Ingolf V. Hertel
ClausPeterSchulz
Atoms,
Molecules and
Optical Physics 1
Atoms and Spectroscopy
Graduate Texts in Physics
Series Editors
Professor Richard Needs
Cavendish Laboratory
JJ Thomson Avenue
Cambridge CB3 0HE, UK
rn11@cam.ac.uk
To my Wife Gudrun
CPS
Preface
Atomic, Molecular and Optical physics short AMO physics is one of the canon
ical fields of physics, a profound knowledge of which is essential for understanding
almost any other area of modern physics. And while its roots reach back over a
century and are closely connected with the early days of modern physics, current
research in AMO physics is still highly productive in respect of both, cutting edge
applications and fundamental insights as several N OBEL prizes in recent years
have documented convincingly.
Looking back at the technical development of modern industrial society which
is closely connected with modern physics one may refer (COSE, 1998) to the
20th century as that of the electron while the 21st is the century of the photon. This
interesting particle, the essential ingredient of modern optics and quantum optics,
surprises humankind since N EWTON with its waveparticle dualism. It does not only
play a key role in todays information technology but is, from a general point of
view, also the primary carrier of any information which can be obtained about the
constituents of matter and materials. Even collisions of particles with mass under the
influence of the C OULOMB force may be viewed as exchange of virtual photons.
The textbooks presented here try to give a fairly comprehensive overview on the
whole field. They cover state of the art experimental methods, and combine this
with preparing the basis for a serious, theory based understanding of key aspects in
modern AMO research. The two volumes, originally written in German language
(H ERTEL and S CHULZ, 2008), are a genuine authors translation not just an En
glish mirror image of the original. We have rewritten much of the text, extended
it wherever appropriate, and updated a number of aspects to catch up with recent
progress in the field.
On the one hand we address advanced students of physics, chemistry and other
neighbouring fields, typically at the end of their undergraduate studies, or during
their doctoral work. On the other hand we also wish to reach young postdocs or
even mature scientists, who feel it is time they connect freshly with the topics ad
dressed here. We consider the basics of classical geometrical optics and wave optics
as well as electrodynamics to be well known by our readers. We also expect a cer
vii
viii Preface
the final, physically important results which we discuss and illustrate usually in
some detail. In addition, we provide several appendices for the reader interested
in more detail. We have e.g. collected a toolbox for angular momentum algebra in
atomic and molecular physics without any claim for full mathematical consistency,
but quite compact and possibly useful in practice.
Some words about formats, notation, units, typography appear in order:
Each chapter begins with a brief motto setting the tune of the chapter, followed
by short abstract guiding the reader through the text. At the end of each section a
short summary recalls what the readers should have learned from the preceding
text. All chapters build upon each other, but may be read by advanced readers
also individually: this is facilitated by intensive cross referencing of formulas and
figures, extended indices covering both volumes, a list of acronyms and important
terminology as well as references at the end of each chapter.
For clarity and homogeneity we do not reproduce original drawings or other
material from the literature. Rather, all published data have been redrawn (after
digitalization if necessary), are presented in a standard format, and all sources
used in the figures and text are properly quoted.
We consequently use the SISystem for all measurable quantities, and we empha
size the pedagogical and practical value of a dimensional analysis for complex
physical formulas.1 On the other hand, atomic units (a.u.) facilitate the writing
of many relations in atomic and molecular physics dramatically. Hence, we use
them intensively considering, however, Eh , a0 and t0 etc. simply as abbrevia
tions for quantities with dimensions. Phrasings such as we set , e, me , c equal
to unity are avoided, since they are highly misleading.
The finite number of letters in the Latin and Greek alphabets makes some incon
sistencies or unusual designations unavoidable: we mention specifically, that in
order to allow the use of E for the electric field strength (an important quantity
in AMO) we use W (with appropriate indices) for energies of various types (with
the exception of the atomic unit of energy which is internationally defined as Eh ).
Occasionally we use the letter T for kinetic energy and try to avoid the neigh
bourhood of time and temperature which are often also designated by T . Vectors
are written as r or k, unit vectors in these directions are er and ek , respectively.
We write operators as H , vectoroperators as p and tensors of rank k as Ck . For
the unit operator and unit matrix we use 1. For integer numbers we mostly use
calligraphic letters such as N , while number densities are simply N to distin
guish them from the index of refraction n which is also an often used quantity
throughout this text. Oscillations and other periodic processes are mostly char
acterized by their angular frequencies (sometimes also by their frequencies )
and the corresponding energies are (or h).
1 We make, however, use of allowed prefixes (NIST, 2000a), such as cm1 as unit of wavenumbers
(which appears ineradicable in the literature). We also use accepted units outside the SI (NIST,
2000b), such as the enormously practical energy unit eV (electronvolt), or b (barn) as unit for cross
sections.
x Preface
Finally, we hope that these books will become a continuing source of reference
for the fastidious reader, working in or just needing to use AMO physics in her or
his special field. We ask all of you to kindly provide us with the necessary feedback.
We shall try to react to useful suggestion promptly. At the home page of the books,
http://www.mbiberlin.de/AMO/bookhomepage, we shall continuously report on
the status, list errata and possibly present additions. For additional reading and
cross referencing we have collected a few related textbooks and monographs in the
reference list below, just as typical examples without any claim for completeness.
Berlin Adlershof, Germany Ingolf V. Hertel
January 2014 ClausPeter Schulz
References
ATKINS, P. W. and R. S. F RIEDMAN: 2010. Molecular Quantum Mechanics. Oxford: Oxford Uni
versity Press, 2nd edn.
B ERGMANN, L. and C. S CHAEFER: 1997. Constituents of Matter Atoms, Molecules, Nuclei and
Particles. Berlin, New York: Walter der Gruyter, 902 pages.
B LUM, K.: 2012. Density Matrix Theory and Applications. Atomic, Optical, and Plasma Physics
64. Berlin, Heidelberg: Springer Verlag, 3rd edn., 343 pages.
B ORN, M. and E. W OLF: 2006. Principles of Optics. Cambridge University Press, 7th (expanded)
edn.
Preface xi
B RANSDEN, B. H. and C. J. J OACHAIN: 2003. The Physics of Atoms and Molecules. Prentice Hall
Professional.
B RINK, D. M. and G. R. S ATCHLER: 1994. Angular Momentum. Oxford: Oxford University Press,
3rd edn., 182 pages.
COSE (Committee Optical Science and Engineering): 1998. Harnessing Light: Optical Science
and Engineering for the 21st Century. Washington, D.C: National Academy Press, 360 pages.
D EMTRDER, W.: 2010. Atoms, Molecules and Photons. Berlin, Heidelberg, New York: Springer,
2nd edn.
D RAKE, G. W. F., ed.: 2006. Handbook of Atomic, Molecular and Optical Physics. Heidelberg,
New York: Springer.
E DMONDS, A. R.: 1996. Angular Momentum in Quantum Mechanics. Princeton, NJ, USA: Prince
ton University Press, 154 pages.
H ERTEL, I. V. and C. P. S CHULZ: 2008. Atome, Molekle und optische Physik 1; Atomphysik
und Grundlagen der Spektroskopie. SpringerLehrbuch. Berlin, Heidelberg: SpringerVerlag,
1st edn., 511 pages.
H ERTEL, I. V. and C. P. S CHULZ: 2010. Atome, Molekle und optische Physik 2; Molekle und
Photonen  Spektroskopie und Streuphysik, vol. 2 of SpringerLehrbuch. Berlin, Heidelberg:
SpringerVerlag, 1st edn., 639 pages.
NIST: 2000a. Reference on constants, units, and uncertainties: SI prefixes, NIST. http://physics.
nist.gov/cuu/Units/prefixes.html, accessed: 8 Jan 2014.
NIST: 2000b. Reference on constants, units, and uncertainties: Units outside the SI, NIST.
http://physics.nist.gov/cuu/Units/outside.html, accessed: 8 Jan 2014.
S TEINFELD, J. I.: 2005. Molecules and Radiation, An Introduction to Modern Molecular Spec
troscopy. Mineola, NY: Dover Edition, 2nd edn.
W EISSBLUTH, M.: 1978. Atoms and Molecules. Student Edition. New York, London, Toronto,
Syndey, San Francisco: Academic Press, 713 pages.
Acknowledgements
Over the past years, many colleagues have encouraged and stimulated us to move
forward with this work, and helped with many critical hints and suggestions. Most
importantly, we have received a lot of helpful material and state of the art data for
inclusion in these textbooks.
We would like to thank all those who have in one or the other way contributed
to close a certain gap in the standard textbook literature in this area that is at
least what we hope to have achieved. Specifically we mention Robert Bittl, Wolf
gang Demtrder, Melanie Dornhaus, Kai Godehusen, Uwe Griebner, Hartmut Ho
top, Marsha Lester, John P. Maier, Reinhardt Morgenstern, HansHermann Ritze,
Horst SchmidtBcking, Ernst J. Schumacher, Gnter Steinmeyer, Joachim Ullrich,
Marc Vrakking und Roland Wester; their contributions are specifically noted in the
respective lists of references.
Of course, all other sources are also documented there which we have used for
information and which have provided the data used to generate the figures in these
books.
One of us (IVH) is particularly grateful to the MaxBornInstitute for provid
ing the necessary resources (including computer facilities, library access, and office
space etc.) for continuing the work on this book after official retirement.
xiii
Contents of Volume 1
1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Overview, History and Magnitudes . . . . . . . . . . . . . . . . 1
1.1.1 Quantum Nature of Matter . . . . . . . . . . . . . . . . . 2
1.1.2 Orders of Magnitude . . . . . . . . . . . . . . . . . . . . 5
1.2 Special Theory of Relativity in a Nutshell . . . . . . . . . . . . . 10
1.2.1 Kinematics and Dynamics . . . . . . . . . . . . . . . . . 10
1.2.2 Time Dilation and LORENTZ Contraction . . . . . . . . . 13
1.3 Some Elementary Statistics and Applications . . . . . . . . . . . 14
1.3.1 Spontaneous Decay and Mean Lifetime . . . . . . . . . . 15
1.3.2 Absorption, LAMBERTBEER Law . . . . . . . . . . . . 17
1.3.3 Kinetic Gas Theory . . . . . . . . . . . . . . . . . . . . 18
1.3.4 Classical and Quantum Statistics Fermions and Bosons 20
1.4 The Photon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.4.1 Photoelectric Effect and Quantization of Energy . . . . . 26
1.4.2 COMPTON Effect and Momentum of the Photon . . . . . 28
1.4.3 Pair Production . . . . . . . . . . . . . . . . . . . . . . 30
1.4.4 Angular Momentum and Mass of the Photon . . . . . . . 30
1.4.5 Electromagnetic Spectrum . . . . . . . . . . . . . . . . . 31
1.4.6 PLANCKs Radiation Law . . . . . . . . . . . . . . . . . 31
1.4.7 Solar Radiation on the Earth . . . . . . . . . . . . . . . . 34
1.4.8 Photometry Luminous Efficiency and Efficacy . . . . . 37
1.4.9 XRay Diffraction and Structural Analysis . . . . . . . . 40
1.5 The Four Fundamental Interactions . . . . . . . . . . . . . . . . 43
1.5.1 COULOMB and Gravitational Interaction . . . . . . . . . 44
1.5.2 The Standard Model of Fundamental Interaction . . . . . 46
1.5.3 Hadrons . . . . . . . . . . . . . . . . . . . . . . . . . . 48
1.5.4 The Electron . . . . . . . . . . . . . . . . . . . . . . . . 49
1.6 Particles in Electric and Magnetic Fields . . . . . . . . . . . . . 51
1.6.1 Charge in an Electric Field . . . . . . . . . . . . . . . . 52
1.6.2 Charge in a Magnetic Field . . . . . . . . . . . . . . . . 53
xv
xvi Contents of Volume 1
xxvii
xxviii Contents of Volume 2
xxxvii
Basics
1
Overview
Section 1.1 gives a brief survey of the canonical subject areas in physics, of
physics history and the quantum nature of atomic phenomena. An introduc
tion to orders of magnitude of length, time and energy follows. Section 1.2
summarizes some essentials of special relativity. Section 1.3 introduces some
elements of statistical mechanics and thermodynamics. The photon, key par
ticle in this text book, enters the scene in Sect. 1.4. Section 1.5 makes a very
short excursion into the nature of the four fundamental interactions and to
the standard model of elementary particles. Section 1.6 deals with the mun
dane subject of how free, charged particles move under the influence of an
external electromagnetic field. Particles and waves (Sect. 1.7) and the B OHR
model of the H atom (Sect. 1.8) lead us to the foundations of modern physics.
Section 1.9 introduces one of the key concepts of quantum mechanics: space
quantization as discovered in the famous S TERN G ERLACH experiment
which in turn led, more or less directly, to the discovery of the electron spin,
treated in Sect. 1.10.
Table 1.1 gives a compact overview of the canonical subject areas of modern
physics and connects them to the content of these textbooks: marked in italics
are those themes that are at least partially treated here. Otherwise, the table is
more or less self explaining.
At the beginning of the past century, the history of atomic, molecular and optical
(AMO) physics was almost identical to the history of physics at large. Without any
claim for completeness, we collect in Table 1.2 some milestones in the topical fields
of this textbook, and Table 1.3 highlights some key developments in theoretical
particle physics which we shall touch very briefly in Sect. 1.5. Of course, a host
of fascinating details during the development of modern physics cannot even be
mentioned here.
Table 1.2 Highlights of science history from the idea of the atom to modern physics of atoms,
molecules and quantum optics (incomplete list)
400 BC D EMOCRITOS oo (indivisible)
1808 AD DALTON Multiple proportions
1811 AVOGADRO Molecular theory of gases
1814 F RAUNHOFER First useful spectrometer
1834 FARADAY Electrolysis (FARADAY constant)
1868 M ENDELEEV Periodic table of elements
1869 H ITTORF Cathode rays
1886 G OLDSTEIN Channel rays
1895 RNTGEN Xrays
1896 B ECQUEREL Radioactivity
1897 J.J. T HOMSON e/m for electrons
1898 Marie & Pierre C URIE Polonium, radium
1898 W IEN e/m for ions
1900 P LANCK E = h
1903 RUTHERFORD Atomic nuclei
1905 E INSTEIN E = mc2
1913 B OHR Atom model
1913 M ILLIKAN edetermination
19211922 S TERN & G ERLACH Space quantization
1925 Max B ORN (N OBEL prize 1954) Fundamental research in quantum
mechanics
1926 S CHRDINGER Wave equation
1927 H EISENBERG Uncertainty relation
1947 L AMB and R ETHERFORD L AMB shift for excited H
19581966 S CHAWLOW, T OWNES, Basov, P ROKHOROV, Maser, laser and spectroscopy
M AIMAN, JAVAN, K ASTLER
1986 N OBEL prize Dudley R. H ERSCHBACH, Dynamics of chemical elementary
Yuan T. L EE and John C. P OLANYI processesa
1989 N OBEL prize Norman G. R AMSEY, R AMSEY fringes, atomic clocksb , ion
Hans D EHMELT and Wolfgang PAUL trapsa
1996 N OBEL prize R. F. C URL Jr., H. K ROTO, Discovery of fullerenesa . . . C60 etc.
R. E. S MALLEY
1997 N OBEL prize S. C HU, Methods to cool and trap atoms with laser
C. C OHEN TANNOUDJI, W. D. P HILLIPS light
1999 N OBEL prize Ahmed Z EWAIL Femto(second) chemistrya
2001 N OBEL prize Eric A. C ORNELL, Cold atoms and
Wolfgang K ETTERLE, Carl E. W IEMAN B OSE E INSTEIN condensationa
2002 N OBEL prize John F ENN, Koichi TANAKA Electro spray, molecular beamsa , MALDI
mass spectroscopya
2005 N OBEL prize Roy G LAUBER, Theory of optical coherencec and
John H ALL and Theodor H NSCH laser precision spectroscopya
2007 N OBEL prize Gerhard E RTL Chemical processes at surfaces
a Work from several preceding years
b Work from the 1950s
c Work from the 1960s
4 1 Basics
Table 1.3 Theory on the way from electrodynamics to the standard model of the fundamental
interactions (the dates of N OBEL prizes given here refer to discoveries and developments which
typically happened much earlier)
ca. 1850 James Clerk M AXWELL Electrodynamics
1918 N OBEL prize Max P LANCK Energy quanta
1921 N OBEL prize Albert E INSTEIN Law of the photoelectric effect
1932 N OBEL prize Werner H EISENBERG Creation of quantum mechanics
1933 N OBEL prize Erwin S CHRDINGER, Paul D IRAC Wave equations for matter
1949 N OBEL prize Hideki Y UKAWA Prediction of mesons
1954 N OBEL prize Max B ORN Statistical interpretation
of quantum mechanics
1963 N OBEL prize W IGNER, G OEPERTM AYER, J ENSEN Structure of the nucleus
1965 N OBEL prize T OMONAGA, S CHWINGER, F EYNMAN Quantum electrodynamics
1967 N OBEL prize B ETHE Theory of nuclear reactions
1969 N OBEL prize G ELL M ANN Quark model
1979 N OBEL prize G LASHOW, W EINBERG, S ALAM Theory of weak interaction
1982 N OBEL prize W ILSON Renormalization, critical
phenomena
1999 N OBEL prize T H OOFT, V ELTMAN Quantum structure
of electroweak interaction
2004 N OBEL prize G ROSS, P OLITZER, W ILCZECK Asymptotic freedom of quarks
Closely related and supplementing the above we shall also have to discuss the quan
tization of energy. As examples we mention the
Wph = h = (1.3)
Jz = mj (1.8)
mj = j, j + 1, . . . , j. (1.9)
Here J is the absolute value of the angular momentum, Jz its projection on a given
axis (here z) in space. One distinguishes orbital angular momentum (integer quan
tum number) and intrinsic angular momentum or spin (integer for so called bosons,
half integer for so called fermions, see Sect. 1.5). To characterize these properties
of a given orbit or particle we shall refer to its angular momentum (or spin) as being
0, 1/2, 1, etc.
Before going into details, we want to give an orientation about orders of magnitude
for some relevant physical observables with which we shall have to deal. Orders
of magnitudes are always important in physics and the student is well advised to
roughly memorize some of these data. This will turn out to be very helpful when
trying to find her or his bearings later on when indulging into real, own measure
ments. (A list of the most important fundamental constants of nature is provided in
Appendix A.)
diameter of
Planck length
our galaxy
bacteria
atomic nuclei
next star
universe
insects
human
atoms
virus
cells
1035 1015 1010 105 100 105 1010 1015 1020
dominant forces: length / m
strong interaction gravitation (among quasi neutral objects)
electroweak interaction
electromagnetic interaction
Fig. 1.1 Length scales in the universe and the range of dominance of the four fundamental forces:
strong, electroweak, electromagnetic and gravitational interaction (see Sect. 1.5)
1018 m quark 1 am
electron
? ? (lepton)
1.1 Overview, History and Magnitudes 7
homo sapiens
Planck time
universe
lifetime
0 meson
human life
one year
neutron
lifetime
earth
see enlarged scale
1044 1016 1012 108 104 100 104 108 1012 1016
time / s
Fig. 1.3 Time scales in the universe. The range marked with the fat double arrow is shown in
Fig. 1.4 on an enlarged scale
pointed out that the finite extension of the atomic nuclei play a role (proton radius
ca. 0.881015 m) for ultrahigh precision spectroscopy (see Chap. 9). On the other
end of the atomic length scale we have to be aware that the wavelength of electro
magnetic radiation (spectral range of the visible, VIS, from 380 nm to 760 nm, see
Sect. 1.4.5) is of high relevance in all spectroscopic investigations.
primary
processes fast
of photo digital electronics
synthesis life time of excited
atomic states fast camera shutter
ultrafast physics
1018 1015 1012 109 106 103 1
atto femto pico nano micro milli s
characteristic time scales / seconds (s)
Fig. 1.4 Enlarged time section from seconds to attoseconds (1018 s) from Fig. 1.3. A presently
very active area of research in AMO is ultrafast and attosecond physics
electronic
W & Z boson
rest mass p+
excitation in atoms,
molecular molecules,
molecular vibrations solids
rotation
1 nK 1 K 1 mK 1K 1000 K 1 MK 1000 MK
LHC lead
room
protons
Fig. 1.5 Energy and equivalent temperature of quantum systems: the scale starts today at one
or several 100 pK, referring to the coldest B OSE E INSTEIN condensates (BEC) and cosmic mi
crowave background radiation (CMB(R)) at 2.725 K on the one end of the scale. On the other end
we find collisions artificially generated at the large hadron collider (LHC) with 14 TeV for protons
and more than 1000 TeV for lead nuclei. AMO is interested mainly in the red marked energy range
1.1 Overview, History and Magnitudes 9
Table 1.4 Orders of magnitude of atomic interactions here for typical examples such as H, alkali
metal atoms and He; the structure of volume 1 of these textbooks essentially follows this energetic
scheme
Interaction Order of magnitude See
cm1 eV kHz K
Pure C OULOMB Z/r 30000 4 1015 43000 Chap. 2
Exchange (in He n = 2) 1000 to 6000 0.12 to 0.7 3 1010 1400 Chap. 7
to 1.81011 to 8600
C OULOMB screening 3000 0.4 1014 4300 Chap. 3
Fine structure (FS) 1 to 1000 104 to 0.1 3 1010 1.4 Chap. 6
to 3 1013 to 1400
External el. mag. fields 1 104 3 1010 1.4 Chap. 8
Hyperfine structure 103 to 1 107 to 104 3 107 1.4 103 Chap. 9
to 3 1010 to 1.4
which should be memorized. We shall often come across this very important con
stant, typically indicating some influence of, or connection with special relativity.
The currently best value of , measured with very high accuracy, is reported in
Appendix A, based on 2010 CODATA (NIST 2010). There, one also finds precise
conversion factors between different units of energy.
We finally mention that one may, from fundamental constants, also construct a so
called P LANCK energy WP = c2 (c/G)1/2 = 1.221 1019 GeV. This is an energy
of cosmic magnitude which may be related to the first moments after the big bang.
There are indications for unification of the fundamental forces at such energy, i.e. at
least three coupling constants might become equal at this energy.
Section summary
The history of AMO physics is identical to the early history of physics at
large. We have recalled some basic observations documenting the quantum
nature of submicroscopic matter.
The essence of (any) physics is to obtain a quantitative description of nature.
Numbers and orders of magnitude are thus essential, and a good feeling for
the scales of lengths, time and energy is important for working in physics.
Section 1.1.2 gives a summary.
Specifically, we memorize P LANCKs constant h 6.63 1034 J s and the
dimensionless fine structure constant 1/137 which will be steady com
panions on our voyage through AMO physics.
10 1 Basics
We cannot give here a serious introduction to the theory of relativity and assume
the reader to be somewhat familiar with E INSTEINs special theory of relativity.
Generally speaking, throughout this textbook relativistic effects will be treated as
a kind of afterthought where necessary (e.g. in Chap. 6). It is, however, useful to
recall here some formulas for later use.
We first remember that N EWTONs equations, specifically his second axiom, re
main fully valid under relativistic conditions. Thus,
dp
=F (1.11)
dt
at high velocities v, has just to be applied to the relativistic momentum:
mv
p= = mv and p = mc = mc 2 1. (1.12)
1 2
Here m is the rest mass of a particle (its intrinsic mass in its rest frame, also called
invariant mass) and
1 v
= with = (1.13)
1 2 c
is the so called L ORENTZ factor. One may also rewrite (1.13) as
2 2 2 = 1 (1.14)
which is invariant under L ORENTZ transformation (it is a constant). For highly rel
ativistic particles 1 one obtains
1 1/ 2 2 . (1.15)
Next we recall the equivalence of mass and energy. In the particles rest frame
the famous E INSTEIN relation reads
2 m2 c4 2 2 m2 c4 = m2 c4 ,
2 m2 c4 p 2 c2 = m2 c4
W 2 p 2 c2 = m2 c4 . (1.17)
1.2 Special Theory of Relativity in a Nutshell 11
holds with energies Wi according to (1.19) to which potential energies may have
to be added due to internal or external fields. In addition
relativistic momentum conservation p i = const
Note that
P 1 P 2 = W1 W2 /c2 p1 p 2
is the inner product of two four vectors P 1 and P 2 . The length of a fourvector is
invariant under L ORENTZ transformation (M INKOWSKI Norm):
respectively. In modern theoretical literature the introduction of this relativistic mass is, however,
usually omitted to avoid confusion: it is the energy that changes with velocity, while the rest mass
m is L ORENTZ invariant.
12 1 Basics
W2
P2 = p2
c2
W 2 p 2 c2 = W 2 p 2 c2 = const. (1.20)
This expression describes a system of particles with total energy and total momen
tum W , p in one frame, and W , p in an other frame of reference, respectively.
This is completely equivalent to (1.17), which was referring to a single particle and
its rest frame on the right hand of the equation. As an example, for two (or more)
particles the transformation from the laboratory system (lab) to the centre of mass
system (cm) would read
2
Wlab plab
2 2
c = Wcm
2
, (1.21)
where in the laboratory frame Wlab and p lab are the sums of relativistic energies and
momenta of all particles, respectively, while Wcm is the sum of all particle energies
in the centre of mass system in which per definition the sum of all momenta is
p cm = 0.
The kinetic energy T of a relativistic particle is defined by
1 + wk /2 3
= 2wk 2wk 1 wk + . (1.24)
1 + wk 4
The approximate equality is an expansion for small kinetic energies. The relation
between the magnitude of the momentum p and the kinetic energy T is derived by
inserting (1.22) into (1.19) and squaring, so that
For small kinetic energies one recovers the classic relation T = p 2 /2m.
A nonrelativistic treatment is only possible if T mc2 . For an electron this is a
rather limited range of energies since
Wrest e = me c2 = 0.511 MeV. (1.26)
1.2 Special Theory of Relativity in a Nutshell 13
We want to transform times and positions that are known in one frame of reference,
say the rest frame of a particle with coordinates x , y , z , t , moving at a velocity
v in respect of another frame, say the laboratory system with coordinates x, y, z, t.
A distance
x in the moving system is seen from the laboratory system at rest as
much shorter (L ORENTZ contraction):
x = x / . (1.27)
Conversely, the time difference
t of two events measured in the moving system
will appear to be stretched in a measurement in the laboratory system (so called time
dilation):
t =
t . (1.28)
We recall the so called twin paradox: the twin brother travelling in a spacecraft at
nearly the speed of light returns only a little bit older, while his twin who remained
on earth has already become an old man.
Closely related is the relativistic D OPPLER shift. Assume, as indicated in
Fig. 1.6, a source moving with velocity v to emit radiation of angular frequency
(wave vector k with k = /c, wavelength = c/ ). In the laboratory sys
tem at rest one registers an angular frequency (wave vector k) at an angle with
respect to v:
k 1 1
= = = = = . (1.29)
k (1 cos ) 2 1 cos
= 2 . (1.31)
14 1 Basics
We shall come back to this remarkable fact in the context of synchrotron radiation
in Sect. 10.6.2.
In the limit of very small velocities v one expands (1.29) in powers of = v/c
to recover the classical D OPPLER shift:
= k v. (1.33)
For later use we also give an expansion in terms of the kinetic energy. For wk =
T /(mc2 ) = 1 1 we obtain from (1.30) (at = 0 and ):
wk
= wk 2wk + wk 2wk 1
2 + . (1.34)
2
Section summary
Most of the relations from special relativity communicated here will be used
quite often throughout this textbook. The reader may want to memorize at
least (1.12), (1.13), (1.17), (1.20), (1.22), and (1.26)(1.28).
Thermodynamics and statistics are together a big and important theme in physics
and physical chemistry. Many substantial textbooks exist on the subject and a vari
ety of important aspects may be found well presented by online scripts in the Inter
net. Here we present only a collection of topics and formulas from this wide field
with particular relevance to atomic and molecular physics. We start with some rather
elementary remarks about exponential probability distributions. They will be exem
plified for spontaneous decay of unstable (excited) states of quantum systems and
for absorption of radiation. We then present a collection of formulas from kinetic
gas theory and end with probability distributions for classical particles, fermions
and bosons. But before going into specific examples we define some general termi
nology.
A probability distribution w(x) describes the probability w(x)dx to find a ran
dom variable between the values x and x + dx. Properly normalized the probability
to find the system with any value of x must be unity:
w(x)dx = 1. (1.35)
0
1. The decay constant A does not depend on the number of particles that are in
cidentally present in the experiment but only on the properties of the object(s)
studied.
2. It does also not depend on the time at which the particle or state decays: the decay
occurs at some arbitrary, a priori unknown time and we cannot predict what that
time will be; we only know the probability Adt that a decay may happen within
a time interval dt.
No predictions can be made about the destiny of a specific particle or state. How
ever a quite accurate prediction can be made for a large ensemble of, say N , parti
cles or objects in unstable states.2
Here and in the following N gives the number of particles in a volume V of interest while
2 Note:
N = N /V refers to the number density (or particle density) measured in units [N] = m3 .
16 1 Basics
dN = N Adt. (1.39)
The minus sign here indicates reduction. If we start at time t = 0 with N0 particles
in one particular initial state, the number N (t) of particles that at time t are still
found in their initial state is obtained by integration:
t
dN
= Adt ln N (t) ln N0 = At
N 0
where we have introduced the mean lifetime = 1/A and the so called half lifetime
1/2 = ln 2/A = ln 2 = 0.692 . The latter is the time during which half of the ini
tial particles have decayed. Often one is interested in the number of decay processes
per unit time in the sample (at time t):3
dN N (t) ln 2 N0 t/
A= = AN (t) = = N (t) = e . (1.41)
dt 1/2
The exponential decay law is of fundamental nature and describes the statistical
(also called spontaneous or natural) decay of excited atomic or molecular states or
of electron hole pairs (excitons) in a solid as well as the decay of radioactive nuclei
(and thus the decrease of radioactive radiation with time) or of elementary particles
(barions, mesons, etc.) such as the decay of the neutron which we shall describe
in Sect. 1.5.3.
The exponential distribution or decay law (1.40) is displayed in Fig. 1.7. One
directly recognizes the significance of the half lifetime 1/2 . Alternatively to a linear
display shown in Fig. 1.7(a) one often uses a logarithmic scale for the probability.
As seen in Fig. 1.7(b) the exponential decay then gives a straight line. This is often
used to recognize exponential decays and to estimate the half lifetime.
Finally, we may recast (1.40) into a probability w(t)dt for any of the initial N0
particles to decay between a time t and t + dt:
1 dN 1
w(t) = = et/ . (1.42)
N0 dt
We note that this probability distribution is normalized according to (1.35): any
given particle will definitely decay at some time between t = 0 and . Conversely,
the probability that a particle has not yet decayed before a time t is exp(t/ ).
And with (1.37) we verify indeed that the mean lifetime (also mean or average
decay time) is t
= = 1/A as introduced in (1.40). The variance (1.38) for the
exponential decay function is 2 .
3 In nuclear physics this is called activity which must not be confused with the decay constant (or
0.8 1/2
1/4
0.6
1/2 1/8
0.1
0.4
1/4
0.2
1/8
0.0 0.01
0 1 2 3 4 5 6 t / 1/2 0 1 2 3 4 5 6 t / 1/2
Fig. 1.7 Exponential decay law (a) in linear, (b) in logarithmic display. Note that the time axis is
scaled here in units of the half lifetime 1/2 so that for t = 1 and t = 2 and t = 31/2 the probability
decreases to 1/2, 1/4 and 1/8, respectively as indicated by the dashed lines
Exponential distribution functions of the kind just discussed do not only describe
probabilities as a function of time. Another important application is the absorption
of fast moving particles or electromagnetic radiation (i.e. photons, including visible
light as well as rays) when passing through matter.
We discuss prototypically the absorption of visible light of intensity I (z) when
passing over a distance z through a medium. The intensity is defined as the total
energy transported in the light beam per unit time and area. It is measured in [I ] =
J s2 m2 = W m2 .
Alternatively, in the particle picture, we may describe the photon flux (z) =
I (z)/(h) with [] = particles s1 m2 as a function of position z (each photon
having an energy h). As indicated in Fig. 1.8, light with intensity I (z) at position
z is absorbed on the way through a medium. In analogy to (1.39) and (1.40), the
intensity dI absorbed over a short distances dz is proportional to that distance and
z
18 1 Basics
at position z. Thus, the result is again an exponential decay law here it is the light
intensity which decays with the optical path z travelled.
The proportionality constant is called absorption coefficient and is measured
in units [] = m1 . It may be understood on an atomistic, statistical basis: the ab
sorber medium may e.g. be an atomic gas in its electronic ground state, it may also
be glass or a liquid. In any case it consists of many absorbing particles (atoms,
molecules, defects in a solid), each of which has a characteristic absorption cross
section ([ ] = m2 ) for the photons which pass through the medium. The absorp
tion coefficient is then the product of cross section and particle density (of the
absorbers), = N .
For a medium of finite thickness d the transmitted intensity is thus given by the
so called L AMBERTB EER law
The probability w(z)dz that any of the photons arriving in the initial flux 0 at z = 0
is absorbed between z and z + dz is given by the probability distribution
1 d 1
w(z) = = ez/ l with l = 1/( N ). (1.46)
0 dz l
In analogy to the notation used in Sect. 1.3.1, here l is called mean free path length
of the photon. It is the distance a photon can on average move freely in the
medium without being absorbed. And exp(z/ l) is the probability that a photon
has survived up to distance z.
The statistical interpretation of the properties of ideal and real gases by the kinetic
theory of gases has played an important role in the history of atomic and molecular
physics. Here we just communicate some basic terminology and results without
derivation.
4 Inthe chemical literature one often writes = ln 10C 2.303C, with C being the con
centration of the absorbent (e.g. in dilute liquids or gases), measured in [C] = mol L1 . The
molar absorption (or extinction) coefficient thus has the dimension [] = L mol1 cm1 =
1000 mol1 cm2 . The so called absorption (or extinction) is then log(I0 /I (x)) = Cx.
1.3 Some Elementary Statistics and Applications 19
One mol of a gas (an SI unit) is defined as the amount of a substance that
contains as many elementary entities (e.g. atoms, molecules, ions, electrons) as there
are atoms in 12 g of the carbon isotope 12 C. The corresponding number of particles
is the AVOGADRO constant:
The relative atomic (or molecular) mass Mr , formerly called standard atomic
(molecular) weight, of an isotope (substance) X is defined as Mr (X) = m(X)/
[m(12 C)/12] where m(X) is the mass of the corresponding atom and m(12 C) that
of 12 C. Note, that one typically finds Mr given in [Mr ] = g mol1 so that the mass
of one atom (molecule) of this substance is m = Mr /NA . Formally, however, Mr is
a dimensionless number and m = (Mr /NA ) g mol1 .
The classical equipartition theorem states that in a system at thermodynamic
equilibrium each degree of freedom that enters quadratically into the total energy
of the system contributes equally to it with kB T /2. The total internal energy of the
complete system is then given by5
kB T RT
U = f NA =f (1.48)
2 2
where f is the number of degrees of freedom per particle, T the absolute tempera
ture of the gas, kB the B OLTZMANN constant (1.83) and R the molar gas constant
R = N A kB .
In an ideal (atomic) gas only the kinetic energy is of relevance and f = 3. Tem
perature then corresponds to the average internal kinetic energy u of the particles
according to
1 3
u = mv 2 = kB T . (1.49)
2 2
These particles
move in the gas with an average velocity v 2 , their average momen
2
tum being m v . From this, one may calculate the pressure p of an ideal gas as the
momentum transfer per unit time and area by elastic collisions and back reflections
with and from the walls containing the gas. Since 1/6 of all atoms move into one
direction we obtains with the particle density N
p = N mv 2 /3 = N kB T . (1.50)
pV = NA kB T = RT (1.51)
5 We use the traditional letter U for this energy/ mol and u for its average per particle.
20 1 Basics
In a real gas particles do collide not only with the walls but also with each
other. With the particle velocity v, the particle flux (i.e. the number of particles
passing through a unit area per unit time) is vN . If is the gas kinetic cross section,
measured in [ ] = m2 , the time tcol and the distance l a particle moves on average
freely between two collisions are
1
tcol = and l = v
tcol , (1.52)
vN
respectively. The bracket
indicates that one has to average over the velocity dis
tribution N(v) of the gas particles. Somewhat more precisely, in the denominator v
is the relative velocity between the colliding particles.
If only one particle species
is of interest their average relative velocity is 2v
. Thus, the so called mean free
path becomes
1
l= , (1.53)
2 N
where is an average gas kinetic cross section. For typical elastic collisions between
atoms and molecules it is on the order of 1019 m2 . Relations similar to (1.53) may
also be written for the absorption of ions and nucleons or of light, X or rays, see
(1.46).
Finally, we come back to the equipartition theorem. For molecules one has to
add 1 to the number of degrees of freedom f for each accessible rotation, and 2
for each vibrational mode (for kinetic and potential energy). Thus, in a solid f =
6 per atom, in a diatomic molecular gas f = 7 per molecule (three translational
coordinates, two rotational axes, one vibrational mode), while for linear and bent
triatomic molecules we have f = 9 and 10, respectively and so on. However, in
these cases one has to account for quantization of rotational and vibrational energies:
they become inaccessible (frozen) at very low temperature. Thus, the equipartition
theorem in its simple form (1.48) holds only at sufficiently high temperature.
When speaking of mean energy, velocity, lifetime or free pathway of particles (pho
tons, atoms, molecules, ions, electrons, nuclei etc.) one implies that these quantities
are described by a statistical distribution. Classically, B OLTZMANN statistics pro
vides the statistical distribution of energies in all fields of physics. It refers to the
probability for finding a certain energy u per particle in an ensemble. This energy
may be kinetic energy or internal excitation of the particles (e.g. electronic, vibra
tional and rotational excitation in a molecule). Quantum physics requires certain
modifications, but for low enough particle densities and sufficiently high tempera
tures these are very small, as we shall see in a moment.
In view of the possible quantization of energy we have to distinguish discrete
and continuous energy states. Energies of the system may be realized by different
quantum mechanical states. Different states i with identical energies ui are denoted
1.3 Some Elementary Statistics and Applications 21
Ni gi
= exp(ui /kB T ) (1.54)
N Z(T )
or dN g(u) exp(u/kB T )du (1.55)
with the total number density N of the particles and the so called partition function
Z(T ) = gi exp(ui /kB T ) which ensures that summation over all states i on the
right hand side of (1.54) gives 1.
