\- + m * 2)(Yf*'
t\ l vf*'>. u
lCy*rt
It follows that
$rlyr> :l#,''. rt - m)(/ - m - t)(t + m * L)(t + -_;;),

lyr*,>.
Ultimately,
grlyr> :Wf V - *)(t - m _ 1)''' (1)( / + m+ 1) (/ + m -t2)

x " ' (2/)(YtlYh.
Thus'
.,
\ v,r YT\ :lq
uy l'
,;l'(t-m)t'ffi<Y',lYr>,
ANGULAR MOMENTUM EIGENFUNCTIONS 267
and setting both norms equal to unity, we obtain
CT :,,J (2/)t(/ I IWI l(r +ml

(/ a m)l :W.ttt'16I' (7'55)
m)t -
where Equation 7.52 has been used in the last expression. The value of this
coefficient may now be put into our earlier expression (Equation 7.+7) to
obtain:
Yr :(ir-;ffi m)l
"ol '
- m)l
rl-mvlr
l-- tt (7.s6)
--r
YT?r 6) (7.57)
ffiPT?t),'*d,
and
Y7?1,6) : #; - t)'
(7.58)
where rl : cos 0.
The above results were obtained for m positive. In order to obtain the
functions for negative m, it is most convenient to use the relations
P|*?i : PT?i (7.5e)
and
Y7* (ri : (
- 1)- ( YT)* ?i . (7.60)
Of course, these expressions could have been obtained by following the pro-
cedure above, using Y;t as the starting point and operating with Z*. In this
case the counterparts of Equations 7.54 and 7.56 are:
Yr :iffiL*YT-" (7.61)
and
YT : (i)'.- J (2t)t(/
(/ - m)t Lf*Yr'. (7 .62)
m)l |
The angular momentum eigenfunctions, as given in Equations 7.56 and
7.58, are known as the spherical ltarmonics. They comprise a complete, ortho-
normal set of functions on the unit sphere. A few of the normalized spherical
268 THE WAV.E EOUATION IN THREE DIMENSIONS
harmonics are tabulated below because of their frequent use:
vo- I
ro
- 2\/;
n:1/1
' 2rl.r .oro
r,*':Tl zrrl/Trin o'e+io
2n
vE ::^lt. (3 cos2 o - I)
^v?7
v+r : -- I
12 - r 2rl/E
z,
. sin 0 .cos 0 . erio
- :! +'Jit5
Y"L'
2n
. sin2 o . e+zio
4:l_
,3
+l;17'ftcos'o-3coso)
Y*':+l
rs -
-f ; I.lT sin0.(5cos2
dV
0_ l).e+to
rr
rs :! /to5.sinz o.cos o. el;zio

Y+z _
4rl*
UV 7t
m 7-15 to prove the orthonormality

Use the results of Proble
of the Y{ as expressed by Equation 7.57.
7. THE ANGULAR MOMENTUM MATRICES

Since the spherical harmonics are simultaneous eigenfunctions of the
operators L2 and L", they constitute a basis for a matrix representation of
these operators in diagonal form. The matrix elements are given by:
(YF l L' Vr> : h'/(l + t) 6,,,, 6^,,*

and
<YF'l L, Vf ) : mh 6r,t 6*,,*.
THE ANGULAR MOMENTUM MATRICES 269
Since the ofldiagonal elements are characterized by either d' a { or m' t' m
(or both), the Kronecker deltas ensure that all off-diagonal elements are zero.
If we order the basisfunctionsasfollows, I?, Yl,Yl, Yrt, Y3, Yl, Y3,..., then
we can write the matrix of L2 as:
io
i1
i9
t1
lo
t/
t-
(L'.) : h' i6 i
(7.63)
t2
where all of the off-diagonal elements are zero. The boxes correspond to the
2l + 1 functions associated with the same eigenvalue of L2. That is, the one-
by-one submatrix corresponds to / : 0, the three-by-three to { : l, the
five-by-five to / : 2, and so forth. In like manner, the matrix representation
for the operator L, is
_1
i
: I
(L,) h (t.a+1
-1 cri
j
-1
Although the matrices shown in Equations 7.63 and 7.64 can be of infinite di-
mension, each set of 2/ + I functions associated with a given angular mo-
mentum forms a subspace of dimension 2/ + t. Furthermore, tlte 2/ + I
spherical ltarmonics associated with that f-ualue constitute a complete, orthonormal basis
in that subspace.
The operators L*, L-, L,, and Lo can also be represented by matrices,
although their representations are more difficult to obtain. One approach is
to begin with Equations 7.54 and 7.55 and to write:
L_Yr : h{v + m)(f - m + D Y -t (7.6s)

Then, operating on Equation 7.65 with L*, we obtain:
L+L_Y\:hf@ L,YT_1
TO
The left member may be simplified by using Equation 7.27.
(L' + h.L" - L)Yr : hf@L*yT-'

_1 : hll(l + 1) + m - m'f
L,Yi,
T
rM Yr
:h'\/@Yr
or
L+YT:hf@Yry+r (7.66)
Equations 7.65 and 7.66 may be combined in the form
L|YT:h.f6Yr*t. (7 .67 )
Now it is possible to find the matrix elements of L*. Thus:
<Yr' lL*YT) : hf@ 6t',, 6*',nt!\t (7'68)
because of the orthonormality of the spherical harmonics. Hence, the orthog-

onality property requires that all matrix elements of (Zn) be zero except for
those along one diagonal adjacent to the principal diagonal. By way of
illustration, the matrices for the case of { : 2 are shown below:
(L*):hfi
(L*)-n\: l': ',l l\
and
:3:U'
lo o o o
l, o o o
(L-\:hl o \/6 o o
\o o fG
\o o o
o
2 il
where all other elements are zero. Then, from the definitions of Equation
THE ANGULAR MOMENTUM MATRICES 27t
7.24, for f: 2 we see that

0 20 00
2 o f6 00
(L,) : ,r(Lu) + rr(L-) :* 0 {6 o l,/o o
0 u^/=y6 02
0 00 20
and
0-2 00
20 -t/a o
(t,) :42 o f6 o -f6

00 \/G o
00 02
PROBLEM 7-17 ']+,.ril', .1 .+:t:' i.;.i;,,'.,i11,:,'1,'.;:::1,,,:,ii::..:."i:
Derive Equation 7.65.
PROBLEM 7.18
(u) Write the matrix representations of L*, Lo, and L, for

0/a J.
-
(b) Using matrix algebra, verify that the matrices obtained
in (a) satisfy the equation : (L') : (Li) + Q?) + (L).
PROBLEM T.'9
(u) Operate on Yl with L- to obtain Yl and Yl.

(b) Operate on Y;2 with L* to obtain Y;r and Yl. Hint:
Make use of Equations 7.24 and 7.67.
PROBLEM 7.20
(u) Find the matrix (t/) that will diagonalize (L,). (Do
this for the 3 x 3 matrix).
(b) What does this transformation do to (Ir) and (L,)?
(.) Show that the transformed matrices still satisfy
(L,)' + (L,)' + (L,)' :2h'(l). Physically, this is
equivalent to changing the axis of quantization from
the z-axis to the x-axis by, for example, rotating a
magnetic field from the z-direction to the x-direction.
272 THE WAYE EEUATION IN THREE DIMENSIONS
8. HYDROGENIC ATOMS ,
As in section 4 of Chapter 3, we will adopt the model of an atom consisting

of a fixed nucleus of charge Z and a single electron of reduced mass p moving
in a spherically symmetric Coulomb potential. However, we have refined
our earlier model in two important ways. First, the electron is no longer
viewed as a hard sphere of charge -e whose precise position can be specified
in space and time. Instead, we describe it by means of a wave function
Y(r, 0, 6, t), whose square is a measure of the probability that the electron can
be found at position (r, 0, 4) ut time /. In the second place, we now understand
the intimate relationship between coordinate space and angular momentum.
Therefore, instead of artificially imposing Bohr's quantum condition on circular
orbits, we now know that the allowed wave functions for the electron must be
eigenfunctions of the square of the total angular momentum operator.
Our task is to solve the time-independent Schrcidinger equation for the
allowed energies and wave functions of the electron. Thus, we begin by
setting V(r): -(Ze,lr) in Equation 7.6. Since the Coulomb potential is
spherically symmetric we can immediately write:
Vr,*(r,0, +) : ft(t) 'YT(0, +),
where the wave function has been labelled with the angular momentum
quantum numbers corresponding to those of the angular function. We already
know that the angular functions are the spherical harmonics given by Equation
7.57. It now remains to determine the radial functions. Rewriting Equation
7.9, we obtain:
/42 2pf(t(/+L)h' Zez

:
(z* + ?*)^0, h2 L\ 2pr' ) - "]^r,) o. (7.6e)
Before proceeding with the solution of Equation 7 .69, it should be pointed
out that the quantity in parentheses in the second term is often called the
ffictiue potential seen by the electron. A non-zero angular momentum results
in a so-called "centrifugal barrier," which must be overcome as the electron
approaches the origin. Hence the effective potential is the algebraic sum of the
Coulomb potential and the so-called "centrifugal potential." (See Figurc 7'4.)
Equation 7.69 can be simplified by means of the substitution
R(r) : +,
which, upon differentiation with respect to r, has the derivatives:
g'(r) _ g(r)
R'(r) --
r 12
HYDROGENIC ATOMS 273
:-, I
=l-
+lE
IL
\
'-,
lc.l
\l I
b0
li
Figure 7-4 (a) The centrifugal Potential for

several values of L (b) The effective hydrogenic Poten'
tial due to both the Coulomb and centrifugal terms.
O
I
Coulomb potential
and
R"(r)--+-ryP+ry
s,,(r) -#1ry-+ -lsot -o (7 .70)
Since this cannot be solved closed form, we must employ either the
in
power series methodl or an analytical approach which utilizes a knowledge
fttn. asymptotiCs6lutions in the limits of very'small and very large r. Choosing
s See L. Pauling and E. B. Wilson, Introduction to Quantum Mechanics. McGraw-Hill Book Co., Inc',
New York, 1935, p. 12l.
the latter approach, we note that for very large r, Equati on 7 .7 0 may be approxi-
mated by
s"(r) +ff s?) - o,
whose solutions are

g(r) ^'.*o ( +lht-t E )
Now, for bound solutions the total energy .E must be negative. Therefore, we
write
-'-
-h ,r'-z,tEl:+ilWt-*A,
and then
g(r) '.-' e*7''
Since the plus sign leads to a function which diverges for large r, we cannot
use it for a physical system. Hence our only acceptable solution is
g(r) '.-.' e-^'' (7.71)
The oscillatory solutions which result from positive total energies (which were
excluded above) correspond, physically, to scattering states in the energy
continuum for free particles. Each continuum state is infinitely degenerate
since it can be assignid un infinite set of {-values. A detailed discussion of these
states will be deferred until the section on the quantum mechanical treatment
of scattering.
In the li*it of very small r, the dominant term in the bracket of Equation
7.70 is the llr2 term. Thus, in the asymptotic limit for small r, the solutions of
Equation 7.70 will be dominated by the solutions of
s,,(r) _ry.e(r) :0.
Trying a solution of the form

g(r) '-'r'
leads to the relation:
p(p - 1) : /(/ +t).
Thelatterissatisfiedonlyforp:-lotp:/+l'whichcorrespondtothe
asymptotic solutions (for small r), r-t or rt*t. Since the first of these diverges
at th; origin, whereas the second is well-behaved, we choose for small r the
solution
g(r) ,-.'' Yt+r. (7.72)
Our present knowledge of the final solution is shown schematically in Figure

7-5. We will now seek a polynomial solution for the intermediate region.
It is convenient at this point to change to a dimensionless variable by
Figure 7-5 Schematic representation

of acceptable radial wave functions in the
regions where r --* 0 and r -+ oo.
means of the substitution

P : 2rr'
Then, g"(r) : +1'g" (p), and Equation 7.70becomes:
s"(p) -W;!-i+l]r,p) :o (7.7s)
Here z is an unspecified number defined as
pZe' z (7.7+)
'o - h2)" aoA
We construct a product solution from our asymptotic solutions and a power

series in p:
g(p) : e-ptz . pt,+t . pi : e-ptz . p/+r+i.
Ao, Ao,
Differentiating and substituting into Equation 7.73, we obtain the following
recursion relation on the coefficients a,:
att+t : alt
l+k*L-n
(k+L)(2/+k+2)'
In order to obtain a satisfactory wave function, the infinite power series must
be truncated to a polynomial so that it can be normalized. If the polynomial
is to terminate with the kth term, it is necessary that ar*, - 0. That is, we re-
quire that
n : / + k + l,
wherek:0,112,... Hence, z takes on integral values given by / * l,
/+2, /+3,.... It is customary to callntheprincipal or total quantum
number; it is allowed to take on integral values but not zero. For a given value
of n, the angular momentum quantum number / can have any one of the set
of values 0r Ir2, n - l.
The polynomial solutions of Equation 7.73 may be obtained by means of
series solutions in a' manner analogous to the treatment of the harmonic
oscillator in Chapter 5. .For a detailed treatment the reader is referred to one
of the references listed at the end of this chapter. The polynomials are members
of a class of well-known functions called the associated Laguerre polynomialsra
L?:i;( p) : (#rl'" lr (kl.' 1,**',-o1f (7 .7 5)
To obtain the complete radial solution for the hydrogenic atom we must
retrace our steps through all of the substitutions and obtain
R*,(p):9 : s-Ptl P'LYI/(P),

P
where p :2),r : (2Zlnao)r. Normalization of the radial functions over P

may be achieved by evaluating the integral
Io-'o
P"lLYf;(p)l' p' d P :ffffi
Normalization over r is obtained as follows:
rco 'zrz
dr : (fJI*t R*,(p)1, p, dp : W I,-l*^,(p)f,p, dp.
J. lR",(r)l
Then the normalized radial wave functions are given by:
R*,(,) : -.2(# Jffi (#Y

"",tnaoLzt+'\
(#,) ( 7 76)
Figure 7-6 shows the general behavior of some of the radial functions. A few
of the normalized radial functions are given below for convenience.
ft,'(,) : (ff.2 exp (- t;)

R,o(,) : (Af (z -
n) (-#,)
*o
ft,,(,) :(ef
+ 7.*o (-#)
R,o(,) :61 ,[' -? fl*+ffi] exp (-#,)
ft,,(r) :(#,r + ';(, -L? exp
?a)
R,,(,) :(#: m (#.*o(- ?)
a See, for example, M. R. Spiegel, Mathemntical Handbook of Formulas and Tables. McGraw-Hill
Book Co., New York, 1968, p. 155.
Zrlas
Zrl as
l5-
Zr/ ao
Figure 7-6 Graphs of the radial wave functions R,,t|l for n : 1,2, and 3 and /:0,1,2.
(From Principlesof Modern Physics by R. B. Leighton. Copyright 1959 by McGraw-Hill BookCompany.
Used with permission of McGraw-Hill Book Company.)
The radial solutions obtained above should be recognized as the energy

eigenfunctions for the hydrogenic atom. This result could have been anticipated
from Equations 7.9 and 7.10, where the energy is seen to appear only in the
radial equation. The energy eigenvalues corresponding to the radial solutions
are readily obtained from Equation 7.74. Thus:
1_z
1:;{2plE,l: f,oo,
of,
n Z2e2 woZ'
un - 2a;on'- n2
The allowed energies are precisely the same as those obtained for circular
Bohr orbits and for Sommerfeld's non-relativistic elliptical orbits. Such good
agreement between classical and quantum mechanical calculations is fortuitous
in the case of the Coulomb potential, and should not be expected in general.
Figure 7-7 Photographic representation of the electron probability density distribution

?*g for-several energy uigenrt"tes. These may be regarded as sectional views of the distributions
inaplanecontaininglhepolaraxis,whichisverticalandintheplaneofthepaper. Thescalevaries
from figure to figuie. (From Principles of Modern Physics by R. B. Leighton. Copyright 1959 by
McGraw-Hill Book company, Used with permission of McGraw-Hill Book company.)
Since the energy levels are independent of the quantum numbers / and m,
we can express the degeneracy of each level by the sum
n-L
>(2t+ 1) : I +3 +5 + "' + 2n -r :n2-
f:o
The (21 + I )-fold degeneracy of the m-values is characteristic of all spherically
symmetric potentials, while the z-fold degeneracy of the f-values occurs only
for a poteniial such as the Coulomb potential, which satisfies a specific invari-
ance iondition.s Historical usage has coined the unfortunate phrase "acci-
dental degeneracy" to describe the latter. The reader can anticipate some of
the results of later sections by speculating that any alteration of the physical
system which destroys the spherical symmetry or the pure Coulombic potential
will lead to the removal of some or all of the degeneracies'
5 The n-fold degeneracy will occur for any potential that is invariant under the group O(4). See
L. I. Schiff, Quantum Mechanics, 3rd. ed. McGraw-Hill Book Co., New York, 1969, p.237.
Figure 7-7 Continued.
The total hydrogenic wave functions can now be written as

Vu*(r, 0, 6, t) : Vnr*(r, 0, +) : R,r(r)yle1, $)e-iu"ttn,
. e-iuntth
where n canbe any integer, I can take on the values 0rlr2r...r(n _i),
?,nd
*,:1n have any of the values /, I 1,...,0,
- given .'..',' f. 'ilthough
-1,
the radial and angular functions have been above, a few of the norm ali'i"d
spatial, hydrogenic wave functions are tabulated below to facilitate their use
in the problems:
Vrc- : ''-zrtas
Vlr,o
*Gf
Vzt - l?zoo :#c€)r (z 7).,-zr,za.
I /Z\z Zr
Vze o
a {Zn\aol ao
1Pze
I /Z\z';Zr ' sin o'e=io
m(;/ '-zrt,oo
Vst: lpsao
280 THE WAVE EAUATION IN THREE DIMENSIONS
?,-"''"'cos t;)
ege
rJ(, -
--?sro: *J; o
vsp :?arrr --
t;)
#rffi(t - 7o''-"'''oo
sin .'e+ao
ew: llszo:
#(tr # '
'-zrt,ao'
(3 cos2 0
- 1)
vsa : vsztr : sin 0' cos 0' etno

#(tr #',-zrtBas'
rpsa: lpsztz:
*U*ffi ' r-zrtBao'sin2 0'e*zto
T
The states associated with wave functions for which the angular momentum
iszero (that is, l:0) arecalleds-states. Notethatalls-statesarespherically
symmetric. States associated with / : 1,21 3, +, . .. , are called f', d'r-f , g-r. . .
states, respectively, and their degeneracies are 3-, 5-, 7',9-r... fold. This
terminology had its origin in optical spectroscopy where certain sets of spectral
lines were characterized as sharp, frincipal, diffuse, and;fine. The ground
state of hydrogen is the gr, state .
For a hydrogenic system, the probability that the single electron will be
found in the volume element dt is
lrltu*l'dr : lrpnr*l'r'dr dQ.
Pictorial representations of a few of these probability densities are shown in

Figure 7-7. The probabitity of finding the electron in a spherical shell of radius
r and thickness dr is obtained by integrating the above expression over solid
angle; thus,
dr d.{l : lR,r(r)t'r'dr!* Vf (O)|'z dO : r2 lRuQ)1z dr. (7.77)

J: lrp**12r2
The quantity r2 lR,7Q)12 is called the radial probabilit2.densit2 or the radial
distribution function. Figure 7-B shows sketches of this quantity for a few
hydrogenic
' states.
It is often convenient to express the angular parts of the hydrogenic
functions in rectangular coordinatei. By taking appropriate linear combinations
of a set of degen.it. functions, it is possible to obtain a new set of functions
having easily--visualized symmetries with respect to the rectangular axes. F9t
.*u*i1., thi three degenerate p-functions for any value of n may be written in
the form, '
IpnL,o : -f(r)' r cos 0 : f(r)' z
and
{2 y,*r,*, : -f(r)' r sin 0' erif : "f(r)(x I d.
HYDROGENIC ATOMS 28t
.,a.2s
tt""..
I ,'- -- -j-"''.
\
[4,. ..iri=-_...,_._ -l=.
...-....3P
\.,
.rt
,/
i \..
'tr
\
-t'.
s.rE6A212tt?t6$
cE6A212tr?!6
Figure 7-8 The radial probability distribution function lrRntlz for several values of the
quantum numbers n, /. (From E. U. Condon and G. H. Shortley, The Theory of Atomic Spectro,
Cambridge University Press, Cambridge, 1953. Used with permission.)
Then we may define:
I
IPr, -- f (r) ' r t/z
(rpn , * ,ltnt,-t)
I
Voo : -f (r) 'l (rlrnr, - Vnt,-r)
i\/,
vro:-f (r)' z : gnto- (7.78)
THE WAVE EGIUATION IN THREE DIMENSIONS
Each of these wave functions resembles a figure B having rotational symmetry
about one of the coordinate axes. Such wave functions have applications in
crystal lattices and certain molecular structures when rectangular symmetry
exists.
PROBLEM 7.21
Find the appropriate linear combinations of the d-functions

which will lead to the following d-orbitals:
d*y, dr", d'rp drr-or1 ds2z-,72'
PROBLEM 7-22
Calculate the expectation value (r) for the g16s, the rprro
and the Vra20 states of hydrogen. Compare with the Bohr
theory results.
PROBLEM 7.23
Find the values of r for which the radial probability density

is a maximum for each of the states !)no, Vzto, and ?azo of
hydrogen. Compare with the preceding problem and the
Bohr theory result.
(Ans. i ao, 4aor 9ao.)
PROBLEM 7.21
(u) For the state !)zto of hydrogen calculate the expectation

values (llr) and (pz).
(b) From the results of part (a), find ( Z) and (V ) and
show that (T) : -+(V >. Compare with the Bohr
theory. (This is an example of the virial theorem.)
PROBLEM 7.25
Show that gros and gro6 are orthogonal.
9. MAGNETIC MOMENTS OF HYDROGENIC

ELECTRONS
It is instructive to examine the probability density currents associated with

electronic motion in hydrogenic states. By way of illustration, let us consider
the 2s and 2p states of hydrogen. From Equation 4.47 we have for the prob-
ability density current,
,s--- h
mo
Imag. [E(Vp*)],
MAGNETIC MOMENTS OF HYDROGENIC ELECTRONS 283 }
where
Y(r,o,s):*,*+fia +# fia
Then, for the 2s state of hydrogen, ?(VV*) is real, and ,i : 0. It should be
evident that this will be true for all s-states. For the 2p states, however, we
write:
lpzp : Rrt(r)Yf Q, 6),
and
Vt, : Rrr(')Yt*(0, O)'
Then,
,pro(Ygto) : Rlrr(r) ' Rr, (r) lYil'z i
+ | rn,,t r)pYr fi rtne
+#tR,,(,) rYrf,rY,*)$.
Since the radial component has no imaginary part, it can make no contribution
to the current, as in the case of s-states. Let us now examine the d term. Its
angular factors may be written as the product YT : Pr(O)ei*o. Then,
rrm
YT
A
: Pt(0) 'P',t(O),
a0Yr*
which is also real. Therefore, the 6 term does not contribute to a current. It
remains to examine the fi te.-, which may be written in the form:
-tm
m-o ' lR,'(')l' ' lP'(0)l'.
Since this is an imaginary term it does contribute to a probability current
density in thef'ait.ciion, given by
a
Dd:
hm lRrr(r)Pr(0)l' a
*06-9'
This can be converted to an electrical current density by multiplying by the

charge of the electron, that is,
io : _tSo'
Now, the magnetic moment due to this current density is
dilt : +(i x j) a"

THE WAVE EEUATION IN THREE DIMENSIONS
of,
dtr[-" :4lR"(r.)Pt?)1'z
2moc srn 0
6
lRrr(r)P'(0)1'?
- -mltn d,,
sin 0
and
l@
M, : -mtta I lRr'(t) Pr(o)12 4m2 dr. (7.7s)
Jo
PROBLEM 7.26 +c.sl+l[";i'triiEri.i
(u) From Equation 7.79 show that the magnetic moments

associated with the states gzr+r are f pB, resPectively.
(b) What are the magnetic moments for the 3/ states ?
PROBLEM 7.27 ".illl1+rt.jii$i:$5
The magnetic moment operator may be written in the com-

pact form,
P - -FaL'
Show that this operator provides a direct means of obtaining

magnetic moments from the angular momentum matrices.
;,qrt1)1'lrii,;:4ii;:i
IO. THE PARITY OPERATOR
In Chapter 5 we noted, for both the square well and the harmonic
oscillator, that each eigenfunction can be either symmetric or antisymmetric in
the coordinates. In quantum mechanics the term parity is used to characterize
the symmetry properties of wave functions with respect to inversion (that is,
reflection through the origin), which is equivalent to changing the sign of each
coordinate. Thus, euen and odd parity refer to the symmetric and antisymmetric
cases, respectively.
It is convenient to define a parity operator II which changes the signs of
all coordinates. For example,
Iltp(x,y, z) : V(-x, -), -z).

Then,
fly(-r, -), -z) : hP(x,2t, z).
il'rp(r,), z)' :
Since the eigenvalue of II2 is f 1, the eigenvalues of the parity operator, for
states which have definite parity, are * l, where f I corresponds to even parity
THE PARITY OPERATOR
and - I corresponds to odd parity. This leads to the equivalent statements:
:
plus parity even : symmetry under inversion
parity
minus parity : odd parity : antisymmetry under inversion
Let us now examine the parity of the hydrogenic wave functions. The
parity operator transforms the variables as follows:
r--> r, 0 --> n - 0, and {- Q ! n.
Then, from Equation 7.+9,
ttPT(r) : (-l)'-*Pf ?i,

which results from converting the derivatives dld(cos 0) to dldlcos (- 0)]. Also,
u (eo*01 - rimbrimo : ( _ l)*si'm$.
Using the above and Equation 7 .57, we obtain:
ilYf (rt, O) : ( - t)'Y T ?6 {) .
It follows that
frrpnr*:fIRntYT : RntfrYT : (-L)'rl,nr*,
which says that the parity of the hydrogenic functions is the same as the parity
of the spherical harmonics, namely, (-l)'. Using the same terminology, the
parity of a harmonic oscillator function it ( - I ) " and the parity of a square well
function is (-l)"+t.
When states of different parity have the same energy, the degeneracy is
called accidental degeneracy. This occurs for the different l-values in a hydro-
genic potential when all non-Coulombic interactions are neglected. * If these
interactions are included in the Hamiltonian, the f-degeneracy is removed and
all of the hydrogenic states become states of definite parity.
One might ask whether the parity of a state is a constant of the motion.
To determine the answer we examine the commutator,
Itr,nl :,fff1 - n"/f :,ff(x,), z) - lf (-x, -), -z).

This commutator is zero whenever the Hamiltonian is even in the coordinates.
Therefore, from Equation 6.18, we conclude that tfte paritlt of a state is constant
in time if the Hamiltonian is euen in each coordinate. There is, of course, the important
example of the weak nuclear interactions in which parity is not conserved. The
* See footnote 5.
286 THE WAYE EEUATION IN THREE DIMENSIONS
Nobel prize was awarded to Lee and Yang in 1957 for their theoretical pre-
diction of this phenomenon.o
Frequent use of the concept of parity is made in calculating matrix
elements between states having definite parity. Since an integral of an odd
integrand over symmetric limits is zero, one immediately knows that all matrix
elements of an odd operator (antisymmetric in the coordinates) between states
of the same parity are zero. Likewise, matrix elements of a symmetric operator
between states of opposite parity are zero. This will be very useful in later
discussions of the selection rules governing transitions under the influence of
perturbations. Since the parity of an isolated system is conserved for all but
the weak nuclear interactions, it is often useful for obtaining information about
a state where the exact wave function is unknown.T
SU|'|/YIARY
The two-body problem in three dimensions is shown to be separable in

the coordinates of the center of mass of the system and in the relative coordinates
which give the position of one body with respect to the other. The system
can be then treated as a one-body problem by using the reduced mass in the
relative coordinate system. For a spherically symmetric potential, the Schro-
dinger equation can be separated into a radial and an angular equation. The
differential operator of the angular equation (that is, the angular part of the
Laplacian operator), is shown to be the square of the total angular momentum
operator. Hence, the angular momentum eigenfunctions (the spherical
harmonics) are the solutions of the angular part of the Schrodinger equation
when space is spherically symmetric (isotropic). In other words, when the
Hamiltonian is invariant under a rotation, it follows that: space is isotropic,
angular momentum is conserved, the spherical harmonics are the eigenfunctions,
and I and ffi' ?te good quantum numbers. This close connection between
space and momentum is also apparent in translational motion. For if the
Hamiltonian is invariant under a translation, space is homogeneous, linear
momentum is conserved, the momentum wave functions are the eigenfunctions,
and p is a good quantum number.
As in the case of the harmonic oscillator, there are raising and lowering
operators for the angular momentum. Using these operators, as well as the
operators for the square of the total angular momentum and its z-component,
simultaneous eigenfunctions of Zz and L, are obtained by means of operator
algebra. These functions, which are the spherical harmonics, are the angular
solutions of all physical problems having spherical symmetry.
The solutions of the radial equation for a hydrogenic atom are shown to
be the associated Laguerre polynomials. Hence, the total wave function for a
6 T. D. Lee and C. N. Yang, Ph1ts. Rea. 1041 254 (1956) ; C. S. Wu, E. Ambler, R. W. Hayward,
D. D. Hoppe, and R. P. Hudson, Ph2s. Reu.105, l4l3 (1957); R. L. Garwin, L. M. Lederman, and
M. Weinrich, Ph2s. Reu.lO5, 1415 (1957).
for example, J. M. Blatt and V. F. Weisskopf Theoretical Nuclear Ph2sics, John Wiley and
? See,
Sons, New York, 19S2, or R. D. Evans, The Atomic Nucleus, McGraw-HillBookCo., Inc.,NewYork,
1955.
SUMMARY 287
hydrogenic electron (neglecting spin) may be written as the product of a

Laguerre polynomial, a spherical harmonic, and a time factor depending on
the energy of the state. Each wave function is characterized by three spatial
quantum numbers, z, /, and m. The principal quantum number z takes on
integral values (not zero) and denotes the "orbit" or "shell" in which the
electron is moving. The angular momentum quantum number I can have
any one of the values 0,Ir2, fl - l, for a given value of n. The states
associated with the different f-values are sometimes called "sub-shells," and
their wave functions are frequently referred to as orbitals. The spectroscopic
designations of these orbital states are .r, P, d, -f, gr. . . , respectively, for I :
0, 1,2,3,+, When solving molecular or solid state problems it is often
convenient to form linear combinations of the spherical harmonics so as to
remove all of the angular variables. They may then be written in terms of r
and the rectangular coordinates. The magnetic quantum number m (or m)
can have 2/ + I values for each value of f, ranging through /, / - 1,. . .,0,
1 t
- Lt. . ., '. Each value of ma prescribes an allowed orientation for an
orbital plane, since mt represents the projection on the quantization axis of
the normal to the plane. The name "magnetic" quantum number derives
from the fact that each orbital state of a charged particle possesses an effective
magnetic moment proportional to its z-value. Thus, for an electronic state
in a constant magnetic field, the component of the orbital magnetic moment
parallel to the field it pr: ffitl-tB, where pais the Bohr magneton. The torque
produced by the magnetic field causes the orbital plane to precess about the
field direction at a fixed polar angle. The energy of each state is altered by an
amount given by *mtpn#. The removal of the orbital degeneracy in this
fashion is what gives rise to the normal Zeeman effect.
The absolute square of a hydrogenic wave function integrated over a solid
angle is called the radial probability density of the electron. For the case of
hydrogen, the radii corresponding to the maxima of the radial probability
density function are in good agreement with the values given by the Bohr
theory.
The parity of a hydrogenic state is given by (-I)', where * I are the
eigenfunctions of the parity operator. Since the parity operation is a reflection
through the origin (inversion), the eigenvalue fl corresponds to an even or
symmetric function of the coordinates, whereas the eigenvalue - I corresponds
to an odd or antisymmetric function. The terms "plus" and "minus" are
often used for even and odd, respectively. The efficacy of the concept of parity
derives from the fact that parity is conserved when the parity operator commutes
with the Hamiltonian, that is, when the latter is symmetric in the coordinates.
CHAPTER 7 APPENDIX A
Average Valuess of Powers of r for a Hydrogenic Atom in the State !)ntm.
: t(/ + r)\1
(r)
ryj * i{' _
nz Jl
I
(r') :
%l' * 1{' _r(r+r)-l\l
n2 l) -)
11\ Z
,h"'
11\
\,'/
/1\
m 22
z3
\d/ aflnsl(t++)(/+r)
,_1, _ gr, I
ur^h-t(t+L)\
11\ onr'(/+ +)(/ '+ r)(t + +)t(f
\rn / - +)
SUGGESTED
REFERENCES
David Bohm, Qtnntum Theorlt, Prentice-Hall, Inc., New York, 1951.
Sidney Borowitz, Fundamentals of Quantum Mechanics. W. A. Benjamin, Inc.,
'_ New York, 1967.
Wesley'Publishing Co., Inc., Reading, Mass., 1960.
P. T. Matthews, Introduction to Quantum Mechanics. McGraw-Hill Book Co.,
New York, 1968.
E. Merzbacher, Qtantum Mechanics. John Wiley and Sons, Inc., New York,
1961.
Albert Messiah, Qwntum Mechanics. North-Holland Publishing Co., Amster-
dam, 1958.
D. Park, Introduction to the Qrcntum Theory. McGraw-Hill Book Co., Inc., New
York, 1964.
L. Pauling and E. B. Wilson, Introduction to Qtnntum Mechanics. McGraw-Hill
,/ Book Co., Inc., New York, 1935.
Powell and B. Crasemann, Quantum Mechanics. Addison-Wesley Publish-
t,/J.t-. ing Co., Inc., Reading, Mass., 1961.
Ph2sics,6th ed. McGraw-Hill Book Co., New York, 1969.
D. S. Saxon, Elementary Quantum Mechanics. Holden-Day, fnc., San Francisco,
1968.
Leonard I. Schifl Qtnntum Mechanics, Srd ed. McGraw-Hill Book Co., New
York, 1969.
Robert L. White, Basic Quantunt Mechanics. McGraw-Hill Book Co., New
York, 1966.
Klaus Ziock, Basic Quantum Mechanics. John Wiley & Sons, Inc., New York,
1969.
8 L. Pauling and E. B. Wilson, op. cit., section 2lc.

288
CHAPTER 8
$ptsf;s dhnmilTtsffi 0F
JTHffi T,S ilAK MSil?T Hhf, TAu
A3{m tmHs*Tr*AL
PEKYffiffi&H$
We have freely alluded to the fact that the elementary particles possess an
intrinsic angular momentum which we caLI spin. Thus, we noted in Chapter 3
that a magnetic moment is associated with the spin of a nucleon, as well as
that of an electron, and that this moment can be accurately measured by beam
deflection or magnetic resonance experiments. It was from the number of
components in the beam experiments that the halflintegral value of $h for
the spin of the electron became apparent (see section ll of Chapter 3). The
existence of an electronic spin angular momentum of $h was postulated by
Uhlenbeck and Goudsmit in l92B in order to explain the fine structure splitting
of hydrogen.l It was soon observed that this assumption also explained the
"anomalous" Zeeman effect, the doublet structure of the spectra of the alkali
metals, and the anomalous magnetic moment reported by Compton,2 provided
that one also assumed the "anomalous" gyromagnetic ratio of roughly twice
the classical value.s
The Schrcidinger formulation of quantum mechanics is built, as we have
seen, on the classical Hamiltonian function. Thus, the Schrcidinger equation
does not include spin as an intrinsic variable, since spin is a relativistic phenom-
enon having no classical counterpart. For example, there are no additional
generalized coordinates that will permit us to write the spin in the classical
form,
î x F,.:i
This conceptual disadvantage turns out to be an advantage, however, since the
uniqueness of the spin permits Sz to commute with all other dynamical vari-
ables. Hence, the spin is a constant of the motion and s is a good quantum
1 G. E. Uhlenbeck and S. A. Goudsmit, Naturwiss. 13' 593 (1925).

