Professional Documents
Culture Documents
2015 Bookmatter AtomsMoleculesAndOpticalPhysic PDF
2015 Bookmatter AtomsMoleculesAndOpticalPhysic PDF
Overview
These appendices offer a collection of concepts and relations which belong
to the standard mathematical ‘tools’ in atomic, molecular and optical physics.
On many occasions we refer to these in the main text. We do not attempt here
to derive them in a closed, axiomatic manner – rather a compact summary of
useful and important formulas and equations is presented. Usually they are
accompanied by some hints about their origins and connections to make them
plausible for the reader without requesting the substantial efforts needed to
go into the depth of derivations or proofs. Literature references are given for
further information.
Appendix A offers a tabulated overview of current values for the most important
physical constants used in atomic, molecular and optical physics – and in these
textbooks. Some remarks about systems of units and dimensional analysis are also
included.
Appendix B defines the angular momentum operators and introduces some ba-
sics about angular momentum algebra with 3j symbols (respectively C LEBSCH -
G ORDAN coefficients), 6j symbols (respectively R ACAH functions), and 9j sym-
bols.
Appendix C is concerned with the rules of evaluating and transforming matrix
elements of irreducible tensor operators. This includes the very important W IGNER -
E CKART theorem, matrix elements for electromagnetically induced transitions and
the reduction of coupling schemes as well as some useful geometrical relations.
Finally we give some directions for the evaluation of radial matrix elements.
Appendix D formulates the concept of parity and introduces – as an alternative
to the usual complex spherical harmonics – real basis states for angular momenta,
typically used in quantum chemistry. The concept is also extended to half integer
angular momenta.
Appendix E offers a recipe type of access to the important theme of coordinate
rotation for angular momenta and irreducible tensor operators.
550 Appendices
Appendix F introduces the concept of multipole expansions and gives the most
important relations derived from it. Two alternatives to define multipole tensor op-
erators are discussed: based on spatial coordinates and on combinations of angular
momentum operators.
Appendix G describes the concept of convolution and illustrates it for the most
important distribution functions.
Appendix I presents some essentials about F OURIER transformations with spe-
cial emphasis on electromagnetic fields.
Appendix H provides the formal basis for treating electromagnetically induced
transitions – are presented in the main text somewhat heuristically. It also derives
the important T HOMAS -R EICHE -K UHN sum rule.
In Appendix J, finally, we discuss the behaviour and the proper normalization of
continuum wave functions. Both are important for a correct treatment of ionization
and scattering processes.
Constants, Units and Conversions
A
Table A.1 2010 CODATA recommended values of physical constants extracted from NIST
(2010), published by M OHR et al. (2012)
Constant Value Equation
Universal constants
speed of light in vacuum c 2.99792458 × 108 m s−1 defined
magnetic constant μ0 4π × 10−7 N A−2 defined
107
electric constant ε0 8.854187817 . . . × 10−12 F m−1 1
μ0 c2
= 4π c2
characteristic vacuum Z0 376.730313461 . . . cμ0 = 1
impedance √ ε0 c
= μ0 /ε0
Newtonian constant of G 6.67384(80) × 10−11 m3 kg−1 s−2
gravitation G/c 6.70837(80) × 10−39 (GeV /c2 )−2
P LANCK constant h 6.62606957(29) × 10−34 J s
4.135667516(91) × 10−15 eV s
angular momentum quant 1.054571726(47) × 10−34 J s h/2π
c 197.3269631(49) MeV fm
√
P LANCK mass mP 2.17651(13) × 10−8 kg c/G
P LANCK length P 1.616199(97) × 10−35 m /(mP c)
P LANCK temperature TP 1.416833(85) × 1032 K mP c2 /kB
P LANCK time tP 5.39106(32) × 10−44 s P /c
Electromagnetic constants
elementary charge e 1.602176565(35) × 10−19 C
In general, throughout these text books we consequently use the SI units, meter
(m), kilogram (kg), second (s), ampere (A), kelvin (K), mole (mol) and candela
(cd) – the international standard of units.
In addition to these 7 “base units” (NIST, 2000a, Table 1), we employ
We point out here that all values of constants and physical quantities given in the
table above and elsewhere in these text books refer to the currently valid definitions
of the SI. These definitions may possibly change in 2014 and be replaced by a new
set of definitions, completely based on fundamental constants (see Table A.1) as has
been decided by the CGPM (2011) (see also W IKIPEDIA CONTRIBUTORS, 2013).
Alternatively, we often use atomic units (a.u.) which are commonly used in
atomic and molecular physics. We consider these, however, to be just simplifying
554 A Constants, Units and Conversions
abbreviations: one reformulates equations such, that for lengths, times and energies
only the combinations
occur in the final expressions – which then become otherwise free of units and
include only some dimensionless parameters, such as the fine structure constant α.
It is often helpful to leave the equations in this form which simultaneously includes
a dimensional analysis. However, for simplicity of writing we occasionally omit the
respective units a0 , t0 , and Eh (e.g. in a S CHRÖDINGER equation), if no ambiguities
can arise. The respective quantities are then measured in a0 , t0 , and Eh . For more
details on a.u. see Sect. 1.8.3.
We emphasize, however, that it is rather misleading, to talk about “setting =
e = me = 1” – as occasionally formulated in (mostly theoretical) literature. The
same holds for relativistic expressions where the confusing notion that “one sets
c = 1” is often used – even in general texts, specifically on particle physics.
The attentive reader will find a number of useful references to standard (mostly
older) and sometimes important literature, where G AUSS, electrostatic (esu) or elec-
tromagnetic (emu) units are still used (G AUSS units are equivalent to esu and emu
for electrostatic and electromagnetic units, respectively). Even in current literature,
specifically in theoretical papers, these systems are still used. Obviously, the seem-
ing advantage of writing a few equations in a simpler way than SI demands, ap-
pears so attractive to some authors that they renounce general compatibility. Thus,
Table A.2 communicates some conversion rules between the SI and the G AUSS
systems, following essentially the Appendix on units and dimensions in JACKSON
(1999).
Radian (rad) and steradian (sr) are two somewhat ‘conspicuous’ units which are fre-
quently used in AMO physics. They are, however, genuine “SI derived units (NIST,
2000a)”. A brief clarification may be useful.
Radian and steradian may be visualized as shown in Fig. A.1. The radian (rad) is
defined in two dimensions for the plane angle ϕ by the length of an arc on a circle
of radius r, divided by this radius r:
ϕ= . (A.2)
r
With the full circumference of the circle being 2πr, the full angle thus becomes
ϕ = 2π rad = 360 ◦ , so that 1 rad = 360 ◦ /(2π) = 57.296 ◦ .
A.4 Radian and Steradian 555
Table A.2 Relations and conversions between the SI and the Gaussian system
Quantity/relation SI G AUSS
√
velocity of light c= 1/μ0 ε0 c
dielectric permittivity εr ε0 εr
magnetic permeability μr μ0 μr
electric field E, displacement D, D = ε0 E +√P = εr ε0 E D = E + 4πP √ = εr E
polarization P unit E esu = E SI 4πε0 E SI = E esu / 4πε0
magnetic H and B field, B = μ0 (H + M) B = H + 4πM √ = μr H
magnetization M unit = μr μ0 H √ B SI = B emu μ0 /4π
B emu = B SI 4π/μ0
M AXWELL equations ∇D = ρ ∇D = 4πρ
∇ × E = − ∂B
∂t ∇ × E = − 1c ∂B
∂t
∇B = 0 ∇B = 0
∇×H =J + ∂D
∂t ∇ × H = 1c (J + ∂D
∂t )
L ORENTZ force F = q(E + v × B) F = q(E + × B) v
c
√ √
charge, current, etc. q esu = q SI / 4πε0 q SI = √
4πε0 q esu
unit 1C = 1As 1esu =1 √dyn cm
√ √
=
2.9979 × 109 esu 4πε0 10−9 N m
=
= 3.3356 × 10−10 C
q1 q2 q1 q2
C OULOMB potential V (r) = 4π ε0 r V (r) = r
unit electric dipole moment Cm esu cm
D EBYE 1D = 3.33564 × 10−30 C m 1D = 10−18 esu cm
Polarizability esu = α SI /(4πε )
αE SI = (4πε )α esu
αE
E 0 0 E
B OHR magneton μB = e/(2me ) e/(2me c)
(a) (b)
rad = —
m
In three dimensions the steradian (sr) is defined as the solid angle Ω which char-
acterizes any surface area of the size S on a sphere of radius r. In other words, the
solid angle Ω is given as the surface S on a sphere of radius r divided by the square
556 A Constants, Units and Conversions
of this radius:
S
Ω= . (A.3)
r2
For a cone with a full opening angle 2θ as shown in Fig. A.1 we have
θ
Ω= dΩ = 2π sin θ dθ = 2π(1 − cos θ ). (A.4)
0
With the surface of a sphere being 4πr 2 the full solid angle is 4π sr, and 1 sr cor-
responds to an angle θ = arccos(1 − 1/(2π)) = 0.57195 rad = 32.77◦ – or equiva-
lently to a cone with a full angle ϑ = 2θ = 65.54◦ .