Normalization in the continuum case needs some more detailed consideration.
We exemplify this by the distribution of velocities vx , vy , vz in an ideal gas. With
the particle mass m the kinetic energy is u = m(vx2 + vx2 + vx2 )/2 and g(vx , vy , vz ) =
const (since for < vx,y,z < no velocity vector is a priori more probable than
another) one may write the velocity distribution
3/2
which is normalized such that integration over all velocities gives 1. If, on the other
hand, one is interested in the probability of finding a particle with a certain mag
nitude of velocity in a range v and v + dv one has to integrate over all angles so
that dvx dvy dvz = 4v 2 dv. This leads to the well known M AXWELL B OLTZMANN
velocity distribution
dN 2 m 3/2 2 mv 2
= v exp dv, (1.57)
N kB T 2kB T
which is again properly normalized so thatthe integral over all velocities 0 v <
is 1. The most probable velocity is vm = 2kB T /m. One may rewrite (1.57) as an
energy distribution by substituting u = mv 2 /2:
3/2
dN 2 1 u
= u exp du. (1.58)
N kB T kB T
state 1 2 3 1 2 3 1 2 3
Fig. 1.9 How two particles may be distributed onto three states; this illustrates the key difference
between the statistics according to B OLTZMANN (distinguishable classical particles, here red and
black), B OSE E INSTEIN (indistinguishable particles, no further limitations), and F ERMI D IRAC
(indistinguishable particles, PAULI principle)
and thus fully recovers the classical equipartition theorem for three degrees of free
dom in the
form (1.49). It also confirms the expression (1.49) for the averaged square
velocity v 2
= 3kB T /m, while v
= 8T kB /m.
So much about classical statistics. When reconsidering this from a quantum me
chanical point of view one has to account for phase space quantization as well as for
the indistinguishability of identical particles. The former aspect implies that even
the continuum is not completely continuous. Rather, the 6 dimensional phase space
(3 position and 3 momentum coordinates) has a finite cell size h3 . From this and
with the degeneracy ge = 2s + 1 due to the spin s of the particles under consid
eration (see Eq. (1.8)) follows the density of states for a gas of noninteracting
particles in the continuum:
gs (2m)3/2 4 2m3/2
g(u) = u = gs u. (1.60)
4 2 3 h3
Its dimension is Enrg1 L3 . We shall give a derivation of g(u) for the model of a
free electron gas in Sect. 2.4.3 and discuss the quantization of electromagnetic radi
ation in Sect. 2.2.2, Vol. 2. Here we refrain from elaborating on the derivation of the
statistical distributions, refer to the standard textbooks on statistical thermodynam
ics and quantum statistics, and just summarize some key results.
In respect of indistinguishability, quantum mechanics knows two kinds of par
ticles that behave differently: Bosons and fermions which we have introduced al
ready in Sect. 1.1.1 as particles with integer and half integer spin s, respectively.
For fermions (e.g. e , e+ , p, 3 He, etc.) the PAULI exclusion principle (N OBEL prize
1945) holds so that each (discrete or continuum) quantum state can only be occu
pied by one particle at most. In contrast, bosons (e.g. photons, 2 H = D, 4 He, 12 C,
etc.) are not restricted by the PAULI principle, i.e. each state may be occupied by
many particles. But again, identical bosons are indistinguishable in contrast to clas
sical theory. Figure 1.9 illustrates the fundamental differences between the three
1.3 Some Elementary Statistics and Applications 23
statistics for the most simple example: how can two particles be distributed onto
three states?
Detailed considerations on the probabilities for populating (many) energy levels
for many particles lead to the different statistics: the B OLTZMANN distribution for
classical particles, the B OSE E INSTEIN distribution6 for bosons, and the F ERMI 
D IRAC distribution7 for fermions. To compare all three statistics (concentrating on
a quasi continuum of states) we also rewrite the B OLTZMANN distribution (1.55)
suitably:
1
B OLTZMANN dN = g(u)du (1.61)
exp[(u )/(kB T )]
1
F ERMI D IRAC dN = g(u)du (1.62)
exp[(u )/(kB T )] + 1
1
B OSE E INSTEIN dN = g(u)du. (1.63)
exp[(u )/(kB T )] 1
6 B OSE first applied this to photons while E INSTEIN generalized it to any bosons.
7 F ERMI and D IRAC developed it independently from each other in 1926, F ERMI somewhat earlier
than D IRAC.
8 Inthermodynamics, the chemical potential is defined as the partial derivative G/ N of the free
enthalpy G (G IBBS potential) with respect to the particle number N at constant temperature and
pressure. Thus, gives the amount of energy that is necessary to change the number of particles
of a system (by 1) without disturbing the equilibrium of the system.
24 1 Basics
For a gas of free, noninteracting particles, we may insert the density of states
g(u) from (1.60) explicitly, introduce a quantity
4 2m3/2
A = gs (1.64)
N h3
which emphasizes the quantum nature of these statistics, and obtain:
dN udu
B OLTZMANN =A (1.65)
N exp( u
kB T )
dN udu
F ERMI D IRAC =A u (1.66)
N exp( kB T ) + 1
dN udu
B OSE E INSTEIN =A u . (1.67)
N exp( kB T ) 1
gs (2mkB T )2/3
exp( ) = (1.69)
N h3
gs (2mkB T )3/2
or = = ln . (1.70)
kB T N h3
At high temperatures and not too high densities the chemicalpotential is thus neg
ative. Just to obtain some feeling for typical values of this quantity we note that at
normal conditions N = NL = 2.687 1025 m3 and 273 K for He with m 4 u,
s = 0 and gs = 1 one finds exp( ) 252106 compared to which the additive
constant = 1 in the denominators of (1.66) and (1.67), respectively, is irrele
vant. This is characteristic for gases under standard conditions where no difference
among the three statistics is observable, and 0.293 eV is virtually identical
for all three statistics.
However, at low temperatures and/or high particle densities significant differ
ences exist. For F ERMI D IRAC and B OSE E INSTEIN statistics has to be deter
mined from a numerical integration of (1.68) with = 1, respectively.
It is important to note, that for B OSE E INSTEIN gases must not be positive to
avoid singularities in (1.67) (see, however, Sect. 2.2.2, Vol. 2). For = 0 the integral
on the left side of (1.68) reaches its maximum and can be evaluated in closed form.
1.3 Some Elementary Statistics and Applications 25
(a) (b)
Tc = 0.18K Tc = 0.32K
u
0 1 2 3 4 5 6 0 1 2 3 4 5 6
u / k BT
Fig. 1.10 Comparison of the three statistical energy distributions for an atomic mass 4 u
at 100 mbar and two different temperatures (a) 2.5 K and (b) 1 K. The red lines refer to
B OSE E INSTEIN statistics as applicable to He atoms, the grey line illustrates the classical
M AXWELLB OLTZMANN distribution and the black line represents the energy distribution ob
tained from F ERMI D IRAC statistics
Section summary
Statistical distributions are important in many areas of classical and quantum
physics. They describe the probability to find an observable at a certain posi
tion in space or time or with a particular energy, frequency etc. Characteristic
are the mean value (1.37) and the variance (1.38).
Most common is the exponential distribution which we have introduced in
the context of spontaneous decay of excited states (1.40). It may be charac
terized by the halflifetime of the excited states (the time after which half of
the excited states have decayed) which is t1/2 = ln 2/A = ln 2, with the
mean lifetime and A the transition probability. Analogue relations hold for
the absorption of radiation though matter, described by the L AMBERTB EER
absorption law (1.44).
We have reviewed a few basic concepts from kinetic gas theory: the average
free energy per atom and degree of freedom is kB T /2; the mean free path
between two collisions in a gas is l = 1/( 2 N ), with 1015 m2 the gas
kinetic cross section and N the particle density.
We have also discussed the three relevant statistics (1.61)(1.63) for energy
distribution in an ensemble of particles: the classical B OLTZMANN statistics,
F ERMI statistics (valid for fermions, i.e. particles with half integer spin) and
the B OSE E INSTEIN statistics (valid for bosons, particles with integer spin
quantum number). At temperatures a few degrees above the critical tempera
ture (1.71) for B OSE E INSTEIN condensation all three distributions are almost
identical.
From classical wave optics we know that light can be described as electromagnetic
waves: diffraction and interference are the experimental observations onto which
this viewpoint is based. In terms of geometrical optics, light propagation may even
be described simply by so called light rays or beams a notion which can be based
on wave optics as a special case (more in Chap. 1, Vol. 2).
However, light has also particle properties. The key observations documenting
this aspect are summarized in this section. Quantum mechanics, to be addressed in
some detail in the next chapter, reconciles both points of view or rather: it provides
a set of rules for a consistent interpretation of the experimental observations.
One of the fundamental observations on the quantum nature of light is the photo
electric effect. With light of a wavelength (frequency = c/) one illuminates a
metal surface from which electrons emerge. One measures the kinetic energy T of
1.4 The Photon 27
potential energy
e
T
h
WA
electron
sea x
Fig. 1.11 Simple potential well model to explain the photoelectric effect. The photon of energy
h rises an electron from the electron sea in the metal (bound) into the continuum (free). A hole
is left in the sea
these emitted electrons and makes some quite remarkable observations (astonishing
at least at the time early in the 20th century when this was discovered):
T (max) = h WA , (1.72)
where h is the P LANCK constant (1.2) and WA the so called work function or
electron affinity of the metal surface from which electrons emerge (for experi
ments in the gas phase WA has to be replaced by the ionization potential WI of
the atoms or molecules studied).
E INSTEIN (1905) in his annus mirabilis presented the interpretation for the photo
electric effect one of the key steps in the early days of modern physics for which
he received the N OBEL prize in physics9 in 1921: Light energy exists only in well
defined energy packets of
Wph = h = . (1.73)
This energy packet is the elementary quantum of light, called photon. Light obvi
ously has both: wave and particle character.
To get some numerical feeling, let us consider yellow light (from the sun or
from a sodium street lamp) at a wavelength of = 589 nm. With c = we have
= 5.09 1014 Hz. Thus, Wph = h = 3.37 1019 J = 2.10 eV is the energy of
the photon!
One may visualize the photoelectric effect by a very simple potential well model
for quasi free electrons in a metal. The energetic relations between T , WA and h
are illustrated in Fig. 1.11. If the photon has an energy h > WA it may eject an
electron from the electron sea (where it is bound but may freely move within the
metal) into the continuum (where it is unbound). If the electron originates from the
9 Note: not for his at least equally important theory of special relativity.
28 1 Basics
Fig. 1.12 (a) Experimental scheme to study the C OMPTON effect. (b) Kinematics of the scattering
process. Prior to collision energy (momentum) are W (p) and We (pe ) for photon and electron,
respectively; the dashed quantities refer to the situation after the process
The C OMPTON effect (N OBEL prize in 1927) may be observed with an experi
mental setup as sketched schematically in Fig. 1.12(a). Highly energetic photons
( radiation) are scatted from quasi free metal electrons. The momentum of the
photon is
= 0 incoming radiation
(dashed) = 90
= 45 intensity of the
scattered = 135
 radiation
= 0.710 = 0.710 ' = 0.751
tered radiation is shifted in respect of the incoming . The shift depends on the
scattering angle :
= C (1 cos ). (1.75)
10 Often the reduced C OMPTON wavelength /me c = a0 = 3.8110 1012 m is used. In rel
ativistic quantum mechanics one typically measures lengths in units of the reduced C OMPTON
wavelength /me c, and atomic energies in units of me c2 .
30 1 Basics
The photoelectric effect and the C OMPTON effect are two major mechanisms by
which high energy photons interact with matter. For completeness we also mention
pair production: in the vicinity of a nucleus a photon can be converted into an elec
tron and a positron. This process is symbolically written as e +e+ and has the
energy balance
h = 2me c2 + Te + Te+ , (1.77)
so that this process becomes possible if and only if the energy of the photon
h > 2me c2 1.022 MeV, the rest mass of one electron and one positron. The ex
cess energy is converted (essentially) into kinetic energy Te+ +Te of the two emerg
ing particles. For momentum conservation (at threshold the electron and positron
momenta are very small) the process can only occur in the presence of a nucleus
which by C OULOMB interaction carries away the surplus momentum of the
photon. Pair production may also be viewed as excitation of an electron into the
world from the D IRAC sea in which a whole is created (the positron).
All three processes, photoelectric effect, C OMPTON effect and pair production,
are the key mechanisms for absorption of high energy photons in by atoms (specifi
cally in the solid state). We shall come back to this in Sect. 10.5.3.
For completeness we mention that the exactly inverse process is extremely im
probable due to phase space considerations. In contrast, the generation of two pho
tons by positronelectron annihilation e +e+ 2 is a well known process, in
which energy and momentum conservation is easily realized. The two photons are
emitted in exactly opposite direction. This process is exploited in positron emission
tomography (PET), today a widely used medical technique for precise tumour imag
ing. The positron in this case originates from an artificial isotope attached to a drug
which is accumulated specifically in tumour cells. Detecting the two photons after
e e+ annihilation in coincidence allows one to localize their origin in the human
body.
For completeness we mention already here that the particle photon also has an
intrinsic angular momentum , called the photon spin with a quantum number s = 1.
We shall learn more about the experimental evidence in Sect. 4.1.4. The photon
spin will play an important role in various contexts, and a quantum mechanical
description of the photon will be presented in Sect. 2.2, Vol. 2.
We may even attribute a mass to the photon, the equivalent of its energy:
mP h = h/c2 . (1.78)
Note, however, that the rest mass of the photon is zero: it exists only as a particle
moving with the speed of light. We shall later on discuss that this has serious conse
quences for space quantization of its angular momentum. Briefly, the massless parti
1.4 The Photon 31
cle photon has only two substates with sz = jz = , while according to (1.9) a par
ticle with rest mass and angular momentum j = s = 1 has three possible substates.
Electromagnetic radiation is the key for most spectroscopic studies in atomic and
molecular physics. The relevant radiation ranges from radio frequency photons
whose energy is in the eV energy range up to the hard Xray region with energies
up to MeV. Figure 1.14 gives an extended overview on the whole electromagnetic
spectrum of relevance. Note that slightly different definitions are used for the spec
tral ranges shown in Fig. 1.14, depending on the field of application. We follow the
specifications of ISO 21348 (2007).11 Different units are used in different spectral
regions: Frequencies in the very low energy range, wavelengths in the infrared
(IR), in the visible (VIS), ultraviolet (UV) and vacuumultraviolet (VUV) spectral
range.12 For still shorter wavelengths, i.e. in the extreme ultraviolet (XUV), in the
soft and hard Xray region as well as for rays one uses energy units ( in eV,
keV, MeV). In spectroscopy a convenient measure is the reciprocal wavelength, the
so called wavenumber
= 1/ (1.79)
that is proportional to the photon energy
The SI unit of wavenumbers is m1 but commonly one still uses [ ] = cm1 , often
literally called wavenumber. Up to date energy conversion factors are found e.g.
at NIST (2011).
Quantum mechanics, one may say, was triggered by understanding the photoelec
tric effect due to E INSTEIN (1905) (N OBEL prize 1921). But before that (1900),
P LANCKs law had already revolutionized the world of physics (N OBEL prize 1918)
giving an accurate interpretation of black body radiation for which the depen
dence on wavelengths and absolute temperature T had been measured with very
high precision. The characteristic behaviour is illustrated for several temperatures
in Fig. 1.15. The interpretation of this fundamental distribution of radiation forced
Cs atomic clock
9192631770 Hz
FIR
soft
RF EUV Xray
microwave IR visible UV hard Xray ray
XUV
A BC VUV
(c) 1 km 1m 1 mm 1 m 1 nm 1pm 1 fm
Fig. 1.14 Spectrum of electromagnetic waves (for acronyms see text and p. 81ff.). (a) Specific sources and properties, (b) terminology, (c) wavelength scale, (d)
frequency scale, (e) energy scale, (f) examples for applications, (g) examples for methods of generation. Note the narrow range of visible radiation. Only a few
special sources of radiation are specifically emphasized such as CMB(R) and H Lyman. The ultraviolet (UV) regions UVA, UVB, and UVC are abbreviated
1 Basics
here by A, B, and C
1.4 The Photon 33
1.0
1.0 5772 K (sun)
1073 K
3400 K 288 K (earth)
0.5 4000
10 106
u( ) / J m3 m1
2856 K
0.5 10 0 5 10 15 m
1873 K
100
0.0
0 1 2 3 4 5
visible spectrum (390 to 750) nm wavelength / m
Fig. 1.15 P LANCKs law at different temperatures: 5772 K effective black body temperature of
our sun, 3400 K special purpose, shortlived incandescent bulbs, 2856 K CIE standard illumi
nant A (about 100 W tungsten incandescent bulb), 1873 K blast furnace at the discharge aperture,
1073 K dark red glow (kitchen stove, grill fire), 288 K earth surface temperature (average)
8h 3 d
u()d = , (1.81)
c 3 exp(h/kB T ) 1
3 d
u()d = (1.82)
2 c3 exp(/kB T ) 1
Alternatively one plots as done in Fig. 1.15 the distribution as a function of the
wavelength so that [u()]
= J m4 = 103 mJ m3 m1 and
8hc d
u()d = . (1.84)
exp(hc/kB T ) 1
5
34 1 Basics
Equations (1.81) and (1.84) are the standard form of P LANCKs law for the black
body radiation density as originally published by P LANCK (1900). Often one is
also interested in the intensity at a given point, direction and wavelength per unit of
projected area, solid angle and frequency interval with [L ] = W m2 sr1 Hz1
(or [L ] . . . per wavelength interval), also called spectral radiance. Since black
body radiation is intrinsically isotropic, the corresponding expressions are obtained
by simply multiplying (1.81) and (1.84) with c/4 , i.e. the prefactor 8hc is re
placed by 2hc2 . Typically measured is the spectral distribution of the intensity,13
i.e. of the radiation power emitted (per area) into the forward hemisphere, with
[I()] = W m2 nm1 . Integration over the cos angular distribution (projection
of the surface area onto the direction of emission) gives a factor of so that
2hc2 d
I()d = . (1.85)
5 exp(hc/kB T ) 1
The wavelength max at which this spectral distribution of radiation reaches its
maximum, decreases with temperature as seen in Fig. 1.15. Explicitly one finds
(from dI()/d = 0 for the maximum) the so called W IEN wavelength displace
ment law:
max T = b with b = 2.8977721(26) 106 nm K. (1.86)
Finally, the total power emitted per unit area from the surface of a black body
is obtained by integrating (1.85) over all wavelengths. This gives the important
S TEFAN B OLTZMANN law for the intensity of black body radiation:
2 5 kB4 4
I (T ) = I()d = T = B T 4 . (1.87)
0 15 h3 c2
Thus, the (spectrally integrated) intensity depends on the fourth (!) power of
the absolute temperature T . The proportionality factor B = 5.670373(21)
108 W m2 K4 is called S TEFAN B OLTZMANN constant.
At this point, a few words are in order on the radiation which we receive every day
from our sun. Some relevant parameters are summarized in Table 1.5. The spectral
distribution of the solar radiation at the top of the earth atmosphere is shown in
Fig. 1.16 (wiggly red line). Since about 2004 it is constantly monitored by satellite.
The solar constant S is the integral over this spectrum. The daily results can be
obtained online from SORCE (2012) (the data reported there are renormalized to
13 In radiometry one uses the terms irradiance or radiant flux (see also Sect. 1.4.8). For consistency
with the general custom in AMO physics we usually call this quantity intensity of the radiation,
measured in [I ] = W m2 .
1.4 The Photon 35
Table 1.5 Some properties of sun and earth in the context of solar radiation (data from SSE 2012,
unless otherwise specified)
Mean radius sun R 6.9551 105 km
Mean radius earth RE 6371.0 km
Mean distance suneartha RSE 149.60 106 km 1 uab
Solar constantc S 1360.8(5) W m2 (from KOPP and L EAN 2011)
Radiation power emitted by sun 384 109 PW (into 4 sr)
Radiation power received by earthd 173.5 PW (from the sun)
Effective temperature TS 5772 K (for the above value of S)
Sun surface temperature Tph 4400 K to 6600 K (top to bottom photosphere)
Albedo (B OND) a 0.306 (fraction of radiation reflected)
Earth temperaturee TE 254 K (effective black body)
TEa 288 K (average at surface)
a Dueto the ellipticity of the orbit the distance varies between parhelion (minimum) and aphelion
(maximum) by about 6.9 % between 4th of January and 4th of July
b Per definition the astronomical unit of length is 1 ua = 149597870700 m; 1 ua is almost identical
O2 PLANCK at 5772 K
I () / W m2 nm1
1.0 0.05
H 2O
solar H2O 20
H2O
~
O3 spectrum
at sea H2O 20
H2O,
level CO2
0.0
0 1000 2000 3000 4000
UV vis IR wavelength / nm
Fig. 1.16 Spectral intensity distribution I() of the solar radiation: as measured by SORCE
(2012) above the earths atmosphere (for > 2400 nm from ASTM 2008); at sea level (AM1.5
global tilt spectrum from ASTM 2008, see text); black body radiator at 5772 K. Also indicated
are the main absorbing molecules in the earths atmosphere
36 1 Basics
Fig. 1.17 Definition of air mass coefficient AM for solar radiation standards. Table on the right:
solar radiation (integrated from 280 to 4000 nm) at normal incidence for different zenith angles
a the astronomical unit of length 1 ua, see Table 1.5). The spectral distribution is
remarkably stable over time, the average changes being less than the width of the
red line in Fig. 1.16. As illustrated, the distribution is approximated reasonably well
by a P LANCK distribution from a black body at 5772 K (smooth black line). The
latter is obtained from (1.85) by multiplying it with (R /1 ua)2 , where R is the
solar radius. No further scaling is needed to obtain this fit.
As we see, the fit is not perfect, but considering the gigantic nuclear fusion re
actor which our sun actually is, and considering its complicated photosphere, the
relative similarity with a black body radiator is quite remarkable and so is the sta
bility of the distribution. A variety of values for the sun temperature may be found
in the literature. From the recent, highly accurate measurement of the solar constant,
S = 1360.8 W m2 , by KOPP and L EAN (2011) (see also SORCE 2012) and with
appropriate scaling of the S TEFAN B OLTZMANN law (1.87), we obtain the effective
black body temperature of the sun as TS = (1 ua /R )1/2 (S/B )1/4 = 5772 K.
Also shown in Fig. 1.16 is the spectral distribution of radiation which reaches the
ground level after partial absorption and scattering of the incoming solar radiation
by atmospheric gases. As sketched in Fig. 1.17, this obviously depends on the zenith
angle of the sun ( = 90 latitude angle). For not too large the optical path
length through the atmosphere is given by hx = h0 / cos , where h0 7.7 km is the
effective vertical height of the earths atmosphere (i.e. the height where the pressure
has dropped to 1/e of its value on ground). The ratio
hx 1
AM = (1.88)
h0 cos
is called air mass coefficient. An approximate empirical formula (see e.g. H ONS 
BERG and B OWDEN 2012) for the intensity reaching the earth surface is
0.678
I = 1.1 I0 0.7AM , (1.89)
1.4 The Photon 37
where I0 is the radiation arriving on top of the atmosphere (essentially the solar
constant S, given in Table 1.5). The prefactor 1.1 accounts for radiation scattered
from the air and retroreflected from the ground, the exponentials reflect somehow
the L AMBERTB EER absorption law. With reference to Fig. 1.17 we also note that
the intensity depends in addition on the angle of incidence ( ) onto the receiving
surface, Isur = I cos( ), with (1.89) referring to normal incidence.
As a standard value, AM = 1.5 (short AM1.5) has been adopted, considered rep
resentative for most industrialized countries at noon time. Based on measurements
and modelling, one also has defined two standards spectral distributions arriving on
the earth surface, both for AM1.5: (i) direct normal incidence of radiation and (ii)
hemispherical (global) incidence on a 37 degree tilted surface, which also includes
scattered and retroreflected radiation (ASTM 2008, G1733).14 The latter spectrum
is shown in Fig. 1.16 and referenced the table with Fig. 1.17.
We note at this point that this spectrum has its maximum at ca. 500 nm, i.e. near
the maximum of the spectral sensitivity of the human eye at 555 nm. The evolution
ary context behind this coincidence is evident.
In contrast, the surface of our earth with an average temperature of ca. 288 K
emits in the IR spectral range with a maximum at about 10 m where the green
house gas CO2 absorbs (we shall come back to this in Sect. 5.3.1, Vol. 2) fortu
nately we have to say: without it, the temperature on the earth surface would cor
respond to the effective black body temperature of the earth TE = 254 K and life
would not be possible.
Black body radiators are very inefficient when used to generate visible light. Even
the sun emits only a fraction of its total radiation energy into the visible (VIS) spec
tral region: only 46 % of the total radiation power are emitted between 380 and
760 nm, as one finds from integrating (1.87). Incandescent light bulbs at typical
temperatures convert much less energy into the visible spectral range (e.g. 6.5 %
for a 100 W bulb with its tungsten wire at 2856 K). Worldwide it is understood that
intense efforts are needed to save energy. Primary energy resources must be used
in a most efficient way and energy efficient lighting is an import potential. As a
consequence, the good old light bulb, which humankind has gotten so used to since
over 100 years, is approaching its last days very rapidly.
The overall efficiency in creating luminosity as registered by the human eye is
even much worse, since u() has to be multiplied by the physiological sensitiv
ity V () of the eye. This so called photopic luminous efficiency function peaks at
555 nm,15 where the human eye has its maximum sensitivity. It is standardized by
14 See also ISO 609043 (2008) or DIN EN 609043. Note that for AM1.5 normal incidence and
Table 1.6 Relation between photometric and radiometric quantities after O HNO (2010)
Photometric Unit Relation to lm Radiometric Unit
Luminous flux lm (lumen) Radiant flux W (Watt)
Luminous intensity cd (candela) lm sr1 Radiant intensity W1 sr1
Illuminance lx (lux) lm m2 Irradiance W m2
Luminance cd m2 lm sr1 m2 Radiance W sr1 m2
Luminous exitance lm m2 Radiant exitance W m2
Luminous exposure lx s Radiant exposure J m2
Luminous energy lm s Radiant energy J (joule)
Color temperature K (Kelvin) Radiance temperature K
the Commission international de lclairage (CIE) and forms the basis for all lu
minosity determinations. For back of an envelope calculation it may roughly be
approximated by a Gaussian, centred at 560 nm with a FWHM 104 nm (from a
least square fit). But for precise work tabulated values from the literature have to be
used, e.g. from CIE (or as plotted by D ICK LYON 2008, with detailed references).
For a brief and concise introduction into photometry we refer to O HNO (2010)
from whom we also have adopted Table 1.6. It compares (physiologically weighted)
photometric quantities with (directly energy related, physical) radiometric quanti
ties.16 The relevant photometric SI unit is the candela, cd, defined as . . . the lumi
nous intensity, in a given direction, of a source that emits monochromatic radiation
of frequency17 540 1012 Hz and that has a radiant intensity in that direction of
(1/683) W per sr.
However, the physically more fundamental unit is the lumen (lm), the unit for the
luminous flux. It measures the photometric equivalent of the total radiation power
emitted by a given light source. Thus, the hypothetical, ideal light source driven by
1 W electric power emits 683 lm at 555 nm. It is said to have an overall luminous
efficacy of 683 lm W1 and its luminous efficiency is defined as 100 % it emits the
maximum possible amount of visible light.
All other light sources have (much) lower luminous efficacy and efficiency. One
may compute the
V ()I()d
luminous efficiency = 0 (1.90)
I()d
0
16 Note that this radiometric terminology somewhat confusingly differs from the standards used
elsewhere in optical and laser physics. A laser beam e.g. is typically characterized by its intensity
measured in units [I ] = W m2 , while here in radiometry the corresponding quantities in Table 1.6
are called irradiance or radiant exitance.
17 That is at a wavelength of 555 nm where the human eye has its maximum sensitivity.
1.4 The Photon 39
ted per unit area. This luminous efficiency is a function of T and has a maximum of
about 14.5 % at T = 7000 K. For our sun it is ca. 14 %, while for the above men
tioned CIE standard Illuminant A the efficiency is only about 2.5 %, correspond
ing to a luminous efficacy of ca. 17 lm W1 . A typical standard 100 W incandescent
bulb emits a total luminous flux of 1360 lm and thus its efficacy is 13.6 lm W1 . Of
all the electric energy used to heat the tungsten wire of the bulb, 98 % is lost as IR
radiation and eventually dissipated as heat! So called halogen lamps are somewhat
more efficient since a special chemical process reduces evaporation of the cathode
material and the temperature of the tungsten wire can be significantly higher. Still,
the general efficiency problem of incandescent lamps remains.
Thus, the challenge is, to exploit more efficient ways for converting electrical
energy into visible light. One direction that currently is followed uses miniaturized
fluorescent lamps. Typically, in a mercury gas discharge UV light is generated at
253.7 and 185 nm. It is then converted into visible light of different wavelengths
by fluorescing materials, so called phosphors. Typically three to five different phos
phors with reasonable efficiency are used, whose fluorescence bands by additive
colour mixing appear more or less as white light. Much current development is
focused onto these materials. By judicious choice of the fluorescent bands one may
obtain a more or less continuous coverage of the visible range thus imitating the
visible part of the black body radiation without loosing energy into other spectral
regions.
The colour temperature of such a source is defined as the temperature of a black
body radiator that best matches that spectrum in the visible. Presently, the lumi
nous efficacy of compact fluorescent lamps (commercial energy saving lamps) is
on the order of (50 to 60) lm W1 (7.3 % to 8.7 % efficiency) but up to more
than 100 lm W1 (14.6 % efficiency) can be achieved with long tubular fluorescent
lamps. For street lighting also high and low pressure sodium discharge lamps are
used, easily recognized by their yelloworange light (around 590 nm). Low pres
sure Na discharges have the highest luminous efficacy in stateoftheart lighting
technology of up to 200 lm W1 (29 % efficiency).
Light emitting diodes (LED) could, in principle, be able to supersede these val
ues: they convert electric current so to say more or less directly into light. Again,
additive colour mixing of at least three LEDs is required for generating white light
in practical devices. The luminous efficiency is presently of a similar order of magni
tude as for fluorescent lamps, somewhere between 15 and 25 %. However, strategies
are discussed to achieve much higher efficiencies (e.g. B RETSCHNEIDER 2007), and
the price for such hightech products typically decreases exponentially with years
of experience and mass production.
Occasionally the question comes up about a theoretical maximum of luminous
efficiency. It is difficult to answer: consider a (very hypothetical) light source con
verting electric power to a 100 % into a spectrum, which in the visible (380 and
760 nm) corresponds exactly to the ideal white light of our sun, and is zero else
where. According to (1.90) this would correspond to a luminous efficiency of 38 %.
Hence, conversion of 100 % electric energy into visible light is per definition
a completely unrealistic goal: it would mean totally green illumination. Thus, any
40 1 Basics
d lattice planes
s
d
claims for a luminous efficiency higher than 3438 % have to make massive com
promises on the effective colour temperature. We shall certainly witness an exciting
development in the lighting industry over the coming years.
Electromagnetic radiation in all spectral ranges light in the widest sense is today
one of the most important tools to reveal the structure and dynamics of matter. In
later chapters we shall learn a lot about the various spectroscopic techniques that are
used in this context.
At this point we want to mention, at least briefly, one of the key methods for struc
tural analysis: Xray diffraction, i.e. scattering and interference of electromagnetic
radiation of very short wavelength from crystalline matter. It is treated systemati
cally in textbooks and monographs on solid state physics (see also Sect. 1.7.2).
The basis for a variety of such methods is multiple beam interference from the
crystal lattice structure studied. As sketched in Fig. 1.18 a multitude of parallel, so
called lattice planes may be thought to reflect the Xrays. According to Fig. 1.18
the optical path difference between two rays reflected from neighbouring planes is
2s = 2d sin , where d is the distance of two lattice planes in the crystal and the
so called B RAGG angle.
Thus, the reflected Xray beams (shown in Fig. 1.18 are only two of them) inter
fere constructively if and only if B RAGGs law (also called B RAGG condition)
holds. Here is the wavelength of the scatted Xray radiation and z is an integer.
For reference we also report some further quantitative relations relevant to Xray
scattering. One defines a reciprocal lattice vector
G = hg 1 + kg 2 + lg 3 (1.92)
1.4 The Photon 41
constructed with the basis vectors of the unit cell g 1 , g 2 , g 3 in the reciprocal lattice,
using the so called M ILLER indices h, k, l that characterize the lattice planes in the
crystal. Without entering into details of crystal lattice theory we just note that these
basis vectors in the reciprocal lattice relate to the ordinary basis vectors a 1 , a 2 , a 3
in the crystal lattice by
g i a j = 2ij , (1.93)
and in Fig. 1.18 we identify d 2/G. With wave vectors k and k of incoming
and outgoing Xray radiation B RAGGs law (1.91) is now written
k = k k = G. (1.94)
Normally one investigates only elastic scattering where k = k  and one may
rewrite (1.94) as (k G)2 = k 2 or G2 2kG = 0. We may finally write the condi
tion for diffraction:
2
G G
2kG = G 2
or k = . (1.95)
2 2
In the latter form the diffracted wave vectors describe a plane bisecting the recip
rocal lattice vector G and being perpendicular to it. One may construct such planes
for all elementary reciprocal lattice vectors and combine them to a closed surface
in reciprocal lattice space. They form the so called B RILLOUIN zone (BZ) that rep
resents all wave vectors of radiation that can be B RAGG reflected by the crystal.
BZs are a very important concept in solid state physics, specifically relevant for the
theory of band structure as we shall briefly discuss in Sect. 2.8. Since several dif
ferent lattice vectors with different alignment in space exist (typically more than 3,
see Eq. (1.92)), BZs may be rather complex surfaces. Corresponding surfaces may
also be constructed from 2G, 3G etc. and one distinguishes the 1st, 2nd, 3rd, etc.
B RILLOUIN zone.
The intensity of the diffracted Xray radiation depends on the (electron) charge
density distribution (r) in the unit cell of the crystal and is proportional to the
absolute square of the so called structure factor
F(hkl) = d3 r(r) exp(iG r) (1.96)
cell
= Fj (G) exp i2(xj h + yj k + zj l) .
j
The summation has to be carried out over all atoms in the unit cell of the crystal,
each of which is characterized by its atomic form factor18
Fj (q) = d3 rN (j ) (r) exp(iq r). (1.97)
atom
(a) (b)
Fig. 1.19 (a) Xray diffraction image from a large single crystal (80 100 50 m) of the human
enzyme prolidase, recorded with h = 13.05 keV at the beamline BL14.1 of the Free University
Berlin at BESSY.The maximum resolution of the diffraction image corresponds a lattice plane
distance of the crystal lattice of 0.25 nm. In the squares magnified sections of the image are shown.
(b) Secondary structure mode of the enzyme in dimer form with unbound Mn2+ (red spheres).
With kind permission from M UELLER et al. (2007) and private communications
These form factors have to be determined for each atom by integration over the par
ticle density N (j ) (r) of all its electrons (the dimension of N being L3 ). Without
entering into the details we note here that the imaginary part of the atomic form fac
tor is related to the photoabsorption cross section, while the real part characterizes
the elastic (coherent) photon scattering. For radially symmetric charge distributions
it is
(j )
N (r) sin(qr) 2
Re Fj (q) = 4 r dr, (1.98)
0 qr
Section summary
Photons have (i) a well defined energy W = h = (with = c/ and
= ck) as determined by the photoelectric effect, (ii) a momentum p = k
documented by the C OMPTON effect, and (iii) an angular momentum . They
have no rest mass and exist only moving with the speed of light.
The spectrum of electromagnetic radiation (we call it light in a general
sense) ranges from radio frequencies ( km, h 109 eV) to rays
(105 nm, h 108 eV). The visible spectrum is only a very small part of
it (the wavelength range from 380 and 760 nm).
P LANCKs radiation law (1.81) was a corner stone in the development of quan
tum physics. It describes the spectrum of a black body such as our sun or
incandescent light bulbs. The maximum of the spectrum shifts with tempera
ture according to W IENs displacement law max T 2.9 106 nm K, the total
intensity (per unit area) follows the S TEFAN B OLTZMANN law I (T ) = B T 4 .
A hypothetical, ideal light source with a luminous efficiency of 100 %, driven
by 1 W electric power emits 683 lm (lumen) at 555 nm. Incandescent light
bulbs have a luminous efficiency of only 23 %.
Xray radiation is a very powerful tool for structural analysis. Constructive
interference occurs at angles of incidence (in respect of the lattice plane)
according to the B RAGG law 2d sin = z. Other formulations of the B RAGG
law make use of the reciprocal lattice vector (1.92) and M ILLER indices. The
structure factor (1.96) and the atomic form factor (1.97) describe the intensi
ties in the diffraction pattern.
We cannot give here an even brief introduction into particle physics and the un
derlying theory. The so called standard model (SM) of quantum chromodynamics
(QCD) is a sophisticated theory which cannot be treated in passing. However, it is
important to know a few basics when discussing the interactions relevant in atomic,
molecular and optical physics. The four fundamental interactions
1. Gravitation
2. Electromagnetic interaction
3. Weak interaction
4. Strong interaction
may be seen as what holds the world together in its innermost folds as far as we
understand it today.
In our daily life we are confronted almost exclusively with the first two of these
forces in particular gravitation plays a key role in our everyday experience, while
electromagnetic interactions are perceived commonly in a more indirect manner:
through the action of various machines and equipment, via lighting gears (or occa
44 1 Basics
Table 1.7 The four fundamental interactions, exchange bosons and coupling constants
Interaction Fermion Exchange Mass Couplingb Range Dependence
(e.g.) boson / GeV c2 /m on distance
Gravitation e , p, n Gravitona 0 5.9 1039 1/r 2
Electromagn. e , p Photon 0 7.30 103 1/r 2
Weak e , W boson 80.4 105 1018 1/r 5
Z0 boson 91.2 107 1/r 7
Strong p, n meson 135.139 1 1015 1/r 7
Quarks Gluons 0 0.119
a Hypothetical, not yet found
b Values for gravitation and electromagnetic interaction refer to a pair of protons
The interaction of two particles depends in general in a complex manner from their
distance r. Only the well known force laws for gravitation
1.5 The Four Fundamental Interactions 45
m1 m2 r
F g = G (1.99)
r2 r
1 q1 q2 r
Fe = (1.100)
40 r 2 r
can be written in this simple, closed form, with masses m1 , m2 and charges q1 , q2 of
the interacting particles, respectively. They may be derived from a scalar potential
V (r) 1/r by
19 We note here that for very high energies the electromagnetic coupling constant em changes
(essentially ln W ); at energies 90 GeV (roughly corresponding to the mass of the Z boson)
experiments have determined 1/em 1/128.