I A. H. Compton, J. Franklin Inst.l92r l44 (1921).
8 W. Pauli, Z. Ph2sik 43,601 (1927).
289
290 SPIN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
number. (S, may or may not be a constant of the motion, depending upon the
nature of the spin-dependent interactions, if any, that are included in the
Hamiltonian.)
Although it is an ad hoc assumption in the Schrodinger treatment of quan-
tum mechanics, the half-integral spin of the electron arises quite naturally out
of the relativistic theory of Dirac.a Thus, its validity as an additional coordi-
nate has been firmly established. Having this assurance we can, in many
instances, treat spin phenomenologically by including it in the Schrodinger
wave functions u"a i" the Hamiltonian in such a way as to account for the
physical effect being considered. The success of this ad hoc method is attested
ty the frequent use of a "spin Hamiltonian" in the Schrodinger equation for
solving problems.s
I. SPIN IN THE SCHRODINGER FORMULATION
The spin hypothesis may be incorporated in the Schrodinger formalism

by regarding the ipin angular momentum as an additional degree of freedom.
Ii is issu*.d thui the components of the spin obey the same commutation
relations as the components of the orbital angular momentum. However, in
the case of a single elictron, the spin quantum number s has a fixed value of 6,
unlike the value of f , which can be zero or any integer up to n - l. The
projection of the magnitude, f/tit 1t;, of the spin vector on an axis of
quantization behaves in exactly the same manner as the projection of the length,
\/WT, of the angular momentum vector. That is, successive values of m,
must differ by unity. For r : * there are only 2s * I :2 allowed values of
m,, namely, ++. (See Figure B-1.)
dl
m.
Figure 8-l The two spin orientations for the elec-

tron. lf the axis of quantization is defined by a magnetic
field, the spin precesses as shown.
P. A. M. Dirac, Proc. Royal Societ2 (London) 117' 610 (1928) and 118' 351 (1928).
4
For examples in the fieli of magnetism see K. W. H. Stevens, "Spin Hamiltonians,"
5 in G. T.
Rado and H. S;hl, Magnetism, Vol I, Academic Press, New York, 1963'
SPIN IN THE SCHRODINGER FORMULATION 291
In order to include spin in the Schriidinger wave function we must use the
fact that the spin coordinates are independent of the cogrdinates of con-
figuration space. This enables us to write the total wave funCtion as a product
function,
Yroru, : Vu*(r, 0, O) . 1(spin) . ,-iEntltt : lRu . t-iEntlh\ lf, mt) ls, m,). (B.l)
Here the radial and time functions have been lumped together in the same
eigenket, since our main concern in this chapter will be with the last two kets,
representing the angular momentum and spin eigenfunctions.
For spin * *e know that there are two basis functions for the two-
dimensional spin space. If we adopt the convention that an ".rp" spin has a
positive projection on the axis of quantization (usually the z-axis), these
spin functions may be written in the following equivalent ways:
X+ : r(uP) : (i) : ,*, +)
(8.2)
x- :1(down) : (:) : t*, -+).
Since either of these functions may be used in Equation 8.1, we see that each
is doubl2 degenerate in spin. The spin wave functions, or
h2drogenic spatial state
spinors, given in Equation 8.2 are simultaneous eigenfunctions of the spin
operators 52 and S,. Thus, using ket notation:
: + r)h, li-, +) :
3h2 l1 1\
s, l+, *) 7 lEt 2/
+
'.(*
^t, l+, -*) : *(* +t)h, l+, -+) :7li'
3h2
-2/
1\
,s, l+, *) :
h
il+,,+>
^s, l+, -#) : - it+, -+).

In spite of the non-classical origin of spin, it is assumed that the algebra of
the angular momentum operators can be applied directly to the spin operators.
These expressions will not be repeated here since they will be summarized for
a general angular momentum in the next section.
The matrix elements for the spin $ operators are :
(i, m,,l S' l+, ffi,) : h'[+(+ + I)] 6*,,,*, : 3h2

4
6*",,*"
(t, m,,l.S, l+, m,) -- m,h 6*,,,*,
for which the full matrices are
ghzlt 0\
s2: z(o
r) t':;(;nlr 0\
-)
l0 l\ t0 0\
t*:n(o o/ s-:n(r ;/
s,:*(s*+s-) :X('r;) S,:irr--s*) :i(0, -;)
It is convenient to define the Pauli spin matrices:
6,:? r,:(l ;)
6v : i, t' : (o
/0 -i\ (B'3)
,)
o" :? ., : (l
-:)
PROBLEM 8'l ii.$fii*il$'4-\1l,'l.t
(u) Show that the Pauli matrices anti-commute, that is,

that any pair of them satisfies
{o* or} : 6r6u * oro* - 0.
(b) Show that oron : iour 6ao, : io, and oror : io,
(.) Show that o,oro, : il.
(u) Show that the Pauli matrices are unitary.

(b) Operate on X+ and y-with each of the following matrix
operators:
cr, oa, o, S+ and S-.
(.) Write the eigenvectors of or, 6a, 6u as linear combina-

tions of the basis vectors y* and 7-.
(u)
(b)
Show that o| - o', - oli t.
Show that the commutation relations for the Pauli
THE SPIN.ORBIT INTERACTION
matrices can be summarized by the expression
o X o :Zt,o
where the cross product has the same significance here

as it does in Proble m 7-6.
2. THE SPIN.ORBIT INTERACTION
The total wave function given in Equation 8.1 for an electron possessing
both orbital momentum and spin assumes that there is no interaction between
L and $ that is, that [r, S] - 0. Hence, Equation 8.1 is an eigenfunction of
both L, and S,, which means that both ms arrd ms are good quantum numbers;
in other words, the projections of i and S are constants of the motion. In
reality, however, there is an interaction between f and ,S-called the spin-orbit
interaction-which removes the spin degeneracy, since it produces an energy
difference between the states for which the orbital and spin magnetic moments
are parallel and antiparallel. We will see later that this interaction can be
expressed in terms of the quantity t. S. Since t' S does not commute with
either i or ,S, Equation 8.1 is no longer the correct wave function, and lre, and
rts vte no longer independently quantized; hence, they cease to be good
quantum numbers.
An elementary way to picture the spin-orbit interaction is to regard the
stationary spin magnetic moment as interacting with the magnetic field
produced by the orbiting nucleus (see Problem 3-lB). In the rest frame of the
electron there is an electric field, A : Zeilrz, and a magnetic field
H":T
i xi
t
where f is directed from the nucleus toward the electron. Assuming that d is
the velocity of the electron in the rest frame of the nucleus, the current produced
by the nuclear motion is/ : -(Zelc)i in the rest frame of the electron. Then
!l- fr x 8.
2
Leu xr
E": - c
The spin moment of the electron precesses in this field at the Larmor
frequency (see section 10 of Chapter 3),
dr:TE": - 4o x&,
flloc"
(8.4)
with potential energy, r
Eu : -fr'' E' @r'J' (8.5)

294 SP|N, ANGULAR MOMENTA, AND |DENT|CAL PART|CLES
Equations 8.4 and 8.5 hold in the rest frame of the electron. Since the rest
frame of the nucleus is of more importance experimentally, we must make a
transformation to that frame. This introduces a factor of $ which is called the
Thomas factor.6 This factor results from a relativistic effect known as the
Thomas precession Hence, an observer in the rest frame of the nucleus would
observe the electron to precess with an angular velocity of
dr,: -Wd x&, (8.6)
and an additional energy given by
LÊ: _+d".s. (8.7)
Equations 8.6 and 8.7 can be put in a more general form if we restrict Z
to be any central potential with spherical symmetry. Then,
+AV+
F:-i+- or -eE,
and
-.i.. Z IAV+ -
DXd llaV- ,- I IAV--
--Tixi:--
e or ^-DXF: emor :-L.
er 0r dr
Using Equation 8.6,
+:
QL
I lav-_
- 2^rrrri A, "
and
L,E:- av-
.^l'S. (B.B)
2mficz r dr
Equation B.B illustrates the. explicit dependence of the spin-orbit energy on
the relative orientations of /] and,S. For spin : f the energy splitting is:
:
tAE;
ffii#
For the Coulomb potential this can be approximated by L,E : fr\mt. Ze2lrs.
Since the average value of 1/rB is?
z3
alns/(/++)(/+t) (for / + 0), (8.e)
the spin-orbit splitting of the 2p state of hydrogen is approximately
LE,,
-m- (*)' .
# -,ro-a ev.
6 L. H. Thomas, Nature ll7,514 (1926).
? See Appendix A
of Chapter 7.
VECTOR MODEL FOR COMBINING ANGULAR MOMENTA
This is the same order of magnitude as the fine structure correction which goes
as a2 ^, l0-a (see section 2 of Chapter 4).
p.r6 1 i td L: $l.jaola.:, a 493:4 e $):r tr inj an:1 :6 ar!ti€..$rt

PROBLEM 8.1 t;;ry. ri;i,;$ dioea j-iit*.4i1!i+;,'.4nid.'.:iAd);it-Sr1
Assuming a circular orbit, estimate the angular velocity of

precession for the electron in the ground state of hydrogen.
How does this compare with the electron's orbital angular
velocity ?
PROBTEM 8.5
Use the relativistic velocity transformation to derive the

Thomas factor by considering three inertial frames, one at
the nucleus and one each for two successive positions of the
electron. Assume circular motion.
PROBLEM 8.6
Derive the Thomas factor by calculating the time dilation

between the rest frame of the nucleus and that of the electron.
PROBLEM 8.7
(u) Using the classical expression for the energy of two

interacting dipoles, estimate the spin-orbit energy
splitting for the 2p state of hydrogen.
(b) If a nucleus has a non-zero spin I, there will be a
spin-spin interaction between the magnetic moments
associated with the nuclear and electronic spins.
Using the classical dipole expression, estimate the
energy of this interaction (known as the lryperfne
interaction) for the proton and a 2p electron, and com'
pare it with the spin-orbit interaction.
(Ans.: (a) ^,10-5 eV; (b) .-,10-t .V.)
3. THE VECTOR MODEL FOR COMBINING

ANGULAR MOMENTA
As a result of the spin-orbit interaction,l and ,i exert torques on each other
via their magnetic moments, and neither can be treated independently as a
constant of the motion. However, if there is no external torque acting on the
system the total angular momentum must be a constant of the motion. Using
the so-called uector model we define the total angular momentum i to be just the
295 sPlN, ANGULAR MOMENTA, AND IDENTICAL PARTtcLEs
(a)
(b)
Figure 8-2 two possible orientations of the spin and orbital moments for a single elec-
(a) The
tron' (b) spin'orbit coupling causes the vectors i and ito pr"."ss about
velocity. J'at the same angular
vector sum,
j:i +.9.
The vectors t an{S precess about the vector i at the same angular velocity
(see Figure 8'2). The conditions on angular momentum now apply
to J2 and J, instead of-quantum
to L,, sr, L, and s, separitely. That is, we ca.r d.nrrl
the eigenvectors U, m,) such that
: j(j + t)hz lj, m,),

J, U, m)
and
J" lj, mt) : m,h lj, m,).
The good or conserved quantum numbers are j and m,, where jl can have the
values ll *sl, ll + "i
- 11,...,1/ - sl.ForeachvalueofTthere are2j {I
VECTOR MODEL FOR COMBINING ANGULAR MOMENTA 297
possible values of m,. In the case of a single electron there are always just two
values of7, namely, j : f + +. For an s-electron the only possible value of7
is f, which has the two projections, mi : **. For a p-electron, j can have the
values,j : | * * : I or *; for a d-state, j :2 _[ * :8 or g.
The new angular momentum eigenfunctions to replace those of Equation
8.1 are not obtained so easily, since they must be the appropriate linear com-
binations of the old functions which diagonalize the matrices J2 and J,. We
know, in principle, how to determine these functions but it will suffice for the
time being merely to denote the eigenfunctions by their eigenvalues in ket
notation as follows:
Orbital state 7 value Eigenkets lj, mi)
J 1 t1 1\ 11 1\
z lZt 2/t lZt -2/
p 3
q
13 3\ 13 1\ 13
lEt E/t lE> 2/t lT> -2/t
1\ 13 3\
lEt -E /
1 11 1\ tt 1\ (8.10)
2 lZt 2/t l2t -2/
le 5\ 3\ 1\
l2) 2/) l5
l21 2/) l5
1\
l21 2/t l5
:)
z 12, -2,/,
15
lt, _3\ l_5 _ 5_\
-T /, lEt -E /
t3 3\ 13 1\ 13 1\ 13 s\
12>T/t lEt 2/t lT> -2/t lEt -E /
PROBI.EM 8.8
Assume that the spin-orbit interaction can be "turned off"

for the orbital states enumerated in Equation 8.10. Since
m7 and m, would then be good quantum numbers, write the
eigenkets for the above states in either the form l/, m) ls, m,)
or lms, m).
We will now summarize the properties of the angular momentum oper-

ators, using the symbol J which can have either integral o'r half-integral values.
Thus, all of the orbital and spin cases are incorporated in one formalism.
J+ : J, +iJ,
J2 - ti + ti + J2 : t(J*J_ + J_J+) * J"
i x i:ihi
[Jt, J,] : [J', Jr] : [J2, J,] : 0
[J*, J-] : 2hJ " (B.l l)
[J,, J*] : h,J
+
[J,, J-] : -h; -

[J', J*] : [J', J-] : 0
Jnlj, m1) : h UT m,)(j Imr + l) lj,mi + t)
Hereafter, we will drop the subscript.l on m, since it is clear that when 7 is a

good quantum numbei, m refers to one of the allowed projections of7.
The matrix elements of these operators are :
(i', m' lJ'l j, m): hrj (j + 1) 6,, ,,6*, ,*

(j' , m' lJ,l j, m): mh6o',06*,,*
(j' , m' lJ*|,1, m): h.{U+m)(j +m * 1)d,,,i6*,,*+r.
Then the matrices are as follows:
0l-il--o
I
li
io rl
l______-__i
J2:hz
lb
ib
r0
io
i+ oi
i 0 -l i
ir
i0
l0
Ju: h
VECTOR MODEL FOR COMBINING ANGULAR MOMENTA 299
0,
----l---------,
io ri
il
i0
ll
0i
t--------- i ----------_________,
t-i
io \/2 0i
i 0 y2i
lo lt
oi
io t/3 0 0
it- oz o
i o \/3
ioo
J,:
-r
I ---------___-____-__-_--
02 --------- |
i
0/6- i
o{6
02
0
o\E 0
o 2fr,
i0
ii
0i
it oi
,-------i------
i0 0
i {z o
i 0 y2
tt-
J-: h, i0
I
i2
I
I
o 0
i o {6
I
0
I
Go
I
I
i0 0
i0 0 o2
iU
i{s o 0
i02\/20
i'
0i
t9
o)E 0
\/i o fr.
o lz o
iri t-r t/t o oi
i v3 0 2 0
l-
i
il
i o 2 o ,4i
i-i
ii, o o rz oi
- l0
2 o on
i2 o rG oo
,0 v6 0 v6 0
r0 oG o2
i0 0 0 20
o -{z oi
Y2 rl
O -Y2l
o t5 oi
ih
JU
2
0--2 0 0 0i
ro-.nr6o oi
o.vG o -.G 0r
oo\Go _t!-i
0002 oi
i
4. ADDITIONAL INTERACTIONS IN THE

ONE-ELECTRON PROBLEM
If one wishes to account for the observed fine structure of the level splittings
in the spectra of one-electron atoms, there are two other effects which should be
considered. The relativistic correction is one of these, since its contribution is
INTERACTIONS IN THE ONE.ELECTRON PROBLEM 3OI
the same order of magnitude as the spin-orbit splitting. Although the Dirac
relativistic theory is the proper theory to use, it is possible to obtain the
relativistic corrections from the classical Hamiltonian by writing the kinetic
energy in the relativistic form,
T - E - ffiscz : *or, (t * - ffi0c2,

#rr,)r
where Equations l.16 and l.l8 have been used. Fory' 4 moc, the radical may
be expanded to give
lp'pn\
T - + t . ..) -
fu - W
mocz ttoc2
\t
o
p' pn
2*o - @''
Then the approximate Hamiltonian becomes,
ff x@-+z)
\2*o t ' I -L.
}mfrcz'
The relativistic correction can be obtained by treating the Jast term as a

perturbation (see Problem 9-15). The result is
AEur:ry(r-&) (8.12)
Note that this would be in exact agreement with Equation 4.13 for k : / * z!.
The second effect, the nuclear hyperfine interaction, is several orders of
magnitude smaller than the spin-orbit and relativistic corrections. Although it
is often difficult to detect in atomic spectroscopy, the hyperfine interaction is of
considerable importance in electron and nuclear magnetic resonance experi-
ments. It was pointed out in section 9 of Chapter 3 that nucleons possess an
intrinsic spin of $fi. Consequently, nuclei have i net spin except for the special
cases in which the nucleon spins pair off to give a total spin of zero. The
interaction of the nuclear spin with the spin of the electron is known as the
hyperfine interaction. As in the spin-orbit case, the energy associated with
the hyperfine interaction may be expressed by the vector coupling model as
AEr,y*.nnu: Ai'S, (8. r3)
where .I is the nuclear spin and z4 is the hyperfine constant. Although the
nuclear spin is frequently larger than the spin and the orbital angular momen-
tum of the electron, it should be recalled that the nuclear magneton is smaller
than the Bohr magneton by a factor of about 2000. Hence, the hyperfine
energy is smaller than the spin-orbit energy by roughly a factor of 10-3.
The nuclear spin variables, operators, and eigenvalues may be incor-

porated into the Schrodinger formulation in the same way as the spin for the
electron. Thus, we may define a nuclear spin state by the ket ld m1), whete I
is the nuclear spin quantum number, and m1 is the spin projection. De-
generacies of high order can occur here, since the nuclear spin can assume large
values. For example, a nuclear spin of $ has l0 possible projections, which are
all degenerate in the absence of an interaction that would remove the de-
generacy. The total wave function, Equation B.l, including nuclear spin is
modified as follows:
Y,o,u, : lRnr. e-iE"tln) lf , m) ls, mr) lI, *r). (8.14)
However, if the hyperfine interaction is included in the Hamiltonian, Equation

8.14 is no longer an eigenfunction, since i ' S does not commute with either .for
S. This is again analogous to the spin-orbit case discussed in section 2. We
must now enlarge our definition of the total angular momentum of the system,
since it is that quantity which is conserved, even in the presence of internal
interactions. Therefore, we use the vector coupling model to define the total
angular momentum,
F:j*7:L+S+i
where F and Ms are the good quantum.numbers and where Mp can take on the
setof values lj +Fl, lj +F - 11,.. .,li -Fl. Theform of the new eigen'
functions which replace Equation B.l4 is then,
Y,o,", : lRnre-iE"tln\ V, Mil. (8.15)
As in the spin-orbit case, the kets lF, Mil may be expressed by linear corl-
binations of the kets I t, mt) ls, m,) lI, mt).
5. IDENTICAL PARTICLES, EXCHANGE, AND

SYMMETRY
In quantum mechanics we regard identical particles as indistinguishable.

This is an important departure from classical mechanics, where it is possible
not only to specify the coordinates of each particle at a given instant but also
to follow the subsequent motion of each. Thus, even identical particles can
be "tagged" in classical mechanics. Identical pirticles are far more elusive
in quantum mechanics, as the following example will show. Suppose that_ we
observe two electrons in the same orbital state and we ignore all interactions
between them. Then we can express the total spin state as the product of the
two one-particle states. In particular, if we know that their total spin is zero,
we could write the two-particle spin function as either Z*(l) 'X,-(2) or as
X,_$) X*Q). Since we do not know which electron has its spin up and which
.
has spin down, either function is as good as the other. Furthermore, for all we
know, the electrons might even be exchanging positions and spin states while
we make our measurements. Therefore, the most general spin function is a
IDENTiCAL PARTTCLES, EXCHANGE, AND SYMMETRY 303
linear combination of these two states, namely,
I
a
' + 'r*(2)1. (8.16)
?Si"
\/2 lx*Q) x-Q) x-(1)
The reader should note that V$i" does not change sign under exchange of the
two electrons, while ?*pi,, does change sign under particle exchange. ?.ft" is
said to be symmetric and tp*,r, antisymmetric under particle exchange.
The foregoing ideas may be extended to more general functions by
defining the two-electron wave functions,
v,o(1,2) : ttt,Q) ' ,pr(2) and y:u,(r,2) : va(l) 'ttt,Q). (8.17)
Here the lette r a or D represents the set of quantum numbers (r, /, m, ffir)
which defines a specific one-particle state described by both spatial and spin
coordinates. The integer 1 or 2 refers to the set of spatial and spin coordinates
associated with a given electron. Then, the symmetric and antisymmetric
linear combinations of these functions are,
I
t[
+ ?rr)
{z 0|ou
and (B.lB)
I
ltz
\/2
0t,ou - ,-r')
We now define a particle exchange operator, Prr, which exchanges the co-
ordinates of the two electrons. Thus:
PuVoo(I, 2) : V,(2) 'Vu(l) : tltu,(|,2),

and
Pr*too(\,2) : VoQ) 'rlt,Q) : rl,ou(\,2).
Performing the same operations on the functions given in Equation B.lB, we

find that:
Pplps : *?s
(8.1e)
Pngt : -9t.
From the foregoing we see that although the functions yo6 and yuo are not
eigenfunctions of the particle exchange operator, Prr, the linear combinations
defined by ?u and yn are eigenfunctions. A symmetric eigenfunction of Pp has
the eigenvalue f 1, while an antisymmetric eigenfunction has the eigenvalue
-1. Furthermore, if the operator P* commutes with the two-particle Hamil-
tonian, then the exchange symmetry of each eigenfunction is constant for all
time.
Operations with the particle exchange operator should not be confused
with those of the parity operator II which was discussed in section 10 of
Chapter 7. Although the eigenvalues of both operators are + 1, fI is concerned

with symmetry under inversion of the coordinates (reflection through the origin),
whereas P is concerned with symmetry under particle exchange.
It has been found experimentally that tlte waae function for a s2stem of
electrons must be antisltmmetric in the exchange 0f the coordinates of ont two electrons.
This is a general statement of the Pauli principle, which was first enunciated in
1925 in a form something like the following: no two electrons can occupy simul-
taneousllt the same quantum state on the same atom. In order to reconcile these two
statements let us address the specific problem of two electrons on the same atom.
From Equations B.l7 and B.lB note that if the two electrons were in the same
spatial and spin quantum states, that is, if ,p" : rpo in violation of the Pauli
principle, then the antisymmetric wave function would be zero, whereas the
symmetric wave function would not. Thus:
I
V.t (poo-tyno):0,
{z
I
11s :- (l,oo*1poo):12rp"".
f,
It follows that the Pauli exclusion principle will be automatically satisfied if we
require that only an antisymmetric total wave function be used to describe a
system of electrons. The antisymmetric wave function for two electrons
(Equation B.18) can bewritten conveniently as the 2 x 2 determinant:
IPz : 1lp,(t) y',(2)l 1

lrp"(l) ' ,po(2) -
l::-
!,0(1) 'v"(2)1.
&lrpr(t) ,p,(2)l ttZ
The generalization of this form, which is called the Slater determinanf,s can be
used to construct the antisymmetric total wave function for any system of /[
electrons. Thus:
,p,(l) y,"(2) qt"(N)
po(l) ,po(2) ,Pr(N)
I
IP,q.:wr. (8.20)
Note that the rule for expanding the determinant insures that the wave function
is antisymmetric; the fact that the determinant is zero if any two rows or
8 See, for example, J. C. Slater, Qtwntum Theor\ of Matter,2nd ed. McGraw-Hill Book Co., New
York, 1968, Chapter ll.
IDENT|CAL PARTTCLES, EXCHANGE, AND SYMMETRY
columns are equal guarantees that the Pauli principle is obeyed. An alter-
native definition which is equivalent to Equation 8.20 is,
:
V,t
#, t r..Pr,r,...,.'[v"(
\'PuQ)''''t',(N)1, (8.2 r )
where Pr.r,'.,,r,,permutes the coordinates labels lr2, .. . , .nf. The factor e is *l

when the permuted arrangement of these integers is equivalent to an eaen
number of two-particle permutations; e : - I when the arrangement is
equivalent to an odd number of two-particle permutations.
Electrons are not the only particles that require antisymmetric total wave
functions. It is an experimentally observed fact that all particles with half-
integral spin are in this category, which includes positrons, protons, neutrons,
mesons, neutrinos, and nuclei with odd mass numbers. As a class these
particles are called fermionr, since they are found to obey Fermi-Dirac sta-
tistics. * On the other hand, all particles with zero or integral spin-which
includes photons, phonons, pions, alpha particles, and nuclei with even mass
numbers-musf have symmetric total wave functions. Particles of this class
are called bosons since they obey Bose-Einstein statistics.*
As in the case of a one-particle wave function, the wave function for a
system of particles can be written as a product of a spatial part and a spin part
whenever the spatial and spin coordinates are independent. Then, for the
total wave function we have the following possibilities:
Fermion states:
y'f-particle spatial function x antisymmetric
f 1tyr"*.tric
| \ff-putticle spin function
Tz :
I lurrtirymmetric i[-particle spatial function x symmetric
( \ff-putticle spin function
Boson states:
f 1ry-*.tric //-p article spatial function x symmetric

| \ff-particle spin function
?s :(
| /antsymmetrrc if-particle spatial function x antisymmetric
\ \nf-putticle spin function
To illustrate the above with a specific example, let us separate the spatial
and spin coordinates 'of the two one-electron wave functions, rpo and ga, vs
follows:
ga : vnx+ and !)o : vn,x+,
where n andn' represent different sets of the quantum numbers n, /, m. Then

* See Appendix A of this chapter.
we can obtain the two spatial functions,
I
?spatiar : (2) I vn' 0)'p
"(2)1,
ft,lrp"(L)rp "'
where the function having the plus sign is symmetric and the one having the
minus sign is antisymmetric in the exchange of the electrons. Likewise, we can
write the spin functions:
x,*Q)'x*(2)
x,-Q) ' x,-Q)
I
'x-Q) + x-0)'x*Q)1,
6lx"(\
where the last function containing the minus sign is antisymmetric and all
others are symmetric. Therefore, four antisymmetric product functions may
be written as follows:
x*Q)
I
-tr)' ) - (I) -(2)l (2) + r-(1) 'r.(2)rl
ftfr rp 2 rt, rp
*,( *,
\'j_';: rrr
rrr_r'
l1 'r.Or.l
frfv"tr)
',t'*,(2) + !,*,(L)'p"(2)l
&[r*(l)
' x-( 2) - u-(1)
(8.22)
PROBI.EM 8-9
(u) Write an expression similar to Equation 8.21 for the

most general s1mmetric wave function for // bosons.
(b) Show that Equations 8.20 and B.2l are equivalent for
the case of three electrons.
PROBLEM 8.'O :i'r,:1r:1! r r::1i;ll;;;Fi:t i.l;iI 1l:l: Il -'ii i il lf:iii ifiE i;i:: l; i ri:l:f;::::*:;;:;;; '::;t I Xii
(u) Apply the particle exchange operator Pn to each of the

probabitity densities yf;6tpo6 and y{oyro. Do they remain
unchanged ?
(b) Operate on ?$?s and y\rpo with Ptr. Do they remain

unchanged ?
(.) Do the above results suggest a physical reason for

choosing gs and Vt. as wave functions rather than yo6
and y6o?
EXCHANGE DEGENERACY AND EXCHANGE ENERGY 307
PROBLEM 8.1'
Show that P* commutes with the Hamiltonian
j/r :!2m,*g2m,-Zez *'1

T1 -t!12 Tp
.
6. EXCHANGE DEGENERACY AND EXCHANGE

ENERGY
If the Hamiltonian for the two-electron system commutes with the particle
exchange operator, then all four of the functions of Equation 8.22 are energy
eigenfunctions. This will occur when the Hamiltonian itself is symmetric in
the coordinates of the two electrons. By way of illustration, consider the follow-
ing Hamiltonian for two non-interacting electrons: *
"/f (r,2) : [- #', * [- #o, * u,] - /r(L) + 'tr(2)'

"]* (8.23)
lPrr, tr(l, 2)l : g. (8.24)
As a result of Equation 8.24 we know that the symmetry of each eigenfunction

& t(1,2) will be constant for all time (see Equation O.f A;. Furthirmore, the
functions given in Equation 8.22 can be shown to be degenerate eigenfunctions
of ,tr (1,2). Assume that the one-electron functions satisfy the following
equations:
tr (l)r1"0) : E,rp*(I)
lf :
(2)rP"(2) E*rP,(2)
tr (l)rp* (l) : E,,rt,,,(l)
.tr (2)V"'(2) : E",rP",(2).
Then:
,tr (1, 2)rpo : (En * En)rp,o,
where Vtisany one of the four functions in Equation 8.22. Although these four
functions are degenerate when there are no spin-dependent forces or inter-
actions between the electrons, we will see that the motions of the two electrons
are correlated with their spin states. Thus, for the S : 0 state (singlet state),
when the spins are antiparallel, the spatial eigenfunction is symmetric and the
electrons can come very close together. In fact, the probability density for this
state has a maximum value when the electrons are so close together that their
coordinates are essentially the same. However, for the S : I state (triplet
* Although this is physically unrealizable, we will even neglect, for the time being, the Coulomb
interactions between the two electrons.
308 SPtN, ANGULAR MOMENTA, AND |DENT|CAL PART|CLES
/ l-
singlet
t/
/' Z/ (3-fold delenerate, .ttturc 8-3 The effect of the Coulomb re-
t)r' -.Triplet 3,:'#:"a:
tb.EEFarue degeneracv for two
En + En,
(4-fold degenerate)
-J
state), when the spins are parallel, the antisymmetric spatial function goes to
zero as the electrons approach each other. Thus, in the triplet state the motions
of the electrons are correlated so that they avoid each other as much as possible,
while in the singlet state they attract each other. That is, they move as if they
were under the influence of a force that is repulsive for parallel spins and
attractive for antiparallel spins. This force is called the exchange force since it
has its origin in the symmetry requirements on the wave function under the
exchange of indistinguishable particles. The exchange force is strictly a
quantum mechanical phenomenon with no classical analog.
In spite of the introduction of this new force, the exchange force, the
functions in Equation 8.22 are still degenerate when the Hamiltonian is
written as in Equation 8.23. Suppose we now add one term to the Hamil-
tonian, namely, the Coulomb repulsion between the two electrons, *e,lrrr.
Although this term does not depend upon the spin coordinates, it is evident
that it will partially remove the excha4ge degeneracy of Equation 8.22. Since
the electrons are closer together, on the average, in the singlet state than in the
triplet state, the Coulomb repulsion increases the energy of the singlet state
more than it increases the energ'y of the triplet state (see Figure 8-3). The
energy difference between the singlet and triplet states is called the exchange
enetgt; it is really an electrostatic energy, but its very existence depends upon
the fact that the exchange force causes the aligned spin state to have lower
energy. Accordingly, it is energetically more favorable for the spins of electrons
in the same orbital state to assume a parallel alignment. It is in this sense that
we say the exchange force aligns the spins. This topic will be discussed further
in Chapter 10 in connection with the helium atom and ferromagnetism.
PROBLEM 8.12
Compute the ground state energy for 3 identical fermions

confined to a cubical box of edge a. How does this energy
compare with the ground state energy for 3 identical bosons
having the same mass ?
PROBtErvl 8-r3
Consider a system consisting of a large number of identical

particles, of the order of Avogadro's number, distributed
over a great number of closely spaced energy levels in a big
box. Would you expect the heat capacity of this system to
depend upon whether these particles are bosons or fermions ?
Explain.
THE PERIODIC CHART AND MANY.ELECTRON ATOMS
PROBLEM 8.11
What is the minimum total energy of 5 neutrons and 3

protons confined to a cubical box of edge a ? Neglect all
interactions between particles.
7. THE PERIODIC CHART AND MANY-ELECTRON

ATOMS
In Chapter 7 we obtained the spatial states for a single electron in the

presence of an atomic nucleus, and we saw how these states can be specified
by the three quantum numbers n, f, and m7. In the present chapter we have
seen how the spin of the electron is included by adding a fourth quantum
number, r,, *hich gives the projection of the spin along the axis of quanti-
zation. In ket form, the state of a single atomic electron may be represented by
ln, lrTttrntr)t or by lnrjrmt,mr), wheie the latter takes into account the effect
of the spin-orbit interaction. When a second electron is added to the atom we
must consider not only the strong Coulomb repulsion between the two elec-
trons, but also the symmetry properties of indistinguishable particles of spin f.
In a neutral, unexcited atom the electons occupy the lowest possible
energy states consistent with the Pauli principle. Thus, for hydrogen the
electron occupies ?!)u state (see section B of Chapter 7) with its spin in either
the frn, or the -m, state. In helium the second electron also occupies the g1,
orbital state but its spin state must be opposite to that of the first electron.
Since these two electrons exhaust the possible electronic states for n : l, we
say that the first shell is full or "closed." Electrons associated with the same z
value are said to be in the same shell, and those having the same value of /
(as well as z) are said to be in the same subshell. For z : I there is only the s
subshell (/ -- 0); for n : 2 there are two subshells, the s and p, corresponding
to f,:0 and /:1, respectively. In the case of lithium, with Z - 3, the
third electron must go into the 2s subshell. Normally, the s states have slightly
lower energies than the states with non-zero angular momentum associated
with the same principal quantum number. For beryllium, with Z :4, the
fourth electron closes the 2s subshell. The fifth electron of boron goes into a 2y'
state. Since there are three p orbitals, there are six possible y' states, including
spin, which are successively filled by the elements from Z - 5 to Z
Neon, for which Z - 10, has a completely closed-shell configuration since the
- lA.
ls, 2s, and 2p subshells are all full. Note that a closed shell or subshell has a net
angular momentum of zero and a net spin of zero, since all of the orbital
momenta and spins cancel in pairs.
In addition to the s and p subshells, the n : 3 shell contains a d subshell
which can hold ten electrons, making a possible total of 18 electrons in that shell.
In general, the total number of electrons that the nth shell can hold is 2n2.
The periodic chart is built up in this fashion, although there are important
exceptions to the orderly filling of the shells and subshells illustrated above.
For example, in the transition metal series (Z : 2l to 28) the 4s subshell fills
3IO SPIN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
IONIZATION POTENTIALS
li 4r i sa i oo
Figure 8-d lonization potential for the outermost electron of each element. (From Introduction
to Atoiic Spectra by Harvey E. White. Copyright 1934 by McGraw-Hill Book Company. Used with
permission.)
before the 3d subshell, and in the rare earths (Z : 58 to 7 t ) the 5s and 5p

subshells fill before the 4/ subshell. These unfilled, inner subshells account for
the unusual magnetic properties of these metals (see section 3 of Chapter 10)
as well as many other aspects of their behavior in the solid state.
The specification of the quantum states for all of the electrons of a given
atom (or ion) is known as the electronic confguration for that atom. Thus, for
copper the electronic configuration is Isz 2s2 2p6 3sz 3p0 34ro 4s1, where the
superscript gives the number of electrons of each type. The configurations for
all of the elements are given in tabular form in Appendix B of this chapter.
The outermost electrons of an atom, called the ualence electrons, are the
electrons that take part in the formation of chemical bonds. In'a solid, these
are the electrons that form the conduction bands of metals and the valence and
conduction bands of semiconductors. Many aspects of the physical and chemi-
cal behavior of substances depend upon how firmly these valence electrons are
bound to their atom. Figure B-4 shows a plot of ionization energy for the last
electron of the configuration versus atomic number. Note the great stability
of the electronic configurations for the noble gases. Since they have no valence
electrons, they are almost chemically inert. However, the element just above
and just below each noble gas is quite active chemically. The additional
electron of the element just above the noble gas is easily removed in chemical
reactions. Likewise, the hole in the closed-shell configuration of each element
just below a noble gas is easily filled. The energy given up by the atom when
an electron fills the hole is called the electron ffinit1. For a detailed discussion
of the relationship of the properties of the elements to their electronic con-
figurations, the reader is referred to a text on general chemistry.
We will now look briefly at the many-electron problem. In order to
obtain accurate values for the ground state energy of the atom, one would have
to include in the Hamiltonian all of the following contributions to the energy:
(1) the single particle kinetic energies, (2) the single electron interactions with
the nucle"i, 1e; the electrostatic interactions (including exchange correlation)
of the electrons, (+) the spin-orbit interactions, (5) the spin-spin interactions
between electrons, (6) the dipole-dipole interactions between the orbital
magnetic moments of the electrons, (7) the hyperfine interactions between the
nuclear spin and the electron spin, and (B) other effects such as relativistic
corrections and quadrupole interactions. We know how to handle (1) and (2)
ANGULAR MOMENTUM VECTOR COUPLING SCHEMES 3il
for any number of independent particles by merely writing the Hamiltonian as

the sum of all of the single particle Hamiltonians; the total wave function is
then expressed as the properly symmetrized combination of products of the
one-particle wave functions. (3) is often neglected in the first approximation,
although it can be a large effect. It can be treated by perturbation theory for a
small number of electrons. In the case of a large number of electrons, one
generally regards each electron as moving in an effective potential which de-
scribes its average interaction with all of the other electrons. This topic wiil be
discussed further in Chapter 10. (7) and (B) were discussed in section 4. Since
they represent small corrections to the energies they are usually neglected
unless a refined calculation is required. The remaining interactions, (4), (5),
and (6), will be treated by means of one of the vector coupling models of the
next section.
8. ANGULAR MOMENTUM VECTOR COUPLING

