Quantum Theory

A short course on
Introductory Quantum Theory

-4
-2
0
2
4
x
0
0.5
1
1.5
2
t
0
0.25
0.5
0.75
1
-4
-2
0
2
4
x
Reinhard Hentschke
version of February 9, 2009
Contents
1 Early developments in quantum mechanics 1
1.1 Quantization conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The black body problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Building a general formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 The hydrogen atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Angular momentum algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2 Formal quantum mechanics 31
2.1 Operator averages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.2 The uncertainty principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3 From wave mechanics to matrix mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.4 The time evolution operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.5 The density operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.6 Path integration and the density operator . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3 Scattering theory 47
3.1 The Lippmann-Schwinger equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.2 The 1st Born approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3 Partial wave expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.4 Some remarks on formal scattering theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4 Solution methods in quantum mechanics 53
4.1 Time-independent perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Variation method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3 The quasi classical approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.4 Time-dependent perturbation theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.5 Interaction of charges with electromagnetic elds . . . . . . . . . . . . . . . . . . . . . . . 64
5 Many particle systems 67
5.1 Two kinds of statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2 Constructing N-particle wave functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.3 Electron spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.4 Simplied theory for helium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.5 Simplied theory for the hydrogen molecule . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.6 Quantum mechanics in chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
iii
6 The quantized radiation eld and second quantization 91
6.1 Quantum theory of radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.2 Second quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.3 Interacting quantum elds: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
6.4 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
A Constants and units 97
B Mathematical appendix 99
B.1 Conversion to generalized coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
B.2 Spherical harmonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
B.3 -function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
B.4 Fourier series and Fourier transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
C The Casimir eect 103
D Problems 105
Foreword
I recall a physics student expressing his opinion
that real physics begins with quantum mechan-
ics. This opinion, to which I cannot agree, proba-
bly is due to the strangeness in quantum theory,
which is summarized in an often cited quote by the
late Richard Feynman
1
expressing his believe that
nobody really understands quantum mechanics.
These lecture notes cover introductory quantum
theory to an extend that can be presented in a
one semester course. The subject is approached by
looking rst at some of the pressing questions at
the end of the 19th century, when classical physics,
in the eyes of many, had come close to explaining
all known physical phenomena. We will focus on
a special question (e.g. the black body problem),
then introduce an idea or concept to answer this
question in simple terms (e.g. energy quantization),
relate the quantum theoretical answer to classical
theory or experiment, and nally progress deeper
into the mathematical formalism if it provides a
general basis for answering the next question. In
this spirit we develop quantum theory by adding in
a step by step procedure postulates and abstract
concepts, testing the theory as we go along, i.e. we
will accept abstract and maybe sometimes counter-
intuitive concepts as long as they lead to veriable
predictions.
Chapter one is devoted to the early developments
in quantum theory
2
; chapter two provides a for-
1
Feynman, Richard Phillips, amerikanischer Physiker,
*New York 11.5.1918, Los Angeles 15.2.1988; verfasste
grundlegende Arbeiten zur Theorie der Suprauiditat und
des ussigen HeliumsII sowie zur Theorie des Betazerfalls.
F ur Beitrage zur Quantenelektrodynamik erhielt er 1965 mit
J.Schwinger und S.Tomonaga den Nobelpreis f ur Physik.
Seine Lehrb ucher begr undeten eine neue Physikdidaktik.
Literatur: Mehra, J.: The beat of a dierent drum. The
life and science of R.Feynman. Neuausgabe. Oxford 1994.
c _2000 Bibliographisches Institut & F. A. Brockhaus AG
2
The student should be aware that these lecture notes,
for the most part, cover only a few years in the development
of quantum theory, i.e. early is appropriate in a limited
sense only. In 1900 Planck presented his treatment of black
body radiation; Einsteins interpretation of the photoelec-
tric eect appeared in 1905/06; Bohrs seminal work On
the constitution of atoms and molecules was published in
1913. But after 1925/26, when a formal framework (matrix
mechanics/wave mechanics) had been suggested to do cal-
culations, the development became extremely rapid! Maybe
not surprising it took only about ve years and a number of
textbooks on quantum mechanics had appeared (e.g, Weyl,
1930 [1]; Dirac, 1930 [2]; von Neumann, 1932 [3]). These
mal basis for quantum mechanics treating bound
state problems; chapter three deals with scatter-
ing; chapter four introduces perturbation theory to-
gether with selected numerical techniques; in chap-
ter ve we turn to many particle systems and ex-
ploit the fact that in nature there are two types
of particles-fermions and bosons; the nal chap-
ter, chapter six, presents a short glimpse at the
quantized radiation eld and second quantization.
Chapters ve and six mark the departure of physics
from chemistry as far as university education is con-
cerned. Atomic or molecular systems with more
than two electrons are usually reserved for students
of quantum chemistry. Solids as well as advanced
quantum theory, which leads to the physics of el-
ementary particles, appear to be the domain of
physics students. Here, in a limited sense of course,
I attempt to satisfy both tastes. Because of this the
lecture does not follow one particular text. How-
ever, in preparing these notes I have been guided
by the references [5, 6, 7, 8]. Additional textbooks
on quantum mechanics containing most of the ma-
terial discussed here include [9, 10, 11, 12, 13].
The initial version of these notes was prepared
for a course on introductory quantum mechanics
during the fall/winter term 2002/2003. Subse-
quently these notes were used and extended during
the summers of 2004, 2006, 2007, and again during
the fall/winter term 2008/2009. I owe thanks to
my secretary, Susanne Christ, for typing these
notes, and to Hendrik Kabrede, Jorg Bartke,
Henning Horstermann, Christa Guse, and Ran Jia
for taking over the responsibility for the recitation
sections as well as for making valuable comments
on the material presented here. Thanks also to
Binh Trieu (May 15, 2003) for pointing out a
number of mistakes in an earlier version.
Wuppertal, December 2008
Reinhard Hentschke
Bergische Universitat
Fachbereich Mathematik und Naturwissenschaften
lecture notes, as almost all texts aimed at beginners, do
(partially!) cover this exciting period only. A good histor-
ical source on important steps in physics in the 19th and
early 20th century leading up to the developments discussed
here is A. Sommerfelds Atombau und Spektrallinien [4],
which rst appeared in 1919!
Gauss-Str. 20
42097 Wuppertal
e-mail: hentschk@uni-wuppertal.de
http://constanze.materials.uni-wuppertal.de
Chapter 1
Early developments in quantum
mechanics
1.1 Quantization conditions
The year 1900 marks the beginning of what we call
Quantum Theory. Until then various physicists
of reputation had failed to derive a valid formula
describing the frequency dependence of radiation
intensity emitted from a black body. A black body
can be a metal cavity with walls kept at a xed
temperature T. A small hole drilled into the walls
of this oven will allow radiation to escape - the
black body radiation. This may sound academic.
But what makes this problem especially interest-
ing is, that all prior calculations were based on
apparently well founded theories like Maxwells
1
equations and classical statistical mechanics. These
pillars of physics, however, as we shall see below,
produced nonsense. In order to obtain a correct
formula describing black body radiation the physi-
cist Max Planck
2
had to introduce a revolutionary
1
Maxwell, James Clerk, britischer Physiker, *Edin-
burgh 13.6.1831, Cambridge 5.11.1879; erneuerte die von
T.Young aufgestellte Dreifarbentheorie des Sehens und
forderte besonders die kinetische Gastheorie (maxwellsche
Geschwindigkeitsverteilung), schloss aus mathematischen
Uberlegungen, dass der Saturnring aus festen Teilchen beste-

hen m usse, erweiterte die Vektor- und Tensoranalysis, zeigte
die elektromagnetische Natur der Lichtwellen. Er en-
twickelte die Theorie des elektromagnetischen Feldes (Elek-
trizitat) durch Mathematisierung des von M. Faraday in
die Physik eingef uhrten Feldbegris (maxwellsche Theorie).
2
Planck, Max, Physiker, *Kiel 23.4.1858, Gottingen
4.10.1947; Professor in Kiel und Berlin, 1930-37 und
1945/46 Pr asident der Kaiser-Wilhelm-Gesellschaft zur
Forderung der Wissenschaften (Max-Planck-Gesellschaft
zur F orderung der Wissenschaften e.V.), einer der be-
deutendsten Physiker des 19./20.Jahrhunderts; Hauptar-
beitsgebiete: Strahlungstheorie und Thermodynamik. Er
concept. The energy of the radiation eld inside
the black body is not continuous. It is the sum of
discrete amounts of energy of size h or h . Here
is the radiation frequency ( = 2), and h is the
now famous Plancks constant (dt.: Plancksches
Wirkungsquantum)
h = 6.6256 0.0005 10
34
Js .
Discrete entities were not new in science at that
time. Chemists had long accepted the notion that
matter consists of atoms. This idea was formulated
by Dalton
3
in his book A new system of chemical
leitete 1900 das Strahlungsgesetz f ur die W armestrahlung
schwarzer K orper ab (plancksches Strahlungsgesetz), pos-
tulierte die Quantelung der Energie und entdeckte dabei
die Existenz einer neuen Naturkonstante (plancksches
Wirkungsquantum). Damit begr undete Planck die Quan-
tentheorie und revolutionierte die klassische Physik. 1918 er-
hielt er den Nobelpreis f ur Physik. Die Deutsche Physikalis-
che Gesellschaft stiftete 1928 die Max-Planck-Medaille,
deren erster Preistrager er war.Literatur: J.L. Heilbron:
M.Planck. Ein Leben f ur die Wissenschaft, herausgegeben
von J.L. Heilbron. Aus dem Amerikanischen. Stuttgart
1988. c _2000 Bibliographisches Institut & F. A. Brockhaus
AG
3
Dalton, John, britischer Chemiker und Physiker, *Ea-
gleseld (County Cumbria) 5. oder 6.9.1766, Manchester
27.7.1844; untersuchte die physikalischen Eigenschaften von
Fl ussigkeiten und Gasen und entdeckte 1801 das spater
nach ihm benannte Gesetz uber den Partialdruck. Dal-
tons chemische Atomtheorie basierte auf der neuen Erkennt-
nis, dass sich Atome verschiedener chemischer Elemente in
ihrem Atomgewicht unterscheiden; f uhrte stochiometrische
Grundgesetze ein (Gesetz der multiplen Proportionen, 1808:
Die Gewichtsverh altnisse zweier Elemente, die sich zu ver-
schiedenen Verbindungen vereinigen, stehen im Verhaltnis
einfacher ganzer Zahlen). c _2000 Bibliographisches Institut
& F. A. Brockhaus AG
1
2 CHAPTER 1. EARLY DEVELOPMENTS IN QUANTUM MECHANICS
philosophy, which appeared in two volumes in 1808
and 1810. Also, Thomson
4
in 1897 had discovered
the electron. However, energy was rmly believed
to be continuous. E.g., we all are used to chang-
ing the speed of our car and thus its kinetic energy
continuously. However, like matter, which macro-
scopically appears continuous but microscopically
must be chemically discrete in order to explain
chemical reactions and their mass ratios, energy
also reveals its discrete nature on the microscopic
scale. That classical physics had worked to every-
bodys satisfaction had to do with the experimental
inability of that time to probe the microscopic (or
atomic) scale. However, in addition to the black
body problem, where a microscopic phenomenon
could be measured macroscopically, scientists al-
ready knew of other troubling facts. Chemists had
come up with the periodic table. But how chemi-
cal reactions really work or the nature of chemical
bounds were beyond comprehension. Spectral lines
of simple atoms, e.g., hydrogen (Balmer series
5
),
had been measured but could not be explained. Be-
sides Planck other scientist were productive too. J.
J. Thomson, who had discovered the electron, sug-
gested a new model for the atom in 1904. Accord-
ing to him atoms are small spheres of certain mass
containing a uniformly distributed positive charge
as well as a corresponding number of electrons em-
bedded into this positive background. An improved
model was suggested by E. Rutherford
6
in 1911
4
Thomson, Sir (seit 1908) Joseph John, britischer
Physiker, *Cheetham Hill (heute zu Manchester) 18.12.1856,
Cambridge 30.8.1940, wies nach, dass die Kathodenstrahlen
Strome freier Elektronen sind, entwickelte eine atomistische
Struktur der Elektrizitat, bestimmte die Elementarladung
und das Verhaltnis von Ladung und Masse f ur Elektro-
nen; erhielt 1906 den Nobelpreis f ur Physik. c _2000 Bib-
liographisches Institut & F. A. Brockhaus AG
5
Balmer, Johann Jakob, schweizerischer Mathematiker,
*Lausen (Kanton Basel-Landschaft) 1.5.1825, Basel
12.3.1898; stellte 1884/85 erstmals eine Formel f ur die Spek-
trallinien des Wasserstos (Balmerserie) auf. c _2000 Bibli-
ographisches Institut & F. A. Brockhaus AG
6
Rutherford, Ernest, Lord Rutherford of Nelson (seit
1931), britischer Physiker, *Brightwater (bei Nelson,
Neuseeland) 30.8.1871, Cambridge 19.10.1937; Professor in
Montreal, Manchester, ab 1919 in Cambridge und Direktor
des dortigen Cavendish Laboratory. Rutherford war einer
der bedeutendsten Experimentalphysiker des 20.Jahrhun-
derts, besonders auf dem Gebiet der Radioaktivitat und
der Kernphysik. Er unterschied die beim radioaktiven Zer-
fall entstehende -, - und -Strahlung, stellte mit F.Soddy
1903 eine Theorie des radioaktiven Zerfalls auf und schuf
durch das nach ihm benannte Atommodell die Grundlage der
based on his experiments in which he bombarded
thin metal foils with -particles. His model was es-
sentially a planetary system with a tiny positive nu-
cleus at the center circled by electrons. This imme-
diately caused a problem. According to Maxwells
theory accelerated charges emit radiation. There-
fore Rutherfords atoms should loose energy and
collapse. Similar to Planck in the case of the black
body Niels Bohr
7
postulated a solution to this
problem. In his model of the atom the electrons are
conned to discrete orbits E
1
, E
2
, . . .. They may
however absorb one of Plancks radiation packages,
if it supplies the proper amount of energy separat-
ing two orbits, e.g. h = E
n+1
E
n
. This package
also may be emitted, and the electron returns to its
previous orbit. However, there always exists an or-
bit whose energy is lowest, and therefore the atom
cannot collapse. These postulates were in accord
with Einsteins
8
earlier explanation of the photo-
heutigen Atomphysik. 1919 gelang ihm die erste k unstliche
Kernreaktion durch -Teilchen-Beschuss von Sticksto, der
dabei in Sauersto umgewandelt wurde. 1908 erhielt er den
Nobelpreis f ur Chemie. c _2000 Bibliographisches Institut &
F. A. Brockhaus AG
7
Bohr, Niels Henrik David, danischer Physiker, *Kopen-
hagen 7.10.1885, ebenda 18.11.1962; Professor in Kopen-
hagen; wandte 1913 die Quantenhypothese (M. Planck 1900,
A. Einstein ab 1905) auf das Atommodell E. Rutherfords
an und schuf das bohrsche Atommodell (Atom), das erst-
mals so genannte Quantenbedingungen enthielt. Bohr ent-
deckte das Korrespondenzprinzip. Auf der Basis seines von
A.Sommerfeld erweiterten Atommodells konnte er das Peri-
odensystem der chemischen Elemente theoretisch erkl aren.
1922 erhielt er den Nobelpreis f ur Physik. Nach der
Aufstellung der Quantenmechanik gelang es Bohr 1926/27
in Zusammenarbeit mit W.Heisenberg, die Entwicklung
der Quantentheorie vorlaug abzuschlieen, wobei er zu
der

Uberzeugung kam, dass zur vollst andigen Beschrei-
bung der atomaren Erscheinungen zwei verschiedene Bilder
(Teilchen- und Wellenbild) notwendig seien. Bohr f uhrte
den Begri des Compoundkerns ein und entwickelte das
Tropfchenmodell des Atomkerns. Seine Theorie der 1938
entdeckten Kernspaltung des Urans war wichtig f ur deren
spatere technische Nutzung. 1943-45 arbeitete Bohr in Los
Alamos (USA) an der Entwicklung der Atombombe. Werke:
Abhandlungen uber Atombau (1913); Drei Aufsatze uber
Spektren und Atombau (1922); Atomphysik und menschliche
Erkenntnis (1957); Essays 1958-1962 (1964). Literatur:
Roseberg,U.: N.Bohr. Leben und Werk eines Atomphysik-
ers. Berlin u.a. 31992. c _2000 Bibliographisches Institut &
F. A. Brockhaus AG
8
Einstein, Albert, Physiker, *Ulm 14.3.1879, Princeton
(New Jersey) 18.4.1955, Professor in Z urich und Prag,
191433 Leiter des Kaiser-Wilhelm-Instituts f ur Physik in
Berlin, seitdem Professor in Princeton. Einstein entwick-
elte um 1905 die spezielle, 1915 die allgemeine Rela-
tivitatstheorie, die die moderne Physik auf neue Grundla-
1.1. QUANTIZATION CONDITIONS 3
electric eect. A metal surface exposed to electro-
magnetic radiation (e.g., visible light) emits elec-
trons if the radiation exceeds a threshold frequency
0
. Below
0
no electrons are emitted, independent
of the radiations intensity! The kinetic energy of
the emitted electrons,
1
2
m
e
v
2
, is described by
1
2
m
e
v
2
+
0
= h (1.1)
with
0
= h
0
(= h
0
). The constant
0
is called
the work function
9
.
Nevertheless, Bohrs postulates were merely pos-
tulates. To justify postulates, which were in dis-
agreement with Maxwells beautiful theory of elec-
tromagnetism, Bohr needed to make veriable pre-
dictions based on his model. He needed a quanti-
zation condition. In 1924 Louis deBroglie
10
sug-
gested the relation
gen stellten; 1905 formulierte er eine Theorie der brown-
schen Bewegung, 1907 eine Theorie der spezischen Warme
fester Korper. Seine Erklarung des aueren Photoeekts
(1905, 1921 Nobelpreis f ur Physik) mithilfe der Lichtquan-
tenhypothese trug zur Anerkennung der Quantentheorie bei,
obwohl Einstein die statistische Interpretation der Quanten-
mechanik nie akzeptierte. Die Arbeiten Einsteins nach dem
Ersten Weltkrieg galten der allgemeinen Relativitatstheorie,
insbesondere einer Theorie der Gravitation und einer ein-
heitlichen Feldtheorie. Mit einem Brief an Prasident Roo-
sevelt (1939) gab Einstein, ein uberzeugter Pazist, aus
Furcht vor einer deutschen Aggression zusammen mit an-
deren den Ansto zum Bau der Atombombe; nach 1945
setzte er sich nachhaltig f ur den Abbau von Kernwaf-
fen ein. Werke: Die Grundlage der allgemeinen Rela-
tivit atstheorie (1916);

Uber die spezielle und allgemeine Rel-
ativit atstheorie (1917); Mein Weltbild (1934); Die Evolu-
tion in der Physik (1950, mit L. Infeld); Lebenserinnerun-
gen (1952). Literatur: Pais,A.: Raniert ist der Herrgott
. . . . A. Einstein. Aus dem Englischen. Braunschweig
u.a. 1986. Kanitscheider,B.: Das Weltbild A. Einsteins.
M unchen 1988. Higheld,R. und Carter, P.: Die geheimen
Leben des A.Einstein. Eine Biographie. Aus dem En-
glischen. Berlin 1994. F olsing,A.: A.Einstein. Eine Bi-
ographie. Taschenbuchausgabe. Frankfurt am Main 1995.
Hermann,A.: Einstein. Der Weltweise und sein Jahrhun-
dert. M unchen 21995. Wickert,J.: A.Einstein. Reinbek
106.108.Tausend 1995. http://www.westegg.com/einstein/
9
The photoelectric eect can be used to measure h.
10
Broglie, Louis-Victor, 7. Herzog von Broglie (seit
1960), genannt L. deBroglie, franzosischer Physiker, *Dieppe
15.8.1892, Louveciennes (Departement Yvelines) 19.3.1987;
begr undete die Theorie der Materiewellen (deBroglie-
Wellen) und erhielt daf ur 1929 den Nobelpreis f ur Physik.
p =
h
(1.2)
connecting the momentum of a particle (e.g., an
electron) to a wavelength via Plancks constant.
DeBroglies idea that an electron may behave as
a wave was proven experimentally by C. Davis-
son and L. Germer in 1927, who observed electron
diraction from a nickel crystal. This introduced
the concept of particle-wave-dualism. DeBroglies
relation immediately allows to write down an ex-
pression for the orbital energy of the electron in
the hydrogen atom, which for instance explains the
Balmer series
11
.
From the Kepler problem in classical mechanics
we borrow the relation
1
2
| =

/, where

| and

/
are the average potential and kinetic energy respec-
tively
12
. For a simple circular orbit of radius a we
therefore have
p
2
2
=
1
2
e
2
a
, (1.3)
where = m
e
m
n
/ (m
e
+m
n
) is the reduced mass,
and e is the electron charge
13
. Requiring that the
circumference of the orbit corresponds to an inte-
ger number of deBroglie wavelengths, a standing
electron wave, we nd the quantization condition
2a = n (n = 1, 2, . . .) . (1.4)
Putting (1.2), (1.3) and (1.4) together we obtain
11
This sounds as if Bohr had to wait until deBroglie had
put forward his idea. But this is not true. We merely choose
this path because relation (1.2) allows to interpret Bohrs
quantization conditions via the wave-picture of particles.
12
This is easy to show in the special case of a circular
orbit. The constant kinetic energy is | = (/2)(r )
2
, where
is the (reduced) mass, r is the orbits radius, and is
the constant angular velocity. The latter may be related
to the constant potential energy, | = /r (with > 0),
via Newtons second law,
r =
| = e
r
|/r =
e
r
/r
2
. Using polar coordinates, i.e. r = r(cos , sin ) =
re
r
we obtain

r = r (sin , cos ) and

r = r
2
e
r
. We
nd the relation (r )
2
= /(r) and therefore | = |/2.
This relation, as we have mentioned, holds for general Kepler
orbits if | and | are replaced by their averages over one
complete orbit (cf. [14] exercise 24).
13
Here we have omitted the factor (4
o
)
1
multiplying
the potential energy, i.e. we occasionally toggle between the
MKSA and the Gau unit systems.
1
a
=
e
2
h
2
1
n
2
. (1.5)
With E =

/ +

| =
1
2
| follows
E
n
=
e
4
2h
2
1
n
2
= 13.53eV
1
n
2
. (1.6)
The quantity n is called quantum number. Thus,
an electromagnetic energy packet or photon emit-
ted during the transition from an excited state
(n = m) to the ground state (n = 1) should be
given by
h
m1
= 13.53eV
_
1
m
2
1
_
. (1.7)
This is an agreement with the experiment!
14
But Bohrs theory must pass another test. When
we increase the orbit of the electron to macro-
scopic dimensions in what we call a Gedankenex-
periment it should obey Maxwells equations and
thus emit radiation according to this classical the-
ory. According to classical mechanics we have
[

L [ L = a
2

= a
2
cl
. Here

L is the an-
gular momentum vector, and is the angle of the
position vector r ([ r [ a) with a xed axis in the
two-body problem. Condition (1.4) immediately
implies
L = nh . (1.8)
Combining this with the above equation as well as
(1.5) we obtain for the classical orbital frequency
cl
=
e
4
h
3
n
3
. (1.9)
For large n this should be equal to
qt
obtained via
a transition from orbit n + 1 to orbit n:
qt
=
e
4
2h
3
_
1
(n + 1)
2

1
n
2
_
14
This particular series of frequencies is called Lyman-
series (1906). Balmers series corresponds to h
m2
=
13.53eV
_
1
m
2

1
4
_
.
=
e
4
2h
3
1
n
2
_
1
_
1 +
1
n
_
2
1
_

e
4
2h
3
1
n
2
_
1 2
1
n
1
_
=
e
4
h
3
n
3
.
This is an example of the important correspon-
dence principle, which describes the classical limit
of quantum theory.
Here we may look at the classical limit from an-
other angle. For large n we always require
E
n
hn

cl
. (1.10)
In the case of Eq. (1.6) for example we have
E
n
n

dE
n
dn
=
e
4
h
2
1
n
3
(1.9)
= hw
cl
.
Integrating Eq. (1.10) we obtain
1
h
_
E
n
0
dE
(E)
=
_
n
0
dn
t
= n . (1.11)
This equation may be cast in another useful form,
i.e.
1
h
_
E
0
dE
t
(E
t
)
=
1
h
_
pdq = n , (1.12)
where the loop integral covers the area in classical
phase space bounded by a trajectory corresponding
to the energy E
15
.
We show Eq. (1.12) utilizing the classical action
(dt.: Wirkung), S = S (q, t), as a function of the
(generalized) coordinate q and time t
16
. It is useful
to introduce a Legendre transformation to a new
function,

S =

S (q, H), where H is the Hamilton
function
17
, via
15
Notice that by using p
=[

L [ we obtain
_
p
d =
_
Ld = nh. This is the generalized form of Eq. (1.8).
16
cf. the section on Hamilton-Jacobi theory in Ref. [14].
17
Hamilton, Sir (seit 1835) William Rowan, irischer Math-
ematiker und Physiker, *Dublin 4.8.1805, Dunsink (bei
Dublin) 2.9.1865; entwickelte die geometrische Optik aus Ex-
tremalprinzipien, ubertrug dieses Konzept auf die klassische
Mechanik (Hamiltonprinzip) und f uhrte die Kraftefunktion
in die Dynamik ein; f uhrte 1843 die Quaternionen als Ve-
rallgemeinerung der komplexen Zahlen ein. c _2000 Bibli-
1.2. THE BLACK BODY PROBLEM 5
d
S = d (Ht) +dS
= tdH+Hdt +
S
q
..
=p
dq +
S
t
..
=1
dt
= pdq +tdH
Setting H = E = const we integrate the last equa-
tion between q and q
o
to obtain
S(q, E)

S(q
o
, E) =
_
q
q
o
pdq
t
.
Now we take the derivative with respect to E:
S(q, E)
E
. .
=t
S(q
o
, E)
E
. .
=t
o
=

E
_
q
q
o
pdq
t
.
Here t and t
o
correspond to q = q(t) and q
o
= q(t
o
).
Extending the integration over a full period of the
(periodic) motion yields
T(E) =

E
_
pdq .
Note that T(E) = 1/(E) is the period of the orbit
corresponding to the energy E. Finally we inte-
grate this equation from 0 to E, which yields the
desired equality
_
E
0
T(E
t
)dE
t
=
_
pdq . (1.13)
This is a result of classical mechanics. We test
Eq. (1.13) using the classical harmonic oscillator,
E = p
2
/(2m) + (k/2)x
2
, as an example. Because
the oscillators period is T = 2
_
m/k we have
_
E
0
T(E
t
)dE
t
= T
_
E
0
dE
t
= 2
_
m
k
E . (1.14)
Now we evaluate the right side of Eq. (1.13), i.e.
_
pdq = 2
_

2E/k
2E/k
dx
2m
_
E
k
2
x
2
_
= 2
2mE
_
2E
k
_
1
1
dz
_
1 z
2
. .
=/2
= 2
_
m
k
E , (1.15)
where
_
2E/k is the amplitude of the oscillator.
Obviously the two results, (1.14) and (1.15), do
agree.
1.2 The black body problem
At this point we want to consider the black body
problem explicitly to gure out what goes wrong as
well as how to correct classical theory
18
.
A black body cavity contains electromagnetic ra-
diation. Its classical energy density is given by
W =

E
2
+

H
2
8
(1.16)
[15] (Eq. (2.49)), where

E and

H are the electric
and the magnetic eld strengths, respectively. The
total eld energy E in the cavity is given by
E =
_
V
dV W , (1.17)
where V is the volume of the cavity. We may use
a gauge, where

E =
1
c

A
t
and thus Eq. (1.17)
becomes
E =
1
8
_
_
_
_
1
c

A
t
_
2
+
_

A
_
2
_
_
dV (1.18)
(cf. [15] (2.18) and (2.19) as well as (2.15)). The
vector potential in turn is expressed as a Fourier-
series
18
Historically, this system has been looked upon from
two, practically identical but conceptually dierent, points
of view: (i) The earlier one, the one we take here, views the
radiation eld as a collection of independent oscillators. (ii)
Later, after the photon had fully established itself as an ele-
mentary particle, the radiation eld was treated as a gas of
Bosons. But this is a matter of statistical mechanics, which
we mention only in order to prevent the reader from getting
confused.
A(r, t) =
1
k,
_
c
k,
u
k,
(r)e
it
+c
k,
u
k,
(r)e
it
_
, (1.19)
where k
x
=
2
L
n
x
, k
y
=
2
L
n
y
, k
z
=
2
L
n
z
(n
x
, n
y
,
n
z
= 1, 2, . . .), V = L
3
, and
u
k,
(r) =
()
e
i
kr
. (1.20)
Here and in the following the asterisk indicates
the complex conjugate.
()
is real unit vector in
the -direction of the plane perpendicular to the
momentum vector

k. Inserting Eq. (1.19) into Eq.
(1.18) we obtain
19
E =
k,
w
2
2c
2
c
k,
c
k,
(1.21)
and via
q
k,
=
1
4c
_
c
k,
+c
k,
_
(1.22)
p
k,
=
i
4c
_
c
k,
c
k,
_
(1.23)
the nal form
E =
k,
1
2
_
p
2
k,
+
2
q
2
k,
_
. (1.24)
This means that the electromagnetic eld energy
inside the cavity corresponds to

k,
uncoupled
one-dimensional (1D) harmonic oscillators.
At this point electrodynamics has done its job,
and we now turn to statistical mechanics. In sta-
tistical mechanics you will learn that the average
energy contained in a system consisting of a xed
number of particles (e.g., oscillators) inside a xed
volume, which is in thermal equilibrium with its
surroundings (here the cavity walls), is given by
E) =

ln
m
e
E
m
, (1.25)
19
this is a homework problem
where
1
= k
B
T. T is the temperature, and k
B
is
Boltzmanns constant. The sum

m
is over all en-
ergy values accessible to the system (cf. [16] section
2.1).
We interrupt briey to explain the dierence be-
tween E) as compared to E used above. Statis-
tical mechanics considers the electromagnetic eld
inside the cavity as in thermal equilibrium with the
cavity walls kept at xed temperature T. In this
sense energy may be exchanged between the eld
and the walls, i.e. the elds energy uctuates.
In classical physics energy is continuous, and
the summation in Eq. (1.25) becomes an inte-
gration. From classical mechanics we know that a
harmonic oscillator at constant energy is described
by a trajectory in phase space. Changing energy
means moving to another trajectory. Summing
over all possible energy states (microstates) of the
system, with their proper statistical weight, there-
fore means integrating over the entire phase space.
Thus, for a single oscillator we have
dpdq .
We neglect a constant factor (the phase space den-
sity), because it drops out of the calculation of E)
which now, for a single oscillator, becomes
ln
_

dpdqe
(p
2
+
2
q
2
)/2
= k
B
T .
Because the oscillators are independent we obtain
for the entire cavity
E) = k
B
T
k,
1 . (1.26)
The problem is now reduced to counting the num-
ber of modes

k,
!
It is useful to transform the summation into an
integration. The general prescription for this is
k
_
d
3
k = 4
k
_

0
dkk
2
. (1.27)
k
is the density in

k-space given by
1.3. BUILDING A GENERAL FORMALISM 7
k
=
V
(2)
3
. (1.28)
This follows because the

k-vectors lie on a cubic
grid, whose gridpoints are given by (k
x
, k
y
, k
z
).
The lattice constant of this grid is 2/L. The com-
bination of Eqs. (1.26) and (1.28) nally gives
E)
V
= k
B
T
_

0
d

2
2
c
3
= . (1.29)
This means that the energy density of the black
body radiation eld is innity. Clearly, this result
cannot be correct. And it is very disturbing that
we have obtained it via correct application of elec-
trodynamics and classical statistical mechanics!
We now make a second attempt at solving this
problem - using the quantization condition (1.12)!
We conclude that the energy of the radiation eld
written in the form of Eq. (1.24) should be quan-
tized as follows: For a single 1D harmonic oscillator
we have
n =
1
h
_
pdx
(1.15)
=
E
n
h
(1.30)
( =
_
k/m). This is exactly what Planck had
assumed in 1900.
Again we consider a single 1D oscillator. This
time energy is discrete, and the summation remains
a summation. The right side of Eq. (1.25) now
becomes
ln
n=0
e
hn
=
h
e
h
1
20
. For the entire cavity we obtain
E) =
k,
h
e
h
1
(1.31)
and therefore
20
Using
n=0
z
n
= (1 z)
1
for z < 1.
E)
V
(1.27)
=
h
2
c
3
_

0
d

3
e
h
1
(cf. [16]; Eq. (3.13))
=

2
15
(k
B
T)
4
(hc)
3
.
This in fact is the correct result in agreement with
the experiment
21
! Notice that in the (classical)
limit h 0 the integrand becomes
3
(1 +
h 1)
1
and we recover the classical result of
Eq. (1.29).
In the next section we learn that the above equa-
tion for the energy levels of a harmonic oscillator
is not quite correct. It neglects the so called zero-
point energy. However, here the zero-point energy
is a constant which does not contribute to the ra-
diation detected outside the black body cavity, be-
cause photons correspond to transitions from and
to excited oscillator states only
22
.
1.3 Building a general formal-
ism
Using energy quantization we are able to explain
the crude structure of the spectral lines of the hy-
drogen atom, and we have solved the black body
problem. But there remains much more to ex-
plain: What happens if a second electron is added,
how can we calculate intensities of spectral lines
etc. Clearly, we need a general theory for electrons,
atoms, molecules, radiation and their interaction.
A rst step may be the following thought: Look-
ing into our mathematical tool chest, we may ask
whether quantization can be formulated in terms of
eigenvalue problems either through the use of ma-
trices or dierential equations. The latter are es-
pecially appealing if we think of particles as waves
- as standing waves possessing a discrete spectrum
21
It is possible to drill a hole into one of the container walls
and observe the photon current density I(T) = cE)/V .
This is Stefans law (I(T) T
4
). Notice that the T
4
-
dependence may be argued on classical grounds - not, how-
ever, the coecient [17] (section 12.1).
In 1989 the cosmic background explorer or COBE-satellite
measured the intensity spectrum of the cosmic background
radiation over a wide range of frequencies nding perfect
agreement with Plancks distribution (cf. Ref. [16]; section
3.1).
22
This does not mean that zero-point energy is not inter-
esting as we demonstrate in appendix C.
of frequencies. What we may aim for at this point
is an eigenvalue dierential equation for a parti-
cle (e.g., an electron) in an external potential, |,
yielding discrete energy levels like
E
n
=
e
4
2 h
2
1
n
2
for | =
e
2
r
in the case of the Coulomb potential or
E
n
= hn for | =
1
2
kx
2
in the case of the harmonic oscillator. Consequently
we seek an equation of the form
H [ n) = E
n
[ n) , (1.32)
where E
n
are energy eigenvalues, and H is an
operator corresponding to the classical Hamilton
function H. But what is [ n)? If we think in terms
of matrix theory, then [ n) is an eigenvector cor-
responding to the eigenvalue E
n
. In the following
we will talk about energy eigenstates, and [ n) will
be called a state vector (dt.: Zustandsvektor).
Matrix mechanics:
The prescription for a quantum theory based
on matrix algebra is as follows (The following
ideas are due to Heisenberg, Jordan, and Born
23 24 25
.): We seek a system of 2k matrices
q
1
, . . . , q
k
, . . . , p
1
, . . . p
k
, which correspond to the
generalized coordinates and momenta of the me-
chanical analog. The key requirement is that these
matrices must obey the commutator relations
_
q
m
, q
n
_
q
m
q
n
q
n
q
m
= 0 , (1.33)
_
p
m
, p
n
_
p
m
p
n
p
n
p
m
= 0 (1.34)
and
_
p
m
, q
n
_
p
m
q
n
q
n
p
m
= ih
mn
. (1.35)
Even though these relations seemingly appear out
of the blue, we remind the reader of the close
23
Heisenberg, Werner Karl, Physiker, *W urzburg
5.12.1901, M unchen 1.2.1976; war 1927-41 Professor
f ur theoretische Physik in Leipzig, 1941-45 Direktor des
Kaiser-Wilhelm-Instituts f ur Physik (Berlin); Mitarbeit
am Uran-Projekt (Bau eines Uran-Reaktors); 1946-70
Direktor des Max-Planck-Instituts f ur Physik und Astro-
physik (Gottingen, seit 1958 in M unchen) sowie Professor
in Gottingen und M unchen. Gemeinsam mit M.Born und
P.Jordan begr undete Heisenberg die Quantenmechanik in
der Matrizenform und stellte 1927 die f ur sie grundlegende
Unscharferelation (heisenbergsche Unscharferelation) auf.
Weitere Arbeiten Heisenbergs forderten die Quanten- und
Wellenmechanik, die Atom- und Kernphysik, die Physik
der kosmischen Strahlung und der Elementarteilchen,
die Theorie der Supraleitung und des Ferromagnetismus.
Seit etwa 1953 arbeitete Heisenberg an einer einheitlichen
Feldtheorie der Materie (heisenbergsche Weltformel).
1932 erhielt Heisenberg den Nobelpreis f ur Physik. Werke:
Die physikalischen Prinzipien der Quantentheorie (1930);
Das Naturbild der heutigen Physik (1955); Physik und
Philosophie (1959); Der Teil und das Ganze (1969); Schritte
uber Grenzen (1971). c _2000 Bibliographisches Institut &
F. A. Brockhaus AG
24
Jordan, Ernst Pascual, Physiker, *Hannover 18.10.1902,
Hamburg 31.7.1980; 1957-61 MdB; war magebend an der
Entwicklung der Quantenmechanik beteiligt und wandte sie
auch auf biophysikalische Fragen an, arbeitete uber Quan-
tenelektrodynamik, allgemeine Relativitatstheorie, Astro-
physik und Kosmologie. c _2000 Bibliographisches Institut
25
Born, Max, Physiker, *Breslau 11.12.1882, Gottingen
5.1.1970; Professor in Breslau, Berlin, Frankfurt am Main,
Gottingen und Edinburgh; grundlegende Arbeiten in der
Quantenmechanik, gab die heute allgemein anerkannte
statistische Deutung der Wellenmechanik, erhielt daf ur
1954 den Nobelpreis f ur Physik (zusammen mit W.Bothe).
correspondence between these so called commu-
tators and the Poisson brackets in Hamiltonian
mechanics (q
m
, q
n
= 0, p
m
, p
n
= 0 and
p
m
, q
n
=
mn
[14]). Secondly the matrix
H
_
q
1
, . . . , q
k
, . . . , p
1
, . . . , p
k
_
, which is the analog
of the Hamilton function H, becomes diagonal.
As an example we consider the 1D harmonic os-
cillator, i.e. H =
p
2
2m
+
1
2
m
2
x
2
. We make a bold
guess and write down the matrices
x =
_
h
2m
_
_
_
0 1 0 0
1 0

2 0
0

2 0

3 0
0 0

3 0
.
.
.
.
.
. 0
.
.
.
_
_
_ (1.36)
and
p = i
_
hm
2
_
_
_
0 1 0 0
1 0
2 0
0

2 0
3 0
0 0

3 0
.
.
.
.
.
. 0
.
.
.
_
_
_ , (1.37)
which indeed obey the relations (1.33) to (1.35).
Inserting (1.36) and (1.37) into
H =
1
2m
p
2
+
m
2
2
x
2
(1.38)
yields
H = h
_
_
_
_
1
2
0 0 0
0
3
2
0 0
0 0
5
2
0
0 0 0
7
2
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
. (1.39)
Thus, we obtain the desired eigenvalues E
n
for
[ n) =
_
_
_
_
_
_
.
.
.
0
1
0
.
.
.
_
_
_
_
_
_
, (1.40)
where 1 stands for the nth element.
There is an interesting dierence to our previous
result, Eq. (1.30), based on the original quantiza-
tion condition. The lowest attainable energy, the
ground state energy, now is
1
2
h
26
! However, the
26
This energy is also called zero-point energy.
increments are still equal to h. We recall that
the original quantization condition,
_
pdq = nh, re-
sulted from (1.10), which is based on large n. Our
new quantum mechanical theory however produces
the proper ground state energy as we shall see.
Notice also that there is a basic problem inher-
ent in the above prescription for constructing the
quantum analog of a mechanical system. The clas-
sical generalized coordinates and momenta may be
interchanged without altering the Hamilton func-
tion. The corresponding operators or matrices,
according to (1.35), may (in general) not be in-
terchanged without altering H. Fortunately we
do not run into this problem often, because for
H = /(p
1
, . . . , p
k
) +| (q
1
, . . . , q
k
) there is no such
problem.
Another point is that the quantities q
m
and p
m
may be complex (cf. Eq. (1.37)). Because we re-
quire real eigenvalues E
n
, the Hamilton operator H
must be hermitian (dt.: hermitesch oder selbstad-
jungiert), i.e.
H = H
+
. (1.41)
The symbol + means: Take the transpose of H
and also take the complex conjugate (i i) of
each element
27
. To prove that (1.41) yields real
eigenvalues we write
H [ n) = E
n
[ n) . (1.42)
Now we do two things to this equations: (a) we
multiply from the left with n [, which by denition
is the transpose plus complex conjugate of [ n)
28
,
n [ H [ n) = E
n
n [ n) ; (1.43)
(b) we take the transpose plus complex conjugate,
i.e. n [ H
+
= n [ E
n
29
, and multiply from the
right with [ n), i.e.
n [ H
+
[ n)
(1.41)
= n [ H [ n) = E
n
n [ n) . (1.44)
Comparing (1.43) and (1.44) we see that E
n
= E
n
,
i.e. the eigenvalues are real. We introduce this
27
We discuss hermitian operator in more detail on page
32.
28
The notation n [ . . . [ n) is known as the bra...ket
notation [2].
29
Here is the conjugate complex
notation in analogy to products of vectors and ma-
trices, i.e.
x
T
A y =
_
x
T
A y
_
T
= (A y)
T
x = y
T
A
T
x .
Analogously we dene
x [ A [ y) = y [ A
+
[ x)
, (1.45)
where A is an operator in the same sense as H .
But even the additional condition H = H
+
does not determine H uniquely starting from H.
We do not want to investigate the origin of the
relations (1.33) to (1.35), which is heuristic in
any case, because quantum mechanics cannot
be derived from mechanics. We rather want
to suggest a second recipe for constructing a
quantum theory. Later, however, we will return to
matrix mechanics and show how it is connected to
the ideas of wave mechanics which we explore next.
Wave mechanics:
We return to the notion that particles may be-
have as waves. A plane wave travelling in the pos-
itive x-direction may be described via
(x, t) cos
_
2
_
x
t
__
, (1.46)
where is the wavelength, and is the frequency.
We may also write this as
(x, t) cos (kx t) . (1.47)
If we think about particles, we think about local-
ized objects. However, our (x, t) thus far is not
localized. Therefore we construct a wave packet by
writing
(x, t) =
_