Note that expressed in SI base units, both rad = m m−1 and sr = m2 m−2 have
the unit 1, i.e. are dimensionless. The symbols rad and sr may thus be used where
advantageous to distinguish the respective observable quantities from others . . . or
be omitted wherever the expressions are unambiguous.
For small angles (A.4) may be expanded as
lim Ω = πθ 2 + O θ 4 .
θ→0
The divergence of well collimated beams (light beams, electron or ion beams
etc.) is often given in mrad2 , i.e. in terms of the full divergence angle ϑ = 2θ mea-
sured in mrad. In the case of axially symmetry, i.e. for paraxial rays in a conical
beam, the solid angle of divergence is thus
2
π ϑ
Ω = 10−6 sr.
4 mrad
Thus, a (full) divergence angle of 1 mrad corresponds to a solid angle of Ω =
(π/4)10−6 sr. Note, however, that for a rectangular beam with a small diver-
gence ϑ 2 , measured in mrad2 , the solid angle is just Ω = 10−6 (ϑ/ mrad)2 sr.
2. In dimensional analysis we try to avoid specific units for mass, length, and so on,
since physical laws must be independent of the specific choice of units. Rather
we use a basic set of dimensions: mass (M), length (L), time (T), charge (Q),
and temperature () – all written in block letters without serifs. For most of our
purposes all other dimensions of physical quantities can be expressed as a prod-
uct of powers of these basic dimensions, e.g. the dimension of force is MLT−2 ,
of energy ML2 T−2 , of electric current QT−1 , or with the thermal energy being
Wtherm = kB T , the B OLTZMANN constant kB has the dimension ML2 T−2 −1
and so on.
3. However, for convenience or for clarity we shall sometimes use more letters or
abbreviations for the dimensions of physical quantities, e.g. for volume V = L3 ,
for energy Enrg = ML2 T−2 or R = L−3/2 for the dimension of a radial wave func-
tion.
4. The units of a physical quantity A or of a dimension A are represented by square
brackets, [A] and [A], respectively. For instance the unit of a potential V (r)
would be [V (r)] = eV – and more generally we have [Enrg] = eV or [Enrg] = J,
both are acceptable.
References
B OUCHENDIRA, R., P. C LADE, S. G UELLATI -K HELIFA, F. N EZ and F. B IRABEN: 2011. ‘New
determination of the fine structure constant and test of the quantum electrodynamics’. Phys.
Rev. Lett., 106, 080801.
CGPM: 2011. ‘Resolution 1 of the General Conference on Weights and Measures (2011) CGPM:
On the possible future revision of the International System of Units, the SI’, Bureau Interna-
tional des Poid et Messure (BIPM). http://www.bipm.org/en/si/new_si/, accessed: 8 Jan 2014.
558 A Constants, Units and Conversions
JACKSON, J. D.: 1999. Classical Electrodynamics. New York: John Wiley & sons, 3rd edn., 808
pages.
M OHR, P. J., B. N. TAYLOR and D. B. N EWELL: 2012. ‘CODATA recommended values of the fun-
damental physical constants: 2010’. Rev. Mod. Phys., 2013, 1527–1605. http://physics.nist.gov/
constants, accessed: 8 Jan 2014.
NIST: 2000a. ‘Reference on constants, units, and uncertainties: SI base and derived units’, NIST.
http://physics.nist.gov/cuu/Units/units.html, accessed: 8 Jan 2014.
NIST: 2000b. ‘Reference on constants, units, and uncertainties: SI prefixes’, NIST. http://physics.
nist.gov/cuu/Units/prefixes.html, accessed: 8 Jan 2014.
NIST: 2000c. ‘Reference on constants, units, and uncertainties: Units outside the SI’, NIST.
http://physics.nist.gov/cuu/Units/outside.html, accessed: 8 Jan 2014.
NIST: 2010. ‘Reference on constants, units, and uncertainties’, NIST. http://physics.nist.gov/
cuu/Constants/, accessed: 7 Jan 2014.
W IKIPEDIA CONTRIBUTORS: 2013. ‘Proposed redefinition of SI base units’, Wikipedia, The Free
Encyclopedia. http://en.wikipedia.org/wiki/Proposed_redefinition_of_SI_base_units, accessed:
8 Jan 2014.
Angular Momenta, 3j and 6j Symbols
B
We note here the most important relations for angular momenta in compact form
for reference and use throughout these books. Closely related are the 3j , 6j and 9j
symbols relevant for the coupling of angular momenta. They play an important role
in atomic and molecular physics. We follow here – with one important exception
concerning the normalization of the spherical components of angular momenta –
strictly the definitions of B RINK and S ATCHLER (1994) and E DMONDS (1996). Ex-
plicit expressions for the 3j , 6j and 9j symbols are somewhat complicated but well
documented in both references as well as in many textbooks on quantum mechan-
ics. For practical evaluation purposes one may use today readily accessible, compact
3j , 6j and 9j calculators (see e.g. S VEN G ATO R EDSUN, 2004). Also, they are of-
ten integrated in standard mathematical programmes. Thus, we refrain here from
extensive tabulations and restrict ourselves to definitions, symmetry properties, or-
thogonality relations, a few matrix elements and some important special cases (as
collected from several sources: E DMONDS, 1996; B RINK and S ATCHLER, 1994;
W EISSTEIN, 2004a,b).
[Jx , Jy ] = iJz , [Jz , Jx ] = iJy , [Jy , Jz ] = iJx and
J , Ji = 0,
or in compact form:
J ×
J = i
J and J ,
J = 0. (B.2)
Alternatively one may construct J from three spherical components J± and J0 (so
called helicity basis) which are an irreducible representation of the rotation group
of rank 1. They are derived from the real components:1
1 1
J+ = − √ (Jx + iJy ), J− = + √ (Jx − iJy ) and J0 ≡ Jz . (B.3)
2 2
The inverse relations are:
1 i
Jx = − √ (J+ − J− ), Jy = √ (J+ + J− ). (B.4)
2 2
We note that these operators are constructed in complete analogy to the spherical
components (4.75) of the position operator r in the helicity basis which we have
introduced in the context of dipole transitions in Sect. 4.4.
The spherical components J± are not Hermitian, with
J , J± = 0 and J , J0 = 0. (B.7)
J 1 and
The scalar product of two angular momenta J 2 is given by
J1 ·
J 2 = J1x J2x + J1y J2y + J1z J2z = −J1+ J2− + J10 J20 − J1− J2+ (B.8)
The operators J+ and J− play a key role in the construction of angular mo-
mentum states. Using their commutation relations (B.6)–(B.7) one may construct
angular momentum states without using spherical harmonics (2.79)–(2.83) – which
1 Very often one finds in textbooks (even in B RINK and S ATCHLER, 1994) these ladder operators
defined as J± = Jx ± iJy . These are, however, not spherical components of the
J vector operator
and lead to asymmetric commutation relations. With the present definition – which is also used
e.g. by B LUM (2012), W EISSBLUTH (1978) – J± fulfill both functions: as ladder (or stepping)
operators and spherical components of J.
B.1 Angular Momenta 561
exclusively represent orbital angular momenta with integer quantum numbers . All
relevant algebra for angular momentum states may be derived in that way. Specifi-
cally, if |J M are eigenstates of Jz = J0 with eigenvalues M, it follows from (B.6)
that J+ |J M and J− |J M are also eigenstates of J0 with eigenvalues (M + 1) and
(M − 1), respectively:
With some algebra one may show that the only nonvanishing matrix elements of the
spherical angular momentum components J0 , J+ and J− are
J M|J0 |J M = M and (B.14)
J M ± 1|J± |J M = ∓ √ a(±M) (B.15)
2
with a(M) = J (J + 1) − M(M + 1). (B.16)
One special kind of angular momenta are the orbital angular momenta introduced in
Sect. 2.5. As described there, their respective operators may be represented in terms
of spatial coordinates – more precisely, the operator of the squared orbital angular
momentum L and the components of
2
L can be expressed as differential operators
with respect to the azimuthal and polar angles in polar coordinates. Of particular
importance are
2 1 ∂ ∂ 1 ∂2
L = Lx + Ly + Lz = −
2 2 2 2
sin θ + 2 (B.18)
sin θ ∂θ ∂θ sin θ ∂ϕ 2
z = −i ∂
and L (B.19)
∂ϕ
with the spherical harmonics Ym (θ, ϕ) as joint eigenfunctions. These operators
have the eigenvalues 2 ( + 1) and m, respectively, corresponding to (B.12). Here
= 0, 1, 2, . . . and m is any integer value between − ≤ m ≤ .