SCHEMES
Thus far we have considered only the coupling of the spin and orbital
momentum of a singie electron by means of the spin-orbit interaction. A
question naturally arises as to how one should proceed in the case of two
electrons where there are four constituent momenta to be considered. One
model, known as the j-j coupting modll, assumes that the spin-orbit interaction
dominates the electrostatic interactions between the particles. Thus, we write
jt : ft + .St and i, : L, + & for each of the two particles, where the
quantum numbers jo are obtained as before. Then the total angular momen-
tum is obtained by combining J, and Jr. That is,
j:jr*j,
andT : ljt * jrl, l,r'. * j, - 11,.. .,Ij, - jlr.
We will illustrate y-7 coupling by applying it to iwo inequivalent* /_
electrons. For each electron we found above that7, : j2 : * or 3. Then the
possible ways of combining these are shown in Table B-1. In a weak magnetic
TABLE 8-l j-j Coupling of Two lnequivalent p Electrons
Number of States in
Spectral a Magnetic Field
Ir iz i Terms (Number of mo Values)
3
z ,a z, 2, l, o (9, 9)r,z,r,o 16
3 1 ,1
o, r!1r
2 2 2t 212,1 B
,1 +
3
l, 0 (*, t)r,o 4
1 9 Ir
at /1 lt\ B
2 2 \!t 2-)Z.L
10 rtut"t 36 states
* Inequivalent electrons occupy different atoms or different orbits of the same atom. Hence,
they automatically satisfy the Pauli principle.
field, each_ state of a given 7 will split into 2j + 1 states corresponding to the
allowed values of m,.
Although yy coupling is used extensivety for the description of the nuclear
states observed in ttn.l.ur spectroscopy, it is not appropriate for many atomic
systems because of the strong electrostatic (and other) interactions between the
two electrons. The model that has been most successful in accounting for
atomic spectra of all but the heavier atoms is known as the Russell-Saunders
coupling scheme.e In essence, this model simply assumes that the electrostatic
inteiaction (including exchange forces) between two electrons dominates the
spin-orbit interaction. In this case, the orbital momenta and the spins of the
tio electrons couple separately to form E - it * F, a1d S : Sl1 + Sr. Then
the total angulai momentum is given by J : i + S, as before. For two
inequivalent"p electrons we have: (:2, !, or 0 and s : I or 0. Fot each /
utd s thej values are | / * sl, ll +s - 11, . . .,1/ -sl and for eachT value
there are-(2j + 1) values of m,. The combinations are shown in Table B-2.
Once again-there are 36 states in a weak magnetic field, although of course,
their energies are not the same as those in the Yy coupling scheme.
Supp6se we now subject the two electrons to an intense magnetic field
such that p6I is much greater than the spin-orbit interaction as well as all
interactions between the electrons. This will uncouple the momenta, regardless
of whether Russell-Saunders or 7-7 coupling holdt it *.uk fields, so that -tr,
ir, ^fr, and.i, will precess independently about d. The good quantum num-
Uits *itt no longei tr.7 and m but will be msr, /rt/r, tTtrrr artd mrr. The total
number of states-that is, the possible combinations of these quantum num-
bers-will be (21, * l)(21, + 1)(2t' + t)(2s, + l) : ?6, as before'
'discussed
We have thus far only inequivalent electrons, whereas equi'
ualent electrons are of great physical importance in atomic spectroscopy. !inc9
equivalent electronr o-".rpythe same orbital state on the same atom, the Pauli
TABLE 8-2 Russell-Saunders Coupling of Two lnequivalent p

Electrons
Number of States in
SPectral a Magnetic Field
i Terms (Number of m, Yalues)
, I 3,2, I tDr,r,, l5
2 0 2
,D, 5
I I 2, I,0 3D
r 0,1, 2
9
1 0 I ,P, 3
I I tS,. 3
0
0 0 0 tso I
l0 rt"t t 36 states
e H. N. Russell and F. A. Saunders, Astrophls, J. 61, 33 (1925). For a more detailed discussion,
see J. H. VanVleck, ih, Throry of Ehctrii and Mignetic Susceptibitities, Oxford University Press, Oxford,
1932.
ANGULAR MOMENTUM VECTOR COUPLING SCHEMES 3I3
TABLE 8-3 The Fifteen States for Two Equivalent p Electrons
^t, ^t, -., -.r, flt s Spectral Term
I I 2
I _l
2 22 0 tD,
I 0 2
1
-2
I ll 0 tD,
I -1 I2 _1o 00 0 ,D,
0 -l 2
1 _12
-l -l 0 tD,
I ,D,
-l -l 1
, -2 -2 -2 0
0 ++2rl ,P,
-l ++ror "P,
0 -+-+ol-r ,P,
-1 -+-*-ro-l
_+ _1, _2 _l _l
,P,
-l "P,
0 -++ll0
+ + o -t l
"p,
tp,
-l
+ -+
+-+ooo -l -1
o tP,
-1
tPo
-l
_l 2 tSo
exclusion principle puts restrictions on the allowed values of m7 and m,. Let us
now re-examine the two coupling schemes just discussed and apply the ex-
clusion principle.
In the case of the Russell-Saunders coupling we find that for two equivalent
y' electrons we must exclude the following terms i ,Dr.r,r, lPr, and 3Sr. The
exclusion of the 3D terms is easily understood, since a D term requires that
/ :2, which means that ms, and, ms, must be the same ; then the Pauli
principle permits only a singlet spin state (opposite spins), or a tD term. Simi-
larly, the 3S term requires that ms, - rnh: 0 and that the spins be parallel,
which violates the Pauli principle. The exclusion of the lP term is not quite so
obvious. However, if a table is constructed such as Table B-3, which shows all
of the allowed states, every state is accounted for by the tDr, tPr,r,o, and 1,So
terms. There are no additional distinguishable states which can be attributed
to a rP term. In a weak magnetic field these five states split into 15 states. It is
easy to see why only l5 of the original 36 states for two inequivalent electrons
remain. Six of the 36 states violate the Pauli principle, and of the remaining 30
states only half are distinguishable.
When applying 7y coupling to two equivalent y' electrons, note first that
forJ, : jr:8 we must exclude the following cases:
,'-3 ,'-s ;-q.
Jt-Et J2-Vt J -.)
'jt:*, jz:*, j:I
jr:t, jz:t, j--1.
The first case is obvious since it would require parallel orbital moments and
parallel spins, whiCh would violate the Pauli principle. The third case would
require lt : lz : 0, as well as parallel spins. The middle case is not so
3t4 sPlN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
TABLE 8-4 j-j Coupling of Two Equivalent p Electrons

Spectral Number of States in
Ir i, Terms a Magnetic Field
33 2ro 13 3t 6
qq \2t 2 ) 2,0
31
qo 2rl /!. l\
\2, 2) 2,7 B
11.
7t6 0 /l 1\ I
\2, 2l o
5 states 15 rt"t*
obvious, but as in the case of the lP term in the example from Russell-Saunders
coupling, it can be shown that no additional distinguishable states are available
to prorrlae for the 7 : I level when jt : j, : ,a. Also, the states given by
jr:8, jz: $ are indistinguishable from those given by.tt : t, jr: ,a. Th91,
for .q"i"alenl electrons, the distinguishable allowed states are shown in Table
B-+.
The above coupling with the empirical rules of
schemes, augmented
Hundrlo enable one to account for many features of the atomic spectra of
many-electron atoms without actually solving the many-body problem. A
quantum mechanical calculation is necessary, however, to obtain the energies
of the states and to verify Hund's rules.11 Hund's rules for the ground state of a
collection of equivalent electrons rrray be summatized as follows:
(1) Electron spins align so as to produce the largest value of S consistent

with the Pauli principle.
(2) The states with largest / values are filled first.
(S) J :/ sfor less than half-filled shells, butT :l + sfor more than
-
half-fitled shells.
To see how these rules are applied, consider the ground state of samarium,
which has 6 electrons in the 4/shell and no other partially filled shell. Since the
4/ shell can hold 2(2 .3 1) 14 electrons, is less than half full in Sm,
+ : it
and all six spins are parallel. Thus, s :
3. A possible set of different m7 ualues
0; -!, -2, which combine to give tnl :3, corresponding to / : 3;
is 3, 2, 1,
that is, an F teim. By the third rule above, i -- / - r :-0, and the term
symboi is tFo. The superscript is the multiplicity,2s * 1, while the subscript
is the valffii*a
example, consider the ground state of iron, which has - 6
electrons in the 3d shell, the only partially filled shell. Since this shell can hold
2(2 .2 + l) : 10 electrons, it is more than half full. Then 5 electrons have
parallel spins and the sixth must be antiparallel, producing a total spin of 2.
ih" *ruuln.t of the 5 parallel electrons are 2, lr 0, -1, -2 which combine to
for example, G. Herzberg, Atomic Spectra and Atomic Structure, Dover Publications, Nbw York,
10 See,
1944,or H. n. Wnii", introductibn to Atomic Spictra, McGraw-Hill Book Co., New York, 1934.
it E. U. Condon and G. H, Shortley, The Theor2 of Atomib Spectra, Cambridge University Press,
Cambridge, 1953.
THE LANDE s FACTOR AND THE ZEEMAN EFFECT 315
zero; the sixth electron has a momentum of 2. Hence, { : 2, which corre-
sponds toaD term. Fromthethirdruleabove, j: I + s --+,andtheground
state term for Fe is 6Dn.
We will close this section by merely stating the selection rules for electric
dipole transitions which are appropriate for the two vector coupling models.
More will be said about selection rules in section 6 of Chapter 9. It is assumed
that only one electron at a time makes a transition.
Russell-Saunders coupling :
For the electron making the transition, L^l : tl, (that is, the parity
must change).
For the atom as a whole,
As:0
L,/:0, +l
Lj:0, *l (butatransition'from j:0 to7:git
forbidden)
Lm, : g, f I (but if Aj - 0, a transition from ffii : 0
to m, : 0 is forbidden)
j-j coupling:
For the electron making the transition, L^/': tl (parity must change),
and A7:0, tl.
For the atom as a whole,
Lj :0, *1 (a transition from-l : 0 to j :0 is forbidden).

L*, - 0, *1 (if Lj :0, a transition from ffii : 0 to m, : 0 is forbidden).
g. THE LANDE g FACTOR AND THE ZEEMAN

EFFECT
We have previously obtained the result that the orbital and spin contri-
butions to the magnetic moment are given by (section 9 of Chapter 3),
+ke,*
lrr: -ntE: -g'Pu{@i
and
+g'oi
Itrs: - #_S: -g,prrlr1r + ji,
where gt : I and 3, : 2.002+ ,--',l. Now, when i and S are coupled we have
J:i+S
and
i :it *7": - Yf (t + 2S). (8.25)

3t6 SPIN, ANGVLAR MOMENTA, AND IDENTICAL PARTICLES
I
lr.. cos 0, s)
: Ferr
+
I
f,cz COS 0, L)
Figure &.5 The efflective magnetic moment of an atom is the comPonent of y' which lies along
the direction of the total angular momentum.
It is evident from the last two expressions that the total magnetic moment is not, in
general, colinear with the total angular momentum. This is illustrated in Figure 8-5.
Since E and,,S precess about i, it is apparent from the figure that y' also pre-
cesses about i. However, the efectiue magnetic moment, that is, the component
of p along J, maintains the constant value,
Fi: rt.i
lJl
: _Fat.J +zS'J
h tJt
puL'(E +S) +23'(i +S)
h tjt
palr+2Sz+3i.S
h tJt
pa Ez + 2Sz + ,L(i' - iz - sr)

h tJt
THE LANDE g FAcroR AND THE ZEEMAN EFFECT 3t7
where we have used a relationship in the last step that will be verified in
Problem 8-16.
Then,
irt:-
pa ljl ftj' + ,S' - i'\
h \W)
: _ rf v,.(, * jr+Jz- 2J2

L:)
: - p"{jffi
.[t + i(i+1)+s(s+t)-/(/+
2i(i + r)
!l
Defining the Landi g factor as
.9:1 +J-(f! *s(s+1) -/(/+r) , (8.26)

2i(i + t)
the efectiae mag\etic moment becomes,
lpil-spu{ju+\. (8.27)
Note that for zero spin, Equation 8.26 reduces to the classical case of g
in a similar fashion 3 : Z It t: 0. - l;
Now it is possible to account for the so-called anomalous Zeeman effect
mentioned in Chapter 3. In a weak magnetic field, the angular momentum
will precess about I such that the projection of J along the"field direction will"/
be one of the allowed-values,-m,h. Thi correspondinginagnetic moment along
the field direction (taken to be the z-directio"; *itt lh.t i.,
Itrz : -9\amr
having a magnetic dipolar energy of
E: gmtpa%. (8.28)
The feature that distinguishes Equation 8.28 from the case of the normal
Zeeman effect is that here we see
that g is no longer unity, but depends upon
the quantum numbers /, s, and7. By way of illustiation lei us calcuiategfor an
electron in a p state and an s state:
orbitalstate / j g
p 134
LE5
p 112
L25
J 01r2
318 sPtN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
(4s, j) g:o g +o m;
J
I
t
(',;,;)
-tI
J
2
I
t
(t,rt,rt) _,L
('' l'l)
12345678910
Figure 8J Zeeman splittings for p and s states.
In a magnetic field 0, such that p,u98 is less than the spin-orbit energy, j and
miare good quantum numbers and the energies of the states split, as shown in
Table B-5. These splittings are shown schematically in Figure 8-6, and the
allowed spectral lines due to electric dipole transitions are shown in Figure B-7.
The lines corresponding to transitions in which m changes by +1 are labeled
o, and those corresponding to Lm: 0 are labeled z. The z lines are plane
polarized with the direction of polarization parallel to the field. The o lines
TABLE 8-5 Energy Splitting in a Magnetic Field

AE : gmi
Orbital State i mi g (in units of P.u%)
p 33+ ,
p 31L
'23
z23 5
p 3
z-23 _1 !
334
-5
p t-25
112
-,
1
p 823 3
p r12 1
2-23
J a1,
d6- I
1 _1 ,
.t oq -1
t
THE LANDE g FACTOR AND THE ZEEMAN EFFECT 3t9
-3 -lE,l
Energy (units
"rl u'a)
Figure 8-7 The spectrallines.resulting from the electronic dipole transitions shown in Figure 8-6.
lf the emitted light is observed perpendicular to the magnetic field, all of the above lines are seen
(transverse Zeeman effect). However, if the emitted light is observed parallel to the field only the
o lines are seen (longitudinal Zeeman effect).
are circularly polarized when observed parallel to the field, and linearly
polarized (perpendicular to the field) when observed at right angles to the field.
The selection rules for these transitions will be derived in the next chapter.
It is evident from the foregoing that the so-called "anomalous" Zeeman
effect is, in reality, what would normally be expected for an electron having
half-integral spin in a weak magnetic field. The "normal" or classical Zeeman
effect discussed in section l0 of Chapter 3 cannot occur for a single electron in a
weak field because of the spin term in Equation 8.26. However, in atoms in
which the spins are paired so that the total spin is zero, the g value for all
spectroscopic states is the classical value and only three spectral lines are ob-
served. The pair of lines whose energies are shifted by tpag are circularly
polarized, and the unshifted line is plane polarized.
fn a sufficiently strong magnetic field such that the spin-orbit splitting
is less than the magnetic splitting, the anomalous Zeeman effect converts to the
normal Zeeman effect. This phenomenon, known as the Paschen-Bach effect,
can be explained in the following manner. When the spin and the orbital
magnetic moments interact with the applied field more strongly than they
nri ml m. m7* 2m.

1
AE
6EI
3
, t!,
4
3
2
I
t oi
I
-tt-t;
Zil;eii;r I
r_l t t-)
-, 0-]'
1
-l
2.0 2.5
-t3 -, -L
ttsS _
(AE;
Figure 8-8 Energy level splittings in a magnetic field for the 2p states associated with j : 3/2 and
i :112. The zero-field fine structure splitting (AE)o is largely due to the spin-orbit interaction. The
weak-field Zeeman effect occurs for values of I below the point where the levels would first cross.
The Paschen-Bach effect occurs for very large fields where the splitting produces five levels that are
nearly equallyspaced. ln the latter case, three spectral lines are seen as in the normalZeeman effect.
interact with each other, E and,S are effectively uncoupled and they precess
independently about fr. The good quantum numbers are then nt1 vrtd flt*
insteid of 7 and mn which hold in weak fields. Hence, in strong fields the
projections of the magnetic moment on the field direction are proportional to
mt * 2m, rather than m,. This is illustrated graphically for / electrons in
Figure B-8, where the energy level splittings are shown as a function of applied
mignetic field strength. -Using the selection rules, L*, - 0, :tl and Lm, : 0,
only three spectral lines are observed, as in the normal Zeeman effect, fot
transitions from the 2p states to the 2s states in high magnetic fields.
PROBTEM 8.I5
(u) Using Russell-saunders coupling, obtain the sp_ectral

terms for the ground states of Nd3+, Gd8+, and Ho3+,
which have, respectively, 3, 7, and 10 electrons in the
unfiIled 4;f shell.lz
(b) Calculate the g values and effective magnetic moments
for these ions.
(Ans.: alg, tSt, ulri 0.738, 2.000, 1.250; 3.68, 7.94,
l0.6ps.)
PROBLEM 8.'6
(u) Show that in the vector coupling model one may define
an operator,
i.S:t(J'-Lz -s').
This operator will be used later to define the spin-orbit
operator (see Equation 8.8).
(b) Does this operator commute with J'rL', and 52?
PROBLEM 8.17
(u) For a single d electron, calculate the g values and the

Zeeman splittings in units of ptuQ.
(b) Do the same for a single;f electron.
PROBtErtl 8.r8
(u) Consider a system for which ff,, L', Lu, S', and S,
form a set of mutually commuting operators. What
are the good quantum numbers for the system ? Write
a possible eigenfunction in symbolic ket form. With
these kets as a basis, which operators have diagonal
matrix representations ?
(b) Now assqme that a spin-orbit interaction is turned on
so that the Hamiltonian of the system becomes if -
tr o I n.d ' ^f. Evaluate the commutators, |,tr ,L,]
ls See American Insti.tute of Phlsics Handbook,2nd Ed. McGraw-Hill Book Co., New York, 1963,
page 5-222, Table 59-29.
EIGENFUNCTIONS OF COUPLED ANGULAR MOMENTA 321
and [ ff, S"]. Do ff,L,, and Sz commute ? Are m6

and m, still good quantum numbers ?
(.) Defining J":L., * S,, show that [ ffrJu]:0, where
,ff : tro * in part (b). Can we now claim
that ff, L2, "^d.îas
Sr, Jr, and J, form a set of mutually
commuting operators ? What are the good quantum
numbers for this more complex system ? Write a
possible eigenfunction in ket form. with these kets as a
basis, which operators have diagonal matrix repre-
sentations ?
IO. EIGENFUNCTIONS OF COUPLED ANGULAR

MOMENTA
fn section B we saw how several angular momenta can be combined using
one of the vector coupling models. This approach, augmented by Hundi
rules, enables one to account for many details of the specira of multi-electron
atoms without knowing the exact eigenfunctions of the total angular momen-
tum. However, it is frequently necessary to know these eigenfunctions in order
t-o use a,representation in which the square of the total angular momentum
is
diagonal.
Consider two angular momenta l, and ir, each of which may be either
a spin or an orbital momentum, or a combination of these. i, is defined in a
space of dimension (271 + 1), whereT, is the maximum projection ofjr, that is,
the maximum value of mt. The eigenkets ljr, mr) constiiuti the basis on which
Jf and Jr, are diagonal matrices. likewisifi, resiaes in a space of dimension
(2j, t 1) and the eigenkets lyr, mr) permit a diagonal representation for J22 and
Jr". We now must find out qhat happjns when we define a new spaie by
combining i, and J, to write i : i, I Jr. The new space is called i produrt
space and it has.dimension (2j, + l).(2j, + I). This product space contains
subspaces of dimension Nr, Nr,..., where ifl : 2j + I :2(f, + jr)
+1,
!, :2(i - 1) + | :2(jr +jr) l, -
have assumed thatjr> jr. For example, if7, :3 and ir: i,-.l,r :2L; there
are three subspaces which have dimensions of gr 7, and 5, respectively.
A simpler example, but one of considerable importance, is the case of
coupling two unit angular momenta, that is, Jr. : j, : I. The product space
in this case has dimension 3 X 3 : 9, consisting of the subspaces of dimension
5, 3, and 1. The matrix representations of J2 and J, can be portrayed sche-
matically as follows:
5x5
ilx
9 x 9 matrix
I22 SPIN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
In order to diagon alize these matrices completely we must obtain the simul-
taneous eigenfunctions of J2 : (Jr * Jr)2 and Jr: Jy I Jzu. Our task,
then, will be to express the new eigenfunctions in terms of functions which we
already know, namely, ljrmr), the eigenkets of Jl and Jt", and ljrmr), the
eigenkets of J! and J2". Since the coordinates of J, and J2 are independent in
the old representation, the eigenkets may be written as the product functions,
litirmrmz) - |i'm'.) Iirmr).
These kets form a complete, orthonormal basis so that we may define a pro-
jection operator,
fnyffl2
' \IILL ) (Vtt'*''<i'*'t) - 1' (B'2e)
Let us designate the new eigenkets by jt jrj*), where I
J' lj,jrj*> : j(j + L)h' lj,jrj*)

and
J, ljrjrjm) : mh I
j'jrj*).
The new eigenkets may be expressed interms of the original eigenkets by means
of the projection operator, Equation 8.29. Thus,
lj,j,jm) : (8.30)
mym2
SinceT, and jrare fixed numbers for a specific calculation, the notation may be
simplified by omitting them from the kets. It should be remembered, however,
that the maximum values of m, and m, are j, and jr, respectively, and that the
maximum value of m is J, *Jr. Let us also include the requirement that m
always be equal to the sum ???1 I mz. Then we have
ljm) : (8.31)
mL+rmz:rm
The scalars, (mrmrl j*>, are merely coefficients that tell how much each old
eigenket lmrmr) contributes to the new eigenket ljm). These coefficients are
called the angular momentum coupling coeficients or the Clebsch-Gordon coffiients.
They are not difficult to evaluate for small values ofTr and j2, but the general
expression for them is quite complicated. Many of the Clebsch-Gordon
coefficients are available in tabular form.l3 We will illustrate how they may be
obtained by means of the angular momentum raising and lowering operators
for a few special cases.
' rs See E. U. Condon and G. H. Shortley, op, cit.; G. J. Ntjgh, A. H. Wapstra and R. van
Lieshout, Nuclear Spectroscop2 Tables,North-Holland Publishing Co., Amsterdam, 1969; M. Rotenberg,
R. Bivins, N. Metropolis, and J. K. Wooten, Jr., The 3-j and 6-j S2mbols, Technology Press, Cambridge,
Mass., 1959.
,j t.

ii
By way of illustration, let us consider the case of jt : j, : l;

that is, we
wish to combine two unit angular momenta. Since the space has dimension-
ality of (2 + 1) . (2 + l) : 9, there are nine eigenkets in both the old and the
new bases. The old kets, jrjr, mtmz) - lmrmr) are;
I
'r \'.1
ll, 1), 11,0), 11, -1), 10, l), 10,0), 10, -l), l-1, l), l-1,0), l-I, -l).
(8.32)
The new kets, ljm), which are primed for convenience when 7 and m are
replaced by numbers, are:
r ; /,/,t,t
For ;l_ i
\,
j : 2: 12,2)', '2, l)', 12, 0)', '2, -L)', ',2, -2>'
j : l: ll, l)', ll, 0)', ll, -l)'
j : 0: 10, 0)'.
Note that state |12,2)' - ll, 1), since they both correspond to the case where
tn1 : t7t2: I and mL and mz are parallel. Likewise, lr2, -2)' - l-1, -l).
Thus, we can find all of the states for j : 2 by multiple applications of the
lowering operator on |12,2>' or by using the raising operator on 12, -2>'.
Choosing the former approach, we utilize Equation B.l I and proceed as follows:
J_12,2>', -- (;r_ + Jr_) 11, 1).
\/Wlz,t)':/ml',t)
+/mll,o)
r>' : r) + ll, o)). ;;;
12,
+r00, 1,,r,
l ;r, :;;
:
fiV* -r Jr-)10, t) +
J-12,1)' 11, o))
@12,0),:#t/@t-r,t)
+/mlo,o) rtt rv,r Jt _n,
+/mro,o) l
+rZ6
Jz 'L fir Ji -.rrl
11,-l)l
12,0>' : I
(l-t, l) + 2 10, 0) + ll, -t)),
and in like manner,
-vb
-l)' :
I
12, .r/ru}, -1) + l-1, 0)),
',2, -2)': l-1, -l).

SPIN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
Notice that all of the nine original kets have been used in forming the first
five primed kets. It turns out that all new kets having the same m value will be
linear combinations of the same group of old kets, each of which satisfies the
condition mr * rt2:7ys. We thus choose 11, l)'to be the linear combination
of 10, l) and ll, 0) which is orthogonal to 12, l)', namely,
11, l)' : ir,,o, 1) - 11,0)).
Using the lowering operator we readily obtain:
ll,0)' :#rrr, -1) - l-1, l))

and
11, -l)' :h,,0, -1) - l-1,0)).

The last function, 10,0)', must be a linear combination of ll, -l), l-1, 1),
and 10, 0), and it must be orthogonal to both 12, 0)' and ll, 0)'. If we denote the
coefficients of 10,0)' as (a, b, c), then its scalar products with 12,0)'and ll,0)'
Table 8-6 Clebsch-Gordon Coefticients for h : jz: I
New Ket Old Ket Clebsch-Gordon Coefficient

Um) lmtmr) (mrmrl jms
12,2>' ll, 1) I
12, t>' 10, 1) U \/,
ll,0) tl\/,
12,0>' 11, -l) U \/6
10,0) 2l\/6
l-1, l) Ll\/6
12,
-l>' 10, -l) U\/2
l-1,0) \ \/,
12, -2>' l-1, -l) I
ll, 1)' 10, l) rl\/,
ll,0) -rt\/,
ll, 0)' ll, - 1) It\/,
10,0) 0
l-1, 1) -U \/,
11, - l)' 10, -1) tl\/2
l- t, 0) -rt\/,
10,0)' ll, -l) tl\/3
10,0) -t l\/i
l-1, l) tl\/g
yield the following relations:
t.*2b+d:0
a-c-0.
since these must be simultaneously satisfied we obtain,
a:-b:C.
Therefore,
0)' : I
10,
ilfll' -t) - l0' 0) + l_1, 1)).
We may summarize the results of the above example by tabulating the

Clebsch'Gordon coefficients that we have obtained (re^fer to Equatlon Allt).
PROBLEM 8-'9
Show that the nine kets obtained above are eigenkets of both
Jz - (Jr * Jr)z and Ju: Jru * Jr".
PROBLEM 8.20 ffi#.#',
Write the six state functions ljm) for a single y' electron in
terms of the kets lmt, m,) - l/, m7) ls, m,). (Answers in
Appendix C of this Chapter.)
PROBLEM 8-2'
Show that the spin-orbit coupling factor t .S nray be

represented by the following operators:
t' S : LoS, * *(L*S- * L-S+).
PROBLEM 8-22 ffiffi##.e
Find the matrix representation of the spin-orbit operator of

the previous problem using:
(u) the six states lms, m,) for a p electron as the basis
functions,
(b) the six states j, m) for a p electron as the basis functions.
I
(Answers in Appendix C of this chapter.)

sPtN, ANGULAR MOMENTA, AND IDENTICAL PARTTCLES
PROBLEM 8-23
(u) Write the fifteen kets ljm)for two equivalentp electrons

in terms of the one-electron kets,
lmrmr) - ljrmr) ljrmr).
I I. APPLICATION TO TWO.PARTICLE SPIN

FUNCTIONS
The method of the previous section is readily applied to the task of ob-
taining the spin wave functions for two or more particles. Let us first consider
the case of two identical particles of spin $, where we will now label the spinors
of Equation 8.2 with a I or a 2 in order to distinguish the particles. Thus,
X*0) means that particle I has its spin up and 7-Q) means that particle I has
its spin down. In ket form these spinors are written as lrtrr, ffirr) - lsrfl?rr) lsrmrr),
so that we have for two identical spin f particles:
:
l1 1\
lZt 2/ r+Q) ' x*Q)
l+, -+) : x+(t) ' x-Q) (8.33)
I 1 1\ :
l-2t 2/ x,-(L)'x*Q)
l-+, -+) : )L(r) ' x-(2).
If we consider the total spin operators 52 - (S, * S2)2 and S, : Sr, * Sr,,
it is easy to show that the kets of Equation 8.33 are not eigenkets of 52, although
they are still eigenkets of S,.
PROBLEM 8.21
Show that the operator Sz for two spin $ particles may be

expressed as:
52 : *h' + 2S1,Sz, + (S1+S2- * St-Sr+). (8.34)
PROBLEM 8.25
Given that the kets of Equation B:33 are eigenkets of Sf, S!,
51" and S2,, show that they are not all eigenkets of 52 :
(S' * Sz)'.
Now let us find the spinors ls, m,) defined in the product spin space. This
is a 4-dimensional space containing a 3-dimensional subspace corresponding to
r :1, and a l-dimensional subspace, for s:0. The new kets are 11, 1)',
11,0)', ll, -1)', and 10,0)'. (Primes are included here although these kets
APPLICATION TO TWO-PARTICLE SPIN FUNCTIONS
can be distinguished from those of Equation 8.33.) It should be apparent that

11, 1)' : /+(1) .X*Q) and 11, -1)' : )1*Q).)LQ). To find lI,0)'we oper-
ate as follows:
* sr-)r+ Q) ' x*Q) : x-0)x*(2) + x*Q)x-Q)
0 -+, +) + l*, -+)).
The last ket, 10, 0)', must be orthogonal to ll, 0)'. Therefore, we have:
_ 11 1\
ll,1)' - l?t 2/
11,0)' : i-r(t-+, +) + l+, -+))

I
(8.35)
ll, -1)' _
-
I
f -Et
1
-2/
1\
10,0)' : l
+) - l+, -+)).
ft,(l-+,
The first three eigenkets in Equation 8.35 correspond to the triplet state (s : 1),
while the last is a singlet state (r : 0).
PROBLEM 8.25
The generalization to particles having higher spin values is straightforward.

Thus, for a spin-one particle there are three spin states which may be sym-
bolized as follows:
)(+ : f (uP) : (i) : r)

'|r'
)(o :1(horizon tar) :(il:11,0) (8.36)
X_ : g(down) : (l) : rr' -r)

328 sptN, ANGULAR MOMENTA, AND IDENTICAL PARTICLES
For two spin-one particles there are nine product functions that can be formed
from the functions given in Equation 8.36. These are: )t*0) ' 1G(2), U*(1) '
Xo(2), )GQ)',GQ), Zo(1) 'X*(2), and so on. The appropriate lindar combin-
ations of these product functions which form simultaneous eigenfunctions of
Jz and J, have already been obtained in the previous section for the case of
jr:jr:I.
PROBLEM 8.27 s.i+,q,##
ji{:i;.fi"i8i,,$1s
(u) Write the expressions for the nine basis functions for
two spin-one particles which provide a diagonal
representation for J2 and Jr.
(b) Verify that these functions are the correct eigenfunctions
of J2 and J r.
ffi.t?,Ei8'g;tF,:l{,1,E3.'s.i$i8
12. APPLICATIONS TO SYSTEMS OF IDENTICAL

PARTICLES
Exchange degeneracy and the characteristic symmetry of indistinguishable

particles are responsible for many interesting physical phenomena as diverse
as ferromagnetism, the superfluidity of liquid helium, and the saturation of
chemical bonds. As an example of the role played by the statistics of indistin-
guishable particles, let us consider the states of the hydrogen molecule. Since
the total wave function may be written as a product of the wave functions for
each type of particle, the molecular wave function may be expressed as,
emorecures
: ffi#illTrin)l*oto'. . .
[e(spatial) e(spin)]"r",t,oo, (8.37)
We already have seen that the two-electron wave function must be antisym-
metric in the exchange of the two electrons. Likewise, the two-proton wave
function must be antisymmetric in the exchange of the protons, since protons
are also Fermi particles. Then the possible proton functions have the same
form as the four electron functions given in Equation 8.22. Hence, ?molecule
has 16 differentforms, each of which is antisymmetric in the exchange of two
identical particles.
Let us now suppose that we are performing an experiment that would
detect a change in the spatial state of the protons. For example, we might
be looking at the infrared spectrum of the molecule. Since the interesting
electronic transitions are in the ultraviolet or visible spectral regions, we can
essentially ignore the electronic part of ?morecur" in the analysis of this experi-
ment. The principal contributions to pproron (spatial) are from the trans-
lational, rotational and vibrational states of the dumbbell-shaped molecule.
We will ignore the translational states since we are interested in changes of the
internal energy of the molecules. The vibrational states will be ignored for
APPLICATIONS TO SYSTEMS OF IDENTICAL PARTICLES 329
two reasons. First, they are symmetric in the exchange of the protons; second,
they require rather large excitation energies. * Our molecular wave function
now reduces to
?moleoule : ?protonr(rotation)' ?p.otoo.(spin),
which must be antisymmetric. The rotational states are the spherical harmonics
which are characterized by the angular momentum quantum number ;1,
where
?nrotoos(rotation) : Vt(l)''P,(2),
and
D
-iu + \h'
Lr:7,
z{ being the moment of inertia of the molecule. Since for a dumbbell molecule
the exchange of the two protons is equivalent to an inversion through the
origin, the symmetry of each state is given by the parity, (-l)r. This leads to
the important result (for hydrogen) that only odd aalues of j are allowed when the
proton spins are parallel and onll euen ualues of j are allowed when tlze proton spins are
antiparallel.
Hydrogen molecules which are symmetric in the nuclear spin are called
ortlto hldrogen, while those that are antisymmetric are called para lrydrogen.
Assuming the same statistical weight for all states and a uniform distribution
over the rotation states, there should be three times as many ortho molecules
as para molecules. This follows from the fact that there are three symmetric
spin functions which can be used with each odd rotation function, while only
one antisymmetric spin function can be used with each even rotation state.
This surprising prediction is confirmed experimentally for hydrogen at ordinary
temperatures. The infrared spectra consist of bands of lines corresponding to
odd-odd or even-even transitions, in keeping with the fact that transitions are
forbidden between states of different exchange symmetry (see Equation 8.24),
and each line associated with an odd-odd transition is roughly three times as
intense as each line in the even-even band.
At lower temperatures, however, when &Z becomes comparable with the
energy intervals between rotation levels, the relative populations of the rotation
levels are greatly altered. In fact, the system tends to go completely into the
para state-corresponding to zero rotational energ/ at about 20'K. This
ortho-para conversionf is an amazing verification of the role of symmetry and
statistics in quantum processes. Since the excited states of para hydrogen lie
lower than the corresponding ortho states, the heat capacity at low tempera-
tures is much greater for para hydrogen than for ortho hydrogen or for any
mixture of the two forms. So-called "ordinary" hydrogen is a mixture whose
concentration of para molecules can be as low as 25 per cent at room tempera-
ture and as high as 100 per cent at low temperatures.
* See section I of Chapter 5, particularly Problem 5-18.
f The conversion process is slow except when catalyzed by adsorption on carbon. The carbon
dissociates the metastable ortho molecule; when the atoms recombine at low temperatures they form
the stable para molecule.
Ortho-para processes are not limited to the hydrogen molecule. In

general, if / is the nuclear spin of a homonuclear molecule, then each nuclear
spin has 2I + I possible orientations. Of these, there are 2I * I states that
are symmetric where both nuclei have the same spin projections. In addition,
there are I(21 + 1) possible combinations of two different spin projections, for
each of which there is a symmetric and an antisymmetric function. Then the
ratio of symmetric to antisymmetric states is:
Number ortho states _ (2/ + t) + /(21 + 1) : !|! (8.38)
Number para states I(21 + 1) I
For the hydrogen molecule , I : $ and the ratio of ortho to para states is 3 :1
as we obtained above.
A further example is provided by the deuterium molecule. Although the
deuteron is composed of two fermions-a neutron and a proton-it is a spin 1
boson. The deuteron does not exist in the spin-zero state. From Equation
8.38, there should be twice as many ortho states as para states. The spin
functions for this case were obtained in section 10 (see Table 8-6). It is evident
that there are six symmetric spin functions corresPonding to S : 2 and,S : 0;
the three functions corresponding to S : I are antisymmetric.
Since deuterons are bosons, the total wave function for the deuterium
molecule must be symmetric. Using arguments similar to those for hydrogen,
we discover that the ortho states are associated With the even rotation states
and para states are associated with odd rotation states. Accordingly, spectral
lines corresponding to transitions between even rotation states have twice the
intensity of those corresponding to odd rotation states. At very low tempera-
tures, deuterium converts to the ortho form for which the zero rotational state
is allowed.
Although a rigorous discussion of the superfluid state of Hea is beyond
the scope of this book, it should be noted that the differences in behavior of
liquid Hes and Hea are due to the f;act that the former consists of Fermi
particles and the latter of Bose particles. The repulsive exchange forces
prevent any further condensation of liquid Het, and accordingly it behaves
like an "ordinary" liquid. However, in liquid Hea the exchange forces are
attractive, and at 2.18'K a transition occurs to a lower state which no longer
contributes to the entropy or the specific heat. This is the state that is known
as the superfluid state.
PROBLEM 8.28
(u) Write the relative energies of the eight lowest rotation

states for a diatomic molecule.
(b) If the allowed transitions are given by Li : *2,
what are the relative energies of the lines of the emis-
sion spectrum ?
(") How would you expect the heat capacities of ortho,

para, and "normal" hydrogen to compare ?
#;:: T:;i:ii:".: ll -"."';:ll;;t:it.iltt;*1*lf**:::i4i li::;;:ti;;:jflI;:;;T"**?:$i#*{s.iflilggg

SUMMARY 33t
PROBLEM 8-29
Apply the particle exchange operator to the nine kets for two
spin-one particles (see Table 8-6 in section 10) and show
that the ratio of the number of ortho to para states is 2 :1.
PROELE'T4 8.30
Describe the characteristics of the rotation spectrum of

oxygen molecules composed only of 016 atoms.
PROBTEM 8.3I
Explain how one can easily determine the spin of a nuclear

species provided that it forms diatomic molecules.
SU'YIMARY
In this chapter we have seen how spin can be incorporated into the
Schrodinger formalism by postulating spin operators which act upon spin
functions in a manner analogous to the angular momentum operators. When
the spin and spatial coordinates are independent, the total wave function is a
product of the spin function and the solution to the spin-independent Schrodin-
ger equation. If any spin-dependent forces are to be included in the Hamil-
tonian, they must be expressed in terms of the spin operators. In particular,
the spin-orbit interaction is discussed and it is shown that it can be expressed in
terms of the operator scalar product t' S. The vector coupling models are
then introduced and applied to simple systems. The product space formed by
coupling angular momenta is described and a procedure given for obtaining
the eigenfunctions in the new space.
The concept of the exchange degeneracy of indistinguishable particles was
introduced. It was shown that a satisfactory wave function for two or more
identical particles must be either a symmetric or an antisymmetric linear
combination of the possible product states. This derives from the requirement
that the wave function be an eigenfunction of the particle exchange operator
and that its probability density remain unchanged under an exchange of
particles. The experimental connection between particle spin and the ex-
change symmetry of wave functions is summarized by the statements: a
collection of bosons (identical particles having zero or integral spin) is de-
scribed by u wave function that is symmetric under the exchange of any two
particles; a collection of fermions (identical particles having half-integral spin)
is described by a wave function that is antisymmetric under the exchange of any
two particles. Fermions tend to aaoid the same quantum state, while bosons
tend to condense into the sarne quantum state. Furthermore, the energy density
of a collection of fermions is greater than that for a collection of the same
number of bosons. It is in this sense that bosons attract one another and fer-
mions repel each other via an exchange force which has no classical analog.
These concepts are shown to account for the infrared spectra of diatomic homo-
nuclear molecules and for the electronic exchange energy which separates singlet
and triplet states in atomic spectra.
CHAPTER I APPENDICES A to C
APPENDIX A. THE DISTRIBUTION FUNCTIONS RESULTING