0
dkw(k) cos (kx t) (1.48)
with
w(k) = e
b
2
k
2
. (1.49)
Setting t = 0 for the moment we obtain for
(x, t = 0) =
_

4b
2
e
x
2
/4b
2
. (1.50)
We see that by adding up a sucient number of
plane waves, putting in many more waves around
k = 0 (long wavelengths), we obtain an object, i.e.
a wave packet localized around x = 0. The exact
form of w(k) does not matter. Here we choose a
Gaussian shape because it is simpler to integrate.
You may try this out yourself using the Mathemat-
ica program shown in Fig. 1.1, e.g. you may sub-
stitute e
]k]
instead of e
k
2
.
As Fig. 1.2 shows, we may put t > 0 and ob-
serve the wave packet travelling along x. Here we
choose = c
p
k, where c
p
= 1 is a constant velocity
identical for all plane waves in the packet
30
. This
representation of a free particle moving in space
appears acceptable.
However we have not used the correct dispersion
relation (here: k). From Eq. (1.2) we have
k = p/h. Together with h = E and E = p
2
/2m
for the classical free particle, we obtain
=
hk
2
2m
, (1.51)
i.e. k
2
. Fig. 1.2 illustrates what happens
to the wave packet (setting
h
2m
= 1). It looses its
localization as t increases!
Nevertheless, let us proceed and write more gen-
erally
(r, t) =
_
a.s.
d
3
pw( p) e
i( prEt)/ h
. (1.52)
Here we use complex waves, which is equivalent to
using cos
_
pr
h

Et
h
_
as long as we apply linear op-
erations to (r, t). Next we apply the operators
2
and
t
to Eq. (1.52):
2
(r, t) =
_
d
3
p
_
p
2
h
2
_
w( p) e
i
h
( prEt)
.
and
t
(r, t) =
_
d
3
p
_
i
h
E
_
w( p) e
i
h
( prEt)
.
30
Exact solution: (x, t) =
_

4b
2
exp
_
(x t)
2
/4b
2

4 2 2 4
x
0.2
0.2
0.4
0.6
0.8
1.0
4 2 2 4
x
0.2
0.2
0.4
0.6
0.8
1.0
wk : Expk^2
PlotSumwk Cos2 k x , k, 0, 1, .01100,
x, 5, 5, PlotRange .2, 1, AxesLabel x, " ",
TextStyle FontSize 12, PlotStyle Black
PlotSumwk Cos2 k x , k, 0, 2, .01100,
x, 5, 5, PlotRange .2, 1, AxesLabel x, " ",
TextStyle FontSize 12, PlotStyle Black
Figure 1.1: Top: A wave packet constructed by
the superposition of cos-waves using Eq. (1.49).
Because the summation over k is nite there appear
secondary maxima. These secondary maxima are
reduced by including additional k-values. Bottom:
Text of the Mathematica program.
.
-4
-2
0
2
4
x
0
0.5
1
1.5
2
t
0
0.25
0.5
0.75
1
-4
-2
0
2
4
x
-4
-2
0
2
4
x
0
0.5
1
1.5
2
t
0
0.25
0.5
0.75
1
-4
-2
0
2
4
x
Plot3D[Sum[w[k] Cos[2 (k (x - k t)) ],
{k, 0, 2, .01}]/100, {x, -5, 5}, {t, 0, 2},
PlotPoints -> 80, PlotRange -> {-.2, 1},
AxesLabel -> {"x", "t", " "},
TextStyle -> {FontSize -> 10}]
Figure 1.2: The wave packet plotted as function of
time using = k (top) and = k
2
(bottom). Also
shown is the Mathematica program which generates
the bottom graph.
.
Multiplication of the upper relation by
h
2
2m
and
the lower relation by ih yields, if both relations are
combined,
_
h
2
2m
2
ih
t
_
(r, t) = 0 , (1.53)
where we have used E = p
2
/2m! This dierential
equation looks very much like a diusion equation
with a complex diusion constant
i h
2m
. It describes
the free particle.
Notice that analogous to matrix quantum me-
chanics, where we have replaced the classical coor-
dinates and momenta by q
m
and p
m
, we now dene
the operators
E ih
t
(1.54)
and
p ih
. (1.55)
The operators r r and p ih
again satisfy
the commutator relations (1.33), (1.34), and (1.35),
i.e.
[x
m
, x
n
] = 0 (1.56)
_
p
m
, p
n
_
= 0 (1.57)
_
p
m
, x
n
_
= ih
mn
. (1.58)
More precisely
_
p
m
, x
n
_

_
p
m
, x
n
_
(r, t)
= ih(
x
m
x
n
x
n
x
m
)
= ih
mn
(r, t) . (1.59)
Notice also that we may satisfy the classical re-
lation E =
p
2
2m
+ | (r, t) (with E = const) by for-
mally adding a term | (r, t) (r, t) to (1.53). The
result is the Schrodinger equation for a single par-
ticle moving in a potential | (r, t)
31
:
_
h
2
2m
2
+| (r, t) ih
t
_
(r, t) = 0 . (1.60)
This equation, which yet has to prove itself, is
the starting point for all calculations in wave
mechanics.
The meaning of (r, t):
The rst case for which we try out Eq. (1.60) is
| (x) =
1
2
m
2
x
2
,
i.e. the 1D harmonic oscillator. First we apply the
Ansatz
(r, t) = (r) e
i
h
Et
, (1.61)
which, in one dimension, yields the time-inde-
pendent or stationary Schrodinger equation
_
h
2
2m
2
x
+| (x) E
_
(x) = 0 . (1.62)
This eigenvalue dierential equation has the solu-
tion
n
(x) =
_
m
h
_
1/4
1
n!2
n
(1.63)
H
n
__
m
h
x
_
exp
_
m
2 h
x
2
_
31
Schrodinger, Erwin, osterreichischer Physiker, *Wien
12.8.1887, ebenda 4.1.1961; Professor in Stuttgart, Breslau,
Z urich, Berlin, Oxford und Graz; 1938 Emigration nach Ir-
land (R uckkehr nach Wien 1956). Nach Arbeiten zur statis-
tischen Thermodynamik und zur Theorie des Farbensehens
baute Schrodinger 1926 die Ansatze L. deBroglies uber Ma-
teriewellen zur Wellenmechanik aus; die von ihm als Wellen-
gleichung aufgestellte Schr odinger-Gleichung wurde durch
die statistische Deutung (Quantenmechanik) zur nichtrela-
tivistischen Bewegungsgleichung quantenmechanischer Sys-
teme. Sp ater bearbeitete er Probleme der relativistischen
Quantenmechanik, der Gravitationstheorie und der ein-
heitlichen Feldtheorie. 1933 erhielt er mit P.A.M. Dirac den
Nobelpreis f ur Physik. c _2000 Bibliographisches Institut &
F. A. Brockhaus AG
with
H
n
(q) = (1)
n
e
q
2 d
n
dq
n
e
q
2
. (1.64)
The H
n
(q) are Hermite polynomials (H
0
(q) = 1,
H
1
(q) = 2q, . . . )
32
. Each index n corresponds to
an energy eigenvalue
E
n
= h
_
n +
1
2
_
n = 0, 1, 2, . . . (1.66)
This means that Schrodingers equation agrees with
(1.39)! Notice in particular that E
0
=
1
2
h. Notice
also that the discrete eigenvalues result from the
requirement
n
() = 0. This is illustrated in
Fig. 1.3 for the case of n = 0, which shows the
numerical solution for three dierent energies, only
one of which corresponds to the possible energy
E
0
=
1
2
h. However, what is the meaning of the so
called wave function
n
(x)?
Fig. 1.4 shows the wave functions
0
(q) to
6
(q). Guided by the original idea behind the wave
packet we might try to interpret
n
(q) as a mea-
sure of likelihood to nd a particle in a harmonic
potential at a certain q-value. But the
n
(q), ex-
cept for the ground state, n = 0, may be negative.
The next best thing to try is
n
(q)
2
or even better
[
n
(q) [
2
=
n
(q)
n
(q), because in general the
wave function may be a complex function (as we
will see). Accepting this idea we must require
_

dq [
n
(q) [
2
= 1 , (1.67)
32
Substitution of x =
_
h/ (m)q into Eq. (1.62) yields
_
2
q
q
2
+ 2
_
(q) = 0
with E = h. The Ansatz (q) = constH
n
(q) exp[q
2
/2]
transforms this equation to its nal form known in mathe-
matics as Hermites dierential equation:
_
2
q
2q
q
+ 2 1
_
H
n
(q) = 0 (1.65)
with 2 1 = 2
_
n +
1
2
_
1 = 2n (n = 0, 1, 2, . . .) [18]
(Chapter 22).
We revisit the harmonic oscillator again in section 1.5.
The approach discussed there will allow us to construct the
solutions more easily. Thus, there is no need to discuss Her-
mites dierential equation in detail at this point.
-4 -2 0 2 4
0
q
=0.5
=0.45
=0.55
Figure 1.3:
0
of the 1D harmonic oscillator com-
puted numerically for dierent starting from q =
4 using the initial values
0
=
t
0
= 0.0001. Here
q =
_
m/hx.
-3 -2 -1 1 2 3
q
1
2
3
4
5
6
7
y
n
HqL
Figure 1.4: The ground state and the rst ve ex-
cited states of the 1D harmonic oscillator (shifted
according to energy) together with the harmonic
potential. Here q =
_
m/hx, whereas energy is
in units of h.
which means that the particle must be somewhere
in space
33 34
. It also means that [
n
(q) [
2
should
be interpreted as a probability density.
Note also that the integration extends from
to and not from a to a given by the classical
condition E | (a) = 0
35
. As Fig. 1.4 shows,
if our interpretation of [
n
(q)
2
[ is correct, then
the quantum particle penetrates into the classically
forbidden area E < | (q)
36
!
Let us again consider the correspondence prin-
ciple, which allows a stringent check of our inter-
pretation of the wave function. For the classical
harmonic oscillator p
cl
(x) x is the probability for
nding the oscillator in the interval x. p
cl
is given
by
p
cl
(x) =
1
a
1
_
1
_
x
a
_
2
. (1.68)
In order to compare this with p
qt
(x) [
n
(x) [
2
we write E =
1
2
m
2
a
2
and inserting E =
h
_
n +
1
2
_
we obtain
a =
_
h
m
2
_
n +
1
2
_
. (1.69)
Fig. 1.5 shows the results for n = 0, n = 1,
n = 10 and n = 20. We see that for large n, i.e.
in the limit where the correspondence principle
should hold, indeed p
cl
(x) p
qt
(x) or better
p
cl
(x) x p
qt
(x) x. This is a nice conrmation
of the assumptions which we have made thus far.
One-dimensional rectangular well potential I:
Before we launch into the discussion of the ap-
plication of Schrodingers equation to problems like
the hydrogen atom, it is wise to exercise our skills
by means of a simpler problem. Imagine the follow-
ing not very realistic but nevertheless instructive
potential in one dimension:
33
This normalization already is build into Eq. (1.63).
34
This explains the above requirement
n
() = 0!
35
Integration from a to a (requiring (a) = 0) does
not yield the correct energy eigenvalues!
36
If the potential barrier is not innite, as in the case
at hand, then there is a nite probability for the particle
to penetrate this barrier, which is not possible in classical
mechanics. This is called tunneling.
-2 -1 1 2
q
0.5
1
1.5
2
n=0
-3 -2 -1 1 2 3
q
0.2
0.4
0.6
0.8
1
n=1
-4 -2 2 4
q
0.2
0.4
0.6
0.8
n=2
-10 -5 5 10
q
0.05
0.1
0.15
0.2
n=20
Figure 1.5: Comparison of p
qt
(x) with p
cl
(x) for
n = 0, 1, 2, and 20. Again the unit of length is
_
h/m. Dashed vertical lines indicate the classi-
cal turning points.
.
-a/2 a/2 x
U(x)
-U
o
Figure 1.6: Rectangular well potential in one di-
mension.
| (x) =
_
|
o
for
a
2
x
a
2
0 otherwise
(|
o
> 0). The potential is depicted in Fig. 1.6.
It is a so called rectangular well potential. There
are two important dierences compared to the har-
monic oscillator. The potential is nite, and its
shape requires the separate treatment of the three
regions x < a/2 (I), a/2 x a/2 (II), and
a/2 < x (III).
We start from the stationary Schrodinger equa-
tion (1.62). Using the substitution
x =
a
2
q ,
where q is dimensionless, yields
_
2
q
u(q) +
_
(q) = 0 . (1.70)
Notice that u(q) is given by
u(q) =
_
u
o
for 1 q 1
0 otherwise
,
where
u
o
=
ma
2
2 h
2
|
o
.
and
=
ma
2
2h
2
E . (1.71)
The general solutions in each of the three regions
are
(q) =
(1)
,o
e
ik
q
+
(2)
,o
e
ik
q
, (1.72)
where
(1)
,o
and
(2)
,o
are constants and =
I, II, III. Inserting (1.72) into (1.70) yields
k
2
+ = 0 ,
where u
I
= u
III
= 0 and u
II
= u
o
, or
k
+ .
In the following we concentrate on the so called
bound states for which < 0, i.e. the particle is
trapped inside the potential well
37
. Thus
k
I
= k
III
= i
_
[ [ and k
II
=
u
o
+
(u
0
+ 0) and therefore (using k
I
= k
III
=
i
I
(q) =
(1)
I,o
e
q
+
(2)
I,o
e
III
(q) =
(1)
III,o
e
q
+
(2)
III,o
e
q
.
In order to ensure normalizability (cf. Eq. (1.67))
we must require
(2)
I,o
=
(1)
III,o
= 0 .
We are left with four unknown coecients
(i)
,o
and
. At this point we assume the following boundary
conditions
I
(1) =
II
(1)
I
(q) [
q=1
=
q
II
(q) [
q=1
37
The situation > 0 we discuss in the next section start-
ing on page 24.
(set 1) and
II
(1) =
III
(1)
II
(q) [
q=1
=
q
III
(q) [
q=1
,
(set 2), i.e. the solution is continuous and dieren-
tiable at q = 1
38
.
From set 1 follows
(1)
I,o
e
=
(1)
II,o
e
ik
II
+
(2)
II,o
e
ik
II
(1)
I,o
e
= ik
II
_
(1)
II,o
e
ik
II
(2)
II,o
e
ik
II
_
and therefore
= ik
II
(2)
II,o
e
ik
II
(1)
II,o
e
ik
II
(2)
II,o
e
ik
II
+
(1)
II,o
e
ik
II
. (1.73)
From set 2 follows
(2)
III,o
e
=
(1)
II,o
e
ik
II
+
(2)
II,o
e
ik
II
(2)
III,o
e
=
k
II
i
_
(1)
II,o
e
ik
II
(2)
II,o
e
ik
II
_
and therefore
= ik
II
(1)
II,o
e
ik
II
(2)
II,o
e
ik
II
(1)
II,o
e
ik
II
+
(2)
II,o
e
ik
II
. (1.74)
Eqs. (1.73) and (1.74) may be fullled simultane-
ously by setting
(1)
II,o
=
(2)
II,o
. .
(+)
or
(1)
II,o
=
(2)
II,o
. .
()
.
The condition (+) yields
u
o
+ tan
u
o
+ , (1.75)
38
We discuss these assumptions at the end of this section.
p
2
p
z
-2
2
4
Untitled-1 1
Figure 1.7: Example graphical solutions for u
o
=
5/3 indicated by the solid circles. Solid lines:
tanz; short-dashed line: cot z; long-dashed line:
_
(u
o
z
2
)/z
2
.
whereas the condition () yields
u
o
+ cot
u
o
+ . (1.76)
Using the substitution z =
u
o
+ we have
_
u
o
z
2
z
2
= tanz (+)
and
_
u
o
z
2
z
2
= cot z ()
Fig. 1.7 illustrates the graphical solution of these
equations.
The z-values at which the long-dashed curve in-
tersects tanz and cot z, respectively, correspond
to discrete energy levels
2n1
= u
o
z
2
2n1
(+) , (1.77)
and
2n
= u
o
z
2
2n
() , (1.78)
where n = 1, 2, . . . , n
max
. Obviously there is a
highest energy level for a given nite u
o
. On the
other hand the smallest possible number of energy
levels is one when u
o
becomes less than /2.
If we use (1.77) and (1.78) to write down corre-
sponding solutions in region II we nd
II,2n1
(q) = c
II,2n1
cos
__
u
o
+
2n1
q
_
and
II,2n
(q) = c
II,2n
sin
_
u
o
+
2n
q
_
,
where c
II,2n1
and c
II,2n
are constants. The com-
plete solution can be constructed by tying these
functions to the solutions for q < 1 and q > 1:
I/III,2n1
(q) = c
I/III,2n1
exp
_
2n1
q
_
and
I/III,2n
(q) = c
I/III,2n
exp
_
2n
q
_
.
The coecients are determined by the required
continuity together with the overall normalization
_

dq [ (q) [
2
= 1 .
Fig. 1.8 shows the two possible solutions for u
o
=
5/3 (cf. Fig. 1.7). As in the case of the har-
monic oscillator there is a nite probability to nd
the particle in the classically forbidden regions I
and III. This probability obviously is larger for the
energy level higher up in the potential well.
An interesting limit is u
o
, i.e. the rect-
angular well becomes an innite rectangular well.
In this case the function
_
(u
o
z
2
)/z
2
depicted
in Fig. 1.7 approximates a horizontal straight line.
The corresponding energy levels are
()
2n1
= u
o
+
2n1
=
_
2
(2n 1)
_
2
(+) , (1.79)
and
()
2n
= u
o
+
2n
=
_
2
(2n)
_
2
() . (1.80)
q
yHqL
Untitled-3 1
Figure 1.8: Ground state and one excited state cor-
responding to the graphical solutions in Fig. 1.7.
The horizontal lines indicate the energy eigenvalues
relative to the potential.
Notice that
()
measures a levels energy from
the bottom of the potential well on an energy scale
for which the bottom is at zero energy. The nor-
malized solutions are particularly simple:
()
II,2n1
(q) = cos
_
2
(2n 1) q
_
and
()
II,2n
(q) = sin
_
2
2nq
_
with n = 1, 2, . . . , . Outside region II, i.e. outside
the well, the solutions vanish due to the exponential
factors
e
2n1
]q]
and e
2n
]q]
.
Note that both
2n1
and
2n
are innitely large
for every nite n.
Before we move on to the next section, we briey
return to the assumptions regarding the continuity
and dierentiability of the wave functions on page
15. Let us approximate the rectangular well poten-
tial shown in Fig. 1.8 using the following function:
u(q) =
5
3
exp[q
m
] . (1.81)
The results for m = 2, 10, and 40 are shown in Fig.
1.9. Obviously, we obtain the rectangular well in
the limit of large m. We use the same numerical
method as in Fig. 1.3 to compute eigenfunctions
and energy eigenvalues for this potential. Selected
examples, i.e. two ground state wave functions,
are shown in Fig. 1.10. The result for m = 40
is not included, because it basically cannot be dis-
tinguished from the corresponding result obtained
above for the true rectangular well (shown here as
dashed curve). The eigenvalues are listed in the
following table:
m
1
2 3.3243 . . .
10 3.9485 . . .
40 4.0457 . . .
4.0706 . . .
Apparently we are able to obtain the rectangu-
lar well results via a continuous deformation of
a smooth potential function thereby avoiding the
sharp edges which bring about the boundary con-
ditions introduced on page 15. In this sense the
above is a numerical justication of the bound-
ary conditions - at least in this special case.
-2 -1 1 2
q
-5
-4
-3
-2
-1
1
uq
Figure 1.9: Approximations to the rectangular well
potential in Fig. 1.8 via Eq. (1.81) with m = 2, 10,
and 40.
1.4 The hydrogen atom
The obvious problem to tackle next is the hydrogen
atom. This is the rst major project, to which
we apply wave mechanics. It also is a two-body
problem. We proceed by writing H in terms of
the cartesian coordinates of the electron r
e
and the
nucleus r
n
:
-3 -2 -1 1 2 3
q
0.2
0.4
0.6
0.8
q
In[91]:= f NDSolve''q 3.32432 q
5 3 Expq^2 q,
6 0.000001, '6 0.00,
, q, 4, 4;
q_ : q . f1;
norm NIntegrateq^2, q, 4, 4;
Plot qSqrtnorm, q, 3, 3
Figure 1.10: Top: Numerical ground state wave
functions for the potential (1.81) with m = 2 and
10. The dashed curve is the previous result for
the rectangular well. Bottom: Mathematica-code
for computing and plotting the ground state wave
function in the case m = 2.
H =
h
2
2m
e
2
e
h
2
2m
n
2
n
+| ([ r
e
r
n
[) .(1.82)
Next we introduce the new coordinates
r = r
e
r
n

R =
m
e
r
e
+m
n
r
n
m
e
+m
n
(in analogy to the Kepler problem in classical me-
chanics; cf. section 5.2 in [14]), which yield
H =
h
2
2
2
r
h
2
2m
2
R
+| (r) , (1.83)
where = m
e
m
n
/ (m
e
+m
n
) is the reduced mass,
and m = m
e
+ m
n
. We disregard the second
term, because we are interested in the solution
for the center of mass frame only. The result-
ing Schrodingers equation corresponds exactly to
Eq. (1.60). The potential energy obviously is
1.4. THE HYDROGEN ATOM 19
| (r) = (4
o
r)
1
e
2
, where e is the electron
charge. Using again (r, t) = (r) e
i
h
Et
we -
nally have in dimensionless (!) units
_
2

+
2
+
_
( ) = 0 , (1.84)
i.e. lengths now are in units of
a =
4
o
h
2
e
2

4
o
h
2
m
e
e
2
a
o
, (1.85)
39
, where a
o
= 0.529177
A is Bohrs radius or the

atomic unit of length, and
=
2a
2
h
2
E 2
_
e
2
4
o
a
o
_
1
E .
Note that
e
2
4
o
a
o
, the atomic unit of energy, is called
the hartree (symbol E
h
) (1 hartree = 27.2114eV =
4.35975 10
18
J)
40
.
It is useful to introduce spherical coordinates,
because we are dealing with a problem possessing
spherical symmetry. In addition we try to simplify
matters using the following separation ansatz:
(, , ) =
u()
Y (, ) . (1.86)
Here u() solely depends on the dimensionless ra-
dial coordinate , whereas Y (, ) is a function of
the angles and . We also need the Laplace op-
erator expressed in spherical coordinates:
2

=
1
_
+

2
,
where
2
,
=
1
2
_
1
sin
2
+
1
sin
(sin
)
_
39
= m
e
1
1 +
m
e
m
n
m
e
_
1
m
e
m
n
_
40
In the following we will omit the factor 4
o
.
(cf. appendix B. 1). Inserting (1.86) into (1.84)
yields
2
u()
_
+
2
+
_
u() (1.87)
=
1
Y (, )
2
,
Y (, ) ,
where we have used
1
_
u()
=
1
u() .
Notice that the left side of Eq. (1.87) depends ex-
clusively on , whereas the right side depends on
the angles and . In order for this equation to
be generally valid, both sides must be equal to a
constant c, i.e.
_
+
2
+
_
u() = c
u()
2
(1.88)
and
2
,
Y (, ) = cY (, ) . (1.89)
The second (eigenvalue) dierential equation,
Eq. (1.89), we investigate in detail in the next sec-
tion. We will nd that c is given by
c = l (l + 1)
with
l = 0, 1, 2, . . . . (1.90)
The corresponding solutions,
Y (, ) = Y
lm
(, ) (1.91)
are the so called spherical harmonics (dt.:
Kugelachenfunktionen), where
m = l, l 1, . . . , 0, . . . , l + 1, l . (1.92)
These values for the index m follow from a sec-
ond (eigenvalue) dierential equation satised by
Y
lm
(, ):
i
Y
lm
(, ) = mY
lm
(, ) . (1.93)
This again is an equation which we will study in
detail in the next section.
Eqs. (1.89) and (1.93) both have distinct phys-
ical meaning. If we dene an angular momentum
operator via
L = r p ,
where p = ih
, it is rather straightforward to
show that
L
2
= h
2
2
,
(1.94)
and
L
z
= ih
(1.95)
(cf appendix B.1.). In other words this means ac-
cording to Eqs. (1.89) and (1.93)
L
2
Y
lm
(, ) = h
2
l (l + 1) Y
lm
(, ) (1.96)
and
L
z
Y
lm
(, ) = hmY
lm
(, ) . (1.97)
Analogous to the energy described by the Hamil-
ton operator H, we now have two additional phys-
ical quantities, the angular momentum square and
the z-component of angular momentum, which are
quantized! The quantization condition (1.96) must
be compared to Eq. (1.83), corresponding to quasi-
classical quantization. As in the case of energy
quantization there is a noticeable dierence for
smaller integer values of l (or n). The quantiza-
tion condition (1.97) is a bit mysterious, because it
tells us, that the direction of angular momentum is
quantized even though we are free to choose the di-
rection of the z-axis of our coordinate system. We
will come back to this point latter in the course.
For the moment we want to return to the radial
equation (1.88).
Using the result (1.89) the radial equation be-
comes
_
l (l + 1)
2
+
2
. .
/
eff
+
_
u
l
() = 0 . (1.98)
Here we dene |
eff
merely to highlight the simi-
larity with the classical two-body problem. At this
point it is useful to investigate the limiting behav-
ior of u
l
() for and 0. In the former
case Eq. (1.98) becomes
_
+
_
u
l
() = 0 and
u
l
() e
; in the latter case Eq. (1.98) be-

comes
_

l(l+1)
2
_
u
l
() = 0 and u
l
()
(l+1)
.
Therefore it is sensible to try
u
l
() = e
l+1
j=0
c
j
j
. (1.99)
Inserting Eq. (1.99) into Eq. (1.98) yields the re-
cursion relation
c
j+1
=
2
_
l +j + 1
1
_
(j + 1) (j + 2l + 2)
c
j
. (1.100)
In order for the series to remain nite we require
[. . .] = 0 at j = j
max
. Therefore
1
must be an
integer number n l + 1 , i.e.
n
=
1
n
2
(n l + 1) . (1.101)
We recognize that our previous solution, Eq.
(1.6), has emerged. However, there is more infor-
mation here. First we note that we get several so-
lutions for each value of n. More exactly, there are
0 l n 1 possible values for l, and each l has
2l + 1 possible m-values. Thus there are
n1
l=0
(2l + 1) = n
2
(1.102)
solutions
n
(, , ) for each
n
. This is called de-
generacy. The high level of degeneracy is a special
feature of the 1/r potential. Later we will learn
that if the Coulomb potential is altered by an ex-
ternal eld, we call this a perturbation, then the
degeneracy is reduced or entirely removed.
We now turn to calculating the wave functions
explicitly:
1. n = 1 according to (1.100) implies 0 = l +
j
max
+11 and therefore j
max
= 0. The solu-
tion
n,l,m
(, , ) is
1,0,0
(, , ) =
c
0
4
e
.
The normalization condition
_
d
3
r [ [
2
= 1
yields c
0
= 2, i.e.
1,0,0
(, , ) =
1
. (1.103)
2. n = 2 implies according to (1.100) 0 = l +
j
max
+ 1 2 , i.e. j
max
= 1 l. For l = 0 we
have
2,0,0
=
c
0
4
_
1

2
_
e
/2
and therefore
2,0,0
=
1
8
_
1

2
_
e
/2
; (1.104)
for l = 1 we obtain
2,1,1
=
1
8
sine
i
e
/2
(1.105)
2,1,0
=
1
4
2
cos e
/2
(1.106)
2,1,1
=
1
8
sine
i
e
/2
. (1.107)
3. . . .
4. The general expression for
n,l,m
(, , ) is
n,l,m
(, , ) = R
n,l
() Y
lm
(, ) , (1.108)
where the radial wave function is
R
n,l
() =
2
n
2
(n l 1)!
(n +l)!
_
2
n
_
l
exp
_
n
_
L
2l+1
nl1
_
2
n
_
(1.109)
[18] (chapter 22). L
2l+1
nl1
_
2
n
_
are generalized
Laguerre polynomials; they can be obtained
via LaguerreL
_
n l 1, 2l + 1,
2
n
in Mathe-
matica.
In spectroscopy n is called the main quantum
number, and the dierent l = 0, 1, 2 . . . are denoted
by small letters, i.e. l = s, p, d, . . . We will call every
n,l,m
(, , ) an atomic orbital. Thus hydrogen
has one s-orbital for every n, three p-orbitals for
every n if l = 1, ve d-orbitals for every n if l = 2
and so on
41
. Fig. 1.11 shows a schematic of the
hydrogen energy levels expressed in this language.
Fig. 1.12 shows the radial part of the ground state
wave function as well as the radial wave functions
corresponding to the rst excited energy level. Fig.
1.13, nally, shows the absolute square of the angu-
lar part of the wave functions shown in the previous
gure. These can be viewed as orientation distri-
bution of the electron cloud.
It is important to note that the complete solution
of Schrodingers equation for a hydrogen atom with
energy E
n
in the absence of symmetry breaking
elds
42
is the linear combination
n
(, , ) =
n1
l=0
l
m=l
c
nlm
n,l,m
(, , ) , (1.110)
where

n1
l=0
l
m=l
c
nlm
= 1. This follows
because the spherical harmonics form a complete
set of orthogonal functions (cf. the appendix).
Magnetic moment of atomic hydrogen:
In electrodynamics we had derived the following
formula for the magnetic moment produced by a
current density e
j:
41
Examples for common notation:
1,0,0

1s
or
2,1,1

2p
1
.
42
The level splitting due to an electrostatic eld is called
Stark eect. If the applied eld is a magnetic eld the anal-
ogous eect is called Zeeman eect.
-1/4
-1
-1/9
-1/16
l=0 l=1 l=2 l=3
E
n
2s
3s
4s
1s
2p
3p
4p
3d
4d 4f
.
.
.
.
.
.
.
.
.
.
.
.
Figure 1.11: Spectroscopic terminology of the hy-
drogen energy levels.
m =
e
2c
_

j () d
3
(1.111)
(cf. [15]; Eq. (3.62)). Here
e
_
= 1.602 10
19
C
_
is the electron charge,
and

j is a current density in units of e. In the
following we rst derive a general expression for
j in terms of the wave function. Then we write

down an expression for m in the case of hydrogen
in a state described by n, l, and m.
We start by adding the equations
t
=
1
ih
H (1.112)
=
1
ih
H
. (1.113)
The result is
t
(
) +
ih
2m
_
_
= 0 , (1.114)
where we have used H =
p
2
2m
+| and assumed that
| is real and independent of velocity. The second
bracket in (1.114) may be transformed, i.e.
(. . .) =
+
_
__
__
_
2 4 6 8 10
r
0.2
0.4
0.6
0.8
1
2 R
1, 0
HrL
2 4 6 8 10
r
-0.4
-0.2
0.2
0.4
0.6
0.8
1
!!!!
2 R
2, 0
HrL,
!!!!
2 R
2, 1
HrL
Figure 1.12: Radial wave functions for the hydro-
gen atom. Top: The ground state 2R
1,0
() includ-
ing the Mathematica-code. Bottom:

2R
2,0
()
and

2R
2,1
(). Note that R
2,1
becomes zero at
the center.
Y
0,0
2
x
z
Y
1, 0
2
y
z
y
Y
1, 1
2
x
y
z
x
y
Untitled-1 1
Figure 1.13: Angular charge distributions. Top:
The ground state [Y
00
(, )[
2
including the
Mathematica-code. Bottom: [Y
10
(, )[
2
and
[Y
11
(, )[
2
.
_
. (1.115)
This yields the continuity equation
t
+

j = 0 (1.116)
with
[ [
2
(1.117)
and the probability current density
j
ih
2m
_
_
. (1.118)
Note that in the case of hydrogen we must replace
m by .
We proceed by evaluating (1.118) for hydrogen
using spherical coordinates, i.e.
= e
+e
1
sin
+e
(cf. the appendix). Because the radial part of the

hydrogen wave functions is real, cf. Eq. (1.108),
we immediately have
= e
i
h
2
_
_
= 0 , (1.119)
which makes sense if hydrogen is supposed to be
stable. The same argument applies for the -part
of the wave functions (cf. appendix), i.e.
= e
i
h
2
_
_
= 0 . (1.120)
Thus, the only non-vanishing contribution is
= e
i
h
2
1
sin
_
_
. (1.121)
For the moment we concentrate on
n,l,m
instead
on
n
given by Eq. (1.110), i.e. in this special state
j
,lm
= e
mh
sin
[
n,l,m
[
2
. .
j
,lm
, (1.122)
where we have used
Y
lm
= imY
lm
(cf. the ap-
pendix). This is a current density corresponding
to a current circling the z-axis in counter clockwise
direction. Inserting Eq. (1.122) into Eq. (1.111)
yields
m
t
=
e
2c
_
j
,lm
e
. .
=e
d
3
=
mhe
2c
_
e
sin
[
n,l,m
[
2
d
3
()
=
mhe
2c
_
[
n,l,m
[
2
d
3
. .
=1
_
_
0
0
1
_
_
,
and thus
m
t
=
mhe
2c
e
z
. (1.123)
The equation () is correct because [
n,l,m
[
2
does
not depend on . The prime in Eq. (1.123) reminds
us that we have used
n,l,m
only to compute m.
We note that the quantity
m
B
=
he
2m
e
c

he
2c
(1.124)
is called Bohrs magneton. Note also that
0 [

m
t
[ lm
B
. (1.125)
From the appendix we have L
z
n,l,m
=
ih
n,l,m
, i.e.
L
z
n,l,m
= hm
..
L
z
n,l,m
(1.126)
(cf. Eq. (1.97)). The quantity g, dened via
m
t
z
h
L
z
=
he
2c
g , (1.127)
is called gyromagnetic factor or g-factor. Here we
have g = 1 obviously. If we insert the full linear
combination (1.110) instead of just
n,l,m
into Eq.
(1.121) then a similar but somewhat more elaborate
calculation (homework problem) yields
m = 0 . (1.128)
This is the result for the completely degenerate
hydrogen atom in the energy eigenstate
n
.
One-dimensional rectangular well potential II:
In order to better understand the meaning of the
probability current density introduced above, we
return to the rectangular potential well (cf. page
14). Previously we had omitted the discussion of
states above the potential well ( > 0; cf. Fig. 1.6),
i.e. the states no longer are bound states. The
general solutions for > 0 still are given by Eq.
(1.72). In addition we employ the same boundary
conditions as before (cf. set 1 and set 2 on page
15). However, there is an important dierence to
the < 0-case when the states are bound. The
overall normalization of the wave function no longer
converges. All k
are positive real numbers and the

solution is oscillatory in all three regions q < 1,
1 q 1, and 1 < q. In addition no longer has
to assume discrete values.
In principle there are two alternatives at this
point. The rst one is illustrated in Fig. 1.14,
where the original potential well now is a small de-
pression embedded in the middle of a vastly larger
normalization potential well with innitely high
walls.
Solving Schrodingers equation for the box-
inside-a-box problem is not dicult. Our previous
bound state solutions, i.e. the solutions for < 0,
are not aected by the normalization well. This is
!>0
!<0
Figure 1.14: Square well potential in one dimension
(small well) including normalization box.
because the bound state wave functions decay to
zero before they begin to feel the walls of the nor-
malization well
43
. However, the solutions for > 0
now must vanish at the walls of the normalization
well and beyond (cf. our discussion of the innite
potential well on page 17). The resulting wave func-
tion is normalizable, and the additional boundary
condition yields discrete energies as before. The
spacing E of these energy values depends on the
width and depths of the original well, but more
strongly it is determined by the width of the nor-
malization well. If we call this quantity a
nw
then,
as we have seen before, we have E a
2
nw
(cf. Eq.
(1.71)). In other words, for large a
nw
the spectrum
of energy eigenvalues for > 0 is almost continu-
ous. Here, however, we want to pursue a dierent
approach. But we will return to the method of the
normalization box.
The second alternative is based on a dierent
idea. The problem is treated as a scattering prob-
lem. The solutions (1.72) are used to construct
one-dimensional probability current densities. In
the following j
i
corresponds to a current (lets say)
incident from q = on the potential well. We
are interested in the transmitted current density, j
t
,
in region III (q > 1) propagating to the right and in
the reected current density, j
r
, propagating to the
left in region I (q < 1). We will calculate the so
called transmission coecient T = [j
t
/j
i
[ and the
reection coecient R = [j
r
/j
i
[ (Note: T +R = 1).
Thus, T measures the probability of an incident
particle to overcome the potential well, whereas R
measures the probability that the incident particle
is reected.
Incident particles are described via
I
(q) =
(1)
I,o
e
ik
I
q
.
Here k
I
is a positive wave vector. The coecient
(2)
I,o
vanished because otherwise there also would be
a wave in the opposite direction with k
I
. Likewise
we have
III
(q) =
(1)
III,o
e
ik
III
q
.
for the transmitted particles. Now we can work
out the current densities via Eq. (1.118). In order
43
Unless of course for energy eigenvalues close zero, when
the tails of the attending eigenfunction decay slowly.
to simplify the computation we use the algebraic
computer program Mathematica. The following is
a list of the various program steps (in bold face)
and their results.
First we express the coecients of the solutions
in the regions I to III in terms of the coecient
(1)
I,o
which is called A in the program:
"Rectangular Square Well H!>0L";
In[1]:= "Wave vectors and wave functions";
k1 =
!!!!
! ;
k2 =
!!!!!!!!!!!!!
u0 + ! ;
k
3
=
!!!!
! ;
y
1
@q_D := A Exp@I k
1
qD + c12 Exp@-I k
1
qD
y2 @q_D := c21 Exp@I k2 qD + c22 Exp@-I k2 qD
y3 @q_D := c31 Exp@I k3 qD
In[8]:= "Solving for the coefficients using the assumption that
the wave function is continuous and differentiable
at q=!1. Notice that A remains undetermined.";
s = Solve@8y
1
@-1D == y
2
@-1D, y
2
@1D == y
3
@1D,
HD@y1 @qD, qD . q -1L HD@y2 @qD, qD . q -1L,
HD@y
2
@qD, qD . q 1L HD@y
3
@qD, qD . q 1L<,
8c12, c21, c22, c31<D;
ss = FullSimplify@Flatten@sDD;
c12 = c12 . ss
c21 = c21 . ss
c22 = c22 . ss
c31 = c31 . ss
Out[11]=
A
-2 I
!!!!
! +
!!!!!!!!!!!!
!+u0 M
I-1 +
4
!!!!!!!!!!!!
!+u0
M u0

2 H2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u
0
D
!!!!!!!!!!!!!
! + u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D H2 ! + u
0
LL
Out[12]=
A
- I
!!!!
! +
!!!!!!!!!!!!
!+u0 M !!!
! H
!!!
! +
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[13]=
A
I-
!!!!
! +
!!!!!!!!!!!!
!+u0 M
H-! +
!!!
!
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 - Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[14]=
2 A
-2
!!!!
!
!!!
!
!!!!!!!!!!!!!
! + u0

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
"Computing the probability current densities";
RectSqWell.nb 1
k1 =
!!!!
! ;
k2 =
!!!!!!!!!!!!!
u0 + ! ;
k
3
=
!!!!
! ;
y
1
@q_D := A Exp@I k
1
qD + c12 Exp@-I k
1
qD
s = Solve@8y
1
@-1D == y
2
@-1D, y
2
@1D == y
3
@1D,
HD@y
2
@qD, qD . q 1L HD@y
3
@qD, qD . q 1L<,
8c12, c21, c22, c31<D;
c12 = c12 . ss
c21 = c21 . ss
c22 = c22 . ss
c31 = c31 . ss
Out[11]=
A
-2 I
!!!!
! +
!!!!!!!!!!!!
!+u0 M
I-1 +
4
!!!!!!!!!!!!
!+u0
M u0

2 H2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u
0
D
!!!!!!!!!!!!!
! + u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D H2 ! + u
0
LL
Out[12]=
A
- I
!!!!
! +
!!!!!!!!!!!!
!+u0 M !!!
! H
!!!
! +
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[13]=
A
I-
!!!!
! +
!!!!!!!!!!!!
!+u0 M
H-! +
!!!
!
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 - Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[14]=
2 A
-2
!!!!
!
!!!
!
!!!!!!!!!!!!!
! + u0

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
RectSqWell.nb 1
Thus the coecients are
k1 =
!!!!
! ;
k2 =
!!!!!!!!!!!!!
u0 + ! ;
k
3
=
!!!!
! ;
y
1
@q_D := A Exp@I k
1
qD + c12 Exp@-I k
1
qD
s = Solve@8y
1
@-1D == y
2
@-1D, y
2
@1D == y
3
@1D,
HD@y
2
@qD, qD . q 1L HD@y
3
@qD, qD . q 1L<,
8c12, c21, c22, c31<D;
c12 = c12 . ss
c21 = c21 . ss
c22 = c22 . ss
c31 = c31 . ss
Out[11]=
A
-2 I
!!!!
! +
!!!!!!!!!!!!
!+u0 M
I-1 +
4
!!!!!!!!!!!!
!+u0
M u0

2 H2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u
0
D
!!!!!!!!!!!!!
! + u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D H2 ! + u
0
LL
Out[12]=
A
- I
!!!!
! +
!!!!!!!!!!!!
!+u0 M !!!
! H
!!!
! +
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[13]=
A
I-
!!!!
! +
!!!!!!!!!!!!
!+u0 M
H-! +
!!!
!
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 - Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[14]=
2 A
-2
!!!!
!
!!!
!
!!!!!!!!!!!!!
! + u0