Without proof we communicate here the spherical harmonics,
m 2 + 1 ( − m)! m
Ym (θ, ϕ) = (−1) P (cos θ ) exp(imϕ) (B.20)
4π ( + m)!
2π π
∗
with dϕ Ym (θ, ϕ)Y m (θ, ϕ) sin θdθ = δ δm m . (B.21)
0 0
1 d 2
P (x) =
x −1 . (B.23)
2 ! dx
As a function of cos θ the first L EGENDRE polynomials are:
P0 (cos θ ) = 1
P1 (cos θ ) = cos θ
3 1
P2 (cos θ ) = cos2 θ − (B.24)
2 2
5 3
P3 (cos θ ) = cos3 θ − cos θ
2 2
1
P4 (cos θ ) = 35 cos4 θ − 30 cos2 θ + 3 .
8
B.1 Angular Momenta 563
The definition (B.20) of the spherical harmonics Ym (θ, ϕ) implies the so called
standard phase convention according to C ONDON and S HORTLEY (1951) with the
factor (−1)m . One finds that the c.c. is
∗
Ym (θ, ϕ) = (−1)m Y−m (θ, ϕ), (B.26)
The spherical harmonics are said to have positive or negative parity depending on
whether is even or odd, respectively (see detailed discussion in Appendix D). In
compact form one obtains – well suited for computational use –
(−1)+m (2 + 1)( − m)! d+m (sin θ )2
Ym (θ, ϕ) = (sin θ )m exp(imϕ)
2 ! 4π( + m)! d(cos θ )+m
(−1) (2 + 1)( + m)! d−m (sin θ )2
= (sin θ )−m exp(−imϕ).
2 ! 4π( − m)! d(cos θ )−m
(B.28)
For simplicity of writing one often also uses renormalized spherical harmonics
4π
Cm (θ, ϕ) = Ym (θ, ϕ). (B.29)
2 + 1
B.2.1 Definitions
J1 M1 J2 M2 |J1 J2 J M. According to the general convention one chooses the phase
such that the coefficients are real:
J1 J2 J
J1 J2 J M|J1 M1 J2 M2 =
J1 M1 J2 M2 |J1 J2 J M = CM 1 M2 M
. (B.33)
B.2 Coupling of Two Angular Momenta 565
They are called C LEBSCH -G ORDAN coefficients. They are non-zero if – and only –
two conditions are fulfilled:
1. for the Ji quantum numbers δ(J1 J2 J ) = 1 must be fulfilled, with the triangular
relation defined by:
1 if |J2 − J1 |, |J2 − J1 | + 1, . . . ≤ J ≤ |J2 + J1 |
δ(J1 J2 J ) = (B.34)
0 else
must hold.
The C LEBSCH -G ORDAN coefficients disappear if the triangular relation is not sat-
isfied, i.e. if δ(J1 J2 J ) = 0, or if M
= M1 + M2 .
Alternative to the C LEBSCH -G ORDAN coefficients the W IGNER 3j symbols
J1 J2 J
M1 M2 M
are used. They are defined completely symmetric in respect of the three angular
momenta, and are related to the C LEBSCH -G ORDAN coefficients by
J1 −J2 +M J1 J2 J
J1 M1 J2 M2 |J M = (−1) (2J + 1)1/2
. (B.36)
M1 M2 −M
Note the minus sign in front of M in the last column. The 3j symbols disappear
unless
δ(J1 J2 J ) = 1 and M1 + M2 = M. (B.37)
J1 J2 J
2
δ(J1 J2 J )
and thus also = . (B.42)
M1 M2 M (2J + 1)
M1 M2
The 3j symbols can be computed either by suitable recursion formulae or from the
R ACAH formula:
J1 J2 J
(B.46)
M1 M2 M
= (−1)J1 −J2 −M δ(M1 +M2 )−M (J1 J2 J )
× (J1 + M1 )!(J1 − M1 )!(J2 + M2 )!(J2 − M2 )!(J + M)!(J − M)!
× (−1)t t! (J − J2 + t + M1 )!(J − J1 + t − M2 )!
t
−1
× (J1 + J2 − J − t)!(J1 − t − M1 )!(J2 − t + M2 )!
One has to sum over all values of t which lead to non-negative terms for the faculty
expressions in (B.47). The 3j symbols can be obtained very fast from simple com-
puting programmes, as already mentioned (e.g. S VEN G ATO R EDSUN, 2004, Java
script, also for 6j and 9j symbols).
Closed expressions for some frequently used cases are often useful. The 3j symbol
becomes specially simple if one of the angular momenta is zero:
J1 J2 0 (−1)J1 −M1 δJ1 J2 δ−M1 M2
= √ . (B.48)
M1 M2 0 2J1 + 1
Often one has to couple three angular momenta. A typical case is the hyperfine
structure (HFS) of atoms in one electron systems, where orbital angular momentum,
electron spin and nuclear spin have to be considered. In the case of two electrons,
one may have no orbital angular momentum (ns orbital), the other electron may
be in a n configuration, e.g. Hg(6sn). Here the spins of the two electrons cou-
ple to a total spin and the latter couples with the orbital angular momentum of the
n electron. In molecular physics too, a host of problems involves similar coupling
requirements, e.g. the molecular rotation may couple with the electron’s orbital an-
gular momentum and its spin. And even in optical transitions between spin-orbit
coupled atomic states the photon adds an extra angular momentum from which the
optical selection rules are derived.
Generally, in order to determine the states of three angular momenta one starts
with the product states
and M. This may be done in two alternative schemes, based on the coupling of
two angular momenta described in Appendix B.2. One may first couple J 1 and
J2
to
J 12 and then combine J 12 with
J 3 to
J . According to (B.31) from the product
states |j1 m1 |j2 m2 one obtains
B.3 RACAH Function and 6j Symbols 569
(j1 j2 )J12 M12 = |j1 m1 |j2 m2
j1 m1 j2 m2 |J12 M12 ,
m1 m2
j1 (j2 j3 )J23 ; J M = |j1 m1
(j2 j3 )J23 M23
j1 m1 J23 M23 |J M. (B.60)
m1 M23
For the 6j symbols not to disappear, the 6 angular momenta must be coupled accord-
ing to the above definitions. This requires triangular relations between the angular
momenta. Schematically the following triangular relations must be satisfied:
, , , . (B.64)
570 B Angular Momenta, 3j and 6j Symbols
The 6j symbols are invariant against exchanging any two columns or any two rows:
j1 j2 J12 j1 J12 j2 j2 J12 j1 J J23 j1
= = = , etc.
j3 J J23 j3 J23 J J J23 j3 j2 J12 j3
(B.65)
Finally, a number of sum rules hold for the 6j symbols. We note
j1 j2 k
(−1) (2k + 1)
2k
= 1, (B.66)
j2 j 1 J
k
One may develop explicit expressions for the 6j symbols from (B.59)–(B.61). One
finds according to R ACAH:
j1 j2 J12
(B.68)
j3 J J23
(−1)t (t + 1)!
= (j1 j2 J12 )(j1 J J23 )(j3 j2 J23 )(j3 J J12 )
t
f (t)
and the triangular function (abc) according to (B.47). The summation is again
over all nonvanishing terms. Again it is commendable to use one of the available
Java-applets for practical evaluations.