FROM QUANTUM STATISTICS
The quantum statistics associated with the two species of identical
particles provide us with two new distribution functions to replace the classical
Maxwell-Boltzmann distribution function discussed in Chaptet 2, Appendix B.
Recall that in that derivation the quantity gn was called the intrinsic probability
of the ftD state. We will now interpret the ith cell as the energy level en and will
regard .g; as the fugeneracJ of that level. If we also invoke the Pauli exclusion
principle: we constrain the number of Fermi particles in an energ'y level to a
value that is either less than or equal to the degeneracy. That is, for fermions,
n; !8*
Then the total number of ways of filling g, states of energy eo with n, fermions is
simply the number of combinations of g, identical states taken nn at a time,
gt!
ffit,
and
W-TTffi
This is the distribution given in Problem B-32.
For bosons the same combinatorial argument cannot be used, since a
state is no longer simply occupied or unoccupied but can contain any number
of particles. Using a somewhat different approach, the distribution of Problem
8-33 is obtained.ra
A more sophisticated method of deriving the distribution functions is to
use the grand partition functionls
Z_ 2 e-nto,
d
(A1)
where n: (ru + ),')lkT. Then, for fermions,ni can be only 0 or l, so
Z-l*e-",
1r R. B. Leighton, Pinciples of Modern Ph2sics, McGraw-Hill Book Co., New York, 1959,
Chapter 10.
16 T. L. H:ill, An Introduction to
Statistical Tlurmodymmics, Addison-Wesley Publishing Co., Reading,
Mass., 1960, section 22-1.
333
and the occupation number of fermions :per state is
: - ln7,: e-d I
tlo:-kT*t" Z:-a iA2;
dAr anlnz- r + -': e'+l'
Defining the Fermi energy by .o .1r, this becomes
I
fi;:F' (A3)
.*p\ l,r )+r
For bosons the grand partition function becomes
I
Z- Ia s-o * e-'o * e-"' + "' : L-e*
Then,
n _*hZ:ft.:*:#
rti
-
-
-.
(A4)
'*P\ rr )-L
Using the method of Lagrange's multipliers (See Appendix

B of Chapter 2), obtain the Fermi-Dirac distribution function
from the probability function
14/:ff-r- nr!(g,
',3n! .'.
- nn)!
PROBIEM 8-33: ]ri,! li;i;j:;rriill.;t4' .ijre'iJiir.i;tu.{'ii i'.i;i+.dur'9'li"ijlili3, :;',ri.xie,rll,r+.i*ill1
In like manner, obtain the Bose-Einstein distribution function

from the probability function
tr[-:nffi
_ _ l)! (_ gi
APPENDICES A TO C 335
APPENDIX B. GROUND STATE ELECTRON CONFIGURATIONS

FOR THE ELEMENTS16
Shell:f K LMNOPe Ground

Element
lt 234s6i State
Subshell: s sp sp d sp d f sp d f sp d s Term
I H hydrogen 1 ,s*
2 He helium 2 tSo
3 Li lithium 2 I ,s+
4 Be beryllium 2 I tSo
5 B boron 2 2r ,P+
6 C carbon 2 22 3D
rO
7 N nitrogen 2 23 aSr
B O oxygen 2 24 rP,
9 F fluorine , 2s ,P*
l0 Ne neon 2 26 tSo
I I Na sodium , 26 l ts+
12 Mg magnesium 2 26 2 tSo
l3 Al aluminum 2 26 2t ,Pt
l+ Si silicon 2 26 22 3D
r0
l5 P phosphorus 2 26 23 nsg
16 S sulfur 2 26 24 3D
L2
17 Cl chlorine 2 26 25 ,Pg
lB Ar argon 2 26 26 tSo
l9 K potassium 2 26 26. I ,s+
20 Ca calcium 2 26 26 2. tSo
2l Sc scandium 2 26 261 2 ,D*
22 Ti titanium 2 26 262 2 rF,
23 V vanadium 2 26 263 2 nFs
24 Cr chromium 2 26 26s I rs,
25 Mn manganese 2 26 26s 2 usg
26 Fe iron 2 26 266 2' uDn
27 Co cobalt 2 26 267 2 aF"
28 Ni nickel 2 26 268 2
2
tFn
29 Cu copper 2 26 2610 I ,sl
30 Zn zinc t 26 2610 2 tSo
3l Ga gallium 2 26 2610 2r ,P+
32 Ge germanium 2 26 2610 22 tPo
33 As arsenic 2 26 2610 23 ns*
34 Se selenium 2 26 2610 24 tP,
35 Br bromine 2 26 2610 2s ,P*
36 Kr krlpton 2 26 2610 26 tSo
37 Rb rubidium 2 26 2610 26 I ,s*
38 Sr strontium 2 26 2610 26 2 tSo
39 Y yttrium 2 26 2610 261 2 ,Dg
40 Zr zirconium 2 26 2610 262 2 rF,
4l Nb niobium 2 26 2610 264 I uD+
42 Mo molybdenum 2 26 2610 265 I ts,
43 Tc technetium 2 26 2610 266 I tst
44 Ru ruthenium 2 26 2610 267 I uFu
45 Rh rhodium 2 26 2610 268 I nFg
46 Pd palladium 2 26 2610 2610. tSo
47 Ag silver 2 26 2610 2610 I ,S*
48 Cd cadmium 2 26 2610 2610. 2 tSo
49 In indium 2 26 2610 2610 2t ,Pb
50 Sn tin 2 26 2610 2610 22 tPo
5l Sb antimony 2 26 2610 2610 2g nSt
52 Te tellurium 2 26 2610 2610 24 rP,
Amelnc,an Institute of Ph\sic.s Handbookr 2nd ed., McGraw-Hill Book Co., New york, 1963,f"StiZi-i',,
p.7-14.
APPEN DIX B. (continued)
Shell:j K L l{t NO P e Ground

z
ltz3 45 6 7 State
Element Subshell:s sp spd spd f spd f sPd s Term
53 I iodine 2 26 26r0 2610. 25 ,P*

54 Xe xenon 2 26 2610 2610. 26 tSo
55 Cs cesium 2 26 2610 2610. 26. l ,s+
56 Ba barium 2 26 2610 2610. 26. 2 tSo
57 La lanthanum 2 26 2610 2610. 261 2 ,D*
58 Ce cerium 2 26 2610 2610 I 26t 2 tHo
59 Pr praseodymium 2 26 2610 26103 26. 2 nI*
60 Nd neodymium 2 26 2610 26t04 26. 2 urn
6l Pm promethium 2 26 2610 26105 26. 2 6Ffe
62 Sm samarium 2 26 26r0 26r06 26. 2 rs,
I
63 Eu europium 2 26 2610 26t07 26. 2 tst

64 Gd gadolinium 2 26 2610 26t07 261 2 ,D,
65 Tb terbium 2 26 26r0 26109 26. 2
66 Dy dysprosium 2 26 2610 261010 26 2
67 Ho holmium 2 26 2610 26l0ll 26. 2
6B Er erbium 2 26 26r0 26t0t2 26. 2
69 Tm thulium 2 26 2610 261013 26. 2 ,Fg
70 Yb ytterbium 2 26 2610 26r014 26. 2 tSo
7l Lu lutetium 2 26 26lp 261014 261 2 ,D*
72 Hf hafnium 2 26 2610 26t014 262 2 rF,
73 Ta tantalum 2 26 2610 261014 263 2 nFt
7+ W wolfram (tungsten) 2 26 2 610 261014 264 2 uDo
75 Re rhenium 2 26 26r0 261014 265 2 tsg
?6 Os osmium 2 26 2610 261014 266 2 uDn
77 Ir iridium 2 26 2610 261014 267 2 nFi
78 Pt platinum 2 26 2610 261014 269 l rD,
79 Au gold 2 26 2610 261014 26r0. I ts+
80 Hg mercury 2 26 2610 261014 2610. 2 tSo
8l Tl thallium 2 26 2610 261014 26r0 2t tP+
82 Pb lead 2 26 2610 261014 2610 22 tPo
83 Bi bismuth 2 2'6 2610 2610t+ 2610 23 ns*
84 Po polonium 2 26 2610 261014 26r0 2+ 8f,'
Lz
85 At astatine 2 26 2610 261014 2610 25
86 Rn radon 2 26 2610 261014 2610 26 tSn
87 Fr francium 2 26 2610 261014 2610 26. t tsr
88 Ra radium 2 26 2610 261014 26r0. 26.2 tSo
89 Ac actinium 2 26 2610 261014 2.610 26t 2 ,Dt
90 Th thorium 2 26 2610 261014 2610 262 2 tFt
9l Pa protactinium 2 26 2610 261014 2610r 262 2
92 U uranium 2 26 2610 261014 26r03 2612 uL.
93 Np neptunium 2 26 2610 261014 2610+ 2612
94 Pu plutonium 2 26 2610 261014 26r06 26.2
95 Am americium 2 26 2610 261014 26107 26 2
96 Cm curium 2 26 2610 261014 26107 26r 2
97 Bk berkelium 2 26 2610 261014 26108 2612
98 Cf californium 2 26 2610 261014 261010 26.2
99 Es einsteinium 2 26 26r0 261014 2610 ll 26.2
100 Rm fermium 2 26 2610 261014 261012 26.2
l0l Md mendelevium 2 26 2610 261014 261013 26.2
102 No nobelium 2 26 2610 261014 26r014 26.2
103 Lw lawrencium 2 26 2610 261014 261014 261 2
APPENDICESATOC 337
APPENDIX C. ANSWERS TO SELECTED PROBLEMS
The results of these problems will be ne eded for problems in later chapters.
PROBLEM 8.20
The new kets U, m) are primed for convenience; the

unprimed kets are lm1, mr).
l3' 3)' : 11, +)
l+, +>' : +) + 11, -+))

+Urlo,
l+, -t)' : jrU-l, +) + \/r lo, -+))
l+, -g)': l-1, -+)

l+, +>' : +) - frll, -+))
+(lo,
l+, - i)' : jrt\/, l-1, +) - lo, -+)).
PROBLEM 8.22
(u) On the basis Il, +), ll, -+), 10, +), 10, -+), l-1, +),
- 1, -+),
l: _? ;
l
: ::\
' :;(l 'l
_4
s h2
l ,,'4'-:il
(b) On the basis ordered as in Problem 8-20 above,
,s :r(
;,, :,)
PROBLEM 8.27
See Table 8-6 in section I0.
SUGGESTED
REFERENCES
David Bohm, Quantum Theorl. Prentice-Hall, Inc., New York 1951.
E. U. Condon and G. H. Shortley, The Theorlt of Atomic Spectra. Cambridge
University Press, Cambridge, 1953.
Wesley Publishing Co., Inc., 1960.
P. A. M. Dirac, The Principles of Quantunt Mechanics, 3rd ed. Oxford University
A. R. Edmonds, Angular Momentum in Quantum Mechanics. Princeton University
Press, Princeton, 1957.
R. M. Eisberg, Fundamentals of Modern Ph2sics. John Wiley and Sons, fnc.,
New York, 1961.
W. R. Hindmarsh, Atomic Spectra. Pergamon Press, New York, 1967.
R. B. Leighton, Principles of Modern Phltsics. McGraw-Hill Book Co., Inc., New
York, 1959.
E. Merzbacher, Quantum Mechanics. John Wiley and Sons, Inc., New York,
r961.
Albert Messiah, Quanturn Mechanics. North-Holland Publishing Co., Amster-
dam, 1958.
D. Park, Introduction to the Quantum Theor2. McGraw-Hill Book Co., Inc., New
York, 1964.
J. L. Powell and B. Crasemann, Qunntum Mechanics. Addison-Wesley Publish-
ing Co., Inc., Reading, Mass., 1961.
M. E. Rose, Elementarlt Theor2 of Angular Momentum. John wiley and sons,
Inc., New York, 1957.
D. S. Saxon, Elementarlt Quantum Mechanics. Holden-Day, Inc., San Francisco,
1968.
Leonard I. Schiff, Quantum Mechanics, 3rd ed. McGraw-Hill Book Co., New
York, 1969.
John C. Slater, Qunntum Theory of Matter. McGraw-Hill Book Co., New
York' 1968.
H. E. White, Introduction to Atomic Spectra. McGraw-Hill Book Co., New York,
1934.
Robert L. White, Basic Quantum Mechanics. McGraw-Hill Book Co., New
York, 1966.
CHAPTER 9
ewwffiffiKMM&wffiffiw
ffiffiffiw##ffiffi$ effiffi
&ffipkrffiewt&ffiffi
It may come as a surprise to the reader to learn that we have already

treated most of the physical systems that have exact solutions. Our repertoire
consists of the solutions for the harmonic oscillator, the rigid rotator, the
hydrogen atom, and the bound states of a particle in a rectangular well or box.
Even here we have begun to detect discrepancies due to additional interactions
such as the spin-orbit and spin-spin interactions. Complications may also
arise when an electromagnetic field is present. A further difficulty appears
when additional particles are added to the system. In such cases we have
obtained an exact solution only for the unrealistic case of completely non-
interacting particles. This might be a good approximation if the only inter-
actions were the weak intermolecular forces; but if the particles are charged,
the addition of the Coulomb forces could have a tremendous effect upon the
nature of the solutions.
If we are to make further headway in the apBlication of quantum me-
chanics to real systems we will need to develop techniques for arriving at
approximate solutions. The most widely used of these approximate methods for
bound systems is perturbation theory.l
I. PERTURBATION THEORY FOR STATIONARY

STATES
The easiest and most direct approach for solving a real physical system is
to regard it as a modification of one of the model systems whose solutions we
already know. In particular, we wish to express the Hamiltonian for the real
system, ff, as the sum of .f o, the idealized Hamiltonian whose solutions we
know, and an additional part Jt'ftt which contains the new interactions. If
it'o, is small comparedto ff0, the corrections to both the eigenfunctions and
I E. Schrddinger, Ann. Physik 80, 437 (1926).
339
APPROXIMATION METHODS AND APPLICATIONS
the eigenvalues resulting from Jt'o, will also be small. In such cases it is possible
to use the methods of perturbation theory.
We begin by writing the Hamiltonian for the real system as
./f -- ffo * lJt'(t), (e.l)

where ,ffois the part that can be solved exactly and ,ff(r)is regarded as the
perturbation. The parameter ,1is not fundamental but is added for convenience
in "turning on" and "turning off" the perturbation. It is restricted to values
between 0 and 1, the former corresponding to the ideal Hamiltonian, and the
latter corresponding to the physical problem to be solved.
The sets of known functions and eigenvalues are given by
trorplot : Elo'ylo', (e.2)
where the rpjo) form an orthonormal set such that (i l.t) : dor. The equation we
wish to solve is
(tro * ldf ttt)rln : E,gn. (e.3)
It is assumed in this section that both the rpjo) and the q)n are discrete, non-
degenerate sets of time-independent functions. We further assume that there
is a one-to-one correspondence between the members of each set. Thus, rp,
goes into ?lo) us ,1 varies from I to 0, Vz + yft, etc. Stated mathematically,
lim y* : y,l:t and lim E, : Ef' (e.4)
In keeping with the spirit of these limits, let us represent the new eigenfunctions
and eigenvalues by the following power series in )";
:
tpn vlf) + tvll' + |ry,l?' + . . .
(e.5)
En : Ell' + lEf' + ffEf) +'''
It is assumed that successive terms of these series get smaller so that the series
converge. Our task will be to find these corrections to the eigenfunctions and
the eigenenergies to any desired order. Substituting Equation 9.5 into Equa-
tion 9.3, we obtain:
(tr, * lffttt)(yl,o' + lrp'"t' + lrrp(,i) + ...)

: (Erc' + lEP a ),zEft + . .)(ql,o' + lrpl-t) + lrrtl,') + . . .).
trorpll) + l(iforvrct * trorpll)) + l,(trotp*t * ,ff<'trot, +...
-- Elf'vf' + + nf',pt')
^(El!"tll'
+ 1, (Ef, rp'," + ELt' rp'"t' * E!i, rpli,) + . . .
Ot,
(trorpll' - El!'rpll') + i( i/f$\v(ot * troy,ll' - El,"rprct - E!f'rpl)')
+ |r(trorpllt * .ff<'trft) - El!,rpl"r, - Ef,rp*, - Ef,rpl,o,) +...- 0.
PERTURBATION THEORY FOR STATIONARY STATES 3{l
Since the parameter 2 is arbitrary, the coefficient of each power of ,1 can be

set equal tozero. Therefore, we have the equations:
tr orpll) : El!'rp'f'
*tttt r<ot * ffi oy,*t : EllrPl,o' + Elf"pll' (e.6)
tr rrp'," +,ff"' rttll' : E!,1' rpl!' + Et' rpt' + n!," tplf'
:
The first of these is simply Equation 9.2, our starting point. The second may be
solved by expanding the first-order correction to the wave function in terms of
the unperturbed eigenfunctions. That is,
,Pll' :|o,ny,lo'.
Then the first order equation becomes:
at'i) vrc' * ff o 1 o*oy,lo' : Eli"plf' + E!f'

l o,orylo' .
Multiplying by glo)- and integrating,
(kl ffttt ln) * troli) : Ef)$lr) + > aniEl!'<kli>.

\o"n(kl
(kl ytrt ln) -f o*nDlo'6t t : Ell'6o* + > anoEf) 6ro,
7
(kl ffrtt ln) * o*rE[ot : El]t 6o* + o,*E!]'.
The quantity (kl lf trt lz) should be recognized as the matrix element, "tr\\,
of the operator Jf o) with the unperturbed eigenfunctions as the basis. Then
we may write,
,tt') * a*o(Eft - Elf') : Elnil 6rn. (e.7)
For ft : n, Equation 9.7 gives us the important result,
E!*t, : ,trL\. (e.8)
In words, Equation 9.8 says: tltc frst order correction to the energ) of the nth eigen-
state is the diagonal matrix element corresponding to tlte nth row and tlte ntk column of the
matrix of the perturbation. Note that the unperturbed wave functions are used for
this calculation. For ft t' n, Equation 9.7 becomes
of,
,trf) + o,*(El,') - Eto)) - 0,
,trf)
Qntc:Mr. (e.e)
342 APPROXIMATION METHODS AND APPLICATIONS
Equation 9.9 tells us how to find the first order corrections to the wave function
for the nth state. Thus,
Ipn: rplf' + k+n

Zo**rpf)
IPn:,ift +ffi,'pfo)+ ffi,'pl;" +"' (e. r0)
Each function that has a nun-zero matrix element with Vft contributes to the new
basis
waae function for the nth state. Hence, when the matrix of the perturbation has
non-vanishing, ofldiagonal elements, this means that the perturbation has
produced some interference or mixing of the original wave functions so that
they are no longer orthogonal. Note that the amount of this mixing depends
not only on the magnitude of the matrix element in the numerator, but also on
the "energy denominator." Thus, states that are widely separated in energy
are not expected to interfere to any large degree. States that are close together
will be expected to have a large amount of mixing. The denominator of
Equation 9.9 can never be zero, since we have excluded the possibility of any
degeneracy in this derivation.
To obtain the second-order corrections we will use the last expression
given in Equation 9.6,
tr orp*t * "f ory!,t' : Elf' rp'*" + El"t' rpli' + E'^" rplf' .
Once again we utilize the expansion postulate and write
!,llt : 2 b^ftt\ot,
i
as well as our previous assumption that
,pflt : o*orp[o'.
7
Making these substitutions,
tro2b*,rt'f' * .ffttt 2o*orplo' : Ef'Zb,,rp'n" + Ef' Zo*np'n" + El"'trplf'.

Tiii
Multiplying by g[o)* and integrating:
Zb*,(kl ,trou> + > ann(kl 77'ttt li)

in
ii
2b*oE'ftdr,
niii
* 2onotrL\'
b**El!'+ f amffl,|.' : bnoElft + on*Ell' + E'n"6on,
ofr
b,*(El!' - Ef') + > o*otrL\' - an*Efl' : Ellt6*n. (e.1 1)
PERTURBATION METHOD TO NON.DEGENERATE STATES 343
For & : n, Equation 9.11 gives the second-order energy correction as
Ell' :2ono,frt) - o,nEfll

i
i
= 2 eno'trr')'
i+n
Then,
r_(2) s
un trr*t, <1 l,trtt)|,
tr:L,
(e.12)
- o?nEf' - E\ot - o?"EY' - E\ot '
H.:
where the results of the first-order theory expressed by Equations 9.8 and 9.9
have been used. The second-order correction to the energy has been obtained
using only the first-order wave function as expressed by the expansion coeffi-
cients arro.
To obtain the wave function to second order, we return to Equation 9.11
and seek solutions for k t' n. Then,
and,
bnt:2 (Ef, #:1, trli, trLl ,trf)

(e.13)
d*n - E;0,) (Ef, - E;o,) @EfJ,f\,,
where n + k.
It
is evident that the corrected wave functions rapidly get quite compli-
cated. To second order, the nth eigenfunction is
Tn: rplf' +le+n

2o,*!tft + > b,,,!,|!t,
1
(e.14)
where the expansion coefficients are given by Equations 9.9 and 9.13. For this
reason, perturbation theory is rarely used for wave function corrections beyond
the first order and for energy corrections beyond the second order. In principle,
however, higher order corrections can be obtained by proceeding to solve, in a
similar manner, the equations corresponding to higher powers of ,1 as in
Equation 9.6.
2. APPLICATIONS OF THE PERTURBATION METHOD

TO NON-DEGENERATE STATES
As our first example, let us determine the effect of an anharmonic term
on the ground state energy of a harmonic oscillator. The unperturbed Hamil-
tonian is
ffo:**tkx', (e.15)
with energy eigenvalues given by Elf' : ah.(n + +). The unperturbed energy
eigenfunctions of this Hamiltonian are tabulated in section 7 of Chapter 5.
Now if we add.an anharmonic term of the form bxs to the potential, the
Hamiltonian becomes
itr : & * ikx, * bxs : tro * it2(t),
2m
where i($t : bxs may be regarded as a perturbation if D is small. Then the

first order correction to the ground state energy is, from Equation 9.8,
E[t, : .tr|tr, : (?[o)l bx, ,rto, ) : 0.
This result follows immediately from parity arguments without performing the
integration. Since each oscillator state has definite parity, the square of any
state has even parity. For an odd perturbation, then, the integrand is odd and
the integral over symmetric limits will vanish.
When the first-order perturbation vanishes, it is necessary to proceed to
second order in order to find the energy correction. A glance at Equation 9.12
tells us that we are going to need the ofldiagonal matrix elements
lfLl' : @l bxs li).
It will be shown below that when (n i 'is the unperturbed ground state wave
function there are only two non-vanishing matrix elements, namely,
3b
ff["t :
t/Bu"
and
b
,ff[tt = 13
tVo.r'
' Then,
ll
83":- Busafi.
LT b2
Buzk'
b2
and
ll b2
Eo:irh- B"'k'
PROBLEM 9.1 :"*1. -: ; l"i;;. i; : : :. :" ll,"l;: f ,:l lliri; ;;:,:i :il1lt; i;,ii ;: :i i:1 J:ri
Verify the above values for ff[t' and ,tr[\' by direct

integration.
We will now use the raising and lowering operators developed in Chapter
5 to obtain the matrix of the perturbation. This will enable us to simply write
down the energy correction for any level without performing the integrations.
Recall that the unperturbed Hamiltonian for the harmonic oscillator can be
PERTURBATION METHOD TO NON.DEGENERATE STATES 345
written in the forms (see Equations 5.43 and S.+S),
ffo:p'+Q,:aaI aata, (e. r6)
where q : \/; x, P : -dQlAq), and u : amlh. From Equation 5.58 we have
q:#@ * oI),
I
q" : + (ot)'+ a'aI I ata2 * a(al)z + (al1za I aata I aIaaIl.
2tilo"
(e.1i)
It is readily shown that:
{ap' I q'r) : fM6n,,n+s
+f@6n,*_s (9.18)
* 3(n + 1) fn + | 6n,,n+L * Snfi A*,,,_r.
PROBLEM 9-2
verify Equation 9.18 with the aid of the operator equations,
alln>:'y'n+llnll),
and
a ln) : \/; ln - I).
We can readily construct the matrix tlt\ from Equation 9.18, srnce
"*t:'" : (n,l bxs ln, : @,1 q" ln).

#
Thus:
0 3 0 f6 0 0
o
J o 6f, o 2{6 o
0 6f, o sfz o 2fts
(trt,"):+ {6 o efi o 2+ o
(e.le)
t/ BoLg 0 2l/602+ors./s
0 o 2fts o Lsfs o
PROBLEM 9.3
Using the matrix elements given in Equation 9.19, calculate

the second-order energy shifts in the first two excited states
of a harmonic oscillator due to the perturbation Dx3. Would
you expect perturbation theory to be valid for states cor-
responding to large z values ?
PROBLEM 9.1
Write the wave functions corrected to first order in the

perturbation bf for the ground state and the first excited
state of a harmonic oscillator.
PROBLEM 9.5
Find the first-order corrections to the energies of the ground

state and the first twd excited states of a harmonic oscillator,
arising from the perturbation 3f
(lt : cx4.
(n"r.: E[') :#; E{':#; Et":#)

PROBLEM 94
Using the raising and lowering operators for the harmonic

oscillator obtain an expression analogous to Equation 9.18
for the matrix elements, (n'l f ln).
PROBLEM 9.7
From the result of the previous problem, construct the matrix

tr9*.
PROBTE/I4 9-8
Use the matrix of the previous problem to obtain the ener-

gies of the ground state and first excited state of the harmonic
oscillator corrected to second order in the perturbation
PERTURBATION METHOD TO NON-DEGENERATE STATES 347
ift) : cx4. Would you expect perturbation theory to be

valid for states corresponding to large z values ?
Ans.: Eo : $ah *#(, - 2"'*h

7c
llc
E, - ,Lwh *#,(' 2uzah)
PROBLEM 9.9 ;*a,:,;auil:i,l,ij+jl.ii ljjirlnji;ili:i,;i,i,.1i.tji..1iri1i1'lijiii!.1.j' 1$:1i.*,ij,:1;1,1i!f1,
Consider the infinite well having a V bottom as shown in the

figure. e is small, the quantity ytrt: (rlo)lrl may be
If
regarded as a perturbation on the square well potential in
the regiort -a 1x 1a. Calculate the ground state energy

and wave function correct to first order in this perturbation.
(Three or four terms are sufficient for the wave function.)
PROBLEM 9.IO
Calculate the first-order correction to the ground state

energy of a harmonic oscillator due to a perturbation of the
form Jf$ : t)dfn'.
(n"r.: EJ'):r(#)*)
PROBLEM 9.' I
An electron is performing simple harmonic motion in one

dimension. If a weak electric field I is applied in the
3{8 APPROXIMATION METHODS AND APPLICATIONS
direction of its motion, find the shift in the energies of

the ground state and first excited state in the presence of
the electric field (the Stark shift).
/ ,^\
(e"*.: E[') - Eltt - 0; Et" : El" : -; e282 \)
PROBLEM 9-T2
Use Equations 5.41 or the method of Problem 5-25 to find

the wave function correct to first order for the nth state of the
charged harmonic oscillator in an electric field discussed in
the previous problem.
/eE-,-\- \/rl,
(e"r. i en: ln) * #W-*l\/n + ln *L> -
r - r>1.)
Show that the potential, V : tkx' - e8x, of the two pre-

vious problems can be solved exactly. (Hint: complete the
square.) Find the energies of the system and interpret your
results.
Derive the energy shift of a harmonic oscillator to second

order in the perturbation bxs for the nth level,
E!," : -+
d"'tt
(3onz * 3on + 1 1).
(u) Calculate the relativistic correction for the ground

state of hydrogen by means of the perturbation term
leading to Equation 8.12,
it'$) : -w"
hn on
(b) Do the same for the 2s state of hydrogen' (Hint:

make use of the formulas given in Appendix A of
Chapter 7.)
INTERACTIONS OF A CHARGED PARTICLE 349
3. INTERACTIONS OF A CHARGED PARTICLE WITH

STATIC ELECTRIC AND MAGNETIC FIELDS
The classical Hamiltonian for a particle of mass m and charge g in the

presence of static electric and magnetic fields is given by
i€:*(r_ iul +v*qe, (e.20)
where l is the ordinary momentu m, i is the vector potential , V isthe mechanical

potential energy, and @ is the electrostatic potential. The potentials and fields
are related by the equations,
+ + 1 AZ
9:ix.Z |-":'.
and 8: -VO- cdt (e.21)
+
Since the potentials (D do not uniquely specify the fields / and 0, we
and, ,i
are at liberty to make an additional assumption relating them. For con-
venience, we assume that
i.A+1*:0,
c0t
(e.22)
which is known as the Lorentz condition. This choice arises quite naturally
during the simplification of Maxwell's equations.
Substitute Equationg.2l into Maxwell's equations and show

where the Lorentz condition arises. How does it simplify
the field equations ?
The Hamiltonian (Equation 9.20) may be readily derived from the

Lorentz force,
I *-1
*;to " q),
f*
F:rlu (e.23)
where d is the velocity of the particle. Substituting Equation 9.21 into 9.23,
F: a{-vo -:W, -i x (v ' r)]} p2+)
Using a standard vector identity and regarding the components of d (as

well as those of i) as independent variables, we have
V(a. A1 : 1o.i1i * ri x (V x 4. (e.25)

It can also be easily shown that

(o.i)t:# -f,. (e.26)
Making use of these relations the Lorentz force becomes,
F : -v(ro -!6 4 - 14r, ,
where we interpret the quantity in parentheses as a generalized potential

function U. Then the Lagrangian for the system isz
L: T - LI : *m6' - qA +!O'A Q.27)

C
: *(*o +t,^l _ #.A, _ qa.

From Equation 9.27 it follows that the generalized momenta are: '
ln:Y*:m*+gA'
ALo
lu:6:mi+!Au
lu:Lun:m2+!Au'
where it is assumed that the components of i are not velocity dependent. Then
we can write
O-
F:md*zrq,'. (9.28)
Thus, the' canonical momentum is now composed of the ordinary linear

momentum plus a contribution from the magnetic field.
Verify Equations 9.25 and 9.26.

,i.:i#ri!:i+rtf!i!..i .i.1tiTitib.l,ii,F,,e;1i.,
The Hamiltonian function is defined as
i/f - 2pnqo - L - F. i - *(*o +: + #A, + qQ,

^f
2 H. Goldstein, Classical Mechanics, Addison-Wesley Publishing Co., Reading, Mass., 1950.
INTERACTIONS OF A CHARGED PARTICLE 35I
wherein we substitute the expression for d from Equation 9.28, namely,
o:!(
- m(o -r.r\
c'- l'
Thereupon, the Hamiltonian simplifies to
'ff : *(, -:'f * qe, (e.2e)
plus any mechanical potentials which may be present. Note that the kinetic
energy term for a particle in a magnetic field can be obtained by merely
substituting the quantity (p - qlc i) for p in the expression p2l2m.
In order to utilize the above Hamiltonian in quantum mechanics, we will
regard f and i as operators and will replace p with -ihV. It is therefore
useful to know under what conditions y' and ,Z commute. Since each com-
ponent behaves as follows,
lA*, f nf - -ih lr e,,

then
li.F,F.Al:ihi.l. (e.30)
Referring to the Lorentz condition, Equation g.22, we note that V . ,4 :0

wlteneuer the electric feld is constant in time. Under this circumstance p and .tr
commute and the Hamiltonian may be written,
' * :#* qa -#,F.i *#i,. (e.31)
It is common practice to represent a uniform magnetic field by the vector

potential,
A:-gixd. (e.32)
To show that this satisfies our original definition of the vector potential,
Equation 9.2I, we operate as follows:
Vxi:-+Vx(;xA)
: -*ti(V.6 + (d.iii - dF.i) - fi.Vdl.
Since the divergenceof 6 is always zero, the first term vanishes. The second
term is(d.Vli : A. Since the divergence of iis 3, the third term becomes
34. The last term is zero because 6 j, u constant vector. Thus,
ixi:-+@-s4-d,
as required. Equation 9.32 must also satisfy the Lorentz condition:
i ..4: - ilA.V x i - i x V. hf :0,

since the curl of i is zero and the divergence of d is zero. Substituting this
vector potential into the Hamiltonian (Equation 9.31),
# : t** qa - hA.i + ffie " d),. (e.33)
Letting q be the electronic charge e, we see that the linear term in the magnetic
field is simply the Zeeman term. If the z-direction is taken along the field
direction, the magnetic dipole or Zeeman energy is just
LÊ : mt\s%,
as discussed in section l0 of Chapter 3. On the other hand, we may incor-

porate spin in Equation 9.33 by rewriting the second term as
h A. (^d + zs),
and by adding the spin-orbit term, Equation B.B. The Zeeman energy then
becomes
L,E : (*, + 2m,)pry% : gmrpB%,
in agreement with section 9 of Chapter B. The quadratic term in I is the

diamagnetic term which is important for the calculation of the diamagnetic
susceptibility. Since it is considerably smaller than the Zeeman term, it can
usually be neglected, unless either the Zeeman term vanishes or extremely
high magnetic fields are used.
As our next example let us treat the weak-field Zeeman effect in hydrogen
by perturbation theory. The Zeeman energy (see section 9 of Chapter B) is
given by
L,E
-ry (L"+2s,; : fft", (e.34)
where the field 0 is taken to be in the z-direction. Since we must regard the
spin and orbital momenta as strongly coupled in weak magnetic fields, the
appropriate representation is lj, m,). Then, by first order perturbation theory,
the Zeeman energy corrections are
tr\!]*,,,,*, : r# U, m) J, I
j, mi) : €pahmr. (e.35)
These energies have already been calculated for p and s states of hydrogen, and
are given in Table B-5 in section 9 of Chapter B.
INTERACTIONS OF A CHARGED PARTICLE 353
The reader might well question the utility of the middle expression in
Equation 9.34 above. Its usefulness arises when the states u.. .*pr.rsed in the
lmt, m,) = l/, m) ls,.m,) representation. Thus, it would be used for the strong-
field Zeeman effect (Paschen-Bach effect) when the dipole energy is comparubTe
to or greater than the spin-orbit energy. It can also be used for the *.uk-fi.Id
Zeeman effect if the kets 17, m) are written in terms of the kets lms, m,). It will
no doubt be instructive to show this in detail for a few cases. Using the kets
for a/-electron given in Problem 8-20 we have:
:T.to(rt
ltaQ (1, L,
tril,g,t3+l * 25, ll, +) :2pn0,
trt,I,*,t:#'[3(0, +lL,+ 2s, 10, +) ++(1 , -+lL, *2s, lt, -+)]
: &png,
and so forth, in agreement with the values calculated from Equation g.35.
The spin-orbit energy correction for hydrogen can also be calculated
readily by means of perturbation theory. Using Equation B.B and the operator
given in Problem 8- 16, we can write
ffrt: -+ ffif,l ,r' - Lz - s,), (e.36)
where 0Vl0r has been replaced by e2fr2. This form of the operator
^d. ,S was
chosen since it has a diagonal representation when the kets U, mi) are used as
basis vectors. Then, for the I states of hydrogen,
(j, ml it'\) li, m,)

(j, mtlU, *,>
#rf U" + r) -+] ",+,:*ffJU,,+ r) -+],

where Appendix A of Chapter 7 was used to evaluate (1/rB). Then, forT : g
andT : +:
(*, mil irt) l&, m) : "--eJ rr.,f2xl}-tEn,
(i, mil Jt'n l*, m) - -En (H'^r,,- 4xro-aln,

3
ingood agreement with our previous calculation. The spin-orbit correction

cannot be obtained so easily for s-states, but requires the relativistic theory.
35{ APPROXIMATION METHODS AND APPLICATIONS
4. PERTURBATION THEORY FOR DEGENERATE

STATES
In the event that degeneracies exist among the unperturbed states, it is

still possible to calculate the first-order perturbation corrections by means of
Equation 9.8. The effect of the first-order correction is frequently such as to
remove some or all of the original degeneracy. This was illustrated by the
examples treating the Zeeman effect and the spin-orbit interaction in the
previous section. On the other hand, it might turn out that the first-order
correction is zero for all states-as in the Stark effect for states of definite
parity-so that perturbation theory fails unless a correction is obtained in the
second-order calculation. It might also be desirable to proceed to second order
in a case where the first-order correction is not zero but where the degeneracy
remains. However, 7\y attempt to apply the second-order theory of the pre-
vious section to a system where degeneracy persists in the first-order correction
is doomed to fail. This is most readily seen by examining Equation 9.7. Iq
for instance, the kth and n'D states are degenerate, then for k # n, both of the
quantities dr,, and (E[ot - Elf') are zero simultaneously. This gives the trivial
result that trft) : 0, which may or may not be true. The difficulty lies in the
fact that we are unable to find out what the actual value of ffonis by means of
the non-degenerate theory when the states z and k are degenerate.
In order to develop a procedure Tor treating degeneracies, let us alter
our previous approach somewhat, and assume that we can find solutions to the
eigenequation
(tro -f #tttlfn: Engn, (e.37)
where : and tro li) : li). The vectors ) : are the eigen-
1t*n li)
1pn Elo' lf ,p:ot
functions of the unperturbed Hamiltonian, as before. Multiplying by another

vector of the setl (,t1, we obtain:
cnoEl,o' + > t,otrLlt : cnt En,
cnn(En
- Ef') (e.38)
For convenience let us assume that we have only two degenerate states, the
ntk alnd /th'. Then the contributions to the energies of these states for i t' n, /
are negligible, and we may write
Ipn : cnn ln) + c*, lf>

and
IPr:cnln) *cul/).
First setting ft : n and then k : / in Equation 9.38, we obtain the two
equations:
t**(trt') + Ef' - E*) * c6ffff : g
(e.3e)
t,*ff,kt * cu(trt)' + Eot - E*) - 0.
PERTURBATION THEORY FOR DEGENERATE STATES 355
The necessary condition for these equations to have a solution is that the deter-
minant of their coefficients be zero. That is,
#r') + Eft - En trt')

,trt!", ,trt), +- Et') - En -0, (e.40)
which has the solutions
En:+l@H+Ef') +(trti'+Eo')l ++{t( trH +Eto))
- (.trt',' + Et\12 + 4:,trr')lltt. (e.41)
If, as we assumed above, Ef' : E)'\, then Equation 9.41 reduces to
En : Ell, + *(trr') + trty) + +t( trL') -,trti\z + + I tr#l,|b.
The meaning of what has been done in solving the degenerate case will
become apparent if one but compares Equation 9.40 with the diagonalization
procedure of Chapter 6. Thus, a matrix which has degenerate diagonal
elements and non-zero ofldiagonal elements in the representation of the
unperturbed eigenfunctions has a diagonal representation when expressed in
the new basis of the eigenfunctions of the perturbed Hamiltonian.
As an illustrative example, let us calculate the Stark splitting for the
n : 2level of the hydrogen atom by an electric field I in the z-direction. The
perturbation term of the Hamiltonian is zf$'t : eE z : e8r cos 0. There are
four degenerate wave functions corresponding to n : 2, !)zoo, Vzror grrr, and
Vzfi. Since each of these functions has definite parity and the perturbation is
odd, all diagonal matrix elements are zeroi that is, there is no first-order
correction to the energy and the degeneracy persists. Now let us obtain the
full matrix of the perturbation, whose elements are given by
(rpu,*,1 e8r cos 0 lrpu*).
Using parity arguments, one of the quantities / and f' must be odd and the
other must be even in order to have a non-zero matrix element. Furthermore,
we learned from the properties of the spherical harmonics that m' must equal
mfor a non-zero integral. Thus the only non-zero matrix elements are:
trLitr,rro : trL\L,roo : e8 (Vrool r cos 0 lrltrro)
:
ffi|,**(' - f,)'-''"' o'1,'sin o cosz o do
: 3aoe8.
The perturbation matrix is:

0 Saoe8
("trt't1 :(
3aoe8 0
00
i;il
00
Since the degeneracy persists, we must use degenerate perturbation theory
for the second-order calculation. This simply amounts to diagonalizing the
2 x2submatrix. Then,
_p(2)
u2 SareE
3aoe8
-0,
-E\zt
and
EL" : !3aoe8.
The diagonalized matrix of the Hamiltonian is:
I 3aoe8 0
0 Bto) * Saoe8
(.tr) : (tr,) -t (./ftu1 :
("*' 0
0
0
0 4i) (s.+2)
Equation 9.42 shows that the four-fold degeneracy is partially lifted by

the Stark effect, in that a single level is split into three levels by the electric
field. There is still a two-fold degeneracy associated with the unperturbed
energy Eto'. The shifted levels are associated with states that are described by
linear combinations of the original wave functions. Thus, the state with energy
ELo' + 3are8 has the wave function O I f2l (proo * ?rro), while the state
Qlfrlprro - gno) has energy Eto' - 3aoef. This behavior can be described
by attributing a permanent electric dipole moment of magnitude Saoe to the
n : 2 state of hydrogen. It then follows that the three energy states correspond
to parallel, antiparallel, and transverse orientations of this dipole with respect
to the electric field.
PROBLEM 9.18
Calculate the Stark effect for the n : 3 state of hydrogen.