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
RectSqWell.nb 1
Now we evaluate the current densities as well as
the transmission (T) and reection (R) coecient.
Figure 1.15 shows that for small positive reection
dominates, whereas for large positive we nd T
1. Also worth noting is the oscillatory intermittent
behavior.
k1 =
!!!!
! ;
k2 =
!!!!!!!!!!!!!
u0 + ! ;
k
3
=
!!!!
! ;
y
1
@q_D := A Exp@I k
1
qD + c12 Exp@-I k
1
qD
s = Solve@8y
1
@-1D == y
2
@-1D, y
2
@1D == y
3
@1D,
HD@y
2
@qD, qD . q 1L HD@y
3
@qD, qD . q 1L<,
8c12, c21, c22, c31<D;
c12 = c12 . ss
c21 = c21 . ss
c22 = c22 . ss
c31 = c31 . ss
Out[11]=
A
-2 I
!!!!
! +
!!!!!!!!!!!!
!+u0 M
I-1 +
4
!!!!!!!!!!!!
!+u0
M u0

2 H2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u
0
D
!!!!!!!!!!!!!
! + u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D H2 ! + u
0
LL
Out[12]=
A
- I
!!!!
! +
!!!!!!!!!!!!
!+u0 M !!!
! H
!!!
! +
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[13]=
A
I-
!!!!
! +
!!!!!!!!!!!!
!+u0 M
H-! +
!!!
!
!!!!!!!!!!!!!
! + u0 L

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 - Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
Out[14]=
2 A
-2
!!!!
!
!!!
!
!!!!!!!!!!!!!
! + u0

2
!!!
! Cos@2
!!!!!!!!!!!!!
! + u0 D
!!!!!!!!!!!!!
! + u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D H2 ! + u0 L
RectSqWell.nb 1
In[15]:= ji =
I
2 m
HA Exp@I k1 qD D@Simplify@
Conjugate@A Exp@I k
1
qDD, ! > 0 && u
0
> 0 && q < 0D, qD -
Simplify@Conjugate@A Exp@I k1 qDD, ! > 0 && u0 > 0 && q < 0D
D@A Exp@I k1 qD, qD L
Out[15]=
A
!!!
! Conjugate@AD

m
In[16]:= jt =
I
2 m
FullSimplify@y3 @qD
D@Simplify@Conjugate@y3 @qDD, ! > 0 && u0 > 0 && q > 0D, qD -
Simplify@Conjugate@y
3
@qDD, ! > 0 && u
0
> 0 && q > 0D
D@y3 @qD, qD, Assumptions 8! > 0, u0 > 0< D
Out[16]=
4 A !
32
Conjugate@AD H! + u
0
L

m I4 !
2
+ 4 ! u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D
2
u
0
2
M
In[17]:= "Transmission coefficient:";
T = FullSimplify@Abs@j
t
j
i
D, Assumptions 8! > 0, u
0
> 0<D
Out[18]=
4 ! H! + u0 L

4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
In[19]:= jr =
I
2 m
FullSimplify@
c12 Exp@-I k
1
qD D@Simplify@Conjugate@c12 Exp@-I k
1
qDD,
! > 0 && u0 > 0 && q < 0D, qD -
Simplify@Conjugate@c12 Exp@-I k1 qDD, ! > 0 && u0 > 0 && q <
D@c12 Exp@-I k
1
qD, qD, Assumptions 8! > 0, u
0
> 0< D
Out[19]= -
A
!!!
! Conjugate@AD Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2

m I4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
M
In[20]:=
"Reflection coefficient:";
R = FullSimplify@Abs@j
r
j
i
0
> 0< D
Out[21]= 1 -
4 ! H! + u0 L

4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
RectSqWell.nb 2
In[15]:= ji =
I
2 m
HA Exp@I k1 qD D@Simplify@
Conjugate@A Exp@I k
1
qDD, ! > 0 && u
0
> 0 && q < 0D, qD -
Simplify@Conjugate@A Exp@I k1 qDD, ! > 0 && u0 > 0 && q < 0D
D@A Exp@I k1 qD, qD L
Out[15]=
A
!!!
! Conjugate@AD

m
In[16]:= jt =
I
2 m
FullSimplify@y3 @qD
D@Simplify@Conjugate@y3 @qDD, ! > 0 && u0 > 0 && q > 0D, qD -
Simplify@Conjugate@y
3
@qDD, ! > 0 && u
0
> 0 && q > 0D
D@y3 @qD, qD, Assumptions 8! > 0, u0 > 0< D
Out[16]=
4 A !
32
Conjugate@AD H! + u
0
L

m I4 !
2
+ 4 ! u
0
+ Sin@2
!!!!!!!!!!!!!
! + u
0
D
2
u
0
2
M
In[17]:= "Transmission coefficient:";
T = FullSimplify@Abs@j
t
j
i
0
> 0<D
Out[18]=
4 ! H! + u0 L

4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
In[19]:= jr =
I
2 m
FullSimplify@
c12 Exp@-I k
1
qD D@Simplify@Conjugate@c12 Exp@-I k
1
qDD,
! > 0 && u0 > 0 && q < 0D, qD -
Simplify@Conjugate@c12 Exp@-I k1 qDD, ! > 0 && u0 > 0 && q <
D@c12 Exp@-I k
1
qD, qD, Assumptions 8! > 0, u
0
> 0< D
Out[19]= -
A
!!!
! Conjugate@AD Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2

m I4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
M
In[20]:=
"Reflection coefficient:";
R = FullSimplify@Abs@j
r
j
i
0
> 0< D
Out[21]= 1 -
4 ! H! + u0 L

4 !
2
+ 4 ! u0 + Sin@2
!!!!!!!!!!!!!
! + u0 D
2
u
0
2
RectSqWell.nb 2
5 10 15 20 25 30

0.2
0.4
0.6
0.8
1
T,R
R
T
Figure 1.15: Transmission (T) and reection (R)
coecient vs. energy, , for u
o
= 5/3 (cf. Fig.
1.8).
1.5. ANGULAR MOMENTUM ALGEBRA 27
1.5 Angular momentum alge-
bra
Raising and lowering operators:
In order to better understand what comes next,
we introduce the key concept via a simple example.
We briey return to the harmonic oscillator in one
dimension. Its Hamilton operator is given by
H =
p
2
2m
+
1
2
m
2
x
2
. (1.129)
Using the denitions
a :=
_
m
2h
x +i
p
2mh
a
+
:=
_
m
2h
x i
p
2mh
it is easy to show that
H = h
_
a
+
a +
1
2
_
(homework problem!). Writing the stationary
Schrodinger equation in the symbolic form used
previously (cf. Eq. (1.32)) we have
H [ n) = E
n
[ n) = h
_
n +
1
2
_
[ n) . (1.130)
In addition we we may work out the commutator
[a
+
, a] (this again is a homework problem). The
result is
_
a
+
, a
= 1 . (1.131)
Eqs. (1.130) and (1.131) allow an entirely new
way of looking at the harmonic oscillator. This
becomes apparent if we work out the result of
H(a
+
[ n)):
H
_
a
+
[ n)
_
= ha
+
aa
+
[ n) +
1
2
ha
+
[ n)
(1.131)
= ha
+
_
a
+
a + 1
_
[ n) +
1
2
ha
+
[ n)
= ha
+
_
a
+
a +
1
2
_
[ n)
. .
(1.130)
= (n+
1
2
)]n)
+ha
+
[ n)
= h
_
n +
3
2
_
_
a
+
[ n)
_
.
Thus
a
+
[ n) = c [ n + 1) .
The constant c we obtain via
n [ aa
+
[ n)
. .
=n]a
+
a+1]n)
=[ c [
2
n + 1 [ n + 1)
. .
=1(Normalization!)
= (n + 1)n [ n)
= n + 1 ,
i.e. c =

n + 1. Notice that we have made use of
the algebra underlying Eqs. (1.42) to (1.44). The
nal result therefore is
a
+
[ n) =
n + 1 [ n + 1) . (1.132)
Analogously we derive
a [ n) =
n [ n 1) (1.133)
(homework problem). Because of the above equa-
tions a
+
and a are called raising and lowering oper-
ator, respectively (dt.: Leiteroperatoren). Because
of a
+
a [ n) = n [ n) the operator a
+
a is also called
number operator.
We may even obtain the explicit solutions
n
(x)
of the harmonic oscillator in Eq. (1.63). Noting
that
a [ 0) = 0
we write
_
_
m
2 h
x +
_
h
2m
x
_
0
(x) = 0 (1.134)
which is the same equation with p = ih
x
, where
[ 0) is replaced by the explicit wave function
0
(x)
of the ground state. Thus
0
(x) =
m
h
x
0
(x) .
The solution is
0
(x) = C exp
_
1
2
m
h
x
2
_
,
which agrees with (1.63) for n = 0. The constant
C follows from
_
dx [
0
[
2
= 1 of course. The
other solutions,
n
(x) (n = 1, 2 . . .), can be worked
out by applying the raising operator (a
+
)
n
to
0
(x) analogous to Eq. (1.134).
Application to angular momentum:
Prior to the solution of the hydrogen atom via
Schrodingers equation we only dealt with problems
characterized by a single quantum number n associ-
ated with an energy level. Now there are two more
quantum numbers. These are l and m associated
with the eigenvalues of

L
2
(cf. Eq. (1.96)) and L
z
(cf. Eq. (1.97)). We will learn in section 2.2 that
only the expectation values of commuting operators
can be measured both exactly and simultaneously.
The signicance of this statement is that the set
of quantum numbers of all (relevant) commuting
operators characterize the state of the system! In
the case of hydrogen these operators are H,
L
2
, and
L
z
. Because we have dealt with H already, we now
must deal with

L
2
and L
z
.
In the following our goal is the solution of the
eigenvalue problems (1.96) and (1.97). And we use
a generalization of the above notation:
L
2
[ n
1
, n
2
) =
L
2 [ n
1
, n
2
) (1.135)
and
L
z
[ n
1
, n
2
) =
L
z
[ n
1
, n
2
) . (1.136)
Why we do not yet use the seemingly obvious no-
tation [ l, m) instead of [ n
1
, n
2
) will become clear
shortly.
First we point out that it is easy to conrm the
following commutator relations involving the com-
ponents of the angular momentum operator:
_
L
i
, r
j
= ih
ijk
r
k
, (1.137)
_
L
i
, p
j
_
= ih
ijk
p
k
, (1.138)
_
L, r
2
_
=
_
L, p
2
_
(1.139)
=
_
L, r p
_
= 0
_
L
i
, L
j
= ih
ijk
L
k
, (1.140)
44
and
_
L
2
, L
i
_
= 0 (1.141)
for i = 1, 2, 3. Here
123
=
231
=
312
= 1,
132
=
213
=
321
= 1 and
ijk
= 0 for all other ijk.
We pursue our original goal by dening raising
and lowering operators via their commutators
_
a
i
, a
+
j
=
ij
(1.142)
_
a
i
, a
j
=
_
a
+
i
, a
+
j
= 0 (1.143)
with i = 1, 2. In accordance with the preced-
ing example we have here
_
a
+
1
_
n
1
_
a
+
2
_
n
2
[ 0, 0) =
n
1
!n
2
! [ n
1
, n
2
). It is somewhat laborious but
otherwise easy to show that the right hand sides of
L
x
:=
h
2
_
a
+
1
a
2
+a
+
2
a
1
_
(1.144)
L
y
:= i
h
2
_
a
+
2
a
1
a
+
1
a
2
_
(1.145)
and
L
z
:=
h
2
_
a
+
1
a
1
a
+
2
a
2
_
(1.146)
are consistent with the above relations (1.140) and
(1.141). In this sense the right hand sides of the
above denitions are truly equivalent to the angular
momentum components. The fact that the angular
momentum components can be expressed in terms
of raising and lowering operators is not a mere com-
plication but can be applied quite eciently to our
original purpose!
44
Again there is a strong resemblance to the Poisson
brackets in classical mechanics.
1.5. ANGULAR MOMENTUM ALGEBRA 29
Using the commutators (1.142) and (1.143) we
obtain
L
2
=
h
2
2
(N
1
+N
2
) (1.147)
+
h
2
4
_
N
2
1
+N
2
2
+ 2N
1
N
2
_
,
where N
1
and N
2
are the number operators
N
i
= a
+
i
a
i
(i = 1, 2) (1.148)
obeying [N
1
, N
2
] = 0. We therefore immediately
obtain
L
2
[ n
1
, n
2
) = h
2
_
n
1
+n
2
2
(1.149)
+
(n
1
+n
2
)
2
4
_
[ n
1
, n
2
)
and
L
z
[ n
1
, n
2
) =
h
2
(N
1
N
2
) [ n
1
, n
2
)
= h
n
1
n
2
2
[ n
1
, n
2
) . (1.150)
Using the substitutions
l =
n
1
+n
2
2
(1.151)
and
m =
n
1
n
2
2
(1.152)
we nd
L
2
[ l +m, l m) = h
2
l (l + 1) [ l +m, l m)
and
L
z
[ l +m, l m) = hm [ l +m, l m) .
With the natural replacement [ l + m, l m) [
l, m) we nally write
L
2
[ l, m) = h
2
l (l + 1) [ l, m) (1.153)
and
L
z
[ l, m) = hm [ l, m) , (1.154)
which partially fullls our original goal. But there
is more information here. From
l +m = n
1
0 and l m = n
2
0
follows
m = l, l + 1, . . . , l 1, l , (1.155)
which is also in agreement with our previous as-
sumption. On the other hand we nd from (1.151)
that
l = 0,
1
2
, 1,
3
2
, . . . (1.156)
Thus, in addition to the expected integer values
we also get half integer values, which requires an
explanation.
In the appendix we had shown L
z
= ih in
polar coordinates. In particular we may write Eq.
(1.154) in a concrete representation
45
, i.e. for m =
l we may write
L
z
Y
l,l
(, ) = hlY
l,l
(, ) ,
and thus
Y
l,l
(, ) e
il
, (1.157)
where we pretend to not know that Y
lm
are spher-
ical harmonics. Relation (1.157) yields Y
lm
(, 0) =
Y
lm
(, 2) only if l = 0, 1, 2, . . .; but this is exactly
what we expect in the case of angular momentum!
In order to nd the explicit form of Y
lm
(, ) we
need to rst calculate a
+
1
a
2
[ n
1
, n
2
) and a
+
2
a
1
[
n
1
, n
2
). Here we go!
45
The exact meaning of representation will be explained
in the next chapter.
a
+
1
a
2
[ n
1
, n
2
)
= a
+
1
a
2
_
a
+
1
_
n
1
_
a
+
2
_
n
2
[ 0, 0)
1
n
1
!n
2
!
=
_
a
+
1
_
n
1
+1
a
2
_
a
+
2
_
n
2
[ 0, 0)
1
n
1
!n
2
!
=
_
a
+
1
_
n
1
+1
a
2
a
+
2
. .
=a
+
2
a
2
+1
_
a
+
2
_
n
2
1
[ 0, 0)
1
n
1
!n
2
!
=
_
a
+
2
a
2
+ 1
_
[ n
1
+ 1, n
2
1)
_
n
1
+ 1
n
2
=
_
(n
1
+ 1) n
2
[ n
1
+ 1, n
2
1) .
With n
1
= l +m and n
2
= l m follows
a
+
1
a
2
[ l +m, l m)
=
_
l (l + 1) m(m+ 1)
[ l + (m+ 1) , l (m+ 1)) .
Using the same replacements as above we nally
have
a
+
1
a
2
[ l, m) (1.158)
=
_
l (l + 1) m(m+ 1)
[ l, m+ 1) .
The operator a
+
1
a
2
acts as raising operator on m!
Analogously we can show
a
+
2
a
1
[ l, m) (1.159)
=
_
l (l + 1) m(m1)
[ l, m1) .
Therefore the operator a
+
2
a
1
acts as lowering oper-
ator on m! Now we put Eqs. (1.158) and (1.159)
to work. We start considering
a
+
2
a
1
Y
l,l
(, ) = 0 . (1.160)
This is sensible because l is the lowest possible
value for m. Using Eqs. (1.144) and (1.145) we
obtain
0 =
1
h
_
L
x
iL
y
_
Y
l,l
(, )
appendix
= e
i
(i cot
) Y
l,l
(, ) .
Inserting Y
l,l
(, ) = f () e
il
(cf. Eq. (1.157))
we get
0 = e
i(l+1)
(l cot
) f ()
and therefore
f () sin
l
or
Y
l,l
(, ) sin
l
e
il
.
Because we use the normalization
_
Y
lm
Y
lm
d = 1 ,
we nally obtain
Y
l,l
(, ) =
1
l!2
l
_
(2l + 1)!
4
sin
l
e
il
.
At this point we can apply Eq. (1.158), i.e.
a
+
1
a
2
Y
l,l
(, )
=
1
h
_
L
x
+iL
y
_
Y
l,l
(, )
appendix
= e
i
(i cot
) Y
l,l
(, )
=
2lY
l,l+1
(, ) (1.161)
or
46
Y
lm
(, ) =
_
(2l)!
(l +m)!
(l m)!
_
1
2
(1.162)
_
e
i
(i cot
l+m
Y
l,l
(, ) .
Thus, combining the results (1.153), (1.154), and
(1.162) we have derived Eqs. (1.96) and (1.97) as-
sumed previously!
46
Using (1.158) to obtain the correct normalization.
Chapter 2
Formal quantum mechanics
The above examples have inspired some con-
dence in the newly developed quantum mechanics.
Now it is time to explore the formal aspects of this
theory.
We begin by introducing the notation
(r, t) r [ (t)) (2.1)
for the wave function. If we Fourier-transform Eq.
(2.1), we obtain the wave function in

k-space, i.e.
(r, t)
F. T.
k, t)
k [ (t)) . (2.2)
We recognize that instead of (r, t) or

(
k, t),
which describe the wave function in r- or

k-
representation, we may view [ (t)) as an abstract
or non-specic representation of the wave function.
In order to tie (r, t) to

(
k, t) using the new no-

tation we introduce the complete sets
1 =
_
d
3
r [ r)r [ (2.3)
and
1 =
_
d
3
k [
k)
k [ . (2.4)
The meaning of (2.3) and (2.4) becomes clear if we
multiply both sides of (2.4) by r [ from the left
and by [ (t)) from the right. The result is
(r, t) = r [ (t)) (2.5)
=
_
d
3
kr [
k)
k [ (t))
. .
=

(
k,t)
,
and we obtain
r [
k) =
1
(2)
3/2
e
i
kr
. (2.6)
Similarly we have
k [ r) =
1
(2)
3/2
e
i
kr
, (2.7)
i.e. r [
k) is the complex conjugate of
k [ r). Of
course, what we have so far is not limited to Fourier
transformation. It applies to every complete set of
functions, which can be used to expand the wave
function.
Note also that using
p = h
k (2.8)
the relations (2.6) and (2.7) may be written as
r [ p) =
1
(2h)
3/2
e
i pr/ h
(2.9)
and
p [ r) =
1
(2h)
3/2
e
i pr/ h
. (2.10)
Hilbert space:
We have already discussed the necessity and the
physical meaning of the of existence of
31
32 CHAPTER 2. FORMAL QUANTUM MECHANICS
[ ) =
_
a.s.
[ [
2
dV = nite (2.11)
for the wave function (a.s.: all space). The set
of all functions for which the above integral ex-
ists span the Hilbert space H. In addition to (2.11)
there are other properties the wave functions must
have:
1. Linearity:
1
,
2
H c
1
1
+c
2
2
H for
c
1
, c
2
complex numbers.
2. Scalar product: The relation
1
[
2
) =
_
a.s.
2
dV denes a scalar product with the
following properties. (i)
1
[
2
)=
2
[
1
)
,
(ii) [ c
1
1
+ c
2
2
)= c
1
[
1
) +c
2
[
2
),
and [ ) 0. (i) through (iii) yield the
Schwarz inequality:
[
1
[
2
) [
_
1
[
1
)
2
[
2
) . (2.12)
Hermitian Operators:
We interpret the quantity [ (r, t) [
2
d
3
r as the
probability of nding the particle in a volume ele-
ment d
3
r at position r. Therefore it is sensible to
require
t
_
a.s.
(r, t) (r, t) d
3
r = 0 , (2.13)
i.e. probability conservation. Using
t
=
1
i h
H
and (
t
)
=
1
i h
(H)
respectively, we do obtain
1
ih
_
a.s.
_
(H) (H)
_
d
3
r = 0 (2.14)
or
_
a.s.
d
3
r
(H) =
_
a.s.
d
3
r (H)
. (2.15)
Every operator obeying Eq. (2.15) for arbitrary
(r, t) is called hermitian. Notice in particular that
a hermitian operator has real eigenvalues.
Eq. (2.15) may be generalized to
_
a.s.
d
3
r
1
(r, t)
(
2
(r, t))
=
_
a.s.
d
3
r (
1
(r, t))
2
(r, t) . (2.16)
In order to prove this, we insert (r, t) =
1
(r, t)+
2
(r, t) into Eq. (2.15). The argument of the inte-
gral on the left side becomes
() = (
1
+
2
)
((
1
+
2
))
=
1
(
1
) +
2
(
2
)
+
1
(
2
) +
2
(
1
) ,
where we assume that is a linear operator
1
.
Analogously we obtain on the right side
()
= (
1
)
1
+ (
2
)
2
+(
1
)
2
+ (
2
)
1
.
Combining these two equations and using the de-
nition (2.15) for hermitian operators we obtain
_
a.s.
d
3
r
_
1
(
2
) +
2
(
1
)
_
=
_
a.s.
d
3
r
_
(
1
)
2
+ (
2
)
1
_
or
_
a.s.
d
3
r
_
1
(
2
) (
2
)
1
_
. .
=2Im
_
1
(
2
)
_
=
_
a.s.
d
3
r
_
(
1
)
2
(
1
)
_
. .
=2Im
_
(
1
)
2
_
.
Therefore
Im
_
a.s.
d
3
r
1
(
2
)
= Im
_
a.s.
d
3
r (
1
)
2
. (2.17)
1
That is (c
1
1
+ c
2
2
) = c
1
1
+c
2
2
, where c
1
, c
2
are constants.
33
Now we repeat this calculation using a dierent
(r, t), i.e. (r, t) =
1
(r, t) + i
2
(r, t). The
result is
Re
_
a.s.
d
3
r
1
(
2
)
= Re
_
a.s.
d
3
r (
1
)
2
. (2.18)
Eqs. (2.17) and (2.18) together show that Eq.
(2.16) is indeed correct.
Let us consider an example, i.e. we want to show
that p
x
= ih
x
is hermitian. We write
_
a.s.
d
3
r
_
ih

x
_
p. i.
= ih
dydz
+ih
_
a.s.
d
3
r
_

x
=
_
a.s.
d
3
r
_
ih

x
.
Here p. i. stands for partial integration. The
[
. . .term vanishes, because of the normalizabil-

ity of the wave function, and we nd that p
x
obeys
Eq. (2.15). Notice that ip
x
on the other hand does
not.
Finally we want to translate the condition for
hermiticity to our new bra-ket notation. Using
(r, t) = r [ (t)) and
(r, t) = (t) [ r),

i.e. Eq. (2.5), we may write the left hand side of
Eq. (2.16) as
_
a.s.
d
3
r
1
(r, t)
(
2
(r, t))
=
_
a.s.
d
3
r
1
(t) [ r)r [
2
(t)) ,
where [
2
(t)) [
2
(t)). Inserting 1, i.e. Eq.
(2.3), between and [
2
(t)) we obtain
_
a.s.
d
3
r
1
(r, t)
(
2
(r, t))
=
_
a. s.
d
3
rd
3
r
t
1
(t) [ r)r [ [ r
t
)r
t
[
2
(t))
=
1
(t) [ [
2
(t)) .
Now we consider the right hand side of Eq. (2.16),
i.e.
_
a.s.
d
3
r (
1
(r, t))
2
(r, t)
=
_
a. s.
d
3
r
1
(t) [ r)r [
2
(t))
=
_
a. s.
d
3
rd
3
r
t
(r [ [ r
t
)
r
t
[
1
)
) r [
2
(t)) ,
where
1
(t) [ r) r [
1
(t))
= r [ [
1
(t))
.
Therefore
_
a.s.
d
3
r (
1
(r, t))
2
(r, t)
=
_
a. s.
d
3
rd
3
r
t
2
(t) [ r)
r [ [

r
t
)
r
t
[
1
(t))
=
2
(t) [ [
1
(t))
(1.45)
=
1
(t) [
+
[
2
(t)) ,
and Eq. (2.16) becomes
1
[ [
2
(t)) =
1
(t) [
+
[
2
(t)) . (2.19)
Because Eq. (2.19) must hold for arbitrary states
[
1
(t)) and [
2
(t)) we conclude from
1
(t) [
+
[
2
(t)) = 0
that
=
+
(2.20)
for hermitian operators.
Expansion of states in eigenstates:
We return to the general eigenvalue equation
[
n
) =
n
[
n
) . (2.21)
Here the
1
,
2
, . . . form a discrete eigenvalue spec-
trum of the hermitian operator . The [
n
) are
the so called eigenstates of .
1. As we have shown already, the eigenvalues are
real.
2. Two eigenfunctions belonging to two dierent
eigenvalues are orthogonal
2
and thus linearly
independent.
3. If is n-fold degenerate, it is still possible to
construct n eigenfunctions to which are or-
thonormal (e.g., via the procedure according to
Schmidt). Therefore each eigenvalue is associ-
ated with a series of orthonormal eigenfunc-
tions which consists of either one element (no
degeneracy), a nite number of elements (nite
degeneracy) or an innite number of elements
(innite degeneracy).
4. We postulate that the set of eigenstates [
n
)
of any hermitian operator that represents a
physical quantity form a complete set. Using
the notation introduced in (2.3) and (2.4) this
can be expressed as
[ ) =
n
[
n
)
n
[ )
=
n
c
n
[
n
) . (2.22)
Therefore
P
n
[
n
)
n
[ (2.23)
can be view as a projection operator, which,
if it is applied to [ ), singles out a certain
component of the state vector:
P
n
[ ) = c
n
[
n
) (2.24)
A simple analogy may serve as illustration.
Consider a vector a expressed in terms of its
components via
a =
3
n=1
e
n
(e
n
a) ,
where the e
n
are unit vectors along the axes of
the coordinate system. Here e
n
(e
n
) is analo-
gous to P
n
.
2
This we prove via
2
[ [
1
)
1
[
+
[
2
)
2
[ [
1
)
1
[ [
2
)
= (
1
2
)
2
[
1
) = 0.
Intuitively all of this makes perfect sense, be-
cause we are used to expressing even compli-
cated boundary conditions, which appear in
the context of eigenvalue dierential equations,
in terms of linear combinations of the eigen-
functions. Completeness of the eigenfunctions
can be proved in 1D and in certain multidi-
mensional cases, but must be postulated for
most multidimensional systems. For details
see [3, 19, 20].
Much of this is very formal, and the usefulness
of the new notation may not be apparent at this
stage. But it will be!
2.1 Operator averages
We have noted that the quantity [ (r) [
2
d
3
r cor-
responds to the probability of nding a particle in-
side the volume element d
3
r. In this spirit we dene
the following average of the position operator
r) =
_
a.s.
d
3
r
(r, t) r (r, t) . (2.25)

Using the new notation this reads
r) = (t) [ r [ (t)) . (2.26)
Here we use r to remind us that r really is an op-
erator. To show the connection between Eq. (2.26)
and Eq. (2.25) we again insert 1, i.e. (2.3), on both
sides of r in Eq. (2.26):
r) =
_
d
3
rd
3
r
t
(t) [ r) (2.27)
r [ r [

r
t
)
r
t
[ (t)) .
Comparison with Eq. (2.25) implies
r [ r [

r
t
) = rr [

r
t
) (2.28)
= r(r
r
t
) . (2.29)
The last equality we prove via
2.1. OPERATOR AVERAGES 35
r [

r
t
)
(2.4)
=
_
d
3
kr [
k)
k [

r
t
)
(2.6,2.7)
=
1
(2)
3
_
d
3
ke
i
k(r
)
= (r
r
t
) .
Next we want to look at the time derivative of
r), i.e.
d
dt
r) =

(t) [ r [ (t)) + (t) [ r [

(t)) .
Using Schrodingers equation in representation free
notation,
ih [

(t)) = H [ (t))
ih

(t) [ = (t) [ H ,
we may write
d
dt
r) =
i
h
(t) [ [H, r] [ (t)) . (2.30)
Inserting H =
p
2
2m
+| (r) we obtain (using (1.56)).
d
dt
r) =
i
2mh
(t) [
_
p
2
, r
[ (t))
=
1
m
p) (2.31)
3
. Here we have used the commutator (1.58) and
[AB, C] = A[B, C] + [A, C] B , (2.32)
which we prove by expanding both sides of the
equation. Notice the analogy between (2.31) and
the classical relation.
To make this analogy more apparent we consider
3
_
p
2
, r
=
_
[p
2
x
+ p
2
y
+ p
2
z
, x]
[. . . , y]
[. . . , z]
_
=
_
_
[p
x
2
, x]
[p
y
2
, y]
[p
z
2
, z]
_
_
=
2(ih)
_
p
x
p
y
p
z
_
= 2(ih) p
d
2
dt
2
r) =
1
m
d
dt
p)
=
i
mh
(t) [
_
H, p
[ (t))
=
i
mh
(t) [
_
|, p
[ (t)) . (2.33)
The last equality results from relation (Eq. ( 1.57)).
In a homework problem we will prove that for ev-
ery analytic function F (x) we have [A, F (B)] =
[A, B] F
t
(B), where F
t
(x) denotes the derivative
of F (x)
4
. Assuming that | (r) is analytic we nd
_
|, p
=
_
_
[|, p
x
]
[|, p
y
]
[|, p
z
]
_
_
=
_
_
_
[x, p
x
]
d/
dx
[y, p
y
]
d/
dy
[z, p
z
]
d/
dz
_
_
_ = ih
|
and thus
m
d
2
dt
2
r) = (t) [
| [ (t))
=
|) . (2.34)
i.e. we obtain Newtons second law for the quantum
mechanical averages. Relations (2.31) and (2.34)
are known as Ehrenfests theorem
5
.
We note that in Eq. (2.30) instead of r we could
have inserted other operators. Replacing r by (t)
we obtain
d
dt
) =
i
h
(t) [ [H, ] [ (t)) +
(t)
t
) .(2.35)
For a which does not depend on time explicitly,
we nd that
d
dt
) = 0 if commutes with H. Be-
cause H commutes with itself, we nd that in a
closed quantum mechanical system energy is con-
served. From Eqs. (2.33) and (2.34) we also nd
d
dt
p) = 0 in the absence of external forces. Thus we
obtain momentum conservation. Finally, for cen-
trosymmetric force systems we have [H, L
2
] = 0
and [H, L
z
] = 0, which is the conservation of angu-
lar momentum.
4
provided that the operators A and B commute with
their commutator, i.e. [B, [A, B]] = [A, [A, B]] = 0.
5
Ehrenfest, Paul, osterr. Physiker, *Wien 18.1.1880,
(Selbstmord) Amsterdam 25.9.1933; ab 1912 Prof. in Lei-
den; arbeitete uber Atomphysik, Quantentheorie, statist.
Mechanik; stellte u.a. die Quasiergodenhypothese (Ergo-
denhypothese) auf. c _2000 Bibliographisches Institut & F.
A. Brockhaus AG
2.2 The uncertainty principle
The relations (2.31) and (2.34) may be pleasing in
the sense that they relate quantum and classical
mechanics. But now we compute the root mean
square of operators, and the result will be entirely
new and unique to quantum theory.
We are going to work out the right side of the
inequality
0 (t) [ (AiB)
+
(AiB) [ (t))
Here A = AA) and B = BB). We assume
that A and B are hermitian operators, i.e. A =
A
+
and B = B
+
. nally is a real parameter.
Expanding the right side of the above inequality
yields
0 (t) [
_
2
A
2
+B
2
i [A, B]
. .
=C
_
[ (t))
= A
2
)
_
2
+
C)
A
2
)
_
+B
2
)
= A
2
)
_
+
1
2
C)
A
2
)
_
2
1
4
C)
2
A
2
)
+B
2
) .
Because is a parameter, we may chose such
that (. . .)
2
= 0 (provided C) is real), i.e.
A
2
)B
2
)
1
4
C)
2
or
AB
1
2
[ C) [ , (2.36)
where A =
_
A
2
) and B =
_
B
2
).
In order to understand the meaning of the in-
equality (2.36) we consider the two operators we
know best - r and p. They do not commute and
thus
xp
x

1
2
h . (2.37)
In the units, which we used to discuss the wave
packet idea (cf. Eq. (1.48)), this reads
xk
1
2
. (2.38)
A simple example may serve to illustrate this in-
equality. First we note that for any operator A we
may write
A
2
= (AA))
2
)
= A
2
2AA) +A)
2
)
= A
2
) A)
2
. (2.39)
In the case of x we have
x
2
= x
2
) x)
2
,
where
x) =
_

(x, t)
x(x, t)dx (2.40)

and
x
2
) =
_

(x, t)
x
2
(x, t)dx . (2.41)
In section 1.3 (cf. Eq. (1.50)) we had constructed
a wave package for which
(x, t = 0) = c exp[x
2
/(4b
2
)] ,
where c is a constant. Inserting this expression into
the above integrals yields
x)
t=0
= c
c
_

xexp[x
2
/(2b
2
)]dx
and
x
2
)
t=0
= c
c
_

x
2
exp[x
2
/(2b
2
)]dx .
Because the corresponding argument under the in-
tegral is an odd function we have
2.2. THE UNCERTAINTY PRINCIPLE 37
x)
t=0
= 0 .
In the case of x
2
)
t=0
we express c
c in terms of the
normalization, i.e.
x
2
)
t=0
=
_
x
2
exp[x
2
/(2b
2
)]dx
_
exp[x
2
/(2b
2
)]dx
=
d
d
_
1
2b
2
_ ln
__

exp[
x
2
2b
2
]dx
_
z=
_
1
2b
2
x
=
d
d
_
1
2b
2
_ ln
_
_
1
2b
2
_

e
z
2
dz
_
= b
2
.
Therefore we have
x
t=0
= b . (2.42)
Now we must compute p
x
, i.e.
p
x
2
= p
x
2
) p
x
)
2
.
This time we have
p
x
)
t=0
= ihc
c
_

exp[
x
2
4b
2
]
d
dx
exp[
x
2
4b
2
]dx
and
p
x
2
)
t=0
= h
2
c
c
_

exp[
x
2
4b
2
]
d
2
dx
2
exp[
x
2
4b
2
]dx .
The argument under the rst integral is again odd
and therefore
p
x
)
t=0
= 0 .
The second integral is simplied by applying partial
integration once which yields
p
x
2
)
t=0
= h
2
c
c
_

_
d
dx
exp[
x
2
4b
2
]
_
2
dx
=
h
2
(2b
2
)
2
c
c
_

x
2
exp[
x
2
2b
2
]dx
=
h
2
(2b
2
)
2
x
2
)
t=0
=
h
2
4b
2
.
And thus
p
xt=0
=
h
2b
. (2.43)
Combining Eqs. (2.42) and (2.43) we nd for the
special case of our wave packet (at t = 0) in section
1.3
xp
x
t=0
=
h
2
in agreement with (2.37).
If we localize a particle by reducing its x then
we must allow for a larger p
x
. Position and mo-
mentum along the same coordinate direction can-
not both be dened or measured with arbitrary pre-
cision! This was rst noted by W. Heisenberg 1928
and is called uncertainty principle. The inequal-
ity (2.36) is a general expression of the uncertainty
principle. If, on the other hand, the state of a phys-
ical system like the hydrogen atom is determined by
a complete set of quantum numbers, e.g. n, l, and
m, then the associated operators must commute.
In the case of hydrogen these operators are H, L
2
and L
z
.
Remark: We may wonder whether in analogy to
xp
x

h
2
we may write
tE
h
2
. (2.44)
(cf. (1.54)). However, contrary to the spatial co-
ordinates time is merely a parameter in quantum
mechanics. It is therefore not straightforward to
give an interpretation of relation (2.44). Below we
will address this question in detail.
2.3 From wave mechanics to
matrix mechanics
Looking at the denition (2.26) we may wonder
whether the expression on the right is meaningful if
we chose dierent state functions, i.e. if we dene
mn

m
[ [
n
) . (2.45)
Here is an arbitrary hermitian operator.
Let us try to evaluate (2.45) for the 1D harmonic
oscillator with = q, i.e.
q
mn

m
[ q [
n
) (2.46)
=
_

m
[ q)q [ q [ q
t
)q
t
[
n
)dqdq
t
.
Using again (2.29) we obtain
q
mn
=
_

m
(q) q
n
(q) dq . (2.47)
With the aid of (1.64) and (1.65) it is easy to show
that
qH
n
(q) =
1
2
H
n+1
(q) +nH
n1
(q) . (2.48)
Now we employ the useful theorem for hermi-
tian operators introduced above: Two eigenfunc-
tions belonging to two dierent eigenvalues are or-
thogonal. Therefore
q
mn
=
1
_
_
n + 1
2
1
(n + 1)!2
n+1
dqe
q
H
2
n+1
(q)
m,n+1
+
_
n
2
1
(n 1)!2
n1
dqe
q
H
2
n1
(q)
m,n1
_
=
_
n + 1
2

m,n+1
+
_
n
2
m,n1
(2.49)
(note that m, n 0) or if we write q in matrixform,
q =
1
2
_
_
_
_
_
0 1 0 0
1 0

2 0
0

2 0

3
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
. (2.50)
We note that (2.50) agrees exactly with Eq. (1.36)
if we take
_
h/(m) as the unit of length, i.e.
we have discovered the bridge between matrix
and wave mechanics! The recipe is easy. Solve
Schrodingers equation for a given potential func-
tion and calculate (2.45) for = r and = p.
Because in our example we have used the eigen-
functions of the Hamiltonian to evaluate the matrix
elements in (2.45) the right side of this equation is
called the energy representation of . Thus (2.50)
is the energy representation of q.
Let us consider a second example dealing with a
dierent representation of the position operator r.
Eq. (2.25) we now call r-space representation of r.
Using Eq. (2.50) in Eq. (2.25) we can easily work
out the p-representation of r:
r(r
r
t
) = r [ r [

r
t
)
=
_
d
3
kd
3
k
t
r [
k)
k [ r [

k
t
)
k
t
[

r
t
)
()
=
1
(2)
3
_
d
3
kd
3
k
t
e
i
kr
k [ r [

k
t
)e
i
In order to satisfy () we must have
k [ r [

k
t
) = i(
k

k
t
)
k
(2.51)
or
p [ r [

p
t
) = ih( p
p
t
)
p
. (2.52)
In p-space the position operator r therefore is given
by ih
p
! Analogously we can show that the p-
representation of p is
p [ p [

p
t
) = p( p
p
t
) . (2.53)
Thus in summary:
2.4. THE TIME EVOLUTION OPERATOR 39
operator r-space p-space
r r ih
p
p ih
r
p
Secondly we remark that matrix elements as
represented by (2.45) also have a physical inter-
pretation. They describe transitions from a state
characterized by [
n
) to a state characterized by
[
m
) under the inuence of a physical pertur-
bation described in terms of . The right side
of (2.45) is sometimes called transition matrix
element.
2.4 The time evolution opera-
tor
A system may be in a state described by [ (t)) at
time t. We may express this formally as
[ (t)) = S (t) [ (0)) . (2.54)
To nd out what the operator S (t) stands for we
insert (2.54) into Schrodingers equation, i.e.
0 = (ih
t
H) [ (t))
= (ih
t
S (t) HS (t)) [ (0)) . (2.55)
Assuming that H is independent of t we obtain
S = e
i
h
1t
1
!
_
i
h
Ht
_
(2.56)
6
. Using H = H
+
we see that
6
Here me make use of the energy representation of S (t),
i.e.
m
[ S (t) [
n
). The [
n
) are Eigenkets of 1. By
multiplication of Schr odingers equation from the left with
m
[ and by inserting 1 =
n
[
n
)
n
[ we obtain
n
_
ih
t
m
[ S (t) [
n
)
m
[ 1 [
n
)
n
[ S (t) [
n
)
_
n
[ (0)) = 0 .
Using
m
[ 1 [
n
) = E
m
m
[
n
) = E
m
mn
this sim-
plies to
n
_
ih
t
m
[ S (t) [
n
)
S
+
S = SS
+
= 1 . (2.57)
Operators with this property are called unitary.
Thus far we always had assumed time-
independent operators and time-dependent state -
or wave functions. This is called the Schrodinger
picture. However, using S (t) we may transform
the Schrodinger picture into the Heisenberg pic-
ture, where the state functions do not depend on
time but the operators do, i.e.
[ )
H
= S
+
(t) [ (t))
S
(2.58)
and
H
(t) = S
+
(t)
S
S (t) , (2.59)
where the index H stands for Heisenberg and the
index S for Schrodinger. Both pictures are com-
pletely equivalent.
Both pictures may be used simultaneously. We
divide the Hamilton operator according to
H(t) = H
0
+H
1
(t) , (2.60)
where H
1
(t) is a small and time-dependent inter-
action term. In this case we write
[ (t))
I
= S
(0)+
[ (t))
S
(2.61)
and
I
(t) = S
(0)+
S
S
(0)
(2.62)
where
S
(0)
(t) = e
i
h
1
0
t
. (2.63)
This is the interaction picture (Index I). But why is
the interaction picture useful? To see this we insert
S = S
(0)
S
(1)
into (2.55) and multiply from the left
with S
(0)
+
:
ih
t
S
(1)
= S
(0)
+
_
HS
(0)
ih
t
S
(0)
_
S
(1)
. (2.64)
E
m
m
[ S (t) [
n
)
_
n
[ (0)) = 0 .
This equation holds for arbitrary [ (0)) and therefore
(. . .) = 0. The solution therefore is
m
[ S (t) [
n
) = exp [iE
m
t/h]
mn
.
Here
mn
results from S (0) = 1.
If (. . .) is close to zero, then Eq. (2.64) is more suit-
able for an approximation schema than the original
Eq. (2.55), because S
(1)
varies slowly with time. In
particular (2.64) can be written as
ih
t
S
(1)
= H
(1)
I
S
(1)
.
In section 4.2 this equation will be discussed in
detail.
Uncertainty principle revisited:
Consider a time-independent Hamiltonian and
assume that the spectrum is nondegenerate. The
state [ (t
0
)) may be written as
[ (t
0
)) =
n
[
n
(t
0
))
n
(t
0
) [ (t
0
)) . (2.65)
The [
n
(t
0
)) are eigenstates of the Hamiltonian
at time t
0
. For the following it is convenient to
consider a continuous spectrum so that the above
sum becomes an integral
[ (t
0
)) =
_
(E) [
E
(t
0
))dE , (2.66)
where
n
(t
0
) [ (t
0
)) = (E) E. The absolute
square of (E) is sketched in gure 2.1. This means
that the state [ (t
0
)) is composed mainly of energy
eigenstates [
E
(t
0
)) belonging to energies close to
E
o
. Application of Eq. (2.54) to [ (t
0
)) yields its
time evolution
[ (t)) =
_
(E) e
iE(tt
0
)/ h
[
E
(t
0
))dE .(2.67)
We now want to know how long it takes for [
(t
0
)) to change signicantly. In order to estimate
the time interval during which the system evolves to
an appreciable extent, we calculate the probability
of nding the system in a dierent state [ (t)) at
time t. This probability is dened via
P
(t) :=[ (t) [ (t)) [