46 1 Basics
The other two forces, strong interaction and weak interaction, play an important role
on a subatomic length scale only: as indicated in Table 1.7, their range of interaction
is finite in complete contrast to C OULOMBs law and gravitation. Strong interac
tion is mainly responsible for the cohesion of nuclear matter, i.e. it takes care of
the stability of nuclei which otherwise under the influence of electrostatic repulsion
would simply explode. Weak interaction on the other hand plays a decisive role in
decay, i.e. when an electron is emitted from an atomic nucleus.
Electromagnetic, strong and weak interaction are described today consistently
and convincingly by the standard model of QCD only gravitation resists so far
a unified interpretation. At the core of the theory is a set of objects that may be
considered point like20 (<1018 m) real elementary particles without any internal
structure but with some well defined properties in terms of mass, charge and spin
(intrinsic angular momentum). They are summarized in Fig. 1.20.
The standard model distinguishes two varieties of elementary particles:
1. Matter particles
6 quarks are subject to strong interaction and carry one of three colour charges
red, green or blue; they also carry a charge +2/3e or 1/3e and are
subject to electromagnetic as well as to weak interaction.
6 leptons are all subject to weak but not to strong interaction i.e. they are
colourless. The electron, muon and tauon carry also an electric charge e
and are subject to electromagnetic interaction. In contrast, the corresponding
neutrinos do not have an electric charge. The have, however, a (very small)
mass.
All of these are fermions with spin 1/2. They are complemented (not shown
in Fig. 1.20) by the corresponding antiquarks (in the colours antired, antigreen,
antiblue) and antileptons, their charges are opposite.
2. Exchange particles (or gauge bosons) are considered to be exchanged between
the matter particles during interaction. They are all bosons with spin quantum
number s = 1 and may be seen as mediators of the forces. They are formed for a
very short time only, i.e. virtually, and are unobservable directly.21
0 0 0 electroweak
colourless
Fig. 1.20 Elementary particles according to the standard model: matter particles (fermions, spin
1/2) and exchange particles (gauge bosons, spin 1). The masses and charges are indicated:
masses are given in terms of their energy equivalent Wm = mc2 . *) According to CERN (2013)
On 4 July 2012, the ATLAS and CMS experiments at CERNs Large Hadron Collider announced
they had each observed a new particle in the mass region around 126 GeV /c2 . . . consistent with
the H IGGS boson but it will take further work . . . . This is based on theoretical work for which
E NGLERT and H IGGS (2013) received the N OBEL prize
We have discussed already in Sect. 1.4 the photon. It is the exchange particle
for electromagnetic interaction by which all charged particles interact (in addition
to weak or strong interaction and to gravitation). Strong interaction is characteristic
for quarks as well as for antiquarks (not shown in Fig. 1.20) with charge 2/3e
and +1/3e. They are never observed as free particles and occur only as compound
systems (confinement). On the other hand, at very close distance their interaction
tends to zero (so called asymptotic freedom).
In analogy to electric charge one attributes to the quarks a colour charge by
which they interact with each other. The colour comes in three varieties red, green
or blue (in our graphs coloured red, grey or black). In each composite particle the
colours have to add in such a manner that their sum leads to white (in analogy to
the colours observed from the visible spectrum of light). Antiquarks are coloured
with complementary colours antired, antigreen or antiblue. The exchange bosons
of the strong interaction are called gluons (reminding us of glue). They come in
8 different varieties encompassing one colour and one anticolour each (while the
48 1 Basics
Proton Mass
mp = 1836.15 me
u
Charge = 1e0 = 1.67261027 kg
u Spin = 1/2 = 938.27231 MeV/c 2
d stable = 1.007 276 47 u
Neutron
Charge = 0 mn = 1838.68 me
u Spin = 1/2 = 1.6749 1027 kg
d unstable = 939.5656 MeV/c 2
d t1/2 = 10 min 14s = 1.008664 92 u
Fig. 1.21 Structure of the two nucleons: proton and neutron. They are built from up and down
quarks (u and d, respectively). The special choice of colours is here without direct relevance and is
supposed to just illustrate the principle of colours in quantum chromodynamics: each of the three
fundamental colours (commonly called red, green, blue) has to be present exactly once. The wiggly
lines indicate exchange of gluons (strong interaction)
photon does not contain any charge). Both, gluons and the photon are massless
particles while the three exchange bosons of weak interaction Z0 , W+ and W are
again colourless but have each a (different) mass.
1.5.3 Hadrons
u u
n d d p
d u
W _
e
Fig. 1.22 F EYNMAN diagram for the neutron decay by weak interaction (half lifetime
1/2 = 10 min +14 s): by emission of an electron e and an antielectron neutrino e a down quark
d (charge 1/3e) is changed into an up quark u (charge +2/3e). The interaction occurs via a
virtual W boson
In contrast to the barions the so called mesons (medium heavy particles) are built
from a quark and an antiquark with complementary colours (colour and anticolour),
which makes them also white. Barions and mesons together are called hadrons.
Hadrons are all subject to strong interaction, to weak interaction and as far as
electrically charged also subject to electromagnetic interaction.
The leptons electron, muon and tauon and their respective neutrinos are, in contrast
to the hadrons genuine elementary particles, at least as far as we know today. They
interact among each other as well as with quarks, protons and other barions and
atomic nuclei by electroweak and (as far as charged) by electromagnetic interac
tion.
In atomic, molecular and optical physics we are almost exclusively concerned
about electromagnetic interactions and by far the most important particle is the elec
tron. Let us have a somewhat closer look at it. Its rest mass is
Generally, free and isolated particles are only observed with an integer multiple
of e (positive, zero or negative). We point out again, however, that the bound quarks
and antiquarks have charges 2/3e or 1/3e.
Directly related to the elementary charge is the energy unit electronvolt
Section summary
The four fundamental interaction responsible for keeping the world at its
innermost together are gravitation, electromagnetic interaction, weak inter
action, and strong interaction. Gravitation and electromagnetic interaction en
ergies are 1/r, i.e. have in principle infinite range, and are dominant in
macroscopic physics and atomic physics, respectively. Weak and strong inter
action are of short range on the order 1018 and 1015 m, respectively, and
from an atomic physics point of view refer to forces inside the nuclei.
The standard model of elementary particle physics describes electromagnetic,
weak and strong interaction in a unified scheme. It comprises 6 quarks (matter
particles, fermions), 6 leptons (including the electron, all fermions, subject
to weak and electromagnetic interaction, if charged) and 4 types of exchange
bosons, including the photon which mediates electromagnetic interaction. The
quarks have charges 2e/3, e/3, spin 1/2 and come with three kinds of
colour charges. They are subject to weak, strong and electromagnetic forces.
They are building blocks of all other massive particles, including the nucleons
(neutron and proton) which in turn are constituents of atomic nuclei.
Leptons are a different kind of elementary particles, subject only to weak
and electromagnetic interaction (if charged). Of particular interest in atomic
and molecular physics is the electron. We memorize its electric charge e
1.6 1019 C. To move it through an electric field requires energy or sets
energy free. The atomic unit of energy is thus 1 eV 1.6 1019 J.
dp
F= = q(E + v B). (1.105)
dt
In this form the equation is correct even in the relativistic limit. However, for sim
plicity we shall restrict the following discussion to the nonrelativistic limit unless
pointed out differently.
Geometry (b and ) as well as the voltage U applied are known. The ratio (and only
the ratio) of kinetic energy T to charge q may be determined according to (1.106) by
the electric deflection method! This is valid in general, completely independent of
how complicated the setup may be. In practical applications one often uses special
capacitor designs e.g. cylindrical or hemispherical capacitors (the latter is described
in some detail in Appendix B.3, Vol. 2) as well as segments of these for determining
the particle energy if its charge is known. In addition to a measurement of the energy,
for extended, divergent particle beams such setups allow to focus the trajectories
from the entrance aperture onto the exit aperture of an energy analyzer.
To also determine the mass of a particle one needs an additional measurable quan
tity. In a purely magnetic field B the movement of the charge q is determined by
dp
F= = qv B
dt
with the force F being perpendicular to v and B (see Fig. 1.25). The change in
energy
dW
= vF = qv v B 0
dt
is thus identical to zero, and v = const. This holds also relativistically so that = 0
and = const.
The relativistic equation of motion thus reads
dp dv
= m = qv B,
dt dt
with the rest mass m of the particle and p = mv. The particle moves on a circle
that is determined by balancing the centrifugal force Fc and the magnetic force
F = qvB:
dp 2 2
= Fc = mv = mv = qvB, (1.107)
dt r r
p
so that mv = p = qBr or = rB (1.108)
q
mv p
and r = = . (1.109)
qB qB
If one wants to use this orbital radius r for mass selection, again only one parameter
can be extracted: with a magnetic field the ratio of momentum and charge p/q can
be measured.
The expressions given above also hold in the nonrelativistic case where 1.
For highly relativistic energies as they are e.g. realized in electron storage rings built
for the generation of synchrotron radiation (see Sect. 10.6.2) one has more or less
v
54 1 Basics
If one combines the two types of fields judiciously one may achieve with T =
p 2 /2m a determination of m/q:
m 1 p 2 /q 2
= . (1.112)
q 2 T /q
In the classical, so called double focussing mass spectrometers, ions pass succes
sively electric and magnetic fields. Such devices are still used today but have lost
much of their importance in relation to the FTICR mass spectrometers discussed in
Sect. 1.6.3.
Another type of mass spectrometers, today widely used, is the so called
quadrupole mass spectrometer (QMS). It exploits the mass dependence of the dy
namic stability regions for ion trajectories in a rapidly oscillating electric quadrupole
field (see e.g. D EHMELT and PAUL 1989): for specific ratios of DC to AC fields only
ions with a specific mass to charge ratio m/q may pass on a stable trajectory through
the setup while all other ions undergo increasing oscillations and do not reach the
exit slit. The interesting mathematics follows M ATHIEUSs differential equation (see
e.g. W IKIPEDIA CONTRIBUTORS 2013).
In time of flight mass spectrometers one accelerates ion in electric fields to select
m/q through their time of flight to the detector. In stateoftheart setups of this type
one exploits a variety of tricks to focus ions of the same mass but from different
starting positions at the ion source (W ILEY and M C L AREN 1955) as well as those
with different initial kinetic energies (M AMYRIN 1994) at the same time on the
detector. With present commercial solutions high mass to charge ratios m/q (up
to 4000 u) may be reached and high mass resolution (m/
m 60 000) can be
achieved.
For special applications two other, classical setups are still useful and deserve to
be mentioned. We already mentioned J.J. T HOMSONs method with parallel electric
(E) and magnetic (B) fields which nowadays is used to analyze particle emission
with high kinetic energies T . Assume the fields to point into +ydirection, the ions
enter the field perpendicular in +zdirection (velocity vz ) and pass through them on
a length l. In this case the L ORENTZ force (1.105) leads to acceleration both in y
as well as in xdirection. After a time t1 = l/vz the particles have passed the fields
and acquired velocities
Behind the fields the ions are detected on a position sensitive detector: J.J. T HOM 
SON used a photographic plate, today sophisticated electronic devices are available
as we shall describe in Appendix B.1, Vol. 2. If the detector is positioned at z = s
far behind the field, the ions hit the detector at a time t2 = s/vz and their deflections
are
qEls qEls qBls qBls
y = v y t2 = 2
= and x = vx t2 = = . (1.114)
mvz 2Wkin mvz 2mWkin
56 1 Basics
v = E/B
the charged particles just fly on straight line trajectories. If the mass is known one
may thus determine the momentum and hence the kinetic energy.
Electrons in plasmas but also in clusters and condensed matter (in particular quasi
free electrons in metals and semiconductors) may carry out collective oscillations,
so called plasma oscillations with a characteristic frequency that plays an impor
tant role in many areas of physics. The simplest model starts with is a quasi neutral
plasma of electrons e and ions with a density N of quasi free electrons, equal to
that of the ions. Assume we displace all electrons in respect of the ions by a distance
x as shown in Fig. 1.26.
This leads to a surplus of surface charge density = eN x that causes an elec
tric field between electrons and ions E = /r 0 . The equation of motion for each
electron in this field is thus
me x = eE = e/r 0 = e2 N/r 0 x.
This differential equation is simply that of a harmonic oscillator with a force con
stant kP = e2 N/r 0 . The angular frequency of the oscillations
kP N e2
p = = (1.116)
me me r 0
1.7 Particles and Waves 57
p2
r () = 1 , (1.117)
2
which for < p becomes negative.
To interpret this result, we consider a plane wave exp[i(kx t)] propagat
ing in +xdirection, with k = n/c being the magnitude of the wave vector in
the medium and n = r the index of refraction. According to (1.117), below the
plasma frequency n becomes imaginary and we write k = i. Correspondingly the
wave, exp(x) exp(it), is damped. Thus, the medium absorbs at all frequen
cies < p and is transparent for > p .
We finally note that in isolated particles, the situation is somewhat more compli
cated, e.g. so called plasmon resonance phenomena will be observed at p .
Section summary
In an electromagnetic field E, B particles with a charge q are subject to the
L ORENTZ force F = q(E + v B), where v is the particles velocity.
This can be used to manipulate and guide the charged particles (typically elec
trons and ions) and to select their energy and/or mass/charge ratio.
We note two interesting frequencies: (1) In a homogeneous magnetic field B
a free electron moves on a circle, perpendicular to the field (or spirales) with
the cyclotron
frequency c = v/r = eB/( me ). (2) The plasma frequency
p = Ne 2 /(m e r 0 ) is characteristic for oscillations of quasi free electrons
in clusters and metals (particle density N ).
screen
cathode +U
aperture
Wkin low Wkin high
Fig. 1.27 D EBYE S CHERRER diffraction schematically: (a) setup and generation of the diffrac
tion cones; (b) typical diffraction patterns for a polycrystalline target: top view onto the screen (a)
as seen from the right
h 1.23 nm
dB = = . (1.120)
2me T T / eV
At a kinetic energy 100 eV the wavelength of an electron is thus on the order of the
size of an atom 0.1 nm. Note, however, that (1.118) also holds in the relativistic
case, with p given by (1.25). Specifically, for electrons that have been accelerated
by a voltage U one has to use
eU
p= 2me eU 1 + . (1.121)
2me c2
With the rest energy me c2 = 0.511 MeV for electrons this leads already at a moder
ate kinetic energy of 50 keV to a reduction of the wavelength of 2.5 % in comparison
to (1.120). In a precision experiment this clearly has to be accounted for.
elastically
diffracted electrons
Fig. 1.28 Low energy electron diffraction (LEED): (a) experimental setup; (b) typical diffraction
pattern from a single crystalline ordered surface
The diffraction structures are generated by many single reflexes from the
microcrystals and appear according to (1.91) under the B RAGG angles 2 =
arcsin(z/2d) with z = 0, 1, 2 . . . The various distances d of the crystals lattice
planes contain information about the crystal structure. Since the crystals are aligned
at random into all directions of space, all electrons diffracted into an angle 2 will
be emitted into a cone around the incoming electron beam. For each m and each
d such a cone exists. For each cone a circle is formed at its intersection with the
plane of observation (photographic plate, or today more likely a CCD camera). In
Fig. 1.27(b) such diffraction patterns for electrons of low and higher energy are
sketched schematically.
10
6 4 2 0 2 4 6
scattering angle / mrad
even be diffracted by quasi macroscopic optical elements, and they interfere. For
this nice experiment a transmission grating was manufactured by stateoftheart
nanotechnology. The diffraction pattern may be completely understood in terms
of K IRCHHOFFs diffraction theory which has been developed more than 150 years
ago.
One may of course ask how far one can push such kind of wave optics with
particles and particle beams. This is an interesting theme which may even be po
tentially relevant for technical applications. Atom lithography and atom optics with
very cold atoms is subject to current research. We have already mentioned briefly in
Sect. 1.3.4 research on B OSE E INSTEIN condensation (BEC). One particular fasci
nating aspect is that the wavelength of the matter waves increasing with decreasing
temperature according to (1.118). It eventually reaches values comparable to the av
erage distance between atoms (particle density N 3 ). With (1.118) and (1.50)
this happens when
h2
T N 2/3 . (1.122)
3mkB
Comparison with (1.71) shows that this estimate agrees with the critical temperature
for BEC apart from a numerical prefactor and the missing spin degeneracy. This
is a nice, plausible result.
One may also push the wave concepts into the opposite direction and ask about
the macroscopic limits of the waveparticle
duality: at constant kinetic energy the
DE B ROGLIE wavelength decreases m, i.e. with the square root of the mass.
Diffraction experiments with larger molecules are thus increasingly more difficult
and much more challenging.
Z EILINGER and his associates (see e.g. A RNDT et al. 1999) were able to show
that even such big objects as Fullerene molecules, C60 , are subject to wave mechan
ics. In a single slit diffraction experiment one observes quite normal diffraction
patterns, as documented in Fig. 1.30 and as one actually would expect. The ul
timate conundrum in this context is about really large objects as sketched in the
cartoon Fig. 1.31 the question is, what will happen? We leave this as a little mental
1.7 Particles and Waves 61
exercise for the reader who may want to estimate orders of magnitude for a pertinent
experiment.
px x (1.123)
22 Note that measurements of microscopic observables are based on probability distributions. Thus,
the exact value on the right hand side of the uncertainty relation (, h/2, h etc.) depends on the na
ture of that distribution and on the exact definition of uncertainty, e.g. FWHM, foot width, position
of the first minimum, etc.
62 1 Basics
Fig. 1.32 Classical Gedankenexperiment according to H EISENBERG why an electron may not
be localized by an optical microscope better than allowed by the uncertainty relation
x
an electron beam by an
aperture leads to diffraction
px
and hence to uncertainties in e
the determination of the p
electrons lateral momentum
x
be small and the opening angle of the microscope has to be kept large, in order
to obtain high resolution. According to A BB the smallest structure
x one may
resolve in an optical microscope is determined by its numerical aperture n sin and
the wavelength
h
x = , (1.124)
n sin
with n = 1 for the optical index of refraction (in vacuum). Small and large sin
imply, however, an intrusion into the experimental situation, since the photon will
thereby lose or change its momentum p by
h
px = p sin = sin ,
h
and transfer it to the electron. Together with the A BB relation (1.124) we obtain
px
x h and thus have derived the uncertainty relation (1.123). Any other
method trying to localize the electron leads eventually to the same result.
Assume e.g. we try to localize an electron beam by a slit this leads to diffrac
tion as indicated in Fig. 1.33. To obtain an estimate for the uncertainty in the x
component of the electron momentum
px = p sin we may use the angle min at
which the first minimum of the diffraction pattern occurs. The latter is well known
from the corresponding formula in wave optics:
sin min = .
x
Using this one reads from Fig. 1.33
h/p
px = p sin = p =p =
px
x = h,
x
x
which again is equivalent to the uncertainty relation (1.123).
1.7 Particles and Waves 63
W t . (1.125)
The big dilemma of physics at the beginning of the 20th century was the inability
to understand stable electron orbits around the atomic nuclei. Electrons on circu
lar orbits are accelerated charged particles which according to classical physics
should continuously radiate electromagnetic waves and hence loose energy. Thus,
they should get slower and slower with time and being negatively charged should
eventually fall into the positively charged nucleus!
Alas! Why do the electrons remain on stable orbits? As we shall see in Sect. 1.8,
B OHR simply postulated that stable electron orbits have angular momenta equal to
integer multiples of . Among theses stationary states according to B OHR there
are also excited ones which decay spontaneously by emitting radiation according to
an exponential decay law exp(At). But still, the ground state is completely stable
even in the B OHR model!
A quantitative solution of this problem is given by quantum mechanics. Inter
estingly, the few basics about the wave nature of the electron which we have dis
cussed so far, already allow us to answer this puzzle in principle. We simply use
the H EISENBERG uncertainty relation to obtain an estimate for the minimum en
ergy without adopting a specific model for the atom: Let a be the mean radius of
the atom. With a high probability electrons will be found within that radius. This
immediately leads to an estimate of the uncertainty
p in a determination of the
electrons momentum by writing (1.123) now
p2 2
T= .
2me 2me a 2
e2
V =
40 a
this gives a total energy W T + V . Assuming the equality to hold one obtains
2 e2
W= . (1.127)
2me a 2 40 a
64 1 Basics
W obviously depends on the atomic diameter a. So let us now find the lowest pos
sible energy! We just apply the standard rule of calculus to find this minimum:
dW 1 42 0 e2 me a !
= = 0,
da 4 me a 3 0
from which we obtain for the radius of the atomic ground state:
42 0 0 h2
a0 = = . (1.128)
e2 me e2 me
Putting this back into (1.127) the total energy of the atomic ground state becomes
1 me e4 me e4
Wmin = 2 2
= 2 . (1.129)
32 2 0 80 h2
This is exactly the value of the B OHR model for the ground state of the hydrogen
atom! Big surprise with such a simple estimate? Well, the idea is obviously right.
One should, however, not overestimate the quantitative assertion: the correct numer
ical result is a direct consequence of the specific form of (1.126) with on the right
hand side, rather than h/2 or something else (see footnote 22). And from this we
have derived the specific estimate for the minimum product of position and momen
tum uncertainties.
Section summary
We have discussed here some elements of quantum mechanics. While the pho
ton was recognized as the particle associated with electromagnetic waves,
we now associate matter waves with the well known particles. The DE
B ROGLIE wavelength of particles dB = h/p also holds for relativistic ve
locities if the correct expression for p is used.
In the nonrelativistic limit, for
slow electrons, we memorize dB 1.2 nm / T / eV. Convincing experimen
tal evidence comes from diffraction experiments with particles.
The uncertainty relation may be said to express the particle wave duality in
terms of minimal uncertainties in a measurement of two canonic conjugate
variable, e.g. for momentum and position
p
x or for energy and time
W
t . The precise value of the lower limit (, h/2 or the like) depends
on the specific problem and on the accuracy which appears tolerable.
The stability of the atomic ground state can be viewed as a direct consequence
of the uncertainty relation. (Why do negative electrons not fall into the positive
nucleus?)
In 1913, during the big time of quantum mechanical discoveries, Niels B OHR
(N OBEL prize in physics 1922) worked as a young postdoc with Ernest RUTHER 
1.8 BOHR Model of the Atom 65
FORD in England. RUTHERFORD had developed an atomic model based on his scat
tering experiments of alpha particles by atoms according to which . . . the atoms
consist of a positively charged nucleus surrounded by a system of electrons kept
together by attractive forces from the nucleus; the total negative charge of the elec
trons is equal to the positive charge of the nucleus. Further, the nucleus is assumed
to be the seat of the essential part of the mass of the atom, and to have linear di
mensions exceedingly small compared with the linear dimensions of the whole atom
. . . . J.J. T HOMSON expanded this by assuming that . . . the atom consists of a
sphere of uniform positive electrification, inside which the electrons move in circu
lar orbits (both quotes from B OHR 1913).
As discussed above, the big challenge in physics was then to explain the highly
problematic fact that these atoms do not collapse: why do electrons circle on sta
ble atomic orbits without loosing energy? Remember, this was 10 years before
DE B ROGLIE postulated his matter waves as discussed in Sect. 1.7.1! B OHR , so
to say, just cut the Gordian knot and developed the fundamental ideas for a the
ory of atomic structure. He knew from BALMERs spectroscopic work, that the
energy levels of atoms follow the phenomenological equation W n2 with in
teger numbers n. These numbers became B OHRs quantum numbers. Since accord
ing to classical mechanics the energy of a particle on a circular orbit scales in
versely to the square of its orbital angular momentum, B OHR suggested that the
angular momenta of stable orbits be directly proportional to these quantum num
bers. And since the P LANCK constant h, then already known from P LANCKs ra
diation law and E INSTEINs explanation of the photoelectric effect, has the unit of
an angular momentum, he postulated that angular momenta be quantized in units
of h/2 = . From a classical view point, this quantization could neither be proven
nor even be justified, but B OHR demonstrated that with these assumptions the ex
perimentally observed spectrum of hydrogen could be explained with unique preci
sion.
Many versions and numbers of B OHRs postulates may be found in the literature
and in a multitude of Web pages on the subject. In his original paper B OHR (1913)
used the following two principle assumptions:
1. The dynamical equilibrium of the systems in the stationary states can be dis
cussed by help of the ordinary mechanics, while the passing of the systems be
tween different stationary states cannot be treated on that basis.
2. The latter process is followed by the emission of a homogeneous radiation, for
which the relation between the frequency and the amount of energy emitted is
the one given by P LANCKs theory.
electron round the nucleus in a stationary state of the system is equal to an entire
multiple of a universal value which he identified as h/2 , today written as . In
concise form, the angular momentum L = r p is quantized and its magnitude on
a stationary circular orbit is
h
L = rme ve = n = n, (1.130)
2
with the velocity ve , the electron mass me and the radius r of the orbit.
The second of B OHRs principle assumptions is about the photon energies emit
ted in a transition from one stationary orbital n to another one n . In mathematical
form it reads:
h = Wn Wn . (1.131)
Thus, we have derived a classical expression for the radius, while the kinetic energy
may be written in terms of the potential V (r)
me ve2 Ze2 1
T= = = V (r).
2 80 r 2
(Note that the latter equality is directly given by the C OULOMB law but also corre
sponds to the classical virial theorem.) Thus, the total energy of the system is
Ze2
W = T + V = T = < 0, (1.133)
80 r
with the minus sign characterizing bound states: one defines zero total energy W = 0
for an electron that is just no longer bound (no kinetic energy at r ).
Finally, the quantization condition (1.130) is introduced by rewriting the kinetic
energy, again using classical mechanics:
L2 (n)2
W = T = = . (1.134)
2me r 2 2me r 2
1.8 BOHR Model of the Atom 67
One may now obtain independent expressions r(n) = an and W (n) = Wn for any
given n = 1, 2, . . . < by solving the set of Eqs. (1.133) and (1.134) accordingly.
One finds B OHRs orbital radii
n2 0 h2 n2
an = = a0 (1.135)
Z e2 me Z
for the stationary states of hydrogen (Z = 1) and hydrogen like ions (Z > 1), with
n called principle quantum number and a0 B OHRs radius:
0 h2
a0 = = 5.2918 1011 m. (1.136)
e2 me
The latter is used as unit of length in atomic physics (we memorize a0 0.05 nm).
The total energy (or term energy) of the stationary states is
Z 2 me e4 Z2 Z2
Wn = = E h = R hc. (1.137)
n2 802 h2 2n2 n2
me e4 e2
Eh = 2R hc = = = 4.359 1018 J =
27.211 eV, (1.138)
402 h2 40 a0
a0 a2 2 2 h3
t0 = = 2me 0 = 0 4 = 2.4189 1017 s. (1.140)
v0 h me e
23 Notto be confused with the rest energy of the electron Wrest = me c2 , also called natural unit of
energy (see e.g. NIST 2010).
68 1 Basics
Fig. 1.34 Orbital radii rn = a0 n2 /Z and energies Wn = Eh Z 2 /(2n2 ) for H (Z = 1, red) and He+
(Z = 2, black) according to the B OHR model
The atomic unit of mass is the rest mass of the free electron:
Note, however, that this must not be confused with the unified atomic mass unit
which is defined as 1/12 of the mass of an unbound atom of the nuclide 12 C at rest
and in its ground state. It is a unit outside the SI but widely used and recommended
by the Comit International des Poids et Mesures.
Ions with only one electron, i.e. He+ , Li++ , . . . , U91+ are called hydrogen like. For
the examples H and He+ Fig. 1.34 illustrates the magnitudes of orbital radii rn =
n2 a0 /Z and energies Wn = (Z 2 /2n2 )Eh .
So far we have always assumed that the nuclear mass is at rest (mn = ). However,
correctly we should describe the electron motion in the centre of mass system.
For a two particle system this kinematic correction is done by simply replacing
the electron mass me with the reduced mass m e:
me mn
me m
e = . (1.143)
me + mn
1.8 BOHR Model of the Atom 69
Z2 Z2
Wn = E h = RH hc, (1.145)
2n2 n2
with Eh and the RYDBERG constant R (1.138) being replaced by
e
m e
m
E h = Eh and RH = R , respectively. (1.146)
me me
From the term energies (1.137) one derives with (1.131) the famous RYDBERG
(RYDBERG R ITZ) formula for the spectra of hydrogen and hydrogen like ions H,
D, He+ , Li++ , Be+++ , . . . , U91+ :
1 1
Wn1 n2 = = h = Wn1 Wn2 = Z R 2 22
(1.147)
n1 n2
1 1 1
= = Z R 2 2
2
in wavenumbers (1.148)
n1 n2
1 1
= Z 2 R c 2 2 as transition frequency. (1.149)
n1 n2
Stateoftheart spectroscopy of the hydrogen atom contributes today decisively to
the most precise measurements of fundamental constants. We shall discuss their so
phisticated methods and their amazing precision in Sect. 6.5.4. In order to compare
these experimental results with various theoretical predictions one has to correct
them of course according to (1.146) with the different kinematic correction factors
for each nuclear mass.
The B OHR model works surprisingly well for H and H like ions. B OHRs term
energies (1.137) in these cases are identical to those derived from nonrelativistic
quantum mechanics. Unfortunately, the model fails for all other atoms in spite of
some serious efforts made in the early days of quantum mechanics. It also fails in
the relativistic case when ve /c = (e2 /20 hc)(Z/n) = Z/n. Here the fine struc
ture constant 1/137 according to (1.10) appears again. Relativistic effects ob
viously become important for large Z and low n. We may neglect them as long as
Z/n 1.
70 1 Basics
So far we have also completely ignored the spin of the electron which is also of
relativistic origin. The electron spin leads to an extension of the concept of angu
lar momentum and one of its practical consequences is fine structure splitting (see
Sect. 6), which the B OHR model of course does not consider.
We have to be aware that the concept of electron orbitals with well defined radii
will have to be reconsidered very critically: strictly speaking, from todays under
standing of quantum mechanics, the B OHR model is simply not correct as we shall
discuss in detail in Sect. 2.6.11. Still, it serves in many cases as a helpful first step
towards formulating and understanding the correct quantum mechanical theory.
Section summary
The B OHR model of hydrogen (and Hlike atoms), historically one of the cor
nerstones in the development of modern physics, makes surprisingly correct
predictions for energy levels, Wn = Eh Z/(2n2 ) and transition frequencies
h = Wn Wn . . . though just based on heuristic postulates. Even the B OHR
radii rn = a0 n2 /Z allow a reasonable interpretation as we shall see in the next
chapter. Thus, the B OHR model remains a reference for quick cross checks
and estimates. Also, its educational value should not be underestimated.
For higher precision, some small correction for finite mass of the nucleus are
needed as summarized in Sect. 1.8.5.
Using the results of the B OHR model, we have introduced the system of
atomic units (a.u.) with energies measured in Eh (twice the binding energy
of the H atoms ground state), lengths in a0 (the radius of the first B OHR or
bit), time in t0 (the time an electrons needs to circle once around that orbit),
and the mass in me (the rest mass of the electron).
One of the key experiments for quantum mechanics has been performed in 1922 by
S TERN (N OBEL prize 1943) and G ERLACH. It showed with then unsurpassed clarity
that classical mechanics and electrodynamics are unable to explain the observations
on the level of atomic dimensions.
Before trying to understand this experiment and its consequences, we have to recall
some basics from mechanics and electrodynamics. According to the B OHR model
the electron circles around the atomic nucleus, a motion that in a classical picture
may be viewed as a spinning top.
As indicated in Fig. 1.35 this orbit is connected with an electric current
e ev
I= = .
t 2r
1.9 S TERN G ERLACH Experiment 71
+I
of
This current, enclosing an area A (A = r 2 ), generates a magnetic moment M
the magnitude
ev evr eL
M = IA = r 2 = = .
2r 2 2me
It is proportional to the magnitude of L and oriented into the opposite direction. The
magnetic moment of an orbital angular momentum is thus
e L
M= L = B . (1.150)
2me
Note: the gyromagnetic ratio derived here, M/L = e/2me = ge B is indepen
dent of the specific geometry of the motion. It does not depend on whether a point
charge or an extended charge distribution orbits around the nucleus. However, this
above classical derivation is valid only for orbital angular momenta. It will have
to be modified in view of the results of the S TERN G ERLACH experiment.
In the B OHR model angular momenta (1.130) are quantized and appear only
in units of . Thus, L/ is an integer number in B OHRs theory, and the unit of
magnetic moment is the so called B OHR magneton
e 1
B = = 927.400915(23) 1026 J T . (1.151)
2me
T = M B. (1.152)
Its potential energy T d is thus given by
VB = M B = MB cos , (1.153)
where is the angle between M and B and zero energy has been assumed for
M B. Minimum potential energy corresponds to M parallel to B, and L an
tiparallel to it.
72 1 Basics
F = grad V (r).
Otto S TERN had invented the so called molecular beam method24 and already ap
plied it to determine the M AXWELL B OLTZMANN velocity distribution in gases
(see Sect. 1.3.4). In his famous paper Ein Weg zur experimentellen Prfung der
24 In this and later context, an atomic or molecular beam source may be thought to just consists of
a reservoir with the species to be investigated (possibly heated, or cooled, possibly mixed with an
inert carrier gas). Through a small orifice in it, the atoms or molecules diffuse or stream into the
vacuum, and their divergence angle is collimated by one or more apertures along the beam axis.
1.9 S TERN G ERLACH Experiment 73
z (a) z (b)
atom
oven magnet B on S
S B off
N x y
magnet B on N
slit photo plate profile
Fig. 1.37 Schematic of the S TERN G ERLACH experiment. (a) The atomic beam from the oven
is collimated through a slit, may then be deflected in an inhomogeneous magnetic field, and will
finally hit a photographic plate. (b) Side view of magnet poles and atom beam
(a) z
y
z
(b) y
Fig. 1.38 Two wire field: it corresponds approximately to the field created by two wires through
which a current flows. The corresponding magnetic field lines are illustrated in (a). In an experi
ment the field is realized by a pair of suitably shaped permanent magnets as indicated in (b). The
red line parallel to the ydirection indicates the position of the atomic beam entering the space be
tween the magnets from the back. It will be deflected by the inhomogeneous field in zdirection
z B
Fz = M . (1.157)
z
z
(a)
x
z
(b) x
(c) x
Fig. 1.39 Classical expectation about the S TERN G ERLACH experiment: (a) View onto the
atomic beam coming out of the yz plane without magnetic field. (b) Alignment of the magnetic
moments prior to entering the magnet, distributed at random. (c) Atomic beam profile after deflec
tion by the magnetic field. The distribution would arise from B/z being strongest for x = 0 and
decreasing for x > 0
What Do We Expect?
Let us assume the profile of the atomic beam without magnetic field arriving in the
detector plane is shaped as indicated in Fig. 1.39(a). Classically, the magnetic mo
ments M are expected to be randomly distributed in all directions as indicated in
Fig. 1.39(b), their projections M z onto the +zdirection ranging from B L/ to
+B L/. The atomic beam enters the inhomogeneous magnetic field and its atoms
are deflected corresponding to their distribution of Mz . Behind the magnet the re
sulting spatial distribution is registered by a suitable detector. With the magnetic
field switched on, the classical expectation is to observe a broadening of the beam
in zdirection proportional to the field and the average magnetic moments reflect
ing the randomly distributed Mz components. The deflection on the zaxis (x = 0)
will be largest since there the field gradient B/z is highest.
25 A quaint note aside: the silver traces only became visible after Otto S TERN blew the smoke of
his (sulfur containing) cigar onto the photographic plate an early, unintended contribution to the
photochemistry and catalysis of the photographic development process.
1.9 S TERN G ERLACH Experiment 75
atom signal
magnetic field bears out the dramatic splitting into two components. This observa
tion, which is also made for many other atoms, is a direct consequence of space
quantization as we shall discuss in a moment.
How can the results of the S TERN G ERLACH experiment be interpreted? We re
member, the B OHR model postulates that angular momenta are quantized according
to (1.130): Thus, we expect L = with to be a positive integer number or zero.
Let us assume now, that a similar relation holds for the components of angular mo
menta, say for Lz in zdirection. Then Lz would assume 2 + 1 integer multiples
of :
Lz = m with m = , + 1, . . . , . (1.158)
76 1 Basics
Fig. 1.42 Schematic of a L ANGMUIR TAYLOR detector: (a) Potential for an electron at an isolated
tungsten surface, (b) potential and energy of the valence electron in a K atom, (c) potential felt by
a K atom when hitting a tungsten surface, (d) simple detector setup
This quantization is called space quantization (in German language somewhat more
precisely: Richtungsquantisierung). The number m is often called magnetic quan
tum number since it was originally observed in a magnetic field. We prefer, how
ever, the more general term projection quantum number.
In contrast to a classical statistical distribution of angular momenta, as illustrated
in Fig. 1.39(b), there are only 2 + 1 allowed projections of L = onto the z
axis. The number 2 + 1 is called multiplicity of the state (in Sect. 1.3.4 we have
introduced it as degeneracy a term used in the context of energies). According to
(1.150) the magnetic moment is M = B L/, thus space quantization of L also
implies quantization of M. The component Mz may thus assume values from M
to M which correspond to the components L to L of the orbital angular
momentum.
A closer quantum mechanical analysis (see Chap. 2) shows that the conjectures
made above are nearly correct except for the fact that the magnitude of the angular
momentum is given by
L = ( + 1), (1.159)
J
magnetic moment for J : MJ = gJ B (1.160)
and its projection onto a given axis, say z, is given by
MJ z = gj B mj with mj = j, j + 1, . . . , j. (1.161)
The so called L AND gfactor is gL = 1 for pure orbital angular momenta and =2
for electron spin states. We shall derive general expressions gj in Chap. 8. A mul
titude experiments confirms that the magnetic moment of atoms and molecules is
indeed proportional to the total angular momentum J and oriented opposite to it.
Equation (1.156) has also to be modified:
e
L ARMOR frequency for J : j = gj B = gj L . (1.162)
2me
Section summary
The S TERN G ERLACH experiment has been another corner stones in the de
velopment of quantum mechanics. It proofs what we call space quantization:
angular momenta J are oriented in space so that their projection on a given
axis is Jz = mj with mj = j, j + 1, . . . , j where j is the angular mo
mentum quantum number.