[Ans.: Ejo), Elo' + laoe8, E[o) ! 9are8.l
PROBLEM 9.19
Find the new basis functions for the diagonal representation

of the complete Hamiltonian in the preceding problem.
TIME-DEPENDENT PERTURBATION THEORY 357
PROBLEM 9AO iiji:trir,l;:r.r:rr,,t,.ll
A two-dimensional, isotropic oscillator has the Hamiltonian
ff : -#(#,.#) **rt + bry)@, *t)

(u) If b :0, what are the wave functions and energies of
the three lowest states ?
(b) If D is a small positive number such that b ( l, find
the perturbation theory corrections to the energies of
the three lowest states.
PROBLEM 9.27 'li.iri-i-:rl:r:i,,,::,,l
A certain electric potential is described by the spin Hamil-

tonian,
jt,r):Ds? +E(s3_si).
Treating this as a perturbation on an ion with three de-
generate spin-one states, find the energy splitting that results.
iir;'r:jiijitiirffiirlir,i!'iiii,l ;i';ili,-tti,t;, ,.i:.rr":irrtli:lr:ti
5. TIME.DEPENDENT PERTURBATION THEORY
We will now consider'the case in which the perturbing part of the Hamil-
tonian is time-dependent; that is, when
tr (i, t) : tr o I i($) (i' t).

As before, .ffttt is regarded as small with respect to ff,. We require solutions
of the Schrodinger equation,
Jry:ih,E,AY
where
tr # o : Elot V n and Ye : tpc-wntnlott -
The yo satisfy the requirement <;lj>:6ot for a complete, orthonormal basis

for the expansion of an arbitrary wave function. In contrast with previous
practice, however, we will now regard the expansion coefficients themselves
as time-dependent. Thus,
yr : )a"(t)y" (e.43)
and
#:j(0"-'ntf'"-)v,. (s.4+)
Substituting Equations 9.43 and 9.44 into the Schrcidinger equation, we obtain
(tr0 * ,tr<rt1 2o,,V*
2 o*(ff' - Ef\Y* + > (anffttt - ihd*)Yn : 0. (e.45)
The first term of Equation 9.45 is trivially zeto since Y,, was defined to be an
eigenfunction of ff o. Multiplying the second term by Vf there results
( (kr a*J(e, rn) : ihdn6kneuth'@k(0'-Eft)t.

;
Defining
I
a)tn: o{A;t' - Ef'),
this may be written as
ihd* : 2 o*,ff[!ei'o"t. (e.46)
Equation 9.46 is impossible to solve exactly, since it relates the derivative of

the kth coefficient to all of the others. However, for a small perturbing potential
it is not a bad approximation to assume that the rate of change of eaih coeffi-
cient a, is so small that an can be taken as a constant on the right hand side.
Suppose that at t:0 the system is in the state Y, so that a,(0) : l. Then,
after a time / we assume that ao(t) is still nearly 1, and so
ihdk(t) x ff[totei'r't. (s.47)

From Equation 9.47 ,
a,(t) : r -; !, *;'; o, (e.48)
and
a*(t) : -;l: ffo,ei'ott dt, for k+ i. (e.4e)
For a constant perturbation that is turned on at t :0, Equations 9.48

and 9.49 may be integrated easily to obtain the development of the expansion
coefficients with time. Thus,
atQ):1- trll't (e.50)
and
a*(t) : - ei'oit), fo, k +i. (e.51)
H(I
The quantity lak(t)12 is the probability that the state Y* is occupied at time /.
Hence, it is a measure of the probability that the system will make a transition
from the initial state Vn to V* in time /. Squaring Equation 9.51, we have
Pt, : la,k4)f :lml' . rrn'

(EIo' -
2h
nlo')t
, for k + j. (9.52)
l"lc "i I
PERTURBATIONS THAT ARE HARMONIC IN TIME
Note that the probability oscillates at the angular frequency given by trro
but that its effect is appreciable only for values of ft near j, where the amplitude
reaches its maximum value. Recall that the derivation of Equation 9.52
assumes that the arremain small; this implies that ,trfr,t 4fr[0t
- Eo'.
For large / the amplitude aoQ) in Equation 9.52 resembles the delta
function, 6(E - E:o'), discussed in section 7 of Chapter 4. Assuming a quasi-
continuum of closely-spaced states, the area under the graph showin g lak(t)|,
versus E represents the total transition probability. Then, letting p@) be the
density of states,
p- p@)dE:12,tr;}'1, p(E)
!**bo{,)l, ,rt[_-#or
:- 2nt ,tr;t'l'
I
- p(E) , (e.53)
h
which says that the total probability of a transition out of the initial state Y, is
directly proportional to the time that the perturbation acts on the system. Both
p(E) and .ffft) are assumed to be slowly varying and are regarded as constants
for the integration of Equation 9.53. This is not a bad assumption, since the
integrand is a sharply-peaked function centered around Er10). The transition
rate from the initial state j be defined as,
ûy
, _dP _2n I tr!,'o'l' p(E)
rL- , (e.54)
dt- h
which is often referred to3 as Fermi's Golden Rule Number 2. It will be used later
in our discussion of scattering theory.
A few words should be said about energy conservation during the trans-
itions discussed in this section. In Equation 9.52, energy is conserved to
within the limits of the uncertainty principle expressed by LE . L,t ,-, h,
Although the system can oscillate between states of widely separated energies,
the greater this separation, the shorter the lifetime of the'excited state. How-
ever, as the resonance condition is approached, AE -* 0 and A, -' oo, implying
that a long-lived transition can occur. In this case the radiation field insures
that the total energy is conserved by creating or annihitating a photon. In
Equations 9.53 and 9.54 energy conservation is built into the formalism by
the assumption that all transitions occur between states which are closely
packed around the energy Ero'.
6. PERTURBATIONS THAT ARE HARMONIC IN TIME
Harmonic perturbations are extremely important in physics, since nearly

all of the weak interactions of electromagnetic radiation with matter may be
approximated by this model. Such diverse phenomena as magnetic resonance
8 E. Fermi, Nuclear Physics, University of Chicago Press, Chicago (1950), p.142.
in solids and the study of resonance absorption in optical spectroscopy are both
easily described in the harmonic approximation.
Since either the electric vector or the magnetic vector (or both) of the
incident radiation rrray interact with an atom, let us omit the details of the
interaction for the time being and write the perturbation as
27fo cos arl - Jt,o)(rn., + t-trot). (9.55)
Substituting this into Equation 9.49 we obtain for the expansion coefficient
a*(t),
a*(t) : - tO *1,,,t (sik"a+att a sib*i-'tt1 dt
lr'
: - *L:){ exp[f (aro,+ or)t] - I +, exp[i(aro, - ,)t] -

h(*r, + ar) h(ro, - ar)
t) (e.56)
It is important to notice that the first term of Equation 9.56 completely

dominates the second term when a) t.-J -(trkit that is, when
h.at ,-, g\ol Blo).

-
This corresponds to the physical process of stimulated emission discussed in
section 4 of Chapter 2. An incident
photon of energy fiar induces a transition
from the state Efo) to a state E[0) of lower energy. This process has a maximum
probability when h.a : B\ot - gtot. In like manner, the second term of
Equation 9.56 dominates the first term when o) t- ao,, that is, when
,-, E!!' nlo'.

h.at
-
This will be recognized as the process of resonant absorption, in which the photon
excites the system to a higher energy state E[o' > E:o'. For values of co which
are not close to *.ar,t neither term of Equation 9.56 gets large and aoQ) is
negligibly small. In the limit as a; goes to zero, Equation 9.56 reduces to the
case of a constant perturbation given by Equation 9.51.
Since only one term of Equation 9.56 is important at any time, we may
write the probability of absorption or emission as
(e.57)
lah?)Iz --
#rl',r,"195"i,1,
where the minus signs correspond to absorption and the plus signs to emission.
Using the delta function approximation for large /, the transition rate to the
kth state from the jth state may be expressed as
2r
Ro: E I tr[",], 6(Eoi + E), (e.58)
where E - ha is the photon energy.

PERTURBATIONS THAT ARE HARMONIC IN TIME 36I
As an example of a harmonic perturbation, let us consider the case of

electric dipole radiation. In the dipole approximation we neglect the spatial
variation of the electric field over the region of interaction (that is, over the
dimensions of the atom) and write the perturbation as
tytD(i,t) : -rF.f (r0., + r-n"r, : Ai .D(r,,, * e-i,tr. (9.59)
Here, Fo : t8, + jE, + fc* the amplitude of the electric field and D : -0i
"is Then, from Equation 9.58, the rate of electric
is the electric dipole operator.
dipole transitions from the jth to the krh state is given by
Rr,:| t4l F.n ll)1, d (Ef' - Et" + E). (e.60)
If the incident radiation is unpolarized, we may write
8r:8n:8",
s2 -- 8tr : E2 : +8t,
since EB : Sl + Si + 92.
Furthermore, since the energy density of an
electromagnetic wave is given by
I(,) :: +77
@3>,
when the permeability is unity, the matrix element of Equation 9.60 may be
expressed as
l<kl e'D li>1' -

4nI-(')
3 t&t D
l\'er st il : 4nl(a\
,r,l' ? lDo,l', (9'61)
where
lio,l, : (D,)r*, * (D,)rr, * (D,)?,,.
Then
R.
tttc
--Bn2I(a)
gh
.lDo,lrd(E[0,
- ptil + E). (9.62)
Recall that Bo,, the Einstein coefficient for induced emission or absorption
of photons, was defined in terms of the radiation energy density as (see section
4 of chaPter 2)'
R t : B*il(ot). (9.63)
By equating 9.62 and 9.63, we see that the Einstein coefficient may be expressed
in terms of the electric dipole matrix elements; that is,
B*r :ff w-,t'' (9'64)

It is interesting to note that the electric dipole model does not permit a direct
calculation of the Einstein coefficient for spontaneous emission, A*,, although
the latter can be obtained from the ratio given in section 4 of Chapter 2. In
order to treat spontaneous transitions rigorously, it is necessary to use a model
in which the electromagnetic field is quantized so that the physical system
consists of the quantum states of the field as well as those of the atom. The
absorption of a photon is then described by the simultaneous annihilation of a
photon state in the field and an upward transition in the atom. Emission is
iegarded as the creation of a photon state in the field, accompanied b.y u
downward transition in the atom. In the harmonic oscillator approximation,
the operators of Chapter 5 may be used. Although this topic belongs in a more
advanced course, the interested reader may wish to consult the excellent book
by Heitler.a
The selection rules for electric dipole transitions have been mentioned
previously, but it will be instructive to discuss them in the present context.
Note thai R* will be zero whenever Equation 9.61 is zero. Let us write the
hydrogenic wave functions in ket form vs !)nt* : ln) l/) lm). Then the matrix
.i.*.trtt of the components of D are:
Q"t D*IP) : (n't r tn) (/'t sin 0 th @' I cos { lrz)
lr)',',:'ir'r::-: rt:'*:'
Q"l Du h) : trrri'r'r'r".'r'u',',ti,"u!),,'1)*,'*'l sin 4lm)
(P' I D' l'D :-t @' | *)

r*ri' r'r'r'.'rt{,,',iltr:,t:'
The electric dipole selection rules are (neglecting spin) :
A,/ : tI
(e.65)
Nm:0, +1.
If spin is included, they become t Li - 0, +l; i : 0 ]'i : 0
Lm :0, +li * :0 +> ffi :Qwhen A.t :0.

PROBLEM 9.22
Verify the following matrix elements:

(u) (/'l sin 0 l/> : 6r,,r+t
(b) (/'l cos 0 l/> : 6r,,r*,
(") (m'l sin 6lm) : 6m,,m*r
(d) (m'l cos 6l*) : 6m,,m*r.
a W. Heitler, The Quantum Theory of Radiation, 3rd ed., Oxford University Press, Oxford, 1958.
PERTURBATIONS THAT ARE HARMONIC IN TIME 363
The treatment of atomic radiation may be cast in a somewhat different

form which has the advantage of showing how multipole radiation arises.
Using the interaction term from Equation 9.31, (elmc)p',4, let us express the
incident radiation as the real part of a plane wave,
i : ioer(i.i-,t1. (9.66)
The plane of polarization of this wave is perpendicular to its direction of

propagation, that ir, 1o .E:0. Now, if E.i < l, the exponential may be
expanded as follows,
A:Aoe-o"lL
1L I lrrv rJ + W+"'-l.
+UE'i) (e.67)
-1L0v L^ ' 2
Such an expansion is certainly valid for atomic radiation in the visible spectrum
' where)"f ao ,--' lgt. Keeping only the first term, the perturbation is
e*
' Ao cos at.
;F
Suppose that the wave is polarized in the r-direction. Then
F'io : !,Ao
and i
trri, --'#, <kt p,l j>. (e.68)
But
fn:m*:fflrr,*1,
from Equation 6. t l. Then,
.tolr : i2rh
.ff\','
ieA,
xtr,I j)l
l&l tror Ii) - (kl
: j2rhl@rkl
ieAo
x Ii) - <kl x l,tr i>l
:#(Ef,
ieA,
-Eo,)(klxlj)
:r#glexlj)
:'+ (D*)a,
where we have assumed the resonance condition h.at : o'.

Btot - E Since the
energy density is
ir , : @'A3
t\a1
Brcr,
we write Equation 9.60 as
L-2
Rr,: i ,@ l(D*)nil, (e.6e)
For an unpolarized wave, the average of cos2 0 over solid angle yields a factor
of fi which makes Equation 9.69 consistent with Equation 9.62.
Returning to Equation 9.67, note that the next term of the expansion
introduces an additional factor of iE 'i which multiplies the quantity (ir'il.
These products are components of a second-rank tensor, some of which contain
orbitaf angular momentum while others contain charge times quadratic terms
in the coordinates.s The former correspond to magnetic dipole moments,
while the latter are called electric quadrupole moments.
Each higher power of iE. i introduces higher-ordered multipoles and a
corresponding new set of selection rules. We will not discuss them further
here except to remark that the appearance of a spectral line which is forbidden
in the dipole approximation merely means that other kinds of poles are playing
a role in the physical process. For example, the selection rules for electric
quadrupole radiation aret
,
Neglecting spin : L'/ : 0, +2

Lm:0, =El, +2
Including spin: Lj - 0, +1, +2
Lm :0, tL, +2
No parity change
For magnetic dipole transitions the selection rules are:
Neglecting : 0
spin: L,/
Lm :0, +1
fncluding spin I Li - 0
Lm:0, +1
A further example of a harmonic perturbation is provided by the phenom-

enon of magnetic resonance. Here the magnetic moment operator produces
resonant absorption by means of magnetic dipole transitions. In a typical experi'
ment the magnetic levels are split by a uniform fixed field, and transitions
u
J. D. Jackson, Classi.cal Electrodynamics, John Wiley and Sons, N.Y., 1962, section
9.3; W. K. H.
panofsiy uni tut. Phittips, Classical Etectricitl, and Magnetrim, Addison-Wesley Publishing Co., Reading,
Mass., 1955, section l3-7.
PERTURBATIONS THAT ARE HARMONIC IN TIME 365
between these levels are induced by an oscillating field. The frequency of the
oscillating field is generally in the radio frequency band for nuclear spin
resonance, in the microwave band for electron spin resonance, and in the optical
band for electronic orbital transitions.
The magnetic term in the Hamiltonian is,
(do + 2d, co, ,r) 'T j : spu*roo rry# d, . icos a;r, (9.70)
where we have taken 9o to be in the z-direction. For a fixed value of 90,
we already know that the solutions for the Hamiltonian
tro -l €Fnmrfio (e.i 1)
are the kets lj,*,). Therefore, we will now regard Equation 9.71 as the
unperturbed Hamiltonian and the kets 17, m,) as the unperturbed wave functions.
It should be pointed out that Equation 9.70 assumes that g is a scalar
quantity. This is a good assumption for free or nearly free atoms, but is not
generally valid for ions or atoms in a crystalline lattice. In solids it frequently
turns out that the g factor is a tensor, and hence the perturbation must be
wrrtten as
jt'l:+d''3'i.
However, for simplicity we will regard g as a scalar in the present example.
Let us further assume that the oscillating field d, is in the x-direction. Then
the perturbation becomes
rt'l :ry e,J, :ttf ee* + J-). (e.72)
The transition rate from the jth to the kth state at resonance is given by Equation
9.58 as
R-:T (W)'t,,t J* ti) + (frt I-ti>t',
where J* governs the absorption process and J- the emission. The rate of
absorption in the state I j, m) is
1T
2h
kpu%,),(j -*)(j +m|_I),
and the rate of induced emission is
fi
2h
kpue,)r(j + *)(j - m -l r).
Of course, the number of transitions per second must take into account the
populations of the states as discussed in section 4 of Chapter 2.
PROBLEM 9-23
Show that the transition rate from state I

j, m) to j, m +
I L>
is equal to the rate from I
j, m + I) to I j, m).
7. ADIABATIC AND SUDDEN PERTURBATIONS
If a perturbation is varied slowly with time so that the physical system

always has adequate time to respond to minute changes in the applied forces,
then the system will remain in its initial state. During such an adiabatic process,
botlz tlze waue function and tlte energlt of tlte s2stem will change graduallt -fro* tlte ualues
associated with the nth state of the original Hamiltonian to those of the nth state of the
f.nal Hamiltonian. This is exemplified during the tuning of a violin when l
ihe string tension is slowly changed while a cirtain harmonic is bowed. The
harmonic used corresponds to a given oscillator state, and the change of pitch
is relate d to the change in the energy of the state. Another example is provided
by the elastic collisions of low-energy gas molecules. Since the electronic
motions are very rapid compared to the molecular motions, the distortions
produced during the collision process may be regarded as adiabatic and the
electronic wave functions are restored to their original form. On the other
hand, a high-speed molecule can induce a transition to an excited state, and
the collision process will no longer be elastic.
The analytical form for the adiabatic approximation may be obtained
by integrating Equation 9.+7. Since ,tr*' is now a function of the time, we
must integrate by parts:
at,: - +
# en'*,'
*,Ifir*;:llei'*t dt. (e.73)
The second term may be made arbitrarily small in the adiabatig approximation,
so that the first term gives the required result. Note that this result would be
identical with Equation 9.9 for a constant perturbation if the time factors were
included in the wave functions ,s'ed in section I of this chapter.
Now suppose that a system undergoes a sudden change at time / : 0
such that its Hamiltonian is given by trofor I < 0 and by tr - ffo * .ffrtt
forl)0,where
tr o li) : E,i li),
and
tr lq) : Er ltl)'
That is,, the orthonormal eigenfunctions of the original Hamiltonian are
designated by Roman letters, whereas the orthonormal eigenfunctions of the
ADIABATIC AND SUDDEN PERTURBATIONS 357
new Hamiltonian are designated by Greek letters. Then a general wave

function may be expressed as,
Since the wave function must be continuous at t :0 (although the derivative

will not be in this case), we have
l ou lr> : 7
bo ln).
Multiplying by (ql we obtain the sudden approximation:
b,t:ZonOtli>. (e.7 4)
1,
Equation 9.74 would give the correct expansion coefficients for an instantaneous
perturbation. In a physical problem, the system will require a finite time A/
in order to make the transformation from the state characterized by ff o to
that described by tr. In such a case,
br: \ &\rl I i\O-tntrttti-E,tt

a^(n
,,
+
u>Lt - *rt, - ar; lt], for Al4 1, (9.75)

1o'<rl
where the exponential has been replaced by the first two termd of its power
series expansion. Comparing Equations 9.74 and 9.75, it is evident that the
error introduced by the sudden approximation is proportional to At and to the
energy difference between the initial and final states.
As an example, consider the harmonic oscillator whose spring constant is
suddenly reduced to one-half its original value. We wish to know whether
this oscillator is likely to make a transition or whether it will remain in its
original state. For simplicity assume that the oscillator was initially in its
ground state, given by
rt:o : (!\
\7r/
r-*'r,,
where : l/f4n. The ground state wave function associated with the new
"
spring constant may be written by replacing a with
"112,
(Y-t
'' " : \T\/ 2l r-naztz{T,
ro-:n
'
Then; from Equation 9.7+,
bo: (,tq:ol,pn:): (;l (#r)t -o.ee,

and the probability that the oscillator will remain in its ground state is pro-
portional to lbol, 0.98. Note that all odd excited states are excluded by
-
conservation of parity. Since the most likely transition would be to the state
for which rl :2r let us calculate the coefficient br:
bz: (t',:,1'po:): (#J(;)'1., {2 ox' - 1)

'*o[ -9#'fno-
,-, -0.12.
Since lDrlz is of the order of 0.01, the probability for a transition to the second
state is roughly I per cent of the probability for the oscillator to remain in the
ground state.
PROBLEM 9.24
Calculate the probability that the oscillator discussed above

will make a transiti.on to the state for11 : +. Be sure that
rpn is normalized.
PROBLEM 9.25
A harmonic oscillator is slowly displaced at constant velocity

so that its equilibrium point is moved from x :0 to x : xr
If the oscillator was initially in its ground state, calculate
the coefficient pertaining to a transition to the first excited
state .
PROBLEM 9.26 !.jii.i.iii:liliir:t:"ii.r4il,l
For the oscillator of the previous problem, make the same

calculation using the sudden approximation.
PROBLEM 9.27
Use the sudden approximation to calculate the probability

that the ls electron of tritium will remain in the ls state of
Hel when tritium decays by beta emission.
PROBLEM 9.28
A particle of mass m is in the ground state of an infinitely

deep, one-dimensional, square well.
THE VARIATIONAL METHOD
(u) If the wall separation is suddenly doubled, what is the

probability that the particle will remain in the ground
state ?
(b) Ifthe wall separation is suddenly halved, what is the
probability that the particle will remain in the ground
state ?
8. THE VARIATIONAL METHOD

It frequently happens that a physical system which cannot be solved
directly is also not amenable to the perturbation method. This may be due to
the fact that the system either does not resemble a system which can be solved
exactly, or that the perturbation term is too large in relation to the exactly
soluble part of the Hamiltonian. In such cases it is possible to apply a variational
principle to obtain an upper bound for each of tkre lowest statis of the system.
Suppose that the system has a non-degenerate set of eigenvalues,
Eo 1 E, I E, 1. . . , corresponding to an orthonormal set of eigenfunctions
li) such that
,r V> : Et li).
Then an arbitrary function g may be expressed by the superposition,
g' : >anli),
and the expectation value of the energy is
2lanl' Eo
<./f> _<tt'l (lqt) (e.76)
\rp I vl 2lo)'
i
If each eigenvalue on the right is replaced by the lowesl eigenvalue, -Eo, we

obtain the inequality,
Eo Z lool,
<tr>rfu-Eo. (e.7 7)
Thus, Equation 9.76 provides an upper bound on the ground state of the system,
although not a very useful one. If Equation 9.76 can be minimized, however,
then the upper bound so obtained might well be a good approximation to the
actual ground state energy. The usefulness of the method lies in the fact that a
trial waue function V can be chosen such that it contains one or more variational
parameters for the minimization procedure. The effectiveness of the method
is dependent upon a judicious choice of the trial function. After the ground
state energy and wave function are obtained, the method can be applied to the
first excited state by choosing a trial function that is orthogonal to the ground
state function.
For the purpose of illustrating the method, let us apply it to a problem
whose solution we already know, that of finding the ground state energy and
wave function for the harmonic oscillator. Choosing a trial function of the
form
yt : Ce-"', i
we wish to minimize the expectation value of the energy with respect to u.

First, normalizitg ? we have:
r : (tttl,D : r, l7^, or ' : (7f

Then, with
ff : -#fi* Lkx,,
</f> : (vl tr lv) : !* +H (e.78)
Letting
!<*>-0,
oa
km
'-+h,
zt2-
-
,:T:I,,
\/k* d.
where a is defined as in Chapter 5. Thus the ground state wave function is
eo : (t)* ,-auztz,
\7rl
and the ground state energy from Equation 9.78 is
Eo: (ff)în:+-+H:ry.
Because of our choice of the trial function, we obtained the exact solution in
this case.
THE JWKB SEMICLASSICAL APPROXIMATION 37t
PROBLEM 9.29
Calculate an upper bound on the ground state of hydrogen

by using a Gaussian trial function of the form
y, : Ce-f,',
where B is the variational parameter.
PROBI.EM 9.30
Use the variational method to calculate the lowest energy

state of the anharmonic oscillator whose potential energy is
V: *r'mxz -f bxa. Takethetrialfunction asq - Ipo * |Vr,
where rpo and Vz are oscillator eigenfunctions. Let both ro
and p be variational parameters.
PROBLEM 9.31 It*.EjF.f+l+#, jif;;rd,tffi8,-.$n.F#",9-di9re'e;"$-l**:Ea.
Use the variational method to estimate the ground state

energy of a particle of inass m in the potential V : Cx,
where .r > 0. The particle is confined to the semi-infinite
space defined by 0 < x < * o. Take !) : xe-o' as the trial
function.
[Ans.: Upper bound for Eo is !(2hzC'pm)+.1
i"',-?.,riigidPi.Friu"3.;i.}d i#'d.9##F,#ffi4.$itri"!w'e.FJ#ffiPd##'##ffi####ffiw
9. THE JWKB SEMICLASSICAL APPROXIMATION
It was seen in Chapters 4 and 5 that the motion of a particle in a region

having a constant potential can be represented by a superposition of plane
waves traveling to the right and to the left. Each such wave has the form
Ip t\-, exp( fikx),

where
hk:\/W.
A similar solution may often be used to approximate the real solution in physical
cases where the potential is not constant, provided that the potential is slowlt
uarlingin space. This approximation is quite useful in one dimensional problems
and in problems with radial symmetry. Although it has been credited to a
number of people, the method is usually called the WKB method or the JWKB
metltod, after Jeffreys, Wentzel, Kramers and Brillouin6 who first applied the
mathematical techniques to physical problems.
By u "slowly varying potential" we mean a potential whose energy change
in a deBroglie wavelength is very much smaller than the kinetic energy of the
particle. Thus, the fractional change in kinetic energy per wavelength is much
less than one:
I I AVI
(e.7e)
(E - v)lax I
l2h ap
<1.
lp'a
This is analogous to an optical medium in which the index of refraction varies
continuously, but sufficiently slowly so that no reflection occurs.
Let us designate the slowly-varying potential by S(r) and require that
S(") : px when Z is constant. We will assume that S(r) may be represented
by an asymptotic expansion in powers of h as follows:7
S(r) :S'o(r) f fiS,(r) +trS,(") *..., (e.80)
where we also assume that only a few terms will be required.

Writing the wave function as,
..p F-,*o(f) ,
we find upon substitutitg ? into the Schrodinger equation,
firt'l'+ (v -E) -!*s":0, (9.81)
where the primes denote differentiation with respect to x. Substituting

Equation 9.80 into 9.81:
*ur' + L2*(si)' + + *(2s;si + nsds; + "')

+ (v - E) - *(sff + hs i +trt;+''') : o.
Jeffreys, Proc. London Math. Soc. (2) 23, 428 (1923); G. Wentzel, Z. Physik 38' 5lB (1926);
6 H.
H.A.Kramers, Z.Ph2sik39r82B (1926); L.Brillouin,Compt.rend.lSSr24(1926) andJ. phls.etradium

7,353 (1926).
? This treatment will follow that of D. Bohm,
Quanturn Theorl, Prentice-Ilall, Inc., Englewood
Cliffs, N.J., 1951, Chapter 12. Also, see A. Messiah, Quantum Mechani,cs, North-Holland Publishing
Co., Amsterdam, 1958, Chapter 6.
THE JWKB SEMICLASSICAL APPROXIMATION 373
Combining like terms and setting the coeffi.cients of each power of fi separately
equal to zero, we obtain the following equations:
I
2*
(s6)' + V- E:0 (e.82)
,s6^ti
- ,S'; -0 (e.83)
S,iSj + (Si), - is'i - 0, (9.84)
where all terms beyond the quadratic terms in hhavebeen dropped. Rewriting
Equation 9.82 as
dso : t/Z*1n - V1 ar,
its solution may be written as
So:*lJ,, t/zm1n-v1
Pa
ax. (e.85)
fn a similar fashion,
sr:!r"1s6y
or
exp(fS') : l2m(E - V)l-+,
and
c
'Jz- ; # . t2m(E - v)t-B - { [ rr*rE - n]-t (#)' *
If V is a slowly-varying function of # so that Equation 9.79 is satisfied
and all higher derivatives of V are also small, then it usually is possible to
retain only the first two terms of the expansion given by Equation 9.80. Then
the wave function may be written as
v : lE - V)-tlAerso * Be-iso1, (e.86)
where So is expressed by Equation 9.85.

The reader will note that y, as given by Equation 9.86, will not be valid
as (E - V) approaches zero. This creates serious difficulties since one of the
principal applications of the JWKB method is to problems involving barriers,
where it is important to match wave functions in two regions. Thus, for the
potential shown in Figure 9-1, a classical particle with energy Ê would have
a turning point at x : a, where E - V : 0. In the quantum mechanical
description, however, this particle will have an oscillating wave function for
x < a and an exponentially decaying function for x > a. The total wave
function requires the matching of these two functions at the very point where
the JWKB method fails !
The way out of this dilemma is to exclude the two functions from the
immediate vicinity of x : a and to construct a connection formula to join Vn>y
Figure 9-l Schematic one-dimensional Potential.

E is the total energy of a Pafticle.
and ysqv. If the region where the JWKB solutions break down is small
.tro.rgh, the potential can be represented by u straight line within the region.
Taking its slbpe to be that of the potential curve at x - 0, the equation of
the line is
v-E:(#),_"(x-a). \
Then the Schrcidinger equation in this region can be solved with the aid of
Bessel's functions of order $. The details of deriving the connection formulas
will not be given here,8 but for a simple barrier they may be summarized as
follows.
(1) BeRrlBn to t";}cHr.

lPn<v
#,.*(f k,dx-) =+E*o( -f r,ar)

(e.87)
+r'i"(fr' * -;) - +,"*o(fu*)

r' (e.BB)
V(x)
k, : ,_l
Vem(n-v) kr:
I-
Vzm(v-P)
6a | fra
(2) Bennrnn ro rHE Lnrt.

vn<v 'l'urr-
;-8"*o(- I:r,
*) o
ft",,ff,
-, dr
- t (e.8e)
h..o(,[0,
or)
= - +^'t(fo,
d, -) (e.e0)
t See E. C. Kemble, The Fundamental Principles of Quantum Mechanics, McGraw-Hill Book Co.,
NewYork, 1937, section 21; N. Frciman and P. O. Frcjman,JWKB Approximation,Contributiotu to the
Theory,North-Holland Publishing Co., Amsterdam, 1965; P. M. Morse and H. Feshbach, M*hods of
Theoieticat Physics, McGraw-Hill Book Co., New York, 1953, pp. 1092 tr
THE JWKB SEMICLASSICAL APPROXIMATION 375
.]'\
k, :;it_!Vzm(v-n) | k, : O:l_ V2m(E-V)

We will illustrate the application of the connection formulas by considering
the one-dimensional barrier shown in Figure 9-2. A particle of energy E will
have the classical turning points indicated by a and b. For simplicity we will
assume that the barrier is symmetric so that the momentum of the particle
isgivenbyk,:(|lh')\/winbothregionsIandIII.Inorderfor
the connection formulas to be valid, the barrier must be high enough and thick
enoughsothat!2k,dx-(|lh)!i\/WE)>l.Ifparticlesareincident
only from the left, the solution in region III will consist simply of a wave
traveling to the right, having the form
gm - ertio,u,,
where
t_
kt::Ot/Z*p-V).
For convenience later in applying the connection formulas, we will add a
phase factor to ?r'; thus,
lpnr : +;(litcou-ttrPt) . (e.e1)

!k,
Since ,4 is complex, it can readily absorb the factor eiol4. Equation 9;91 may
be expanded in the form
e,r : dx -) *;'i"(fr, dx -;)]

h[."'(f*,,
Using connection formulas 9.89 and 9.90, we obtain for rprr,
v,:h[+.*o(-I r,,ar)- iexp (tn,-)] (e.e2)
In order to find Vr we must use the connection formulas for the barrier to the
Figure 9-2 A symmetric one-dimensional barrier.

right. First, let us rewrite Equation 9.92 in the following convenient form.
en: +[+."0 (-to,dx -to,o4 - r exp (1.r,* *!'o,r4]

!hz
: .*o (- Io, n),"0 (I n, a)
h[+
- iexp (tn,ar) .*o( -l o,*))
Applying connection formulas 9.87 and 9.88, this becomes:
vr: -+[+.*o
!krt
(-l:r,ar) ,t"(fo, or -f
+ i2'"0 (fn , o,). ."'(fo , dr -;)] (e.e3)
It wilt be easier to interpret Equation 9.93 if we simplify the notation as follows:

Let
n :f n'or
and
P:fo,dx -;.
Ipr : - * i2e"cos P)
#r(Er-'sin P
A l- r
- T^l+t-"';("u - e-il) * 2ie"' iQnP +

'-of)
: - {ffu | f,e-d)e-ol + (e" - Le-,)eif1.

tk,
Since propagation to the right is represented by negative values of B, the
amplitude of the incident wave in region I is
Le-')
l#,(e'+
Then the transmission coeffi.cient may be written:
A ,
T- A
lkr : ['"0(.ll lr, a*) + ie*P ( I"n,*)f-'
Gr@ * Le-")
(e.e4)
However, since we require 12 k, dx ) l, the negative exponential can be

neglected with respect to the first term in Equation 9.94. Then Z is approxi-
mately given by
T ,--' ,-z!b nz aa.
For a rectangular barrier of width 2a this is approximately s-4k,a, which is in

order-of-magnitude agreement with the approximate result obtained in section
2 of Chapter 5 for the example of tunneling through the nuclear barrier.
The JWKB method can be applied to bound states as well as to scattering
problems in unbound systems. Although it has applications in many areas of
physics, it is probably fair to say that this method is used onlyas a last resort;
that is, if other methods fail. For this reason, we will not pursue it further
here, but will refer the interested reader to the references cited earlier.
SU/TÎHARY
In spite of the fact that very few physical problems in the real world can
be solved exactly, methods for approximating complex systems have been
developed which are responsible for the widespread application and success
of quantum mechanics. The most important and simplest of these approximate
methods is perturbation theory, which can be used whenever the actual system
can be regarded as a slight modification of a system whose solutions are known.
Special techniques are required for the cases of degeneracy and time-depend-
ence of various sorts. These methods are illustrated by numerous examples.
When the physical system is not amenable to perturbation theory, a variatibnal
method can often be used to obtain an upper bound on the energies of the
lowest states. The success of this method hinges upon judicious choices of the
trial wave function and the number and kind of variation parameters. A semi-
classical method is also described and is applied to a problem in barrier
penetration.
SUGGESTED
REFERENCES
David Bohm, Quanturn Theory. Prentice-Hall, Inc., New York, 1951.
Sidney Borowitz, Fundamentals of Quantun Mechanics. W. A. Benjamin, fnc.,
New-York, 1967.