2
. (2.68)
This interpretation more or less follows if we ex-
pand the normalized scalar product [ ) in
|(E)|
2
E E
o
E
Figure 2.1: [ (E) [
2
vs. E
eigenstates of some hermitian operator (cf. Eq.
(2.22)), i.e.
1 = [ )
=
[ )
=
[ ) [
2
.
Each Term [
[ ) [
2
is the contribution of the
respective projection to the total probability.
Inserting Eq. (2.67) into (2.68) we have
P
(t) =[ (t) [ (t)) [

2
=
_
(E) e
iE(tt
0
)/ h
(t) [
E
(t
0
))dE
2
.
Let us assume that the width of [ (E) [
2
is
small compared to the energy variation of [ (t) [
E
(t
0
)) [
2
. With this assumption we may approxi-
mate P
(t) as
P
(t)

= [ (t) [
E
o
(t
0
)) [
2
_
(E) e
iE(tt
0
)/ h
dE
2
.
For the sake of concreteness we may further assume
that the E-dependence of (E) is Lorentzian
7
, i.e.
7
This is not an essential assumption, e.g. we may use a
Gaussian instead.
2.5. THE DENSITY OPERATOR 41
(E)
(E/2)
2
(E E
o
)
2
+ (E/2)
2
. (2.69)
This yields
_

(E) e
iE(tt
0
)/ h
dE exp
_
E [ t t
0
[
2h
_
8
and therefore
P
(t) exp
_
E [ t t
0
[
h
_
, (2.70)
still under the assumption that the factor (t) [
E
o
(t
0
)) [
2
varies slowly by comparison. We may
therefore use the relation
tE h (2.71)
to dene the time interval t =[ t t
0
[ beyond
which the state [ (t
0
)) has changed signicantly,
i.e. t is a measure of the lifetime of [ (t
0
)).
The lifetime is short if E is large and long if E
is small.
2.5 The density operator
Again we consider an operator average, i.e.
) = (t) [ [ (t)) . (2.72)
We expand the state vector [ (t)) in orthonormal
eigenstates of , i.e.
[ (t)) =
(t))
(t) [ (t))
=
(t)) .
In the following we dene p
[c
[
2
, which is
the probability for nding the system described by
[ (t)) in the state [
(t)). Thus, ) may be

written as
8
_
dx
e
ipx
a
2
+x
2
=

a
exp[ [ ap []
) =
(t) [ [
(t))
=
,n
p
(t) [
n
(t))
n
(t) [ [
(t))
=
(t) [
n
[
n
(t))p
n
n
(t) [ [
(t))
=
(t) [ [
(t))
) = Tr
_
_
. (2.73)
The operator dened via

n
[
n
(t))p
n
n
(t) [ . (2.74)
is called density operator
9
. The operation Tr is
dened via
Tr (. . .)
(t) [ . . . [
(t)) ; (2.75)
it is called trace (dt.: Spur). Eq. (2.73) may be
generalized to an arbitrary function of , i.e.
F ()) = Tr
_
F ()
_
. (2.76)
An important property of the trace is that it is
invariant under cyclic permutation of operators:
Tr (ABC) = Tr (BCA) = Tr (CAB) . (2.77)
Proof:
Tr (ABC) =
,a,b,c
[ a)aa [ b)bb [ c)cc [ )
=
a,b,c
aa [ b)bb [ c)cc [ a)
=
,a,b,c
aa [ ) [ b)bb [ c)cc [ a)
=
,a,b,c
[ b)bb [ c)cc [ a)aa [ )
= Tr (BCA) . (2.78)
9
In this example the [
n
) are orthogonal but this is not
essential for the denition of the density operator.
Notice in particular that Tr () is invariant under
a unitary transformation of , i.e.
Tr () = Tr
_
SS
+
..
=1
_
= Tr
_
S
+
S
_
.
The signicance of this is that the trace remains
unaltered when the representation is changed. E.g.,
we may write
n
[ [
n
)
=
n,,
n
[
[ [
[
n
)
=
[ [
[
n
)
n
[
)
. .
=
[ [
)
Here the [
) are (normalized) eigenkets of ,

whereas for the [
n
) this may not be the case!
A system is considered to be in a pure state if
the we can write as
=[
pure
)
pure
[ . (2.79)
Otherwise it is a mixed state.
The density operator is very useful in the con-
text of many particle systems. A gas, for instance,
contained in a xed volume in thermal equilibrium
with its walls at temperature T, is described via
n
[ [
n
) =
1
Q
e
E
n
, (2.80)
where the [
n
) are eigenkets of H. Here Q is the
canonical partition function. We may obtain Q via
1 =
n
[ [
n
), i.e.
1 =
1
Q
n
e
E
n
=
1
Q
Tr
_
e
1
_
, (2.81)
and therefore
Q = Tr
_
e
1
_
=
n
[ e
1
[
n
) . (2.82)
The partition function can be used to derive all
quantities of interest for many particle systems,
provided of course that we can diagonalize the sys-
tems Hamilton operator (for details see Ref. [16]).
Notice that in this example we consider a system,
a gas conned in a container, which only is a small
portion of the universe. However, we do not need
the state function of the universe; the state function
of the gas is sucient for the construction of .
Finally we want to again consider the time
derivative
d
dt
) =
d
dt
(t) [ [
(t)) . (2.83)
Here the statistical weights p
are assumed to be
independent of time
10
. According to Eq. (2.35)
we may write
d
dt
) =
d
dt
(t) [ [
(t))
=
i
h
(t) [ [H, ] [
(t))
+
t
) .
Inserting 1 =
n
[
n
)
n
[ we obtain
d
dt
) =
,n
i
h
[
n
)p
n
n
[
_
H,
_
[
)
+
t
)
=
i
h
Tr
_
_
H,
__
+
t
)
=
i
h
Tr
__
, H
_
_
+
t
)
= Tr
_

_
+
t
) , (2.84)
where is given by
=
i
h
_
H,
n
[
n
)p
n
n
[
_
i
h
_
, H
. (2.85)
10
This refers to a condition called equilibrium later in Sta-
tistical Mechanics. In particular Eq. (2.80) holds in equilib-
rium.
2.6. PATH INTEGRATION AND THE DENSITY OPERATOR 43
Eq. (2.85) follows from
d
dt
(2.56)
=
n
d
dt
_
S [
n
(0))p
n
n
(0) [ S
+
_
=
i
h
H +
i
h
H .
If we assume that the [
n
(t)) in Eq. (2.85) are
stationary, i.e.
[
n
(t)) = e
i
h
E
n
t
[
n
(0)) , (2.86)
it follows that
= 0 . (2.87)
2.6 Path integration and the
density operator
In classical mechanics the path followed by a par-
ticle in space and time can be derived from a
least action principle, S = 0, where the action
is S =
_
t
2
t
1
L(q, q, t)dt, and L is the Lagrangian de-
pending on the coordinates q, velocities q, and time
t
11
. We may ask whether the action does play any
related role in quantum mechanics. It is this ques-
tion we want to address here.
Notice that the density operator , given by Eq.
(2.80), may be written as
() =
e
1
Tr (e
1
)
(2.88)
12
. However, in the following we want to use the
unnormalized dened by
u
() = e
1
. (2.89)
In order to avoid the additional index u we just
omit it from here on throughout the remainder of
this chapter unless noted otherwise.
First note that () obeys the dierential equa-
tion
11
Do not confuse the action in this section with the time
evolution operator.
12
Note: =
n,n
[
n
)
n
[ [
n
)
n
[ and
n
[ [
n
) = (Tr(exp[1]))
1
exp[E
n
]
nn
.
= H . (2.90)
We can see this by writing in energy representa-
tion as
nm
=
nm
e
E
n
(2.91)
(cf. page 39). Dierentiating this equation we have
nm
=
nm
E
n
e
E
n
= E
n
nm
,
which is the energy representation of Eq. (2.90).
As an example application of (2.90) we calculate
the r-representation of , (x, x
t
; ) = x [ [ x
t
),
for a 1D free particle with H =
p
2
2m
, i.e.
(x, x
t
; ) =
h
2
2m
2
x
(x, x
t
; ) . (2.92)
Eq. (2.92) is a 1D diusion equation which has the
solution
13
(x, x
t
; ) =
_
m
2h
2
exp
_
m
2h
2
(x x
t
)
2
_
(2.93)
for the initial condition (x, x
t
; 0) = (x x
t
). No-
tice that here assumes the role of time.
Now assume that we are interested in H)
14
,
which according to Eq. (2.73) is given by
H)
()
= Tr
_
H
_
(2.82)
=
lnQ (2.94)
(: Note that in this formula is the full not
u
!). If Q is evaluated in r-representation
15
instead
of in energy representation as in Eq. (2.82) we have
13
See any book on partial dierential equations in physics!
14
Clearly, we know that the energy of an ideal gas consist-
ing of N point-like particles is 1) =
3
2
Nk
B
T, where each of
the 3N degrees of freedom contributes
1
2
k
B
T. This is what
we want to show here!
15
Q =
n
_
dxdx
n
[ x)x [ exp[1] [ x
)x
[
n
) =
_
dxdx
x [ exp[1] [ x
)(x x
).
H) =
ln
_
(x, x; ) dx
=
_

(x, x; ) dx
_
(x, x; ) dx
=
1
2
k
B
T .
This means that the average thermal energy of a
1D free particle is
1
2
k
B
T as promised.
After this brief excursion into Statistical Me-
chanics we now turn to the idea of path integration.
At this point you should have noticed the resem-
blance of Eq. (2.90) to the Schrodinger equation,
if is replaced by
i
h
t. Thus, instead of (2.89) we
now consider
(t) = e
i
h
1t
(2.95)
16
, which can be broken up into n time slices ac-
cording to
(t) = e
i
h
1
e
i
h
1
. . . e
i
h
1
(2.96)
= () () . . . () (n factors) .
In x-representation this becomes
(x, x
t
; t) =
_
. . .
_
(x, x
n1
; ) (2.97)
(x
n1
, x
n2
; ) . . . (x
1
, x
t
; )
dx
1
. . . dx
n1
.
Figure 2.2 illustrates how to interpret this equation.
The 1D particle travels from x
t
through a series of
intermediate steps, x
1
, x
2
,. . . , x
n1
, which dene
a path. The total (x, x
t
; t) for the particle to be-
gin at x
t
and end up at x is given by a sum over
all possible paths, i.e. for all possible intermediate
positions x
i
. As the time increment approaches
zero, the number of integrations on the intermedi-
ate variables become innite and the last equation
can be written symbolically
(x, x
t
; t) =
_
[x(t
t
)] Tx(t
t
) , (2.98)
16
Notice that our (t) is nothing else but the time evolu-
tion operator S(t) introduced above.
x
x
x'
x
1
x
2
x
3
!
2!
3!
t
0
Figure 2.2: A particle travelling from x
t
to x along
a path x
1
, x
2
etc.
where
[x(t
t
)] = lim
0,nt
(x, x
n1
; ) (2.99)
(x
n1
, x
n2
; ) . . . (x
1
, x
t
; )
and
Tx(t
t
) = lim
n
dx
1
dx
2
. . . dx
n1
.
To better understand the meaning of Eq. (2.98)
we consider the 1D free particle again, i.e. we must
solve Eq. (2.92), where now is replaced by i
t
h
.
From (2.93) we see that
(x, x
t
; ) =
_
m
2ih
exp
_
m
2 hi
(x x
t
)
2
_
. (2.100)
According to (2.99) we may write
17
(x, x
t
; t) (2.101)
= lim
0
_
. . .
_
exp
_
m
2ih
_
_
x x
n1
_
2
+
_
x
n1
x
n2
_
2
+. . .
17
We show this explicitly in a homework problem.
2.6. PATH INTEGRATION AND THE DENSITY OPERATOR 45
. . . +
_
x
1
x
t
_
2 __
dx
1
_
2ih/m
. . .
dx
n1
_
2ih/m
.
Now we let approach zero:
x
k
x
k1

dx(t
t
)
dt
t
=k
x(t
t
)
=k
.
Therefore the integrand in (2.101) becomes
exp[. . .] = exp
_
i
h
_
t
0
m
2
x(t
t
)
2
dt
t
_
. (2.102)
Eq. (2.98) in the case of the free particle now reads
(x, x
t
; t) (2.103)
=
_
exp
_
i
h
_
t
0
m
2
x(t
t
)
2
dt
t
_
Tx(t
t
) ,
where Tx(t
t
) now is given by
Tx(t
t
) = lim
0
n1
i=1
dx
i
_
2hi/m
The argument of the exponent in Eq. (2.103) be-
gins to look interesting. The integral we identify as
the action (dt.: Wirkung) introduced in classical
mechanics S =
_
t
2
t
1
L(x, x, t) dt. To complete this
analogy we have to place our 1D free particle in a
potential | (x). The equation for (x, x
t
; t) now is
ih
t
(x, x
t
; t) =
h
2
2m
2
x
(x, x
t
; t)
+| (x) (x, x
t
; t) . (2.104)
It can be shown
18
that Eq. (2.102) becomes
exp[. . .] = exp
_
i
h
_
t
0
L(t, x(t
t
), x(t
t
)) dt
t
_
, (2.105)
where indeed
18
e.g., R. P. Feynman (1972) Statistical Mechanics. Ad-
dison Wesely or P. Ramond (1981) Field Theory - A mod-
ern primer. Benjamin/Cummings; both books are advanced
texts!
L(t, x(t
t
), x(t
t
)) =
m
2
x
2
(t
t
) | (x(t
t
)) . (2.106)
Earlier we had introduced the probability for
nding a system (or particle) in a state (t) [
at time t, when it was in a state [ (0)) at time
zero (cf. page 40). This probability was [ (t) [
(t)) [
2
, which we may express via
(t) [ (t))
2
(2.107)
=
(t) [ [ (0))
2
=
_
d
3
rd
3
r
t
(t) [ r)r [ [

r
t
)
r
t
[ (0))
2
=
_
d
3
rd
3
r
t
(r, t) (r,
r
t
; t)(
r
t
, 0)
2
.
The quantity inside [ . . . [
2
is called transition am-
plitude. Thus, for any two xed points r and

r
t
the
propagator (r,
r
t
; t) sums up all possible paths,
which the system does follow in getting from

r
t
to
r.
Whereas in classical mechanics there is only one
path dened via S = 0, there are innitely many
paths followed simultaneously by the quantum me-
chanical system. However, we would expect that
paths very close to the classical path are more im-
portant than those deviating more strongly from
this path. One may understand this as follows. A
system may be well described by the classical least
action path S = 0. For every deviation x from
the path the quantity S/h is large, and the cor-
responding phase angle = S/h varies strongly
with x. This results in cancellations among the
many possible phase factors exp[iS/h] - except for
the contributions from the classical path!
Clearly, Eq. (2.107) describes a key quantity in
quantum theory. Therefore it is highly desirable
to compute (r,
r
t
; t) or in the 1D case (x, x
t
; t).
However, the path integrals,
_
. . . Tx(t
t
), are di-
cult, and this is not the place to discuss the various
solution techniques. The interested reader is re-
ferred to R. P. Feynman and A. R. Hibbs (1965)
Quantum mechanics and path integrals. Mc Graw-
Hill or to H. Kleinert (1993) Pfadintegrale. B. Wis-
senschaftsverlag (again, this is advanced reading!)
19
.
19
The origin of the procedure discussed in this section
We close with a quote from Feynmans book: In
this book we shall give the laws to compute the
probability amplitude for nonrelativistic problems
in a manner which is somewhat unconventional. In
some ways, particularly in developing a conceptual
understanding of quantum mechanics, it may be
preferred, but in others, e.g., in making computa-
tions for simpler problems and for understanding
the literature, it is disadvantageous.
is a remark of Dirac ([2]; section 32, p. 125 The action
principle; the book was rst published in 1930!) developed
by Feynman (R.P. Feynman (1949) Phys. Rev. 76, 769;
(1950) 80, 440).
Chapter 3
Scattering theory
3.1 The Lippmann-Schwinger
equation
Scattering experiments are among the most impor-
tant to verify quantum theory. The central quan-
tity is the scattering cross section (dt.: Streuquer-
schnitt). We imagine a homogeneous (particle) cur-
rent density

j
in
directed at a target at innite
distance. Innite means that the interaction be-
tween target and current can be neglected at this
distance. In the same sense we place a suitable de-
tector at an innite distance behind the target.
In classical mechanics we had dened the dier-
ential scattering cross section via
d =
dN
n
(3.1)
(cf. [14]; section 5.3)). Here dN is the number of
particles detected in an angular interval (, +d),
where is the scattering angle, and n is the area
density measured perpendicular to the incoming
current. More generally, however, dN is given by
the number of particles detected to hit an area
r
2
d = r
2
dsind, where is the angle in the
plane perpendicular to the incoming current den-
sity

j
in
. Thus,
d =
j
scat
r
2
d
j
in
, (3.2)
where
j
scat
is the scattered current density. Notice
that we consider r always to be large in the above
sense!
In order to calculate j
scat
we need to nd the
solution of
(H
0
+H
1
) [ ) = E [ ) , (3.3)
or
(E H
0
) [ ) = H
1
[ ) , (3.4)
i.e. we treat the scattering problem as a stationary
problem. This is sensible if E is the eigenvalue of a
time-independent operator
1
. Here H
0
=
h
2
k
2
2
de-
scribes the free particle in the center of mass frame,
where is the reduced mass of target and beam
particle. H
1
on the other hand is the interaction
Hamilton operator. We may solve (3.4) in a formal
sense by writing
[ ) =[ ) +
1
E H
0
H
1
[ ) (3.5)
2
, where [ ) satises
(H
0
E) [ ) = 0 . (3.6)
1
In particular we consider the beam-target interaction to
be elastic!
2
The operator (E1
0
)
1
we do understand as the power
series expansion of (1 x)
1
, i.e.
1
E
_
1
1 E
1
1
0
_
=
1
E
_
1 +
1
0
E
+
_
1
0
E
_
2
+ . . .
_
.
We note that multiplication with E 1
0
yields
_
E 1
0
_
1
E 1
0
= E
1
E
_
1
1
0
E
_
_
1 +
1
0
E
+
_
1
0
E
_
2
+ . . .
_
=
_
1 +
1
0
E
+
_
1
0
E
_
2
+ . . .
1
0
E
_
1
0
E
_
2
. . .
_
= 1 .
47
48 CHAPTER 3. SCATTERING THEORY
The rst term on the right in Eq. (3.5) corre-
sponds to particles passing the target area without
interaction. The second term, however, will yield
j
scat
. For reasons which will be discussed below we
rewrite Eq. (3.5) as
[
(+)
) =[ ) +
1
E H
0
+i
H
1
[
(+)
) . (3.7)
is a small positive number, and in the limit
+0 we recover Eq. (3.5). Eq. (3.7) is called the
Lippmann-Schwinger equation
3
.
In order to obtain j
scat
we need the r-
representation of (3.7), i.e.
r [
(+)
) = r [ ) +
_
d
3
r
t
d
3
kd
3
k
t
r [
k)
k [
1
E H
0
+i
[

k
t
)
k
t
[

r
t
)
r
t
[ H
1
[
(+)
) . (3.8)
We have inserted the

k-states, because in

k-
representation (E H
0
+i)
1
is diagonal
4
, and
we obtain
r [
(+)
) = r [ ) +
2
h
2
_
d
3
r
t
d
3
k
e
i
kr
(2)
3/2
e
i
(2)
3/2
1
k
2
0
k
2
+i
r
t
[ H
1
[
(+)
) , (3.9)
where E =
h
2
k
2
0
2
and =
h
2
2
. Introducing the
denition
G
(+)
0
(r
r
t
) =
1
(2)
3
_
d
3
k
e
i
k(r
)
k
2
0
k
2
+i
(3.10)
3
Schwinger, Julian Seymour, amerikanischer Physiker,
*New York 12.2.1918, Los Angeles (Kalifornien) 16.7.1994;
arbeitete uber Atombau und am Ausbau der Quantenelek-
trodynamik; erhielt daf ur 1965 mit R.P. Feynman und
S.Tomonaga den Nobelpreis f ur Physik. c _2000 Bibli-
4
Notice that we understand
_
E 1
0
+ i
_
1
in terms of
an expansion in powers of E
1
(1
0
i). We further note
that our previous result (cf. (2.53)),
k [
k [

k
) =

k(
),
may be generalized to
k [ k
2l
x
k
2m
y
k
2n
z
[

k
) = k
2l
x
k
2m
y
k
2n
z
(
k

k
) ,
where l, m, n = 0, 1, 2, . . ..
we have
r [
(+)
) = r [ ) (3.11)
+
2
h
2
_
d
3
r
t
G
(+)
0
(r
r
t
)
r
t
[ H
1
[
(+)
) .
G
(+)
0
(r
r
t
) is the so called Green function of the
free particle.
When we do the integral in Eq. (3.10) (cf. be-
low), we will nd that G
(+)
0
(r
r
t
) is such that Eq.
(3.11) may be written as
r [
(+)
) = r [ )
2
h
2
e
ik
0
r
4r
_
d
3
r
t
(2)
3/2
k
t
[

r
t
)
r
t
[ H
1
[
(+)
)
. .
=(2)
3/2
k
]1
1
]
(+)
)
= r [ ) +
A
(2)
3/2
e
ik
0
r
r
, (3.12)
where
A =
_
2
h
_
2
k
t
[ H
1
[
(+)
) (3.13)
is the scattering amplitude.
Using Eq. (1.118) we obtain
j
scat
=
h
2i
[ A [
2
(2)
3
_
e
ik
0
r
r

r
e
ik
0
r
r
(3.14)
e
ik
0
r
r

r
e
ik
0
r
r
_
e
r
=
hk
0
r
2
[ A [
2
(2)
3
e
r
.
And what about j
in
? We do require that the incom-
ing particles possess sharply dened momentum
k
0
.
Thus [ ) =[
k
0
). The corresponding current den-
sity, which we obtain by inserting r [

k
0
) into Eq.
(1.118) is
j
in
=
h
k
0
(2)
3
. (3.15)
Inserting Eqs. (3.14) and (3.15) into (3.2) yields
d =[ A [
2
d . (3.16)
3.1. THE LIPPMANN-SCHWINGER EQUATION 49
At this point we want to interrupt for a moment
and compute G
(+)
0
(r
r
t
) in Eq. (3.10) explicitly.
We use polar coordinates dening te z-axis along
r
r
t
, i.e.
_
d
3
k =
_

0
dkk
2
_
2
0
d
_

0
d sin
=
_

0
dkk
2
_
2
0
d
_
1
1
dy , (3.17)
where y = cos . This yields
G
(+)
0
(r
r
t
) =
1
(2)
2
i [ r
r
t
[
_

0
dk
k
_
e
ik]r
]
e
ik]r
]
_
k
2
0
k
2
+i
=
1
(2)
2
i [ r
r
t
[
_

dk
ke
ik]r
]
k
2
0
k
2
+i
We evaluate the integral by the method of
residues
5
along the path in the upper half plane
shown in gure 3.1. Note that the loop does not
contribute to the integral, because for ik = ie
i
=
i cos sin the resulting factor e
sin
sup-
presses this contribution in the limit for
0 . Thus only the contribution from the
real axis remains. But according to the residue the-
orem we may write
G
(+)
0
(r
r
t
) =
1
4
2
i [ r
r
t
[
lim
_
dk
ke
ik]r
]
k
2
0
k
2
+i
5
The residue theorem: Let f(z) be single valued and an-
alytic on a simple close curve C except at the singularities
a
1
, b
1
, . . .. Then the residue theorem states that
_
C
f(z)dz = 2i (a
1
+ b
1
+ . . .) . (3.18)
Calculation of residues: The function f(z) has a pole of order
k at z = a. a
1
is given by
a
1
= lim
za
1
(k 1)!
d
k1
dz
k1
_
(z a)
k
f(z)
(3.19)
(cf. [21]).
k
k
2
k
1
Figure 3.1: Integration path in the upper com-
plex plane. The poles are indicated by k
1

k
0
+i/(2k
0
) and k
2
= k
1
.
.
=
2i
4
2
i [ r
r
t
[
all residues of
ke
ik]r
]
k
2
0
k
2
+i
at all poles inside the integration loop. The poles
occur at k
1,2
=
_
k
2
0
+i. But only the one with
the positive sign lies within the integration loop.
We therefore have
Res
ke
ik]r
]
k
2
0
k
2
+i
k
1
= Res
ke
ik]r
]
_
_
k
2
0
+i k
_
_
. . . +k
_
= lim
kk
1
(k k
1
) ke
ik]r
]
_
. . . k
_ _
. . . +k
_
=
e
ik
1
]r
]
2
.
Letting 0 we obtain nally
G
(+)
0
(r
r
t
) =
e
i]
k
0
]]r
]
4[r
r
t
[
. (3.20)
Now we consider the case when [ r [[

r
t
[. We
obtain
(r
r
t
)
2
= r
2
+r
t2
2rr
t
cos(r,
r
t
)
= r
2
_
1
2r
t
r
cos(r,
r
t
) +
_
r
t
r
_
2
_
from which we have
[ r
r
t
[= r
_
1
r
t
r
cos(r,
r
t
) +. . .
_
.
Therefore
G
(+)
0
(r
r
t
)
e
ik
0
r
r
1
4
e
ik
0
r
cos(r,
)
+. . . (3.21)
If we now introduce

k
t
according to
k
t
= k
0
r
t
r
t
we get
e
ik
0
r
cos(r,
)
= e
i
(3.22)
= (2)
3/2
k
t
[

r
t
) .
Combining Eqs. (3.21) and (3.22) we now see that
Eq. (3.12) was indeed justied.
3.2 The 1st Born approxima-
tion
The simplest approach to solving the Lippmann-
Schwinger equation is iteration. Replacing [
(+)
)
on the right side by [ ) we obtain the rst Born
approximation to the scattering amplitude:
A
(1)
=
4
2
h
2

k
t
[ H
1
[ ) . (3.23)
Setting [ ) =[

k) yields the scattering amplitude
to lowest order in terms of the

k-representation of
H
1
.
We now consider the special case of a spherically
symmetric potential:
H
1
= | (r) . (3.24)
We now obtain
A
(1)
=
4
2
h
2
_

k
t
[ r)d
3
r| (r) r [
k)
=

2h
2
_
d
3
r| (r) e
ir(
)
=
2
h
2
q
_

0
drr| (r) sin(rq) . (3.25)
The quantity
q = [
k

k
t
[=
_
k
2
+k
t2
2kk
t
cos
(k=k
)
=
_
2k
2
(1 cos )
= 2k sin(/2) (3.26)
is called momentum transfer
6
.
If we insert the so called Yukawa potential
| (r) =
e
r
r
, (3.27)
where
1
is a measure for the range of the inter-
action, we obtain
7
A
(1)
=
2
h
2
1
q
2
+
2
. (3.28)
Setting = 0 (innite range) yields the scattering
cross section for the Coulomb potential:
d =
4
2
2
h
4
1
q
4
d
=

2
2
4h
4
k
4
sin
4
(/2)
d . (3.29)
Because
h
2
k
2
2
=
1
2
v
2
is the kinetic energy in the

center of mass system, where v
is the relative ve-

locity for large distances, this can be written as
d =

2
4
2
v
4
sin
4
(/2)
d . (3.30)
This expression is known as the Rutherford scat-
tering from classical mechanics (cf. [14]; section
6
This particular , which is the scattering angle, should
not be confused with other s in this text.
7
Notice that the resulting integral is the Laplace trans-
form of sin (qr).
3.3. PARTIAL WAVE EXPANSION 51
5.3)! The agreement must be considered acciden-
tal, however. The limit 0 collides with our pre-
vious assumption, that at large distances between
beam particles and target the interaction should
not be felt. If higher approximations are included
this leads to problems with divergencies (for a more
detailed discussion see Ref. [5]). Thus (3.30) is a
computational example of the rst Born approxi-
mation, and not so much an important result.
We close this discussion with a remark regarding
[
(+)
). The same calculation could have been car-
ried out with [
()
), which leads to an integration
in the lower complex half plane. The result is an
unscattered plane wave plus an incoming spherical
wave instead of an outgoing spherical wave.
3.3 Partial wave expansion
Again we consider a potential with spherical sym-
metry H
1
= | (r). In this case we may write the
right side of Eq. (3.12) as an expression in terms
of Legendre polynomials P
l
(cos ):
r, [
(+)
) =
l=0
c
l
l
(r) P
l
(cos )
r
. (3.31)
Assuming the same is possible for the scattering
amplitude we obtain
A() =
l=0
(2l + 1) A
l
P
l
(cos ) . (3.32)
The coecients c
l
and A
l
and the function (r)
can be determined as follows. Comparing with the
case of hydrogen we observe that
l
(r) satises the
radial Schrodinger equation
_
2
r
+k
2
| (r)
l (l + 1)
r
2
_
l
(r) = 0 (3.33)
8
, where the boundary conditions are
l
(0) = 0.
For r we have
l
(r) = B
l
sin
_
kr

2
l +
l
_
. (3.34)
8
Note that k is the magnitude of the wave vector.
The quantity
l
is called phase shift. Similarly we
can expand the plane waves in terms of Legendre
polynomials:
e
ikz
= e
ikr cos
(3.35)
=
l=0
i
l
(2l + 1) j
l
(kr) P
l
(cos ) .
Here j
l
are spherical Bessel functions. If we substi-
tute the Eqs. (3.31), (3.32), (3.33) and (3.34) into
Eq. (3.16) we obtain
c
l
= (2l + 1) i
l
e
i
l
and
A() =
1
2ik
l=0
(2l + 1)
_
e
2i
l
1
_
P
l
(cos ) . (3.36)
Thus, the dierential cross section is given by
d
d
=
1
k
2
l=0
(2l + 1) e
i
l
sin
l
P
l
(cos )
2
, (3.37)
and the total cross section is
= 2
_

0
A()
2
sind
=
4
k
2
l=0
(2l + 1) sin
2
l
. (3.38)
From (3.36) and (3.38) we nd that
=
4
k
ImA(0) . (3.39)
This is called the optical theorem. Notice that for
suciently weak potentials for which the Born ap-
proximation holds, all phase shifts are small and
are given by
sin
l

l
=
2
h
2
_
| (r) j
2
l
(kr) dr . (3.40)
3.4 Some remarks on formal
scattering theory
The Lippmann-Schwinger equation
[
(+)
) =[
k) +G
(+)
0
H
1
[
(+)
) (3.41)
with
G
(+)
0
= (E H
0
+i)
1
, (3.42)
where G
(+)
0
is also called the free propagator, can be
solved formally by multiplication with the inverse
of G
(+)
0
from the left and using H = H
0
+H
1
. The
result is
(E H+i) [
(+)
)
= (E H+i) [
k) +H
1
[
k)
or
[
(+)
) =
_
1 +G
(+)
H
1
_
[
k) , (3.43)
where
G
(+)
= (E H+i)
1
. (3.44)
Inserting Eq. (3.43) into Eq. (3.13) yields
A =
_
2
h
_
2
k
t
[ H
1
_
1 +G
(+)
H
1
_
[
k)

_
2
h
_
2
k
t
[ T [
k) , (3.45)
where
T = H
1
_
1 +G
(+)
H
1
_
. (3.46)
Multiplication of Eq. (3.41) by H
1
and using H
1
[
(+)
) = T [
k) yields another more useful operator

equation for T, i.e.
T = H
1
_
1 +G
(+)
0
T
_
. (3.47)
Obviously (3.47) may be iterated,
T = H
1
_
1 +G
(+)
0
H
1
_
1 +G
(+)
0
T
__
= H
1
_
1 +G
(+)
0
H
1
_
1 +G
(+)
0
H
1
_
1 +G
(+)
0
T
___
= . . . (3.48)
9
, which allows to obtain the higher order correc-
tions to the 1st Born approximation!
The operator T is also useful in scattering prob-
lems involving crystals. To be specic, in practical
calculations of electron beams scattered o crystal
surfaces
10
. The target crystal is assumed to be an
ordered lattice of non-overlapping spherically sym-
metric atomic potentials u(r r
i
) centered around
nuclear positions r
i
. Therefore H
1
(r) becomes
H
1
(r) =
i
u(r r
i
)
and Eq. (3.48) yields
T =
i
u
i
+
ij
u
i
G
(+)
0
u
j
(3.49)
+
i,j,k
u
i
G
(+)
0
u
j
G
(+)
0
u
k
+. . . .
We note that G
(+)
0
indeed describes the free prop-
agation of the electron waves between the scat-
tering potentials u
i
!
9
Note that formally this geometric series may be summed
up to yield T = 1
1
(1 G
(+)
0
1
1
)
1
.
10
This technique is used to investigate the surface struc-
ture of metal surfaces, because electrons of typically 100eV
do not penetrate more than a few atomic layers into the sur-
face. The technique is called LEED (Low Energy Electron
Diraction). Literature: M. A. van Hove and S. Y. Tong
(1979) Surface Crystallography by LEED. Springer
Chapter 4
Solution methods in quantum
mechanics
In most practical cases the Schrodinger equation
cannot be solved analytically. Therefore it is neces-
sary to develop approximation methods, which can
be applied in as many as possible situations. Here
we discuss four such methods.
The rst two assume that the Hamiltonian may
be split into two parts, the rst of which can be
solved exactly, whereas the second is a small pertur-
bation. The latter may be either time-independent
or time dependent. In both cases the corrections
to the unperturbed eigenvalues and eigenstates are
written as power series in terms of the perturbation.
The rst approach will tell us something about the
ne structure of spectral lines. The second ap-
proach is more general and encompasses the cal-
culation of transition probabilities as well as prob-
lems in many particle quantum theory or even in
relativistic quantum eld theory.
The above decomposition of the Hamiltonian
may not be possible however. In this context we
look at two more approximation schemes. The rst,
the variation method, is particularly useful for the
computation of eigenvalues and eigenfunctions of
larger atoms or molecules. The second method is
the WKB method, a glimpse of which was con-
tained in problem 8.
4.1 Time-independent pertur-
bations
We start by decomposing H according to
H = H
0
+H
1
, (4.1)
where H
0
can be diagonalized exactly, i.e.
k [ H
0
[ l) = E
k,0
kl
, (4.2)
with k [ l) =
kl
and 1 =

n
k=1
[ k)k [. We as-
sume that all E
k,0
are distinct, and that all matrix
elements of H
1
in this representation are small. We
express this fact via
H
1
= J , (4.3)
where is small. Thus, we expand the eigenvalue
E
k
and the eigenstate [
k
) of H in powers of
E
k
= E
k,0
+E
k,1
+
2
E
k,2
+. . . (4.4)
and
[
k
) = [
k,0
)
. .
=]k)
+ [
k,1
) +
2
[
k,2
) +. . . . (4.5)
Inserting this into H [
k
) = E
k
[
k
) yields
(H
0
+J)
_
[
k,0
) + [
k,1
) +
2
[
k,2
) +. . .
_
= (E
k,0
+E
k,1
+. . .) ([
k,0
) + [
k,1
) +. . .)
and therefore
H
0
[
k,0
)
+
_
J [
k,0
) +H
0
[
k,1
)
_
53
54 CHAPTER 4. SOLUTION METHODS IN QUANTUM MECHANICS
+
2
_
J [
k,1
) +H
0
[
k,2
)
_
+. . . (4.6)
= E
k,0
[
k,0
)
+
_
E
k,1
[
k,0
) +E
k,0
[
k,1
)
_
+
2
_
E
k,2
[
k,0
) +E
k,1
[
k,1
)
+E
k,0
[
k,2
)
_
+. . . .
If we consider the terms in brackets as approxima-
tions of E
k
and [
k
), then we obtain the recursion
relations
H
0
[
k,0
) = E
k,0
[
k,0
)
and
J [
k,m1
) +H
0
[
k,m
) (4.7)
= E
k,m
[
k,0
) +E
k,m1
[
k,1
) +. . .
+E
k,0
[
k,m
)
with m = 1, 2, . . .. The solution of the rst equation
is [ k) as we already know. In order to solve the
second equation for n = 1 we expand [
k,1
) in
terms of the basis of the unperturbed system:
[
k,1
) =
n
r=1
[ r)r [
k,1
) .
This yields
l [ J [ k) +E
l,0
l [
k,1
) = E
k,1
lk
+E
k,0
l [
k,1
) ,
and for l = k we obtain the rst correction to the
eigenvalue
E
k,1
= k [ J [ k) . (4.8)
The rst correction to the eigenvector follows for
l ,= k:
l [
k,1
) =
l [ J [ k)
E
k,0
E
l,0
. (4.9)
This, however, does not yet determine the ampli-
tude, because we still must normalize [
k
), i.e.
1 =
k
[
k
). Insertion of the expansion (4.5)
yields
1 =
k,0
[
k,0
)
+
_
k,0
[
k,1
) +
k,1
[
k,0
)
_
+
2
_
k,0
[
k,2
) +
k,1
[
k,1
)
+
k,2
[
k,0
)
_
+. . . .
Because is an arbitrary number, the terms in
brackets must vanish, i.e.
k,0
[
k,1
) +
k,1
[
k,0
) = 0
k,0
[
k,2
) +
k,1
[
k,1
) +
k,2
[
k,0
) = 0
etc.
Therefore
k [
k,1
) = i ,
where is real. To rst order we nd
[
k
) = (1 +i) [ k) +
r,=k
[ r)
r [ J [ k)
E
k,0
E
r,0
+. . .
To this level of approximation we may also write
1 +i e
i
.
The second order approximation may be found
with additional work, i.e., after replacing W by
H
1
, we obtain
E
k
= E
k,0
+k [ H
1
[ k)
+
l,=k
[ l [ H
1
[ k) [
2
E
k,0
E
l,0
+. . . (4.10)
for the eigenvalue and
[
k
) =[ k)T
k
+
l,=k
[ l)R
lk
(4.11)
with
T
k
= 1
1
2
m,=k
[ m [ H
1
[ k) [
2
(E
k,0
E
m,0
)
2
+. . . (4.12)
4.1. TIME-INDEPENDENT PERTURBATIONS 55
and
R
lk
=
l [ H
1
[ k)
E
k,0
E
l,0
+
m,=k
_
l [ H
1
[ m)m [ H
1
[ k)
(E
k,0
E
l,0
) (E
k,0
E
m,0
)
l [ H
1
[ k)k [ H
1
[ k)
(E
k,0
E
l,0
)
2
_
. (4.13)
In principle we may carry on like this, but the
degree of complexity quickly increases.
Example anharmonic oscillator:
As an example we consider the anharmonic oscilla-
tor in 1D of problem 8, i.e.
H = h
_
1
2
2
q
+
1
2
q
2
+

4
q
4
_
,
where q =
_
m/hx and = (2h/m)
2
is a pa-
rameter.
The rst order energy correction due to the per-
turbation

4
q
4
is given by
4
k [ q
4
[ k) .
Instead of working in an explicit representation
we choose to express q
4
in terms of the raising and
lowering operators introduced on page 27, i.e.
q =
1
2
_
a +a
+
_
.
Using the commutator (1.131) it is somewhat labo-
rious but straightforward to work out the following
expression for 4q
4
:
4q
4
= 3(2N
2
+ 2N + 1) +a
4
+ (a
+
)
4
+(2N + 1)
_
a
2
+ (a
+
)
2
_
+
_
a
2
+ (a
+
)
2
_
(2N + 1) . (4.14)
Here N = a
+
a is the number operator (cf. Eq.
(1.148). We notice that due to the othogonality of
dierent energy states we may write
k [ q
4
[ k) = k [ 3(2N
2
+ 2N + 1) [ k)
= 3(2k
2
+ 2k + 1) .
To rst order in we therefore obtain
E
k
h
_
k +
1
2
+
3
8
(k
2
+k +
1
2
)
_
.
The second order correction may be com-
puted based on the other terms in Eq.
(4.14). For the ground state we obtain
E
0
h
_
1/2 + (3/16) (21/128)
2
_
(exer-
cise!).
At the beginning of the above calculation we
had ruled out degeneracy between the E
k,0
. But
what happens, if there are degenerate eigenvalues?
Obviously, Eq. (4.10) no longer applies. Let us
assume that an eigenvalue is m-fold degenerate.
We then try to construct m new eigenfunctions
from linear combinations of the m original eigen-
functions. Our goal hereby is to construct the new
eigenfunction in a way that the matrix elements
of H
1
between the new eigenfunctions become
zero. This procedure corresponds to a unitary
transformation in the subspace of the degenerate
eigenfunctions. The resulting representation of
H
1
can be treated with the above formalism. In
general this leads to the removal of the degeneracy,
and the eigenvalues of the perturbed problem are
distinct.
Degenerate case - 1st order energy correction:
We consider one particular energy level E
0
which
is m-fold degenerate. The degenerate states [
,0
)
satisfy
H
0
[
,0
) = E
0
[
,0
) (4.15)
for = 1, . . . , m. Obviously, Eqs. (4.10) and (4.11)
show that our previous treatment must be modied
to avoid the divergencies. This modication con-
sists in writing down the most general solution to
Eq. (4.15), which is the linear combination of the
[
,0
), i.e.
[