The 6j symbols may also be expanded as sums over products of 3j symbols. An
important relation is the contraction formula:
j1 +j2 +j3 +μ1 +μ2 +μ3 1 j2 j3
(−1)
m1 μ2 −μ3
μ1 μ2 μ3
j1 2 j3 j1 j2 3
×
−μ1 m2 μ3 μ1 −μ2 m3
1 2 3 1 2 3
= . (B.69)
m1 m2 m3 j1 j2 j3
B.4 Four Angular Momenta and 9j Symbols 571
For the special case that 1 + 2 + 3 is even and j3 = 1/2, a particularly simple
relation emerges:
j1 j2 3
= −(−1)−1 −2 −j2 −j1 (21 + 1)(22 + 1) (B.70)
−1/2 1/2 0
1 2 3 1 2 3
× .
j2 j1 1/2 0 0 0
Again, some closed expressions for special cases are occasionally useful. With s =
a + b + c and X = b(b + 1) + c(c + 1) − a(a + 1):
a b c (−1)s δ(abc)δbe δcd
=√ (B.71)
0 d e (2b + 1)(2c + 1)
a b c 2(−1)s+1 X
=√ (B.72)
1 c b 2b(2b + 1)(2b + 2)2c(2c + 1)(2c + 2)
a b c 2(−1)s
=√ (B.73)
2 c b (2b − 1)2b(2b + 1)(2b + 2)(2b + 3)
[3X(X − 1) − 4b(b + 1)c(c + 1)]
×√
(2c − 1)2c(2c + 1)(2c + 2)(2c + 3)
a+ 1
a 1
(−1)s+1 (s − 2c + 1)(2 + s)
2 2 = (B.74)
b+ 1
2 b c + 12 2 (2a + 1)(a + 1)(2b + 1)(b + 1)
a+ 1
2 a 1
2 = (−1) s+1 (s − 2c + 1)(2 + s)
. (B.75)
b b+ 2 c
1 2 (2a + 1)(a + 1)(2b + 1)(b + 1)
For later use, we also communicate here one useful formula. In the case that one of
the j ’s is zero, the 9j symbol reduces to a 6j symbol:
⎧ ⎫
⎨a b c ⎬
δcf δgh (−1)b+d+c+g a b c
d e f = . (B.78)
⎩ ⎭ [(2c + 1)(2g + 1)]1/2 e d g
g h 0
References
B LUM, K.: 2012. Density Matrix Theory and Applications. Atomic, Optical, and Plasma Physics,
vol. 64. Berlin, Heidelberg: Springer Verlag, 3rd edn., 343 pages.
B RINK, D. M. and G. R. S ATCHLER: 1994. Angular Momentum. Oxford: Oxford University Press,
3rd edn., 182 pages.
C ONDON, E. U. and G. S HORTLEY: 1951. The Theory of Atomic Spectra. Cambridge, England:
Cambridge University Press, 441 pages.
E DMONDS, A. R.: 1996. Angular Momentum in Quantum Mechanics. Princeton, NJ, USA: Prince-
ton University Press, 154 pages.
R ACAH, G.: 1942. ‘Theory of complex spectra II’. Phys. Rev., 62, 438–462.
S VEN G ATO R EDSUN: 2004. ‘3j 6j 9j -symbol java calculator’, Sven Gato Redsun. http://www.
svengato.com/, accessed: 8 Jan 2014.
References 573
W EISSBLUTH, M.: 1978. Atoms and Molecules. Student Edition. New York, London, Toronto,
Sydney, San Francisco: Academic Press, 713 pages.
W EISSTEIN, E. W.: 2004a. ‘Wigner 3j -symbol’, Wolfram Research, Inc., Champaign, IL, USA.
http://mathworld.wolfram.com/Wigner3j-Symbol.html, accessed: 8 Jan 2014.
W EISSTEIN, E. W.: 2004b. ‘Wigner 6j -symbol’, Wolfram Research, Inc., Champaign, IL, USA.
http://mathworld.wolfram.com/Wigner6j-Symbol.html, accessed: 8 Jan 2014.
Matrix Elements
C
In this appendix some tools are collected which are useful for evaluating matrix ele-
ments of irreducible tensor operators. These matrix elements are needed to compute
transition probabilities for optical excitation of atomic and molecular states as well
as for the analysis of collision induced processes.
C.1.1 Definition
1 We refer to J± = ∓ √1 (Jx ± iJy ) while R ACAH gives the relation for ∓(Jx ± iJy ). See also
2
footnote 1 in Appendix B.
Here Tkq is the component of the tensor in the new coordinate system which arises
by rotating the old, un-dashed system through the E ULER angles αβγ . It is rep-
resented in term of the components Tks of the tensor in the old system. Important
examples of tensor operators are the spherical harmonics Ym (θ ϕ) (rank ) them-
selves, or somewhat more general the eigenstates |j m of the angular momentum
operator (rank j ).
The adjoint (or Hermitian conjugate) operator Tkq †
of a tensor operator Tkq is
defined by (see Sect. 2.3.1)
†
†
∗
J M
Tkq |J M = Tkq J M
J M =
J M|Tkq JM , (C.4)
Tkq
†
= (−1)q Tk−q . (C.5)
±1 = ∓ √1 [V
V y ];
x ± iV 0 = V
V z . (C.6)
2
When evaluating matrix elements for a tensor operator, the W IGNER -E CKART theo-
rem is very useful. It says, that for any irreducible tensor operators of a given rank k,
the matrix elements between states |J M and |J M depend in identical form on
the projection quantum numbers M, M , and q:
γ J M
Tkq |γ J M
#
= J Mkq
J M γ J #Tk γ J (C.7)
√ J J k #
= (−1)J −k+M 2J + 1 γ J #Tk γ J (C.8)
−M M q
J −M
√ J k J #
= (−1)
2J + 1 γ J #Tk γ J . (C.9)
−M q M
C.1 Tensor Operators 577
J 0|Tk0 |J 0/
J 0|J k00 if J + J + k is even.
An important example is the reduced matrix element of an angular momentum.
With
J M
Jz |J M = δJ J δM M M (C.10)
√ J 1 J #
= (−1)J −M 2J + 1 J #
J J
−M 0 M
The W IGNER -E CKART theorem allows one to relate two matrix elements of dif-
ferent tensor operators, Sk and Tk of the same rank k with each other in a simple
manner. The ratio R of the reduced matrix elements of two tensor operators of the
same rank k may be computed from any pair of (nonvanishing) matrix elements of
their components:
γ J Tk γ J
γ J M |Tkq |γ J M
R J , J, T, S = = . (C.12)
γ J Sk γ J
γ J M | Skq |γ J M
Our definition of reduced matrix elements follows B RINK and S ATCHLER (1994), as-
suming, however, integer rank only, so that (−1)2k = 1. √ Note that these reduced
matrix elements are not symmetric in J and J , rather (2J + 1)
γ J T γ J =
√ k
(2J + 1)
γ J Tk γ J . R ACAH √ (1942), E DMONDS (1996) and others use a symmetric
definition:
γ J Tk γ J Racah = 2J + 1
γ J Tk γ J .
578 C Matrix Elements
·
γ J M|V J |γ J M = R J
γ J M γ J M
γ J M| J |γ J M (C.15)
M
= R
γ J M|
2
J |γ J M = RJ (J + 1)2 .
γ J M|V·J |γ J M
R= . (C.16)
J (J + 1)2
Inserting this back into (C.13) gives the very useful projection theorem
γ J M|V·J |γ J M
γ J M
Vq |γ J M = γ J M
Jq |γ J M. (C.17)
J (J + 1) 2
γ J M|V·J |γ J M
z |γ J M =
γ J M|V M. (C.18)
J (J + 1)
Here
k1 k2 q1 q2 |KQ is a C LEBSCH -G ORDAN coefficient as introduced by (B.31)
in the context of angular momentum coupling.
Specifically, with (B.36) and (B.48) the scalar product (rank K = 0) of two ten-
sor operators of the same rank k becomes simply
Rk · Uk = kq U
(−1)q R k−q . (C.20)
C.2 Products of Tensor Operators 579
This is the generalization of the standard scalar product of two vectors R and S (see
e.g. (B.8)). The result is a single scalar quantity.
(1) (2)
C · C = (−1)m Cm (θ1 , ϕ1 )C−m (θ2 , ϕ2 ). (C.21)
m=−
If the coordinate system is such that θ2 ϕ2 defines the z-axis, we have C−m (θ2 , ϕ2 ) =
δm0 , and the sum reduces to C0 (γ 0) = P (cos γ ), with γ being the angle between
the two direction in space characterized by (θ1 , ϕ1 ) and (θ2 , ϕ2 ). Rewriting
∗
(−1)m C−m = Cm
∗
P (cos γ ) = Cm (θ1 , ϕ1 )Cm (θ2 , ϕ2 ) (C.22)
m=−
= Cmp (θ1 , ϕ1 )Cmp (θ2 , ϕ2 ). (C.23)
m=0,p=±1
The second line is easily verified with the definition of real tensor operators (D.5)
applied to the (renormalized) real spherical harmonics Cmp (θ, ϕ) (D.7). One also
derives from (C.19) a useful expansion formula for the product of two spherical
harmonics at the same angle
kq
m m C m (θ, ϕ)Cm (θ, ϕ) = k0
00 Ckq (θ, ϕ), (C.24)
m m
The often used matrix elements of the spherical harmonics in an |m basis, nor-
malized as
√
(2 + 1)(2 + 1)
m
Ckq |m = C∗ m (θ, ϕ)Ckq (θ, ϕ)Cm (θ, ϕ)dΩ,
4π
(C.26)
may be evaluated by replacing the latter two functions with the expansion (C.25):
√
(2 + 1)(2 + 1)
m Ckq |m =
4π
× C∗ m (θ, ϕ)Ck q (θ, ϕ)dΩ k q
kqm k 0
k00
k q
2 + 1
=
δk δq m k q
kqm k 0
k00 .