P. A. M. Dirac, The Principles of Quantum Mechanics, 3rd ed. Oxford University
R. B. Leighton, Principles of Modern Phltsics. McGraw-Hill Book Co., fnc.
New York, 1959.
E. Merzbacher, Quantum Mechanirs. John Wiley and Sons, Inc., New York,
1961.
Albert Messiah, Qmntum Mechanics. North-Holland Publishing Co., Amster-
dam, 1958.
D. Park, Introduction to the Quantum Theor2. McGraw-Hill Book Co., Inc., New
York, 1964.
L. Pauling and E. B. Wilson, Introduction to Quantum Mechanics. McGraw-Hill
Book Co., Inc., New York, 1935.
J. L. Powell and B. Crasemann, Quantum Mechanics. Addison-wesley Publish-
D. S. Saxon, Elementary Quantum Mechanics. Holden-Day, fnc., San Francisco,
1968.
Leonard I. Schiff, Quantum Mechnnics,3rd. ed. McGraw-Hill Book Co., New
York, 1969.
CHAPTER TO
&ffiffiffiw$ffiM&il"
&Pffifuffiffi&Yffiffiruffi
I. THE GROUND STATE OF THE HELIUM ATOM

As we have discovered in the previous chapters, quantum mechanics has
been very successful in accounting for the ground state and the excited states
of the hydrogen atom. This success may be attributed, in the first place, to
the fact that the two-body problem is relatively easy to solve exactly. Secondly,
refinements to the theory are readily incorporated by means of simple per-
turbation techniques, as we have seen for the spin-orbit correction, the rela-
tivistic effect, the hyperfine interaction, the Stark shift in an electric field,
and the Zeeman splittings in a magnetic field.
However, element number two, helium, is not so simple. Since it has two
electrons, we are now faced with a three-body problem. Furthermore, the
helium nucleus contains four nucleons in contrast with the single proton of the
hydrogen nucleus. Because of this additional nuclear structure, the nucleus
no longer looks like a point charge and the Coulomb law does not hold rigor-
ously (although it is still a good approximation). There'is an additional
complication due to the fact that each electron tends to screen some of the
nuclear charge "seen" by the other. We must also take into account the
strong electrostatic interaction between the two electrons, and since they are
indistinguishable particles we must include the quantum mechanical phenom-
enon of exchange degeneracy discussed in Chapter B. The electrostatic and
exchange effects are large effects, not just refinements like the spin-orbit and
spin-spin interactions.
As our point of departure, let us approximate the ground state energy of
the helium atom by ignoring all interactions between the electrons, as well as
their screening effects upon each other. The Hamiltonian for this system is
ff : (- #,vi - T,). (- #,r, -T):2ff,, (r0.1)
where ,ffo is simply a one-electron Hamiltonian for a hydrogenic atom with

Z --2. The ground state energy associated with Equation 10.1 is thus
E{o': -2Zzwo- -Bwo: -l0B.8eV, (10.2)
379
ADDITIONAL APPLICATIONS
where wo : 13.6 eV, the ionization energy of hydrogen. Since the experi-
mental value for the ground state of helium is -78.62 eV, the reader can
readily see how large an error is produced by the effects which have been
neglected in Equation 10.1. In spite of the magnitude of the corrections
needed, it is an attractive idea to try a perturbation calculation, since the
interaction between the electrons is an obvious choice for the perturbing
potential. Pursuing this approach, the Hamiltonian becomes,
,ff :2tro + (10.3)

tr,
where rLz : li, - irl is the distance between the two electrons. The unper-
turbed wave functions are eigenfunctions of Equation 10.1, namely, products of
the two one-electron wave functions. Let us designate the unperturbed ground
state by *
l0) : ?',(1) 'y',(2) 'l', m,) : -z(r'*r'tlao ls, m,). (10.4)

:(Z)"
The nuclear charge Zwill be carried for greater generality in spite of the fact
that Z : 2 for this example. The ket ls, m,) stands for the normalized anti-
symmetric two-electron spin function given in Equation 8.22. This is the only
spin function that can be used for the ground state of helium, since the spatial
wave function must be symmetric. Therefore, the ground state of helium is a
non-degenerate singlet state. Using the wave function given in Equation 10.4,
the expectation value of the energy is
o2
(01 ,tr l0) : -2Z2wo { (01 :- l0), (10.5)
Ttz
where
etf*r') : #€,)' (.0 l-ta(,, +',,] l-;(,,+',,])

l,ll."o
(s, mrl t, *,>
--#Gl In,on, IJo,',r.*o (-+) In,on, Io*

0,,r7,,0(-+)
(10.6)
For spherically symmetric wave functions of the type used here, this
integration may be easily performed by regarding it as the electrostatic energy
of two overlapping spherical charge distributions. Thus, we first calculate the
potential at a point i, due to the infinite charge distribution defined by i, and
having charge density given by exp [ -(2Zrrlao)]. Since this potential is
constant for r, lrn but varies as l lrrfor rz ) rtz, we have the two integrals
*Although the simple product function is satisfactory for the present discussion, the reader
should note that the proper form is the symmetrized spatial function given on page 306.
THE GROUND STATE OF THE HELIUM ATOM 38I
(except for the constant factor) :
vr : 4nrf; dr, * I),-r",,,", ,!,. 4nrf, dr,

["r-r",,,",:,.
vt : u" (9' - * 2x,+ 2)l * (r, * t)e-,,},
{* tr e-,,(xi
where
2Zrt
^1_
",
Now we calculate the energy of the system by integrating the charge
density e-ar over all of space,
,0,*t0)
r
: 5F"l .[o-,-,,vftnxz,dx,:W : *Zwo. (r0.7)
PROBLEM ti{it"iixl'lirrr+i"r
'0'l
Verify the integrations leading to Equation 10.7.
iiij.l/j.1i,:li1rrj .l :'t-: . i., ."-,.. ." ,., .t.,.' ,.-., , :,1ir+";J.jii!,r.1ii
A method of integrating Equation 10.6 which can be extended to non-

sPherical charge distributions is to expand I f r* in spherical harmonics as
follows:1
:*:i|(;l ''n*
r' ror r'r ) rz
1:l - /r-\t
rrz ,rA\r)rt3os o)' for rt I rzt
where 0 is the angle between i, and /r. Because of the orthonormality of the
spherical harmonics, performing the integration over Q, and O, before the
radial integrations will eliminate all of the terms containin g Px except for / : 0.
etf,ro) :f
A[o*or,rl,-,,{['i !l'ar,rz,-,, +!),a***-,ll
5 ezZ
: :tZwo' (10.8)
B
'h
in agreement with Equation 10.7.
1 L. r. schifl Quantum Mechanics,3rd ed. McGraw-Hill Book co., N.y., 1969, p. 258.
ADDITIONAL APPLICATIONS
PROBLEM TO.2 'f#.i"9,5i# fi r+.t
Verify Equation 10.8.
':i:.-:r
::i,Y#,.i)r+,1"ti6ij? ti;rirtilli .:ii'E:gli:;L:-"1;r
"r't1 ),i;iitii i.i, #;p;li6i.5lj*iiil6r
The corrected ground state energy, Equation 10.5, is now
l; *, :
Ee,ou,,a : -2z2uo *1r*, : _
-74.8 eY, (10.e)
for Z : 2. This is in surprisingly good agreement with the experimental value

of -78.62 eV, particularly since perturbation theory would not be expected
to be valid when the perturbation term and the initial term are of the same
order of magnitude.
Let us now attempt to take into account the effect of the partial screening
of the nucleus by the second electron. We will proceed by treating the
nuclear charge Z as avariational parameter in the minimization of the ground
state energy. Using Equation 10.4 as the normalized trial wave function, the
expectation value of the ground state energy is
(Egroono) : (01 2tr0 + !W>

Tn
:2(ol trolo) + (ol Lp>.

Ttz
The second integral was just calculated above, and the first will be left as a
problem. The result is
(Eg.oo,,a) : 2(Zz - 4z)wo * *Zwo' (10. r0)
Minimizing Equation 10.10 by setting (dldZ)(Eg,oo,a) :0, we obtain an

effective nuclear charge of Z : ++ for the screened helium nucleus. Using
this value in Equation 10.10 the upper bound on Egroool becomes -5.7wo:
-77.5 eV, which is in good agreement with the experimental value of -78.62
eV.
It is also possible to make an estimate of the first ionization -energy of
helium from the difference in the energies of the lsz and the ls configurations.
This amounts to
5.7w0-4wo-l.7wo,
whereas the experimental value is l.Blzoo.
Show that
(01 trol0) : (2, - +4 .#r, (10.1 1)
THE LOWEST EXCITED STATES OF HELIUM
where l0) is given by Equation 10.4 and ,ffs is given by
Equation 10.1.
PROBLEM IO-1
(u) Apply the model used for helium to singly ionized

lithium in the 1s2 configuration and set up the integrals.
Note whether any are the same as in the helium
problem.
(b) Show that (01 trol}) : (Z' - 6Z)wo in this case.
(.) Find an upper bound on the ground state of Li+.
(Ans. : -14.4w0.)
PROE|-EM
'0-5
The first ionization energy of lithium can be calculated by
finding the difference in energy for the configurations 1s2
and 1s22s. Choose a product wave function for the ls22s
configuration, set up the integrals, and explain the procedure
to be followed. Do not evaluate the integrals.
PROBLEM 10.6
Find the first ionization energy for Li+. (This is the energy
difference between the ground states of Li+ and Li++.) The
experimental value is 5.56210.
PROBLEM IO.7
Find the first ionization energy for Be++ The experi.mental

value is ll.3lwr.
2, THE LOWEST EXCITED STATES OF HELIUM
The lowest excited states of helium are those for which one electron
remains in the ls state and the other electron is raised to a 2s or 2p state.
Since the electrons are indistinguishable, for each pair of quantum states there
will be four possible spin wave functions as given in Equation 8.22. There
are also four two-electron spatial quantum states represented by n and n',
namely, Vnogzoo, T-oq'', Vnogzn, and tp.,s1p2..5. Neglecting all interactions,
then, there are 16 degenerate wave functions representing the first excited state
of helium. In reality, of course, we cannot turn off the Coulomb interactions,
so that the 16-fold degeneracy mentioned above does not physically exist. It
turns out that the four states associated with Vrcgz, are not raised as much as
384 ADDITIONAL APPLICATIONS
t Bohr Theorv Coulomb Interaction Exchange Coulomb Interactions

E
(eV)
States
Singlet Triplet States
(para) (ortho)
-54.t6 Exp. lst Ionization Energy
2'P,
Ls 21t (3ffi)
(3 states) .,^.-Irs.t j,
/(12 states) -J
|
-
K,*. ,, 2'S,,
/
// ls 2s (l state) t
ls 2s / -./ (4 states ) 4J t'' :* -2"s"l ,

-.-a I
ii ,, L/*, K,". ,* - (3 states)
-68.0 Erlttr
(16 states)
--
lsl I tS,,
fI state) (l state)
-78.62 Exp. ground state
/- -
/
I
/*,..,^
-108.8 E,tu'
Figure l0-l Schematic diagram of the lowest states of helium. The spin-orbit interaction has
been neglected since it is only eV. Degeneracies of leyels are given in parentheses.
-10-a
the twelve states associated with the yrrrprp wave functions. This is shown in
Figure 10-1. The remaining degeneracies are split further by spin-orbit
interactions and the correlation effects arising jointly from symmetry and the
Coulomb repulsion as discussed in sections 5 and 6 of Chapter B. fn order to
see this more clearly, let us look in detail at a perturbation treatment of the
four degenerate wave functions formed from !)*Vzr. We write g zs,
V :7,I [v,,(I)rpr,(2) * ?r,(l)'p',(2)] ' ls, m,), (10.12)
where the plus sign refers to the singlet state and the minus sign to the triplet
state. The ket ls, m,) can be any one of the three symmetric spin functions
when the minus sign is used; for the plus sign, the ket ls,m,) must be the one
antisymmetric spin function. It will not be necessary to express the spin
THE LOWEST EXCITED STATES OF HELIUM
functions exactly because the Coulomb perturbation is independent .of the

spin coordinates. Then the first order correction to the energy of the first
excited state of helium is
EL" : <vlfiW>
ti,
-- rrr,( t),pr,(2) + r)pr,(2) lpt,(l)rpr,(2) + q)r,(l)vt,(2)) (s, nt,l t, *,)
?2,( I
*
- K +J, (10.13)
where
I
K - e2 (p*(1),pr,(2)l
fn l,pr,(1),pr,(2)),
(10. r4)
I
J: ez (,pt,(1)pr,(2)l t ) ).
fn lrltr,(2)y,r,(
The integral K is the direct or ordinary Coulomb integral and the integral J is
called the exchange Coulomb integral. Notice that when ,K is evaluated, the
integration over the coordinates of one of the electrons involves only one of the
functions !)u ot gz* On the other hand, when J is evaluated each configuration
integral involves both yy and gr, and, in fact, is a measure of the degree of
overlap of these functions.
The energy of the first excited state of helium is, then,
Ez:ELo'+K+J,
where the plus sign refers to the singlet state and the minus sign to the triplet
state. The calculation of the following energies will be left to the problems:
K,,,rn : ez (Ls(l) .2p(2)r!*rLs(t) ) rp(r))

- 10.0 eV
(r,,r, : ez (ls(L) .2s(2)tl tls(l) .2s(2))

Ttz - 9.1 ev
( 10. r5)
Jr,,r,: e2 (1s(t) . 2s(2)t 1 tls(2) . 2s(1)) - 0.4 ev
Ttz
Jr,,ro : ez L trr(z) .2p(r)) ^,0.r

(ls(t) ' 2p(2)l; eV.
t2
Note that the energies of both the singlet and triplet states are raised by the
direct Coulomb interaction, but that the exchange interaction raises the
singlet and lowers the triplet (see Figure l0-l).
3. THE HEISENBERG EXCHANGE INTERACTION
AND MAGNETISM
The excltange splitting of Equation 10.13 is given by the quantity,
AEor"1 :2J. ( 10.16)
Heisenberg and Dirac have shownz that the phenomenological vector coupling
model provides a very simple spin Hamiltonian which gives the correct energy
splitting for this case. The Hamiltonian is
tr\"n: -2J&
.Sr, (10.17)
where ,31 and ,S, are the spin vectors for the two electrons. Since the total
spinisS:&*Sr,
,S, : (S, * Sr), : si + sr' + 23, .Sr,
and
23''S, :s(s + 1) -sr(s, f 1) -rr(r, + 1) :s(s + I) - g.
Now, for the singlet state, where the spins are antiparallel,
(St 'Sr).tngtet : -?
and for the triplet state where the spins are parallel,
(S,' Sr)rriplet : +Q - il : +'

Then,
AE *.r, : -2Jt(S, .Sz)ri.gret

- (Sr .Sr)rr,n,.r] : -2J' (-? - +) :2J,
in agreement with Equation 10.16. Thus the vector model gives the correct
energy difference between the singlet and triplet states, and says that the
lower energ"y state is achieved by parallel spins when J is positive. It should
be evident that this phenomenon, applied here to a simple two-electron atom,
is the clue to the mystery surrounding the Hund rule for parallel spins in
many-electron atoms (see section B of Chapter B). This, in turn, accounts
for the presence of permanent magnetic moments due to uncompensated spins
2 W. Heisenbdrg, Z. f. Physik 49, 619 (1928); P. A. Dirac, Proc. Roy.
^9oc.
A 123,714 (1929). See
J. H. Van Vleck, The Theor2 of Electric and Magnetic Suscepti.bilities, Oxford University Press, Oxford,
1932, Chapter 12.
THE HARTREE SELF.CONSISTENT METHOD
in certain atoms of the transition metal and rare earth series of the periodic
table. Moreover, it is now generally believed that this same exchange mecha-
nism is the origin of the intense internal field postulated by Weiss to account
for the parallel ordering of the spins in ferromagnets and the antiparallel
ordering in antiferromagnets (where the exchange integrcl J is negative).
A serious defect of Equation 10.17 is that its form is apt to lead one to
believe that the exchange energy has its origin in a spin-spin interaction (that
is, a dipole-dipole interaction). This certainly is not the case, and it was known
even at the time of Weiss that the dipolar forces were three or four orders of
magnitude too weak to account for spontaneous magnetic ordering. The reader
should bear in mind that the energy difference is due to an electrostatic interaction
which arises from the symmetry requirements of fermions. The fact that this
energy difference can be accounted for by means of a scalar product of spin
vectors should be regarded as fortuitous. In spite of this minor shortcoming,
the Heisenberg-Dirac-Van Vleck Hamiltonian, Equation 10.17, is of such great
importance that the overwhelming majority of papers in the field of magnetism
use it as a starting point.
4. THE HARTREE SELF-CONSISTENT METHOD FOR

MULTI-ELECTRON ATOMS
In the discussion of perturbation theory it was pointed out that the smaller
the perturbation, the better the approximation. Therefore, since the Coulomb
repulsion between electrons is so large, we cannot hope to apply perturbation
theory alone to atoms with a large number of electrons-in spite of its fortuitous
success in the case of helium. On the other hand, the inclusion of the electron
interactions in the unperturbed part of the Hamiltonian would make it an
insoluble many-body problem.
A way out of this dilemma was suggested by Hartree,s who showed that
the spherically symmetric part of the electron interaction energy can be included
in the unperturbed Hamiltonian by means of a potential derived from the
average charge density of all electrons but one. That is, a single electron is
visualized as moving in an effective potential du'e to all of the other electrons
as well as the nucleus. The contribution to the potential seen by the ft'a
electron due to all the other electrons is calculated from the spatial average
of each unperturbed one-electron wave function ?pl.o), or
V*:2 i 'fia", :2@[o'tftwl"y

illc Jri
(10. r8)
Then this potential is used along with the nuclear potential to solve Schrcidinger's
equation self-consistently. That is, the equation
t-#,,, -T + vlvL" : Er!,!j', (10. le)
8 D. R. Hartree, Proc. Cambridge Phil. 50c.24r 89, I I1,426 (1928).

is solved for each electron. Then the new wave functions ylt) are used in
Equation 10.18 to calculate an improved potential V;. The iterations are
continued until the potential ceases to change appreciably; that is, until it
becomes self-consistent.
fn recent years the self-consistent method has been used extensively
because it is readily
programmed for digital computation.a Some of the short-
comings of the model are that it neglects correlations and the Pauli principle,
and it retains only the spherically symmetric part of the Coulomb interactions.
An improvement due to Focks consists of including exchange by using properly
symmetrized product functions for the initial wave function.
5. THE THOMAS-FERMI STATISTICAL MODEL

OF THE ATOM
Another central-field approximatior] that has been applied to atoms of
large Z is due to Thomas and Fermi.6 The model is more crude than the
Hartree model, but it has the advantage of being much easier to solve. It
assumes that the central potential varies so slowly that it can be regarded as
essentially constant over several de Broglie wavelengths. Thus, several electrons
can be localized in a region of quasi-constant potential. Under such con-
ditions the electrons may be treated as aR ideal gas of non-interacting particles
known as a Fermi gas, which fills the potential.
The de nsity of states of a free electron gas is obtained by a.pplying periodic
boundary conditions and box norm alization to a cell that is small with respect
to the volume of the gas, but large with respect to a de Broglie wavelength.
This has been done in Problem 7-2, with the result that
,(E) : #(2m)tr*, (10.20)
where an additional factor of 2 has been included here because there are two
spin states associated with each spatial state. .Notice that the energy of the gas
increases as the number of electrons increases. The energy of a collection of Z
electrons is obtained as follows:
fEr
':)o "@)
dE:t#f'E+dE --ffi"',,
Ep : fi{z,, e)*. (10.21)
In Equation 10.21 the quantity p : zlls is the number of electrons per

unit volume, and E3, is called the Fermi energy. Note that the Fermi energy
aSee D. R. Hartree, TheCalculation of Atomic Structures, John Wiley & Sons, New York, 1957;
J. C. Slater, Quantum Theory of Matter,2nd ed., McGraw-Hill Book Co., N.Y., 1968.
5 V. Fock, Z. Plrysik6l' 126 (1930).
6 L. H. Thomas, Proc. Cambridge Phil. ioc.23r 542 (1927); E. Fermi, Z. Pfutsik 48r 73 (1928).
THE THOMAS.FERMI STATISTICAL MODEL OF THE ATOM
is simply the energy of the highest filled state. In the many-electron atom
which we are treating, we have Ep: -V(r), and from Equation 10.21,
P: L:
z em)* (10.22)
g*r1t ?vg11,,
Now, the self-consistency condition is that the potentials due to the electron
density in Equation 10.22, as well as that due to the nuclear charge, properly
reproduce the potential energy, -V(r). Consequently, the charge density t
-0 p,
and the electrostatic potential, -lV(r)lel, must satisfy Poisson's equation,
_.4n(-ep),
-1VrV:
e
v2v : l,#,(,,#) : 4nez p. (10.23)
Equations 10.22 and 10.23 may be solved simultaneouslyT for p and V, imposing
the boundary condition V(r) : -(Zezlr) as r--+0, and rV(r) -'0 as r --+ oo.
The result is that the "size" of a heavy atom is approximately
h2 ao
me'Ttr z+
and, accordingly, the screened Coulomb potential seen by an outer electron

may be represented by
(10.24)
The validity of the Thomas-Fermi model improves as the number of electrons

increases. However, at best, its description of electron densities is very crude,
since all of the structure of the radial probability density curves is smoothed
out in the statistical averaging process. For light atoms it overestimates the
potential and the charge density at small radial distances.
PROBLEM IO.9
Using Equation 10.20, show that the average energy of a

particle in a Fermi gas is $Ep. What is the total energy of a
Fermi gas of /f particles ?
? L. I. Schifl Qtnntum Mechani,csr 3rd ed., McGraw-Hill Book co., New York, 1968, p. 430.
PROBLEM IO.TO
Taking the density of the gas of conduction electrons in

copper to be ,-.,lQza cffi-3, estimate the Fermi energy. How
does this energy compare with the thermal energy of an
electron at room temperature ? Discuss the consequences of
this with respect to the participation of electrons in thermal
processes in metals.
6. THE HYDROGEN MOLECULE:

THE COVALENT BOND
We have seen in the foregoing sections how, in principle, the eigenfunctions
for the electronic states of a many-electron atom can be constructed from one-
electron eigenfunctions. A similar approach is often used for the electronic
states of molecules, although with much less success since even the simplest
molecules deviate drastically from spherical symmetry. As an illustrative
example of a possible approach to this problem, let us look at the method used
for the first successful calculation of the dissociation energy of the hydrogen
molecule.s
The Hamiltonian for the molecule, which is shown schematically in
Figure 10-2, is
,ff:ffc**trx*fft-f#z*fft, (10.25)
where
ff crr : - h2 vt2ctut v
2*r*,
3fN --h, mp vk +t;
hz y12 ez ez
ffL- - 6"'t - i- ,*
nz hz ez ez
hfz- -
2*u'2 - ,z - ,*
t'
ffrr:
ftz
Here mnl2 is the reduced mass of the two protons and m" is the reduced mass
of the electron relative to a stationary proton. For simplicity we will neglect
the translational and rotational energy of the molecule and will regard the
distance between the protons, R, to be fixed. (W. are also ignoring the
exchange symmetry of the protons.) We expect both electrons to be in ls
9 W. Heitler and F. London, Z. Physi.k 44b 455 (1927).

THE HYDROGEN MOLECULE: THE COVALENT BOND 39t
Figure l(l-2 Schematic diagram of the hydrogen molecule.
statesin the ground state of the molecule. Then, taking into account the
exchange of the electrons, there are two possible approximate electronic wave
functions which are products of hydrogenic functions, namely,
gau : rttr!) 'rpu(2)

(r0.26)
rpbo:rpu!)'tpr(2),
where yo and Vb are both ls wave functions. Recall that in the helium atom,
where there is only one nuclerrs2 po6 and tpuo are degenerate eigenfunctions of
the same Hamiltonian. Here, however, you and y)ba are not degenerate, since
one is regarded as an eigenfunction of ,ff, and the other is an eigenfunction
of .ff r.
A variation method may be used to solve the problem by writing a trial
function that is a linear combination of the functions in Equation 10.26.
Thus we may write
V:Pao*A'Puo'
where / is the variation parameter. Thenre
(1 + A')B +2AD
Egroorro { (10.27)
| 4 trz +2AC )
where
B- (rl,,ul tr lrpou) : (rttu"l tr lrpor)
C: (rl,oulrpu")
D- (rl,,ul tr ly,uo) : (rl,url tr lrp,r).
For fixed R, the derivative of Equation L0.27 with respect to z4 is
2(1 -A')(D-BC) :0,

(1 + A2 +2AC)z
which yields a minimum of Egroooo for A : +1 and a maximum for A : -1.

-4
!o z
^,
a.
t"o
X
u
L
q)
t-1 -z
Proton Separation, angstroms
Figure lG3 Energy per atom for the symmetric and antisymmetric electron sPatial wave
functions for the hydrogen molecule.
WritingV:gm*rpoo,
Egroooa=
B+D (10.28)
f *"
Upon evaluation of the integrals B,C, and D, the difference between the energy
obtained from Equation 10.28 and the electronic energy of two separated hydro-
gen atoms (about 27 eY) should be the dissociation energy of the hydrogen
molecule. The experimental value for the latter is 4.72 eY.
Physically, one could have guessed at the proper linear combination of the
wave functions in Equation 10.26, since an antisymmetric spin function would
require a symmetric spatial function in order that the total wave function be
antisymmetric. That the symmetric spatial function has a lower energy than
its antisymmetric counterpart can be seen by reviewing the example of the
double wells in Chapter 5 (see section 4). The energies of these functions
versus nuclear separation are shown schematically in Figure 10-3. Since the
antisymmetric function can have no bound state, it is often called the anti-
bonding orbital, and the symmetric function is called the bonding orbital.
F- --i
b
Figure l0-{ Schematic cross sections of the spatial wave functions for the hydrogen molecule.
7. MAGNETIC SUSCEPTIBILITY OF AN ATOM

The behavior of an atom in a magnetic field may be determined by means
of the Hamiltonian given in Equation 9.33. Calling the direction of I the
MAGNETIC SUSCEPTIBILITY OF AN ATOM 393
z-direction, the Hamiltonian becomes,
.r - tro - spsmp + #+ @ + fi)s,
-afo*trrq*ffr%r, (10.2e)
where the summation is over all of the i electrons in the atom and where ffo
includes all of the contributions to the total energy which persist when the
magnetic field rt it turned off. Classically, the magnetic moment along the
field direction may be defined as
AE
(10.30)
t-z - ag'
where the energy E is
E - -rt'4.
The quantum mechanical equivalent of Equation 10.30, which is
(l't,)--:
/nn - - Ln
69 ,-)nn,
(10.31)
has been verified by Van Vleck.l0 The quantity (E),, is calculated from
perturbation theory as follows:
(E>*,: (nl troln) + g (nl trrln) * 9, (nl trrln) + g, > Wl![

;, E#t - Ef'
:Ef'+sE::t +s2Ef' +... (10.32)
The perturbation ,ff, has been carried to second order, since all quadratic
terms in 0 should be considered if any are retained. Then Equation 10.31
becomes:
(p")nn : -Et' - 2gEL'z) +'''

: - (nltrrln) -29 (nltrrln) -29 tu*r7'
>ryH unt un
oz
: (nls,pmrln) -g>m@lxl +filn) *rU,nffi ln')|,
gpamt t /,nl
: (t'o)nn-f-@lx? +filn) *r*^7,,ffi,' (10.33)
The middle term in Equation 10.33 is always negative and is the source of the
diamagnetec contribution to the magnetic moment; this term will be neglected
in what follows. Let us now obtain the paramagnetic susceptibility associated
with the induced moment of the last term of Equation 10.33. Assuming a
to
J. H. Van Vleck, op. ci.t., section 36 of Chapter VI.
Boltzmann distribution* over the states z', the atomic susceptibility may be
expressed as
Qtu)nn I
r"-'"'r'
x:V: s>:@' (10'34)
But, from Equation 10.32,

*H'\.
'
e-EnlkT : e-EnorlkT . e-gLnttt1tcT... A/ e-EntotlkT(t -
\' kr l'
where second order terms have been dropped in the expansion of the exponential.
Then, for the numerator of Equation L0.34, we write
] ,,t-u"ttcr ,^-, > (-E*t - zanft\l', -#)r-Entotltcr
,-,
tl_.tf, +s ry - 2sEfilt-'*'o'ro',
where only first order terms in CI are retained. Now the first term in the
bracket represents the net moment in zero field, which is zero. Then,
Z ,ur-',tkr
,-.,s >l% - ru*'fe-Ento)tkr,
and
r: tl%
-rrf,),-Bntotlrcr
(10.35)
In e-E*totlnr
where only the field-independent term is retained in the denominator. Sub-

stituting the values for Ef,rt and Ef into Equation 10.35,
x:
>lW + z
_Z,WH1r,-'""',r, (10.36)
le-8"(')lttr
If the separation of the states is much greater than kT, only the ground state
contribuies to the first term of Equation 10.36, and it becomes
J
Zl@i polmi>12
:- s,rrL 4*f s'\LJ(J +
skT
1)
(10.37)
KT kr 2J +t '
which represents the susceptibility due to the permanent dipole moment.

* See Appendix B of Chapter 2.
The second term in Equation 10.36 gives rise to the so-called Van Vleck
temperature-independent susceptibility.
Here there are two contributions, one due
towidely separated values of z and n', and the other due to widely separated J
values.lr Although these contributions are usually neglected, it is woith noting
that they arise when the Zeeman effect is carried to second order in perl
turbation theory.
SUGGESTED $P.r##i,.il#, n#ffi*,p-
REFERENCES
David Bohm, Qwntum Theorlt. Prentice-Hall, fnc., New York, 1951.
R. H. Dicke and J. P. wittke, Introduction to Quantum Mechani.cs. Addison-
Wesley Publishing Co., fnc., Reading, Mass., 1960.
H. Eyring, J. Walter, and G. E. Kimball, Quantum Chemistry. John Wiley and
Sons, Inc., New York, 1944.
E. Merzbacher,
Qtnnturn Meclwnics. John Wiley and Sons, fnc., New york,
t961.
Albert Messiah, Qwntum Mechanics. North-Holland Publishing Co., Amster-
dam, 1958.
L. Pauling and E. B. Wilson, Introduction to Quanturn Mechanics. McGraw-Hill
Book Co., New York, 1935.
D. Park, Introduction to the Qunnturn Theoryt. McGraw-Hill Book co., New york,
1964.
J. L. Powell and B. crasemann, Quanturn Mechanics. Addison-wesley publish-
D. s. Saxon, Elementary Quantum Mechanics. Holden-Day, Inc., San Francisco,
1968.
Leonard I. schifl Quntum Mechanics..3rd ed. McGraw-Hill Book co. New
York, 1969.
rr
J. H. Van Vleck, op. cit., section 55, Chapter IX.
CHAPTER IT
sffiewwffimffiffiffiffi wffi#ffiffiffiw
I. THE PARTIAL WAVE TREATMENT OF SCATTERING

An introduction to the classical study of unbound systems of particles
by means of scattering experiments was given in sections 2 and 3 of Chapter 3.
We will now discuss a useful quantum mechanical approximation for the case
of elastic scattering. fn analogy with the phenomenon of diffraction in optics
and acoustics, one may visualize the incident particle flux as a superposition
of plane waves. Furthermore, the scattering center itself may be regarded as
the source of a weak spherical wave similar to a Huygens wavelet. The effect
of the scattering event near the scatterer may then be determined by the
superposition of the incident plane wave and the newly generated spherical
wave. Far from the scatterer, in the forward direction, this solution asymptotic-
ally approaches a plane wave again. In the case of elastic scattering, the prop-
agation constant for this asymptotic plane wave has the same magnitude as
that for the incident plane wave, but its direction is altered
A spherical wave may be represented by (l/r) exp (ikr),where the factor r
in the denominator insures that the intensity obeys the inverse square law.
In order to provide for some angular asymmetry, it is common practice to
include an angular factor ,"f (0), in the amplitude. Then the outgoing spherical
wave has the form,
?sph : -f (0) +
Letting the particle beam be directed along the z-axis, the incident plane wave
is given by
Vin:Aeik'i-7titcz.
Hence the scattered wave is the superposition
vac: Aeikz +f A)'+ (11.1)
Since the interaction region is assumed to be restricted to the immediate

vicinity of the scattering center, the beam fluxes far from the scatterer do not
interfere and may be considered separately. The intensity of the incident
396
THE PARTIAL WAYE TREATMENT OF SCATTERING 3g7
beam is
I: lrl,rol,.nn
mm
:U,
fot ?,r' normalized, while the intensity of the beam scattered at an angle 0 is
/r( o) :
Iryl *: t -r (o) t''
-,Y,,'
Then the number of particles Scattered through the angle 0 into an element of
area da, normal to the beam, is
dN(o) : t-f (o)r.## : t-f @r.U act.
From Equation 3.7, the differential scattering cross section is then
h:t-f e)r. (11.2)
The factor-f (0) is called the scattering amplitude, since its square gives the differ-
ential scattering cross-section.
. Eqtation I 1.1 is a solution of the free particle Schrodinger equation. It
is convenient to write it in a different form, however, by expinding the plane
wave in terms of the Legendre polynomials. Thus,
eitcz
- Alt(rf + I)jr(kr) . Pr(cos 0), (11.3)
where the jr(kr) are the spherical Bessel functions.l

A general formula for these functions is
it(k )
: (- \'(Lr)##J'#, (r r.4)
and the first two are:
jo&r)
sin *r
j,(k):t1?fl-ry
(k kr )'
The asymptotic form of these functions for kr ) / is
js(kr) --*
,ir, (t, -+) (1r.5)
kr
1G. N. Watson, Theory of BesselFunctions, Macmillan, N.y., 1944, rev. ed., p. l2B.
SCATTERING THEORY
Thus, at large distances from the scattering center, Equation 11.3 may be
written as
@ ,ir, (r, -+)

silcz > i'(2/ + l)
- t:o .Ps(cos 0) .
kr
(11.6)
Since Equation 11.6 is the free particle solution, it must also be the solution
for a non-zero potential at large distances.fro* the scatterer. We assume that in
the interaction region its form cannot change radically, but that the potential
must only alter the phase of the sinusoidal function. For the asymptotic form
of the scattered wave we then assume
l|sc
,i., (*, -+ *4)
- kr'
where d7 represents a phase shift of the (th partial wave. Rewriting Equation
1I.1,
oo
bxPl@os 0)
,i,'(r, -+*r)
t:0 kr
,i,' (r, - +)
: + l) ' Px@os o) +f(0)
t:
(1 I .7)
Zitel kr r
we find that
br : i/(2/ + l)ei6t (l 1.8)
and
-f (0)
:IAQI + t)darsin d7 . Pr(cos 0). (1 1.e)
Note that the scattering amplitude is zero if all of the phase shifts vanish. The
total scattering cross-section calculated from Equation 11.8 takes the simple
form
' :J.' -f (o)t, de :7,2e/+ r) sin2 d7. (11.10)
PROBLEM I ifSBi.ariF:'iP,;f*i-t:.F }.irlr; ;-1.1t ii,-11_E_lx3-! 8 "*E*jprt;r,;ur.s.*ri
'.1
Show that Equation 11.5 is a solution of the Schrodinger
equation for a free particle.
PROBLEM I1.2
Verify Equations 11.8 and 11.9.
,P'H.*xfiB,'Sy.6e,i}tl-iiE: i,tjj-8.ryr.*iBFi.EHi 43#x,.-;-c+;8F:*Y).T.d.? :*t?Frjf:i?..4.rF.,F/,