Orbital angular momenta have integer quantum numbers, intrinsic quantum
numbers of particles may be half integer (fermions) or integer (bosons).
78 1 Basics
We emphasize that this value of the electron g factor cannot be explained by any
kind of classical charge distribution on an orbit, for which always (1.150) holds,
i.e. gL = 1.
In contrast, the relativistically correct D IRAC theory leads exactly to ge = 2 for
the electron spin. Thus, according to (1.161) the component of the magnetic moment
of the electron in respect of a given axis, say z, becomes indeed Mz = ge B /2 =
B as experimentally observed.
It is interesting to note that the L ARMOR frequency (1.162) for an electron with
spin s and ge = 2 becomes
e e
s = ge L = ge B B, (1.166)
2me me
and is, according to (1.111), (nearly) identical to the cyclotron frequency c =
(e/me )B of an electron in a magnetic field (in the nonrelativistic limit).
We point out, however, that high precision measurements show a small, but signif
icant difference between L and c , i.e. document a deviation from (1.165), in fact
one finds ge 2.0023 . . . . A theoretical understanding of this so called anomalous
magnetic moment of the electron is provided by quantum electrodynamics (QED)
for which T OMONAGA, S CHWINGER and F EYNMAN received the N OBEL prize in
1965 as we shall explicate in Sect. 6.6.
The so called E INSTEIN  DE H AAS effect gives one further impressive piece of ev
idence for the nonclassical nature of the magnetic moment of the electron based
80 1 Basics
on a macroscopic measurement. In this experiment one uses the fact (not to be dis
cussed here in detail) that ferromagnetism is caused by many electron spins which
are oriented in parallel.
One determines the torque exerted onto a (soft) ferromagnetic iron core (cylin
der) due to a change of magnetization: the magnetic field of the solenoid tends to
orient the magnetic moments of the electron spins in parallel to this magnetic field
and hence there will also be a preferential orientation of the electron spins. These
spins all together constitute a total inherent angular momentum of the cylinder. In
order to change it, a torque has to act on this cylinder which in the setup shown
in Fig. 1.44 arises from the external magnetic field. The torque is measured by the
twisting angle of a thin spring wire (silica).
In a quantitative experiment one starts with a completely demagnetized probe
and allows a well defined current pulse to flow through the solenoid. This leads to
a changing magnetic field Hsolenoid (t) as sketched in Fig. 1.45. During this pulse a
torque acts on the system which was originally at rest.
After the magnetic field has decayed, the remanent magnetization Mrem is as
sumed to be N times the zcomponent of the magnetic moment Mz of the individ
ual electrons. With (1.161) one obtains
e e
Mz = ge B ms = ge Sz Mrem = Brem = N ge Sz , (1.167)
2me 2me
where N is the number of electrons that after the pulse remain oriented (magnetized)
in the direction of the applied magnetic field (remanence). In turn, Brem may easily
Hsolenoid
Bmagnet
Brem
t
Fig. 1.45 Temporal evolution of the magnetic field generated by the solenoid in Hsolenoid after
starting to discharge the capacitor C along with the magnetic induction Bmagnet resulting from it
in the ferromagnetic cylinder leading to a remanent part Brem
Acronyms and Terminology 81
with Irod = mR 2 /2, the moment of inertia of the ferromagnetic cylinder, m being its
mass and R its radius. The angular frequency of the cylinder rod after application
of the magnetic field may be determined from the maximum twist angle max of the
silica wire onto which the cylinder is mounted: the initial kinetic energy has then
been transformed into potential energy:
Irod 2 kr 2
= max . (1.169)
2 rod 2
From the ratio of Brem according to (1.167) and the angular momentum Irod rod
according to (1.168) we obtain
Brem N ge S z e e
= = ge .
Irod rod N Sz 2me 2me
Section summary
The spin related properties of the electron have again be summarized in
(1.163)(1.165).
The E INSTEIN  DE H AAS experiment demonstrates the electron spin on a
macroscopic level using in a ferromagnetic rod. When its magnetization is
reversed in external magnetic field, the magnetic moments of N the electrons
have to reversed from +/2 to /2. Thus, the total angular momentum of
all electrons in the rod changes by N . Since N is a large number, this in
duces a macroscopic rotation of the rod, which can be measured, confirming
the microscopic measurements of ge .
CCD: Charge coupled device, semiconductor device typically used for digital
imaging (e.g. in electronic cameras).
chemicalpotential: In statistical thermodynamics defined as the amount of energy
or work that is necessary to change the number of particles of a system (by 1)
without disturbing the equilibrium of the system, see in Sect. 1.3.4.
CIE: Commission international de lclairage, International Commission on Illu
mination, provides e.g. colorometric tables (http://files.cie.co.at/204.xls).
CMB(R): Cosmic microwave background, radiation at 2.725 K from the origin of
the universe.
DC: Direct current, unidirectional electric voltage and current.
ESS: European Spallation Source, large scale facility for generating neutrons for
structural research, see http://europeanspallationsource.se/.
EUV: Extreme ultraviolet, part of the UV spectral range. Wavelengths between
10 nm and 121 nm according to ISO 21348 (2007).
FIR: Far infrared, spectral range of electromagnetic radiation. Wavelengths be
tween 3 m and 1 mm according to ISO 21348 (2007).
FT: F OURIER transform, see Appendix I.
FWHM: Full width at half maximum.
HERA: HadronElektronRingAnlage, for collision experiments between elec
trons of 30 GeV and protons of 820 GeV operated at DESY very sucessfully
until 2007.
HF: High frequency, part of the RF spectrum. Wavelengths from 10 m to 100 m
or frequencies from 3 MHz to 30 MHz according to ISO 21348 (2007).
ICR: Ion cyclotron resonance, spectroscopy (specifically mass spectroscopy)
based on irradiation with a radio frequency in resonance with the cyclotron fre
quency of an ion in magnetic field (see Sect. 1.6.3).
IR: Infrared, spectral range of electromagnetic radiation. Wavelengths between
760 nm and 1 mm according to ISO 21348 (2007).
LED: Light emitting diode.
LEED: Low energy electron diffraction, see Sect. 1.7.2.
LF: Low frequency, part of the RF spectrum from 30 kHz up to 300 kHz.
LHC: Large hadron collider (not to be confused with left hand circularly polarized
light), high energy physics facility at CERN (Geneva) providing particles with
collision energies up to 14 TeV for protons and up to 1 PeV for heavy ions.
MF: Medium frequency, part of the RF spectrum from 300 kHz up to 3 MHz.
MIR: Middle infrared, spectral range of electromagnetic radiation. Wavelengths
between 1.4 m and 3 m according to ISO 21348 (2007).
MW: Microwave, range of the electromagnetic spectrum. In spectroscopy MW
usually refers to wavelengths from 1 mm to 1 m corresponding to frequencies
between 0.3 GHz and 300 GHz; ISO 21348 (2007) defines it as the wavelength
range between 1 mm and 15 mm.
NIR: Near infrared, spectral range of electromagnetic radiation. Wavelengths be
tween 760 nm and 1.4 m according to ISO 21348 (2007).
NIST: National institute of standards and technology, located at Gaithersburg
(MD) and Boulder (CO), USA. http://www.nist.gov/index.html.
Acronyms and Terminology 83
References
A RNDT , M., O. NAIRZ, J. VOS A NDREAE, C. K ELLER, G. VAN DER Z OUW and A. Z EILINGER:
1999. Waveparticle duality of C60 molecules. Nature, 401, 680682.
ASTM: 2008. G17303 Reference Spectra Derived from SMARTS v. 2.9.2, American Soci
ety for Testing and Materials (ASTM). http://rredc.nrel.gov/solar/spectra/am1.5/ASTMG173/
ASTMG173.html, accessed: 7 Jan 2014.
B OHR , N.: 1913. On the constitution of atoms and molecules. Philos. Mag., 6, 26, 125.
B OHR , N. H. D.: 1922. The N OBEL prize in physics: for his services in the investigation of the
structure of atoms and of the radiation emanating from them, Stockholm. http://nobelprize.org/
nobel_prizes/physics/laureates/1922/.
B RETSCHNEIDER , E.: 2007. Efficacy limits for solidstate white light sources, Laurin Publishing
Co., Inc., Pittsfield MA, USA. http://www.photonics.com/Article.aspx?AID=28677, accessed:
7 Jan 2014.
DE B ROGLIE , L.: 1929. The N OBEL prize in physics: for his discovery of the wave nature of
electrons, Stockholm. http://nobelprize.org/nobel_prizes/physics/laureates/1929/.
CERN: 2013. The search for the H IGGS boson, Geneva. http://home.web.cern.ch/about/
physics/searchhiggsboson, accessed: 8 Jan 2013.
C HANTLER , C. T., K. O LSEN, R. A. D RAGOSET, J. C HANG, A. R. K ISHORE, S. A. KO 
TOCHIGOVA and D. S. Z UCKER : 2005. Xray form factor, attenuation, and scattering tables
(version 2.1), NIST. http://physics.nist.gov/ffast, accessed: 7 Jan 2014.
C OMPTON , A. H.: 1927. The N OBEL prize in physics: for his discovery of the effect named after
him, Stockholm. http://nobelprize.org/nobel_prizes/physics/laureates/1927/.
C ORNELL , E. A., W. K ETTERLE and C. E. W IEMAN: 2001. The N OBEL prize in physics: for
the achievement of BoseEinstein condensation in dilute gases of alkali atoms, and for early
fundamental studies of the properties of the condensates, Stockholm. http://nobelprize.org/
nobel_prizes/physics/laureates/2001/.
D EHMELT , H. G. and W. PAUL: 1989. The N OBEL prize in physics: for the development of the
ion trap technique, Stockholm. http://nobelprize.org/nobel_prizes/physics/laureates/1989/.
D ICK LYON: 2008. Luminosity function, Wikimedia Commons. http://commons.wikimedia.
org/wiki/File:Luminosity.png, accessed: 7 Jan 2014.
E INSTEIN , A.: 1905. ber einen die Erzeugung und Verwandlung des Lichtes betreffenden
heuristischen Gesichtspunkt. Ann. Phys., 17, 132.
E INSTEIN , A.: 1921. The N OBEL prize in physics: for his services to theoretical physics, and
especially for his discovery of the law of the photoelectric effect, Stockholm. http://nobelprize.
org/nobel_prizes/physics/laureates/1921/.
E NGLERT , F. and P. W. H IGGS: 2013. The N OBEL prize in physics: for the theoretical discovery
of a mechanism that contributes to our understanding of the origin of mass of subatomic par
ticles, and which recently was confirmed through the discovery of the predicted fundamental
particle, by the ATLAS and CMS experiments at CERNs large hadron collider, Stockholm.
http://nobelprize.org/nobel_prizes/physics/laureates/2013/.
H ERZOG , R.: 1935. Berechnung des Streufeldes eines Kondensators, dessen Feld durch eine
Blende begrenzt ist. Arch. Elektrotech., 29, 790802.
H ONSBERG , C. and S. B OWDEN: 2012. PVCDROM Air Mass, UNSW and Solar Power
Labs at ASU, Australia. http://www.pveducation.org/pvcdrom/propertiesofsunlight/airmass,
accessed: 7 Jan 2014.
ISO 21348: 2007. Space environment (natural and artificial) process for determining solar
irradiances. International Organization for Standardization, Geneva, Switzerland.
KOPP , G. and J. L. L EAN: 2011. A new, lower value of total solar irradiance: evidence and
climate significance. Geophys. Res. Lett., 38, L01706.
M AMYRIN , B. A.: 1994. Laserassisted reflectron timeofflight massspectrometry. Int. J. Mass
Spectrom. Ion Process., 131, 119.
References 85
M ATHER , J. C. and G. F. S MOOT: 2006. The N OBEL prize in physics: for their discovery of the
blackbody form and anisotropy of the cosmic microwave background radiation, Stockholm.
http://nobelprize.org/nobel_prizes/physics/laureates/2006/.
M ILLIKAN , R. A.: 1923. The N OBEL prize in physics: for his work on the elementary charge
of electricity and on the photoelectric effect, Stockholm. http://nobelprize.org/nobel_prizes/
physics/laureates/1923/.
M UELLER , U., F. H. N IESEN, Y. ROSKE, F. G OETZ, J. B EHLKE, K. B UESSOW and U. H EINE 
MANN : 2007. Crystal structure of human prolidase: the molecular basis of PD disease,
Hinxton, UK: PDB entry 2okn. The European Molecular Biology Laboratory (EMBLEBI).
http://www.ebi.ac.uk/pdbesrv/view/entry/2okn/summary.html, accessed: 7 Jan 2014.
NIST: 2010. Reference on constants, units, and uncertainties, NIST. http://physics.nist.gov/cuu/
Constants/, accessed: 7 Jan 2014.
NIST: 2011. Conversion factors for energy equivalents, NIST. http://physics.nist.gov/cuu/
Constants/energy.html, accessed: 7 Jan 2014.
O HNO , Y.: 2010. Radiometry and photometry for vision optics. In: M. BASS, ed., Handbook of
Optics, vol. II, 37.1. New York: McGrawHill.
PAULI , W.: 1945. The N OBEL prize in physics: for the discovery of the exclusion princi
ple, also called the Pauli principle, Stockholm. http://nobelprize.org/nobel_prizes/physics/
laureates/1945/.
P LANCK , M.: 1900. Zur Theorie des Gesetzes der Energieverteilung im Normalenspektrum.
Verh. Dtsch. Phys. Ges., 2, 235245.
P LANCK , M. K. E. L.: 1918. The N OBEL prize in physics: in recognition of the services
he rendered to the advancement of physics by his discovery of energy quanta, Stockholm.
http://www.nobelprize.org/nobel_prizes/physics/laureates/1918/.
S CHLLKOPF , W. and J. P. T OENNIES: 1996. The nondestructive detection of the helium dimer
and trimer. J. Chem. Phys., 104, 11551158.
SORCE: 2012. SORCE Solar Spectral Irradiance, Boulder, Co.: Laboratory for Atmo
spheric and Space Physics, University of Colorado and NASA. http://lasp.colorado.edu/
lisird/sorce/sorce_ssi/, accessed: 7 Jan 2014.
SSE: 2012. Solar System Exploration Our Solar System, NASA. http://solarsystem.nasa.gov/
planets/profile.cfm?Display=Facts&Object=Sun, accessed: 7 Jan 2014.
S TERN , O.: 1921. Ein Weg zur experimentellen Prfung der Richtungsquantelung im Magnet
feld. Z. Phys., VII, 249253. Nachdruck: Z. Phys. D, Atoms, Molecules and Clusters 10, 111
116 (1988).
S TERN , O.: 1943. The N OBEL prize in physics: for his contribution to the development of the
molecular ray method and his discovery of the magnetic moment of the proton, Stockholm.
http://nobelprize.org/nobel_prizes/physics/laureates/1943/.
T HOMSON , J. J.: 1906. The N OBEL prize in physics: in recognition of the great merits of his the
oretical and experimental investigations on the conduction of electricity by gases, Stockholm.
http://www.nobelprize.org/nobel_prizes/physics/laureates/1906/.
T OMONAGA , S.I., J. S CHWINGER and R. P. F EYNMAN: 1965. The N OBEL prize in physics:
for fundamental work in quantum electrodynamics, with deepploughing consequences for
the physics of elementary particles, Stockholm. http://nobelprize.org/nobel_prizes/physics/
laureates/1965/.
W IKIPEDIA CONTRIBUTORS: 2013. Mathieu function, Wikipedia, The Free Encyclopedia.
http://en.wikipedia.org/wiki/Mathieu_function, accessed: 7 Jan 2014.
W ILEY , W. C. and I. H. M C L AREN: 1955. Timeofflight mass spectrometer with improved
resolution. Rev. Sci. Instrum., 26, 11501157.
Elements of Quantum Mechanics
and the H Atom 2
Overview
If the reader is already familiar with quantum mechanics he may just want
to browse this chapter and return later if necessary. However, readers who
have experienced quantum mechanics up to now only as compulsory math
ematical exercise may perhaps read this chapter with advantage and find it
helpful to approach the indispensable instruments without big formal hurdles.
In Sects. 2.12.3 we summarize a minimum of formalism. Section 2.4 treats
as a first example the well known particle in a box and the free electron gas
which in atomic and solid state physics is an important elementary model.
Section 2.5 gives an overview of how to treat angular momenta, needed in
all following text, specified in Sect. 2.5.4 for the electron spin s = 1/2. Sec
tion 2.6 offers a crash course in nonrelativistic quantum mechanics of the
H atom essential knowledge for all the following chapters. We refrain here
from formal derivations in favour of a plausible, possibly somewhat hand wav
ing introduction. Finally, Sect. 2.7 presents a first, elementary approach to in
teractions of atomic electrons with external fields, which will be extended and
deepened in Chap. 8. Clearly, such a brief introduction into quantum mechan
ics cannot substitute a profound study of the formal theory. It should, however,
facilitate the access to it and make the reader fit for the following chapters.
The classical picture of a well defined trajectory with known coordinate x(t) and
momentum p(t) looses its validity in quantum mechanics, as sketched in the phase
diagram Fig. 2.1: position and momentum cannot be measured simultaneously. They
may only be determined with an accuracy in accordance with the uncertainty rela
tion
pi
xi h/2 . Quantum mechanics only makes statements about probability
amplitudes (r, t). These may assume the form of so called Wavepackets which
localize particles in space and time as well as the uncertainty relation allows it. One
finds a particle at position r and time t with the probability
2
w(r, t) = (r, t) . (2.1)
One may easily visualize the concept of a probability amplitude for the example
of photons in a YOUNGs double slit experiment. The probability to find a photon
at position r and time t is proportional to the intensity I (r, t) of the light, and the
latter is proportional to the square of the field amplitude. Let us consider only one
component of polarization, say Ex . The dependence of the intensity on position may
then be written as
2 2
I (r) Ex (r) = (r) = w(r).
The last two equalities can help us to get used to quantum mechanical terminology:
we call the quantity (r) now the (position dependent) probability amplitude or
wave function. In the case of light, it is simply represented by the electric field
component Ex . One determines it according to the laws of optics as solution of the
corresponding wave equation
We may see this expression as a reinterpretation of the classical YOUNG double slit
experiment in terms of quantum mechanical probability amplitudes.
In an actual experiment one may now reduce the light intensity such that only
one single photon at a time is near the double slits and participates to the observed
interference pattern. One may easily verify such a setup with a particle counter
detecting single photons one may even hear the individual photons click.1 If a
sufficient number of such single photon events are recorded and added up, one finds
quite counter intuitively that the classical diffraction pattern known from optics
is recovered! Thus, the probability distribution of each individual photon behind the
double slit is determined by the wave amplitudes = 1 + 2 and it is impossible
to say through which slit the particle has passed (see Fig. 1.31). One also says that
one photon always interferes only with itself. We shall discuss this statement in some
detail in Chap. 2, Vol. 2 and quantify it statistically.
From the perspective just discussed let us have a look at the matter waves intro
duced in Sect. 1.7.1. We have already familiarized ourselves with the relation be
tween momentum and wavelength as postulated by DE B ROGLIE. Matter waves are
1 For the experts: One may invest a lot of thought and substantial experimental efforts to make
absolutely sure that only isolated photons reach the double slit setup (for a recent review on single
photon sources see E ISAMAN et al. 2011). In the present context we shall be content if the average
time tav between each interference event (photon counted) is long compared to the coherence time
of the photon source c = 1/
tav , with
being the bandwidth of the source (see Sect. 2.1.4,
Vol. 2).
90 2 Elements of Quantum Mechanics and the H Atom
And as in optics, diffraction and interference occurs, e.g. at a double slit for which
(2.4) describes the experimentally observable signal of particles.
However, in contrast to electromagnetic radiation, where was identified with
the electric (or magnetic) field, for matter waves (r) cannot be attributed to any
directly measurable physical quantity: the observable physics is described by the
probability w(r). So we call (r) simply the probability amplitude for finding a
particle at position r.
Otherwise, analogue considerations are valid for photons and particles of matter.
If e.g. we try to localize them on their pathways, we loose the interference patterns!
One important general rule holds: interference phenomena are observed if differ
ent but indistinguishable pathways exist on which the particles may proceed toward
the detector. In contrast, no interference is observed if the two pathways may in
principle be distinguished even when no distinction is made in the actual experi
ment.
Section summary
In classical mechanics well defined trajectories in position r(t) and mo
mentum p(t) space describe the motion of particles. In contrast, quantum
mechanics describes probability amplitudes (r) and probabilities dw(r) =
(r)2 d3 r for finding a particle at a certain position r in a volume element
d3 r.
Diffraction and interference of matter waves can be described by concepts in
analogy to those used in wave optics. The big difference is that for photons
the probability amplitude is also a directly measurable quantity (electric or
magnetic field) while for matter waves (r) cannot be measured directly
only probabilities (r)2 refer to the real world.
p2 2m(W V (r))
(r) + (r) =
(r) + (r) = 0.
2 2
Rewritten, this is already the stationary S CHRDINGER equation
2
(r) + V (r)(r) = W (r), (2.6)
2m
or somewhat more compact
(r) = W (r)
H (2.7)
with the eigenfunction (r) and the eigenenergy W . The H AMILTON operator H
represents the total energy (briefly just Hamiltonian):
=
+ V (r).
2
H (2.8)
2m
In the often encountered onedimensional case the S CHRDINGER equation (2.6) is
further simplified:
2 d2 (x)
+ V (x)(x) = W (x). (2.9)
2m dx 2
= 2 + V (r) =
p2
2
H + V (r). (2.10)
2m 2m
Here we have introduced the momentum operator (a vector operator)
92 2 Elements of Quantum Mechanics and the H Atom
/x
p = i = i /y
(2.11)
/z
so that
2 2 2
p2 =
p
p = 2 2 = 2
= 2 + + . (2.12)
x 2 y 2 z2
Thus, (2.10) is the operator form of the classical energy conservation law
p2
W = Wkin + V = + V (r).
2m
(r, t) = i (r, t)
H (2.13)
t
2 (r, t)
or more explicitly:
(r, t) + V (r) (r, t) = i .
2m t
It cannot be derived, and we just communicate it here as it was found by Erwin
S CHRDINGER in the beginning of 1926 by the way: during a winter ski holiday
in the Swiss Alps. We point out some key aspects:
The S CHRDINGER equation is a linear PDE of 2nd order in space and 1st order
in time! As a consequence, time dependence of stationary states is truly complex.
The linear superposition principle may be applied to the solutions.
The statistical interpretation of quantum mechanics understands solutions
(r, t) of this PDE as probability amplitudes for finding a particle at a posi
tion r in space at time t according to (2.1).
Although the S CHRDINGER equation cannot be derived, it is able to describe
a wealth of atomic phenomena and to predict observables quantitatively (under
nonrelativistic conditions) with excellent accuracy: we emphasize that is this very
fact which defines the validity of a physical theory.
Equivalently, formal (algebraic) quantum mechanics can be deduced from a con
sistent set of axioms, which are, however, also heuristically assumed.
2.2 SCHRDINGER Equation 93
Consistent alternatives of wave equations for matter waves are the D IRAC equa
tion for fermions, a multi component spinor equation and the K LEIN G ORDON
equation, a single component PDE of 2nd order in time, which turns out to be
valid for bosons both are relativistically correct.
(r, t) = H
If the Hamiltonian itself does not depend on time, H (r), the time depen
dence of the wave function may be factored by a product ansatz:
(r, t) = i (r, t)
H (r)(t) = i (r)(t)
H
t t
H (r) i (t)
= W.
(r) (t) t
The last identity (with the constant W which has to be determined) must hold so
that the former equality can be valid for all values of r and t. One thus has to
solve id(t)/dt = W (t) and H (r) = W (r). In this case, the time dependent
equation leads to the trivial solution
W
(t) exp i t . (2.15)
The position dependent part is nothing but the stationary S CHRDINGER equation
(2.6) and the parameter W introduced is the total energy of the system. The overall
wave function is given by
W
(r, t) = (r) exp i t . (2.16)
Note: The time dependence is truly complex and the imaginary unit i in the prefactor
= H
is needed for the solution! In the present case, however, with H (t) the time
dependence is trivial in the sense that only
2 2
w(r, t) = (r, t) = (r) (2.17)
can be measured, which contains only information about the stationary state.
In order to find these stationary atomic states for a given potential V (r)
which are the equivalent to B OHRs stationary orbits one has to solve the station
ary S CHRDINGER equation (2.6) under suitably chosen boundary conditions. For
bound states this typically leads to a whole series of discrete total energies W < 0
with a set of quantum numbers characterizing the states. The corresponding series
of wave functions for individual electrons called orbitals describe the probability
of finding the particles at a given position in space.
94 2 Elements of Quantum Mechanics and the H Atom
k
(r, t) = C exp i(t kr) = C exp i t kr (2.19)
2m
W pr
= C exp i t .
With this we obtain the important dispersion relation for free matter waves:
W p2 k 2 2 2
= = = = (k) or W = k . (2.20)
2m 2m 2m
Note: The probability to find this particle, w(r, t) =  (r, t)2 = C2 , is indepen
dent of position and time as expected for an infinitely extended plane wave. In
other words, a particle with well defined momentum may not be localized at all as
expected according to the uncertainty relation (1.123).
Section summary
The stationary S CHRDINGER equation (2.6) may be gleaned from the classi
cal wave equation combined with a free interpretation of the DE B ROGLIE
wavelength in a conservative potential. It is validated by the excellent
agreement of its predictions with experimentally observed data in the sub
microscopic world at nonrelativistic energies.
In its most compact form it reads H = W , where the H AMILTON operator
(2.10), also called Hamiltonian, is constructed in full analogy to its classical
counterpart, just replacing the momentum by the quantum mechanical mo
mentum operator p = i.
The time dependence of the wave function is described by the time dependent
S CHRDINGER equation (2.13). In the case that the Hamiltonian itself is not
time dependent, the (stationary) solutions are given by the product of the solu
tions (r) of the stationary equation (2.6) and a simple exponential function
exp(i(W/)t).
The most simple solution of the time dependent S CHRDINGER equation is a
plane wave exp[i((W/)t kr)], with the energy W = k 2 /(2m).
2.3 Basics and Denitions of Quantum Mechanics 95
Here we summarize briefly the fundamental axioms of quantum mechanics and re
call some terminology and rules which we shall use later on:
Note: the sum includes all bound (discrete) and free (continuum) states.
One calls a basis orthonormal if
fk fi = ki , (2.22)
where 
is the scalar product of two state vectors 
and 
. With this, the
projection of the state 
onto the basis vector fk
is
fk 
= ci fi fk
= ck . (2.23)
i=1
Wave functions, the most commonly used representation of states, are formally
obtained by expanding 
in a continuous position basis {r
} where r extends
over all points in 3D position space. We write (2.21) as ket

= d r r r
3
or as bra  = d 3 r r r . (2.24)
and obtain from (2.24) the definition of a wave function in position space,
r
= r
= d3 r r r r = (r), (2.26)
96 2 Elements of Quantum Mechanics and the H Atom
and (r) = r
. In practice, wave functions are determined by solving the sta
tionary S CHRDINGER equation. With (2.24) and (2.25) we derive (after one 3D
integration) the scalar product in terms of wave functions:2

= (r)(r)d3 r = 
. (2.27)
We finally note that, equivalent to the state expansion in position space, one may
define wave functions (p) in momentum space by

= d3 p p p and
p
= d3 p p pp = (p).
Operators
Linear operators play a key role in quantum mechanics: in general an operator
changes an object onto which it acts (e.g. a state vector, a wave function, another
and let it act on a ket
operator) into something different. Let us call the operator A
vector 
:
A
= 
.
Linearity implies that for a superposition of states 
= c1 1
+ c2 2
+
c1 1
+ c2 2
+ = c1 A
A 1
+ c2 A
2
+ .
B
The product of two operators A is defined by
B)
(A B
=A . (2.29)
B
Operator multiplication is distributive, i.e. A C
= (A
B)
C = A(
B C)
but not neces
sarily commutative. In general AB = B A and one defines a commutator
B]
[A, =A
B B
A (2.30)
2 We shall use the bra and ket symbols rather loosely. In particular, we shall often write wave
which only in special cases may vanish (see Sect. 2.3.3). With 
and 
being
two state vectors or wave functions one defines as matrix element of A between
states 
and 
(which may or may not be basis states of an operator A:
= (A)d
A = A
3 r. (2.31)
Without going into details we define the so called adjoint (or Hermitian conjugate)
by
operator A
= A
A or A = A
(2.32)
3
or d r = (A)d
A 3 r. (2.33)
Of particular importance are the so called Hermitian operators, let us call them O.
They are selfadjoint:
O
O. (2.34)
By this definition and with (2.31)(2.33) the matrix elements
O
= (O)
= O
d3 r = (O)d 3r (2.35)
= O
= O
(2.36)
Observables
Observables are all physical quantities which can in principle be measured (ob
served). Every quantity which is observable in classical physics is represented quan
tum mechanically by a linear Hermitian operator, let us call it O. A quantum sys
tems can be characterized by a set of eigenstates (eigenvectors) fk
of an observable
O which it may in principle assume. From the eigenvalue equation
k
= k fk
Of (2.37)
one determines the eigenvalues k of the observable O for the eigenvectors fk
. In
any individual physical measurement of the observable O only one of its eigenvalues
k can be observed.
From (2.37) and the orthonormality of the fk
basis one sees that the matrix
in the basis of its eigenstates between fi
and fk
are
elements of O
k
= k ik ,
Oik = fi Of (2.38)
3 To verify these relations one simply expands 
and 
in a basis of eigenvectors (eigenfunc
tions) of O.
98 2 Elements of Quantum Mechanics and the H Atom
The latter relation follows directly with (2.22). If we measure now the observable
many times (as one does in a real experiment), the result of each individual mea
O
The probability to detect this particular
surement is one of the eigenvalues i of O.
eigenvalue i is determined by the probability amplitude ci :
O
=O ci fi
= ci i fi
. (2.40)
The average value measured for this observable, i.e. the result of many individ
ual measurements applied to the same state 
, is called expectation value of the
observable:
O
ci 2 i = O
. (2.41)
()
Somewhat more general, if the state of the system is 
= i ci gi
, i.e. it is
its expectation
given in an arbitrary basis {gi
} and gi
are not eigenstates of O,
value is
()
()
= O
O
= ci
gi O
()
ck gk
= ci k
c() . (2.42)
gi Og k
i k ik
Unit Operator
We note in passing a nice mathematical trick by rewriting (2.39) as

= fi
ci = fi
fi 
= fi
fi  
.
i i i
2.3 Basics and Denitions of Quantum Mechanics 99
From this we deduce the often very useful fact that the quantum mechanical unit
operator4 (which does not change the operand) may be written as
1= fi
fi , (2.43)
i
as long as the states fi
represent a complete orthogonal basis and the summation
goes over all states of this basis.
Quantization
Experimentally, when an observable O is determined in an individual measurement
one always finds one of its eigenvalues. With the measurement one prepares also
One
the corresponding eigenfunction (eigenstate, eigenvector) of the observable O.
may say that by a measurement one projects the eigenvector out of the state 
under investigation.
n
= Wn n
.
H
Sz sms
= ms sms
.
2.3.2 Representations
We have already made use of different ways to describe quantum systems and their
changes. The standard terminology speaks about representations of states and oper
ators.
SCHRDINGER Representation
In the S CHRDINGER representation (or picture) the operators are differential oper
ators. The states are represented by wave functions. The scalar product is an integral
according to (2.27) and the orthogonality of basis states is defined by (2.28). The
matrix elements of an operator A in respect of an arbitrary basis set fk
, fi
, . . .
are
(in general not eigenfunctions of A)
k
= fi (r)Af
Aik fi Af k (r)d3 r = fk A
fi = A , (2.44)
ki
where we have used the definition of adjoint operators (2.32). If the operator is
Hermitian, i.e. represents and observable, this relation reads
Oik fi Ofk
= fi (r) Of k (r) d3 r (2.45)
= k (r) fi (r)d3 r
Of i
= Oki
= fk Of
. (2.46)
HEISENBERG Representation
In the H EISENBERG representation (or picture) the operators (we mention in this
context the N OBEL prize for H EISENBERG 1932) A are matrices, which are de
termined by their matrix elements Aik . The states, say 
or 
, are vectors in
H ILBERT space5 which we write

= = b1 f 1 + b2 f 2 + b3 f 3 +

= = c1 f 1 + c2 f 2 + c3 f 3 +
i
= i i
and B
A i
= i i
.
Thus, also the following relations must be valid for simultaneous measurability of
and B:
the operators A
B
A i
= A
i i
= i A A
i
= i i i
= B i
.
B
Hence, the operators must commute, A =! B
A.
Equivalently:
and B
Simultaneous measurement of two observables A is possible if and only
if their commutator vanishes:
B
A B
A = [A,
B]
= 0. (2.48)
From the above follow some simple recipes of how to translate classical quantities
into quantum mechanical operators. One simply has to substitute:
r r and pi i =p
i
xi
(2.49)
or p i , i , i = i =
p.
x y z
All other substitutions are derived from these rules. In particular, the classical
Hamiltonian total energy
p2
Hclass = + V (r) = Wkin + V with p 2 = p p
2m
becomes the H AMILTON operator:
We shall now find the eigenfunctions and eigenvalues of the momentum in the
S CHRDINGER picture, starting with the one dimensional case:
d(x)
x (x) = px (x)
p i = px (x).
dx
One easily verifies that (x) = exp(ipx x/) = exp(ikx x) are solutions of this eigen
value problem. Each value of px (with < px < ) is an eigenvalue of the mo
mentum operator p x in xdirection: the eigenfunction is thus a plane wave with a
continuum of eigenvalues.
This is easily extended into 3D space. The eigenvalue equation
p (r) = i(r) = k(r) is solved by
a plane wave (r) = C exp(ik r).
with p = k and the angle between e and k just as one would guess.
Section summary
In the S CHRDINGER picture, states (bra 
and ket ) of quantum sys
tems are represented by wave functions (r) and (r), respectively. The
H EISENBERG picture uses state vectors in H ILBERT space.
Quantum states may be expressed as a linear superposition of states (2.21)
from a complete, orthonormal basis with fi fk
= ik . The unit operator
may be written i fi
fi .
The S CHRDINGER representation uses differential operators. Classical the
ory is translated by replacing r r and p p = i.
Matrix elements of an operator A k
= f (r)Af
are Aik fi Af k (r)d3 r
i
in the S CHRDINGER picture. In the Heisenberg representation operators are
defined by the corresponding matrices.
The adjoint A of an operator A is defined by A

= A
.
Hermitian
operators are selfadjoint A = A, and Aik = Aki holds.
2.4 Particles in a Box And the Free Electron Gas 103
Eigenstates of the momentum operator p are plane waves exp(ik r), with
p = k.
2 d2 n (x)
= Wn n (x).
2m dx 2
Solutions can in principle be (x) = sin(kx) or cos(kx). However, since the wave
function cannot penetrate into the wall, it must vanish on the walls which at x = 0 is
only possible for the sin(kx) solution. For continuity on the other wall sin(kL) = 0
and thus k = n/L must hold, with n = 1, 2, 3, . . . . In summary, the solutions
(eigenfunctions) are standing waves with nodes on both walls of the box. The
eigenenergies assume discrete values Wn :
2 nx 2 k 2 h2 n2
n (x) = sin with Wn = = . (2.52)
L L 2m 8mL2
The expectation value of the momentum p x is derived from
x n
= n (x)
px
= n p px n (x)dx
2 nxL d sin nx
= i
sin L
dx
L0 L dx
i2n L nx nx
= sin cos dx 0.
L2 0 L L
This corresponds to the fact that the particle in the box moves with equal probability
back and forth. In contrast, the square of the momentum,
2 2 L nx 2 d2 sin nx
x = n p
p x2 n = n p x2 n dx = sin L
dx
L 0 L dx 2
104 2 Elements of Quantum Mechanics and the H Atom
2(n)2 L nx 1 h2 2
= sin2 dx = n ,
L3 0 L 4 L2
=
is not zero. With this and H px2
/2m we verify Wn in (2.52).
A next step towards reality is the extension into the 3D space. The movement of a
particle is now restricted to a large but finite 3dimensional box, with rigid walls
beyond which the probability to find a particle is zero. For simplicity we assume
the box to be a cube with edge length L as illustrated in Fig. 2.2(a). Inside the box
particles move freely. Stationary solutions are plane waves (2.18) which we write in
the box as real functions now as product in three dimensions:
To be continuous, the wave function must vanish on the walls of the cube:
sin(kj L) = 0 kj = n j for j = x, y, z. (2.54)
L
With these boundary conditions, and in analogy to the 1D case (2.52),
2 2 2 k 2 2 2 2
W= kx + ky2 + kz2 = = n , (2.55)
2m 2m 2mL2
now with three integer quantum numbers nx , ny , nz and n2 = n2x + n2y + n2z .
One may view this in k or n space as indicated in Fig. 2.2(b). Equation (2.54) says
that precisely one solution exists for each lattice point with integer nx , ny and nz .
Thus, one reads from the figure that the total number of states with quantum num
bers 1 to nx , 1 to ny and 1 to nz , i.e. the number of states with n n2x + n2y + n2z is
given by NZ (n) = 1/8 4/3 n3 . Expressing n by the energy W , the total number
of states with energies W is
1 (2mW )3/2 3
NZ (W ) = L .