2 + 1
kq
The last equation is simply the W IGNER -E CKART theorem for the matrix element.
Comparing the latter two equalities gives the reduced matrix element of Ck :
# k
#
Ck = (−1) (2 + 1) . (C.30)
0 0 0
Note that this expression only holds for even k. The reduced matrix element vanishes
for odd k. For the special case C2 one obtains:
√
( + 1)
C2 = − √ . (C.34)
(2 − 1)(2 + 3)
Note the minus sign!
Of special importance in respect of the selection rules for electric dipole transi-
tions is according to (4.79) the case k = 1. Because of (B.34) and (B.49)
(see also Appendix D). With (B.50) the only nonvanishing matrix elements are
⎧
# ⎨ +1
for = + 1
#C1 =
2+3
(C.36)
⎩−
(2−1) for = − 1,
Hence, in dipole transitions the selection rule for the projection quantum number
becomes m = q. The matrix elements vanish unless m = m ± 1 or m = m. A de-
tailed evaluation of (C.29), using (B.52)–(B.55) leads to:
( − m + 1)( + m + 1)
( + 1)m
C10 |m = (C.38)
(2 + 1)(2 + 3)
( + m + 1)( + m + 2)
( + 1)(m + 1)
C11 |m = (C.39)
2(2 + 1)(2 + 3)
( − m + 1)( − m + 2)
( + 1)(m − 1)
C1−1 |m = (C.40)
2(2 + 1)(2 + 3)
( − m)( + m)
( − 1)m
C10 |m = (C.41)
(2 − 1)(2 + 1)
( − 1 − m)( − m)
( − 1)(m + 1)
C11 |m = − (C.42)
2(2 − 1)(2 + 1)
582 C Matrix Elements
( − 1 + m)( + m)
( − 1)(m − 1)
C1−1 |m = − . (C.43)
2(2 − 1)(2 + 1)
For the particularly simple case of s ↔ p transitions one finds that all allowed tran-
sitions have the same amplitudes:
10|C10 |00 =
00|C10 |10 =
11|C11 |00 (C.44)
=
1 −1|C1−1 |00 =
00|C11 |1 −1 =
00|C1−1 |11 = 1/3.
For evaluating the reduced matrix elements of product tensors angular momentum
algebra provides a useful toolbox. In simple cases, with some algebra and exploiting
6j symbols (B.61), the reduced matrix elements in the coupled scheme |(LS)J M
may be recast into the uncoupled scheme |LML |SMS . We communicate here only
(1)
some results. If Rk acts only on the first angular momentum of the coupled scheme
(2)
and Uk ≡ 1, we have:
# (1)
L S J #Rk LSJ = δS S (−1)k+L +S+J (2J + 1) 2L + 1
L L k #
× L #Rk L. (C.46)
J J S
Conversely, if Rk ≡
(1)
1 and U acts only onto the second angular momentum in
the coupled scheme, we obtain:
# (2)
L S J #Uk LSJ = δL L (−1)k+L+S+J (2J + 1) 2S + 1
S S k #
× S #Uk S. (C.47)
J J L
C.3 Reduction of Matrix Elements 583
Note: due to the well defined phases of the C LEBSCH -G ORDAN coefficients the se-
quence of the angular momenta is not arbitrary. For the scalar product (rank K = 0)
of two tensor operators of rank k the reduction becomes particularly clear. We first
notice that according to (C.7)
#
j1 j2 J M
(Rk · Sk )00 |j1 j2 J M = δM M δJ J j1 j2 J #Rk · Sk j1 j2 J , (C.48)
where we have used the explicit expression (B.48) for the 3j symbol. Explicitly
(C.48) may be evaluated by inserting on the left side into the scalar product Rk · Sk
the quantum mechanical unit operator. Then each part may be reduced according
to (C.46). By applying the orthogonality relations according to R ACAH (1942) (if
one has only to average over the angular momentum part) one finds for the reduced
matrix element of the scalar product:
#
j1 j2 J #Rk · Sk j1 j2 J = 2j1 + 1) 2j2 + 1 (C.49)
j j k # #
× (−1)J +j1 +j2 1 1 j1 #Rk j1 · j2 #Sk j2 .
j2 j 2 J
Most often required√ are the matrix elements of the spherical harmonics Yk (or of the
renormalized Ck = 4π/(2k + 1)Yk ). We have described these in Appendix C.2.2
already in a basis of pure n states. For LS coupled orbital angular momenta with
orbital quantum numbers and the reduced matrix elements
Ck are given
by (C.28).3 In the coupled scheme with a total electron spin S the reduced matrix
element (C.46) is:
3 The following expressions are also valid in multi-electron systems as long as there is only one
active electron (while further electrons have zero orbital angular momentum as it is the case for
alkali atoms, or for He below the first ionization limit). Here and is identical to the total orbital
angular momenta L and L, respectively, while the total spin S ≥ 1/2 of the system can in principle
be composed from several particles.
584 C Matrix Elements
#
S J #Ck SJ = δS S (2J + 1) 2 + 1 (2 + 1) (C.51)
k k
× (−1)−S− −−J .
J J S 0 0 0
For the case S = s = 1/2 (e.g. for an effective one particle system with spin 1/2 –
i.e. electron in an atom or a single active proton in a nucleus), using (B.70) reduces
(C.51) to:
$ # # %
# #
1 # # 1 j −1/2+k
j j k
j #Ck # j = (−1) (2j + 1) × . (C.54)
2 2 1/2 −1/2 0
For the evaluation of line strengths in E1 transitions, we report here some values for
the reduced matrix element of C1 in the case of a one electron systems:
j j
# # 1/2 3/2 5/2
$ % √ √
# # 1
1 # 1/2 − 1/3
j #C1 #
1/3 0
2 # 2 :
j √
3/2 − 1/3
√
− 1/15
√
2/5
(C.56)
√ √
5/2 0 − 2/5 − 1/35
C.3 Reduction of Matrix Elements 585
total angular momentum J = L+S. In the scheme of coupled angular momenta, i.e.
when the system is described by eigenstates |LSJ MJ , the good quantum numbers
are L, S, J and MJ . From the first binomial formula one obtains:
LSJ MJ
L ·
S|LSJ MJ
1
2
LSJ MJ
J −L −
2 2
= S |LSJ MJ
2
2
and uses the matrix elements (B.14)–(B.17) for the angular momenta. Since the
q act only onto the LML component of the state vector and
operators L Sq only onto
the SMS component one sees that the matrix elements are only nonvanishing for
MJ = ML + MS = ML + MS = MJ . (C.59)
L ·
Table C.1 Matrix elements of S/2 for a 2 P state in the uncoupled basis |LML SMS , with
υ = ML + 2MS .
ML MS |1 12 |0 12 |1 − 12 |−1 12 |0 − 12 |−1 − 12
υ 2 1 0 0 −1 −2
MJ 3/2 1/2 1/2 −1/2 −1/2 −3/2
1 12 | 1/2 0 0 0 0 0
√
0 12 | 0 0 1/ 2 0 0 0
√
1 − 12 | 0 1/ 2 −1/2 0 0 0
√
−1 12 | 0 0 0 −1/2 1/ 2 0
√
0 − 12 | 0 0 0 1/ 2 0 0
−1 − 12 | 0 0 0 0 0 1/2
L ·
LML SMS | S|LML SMS
z
= −
LML SMS |L Sz |LML SMS = 2 ML MS . (C.62)
where in the last step (C.11) was used. The nonvanishing matrix elements of the
components of J may now be written with the W IGNER -E CKART theorem (C.7),
e.g.
J I F MF
Jz |J I F MF (C.64)
√ F F 1 #
= (−1)F −1+MF δMF MF 2F + 1 J I F #J J I F .
−MF MF 0
(a) F = F
J I F MF |Jz |J I F MF
√
3F +I +J MF (2F + 1)(2J + 1)(J + 1)J J J 1
= (−1) √
F (F + 1) F F I
(−1)2I +2J +1 F (F + 1) + J (J + 1) − I (I + 1)
= MF (C.66)
2 F (F + 1)
except for F = 0 where
J I F MF |Jz |J I F MF / = 0.