THE PARTIAL WAVE TREATMENT OF SCATTERING 399
PROBLEM I1.3
Derive Equation I 1.10.
From the notion of the conservation of probability, it is evident that

when scattering occurs, the removal of particles from the beam should affect
the forward scattering amplitude,/(0). We therefore expect to find a relation-
ship between the total cross section and/(O). Using Equation 11.9, we write
/(0) :IrZQ/+ t) cos 6,sin il t;2O/ + r)sinz d7.
Comparing the second term with Equation 11.10, we observe that
/--
o :7Im t/(0)1. (11.1r)
This relationship is known as the optical theorem; its validity is far more general
than the present derivation implies.
In order to apply the partial-wave method of analysis to a scattering experi-
ment it is necessary to be able to calculate the phase shifts, il. In general,
the phase shifts will be positive for an attractive potential, which means that
the phase is advanced and the wave is pulled in toward the scatterer in the
interaction region. The peak of the outgoing wave is thus spatially behind
the free particle wave (see Figure l1-l). The reverse is true for a repulsive
potential. The scattering cross-section, however, does not depend upon the
sign of d7. The d, are obtained by matching the radial solutions of Equation
7.70 for V : 0 to the radial functions which include the effect of the scattering
potential, Equation 11.7. An explicit expression relating the phase shifts to
the scattering potential and the radial eigenfunctions can be obtained by means
Figure I l-l Phase shifts pro-

duced by scattering from attractiye
and repulsive potentials. V:0 for
the top figure, V is attractive for
the middle figure, and V is repulsive
for the bottom figure.
SCATTERING THEORY
of the Green's function treatment of the scattering problem.z However, the

partial-wave method is frequently applied to situations where only the first
one or two phase shifts are required, as the following argument will show.
Each partial wave corresponds to a definite value of angular momentum
which may be related to the impact parameter and the linear momentum of
the incoming particle by the expression, / : h.ks.* Since there will be essentially
no scattering for values of s greater than the range of the potential, ro, it is
evident that the relationship
/ < hkro (1 l. l2)
provides a simple guide for determining how many values of / will be required
to represent the scattering. Thus, for low energy particles (small ft) or a
short-range potential (small ro) only the lowest / values are required. fn fact,
it often turns out that kro 11 so that / : 0 is the only angular momentum
state that need be considered. Such an event is called S-wave scatterin$, in
analogy with the terminology of spectroscopy. S-wave scattering is isotropic
(in the C-frame), since from Equation 11.9 we note that
-f (o)
: ! sin do'
'nuo
when all phase shifts vanish except for that associated with / : 0. The total
scattering cross section for S-wave scattering is
:#sinz do. (11.14)

"
The isotropy of low energy scattering may easily be predicted from the wave
picture. Since the de Broglie wavele ngth of the incident particle is much
greater than the range of the scattering, the detector must be able to resolve
phase shifts of the order of 2n'(roll) : kro if anisotropies are to be observed.
Hence, if ,t and ro are both small, no phase shifts will be detected.
Applying the S-wave approximation to the classical collision of two hard
spheres, the maximum phase shift is do : ft.R, where R is the sum of the radii
of the spheres. For frR ( 1, sin do - do, and the total scattering cross section
is 4nR2, which is four times the classical value obtained for the geometrical
cross section in Problem 3-1. The additional cross section in the quantum
mechanical case is due to interference and diffraction effects.
A more important example is that of square-well scattering. Let the
potential beV(r): -Voforr <aandV(r):0 for r>a. Then a suitable
wave function inside the well for the case of / :0 is
gin:Asinkrr, for r1a,

Quantum Mechanics, John Wiley and Sons, N.Y., 1961, section
2 See, for example, E. Merzbacher,
12.7. For a diffegent method of evaluating the phase shifts see L. I. Schiff, Qwntum Mechanics, McGraw-
Hill Book Co., N.Y., 3rd ed., 1968, p. l2l.
* See section 2 of Chapter 3.
THE PARTIAL WAVE TREATMENT OF SCATTERING 40t
where
hk,--\/Wm.
fn a similar manner, the S-wave solution outside the well is
Voat : B sin(krr * do), for r ) a,
where hk, -- {2rrrE. The boundary condition requires that (1 ftDQrpQr) be

continuous at r : e., Thus,
k, cot k1a : k, cot (kra { 6o)
or (1 1.15)
tan(kra * do) : kp.

frtan
At low energies k, and do are both small so that
tan(kra * do)
-kz& * do,
and then
do 1 k1a - k2a
!rr^"
or
sindo-aa(W-t).
From Equation 11.15, the total scattering cross section for S-waves is
d:4na',(w- tl (1r.16)
The scattering cross section becomes infinite here for kra : nl2, 3n12, . . .
where the tangent function becomes infinite. Since Equation I l. l6 was derived
from the assumption that do 4 kfi, the infinite cross section is inadmissible.
Returning to the exact expression, Equation 11.15, we see that sin do
k1a: nl2 andkra 41, and so the maximum cross sectionis - I when
4r
ures-7. (11.17)
fv2
Such maxima are called S-waue resnna.nces.

ff the energy of the incident particles is high enough to overcome the
centrifugal barrier, a similar procedure may be carried out for higher / values
and one may find P-waae resznances, and so on. Such resonances are often called
uirtual states, since the incident particle appears to be temporarily bound in
spite of the fact that its total energy is positive.
SCATTERING THEORY
PROBLEM I1.1
By means of the optical theorem, Equation I l. I 1, show that

the upper bound for o for ,S-wave scattering is Equation
I 1.17, regardless of the potential.
The Ramsauer-Townsend ffict is also predicted by Equation 11.16, since

o:0 when k{-tankra, and 6o:nr2nr.... Nearly complete trans-
mission of electrons is observed for scattering by rare gas atoms at energies
bround 0.7 eV. For energies higher and lower than this value a scattering
cross section appreciably greater than zero is measured. This phenorhenon
is similar to the transparency of barriers (for wave functions that match up
properly) discussed in section 2 of Chapter 5.
2. SCATTERING AS A PERTURBATION:
THE BORN APPROXIMATION
Time-dependent perturbation theory may be applied to the scattering
problem by treating the perturbation potential as being "turned on" when the
incident particle gets within its range, and "turned off"'again when the scattered
particle exceeds its range. The scattering process is thus regarded as a time-
dependent transition from an initial plane wave state lf) to a final plane wave
state l;f). The transition rate per incident particle is given by Fermi's Golden
Rule, Equation 9.54, as
R: Ir<ttvti>t, p(Et), (11.18)
where Z is the scattering potential and p(E) is the density of final states.
Since the final states here are free particle states, we may use the electron
gas density of states, Equation 10.20, omitting the factor of two for spin states,
p(E):mw:m (r1.1e)
We are interested in those states for which the vector E, ties within a small
solid angle centered around Er. Therefore, the density of states, Equation
'11.19, must be multiplied by the factor dQl4r. Then the transition rate per
unit solid angle centered around li, due to a particle flux of intensity
',ruhko
:-:-
Ls Lsm
ls
R -do
do - do'
-:l-
SCATTERING AS A PERTURBATION 403
Thus, the differential scattering cross section is
R
do
-
do- /do -- \ffi)' *,. t<-n v ti>t, (r 1.20)
For elastic scattering of plane waves, kr : ko and Equation I 1.20 becomes
and
h: ffill,t-.nL',v(,)aEni d,f ,
-f (o): - #.[,ng1rn'Eo-E,t'r
4,. (1 1 .21)
This integral is known as the Born approximation. Since the volume normalization
factor Z-8 drops out, the integruiiot is independent of the size of the box.
Hence, contributions to the scattering cross section arise only where V(r) is
not regardless of the limits of integration on r. This approximation is
-small,
equivalent to assuming that only single scattering events u.. i*portant, and
thus the first-order term in V(r) is all that app.utrl.t Equation I 1.21. We will
see in the next section how multiple scattering events are treated.
Equation 11.21 may be simplified by means of the substitution
R:Eo-Et
^R is given by K :
Since fo and E, are the same length, the magnitude of
2,to_'.sin 012, where 0 is
the angle of scattering between Eo und, E, (see Figure
11-2). By defining an auxiliary set of polar ioordinates with the'polar"axis
directed along R,
V?)stil'i dr : z, v{")r2 dr "o"

o sin o do
!* Ioro",
Yo
[, no) sin K' rdr'
Then the scattering amplitude is
-f(o) kl:r(r) .sin Kr.dr. (1 1.22)
Equation 1I.22 is sometimes referred to as the standard -for* of the Born

approximation.
A discussion of the range ofvalidity of the Born approximation is a complex
Figure ll-2 The wave vectors used for the Born

approximation.
4U SCATTERING THEORY
subject, but we may make a few general remarks in the context of the assump-
tions used here. First, we have used perturbation theory, which presupposes
that the amplitude of the scattered wave is small compared to that of the in-
cident wave and that lv(r)l < E, the energy of the incident particles. Thus,
the energy of the incident particles must be large enough so that they may be still
represented by plane waves after the scattering occurs.
As an example, let us apply the Born approximation to the scattering of
electrons by u screened Coulomb potential,
vb\:
\'./ - -Zt'r-'/o
r " t
where a is the screening radius. From Equation Ll'.22,
-f(o): - WI:sin (Kr) s-rlaiy.

The integral has the value,
- K "4]; cos
K2+0'
ful'-''"(-:sinKr
Thus, the scattering amplitude may be'written in the form
-f(o):w[,_.#l (r 1.23)
PROBLEM IT.5
(u) Apply a limiting process to Equation 11.23 to obtain

-f (0) for the unscreened Coulomb potential. How does
this compare with Equation 3.15 ?
(b) Sketch the curves of (doldQ) vs 0 for these two cases
on the same graph.
PROBLEM I I-6
Calculate the differential scattering cross section for the

attractive square well potential,
V(r):-Vo' for 0<r--ro'
V(r) :9, for r l ro'
using the Born approximation. Sketch (doldQ) vs Kro.
[4"''' #:Wl .fsin (K,o)- Kro cos (/(rr)]']

SCATTERING AS A PERTURBATION 405
PROBLEM I I-7 f#idri:iitit#.fiBiti,nfi *,.{oji.1*',i;i:ir?;#iiiir;:uqoji::iF;.ij"i.:id i;:i1i!ji,]i t:iii 3',ifi.i;:i,:;.f$
Find the total scattering cross section in the limits of high

and low energies for the square well of the previous problem.
PROBtEtr4 tl-8
Calculate the differential scattering cross section for the

attractive Gaussian potential,
v(,) -- *vo"*o
[-(il'
using the Born approximation.
[A", ' h:,Wl exp [-+(r,,),]]
PROBLEM I1.9 *.{,1itii--tilr:.tJrt"{ffi ;iii*i*i*i+:,1,;'i]{,
A particle is scattered by the repulsive potential,
V(r) : lbrz exp[-(;)],

where 6 is positive.
(u) Use the Born approximation to obtain the differential
scattering cross section.
(b) Estimate the energy for which phase shifts associated
with / > 0 will become appreciable.
PROBLEM I I.IO ffi$".iji*irfirti,ri+$
Using the Born approximation, find the scattering amplitude

for the spherical, triangular potential shown in the figure.
406 SCATTERING THEORY
3. THE GREEN'S FUNCTION METHOD

FOR SCATTERING
A solution of the equation
(V' + k')rp(i) : -4rp(f) (11.24)
can be constructed from a superposition of the solutions for point sources of

unit strength, p(i'), if we define p(i) by
pQ):lrtt-
J
r')p(i')di', (11.25)
where "f d(i - i') di' : !. Suppose that a solution G(i,i') can be found for
the equation
(V, + kz)G(i,f') : -4tr6(i - i'). (11.26)
Then ,pe), the desired solution of Equation 11.2+, is readily shown to be
,t,V) : I te, i') p(i') di' .

J
(11.27)
To verify that Equation LI.27 is a solution of ll.2+, we utilize Equations I 1.25

and I 1.26 as follows:
(v, + r,1!cg,i') p(i') or' :l(v, + kz)G(i,r) pQ) dl
: | _'nnd(i - i') p(i') di'

J
: -4np(i).
In order to apply the above technique to the scattering problem, we will
write the Schrodinger equation in the form,
(v' + k'),t|) :T VQ)4Q), (11.28)
where hzkz :2m8. Then a solution of Equation 11.28 is
,t,Q) : -TIre,i')v(i')y(i') di, (11.2e)
where G(i, f') is the solution of
(v, + kz)G(i, i') : -d(i - i'), (11.30)

THE GREEN'S FUNCTION METHOD FOR SCATTERING 4OI
which must now be determined. Using the solutions of Poisson's equation in

electrostatics as a model, we write the solution as,
G(i,i'):+#, (11.31)
PROBLEM I I.I I ffF#lil#.1,$,ii *" $S$,F;lliir
Show that Equation 11.31 satisfies 11.30.
1.yi3a;;IS;13";;J;fjg;q;fiflgf*g3u;ggl*.fJ*;:;.ff$t.rif,+j:i*"_e"';11lS.jfrf:riJ.:i;#:fff:jij;y:,efFj
Using Equations 11.29 and 11.31, Equation 11.1 may be written as
lpsc : ,ik,z - +l("-': vQ,)tp(i,) di,. (n.32)

2trh.zJ li - i'l
In the limit as i gets very large, the directions of f and i tend to coincide and
kli - i'l - k1r.- i' i'l : kr - E'' i',

where E' : frl (see Figure 11-3). That is, E'has the same magnitude as the
initial wave vector f, tut a different direction. Then, for large r,
,ttcli-i'l,ittcr-i'.i'I
A/-
lf-i'l r '
and
._m
Vsc : eLkoz
- 6-
znhz
' '#l--' i'v(i,)ve,) di'. (11.33)
Scattering regiou
Figure I l-3 Schematic diagram of the scattering process for an extended scatterer.
SCATTERING THEORY
Comparing this with Equation 11.1, we see that the scattering amplitude is
given by
-f (o): - #l.,u"r'v|)v(i') di', (11.34)
where the scattering angle, 0, is the angle between the incident beam and i.
A calculational difficulty still exists in that the total scattered wave function
appears in the integrand of Equation I 1 .3+. Note that if this total wave function
is replaced by only the frst term of Equation LL.32, namely, the incident plane
*urri, then Equation Ll.3+ is identical with the result for the Born approxi-
mation, Equation 11.21. An analytical justification for this substitution may
be sketched as follows. Assuming the convergence of iterative solutions, the
solution for g(i') may be written as
,t,e) : ,iE'i' - #.1 ,ro',i")v(i")y(r') di',
which, upon substitution into Equation 11.33, gives:
lPsc : eiLo'i -# [ tr', i')v(r)sirt'i' dl
. (h)' i')v(i')G(i" i")v(i")tp(i") di"' (1 1'35)

I
"o'
and so on. Each successive iteration replaces the wave function in the integrand
by the solution in the next lowest approximation. Thus, the first integral
contains the wave function in the zero-order approximation, which is the
incident plane wave. This has already been described as the Born approxi-
mation. The second integral in Equation 11.35 may then be evaluated by
replacing y,(i") by the wave function obtained from the Born approximation.
T[is is called the second Born approximation. In general, the nth Born approxi-
mation consists of replacing the wave function in the ntk integrand of the iter-
ative expansion by the solution obtained in the (n - 1)ta approximation.
The physical interpretation of the iterative expansion, the first few terms
of which are given in Equation 11.35, is that the higher order terms represent
multiple scatterings.s An integrand containing one Greent function is an
integration over single scattering events; two Green's functions characterize
doufite scatterings; and so forth. The Green's functions may be regarded as
"propagators" which carry the wave to the next scattering point. Since one
would not expect many multiple scattering events for a weakly interacting,
potential, it is easy to see why the first Born approximation has such a wide
range of applicability.
3 R. P. Feynman, Phys. Rea.761749 (1949).
SCATTERING OF ELECTRONS BY ATOMS N'
4. SCATTERING OF ELECTRONS BY ATOMS
The atomic potential seen by an incoming electron may be represented

by the function
: I e,q,(i')
v(r) -zez J li-i'l!:' '
.
where pr?') : d(i') I p(i'), is the total charge density, that is, nuclear plus
electronic, at the point i'. Then Equation ll.2l becomes
-f(0):#, IIffidf,di
mZez f I ,rt.r:f.r'r:,!"
:-2m))er7)- di, di
V_i,l
:# ! o'{'')'rt''' di' ['#u "
The integral on the right is evaluated by taking f' :0 as the origin for the
integration over i and by including the inlegrating factor e-nr. Then,
"i
I'+di -:l
n "it.i PZt Ptt r.a
,[ J*,-",frirdrdodo
- tz'lim f*r-,,,'*'Joo
-'";5Jro a, lt ,ifl'r"or o 4\rr\
d(cos 0)
:2nl1 rrl"*' i<!, f

,[ "r
,-
:+:T Kr dr
I-'-o'sin
: 4n-. K : 4n
t"{
T o,t + r, K,
It follows that
-f (o)
:ffi2mZe2 r .?
Jer{r,)riK.i'4i,
:WJ tat'l - p|)re&,'di'
:ffif'- Ip4)eikr di'l
:Ytr-r(,qr. (11.36)
4IO SCATTERING THEORY
Here F(K) is the atomic scattering factor, which is a measure of the amount of
shielding of the nuclear charge by the electrons. It is given by,
l+
F (K) : I p1)tiK i'7i' . (11.37)
J
Note that the atomic scattering factor is the Fourier transform of the electronic
density. For a spherically symmetric charge distribution, Equation 11.37 may
be integrated over the angle variables and simplified to
F(K) :+ [ e?)sin (Kr') r' dr'. (1 1.38)
PROBLEM I I.I2 i+.i4q--h-1-9&W$.F,.Xb--1r9l!i'1,.h
Integrate Equation II.37 over solid angle to obtain

Equation 11.38.
PROBLEM I I.I3 l.*&9;B:o--.i-s,, .-qiru|-B
Assuming an electronic charge density of the Gaussian form,
P?) : (#,1 ,-0/ro)2,
obtain the atomic scattering factor F(K). Here ro is the

atomic radius.
[n",.:
r(^K) : eXp t- (?l] ]
,*i.s,-&iS:.Ei#i1lSF.fd.+.SrS1E#9
The quantity hzKz in Equation 11.36 can be written in terms of the energy
of the incident electrons by means of the relationship derived from Figure LL-2,
hzKz : rrr 9
+hzkzrirr -- v,'eu 9LLL 9
BmEsirrr .
2 - 2.
Then,
f(o):frt 4E sinz ,
-F(K)1. (r 1.3e)
This gives the Rutherford result for large angle scattering, Equation 3.!5,
when th. .n.. gy E is much greater than the ionization energy so that the
shielding is ineffective. However, for very small scattering angles the shielding
THE EFFECTS OF EXCHANGE SYMMETRY AND SPIN 4il
effect is great enough to prevent the singularity in I J(0)12 at 0 - 0, which is

predicted by the Rutherford formula.
5. THE EFFECTS OF EXCHANGE SYMMETRY AND

SPIN ON SCATTERING CROSS SECTIONS
In the case of a two-particle scattering event in the C-frame, the two

particles wilL go off in opposite directions after they interact (see Figure 3-5b).
That is, one particle appears at the angle 0 and the other at the angle (, - 0).
Now, if the particles are indistinguishable, it is impossible to tell which is the
scatterer and which is the scattered particle. Thus, it is necessary to combine
the scattering amplitudes,/(0) andf (r - 0), before calculating the scattering
cross sections. For spinless particles, such as the a-particle, the wave function
must be symmetric (section 5 of Chapter B) and the scattering amplitudes
add with the same sign. Hence, in the C-frame,
(symmetric spatialfunction) doldA: lf(0) +_f (n - 0)1,, (11.40)
for two-particle scattering events involving identical, spinless bosons. The

reader should note that the classical result for this case would be
do
-'"f(0)l'+l"fb-0)1"
dC-- |
where no interference terms are present.

On the other hand, suppose we consider the elastic scattering of identical
spin I particles by particles of the same kind. For the symmetric spin case
we must also use the symmetric spatial function (since the total wave function
must be symmetric), and the differential cross section is given by Equation
t 1.40. However, we can also obtain a total wave function that is symmetric
by using both antisymmetric spatial and spin functions. For this case the
differential scattering cross section is given by
_ (n 0)l'.
spatialfunction)
fO: l-fQ) -f -
(antis2mmetric (11.41)
Since the symmetric and antisymmetric scattering events are independent

events, we expect the differential scattering cross sections to be the sum of
Equations 11.40 and 11.41, with proper consideration being given to their
statistical weights. From the table in section 10 of Chapter B (see also Problem
B-29), we note that the ratio of symmetric to antisymmetric spin states is 2:I.
Therefore, the differential scattering cross section for the elastic scattering of
unpolarized, indistinguishable, spin I particles is
#: zrfe) +-f (, - o)r' + + t-fe) --f (, - o)t' (11.42)

412 SCATTERING THEORY
A similar procedure is followed for fermions. For example, for either

0 - 0,
- p
p, or n - n scattering, where the spin is *, we have already seen that
the symmetric spatial function can be combined with only one antisymmetric
spin function, whereas the antisymmetric spatial function can combine with
three different symmetric spin functions. Therefore, the differential scattering
cross section for the elastic scattering of identical, unpolarized, spin $ particles
is
do
da
: + I "f(0) +"f (n - 0)l' + il"f@ --f (, - 0)l'. (1I.43)
su/vl/vlARv
Three approaches to scattering problems have been introduced in Chapter
I 1. In the partial wave method, the incident particle is represented by a
plane wave which is expanded in the Legendre polynomials and the spherical
Bessel functions. The scattering event is then treated as the superposition of
this incident wave and a spherical wave generated by the scattering center.
By comparing the forms of the scattered and incident waves at large distances
from the scatterer (the asymptotic limit), the effect of the scatterer can be
described in terms of phase shifts in the arguments of sine and cosine functions.
The calculation of these phase shifts constitutes a solution to the problem.
In the perturbation method, a scattering event is regarded as a transition from
one state in the continuum of positive energy states to another. Hence, the
mathematics used in calculating transition probabilities, including Fermi's
Golden Rule, can be applied directly. The Green's function method is equiv-
alent to using an infinite perturbation expansion. It can be applied to both
elastic and inelastic scattering, and multiple scattering events to any order can
be included. The Green's function is regarded as a propagator which carries
the particle from one scattering center to another. The Born approximation;
whiih is valid when the particle energy is large with respect to the interaction
energy, is derived and applied to simple potentials.
SUGGESTED
REFERENCES
David Bohm, Qwntum Theorlt. Prentice-Hal, fnc., New York, 1951.
R. H. Dicke and J. P. Wittke, Introduction to Qtnntum Mechanics. Addison-
E. Merzbacher, Quantum Mechanics. John Wiley and Sons, fnc., New York,
1961.
Albert Messiah, Qtnntum Mechanics. North-Holland Publishing Co., Amster-
dam, 1958.
D. Park, Introduction to the Qu.antum Theor2 McGraw-Hill Book Co., Inc., New
York, 1964.
J. L. Powell and B. Crasemann, Quantum Mechanics. Addison-Wesley Publish-
D. S. Sixon, Elementarlt Quantum Mechanics. Holden-Day, fnc., San Francisco,
1968.
Leonard I. Schifl Quantum Mechanics,3rd ed. McGraw-Hill Book Co., New
York, 1969.
ffipnm*sffi,$ffi
It is difficult to terminate a book at this point because there are so many

additional topics that one would like to include. There is no natural end to
the subject, since quantum mechanics is continually being applied either to
new phenomena or to old phenomena in new ways.
My hope is that those readers who have come this far will have gained
new insight into the physical behavior of our universe on the microscopic
level, as well as some degree of mastery of the mathematical techniques neces-
sary for the solution of problems hitherto undreamed of. It is my further
hope that the reader will proceed either to apply quantum mechanics to
specialized problems in solids, nuclei, biological systems, and so forth, or to
move forward into the realms of relativistic quantum mechanics and quantum
field theory.
&ffiwffiffiffi ffirumffiK
#*iqi'tJ.#H:{$iffi
Alviiger, T., 5 Edmonds, A. R., 338

Alvarez, L. W., 103 Ehrenfest, P.,214
Ambler, E., 286 Einstein, A., 8, 42,50, 55
Eisberg, R. M., 72,110,202,338
Elton, L. R. B.,93,98, ll0
Evans, R. D.,93,98, 110,286
Beams, J..W., 55
Eyring, H., 395
Beiser, A.,72, Il0
Bergmann, P. G.,42
Bivins, R., 322
Blanchard, C. H., 72,110
Blatt, J. M.,2gO
Bless, A. A.,59
Bloch, F., 103, 175
Bohm, D.,244,288, 338, g7Z, 977 , gg5, 412
Bohr, N.,83
Bollman, V. L.,90
Born, M.,72,110, lll
Borowitz, S., 157, 202,288,277
Bothe, W.,59
Bradley, J., 3
Brillouin, L.,372
Brown, A. B, 32
Burnett, C. R., 7Z,ll0
Gamow, G., 164
Garwin, R. L.,286
Geiger, H.,73,74
Gerlach, W., 107
Compton, A. H., b8, llS, Zgg
Germer, L. H., 122,124
Condon, E. U., 164, 281, Zl4, B22, gZB
Goldberg, A., 170, 17l, L72
Cooper, J. N., 42, 72,ll0,lb7,20B,2gB
Goldstein, H.,350
Courant, R., 130, 184, 216
Goudsmit, S. A., 102, 289
Crasemann, B., 140, lb7, Z0g, 24b, 288, BB8,
Gurney, R. W., 164
378,395,4t2
Hartree, D. R., 387, 388

Davisson, C., 122 Hayward, R. W.,286
de Broglie, L., lll, 120 Heisenberg,'W., 63, 64, lll, 386
De Sitter, W.,4 Heitler, W., 362, 390
Dicke, R. H., 157,202, 244,288,338, 978, gg5, Hertz, G., 88, 89
4r2 Herzberg, G.,314
Dirac, P. A. M., 102, l9l, 245,290,338, 978, 986 Hilbert, D., 130, 184, 216
Duane, W., ll3 Hill, T. L.,3.33
Duke, C. B., 166 Hindmarsh, W. R., 338
DuMond, J.,90 Hofstadter, R., 93
415
416 AUTHOR INDEX
Hoppe, D. D.,286 Nijgh, c. J.,322

Hudson, R. P.,286 Noble, H. R.,4
Huggins, E. R., 127
Panofsky, W. K. H.,26,42,3M
Illingworth, K. K.,4 Park, D., 157, 203, 288, 338, 378, 395, 412
Ives, H. 8.,37 Parker, W.H.,47
Pauli, W.,289
Pauling, L.,157 , 189, 203, 273,288,378, E95
Penney, W. G., 174
Phillips, M., 26, 42,364
Jackson, J. D.,42, 864 Phipps, T. E., 108
Jeans, J. H., 46 Planck, M.,47
Jeffreys, H.,372 Powell, J. L., 140, 157, 203,245,288, 338, 378,
Jiinsson, C.,127 395,4r2
Jordan, P., lll Pringsheim ,8., 44
Josephson, B. D.,47
Joy, A. H.,5
Rado, G. T.,290
Rainwater, J.,86
Kiillen, G., 162 Ramsey, N. F., 103
Kemble, E. C.,374 Rayleigh, Lord, 46
Kennard, E. H.,42,72,89, ll0, 157,203,288 Resnick, R.,42
Kennedy, R. J., 4,6 Richtmyer, F. K.,42,72,89, ll0, 157,203,288
Kimball, G.8., 395 Rojansky, V.,157,245
Kjellman, J., 5 Rose, W. G.V.,42
Kramers, H. A.,372 Rosser, W. G. V., 42
Kronig, R. de L., 174 Rotenberg, M.,322
Royds, T.,73
Rusell, H. N.,312
Rutherford, 8., 7 3, 75, 80
Langenberg, D. N.,47
Lauritsen, C. C., 32, 89
Lawrence, E. O.,55
Sadeh, D.,5
Lederman, L. M.,286
Sanders, J. H., 8
Lee, T. D.,286
Sands, M., 72, ll0
Leighton, R. B., 42,72,110, 202, 245,277,279,
Sanford, R. F., 5
333, 339, 378
Saunders, F. A., 312
Lengyel,8.,54
Saxon, D. S., 157, 203, 288, 338, 378, 995,412
London, F., 390
Schey, H. M., 170,171,172
Lummer, O.,44
Schiff, L. 1., 157,203,245,278, 288, 338, 378,
381, 389, 391, 395, 4W,412
Schriidinger, E., lll, 142, 339
Schroeer, J. M.,57
Marsden, 8.,13,74 Schwartz, I. L., l70t 17 l, 172
Matthews, P. T., 202,245,288 Scott, G. D., l4
May, J. W.,124 Semat, H., 72, I l0
McDonald, D. F.,8 Shadorvitz, A.,4L,42
Merzbacher, 8., 157, 190, 203, 245, 264, 288, Shankland, R. S.,6
339, 378, 995,400,4t2 Sherwin, C. W., 198
Messiah, A., 120, 149, 157,245,288,338, 372, Shortley, G. H.,281, 314, 322,338
378,395,412 Slater, J. C., 804, 338, 388
Metropolis, N., 322 Slichter, C. P., 106
Michelson, A. A.,4 Smith, J.H.,42
Millikan, R. A., 54,56 Smith, R. L.,57
M6ller, C.,42 Snyder, C. W., 32
Morley, E. W.,4 Sommerfeld, A., I12, 176
Morse, H. G. J.,90 Sparks, M., 106
Miissbauer, R., 96 Spiegel, M. R.,276
Mulligan, J.F.,8 Stern, O., 107
AUTHOR INDEX 417
290
Stevens, K.W.H., Wapstra, A.H.,322
3
Stewart, A. B., Watson, G. N., 397
37
Stilwell, G. R., Weber, J., 100
72,ll0
Stoner, R. G., Weber, R. L., 72, 100, ll0
Suhl, H.,290 Weinrich, M.,286
Weisskopf, V. T., 14, 386
Wentzel, G.,372
Taylor, B. N.,47 \Vertheim, G. K.,96
Taylor, J.8., 108 Whecler, J. A., 86
Teich, M. C., bZ White, H. E., 310, gl4, g3g
Thomas, L. H., 294, Bgg White, R. L., lS7 ,209,248,299, Bgg
Thomson, G. p., 126 Wien, W.,44
Thomson, J. J.,54,73 Wilson, C. T. R., 59
Thorndike , E. M., 6 Wilson, E. B., 157, 189, 203, 27g,2gg, Z7g, ggb
Tipler, P. A.,72, ll0 Wilson, W., llz
Trouton, F. T.,4 fVittke, J. p., 1b7,202,244,2gg, B3g, g?g, 3g5,
4t2
Wolga, G. J., b7
Wollan, E. O., 125
Uhlenbeck, G. E., 102, 289 \ivooren, J. K., !r.,322
Wu, C. S.,286
Van Lieshout, R., 322

Van Name, F. W., Jr.,42,72,I10 yang, C. N., 286
Van Vleck, J. H., 104, 312,386, 393, 395 young, H. D., 72, lI0, 126
Viner, M. R., 14 yukawa, H., 94, 100
Wallin,I.,5 Zeeman, P., 105

Walter, J., 395 Ziock, K.,203, 245,288
I
-*i.&qffi
SUffiJffiCT ilNPffiK
(Italics refer to illtrstration page numbers.)

Aberration of starlight , 3, 4,4 Annihilation operator, see Creation-annihila-
relativistic equation, I 7 tion operators.
Absorption coefficient, photons, 51, 360, 361 Anomalous gyromagneric ratio for spin, 102.
Action integral, l12 106, l0g,2gg
Accidental degeneracy, 27 8, 285 Anomalous magnetic monrent, 102, 106,289
Addition of angular momenta,295ff, 302, 311ff, Anomalous T,eeman effect, see Zceman cffcct.
321tr Antillonding orbital of hytlrogcn, 392
Additional interactions in the one-electron Anticommutator, 292
problem,300-302 Anti-Hermitian, 2l I
Adiabatic and sudden perturbations, 366-369 Antiproton, 33
Adjoint Antisymmetric wavc funcrion, 168, 169, 172,
of a matrix,226 t74,284, 303ft
of an operator, 196, 197 Application of matrix mechanics to lrarmonic
Alpha emission, 164- 166 oscillator, 240-243
Alpha particle Applications of perturbation theory, see Per-
binding energy, 93-95 turbation theory.
scattering, 74, 77 -80 Approximation methocls in quantum nle-
Ammonia molecule, 190, chanics, 339-378
191
Associated Laguerre polynomials, 276, 286
Angular momenta (momentum), addition of,
Associated Legendre functions, 264
295fr.,302,3utr, 32tff
Asymptotic form
and the rotation operator, 256 for harmonic oscillator rvave function,177
commutation relations, 253, 257 for hydrogenic radial function, 273-275
conservation of., 254, 256
for scattered wave, 398
coupling coefficients, 322ff Atomic magnetic susceptibility, 392-395
definition ot,252 Atomic models, see Bohr, Ruthcrford, and
in rectangular coordinates, 252 Thomson.
in spherical coordinates, 254 Atomic nucleus, see Nucleur
eigenfunctions, 26 I -264 Atomic number,92
normalization of, 264 Atomic orbitals, 280,282,297 , 309
intrinsic,see Spin. Atomic radius, 83, 84, I l0
matrices, 268-271 Atomic scattering factor, 410
operators, 252-257 Atomic spectra, Bohr theory, 83-91
eigenvalues of,257 Atomic units, 157
orbiral, 83, 94, l0l, 102, l0g, 252ff, 297,309 Auger effect, 9l
quantization of, 83, 84, l0l, 102, 105, 107- Average value, see Expectation value.
It0
quantum number, l0l, 287
raising and lowering operators, 25?
spin, 102, 106, 108, 289, 290, 301, 302 Balmer series, 85, 87
total,254,258,295, 302, 3l I Band spectra of molecules, 329
uncertainties in meaurement, 254 Band structure of solids, 175,176
vector coupling schemes, 295, 302,31l-315 Barrier penetration, 162-167
Anharmonic oscillator, 343-846 alpha decay, 164-166
Anisotropic oscillator, 343-347 Basis, change of, 232-235
419
420 SUBJECT INDEX
Basis functions, Hilbert space, 215 Collision diameter, 78

completeness, 216 Collisions, see Scattering.
linear independence, 216 Commutation relations
normalization, 217 angular momenta, 253, 257, 297
orthogonality, 2Lb coordinates and momenta, 150, 214, 252
Beta functions,265 oscillator raising and lowering oPerators,
Binary stars, light from,4, 5,37 193
Binding energy Pauli matrices, 292
of alpha particle, 93-95 rotations, 256,257
of deuteron, 93 Commutative proPerty, 146, 147
of electrons, 84 Commutator bracket, 147, 213
of solids, 97 Commuting operators, and compatible observ-
per nucleon, 93 ables, 212-214
Blackbody radiation, 43-45,50, 53 simultaneous eigenfunctions, 2l 3
experimental cuwe,44 Compatible observables, 2L2, 213

Planck law, 46-50 Complementarity princiPle, 61
Rayleigh-Jeans law, 45, 46, 49 Completeness, of basis functions, 216
spectral distribution, 43 Compton, effect, 57-60, 100
Stefan-Boltzmann law, 44, 45, 49, 50 scattering, 57-60, 100
Wien displacement law, 44, 45, 49 wavelength for the electron, 59, l0l, 156
Bloch function, 175 Concept of the nuclear atom, 73-ll0
Bohr, correspondence principle, see Corre- Configuration, electronic, 310, 335ff
spondence principle. Conjugate variables, 63, I l2
magneton, l0l, 102, 105, 108 Connection formulas for semiclassical rnethod,
radius,84, 156 37ltr
principle of complementarity, 6l Conservation
theory of atomic spectra, 83-91 of angular momenturn , 33,254'256
Boltzmann factor, 48, 51, 53, 67'70, 394 of energy, during transitions, 359
Bonding orbital, hydrogen, 392 in virtual processes, 99
Born approximation in scattering theory, 403- of linear momentum, 33, 254
405, 408 of parity, 285
Bose-Einstein statistics , 305, 334 of probability density, 143'146, 152
Bosons, 305, 308, 333fi of relativistic mass-energY, 20, 33
Bra vector, 222 of spin, 35
Bra-Ket notation, 221 -n4 Continuous spectrum, 89, 90
Brackett series, 85 Contraction of lengths, 5,ll, 12,40
Bragg equation, 11J, ll4,122 Coordinate representation, 139
Bragg reflection, ll8, 122, 123, 126 Coordinate-momentum domain, I32
Bremsstrahlung, 89 Coordinates, for the two'body problern, 249
planar and spherical, 115
Correspondence principle, 91, ll3, 135, L42,
146,148, 176
C frame,28,29 Coulomb, batier,165
scattering cross section in, 80-82 integrals, direct and exchange, )84,385
Canonical momentum, 63, l12, 350 potential, 94, 95, 248, 272, 294
Center of mass coordinates, two-body problem, scattering, 77-80, 410
249,286 Coupled angular momenta, 321ff
Centrifugal barrier potential, 272, 27 3 Coupled oscillators, 189-l9t
Change of basis, unitary transformation, 232' Coupling, j-j, 3l l, 313ff'
235 Russell-Saunders, 312, 320
Characteristic equation of a matrix, 23I Covariance, 3
Characteristic x-ray spectrum, 89, 90 Covariant laws, l, 2, 3, 8, l0
Charged particle in static electric and mag- Covalent bond, 390-392
netic field, 349-353 Creation and annihilation of particles, S2-35
Chopping an infinite sine wave, 136-138 Creation -annihilation oPerators
Classical principle of relativity, l-8 for angular momentum states, 257ff
Clebsch-Gordon coefficients, 322, 824 for oscillator states, 202
Clocks for photons, 362
running slow, 13, 14,37 Cross iection, classical scattering, 75-80
synchronization of,9, 13, l5 differential, 76, 397, 403, 4ll
Closed shell structure, 309 in C and L frames, 80-82
Cofactor matrix, 226 total, 77, 398ft
Coherent radiation, 5I, 53, 129 Crystal diftraction, I 13, I14, 122, 123' 126
SUBJECT INDEX 421
Crystalline electric field splitting, 357 Ehrenfest's theorem, 214