6 2 3
If the particle has a spin s, we also have to account for the energy degenerate ge =
2s + 1 possible orientations in space. Dividing by the volume of the box L3 , we
obtain the total number of available states per unit volume:
ge (2mW )3/2
NZ (W ) = . (2.56)
6 2 3
From this we derive the number of states in an energy interval from W to W + dW ,
called density of states (DOS), here per unit volume:
dNZ (W ) ge (2m)3/2 4 2m3/2
g(W ) = = W = ge W. (2.57)
dW 4 2 3 h3
We note in passing, that assuming the phase space to be quantized in unit cells of
size h3 leads to exactly the same result.6 For later use we also give the density of
states in respect of a specific element of solid angle d and express the energy by k,
the magnitude of the wave vector:
d2 NZ (W ) d ge mk
dg = = d. (2.58)
dW 4 (2)3 2
w(W ) / eV 1 (a) W
w (W ) / eV 1 (b)
W
0.2 0.2
T = 0K
T = 50 K
0.1 0.1 T = 500K
Ne Ne
T = 293 K
0 0
0 2 4 6 8 6.8 6.9 7.0 7.1 7.2
F W / eV F
Fig. 2.3 F ERMI D IRAC probability distribution according to (2.61) as a function of energy for
electrons in a metal with a F ERMI energy F = 7 eV at different temperatures T : (a) energy range
from 0 eV to 9 eV, (b) expanded scale around F ; at T = 0 the electron density Ne extends up to
F (grey shading up to the dashed vertical red line); the full red line indicates the density of states
W
total number of electrons per volume with kinetic energies between 0 and F is
thus
1 2me F 3/2
Ne = 2NZ (F ) = . (2.59)
3 2 2
We may invert this to obtain the so called F ERMI energy F . At absolute zero tem
perature F is identical to the chemicalpotential introduced in Sect. 1.3.4, and
corresponds to the maximum electron energy. One state after the other is filled, each
by two of electrons, up to the F ERMI energy:
2 2 2/3
F = 3 Ne . (2.60)
2me
Typical F ERMI energies for metals range from 1.6 eV (Cs) to 14.3 eV (Be).
Expressing the prefactor A given by (1.64) with ge = 2 in terms of the F ERMI
energy F one may write the probability distribution for finding a given energy W
in the electron gas in a box:
dNe 3 1 3/2 W
w(W )dw = = dW (2.61)
Ne 2 F exp[(W )/(kB T )] + 1
with F F as long as kB T F .
Section summary
The particle in a box model allows a most simple description of electrons
moving freely in a metal. The wave functions (2.53) in the 3D box (vol
ume L3 ) have nodes at theboundaries of the box. The energies are W =
2 2 n2 /(2mL2 ) with n = n2x + n2y + n2z and ni representing positive val
ues in 3D integer number space.
From this the DOS (2.57) is derived, which according to (2.57) is
ge m3/2 W , where m is the particles mass and ge = 2s + 1 the degen
eracy due to its spin.
In the case of electrons (Fermions) each state can be filled with up to 2 elec
trons. Then, at T = 0 K the highest energy with occupied states is the F ERMI
energy F = 2 (3 2 Ne )2/3 /(2me ).
At temperatures T > 0 the boundary between occupied and unoccupied states
smears out according to (2.61). The width of the boundary layer is on the
order of kB T .
Angular momenta play a central role in atomic and molecular physics and Ap
pendix B gives a summary on the essentials: their abstract definitions, properties,
combinations and the relevant algebra. Here we introduce the S CHRDINGER pic
ture of orbital angular momenta as used in the quantum mechanics of the H atom.
The orbital picture emerging from this concept is directly accessible to physical
imagination and visualization. At the end of this section we shall generalize the
basic concepts, including electron spin.
Quantum mechanical problems may be treated with advantage in polar rather than in
Cartesian coordinates if they have a symmetry centre, e.g. if the potential depends
only on the distance r from origin V (r) = V (r) as in the C OULOMB case. The
108 2 Elements of Quantum Mechanics and the H Atom
y
x
transformations between Cartesian (x, y, z) and polar coordinates (r, , ) are read
from Fig. 2.4,
x = r sin cos
y = r sin sin (2.62)
z = r cos ,
2 1 1 2
L = 2
sin + 2 . (2.65)
sin sin 2
In full analogy to classical mechanics, the form of (2.64) suggests two components
of kinetic energy:
2 1 2
radial energy Hr = r and (2.66)
2m r 2 r r
2
rotational energy rot = L .
H (2.67)
2mr 2
L = r
p. (2.68)
This has to be expressed in polar coordinates. We just show this for one coordinate
by way of example:
z = i x y .
L (2.69)
y x
With (2.62) one transforms the expression in brackets into polar coordinates:
x y z
= + + = r sin sin + r sin cos +0
x y z x y
= y +x =x y .
x y y x
z = i .
L (2.70)
We are now ready to discuss eigenvalues and eigenstates of angular momentum op
erators. They will be used in more or less all of the following chapters. In the spirit
of this textbook, again we only sketch the basic concepts, present a collection of nec
essary tools, and refer the reader to the standard textbooks on quantum mechanics
and angular momentum algebra for details.
1
with eigenfunctions m = exp(im), (2.74)
2
and eigenvalues m where m = 0, 1, 2, . . . .
x , L
[L y ] = L
x L y L
y L z ,
x = iL
(2.75)
[L z ] = iL
y , L x z , L
and [L x ] = iL
y .
2
L Y (, ) = L2 Y (, ) (2.76)
with the eigenvalue L2 and use a the product ansatz towards its solution:
Y (, ) = ( )(). (2.77)
z and substitute ( )m ()
As azimuthal part we try the eigenfunctions (2.74) of L
into (2.65). This leads to
1 ( ) 1 2
2
sin m () + 2 ( ) 2 m ()
sin sin
= L2 ( )m () = L2 Y (, )
1 m2
2 sin 2 = L2 . (2.78)
sin sin
Thus, only one ODE remains to be solved. Several procedures lead to the cor
rect solutions. One may directly use the associated L EGENDRE polynomials, known
from the mathematics of ODEs, or (perhaps more elegantly) by exploiting the prop
erties of the angular momentum operators to find appropriate recursion formulas.
In any case, one has to demand physically reasonable boundary conditions. In anal
ogy to (2.72) for the zcomponent, the wave functions must be finite and unique for
0 . Without proof we communicate here that such physically meaningful
solutions exist for which the following relations hold (for completeness we include
again the zcomponent):
x 2
z only acts
The validity of (2.81) follows directly from (2.74) and (2.77), since L
onto the component of Y m (, ). This implies
2
2
L L z
z=L L
2
or z = 0.
L ,L (2.84)
Vector Diagram
With (2.79) one may write the magnitude of the angular momentum as:

L = ( + 1). (2.85)
For a given set of quantum numbers m the exact direction of the total angular
momentum is undefined. Precisely defined is only the magnitude and component
Lz = m in respect of the zaxis. One visualizes these relations with the help of
a vector diagram shown in Fig. 2.5. ofthe electron spin (see Fig. 1.43). Figure 2.5
illustrates the example = 2, 
L/ = 6 2.45 with L z / = m = 2, 1, 0, 1, 2.
One may consider the vector arrows statistically
distributed around the zaxis, i.e.
on cones of height m with a side length ( + 1).
Spherical Harmonics
The eigenfunctions of
2
L and Lz are called spherical harmonics Y m (, ). General
formulas and properties are summarized in Appendix B.1.2, and specific expres
sions up to = 3 are tabulated in Table B.1. A graphical illustration of the angular
dependence is shown in Fig. 2.6.
The Y m (, ) are orthonormalized:
2
d Y m (, )Y m (, ) sin d = m m . (2.86)
0 0
2.5 Angular Momentum 113
x y
Fig. 2.6 3Dplot of the s, p and d spherical harmonics; plotted are the squared moduli as a func
tion of angles, the shading colours indicate the sign of Y m (, 0). For an alternative representation
in the real basis see Appendix D.3 and in particular Fig. D.1
Y m (, ) = (1)m Y m (, ), (2.87)
Y m ( , + ) = (1) Y m (, ), (2.88)
In the following text we shall, for compactness, usually write the spherical harmon
ics in bra and ket form, substituting
Y m (, )  m
and Y m (, ) m. (2.89)
In this notion, the orthogonality relations (2.86) and the matrix elements of an oper
ator A are written as
m m = mm m .
and A m, m = mA (2.90)
As shown by the S TERN G ERLACH experiment, the electron has in addition to mass
and charge one further property which we have identified in Sect. 1.10 as an intrinsic
angular momentum, called spin. The spin is characterized by the spin quantum num
ber s = 1/2. Its magnitude is S = s(s + 1), and two orientations with angular
momentum components /2 and /2 are possible. Closely related to the spin, the
electron has also a magnetic moment with a g factor as defined by (1.162) close
to ge 2.
One simply transfers the formal rules which we have introduced in Sects. 2.5.2
2.5.3 onto the properties of the spin. Clearly, the spin may not be imaged in position
space. However, we may define quite formally in analogy to the orbital angular
momentum L a new vector operator S with a square magnitude operator S 2 and a
component Sz in zdirection for which the general commutation rules for angular
momenta (2.75) and (2.84) are valid:
2
[
Sx ,
Sy ] = i
Sz , [
Sy ,
Sz ] = i
Sx , [
Sz ,
Sx ] = i
S ,
Sy ,Sz = 0.
(2.91)
They imply, as in the general case, that the components of the spin cannot be mea
sured simultaneously. However, it is possible to determine its magnitude together
with one of its components is (e.g. the zcomponent
Sz ).
We now introduce spin states sms
. In analogy to (2.79)(2.83) (for the orbital
angular momentum)
for which the spin points into +z and zdirection, respectively. We have intro
duced here three equivalent notations which are commonly used for compactness.
One also finds in the literature the notation spin function and without bra or ket.
Alternatively, one speaks about spin up () and spin down () states. In any
case, the relations
2 
= 3 2 
S 2 
= 3 2 
S
4 4
(2.94)
Sz 
= 
Sz 
= 
2 2
hold, together with the orthonormality relations

= 
= 0 and 
= 
= 1. (2.95)
2.5 Angular Momentum 115
From the commutation rules (2.91) one may derive how the other components of
S
act onto the basis (here without proof):
i i
Sx 
= 
Sx 
= 
Sy 
= 
Sy 
= 
. (2.96)
2 2 2 2
In this basis each arbitrary spin state of an electron may be expressed as
 = +  +  , (2.97)
which are normalized by definition. As an exercise, the reader may show with the
aid of (2.94) and (2.96), that the parameters and give the polar and azimuthal
angles, respectively, at which the so defined spin state is oriented in space.
The above provides a toolbox which is fully sufficient to describe the properties
of the spin states. Nevertheless one often writes perhaps for historical reasons or
better visualization operators in the form of matrices and eigenstates as vectors,
the so called spinors
+
= and = + (2.99)
1 0
with the basis = and = .
0 1
116 2 Elements of Quantum Mechanics and the H Atom
S= , (2.100)
2
with
being a vector operator, the so called PAULI vector, which is built from the
PAULI matrices:
0 1 0 i 1 0
x = y = and z = (2.101)
1 0 i 0 0 1
3 1 0
S = Sx2 +
Sy2 +
2
Sz2 = 2 . (2.102)
4 0 1
For later use we note here that the PAULI matrices anticommutate
i j + j i = 2ij , (2.103)
and thus
x2 = y2 = z2 =
1 and x y = y x = iz . (2.104)
Section summary
Orbital angular momentum operators may be derived from L=r p , with
p = i, or be constructed from the commutation rules (2.75).
From the spatial representations of the operators z according to
2
L and L
(2.65) and (2.70), respectively, one obtains eigenvalues and wave functions
as summarized in (2.79)(2.83), and schematically illustrated in the vector
diagram Fig. 2.5.
The shape of the orbitals with lowest angular momentum s, p and d shown
in Fig. 2.6 should be memorized. General formulas and properties are sum
marized in Appendix B.1.2, specific expressions are tabulated in Table B.1.
The electron spin obeys the same commutation rules as orbital angular mo
menta. Its intrinsic angular momentum, the spin, is however s = 1/2 and the
projection quantum number ms = 1/2. A frequently used representation of
the spin operators are the PAULI matrices (2.101) and (2.102) which act on the
so called spinors two component representations of the spin eigenfunctions.
2.6 One Electron Systems and the Hydrogen Atom 117
2
=
H
p2
+ V (r) = Hr + L + V (r) (2.106)
2me 2me r 2
2 1 2
with Hr = r , (2.107)
2me r 2 r r
r + L
H + V (r) n m (r, , ) = Wn m n m (r, , ). (2.108)
2me r 2
2 1 d 2 d 2 ( + 1)
r + + V (r) Rn (r) = W Rn (r). (2.110)
2me r 2 dr dr 2me r 2
2 ( + 1)
Veff (r) = V (r) + . (2.111)
2me r 2
With this and the substitution
2 d2 un
2
+ Wn Veff (r) un (r) = 0. (2.113)
2me dr
Note: The total energy does not depend on the projection quantum number m and is
thus be written W = Wn . Zero energy is usually set for the completely unbound
system, i.e. for electron and nucleus at infinite distance with no kinetic energy.
Bound electrons have negative energies Wn < 0, while free electrons have total
energies W > 0. Extending (2.89) to the full electron wave function one often ab
breviates
Rn (r)Y m (, ) n m
. (2.114)
More specifically, one even writes these atomic orbitals shorthand as n = 1s, 2s,
2p, 3s, 3p, 3d, etc. in the notation introduced in Sect. 2.5.3.
We recall here the concept of atomic units (a.u.), introduced in Sect. 1.8.3:
energy Eh = me e4 02 h2 /4
length a0 = 0 h2 e2 m1
e / = / me Eh (2.115)
time t0 = 202 h3 e4 m1
e /.
Numerical values of these quantities are given in Appendix A, and the most recent,
accurate updates can be found at NIST (2010). We use these definitions here to
rewrite the radial S CHRDINGER equation 2
(2.113). We multiply (2.113) by me /
and a0 , and apply the identity a0 = / me Eh to obtain in dimensionless form:
2
1 d2 un ( + 1) V (r/a0 )
+ W n /E h + un (r) = 0. (2.116)
2 d(r/a0 )2 2(r/a0 )2 Eh
1 d2 un
2
+ Wn Veff (r) un (r) = 0 (2.117)
2 dr
( + 1)
with Veff (r) = V (r) + and for the H atom V (r) = Z/r.
2r 2
Following this scheme one may rewrite all atomic equations in a dimensionless,
rather clean looking form. Theoreticians, in particular, like this kind of equation very
much and even give the recipe to just set = e = me = 1 which really oversimpli
fies what has to be done. The procedure has one serious disadvantage: a dimensional
analysis is no longer possible which is often highly commendable to check com
plex calculations. Thus, we typically try to use equations in a form indicated by
(2.116), and carry the a.u. a0 , Eh , and t0 explicitly along. Sometimes one may even
be able to combine elementaryconstants to truly dimensionless quantities, such as
the fine structure constant = Eh /me c2 according to (1.10).
Up to now we have treated the problem as if the electron would orbit around a space
fixed centre. As the nuclear mass is much larger than the electron mass me in the
case of the H atom with a proton as nucleus mp 1840me the centre of mass
is indeed very close to r = 0. For more demanding precision one has, however, to
correct for the difference. As in classical mechanics (see corrections to the B OHR
model described in Sect. 1.8.5), in quantum mechanics too, one transforms the two
body problem into an effective one particle problem by replacing the electron mass
me with the reduced mass m e (1.143) of the system. All a.u. have, in principle, to
be replaced correspondingly (kinematic correction):
me e
m e
m
a0 a 0 = a0 , Eh E h = Eh , and t0 t0 = t0 . (2.118)
e
m me me
For simplicity we shall, however, in the following text continue to use me and the
units a0 , Eh and t0 , and refer to the exact calculations if relevant.
While we know already the angular part of the hydrogenic wave functions, we
are still left with the task to find physically meaningful solutions to the radial
S CHRDINGER equation (2.117). They have to behave reasonably at r = 0 (see
below) and must not diverge for r . From this follows necessarily that only a
particular set of discrete total energies Wn < 0 leads to such solutions. To find these
is the task at hand.
Before applying mathematics we want to obtain a qualitative picture to support
our physical intuition. Figure 2.7 illustrates this for the = 0 case in a C OULOMB
potential V (r) = Z/r. We derive the kinetic energy of the electron at different
120 2 Elements of Quantum Mechanics and the H Atom
W,V(r )
0
Wn r
Wkin
rcl
V
un (r ) large classically for
bidden region:
= h /p exponential
small decrease of
wavefunction r
Fig. 2.7 Bound state radial wave functions for s states, schematic. Top: C OULOMB potential
V (r) 1/r (red) and total energy Wn < 0 (black) determine the classical turning point rcl
(onset of the classically forbidden region). Bottom: The characteristic behaviour of radial wave
functions un is explained by changes of the kinetic energy Wkin in different regions of the poten
tial (see text)
positions in the potential from Wkin = Wn V (r) and take the corresponding DE
B ROGLIE wavelength = h/p = h/(2me Wkin )1/2 as an indication for changes in
the radial wave function un (r). It obviously will change more rapidly for small
r (large Wkin ) than in the neighbourhood of the classical turning point rcl , where
Wkin = 0. In the classically forbidden area with r > rk (negative kinetic energies)
we expect exponential damping of the wave function, as illustrated in Fig. 2.7.
In a next step we explore the limits for very large and very small r. For the
limiting case r we may neglect the potential altogether and (2.117) becomes
a simple oscillator equation:
1 d2 un
+ Wn un (r) = 0.
2 dr 2
The classical solution is un (r) exp(i 2Wn r). Since for bound states Wn < 0,
we note for large r
lim un (r) r n exp 2Wn r . (2.119)
r
In the opposite limit r 0 the centrifugal term ( +1)/2r 2 dominates the potential
in (2.117) and
1 d2 un ( + 1)
un (r) = 0
2 dr 2 2r 2
r0
has the solution un (r) = Ar +1 as one easily verifies. Thus, we note for small r
lim Rn (r) = lim un (r)/r r . (2.120)
r0 r0
2.6 One Electron Systems and the Hydrogen Atom 121
We now specialize to the H atom. In the present chapter we neglect the size of the
atomic nucleus (positively charged with +Ze) since nuclear radii rnuc are much
smaller than atomic radii, typically ratom 105 rnuc . Thus, we are dealing with a
pure C OULOMB potential
1 Ze2
V (r) = , (2.121)
40 r
apart from small and very small effects which will be treated in Chaps. 6 and 9,
respectively. The general solution of the radial S CHRDINGER equation (2.110) is
found by using a power series of the type
...
Rn (r) = exp 2Wn r Ak r k ,
k=
which includes the limiting cases just discussed. Well known results from mathemat
ics are used and we summarize here the results, again without proof. For hydrogen
like systems i.e. for one electron in the C OULOMB potentialof a Z fold charged
nucleus the radial function is
n 1
[(n + )!]2 k
2 +1
Ln+ () = (1)k+1 . (2.123)
(n 1 k)!(2 + 1 + k)! k!
k=0
We introduce here finally the often used term good quantum numbers: They char
acterize the eigenvalues of such observables that may be measured simultaneously
with the H AMILTON operator, i.e. their operators commutate with the Hamiltonian.
We already know and m as typical examples: they are part of the set of quantum
numbers characterizing the total energy of the system; z are simultaneously
2
L and L
measurable with H .
122 2 Elements of Quantum Mechanics and the H Atom
Wn / eV n
=0 =1 =2 =3 =4
0
8
6
5s 5p 5d 5f 5g
 0.85 4
4s 4p 4d 4f
 1.51 3
3s 3p 3d
 3.40 2
2s 2p Z2
Wn = E h
2n 2
independent of
13.6 1
1s
Fig. 2.8 Term energies of the hydrogen atom (Z = 1) for different n and
Table 2.1 The lowest atomic levels, their energies in the H atom and the degeneracy of the states
(with Eh = 27.2 eV)
Shell Orbital n m Wn Degeneracy
Without spin With spin
sum in shell sum in shell
K 1s 1 0 0 Eh /2 1 1 2 2
L 2s 2 0 0 Eh /8 1 2
2p 2 1 0, 1 3 4 6 8
M 3s 3 0 0 Eh /18 1 2
3p 3 1 0, 1 3 6
3d 3 2 0, 1, 2 5 9 10 18
N 4s, p, d, f 4 Eh /32 16 32
These solutions of the S CHRDINGER equation are the quantum mechanical equiv
alent to B OHRs stationary orbits. Substituting unl (r) according to (2.122) into the
radial equation (2.113) one finds the eigenenergies Wn for the H atom. Remark
ably, they are identical to the energies (1.137) from the B OHR model.8 These results
are summarized in Fig. 2.8 and Table 2.1.
We recall that the states are characterized by the principle quantum number
n = 1, 2, 3, . . . , the angular momentum quantum number (0 n 1) and
the projection quantum number m ( m ). Each set n m of quantum num
bers refers to a different wave function (atomic orbital). In addition, we also have to
consider the spin of the electron, with projections ms = 1/2.
8 The corrections for finite mass of the atomic nucleus (Sect. 1.8.5) also apply.
2.6 One Electron Systems and the Hydrogen Atom 123
Veff (r ) / E h
=1 =2
0.2
5 10 15 20
0
n=2 n=3
r /a 0
COULOMB potential
 0.5 n =1 = Veff for = 0
Fig. 2.9 Illustration of degeneracy: Shown are the pure C OULOMB potential (red line) and the
effective potentials (black) for the H atom. The term energies Wn are indicated by horizontal
lines in their respective effective potentials: for = 0 (dotted red), for = 1 (dashed grey) and for
= 2 (heavy, full red). As indicated, the C OULOMB potential leads to a characteristic degeneracy
of states with equal n but different
However, for the H atom the energies Wn depend only on the principle quantum
number n. Thus, the energy levels are degenerate, the total degeneracy in a shell
n being gn = 2n2 . This is summarized in Table 2.1 for n = 1 to 4, also showing
the assignment of orbitals with equal principle quantum number n (i.e. with equal
energies and similar orbital radii) to specific shells (K, L, M, N . . . corresponding
to n = 1, 2, 3, 4, . . .).
It is important to note that degeneracy is a special property of the pure
C OULOMB potential, while m degeneracy occurs for all atoms if no external field
is present. Figure 2.9 illustrates degeneracy in the potential energy diagram, also
showing the effective potentials (2.111) for = 1 and = 2 (for = 0 effective and
C OULOMB potential are identical).
The radial wave functions have very specific shapes, which may be understood from
the properties of the effective potentials. Figure 2.10 illustrates this schematically by
way of example for the n = 3 level and orbital angular momenta = 0 and 1. Shown
are C OULOMB potential, centrifugal potential and effective potential for = 1. The
classically forbidden areas (Wkin < 0) are grey shaded, the classical turning points.
While the radial functions for = 0 start with a finite value at r = 0, for = 1
the probability there is = 0 (since ( + 1)/(2r 2 ) ). Oscillations of the radial
functions are expected only in between the classically turning points.
Table 2.2 presents the radial wave functions Rn (r) for the six energetically low
est states (n 3) of atomic hydrogen (Z = 1) and Hlike atoms (Z > 1) in closed
form.
From the radial wave functions one derives the probability distributions
2
w(r)dr = Rn (r) r 2 dr (2.125)
124 2 Elements of Quantum Mechanics and the H Atom
1 R21 (r) = 1
( Z )3/2 e/2
2 6 a 0
3 0 R30 (r) = 1
( Z )3/2 (6 6 + 2 )e/2
9 3 a 0
1 R31 (r) = 1
( Z )3/2 (4 )e/2
9 6 a 0
for the electron to be found between r and r + dr. For comparison with the classical
picture of an orbiting electron one has to consult these probability distributions as
a function of distance from the nucleus. A graphical illustration of the radial wave
functions Rn (r) and the radial probability distributions w(r) is shown in Fig. 2.11.
Closer inspection of the probability distributions shows, that their maxima match ex
actly the radii of the respective B OHR orbits for the largest = n 1 at any given n!
The square of the wave function n m (r, , )2 represents the probability distribu
tion for finding electrons per volume element, or equivalently the electron density
within the atom as a function of position in space. Since these atomic orbitals form
2.6 One Electron Systems and the Hydrogen Atom 125
Fig. 2.11 Radial wave functions of the H atoms Rn (r) and probability distributions r 2 Rn 2 (r) for
the K, L and M shell. The dashed, vertical lines in the probability plots for the highest at a given
n indicate the corresponding B OHR radii. Note the drawing for the 1s orbital on a blown up rscale
3
z 1s 0 2p 1 3s0 3p 1
5 10 10
0 0 0 0
5 10 10
3
3 0 3 5 0 5 10 0 10 10 0 10
3d 1 3d 2 20 4 s 0 40 5s 0
10 10
0 0 0 0
10 10
20 40
10 0 10 10 0 10 20 0 20 40 0 40
x
Fig. 2.12 Density plots for some characteristic H atom wave functions n m . Plotted are equiden
sity lines (red high, grey low probability density). The distances are given in a0
a base for solving many key problems in atomic and molecular physics, we recom
mend our readers to visualize and memorize their general shape intensively. One
finds a host of instructive Internetsites, e.g. with Java applets to generate the H or
bitals in a variety of displays. Thus, we present in Fig. 2.12 only a small selection of
cuts through the 3D density distributions. Plotted are the contour lines of the den
sity n m (x, y = 0, z)2 = Rn (r)Y m (, )2 in the zx plane on a linear scale (in
contrast to many presentations in the WWW, where one finds the density plotted on
a logscale). For clearness we have indicated the highest densities by red areas.
126 2 Elements of Quantum Mechanics and the H Atom
We finally note again, that only the ns orbitals have a finite density at the origin,
ns (0) > 0, while
n m (0, , ) 0 for > 0. Specifically, from (2.122) and with
Y00 (, ) = 1/ 4 we record for later use:
n00 (0)2 = Z 3 / a 3 n3 . (2.127)
0
according to n (with n < n). Clearly, due to the m and (for the H atom) degener
acy, the angular momentum quantum numbers do not influence the position of the
spectral lines as one also verifies by a glance at Fig. 2.8. This holds at least in 1st
order approximation. Finer effects will be discussed in Chap. 6.
A lot of scientific detective work of the early pioneers went into discovering the
connection between spectra and term energies, and the series are named after those
who originally discovered them. Most prominent are the LYMAN (n = 1 in the
VUV), BALMER (n = 2 in the VIS and near UV) and PASCHEN series (n = 3 in
the near IR). Within the series, the BALMER lines are historically referred to as H
alpha, Hbeta, Hgamma etc., more generally the lines of the different series
are designated as Ly, and Ly, Ly , . . . , Ba, Ba, Ba , . . . and so on, with
= n + 2, = n + 3, etc.
One often needs to know the expectation values of the electron distance from origin
r to a certain power k. In principle, they could be determined by a large number of
individual measurements of this value in an suitably designed experiment. Quantum
mechanics provides:
k
r = n r k n
= Rn (r)r k Rn (r)r 2 dr = 2
Rn (r)r 2+k dr (2.129)
0 0
with 2
Rn (r)r 2 dr = 1.
0
These integrations are trivial but somewhat tedious. Using (2.122) and (2.123) one
obtains (for later reference):9
n2 1 ( + 1)
r
n = a0 1+ 1
Z 2 n2
"
#
2 n4 3 ( + 1) 1/3
r n = a02 2 1 + 1
Z 2 n2
1 1 Z
= (2.130)
r n a0 n2
1 1 Z2
=
r 2 n a02 n3 ( + 1/2)
1 1 Z3
= .
r 3 n a03 n3 ( + 1/2)( + 1)
One often hears the verdict that the B OHR model is basically wrong albeit predict
ing the correct term energies Wn and spectra (1.131) for the H atom in agreement
with quantum mechanics and thus should be abolished when teaching modern
atomic physics. We do not adhere to such dogmatic view.
Of course, B OHRs orbits have to be replaced by atomic orbitals, i.e. by den
sity distributions of the electrons within the atoms, and B OHRs quantization condi
tion (1.130), L = n, has to be confronted with its quantum mechanical analogue,
z () = m(). Obviously, m n 1 is not equal to n as postulated in the
L
B OHR model but it is a good first guess.
On the other hand, the concept of angular momentum quantization, of stationary
states, and of radiation emitted upon transition between them with h =
Wn was
a brilliant, daring and instrumental step on the way to understand the quantum nature
of microscopic phenomena even though the fact that states without orbital angular
momentum ( = 0) do exist, does not fit at all into the B OHR model. However, the
higher the angular momenta, the closer the atomic orbitals correspond to B OHRs
orbits. And in general, the correspondence principle holds:
However, the maxima of the probability distributions are of course not directly
observable; rather one may compare expectation values such as n rn
= r
. For
the largest angular momenta = n 1 one obtains from (2.130)
1 2 a0 a0
r
= lim n + n = n2 , (2.131)
n 1 2 Z Z
which obviously agrees in the limit with B OHRs prediction (1.135). Conversely,
for the smallest values of = 0, the average radius r
is distinctively larger than
predicted by the B OHR model, namely (3/2)n2 a0 /Z.
In summary, we do not feel that the B OHR model should be completely forgotten.
Apart from its outstanding historical impact, we shall come across a number of spe
cific aspects in modern AMO physics for which useful, simple models or concepts
have been stimulated by images of electrons moving on classical trajectories com
bined with appropriate quantization rules. We may think of B ORN O PPENHEIMER
approximation for molecular physics and of semiclassical trajectory calculations in
atomic scattering theory, to mention just two important, and very successful exam
ples or, as we shall see in the following section, the derivation of the magnetic
moment of an electron associated with its orbital angular momentum.
Section summary
The eigenenergies of the H atom, Wn = Z 2 /(2n2 )Eh , depend only on the
principle quantum number n . . . a specific consequence of the pure C OULOMB
potential. For precision measurements this has to be corrected by m e /me
where m e is the reduced mass of the electron.
Electron wave functions for the H atom can be expressed in analytical form
n m (r, , ) = Rn (r)Y m (, ) with the spherical harmonics Y m (, ) and
the radial function Rn (r) being proportional to the L AGUERRE polynomials.
The asymptotic behaviour of the wave functions should be memorized:
limr Rn (r) exp( 2Wn r) and limr0 Rn (r) r .
good quantum numbers characterize the eigenvalues of such observables that
may be measured simultaneously with the H AMILTON operator.
The so called normal Z EEMAN effect is actually not normal at all. It is observed
only in special cases. It concerns atoms in an external magnetic field. The word
normal simply alludes to the classical interpretation of such a situation ignoring
the electron spin. We shall treat atoms in external fields in great detail in Chap. 8.
Thus, the subject is touched here just briefly, and only since it confronts us for the
first time with the removal of a specific energy degeneracy, the m degeneracy in this
case.
2.7 Normal ZEEMAN Effect 129
with
W being the change of total energy in respect of the unperturbed state. Here
and in the following we use state vectors instead of wave functions n m n m
for compact writing.
The present problem is a particular simple case for the so called perturbation the
ory to which we shall come later in some detail. At this point we just remember that
according to (2.79) and (2.81) the eigenfunctions n m (r, , ) = Rn (r)Yn (, )
of the H atom, or their eigenstates n m
, respectively, are also eigenfunctions
(eigenstates) of z with
2
L and L
z n m
= mn m
.
L (2.136)
130 2 Elements of Quantum Mechanics and the H Atom
m=0
B
m =1
The latter relation obviously implies that in a magnetic field the energy degeneracy
for different m is removed. The originally identical energies of the 2 + 1 states
n m
which correspond to a given value of n and , now split into 2 + 1 different
energy sublevels. According to (2.137) the splitting is proportional to m and B.
The origin of this splitting is that the magnetic field breaks the spherical symmetry
characteristic for the unperturbed H atom.
We illustrate this for the example of an np state npm
with = 1 and sublevels
m = 1, 0, and 1. Figure 2.14 shows the energy changes
Wm as a function of the
magnetic field B.
One may observe this splitting e.g. in optical emission spectra. Figure 2.15 com
pares the emerging spectra for (a) a p s and (b) a d p transition. The individ
ual transitions are indicated by black, downward arrows. The selection rules applied
in this plot are
= 1 and
m = 0, 1. They will be derived and discussed in
Sect. 4.4.
As the degeneracy and hence the number of split levels is 2 + 1, a manifold
of transitions may occur if states with > 1 are involved. This is illustrated in
Fig. 2.15(b) where the upper levels correspond to a d state. Since, however, for
the normal Z EEMAN effect the splitting between neighbouring levels is always
B B according to (2.137), independent of and m, and since for all transitions
m = 0, 1, one nevertheless sees only a line triplet in all cases.
As mentioned at the beginning of this section: in reality this kind of Z EEMAN
effect is seen only in special situations (see Sect. 8.1.2), since usually the spin of
electrons plays an important role and complicates the observations.
We note here an important message from a situation which one typically en
counters in quantum systems with two or more degenerate states 1
, 2
, 3
: The
degeneracy is removed as soon as some additional, perturbing interaction V1 has to
be considered for which the matrix elements iV1 j
between some of these states
do not vanish.
2.8 Dispersion Relations 131
m=+2
(a) (b) +1
+1
BB 0
p 0
d 1
1 BB
2
+1
BB
p 0
s m = 1 0 +1
1
m = 1 0 +1
Fig. 2.15 Normal Z EEMAN effect for (a) p s and (b) d p transitions. In case of equal
splitting in the excited and ground states one observes in each case a line triplet in spite of the
5fold splitting in the d state
Section summary
A magnetic field B removes the central symmetry and hence the m degener
acy. For the so called normal Z EEMAN effect, theory predicts a level splitting
Wm = B mB.
The selection rules for optical transitions are
= 1 and
m = 0, 1.
We make a brief detour here, illuminating the borderline between atomic, molecular,
optical and solid state physics. Traditionally, dispersion relations are used in optics
and characterize an important material property: the dependence of the wavelength
(or wavenumber k = 2/) of electromagnetic radiation on its frequency . From
a quantum mechanical point of view one may generalize this to describe the relation
between the energy of a photon W = (or in fact the energy of any other particle)
and its wave vector k.
For the massless particle photon in vacuo, with c = = /k, the dispersion
relation
W = = ck (2.138)
is obviously a linear relation between energy and wave vector. In contrast, the en
ergy of a freely moving, nonrelativistic electron (whose mass is me ) is
me 2 p2 2 k 2
W (k) = WP + v = WP + = WP + , (2.139)
2 2me 2me
where we have used the DE B ROGLIE relation (1.118). WP allows for arbitrary en
ergy calibration and may e.g. account for a potential energy or the rest mass energy.
Thus, in this case the dispersion relation is quadratic. Both cases are illustrated
graphically in Fig. 2.16.
132 2 Elements of Quantum Mechanics and the H Atom
k k
It should be noted that the quadratic relation (2.139) for particles with mass is
also in accord with the quantum mechanical calculation for a free particle in a 1D
or 3Dbox according to (2.52) or (2.55). Providing the box is large enough, the
energies may be considered continuous, i.e. they generally follow (2.139). However,
this is only the most simple model for the electronic band structure in a solid. The
particle in a box model does not account for the fact that the electrons in a solid
do not really move freely: rather, the electrons move in the lattice of atomic ions
and experience a periodic potential with strong attraction close to the ionic cores.
Elsewhere the potential is partially screened by bound and other free electrons in
the solid. We mention here two crucial consequences arising from this situation.
First, the dispersion relation (2.139) will have to be modified, which in principle
requires a serious band structure calculation. Many phenomena may, however, be
explained by the so called parabolic approximation. It parameterizes the band en
ergy by introducing an effective mass me of the electron which may even depend on
the direction into which the electron moves:
2
kx ky2 kz2
W (k) = WP 2
+ + . (2.140)
2mx 2my 2mz
The sign allows to extend the concept also to electron holes (positive charges in
the latice) and both these quasiparticles may have different masses. This mass
enters into all further calculations on the dynamics and statistics of electrons and
holes in a solid, e.g. into the F ERMI D IRAC statistics discussed in Sect. 2.4.3.
Second, one has to account for the periodicity of the motion in the lattice. The
solutions for this problem are so called B LOCH waves
which are the product of a plane wave exp(ikr) and a periodic function uk (r). The
latter has to obey the periodic boundary condition
uk (r + T ) = uk (r), (2.142)
where T is any translation from one elementary cell of the lattice into another. Over
all we still expect the dispersion relation (2.139) for electron energies to be more or
less valid. As detailed studies show, one may have to replace the electron mass me
by a (different) effective mass meff but the quadratic dependence of the energy W
2.8 Dispersion Relations 133
on the wave vector typically holds. The electron motion in general averages over
the periodic potential. However, the energies and eigenfunctions will be influenced
strongly by the lattice potential if the electrons are particularly close to the ions or
particularly far away from them. This is specifically relevant when the electron wave
vector k = k BZ is at (or close to) the B RILLOUIN zone (BZ), i.e. if the correspond
ing wave functions are constructively interfering (see Sect. 1.4.9). Such electrons
feel the periodic potential strongly, all others experience only an average.
We cannot go into detail of these concepts which are fundamental for the the
ory of band structure in solid states. We just emphasize some aspects which are
also of importance in molecular physics. Thus, let us discuss the particularly simple
situation of the 1D case, which e.g. describes a chain of atoms with a distance a
from each other (also, with slight modifications, a ring like molecule). According to
(1.95) and (1.93) the B RAGG condition for the ntn BZ then reads
k = kBZ = n . (2.143)
a
The corresponding wave functions are essentially exp(ikBZ x) and exp(ikBZ x), and
for free travelling electrons the energy in that situation would be 2 k 2BZ /2me in both
cases. We have two energetically degenerate solutions. However, the most general
solutions are linear superpositions of both, i.e. standing waves, with the two physi
cally reasonable cases:
These correspond to cos kBZ x and sin kBZ x. The probability (x)2 to find the elec
tron close to the lattice ions is maximal in the first case, minimal in the latter. The
consequence of the perturbing periodic potential is removal of the degeneracy, sim
ilar to Z EEMAN splitting discussed in the last section. And since the perturbing
potential is highly attractive, we expect the energy to be lowered in the first case,
and to be risen in the second case. In effect, at the boundary of the BZ we expect
the energies to split into two, and a gap to arise between two bands of the otherwise
continuous 2 k 2 /2me distribution.
This is illustrated in Fig. 2.17(a). Since the periodicity of the system does not
favour any particular origin in the reciprocal lattice, one projects all the possible
energies onto the first BZ as shown in Fig. 2.17(b). Electrons in the system may as
sume all energies W in the grey shaded energy bands. The energy regions marked
band gap are energetically forbidden.
In summary, in the solid state case continuous energy bands, with gaps in be
tween them, replace the discrete energy levels which we have discussed for the
atomic case (specifically for the H atom in the present chapter). One must, how
ever, be aware of the fact that Fig. 2.17 shows only a particularly simple situation
(1D case, one valence electron only). In general, the band structure of solid states
is much more complicated, and depending on how many electrons are available to
fill the bands, it provides the basis for such different materials as metals, isolators or
semiconductors.