(b) F = F + 1
J I (F + 1)MF |Jz |J I F MF
J J 1
= (−1)3F +I +J +1
F F +1 I
2(F + MF + 1)(F − MF + 1)(2J + 1)(J + 1)J
×
(2F + 2)
(−1)4F +2I +2J +2 MF 2
= 1− (C.67)
2 F +1
(J + I − F )(F + J + I + 2)((F + 1)2 + (J − I )2 )
× .
(2F + 1)(2F + 3)
In the last step we also have inserted the explicit expression for the 6j symbols
(B.72) to obtain these useful, compact relations.
Since C1 may be seen as a tensor of rank 1, this expression contains in principle the
(2 × 1 + 1)(2 × 1 + 1) components of the product tensor. One may express these
in irreducible form by tensors of rank 0 (basically the total intensity), rank 1 (so
called orientation) and rank 2 (so called alignment). We shall come back to this in
Appendix C, Vol. 2.
For low Z atoms, LS coupling is characteristic according to the scheme
|γ LSJ M and |γ L S J M . In this case the selection rules are determined by
the matrix elements
S L J M |C1q |SLJ M. Equation (C.53) factors this matrix
element into its important components. In an electromagnetically induced transi-
tion, the total electron spin does not change, i.e. S = 0. For E1 transitions we have
k = 1 and with q = 0, ±1 one reads in (C.53) from the first 3j symbol the selection
rules for J and M: M = q and δ(J 1J ) = 1. Thus M = M ± 1 or M = M and
J = J ± 1 or J = J – however, 0 0 is strictly forbidden. Finally, the last 3j
symbol recovers the parity selection rule L = L ± 1.
a|T0 T|b(E2) ∝
b|Q2q+ |a = √
b|r 2 |a (−1)q
b|C2q |a +
b|C2−q |a
2
(C.69)
1
and ∝
b|Q2q− |a = √
b|r 2 |a (−1)q
b|C2q |a −
b|C2−q |a ,
2i
where the definitions (D.5) have been inserted. Since k = 2, we note that q = 0,
±1, or ±2. In LS coupling we use again (C.53) for evaluation. Again S = 0,
and considering the triangular rule δ(J J 2) = 1, we read from the 3j symbols that
the total angular momentum J may now change by 0, by ±1 or ±2 units of ,
now with the restriction 0 0 and 1/2 1/2. With M = q follows in this case,
that transitions are possible with M = M, as well as with M ± 1 and even with
M ±2. This leads to correspondingly structured quadrupole absorption and emission
patterns. In addition, for quadrupole transitions, parity conservation now implies
L = L ± 2.
C.4 Electromagnetically Induced Transitions 589
Selection rules for M1 transitions are derived from the matrix element (5.53)
a|eE0 L + 2
D|b(M1) ∝
b| J +
S|aB =
b| S|aB , (C.70)
where the index B indicates that the coordinate frame for orbital angular momentum
L and spin
S is chosen here with its z-axis parallel to the magnetic RF field inducing
the transition. This particular polarization dependence can be expressed again by the
W IGNER -E CKART theorem (C.9). Since we discuss a dipole transition with k = 1
we obtain:
√
L + 2
b| S|aB = γ
γ (−1)J −1+M 2J + 1 (C.71)
J J 1 # #
× J # LJ + 2 J #
SJ .
−M M q
With
γ |γ we describe the overlap integral of radial part of the wave function
between initial and final state. The 3j symbol determines the selection rule for the
magnetic quantum numbers M, with q referring in this case to the direction of the
magnetic field vector (rather than the electric field in the E1 case). In spectroscopic
experiments, typically, radio frequency (RF) or microwave radiation (MW) is used
in external, static magnetic field B st , defining the z-axis. If the radiation field vector
B points into this z-direction (i.e. if the polarization vector e and the wave vector k
lie in the xy plane), then q = 0 and transitions with M = M occur. To induce M =
±1 transition (q = ±1), the B vector of the inducing field has to be perpendicular
to static field B st .
The two reduced matrix elements in (C.71) determine the selection rules for the
quantum numbers J , L and S. They have to be evaluated (for not too high static
magnetic fields) according to √(C.46) and (C.47) in the coupled√ system. With (C.11)
we have
L LL = δL L L(L + 1) and
S SS = δS S S(S + 1). Thus we
obtain
#
L S J # LLSJ = δS S δL L (−1)k+L +S +J (2J + 1) 2L + 1
L L 1
× L(L + 1) and (C.72)
J J S
#
L S J # SLSJ = δS S δL L (−1)k+L+S+J (2J + 1) 2S + 1
S S 1
× S(S + 1). (C.73)
J J L
Hence, the total spin and the total orbital angular momentum do not change during
a M1 transition in LS coupling! In contrast, transitions with J = J ± 1 or J = J
are possible, provided 0 0. We recall that the radial wave functions for different
principal quantum numbers and otherwise identical configuration are orthogonal.
590 C Matrix Elements
For not too large values of n and one may evaluate (C.75) in closed form,
using standard analytical programs like Maple, Mupad4 or Mathematica on a PC.
We assemble in Table C.2 some characteristic results for
n |r|n in (atomic,
Z scaled) units a0 /Z.
If very high values for n and are required (RYDBERG atoms), there are suitable
alternatives. Typically, it is sufficient to compute
n , − 1|r|n, for n
= n. Based
on earlier work a closed form has been given e.g. by T OWLE et al. (1996) (again in
units of a0 /Z):
(−)n − (n + − 1)!(n + )! 1/2
n , − 1
r|n, =
4Z (n − )!(n − − 1)!
(4n n)+1 (4n n)ν (n − n )N −ν−1 N !
× (C.76)
(n + n)n +n (2 − 1 + ν)!(N − ν)!
× 2 F1 (−ν, −2 + 1; N − ν + 1; x)
(−4n n)ν (n − n )N −ν −1 N !
−
(n + n )2 (2 − 1 + ν )!(N − ν )!
× 2 F1 −ν , −2 + 1 − ν ; N − ν + 1; x .
coefficient A(jj ) corresponding to (4.114)) are found in T OWLE et al. (1996) for H
like RYDBERG atoms. Since for sufficiently high n and all atoms become hydrogen
like, these results are of general relevance.
References
B LUM, K.: 2012. Density Matrix Theory and Applications. Atomic, Optical, and Plasma Physics,
vol. 64. Berlin, Heidelberg: Springer Verlag, 3rd edn., 343 pages.
B RINK, D. M. and G. R. S ATCHLER: 1994. Angular Momentum. Oxford: Oxford University Press,
3rd edn., 182 pages.
C ONDON, E. U. and G. S HORTLEY: 1951. The Theory of Atomic Spectra. Cambridge, England:
Cambridge University Press, 441 pages.
E DMONDS, A. R.: 1996. Angular Momentum in Quantum Mechanics. Princeton, NJ, USA: Prince-
ton University Press, 154 pages.
ISO 21348: 2007. ‘Space environment (natural and artificial) – Process for determining solar
irradiances’. International Organization for Standardization, Geneva, Switzerland.
R ACAH, G.: 1942. ‘Theory of complex spectra II’. Phys. Rev., 62, 438–462.
T OWLE, J. P., P. A. F ELDMAN and J. K. G. WATSON: 1996. ‘A catalog of recombination lines
from 100 GHz to 10 microns’. Astrophys. J. Suppl. Ser., 107, 747–760.
Parity and Reflection Symmetry
D
The concept of parity conservation plays a key role in modern physics and is used
throughout this book. We present here a short introduction to parity and reflection
symmetry from an atomic and molecular physics perspective.
D.1 Parity
refers to spatial symmetry of states and is defined by:
The parity operator P
Pψ(r) = ψ(−r). (D.1)
H
[P, ] = 0.
Thus, the eigenfunctions of the Hamiltonian are also eigenfunctions of the parity
operator. We distinguish states with positive and negative parity (alternatively also
has eigenvalues Pat = ±1:
called odd and even), i.e. P
Pψ(r) = ψ(−r) = ±ψ(r).
The product of two functions with unequal parity has negative parity, while the
product of two functions with the same parity has positive parity. We note that the
integral over all space disappears for a function F (odd) (r) with odd parity:
2π π
F (odd) (r)dΩ ≡ 0. (D.2)
0 0
so that the parity of the states ψnm (r) is Pat = (−1) , hence, spherical harmonics
which differ by = ±1 have opposite parity.
Specifically, the relation
= −r
Pr (D.3)
holds, and we may say that the dipole operator D = −er has odd parity. As trivial
as that statement may seem, it has significant consequences in the context of selec-
tion rules for optical transitions. As discussed in detail in Chap. 4, for an optical
& 2π & π
transition to occur, the integral 0 0 Yb mb (θ, ϕ)rYa ma (θ, ϕ)dΩ must not disap-
pear. Thus, with (D.2) and (D.3) the product Yb mb (θ, ϕ)Ya ma (θ, ϕ) must have odd
parity and transitions are thus only allowed if the parity of initial |a ma and final
state |b mb is different. A dipole transition thus connects only states with opposite
parity. The selection rule for single electron transitions is = ±1.