Current, probability density, 144, 145, l6l, 173 Eigenfunction(s), definition, Lll, 152
Current density four-vector, 25 of angular momentum operators, 261
of commuting operators, 213
of coupled angular momenta, 321-326
of harmonic oscillator, I88
d states, 280, 282, 297, 310 of hydrogenic atoms, 279
Davisson-Germer experiment, 122 of particle exchange operator, 303
de Broglie, relation, 120, 128, 135,142 of spin operators, 29I,302,326fr
wavelength , 75, ll4, 120, l2l, 122, 130 of the square well, 168, 169
Degeneracy, accidental, 27 8, 285 Eigenstate, definition, 153
exchange, 307ff, 328, 383ft Eigenvalue(s), definition, 65, l5l, 152
for particle in a box,248 energy, see Energy eigenvalues.
of free particle states, 159 of the angular momentum operators, 257-
of hydrogenic wave functions, 278 261
of isotropic oscillator states, 113,248 of the hydrogenic radial equation, 277
spin, 291 of a matrix,23l,236
Degenerate states, I 13, I 19, 218, 307 problem, matrix form of, 230,231
perturbation theory for, 354-357 result of a physical measurement, 146, 207,
Delta function, 130-135 208
normalization, 138, 139 Eigenvectors of a matrix,236,237
Density of states, 248,359,388, 402 Einstein, coemcients, 50-54, 361, 362
Determinant of a matrix,226 photoelectric equation, 55
Deuteron, binding energy, 93 special theory of relativity,8
magnetic moment, 103 Elastic scattering, 30, 75'82
spin, 103 Electric dipole, matrix elements, 361, 362
Development of wave mechanics, ll l-155 moment, 356
Diagonal matrix, 231 operator, 361
Diagonalization of rnatrices, 235 -240
' radiation, 52, 361-364
Diamagnetic term of the Hamiltonian, 352, selection rules, I 13,315, 362
393 Electric quadrupole moment, 364
Diatomic molecule, potential energy of, 189 Electromagnetic field four-tensor, 26
symmetry and statistics, 328-330 Electromagnetic potential four-vector, 25
Difierential scattering cross section, 76 Electromagnetic theory and Galilean trans-
in C and I, frames, 80-82, 397,403 formation, 3, 8
Diffraction, Il3, ll4 Electron(s), affinity for, 310
Bragg equation, I l4 charge,54
of electrons, 122-124, 1 27 de Broglie wavelength, 122
of neutrons, 12J equivalent, 312, 313, 326
of particles, 12l-126, 127 gas, 388, 389,402
of photons, I l3 inequivalent, 3l I
Dirac, bracket notation, 154, 184 in a periodic potential,lT4, 175
bra-ket notation, 221 -224 nragnetic moment of, l0l, 102
delta function, I33-135 mass of, 54
Dispersion relation, 62 photoemission of, 54-57
Distribution function, Boltzmann, 48, 51, 53, recoil in Compton effect, 59, 60
67-70, 394 spin, 35
Bose-Einstein, 334 statistics, 305
Fermi-Dirac,334 valence, 310
Doppler shift, binary stars, 4, 5 Electronic configuration, 310
relativistic, 35-38 of the elements, 335ft
Double oscillator p,otential, 189, 190 Elements, table of, see inside front cover.
Double slit interference,126, 127 Emission, of photory, induced or stimulated,
Dual nature of the photon, 60-62 51,360,361
Dulong-Petit law of specific heat, 50 spontaneous, 5 l, 362
theory for velocity of light,4, 5
Endoergic and exoergic reactions, 3l
Energy, available in C and l, frames, 33,34
Effective magnetic moment, 316,317 bands in solids, 175,176
Effective mass, 87 binding, see Binding energy.
Effective radial potential, 272, 27) conservation during transitions, 359
Efiects of exchange symmetry on scattering crystal field splitting, 357
cross sections, 4l I -412 diamagnetic, SS2
422 SUBJECT INDEX
Energy (Continued) Force (Continued)

eigenequation, 150, 152, 153, 206 nuclear,80,93-95, 100
eigenvalues, 150-153 spin-orbit, 293-295
coupled oscillators, 190 Formal structure of quantum mechanics, 204-
free particle,246 245
harmonic oscillator, l7g, 180 Fourier integral theorem, 129-136
hydrogenic atoms,277 Fourier transforms, 132, 139, 140,410
infinite square well, 172 Four-scalar, 27
molecular rotations, 329ft Four-vector notation, 2l-?7
molecular vibrations, 189 Frame of reference, C and L,28-30
particle in a box, ll7 ,247 inertial, 2
rigid rotator, 114 Franck-Hertz experiment, 88, 89
upper bound from variational method, Free particle, degeneracy of states, 159
369-371 wave functions for, I38,246
equipartition of, 46 Fundamental forces and exchange particles,
exchange, 307ff, 385, 386 99-r0t
forbidden bands of, 176 Function space, 215
operator, 148, 149
recoil, 59fi, 96ff
relativistic, 19
rest mass 19
g factor, 102, 315ft
spin-orbit, 294,353 Galilean transformation, 1,2, 6
Stark splitting, 355, 356 and Maxwell's equations, 3, 8
Zeeman, see Zeeman eftect.
Gamma emission, decay of positronium, 35
Equipartition of energy; 46 from excited nuclei, 95-97
Equivalent electrons, 312, 313, 326 pion decay, 38
Ether theory, 3,4,6,7 recoil-free, 96-97
Exchange, Coulomb integral, tS4' 385 Gamma functions,265
degeniracy and exchange energy, 307-309 Gamma rays, and pair production, 34,99
force, 308 Gaussian. packet, 138, 139, 140, 141, 146, 149
operator, 303, 306, 307 breadth of, 1)9
particles, 99-101 scattering from barrier, 170
symmetry, 303, 307 tr, 328 scattering from square well, 171
Excited states of helium, 383-386 Gauss' probability integral, 71
Exclusion principle, see Pauli principle. Generalized coordinate, 1 12
Expectation value, definition, 147 Generating function, Hermite polynomials,
from Hermite polynomials, 186-188 180-193, 196
in the operator formalism, 200-202,210 Geometric representations of space-time, 38-41
time-dependence of., 213, 214 Golden rule, 359, 402
Euler identity, l3l Gradient operator in spherical coordinates, 283
Event, space-time, 10, 38
Grand partition function, 333
Graviton, 100
Green's function method for scattering, 400,
406-408
Factorization method, for angular momentum, Green's theorem, 144
26t-264 Ground state electron configuration for the
for harmonic oscillator, l9l'200 elements, 335-336
Fermi energy, 388-390 Ground state energy
Fermi gas of electrons, 388, 389, 402 harmonic oscillator, I dimension, I95
Fermi unit, 80 2 dimensions, 113,220
Fermi's Golden Rule, 359,402 3 dimensions,24S
Fermi-Dirac statistics, 305, 334 Jrelium atom, 380, 382
Fermions, 305, 308, 333ft hydrogen, 84, 156, 348
Feynman diagram, 100 '' particle in a box, 248
Fine structure, of atomic sPectra, 114,289,295 singly ionized lithium, 383
constant, 119, 156 Ground state of the helium atom, 379-383
Finite potential barrier, 162'L67,
Group velocitY,6l,62, l2l, 130, 135
eigenfunctions for, I 63 Gyromagnetic ratio, 102
FitzGerald-Lorentz contraction ,5,7 , ll,12, 13,
40
Force, exchange, 308
four-vector, 24
fundamental, 99-101 Half-life, 166
SUBJECT INDEX 423
Hamiltonian, for a charged particle in an elec- Hydrogen (Continued)

tromagnetic field, 349 energy levels, 84,277
harmonic oscillator, 46, 176 hyperfine correction, 301
hydrogen atom or hydrogenic atom, 250ff, ionization energy, 85
272 molecule, 328ff, 390-392
hydrogen molecule, 328ff orbital velocity of electron, 156
two-electron atom, 379ft ortho and para, 329
Hamiltonian operator, 143 relativistic correction, ll9, 300, 301, 348
Harmonic oscillator, 45-48,91, l12, ll3, 343- Schriidinger theory, 272-282
348 Sommerfeld theory, I 14-120
anharmonic perturbations, 343 spectral series, 85
applications to molecule, 189-l9l spin-orbit correction, 293-295
average energy, 48 rvave functions, 279
blackbody r adiation, 45 -47 normalization, 276
double oscillator, 189-l9l parity,285
energy levels, I79,180 table of,279,280
Hamiltonian for, 46, 176 Hydrogenic, atoms, 83-88, 272-282
matrix elements, 185-188, 202, 240fr energy corrections,
normalization of eigenfunctions, 183-188 hyperfine, 295,30L
operator method, l9l -200 relativistic, 301
parity of eigenfunctions, 285 spin-orbit, 294ff
polynomial solution, 176-180 radial wave functions,
27 6, 277
raising and lowering operators, 193, 194 states, magnetic moments of,282ft.
three dimensional, 248 Hyperfine interaction, 295
two dimensional, ll3, 220
variational method, 370
wave functions, 184, 185, 188
Wilson-Sommerfeld method, I 12 ldentical particles, exchange, and symmetry,
Harmonic perturbations, 359-366 302-207
Hartree self-consistent method, 387 probability density, 306
Heisenberg, equation of motion for an opera- wave function for, 302ff, 328tr
tor,2l4 Identity operator,220
exchange interaction, 386-387 Imaginary matrix, 228
formulation of quantum mechanics, lll, Impact parameter, 76, 77 , 80,94
230-243 lndeterminacy principle, see Uncertainty prin-
nncertainty principle, see lJncertainty principle.
ciple. Index of refraction for electrons, 123
Helium, ortho and, para, )84 Inelastic scattering, Q equation, 30, 3l
atom, ground state, 379-383 Inertial frame, 2, 8
lowest excited states, 383-386 Infinite square well, 169, 172
Hermite equation, 178, 182 Infinitesimal rotation, 256ff
Hermite polynomials, calculation of expecta- Infinitesimal translation, 255
tion values, 186-188 Interactions of charged particle with static
generating functions, 180-183, 196 electric and magnetic fields, 349-353
identities, 182, 183 Interval, space-like, 27 , 41
normalization, 180 time-like, 27, 4l
orthogonality, 184 Invariance, Lorentz, 22, 27
parity of, 180 rotational, 256
recursion relation, 178 translational, 255
Hermitian, adjoint, operator, 208, Inverse, of a matrix product,227
matrix, 226 matrix, 226
conjugate operators, 196, 222 operators, 221
vectors, 222 " photoelectric effect, 89
form matrix,228 Ionization energy, helium, 87
operator, 208 Y\ hydrogen, 85
matrix, 2?8 hydrogenic atom, 85
operator, 154,207 -2ll lithium,383
property, 154,208 Isomorphism between operators and matrices,
symmetry;215 23t
Hilbertspace, 215
Hund's rules, 314, 321
Huygens wavelet, 137, 396
Hydrogen, Bohr theory, 83-88 Jeans' number, 46, 49, 65-67
424 SUBJECT INDEX
j-jcoupling,98, 311ff Magic numbers, 98

two equivalent p electrons, 313 Magnetic, dipole energy, 105, 317
two inequivalent p electrons, 3ll dipole transitions, 364-365
JWKB method, 371-377 exchange energy, 386, 387
moment, anomalous, 102, 106,289
definition, l0l
deuteron, 103
Ket vector, 221 effective, 316, 317
Kronecker delta,2l electron, l0l, 102
Kronig-Penney model, 174, 175 neutron, 103
operator, 284
of hydrogenic electrons, 282-284
orbital, see Orbital magnetic moment.
L-frame, 28,29, 30 proton, 103
scattering cross-section in, 76-82 spin, 102, 103
Lagrange's multipliers, method of, 68 quantum number, II7 ,287
Lagrangian for charged particle in an electro- resonance, 359
magnetic field, 350 susceptibility of an atom, 392-395
Laguerre polynomials, associated, 276 Magnons, 202
Landd g-factor, and the Zeeman effect, 315-321 Many-electron atoms, 309ff
Laplacian operator in spherical coordinates, Hartree model, 387
251 Thomas-Fermi model, 388, 389
Larmor, frequency, 104-106, 293 Maser, 52-54
precession, 104, 105, 293, 296 Mass, eftective, 87
theorem, 103-106 reduced, see Reduced mass.
Laser, 52-54,57 relativistic change of, l8
Laue diffraction pattern, 125,126 Mass defect,93,94
Legendre functions, associated, 264 Mass-energy, conservation of, 20, 33
Legendre polynomials, normalization, 397 equivalence, 19, 32
table of, 264 Mass number, 92
Length, contraction of, ll, 12,40 Matrix (matrices), adjoint, 226
Light, Doppler shift, 4, 5 adjoint of a product, 228
emission theory,4,5 angular momentum, 267 ft , 298fr
ether theory, 3, 4, 6, 7 cotactor,226
speed of, 8 column,225
limiting speed, I6, 20 complex conjtgate,227
Linear independence, of a set of functions, 216 of a product, ?28
Linear operator, definition of ,146,207 deierminant of ,226
matrix representation of, 224-230 diagonal,2Sl
Linear transforma tions, 220, 221 eigenvalues of,23l
Linear vector spaces, Zl5-217 eigenvectors of , 236, 237
Linewidth, natural, 63, 96, 97 element, 185-188, 202, 228
Doppler broadening, 37, 97 harmonic oscillator, 185-188, 240ff
recoil broadening, 59, 60, 96, 98 Hermitian,228
Liquid drop model of nucleus,98 imaginary, 228
Lithium, ionization energy, 383 inverse, 226
singly ionized ground state energy, 383 of product, 228
Lorentz, condition, 349, 351, 352 Lorentz transformation, 22, 23
covariance, 10,27,41 non-singular, 225
force,3, 106, 349, 350 orthogonal, 228
transformation, 9, 10 rcal,228
matrix of, 22 representation of a linear operator, 224-230
matrix of inverse, 23 row,229
triplet, 105 secular equation, 231
Lorentz-FitzGerald contraction , 5,7 ,lL, 12, 13, skew-Hermitian, 228
40 skew-symmetric, 228
Lowering operator, 194 spin, 291ff
angular momentum, 257 spin-orbit operator, 325
harmonic oscillator, 193 symmetric,228
Lowest excited states of helium, 383-386 trace of,234
LS coupling scheme, see Russell-Saunders transpose, 226
coupling. of a product,228
Lyman series, 85, 96 unitary, 228
SUBJECT INDEX 42s
Matrix form of the eigenvalue problem, 230, Normalizat ion (C ont inue d)
23r of harmonic oscillator wave functions, 183-
Matrix mechanics, lll 188, 196, 197
application to the harmonic oscillator, 240- Nuclear, atom, 73-ll0
243 barrier,164-166
Maxwell velocity distribution ,70-72, 109 binding energy, 93
Maxwell's electromagnetic equations, 3, 8, 349 force, 80, 93, 94, 95, 100
Maxwell-Boltzmann distribution, 48, 51, 53, magnetic moment, 103
67 -70, 394 magneton, 103
Measurement, of compatible observables, 2I2- models,98, 99
2t4 liquid drop,98
Mesic atoms, 86 optical, 98
Meson, 100 shell, 98
mu, 16,86, 87 radius, 80, 92, 93, 94, ll0
pi,34, 35,38, 100, 101 spin, 104
Meson theory of nuclear force, 100, l0l structure and spectroscopy, 92-99
Metastable state, 52, 53 Nucleon, 92
Methods of generating Hermite polynomials, Nucleon number, 92
180-183 Null operator,220
Michelson-Morley experiment, 4-8 Null vector,2l5
Minimum uncertainty product, see Uncer- Number operator, 202
tainty product.
Minkowski diagram, 40, 4 I
force,24
Mirror nuclei,94 Observables, operators, and expectation values,
Mixed state, 109, 210 t46-I50,207
Molecular vibrations, 189-t9l Odd and even functions, see Symmetric and
Molecules, homonuclear antisymmetric.
nuclear spin eftects, 329, 330 Old quantum theory, l12-114, ll9
ortho and para forms, 329,330 One-electron atom, see Hydrogen atom and
rotational angular momentum, 329 hydrogenic atom.
rotational states, 329ff Operator, angular momentum, 252-257
Momentum, conservation of, angular, 254,256 anti-Hermitian, 2l I
linear,254,255 creation and annihilation, 202, 257
four-vector, 24 electric dipole, 361
operator, 148, 149 energy, 148
representation, 139 equation of motion,2l4
Morse potential, 189, 190 gradient in spherical coordinates, 283
Moseley's law,90 Hermitian, 154, 207 -2ll
M<issbauer eftect, 96, 97 idenriry, 220
Multiple square wells, 174, 17 5 inverses, 221
Multiplicity, 814 Laplacian in spherical coordinates, 251
Multipole radiation, 363ff linear, 207
Muon, 16,86,87 maBnetic moment,284
Muonium, 86 matrix representation of, 224-230
Mu-mesic lead,87 non-singular, 221
null,220
parity, 284ft
particle exchange, 303, 306, 331
Natural linewidth, see Linewidth. projection, 223, 224, 322
Natural units, 156,157 raising and lowering, angular momentum,
Neutrino, 305 257
Neutron, magnetic moment of, 103 harmonic oscillator, 193
number, 92 right and left inverses, 221
spin, 102, 305 rotation,256
statistics, 305 spin-orbit, 320,325
Non-singular matrix, 225 total angular momentum, 254
Non-singular operator, 221 translation, 255
Norm of a vector,2IT Operator solution of the harmonic oscillator,
Normalization, 129, 130 l9t-200
angular momentum eigenfunctions, 264-268 Optical model of nucleus, 98
box,247 Optical theorem in scattering theory, 399
delta function, 138, 139 Orbital angular momentum, 83, 84, l0l, 104,
426 SUBJECT INDEX
Orbital angular momentum (Continue d) Perturbation theory (C ontinue d)

105, 108, 252tr,287 , 309 to grotrnd state of helium, 380-382
Orbital magnetic moment, l0l, 104, 105, 287' to harmonic perturbations, 359-366
315 to isotropic oscillator, 357
Orbits, in Bohr theory, 83-91 to magnetic susceptibility of an atom, 392-
in Sommerfeld theory, ll4'120 395
Ortho, deuterium,330 to relativistic correction, 348
helium, J84 to scattering,402-405
hydrogen,329 to spin-orbit energy, 353
positronium, 36 to Stark eftect,348, 355, 356
Orthogonal matrix, 228 to two-electron problem, 384ft
Orthogonal transformation, 235 to Zeematr effect, 352,353
Orthogonality, 215 degenerate states, 354'357
of associated Legendre functions, 264 first-order corrections, 341, 342
of Hermite polynomials, 184 second-order corrections, 343
of the spherical harmonics, 267,268 stationary states, 339-348
of vectors,215 sudden approximation, 366-369
Orthonormality, 206 time-dependcnt, 357-366
Oscillator, see Harmonic oscillator. Pfund series, 85
Phase shift, partial wave, 398tr, )99
Phase velocity,6l,62, l2l, 130
Phonons, 50,97 , 202
p states,280,297, 309 Photoelectric effect, 54-57
P-wave resonances, 401 Photon, as a gas, 50, 99
Pair production, 34 as a quantum of radiation, 34, 47, 64, 360-
virtual, 99 362
Para, deuterium, 330 as a wave packet,60, 64, 120, 126
helium, J84 dual nature of, 60-62, 64
hydrogen, 329 energy of, 35
positronium, 86 momentum of,35
Parity, 180,284ft, 315 particle nature of, 54-60
conservation of, 285 spin, 35, 64,102
of harmonic oscillator eigenfunctions, 285 virtual, 99
of Hermite polynomials, 180, I81,285 Pion,34,35,38, 100
of hydrogenic wave functions, 285 Planck's constant, 47,56,63, 83, 90
of spherical harmonics, 285 Planck's theory of radiation, 35, 46-50, 52, 83,
of square well eigenfunctions, 285 120
operator, 284-286 Poisson's equation, 389
Partial wave treatment of scattering, 396-402 Position four-vector, 22
Particle(s) Positron, 34,35
as wave packets, 135-141 emission of, 94
charged, in an electromagnetic field, 349 virtual, 99
creation and annihilation of, 32'35, 38, 144 Positronium, 86
exchange, 99-101 decay of, 35
operator, 303, 306, 331 Postulates, of Bohr theory, 83
free, 138, 159, 246 of Planck theory of radiation, 47
identical, 302ft, 328ft of quantum mechanics, 204-212
in a box, Il4,172,2.47 of special relativitY, 8
nature of photons, 54-60 Potential, centrifugal, 272
resonances, 173 double oscillator, I89, 190
symmetry and statistics, 328ff finite barrier, 162-167
Partition function, 69, 333 four-vector, 25
Paschen series, 85 lVlorse, 189, 190
Paschen-Bach efiect, 3L9, 353 multiple wells, 174ff
Pauli principle, 304, 809, 312,313, 314 periodic lattice, 87
Periodic chart, 309-31 I rectangular coordinates, 247
table, see inside back cover. scattering, see Scattering.
Perturbation theory, adiabatic approximation, spherically symmetric, 248, 272
366
square well, 167ff
application, to anharmonic oscillator, 343- step, 158-162
347 Precession, Larmor, 104, 105, 293, 296
to crystal field sPlitting,357 Thomas, 294
to electric dipole radiation, 361-364 Principal axes theorem, 231
SUBJECT INDEX 427
Principle of relativity, see Relativity principle. Ramsauer-Townsend efiect, 164, 402

Principle of superposition, see Superposition Range of nuclear force, 94, 95,100, l0l
principle. Rayleigh-Jeans theory, 45, 46, 49
Probability , 126, 128, l4l, 143, I47 ,205 Real matrix,228
densiry, I28,205 Recoil energy, atom in photon emission, 96-98
conservation of, I43, 145,152 electron in Compton scattering, 59, 60
current, L44, L45, I6l, 173 Rectangular coordinates in three dimensions,
harmonic oscillator states, .198 246-248
radial,278 Recursion relation, for associated Laguerre
of transition, see Transition probability. polynomials,2T5
Product space, 321,326 Reduced mass, 80, 83, ll5
Projection operator, 223, 224, 822 Reflection coefficient, at potential barrier, 160,
Propagators in scattering theory, 408 16l, 163
Proper frequency, 36 Relative coordinate, two-body problem, 249ft,
Proper length, 11, 12, 13,40 286
Proper time interval, 11, 13,27, 40 Relativistic, collisions, 27-32
Proper velocity, 17 ,24 Doppler shift, 35-38
Proton, magnetic moment, 103 dynamics, l5-21
spin, 102 energy, kinetic, l9
statistics, 305 rest mass, 19
Pure state, 109,210 total, 19
energy correction, in one-electron atoms,
301, 349
kinematics, 6-15
Q equation, 3l mass, 18
Quadrupole, electric, selection rules, 364 momentum,20,l2l
Quantization, of action, ll? theory of hydrogen atom, 114-120
of angular momentum, ll7 Relativity principle, 10, 87
spatial, 107-109 , classical or Galilean, l-8
Quantum constant, 47, 56, 63, 83, 90 Einstein's special theory, 8-42
Quantum mechanics, postulates of , 204-212 Representation, of an operator by a matrix,
Quantum number, angular momentum, l0l, 224-230
287 of a vector, uniqueness of, 216
azimuthal, ll8 Resonance, 98
good, 286, 289, n3,296, 3I2, 320 s- and p-wave,40l
magnetic, 1L7,287 Resonant absorption of photons, 360
orbital, 261 Rest mass energy, 19, 123
principal, 275,287 Right and left inverse operators, 221
spin of electron,289, 290 Rigid rotator, il4
spin of nucleus, 301 Rotation operator,256
total, 118,275 Rotational states of dumbbell molecules, 328
Quantum statistics, 333ff Russell-Saunders coupling, 312ff, 320
Quantum theory, beginnings of, 43-64 two equivalent p electrons, 313
Bohr theory of the atom, 83-91 two inequivalent p electrons, 312
Heisenberg method, lll, 230-248 Rutherford model of the atom,73-75,92
Planck theory of radiation, 46-50 Rutherford scattering, 77-80, 410
Schrddinger theory, 141ff Rydberg unit, 85, 156
Wilson-Sommerfeld theory, I t2-120
Radial probability density, 278, 279, 280, 281, s states, 280,297,309

282,287 S-wave, resonances, 401
Radial wave function,276, 277 scattering, 400
normalization of,276 Saturation of nuclear force, 95
Radiation, blackbody, 43-45 Scalar product, 215
electric dipole, 361-364 Scattering, 28, 7 5-82, 396-412
Planck theory, 46-50, 52 alpha particles, 74, 77 -80
Radioactivity, 95 Born approximation, 403-405, 408
alpha decay, 164-166 Compton,5T-60
Raising and lowering operators, angular mo- Coulomb, TT-80
mentum,257 elastic, 30,75-82, 403
harmonic oscillator, 193, 194 electrons by atoms, 409-4ll
428 SUBJECT INDEX
Scattering (C ontinue d) Spectroscopy (C ontinue d)

Gausian packets, 170, 17 1 molecular
Green's function method, 406-408 rotation states, 328-330
hard sphere,27,77 vibration states, 189-191
inelastic, 30, 31,88 natural linewidth, 63,96, 97
partial wave treatment, 396-402 notation, 314
perturbation treatmen t, 402-405 recoil broadening, 59, 60, 96, 98
propagators, 408 Russell-Saunders coupling, 31Zff
resonances in, 98, 401 selection rules, I13, 315, 362, 364
S-wave, 400 series, 85, 87
spin dependence,4llft splitting factor, 102
square'well, 17,400ft Stark effect,' M8, 355, 356
Scattering amplitude, 397 Zeeman efiect, 103-106, 317-319
Scattering cross section, classical, 75-80 Spectrum of states, 138
difierential,76, 397 , 4ll Spherical Bessel functions, 397
C and Z frames, 80-82 Spherical harmonics, 267, 268, 881
total,77, 398ff as basis functions, 267
Schmid t orthogonalization procedure, 218 -220 as eigenfunctions of the Laplacian operator,
Schriidinger wave equation, 254fr,286
introduced , l4L-143 Spherically symmetric porentials, 248-252
postulated,206 Spin, and scattering cross sections, 4l lff
separation of, 150-153 degeneracy, 291
time-independent, 151, 152 functions, two particle, 302fi, 326fi
in rectangular coordinates, 246 Hamiltonian,290, 386
in spherical coordinates, 249ft in the Schrfiinger formulation, 290-293
Schwartz inequality, 217 intrinsic, l0?, 106, 108, 115,289,301, 302
Secular equation, of a matrix, 231 mlgnetic tnoment, 102, lOt
Selection rules, ll3 precereion of, ?90
electric dipole transitions, 315, 362 quantum number, 289, 290
electric quadrupole transitions, 364 wave functions, ?91
magnetic dipole transitions, 364 Spin-lattice relaxation time, 106
Self-consistent method, 387 Spin-orbit, energy, 294,295, 3I9
Semiclassical approximation, 37 l-37 7 interaction, 86, I15, 2W-n5, 320
Separation of space and time, in the Schriid- operator, 293,320,325
inger equation, 150-153 Spinors, one-particle, 29L
Shell model of nucleus, 98, 104 two-particle, 302ff ,306, 326ff
Shell structure of atoms, 309 Spontaneous emission of photons, bl,3Oz
Similarity transformation, 235 Spread of a packet, 138ff
Simultaneity, 9, I l,12,27,39, 40 Square well, in one dimension,16T-I74, 167
Skew-Hermitian, matri x, 228 eigenfunctions for, 168, 169
operator,2ll multiple wells, 174-I76
Skew-symmetric matrix, 228 scattering, l7l
Slater determinant, 304 Stark efiect
Solid angle transformations for C and L Iinear oscillator, 348
frames, 80-82 hydrogen, 355, 356
Solids, band theory, L75, 176 State, see Wave function.
Solutions of some one dimensional systems, Stationary state, l5l-153
158-202 Statistics
Sommerfeld's relativistic theory of hydrogen, Boltzman, 48, 51, 53,67-70,394
I 14-120 Bose-Einsfein, 305, 334
Space, homogeneity and isotropy of ,254,286 Fermi-Dirac, 305, 384
Space-like interval, 27, 41 Stefan-Boltzmann law, 44, 45, 49, 50
Space-time, geometric representations of, 38- Step potentials, 158-162
4L eigenfunctions for, 1 60
Spatial quantization, 107-109 Stern-Gerlach experiment, 107-109
Special theory of relativity, l-42, Lzl Stimulated emission, see Emission of photons.
Einstein's postulates, 8, 4l Sudden approximation, 366ff
Spectroscopic splitting factor, 102 Superposition principl e , 6l , 129 , I42
Spectroscopy, atomic, 83-91 for physical states, 206,207
Doppler shift, 37, 97 Susceptibility, Van Vleck, 395
fine structure, ll4 Symmetric and antisymmetric wave functions,
hyperfine structure, II4-120, 295 169, 169, 172, 174,284, 303ff
j-j coupling,3llfi Symmetric matrix, 228
SUBJECT INDEX 429
Symmetry Uncertainty product, 138, 149, 173, 199

and degeneracy, 285 absolute minimum, 138, 149
inversion, 284,287 , 304 time-dependence of, 138
rotational, 256ff Unitary matrix,228
spherical, 248, 272, 286 Unitary transformation, 231, 232-234
spin and statistics, 305
translational, 255fr
under particle exchange, 303,304
Synchronization of clocks,9, 13, l5
Valence electrons, 310
Systems of identical particles, 328-330
Van Vleck susceptibility, 395
Variational method, 369-37 I
for helium, 382ff
Temperature, statistical, 53 for hydrogen molecule, 391
Terms, spectroscopic, 3I l -315 for lithium ionization energy, 383
Thomas, factor, 294, 295 Vector, axial,252
precession, t94 coupling coemcient, 322, 324
four-dimensional, 2l -27
Thomas-Fermi statistical model of the atom, model for combining angular momcnta,295-
388, 389
Thomson model of the atom,73-7b 300,31Iff
Time, and simultaneity, 9, LL,12,27 polar,252
dilation, 13,37, 40 potential, 25,349,351
proper, ll, l3 pseudo-, 252
Time-dependent perturbation theory, space, linear, 215,217
Bb7 -266
Time-frequency domain, 132 Velocity, four-vector, 24
Time-independenr Schrddinger equation, of light, 8, l2l
lbl- limiting, 16,20
r52
in rectangular coordinates, 246 proper, 17
in spherical coordinates, .relativistic transformation, l5
249ft
Time-like interval, 27, 4l Vibrational states of molecules, 189-191, 328
Total angular momentum operator, ZS4,2b8 Virial theorem, 282
Trace, of a commutator,2ZE Virtual photons,99
of a matrix, 234 Virtual scattering states, 401
of a product, 234
of a sum, 235
Transformation
Lorentz, 9 Wave, amplitude of,6l, 128
orthogonal, 235 equation, electromagnetic, 10, 6,1'r, 66
similarity, 23b Schriidinger, in one dimension, 158-203
unitary, 231,232-234 in three dimensions, 246-288
Transirion probability, 958, 3bg function(s), definition of, 128, 141,204
Einstein coefficients, 50-54 double squarc well, 189, 190
harmonic perturbation, 360 electron, 126,291
Transition rate, induced and spontaneous, bl, for N electrons, 3M
359, 360 for two electrons, 304
Translation operator, 255 free particle, 138, 246
Transmission coefficient, barrier, 160, 16l, l6), harmonic oscillator, 184, 185, 188
164, 165 hydrogenic atoms, 279, 280
Transpose of a matrix, 226 isotropic oscillators, ll3, 220, 248
Trouton-Noble experiment, 4 orthonormality condition, 205, 206
Tunneling particle in a box, ll4,172,247
in ammonia molecule, 190. l9l periodic potential, I 75
through barriers, 162-167 rigid rotator, ll4
Two-body problem treated as one-body prob- spin, electron, 291
Iem,248, 249,286 nucleus, 302
Tu'o-particle spin functions, 802ft, 306, 326-BZ8 superposition postulate, 129, 142, 206
symmetry of, 284, 304, 328ft
mechanics, development of, I I l-155
nature of particles, 120, l2l
Ultraviolet catastrophe in radiation theory, 46, packet, 62, l2l, I 35-l4l
48 superposition of, see Superposition principle.
Uncertainty principle, 62-64, 99, 100, 137, 138, Wein's displacement law, 44,49
t40, I4l, l4g,172, l74, g5g radiation law,45,49
430 SUBJECT INDEX
Wilson-sommerfeld quantization rules, r l2-120 Young's double slit experiment,126, 127

WKB method, 371-377 Yukawa, particle, 100
Work function, S5 potential, 94,95
World line, 38
World-time diagram, 40, 4 I
Zceman, effect, anotnalotts, 106, 289, 317, 318,

319, 352, 365
X-ray cutoff wavelength, 89,90
normal, 103-106, 287 , 317 ,319, 352
X-ray spcctra, 89-91
by perturbation thcory, 352, 3l'r3, 365
Auger effect,9l
Zeeman term of the Hamiltonian, 352,365
Bremsstrahlung, 89
characteristic spcctrum, 90, 9I Zero-momentutn fram e, see C-frame.
energy levels, 9l Zero-point motion, 47, l7g

Elmer E. Anderson - Modern Physics and Quantum Mechanics-W. B. Saunders Company (1971)

Uploaded by

Copyright:

Available Formats

Elmer E. Anderson - Modern Physics and Quantum Mechanics-W. B. Saunders Company (1971)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Elmer E. Anderson - Modern Physics and Quantum Mechanics-W. B. Saunders Company (1971)

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the first four chapters?

What topics are covered in the first four chapters?

MODERN

W. B. SAUNDERS COMPANY PHILADELPHIA . LONDON . TORONTO

Modern Physics and Quantum Mechanics sBN 0-7216-1220-2

This book grew out of my experience in teaching a one-year course in

Sometimes the over-dramatization of the concepts of modern physics

Appendix A. Calculation of Jeans' Number b5

THE DEVETOPMENT OF WAVE ,\,IECHANICS llr

Wffiffi ffiffiffiffiffiffiffi WWffiffiffiffiW

The special theory of relativity is much more than a conceptual revolution

I. CLASSICAL PRINCIPLE OF RELATIVITY:

provided that t : t' -- 0 when the origins coincide. Differentiating with

where 4, : du*ldt, and so forth. Then,

2. ELECTROMAGNETIC THEORY AND THE

Long after the Galilean principle of relativity was well established,

Another serious difficulty arose from electromagnetic theory, in connection

Figure l-2 The angular position of a star is

6 A. H. Joy and R. F. Sanford, Astrophls. J. 64, 250 (1926).

3. THE MICHELSON-MORLEY EXPERIMENT

As mentioned, in the previous section, if the ether is assumed to be at rest

If the rower's rp.id is c, his path is actually of length

Kennedy and E. M. Thorndike, Plryts. 400 (1932)'

Show that the Lorentz-FitzGerald contraction can account

4. THE POSTULATES OF SPECIAL RELATIVffY

5. THE LORENTZ TRANSFORMATION

time and space coordinates, the simplest form to try is,

where A -* I and B -- 0 as a --+ 0.

If the new transformation satisfies Equation 1.5, it will be consistent with

PROBTE/U l-3 :,1.;jii.ri;!liil,iffi;tir:'i;.i::,;.'i,1l,:,'i iii;ji.iiljirli#

By means of Equations 1.3, L.4, and 1.5, determine the

Y'6(*,), z, t) ::r# 6@,), z, t),

is covariant under a Lorentz transformation.

Figure 1.5 Length measurements per-

x', - x'r: y(xz - rr) - FyrU, tr).

Ls:yL or L:41.-PzLo. (1 .i)

xL: y@l + fcti)

which reduces immediately to Equation 1.7.

Letting t; - ti: z ?nd t, tt : rot

Figure 1.7 Comparison of time intervals measured in two frames.

(!cr)z : /2+ (*r")'

, - 0'r: (1 - p')" : (l - P')Tro: lI - P'"0,

A meter stick lies at rest in the xy-plane at an angle of 45"

transformations as given by the Galilean principle of relativity; these are the

In the rest frame of an electron, another electron approaches

A cosmic ray mu meson moves vertically through the atmos-

A stream of electrons has velocity components V* :0.9a,

electrons with respect to an observer approaching the

Consider three inertial frames such that S and ,S' have

(u) Derive the result expressed by Equation I .7 by allowing

A radioactive atom is moving with respect to the laboratory

An observer in S places a laser so that the beam makes an

The velocities defined by Equations 1.9 were obtained by differentiating

where u,2 : uI + ui + zl is the square of the local velocity.

velocity a relative to an observer. After squaring Equation 1.12, it can be

Differentiating with respect to both m and u, we obtain

c2 dm - t12 dm * mu du. (1.13)

Q : r'* - T: (; ,), - (; *o)', (1 .38)

E:yE :28* :BMocz.