134 2 Elements of Quantum Mechanics and the H Atom
(a) W W (b)
3rd 2nd 1st 1st 2nd 3rd band gap
band gap
band gap
k k
 3 a
2  2 3
a 
a a 0
a a a
Fig. 2.17 Emergence of energy bands in a periodic system (lattice constant a) and band gaps.
(a) Allowed energies ( ) with distortions at the boundaries of the 1st, 2nd, and 3rd B RILLOUIN
zone in comparison with the free particle ( ) as a function of k. (b) Projection onto the first
B RILLOUIN zone
Section summary
Dispersion relations describe how the energy W of the system depends on
the wave vector k. The most simple cases are (a) the photon (a massless
particle) W = ck, and (b) an electron (a particle with mass) moving freely
W = 2 k 2 /(2me ).
The influence of an average potential can be accounted for by an effective
mass, replacing me which even may depend on the direction of the electrons
momentum.
B LOCH waves, (r) exp(ikr)uk (r), are constructed to include the period
icity of the lattice in the function uk (r) = uk (r + T ).
References
B ORN , M.: 1927. Das Adiabatenprinzip in der Quantenmechanik. Z. Phys., 40, 167192.
B ORN , M.: 1954. The N OBEL prize in physics: for his fundamental research in quantum
mechanics, especially for his statistical interpretation of the wave function, Stockholm.
http://nobelprize.org/nobel_prizes/physics/laureates/1954/.
E ISAMAN , M. D., J. FAN, A. M IGDALL and S. V. P OLYAKOV: 2011. Invited review article:
singlephoton sources and detectors. Rev. Sci. Instrum., 82, 071101.
H EISENBERG , W. K.: 1932. The N OBEL prize in physics: in recognition of the great merits
of his theoretical and experimental investigations on the conduction of electricity by gases,
Stockholm. http://www.nobelprize.org/nobel_prizes/physics/laureates/1932/.
ISO 21348: 2007. Space environment (natural and artificial) Process for determining solar
irradiances. International Organization for Standardization, Geneva, Switzerland.
NIST: 2010. Reference on constants, units, and uncertainties, NIST. http://physics.nist.gov/cuu/
Constants/, accessed: 7 Jan 2014.
S CHRDINGER , E. and P. A. M. D IRAC: 1933. The N OBEL prize in physics: for the discov
ery of new productive forms of atomic theory, Stockholm. http://nobelprize.org/nobel_prizes/
physics/laureates/1933/.
Periodic System and Removal of Degeneracy
3
Overview
This is a quite compact and important chapter. The attentive reader should
be able to work through it rather quickly after having refreshed her or his
basic knowledge in the previous two chapters. We shall make here the first
steps to generalize the methods successfully applied for the H atom, and
allow for an interaction potential which is no longer strictly proportional
to 1/r. This forms the basis for understanding the physics behind the peri
odic system of the elements which is summarized in Sect. 3.1. The most sim
ple multielectron systems are the alkali atoms. Their energy levels are dis
cussed in Sect. 3.2 phenomenologically and analyzed qualitatively and quan
titatively, briefly explaining quantum defect theory. In Sect. 3.3 we introduce
time independent perturbation theory a tool that will be used later on quite
frequently and illustrate it by way of example for the energies of alkali
atoms.
The periodic system provides the structural basis for our understanding of atoms
and molecules. Thus, albeit fairly well known, it appears worthwhile to summarize
here the underlying concepts, observations and definitions.
The theory of the hydrogen atom as a prototype atom contains already all ingredi
ents needed for understanding of how the more complex atoms are built up. The
periodic system of the elements follows naturally from the so called Aufbau princi
ple (from German Aufbau = buildup): to a first order approximation one treats
the N electrons of an atom (with a nuclear charge Z) as independent of each other
and assumes that their respective wave functions look very similar to those of the
electron in a hydrogen atom. It turns out that this approach serves astonishingly well
for a first guess. Of course one has to modify the potential that each individual elec
tron sees with all the other electrons around it clearly cannot simply be Z/r:
the nuclear charge will partially be screened off by (N 1) electrons.
We shall discuss this screening and its consequences in detail later. Here we
simply note, that each electron (numbered as i = 1, 2, . . . , N ) is characterized by a
set of quantum numbers:
They correspond to the quantum numbers of the electron in an H atom. The entirety
of quantum numbers for all electrons of an atom in a particular state is called its
In its best known form the PAULI exclusion principle (short PAULI principle,
N OBEL prize 1945) states that no two electrons in the same atom can have the
same four quantum numbers, briefly
More generally, no two identical fermions (i.e. particles with half integer spin
s = 1/2, 3/2, etc.) can be in the same quantum state. This is an empirically con
firmed property of fermions. In Sect. 7.3.1 we shall discuss and use the quantum
mechanical formulation: The total wave function of identical fermionsis antisym
metric in respect of the exchange of two particles. We shall see there that both for
mulations are completely equivalent.
3.1 Shell Structure of Atoms and the Periodic System 139
One defines so called electron shells which are occupied by all electrons of an atom
with the same principle quantum number n. The shells are denoted by the letters K,
L, M, N, . . . , as summarized in Table 3.1. As already noted in Sect. 2.5.3, the letters
s, p, d, f, g stand for the angular momentum quantum numbers = 0, 1, 2, 3, 4. The
number of states in a shell with the principle quantum number n is 2n2 (including
spin states with ms = 1/2) all of which are degenerate in the H atom case.
According to the PAULI principle, each state may be occupied with no more than
one electron. The electron configuration of an atom is written in compact form as
illustrated in Table 3.2 for the ground state of some light atoms. For larger atoms
(e.g. Na) one summarizes the inner shells by the symbol of the next smaller rare gas
atom in brackets [ ].
The Aufbau principle of the periodic system assumes that these shells are sub
sequently filled as the number of electrons increases. Figure 3.1 illustrates the shell
filling scheme up to neon graphically.
140 3 Periodic System and Removal of Degeneracy
s p s p
L L
K H K He s p s p
L L L L
K Li K Be K B K C
L L L L
K N K O K F K Ne
Fig. 3.1 How the K and L shell of the periodic system of elements are filled with electrons. Arrows
indicate the spin projection (1/2) of the electrons
Table 3.3 gives a complete overview of the periodic table of elements, indicating by
colour shadings how the electron shells are filled. The elements within one group
(vertical columns) have equivalent outer shell electrons (differing only by n) and
typically show corresponding similarities in their chemical and physical properties.
Each period (horizontal row) corresponds to the filling of one particular shell and
comprises elements with different electron configuration and usually different prop
erties. Exceptions are the d block of transition elements, the Lanthanides and the
Actinides, for which deferred inner electron shell filling occurs. The radioactive el
ements of the 7th period have been (except for Fr and Ra) generated artificially
at heavy ion storage rings (in very low concentration). They typically have mean
lifetimes of only seconds to minutes, and little is known about their electron config
uration or even about their physical and chemical properties. Nevertheless, they are
interesting objects for fundamental studies.
We refer the reader to numerous, well linked presentations in the internet, such
as W IKIPEDIA CONTRIBUTORS (2014). The source par excellence for further infor
mation is, however, NIST (2011) from where a host of tabulated properties of the
elements including all available spectroscopic data can be retrieved. Quite instruc
tive is also the animation from the U NIVERSITY OF C OLORADO (2000).
Table 3.3 Periodic system of the elements. Left on the top of the elements the atomic number is given, below each element the configuration of the last
3.1
built in electron is shown. The filling of different shell is marked by colour shadings: s electrons and p electrons determine the main groups, adding
d electrons leads to the subgroups in the middle of the periodic table that contain all transition metals. The deferred addition of 4f and 5f electrons
occurs for Lanthanides and Actinides
Shell Structure of Atoms and the Periodic System
141
142 3 Periodic System and Removal of Degeneracy
Be Mg Zn Cd Yb Hg
25
3d 104s 24p 6
[Ar]3d 104s 2
[Kr]4d 105s 2
[Xe]4f 146s 2
20
3s 2
WEA and WI / eV
15
10
0
Z= 0 20 40 60 80
He Ne Ar Kr Xe Rn
Fig. 3.2 Ionization potentials WI (red) and electron affinities WEA (grey) of the atoms as a func
tion of nuclear charge Z. The full vertical lines indicate shell closures, the dashed lines the closure
of subshells, corresponding to the electron configuration given
The essential basis for the experimentally well confirmed shell structure of the
elements is (beyond the PAULI principle) the astonishingly good model of quasi
independent electrons in the so called central field approximation. We shall explain
this in detail in Chap. 10 where a profound understanding of the periodic system
will emerge.
The ionization potentials (IP) WI for all natural elements of the periodic system
are presented in Fig. 3.2. The diagram illustrates impressively the shell nature of
atomic structure: The IPs assume very pronounced maxima for the rare gas atoms
(He, Ne, Ar, Kr, Xe, Rn) that are characterized by closed shell. They are chemically
particularly inert. Smaller maxima are also in between, whenever a subshell for one
particular is completely filled.
A complementary behaviour is seen in the electron affinities WEA (grey line),
i.e. the negative of the binding energies, which may be gained by the formation of
a negative ion (anions) energies. They correspond to the negative binding energies
of the electrons in the anion (WB = WEA ) and are particularly high, if the corre
sponding neutral atom has just one vacancy in the outer shell for the electron to be
accepted which holds for the H atom and the halogens (F, Cl, . . .) and becomes
zero for the rare gases with their closed shells: no stable rare gas anions are known
to exist (see also Sect. 6.4.5, Vol. 2).
The atomic radii also support the shell structure of atoms very clearly. They are
plotted in Figs. 3.3 and 3.4 as functions of nuclear charge Z. Of course, the term
3.1 Shell Structure of Atoms and the Periodic System 143
r / nm
methods: W IGNER S EITZ
radii (red) and VAN DER
WAALS radii (black) 0.2
0.1
He Ne Ar Kr Xe
Z
0.2
0.1
0
He Ne Ar Kr Xe
Z
atomic radius is not very well defined as we have learned in Chap. 2 the size of
an atom is characterized by the probability to find its electrons in a certain distance
of the nucleus, and a limit cannot be defined uniquely. One may e.g. use the so called
W IGNER S EITZ radius, rWS . That is the radius of a sphere of the same volume that
the atom occupies on average in the condensed phase if such a phase exists for that
particular atom. From particle density N or mass density , relative atomic mass Mr
and AVOGADRO number NA one finds
3 3 3Mr
rWS = = 3 . (3.4)
4N 4NA
A similar quantity is the VAN DER WAALS radius, that gives the distance up to
which chemically not bound atoms can approach each other (for a more precise def
inition see Sect. 3.2.6, Vol. 2). Both quantities are shown in Fig. 3.3 for the elements
H to Ba.
Alternatively, Fig. 3.4 shows calculated atomic radii. They are extracted e.g. from
quantum mechanical expectation values, as discussed in Sect. 2.6.11 relying on
good computations of the respective atomic wave functions. An empirical deter
mination is based on so called covalent radii that are derived from a set of well
known binding lengths of molecules, preferably diameters. By comparison with
144 3 Periodic System and Removal of Degeneracy
other molecules one may improve these estimates without big efforts as also shown
in Fig. 3.4 (asterisks). In spite of all ambiguities in defining the term atomic radius
one recognizes very clearly the general trends as a function of atomic number. Rare
gases typically have the smallest radii, alkali metals the largest: electrons in closed
shells all see essentially the same, high charge, while in the outermost shell of an
alkali atom this charge is strongly screened for the one valence electron due to the
core electrons.
We shall discuss this in detail in the following sections of the present chapter.
Before doing so we mention one more little difference between Fig. 3.3 and Fig. 3.4:
while the minima of the atomic radii for rare gases are quite pronounced in the
latter case, such clear cut minima cannot be identified for the VAN DER WAALS and
W IGNER S EITZ radii. Obviously, in addition to the electron density the influence
of polarizability plays here an important role.
Section summary
According to the PAULI principle each identical fermion in a quantum system
must differ by at least one quantum number.
The periodic system of the elements is based on the PAULI principle. The
PAULI principle prevents electrons from falling all into the energetically low
est level and thus ensures the variety of the chemical elements. One may say
that it is behind the mystery of life.
The atomic shells are denominated by K, L, M, N, . . . according to their prin
ciple quantum number n = 1, 2, 3, 4, . . . , respectively. Each shell may contain
up to 2n2 electrons. The Aufbau principle (building up principle) of the peri
odic system says that electrons fill states with quantum numbers n mms one
after the other (essentially) in numerical sequence.
Chemical and physical properties of the elements depend crucially on the
number of electrons in the outermost shell. Completely filled shells corre
spond to the rare gases; within each main period they have the highest ion
ization potential and the smallest radius (as represented e.g. by the W IGNER 
S EIZ radius Eq. (3.4)).
The most simple cases of multielectron systems are those where one or several
inner shells are completely filled with electrons (closed shells) and only one electron
is found in the outermost shell. A look at the periodic system in Table 3.3 identifies
these as the elements of the 1st group, the alkali atoms and the alkali like ions. Their
electronic ground state configuration is {[Rg]ns}, where [Rg] stands for a rare gas
configuration of the atomic core, e.g. Li: {[He]2s} Na: {[Ne]3s} K: {[Ar]4s}, etc.
In the following we shall focus on the active electron in the respective new shell,
filled only with this one electron of particular interest. It is called valence electron
(in the original German literature also Leuchtelektron = lighting electron).
3.2 QuasiOneElectron System 145
497
transitions (double arrows) 3p 3d
.2
are indicated with
.2
wavelengths given in nm. The 3s
460
3
81
.3
diagram has been generated
2.6
610
from the NIST data bank. The
term energies are related to
2
2
the respective (negative)
323.
2p
binding energies Wn of the
valence electron by
0.8
Wn W2s , with WI = W2s 1
67
being the IP of the system
0
2s angular momentum
The most detailed and precise information comes again from spectroscopy. The
spectra and hence the term schemes of the alkalis turn out to be similar to those
of the H atom with its 1s ground state electron with two important modifications:
(i) the active valence electron is now initially in an ns state with n > 1 and the inner
shells are filled (thus no VUV radiation is observed) and (ii) the degeneracy is
now removed (which makes the spectra richer, i.e. more complicated to interpret,
than that of the H atom).
As a characteristic example Fig. 3.5 shows the term diagram of Li(1s)2 n as
derived from a host of spectroscopic information collected by generations of spec
troscopists. This type of term scheme indicating the possible transitions is usually
called G ROTRIAN diagram.
The general relation between the energies Wn of stationary states and spectral
lines observed (in wavenumbers) is again
1 1
n n = = (Wn Wn ), (3.5)
hc
now, however, with different energies for each .1
The basics of emission, absorption and fluorescence spectroscopy will be intro
duced briefly in Sect. 4.2.2 and a variety of more refined modern methods will be
described later, e.g. in Sect. 6.1. Here we only give a survey over the collected re
1 Strictly speaking, one should even introduce an additional quantum number j for the total angular
momentum, which becomes increasingly relevant as Z gets larger. We have already familiarized
ourselves with it in the context of the S TERN G ERLACH experiment in Sect. 1.9.5 and shall come
back to it in detail in Sect. 6.2.5.
146 3 Periodic System and Removal of Degeneracy
Wn / E h H Li Na K Rb
ionization continuum
0
n=8
 0.020
 0.031 n=4 6p 4f 5f
4s 4p 4d 4f
4s 4p
4d 4f 5s
5p 4d 4f 6s 4d 7s 7p 5d
 0.056 n=3 4p 3d 5p 6p
3s 3p 3d 3p 3d 6s
3s 4s 5s
4p 5p
 0.125 n=2 3p
2s 2p 2p
4s 5s
2s 3s
 0.500 n=1
1s
Fig. 3.6 Overview on the term energies of the alkali atoms in comparison to the H atom. Charac
teristic is the removal of degeneracy and the decrease of the energies in respect of the H atom for
small angular momenta
sults for the alkalis and refer the interested reader to the spectroscopic data bank of
K RAMIDA et al. (2013) which we have already mentioned.
A comparison of the energies of the H atom with those of all alkali atoms (va
lence electron) shows Fig. 3.6. The characteristic removal of degeneracy lets the
term energies of the alkali valence electrons always lie below those of the H atom,
Wn < Eh /(2n2 ), owing to the higher nuclear charge Z. However, this lowering of
energy gets smaller as the orbital angular momentum gets larger. Figure 3.6 clearly
documents that Wns < Wnp < Wnd < Wnf . For the nf terms the energies of the al
kali atoms are practically identical to those in atomic hydrogen. We shall understand
in a moment why that is so.
Eh
Wn = with n = n . (3.6)
2n2
The parameter is called quantum defect and n the effective quantum number. For
the moment this is simply a parameter, determined experimentally, which allows
one to order the spectroscopic data in a systematic manner.
A comparison of experimental data shows that strongly depends on the orbital
angular momentum and upon closer inspection also slightly on the principle
3.2 QuasiOneElectron System 147
(n, )
ns
function of binding energy 1.36
Wn for orbital angular
momenta = 0, 1, 2, 3 and 4. 1.35
Full black circles are 0.89
experimental data taken from np
0.87
K RAMIDA et al. (2013), open 0.85
circles are denoted there as  4  2  0.8  0.4 0
determined by interpolation
or extrapolation . . . or by 0.014
semiempirical
calculations. . . . Full lines nd
represent linear least square 0.012
fits to the measured data.
Note the break in energy 0.010
scale for the ns and np series, 0.0015
and the different scale for the nf
nd, nf and ng series 0.0010
0.0004 ng
0.0000
 1.6  1.2  0.8  0.4 0
Wn / eV
quantum number n. Figure 3.7 illustrates this for Na as example. Obviously, for
large n the quantum defect approaches a constant value that depends only on . We
shall come back to the theoretical interpretation of in the framework of the so
called quantum defect theory (QDT) in Sect. 3.2.6. For a quantitative comparison
one fits the experimentally determined term energies, i.e. the excitation energies
in respect of the ground state, by an extended RYDBERG R ITZ formula (see e.g.
W EBER and S ANSONETTI 1987):
Eh
term energy Wn = Wn0 0 (3.7)
2(n (n, ))2
with (n, ) = + B/(n )2 + C/(n )4 + (3.8)
or (n, ) + DWn . (3.9)
Here Wn0 0 is the ground state energy, which is related to the IP by WI = Wn0 0 .
The quality of present day spectroscopic data (K RAMIDA et al. 2013) allows a very
precise determination of the parameters. The thus derived quantum defects of all
alkali atoms for large n are summarized in Table 3.4 and Fig. 3.8.
One sees very clearly that the quantum defect increases strongly with the nuclear
charge Z and decreases with increasing orbital angular momentum. We also want
to emphasize that for large n (small Wn ) decreases indeed linearly with Wn
(not shown here) according to (3.9) which allows for (i) a determination of WI with
spectroscopic accuracy and (ii) analytic continuation of (n, ) (W, ) into the
continuum where W > 0. In the following we shall try to understand these findings
148 3 Periodic System and Removal of Degeneracy
(Z1)e0
qualitatively and quantitatively by looking at the potentials and wave functions, and
finally we shall introduce QDT which explains that the quantum defect is not just
some kind of fudge parameter.
A quite reasonable model for the alkali atoms assumes that the Z 1 core elec
trons (which fill the inner shells of the core ion completely) do influence the energy
terms of the one outermost electron (the valence electron) only by screening the
pure C OULOMB potential of the nucleus. As illustrated in Fig. 3.9, we consider an
N electron problem (nuclear charge Z = N ). At large distances, the valence elec
3.2 QuasiOneElectron System 149
2 4 6
0
1/r VS (r) r / a0
2
VS (r ) / E h
4
6  Z/r
8
10
Fig. 3.10 Effective potential VS (r) in which the valence electron of an alkali atom moves (red),
compared to a fully screened C OULOMB potential 1/r and the unscreened potential Z/r that
the electron experiences close to the nuclear charge Z (here with Z = 11 representing the Na
atom). The grey shaded area indicates the radius of the ionic core
tron sees only the screened nuclear charge 1e and experiences something about
the nucleus only if it dives into the ionic core. One may treat this simplified problem
with nearly the same methods as the H atom except that we no longer have a pure
1/r potential.
Rather we have to solve the problem for a screened Z/r potential (in a.u.)
Z/r r 0
VS (r) = 1/r + VC (r) in between (3.10)
1/r r
where the ionic core is accounted for by a suitable, smooth potential, in the simplest
case e.g.
VC (r) = (Z 1) exp(r/rS )/r (3.11)
as illustrated schematically in Fig. 3.10.
n m (r) = Rn (r)Y m (, ).
Again we substitute un (r) = rRn (r) and obtain in complete analogy to (2.113) the
radial differential equation (in a.u.)
150 3 Periodic System and Removal of Degeneracy
d2 un ( + 1)
+ 2 W n VS (r) + un (r) = 0 (3.13)
dr 2 2r 2
with the important difference that VS (r) is no longer the pure C OULOMB potential
but rather a suitably screened one as described by (3.10).
The task at hand now is to identify from the infinite manifold of solutions for
(3.13) those which are physically meaningful. They have to reproduce the correct
asymptotic behaviour both for large and small distances r. Since we have derived
the corresponding relations in the case of the H atoms without any reference to
the interaction potential, (2.119) and (2.120) must be valid in the present case too.
r the radial wave function has to be damped exponentially, un
Thus, for
exp( 2Wn r), and Wn is the soughtafter energy of the stationary states of the
system. On the other hand, for r 0 the radial function has to follow un r +1 .
This very behaviour allows us to explain the experimentally observed depen
dence of the quantum defect (Table 3.4): according to (2.125) the probability to find
an electron at a distance r from the nucleus is w(r) = 4r 2 Rn 2 (r) = 4u2 (r).
n
Specifically for small distances it is thus r 2 +2 . Hence, the higher its angular mo
mentum, the less the electron is influenced by the ionic core. For very large the
wave function will be essentially hydrogen like, and the quantum defect correspond
ingly small, just as documented in Sect. 3.2.2.
For a truly quantitative analysis of wave functions and eigenenergies we have
to integrate (3.13) explicitly. Even though the radial S CHRDINGER equation for
a screened potential of the type shown in Fig. 3.10 cannot be solved analytically,
numerical integration of (3.13) may be done today on any PC without problems.
A number of robust and simple procedures are available, e.g. the often used RUNGE 
K UTTA method. Typically one integrates for a given trial energy Wn from the inside
outwards as well as from the outside inwards, starting with the asymptotic forms just
discussed. One accepts only such solutions that can be matched continuously and
differentiable and by variation of Wn one thus obtains the discrete eigenstates and
eigenenergies of the system.
Of course, the result of such calculations can only be as good as the potential
VS (r) which so far we have only described qualitatively. There are a number of
useful approximations to estimate VS (r). In the most simple approach one guesses a
parametric form which is fitted such that the experimentally observed energies of a
few states are reproduced exactly. With the thus determined empirical core potential
one may derive the wave functions for any state of the system and compute other
quantities, such as the energy terms, optical transition probabilities, polarizabilities
and so on.
0 3s
4 r 2Rn (r) / a 0
0.2
1 2s 3p
2
0
0.1
5 3d
1s
0 0.0
0 1 2 0 5 10 15 r / a0
Fig. 3.11 Radial electron probability distributions w(R) in Na for (a) the core electron orbitals
and (b) orbitals of the valence electron (alternatively in the 3s ground state and in the 3p and 3d
excited states). Also shown is the cumulated radial electron probability in the Na+ core (Z = 11,
closed [Ne] shell), downscaled by a factor of 40. Grey shaded is the radius of the ion core (1.8a0 ).
The distributions shown by red lines have been calculated ab initio using a finite difference atomic
structure code, FDAlin, conveniently provided by S CHUMACHER (2011). The black dashed and
dash dotted lines are derived from QDT and are not expected to be valid inside the ionic core (see
Sect. 3.2.6). Note in particular the excellent agreement for 3s
2 This holds at least for the smaller alkali atoms where relativistic effects play a minor role and
quantitatively that the valence electron stays predominantly outside the core. This
holds a fortiori for excited states, here very clearly seen for the 3d state the orbital:
its radial wave function (R3d (r) r 2 for small r) leads to very little probability for
finding the electron inside the core as Fig. 3.11 illustrates.
This explains again convincingly why the quantum defect decreases so rapidly
with as illustrated in Fig. 3.8. It also clarifies the overall striking similarity of the
alkali spectra with those of the hydrogen atom which we have seen in Fig. 3.6. The
nf electrons, for which at the origin R4f (r) r 3 and thus w(r) r 8 , practically
never come close to the nucleus and the quantum defect can essentially be neglected
as for all higher values of (see Fig. 3.8).
We recall again: degeneracy observed in H and H like ions is just a very specific
consequence of the pure C OULOMB potential. Deviations from the C OULOMB po
tential lead to different energies for different , i.e. to a removal of this degeneracy.
The deviations are the larger the more the electron feels of the ionic core. For very
large the situation is practically identical to the pure 1/r potential in the H atom
case.
Even though in principle the full quantum mechanical problem of bound states in
small alkali atoms may be treated numerically, it is useful to return once more to the
model of a single electron in an effective potential. So far, we have not yet under
stood why the extended RYDBERG R ITZ formula works so well with essentially
constant quantum defect (3.8) for large n as documented by Fig. 3.7. This remark
able observation hints at some genuine physical background, even though the ef
fective quantum number n and the quantum defect were introduced in (3.6) just
as useful empiric parameters. It turns out that understanding this physical origin of
the quantum defect allows one to extract useful concepts and further quantitative
information.
The roots of the so called quantum defect theory (QDT) go back to the early
days of quantum mechanics and H ARTREE (1928), while between 1950 and 1990
S EATON (1983), FANO (1986), J UNGEN (1996) and their students have developed
it into a powerful theoretical framework and extended it to multichannel problems
(MQDT). It has been applied successfully to calculate oscillator strengths for atomic
and molecular transitions including high lying RYDBERG states, to determine accu
rate ionization potentials and photoionization cross sections (see also Sect. 5.5), to
understand autoionizing series and perturbations in multielectron spectra, as well
as in electron ion scattering theory and is used even in solid state structure the
ory.
The key to QDT is that for large r far outside the atomic core electronic wave
functions evolve in a pure C OULOMB potential and may be derived as analytic func
tions also for noninteger quantum numbers with the correct damping at large r
(in a.u.)
3.2 QuasiOneElectron System 153
un (r) r n exp 2Wn r = r n exp r/n , (3.14)
r
3 The r n factor used here in contrast to (2.119) can improve the convergence.
4 This is a rather crude choice. It leads, however, to qualitatively correct wave functions. The radius
of the ionic core for Na+ is typically given in the literature as 0.095 nm = 1.8a0 . At this distance
this screened potential VS (r) is about 1.1/r.
154 3 Periodic System and Removal of Degeneracy
(a) (c)
0.6 0.2
H 18s
r Rn / arb. un.
0 0
H 20s
0.2
0 10 20
50 H 18s H 20s
1.0
Na+ core
(b) (d) Na 18s
0.6 0.2
Na 20s
0 0
0.2
0 10 20
50 Na 18s Na 20s
1.0 r / a0
0 500 1000
Fig. 3.12 Illustration of the quantum defect as phase shift = for Na in the 18s and
20s states: (a) The radial wave functions rRn (r) for a pure C OULOMB potential (H atom) are
compared (d) to those in an Na+ pseudopotential VS (r) (see Fig. 3.10). All wave functions are
normalized to their minimum, which is set to 1. The blow ups of the r scale (c) and (d), respec
tively, illustrate the strong influence of the Na+ ionic core (grey shaded area) onto the Na wave
functions. It leads to a shift of the rRn maxima in respect of the H atom; this shift is directly
related to the quantum defect (n, )
deeper attractive potential VS . While the Na 18s and 20s wave functions are nearly
identical in this small r range (apart from their magnitude owing to normalization),
they differ dramatically from their counterparts in the H atom. Outside the core the
potential is purely C OULOMBic for both Na and H hence the two pairs of wave
function differ essentially only by this phase shift .
QDT treats the outside wave functions as a linear superposition of two pure
C OULOMB functions. The mathematics involved is not completely trivial and
we can only outline here just a few basics, following the excellent review of
S EATON (1983) and his concise mathematical summary from 2002. The radial
S CHRDINGER equation (3.13) is solved in two parts:
"
FI (r) for 0 r < r0 with VS (r)
u(r) = (3.15)
FII (r) for r0 r with 1/r.
(If alkali like ions are discussed with a remaining charge ZC of the ionic core, the
1/r potential has to be replaced by ZC /r.)
3.2 QuasiOneElectron System 155
Since this is an ODE of 2nd order it has two linearly independent sets of C OULOMB
functions as solutions for which one finds various forms in the literature. For QDT
S EATON (1983) defines s(, ; r) as the regular solution, and c(, ; r) as the irreg
ular solution. When treating the H atom in Sect. 2.6.5 we have considered only the
regular solution un s(n , ; r) for n = 1/n2 (n being an integer); asymptoti
cally they are r +1 for r 0 and r n exp(r/n) for r . It is important to
note that for all other values of the regular solution s(n , ; r) diverges at large r
hence the n give the eigenvalues of the H problem.
The irregular solutions c(, ; r) are r at the origin so that any radial
wave function R(r) rc(, ; r) would diverge hence they were considered non
physical for the H problem. It is interesting to note, however, that for large r and
n = 1/(n + 1/2)2 (and only for these) they are damped r n exp(r/n).
Hence, it is plausible as shown already by H ARTREE (1928) that at arbitrary
energies = 1/n2 a physically meaningful solution of the problem for r > r0 is:
FII (r) = cos n s(, ; r) + sin n c(, ; r)
(3.18)
= (1)n cos()s(, ; r) + sin()c(, ; r) .
Bound States
Specifically for bound states = 1/(n )2 < 0, S EATON shows
sin(n ) n3 1/2
s(, ; r) = (1) cos n K
(2)1/2 K 2
(3.19)
n3 1/2
cos(n )
c(, ; r) = (1) + sin n K
(2)1/2 K 2
Since n = (n+1 n ) 2/n3 in the limit of large n, the bound state functions
s(n , ; r) are said to be normalized per unit energy in this limit.
Since the component of the solutions (3.19) diverges for large r it is not suit
able for the description of any realistic physical wave function. And indeed, when
inserting (3.19) with (3.21) into the general solution (3.18) the terms cancel and
one obtains for r > r0 simply:
3
1/2
n 2r
FII n , ; r = (1) K n , Wn , +1/2 . (3.24)
2 n
This is the soughtafter general wave function in a pure C OULOMB potential outside
the atomic core for arbitrary energy! The W HITTAKER functions are included in
advanced mathematical programmes, e.g. in Mathematica, and we have computed
these functions for the 18s and 20s case of Na with n = n 1.3848. The result
is completely identical to the wave functions shown in Fig. 3.12(b) and (d) within
the limits one could see in the graphs except inside the ionic core where it is not
expected to be valid.
3.2 QuasiOneElectron System 157
Continuum States
We have already indicated that the behaviour of continuum wave functions may be
gleaned from analytic continuation of the quantum defect (n, ) (, ) for =
k 2 > 0. What we have vaguely addressed as phase shift between the oscillations
of the bound state wave functions will then become indeed a genuine phase shift
between a pure outgoing C OULOMB wave and the continuum wave function dis
torted by the screened Z/r potential of the ionic core. Here too, (3.18) applies for
the outside part of the wave function, FII , and at a suitable r0 one has to join FI
and FII . Again various forms for the regular and irregular pure C OULOMB contin
uum wave functions are used. According to S EATON (2002), Eqs. (86)(90) and
(113)(118)
' (
s(, ; r) Im exp i( /2) Wi/km +1/2 (2ikr)
' ( (3.27)
c(, ; r) Re exp i( /2) Wi/km +1/2 (2ikr)
158 3 Periodic System and Removal of Degeneracy
0.4
s (, ;r)
0.0
c( , ;r)
0.4
0 5 10
r
Fig. 3.13 C OULOMB continuum s waves ( = 0) at = 4 for attractive potential: regular solution
(red line, zero at origin), irregular (black line, finite at origin) adapted from S EATON (2002). The
asymptotic behaviour (3.30) (pink and grey dashed lines, respectively) gives a nearly perfect match
to the exact solutions already at r 3
where Wi/km +1/2 (2ikr) is again a W HITTAKER W function, this time to a com
plex argument, and for the presently considered attractive C OULOMB potential we
have r > 0. The C OULOMB phase shift
decreases rapidly with k and corresponds to the phase difference of the C OULOMB
in respect of a free spherical wave (not to be confused with the phase shift () due
to the ionic core potential VS being different from purely C OULOMB).
We have suppressed here a discussion on the important question of normalization
and simply note that
s(1 , ; r)s(2 , ; r)dr = (1 2 ), (3.29)
0
i.e. s(1 , ; r) is normalized to scale (see Appendix J). Again, the regular solution
goes to 0 at the origin while the irregular solution stays finite or diverges. For large r
these wave functions are essentially outgoing spherical waves apart from a slowly
varying logarithmic phase shift characteristic for the C OULOMB potential (see (J.8))
i.e. they behave essentially like sine and cosine functions:
2 2
lim s(, ; r) = sin and lim c(, ; r) = cos (3.30)
r k r k
1
with = kr + ln(2kr) /2 + (k, ). (3.31)
k
Figure 3.13 illustrates these continuum functions for = 0 at a moderate energy =
4 or k = 2 where 0 0.078 102 . The exact solutions (adapted from S EATON
(2002), Fig. 1) are apparently matched rather perfectly by their asymptotic form
(3.30) already at surprisingly small values of r 3.
We now come back to the general QDT theme of constructing a wave function
for r > r0 , this time in the continuum, using the analytic continuation () de
rived from bound state quantum defects. Being interested in large r we insert the
3.2 QuasiOneElectron System 159
Z 2 (Z qs )2
Wn = 2
Eh = Eh , (3.33)
2n 2n2
or in wavenumbers
(Z qs )2 (Z qs )2
n = E h /2 = R
hcn2 n2
2Wn  n Z Z qs
= = = . (3.34)
Eh R n n
Plots of the square roots of the energies as a function Z are called M OSLEY di
agram and should give according to (3.34) a straight
line. The screening parameter
qs is then obtained from the axis intercept of 2Wn /Eh where Z qs = 0. Full
screening of the C OULOMB potential would imply qs Z 1. By way of exam
ple we discuss the energy terms for Na and Nalike ions with data taken from the
K RAMIDA et al. (2013) data bank. The series starts with Z = 11 (Na) and continues
with Mg, Al, Si, P, S, Cl, etc.
Figure 3.34 presents the data for the ns, np and nd terms with n = 3, 4 and 5.
The graph documents that indeed the square root of the energies follows to a very
good approximation a straight line as a function of the nuclear charge Z as predicted
by (3.34). Admittedly we still need two parameters, qs and n n to fit a whole
series well. As expected, for states with the highest orbital angular momentum the
160 3 Periodic System and Removal of Degeneracy
4
3s
3p
Z10.03
3d
3 2.91
4s
2Wn / E h 2 4p
Z10.02 4d 5s
3.90 5p
1 5d
Z10.02
4.97
0
10 11 12 13 14 15 16 17 18 19 20 Z
Na I Al III P V Cl VII Ca X
Fig. 3.14 M OSLEY diagram for Na like ions and different n states. Following spectroscopic
traditions we designate the degree of ionization with roman numbers: I for neutral atoms, II for
singly ionized, III for doubly ionized atoms, etc. Symbols correspond to the experimental data
according to K RAMIDA et al. (2013), the lines are fits following essentially (3.34). Nearly perfect
screening (qs = 10) is shown for the d states
energies agree best with the straight lines as validated in Fig. 3.14 particularly well
for the 3d, 4d and 5d series. The screening in all three cases is nearly perfect (qs =
10) and die slope of the lines is almost ideally 1/n. For the p and s states the
screening is obviously not perfect: in particular for the 3s states the screening is
only just above qs = 9, i.e. the electron still sees effectively nearly two charges of
the nucleus. In any case, the surprisingly good agreement of experimental data with
the prediction (3.34) supports very impressively the model of a quasioneelectron
system over a whole isoelectronic series.
For higher Z the lines typically fall into the Xray region. M OSLEY diagrams are
therefore also used successfully to characterize Xray spectra from inner shells (see
Sect. 10.5.2).
Section summary
The spectra of alkali atoms resemble those of the H atom in the VIS except
that now degeneracy is removed and consequently several series of lines are
observed for each principle quantum number n.
Thus, a good concept to understand these spectra is a quasioneelectron
model where all but one active electron (the valence electron) constitute the
closed shell of the ionic core. They (partially) screen the charge of the nucleus
but otherwise do not participate in standard spectra.
The energy terms of the alkali atoms (and alkali like ions) depend thus
on n and of the valence electron and may be described by Wn =
Eh /[2(n )2 ]. The quantum defect (n, ) depends strongly on and
weakly on n. For large n it becomes nearly independent it. The higher , the
smaller is , i.e. the closer the energy levels are to those of the H atom (and
H like ions, respectively).
3.3 Perturbation Theory for Stationary Problems 161
These findings are explained by the low probability of the valance electron to
be found close to the nucleus, being r 2 +2 for r 0.
Quantum defect theory (QDT) gives a very useful analytic foundation to the
observed behaviour of quantum defects: it turns out that the quantum defect
relates to the phase shift = between a pure C OULOMB wave function
(H atom) and the wave function for the alkalies.
M OSLEY diagrams summarize the spectral lines of a whole series of ions with
equal electron configuration as a function of the nuclear charge Z. Plotting
2Wn /Eh vs. Z leads to essentially straight lines.
is assumed to be solved. Let the Hamiltonian for the new problem be given by
=H
H (r, p
0 + U ), (3.36)
162 3 Periodic System and Removal of Degeneracy
with the perturbation operator U (in the most simple case just a scalar potential).
For perturbation theory to be applicable we have to assume that for the averaged per
 H
turbation U 0
 holds. We thus write the S CHRDINGER equation (3.12)
in the form
0 + U
(H )k = Wk k , (3.37)
which has to be solved by the perturbation approach. One then expands the ener
gies and wave functions into a series in terms of small quantities Wk and k
assuming that U is also small on the order of . The formally correct procedure is
to compare quantities of the same power in the smallness parameter and thus to
obtain correction terms of the order , 2 , etc. These are the corrections in 1st, 2nd,
etc. order perturbation theory. The key quantities characterizing the perturbation are
the matrix elements:
(0) (0)
Uj k = j U k = j(0) (r)U (r, p)k(0) (r)d3 r. (3.38)
We abbreviate the procedure just outlined somewhat. Since the eigenfunctions i(0)
of the unperturbed Hamiltonian H 0 , derived from (3.35), form a complete, orthonor
mal basis set, any function may be constructed as linear superposition from them
including the solution of (3.37). Thus, we use the perturbation ansatz
(0)
k = ai i with ak  1 and ai  1 for i = k. (3.39)
i
The conditions ak  1 and ai  1 are crucial; they establish the essence of the
perturbative approach.