,0
) =
m
=1
c
[
,0
) . (4.16)
Now we repeat the above calculation, but with this
set of m unperturbed states! We write
[
) =[

,0
) + [
,1
) +
2
[
,2
) +. . . (4.17)
for the perturbed state and
E
= E
0
+E
,1
+
2
E
,2
+. . . (4.18)
for the perturbed energies. Inserting (4.17) and
(4.18) into H [
) = E
) yields
H
0
[

,0
) = E
0
[

,0
) (4.19)
and (rst order only)
(H
0
E
0
) [
,1
) = (E
,1
J) [

,0
) . (4.20)
As before we write down an expansion of [
,1
) in
terms of the solutions of the unperturbed problem:
[
,1
) =
m
=1
[
,0
)
,0
[
,1
)
+
n
r=m+1
[ r)r [
,1
) . (4.21)
The rst summation includes the degenerate states,
where we assume only one degenerate energy level.
The second term includes all remaining states for
all other energy levels. Upon inserting (4.21) into
(4.20) we nd
n
r=m+1
(H
0
E
0
) [ r)r [
,1
)
= (E
,1
J) [

,0
) . (4.22)
Finally we multiply with
,0
[ from the left to
obtain
0 = E
,1
,0
[

,0
)
,0
[ J [

,0
) , (4.23)
where we have used
,0
[ r) = 0 (because [
,0
)
and [ r) belong to distinct eigenvalues). At rst
glance this equation looks like Eq. (4.8), but we
should not forget to insert Eq. (4.16). The result
is
0 =
m
=1
c
E
,1
,0
[
,0
) (4.24)
,0
[ J [
,0
) ,
which is a set of linear coupled equations. The non
trivial solution requires
0 = det
_
E
,1
,0
[
,0
) (4.25)
,0
[ J [
,0
)
_
.
In general we do not have
,0
[
,0
) =
, but
we can always construct such [
,0
) based on a
complete solution of the unperturbed problem,
i.e. without loss of generality we may indeed
assume
,0
[
,0
) =
. Thus, we obtain the

m rst order corrections to E
0
by computing the
eigenvalues of an mm matrix with the elements
,0
[ J [
,0
) in this case.
To illustrate this procedure we calculate the cor-
rections to the eigenvalue to 1st order and the cor-
responding eigenfunctions to 0th order for a two-
fold degenerate energy level E
0
. According to Eq.
(4.25) we have
J
11
E
1
J
12
J
21
J
22
E
1
= 0 ,
where J

,0
[ J [
,0
). The solution is
E
,1
=
1
2
_
J
11
+J
22
_
(J
11
J
22
)
2
+ 4[J
12
[
2
_
.
The dierence between these values, i.e. 2
. . .,
is the level splitting to 1st order due to the pertur-
bation.
The (not normalized) 0th order eigenstate [

,0
)
is obtained via Eq. (4.16) using m = 2. The c
are
4.1. TIME-INDEPENDENT PERTURBATIONS 57
computed via (4.24) with the now known E
,1
=
E
,1
, i.e. we solve
0 =
2
=1
c
E
,1
for c
1
and c
2
. Including the normalization the
coecients are given by
c
1
=
_
J
12
2 [ J
12
[
_
1
J
11
J
22
. . .
_
c
2
=
_
J
21
2 [ J
12
[
_
1
J
11
J
22
. . .
_
.
This example including how to obtain the next
higher approximations can be found in reference
[8] (problems 1 and 2 in 39).
The 1/r
6
-dispersion interaction:
We consider two noble gas atoms separated by a
distance r. This separation refers to the nuclei of
charge Ze located at 0 and r along the x-axis, each
surrounded by Z electrons. The electrons bound
to the nucleus at 0 have coordinates
i
; and those
bound to the nucleus at r have coordinates
i
t
.
We assume r to be so large that there is negligi-
ble overlap between the electronic charge distribu-
tions about the two nuclei. The Hamiltonian for
this two-atom system will be
H = H
1
+H
2
+| ,
where |, the atom-atom interaction energy, is given
by
| = e
2
_
Z
2
r

Z
i=1
_
Z
[

t
i
[
+
Z
[
i
r [
_
+
Z
i,j=1
1
[
i
t
j
[
_
.
To second order in perturbation theory, the inter-
action energy between the two atoms will be
E = 0 [ | [ 0) +
[ 0 [ | [ ) [
2
E
0
E
.
Here [ 0) is the ground state of the unperturbed
two-atom system, and [ ) are corresponding ex-
cited states.
The rst term,
0 [ | [ 0) =
(eZ)
2
r
0 [ 0)
. .
=1
+0 [
Z
i=1
_
. . .
_
+
Z
i,j=1
1
[
i
t
j
[
[ 0) ,
expressed in r-representation becomes
0 [ | [ 0) =
_
d
3
d
3
0
()
1
[
t
r [
0
(
t
) ,
where
0
denotes the ground state charge distri-
bution. Note that
0
includes the nuclear charge!
From classical electrodynamics we know that the
integral may be expressed as
_
d
3
d
3
t
. . . =
_
d
3
0
()
0
()
[15] (section 3.1), where
0
() is the electrostatic
potential at due to the second atom at r. A simple
argument shows, however, that this term vanishes.
Consider an isolated neutral and radially symmet-
ric charge distribution
0
. According to Gausss in-
tegral theorem we have
_
Sphere
d
f

E
0
= 0, where
0
is at the center of the sphere. By symmetry
we conclude

E
0
= 0 everywhere outside the charge
distribution. Therefore
0
= const, which follows
from

E
0
=
0
. Thus we nd that the above
integral vanishes, because the isolated charge dis-
tribution may be brought in from innity without
altering

E
0
= 0!
This means that we are left with
E =
[ 0 [ | [ ) [
2
E
0
E
. (4.26)
This time the matrix element, i.e.
0 [ | [ ) =
_
d
3
d
3
0,
()
0,
(
t
)
[
t
r [
,
does not vanish. Note that the function
0,
()
no longer must be radially symmetric. Using the
multipole expansion [15] (cf. Eq. (3.15) pp.) we
nd to lowest order for large r:
[ 0 [ | [ ) [
2
_
1
r
3
_
2
(dipole-dipole interaction). Notice that the numer-
ator in Eq. (4.26) is positive, whereas the denom-
inator is negative
_
E
> E
0
!
_
. We therefore con-
clude
E =
C
r
6
, (4.27)
where C is positive, for large r!
Eq. (4.27) is called dispersion attraction. It is
build into many phenomenological atom-atom or
molecule-molecule potentials. One example is the
well known Lennard-Jones pair potential
u
ij
= 4
_
_

r
ij
_
12
_

r
ij
_
6
_
, (4.28)
where and are positive constants. Notice that
the repulsive (/r)
12
-term merely represents a
convenient numerical form of the rather complex
overlap repulsion at short distances. In contrast to
the (/r)
6
-term it is not justied by a systematic
expansion.
4.2 Variation method
This method can be applied when it is not possible
to split the Hamiltonian into two parts, one of
which is exactly solvable whereas the remaining
piece is a small perturbation. Here we discuss the
method applied to the stationary case.
Application to the ground state:
First we note that the average energy in an arbi-
trary normalized state [ ) of the system is always
larger or equal to the smallest energy eigenvalue
(ground state). Proof:
[ H [ ) =
n
[ H [
n
)
n
[ )
=
n
E
n
[ [
n
) [
2
= E
0
n
[ [
n
) [
2
. .
=1
+
n
(E
n
E
0
) [ [
n
) [
2
E
0
.
The idea of the method is as follows. First one
decides to work in a certain representation; here we
use the r-space representation. Then the unknown
ground state amplitude r [
0
) is replaced by an
Ansatz r [
0,A
), which is a function depending
on one or more parameters. This function should
already include the known properties of the correct
solution like symmetries or asymptotic behavior.
The goal is to minimize the average ground state
energy by variation of the parameters, i.e.,
0,A
[ H [
0,A
) = 0 (4.29)
yields a number of coupled equations from which
the variation parameters may be determined. No-
tice that
[ H [ ) (4.30)
=
_
[ r)d
3
rr [ H [ )
=
_
d
3
r [ r)
_
h
2
2m
2
+|
_
r [ )
The idea of the above procedure is an approxima-
tion to r [
0
) including the approximate ground
state energy E
0,A
.
Example anharmonic oscillator:
As an example we consider again the anharmonic
oscillator in 1D of problem 8, i.e.
H = h
_
1
2
2
q
+
1
2
q
2
+

4
q
4
_
,
4.2. VARIATION METHOD 59
where q =
_
m/hx and = (2 h/m)
2
is a
parameter. Our normalized Ansatz or normalized
trial function (dt.: Versuchsfunktion) is
0,
(q) =
_
_
1/4
exp
_
2
q
2
_
. (4.31)
For = 1 this is the ground state of the harmonic
1D oscillator.
A not too dicult integration yields
H)
0,
=
_

dq
o,
(q)H
o,
(q)
= h
4(
3
+) + 3
16
2
.
The best estimate of H)
0,
based on our trail func-
tion follows via
0 =
d
d
H)
0,
min
= h
2
3
2 3
8
3
min
.
Rather than solving this equation exactly, which
of course is possible, we consider two limiting cases,
i. e. 1 and 1. In the case 1 we may
insert
min
= 1 + into the above equation and
keeping terms linear in only we obtain = 3/4
or
min
1 + (3/4) and thus
min
H)
0,
h
_
1
2
+
3
16
_
.
1
. In the opposite limit, 1, we obtain via
0 = 2
3
min
2
min
3 2
3
min
3
min
(3/2)
1/3
and therefore
min
H)
0,
h
3
8
_
3
2
_
1/3
0.4208 h
1/3
.
We note that the numerically exact solution
is 0.429h
1/3
. The attendant ground states
1
Including the second order in the bracket becomes
_
1
2
+
3
16

27
32
2
_
. Compare this result to the perturbation
theory on page 55.
q [
0,
min
) are obtained via insertion of
min
into Eq. (4.31).
Application to excited states:
This method may also be applied to calculate
approximations to the excited states. We con-
sider the nth excited state and assume that the
E
i
(i = 1, . . . , n 1) and the corresponding states
[
i
) (i = 1, . . . , n 1) are known in one particular
representation (i.e., the r-representation r [
i
)).
The states must be orthogonal to each other, be-
cause they belong to dierent eigenvalues:
i
[
k
) =
_

i
[ r)d
3
rr [
k
) = 0 (i ,= k) .
Again we construct a reasonable r [
n,A
), which
depends on certain parameters. It is important
that [
n,A
) is orthogonal to all other states. Now
the procedure is the same as above, and we mini-
mize
n,A
[ H [
n,A
) i.e.
n,A
[ H [
n,A
) = 0 .
That the resulting minimal [
n,A
) =[ ) is reason-
able we may see via
[ H [ )
=
n1
i=0
[ H [
i
)
i
[ )
. .
=0
+
in
[ H [
i
)
i
[ )
=
in
E
i
[ [
i
) [
2
= E
n
in
[ [
i
) [
2
+
in
(E
i
E
n
) [ [
i
) [
2
= E
n
+
in
(E
i
E
n
) [ [
i
) [
2
E
n
.
Notice that 1 = [ ) =

i0
[
i
)
i
[ ) =
in
[ [
i
) [
2
. A disadvantage of the variation
approach is, that it is not a systematic expansion,
i.e. we obtain an approximation whose quality is
dicult to judge. Nevertheless, with some experi-
ence it produces good results, and it is widely used
also outside quantum mechanics.
4.3 The quasi classical ap-
proximation
This approximation procedure is also called WKB-
approximation according to its authors Wentzel,
Kramers and Brillouin, who invented it indepen-
dently in 1926. The WKB method is mainly appli-
cable to one dimensional or radial symmetric prob-
lems. We consider both cases via
x [ (t)) = e
i
h
Et
() , (4.32)
where x and | () = | (x), and
r [ (t)) = e
i
h
Et
1
r
Y
lm
(, ) () , (4.33)
with r and | () = |
eff
(). The Schrodinger
equation now has the form
_
h
2
2m
E +| ()
_
() = 0 . (4.34)
The WKB method is based on the following con-
siderations. The solution of the Schrodinger equa-
tion for a constant potential (free particle) has
the form of a plane wave exp
_
i
h
p
_
, where p =
_
2m(E |). If the potential contains slight spa-
tial variations then one might expect that this solu-
tion changes to exp
_
i
h
_
p (
t
) d
t
_
. Slight vari-
ations refers to cases when the wavelength is short
compared to length scale of the changes in the po-
tential. This corresponds to high quantum numbers
(remembering for instance the harmonic oscillator).
Thus, we insert the Ansatz = A
+
+
+A
with
() = exp
_
i
h
_

(
t
) d
t
_
(4.35)
into the Schrodinger equation to obtain a Riccatis
dierential equation:
( ())
2
2m
(E | ())
h
2im
t
() = 0 . (4.36)
We now expand () in h, i.e.
() =
0
() + h
1
() + h
2
2
() +. . . . (4.37)
Inserting this into the above equation yields
0
() =
_
2m(E | ()) Q()
1
() =
i
2
t
0
0
=
i
2
lnQ()
2
=
Q
tt
4Q
2
+
3 (Q
t
)
2
8Q
3
=
m
4Q
3
_
|
tt
+
5m
2
_
|
t
Q
_
2
_
.
.
.
Inserting this we notice that the term
1
can be
integrated and yields a factor. The nal result is
() =
1
Q
exp
_
i
h
_

d
t
_
Q(
t
)
+
mh
2
4Q
3
_
|
tt
+
5m
2
_
|
t
Q
_
2
_
_
+. . .
_
. (4.38)
This approximation is a good one if
mh
2
4Q
4
_
|
tt
+
5m
2
_
|
t
Q
_
2
_
1 . (4.39)
If | > E we obtain analogously (use i)
via = B
+
+
+B
=
1
R
exp
_
1
h
_

d
t
_
R(
t
)
+
mh
2
4R
3
_
|
tt
+
5m
2
_
|
t
R
_
2
_
_
+. . .
_
, (4.40)
where
R() =
_
2m(| E) .
4.4. TIME-DEPENDENT PERTURBATION THEORY 61
Next we want to compute the energy eigenvalues.
For this purpose we count the number of zeros, n,
of the wave function. We consider in the complex
plane (not just on the real axis). If (z
0
) = 0 then
the function
t
/ has a simple pole with residue 1,
because = A(z z
0
) +. . . and
t
= A+. . .. i.e.
2in =
_
C
t
(z)
(z)
dz =
_
C
d
dz
lndz
(cf. above). Here C is a closed curve including all
zeros. Note that all poles are in the range dened
by E | > 0, because for E | < 0 the solution
is monotonous. Inserting
+
according to (4.35)
(using
is analogous),
+
() = exp
_
i
h
_

d
t
_
Q(
t
)
+
ih
2
d
d
t
lnQ(
t
) +. . .
__
, (4.41)
we obtain
d ln (z)
dz
=
i
h
Q(z)
1
2
Q
t
(z)
Q(z)
+. . . . (4.42)
The integration of the second term is easy, because
Q
Q
has simple poles of order 1 with residue 1/2 at
the classical turning points. Thus
_
Q
t
Q
dz = 2i , (4.43)
and we obtain
2h
_
n +
1
2
_
=
_
Q(z) dz +. . . ,
which yields
2h
_
n +
1
2
_
= 2
_

2
1
d
_
2m(E | ()) .(4.44)
Here we have used
_
C
2
_
1
, where
1
and
2
are the solutions of E | () = 0. Eq. (4.44) is
identical with the enhanced quantization condi-
tion, which we had applied to the anharmonic 1D
oscillator in problem 8.
It looks as if this procedure will always yield
an approximation for two special solutions of the
Schrodinger equation,
, which can be combined

to give a linear combination fullling the boundary
conditions. In general, however, this is not the case,
because
1
and
2
produce singularities. Thus, it is
not possible to tie the approximations for | < E
and | > E together. This diculty may be avoided
via an exact solution near the points
1
and
2
. We
do not want to do this here and refer the reader to
Ref. [5] (section 41). The resulting wave functions
are
() =
2B
Q
cos
_
1
h
_

1
Q(
t
) d
t

4
_
(4.45)
( <
1
) and
() =
B
R
e
1
h
_
1
R(
)d
(4.46)
( >
1
) for the upper turning point, and
() =
2B
t
Q
cos
_
1
h
_

2
Q(
t
) d
t

4
_
(4.47)
( >
2
) and
() =
B
t
R
e
1
h
_
R(
)d
(4.48)
( <
2
). Note that B
t
= (1)
n
B
.
4.4 Time-dependent pertur-
bation theory
This method is due to P.A.M. Dirac
2
, and was in-
vented to treat spontaneous as well as forced emis-
sion and absorption of radiation. But it is a more
versatile method than that.
Again the Hamilton operator consists of two
parts,
2
Dirac, Paul, Adrien Maurice, britischer Physiker, *Bris-
tol 8.8.1902, Tallahassee (Florida) 20.10.1984; trug durch
grundlegende Arbeiten wesentlich zum Aufbau der Quanten-
mechanik und -elektrodynamik bei; 1933 erhielt er zusam-
men mit E. Schrodinger den Nobelpreis f ur Physik. c _2000
Bibliographisches Institut & F. A. Brockhaus AG
H = H
0
+H
1
(t)
= H
0
+(t) J , (4.49)
where the second describes a small, time-dependent
perturbation. Here we chose
(t) = (t t
0
) =
_
t > t
0
0 t < t
0
, (4.50)
i.e. (t) switches the perturbation on at t = t
0
.
We also use the time evolution operator introduced
in section 2.4, i.e. [ (t)) = S (t, t
0
) [ (t
0
)), and
we insert
S (t, t
0
) = S
0
(t, t
0
) S
1
(t, t
0
) , (4.51)
where
S
0
(t, t
0
) = exp
_
i
h
(t t
0
) H
0
_
, (4.52)
into the dierential equation for S, i.e.
ih
t
S (t, t
0
) = (H
0
+J) S (t, t
0
) . (4.53)
After multiplication with S
1
0
we obtain
ih
t
S
1
=

JS
1
, (4.54)
with

J = S
1
0
JS
0
. (4.55)
Integration yields
S
1
(t, t
0
) = 1
i
h
_
t
t
0
J() S
1
(, t
0
) d , (4.56)
where we have included the step-function behavior
of (t). For small Eq. (4.56) may be iterated.
We insert
S
1
(t, t
0
) =
n=0
n
S
1,n
(t, t
0
)
into (4.56), and by comparing coecients we obtain
S
1,0
= 1 (4.57)
and
S
1,n
(t, t
0
) =
i
h
_
t
t
0
d

J() S
1,n1
(, t
0
) (4.58)
for n > 0. Combining Eq. (4.58) with Eq. (4.51)
nally yields
S (t, t
0
) = exp
_
i
h
(t t
0
) H
0
_
_
1
i
h
_
t
t
0
de
i
h
(t
0
)1
0
H
1
() e
i
h
(t
0
)1
0
+
_
i
h
_
2
_
t
t
0
d
_

t
0
d
t
e
i
h
(t
0
)1
0
H
1
() e
i
h
(
)1
0
H
1
(
t
)
e
i
h
1
0
(
t
0)
+. . .
_
. (4.59)
Now we are able to compute the transition proba-
bilities between eigenstates of the unperturbed sys-
tem. Starting with a state [ i) at t
0
, i.e. H
0
[ i) =
E
i
[ i), we look for the probability to nd the sys-
tem in a state [ f) at time t, i.e. H
0
[ f) = E
f
[ f).
This probability is given by
P
f
(t) =[ f [ S (t, t
0
) [ i) [
2
, (4.60)
(cf. Eq. (2.68)), where the transition amplitude
has the form
f [ S (t, t
0
) [ i) = e
i
h
(E
f
tE
i
t
0
)
fi
i
h
_
t
t
0
df [ H
1
() [ i)e
i
h
(E
f
E
i
)
+
_
i
h
_
2
_
t
t
o
d
_

t
0
d
t
f [ H
1
() [ )
[ H
1
(
t
) [ i)
e
i/ h[(E
f
E
)+(E
E
i
)
]
+. . .
_
. (4.61)
Special case of a time-independent perturbation:
For simplicity we set t
0
= 0 and calculate the lead-
ing approximation only. For i ,= f we obtain
4.4. TIME-DEPENDENT PERTURBATION THEORY 63
P
f
(t) = 4 [ H
1,fi
[
2
sin
2
_
fi
t
2
_
( h
fi
)
2
.
where H
1,fi
f [ H
1
[ i) and h
fi
= E
f
E
i
.
Special case of an oscillating perturbation: For
example an atom exposed to electromagnetic radi-
ation in a spectrometer or in sunlight may feel such
a perturbation. We assume the form
H
1
(t) = 2H
1
cos(t) . (4.62)
Writing 2 cos(t) = exp(it) +exp(it) and after
a simple integration we nd to rst order
P
f
(t) =
4
h
2
[ H
1,fi
[
2
e
i
(
fi
+)t
2
sin(
(
fi
+)t
2
)
fi
+
+e
i
(
fi
)t
2
sin(
(
fi
)t
2
)
fi
2
(4.63)
4
h
2
[ H
1,fi
[
2
sin
2
(
(
fi
)t
2
)
(
fi
)
2
.
This approximation assumes that the frequencies
are high, and the resonant behavior for
fi
dominates, i.e.
lim
fi
P
f
(t) =
1
h
2
[ H
1,fi
[
2
t
2
. (4.64)
Notice that this equation cannot be valid for all t,
because the probability will exceed unity. Notice
also that the above resonance corresponds to ab-
sorption. We may change the sign of to obtain
an expression valid for emission.
We may use the following limiting representation
of the -function,
(x) = lim
sin
2
(x)
x
2
, (4.65)
to rewrite Eq. (4.63) into
P
f
(t)
2
h
2
[ H
1,fi
[
2
t(
fi
) . (4.66)
This form of Eq. (4.63) is also useful for dening
the lifetime of an excited state [ i) via
1

2
h
2
[ H
1,fi
[
2
(
fi
) . (4.67)
Notice that in cases when there are more than one
decay channel, i.e. one nal state, we must sum
over all decay channels
34
.
Transition rates to continuum states: We may
still use Eq. (4.63), but the observed transition
rate is an integral over all transition probabilities
to which the perturbation can drive the system.
Specically, if the density of states is written (E
f
),
where (E
f
)dE
f
is the number of states in the
range E
f
to E
f
+ dE
f
, then the total transition
probability, P(t), is
P(t) =
_
range
P
f
(t)(E
f
)dE
f
. (4.68)
In this expression range means that the integra-
tion is over all nal states accessible under the in-
uence of the perturbation.
Inserting Eq. (4.66) into Eq. (4.68) and treating
H
1,fi
as constant yields
P(t)
2
h
[ H
1,fi
[
2
(E
i
+E)t (4.69)
5
. The transition rate, which determines the inten-
sity of spectral lines, is given by
dP(t)
dt

2
h
[ H
1,fi
[
2
(E
i
+E) . (4.70)
This equation is known as Fermis golden rule
6
.
3
In this discussion we assume that our considerations
thus far remain valid for such processes involving one or few
photons. The correct treatment requires the quantization of
the radiation eld.
4
You should also realize the relation between and t in
our discussion of Heisenbergs uncertainty principle in sec-
tion 2.4.
5
Note: (
fi
) = h(E
f
(E
i
+ E)).
6
Fermi, Enrico, italienischer Physiker, *Rom 29.9.1901,
Chicago (Illinois) 28.11.1954; Professor in Rom, spater in
den USA, verwendete als Erster Neutronen zur Umwandlung
schwerer Atomkerne (daf ur 1938 Nobelpreis f ur Physik),
war magebend am Bau des ersten Kernreaktors (1942)
und der Entwicklung der Atombombe (Manhattan-Projekt)
beteiligt. c _2000 Bibliographisches Institut & F. A. Brock-
haus AG
4.5 Interaction of charges
with electromagnetic
elds
In the following application of time-dependent per-
turbation theory we consider a charged particle in
an external electromagnetic eld. Because we do
not know otherwise we are going to assume that
this eld can be treated classically. From electro-
dynamics we remember that the Lagrange function
of a charged particle, its charge is e, in such a eld
is given by
L =
mc
2
(v)
+
e
c
A v e (4.71)
([15]; Eq. (2.9)), and therefore its generalized mo-
mentum is given by
p =
v
L = (v) mv +
e
c
A (4.72)
7
. Here (v) =
_
1
v
2
c
2
_
1
2
, where v is the particle
velocity, and c is the velocity of light. In addition,
m is the particles rest mass, and e is its charge.
The quantities

A and are the vector potential and
the scalar potential, respectively. Notice that Eq.
(4.72) describes the relativistic momentum. In the
following we use the nonrelativistic limit v/c 0,
and p becomes
p = mv +
e
c
A . (4.73)
The resulting quantum mechanical Hamilton oper-
ator is
H =
1
2m
_
p
e
c
A(r)
_
2
+e(r) +| (r) (4.74)
8
. Here we have added an additional potential
| (r), which has nothing to do with the external
7
Note:

_
a
b
_
=
_
a

b +
_
_
a +
b
_
a
_
+
a
_
b
_
.
8
We calculate the Hamilton function via 1 = p v /,
which yields 1 = mc
2
+
1
2m
_
p
e
c
A
_
2
+e, where we have
used
1
(v) 1
1
2
v
2
c
2
.
eld. Notice that we also omit the rest energy
term, mc
2
, because it is just a constant.

A(r) and
(r) do not commute with p. Using
_
p, f (r)
=
ih
f (r) (cf. page 35), where f (r) is a function

of r, yields
H =
p
2
2m
+| (r) +e(r)
e
mc
A(r) p (4.75)
+
ihe
2mc

A(r) +
e
2
2mc
2
A
2
(r) .
Before discussing this Hamiltonian we must
examine its transformation properties under the
gauge transformations
A

A
t
=

A+

f (4.76)

t
=
1
c
f , (4.77)
where f is a scalar function. We know (cf. [15];
Eqs. (2.18) and (2.19)) that the gauge transforma-
tions (4.76) and (4.77) do not change the electric
and magnetic eld strengths, i.e., they do not in-
uence physical observations. The same should be
true in quantum mechanics. And indeed, as we
show in a homework problem, if [ (t)) is an eigen-
ket of H in (4.75) then
[
t
(t)) = e
i
f(r,t)
[ (t)) (4.78)
is an eigenket of H
t
with

f =
e
hc
f. The complex
phase factor does not alter the transition probabil-
ity to another state, i.e. [(t) [
t
(t))[
2
= [(t) [
(t))[
2
.
Now we return to Eq. (4.75). In many cases
the last two terms turn out to be small (notice
for instance that they contain the factor
1
c
2
9
),
and they may be neglected. In the next chapter,
where we add an intrinsic angular momentum
to our particle, additional terms will have to be
considered, however.
Induced emission and absorption:
We want to study the important case of an elec-
tromagnetic radiation eld considering a linearly
polarized plane wave
9
For the second-to-the-last term we see this using the
Lorentz condition

A =
1
c
(cf. [15]; section 4.1).
4.5. INTERACTION OF CHARGES WITH ELECTROMAGNETIC FIELDS 65
A = A
0
e cos
_
k r wt
_
and = 0 . (4.79)
Here e is a unit vector characterizing the polariza-
tion direction. Notice that

k e = 0 (Lorentz condi-
tion) and

k
2
= w
2
/c
2
(using the wave equation for
A). Using
H
1
=
e
mc
A(r) p (4.80)
=
eA
0
2mc
_
e
i(
krwt)
+e
i(
krwt)
_
(e p)
and neglecting the term

A
2
in Eq. (4.75) yields
the transition matrix element
f [ H
1
(t) [ i)
=
eA
0
2mc
f [ e
i(
kr)
_
e p
_
[ i)e
iwt
eA
0
2mc
f [ e
i(
kr)
_
e p
_
[ i)e
iwt
.
Now we can apply our results obtained above for a
periodic perturbation. The same simple calculation
yields the transition amplitude to rst order for i ,=
f and t
0
= 0:
f [ S (t, t
0
) [ i) = e
i
f
t
ieA
0
2hmc
e
i
tw
2
sin(tw/2)
w/2
f [ e
i
kr
_
e p
_
[ i) . (4.81)
Note that
fi
= , where the upper sign cor-
responds to absorption and the lower sign to emis-
sion.
Now we investigate the matrix element, i.e.
f [ e
i
kr
_
e p
_
[ i)
=
_
f [ e
i
kr
[ r)r [ e p [ i)d
3
r
=
_
f [ r)e
i
kr
r [ e p [ i)d
3
r . (4.82)
We notice that only for r a
0
(Bohrs radius)
there should be sizeable contributions, because the
wave function f [ r) quickly drops o towards
zero for larger r. For visible light we estimate
kr <
2a
0
10
3
, which means that we may ex-
pand the e-function:
f [ e
i
kr
_
e p
_
[ i)
= f [ e p [ i)
if [
_
k r
_
_
e p
_
[ i) . . . . (4.83)
Assuming an unperturbed Hamiltonian,
H
0
=
p
2
2m
+| (r) , (4.84)
it is easy to show that
p =
m
ih
[H
0
, r] . (4.85)
Using this relation in Eq. (4.83) we nd
f [ e p [ i) = im
fi
f [ e r [ i) . (4.86)
The quantity ef [ r [ i) is called the transition
dipole moment
10
, and the emitted radiation is
called electrical dipole radiation. Note again that
e is the polarization vector, i.e. its orientation rel-
ative to r is important! In order to calculate the
emitted intensity we have to evaluate (4.86) explic-
itly (e.g. for hydrogen states in r-representation).
However, more interesting is the question whether
a particular transition is at all possible. It turns
out, as we will see later, that the quantum num-
bers, which characterize the initial and nal states
must obey selection rules, i.e. only certain dier-
ences between them are allowed.
10
Here e is the charge and not [e[!
Chapter 5
Many particle systems
5.1 Two kinds of statistics
We consider an N-body system of identical parti-
cles. Identical means that the Hamiltonian opera-
tor H is invariant under the exchange of any two of
the particles. An N-particle state vector we might
write as an ordered product
[
) [
1
,
2
,...,
N
) (5.1)
= ([
1
)[
2
) . . . [
N
)) .
where the [
i
) denote single particle states
1
. The
mathematical expression of the above invariance by
denition is
H
ij
=
ij
H , (5.2)
or
1
ij
H
ij
= H , (5.3)
where
ij
, the permutation operator, interchanges
particles i and j, i.e.
ij
[
...,
i
,...,
j
,...
) = [
...,
j
,...,
i
,...
) . (5.4)
If [
) is an eigenstate of H with the eigenvalue

E
, then Eq. (5.3) implies

2
H
_
ij
[
)
_
= E
ij
[
)
_
. (5.5)
1
Notice that here . . . [
i
) [
j
) . . . is dierent from
[
j
) [
i
) . . ..
2
by applying
ij
from the left
This equation allows two alternative conclusions:
(a)
ij
[
) = c [
), where c is a constant.
And therefore
ij
[
) = [
) , (5.6)
because
2
ij
[
) =[
).
(b)
ij
[
) is a linear combination of eigen-

states [
(1)
), . . . , [
(s)
), which all produce the

same eigenvalue E
. In this case the eigenvalue

has an intrinsic degeneracy that cannot be re-
moved by any perturbation | consistent with
our above symmetry assumption regarding H, i.e.
1
ij
|
ij
= |. However, experiments thus far show
that only (a) is realized in nature.
An immediate consequence of Eq. (5.6) is that
state space separates into two completely disjoint
spaces in the sense of the superselection rule
(+)
[[
()
) = 0 . (5.7)
Proof:
(+)
[[
()
) =
(+)
[
1
[
()
) =
(+)
[
+
[
()
) =
(+)
[[
()
). Here the in-

dices (+) and () stand for the respective signs in
Eq. (5.6), and we have used
+
=
1 3
In addition the operator obeys the particle ex-
change invariance, and only such operator makes
3
Proof: Here and in the following we extend the de-
nition of the permutation operator
ij
. The permutation
operator now transforms the ordered set
1
,
2
, . . . ,
N
into the new ordered set
1
,
2
, . . . ,
N
via a suc-
cession of pair-permutations as dened before. Thus
we have
[[
)=
1
,
2
,...,
N
[
1
,
2
,...,
N
) =
67
68 CHAPTER 5. MANY PARTICLE SYSTEMS
sense in systems of indistinguishable particles. As
a consequence of the above we are always dealing
with physical problems, where we have (+) or ()-
states only; one may say also even or odd states
4
! But what is the signicance of this?
As a consequence ()-states with
i
=
j
for
i ,= j cannot exist! Otherwise Eq. (5.6) is incor-
rect. However, (+)-states with
i
=
j
for i ,= j
can exist! Because
i
denotes the quantum num-
bers dening the single particle state [
i
), this
means that in ()-states no two single particles
may be in the same single particle state. In this
case we call the particles fermions. In (+)-systems
two or even all N particles may be in the same
single particle state. These particles are bosons
5
.
Thus, (5.6) introduces two ways of sorting particles
1
,
1
2
,...,
1
N
[
1
,
2
,...,
N
). And therefore
+
=
1
. Note that with
2
=
1
= 1 we have also
+
= .
A simple example may serve to illustrate this. We com-
pute the matrix element 1234 [ [ 5678), where the integer
numbers replace [
1
,
2
,...
). The permutation operator is
dened as follows: rst - interchange digit at position 3 with
digit at position 4; second - interchange digit at position 3
with digit at position 1. Thus
[ 5678) =[ 8657)
and therefore
1234 [ [ 5678) = 1 [ 8)2 [ 6)3 [ 5)4 [ 7) (x) .
Now we want to check whether
+
=
1
. We note that
1234 [ [ 5678) = 5678 [
+
[ 1234)
(cf. Eq. (1.45)). First we work out

1
[ 1234) by reverse
application of the above denition of , i.e.
1
[ 1234 =[ 3241) .
We therefore obtain
5678 [
1
[ 1234)
= (5 [ 3)6 [ 2)7 [ 4)8 [ 1))
= 3 [ 5)2 [ 6)4 [ 7)1 [ 8)

in agreement with Eq. (x). Thus, we indeed nd (in this
special case)
1
=
+
. Notice that the . . . [ . . .) are
merely complex numbers which do commute.
4
(+)-states are also referred to as symmetric; ()-states
are thus antisymmetric.
5
Bose, Satyendra Nath, indischer Physiker, *Kalkutta
1.1.1894, ebenda 4.2.1974; wichtige theoretische Arbeiten
zur statistischen Thermodynamik. c _2000 Bibliographisches
Institut & F. A. Brockhaus AG
into available states. If n
is the number of parti-

cles in the single particle quantum state , then n
is called occupation number and we have
= N (5.8)
where
n
=
_
0, 1, . . . , N bosons
0, 1 fermions
. (5.9)
In statistical mechanics the upper row in (5.9)
yields Bose-Einstein statistics and the lower row
yields Fermi-Dirac statistics.
In 1940 W. Pauli
6
(Phys. Rev. 58, 716) showed
that all particles possessing integer spin, which is an
intrinsic angular momentum variable, are bosons,
and all particles with half integer spin are fermions.
This is the so called spin-statistics-theorem which
can be proven in relativistic quantum eld theory.
The fact that no two fermions occupy the same
state is known as Pauli principle.
5.2 Constructing N-particle
wave functions
Now we want to construct the normalized N-
particle states [
()
) which fulll the symmetry re-

quirements for bosons (+) and fermions (). We
start from the Ansatz
[
()
) = K
()
1
,
2
,...,
N
) . (5.10)
As mentioned above,
1
,
2
, . . . ,
N
de-
notes a certain permutation of the original set
1
,
2
, . . . ,
N
, which is obtained via an even (+)
or odd () number of pairwise particle exchanges.
The respective sums,

()
, encompass all dis-
tinct permutations for the two dierent cases in-
cluding the identity. In addition
6
Pauli, Wolfgang, schweizerisch-amerikanischer Physiker
osterreichischer Herkunft, *Wien 25.4.1900, Z urich
15.12.1958; Professor in Z urich, arbeitete uber Relativitats-
und Quantentheorie und stellte 1924 das nach ihm benannte
Ausschliessungsprinzip auf (Pauli-Prinzip); erhielt 1945 den
Nobelpreis f ur Physik. c _2000 Bibliographisches Institut &
F. A. Brockhaus AG
5.2. CONSTRUCTING N-PARTICLE WAVE FUNCTIONS 69
=
_
1 bosons
(1)
P
fermions
, (5.11)
where P is the number of pairwise exchanges
generating the set
1
,
2
, . . . ,
N
based on
1
,
2
, . . . ,
N
. The normalization constant K
follows as usual:
=
()
[
()
)
= [K
[
2
()
1
,
2
,...,
N
[
[
1
,
2
,...,
N
) .
Fermions require =
t
, because all
i
are dis-
tinct. The remaining sum consists of N! terms
7
of
the form [K
[
2
i
[
i
) = [K
[
2
. Therefore
we have K
=
1
N!
. Bosons do not require =
t
,
because this time the
i
need not be distinct. Sup-
pose therefore n
of the
i
(= ) in
1
, . . . ,
N
are
identical. In this case we have
1 =
(+)
[
(+)
) = [K
+
[
2
N!n
! .
The general form of this equation is
1 =
(+)
[
(+)
) = [K
+
[
2
N!
! ,
from which follows: K
+
= (
_
N!
!)
1
. Thus,
we may represent both normalization constants via
K = K
+
= K
=
1
_
N!
!
, (5.12)
where the product for fermions always is equal to
unity.
Let us look at an example to see whether this
really works. Here we set N = 2. We obtain
[
()
) =
1
_
2!
!
([
1
,
2
) [
2
,
1
))
and therefore
7
N! is the number of all possible permutations of N dis-
tinct objects.
()
[
()
) =
1
2
!
_
1
,
2
[
1
,
2
)
1
,
2
[
2
,
1
)
2
,
1
[
1
,
2
)
+
2
,
1
[
2
,
1
)
_
.
For
1
,=
2
follows

! = 1!1! = 1 and thus
()
[
()
) =
1
2
_
1
,
2
[
1
,
2
)
. .
=1
+
2
,
1
[
2
,
1
)
. .
=1
_
= 1 .
For bosons
1
=
2
is allowed. In this case we have
! = 2!0! = 2 and thus
(+)
[
(+)
) =
1
4
(1 + 1 + 1 + 1) = 1 .
In the next section we need the completeness re-
lations for the ()-subspaces
1
()
=
1
N!
1
,...,
N
[
()
)
()
[ , (5.13)
which are analogous to the completeness relations
used above (cf. (2.3) and (2.4)). The proof is left
as an exercise.
The ideal quantum gas:
As an application of what we have said thus far
we want to work out the classical limit of the canon-
ical partition function given by Eq. (2.82) for an
ideal gas. In order to keep things simple we set
N = 2 for the moment. We therefore have
H =
h
2
2m
_
k
2
1
+
k
2
2
_
. (5.14)
Here h
k
i
is the momentum operator corresponding
to particle i. Accordingly we write for the 2-particle
states
[
()
k
) = K
([k
1
, k
2
) [k
2
, k
1
)) . (5.15)
Using this we calculate the trace in Eq. (2.82), i.e.
Q = Tr
_
e
1
_
=
n
[e
1
[
n
)
=
n
[1
()
e
1
[
n
)
()
=
1
2!
n
_
V
(2)
3
_
2
_
d
3
k
1
d
3
k
2
n
[
()
k
)
()
k
[e
1
[
n
)
=
1
2
_
V
(2)
3
_
2
_
d
3
k
1
d
3
k
2
()
k
[e
1
[
()
k
)
=
1
2
_
V
(2)
3
_
2
_
d
3
k
1
d
3
k
2
e
h
2
2m
(k
2
1
+k
2
2
)
()
k
[
()
k
)
(*): Here we have inserted the completeness rela-
tion (5.13) as
1
()
=
1
2!
k
1
,
k
2
[
()
k
)
()
k
[ .
The index

k represents the set

k
1
,
k
2
. Using the
rule (1.27) for transforming the summation into an
integration we write the above relation as
1
()
=
1
2!
_
V
(2)
3
_
2
_
d
3
k
1
d
3
k
2
[
()
k
)
()
k
[ .
Note that V is the volume to which the gas is con-
ned.
In order to transform the integral into a more
useful form we now work out the r-representation
of
()
k
[
()
k
). Using the terminology developed in
the last section we may write
()
k
[
()
k
) = K
2
k
1
,
k
2
[
k
1
,
k
2
)
k
1
,
k
2
[
k
2
,
k
1
)
k
2
,
k
1
[
k
1
,
k
2
)
+
k
2
,
k
1
[
k
2
,
k
1
)
_
= K
2
_
d
3
r
1
d
3
r
2
_
k
1
,
k
2
[r
1
, r
2
)
r
1
, r
2
[
k
1
,
k
2
)
k
1
,
k
2
[r
1
, r
2
)r
1
, r
2
[
k
2
,
k
1
)
k
2
,
k
1
[r
1
, r
2
)r
1
, r
2
[
k
1
,
k
2
)
+
k
2
,
k
1
[r
1
, r
2
)r
1
, r
2
[
k
2
,
k
1
)
_
.
Here we have used the relation (2.3) twice to-
gether with [r
1
)[r
2
) [r
1
, r
2
); i.e. r
i
, r
j
[
k
i
,
k
j
) =
r
i
[
k
i
)r
j
[
k
j
). We may now be tempted to use Eq.
(2.6), i.e. r
i
[
k
i
) =
1
(2)
3/2
e
i
k
i
r
i
. However, this
r-representation of a [
k)-state (free particle) was

based on an innite integration volume. In turn
it represents a free particle state, which cannot be
normalized, i.e. the free particles is not localized
in a particular region in space. Our gas, however,
is conned to the volume V . In order to nd the
proper r
i
[
k
i
) in this case we write down the nor-
malization condition for the [
k)-states in a box of
volume V using the r-representation:
k
=
k
t
[
k) =
_
d
3
r
k
t
[ r)r[
k) .
From
1
V
_
V
d
3
re
i(
)r
=
k,
(cf. problem 4) we
nd
r[
k) =
1
V
e
i
kr
. (5.16)
Using Eq. (5.16) we obtain
()
k
[
()
k
) = K
2
_
d
3
r
1
V
d
3
r
2
V
_
2
e
i(
k
1
k
2
)(r
1
r
2
)
e
i(
k
1
k
2
)(r
1
r
2
)
_
.
This we substitute into the above expression for Q
and obtain
Q =
K
2
2!
2
_
d
3
r
1
(2)
3
d
3
r
2
(2)
3
d
3
k
1
d
3
k
2
e
h
2
2m
(k
2
1
+k
2
2
)
_
1 e
i(
k
1
k
2
)(r
1
r
2
)
_
5.3. ELECTRON SPIN 71
=
1
2!
_
d
3
r
1
(2)
3
d
3
r
2
(2)
3
d
3
k
1
d
3
k
2
(5.17)
e
h
2
2m
(k
2
1
+k
2
2
)
_
1 f
2
(r
12
)
_
with r
12
= [r
1
r
2
[ as well as
f(r
12
) =
_
d
3
k exp
_
h
2
k
2
2m
+i
k r
12
_
_
d
3
k exp
_
h
2
k
2
2m
_
= exp
_
r
2
12
2
T
_
. (5.18)
Here
T
=
_
2h
2
/m is the thermal wavelength.
For N > 2 the calculation is analogous. The result
is
Q =
1
N!
_
d
3
r
1
. . . d
3
r
N
d
3
p
1
. . . d
3
p
N
(2h)
3N
(5.19)
exp
_
i=1
p
2
i
2m
_
_
1
i<j
f
2
(r
ij
)
+
i,j,k
f(r
ij
)f(r
ik
)f(r
kj
) . . .
_
,
where the quantum corrections are expressed in
terms of an expansion involving the number of
particle exchanges (cf. reference [17]). The lead-
ing correction corresponds to our result obtained
above. Again (+) means bosons and () means
fermions.
The classical limit of the partition function fol-
lows from f(r) 0. The quantum corrections for
f(r) > 0 become important only if the particles are
on average closer that the thermal wave length
T
,
i.e. for
T
(V/N)
1/3
(5.20)
the classical limit is a very good approximation.
When the particles approach each other, whether
they are bosons or fermions, they feel a statistical
interaction potential, which to leading order looks
like this:
1
i<j
f
2
ij

i<j
_
1 f
2
ij
_
= exp
_
_
i<j
u
ij
_
_
- 1
0
1
2
3
4
0 0.2 0.4 0.6 0.8 1 1.2
r
ij
/
T
u
ij
( )
fermions
bosons
Figure 5.1: The statistical potential between
bosons and fermions, respectively.
with
u
ij
k
B
T ln
_
1 f
2
ij
_
(5.21)
= kT ln
_
1 exp
_
2 [ r
i
r
j
[
2
2
T
__
This statistical potential, which arises solely as a
consequence of the symmetry of the state function,
is shown in gure 5.1. We see that bosons feel an
attraction, whereas fermions feel repulsion at dis-
tances below or roughly equal to the thermal wave-
length. Because
T
(mT)
1/2
, the potentials ef-
fect will become important especially for light par-
ticles and at low temperatures.
5.3 Electron spin
Examining the spectrum of for instance sodium
(Na), one discovers that the strongest yellow line
(the D line) is actually two closely spaced lines.
The D line arises from a transition from a par-
ticular excited conguration to the ground state.
The doublet nature of this and other lines in the
Na spectrum indicates a doubling of the number
of states available to the valence electron predicted
by Schrodingers equation as we know it. To ex-
plain this ne structure of atomic spectra, Uhlen-
beck and Goudsmit
8
in 1925 proposed an intrinsic
angular momentum,