Somewhat more general – expressing the particle point of view – the photon is
said to have odd parity, Pph = −1. The selection rule (4.86) for dipole transitions
may thus be inferred directly and elegantly from a general parity conservation rule
for the whole system – including atom and photon. Since the photon with its negative
parity disappears during absorption and appears in emission, absorption or emission
of a photon must be accompanied by a change of parity in the atomic system, hence
the selection rule = ±1.
The above discussion tacitly assumed a single electron, characterized by |m, ig-
noring its spin. The parity concept may, however, be extended naturally to multi-
electron systems, assuming that the spatial part of the total wave function |LM can
be written as a product of single electron functions (or as a linear combination of
products such as a Slater determinant). The total parity of an N electron system
is then the product of the parities of all its electron orbits. Thus, a configuration
{n1 1 n2 2 . . . nk k . . . nN N } has even or odd parity depending on whether
N
(−1) 1 i
= (−1)λ = ±1
At this point some general remarks about the choice of basis states or wave functions
and their symmetry properties are in order. For simplicity we first discuss again a
single electron and ignore its spin.
To study different atomic problems some choices of basis states may be more
suitable than others. The one most commonly used in atomic physics for the angu-
lar part of the wave functions (states) is the representation by (complex) spherical
harmonics.
There are, however, a variety of cases where another symmetries may be more
adaptable to a given problem, such as the interaction with an external electric field
(see Sect. 8.2). In that case the eigenstates are linear combinations of states with
±M and |M| remains a good quantum number, while the sign of M is indefinite.
Similarly, in optical transitions the polarization of the exciting light may lead to
particular linear superpositions of different basis states as discussed in Chap. 4. All
these linear combinations are per definition also pure states, and in principle, one
can find an infinite number of such states which, properly orthonormalized, could
also serve as a basis set. In standard quantum mechanical textbooks the |m (or
|LM) states are somewhat overemphasized and one is led to believe that these are
the only suitable basis states for atomic systems. That is only true if a system is
characterized by the magnitude of angular momentum z .
2
L and its z-component L
In addition, the measurement to be described must select states which diagonalize
both operators simultaneously, so that
2
L |LM = L(L + 1)2 |LM
(D.4)
z |LM = M|LM.
L
i(p−1)/2
i(p−1)/2
i(p−1)/2
Note that the relations between real and complex unit polarization vectors (4.7) are
also constructed according to that scheme. And by inserting (B.20) into (D.5) we
obtain the real spherical harmonics
2k + 1
Ykqp (θ, ϕ) = Ckqp (θ, ϕ) where
4π
2(k − q)! q
Ckq+ (θ, ϕ) = P (cos θ ) cos(qϕ) for 0 < q ≤ k (D.7)
(k + q)! k
2(k − q)! q
Ckq− (θ, ϕ) = P (cos θ ) sin(qϕ) for 0 < q ≤ k
(k + q)! k
Rotation of the real tensor operators, real states and real spherical harmonics is
achieved by (E.25) using the real rotation matrices (E.26)–(E.28).
D.3 Reflection Symmetry of Orbitals – Real and Complex Basis States 597
2
L Ykqp = k(k + 1)2 Ykqp
2z Ykqp = q 2 2 Ykqp
L with q = 0, 1, . . . , k, (D.9)
σv (xz)Ykqp = pYkqp with p = ±1, (D.10)
Often one wants to express (D.7) in Cartesian coordinates. We apply a very use-
ful relation communicated by C AOLA (1978). In our notation the so called solid
harmonics Ykq (x, y, z) are (renormalized):
Ckq (x, y, z)
4π
= Ykq (x, y, z) = r k Ckq (θ, ϕ) (D.11)
2k + 1
(k−q)/2
(−1)j (x + iy)j +q (x − iy)j zk−q−2j
= (−1)q (k − q)!(k + q)! .
22j +q (j + q)!j !(k − q − 2j )!
j =max(−q,0)
Inserting this into (D.5) leads to explicit expressions for the renormalized real solid
harmonics Ckqp . Table D.2 summarizes the most important cases.
Figure D.1 illustrates graphically the angular dependence of the three real basis
orbitals for k = = 1: |px , |py , and |pz (short: px , py and pz orbital, respec-
tively) and compares them to the usual complex basis orbitals |1 ±1 and |10. Ob-
viously, the real spherical harmonics are represented by dumbbell like states parallel
to the x-, y-, and z-axes of the atom, respectively. In contrast the complex basis con-
sists of two doughnut like orbitals (|1 ±1 with opposite sense of orientation) while
the |10 orbital is also dumbbell like and identical to |pz .
One may think of preparing these basis states by optical dipole excitation.
However, since electrons have also a spin, and m are typically not
good quantum numbers. Thus, a complete optical preparation of the orbitals shown
in Fig. D.1 is only possible in very special cases. If e.g. one can make use of a
598 D Parity and Reflection Symmetry
⊕z ⊕z
Fig. D.1 Angular part of
atomic p orbitals (spherical p =+1
harmonics) in the complex z
(top row) and real (bottom |+1〉 |0〉 |-1〉
row) basis. Plotted are the
squared moduli |Y1q |2 and
|Y1q± |2 , respectively. The
colour shading indicates x y x y x y
regions of different signs of
the functions. The ⊕ and
signs indicate positive and
negative reflection symmetry
in respect of the xy plane,
⊕z p =+1
z
p =+1 ⊕z p = – 1
| px 〉 | pz 〉 | py 〉
while reflection symmetry in
respect of the xz plane
according to (D.10) is
characterized by red p = +1
and p = −1 x y x y x y
(a) The complex basis set requires circularly polarized light for its preparation: left
and right hand circularly polarized, propagating along the z-axis to excite the
|1 +1 (equivalent to Y1+1 ) and the |1 −1 (equivalent to Y1−1 ) states, respec-
tively. Linearly polarized light propagating perpendicularly to the z-axis and
with its E vector aligned parallel to the z-axis excites the |10 state (equivalent
to Y10 ).
(b) The real basis set requires linearly polarized light for its preparation: the |pz
state (identical to |10) is excited as just described, while the preparation of |px
D.4 Reflection Symmetry in the General Case 599
and |py requires linearly polarized light with its E vector parallel to x and y,
respectively.
Generally speaking, the real basis functions are the basis of choice when the sign
of the angular momentum is irrelevant, i.e. whenever there is no specific preferred
orientation. Thus, they are relevant if external electric fields are present. They play
a particularly important role in molecular physics and quantum chemistry as dis-
cussed in detail in Chaps. 3–4 of Vol. 2, but also in collision problems. There, the
electrostatic field between different, neighbouring atoms defines a suitable z-axis
and is typically very strong so that spin quantum numbers can be treated as a mere
afterthought. Thus, the orbital angular momentum quantum numbers and the abso-
lute value of their projection onto this axis become good quantum numbers. Real
atomic orbitals (AO’s), as depicted in the bottom row of Fig. D.1, and linear com-
binations of them – attached to different atoms – are the basis for a quantitative
treatment of electronic wave functions in molecules.
Reflection symmetry is also an important property in the context of collision
problems. As we shall discuss in Chaps. 6–8, Vol. 2, reflection symmetry in respect
of the scattering plane is conserved during a binary collision for the combined wave
function of the interacting particles. Thus, either reflection symmetry of both parti-
cles remains constant during the collision, or both particles change their reflection
symmetry in respect of the collision plane.
J
σv (xy)|J M = (−1)λ DJM M (π, 0, 0)
J M = e−iMπ (−1)λ |J M
M
We note here in passing that with this relation the |+1 and |−1 p states in Fig. D.1
have both positive reflection symmetry in respect of the xy plane, while the |0 state
has negative reflection symmetry – as intuitively obvious.
Reflection at the xz plane leads to
σv (xz)|J M = (−1)λ DJM M (0, π, 0)
J M
M
= (−1)λ d−MM
J
(π)|J −M = (−1)λ+M+J |J −M. (D.13)
Here we have used the symmetry relation (E.9) for the reduced rotational matrix.