Introducing (3.39) into the S CHRDINGER equation (3.37) we obtain
0 + U
[H Wk ] ai (0) = 0
i
i
0
H
(0) ai (0) Wk ai (0) = 0
ai i + U i i
i i i
(0) (0)
ai (0) = 0.
ai W i W k i + U (3.40)
i
i i
In the last step we have made use of 0th order solution (3.35). Next, (3.40) is multi
(0) (0) (0)
plied from the left with k and integrated over all space. With k i
= ki
one obtains
(0) (0) (0)
ak Wk(0) Wk + ak k(0)
U k + ai k U i = 0.
i=k
3.3 Perturbation Theory for Stationary Problems 163
Since ai  1 for i = k, and since the matrix elements of the perturbation operator
are small too, one can neglect to 1st order the whole sum over i = k in this
U
expression. With this ak may be factored out, and one arrives immediately at the 1st
order correction for the energy
(0) (0)
W = Wk Wk = Ukk = k U k ,
(0)
(3.41)
Inserting the 1st order solution (3.41) for the energy correction this becomes
(0) (0) (0) (0) (0)
0 = aj Wj Wk k U k ai j U i .
(0)
+
i
If again one neglects all terms that are quadratically small, the third term can be
dropped and of the sum only the term with i = k remains (since ak 1), so that
(0) (0) (0) (0)
0 = aj Wj Wk + j U k ,
 (0)
j(0) U Uj k
(0) k (0) (0) (0)
k = k + (0) (0)
j = k + (0) (0)
j . (3.42)
j =k Wk Wj j =k Wk Wj
For the next step we insert the results of 1st order perturbation theory again into
the S CHRDINGER equation (3.37) and repeat the whole procedure. The 2nd order
correction for the energy is thus derived as follows:
 (0)
j(0) U
k(0) H0 (0) + k (0)
Wk = +U k (0) (0)
j
j =k Wk Wj
164 3 Periodic System and Removal of Degeneracy
 (0)
j(0) U
(0) (0) (0) (0)
k k
(0) k
= Wk + U k + (0) (0)
U j
j =k Wk Wj
(0)
Uj k 2
Wk = Wk + Ukk + (0) (0)
. (3.43)
j =k Wk Wj
1. We see that many states (in principle an infinite number) may contribute to the
determination of wave functions as well as of energies. Practical considerations
will limit the efforts, and one has to choose judiciously which states are included
in a perturbation expansion.
2. The contribution of individual states j depends both on the magnitude of the
(0) (0)
matrix elements Uj k  and on the resonance denominators 1/(Wk Wj ). The
closer a the energy of a perturbing state j to that of state k which is perturbed,
the stronger the modification.
(0) (0) (0) (0)
3. If several states 1 , 2 , 3 are degenerate, i.e. have the same energy W1 =
(0) (0)
W2 = W2 , one has to be careful with the perturbation approach because of
the resonance denominators. Only if the nondiagonal matrix elements of the
perturbation disappear, i.e. only if U12 = U13 = U23 0, one may follow the
perturbation procedure just described.
In other words, we have now to solve a set of g linear algebraic equations, which
can lead to a set of g eigenvalues W and state vectors. Generally speaking, one
has to diagonalize the Hamiltonian matrix by a unitary transformation. The matrix
elements of the diagonalized matrix are the eigenvalues of the system.
As well known, a solution of (3.45) only exists if the determinant of the matrix
vanishes. Thus, one first has to find the solutions of
W
det(H 1) = 0. (3.46)
To illustrate how to use 1st order perturbation theory, we briefly return to the alkali
atoms as a simple example. We emphasize the pedagogical aspect of this subsection
that is not meant to generate accurate numbers for energy levels. With the interaction
potential (3.10) discussed in Sect. 3.2.3 the Hamiltonian of the perturbed problem
(in a.u.) is written
= 1
2 + VS (r) = H
H 0 + VC (r) with VS (r) = 1/r + VC (r), (3.47)
2
and we recognize
0 =
2 /2 1/r
H
being the unperturbed Hamiltonian of the H atom.
The spherically symmetric perturbation potential U (r) = VC (r) originates from
the interaction of the valence electron with the full nuclear charge Z at small r and
screening at larger distances. For demonstration we use again the particularly simple
model potential VC introduced in (3.11). Figure 3.15 illustrates this potential for Na
(Z = 11). The screening radius has now been calibrated, as described below, to fit
the experimentally determined ground state binding energy W3s .
The matrix elements of the perturbation (3.11) are computed with the hydrogen
(0)
eigenfunctions Rn (r)Y m (, ) (0th order solutions), which have been treated in
detail in Sect. 2.6.1. Since the perturbation potential is spherically symmetric and
166 3 Periodic System and Removal of Degeneracy
4
VC (r) =
6 (Z 1) r/rS
e
8 r
acts only on the radial part of the wave function, the matrix elements simply are
(0) (0)
VC n m,n m = mm VC (r)Rn (r)Rn (r)r 2 dr = VC n ,n . (3.48)
0
In 1st order perturbation theory only the diagonal matrix elements are needed. We
(0)
make explicit use of the radial wave functions Rn (r) of the H atom as given in
(2.122), using the series expansion (2.123) for the L AGUERRE polynomials. The
numerical integration of (3.48) can be done with standard desktop computing pro
grammes within seconds.
(0)
Inserting the binding energies Wn = 1/2n2 of the hydrogen atom into (3.41)
and noticing that with VC (r) < 0 the matrix elements are also negative, the energies
of the alkali atoms in 1st order perturbation theory finally are (in a.u.)
(0) 1
Wn = Wn + VC n ,n = VC n ,n . (3.49)
2n2
All energy terms are indeed lowered in respect of the H atom as experimentally
observed.
In Table 3.5 we communicate some numbers that we have derived in this man
ner for Na and compare them with the experimental data from the NIST data bank
(K RAMIDA et al. 2013). For comparison, H atom binding energies are also shown.
Acronyms and Terminology 167
The data illustrate quite clearly what may be achieved with such a simple approxi
mation. Of course, the results can only be as good as the perturbation potential that
describes the additional interaction and one cannot expect miracles from (3.11) with
only one free parameter a rather crude guess of the perturbation potential. An ad
ditional problem is, that the wave functions experience a strong phase shift in the
ionic core, as we have seen in Sect. 3.2.6. This is of course not taken into account
in 1st order perturbation theory where the energy is computed from 0th order wave
functions for the H atom.
In view of these difficulties, the results shown in Table 3.5 may be considered
reasonable. But it becomes also obvious that 1st order perturbation theory should
only be used for a first orientation in such problems with several competing influ
ences. If precision is demanded one has to resort at least to the methods outlined in
Sect. 3.2.
Section summary
Perturbation theory can often give important qualitative and semi quantitative
insights into the physics expected from exposing atoms or molecules to spe
cific interactions as long as the averaged perturbation is small compared to
 H
the Hamiltonian defining the basic interaction in the system, U 0
.
In 1st order perturbation theory the change of energy (of nondegenerated
(0) (0)
states) is
W = Ukk = k U k
, while the 1st order wave function is
given by (3.42).
Higher order approaches are needed if first order (diagonal matrix elements)
vanishes.
For degenerate states a different approach has to be taken. One expresses the
perturbed Hamiltonian in matrix form (3.44), using as many and as good basis
functions as appropriate. One then has to diagonalize this matrix, using the
standard procedures of linear algebra. The procedure can be extended to quite
general problems, providing suitable basis functions are available.
References
FANO , U. and A. R. P. R AU: 1986. Atomic Collisions and Spectra. Orlando: Academic Press Inc.,
409 pages.
F REEMAN , R. R. and D. K LEPPNER: 1976. Core polarization and quantum defects in high
angularmomentum states of alkali atoms. Phys. Rev. A, 14, 16141619.
H ARTREE , D. R.: 1928. The wave mechanics of an atom with a nonCoulomb central field. Part I.
Theory and methods. Proc. Camb. Phil. Soc., 24, 89110.
I DZIASZEK , Z. and P. S. J ULIENNE: 2010. Universal rate constants for reactive collisions of
ultracold molecules. Phys. Rev. Lett., 104, 113202.
ISO 21348: 2007. Space environment (natural and artificial) Process for determining solar
irradiances. International Organization for Standardization, Geneva, Switzerland.
J UNGEN , C.: 1996. Molecular Applications of Quantum Defect Theory. New York, London: Taylor
& Francis, 664 pages.
K RAMIDA , A. E., Y. R ALCHENKO, J. R EADER and NIST ASD T EAM: 2013. NIST Atomic
Spectra Database (version 5.1), NIST. http://physics.nist.gov/asd, accessed: 7 Jan 2014.
M ARGENAU , H.: 1984. The Miracle of Existence. Woodbridge, CT, USA: Ox Bow Press.
NIST: 2011. NIST physics laboratory holdings by element, NIST. http://physics.nist.gov/
PhysRefData/Elements/per_noframes.html, accessed: 7 Jan 2014.
O SPELKAUS , S. et al.: 2010. Quantumstate controlled chemical reactions of ultracold potassium
rubidium molecules. Science, 327, 853857.
PAULI , W.: 1945. The N OBEL prize in physics: for the discovery of the exclusion princi
ple, also called the pauli principle, Stockholm. http://nobelprize.org/nobel_prizes/physics/
laureates/1945/.
S CHMIDT , K. and M. A. L EE: 1998. Visual Schrdinger: A visualizersolver. http://fermi.la.asu.
edu/Schroedinger/, accessed: 7 Jan 2014.
S CHUMACHER , E.: 2011. FDAlin programme, computation of atomic orbitals (Windows and
Linux), Chemsoft, Bern. http://www.chemsoft.ch/qc/fda.htm, accessed: 5 Jan 2014.
S EATON , M. J.: 1983. Quantum defect theory. Rep. Prog. Phys., 46, 167257.
S EATON , M. J.: 2002. Coulomb functions for attractive and repulsive potentials and for positive
and negative energies. Comput. Phys. Commun., 146, 225249.
U NIVERSITY OF C OLORADO: 2000. Davids wizzy periodic table, Physics 2000. http://www.
colorado.edu/physics/2000/applets/a2.html, accessed: 7 Jan 2014.
W EBER , K. H. and C. J. S ANSONETTI: 1987. Accurate energies of ns, np, nd, nf, and ng levels
of neutral cesium. Phys. Rev. A, 35, 46504660.
W IKIPEDIA CONTRIBUTORS: 2014. Periodic table, Wikipedia, The Free Encyclopedia. http://en.
wikipedia.org/wiki/Periodic_table, accessed: 7 Jan 2014.
Nonstationary Problems: Dipole Excitation
with One Photon 4
Overview
The present chapter concentrates on electric dipole transitions (E1), while
Sect. 5.4 will also treat electric quadrupole (E2) and magnetic dipole (M1)
transitions. After some basics and terminology on electromagnetic radiation,
polarization, and photon spin (Sect. 4.1), the essentials of spectroscopy are in
troduced in Sect. 4.2, the E INSTEIN A and B coefficients are defined, and the
classical model of a radiating oscillator is reviewed. The advanced reader may
jump over this section and ignore also Sect. 4.3.14.3.5, where the elements
of time dependent perturbation theory are summarized. However, Sect. 4.3.6
with terminology and some key results as well Sect. 4.4 with essentials on
selection rules for dipole (E1) transitions are needed in the following sections
and should be read carefully. The same holds for Sect. 4.5 where the angular
characteristics of dipole radiation are presented. Section 4.6 may be used by
the expert reader just as a source of reference with details on the evaluation
of matrix elements and E INSTEIN coefficients. In Sect. 4.7 photoinduced lin
ear combinations of states are discussed a theme of broad relevance. In this
context we also introduce quantum beats and indicate some spectroscopic per
spectives. Finally, we ask the very fundamental, almost philosophical question
whether electrons may really jump from one stationary state into another
and present experiments illuminating this profoundly.
Before discussing radiation induced transitions we define the terminology for de
scribing electromagnetic waves used here and in most of the following text. For
brevity, we shall speak about electromagnetic waves, electromagnetic radiation
and light more or less synonymously, although the latter is often used more specif
ically for the visible part of the spectrum.
For the present purpose it is sufficient to consider only plane, monochromatic waves
extending over all space, and to concentrate on the electric field component of the
wave which is responsible for E1 transitions.1
The electric field vector E(r, t) is an observable in the real world depending on
position vector r and time t. We thus write it as2
i
E(r, t) = E0 eei(krt) e ei(krt) (4.1)
2
with the (real) field amplitude E0 , the unit polarization vector e, and the wave vector
k with k = 2/ = /c.
Even though it is sometimes convenient to use a complex representation (and take
its real part as observable after all calculations are done), we emphasize strongly that
it is important to write E(r, t) as a real quantity in order to be able to describe all
observed physical phenomena as done here neither of the two summands in (4.1)
can be ignored as we shall see shortly!
From classical electrodynamics we know that the field amplitude E0 is related to
the (time averaged) intensity I of the electromagnetic wave by
E0 = 2I /(0 c) = 2I Z0 = 27.45 I "1/2 (4.2)
with the electric constant 0 , the speed of light c and Z0 the characteristic vacuum
impedance. For practical purposes we communicate a handy numerical expression:
I
E0 = 2745 V m1 . (4.3)
W cm2
1 We find this approach (leading to correct results for E1 transitions) conceptually more accessible
than the general, rigorous treatment of all transition types based on the vector potential. The latter
is outlined in Appendix H, while in Chaps. 1 and 2, Vol. 2, we shall generalize (4.1) and learn how
to treat spatial distributions and quasimonochromasy of real light beams adequately.
2 Theoverall phase angle 0 does not play a role in the present discussion and will generally be
ignored. However, we shall have to come back to it in Chaps. 1 and 2, Vol. 2.
4.1 Electromagnetic Waves: Electric Field, Intensity, Polarization 171
In reality, one usually has to deal with light of a certain bandwidth, i.e. with an
intensity distribution I() (intensity per unit angular frequency). Then one has to
replace I I()d and to integrate expressions derived for transition probabilities
over the whole available spectral range.
In this basis, the polarization of light which propagates in the +zdirection (e0 ) may
again be expressed by only two basis vectors (e+1 , e1 ). One easily verifies again
the orthonormality
eq eq = qq with eq = (1)q eq . (4.6)
For atomic problems the helicity basis is often better adapted, since atoms too are
usually described in a spherical coordinate system. For reference, we express the
Cartesian basis in terms of the helicity basis:
1 i
ex = (e+1 e1 ) and ey = (e+1 + e1 ). (4.7)
2 2
For later use we also give the unit polarization vectors for linearly polarized light in
45 and 135 direction in respect of the xaxis:
1 1
e 45 = (ex + ey ) and e 135 = (ex ey ). (4.8)
2 2
172 4 Nonstationary Problems: Dipole Excitation with One Photon
The unit vector e = e+1 describes left hand circularly polarized light (LHC), also
called + light. Inserting (4.5) into (4.1) gives
1
E +1 (r, t) = E0 sin(kr t)ex + cos(kr t)ey , (4.12)
2
and e1 stands for right hand circularly polarized light (RHC) or light:
1
E 1 (r, t) = E0 sin(kr t)ex cos(kr t)ey . (4.13)
2
An illustration of + light gives Fig. 4.1. Shown is the E vector at a fixed time
t = /2 along the zaxis. As indicated E rotates at a fixed position clockwise
around the wave vector k of the light, i.e. with positive helicity. One verifies this
most directly in (4.12) with kr = 0, or correspondingly in Fig. 4.1, if one considers
the time propagating.3
The most general unit polarization vector for light propagating into an arbitrary
direction can be written in either Cartesian or spherical basis:
1
1
e = ax e x + ay e y + az e z = aq e q with aq 2 = 1. (4.14)
q=1 q=1
In the spherical basis it may be specialized for any elliptically polarized light prop
agating parallel to the +z axis (i.e. into e0 direction) by
3 Methods for generating and detecting polarized light will be discussed in Chap. 1, Vol. 2.
4.1 Electromagnetic Waves: Electric Field, Intensity, Polarization 173
z
+ light (LHC)
k
+t
+t 0
k y
+t
Fig. 4.1 Schematic illustration of left hand circularly polarized light + (LHC). The grey arrows
indicate the direction of the electric field perpendicular to the z k axis at a fixed time, correspond
ing to t = /2. For larger times the sense of the rotation is indicated. The somewhat surprising
definition for left hand circular polarized light is of historical origin: in times prior to the laser,
people used to look into the oncoming light beams, hence they considered + light as rotating
anticlockwise, i.e. LHC polarized
with the component a0 = 0 in this coordinate system). The ellipticity angle de
scribes the degree of ellipticity, the alignment4 angle gives the direction of the
ellipse in respect of ex . Converting (4.15) into Cartesian coordinates with (4.5) and
inserting into (4.1) gives an explicit, parameterized expression for the elliptic field
vector in terms of and :
E el (r, t) = (E0 / 2)
'
cos sin(kr t ) + sin sin(kr t + ) ex (4.16)
(
+ cos cos(kr t ) sin cos(kr t + ) ey .
This describes the standard form of an ellipse on which the E vector rotates if (for
fixed r) the phase kr t + varies between 0 and 2 . As sketched in Fig. 4.2,
this ellipse is inclined by an angle in respect of the xaxis, and for 0 /2
its major and minor axes a and b, respectively, are:
a = E0 (cos + sin )/ 2 = E0 sin( + /4)
(4.17)
b = E0 (cos sin )/ 2 = E0 cos( + /4).
4 Unfortunately the terms alignment and orientation are mixed up time and again in the liter
ature: Alignment refers to the direction of a polar vector (e.g. the E vector in the case of linearly
polarized light). In contrast, orientation specifies the sense of rotation of an axial vector (e.g. the
angular momentum of left and right circularly polarized light).
174 4 Nonstationary Problems: Dipole Excitation with One Photon
Note that a 2 + b2 = E02 . In the literature instead of the ellipticity angle often a
(somewhat ambiguously defined) so called ellipticity is used:
The main part of this chapter will be concerned with absorption and emission of
electromagnetic radiation from atoms. In general, one has to distinguish two dif
ferent coordinate systems: one in which the atom is best described we call it the
atomic frame (at) while the photon may possibly be better described in another
frame we call it the photon frame (ph).
These coordinate systems may differ from each other. In the spirit of Sect. 4.1.2
a convenient choice for the photon frame is illustrated in Fig. 4.3. We shall use this
henceforth unless otherwise stated. The atomic frame may e.g. be defined by an ex
ternal electric or magnetic field in respect of which the projection quantum numbers
m are defined. The photon frame is defined as shown in Fig. 4.3 by the direction of
light propagation (wave vector k), with the axes z(ph) k and y (ph) z(at) .
(at)
The unit vectors eq in the atomic frame (helicity basis) indicated in Fig. 4.3
by heavy red arrows may in a classical picture be seen to represent three different
classical dipole oscillators. This will be discussed in Sect. 4.5.1.
4.1 Electromagnetic Waves: Electric Field, Intensity, Polarization 175
m = 1 k
x (at)
At this point we want to extend our knowledge about the properties of photons sum
marized in Sect. 1.4. In addition to momentum and energy the photon also possesses
an intrinsic angular momentum, the so called photon spin with a quantum number
sph = 1. Projected onto the wave vector k LHC and RHC polarized light have angu
lar momenta ms with ms = 1 and 1, respectively. Owing to the transverse nature
of electromagnetic waves the third component with ms = 0 does not exist.
In a particularly beautiful and fundamental experiment this has been observed for
the first time by B ETH (1936). The setup and the results are shown schematically
in Fig. 4.4. LHC photons are prepared by passing linearly polarized light through
a /4 plate. These photons then pass through a /2 plate where they are converted
into RHC. This implies an angular momentum change of 2 per photon which is
transferred to the /2 plate in order to conserve the total angular momentum of the
system. For more efficiency the light beam is retroreflected behind another /4 and
/2 plate
RHC RHC 0
/4 plate
x y xypolarization plane 5
of linear polarization
linearly polarized light in = 135 90 45 0 45
Fig. 4.4 Experiment of B ETH (1936) on mechanical detection and measurement of angular mo
mentum of light
176 4 Nonstationary Problems: Dipole Excitation with One Photon
Section summary
We describe electromagnetic radiation by a monochromatic plane wave. As
a physical observable it is represented by a real valued function (4.1). Both
exponential terms will turn out to be relevant when describing the interaction
of the electromagnetic field with quantum systems.
The helicity basis (4.5) is convenient for describing orthonormal (4.6) unit ba
sis vectors for problems in atomic physics. The most general unit polarization
vector for light propagating into +zdirection is given by
at physically reasonable boundary conditions (in the following we shall often ab
breviate all quantum numbers by one letter, e.g. n j = j, a or b). Energy zero is
defined such that for bound states Wj < 0 while the free electron has a continuous
spectrum of energies Wkin = W 0. As discussed in the previous two chapters, the
asymptotic behaviour of the radial wave function of the bound states is given by
lim Rj (r) exp 2Wj r
r
while the continuum states are described essentially by C OULOMB wave functions.
4.2 Introduction to Absorption and Emission 177
W2  2
WI
W1  1
Here ba = 2ba is the angular frequency of the radiation, ba = c/ba its vacuum
wavelength and ba = 1/ba = ba /c its wavenumber. Figure 4.6 illustrates three
basic types of setups for spectroscopic investigations. Specialties and refinements
will be discussed in detail later, but essentially all spectroscopic measurements fol
low in principle one of these three concepts:
hot gas,
plasma
emission spectroscopy
cold gas
white light absorption spectroscopy
lamp
fluorescence spectroscopy,
las
Fig. 4.6 3 basic methods of spectroscopy very schematic. Left: experimental setups, middle: type
of observed spectra, right: term schemes and transitions
Absorption
As derived in Sect. 1.3.2 the L AMBERTB EER absorption law (1.44)
dI
= Na I I (z) = I0 e Na z = I0 ez (4.21)
dz
describes the exponential reduction of the intensity I (z) of electromagnetic radia
tion when passing through an absorbing medium of particle density Na with [Na ] =
number of absorbing particles m3 . Here Na is assumed to be constant over dis
tance z and time t.
We now want to understand the process on a microscopic level and derive quan
titative expressions for the absorption cross section ([ ] = m2 ) and the absorp
tion coefficient = Na ([] = m1 ). Since the energy of a photon is we can
4.2 Introduction to Absorption and Emission 179
n (n1)
a Na
translate the intensity of the radiation field I ([I ] = W m2 ) into its photon density
Nph ([Nph ] = number of photons m3 ) or energy density u = Nph ([u] = J m3 )
by
I = cNph = cu. (4.22)
As schematically indicated in Fig. 4.7, on an atomistic level the absorption process
corresponds to a loss of photons from the field, i.e. a reduction of Nph by dNph
which is identical to a loss dNa of particle density Na in the initial state a
and gain
dNb of density Nb in the final excited state b
.
In contrast to the situation described by (4.21), we are now interested in the
change of N with time t at fixed position z for a constant intensity I . Since a change
over a distance dz corresponds to a change over a time dt = dz/c we may rewrite
(4.21) with (4.22) as
1 dNa I
R ba = = ba = ba (4.24)
Na dt
for the transition into the state b
from state a
. We have introduced here also the
photon flux = I /.
Note that the rate R ba has the dimension T1 ; it is the probability for absorption
of photons by an atom per unit of time. To obtain the total number of events per unit
time one has to multiply it by the number of atoms in the volume under observation.
In the above discussion we have tacitly assumed that the radiation to be absorbed
is strictly monochromatic and tuned into resonance with the absorption line. In real
ity, however, the intensity has a spectral distribution, which we indicate by I() with
[I()] = W m2 s1 . Thus, only that fraction of intensity can be absorbed which is
in resonance with the angular frequency ba of the transition. Usually one refers the
transition probability to the spectral radiation density u(), i.e. the energy density
5 Forclarity, throughout this chapter we shall use R ba , B ba etc. for rates and coefficients averaged
over initial and summed over final substates, since in general we have to account for degenerate
energy levels. In contrast, Rab (e) refers to transitions between specific substates b and a with
polarization e.
180 4 Nonstationary Problems: Dipole Excitation with One Photon
Table 4.1 Definitions in the context of absorption and emission of electromagnetic radiation
Symbol Eq. Term Unit Remarks
absorption cross section m2 effective absorbing area
N particle density m3
=N absorption coefficient m1 I = I0 exp(x)
= amplification coefficient m1 if < 0
I = c Nph light intensity W m2
Nph = I /(c ) photon density m3
I = E02 /(2Z0 ) intensity E0 field amplitude
Z0 = 1/(c0 ) impendence in free space 376.73 also vacuum impedance
Nph = r 0 2
2 E0 energy density J m3 radiation field
Nph
u() =
spectral radiation density J m3 Hz1
0 E02
= energy
Vol
= I
c
= 2
= I ()
2c W s2 m3
u() = 2u()
bandwidth (frequency) Hz
. . . (angular frequency) s1
du per spectral range d, which is related to the spectral intensity distribution I()
by
I() du d du u()
u() = = = = , (4.25)
c d d d 2
with the spectral radiation density given alternatively per unit angular frequency
= 2 or per unit frequency of the electromagnetic radiation. (We shall use the
former unless mentioned otherwise.)
Relevant for the transition probability is the intensity at the transition frequency.
Consequently, we now rewrite (4.24) as:
1 dNa I(ba )
R ba = = B ba = B ba u(
ba ). (4.26)
Na dt c
The thus defined constant B ba is called E INSTEIN coefficient for absorption with
[Bba ] = m3 s2 J1 . Note that rates, transition probabilities and matrix elements
from state a
to state b
are typically indexed from right to left.
Table 4.1 summarizes the terms and definitions most frequently used for the
quantitative description of radiation induced dipole transitions. Most of the discus
sion to follow in this chapter will focus on the E INSTEIN coefficients. They contain
the essence of the transitions in quantum systems. Polarization and frequency de
pendence play an important role.
Stimulated Emission
Up to now we have assumed that only one dominantly populated ground state level
contributes to the absorption process. And the absorption as such was thought to be
weak enough so that it does not change the population significantly. We now con
4.2 Introduction to Absorption and Emission 181
n (n+1)
a Na
sider the inverse process, the so called stimulated emission which we have schemat
ically illustrated in Fig. 4.8. For this to happen, the excited state must be populated.
For simplicity we assume for the moment that all atoms are initially found in the
excited state b
as indicated in Fig. 4.8. In complete analogy to (4.24) and (4.26)
the corresponding emission rate is
1 dNb I(ba )
R ab = = B ab = B ab u(
ba ). (4.27)
Nb dt c
We thus find that more photons may come out of the system than going into it if
the excited state has a higher population than the ground state so that this process
dominates. B ab is called E INSTEIN coefficient for stimulated emission.
In general, both processes absorption and stimulated emission happen in a
quantum system. According to (1.54) the ratio of the population densities are given
by Nb /Na = (gb /ga ) exp[(Wb Wa )/(kB T )] in thermodynamic equilibrium. At
room temperature kB T 25 meV is small, while electronic excitation in atoms typ
ically requires some eV; hence Wb Wa kB T and Nb Na . In contrast, if the
energy gaps are much smaller, as it is the case e.g. for molecular vibrational and ro
tational states, one has to consider induced emission also in standard spectroscopic
absorption experiments.
A completely different situation arises when skillful schemes are applied to
achieve a significant population of some specific excited state b
such that Nb > Na
(for some state a
, not necessarily the ground state). Then, stimulated emission can
be stronger than absorption and laser action becomes possible. We shall discuss this
in some detail in Chap. 1, Vol. 2.
From experimental experience we know that excited atoms may also decay sponta
neously. Quantum electrodynamics (QED) explains this by the interaction of quan
tum systems with the vacuum field: In a fully quantized description the electromag
netic field is represented by harmonic oscillators, their energy being (Nj + 1/2)j .
Here Nj is a (typically very large) integer number proportional to the intensity of
the radiation at the frequency j , and j /2 is the so called zeropoint energy. The
latter represents the eigenenergy of the ground states of the field oscillators at zero
intensity of the radiation. This vacuum field is essentially a consequence of the
uncertainty relation. One may say, somewhat figuratively, that it is this (isotropic)
182 4 Nonstationary Problems: Dipole Excitation with One Photon
zero point oscillation of the quantized radiation field, which forces an excited quan
tum system to decay spontaneously. We shall present a quantitative description of
spontaneous emission based on this concept in Chap. 2, Vol. 2.
Here we consider a heuristic, classical approach in which the atomic electron is
treated as an accelerated charge with a dipole moment D(t) = er(t). This clas
sical picture does not allow to extract quantitative results. It leads, however, to an
intuitive understanding. According to classical electrodynamics such a dipole emits
electromagnetic waves. In the nonrelativistic limit the electric field at large distances
R from the source is given by
E(R, t) =
1 k
k
t = er (t ) ,
D (4.28)
2
40 c R k k 40 c R 2
with t = t R/c being the retarded time, and r representing the components of
the dipole acceleration perpendicular to the wave vector k. We identify this with
projecting r onto the x (ph) y (ph) plane of the photon frame introduced in Fig. 4.3.
For a harmonic dipole oscillator r(t) = r (at) exp(iba t) with amplitude r (at) in the
atomic frame and angular frequency ba the field amplitudes are given by
2
eba (ph)
Ex (R, t) = r (at) ex ei(kRba t) (4.29)
40 c2 R
2
eba (ph)
Ey (R, t) = r (at) ey ei(kRba t) . (4.30)
40 c2 R
In Sect. 4.5.1 these expressions will be used for a semiclassical derivation of angular
and polarization characteristic of atomic dipole radiation. Here we are just interested
in the overall power emitted. From (4.28) we obtain with (4.2) the intensity I =
0 cE(R, t)2 , and on average6 an energy
dW
D(t) 2
d = I R 2 d = sin2 k dk (4.31)
dt (4)2 0 c3
is emitted per unit time into a solid angle dk = sin k dk dk . Here k is the angle
between D and k. Integration of sin2 k over all solid angles gives a factor 8/3 so
that the total radiation power emitted is
2
dW 1 e2 dp
P= =  2=
D (4.32)
dt 60 c 3 60 me c dt
2 3
dW 1 e2 r (at) /22 ba
4
P= = . (4.33)
dt 6 0 c3
Due to this radiation the excited state of the atom decays. In a classical picture
the oscillation amplitude (i.e. the orbital radius of the electron) continuously de
creases contrary to the spectroscopic observation of discrete, stationary states:
the atom (or any other quantum system) is either in the excited or in the ground
state.
Thus, quantum mechanically (4.33) is interpreted as a probability statement:
The probability that one excited atom emits a photon of energy during the time
interval dt is
dW 1 e2 r (at) /22 ba
3
R ab dt = A ab dt
(spont)
dt = = (4.34)
3 0 c 3
with 1/137 being the fine structure constant (1.10). The exponential decay law
ensuing form (4.34) has already been discussed in Sect. 1.3.1. For the excited state
density Nb we obtain according to (1.39) and (1.40) in the absence of other pro
cesses
1 dNb 1
= A ab = Nb (t) = Nb0 eAab t = Nb0 et/ab
Nb dt ab
with the average excited state lifetime ab = 1/A ab , the half lifetime 1/2 =
(ln 2)ab , and Nb0 the excited state density at time t = 0.
We note at this point, that (4.35) is almost identical to the exact expression
for the so called E INSTEIN Acoefficient for spontaneous emission, which will be
discussed in Sect. 4.6.2. The specific properties of the emitting atom, here repre
sented by r (at) /2, require of course a stringent quantum mechanical interpreta
tion. For a prominent case, the 3p 3s transition in sodium (the so called Na
D line at = 589 nm), we try an intelligent guess by setting r (at) 0.190 nm
(the atomic radius, see Fig. 3.4). From (4.35) we obtain A 3s3p 3.2 107 s1 or
= 1/A ab 31 ns which is at least on the same order of magnitude as the cor
rect, experimentally determined value 16.2 ns. Of course, we do not expect exact
results from this classical model.
184 4 Nonstationary Problems: Dipole Excitation with One Photon
a Na
initial final initial final initial final
I I I
ab ab
characteristics
of spectra ab
Fig. 4.9 Absorption, induced and spontaneous emission schematically. In the top row the atom
istic view point is indicated, below typical spectroscopic patterns are illustrated: the detected in
tensity as a function of angular frequency of the radiation
Figure 4.9 summarizes the above considerations and findings. In a real experiment
one has to consider all three processes simultaneously for describing an atomic sys
tem in the presence of an electromagnetic field correctly. A special case is a system
of atoms, molecules or condensed matter oscillators in thermodynamic equilibrium
with its own radiation field.
E INSTEIN has taken this thermodynamic equilibrium as a starting point for a very
elegant and simple derivation of P LANCKs radiation law Fig. 1.15. It is based on
the so called principle of detailed balance: under stationary conditions each process
must be in equilibrium with its inverse. One treats the problem as a representative
two level system sketched in Fig. 4.10. The kinetics of this system is described by so
called rate equations. The population of excited and ground state may change due
to the three processes as follows:
dNb dNa
= A ab Nb Nb B ab u(
ba ) + Na B ba u(
ba ) = . (4.36)
dt dt
Stationarity means that the particle densities do not change:
dNb,a
= 0 = A ab Nb Nb B ab u(
ba ) + Na B ba u(
ba )
dt
A ab Nb A ab
=1 = u(
ba ).
(Na B ba Nb B ab )u(
ba ) B ba (Na /Nb ) B ab
4.2 Introduction to Absorption and Emission 185
A ab A ab
ba ) =
u( = .
Bba (Na /Nb ) Bab Bba (ga /gb )eba /(kB T ) B ab
3 1
u() = /(k
,
2 3
c e BT ) 1
3
A ab = R ab = 2 3 B ab and gb B ab = ga B ba .
(spont)
(4.38)
c
Note the 3 factor interestingly, it agrees with the classical prediction (4.35).
We point out that the B coefficients used here refer to spectral radiation densities
Section summary
We have introduced three prototype methods for obtaining information from
quantum systems by interaction with an electromagnetic field: emission, ab
sorption and fluorescence spectroscopy.
Induced processes (absorption and induced emission) are introduced phe
nomenological. The rates for both processes are proportional to the spectral
ba ) = I(ba )/c at the transition frequency studied.
radiation intensity u(
A classical interpretation of spontaneous emission encounters severe prob
lems which have led to B OHRs second theorem and to the probabilistic
interpretation of the emission process. Nevertheless, it provides an informa
tive first guess on the spontaneous emission probability and useful information
(4.32) for later use.
E INSTEINs has derived P LANCKs radiation law from rate equations for a
system of quantum oscillators in thermal equilibrium with the radiation field.
This leads immediately to quantitative relations between the E INSTEIN coef
ficients for spontaneous and induced emission.
The readers should memorize the famous factor 3 (or 3 ) between them:
A ab 3 B ab = 3 (ga /gb )Bba it will play an important role on several oc
casions.
7 In
the literature often u() = 2 u()
per unit frequency is used. Then one has to replace B ba
B ba
()
= B ba /2 in all relevant equations.
186 4 Nonstationary Problems: Dipole Excitation with One Photon
where cj (t) is the time dependent probability amplitude of state j (r) and the prob
ability to find a certain final state j
at time t is given by
2
wj (t) = cj (t) . (4.42)
If the summation (4.41) is carried out over sufficiently many states and the basis j
is complete, it represents a solution as close to exact as desired.
Using this ansatz and the time dependent Hamiltonian (4.40), the time dependent
S CHRDINGER equation (2.13) is written as
 (t)
H0 + U p , r, t) (t) = i
( (4.43)
t
[cj (t)eij t ]j
cj (t)eij t H0 + U
(p , r, t) j
= i .
t
j j
Inserting H0 j
according to (4.39) on the left hand side of this equation, and using
the rules of product differentiation on the right hand side leads to
(
cj (t)eij t Wj + U p , r, t) j
j
dcj (t)
ij t ij t
=i cj (t)(ij )e + e j
.
dt
j
With i(ij ) = Wj the first terms in the sums on either side cancel. One multiplies
from the left with beib t and applies bj
= bj . With Wb = b and the time
dependent perturbation matrix elements
Uij (t) = iU ( (
p , r, t)j
= i (r)U p , r, t)j (r)d3 r (4.44)
Note that this is a general and still exact formulation for atoms or molecules ex
posed to a time dependent interaction. In particular, it is not restricted to the dipole
approximation nor even to monochromatic plane waves.
In practice one can only sum over a finite number of terms. A perturbation approach
(
is possible if U 0
, i.e. if the averaged perturbation is small com
p , r, t)
 H
pared to the averaged Hamiltonian of the unperturbed system (H atom, alkali atom
188 4 Nonstationary Problems: Dipole Excitation with One Photon
etc.). Then the initial conditions will change only slightly with time. To obtain the
probability amplitude cb (t) for the final state b in 1st order time dependent per
turbation theory one starts in 0th order by setting on the right hand side of (4.45)
ca(0) (t) 1 (initial state a) and cj(0) (t) 0 for all j = a. From this one obtains for
b = a in 1st order:
dcb (t) i
= Uba (t)eiba t . (4.46)
dt
If the interaction is switched on at time t = 0, integration over time leads to the time
dependent transition amplitude in 1st order perturbation theory:
i t
cb (t) = Uba t eiba t dt . (4.47)
0
One may iterate the procedure by inserting (4.47) into (4.45) to obtain the 2nd order
solution and so on.
Often one is interested only in how the probability amplitude cb (t) evolves after
many oscillation of the field, i.e. for t 1/ba . In the limit t the 1st order
probability amplitude (4.47)
i
)ba (ba )
cb () = Uba t ei(ba )t dt = U (4.49)
0
is apart from numerical prefactors nothing but the F OURIER transform (I.2)
of the perturbation potential at the frequency ba of the transition. In conclusion
atomic or molecular transitions can only be excited by an interaction potential
which contains F OURIER components U )ba () which are in resonance with the tran
sition frequency ba = .