S, in addition to the orbital
angular momentum,

L. In nonrelativistic quan-
tum mechanics, to which we are conning ourselves,
electron spin must be introduced as an additional
hypothesis. In 1928 Dirac developed the relativistic
quantum mechanics of electrons, where the electron
spin arises naturally.
Analogous to the orbital angular momentum op-
erators,

L
2
, L
x
, L
y
, and L
z
, we postulate the spin
angular momentum operators

S
2
, S
x
, S
y
, and S
z
,
which are assumed to be linear and hermitian. We
also postulate that the spin angular momentum op-
erators obey the same commutator relations as the
orbital angular momentum operators:
_
S
i
, S
j
= ih
ijk
S
k
(5.22)
and
_
S
2
, S
i
_
= 0 (5.23)
for i = 1, 2, 3. Analogous to the angular momen-
tum quantum numbers, l and m, we now introduce
the spin quantum numbers, s and m
s
. Experiment
shows that for electrons s =
1
2
(they are fermions)
and thus m
s
=
1
2
(cf. section 1.5). Protons and
neutrons also have s =
1
2
. Pions have s = 0. Pho-
tons have s = 1. However, because photons are
relativistic particles m
s
= 1 (but not m
s
= 0)!
These two m
s
values correspond to left circularly
polarized and right circularly polarized light
9
.
For electrons the magnitude of the total spin an-
gular momentum is given by
_
1
2
_
3
2
_
h
2
_
1/2
=
3
2
h . (5.24)
The electronic spin eigenstates that correspond to
the S
z
-eigenvalues are denoted [) and [):
8
Goudsmit, Samuel Abraham, amerikanischer Physiker
niederlandischer Herkunft, *Den Haag 11.7.1902, Reno
(Nevada) 4.12.1978; arbeitete v.a. uber die Feinstruktur von
Atomspektren, postulierte zusammen mit G.E. Uhlenbeck
den Elektronenspin; leitete 1944 ein Geheimunternehmen
zur Auskundschaftung des deutschen Atombombenprojekts.
9
In section 6.1 we discus the photon in more detail.
S
z
[) = +
1
2
h [) (5.25)
and
S
z
[) =
1
2
h [) (5.26)
10
. Because S
z
does commute with

S
2
both [) and
[) are also eigenstates of

S
2
:
S
2
[) =
3
4
h
2
[) (5.27)
and
S
2
[) =
3
4
h
2
[) . (5.28)
Notice that [) = 0, because [) and [) belong
to dierent eigenvalues of the hermitian operator
S
z
.
Addition of angular momenta:
In order to learn how to interpret the details of
atomic and also molecular spectra we must learn
how to add angular momenta
11
.
Suppose we have a system with two angular mo-
mentum vectors

J
1
and

J
2
. They might be the
orbital angular momentum vector of a single elec-
tron and its spin
12
, but other combinations like
two spins or two orbital angular momentum vec-
tors are also possible. We already know that the
individual eigenvalues of

J
2
1
,

J
2
2
, J
1,z
, and J
2,z
are
h
2
j
1
(j
1
+ 1), h
2
j
2
(j
2
+ 1), hm
1
, and hm
2
. These
quantum numbers obey the usual restrictions.
Now we dene the total angular momentum
J =

J
1
+
J
2
. (5.29)
The square of

J is given by
10
The terms spin up and spin down refer to m
s
=
1
2
.
11
For a many electron atom, the operator for individual
angular momenta of the electrons do not commute with the
Hamiltonian operator, but their sum does. Hence we want
to learn how to add angular momenta.
12
the case of interest here
J
2
=

J
2
1
+
J
2
2
+
J
1
J
2
+
J
2
J
1
. (5.30)
If

J
1
and

J
2
refer to dierent electrons, they will
commute with each other, since each will aect only
functions of the coordinates of one electron and not
the other. Even if

J
1
and

J
2
are the orbital and
spin angular momenta of the same electron, they
will commute, as one will aect only functions of
the spatial coordinates while the other will aect
functions of the spin coordinates. Thus Eq. (5.30)
becomes
J
2
=

J
2
1
+
J
2
2
+ 2
J
1
J
2
. (5.31)
It only takes some algebra to show that

J obeys the
usual angular momentum commutator relations (cf.
(1.140) and (1.141)). Thus we can simultaneously
quantize

J
2
and one of the components of

J, say
J
z
. Since the components of the total angular mo-
mentum obey the angular momentum commutator
relations, we know, based on the work in section
1.5, that the eigenvalues of

J
2
are
h
2
j (j + 1) j = 0,
1
2
, 1,
3
2
, . . . , (5.32)
and the eigenvalues of

J
z
are
hm
J
m
J
= j, j + 1, . . . , j 1, j . (5.33)
What we need to nd out is how the total angular
momentum quantum numbers, j and m
J
, are re-
lated to the quantum numbers j
1
, j
2
, m
1
, and m
2
of the two original angular momenta.
In order to motivate this we want to briey look
at how the coupling between

L and

S gives rise
to a contribution to the Hamilton operator. Such a
term should be a scalar quantity, because it must be
invariant under rotation and inversion. The quan-
tities, which appear in this term should be p, the
momentum of the electron, its spin,

S, and the po-
tential energy, | (r). Because p is a polar vector
and

S, as angular momentum, is an axial vector,
we need a cross product to build up the invariant
term. The simplest form is
H
L
= K
S
_
| (r) p
_
. (5.34)
We cannot determine the constant K, but Diracs
relativistic theory yields K =
_
2m
2
c
2
_
1
, where m
is the electron mass. Using

| (r) =
/(r)
r
r=r
r
r
(radial symmetry) we nd for (5.34)
H
L
=
1
2m
2
c
2
r
r
| (r) (
L) (5.35)
13
. Clearly, this term will lead to a level splitting,
because its magnitude depends on the relative ori-
entations of

S and

L. According to Eqs. (5.31) and
(5.24) the eigenvalues of

S
L are given by
13
We may also rationalize the form of 1
L
on the basis
of classical electrodynamics. Remember Biot-Savarts law
(cf. Ref. [15]; section 3.3):
H (r) =
1
c
_

j
_
_
r
_
[ r
[
3
d
3
r
.
Here

H (r) is the magnetic eld at position r due to a station-
ary charge current density

j(
). For an observer travelling

with the electron around the nucleus the latter circulates
around the electron creating a magnetic eld
H (0) =
1
c
_

j
_
r
3
d
3
r
at the origin (where the electron is). Using

j =
n
v
n
, the
index n refers to the nucleus, we obtain for a circular orbit
H (0) =
1
c
E
n
v
n
.
Here

E
n
is the electric eld of the nucleus felt by the elec-
tron. With v
n
= p/m, where p and m are momentum and
mass of the electron, we have
H (0) =
1
mc
E
n
p =
1
mec
| p
where e is the electron charge and
| = e
E
n
is the
force on the electron. Finally we use
1
L
= m
e

H (0) =
1
mec
m
e
| p
_
,
where m
e

S is the magnetic moment of the electron.

Notice that according to classical electrodynamics a charge,
e, with mass, m
e
, orbiting in a circular loop produces a
(loop) magnetic moment, m
e
, related to its angular momen-
tum,

L
e
, by m
e
=
e
2m
e
c
L
e
(cf. Eq. (1.111)).
h
2
2
_
j (j + 1) l (l + 1)
3
4
_
. (5.36)
Thus we are back to our question at the end of the
preceding paragraph. But before we continue, here
is a nal remark.
If we substitute (the eective) Bohrs radius a
0
=
h
Zmc
, where Z is the charge number and is the
ne structure constant, for r
1
in Eq. (5.35) we
nd H
L
) (Z)
2
in units of the unperturbed
energy
14
. For Sodium (Na) this is 7 10
3
,
which is of the same order as the above mentioned
level splitting which leads to the proposal of an elec-
tron spin. This type of level spitting is small, and
therefore it belongs to the so called ne structure
of the quantized energy levels. Other ne structure
eects are the coupling of the magnetic moment of
the spin to the orbital momentum, the relativistic
mass eect of the electron, and the so called Dar-
win term. This term arises because the path of a
particle is not sharply dened, and thus it feels the
potential not only at position r. These corrections
correspond to higher derivatives of | (r). Other
eects are still smaller by about a factor of 10
3
;
these are called hyperne structure. An example is
the eect due to the nuclear magnetic moment.
Again the question: How are the quantum num-
bers j and m
J
are related to j
i
and m
i
(i = 1, 2)?
First note that the angular momentum eigenfunc-
tions for particle 1 are [ j
1
, m
1
) and those for par-
ticle 2 are [ j
2
, m
2
). Forming all possible products
of the form [ j
1
, m
1
) [ j
2
, m
2
) yields a complete
set of functions for the two particles [ j
1
, j
2
, j, m
J
),
because it can be shown that the four operators
corresponding to these quantum numbers do com-
mute.
Each unknown eigenstate [ j
1
, j
2
, j, m
J
) can be
expanded according to
[ j
1
, j
2
, j, m
J
) =
m
1
,m
2
c (j
1
, j
2
, j, m
J
; m
1
, m
2
)
[ j
1
, m
1
) [ j
2
, m
2
), (5.37)
where the c (. . . ; . . .) are expansion coecients.
These are called Clebsch-Gordan or Wigner or vec-
14
Use the substitution e
2
Ze
2
in order to transform to
hydrogen-like atoms.
tor addition coecients. We do not want to evalu-
ate them here
15
, but we need the form of Eq. (5.37)
to answer the above question. This is because using
J
z
= J
1,z
+J
2,z
one can prove (homework problem)
that c (. . . ; . . .) = 0 unless
m
1
+m
2
= m
J
. (5.38)
Knowing this we need to nd the possible
values of the total angular momentum quantum
number j. Before discussing the general case, let
us consider the case with j
1
= 1, j
2
= 2. The
following table lists the resulting possible values
for m
1
, m
2
, and m
J
:
m
1
= 1 0 +1
3 2 1 2 = m
2
2 1 0 1
1 0 +1 0
0 +1 +2 +1
+1 +2 +3 +2
The number of times each value of m
J
occurs is
m
J
-values 3 2 1 0 1 2 3
occurrence 1 2 3 3 3 2 1
The highest value of m
J
is +3. Since m
J
ranges
form j to +j, the highest value of j must be 3.
Corresponding to j = 3, there are seven values of
m
J
ranging from 3 to 3. Eliminating these seven
values, we are left with
m
J
-values 2 1 0 1 2
occurrence 1 2 2 2 1
The highest remaining value, m
J
= 2, must
correspond to j = 2; for j = 2, we have ve values
of m
J
, which when eliminated leave
m
J
-values 1 0 1
occurrence 1 1 1
These remaining values of m
J
correspond to j = 1.
Thus, for the individual angular momentum quan-
tum numbers j
1
= 1, j
2
= 2, the possible values
of the total angular momentum quantum number j
are 3, 2, and 1.
15
see E. Merzbacher (1970) Quantum Mechan-
ics (2nd ed.); Section 16.6. The Clebsch-Gordan
coecients are also supplied by Mathematica:
ClebschGordan[j
1
, m
1
, j
2
, m
2
, j, m
J
] gives c(. . . ; ..)
as dened in Eq. (5.37).
A generalization of this line of argument yields
j = j
1
+j
2
, j
1
+j
2
1, . . . , [ j
1
j
2
[ (5.39)
(cf. [22]; sec. 11.4). In particular if j
1
= l and
j
2
=
1
2
we have j = l +
1
2
, [ l
1
2
[. Thus,
except for l = 0, we nd that H
L
in Eq. (5.34)
gives rise to a doublet splitting, which led Uh-
lenbeck and Goudsmit to propose the electron spin!
Spin magnetic moment:
Because spin itself is a relativistic phenomenon,
we cannot expect that the relation (1.127), which
we found in the case of orbital angular momentum,
is valid if L
z
is replaced by S
z
. In fact, Diracs
theory yields
m
s
= g
e
e
2mc
S . (5.40)
Here e is the magnitude of the electron charge, m is
its mass, and g
e
= 2 is the gyromagnetic factor for
the electron. Theoretical and experimental work
subsequent to Diracs treatment has shown that g
e
is slightly larger than 2 (see P. Kusch, Physics To-
day, Feb. 1966, p. 23):
g
e
= 2
_
1 +

2
+. . .
_
= 2.0023 ,
where . . . stands for terms containing higher powers
of the ne structure constant .
Zeeman eect:
In classical electrodynamics we have learned that
in a constant magnetic eld

H we have

A =
1
2
Hr
(cf. [15]; footnote 24 in section 3.3). Therefore
A p =
1
2
_
H r
_
p
=
1
2

H
_
r p
_
=
1
2

H
L . (5.41)
In addition we use Eq. (3.64) in [15] to write down
the potential energy of the spin magnetic moment
in the eld

H:
m
s

H
(5.40)
=
e
2mc
2
S

H . (5.42)
If we now combine Eqs. (4.75) (setting = 0, e
e for the electron and neglecting the two last
terms), (5.35), and (5.42) we obtain
H =
p
2
2m
+ | (r) +
e
2mc
_
L + 2
S
_

H
+
1
2m
2
c
2
r
r
| (r)
_
L
_
. (5.43)
If in addition we insert | (r) = Ze
2
/r for a
hydrogen-like atom we nally get
H =
p
2
2m

Ze
2
r
+
Ze
2
2m
2
c
2
r
3
L

H
with
=
e
2mc
_
L + 2
S
_
.
We now introduce an operator whose eigenvalue is
the Lande factor by requiring
= G
J (5.44)
via multiplication with

J we nd
G =
e
2mc
_
1 +

J

S
J
2
_
=
e
2mc
_
1 +

J
2
+

S
2
L
2
2
J
2
_
. (5.45)
If the magnetic eld strength

H in Eq. (5.43) is
weak compared to the

S

L-term we may apply
perturbation theory using the unperturbed states
[ n, l,
1
2
, j, m
J
). For

H = (0, 0, H) we obtain the
level splitting
E
nj
= n, l,
1
2
, j, m
J
[

H [ n, l,
1
2
, j, m
J
)
=
ehH
2mc
gm
J
, (5.46)
where the Lande factor is
g = 1 +
j (j + 1) l (l + 1) + 3/4
2j (j + 1)
. (5.47)
Notice that 2j + 1-fold degeneracy is completely
gone, which is called anomalous Zeeman eect.
Actually, this terminology has historical reasons,
because the ordinary Zeeman eect corresponds to
g = 1 (i.e. spin zero).
Remark: For strong magnetic elds one observes
a dierent type of level splitting, because the

S
L
coupling becomes negligible and each moment in-
teracts independently with the eld (Paschen-Back
eect.
5.4 Simplied theory for he-
lium
As an application of what we have learned so far
we consider the Hamiltonian
H =
1
2m
_
p
2
1
+ p
2
2
_
2e
2
_
1
[ r
1
[
+
1
[ r
2
[
_
+
e
2
[ r
1
r
2
[
. (5.48)
This simple model for helium is based on the Born-
Oppenheimer approximation
16
, which considers
the position of the nuclei as xed. The approxima-
tion is a good one, because the nuclei are so much
heavier than the electrons. Secondly, spin does not
appear here explicitly. Nevertheless, as we have dis-
cussed above, it enters in terms of a spin part of the
total state vector, which must be anti-symmetric.
In order to nd the quantum numbers character-
izing the helium states we have to nd the operators
which commute with each other. Here it is useful
to write the permutation operator
12
as product
12
= R
12
s
12
, (5.49)
16
Oppenheimer, Julius Robert, amerikanischer Atom-
physiker, *New York 22.4.1904, Princeton (New Jersey)
18.2.1967; 1947-66 Direktor des Institute for Advanced
Study in Princeton. Arbeiten zur relativistischen Quan-
tentheorie und Kernphysik. Unter seiner wissenschaftlichen
Leitung wurde 1943-45 die erste Atombombe in Los Alamos
hergestellt. Nach Kriegsende sprach er sich aus technis-
chen und moralischen Gr unden gegen den Bau der Wasser-
stobombe und f ur eine internationale Kontrolle der Kern-
waen aus. Nach einer Anhorung wegen kommunistis-
cher Umtriebe wurde ihm 1954 die Erlaubnis entzogen, an
geheimen Projekten mitzuarbeiten; 1963 rehabilitiert. Lit-
eratur: Homann, K.: J. R. Oppenheimer. Sch opfer der
ersten Atombombe. Berlin u.a. 1995. c _2000 Bibliographis-
ches Institut & F. A. Brockhaus AG
where s
12
acts on the spins and R
12
on the remain-
ing properties. We note that both, R
12
and s
12
,
commute with H. In addition, s
12
does not com-
mute with the individual spin operators

S
1
and

S
2
,
but it commutes with the total spin
S =

S
1
+

S
2
. (5.50)
The same is true for R
12
(or
12
) with respect to
the total orbital momentum
L =

L
1
+

L
2
. (5.51)
A system of commuting operators therefore consists
of H,
12
, s
12
,

L
2
, L
z
,

S
2
, and S
z
. Notice that this
means: the two angular momenta are coupled and
the two spins are coupled; but there is no cross-
coupling.
First we consider the spin part of the total state
function. The single-spin states are written as [
s
i
, m
i
), i.e. [
1
2
,
1
2
) [) and [
1
2
,
1
2
) [). Using
the terminology introduced in Addition of angular
momenta we may write
[ s, m
s
) [
1
2
,
1
2
, s, m
s
) (5.52)
for the total spin state. According to the above
rules for the addition of two angular momenta we
thus obtain the following four orthogonal spin state
functions:
[ 1, 1) =[)
[ 1, 0) =
1
2
([)+ [))
[ 1, 1) =[)
_
_
symmetric (+)(5.53)
[ 0, 0) =
1
2
([) [)) antisymmetric () (5.54)
Here we have also applied our knowledge about the
symmetry properties of the state function, which
must be either symmetric (+) or antisymmetric
(). Note that even though electrons are fermions
we must also consider the symmetric spin states
at this point, because it is the total state function
which is () for fermions.
5.4. SIMPLIFIED THEORY FOR HELIUM 77
Now we want to consider the remaining part of
the total state function. To simplify matters we
momentarily neglect the electron-electron Coulomb
interaction (5.48). The dimensionless eigenvalues
are then given by
=
Z
2
2
_
1
n
2
1
+
1
n
2
2
_
, (5.55)
where Z = 2 in this case
17
. To build up the total
state function given by
[ n
1
, n
2
, l, m
l
, s, m
s
) = [ n
1
, n
2
, l, m
l
; s)
[ s, m
s
) , (5.56)
we construct the rst factor, [ n
1
, n
2
, l, m
l
, ; s), from
products of single particle states,
[ n
1
, l
1
, m
l
1
) [ n
2
, l
2
, m
l
2
) , (5.57)
consistent with our approximation expressed in
Eq. (5.55).
These linear combinations of products (5.57)
must, however, fulll two requirements: (a) they
must be eigenstates of

L
2
and L
z
; (b) they must
obey the correct exchange symmetry. The sim-
plest case, in which we directly see what is going
on with respect to these requirements, is n
1
= 1,
l
1
= m
l
1
= 0, i.e. l
2
= l , m
l
2
= m
l
, and n
2
n.
We then have
[ n
1
, n
2
, l, m
l
; s)
=
1
2
_
[ 1, 0, 0)
1
[ n, l, m
l
)
2
[ 1, 0, 0)
2
[ n, l, m
l
)
1
_
(5.58)
18
. In order for the total state to be antisymmetric
we must combine the three symmetric spin states
19
with the antisymmetric () state function in (5.58)
or the antisymmetric spin state
20
with the sym-
metric (+) state function in (5.58). Obviously the
ground state, n
1
= n
2
= 1 and l = m
l
= 0, only
17
Notice that here is in hartree!
18
The notation . . .)
i
or
i
. . . replaces the ordered product
convention introduced on page 67. The index i corresponds
to the position in the ordered product.
19
One also calls them triplet states.
20
This case is denoted as singlet state.
exists as singlet state, because the antisymmetric
combination is zero. There are only two electrons
maximum on the inner shell, which dier in the
direction of their spins.
If we now include the electron-electron interac-
tion termH
ee
=
e
2
]r
1
r
2
]
, we see that H
ee
is diagonal
in the representation which we have chosen here
21
.
The expectation value of H
ee
consists of two terms:
1, n
2
, l, m
l
, s, m
s
[ H
ee
[ 1, n
2
, l, m
l
, s, m
s
)
= (5.59)
D A .
D, given by
D =
1
1, 0, 0 [
2
n, l, m
l
[ H
ee
[ 1, 0, 0)
1
[ n, l, m
l
)
2
,
is the direct Coulomb interaction, corresponding to
the screening of the Coulomb eld of the nucleus by
the inner electron. A, given by
A =
1
1, 0, 0 [
2
n, l, m
l
[ H
ee
[ 1, 0, 0)
2
[ n, l, m
l
)
1
,
is due to the exchange of the electrons. It is called
exchange interaction. In general this term is posi-
tive. This is why the triplet states (ortho-helium)
are found at lower energies than the correspond-
ing singlet states (para helium). This may be un-
derstood intuitively as follows: In the triplet state
the spin part of the state function is symmetric
and thus the spatial part is anti-symmetric. There-
fore the electrons avoid each other, and thus the
Coulomb repulsion between them is reduced.
In problem 29 we had calculated the ground state
energy of helium using a simple variation Ansatz.
We had obtained E 2.85
me
4
h
2
= 77.1eV . Ex-
perimentally one nds 78.9eV . Notice that even
though we did not pay attention to the symme-
try properties of the wave function at the time of
problem 29 we obtained the correct result, because
symmetry does not make a dierence for the ground
state.
It is also interesting to calculate the ground state
energy via rst order perturbation theory. In this
case
21
Note that 1
ee
commutes with all operators whose eigen-
values characterize the states.
=
Z
2
+Z
2
2
+D
ma
2
0
h
2
(5.60)
In r-representation D becomes D = e
2
J
1s1s
with
J
1s1s
=
Z
6
2
a
6
0
_
d
3
r
1
d
3
r
2
e
2Zr
1
/a
0
1
r
12
e
2Zr
2
/a
0
(5.61)
This is an example of a Coulomb integral, which
has the general form
J
ij
=
_
d
3
d
3
t
[
i
() [
2
1
[
t
[
[
j
(

R
ij
) [
2
. (5.62)
Here R
ij
is the separation of the two nuclei on
which the orbital functions
i
and
j
are centered.
In the present case R
ij
= 0. For
i
() =
(2
i
)
2n+1
4 (2n)!

n1
e
, (5.63)
the so called Slater orbitals, one obtains
J
i
ns
j
n
s
=
1
R
ij
4
2n+1
i

2n
+1
j
(2n)! (2n
t
)!
(5.64)
d
2n2
d
2n
2
d
2n2
i
d
2n
2
j
1
3
i

3
j
_
1
_
3
2
i

2
j
_
4
j
(
i
j
)
3
(
i
+
j
)
3
e
2
j
R
ij
2
i
3
2
j
_
4
i
(
i
j
)
3
(
i
+
j
)
3
e
2
j
R
ij
4
j
(
i
j
)
2
(
i
+
j
)
2
R
ij
e
2
i
R
ij

4
i

j
(
i
j
)
2
(
i
+
j
)
2
R
ij
e
2
j
R
ij
_
.
In the special case n = n
t
= 1 we nd
J
i
1s
j
1s
x
=
=
e
2R
ij
R
ij
(5.65)
_
e
2R
ij
1
11
8
R
ij
3
4
2
R
2
ij

1
6
3
R
3
ij
_
=
_
1
R
ij
(R
ij
)
5
8

1
12
3
R
2
ij
+. . . (R
ij
0)
(cf. Kapitel V in R. Hentschke et
al. Molekulares Modellieren mit Kraft-
feldern. http://constanze.materials.uni-
wuppertal.de/skripten).
Thus we obtain from (5.65) for R
ij
= 0
D =
5
8
Z
a
0
e
2
(5.66)
and therefore
= Z
2
+
5
8
Z = 2.75 , (5.67)
which is E = 2.75
me
4
h
2
74.4eV . Overall the
agreement is not too bad. As one might have
expected, the variation approach yields the better
result.
5.5 Simplied theory for the
hydrogen molecule
In this section we consider an approximation of the
hydrogen molecule dating essentially back to the
early days of quantum mechanics (Heitler and Lon-
don, 1927). The idea is as follows. We consider the
nuclei, which are heavy and slow compared to the
electrons, at xed positions

R
A
and

R
B
. The hy-
drogen molecules Hamiltonian then becomes
H
H
2
=
p
2
1
2m
+
p
2
2
2m
e
2
_
1
r
A1
+
1
r
A2
(5.68)
+
1
r
B1
+
1
r
B2
1
r
12
1
R
AB
_
Here r
Ai
is the distance of electron i from nucleus
A (analogous for r
Bi
), and r
12
and R
AB
are the
5.5. SIMPLIFIED THEORY FOR THE HYDROGEN MOLECULE 79
electron-electron and nuclear separations, respec-
tively.
Our goal will be the calculation of the electron
distribution and the electronic energy levels. Hav-
ing solved this problem, we may compute the forces
between the nuclei and thus their vibrational mo-
tion (cf. below).
The electronic problem cannot be solved exactly.
We, in particular, will use a simple approximation,
i.e. the H
2
electronic state function is taken as
[
III/I
) = [ s, m
s
)f
III/I
(5.69)
_
[
A
)
1
[
B
)
2
[
B
)
1
[
A
)
2
_
.
Here [ s, m
s
) is the spin part dened in Eq. (5.52)
and written down explicitly in Eqs. (5.53) and
(5.54). Notice that the [
A
) and [
B
) are sim-
ply atomic hydrogen states associated with nucleus
A and B. Notice also that the ()-sign corresponds
to the symmetric spin triplet states (III), and the
(+)-sign corresponds to the antisymmetric spin sin-
glet states (I). The factor f
III/I
is obtained from
the normalization condition
III/I
[
III/I
) = 1,
i.e.
1 = [ f
III/I
[
2
(5.70)
_
1
A
[
A
)
1
. .
=1
2
B
[
B
)
2
. .
=1
+
2
A
[
A
)
2
. .
=1
1
B
[
B
)
1
. .
=1
A
[
B
)
12
B
[
A
)
2
B
[
A
)
12
A
[
B
)
2
_
.
The
i
A
[
B
)
i
do not vanish, because the r-
representations of [
A
) and [
B
) are centered on
dierent nuclei. The quantities
K
AB
(R
AB
)
1
A
[
B
)
1
=
_
d
3
r
A
[ r)r [
B
) (5.71)
are called overlap integrals. Consequently
f
III/I
=
1
_
2 (1 [ K
AB
[
2
)
. (5.72)
x
A
A

B
R
AB
e
-
z
B
1
0
Figure 5.2: Coordinates used in the example com-
putation of the overlap integral (5.73).
In order to clarify the notation we compute K
AB
based on the hydrogen ground state wave functions
(1.103). For simplicity we use Bohrs radius as the
unit of length, i.e.
1
A
[
B
)
1
=
1
_
d
3
e
A
e
B
. (5.73)
The electron-to-nucleus separations
A
and
B
are
shown in Fig. 5.2. The integral may be solved
using elliptical coordinates , , and : = (
A
+
B
)/R
AB
and = (
A
B
)/R. Note that
A
is the
separation of one particular electron (here electron
1) from nucleus A, and
B
is the separation of the
same electron from nucleus B. is the angle of
rotation around the line connecting A and B. In
elliptic coordinates the volume element is d
3
=
(R
3
AB
/8)(
2
2
)ddd (1 , 1 1,
0 2). After a not too dicult integration
we obtain
1
A
[
B
)
1
=
_
1 +R
AB
+
R
2
AB
3
_
e
R
AB
. (5.74)
At this point we may pursue dierent avenues.
Based on [
III
) and [
I
) we may either do pertur-
bation theory or, introducing again as in the case of
helium an eective charge, we may take the varia-
tion approach. Here, however, we simply calculate
the expectation values
E
I
=
I
[ H
H
2
[
I
) (5.75)
= E
A
+E
B
+
D +A
1+ [ K [
2
and
E
III
=
III
[ H
H
2
[
III
) (5.76)
= E
A
+E
B
+
D A
1 [ K [
2
,
where
D(R
AB
) = e
2
1
A
[
2
B
[
_
1
r
12
+
1
R
AB
1
r
A2
1
r
B1
_
[
A
)
1
[
B
)
2
(5.77)
is the direct Coulomb energy, and
A(R
AB
) = e
2
1
A
[
2
B
[
_
1
r
12
+
1
R
AB
1
r
A2
1
r
B1
_
[
A
)
2
[
B
)
1
(5.78)
is an exchange interaction. Computing these
expressions in r-representation yields, for the
ground state (A) = (B) = (1, 0, 0), the curves
shown in gure 5.3 (cf. problem 37). Notice that
I corresponds to the symmetric spatial part of the
wave function, whereas III corresponds to the
anti-symmetric part. The latter yields a smaller
probability for nding the electrons, between the
nuclei. Here this leads to a larger nuclear repulsion.
In the other case there is an attraction, which
yields a minimum and thus a stable chemical bond.
The Hellmann-Feynman theorem:
Consider a system with a time-independent
Hamiltonian H that involves parameters
i
(i = 1, . . .). An example is the harmonic
oscillator, where the force constant is such a
parameter. It is then relatively easy to prove the
generalized Hellmann-Feynman theorem:
E
n
=
_

n
H

n
d
3
r . (5.79)
Here the
n
and E
n
are eigenfunctions and eigen-
values of H, which in this case depends only on one
parameter, . Notice that
n
also depends on
(cf. [22]; section 14.3)!
1 2 3 4 5 6
Ra
o
1.6
1.8
2.2
2.4
2.6
2.8
3
e
Figure 5.3: Simplied ground state theory for H
2
(cf. problem 37). The bonding curve corresponds
to the singlet (I) and the non-bonding curve to the
triplet (III) spin state. Note that the exchange in-
tegral here was replaced by [K[
2
/a, where a is a pa-
rameter. Solid lines: a = 100; dashed lines: a = 5.
An important application of Eq. (5.79) is the
electrostatic theorem. This theorem, which is based
on the Born-Oppenheimer approximation, reads:
T
i
=
i
_
i<j
q
i
q
j
r
ij
(5.80)
+
_

E
(; l)
q
i
[ r
i
[
d
3
_
.
Here q
i
is the charge on nucleus i, and
E
(; l) is
the quantum electron density in the electronic state
l. Thus, the eective force acting on a nucleus i in
a molecule,

T
i
, can be calculated by simple elec-
trostatics as the sum of the Coulombic forces ex-
erted by other nuclei and by a hypothetical electron
cloud, whose charge density,
E
(; l), is found by
solving the electronic Schrodinger equation. Again,
the reader is referred to [22] (section 14.4) for the
details. Notice that this theorem opens up a way
to do dynamic calculations at temperatures T > 0
(cf. R. Hentschke et al., Molekulares Modellieren
mit Kraftfeldern).
5.6 Quantum mechanics in
chemistry
In the late 1920s Dirac remarked: The underlying
physical laws necessary for the mathematical the-
5.6. QUANTUM MECHANICS IN CHEMISTRY 81
ory of a large part of physics and the whole of chem-
istry are . . . completely known, and the diculty is
only that the exact application of these laws leads
to equations much too complicated to be solved.
In 1998 Walter Kohn and John Anthony Pople re-
ceived the Nobel prize in Chemistry for their con-
tribution to the falsication of the latter part of
Diracs statement. Both Kohn and Pople were hon-
ored for their work on (computational) quantum
chemistry made possible due to the advances in
computer technology, which of course Dirac could
not have foreseen.
22
Quantum mechanics describes molecules in
terms of interactions among nuclei and electrons,
and molecular geometry in terms of minimum en-
ergy arrangements of nuclei. All quantum me-
chanical methods ultimately trace back to the
Schrodinger equation, which for the special case of
the hydrogen atom may be solved exactly (as we
have done in section 1.4). Wave functions for the
hydrogen atom are the s, p, d, . . . atomic orbitals,
familiar to most chemists. The square of the wave
function times a small volume gives the probabil-
ity of nding the electron inside the volume. This
is termed the total electron density (or more sim-
ply the electron density), and corresponds to the
electron density measured in an X-ray diraction
experiment. It is straightforward to generalize the
Schrodinger equation to a multinuclear, multielec-
tron system:
H = E (5.81)
Here, is a many-electron wave function, and H
is
H =
h
2
2
nuclei
A
1
M
A
2
A
h
2
2m
electrons
2
a
22
The following section largely is taken from W.J. Hehre,
J. Yu, P. E. Klunzinger, and L. Lou (1998) A brief guide to
Molecular Mechanics and Quantum Chemical calculations.
wave function, Inc. It is supposed to provide a brief outline
of Quantum Chemistry for beginners without going into
details, which is not possible in a single course on introduc-
tory quantum theory. Nevertheless, I have added additional
detail indicated by (*) and I continue to do so as these
lecture notes evolve. For the interested reader I recommend
references [22] and [23]. In Wuppertal courses in quantum
chemistry currently are oered by Prof. Jensen in Chem-
istry.
e
2
nuclei
A
electrons
a
Z
A
r
Aa
(5.82)
+e
2
nuclei
A > B
Z
A
Z
B
R
AB
+e
2
electrons
a > b
1
r
ab
The rst two terms in (5.82) describe the kinetic en-
ergy of the nuclei, A, and the electrons, a, respec-
tively, and the last three terms describe Coulom-
bic interactions between particles. M are nuclear
masses, and R
AB
, r
ab
and r
Aa
are distances sep-
arating nuclei, electrons, and nuclei and electrons,
respectively.
Unfortunately, the many-electron Schrodinger
equation cannot be solved exactly (or at least
has not been solved) even for the simplest many-
electron system. Approximations need to be
introduced to provide practical methods.
Born-Oppenheimer approximation:
One way to simplify the Schrodinger equation
for molecular systems is to assume that the nu-
clei do not move. This is termed the Born-
Oppenheimer approximation, and leads to an elec-
tronic Schrodinger equation:
H
el
el
= E
el
el
, (5.83)
where
H
el
=
h
2
2m
electrons
2
a
e
2
nuclei
2
a
electrons
a
Z
A
r
Aa
+e
2
electrons
a > b
1
r
ab
.
The term in (5.82) describing the nuclear kinetic
energy is missing in (5.83) (it is zero), and the
nuclear-nuclear Coulomb term is a constant. It
needs to be added to the electronic energy, E
el
,
to yield the total energy, E, for the system.
E = E
el
+e
2
nuclei
A > B
Z
A
Z
B
R
AB
Note that there is no reference to (nuclear) mass
in the electronic Schrodinger equation. Mass
eects (isotope eects) on molecular properties
and chemical reactivities are of dierent origin.
Even with the Born-Oppenheimer approximation,
the Schr odinger equation is not solvable for more
than a single electron. Additional approximations
need to be made.
()
Spatial orbitals and spin orbitals:
To completely describe an electron, it is neces-
sary to specify its spin. A complete set for describ-
ing the spin of an electron in the following consists
of the two orthonormal functions (w) and (w);
i.e. spin up () and spin down ()
23
. The wave
function for an electron that describes both its spa-
tial distribution and its spin is a spin orbital, (x),
where x indicates both space and spin coordinates
(x = r, w). From each spatial orbital, (r) , one
can form two dierent spin orbitals, i.e.
(x) =
_
(r) (w)
(r) (w)
Given a set of K spatial orbitals,
i
[ i =
1, 2, . . . , K, one can thus form a set of 2K spin
orbitals
i
[ i = 1, 2, . . . , 2K via
2i1
(x) =
i
(r) (w)
2i
(x) =
i
(r) (w)
where i = 1, 2, . . . , K.
()
Slater determinants:
We now consider wave functions for a collection
of electrons, i.e., N-electron wave functions. We
begin by dening a Hartree product:
HP
(x
1
, x
2
, . . . , x
N
) =
i
(x
1
)
j
(x
2
) . . .
k
(x
N
) .
Notice that
HP
is the solution of
H
HP
= E
HP
,
23
(w) and (w) are analogous to the r-representation of
the spatial wave function.
where
H =
N
i=1
h(i)
and
E =
i
+
j
+. . . +
k
.
Here h(i) is the Hamilton operator describing elec-
tron i.
The Hartree product does not satisfy the an-
tisymmetry principle, however. We can obtain
correctly antisymmetrical wave functions using so
called Slater determinants(SD). For instance in the
two electron case we may write
(x
1
, x
2
) =
1
i
(x
1
)
j
(x
1
)
i
(x
2
)
j
(x
2
)
(5.84)
=
1
2
_
i
(x
1
)
j
(x
2
)
i
(x
2
)
j
(x
1
)
_
.
The generalization to N electrons is straightfor-
ward:
(x
1
, x
2
, . . . , x
N
) =
1
N!
i
(x
1
)
j
(x
1
)
k
(x
1
)
i
(x
2
)
j
(x
2
)
k
(x
2
)
.
.
.
.
.
.
.
.
.
i
(x
N
)
j
(x
N
)
k
(x
N
)
.
Interchanging the coordinates of two electrons cor-
responds to interchanging two rows in the SD,
which changes its sign. Notice also that having
two electrons occupying the same spin orbital cor-
responds to having two identical rows in the SD
which makes it zero. Thus, the SD meets both the
antisymmetry requirement and the Pauli principle.
It is convenient to introduce a short-hand nota-
tion for a normalized SD:
(x
1
, x
2
, . . . , x
N
) (5.85)
= [
i
(x
1
)
j
(x
2
)
k
(x
N
)) .
If we always choose the electron labels to be in the
order, x
1
, x
2
, x
3
, . . . , x
N
, then Eq. (5.85) can be
further shortened to
(x
1
, x
2
, . . . , x
N
) = [
i
j
. . .
k
)
24
.
()
Exchange correlation:
We consider the 2-electron wave function,
(x
1
, x
2
) = [
1
(x
1
)
2
(x
2
)) ,
to compare two situations
a) the two electrons have opposite spins and oc-
cupy dierent spatial orbitals:
1
(x
1
) =
1
(r
1
) ( w
1
)
2
(x
2
) =
2
(r
2
) (w
2
) . (5.86)
b) the two electrons have the same spin (say )
and occupy dierent spatial orbitals:
1
(x) =
1
(r
1
) (w
1
)
2
(x) =
2
(r
2
) (w
2
) . (5.87)
We are interested in the quantity
P (r
1
, r
2
) d
3
r
1
d
3
r
2
, which is the simultaneous
probability of nding electron one in the volume
element d
3
r
1
at r
1
and electron two in the volume
element d
3
r
2
at r
2
:
P (r
1
, r
2
) d
3
r
1
d
3
r
2
=
_
dw
1
dw
2
[ [
2
d
3
r
1
d
3
r
2
.
Using Eq. (5.84) we have
[ [
2
=
1
2
1
(r
1
) (w
1
)
2
(r
2
) (w
2
)
2
( r
1
) (w
1
)
1
(r
2
) (w
2
)
2
.
Now we apply
_
dw
(w) (w) =
_
dw
(w) (w) = 0
24
This terminology, which we analyse in this chapter, is
not to be confused with the multiparticle state vectors in-
troduced previously!
to obtain for (a)
P (r
1
, r
2
) d
3
r
1
d
3
r
2
=
1
2
_
[
1
(r
1
) [
2
[
2
(r
2
) [
2
+ [
2
(r
1
) [
2
[
1
(r
2
) [
2
_
d
3
r
1
d
3
r
2
.
In particular for
1
=
2
we nd
P (r
1
, r
2
) =[
1
( r
1
) [
2
[
1
(r
2
) [
2
.
Here P (r
1
, r
2
) ,= 0 even if the electrons have the
same r. This is completely dierent in case (b),
which is
P (r
1
, r
2
) =
1
2
_
[
1
(r
1
) [
2
[
2
(r
2
) [
2
+ [
2
(r
1
) [
2
[
1
(r
2
) [
2
1
(r
1
)
2
(r
1
)
2
(r
2
)
1
(r
2
)
+
1
(r
1
)
2
(r
1
)
2
(r
2
)
1
(r
2
)
_
.
Notice that now P (r
1
, r
2
) = 0 if
1
=
2
as expected. In addition we also nd
P (r
1
, r
2
r
1
) 0. A Fermi hole is said to
exist around an electron. The electrons are corre-
lated.
A simple example consists of two spinless non-
interacting fermions in the potential well of the 1D
harmonic oscillator. The ground state wave func-
tion of this system is
0
(q
1
, q
2
) =
1
2
_
0
(q
1
)
1
(q
2
)
0
(q
2
)
1
(q
1
)
_
,
where
0
(q) =
1
1/4
exp
_
q
2
/2
and
1
(q) =
1/4
q
2
exp
_
q
2
/2
(cf. (1.63); q =
_
mw/hx). We are interested
in the probability density P
F
(q), that the two
fermions are separated by a distance q, i.e.
P
F
(q) =
_