8A somewhat heuristic derivation of eE0 and p, r) for electric dipole transitions will be pre
D(
sented below in Sect. 4.3.4, while a more rigorous, general derivation and specialization is found
in Appendix H.
4.3 Time Dependent Perturbation Theory 189
i it
(
Uba (t) = bU p , r, t)a
= b eE0 De D eit a
2 (4.50)
i
= eE0 Dba eit
Dba eit with
Dba = bDa
.
2
The transition amplitude (4.47) now becomes
eE0 t
Dba ei(ba )t
Dba ei(ba +)t dt
cb (t) = (4.51)
2 0
eE0 Dba ei(ba )t D ei(ba +)t
= ba . (4.52)
2 i(ba ) i(ba + )
We note here that both terms, exp(it) and exp(+it), are relevant. And one
sees already in (4.51) that significant contributions are expected only for stationary
phases, i.e. for ba = 0, respectively. Otherwise the contributions to the integral
oscillate rapidly and disappear in the limit t . With the angular frequency of
the radiation > 0 (per definition), and a referring to the initial, b to the final state,
the first term in (4.52) describes absorption, the second induced emission:
absorption b a: ba > 0
D exp(it) term relevant
(4.53)
emission a b: ba < 0
D exp(+it) term relevant.
We now apply this general formalism to the so called electric dipole transitions
(E1 transitions) and present here a slightly heuristic introduction of the interaction
potential. It leads to the same result as the more stringent (but also less evident)
derivation given in Appendix H.1.6. In both cases the key simplification exploits the
fact that the wavelengths of interest are typically large compared to atomic dimen
sions ( a0 ). Hence, the electromagnetic field E(r, t) described by (4.1) may be
expanded in powers of r/ or k r 1. To 1st order exp(ik r) reduces to 1, and
only the temporal change of the electric field E(t) needs to be considered.
The force on an electron from this oscillating field is eE(t), and for the
electronnucleus pair the
i
U (r, t) = D E(t) = er E(t) = E0 er eeit e eit . (4.55)
2
190 4 Nonstationary Problems: Dipole Excitation with One Photon
It depends on the electron position r and on time t. The field amplitude E0 is related
to the radiation intensity I via (4.2), and e describes the polarization of the field.
Hence, by comparing (4.55) with (4.50) one finds that the dipole transition operator
for absorption is given by
D = r e = D e/e and eE0 = e 2I /(c0 ), (4.56)
In the following we evaluate explicitly the probability amplitude for absorption, i.e.
we focus for the moment only on the first term in (4.52), with ba > 0 (Wa < Wb ).
If the electromagnetic field is switched on at t = 0, we have cb (t) 0 for t < 0
while for t > 0 the transition amplitude evolves as
eE0 ei(ba )t 1
cb (t) =
Dba . (4.58)
2 i(ba )
4.3 Time Dependent Perturbation Theory 191
Fig. 4.11
g( ) / (t/2 ) 1
g() = sin2 [(ba )t/2]/(ba )2
as a function of ba in
units of 2/t . For t it
becomes proportional to the
delta function
3 2 1 0 1 2 3
( ab  ) / (2 /t)
Figure 4.11 illustrates the characteristic frequency dependence g(). For large times
t 1/(ba ) it becomes arbitrarily narrow, and at the same time arbitrarily high
(t/2 at = ba ). This line profile is normalized such that g()d = 1, and
is given by
2 sin2 ( ba2 t) t
g() = (ba ). (4.60)
t (ba )2
Thus, g() becomes a representation of the D IRAC delta function in the limit
t .
(abs)
According to (4.59) wba (t) grows linearly with time. The approximation is of
course only valid for not too large fields and not too large times so that the general
assumption of perturbation time cb (t)2 1 is fulfilled for all b = a. At the same
time we are interested in the stationary state, i.e. in times t 1/(ba ).9
(abs) (abs)
We now divide wba (t) by time t to obtain the transition rate Rba = wba (t)/t,
i.e. the transition probability per unit time which in this approximation obviously
becomes independent of time. This is indeed the quantity one determines in an ab
sorption experiment.
Finally, we have to account for the fact that in practice strictly monochromatic
electromagnetic waves do not exist. Each radiation has a certain bandwidth
=
2
. We introduce this frequency dependence by replacing the intensity I in a
spectral range from to + d by I()d, where I() = cu() is the spectral
9 We shall see in Chap. 5 that these somewhat contradictory requirements can already be overcome
by accounting for a finite lifetime of excited states, and a full treatment will be given in Chap. 10,
Vol. 2 in the framework of the optical B LOCH equations.
192 4 Nonstationary Problems: Dipole Excitation with One Photon
intensity distribution. Thus, in (4.59) the square of the perturbation amplitude e2 E02
according to (4.56) has to be replaced by
2
e2 E02 e2 I ()d. (4.61)
c0
(abs)
The absorption rate dRba = wba (t)d/t for radiation of angular frequency be
tween and + d is then given by
e2 I()
dRba = I ()
Dba 2 g()d = 4 2 
Dba 2 g()d, (4.62)
0 c 2
which is identical to the excitation probability of state b
from a
per unit time,
with the fine structure constant 1/137 according to (1.10).
So far we have tacitly assumed that the absorption occurs between two isolated
states a
and b
with sharp energies Wb and Wa , i.e. at a well defined sharp line of
angular frequency ba = (Wb Wa )/. The total transition rate Rba is then obtained
by integrating (4.62) over all frequencies:
I() 4 2
Rba = dRba = 4 2 
Dba 2 g()d = Dba 2 I(ba ). (4.63)

shows that Rba is indeed a rate, i.e. gives the probability for an absorption process
per atom per unit of time. In the present perturbative approach we have assumed
that Rba tobs 1 during the relevant observation time tobs .
We see now explicitly that the excitation probability is proportional to the
F OURIER component of the spectral radiation density u() = I()/c at the tran
sition frequency ba . The integration (4.63) is of course based on the assumption
that I() is constant over the absorption line. For typical classical radiation sources
this is trivially a correct assumption. In laser spectroscopy, however, this is not nec
essarily true as we shall explain in Sect. 5.2.3 and more profoundly in Chap. 10,
Vol. 2.
For the integration in (4.63) we have assumed that the electromagnetic wave
interacts for a sufficiently long time with the system, so that the limit 1/ba t
in (4.60) is well approximated. With typical periods of the light field on the order
of femtoseconds (fs) this approximation may even be applied to excitation by ns
pulses often used in laser spectroscopy. But (4.62) is still valid for much shorter
interaction times: according to (4.60) this does, however, imply a broadening of the
line. We shall come back to this aspect in Sect. 6.1.7.
4.3 Time Dependent Perturbation Theory 193
To derive (4.63) for the absorption process we have evaluated the first exponent
in (4.51). The second becomes only relevant if ba < 0, i.e. for induced emission
(Wa > Wb ). This leads in complete analogy to the rate Rab  Dab 2 = r ab e 2 for
induced emission, with identical prefactors. Because of hermicity r ab = r ab holds
and thus Rba = Rab . The rates for induced emission and absorption for a specific
transition between the states b
a
are identical.
In contrast to the general discussion in Sects. 4.2.34.2.5 we refer here to indi
vidual substates described by a set of quantum numbers e.g. j m and not to energy
levels which may consist of several such (degenerate) sublevels. The key quantity
which characterizes an electric dipole transition is the dipole transition operator, ac
cording to (4.56) Dab = r ba e, i.e. essentially the scalar product between dipole
matrix element D ba = er ba and polarization vector e with the dipole transition
amplitude
r ba = bra
= b (r)ra (r)d3 r = arb
= r ab . (4.64)
Note that r ba is a vector, specific for each system and each transition,10 and the
transition probabilities also depend critically on the polarization e of the radiation.
Before evaluating this in detail, we briefly summarize the results for electric dipole
transitions obtained so far.
We rewrite the probability (4.63) for inducing E1 transitions per unit time be
tween the substates a
= ja ma
and b
= jb mb
with polarization e:
4 2
Rba = I (ba )r ba e2 = Rab (4.65)
4 2 c
= r ba e2 u(
ba ) = B(jb mb ; ja ma ; e)u(
ba ).
In full analogy to (4.26) and (4.27) we have introduced the substate and polarization
specific E INSTEIN B coefficients:
4 2 c
B(jb mb ; ja ma ; e) = r ba e2 = B(ja ma ; jb mb ; e). (4.66)
Again, these B coefficients are defined in respect of a spectral intensity distribution
I(ba ) = cu(
ba ) per unit angular frequency at the transition frequency ba =
Wb Wa /.
10 More precisely, er
ba is called dipole length matrix element. According to (H.25) the correspond
ing dipole velocity matrix element is
pa
= iba me ebra
.
eb
Both formulations lead to identical results if exact wave functions are used. For approximate solu
tions (i.e. quite generally) significant differences may occur.
194 4 Nonstationary Problems: Dipole Excitation with One Photon
(spont) 3 d
dRab = B(ja ma ; jb mb ; e) = Cr ba e2 d
2 c3 8
(4.67)
2 3
ba e2 ba3
= C r ab e d with C = = .
2c2 8 2 0 c3
A discussion of the corresponding E INSTEIN A coefficients will be postponed un
til we have derived the angular distributions of the radiation patterns in Sect. 4.5.
A clean derivation of (4.67) will be given in Chap. 2, Vol. 2.
In summary, absorption and emission of electromagnetic radiation due to E1
transitions is characterized by
According to (4.57) the matrix elements for transitions ja ma
jb mb
are
Dba =
r ba e and Dab = r ab e = Dba for absorption and emission, respectively. They
determine
whether the transition may occur at all (so called selection rules),
its dependence on polarization and angular distribution,
and its overall strength.
In the following sections these aspects will be discussed in detail, in particular the
key quantity r ba e in (4.68) will be evaluated for different polarizations and ge
ometries. As we shall see, this may be somewhat involved since in general, r ba and
e are described most conveniently in different coordinate frames, e.g. in the atomic
(at) and photon (ph) frame, respectively, as introduced in Sect. 4.1.3.
However, we do not want to end this section without treating at least one par
ticularly simple example. We choose excitation (and induced emission) by linearly
polarized light and exploit the fact that r ba = (xba , yba , zba ) is a well defined vec
tor which may be represented by its Cartesian components. Let (r) be the angle
4.3 Time Dependent Perturbation Theory 195
I(ba )
Rba = Rab = 4 2 r ba 2 cos2 (r) (4.69)
with r ba 2 = (xba
2 + y 2 + z2 ). This expression may be somewhat misleading in as
ba ba
far as depending on the alignment of the polarization vector not all components of
r ba contribute equally; e.g. for the simplest geometries where the linear polarization
is chosen parallel to any of the axes (4.69) becomes
I( ) xba 2 for e x
ba
Rba = Rab = 4
2
yba 2 for e y (4.70)
zba 2 for e z.
Section summary
The semiclassical approach describes radiation induced processes by a classi
cal, time dependent electromagnetic field and treats the interacting atomic or
molecular system quantum mechanically.
Time dependent perturbation theory is usually a good approximation for low
enough radiation intensities. Time dependent transition amplitudes (4.47) are
computed from the matrix elements of the perturbation operator U (t).
In the limit t , the transition amplitude (4.49) between initial (a) and
final state (b) is essentially the F OURIER component of Uba (t) at the transition
frequency ba .
The dipole approximation describes so called E1 transitions and is usually an
excellent approximation for a quantitative description of transition probabili
ties as long as the wavelength of the radiation is a0 . It neglects the change
of the field with position over atomic distances. In the dipole length formu
lation the interaction operator is E0 D where E0 is the field amplitude and
Dba = r ba e, with e being the unit polarization vector of the radiation.
Both exponential terms of the real field amplitude (4.1) matter: the
D exp(it) term is responsible for absorption, the
D exp(it) term for in
duced emission. While the semiclassical treatment gives quantitatively cor
rect results for absorption and induced emission, spontaneous decay requires,
strictly speaking, the use of QED to be discussed in Chap. 2, Vol. 2.
196 4 Nonstationary Problems: Dipole Excitation with One Photon
As we have learned in Sect. 4.1.4 the photon has a spin sph = 1. But only two
projections ms q = 1 in respect of the direction of propagation, sph q
= 1 +1
and 1 1
are realized, describing LHC and RHC polarized light, respectively. One
speaks of positive and negative helicity. The fact that the third component q = 0 is
missing, reflects the transverse nature of free electromagnetic waves. According to
relativistic quantum field theory this is a consequence of the photon being a massless
particle.
Quite downtoearth, the angular momentum properties of the photon allow us
to derive selection rules for E1 transitions without any computational efforts: we
simply apply the classical conservation law for angular momenta. For the system
atom + photon the total angular momentum and its projection onto a particular
direction in space is conserved during absorption or emission of a photon. Thus,
transitions may only be observed if the total angular momentum of the whole system
J is conserved. Denoting the total angular momentum operators of the atom with Ja
and J b in the lower and the upper state, respectively, the following rules in vectorial
form hold:
for absorption
J =Ja +
S ph Jb
(4.71)
for emission
J = J b J a + S ph .
absorption
(bottom, Dab = r ab e ) of
a
circularly polarized light ma ma
(left: LHC + ) and y (at) y (at)
(right: RHC ). The k  z k  z
selection rules
m = 1 are
summarized in (4.74). Here x (at) in x (at) in
the atomic at and photon (ph) +(LHC) light (RHC) light
coordinate frames are
identical, with the photon k mb k mb
wave vector k z(at) b
emission
a
out ma = m b  1 out ma=mb +1
y (at) y (at)
x (at) x (at)
j = 0, 1 with ja , jb 0, but 0 0
abbreviated: (ja jb 1) = 1, (4.72)
while (ja jb 1) = 0 indicates that the triangular relation does not hold.
Clearly, the sum of the projections of the angular momenta before and after the
process have to be conserved as well. Hence, for dipole transitions the corresponding
selection rule is
q = m = mb ma = 0, 1. (4.73)
Schematically this is illustrated in Fig. 4.13 for absorption and emission of pure
circularly polarized + (LHC) and light (RHC), propagating in +z(at) direction.
Figure 4.14 shows the situation for linearly polarized light.
Note that due to the transverse nature of electromagnetic waves the selection rule
m = 0 refers to a coordinate system perpendicular to that in the two other cases:
for linearly polarized light it is convenient to choose the photon frame (ph) in respect
198 4 Nonstationary Problems: Dipole Excitation with One Photon
mb
z (at) b
a
ma= mb
y (at)
emission
x (at) k
e 0  z (at)
of the atomic coordinate frame (at) with the electric field vector E z(ph) z(at) and
the light travelling in the x (at) y (at) plane, as illustrated in Fig. 4.14.11
The selection rules just outlined are necessary but not sufficient conditions for E1
transitions. To obtain quantitative expressions for the transition probabilities sum
marized in (4.68) one has to evaluate  Dba 2 = r ba e2 explicitly. To do so, a spe
cific coordinate system has to be chosen. For the unit polarization vector e we have
discussed suitable basis sets in Sect. 4.1.2. In principle, one may use Cartesian (xyz)
or polar coordinates (r ). Depending on the geometry of the experiment one or the
other choice may be advantageous. For linear polarization we have shown by way
of example that a Cartesian basis with e parallel to one of the axes leads directly
to the particularly simple result (4.70). However, since the angular part of atomic
wave functions is usually presented in polar coordinates, for a general description
the corresponding helicity basis (introduced in Sect. 4.1.2) is more appropriate for
expressing the polarization vector.
We introduce spherical components rq of the position vector r. With x =
r sin cos , y = r sin sin and Table B.1 one easily verifies
1
1
r = r = r = xex + yey + zez =
rq eq =r C1q (, )eq (4.76)
q=1 q=1
1
1
1
r C1q (, )eq = r (1)q C1q (, )eq = (1)q rq eq = r .
q=1 q=1 q=1
1
r ba e = bra
e = b jb mb ra ja ma
e = mb rq ma
eq e
q=1
1
= b ra
jb mb C1q ja ma
eq e. (4.79)
q=1
12 More details for evaluating these matrix elements are made available in Appendix C.
200 4 Nonstationary Problems: Dipole Excitation with One Photon
Note that emission differs from absorption as described by (4.57). The polarization
vector e (creation of a photon) replaces e (annihilation of a photon):
1
r ab e = rab
ja ma C1q jb mb
eq e = (r ba e)
q=1
1
= rab (1)q ja ma C1q jb mb
eq e . (4.80)
q=1
We now discuss the simplest case, assuming that the light polarization is given by a
basis vector in the atomic frame, e := eq . All transition rates (4.68) are now propor
tional to 
Dba 2 = r ba eq 2 . With (4.79) this implies
2 2
b ra
jb mb C1q ja ma
.
(abs) (ind) (spont)
Rba = Rab dRab (4.81)
The selection rules for any E1 transition are essentially contained in the angular
part of the matrix elements. With (4.77) the angular part of the transition probability
(4.81) is
2
jb 1 ja
jb mb C1q ja ma
2 = (2jb + 1) jb C1 ja
2 . (4.82)
mb q ma
4.4 Selection Rules for Dipole Transitions 201
Whether a dipole transition is possible at all between two levels with total angular
momenta ja and jb depends on reduced matrix element jb C1 ja
. It does not
vanish if and only if the triangular relation (4.72) holds, as already gleaned from
angular momentum conservation.
Whether transitions can occur between specific sublevels depends on the 3j sym
bol in (4.82). It can only be finite if the sum of its projection quantum numbers in
the second row vanishes (see Appendix B.2). Hence, (4.73) is confirmed as addi
tional selection rule. The underlying conservation of total angular momentum has
been illustrated in Fig. 4.13.
Note: We point out again the differences between absorption and emission as a
consequence of replacing e in (4.79) by e to obtain (4.80) with r = r (real valued
vector). E.g., as spelled out in (4.74), + light absorption increases the projection
quantum number in the nal state by q = 1: mb = ma + 1; in contrast, emission
of + decreases of m in the nal state: ma = mb 1. But in both cases the se
lection rule (4.73) holds per denition. This is a consequence of Dab = r ab e =
(r ba e) =
Dba : the transition probabilities depend only on the absolute squares of
the matrix elements. A corresponding consideration holds for light with q = 1.
For linearly polarized light, according to (4.74) and Fig. 4.14, the unit polar
ization vector becomes e = e0 so that with (4.81) and (4.82) the transition rate is
simply proportional to
2
jb 1 ja
for E z(at) mb = ma . (4.83)
mb 0 ma
Thus, with such linearly polarized light one observes transitions with
m = 0, often
called transitions. In contrast to (4.74) the E vector of the photon and not its wave
vector k points into the z(at) direction. The propagation angle within the x (at) y (at)
plane does not play a role in this case.
For reference we also note that
1
r ba 2 = mb rq ma
2 = xba 2 + yba 2 + zba 2 . (4.84)
q=1
The explicit evaluation of the matrix elements in (4.81) may become quite com
plicated, depending on the angular momentum coupling schemes in which the states
ja ma
and jb mb
are described. In later chapters we shall have to discuss this in
more detail for various examples.
For the moment we just evaluate the most simple case for single electron transitions
between uncoupled, pure orbital angular momentum states, na a ma
and nb b mb
,
ignoring the electron spin. Inserting (C.28) into (4.81) one obtains
202 4 Nonstationary Problems: Dipole Excitation with One Photon
Wn / eV
n =0 =1 =2 =3 =4
0
8
6
5s 5p 5d 5f 5g
0.85 4 4s 4p 4d 4f
1.51 3
3s 3p 3d
3.40 2
2s 2p
13.6 1 1s
Fig. 4.15 G ROTRIAN diagram for the H atom illustrating the selection rule
= 1 in E1 tran
sitions (shown are only emission lines, the electron spin is neglected)
2 2
Rba na rnb
a ma C1q  b mb
2
= na rnb
(2 a + 1)(2 b + 1)
2
2
a 1 b a 1 b
ma mb +q ( a 1 b ) . (4.85)
ma q mb 0 0 0
The first 3j symbol confirms again the selection rules (4.72) and (4.73). The second
3j symbol, however, adds an additional selection rule: since its projection quantum
numbers all vanish, a + 1 + b must be even, lest it disappears according to (B.49).
Thus with the constraint ( a 1 b ) = 1 a new selection rule holds:
a = b 1. (4.86)
The physical origin of this rule is parity conservation which is discussed in more
detail in Appendix D. Briefly, the overall parity of the system has to be conserved,
i.e. its symmetry in respect of inversion at the origin has to be the same before and
after absorption or emission of the photon. The angular part of the wave functions
for such pure states is completely described by spherical harmonics, their parity
being (1) . The photon which is represented by C1 has negative parity. The parity
of the whole system before absorption of the photon is thus (1) a +1 while it is
(1) b after the photon has been absorbed. Thus, for parity conservation a + 1 + b
must indeed be even.
Up to now, we have used pure states for the hydrogen and alkali atoms. As an
example, Fig. 4.15 shows the term scheme of the H atom with some E1 transitions.
Specifically, for an s p transition three processes are allowed: 0ms
1mp
with
ms = 0and mp = 0 or 1. According to (C.44) all three angular matrix elements
are 1/ 3, independent of the substates, so that
ns rnp
2
for an = 0 = 1 transition: Rab . (4.87)
3
4.5 Angular Dependence of Dipole Radiation 203
The electron spin, so fare neglected, will modify details of the term scheme
Fig. 4.15. The corresponding selection rules will be discussed in Chap. 6. In Ap
pendix C one finds compact formulas for the evaluation of (4.82) in the general case.
In Chap. 7 we shall see, however, that transitions between states with angular mo
mentum quantum numbers 0 and 1 are characteristic for closed shell multielectron
systems. They occur e.g. among so called singlet states where the spins of several
electrons compensate each other. E1 transitions may then indeed be described by
the orbital angular momenta of the optically active electron, e.g. p s. In the
remainder of the present chapter we shall identify jb = 1 ja = 0 processes as
1 P 1 S transitions within a singlet system, indicated by the superscript, while
1 0
the subscript stands for j .
To actually calculate the transition probabilities, in addition to the angular part
of the matrix elements, which we have so far discussed in detail, one also has to
compute the radial transition matrix elements. For the H atom Appendix C.5 sum
marizes the necessary formulas. A rather comprehensive collection of spontaneous
transition probabilities for many atoms of the periodic system is found at K RAMIDA
et al. (2013).
Section summary
Selection rules for light induced E1 transitions may directly be inferred from
angular momentum conservation.
For the total angular moments the triangular relation (ja jb 1) = 1 must hold,
the photon having a spin 1.
For the projection quantum numbers q =
m = mb ma = 0, 1 reflects the
photon polarization q = 1 (with respect th z(at) k) and q = 0 (with respect
th z(at) E).
Quantitatively, the transition probabilities are proportional to
2
jb 1 ja
2
rab (2jb + 1) jb C1 ja
2 .
mb q ma
We now turn to the angular characteristics for absorption and emission of electro
magnetic radiation in dipole transitions. We make explicit use of the coordinate
frames introduced in Sect. 4.1.3: the atomic frame (at) for describing the atom and
the photon frame (ph) in which the light is best described. The latter is character
ized by the polar (k ) and azimuthal angle (k ) of the wave vector k as shown in
Fig. 4.3. Section 4.5.1 presents a first somewhat heuristical semiclassical ap
proach by interpreting the classical oscillator as dipole transition matrix element
(at)
r ab . In Sect. 4.5.2 we shall show that this may be cast into an exact quantum me
chanical formulation with essentially identical results, now including all necessary
numerical prefactors.
204 4 Nonstationary Problems: Dipole Excitation with One Photon
The three components directly reflect the three types of classical oscillators illus
trated graphically in Fig. 4.3. They correspond to the three types of transitions which
we have already summarized in (4.74):
For q = 0 ( light) the electron oscillates linearly in z(at) direction with an am
plitude ma r0 mb
. From the classical picture we expect zero radiation in z(at)
direction and maximum intensity emitted into the x (at) y (at) plane with linear po
larization z(at) .
The terms with q = +1 and q = 1 represent an electron on a circular orbit in the
x (at) y (at) plane. The emitted light may propagate into the whole solid angle 4 .
Maximum intensity and fully circularly polarized light is intuitively expected
for propagation into +z(at) direction, while light detected in the x (at) y (at) plane
will be linearly polarized z(at) axis.
More quantitatively, one may glean electric field amplitudes from (4.29) and (4.30)
(at)
by projecting r ab onto the x (ph) y (ph) plane. For light emitted from a linear oscil
(at)
lator with amplitude ma rq mb
e0 one reads from Fig. 4.3
(ph) (ph)
Ey = 0 and Ex ma r0 mb
sin k (4.89)
z (at) k
k= 90 k< 90
E,
k
x (at) y (at)
E, k
Fig. 4.16 Angular distribution of dipole radiation from a linear oscillator ( light)
I (k , ) sin2 k . For two different directions of the wave vector k the alignment of the polar
ization vector is indicated (red, full and dashed vector arrows, respectively)
z (at) k
E, +
k
y (at )
k
x (at) E,
k
Fig. 4.17 Angular distribution of dipole radiation from a circular oscillator ( light)
I (k , ) (1 + cos2 k )/2. The polarization vector is indicated for two different directions of
observation: at 0 < k < /2 (full red ellipse with vector arrow), corresponding to elliptically
polarized light, and for k = /2 (dashed red arrow) corresponding to linear polarization
In Fig. 4.18 the standard geometry for observation of the different transition types
is shown, as e.g. realized for observing the so called normal Z EEMAN effect (see
Sect. 2.7). The three cases q = 0, 1 may in principle be observed independently
since the corresponding atomic lines are split in the magnetic field, with frequencies
0 and 1 = 0 L /2 for transitions with
m = q = 0 and 1, respectively.
206 4 Nonstationary Problems: Dipole Excitation with One Photon
(a) (b)  +
(c)

z (at ) +
1 0 1
+1 spectrum
1P
1 0
mb = 1
1 0 1
Fig. 4.18 Normal Z EEMAN effect as realized in a 1 S0 1 P1 transition; (a) energy levels in an
external magnetic field B ex z(at) , (b) experimental setup, (c) spectrum from light emitted at
k = 0, (d) spectrum from and light emitted at k = /2
We emphasize again that the level scheme Fig. 4.18(a) should not be misinter
preted: the arrows indicating transitions with 1 and 0 refer to different coordinate
systems. The external magnetic field defines here the direction of z(at) B ex ; emis
sion parallel to that axis (i.e. k z(at) B ex ) implies transitions with
m = q = 1
( light) only; in contrast,
m = q = 0 transitions imply light, linearly po
larized with the electric field E B ex , emitted into any direction except in z(at)
direction.
What has been deduced above from a more or less classical picture also follows from
a rigorous quantum mechanical treatment. The angular and polarization character
istic of emission and absorption rates are obtained by inserting (4.80) and (4.79),
respectively, into the general expression (4.68). Typically, the matrix element r ba
and the polarization vector e are described in different coordinate frames, most con
veniently in the atomic (at) and photon frame (ph) as illustrated in Fig. 4.3. The
(spont) (ind) (abs)
transition rates dRab , Rab , and Rba between two angular momentum basis
states jb mb
ja ma
are then proportional to
1 2
(at) (ph) 2 (at) (ph) 2
r e = r e = mb rq ma
eq(at) e(ph)
ba ab
q=1
1 2
(ph)
= b ra
jb mb C1q ja ma
eq e .
(at)
(4.93)
q=1
4.5 Angular Dependence of Dipole Radiation 207
We emphasize that the unit vector e(ph) represents the polarization of the
Note that irrespective of the direction of the above arrows (4.93) holds as writ
ten (see also the note on p. 201). To obtain practical relations for the angular dis
(at)
tributions, such as (4.90) and (4.92), we have to express eq and e(ph) in the same
coordinate system. Which of these frames is more convenient depends on the exper
imental situation. For illustration we carry this out explicitly as a kind of exercise
on coordinate rotation, using the rules explained in Appendix E.
m = q = 0 ( component)
m = q = 1 ( + component)
k (ph) k (ph)
e1(at) =e ik
cos2 e1 + sin2 e1 . (4.96)
2 2
m = q = 1 ( component)
k (ph) k (ph)
e1 = eik sin2 e1 + cos2 e1 .
(at)
(4.97)
2 2
208 4 Nonstationary Problems: Dipole Excitation with One Photon
Note that in the photon frame only the q = 1 basis vectors are considered, with
polarization perpendicular to the photon wave vector k. Only these are of physical
(at) (ph)
relevance in the scalar products (4.93) eq e(ph) . The e0 component is dropped
since it would indicate (unphysical) linear polarization in z(ph) direction. Hence,
(at)
eq is no longer a unit vector and its magnitude depends on k . We may, however,
express it in terms of an angle dependent normalization factor fq (k , k ) and the
(ph)
general unit vector (4.15) for elliptically polarized light in the photon frame eel :
(ph)
e(at)
q = fq (k , k )eel , so that (4.98)
(at)
(ph)
r ab e(ph) = ma rq mb
fq (k , k )eel e(ph) . (4.99)
q
Assuming we can detect one specific transition q =
m only, and register all light
(ph)
emitted from this transition without further discrimination, i.e. e(ph) := eel , the
emission rate becomes proportional to
(at) (ph) 2
r e = ma rq mb
2 fq (k , k )2 . (4.100)
ab el
One easily verifies that this probability has indeed the dimension T1 . The intensity
of radiation emitted per atom seen at a detector of an area
A in a distance R is
obtained by multiplying (4.101) with the photon energy ba , dividing it by
A,
and identifying the solid angle with d =
A/R 2 :
4
(spont) ba mb r0 ma
2 sin2 k .
I = 2 2
(4.102)
2c R
This angular distribution corresponds exactly to the previously derived semi
classical pattern (4.90) shown in Fig. 4.16. The angular part jb mb C1q ja ma
of
the matrix element mb r0 ma
according to (4.77) is typically on the order of 1,
while the radial matrix element b ra
may vary over a wide range depending on
the size of the atomic orbitals and the overlap and symmetry of the wave functions.
We now turn to the circular components of the emitted radiation pattern
(q = 1). The atomic basis vectors projected onto the photon frame are
e1 = f1 eel = f1 (k , k ) ei cos e+1 ei sin e1 .
(at) (ph)
(4.103)
(ph)
The parameters and of the unit elliptic basis vector eel (4.15) are found from
a comparison of (4.103) with (4.96) and (4.97). We see immediately that = /2
4.5 Angular Dependence of Dipole Radiation 209
(the main axis of the polarization ellipse is y (ph) ); with cos2 +sin2 = 1 and with
the trigonometric identity 2(cos4 (k /2) + sin4 (k /2)) = (1 + cos2 k ) we derive
(1 + cos2 k )1/2
f1 (k , k ) = ieik (4.104)
2
For emission at k = 0, i.e. along the z(at) axis we have cos = 1 or = 0 and
sin = 0 or = 1 for q = +1 or 1, respectively. As discussed already, this implies
emission of pure circularly polarized +(LHC) or (RHC) light. In contrast,
at k = /2 one finds cos = sin = 1/ 2, corresponding to light emission with
linear polarization vector in the x (at) y (at) plane (see Fig. 4.3).
The angular distributions for
m = q = +1 and
m = q = 1 follow naturally
by inserting f1 (k , k ) from (4.104) into (4.100) and this in turn into (4.67). Again,
(ph)
the overall intensity of the emitted light is found by setting e(ph) := eel . We finally
obtain for q = 1 transitions
3
mb r1 ma
2 1 + cos k d,
ba 2
(spont)
dR = (4.106)
2c2 2
and the intensity per atom at a distance R is
4
(spont) ba mb r1 ma
2 1 + cos k .
2
I = (4.107)
2c2 R 2 2
This confirms again the angular characteristics (4.92) of the radiation found by the
semiclassical model as illustrated Fig. 4.17.
We may integrate the radiation characteristic (4.101) and (4.106) over the full
solid angle and thus derive the overall spontaneous emission probability for one
specific transition. Both, the sin2 k distribution (4.103) for the components as
well as the (1 + cos2 k )/2 distribution (4.107) for the transitions lead to the same
factor 8/3:
8 1
sin k d = 2
2
sin k dk =
3
= 1 + cos2 k d. (4.108)
4 0 3 4 2
Hence, from (4.101) as well as from (4.106) the integral spontaneous emission prob
ability for a specific transition with ma + q = mb becomes:
13 The overall phase factor i exp(ik ) is here of no significance for measurable quantities, since
these are proportional to the absolute squares of the matrix elements.
210 4 Nonstationary Problems: Dipole Excitation with One Photon
3
(spont) 4 ba mb rq ma
2
Rq = A(ja ma ; jb mb ; q) = 2
3 c
3
4 ba b ra
2 jb mb C1q ja ma
2 .
2
(4.109)
3 c
Equations (4.95)(4.109) describe spontaneous emission processes between well
defined states ja ma
jb mb
. This situation is realized experimentally if a spe
cific initial state is prepared and only one specific final state is detected. Then the
angular distributions (4.102) and (4.107) are observed. One method (by far not the
only one) to prepare and detect such specific transitions is the Z EEMAN effect as
illustrated in Fig. 4.18.
We finally mention that a more general theory of radiation emitted from excited
atoms will be presented in Sect. 9.4, Vol. 2 as an interesting example for applications
of the density matrix.
I(ba ) 2 sin2 (k )
Rba = 4 2 mb r0 ma
.
2
4. Finally, if the same transition can also be induced by linearly polarized light
(ph)
x (ph) , i.e. if e(ph) := ex . We find
I(ba ) 2
Rba = 4 2 ma r0 mb
sin2 (k ),
which obviously is a more efficient process. The pattern is the same as shown in
Fig. 4.16.
Section summary
We have derived the angular distributions of radiation emitted for q = 0 (q =
mb ma ) transitions and for q = 1 from a semiclassical picture as well as
from quantum mechanical rates for E1 transitions.
We should memorize these distributions according to Figs. 4.16 and 4.17.
The polarization depends on the emission angle k with respect to the z(at)
axis. For q = 0 transitions, only linearly polarized light ( light) is emitted,
with its polarization vector parallel to z(at) for k = /2. For q = 1 transi
tions, at k = 0 pure circularly polarized light ( light) is emitted, while at
k = /2 one finds linearly polarized light with polarization vector perpen
dicular to z(at) . For other emission angles the light is elliptically polarized.
The three components of radiation may be separated experimentally, e.g. by
exploiting the Z EEMAN effect as illustrated for a 1 S0 1 P1 transition in
Fig. 4.18.
For induced transitions one may derive the angular dependence of rates for
transitions induced between specific sublevels by a specific polarization. The
general case is more complex however, as we shall see in a moment, it
may be greatly simplified for an initially isotropic population of all sub
states.
212 4 Nonstationary Problems: Dipole Excitation with One Photon
According to (4.68) the key parameter for all absorption and emission processes is
r ba e2 . It contains the necessary information on polarization and angular charac
teristics for E1 transitions between specific substates. In contrast, to characterize the
overall strength of a dipole transition between levels ja and jb with several degen
erate ma and mb substates one introduces (see also Appendix H.2) a symmetrically
defined line strength with the dimension L2 :
S(jb ja ) = r ba 2
mb ma
= b jb mb rq a ja ma
2
mb ma q
2
= (2jb + 1)b ra
jb C1 ja
2 (4.112)
2
= (2ja + 1)b ra
ja C1 jb
2 S(ja jb ).
Here we have used (4.84) with (4.77), (4.82) and the 3j orthogonality relation
(B.41). The radial matrix element b ra
is given by (4.78). We recall that b and
a represent all quantum numbers needed to characterize the radial wave function.
We also note that ja and jb may refer to various kinds of the angular momentum
quantum numbers, representative for the states under discussion in the simplest
case these may be orbital angular momenta or L, but could also be total angular
momenta as we shall discuss in Chaps. 6 and 9.
The line strength is a useful reference quantity. However, when describing spe
cific cases one has to distinguish between types of radiation (polarized or unpolar
ized, unidirectional or averaged over all solid angles) as well as between types of
transitions (selected substates, sums over final states, averages over initial states).
This will be discussed below in terms of E INSTEIN coefficients. Figure 4.19 defines
these for transitions between the substates jb mb
and ja ma
with polarizations
characterized by q = mb ma .
jb mb
B( ja ma ; jb mb ;q)
A( ja ma ; jb mb ;q)
B( jb mb ; ja ma ;q)
ja ma
Fig. 4.19 E INSTEIN coefficients for absorption, B(jb mb ; ja ma ; q), for induced emission,
B(ja ma ; jb mb ; q), and for spontaneous emission, A(ja ma ; jb mb ; q), between substates of an up
per and a lower level characterized by the angular momentum quantum numbers jb and ja , respec
tively
4.6 Strength of Dipole Transitions 213
The spontaneous transition probability (4.109) for one specific transition ja ma
jb mb
and polarization q is now rewritten with (4.82), (4.77) and (4.112)
A(ja ma ; jb mb ; q)
3
2
4ba 2 jb 1 ja
=
b ra
(2jb + 1) jb C1 ja
2
3c2 mb q ma
3
4ba 3
2
= mb rq ma
2 = 4ba jb 1 ja
S(jb ja ). (4.113)
3c2 3c2 mb q ma
It has the dimension T1 . The 3j symbol accounts for all dependencies on angular
momenta and polarization while all specific properties of the atom (or molecule) are
included in S(jb ja ).
Total decay probability and spontaneous lifetime of the upper level are deter
mined by summing over all final states and polarizations:
1
A ab = = A(ja ma ; jb mb ; q)
ja jb ma q
jb
2
3 3
(4.114)
4ba 1 ja 4ba S(jb ja )
= S(jb ja ) = .
3c2 mb q ma 3c2 (2jb + 1)
m q a
Here we have again used the orthogonality relation (B.41) for the 3j symbols. It is
important to note that the spontaneous lifetime ja jb is independent of the quantum
number mb of the sublevel in the excited state!
Explicitly, the prefactor in the above equations is given by
3
4ba 32 3 c 19 3 s
2 7.235 108 m
= = 1.083 10 ba 2 = . (4.115)
3c2 33ba m 3ba s
For practical use we insert the line strength (4.112) into (4.114) and introduce atomic
units a0 and Eh :
4 3 2
2
A ab = 2 3 Wba a0 b r/a0 a
jb C1 ja
2
3c
4 5 me c2 2
= (Wba /Eh
Гораздо больше, чем просто документы.
Откройте для себя все, что может предложить Scribd, включая книги и аудиокниги от крупных издательств.
Отменить можно в любой момент.