dq
1
dq
2
[
0
(q
1
, q
2
) [
2
q=const
=
1
2
_

dQ
2
0
(q, Q) .
Here q = q
1
q
2
and Q = q
1
+ q
2
. Notice also
that dqdQ = 2dq
1
dq
2
. Using
0
(q
1
, q
2
) =
1
1/2
(q
1
q
2
) exp
_
1
2
_
q
2
1
+q
2
2
_
_
together with q
2
+Q
2
= 2
_
q
2
1
+q
2
2
_
we nd
P
F
(q) =
1
2
q
2
exp
_
1
2
q
2
_
Fig. 5.4 shows P
F
(q) vs. q. For q
the oscillator potential governs the behavior of
P
F
(q). The minimum at q = 0, however, is
an eect entirely due to the exchange symmetry of
fermions. The same gure shows P
B
(q), the anal-
ogous probability density for two non-interacting
spinless bosons. Here the ground state wave func-
tion is
0
(q
1
, q
2
) =
1
2
_
0
(q
1
)
0
(q
2
) +
0
(q
2
)
0
(q
1
)
_
.
No Fermi hole is observed for the bosons.
Hartree-Fock approximation:
The most obvious simplication to the
Schrodinger equation involves separation of vari-
ables, that is, replacement of the many-electron
wave function by a product of one-electron wave
functions. The simplest acceptable replacement,
termed a Hartree-Fock
25
or single-determinant
wave function, involves a single determinant
of products of one-electron spin orbitals. The
25
Hartree, Douglas Rayner, britischer Physiker und Math-
ematiker, *Cambridge 27.3.1897, ebenda 12.2.1958; en-
twickelte zusammen mit dem russischen Physiker W.A.Fock
(*1898, 1974) die Hartree-Fock-Methode, ein quantenmech-
anisches Verfahren zur n aherungsweisen Berechnung der
Wellenfunktionen und Eigenwerte von Mehrelektronensys-
temen (z.B. in Atomen). c _2000 Bibliographisches Institut
-4 -2 2 4
q
0.05
0.1
0.15
0.2
0.25
0.3
P
F
-4 -2 2 4
q
0.1
0.2
0.3
0.4
P
B
Figure 5.4: Top: Illustration of the Fermi hole for
two spinless and non-interacting fermions in a 1D
harmonic oscillator potential. Bottom: The analo-
gous probability density, P
B
, plotted for two spin-
less and non-interacting bosons.
Hartree-Fock approximation leads to a set of
coupled dierential equations (the Hartree-Fock
equations), each involving a single electron. While
they may be solved numerically, it is advantageous
to rst introduce one additional approximation.
()
The Hartree-Fock theory is based on the vari-
ation method which we have discussed in section
4.4. Our goal is it to nd a set of spin orbitals
a
such that the single determinant formed from these

spin orbitals [
0
) =[
1
2
. . .
a
b
. . .
N
_
is the
best possible approximation to the ground state of
the N-electron system described by the electronic
Hamiltonian H
el
. According to the variation prin-
ciple, the spin orbitals are those which minimize
the electronic energy
E
0
=
0
[ H
el
[
0
)
=
a
a [ h [ a) +
1
2
a,b
[aa [ bb] [ab [ ba] ,
where
[ij [ kl] :=
_
dx
1
dx
2
i
(x
1
)
j
(x
1
) r
1
12

k
(x
2
)
l
(x
2
) .
Using the variation method we can systematically
vary the spin orbitals
a
, constraining them only
to the extend that they remain orthonormal,
a
[
b
) =
ab
until E
0
is a minimum. The resulting equation is
the Hartree-Fock integro-dierential equation
h(1)
a
(1) +
b(,=a)
__
dx
2
[
b
(2) [
2
r
1
12
_
a
(1)
b(,=a)
__
dx
2
b
(2)
a
(2) r
1
12
_
b
(1)
=
a
a
(1) ,
where
h(1) =
1
2
2
1
a
Z
A
r
1A
is the kinetic energy and potential energy for
attraction to the nuclei, of a single electron chosen
to be electron-one. The orbital energy of the spin
orbital
a
is
a
.
LCAO approximation:
This follows from the notion that the one-
electron solutions for many electron molecules will
closely resemble the (one-electron) solutions for
the hydrogen atom. This seems entirely reason-
able. Since molecules are made up of atoms, why
shouldnt molecular solutions be made up of atomic
solutions? In practice, the molecular orbitals are
expressed as linear combinations of a nite set (a
basis set) of prescribed functions known as basis
functions, :
i
=
basis functions
c
i
. (5.88)
y
x
z
R
1
R
2
R
12
r
r-R
2 r-R
1
Figure 5.5: Coordinate system for minimal basis
H
2
.
c are the molecular orbital coecients, often re-
ferred to simply (and incorrectly) as the molecular
orbitals. Because the are usually centered at the
nuclear positions (although they do not need to be),
they are referred to as atomic orbitals, and expan-
sion (5.88) is termed Linear Combination of Atomic
Orbitals or LCAO approximation.
()
We illustrate the LCAO approximation us-
ing the hydrogen molecule. Each hydrogen atom
has a 1s orbital and, as the two atoms approach,
molecular orbitals (MOs) may be constructed via
linear combination of the atomic orbitals (LCAO).
The coordinate system is shown in Fig. (5.5). The
rst atomic orbital,
1
=
1
(r

R
1
), is centered
on nucleus 1 at

R
1
. The second atomic orbital,
2
=
2
(r

R
2
), is centered on nucleus 2 at

R
2
.
The exact 1s orbital of a hydrogen atom centered
at

R has the form
(r

R) = (
3
/)
1/2
exp[[r

R[]
(cf. Eq. (1.103)). Here , the orbital exponent,
has a value of 1.0. This is an example of a Slater
orbital. In praxis, however, Gaussian orbitals are
used, because the evaluation of the relevant inte-
grals is computationally less demanding. The 1s
Gaussian orbital has the form
(r

R) = (2/)
3/4
exp[[r

R[
2
] ,
where is the Gaussian orbital exponent. Here we
need not be concerned with the particular form of
the 1s atomic orbitals. The two atomic orbitals
1
and
2
can be assumed to be normalized, but they
will not be orthogonal. They will overlap, and the
overlap integral is
K
12
=
_
d
3
r
1
(r)
2
(r)
(cf. Eq. (5.71)). From the two localized atomic
orbitals,
1
and
2
one can form, by linear com-
bination, two delocalized molecular orbitals. The
symmetric combination leads to a bonding MO of
gerade symmetry,
1
=
1
_
2(1 +K
12
)
(
1
(r) +
2
(r))
whereas the antisymmetric combination leads to an
antibonding MO of ungerade symmetry
1
=
1
_
2(1 K
12
)
(
1
(r)
2
(r)) .
Using only two basis functions for H
2
is an example
of a minimal basis set and an obvious choice for the
two functions
1
and
2
is the 1s atomic orbitals of
the atoms. The correct linear combination for this
simple choice are determined by symmetry.
Given the two spatial orbitals
1
and
2
, we can
form four spin orbitals
1
(x) =
1
(w)
2
(x) =
1
(w)
3
(x) =
2
(w)
4
(x) =
2
(w) .
The orbital energies associated with these spin or-
bitals must be obtained via explicit calculation.
But, as might be expected,
1
and
2
are degener-
ate and have the lower energy corresponding to a
bonding situation, while
3
and
4
are also degen-
erate having a higher energy corresponding to an
antibonding situation. The Hartree-Fock ground
state in this model is the single determinant
[
o
) =[
1
2
_
3

4
1

2
o
=
=
* *
=
=
=
Figure 5.6: Dierent representations of the
Hartree-Fock ground state of minimal basis H
2
.
shown pictorially in Fig. 5.6.
Roothaan-Hall equations:
The Hartree-Fock and LCAO approximations,
taken together and applied to the electronic
Schrodinger equation, lead to the Roothaan-Hall
equations:
basis functions
(F

i
K
) c
i
= 0 .
Here, are orbital energies, K is the overlap matrix
(a measure of the extent to which basis functions
see each other), and F is the Fock matrix, which
is analogous to the Hamiltonian in the Schrodinger
equation. In atomic units it is given by
F
[
1
2
nuclei
A
Z
A
r
A
[
)+
basis functions
,
P
1
2
)
_
. (5.89)
The rst term in (5.89) accounts for the kinetic and
potential energies of individual electrons, while the
second term accounts for interactions among elec-
trons. P is the so-called density matrix, the ele-
ments of which involve the square of the molecular
orbital coecients summed over all occupied molec-
ular orbitals:
P
= 2
occupied molecular orbitals
i
c
i
c
i
Methods resulting from solution of the Roothaan-
Hall equations are termed Hartree-Fock or ab initio
models. The corresponding energy for an innite
(complete) basis set is termed the Hartree-Fock
energy. The Hartree-Fock energy is not equal to
the experimental energy.
Correlated models:
Hartree-Fock models treat the motions indi-
vidual electrons as independent of one another.
Because of this, electrons get in each others way
to a greater extent than they should. This leads to
overestimation of the electron-electron repulsion
energy and to too a high a total energy. Electron
correlation, as it is termed, accounts for coupling
of electron motions, and leads to a lessening of the
electron-electron repulsion energy (and to a low-
ering of the overall total energy). The correlation
energy is dened as the dierence between the
Hartree-Fock energy and the experimental energy.
Moller-Plesset models:
A number of methods have been developed to
account for electron correlation. With the excep-
tion of so-called density functional methods (see
discussion following), these generally involve mix-
ing the ground state (Hartree-Fock) wave function
with excited state wave functions. Operationally,
this entails implicit or explicit promotion of elec-
trons from molecular orbitals which are occupied
in the Hartree-Fock wave function to molecular or-
bitals which are unoccupied.
empty molecular
orbitals
filled molecular
orbitals
electron
promotion
.
.
.
.
.
.
Among the simplest practical schemes are Moller-
Plesset models which are formulated in terms of a
generalized electronic Hamiltonian, H
, i.e.
H
= H
0
+J . (5.90)
H
0
is dened such that dierent states do not in-
teract. The perturbation J is dened according
to
J = (HH
0
) . (5.91)
H is the correct Hamiltonian and is a dimension-
less parameter.
and E
, the wave function and

energy for a system described by the Hamiltonian
H
, may then be expanded in powers of , i.e.
=
(0)
+
(1)
+
2
(2)
+. . . (5.92)
E
= E
(0)
+E
(1)
+
2
E
(2)
+. . . . (5.93)
The Moller-Plesset energy to rst order is the
Hartree-Fock energy. The second-order Moller-
Plesset energy, E
(2)
, is given by
E
(2)
=
molecular orbitals
lled empty

i < j a < b
[(ij [[ ab)]
2
(
a
+
b
j
)
.
i
,
j
are energies of lled molecular orbitals,
a
,
b
energies of empty molecular orbitals, and integrals
(ijab) account for changes in electron-electron in-
teractions as a result of promotion. Moller-Plesset
theory terminated to second-order, or MP2, is per-
haps the simplest model based on electron promo-
tion which oers improvement over Hartree-Fock
theory. Higher-order models (MP3, MP4, etc.)
have been formulated, but in practice are limited
in application to very small systems.
()
Fig. 5.7 shows the interaction energy of two
methane molecules as function of the distance be-
tween their nuclei. This interaction energy can be
obtained experimentally from thermodynamic data
on the second virial coecient of gaseous methane.
The latter can be computed using the Lennard-
Jones potential (4.28). In the case of methane one
4 5 6 7 8
r
-1.5
-1
-0.5
0.5
1
1.5
2
EkJmol^1
Figure 5.7: Comparison of the interaction en-
ergy of two methane molecules as function of the
separation of their nuclei. Upper set of points:
Hartree-Fock; lower set of points: MP2; solid curve
Lennard-Jones potential.
obtains for the parameters = 1.2364 kJmol
1
and = 3.79

A (cf. Anwendungsbeispiel III.2
in R. Hentschke et al. Molekulares Modellieren
mit Kraftfeldern. http://constanze.materials.uni-
wuppertal.de/skripten). Here we compare the
result to a Hartree-Fock and a MP2 calculation
performed with a commercial quantum chemistry
package called Spartan (version 4; cf. Fig. 5.8).
In both cased the 6 31G
basis set is used.

We see that MP2 already yields a signicant
improvement over Hartree-Fock. The results may
be improved further by more extended basis sets
or higher order MP-approximation. Note that the
interaction energy displayed here is only about
10
4
of the total electronic energy of the molecules!
Density functional models:
Density functional models provide an alternative
approach to the treatment of correlation in many-
electron systems. They follow from a theorem,
which states that the minimal energy of a collec-
tion of electrons under the inuence of an external
(Coulombic) eld is a unique functional (a func-
tion of a function) of the electron density. To see
how this theorem is put to practical use, rst con-
sider the energy of a many-electron system as given
by Hartree-Fock theory:
E
HF
= E
nuclear
+E
core
Figure 5.8: Computer screen shot showing the
mathane molecules as displayed by Spartan.
+E
Coulomb
+E
exchange
.
Here, E
nuclear
accounts for Coulombic repulsion of
nuclei, E
core
for both electron kinetic energy and
Coulombic attraction between electrons and nu-
clei, and E
Coulomb
for Coulombic repulsion of elec-
trons. E
exchange
also involves electron-electron inter-
actions, and has the eect proceed, of reducing the
size of the Coulomb term. The energy according
to density functional theory includes the same nu-
clear, core and Coulomb terms as the Hartree-Fock
energy, yet replaces the Hartree-Fock exchange en-
ergy by an exchange functional, E
X
(), and adds
a correlation functional, E
C
().
E
DFT
= E
nuclear
+E
core
+E
Coulomb
+E
X
() +E
C
()
Both of the latter are functions of the electron
density, . The simplest models are called local
spin density or local density models, and will
be referred to as SVWN (Slater, Vosko, Wilk,
Nusair) models. Here, the form of the exchange
and correlation functionals follows from the exact
(numerical) solution of a many-electron gas of
uniform density, as a function of the density, by
subtracting E
core
and E
Coulomb
from the total
energy (there is no E
nuclear
term). The SVWN
model would not be expected to be satisfactory
for molecular systems where the electron density
is non-uniform. The model may be improved by
introducing explicit dependence on the account
gradient of the electron density, in addition to the
density itself. Such mirror procedures are termed
gradient-corrected or non-local density model,
functional models. An obvious way to do this, for
the case where the density is a slowly varying
function, is in terms of a simple models power-
series expansion. Unfortunately, such an approach
leads to divergent behavior in the limit of large
density gradients. An alternative is the so-called
generalized gradient approximation those proposed
by Becke and Perdew. In the Becke-Perdew (BP
or BP86) model, a new potential is used in place
of the local potential in the self-consistent-eld
(SCF) procedure. This actually comprises a local
part and a gradient correction. The latter needs
to be recalculated at every SCF iteration. An
alternative, and computationally simpler approach,
is to introduce the gradient correction only after
convergence based on the local potential alone
has been achieved. This procedure, referred to
as a perturbative Becke-Perdew model (or pBP),
oers signicant computational advantages over
the corresponding self-consistent procedure. It
yields nearly identical structures and relative
energies to the full self-consistent BP model. The
equations which result from both local density
and self-consistent Becke-Perdew formulations
may be solved iteratively by expanding the set of
one-particle wave functions in terms of a set of
basis functions centered on the individual nuclei.
Note, however, that terms associated with the
exchange and correlation functionals typically
involve forms for which analytical expressions of
the required integrals are either not available,
or would be too complicated to deal with were
they available. It is usually necessary to resort to
numerical integration techniques. Because of this,
calculated quantities show larger uncertainties
than those from Hartree-Fock Coulomb and MP2
models. Total (and relative) energies should be
accurate to 0.0002 hartrees (on the order of a tenth
of a kcal/mol) and bond lengths accurate to 0.005
A. Because density functional models take electron

correlation into account in an explicit manner, re-
sulting properties should more closely mirror those
of conventional correlated models, such as the
MP2 model, than those of Hartree-Fock models.
That is to say, density functional results should
be better than those from Hartree-Fock models.
Because the energy is expressed as a function
of a single variable (the electron density), the
equations which result from application of density
functional theory are conceptually simpler than
those resulting from Hartree-Fock theory (where
the variables are the set of one-electron wave
functions). In fact, a density functional calculation
scales (formally) as the cube of molecular size,
compared to a (formal) fourth power dependence
on size for a Hartree-Fock calculation. In practice,
both density functional and Hartree-Fock calcu-
lations scale more slowly than the formal rules
suggest.
Basis sets for HF and MP2 calculations:
For practical reasons, Hartree-Fock (HF) and
MP2 calculations make use of basis sets of
Gaussian-type functions (cf. above). These are
closely related to exact solutions of the hydrogen
atom, and comprise a polynomial in the Cartesian
coordinates (x, y, z) followed by an exponential
in r
2
. Several series of Gaussian basis sets have
now received widespread use and are thoroughly
documented. Underlying these, are minimal (STO-
3G) and split-valence (3-21G, 6-31G and 6-311G)
representations. These may be supplemented by
sets of polarization and/or diuse functions.
STO-3G minimal basis set:
The simplest possible atomic orbital representa-
tion is termed a minimal basis set. This comprises
only those functions required to accommodate all
of the electrons of the atom, while still maintain-
ing its overall spherical symmetry. In practice, this
involves a single (1s) function for hydrogen and he-
lium, and a set of ve functions (1s, 2s, 2p
x
, 2p
y
,
2p
z
) for lithium to neon. Note that, while 2p func-
tions are not occupied in the lithium and beryllium
atoms, they are required to provide proper descrip-
tions in molecules.
Each of the basis functions in the STO-3G rep-
resentation is expanded in terms of three Gaus-
sians, where the values of the Gaussian exponents
and the linear coecients have been determined by
least squares as best ts to Slater-type (exponen-
tial) functions.
There are two obvious shortcomings of a min-
imal basis set. The rst is that all basis func-
tions are either themselves spherical (s functions)
or come in sets which, taken together, describe a
sphere (p functions). As a consequence, molecules
which incorporate only atoms with spherical envi-
ronments or nearly spherical environments will
be better described than molecules which incorpo-
rate atoms with aspherical environments.
The second shortcoming follows from the fact
that basis functions are atom centered. While this
is not absolutely necessary, there are no other ob-
vious (unambiguous) locations for the basis func-
tions. The choice of atom-centered locations for the
basis functions does, however restrict their exibil-
ity to describe electron distributions between nuclei
(bonds).
Chapter 6
The quantized radiation eld and
second quantization
The following chapter contains merely a glimpse
of what the title promises. Its main purpose is to
take a look ahead at the next steps in the devel-
opment of a quantum theory treating particles and
elds on an equal footing.
6.1 Quantum theory of radia-
tion
So far we have treated radiation elds as classical
elds. Now we want to develop a quantum theo-
retical description. In order to nd a good starting
point we return to section 1.2, where we dealt with
the black body problem. Notice that there is a
striking similarity between Eq. (1.21), the classi-
cal eld energy in the black body cavity expressed
in terms of the Fourier-coecients c
k,
, and the
Hamilton operator of the harmonic oscillator ex-
pressed in terms of the number operator a
+
a dis-
cussed in section 1.5! Thus, why not start by postu-
lating that the c
k,
become operators c
k,
obeying
the commutator relations
_
c
k,
, c
+
_
=
(6.1)
and
_
c
k,
, c
_
=
_
c
+
k,
, c
+
_
= 0 . (6.2)
In order to write down the Hamilton operator de-
scribing the radiation eld we do not use (1.21)
directly. Instead we use the expression,
E =
k,
2
4c
2
_
c
k,
+c
k,
c
k,
_
, (6.3)
which in problem 4 was combined to obtain (1.21),
because we now pay attention to (6.1). In addi-
tion we absorb a factor
_

2c
2
h
in c
k,
and c
+
k,
,
i.e.
_

2c
2
h
c
k,
c
k,
and
_

2c
2
h
c
k,
c
+
k,
, to
obtain the Hamilton operator
H =
k,
h
2
_
c
+
k,
c
k,
+c
k,
c
+
k,
_
(6.4)
(6.1)
=
k,
h
_
c
+
k,
c
k,
. .
n
k,
+
1
2
_
.
Notice that

k,
h
1
2
is the zero-point energy of
the radiation eld, which, as we have discussed al-
ready, is innite. In our treatment of the black
body radiation this did not matter, but there are
cases when it does
1
. The quantity n
k,
is a num-
ber operator counting the number of photons of
type (
k, ) in the volume (of the cavity) at time t.

The state vector is given by
[ . . . , n
k,
, . . . , n
, . . .) , (6.5)
i.e.
1
see appendix C
91
92 CHAPTER 6. THE QUANTIZED RADIATION FIELD AND SECOND QUANTIZATION
c
+
k,
[ . . . , n
k,
, . . .) =
_
n
k,
+ 1 [ . . . , n
k,
+ 1, . . .) (6.6)
creates a photon, analogously c
k,
, annihilates a
photon, whereas
n
k,
[ . . . , n
k,
, . . .) = n
k,
[ . . . , n
k,
, . . .) (6.7)
counts their number (cf. section 1.5). From Eq.
(6.7) we see that the photons in this formalism are
bosons, because we can generate states for n
k,
=
0, 1, 2, . . ..
This quantization procedure looks appealing.
But does it work? As a rst check we want to
nd the momentum operator and its eigenvalues.
From classical electrodynamics we know that the
Poynting-Vector,

S =
c
4
E

H, represents the ra-
diation energy per area and time leaving a volume
containing the radiation eld (cf. Eq. (2.58) in ref-
erence [15]). Consequently, because
]
S]
c
is a energy
density and using the energy-momentum relation
E = pc for photons, we get
p =
_
V
d
3
r
E

H
4c
(6.8)
for the classical momentum of the electromagnetic
eld. Via a calculation completely analogous to the
one in problem 4 and after replacing the Fourier
coecients by operators, as we have done above,
we nd
p =
k,
h
kc
+
k,
c
k,
=
k,
h
kn
k,
. (6.9)
Obviously the eigenvalues of this momentum oper-
ator, which follow from
p [ . . . , n
k,
, . . .) =
k
t
n
[ . . . , n
k,
, . . .)
=
k
t
n
[ . . . , n
k,
, . . .) , (6.10)
are

k,
h
kn
k,
. This is consistent with our
understanding of the radiation eld as a collection
of photons with momentum h
k and polarization
direction described by .
Remark 1: Notice that the relativistic energy-
momentum relation (cf. [15], section 1.2) is given
by
m
2
=
1
c
4
_
E
2
[ p [
2
c
2
_
photon
=
1
c
4
_
(h)
2
_
h [
k [ c
_
2
_
= 0 , (6.11)
i.e., the rest mass of the photon is zero.
Remark 2: The photon state is characterized not
just by its momentum but also by the polarization
vectors
()
. From classical electrodynamics (cf.
[15]; section 4.1) we know that
()
=
1
2
_
(1)
i
(2)
_
(6.12)
characterizes circular polarized light. Under an in-
nitesimal rotation around

k, the propagation
direction,
()
changes by
()
=
2
_
(2)
i
(1)
_
= i
()
. (6.13)
We see this by applying the rotation matrix
D =
_
_
cos sin 0
sin cos 0
0 0 1
_
_
_
_
1 0
1 0
0 0 1
_
_
to
()
, i.e.
D
()

1
2
_
_
_
_
1
0
_
_
i
_
_
1
0
_
_
_
_
=
()
2
_
(2)
i
(1)
_
=
()
i
()
. .
=
()
,
6.1. QUANTUM THEORY OF RADIATION 93
where
(1)
is in x-direction,
(2)
is in y-direction,
and

k is in z-direction. In problem 17 we had seen
that the angular momentum operator is the gen-
erator of rotation. Carrying this idea over to Eq.
(6.13) we associate with
()
the photons spin com-
ponent m = 1, where the quantization axis has
been chosen in the propagation direction. If
()
were along

k, we would associate m = 0 (since

k
is unchanged under innitesimal rotation about

k);
however, the m = 0 state is missing in the expan-
sion of

A (cf. problem 4) because of the transver-
sality condition

k
()
= 0. In other words, the
photon spin is either parallel or antiparallel to the
propagation direction.
Notice that the description of the polarization
state with
()
as the base vectors is called cir-
cular polarization representation. We could have
used
()
instead of
(1)
and
(2)
to expand

A in
problem 4. A single photon state with denite cir-
cular polarization can be constructed by applying
the creation operator
c
+
k,
=
1
2
_
c
+
k,1
ic
+
k,2
_
(6.14)
to the vacuum state. Conversely, c
k,
[ 0) with
= 1, 2 and

k in z-direction can be regarded
as a 50/50 mixture of the m = 1 and m = 1 state.
Fluctuations and uncertainty:
It is easy to check that the quantities

A,

E, and
H do not commute with n
k,
. As a consequence the
number of photons and the aforementioned quan-
tities cannot be measured simultaneously and ex-
actly as we had shown in section 2.2. Explicitly
we may take

E =
1
c

A
t
and calculate the vacuum
uctuation, i.e. no photon is present, of the electric
eld:
0 [

E

E [ 0) [ 0 [

E [ 0) [
2
(6.15)
2
. Here 0 [

E [ 0) vanishes because of c
k,
[ 0) = 0,
but 0 [

E

E [ 0) turns out to be innite for an
unbounded volume (why?
3
). Thus, the expression
2
Notice that [
H[ = [
E[ (cf. section 4.1 in Ref. [15].

Therefore we may write
1
2
0[
E
2
+

H
2
[0) = 0[
E
2
[0).
3
cf. appendix C
(6.15) is innite as expected from the uncertainty
principle. However, it is more sensible to look at
a nite volume V = l
3
, and consider averages
over this volume,

E =
1
V
_
V
Ed
3
r. One can then
show that
0 [

E [ 0)
hc
l
4
(6.16)
(cf. [24]; section 2.3). We do not want to prove
(6.16), but we note that the left side is an energy
density, which typically should be
h
l
3
. Using
= 2 = 2c/l we obtain the right side in
Eq. (6.16). The signicance of (6.16) is that it
tells us: A classical radiation eld must have an
energy density much larger than
hc
l
4
in order
for the quantum uctuations to be unimportant.
Expressing the classical energy density, [
E[
2
(cf.
[15]; section 2.3), in terms of nh, where n is
the average number of photons in the volume
l
3
, we see that the classical description requires
n 1/l
3
.
Emission and absorption of photons by atoms:
In section 4.3 we had considered emission and
absorption using classical radiation elds. The ba-
sic matrix element was f [ H
1
[ i), where H
1
, is
given by Eq. (4.80). The f [ and i [ were atomic
states.
Let us now consider the absorption of a light
quantum characterized by

k, . An atom which
is initially in a state a makes a radiative transition
to state b. We assume, for simplicity, that there
are only photons of type (
k, ) present. If there
are n
k,
photons in the initial state, then there are
n
k,
1 photons in the nal state. Even though the
vector potential

A contains both c
k,
and c
+
k,
, only
c
k,
gives a non-vanishing matrix element (due to
. . . n
k,
i . . . [ . . . n
k,
j . . .) = 0 for i ,= j).
For a similar reason the term proportional to

A

A
in the Hamiltonian (4.75), which we had neglected,
also yields no contribution! It changes the total
number of photons by either 0 or 2. Thus we
have
b; n
k,
1 [
. .
f]
H
1
[ a; n
k,
)
. .
=]i)
= (6.17)
=
e
mc
V
b; n
k,
1 [ c
k,
e
i(
krwt)
()
p [ a; n
k,
)
=
e
m
_
n
k,
V
b [ e
i
kr
p
()
[ a)e
iwt
.
At this point we are back to the semi-classical treat-
ment of induced emission and absorption in section
4.3. The most signicant dierence between our
old calculation and this one is that [

A
0
[
2
is re-
placed by
n
k,
V
. This corresponds to a replacement
of the classical energy density, [

E [
2
, by an en-
ergy density proportional to the number of photons
per volume. The analogous case of emission yields
a factor
_
n
k,
+1
V
instead of the factor
_
n
k,
V
.
With this simple example of an application of ra-
diation eld quantization we conclude this section.
The story continues in courses on advanced quan-
tum theory/mechanics. The interested student
should also consult texts like [24] or [6] (Spezielle
Kapitel).
6.2 Second quantization
The previous section has introduced a formalism
where particles can be created as well as annihi-
lated. Because mass is just another apparent form
of energy according to Einsteins special theory of
relativity, an obvious question is the following: Can
this formalism be generalized to include massiv par-
ticles? Can all particles in nature be described as
eld quanta? Suspecting that this eld has some-
thing to do with the wave function (r, t) we once
again start from the (non-relativistic) Schrodinger
equation:
ih
t
(r, t) (6.18)
=
h
2
2m
2
(r, t) +| (r) (r, t)
= H (r, t) .
Also once again we expand (r, t) in eigenfunctions
of H
(r, t) =
n
b
n
(t)
n
(r) . (6.19)
If we now use (6.19) to write down H) , i.e.
H) = [ H [ )
=
_
d
3
r
(r, t)
_
h
2
2m
2
+| (r)
_
(r, t) ,
we nd
H) =
n
E
n
b
n
b
n
,
where E
n
are the energy eigenvalues. Thus, we
obtain an expression which looks like the sum over
independent harmonic oscillators.
As before we now transform the expansion co-
ecients into operators, b
n
b
+
n
and b
n
b
n
,
obeying the commutator relations
_
b
n
, b
+
n
=
nn
(6.20)
and
[b
n
, b
n
] =
_
b
+
n
, b
+
n
= 0 . (6.21)
With these denitions we obtain a new Hamiltonian
H =
n
E
n
b
+
n
b
n
, (6.22)
where b
+
n
creates an oscillator quantum and b
n
an-
nihilates an oscillator quantum (analogous to c
+
k,
and c
k,
in the previous section). Because the
commutator relations (6.20) and (6.21) are identi-
cal to those for photons in the previous section the
particles created by this formalism are bosons.
This immediately brings us to the question: How
can we extend the method to include fermions?
It turns out that the only change consists in
dening the new commutator relations
b
n
, b
+
n
=
nn
(6.23)
and
b
n
, b
n
= b
+
n
, b
+
n
= 0 , (6.24)
where .., .. now means
6.3. INTERACTING QUANTUM FIELDS: 95
A, B = AB +BA . (6.25)
.., .. is called anti-commutator. To prove this
point we rst work out the relation
_
b
+
n
b
n
_ _
b
+
n
b
n
_
(6.23)
= b
+
n
_
1 b
+
n
b
n
_
b
n
= b
+
n
b
n
b
+
n
b
+
n
b
n
b
n
(6.24)
= b
+
n
b
n
+b
+
n
b
+
n
b
n
b
n
= b
+
n
b
n
. (6.26)
Now we determine the eigenvalues of the number
operator N
n
= b
+
n
b
n
:
N
n
[ ) = b
+
n
b
n
[ ) = [ )
and
N
2
n
[ ) = b
+
n
b
n
b
+
n
b
n
[ ) =
2
[ )
(6.26)
= [ ) .
Thus,
2
= and therefore = 0 or = 1. This
is exactly what we expect for fermions
4
.
We also want to work out b
+
n
[ n
n
) and b
n
[ n
n
).
We start by noting
N
n
[ n
n
) = b
+
n
b
n
[ n
n
) = n
n
[ n
n
) , (6.29)
where, as we have just shown, n
n
= 0, 1. Now we
consider b
+
n
[ n) by multiplication with N
n
, i.e.
N
n
b
+
n
[ n
n
) = b
+
n
b
n
b
+
n
[ n
n
)
= b
+
n
_
1 b
+
n
b
n
_
[ n
n
)
= b
+
n
(1 N
n
) [ n
n
)
= (1 n
n
) b
+
n
[ n
n
) .
4
We also mention that both types of commutator rela-
tions (6.20), (6.21) and (6.23), (6.24), lead to the same equa-
tions of motion for b
n
:
ih
db
n
dt
=
_
b
n
, 1
= E
n
b
n
(6.27)
or
ih
db
n
dt
= b
n
, 1 = E
n
b
n
(6.28)
(show this).
Therefore we nd b
+
n
[ n
n
) = c
n
[ 1 n
n
), where
the constant c
n
follows from
[ c
n
[
2
= n
n
[ b
n
b
+
n
[ n
n
) = n
n
[ 1 b
+
n
b
n
[ n
n
)
= 1 n
n
.
This yields
b
+
n
[ n
n
) = e
i
n
1 n
n
[ 1 n
n
) . (6.30)
Analogously we obtain
b
n
[ n
n
) = e
i
n
n
n
[ 1 n
n
) . (6.31)
In summary, for . . .
. . . bosons:
b
n
[ . . . , n
n
, . . .) =

n
n
[ . . . , n
n
1, . . .)
b
+
n
[ . . . , n
n
, . . .) =

n
n
+ 1 [ . . . , n
n
+ 1, . . .)
. . . fermions:
b
n
[ . . . , n
n
, . . .) = e
i
n
n
n
[ . . . , 1 n
n
, . . .)
b
+
n
[ . . . , n
n
, . . .) = e
i
n
1 n
n
[ . . . , 1 n
n
, . . .) .
Notice the dual character of the operators b
n
and
b
+
n
in the fermion case. Both may act as creation
as well as annihilation operators depending on the
value of n
n
.
6.3 Interacting quantum
elds:
The next logical step is to combine the quantization
of the radiation eld with the second quantization
in terms of the number representation worked out
in the previous two sections. The key quantity is
the Hamiltonian
H =
_
d
3
r
+
(r, t)
_
1
2m
_
p
e
c
A
_
2
+| (r)
_
(r, t)
+
_
d
3
r
1
8
_
E
2
+

H
2
_
, (6.32)
where (r, t) =

n
b
n
(t)
n
(r). Notice that the
term in square brackets corresponds to the right
side of Eq. (4.74) with (r) set equal to zero (us-
ing the proper gauge). Therefore the rst inte-
gral (6.32) describes charged particles interacting
with an electromagnetic eld and with the poten-
tial | (r). The second integral in (6.32) describes
the energy content of the electromagnetic eld it-
self (cf. [15]; section 2.3). Notice that the electric
eld strength,

E, and the magnetic eld strength,
H, may both be expressed in terms of the vector

potential

A (cf. problem 4). In our formalism they
become operators, because the Fourier coecients
c
k,
are turned into operators.
Using (4.75) in conjunction with (6.4) and (6.22)
we may write H as
H =
n
E
n
b
+
n
b
n
+
k,
h
k
_
c
+
k,
c
k,
+
1
2
_
+H
int
, (6.33)
where
H
int
=
_
d
3
r
+
_
i
eh
mc
A

+
e
2mc
2
A
2
_
(6.34)
describes the interaction between the particle eld
and the radiation eld. It is not dicult but some-
what tedious to express H
int
in terms of b
n
, b
+
n
,
c
k,
, and c
+
k,
. The result is a sum of terms propor-
tional to certain products of these operators. An
example is the product b
+
n
b
n
c
k,
. This term cor-
responds to the annihilation of a photon (
k, ) in a
process which simultaneously annihilates a particle
in state [ n) but creates a particle in state [ n
t
).
Another example is the product b
+
n
b
n
c
+
k,
. This
time the annihilation of a particle in state [ n) si-
multaneously creates a particle in state [ n
t
) and
also creates (
k, ). Such scattering processes can

be depicted graphically by the diagrams shown in
gure 6.1. There are still more processes possible
here involving also two photons. However, at this
point we refer the reader to the detailed account in
reference [6] (Spezielle Kapitel).
|n>
|n'>
k,
k,
|n>
|n'>
Figure 6.1: (a) Scattering of a particle from [ n) to
[ n
t
) via the absorption of a photon (
k, ). (b) Scat-
tering of a particle from [ n) to [ n
t
) via emission
of (
k, ).
6.4 Outlook
We want to conclude these notes with a remark
on what to expect from hereon. Clearly, the con-
tent of this chapter must be dealt with more care-
fully and in detail. This is usually done in a sub-
sequent advanced course on quantum theory (e.g.,
[6, 24, 25, 26]). Nevertheless the conceptual devel-
opment thus far should have familiarized the stu-
dent with the language used in intermediate texts
on theoretical solid state physics (e.g., [27]) or sta-
tistical mechanics (e.g., [28]).
Recently McGraw-Hill has begun publishing
their DeMYSTiFieD-series. With the knowledge
on quantum mechanics acquired in this course I
encourage students to take a look ahead into quan-
tum eld theory based on David McMahons text
Quantum Field Theory DeMYSTiFieD [29].
Appendix A
Constants and units
Fundamental Physical Constants:
quantity symbol value unit
speed of light c 299792458 ms
1
Planck constant h 6.62606876 10
34
Js
h =
h
2
1.054571596 10
34
Js
Gravitation constant G 6.673 10
11
mkg
1
s
2
Planck length
_
hG/c
3
_
1/2
l
p
1.6160 10
35
m
Planck time l
p
/c t
p
5.3906 10
44
s
electric constant
0
8.854187817 10
12
Fm
1
elementary charge e 1.602176462 10
19
C
electron mass m
e
9.10938188 10
31
kg
proton mass m
p
1.67262158 10
27
kg
atomic mass unit u 1.66053873 10
27
kg
Boltzmann constant k
B
1.3806503 10
23
JK
1
,
Rydberg constant
_
2
m
e
c/2h
_
R
10973731.568 m
1
inverse ne-structure constant
_
e
2
/4
0
hc
_
1
1
137.03599976
source: http://physics.nist.gov/constants
Energy equivalents:
conversion to J
1kg (1kg) c
2
= 8.987551787 10
16
J
1m
1
_
1m
1
_
hc = 1.98644544 (16) 10
25
J
1Hz (1Hz) h = 6.62606876 (52) 10
34
J
1K (1K) k = 1.3806503 (24) 10
23
J
1eV (1eV ) = 1.602176462 (63) 10
19
J
1E
h
(1E
h
) = 4.35974381 (34) 10
18
J
97
98 APPENDIX A. CONSTANTS AND UNITS
Appendix B
Mathematical appendix
B.1 Conversion to generalized
coordinates
We consider r = r (u, v, w), where u, v, w are gener-
alized coordinates and dene a coordinate system
via
e
u
=
r
u
1
r
u
e
u
=
r
v
1
r
v
e
w
=
r
w
1
r
w
.
We then have
r = e
u
r
u
+e
v
r
v
+e
w
r
w
dr =
_
_
dx
dy
dz
_
_
= e
u
r
u
du +e
v
r
v
dv +e
w
r
w
dw
=
_
_
z
_
_
= e
u
r
u
u
+e
v
r
v
v
+e
w
r
w
w
To obtain the last relation we write
u
=
r
u

r

r
u

i.e.
r
u
u
= e
u
for u as well as v and w. The desired result follows

via
= e
u
(e
u

) +e
v
(e
v

) +e
w
(e
w

) .
Special cases:
(a) Cylindrical coordinates:
r =
_
_
r cos
r sin
z
_
_
e
r
=
_
_
cos
sin
0
_
_
e
=
_
_
sin
cos
0
_
_
e
z
=
_
_
0
0
1
_
_
r
r
= 1
= r
r
z
= 1
dr = e
r
dr +e
rd +e
z
dz
= e
r
r
+e
1
r
+e
z
2
=
1
r
r
(r
r
) +
1
r
2
+
2
z
99
100 APPENDIX B. MATHEMATICAL APPENDIX
(b) Spherical coordinates
r =
_
_
r cos sin
r sinsin
r cos
_
_
e
r
=
_
_
cos sin
sinsin
cos
_
_
e
=
_
_
sin
cos
0
_
_
e
=
_
_
cos cos
sincos
sin
_
_
r
r
= 1
= r sin
= r
dr = e
r
dr +e
r sin
+e
rd
= e
r
r
+e
1
r sin
+e
1
r
2
=
1
r
2
r
_
r
2
r
_
. .
=
1
r
2
r
(r...)
+
1
r
2
sin
2
+
1
r
2
sin
(sin
)
. .
2
,
Using the denition of angular momentum,

L =
r p, where p = ih
is expressed via the above

components of

it is a straightforward matter to
compute the following formulas:
L
x
= ih(cos cot
+ sin
)
L
y
= ih(sincot
cos
)
L
z
= ih
L
2
= h
2
r
2
2
,
The last equation was obtained using the following
Mathematica-code:
B.2 Spherical harmonics
The spherical harmonics form a complete orthogo-
nal system in two dimensions, i.e.
_
dY
lm
(, ) Y
l
m
(, ) =
ll

mm
,
where
_
d. . . =
_
2
0
d
_

0
sind . . .
and
l=0
l
m=l
Y
lm
(, ) Y
lm
(
t
,
t
)
= (
t
) (cos cos
t
) .
Y
lm
(, ) is given by
Y
lm
(, ) =
(2l + 1) (l [ m [)!
4 (l+ [ m [)!
e
im
P
lm
(cos ) ,
where
P
lm
(cos ) =
m
sin
]m]
_
d
d cos
_
]m]
P
l
(cos )
with
m
= (1)
m
for m 0 and
m
= 1 for m < 0.
The P
l
(x) are Legendre polynomials
P
0
(x) = 1, P
1
(x) = x, P
2
(x) =
1
2
_
3x
2
1
_
, . . .
B.4. FOURIER SERIES AND FOURIER TRANSFORMS 101
P
l
(x) =
1
2
l
l!
d
l
dx
l
_
x
2
1
_
l
obeying the dierential equation
_
_
1 x
2
_
d
2
dx
2
2x
d
dx
+l (l + 1)
_
P
l
(x) = 0 .
Some special Y
lm
are
Y
0,0
=
1
4
, Y
1,1
=
_
3
8
sine
i
,
Y
1,0
=
_
3
4
cos , Y
1,1
=
_
3
8
sine
i
.
Additional Y
lm
(, ) can be obtained via
SphericalHarmonicY[l, m, , ] in Mathemat-
ica. For additional P
l
(x) use LegendreP[l, x]. A
general reference covering the above is [18].
B.3 -function
The -function is dened via
_

f (x) (x x
0
) dx = f (x
0
) .
Formally we may write
(x x
0
) =
_
0 x ,= x
0
+ x = x
0
We summarize some of its properties:
(x) = (x)
(ax) =
1
[ a [
(x)
(g (x)) =
n
1
[ g
t
(x
n
) [
(x x
n
)
(with g (x
n
) = 0 and g
t
(x
n
) ,= 0)
x (x) = 0
f (x) (x a) = f (a) (x a)
_
(x y) (y a) dy = (x a)
(x) =
1
2
_

e
ikx
dk
_

(m)
(x) f (x) dx = (1)
m
f
(m)
(0) .
Here
(m)
is the nth derivative of (x).
(m)
(x) = (1)
m
(x)
_

(m)
(x y)
(n)
(y a) dy =
(m+n)
(x a)
x
m+1
(m)
(x) = 0 .
Special properties of the rst derivative:
_

t
(x) f (x) dx = f
t
(0)
t
(x) =
t
(x)
_

t
(x y) (y a) dy =
t
(x a)
x
t
(x) = (x)
x
2
t
(x) = 0
t
(x) =
i
2
_

ke
ikx
dk .
three-dimensional (3D) formulas:
(r r
0
) = (x x
0
)(y y
0
)(z z
0
)
=
1
r
2
(r r
0
)(cos cos
0
)(
0
)
B.4 Fourier series and Fourier
transforms
If f (x) is a periodic function with a period L, then
it can be expanded in a Fourier series:
f (x) =
n=
a
n
e
ik
n
x
,
where k
n
= 2n/L. The coecients a
n
of the series
are given by
102 APPENDIX B. MATHEMATICAL APPENDIX
a
n
=
1
L
_
L
0
f (x) e
ik
n
x
dx .
The Fourier transform of a function f (x) is
f (k) =
1
2
_

f (x) e
ikx
dx .
The inverse transformation is given by
f (x) =
1
2
_

f (k) e
ikx
dk .
Using the denition of Fourier transforms we can
write down an integral representation of the -
function:
(x x
0
) =
1
2
_

e
ik(xx
0
)
dk
or in 3D
(r r
0
) =
1
(2)
3
_
e
i
k(rr
0
)
d
3
k .
An important theorem is the convolution theorem:
If

f (k) and g (k) are the Fourier transforms of f (x)
and g (x) respectively, then
_

f (k) g (k) e
ikx
dk =
_

f (u) g (x u) du .
If we dene the convolution, denoted by f g, of
the functions f and g to be
f g =
1
2
_

f (u) g (x u) du .
then the above cause written
f g =

f g .
This is the convolution theorem for Fourier trans-
forms.
Identity of norms:
_

[ f (x) [
2
dx =
_

[

f (k) [
2
dk .
Parsevals theorem:
_

f (x) g
(x) =
_

f (k) g
(k) dk .
Appendix C
The Casimir eect
An interesting manifestation of the importance
of the zero-point energy is the Casimir eect [30].
Here we illustrate the essence of the Casimir eect
in a simple 1D example (cf. section 5.8 in reference
[31]).
We consider again the black body cavity intro-
duced in section 1.2. To make things simple, we
reduce the 3D cavity to a 1D cavity of length L. In
addition we focus on the limit T = 0, even though
the eect discussed here also shows up at non-zero
temperatures. The zero-point energy of the radia-
tion eld inside the cavity is given by
E
0
(L) =
h
2
(C.1)
with = cn/L(n = 1, 2, . . . , )
1
. This sum is
innite. But we can force it to be nite. We mul-
tiply by a so-called cuto function, which sup-
presses the high frequency modes:
f () = exp[] =
_
1 for 0
0 for
. (C.2)
Here is a small real number. Now Eq. (C.1)
becomes
E
0,cut
(L) =
h
2
f () . (C.3)
As in the case of black body radiation we want to
transform the sum into an integral. This time, how-
ever, we must be careful, because we do need not
1
Notice that E
0
(L) may be expressed via E
0
(L) =
1
4
_
0[
E
2
[0)dV as pointed out on page 93. This means
that E
0
(L) corresponds to the vacuum uctuations of the
electric eld inside the cavity.
only the leading term but in addition the correc-
tion terms. A systematic transition from a sum to
its integral representation is the Euler-MacLaurin
summation formula:
m1
n=1
F (n) =
_
m
0
F (n) dn (C.4)
1
2
F (m) +F (0)
+
1
12
F
t
(m) F
t
(0)
1
720
F
ttt
(m) F
ttt
(0) +. . .
+ (1)
p1
B
p
(2p)!
F
(2p1)
(m) F
(2p1)
(0) +. . .
[18]. The B
p
are the Bernoulli numbers (B
1
= 1/6,
B
2
= 1/30, B
3
= 1/42, . . .). Using (C.4) we obtain
E
0,cut
h
=
1
2
f () =
c
2L
n=1
nexp
_
cn
L
_
=
c
2L
_
_

0
nexp
_
cn
L
_
dn
. .
=L
2
/(c)
2
1
12
+
1
720
3
2
c
2
2
L
2
. . .
_
=
L
2c
2

c
24L
+

3
c
3
480
2
L
3
+. . . . (C.5)
Obviously, for 0 the result diverges! But
now comes the interesting part. We insert two 1D
metal plates into the box. Plate one is located
at a/2 relative to the center of the box; plate
103
104 APPENDIX C. THE CASIMIR EFFECT
two is located at a/2. Thus, the original cavity is
partitioned into three compartments with lengths
L/2a/2, a, and L/2a/2. The overall zero-point
energy is described by (C.5):
E
total
0,cut
= E
0,cut
(a) + 2E
0,cut
_
L a
2
_
. (C.6)
Thus, according to (C.5) we obtain
1
h
E
total
0,cut
L
L
2c
2

c
24a
+

3
c
3
2
480a
3
+. . . ,(C.7)
and taking the derivative with respect to a yields
a
_
1
h
E
total
0,cut
L
_
=
c
24a
2

3
c
3
2
160a
4
. (C.8)
Finally we set = 0 to obtain
E
total
0
a
=
ch
24a
2
. (C.9)
Eq. (C.9) states that the zero-point energy gives
rise to an attractive force proportional to a
2
be-
tween the two inserted plates. A surprising result!
By inspection of Eq. (C.4) we can see easily that
the result (C.9) does not depend on the form of the
cuto function as long as it obeys the limits indi-
cated in Eq. (C.2) and as long as it is reasonably
smooth (no -function!).
The same calculation can be carried out in 3D.
The resulting force in this case is proportional to
a
4
(cf. K. Huang or K. A. Milton)
2
. Since H. B.
G. Casimirs proposal of this eect in 1948 experi-
ments have conrmed the Casimir force for vari-
ous geometries (e.g. plate-plate or plate-sphere).
2
The Casimir pressure for a plate separation of a =
10
6
m is about 1.3 10
3
Pa.
Appendix D
Problems
Problem 1: Some integrals in quantum me-
chanics
Solve the following integral
_
a.s.
d
3
r [ (r) [
2
for
(r) =
1
exp[r]
(r) =
1
8
_
1
r
2
_
exp[r/2]
(r) =
1
8
sin r exp[r/2 i] .
Do not use tables or computers! Note that
r = r(cos sin, sinsin, cos ).
Problem 2: Classical harmonic oscillator
Calculate the probability p
cl
(x)x for nding
the 1D harmonic oscillator within the intervall x
and plot the probability density p
cl
in the range
between a and a, where a is the amplitude.
Problem 3: Generalized coordinates
We consider r = r (u, v, w), where u, v, w are gener-
alized coordinates, and dene a coordinate system
via
e
u
=
r
u
1
r
u
e
u
=
r
v
1
r
v
e
w
=
r
w
1
r
w
.
(a) Show that
=
_
_
z
_
_
= e
u
r
u
u
+e
v
r
v
v
+e
w
r
w
w
.
(b) Work out

both in cylindrical and spheri-
cal coordinates, i.e. r = r(, , z) and r = r(r, , ).
Problem 4: Energy density of the black body
In class we have learned that the classical eld en-
ergy E contained in a black body cavity enclosing
the volume V is given by
E =
1
8
_
_
_
1
c

A
t
2
+
2
_
_
dV .
Show that using the Fourier decomposition of the
vector potential
A(r, t) =
1
k,
_
c
k,
()
e
i(
krt)
+c
k,
()
e
i(
krt)
_
,
where V = L
3
, k
x
=
2
L
n
x
, k
y
=
2
L
n
y
, k
z
=
2
L
n
z
(n
x
, n
y
, n
z
= 1, 2, . . .), yields
E =
k,
w
2
2c
2
c
k,
c
k
105
106 APPENDIX D. PROBLEMS
Hints: (i)
()
is a real unit vector in the -direction
of the plane perpendicular to the momentum vec-
tor

k ( = 1, 2); (ii)
_
V
d
3
re
i(
)r
= V
k,
.
Problem 5: Compton eect
According to quantum theory, a monochromatic
electromagnetic beam of frequency is regarded
as a collection of particle-like photons, each
possessing an energy E = h and a momentum
p = h/c = h/, where is the wavelength.
Suppose that a photon moving along the x-axis
collides with a particle of mass m
o
. As a result of
the collision, the photon is scattered at an angle
, and its frequency is changed. Find the increase
in the photons wavelength as a function of the
scattering angle. Hint: Use the relativistic energy-
momentum relation, i.e. E =
_
p
2
c
2
+m
2
o
c
4
.
Problem 6: Matrix-mechanics
Show that the matrices
x =
_
h
2m
_
_
_
_
_
_
_
_
0 1 0 0 0
1 0
2 0 0
0
2 0
3 0
0 0
3 0
4
.
.
.
.
.
. 0
4
.
.
.

5
.
.
.
.
.
.
.
.
. 0
5
.
.
.
_
_
_
_
_
_
_
_
and
p = i
_
hm
2
_
_
_
_
_
_
_
_
0 1 0 0 0
1 0
2 0 0
0
2 0
3 0
0 0
3 0
4
.
.
.
.
.
. 0
4
.
.
.

5
.
.
.
.
.
.
.
.
. 0
5
.
.
.
_
_
_
_
_
_
_
_
(a) obey the commutator relation
_
p, x
= ih
and,
(b) if inserted in H =
1
2m
p
2
+
m
2
2
x
2
, yield
H = h
_
_
_
_
_
_
_
1
2
0 0 0
0
3
2
0 0
0 0
5
2
0
0 0 0
7
2
.
.
.
.
.
.
.
.
.
.
.
.
_
_
_
_
_
_
_
.
Problem 7: Phase- and group velocity
Calculate both the phase and the group velocity of
a deBroglie wave for a non-relativistic particle.
Problem 8: Anharmonic 1D oscillator in the
WKB-approximation
We have seen that the enhanced quantization con-
dition
1
h
_
pdq = n +
1
2
(n = 0, 1, 2, . . .)
describes the exact energy eigenvalues of the 1D
harmonic oscillator. Assuming that this formula
applies approximately to other systems as well, cal-
culate the energy eigenvalues of an anharmonic 1D
oscillator, whose energy is given by
E =
p
2
2m
+
m
2
2
x
2
+x
4
to leading order in the small parameter . You may
use computer algebra programs like Mathematica
to carry out the necessary integrations and expan-
sions. Hints: Introducing the new quantities and
c via E = h and c = 4h/(m
2
3
) you nd the
simplied equation
2
_
z
o
z
o
dz
_
1 z
2
cz
4
= n +
1
2
,
where z
o
are the solutions of 1 z
2
cz
4
= 0.
Expand the integrant keeping the lead-
ing order in c only. Integrate the result.
Again expand the result of the integra-
tion to leading order in c (you should nd
_
z
o
z
o
dz
1 z
2
cz
4
= /2 3c/16 + O(c
3/2
)).
Note that this approach fails if you try to obtain
higher order corrections (why?). Insert c = and
solve the quadratic equation for . Because the
solution is valid only to leading order of the small
parameter , expand one last time to obtain in
terms of n and to leading order in . Your result
should look like this: n+1/2 +. . . (n+1/2)
2
,
where . . . is a numerical coecient.
107
Problem 9: Wave packet and the harmonic
oscillator potential
(a) Show that whatever superposition of har-
monic oscillator states is used to construct a wave
packet (x, t), it is localized at the same place at
times 0, T, 2T, . . . , where T is the classical period
of the oscillator.
(b) Construct the explicit form of (x, t) at
x = 0, and discuss its time behavior. Hints.
H
2n+1
(0) = 0 and H
2n
(0) =
(1)
n
(2n)!
n!
.
Problem 10: Potential well
Consider the Schrodinger equation
_
1
2
2
q
u(q) +
_
(q) = 0
for the potential
u(q) =
_
_
_
(q 0)
u
o
(0 < q a)
0 (a < q)
,
where u
o
and a are positive.
(a) Determine the bound state eigenfunctions
(you need not calculate the overall normalization
constant explicitly).
(b) What is the condition the quantity a
2
u
o
must fulll in oder for at least one bound state to
exist. Provide an equation that allows to obtain
the eigenvalues numerically.
Hint: The wave function and its derivative are
continuous at q = a.
Problem 11: Raising and lowering operators
We write the stationary Schrodinger equation for
the 1D harmonic oscillator in the symbolic form
H [ n) = E
n
[ n). Using the denitions
a :=
_
m
2h
x +i
p
2mh
a
+
:=
_
m
2h
x i
p
2mh
show
_
a
+
, a
= 1
and that
H = h
_
a
+
a +
1
2
_
agrees with the Hamilton operator of the 1D har-
monic oscillator. In addition show
a [ n) =
n [ n 1) .
Problem 12: The hydrogen atom
When we spoke in class about the hydrogen
atom we left out several steps to be completed as
homework problems. Here they are
(a) Show explicitly how to get from
H =
h
2
2m
e
2
e
h
2
2m
n
2
n
+| ([ r
e
r
n
[)
to
H =
h
2
2
2
r
h
2
2m
2
R
+| (r) ,
where r = r
e
r
n
and m
R = m
e
r
e
+
m
n
r
n
(m = m
e
+m
n
, = m
e
m
n
/m).
(b) Using

L = r p, where p = ih
, show that
L
x
= ih(cos cot
+ sin
) ,
L
y
= ih(sincot
cos
) ,
L
z
= ih
.
(c) Show also that
L
2
= h
2
r
2
2
,
.
Hint: Use

in spherical coordinates. Note that
r
2
2
,
= sin
2
+ sin
1
(sin
).
(d) Derive the recursion relation (1.100):
c
j+1
=
2
_
l +j + 1
1
_
(j + 1) (j + 2l + 2)
c
j
.
Problem 13: Step potential
We look at the step potential
0
x
U(x)
U
o
| (x) =
_
|
0
x > 0
0 x < 0
.
Consider a current of particles of energy E > |
0
moving from x = to the right.
(a) Write the stationary solutions for each of the
regions. Express the fact that there is no current
coming back from x = + to the left, and use the
matching conditions to express the reected and
transmitted amplitudes in terms of the incident
amplitude. Hint: Because the direction of the
current is determined by the wave vector

k (or
by the sign of k in the one-dimensional case) we
may express the incident part of the current, j
i
,
via
I
(x) = A
1
exp[ik
1
x], where A
1
is the incident
amplitude and hk
1
=

2mE. To obtain the
complete
I
(x) we must add to this the reected
contribution (k
1
!) in region I.
II
(x) may be
constructed analogously.
(b) Compute the probability current in the
regions I and II and interpret each term.
(c) Find the reection (r) and transmission (t)
coecients dened as R = [j
r
/j
i
[ and T = [j
t
/j
i
[.
Problem 14: Potential barrier
Consider a square potential barrier
-a/2 a/2
x
U(x)
U
o
| (x) =
_
_
_
0 x < a/2
|
0
a/2 x a/2
0 a/2 < x
.
(a) Assume that the incident particles of energy
E > |
0
are coming from x = . Find the
stationary states. Apply the matching conditions
at x = a/2 and x = a/2.
(b) Find the transmission and reection coef-
cients. Sketch the transmission coecient as a
function of the barriers width a, and discuss the
results.
(c) Find the stationary states describing inci-
dent particles of energy E < |
0
. Compute the
transmission coecient and discuss the results.
Remark: The eect that R = 0 for certain values
of a in part (b) is called resonance scattering. Part
(c) on the other hand is an example of the above
mentioned tunnel eect.
Problem 15: Spherical innite potential well
Consider a particle in a spherical innite potential
well:
|(r) =
_
0 0 r a
a < r
.
109
Write the dierential equation of the radial and
angular parts, and solve the angular equation.
Compute the energy levels and normalized eigen-
functions for l = 0. Hint: The eigenfunction is
supposed to be non-singular at the origin.
Problem 16: The free symmetric top
A symmetric top with moments of inertia I
x
= I
y
and I
z
in the body axes frame is described by the
Hamiltonian
H =
1
2I
x
_
L
2
x
+L
2
y
_
+
1
2I
z
L
2
z
.
Note that moments of inertia are parameters not
operators.
(a) Calculate the eigenvalues and the eigenstates
of the Hamiltonian.
(b) What values are expected for a measurement
of L
x
+ L
y
+ L
z
for any state? The state of the
top at time t = 0 is [l = 3, m = 0). What is
the probability that for a measurement of L
z
at
t = 4I
x
/h we will obtain the value h?
Problem 17: Generators of rotation and
translation
(a) Using the same approach that we had applied
in classical mechanics to show that angular momen-
tum is conserved in an isotropic system, show that
U
R
()(r), where
U
R
() = exp
_
i
h
e
L
_
,
describes a rotation with respect to an axis ori-
ented along the unit vector e
. Here is the angle

of rotation, and

L is the usual angular momentum
operator, which in this context is called the gener-
ator of rotation. Hint: Consider an innitesimal
rotation
rst (just replace the Lagrangian

L(r
i
,

r
i
) by (r); cf. my lecture notes on Classical
Mechanics), then use lim
n
_
1 +
x
n
_
n
= e
x
to
obtain the result for a nite rotation.
(b) Find the analogous translation operator
U
T
(a), where a is a displacement vector.
Problem 18: Commutator algebra
Consider the operators A and B, which
commute with their commutator, i.e.
[B, [A, B]] = [A, [A, B]] = 0. Prove that
(a) for every analytic function F(x) we have
[A, F(B)] = [A, B]F
t
(B) ,
where F
t
(x) denotes the derivative of F(x). Hint:
Start by proving [A, B
n
] = n[A, B]B
n1
. Then use
the fact that F(x) is analytic.
(b)
e
A
e
B
= e
A+B
e
[A,B]/2
.
Hint: Consider the derivative of F(s) = e
As
e
Bs
with respect to the real parameter s, and work
your way towards a rst order dierential equation
s
F(s) = (. . .)F(s).
Problem 19: Changing representations
Consider a two-dimensional physical system. The
kets [
1
) and [
2
) form an orthonormal basis of
the state space. Another basis [
1
) and [
2
) is
dened via
[
1
) =
1
2
([
1
)+ [
2
))
[
2
) =
1
2
([
1
) [
2
)) .
An operator P is represented in the [
i
)-basis by
the matrix
_
1
1
_
.
(a) Find the representation of P in the [
i
)-
basis, i.e. nd the matrix
i
[ P [
j
).
(b) Obtain the representation of the ket e
P
[
1
)
in the [
i
)-basis.
Problem 20: Uncertainty principle
A particle is described by the wave function
(r) =
1
a
3
exp[r/a] .
Calculate x and p
x
and verify the uncertainty
relation for these components.
Problem 21: A simple state space
Consider a physical system with a three-
dimensional state space. An orthonormal basis of
the state space is chosen; in this basis the Hamil-
tonian is represented by the matrix
H =
_
_
2 1 0
1 2 0
0 0 3
_
_
.
(a) What are the possible results when the
energy of the system is measured?
(b) A particle is in the state [ ), represented
in this basis as
1
3
_
_
i
i
i
_
_
. Find H), H
2
), and
H.
Problem 22: Changing representations once
again
The Hamiltonian of a particle in a potential |(r)
is
H =
1
2m
p
2
+|(r) .
Write the Schrodinger equation in both the r- and
the p-representation.
Problem 23: The parity operator
The parity operator is dened by
[ r) =[ r) .
(a) Let [ ) be an arbitrary ket with corre-
sponding wave function (r). Find the wave
function corresponding to [ ). Show that is a
hermitian operator.
(b) Find the operator
2
. What are the possible
eigenvalues of ?
(c) We dene the operators
p
+
=
1
2
(1 +) p
=
1
2
(1 ) .
For an arbitrary ket [ ) we also dene
[
+
) = p
+
[ ) [
) = p
[ ) .
Show that [
+
) and [
) are eigenvectors of .
(d) Prove that the wave functions corresponding
to [
+
) and [
) are even and odd functions,

respectively.
Problem 24: Density matrix
(a) A system is in a pure state [). We had said
that the probability of experimentally nding it
in a state [) is given by [[)[
2
. Show that this
probability can be written as P =
n
p
n
[
n
[)[
2
,
where =
n
[
n
)p
n
n
[.
(b) We describe polarized light using the density
matrix. Consider a beam of light travelling in z-
direction. The state vectors
[
1
) =
_
1
0
_
[
2
) =
_
0
1
_
describe the x- and y-polarized states, respectively.
Any pure state can be written as a linear combina-
tion of these two states, i.e. [ ) = a [
1
) +b [
2
),
where [a[
2
+ [b[
2
= 1. Write down the density
matrix for the pure state for the following cases:
(i) x-polarized case, (ii) y-polarized case, (iii)
45
o
-polarized state, and (iv) 135
o
-polarized state.
111
Show also, that the density matrices for the follow-
ing two mixed states are identical: (v) mixture of
50% x-polarized and 50% y-polarized states; (vi)
50% 45
o
-polarized and 50% 135
o
-polarized.
Problem 25: Path integral of the free particle
Using the x-representation show that the equation
(t) = ()() . . . () (n factors ; t = n)
is correct for
(x
k
, x
k1
; ) =
_
m
2hi
exp
_
m
2 hi
(x
k
x
k1
)
2
_
.
Problem 26: First Born approximation
(a) A particle of mass and momentum h
k
is scattered by the Yukawa potential |(r) =
|
o
exp[r]/(r), where |
o
is real and is positive.
Using the rst Born-approximation to calculate the
scattering amplitude, i.e.
A
(1)
=
4
2
h
2

k
t
[ | [
k) ,
obtain both the dierential and the total scattering
cross section, i.e. d/d = [A
(1)
[
2
and .
(b) Repeat the calculation of d/d
in part (a) for the Gaussian potential
|(r) =
1
4
|
o
exp[(r/2)
2
]. Compare d/d
graphically for the Yukawa and the Gaussian
potential. Comment on the dierences observed
for small and large momentum transfer q = [
k
t
[.
Problem 27: Stark eect
Consider a hydrogen atom placed in a uniform
static electric eld E that points along the z-
direction. The term that corresponds to this in-
teraction in the Hamiltonian is
J = eEz
Note that for the electric elds typically produced
in a laboratory, the condition J H
0
is satised.
The appearance of the perturbation removes the
degeneracy from some of the hydrogen states. This
phenomenon is called the Stark eect. Calculate
the Stark eect for n = 2 in a hydrogen atom.
Problem 28: Variation method applied to 1D
harmonic oscillator
Consider a 1D harmonic oscillator:
H =
h
2
2m
d
2
dx
2
+
1
2
m
2
x
2
(a) For the one-parameter family of wave func-
tions
(x) = e
x
2
( > 0), nd a wave function
that minimizes H). What is the value of H)
min
?
(b) For another one-parameter family of wave
functions
(x) = xe
x
2
( > 0), nd a wave
function that minimizes H) and compute the
value of H)
min
.
(c) Repeat the same procedure for
(x) =
1
x
2
+
( > 0) .
Problem 29: Variation method applied to
helium
(a) Write the Schrodinger equation for the he-
lium atom. What are the solutions for the ground
state if one neglects the interaction between the
two electrons?
(b) Assume that the electrons perform an
electric screening of each other and dene Z as a
variational parameter. Use the variation method
and nd H) and the screening charge.
Problem 30: Charge in an electromagnetic eld
(a) Starting from the relativistic Lagrange func-
tion L =
mc
2
(v)
+
e
c
A v e, where (v) =
_
1
v
2
c
2
_
1
2
, show how one may arrive at the non-
relativistic Hamilton operator
H =
p
2
2m
+e(r)
e
mc
A(r) p
+
ihe
2mc

A(r) +
e
2
2mc
2
A
2
(r) .
for a charge in an electromagnetic eld.
(b) In addition show that if [(t)) is an eigen-
ket of H then [
t
(t)) = e
i
f(r,t)
[ (t)), where
f (r, t) =
e
hc
f (r, t), is an eigenket of H
t
obtained
via the gauge transformations
A

A
t
=

A+

f
t
=
1
c
f ,
applied to H. Hints: (i) use H in its original form
(4.74); (ii) show rst that p[
t
) = e
i
f
( p+ h
f)[),
i.e. ( p
e
c
A
t
)[
t
) = e
i
f
( p
e
c
A)[).
Problem 31: Time-dependent perturbation
theory of a two state system
Consider a quantum system with two stationary
eigenstates [ 1) and [ 2). The dierence between
their eigenvalues is given by E
2
E
1
= h
21
. At
time t = 0, when the system is in state [ 1), a
small perturbation that does not change in time
and equals H
t
is applied. The following matrix el-
ements are given:
1 [ H
t
[ 1) = 0
2 [ H
t
[ 1) = h
0
2 [ H
t
[ 2) = h
21
(a) Using rst-oder time-dependent perturbation
theory, calculate the probability of nding the
system at time t in state [ 1), and the probability
of nding it in the state [ 2).
(b) Solve exactly the Schrodinger equation and
nd [ (t)).
(c) What is the probability that at time t
the system is in the state [ 2)? When is the
approximation used in part (a) a correct one? At
what time (for the rst order) will the system be
with probability 1 in state [ 2).
Problem 32: Field induced ionization of atomic
hydrogen
Consider a hydrogen atom in its ground state
at time t = 0. At the same time a uniform
periodic electric eld,

E =
1
c

A
t

, is ap-
plied to the atom. Assume a simple plane wave,
A = A
0
e cos
_
k r wt
_
, where e is a unit vector
characterizing the polarization direction. In addi-
tion with proper gauge, we may set = 0. Using
rst order time-dependent perturbation theory,
nd the probability of ionization per unit time.
Assume that when the atom becomes ionized, its
electron becomes free:
(a) Find the minimum frequency,
o
, that the
eld needs in order to ionize the atom. Using the
expressions discussed in sections 4.2 and 4.3 of the
lecture notes write down an expression for the ion-
ization probability P
f
(t) and the time-independent
quantity [H
1,fi
[
2
in dipole approximation. Replace
A
0
by E
0
= A
0
/c.
(b) Calculate the matrix element, f[e r[i)
and give an expression for [H
1,fi
[
2
. Hint: Here
r[f) =
1
V
exp[i
k r], where

k is the wave vector
of the free electron. This expression assumes
that the nal state corresponds to an almost free
particle - a particle in a large box with volume
V . The connement ensures that the state can be
normalized.
(c) Write down an expression for the nal density
of states, (E
f
), and use the formula
P(t) =
_
range
P
f
(t)(E
f
)dE
f
d
f
.
to calculate the total ionization probability. Here
the integration over d
f
accounts for the fact
that the free electron wave vector can have any
direction in space. Finally express k in terms of
and
o
and write down an expression for the
probability of ionization per unit time, dP(t)/dt.
Problem 33: Quasi classical approximation
and Morse potential
Using the quasi classical approximation,
1
h
_
pdx = 2
_
n +
1
2
_
(n = 0, 1, 2, . . .) ,
113
numerically calculate the eigenvalues,
n
=
2mE
n
h
2
a
2
,
for a particle conned to a Morse potential
|(x) = D(1 exp[ax])
2
,
with D = 16
h
2
a
2
2m
. You may use a computer
language according to your choice. Your result
should consist of a table listing all eigenvalues, a
short but exact description of your algorithm, and
a printout of your program.
Problem 34: Completeness relation for bosons
and fermions
For the case of N = 2 prove the relation
1
()
=
1
N!
1
,...,
N
[
()
)
()
[ .
by showing
1
()
[
()
) = [
()
) .
Problem 35: Pauli matrices
(a) Calculate the commutator relation [
i
,
j
],
where i, j = x, y, z and the
i
, are the Pauli
matrices given by
x
=
_
0 1
1 0
_
y
=
_
0 i
i 0
_
z
=
_
1 0
0 1
_
The Pauli matrices are related to the spin operator
via
S =
h
2
.
Using the basis vectors of S
z
eigenvectors, calcu-
late S
i
[ +1/2) and S
i
[ 1/2) (i = x, y, z), where
[ +1/2) and [ 1/2) are the eigenvectors of S
z
with eigenvalues +h/2 and h/2, respectively.
(b) If the z-component of an electron spin is
+h/2, what is the probability that its component
along a direction z
t
that forms an angle with the
z-axis equals + h/2 or h/2?
(c) What is the average value of the spin along z
t
?
Problem 36: Clebsch-Gordan coecients
In class we had used that the
c (j
1
, j
2
, j, m
J
; m
1
, m
2
) = 0 unless m
1
+ m
2
= m
J
.
Show this!
Problem 37: The hydrogen molecule
In class we have discussed a simple theory for the
hydrogen molecule. Our result for the electronic
energy was:
E
I/III
= E
A
+E
B
+
D A
1 [K[
2
.
E
A
and E
B
are atomic ground state energies. For-
mulas for the quantities D(R) and A(R), i.e. the
direct Coulomb energy and the exchange energy,
as well as K(R), the overlap integral, were given in
class. R is the distance between the nuclei. Evalu-
ate E
I/III
explicitly using atomic hydrogen ground
state (1s) orbitals, graph your result, and determine
the bond length in units of a
0
.
Hints: You will have to evaluate a number of
integrals. In most cases it is useful to use elliptic
coordinates , , and : = (
A
+
B
)/R and =
(
A
B
)/R. Note that
A
is the separation of one
particular electron from nucleus A, and
B
is the
separation of the same electron from nucleus B.
is the angle of rotation around the line connecting A
and B. In elliptic coordinates the volume element
is d
3
r = (R
3
/8)(
2

2
)ddd (1 ,
1 1, 0 2). For K you should get
K = (1+R+R
2
/3) exp[R] (cf. Eq. (5.74) on page
79). Important: Use the equation given in class to
evaluate the Coulomb integral; replace the electron-
electron separation in the exchange integral by a
constant a (use a = 10; check also what happens
for other values).
Bibliography
[1] H. Weyl (1931) Gruppentheorie und Quanten-
mechanik. 2nd edition
[2] P.A.M. Dirac (1930) The principles of quantum
mechanics. Oxford University Press
[3] J. von Neumann (1932) Mathematische Grund-
lagen der Quantenmechanik. Springer
[4] A. Sommerfeld (1978) Atombau und Spek-
trallinien. Verlag Harri Deutsch
[5] H. Mitter (1979) Quantentheorie, BI
Hochschultaschenb ucher, Vol. 701
[6] W. Greiner (1975) Quantenmechanik - Eine
Einf uhrung; (1980) Quantentheorie - Spezielle
Kapitel. Verlag Harri Deutsch
[7] A. Messiah (1976) Quantenmechanik. de-
Gruyter; Vol. 1,2
[8] L. D. Landau, E. M. Lifschitz (1979) Quanten-
mechanik. Academie-Verlag
[9] P. W. Atkins, R. S. Friedman (1997) Molecular
Quantum Mechanics (3rd Ed.). Oxford
[10] F. Schwabl (2005) Quantenmechanik: Eine
Einf uhrung. Springer
[11] W. Nolting (2007) Grundkurs theoretische
Physik. Bd.5/1: Quantenmechanik. Springer
[12] P. Reineker, M. Schulz, B.M. Schulz (2007)
Theoretische Physik III Quantenmechanik 1.
Wiley-VCH
[13] E. Rebhan (2008) Theoretische Physik: Quan-
tenmechanik. Spektrum
[14] R. Hentschke (2000) Klassische Mechanik.
Vorlesungsskript
[15] R. Hentschke (2001) Klassische Elektrody-
namik. Vorlesungsskript
[16] R. Hentschke (2004) Statistische Mechanik.
Wiley-VCH
[17] K. Huang (1963) Statistical Mechanics. Wiley
[18] M. Abramowitz, I. Stegun; Eds. (1972) Hand-
book of Mathematical Functions. Dover
[19] P. M. Morse, H. Feshbach (1953) Methods of
Theoretical Physics. McGraw-Hill
[20] R. Courant, D. Hilbert (1968) Methoden der
Mathematischen Physik I. Springer
[21] M. R. Spiegel (1964) Complex Variables.
Schaums Outline Series, McGraw-Hill
[22] I. N. Levine (1991) Quantum Chemistry. Pren-
tice Hall
[23] A. Szabo, N. S. Ostlund (1996) Modern Quan-
tum Chemistry. Dover
[24] J. J. Sakurai (1982) Advanced Quantum Me-
chanics. Addison-Wesley
[25] E. Rebhan (2008) Theoretische Physik, Bd.2.
Spektrum
[26] P. Reineker, M. Schulz, B.M. Schulz (2007)
Theoretische Physik III Quantenmechanik 2.
Wiley-VCH
[27] D. Pines (1983) Elementary excitations in
solids. Benjamin Cumming
[28] R. P. Feynman (1972) Statistical Mechanics.
Addison Wesley
[29] D. McMahon (2008) Quantum Field Theory
DeMYSTiFieD. McGraw-Hill
[30] K. A. Milton (2001) The Casimir Eect. World
Scientic
[31] K. Huang (1998) Quantum Field Theory. Wi-
ley
115
Index
-function, 101
g-factor, 24
ab initio models, 87
action, 45
least action principle, 43
angular momentum
total, 72
anti-commutator, 95
basis functions, 85
basis set, 89
minimal, 86, 89
STO-3G, 89
Bernoulli numbers, 103
black body, 1
Bohr
magneton, 24
radius, 19
Boltzmanns constant, 6
Born approximation
rst, 50
higher corrections, 52
Born-Oppenheimer approximation, 76, 80, 81
bosons, 68
Casimir eect, 103
chemical bonding, 80
circular polarization representation, 93
Clebsch-Gordan coecients, 74, 113
commutator, 9
relation, 8
Compton eect, 106
conservation laws, 35
convolution theorem, 102
correspondence principle, 4, 14
Coulomb
integral, 78
potential, 21
cuto function, 103
Darwin term, 74
DeBroglie wavelength, 3
degeneracy, 20
density
functional models, 88
of states, 63
diagrams, 96
dispersion
interaction, 57
relation, 10
Ehrenfests theorem, 35
electron
correlation, 87
density, 81
elliptic coordinates, 79, 113
energy representation, 38
Euler-MacLaurin summation formula, 103
exchange
correlation, 83
energy, 113
interaction, 77, 80
Fermi hole, 83
Fermis golden rule, 63
fermions, 68
ne structure, 53, 71, 74
gauge transformations, 64
Gedankenexperiment, 4
generator
of rotation, 109
of translation, 109
ground state
energy, 9
helium, 77
hydrogen molecule, 80
gyromagnetic factor, 75
Hartree
-Fock approximation, 84
product, 82
unit, 19
116
INDEX 117
Heisenberg picture, 39
Hellmann-Feynman
electrostatic theorem, 80
theorem, 80
Hermite polynomials, 13
Hilbert space, 32
hyperne structure, 74
interaction picture, 39
Lande factor, 75
LCAO approximation, 85
LEED, 52
Legendre polynominals, 100
Lennard-Jones interaction, 58
level splitting, 21, 73
lifetime, 41
of excited state, 63
Lippmann-Schwinger equation, 48
Lorentz condition, 64
Moller-Plesset
models, 87
MP2, 87
Mathematica, 10, 22, 23, 74, 100
Morse potential, 113
occupation number, 68
operator
annihilation, 92, 94
creation, 9294
density, 41
Hamilton, 9
hermitian, 9, 32
lowering, 27
number, 27, 29, 91, 92, 95
parity, 110
permutation, 67
projection, 34
raising, 27, 30
time evolution, 39
optical theorem, 51
orbital
atomic, 21, 85
Gaussian, 85
Slater, 78, 85
spatial orbital, 82
spin orbital, 82
oscillator
anharmonic, 55, 58, 106
classical harmonic, 5, 6, 14, 105
harmonic, 5, 79, 12, 27, 38, 80, 91, 106, 107,
111
overlap
integral, 79, 86, 113
repulsion, 58
Parsevals theorem, 102
particle-wave-dualism, 3
partition function
canonical, 42
Paschen-Back eect, 76
path integration, 44
Pauli
matrices, 113
principle, 68, 82
perturbation, 21
phase shift, 51
photoelectric eect, 3
photon, 4, 63
Plancks constant, 1
Poynting-vector, 92
probability current density, 23
propagator, 45
free, 52
quantization condition, 3
quantum number, 4, 28
rectangular well potential, 15
reduced mass, 3, 18
reection coecient, 25, 108
residue theorem, 49
resonance scattering, 108
Roothaan-Hall equations, 86
Rutherford scattering, 50
scattering
amplitude, 48
cross section, 47
dierential, 47
total, 51
momentum transfer, 50
Schrodinger
equation, 12
equation and second quantization, 94
picture, 39
Schwarz inequality, 32
self-consistent-eld, 89
singlet state, 77
118 INDEX
Slater
determinants, 82
orbital, 85
special theory of relativity, 94
spherical harmonics, 19, 100
spin, 68
spin-statistics-theorem, 68
Stark eect, 21, 111
state
mixed, 42
pure, 42
vector, 8
superselection rule, 67
thermal
equilibrium, 6
wavelength, 71
trace, 41
transition
amplitude, 45, 62
dipole moment, 65
probability, 62
rate, 63
selection rules, 65
transmission coecient, 25, 108
triplet states, 77
tunnel eect, 108
tunneling, 14
uncertainty principle, 37, 63, 93
vacuum
uctuation, 93, 103
state, 93
variation
method, 58, 84
parameters, 58
wave
function, 13
mechanics, 12
packet, 10
WKB-approximation, 106
work function, 3
Yukawa potential, 50
Zeeman eect, 21, 76
zero-point energy, 7, 9, 91, 103

Quantum Theory

Uploaded by

Quantum Theory

Uploaded by

A short course on

Introductory Quantum Theory

Uberlegungen, dass der Saturnring aus festen Teilchen beste-

1.3. BUILDING A GENERAL FORMALISM 11

A is Bohrs radius or the

; in the latter case Eq. (1.98) be-

j in terms of the wave function. Then we write

(cf. the appendix). Because the radial part of the

are positive real numbers and the

k, t) using the new no-

k) is the complex conjugate of

. . .term vanishes, because of the normalizabil-

(r, t) = (t) [ r),

(r, t) r (r, t) . (2.25)

x(x, t)dx (2.40)

In order to satisfy () we must have

(t) :=[ (t) [ (t)) [

(t) =[ (t) [ (t)) [

(t)). Thus, ) may be

) are (normalized) eigenkets of ,

is the kinetic energy in the

is the relative ve-

k) yields another more useful operator

. Thus, we obtain the

, which can be combined

f (r) (cf. page 35), where f (r) is a function

) is an eigenstate of H with the eigenvalue

, then Eq. (5.3) implies

) is a linear combination of eigen-

), which all produce the

. In this case the eigenvalue

). Here the in-

(cf. Eq. (1.45)). First we work out

= (5 [ 3)6 [ 2)7 [ 4)8 [ 1))

= 3 [ 5)2 [ 6)4 [ 7)1 [ 8)

is the number of parti-

is called occupation number and we have

) which fulll the symmetry re-

! = 1!1! = 1 and thus

! = 2!0! = 2 and thus

k)-state (free particle) was

). For an observer travelling

at the origin (where the electron is). Using

S is the magnetic moment of the electron.

such that the single determinant formed from these

, the wave function and

, may then be expanded in powers of , i.e.

basis set is used.

A. Because density functional models take electron

k, ) in the volume (of the cavity) at time t.

H do not commute with n

E[ (cf. section 4.1 in Ref. [15].

H, may both be expressed in terms of the vector

k, ). Such scattering processes can

for u as well as v and w. The desired result follows

is expressed via the above

. Here is the angle

rst (just replace the Lagrangian

) are even and odd functions,

You might also like