From (D.12) and (D.13) we see that the standard complex |J M representation are
eigenstates of the xy plane reflection operator but not of the operator for reflection
in the xz plane. In analogy to (D.5) one may, however, construct also for the general
case (J integer or half integer) states of specified reflection symmetry in respect of
the xz plane:
1. The states defined by (D.14) are eigenstates of Jz2 . One easily derives
+1 for p = +1 or +i
Jz |J Mp = iM J M(−p) ×
−1 for p = −1 or −i
so that Jz2 |J Mp = M 2 2 |J Mp. (D.17)
2. Jz changes the reflection symmetry from +(−1)J to −(−1)J and vice versa, and
so does Jx :
Jx |J Mp = a(M)
J (M + 1)(−p) + a(−M)
J (M − 1)(−p) (D.18)
as one verifies for J ≥ 1 and M > 0 easily with some algebra. In contrast Jy
does not change reflection symmetry:
Jy |J Mp = a(M)
J (M + 1)p − a(−M)
J (M − 1)p (D.19)
a(M) = √ J (J + 1) − M(M + 1).
2
3. As specific examples, we have already discussed the p orbitals |px = |11+,
|py = |11− and |pz = |10+ depicted in the lower part of Fig. D.1. Without
proof we note here in passing the relations:
4. The simplest, and somewhat exceptional case with half integer J is a pure spin
state with λ = 0, J = 1/2, describing e.g. an electron in a 1 S1/2 state. The basis
states in the |J M representation are |1/2 1/2 and |1/2 −1/2, with spin point-
ing in the +z- and −z-direction, respectively. Reflection on the xy plane gives
according to (D.12)
%
%
%
%
1 1
1 1
1 1
1 1
σv (xy)
= i
and
σv (xy)
− = −i
− .
(D.21)
22 22 2 2 2 2
We see that the up and down spin states have opposite reflection symmetry in
respect of the xy plane. The corresponding states with defined xz reflection sym-
metry are obtained from (D.14) (λ = even) with p = ±i:
1 These formulas hold for all J ≥ 1 and M ≥ 1. The other cases are easily derived individually
These states are identical to (E.17) and (E.18), respectively, obtained by coor-
dinate rotation for a spin state pointing into ±y direction. Reflection at the xz
plane gives with (D.13):
σv (xz)|+i = i|+i and
σv (xz)|−i = −i|−i. (D.24)
In this particular case (and only in this one), these xy reflection symmetric states
are also eigenvectors of Jy . With (B.4) and (B.14)–(B.17) one easily verifies that
Jy |±i = ± |±i. (D.25)
2
We finally apply the concept of reflection symmetry to optical transitions treated
in detail in Chap. 4. In Sect. 4.1.2 we have introduced linear (ex , ey ) and circular
(e+1 , e−1 ) polarization vectors. We may write |ex , |ey and |e+1 , |e−1 as the real
and complex representation of the corresponding photon states, respectively. Then
the definitions (4.4)–(4.7) are in complete agreement with (D.14). Since for the pho-
ton J = λ = 1, M = ±1, we have p = +1 for |ex and p = −1 for |ey , i.e. |ex
and |ey have positive and negative reflection symmetry in respect of the xz plane.
Since the (fully quantized) Hamiltonian for the system photon + atom is invariant
under reflection at any plane through the centre of the atom, the reflection symmetry
of the state vectors describing the combined system must conserve reflection symme-
try. Depending on whether we use light which is polarized in the x- or y-direction,
the atom changes or conserves its reflection symmetry in an optical transition as
indicated by the following scheme:
|J Mp; ex ←→ |J Mp; . . .
|J Mp; ey ←→
J M(−p); . . . .
to the transition has +i × (−1) = −i reflection symmetry, thus the final state must
also have −i symmetry, which corresponds to |Jb pointing into +y-direction. In
contrast, for excitation with ex polarized light, the combined initial system |Ja |ex
has +i × (+1) = i symmetry, which must be conserved and corresponds to reversing
the orientation of the spin into |−Jb .
Circularly polarized photons |e+1 or |e−1 propagating into z-direction have
both positive reflection symmetry in respect of the xy plane (in analogy to the |+1
and |−1 p states in Fig. D.1) and cannot change this reflection symmetry of the
atomic states. Formally one sees this from (D.12), since in a dipole transition both λ
and M change by ±1 in respect of the “photon frame” where z k. Thus, (−1)λ+M
remains constant during the excitation process and reflection symmetry in respect
of the xy plane does not change.
good quantum number: ‘Quantum number for eigenvalues of such observables that
may be measured simultaneously with the H AMILTON operator (see Sect. 2.6.5)’.
References
C AOLA, M. J.: 1978. ‘Solid harmonics and their addition theorems’. J. Phys. A, Math. Gen., 11,
L23–L25.
Coordinate Rotation
E
We compile here the most important formulas for rotation of coordinate systems
and their consequences on angular momentum states and tensor operators – without
going into details of the derivation.
(xyz) → x y z → x y z → x # y # z# .
First, one rotates the system around its original z-axis (z = z) through an angle α.
The following rotation through an angle β around the new y axis (y = y ) brings
z into the direction z and x into direction x . The final rotation through an angle
γ is around the thus tilted z = z# axis (x → x # and y → y # ).
The operator achieving this rotation is derived from the angular momentum opera-
tor
J:
D(αβγ ) = exp(−iγ Jz ) exp(−iβ Jy ) exp(−iα Jz ).
The corresponding rotation matrix elements, more precisely the matrix elements
J of the rotation group, may be derived with
of the irreducible representation D
some angular momentum algebra (following the notation of B RINK and S ATCHLER,
1994),1 and are written as
DJMN (αβγ ) =
J M|D(αβγ )|J N = exp(−iαM)dMN
J
(β) exp(−iγ N ), (E.1)
Special cases:
4π
DJM0 (αβγ ) = CJ∗ M (β, α) = Y ∗ (β, α) (E.3)
2k + 1 J M
(J − M)! M
J
dM0 (β) = (−1)M P (cos β) if M > 0 (E.4)
(J + M)! J
J
and d00 (β) = PJ (cos β). (E.5)
1 Here the coordinate system is rotated. An alternative convention, also found in the literature, is to
2
d22 = d−2−2
2
= cos4 (β/2) (E.13)
2 = d2 = d2
−20 = d0−2
2 = −d 2
−1−2 0 = −d−2−1 = d12
d20 2 2 2
02 d21
√
= ( 3/8) sin2 β = − sin β(1 + cos β)/2
2
d2−1 = d1−2
2 0 = −d 2
−21 = −d−12
2
2
d2−2 = d−22
2 = sin4 (β/2)
= sin β(1 + cos β)/2
2
d11 = 2
d−1−1 2
d1−1 = d−11
2
The components Tkq (θ # , ϕ # ) of tensor operators of rank k (e.g. Ykq or a vector with
k = 1) can be expressed by tensor components in the old system
Tkq θ # , ϕ # = Tkq (θ, ϕ)Dkqq (αβγ )
q
= exp −iγ q k
Tkq (θ, ϕ) exp(−iαq)dqq (β) (E.15)
q
with θ , ϕ and θ # , ϕ # being the angles in the old and new system, respectively.
We illustrate the procedure by a particularly simple and important example,
a spin s = 1/2 state |1/2 1/2# pointing into the ϕ = α, θ = β direction in the orig-
1/2
inal coordinate frame xyz. Inserting for this case dMN (β) from (E.11) into (E.14)
we obtain:
608 E Coordinate Rotation
(a)
(b)
Fig. E.2 Rotation to describe a spin s pointing into (a) +y-direction, (b) +x-direction
%#
%
%
1 iγ iα β
1 iα β
1
= exp − exp − cos + exp sin − .
2 2 2 2
2 2 2
2
(E.16)
(For simplicity of writing we have dropped here J = 1/2.) The third rotation angle
γ around the final z# axis is somewhat arbitrary and just contributes a free phase
factor.2 Figure E.2(a) indicates the rotation angles for one specific case where the
spin s points into the +y-direction of the (xyz) coordinate system. Thus, the oblig-
atory rotation angles are α = ϕ = π/2 and β = θ = π/2. We choose γ = −3π/2,
make use of (E.11)–(E.13), and thus obtain from (E.16) for this particular spin state
%
%
%
1 #(y) 1
1
1
+ = √ i
+ −
−
. (E.17)
2 2 2 2
If the spin s points into the −y-direction, we make exactly the same rotation and
find from (E.14)
%
%
%
1 #(y) 1
1
1
− = √ i
+ +
− (E.18)
2 i 2
2
2 .
Somewhat more compact, we designate the spin states parallel and antiparallel to
the y-axis with by |α(y) and |β(y) , respectively, and those which refer to the z-axis
with |α and |β. Equations (E.17) and (E.18) may then be written:
1
1
For completeness we also describe a state where the spin s points into the ±x-
direction. As illustrated in Fig. E.2(b) the E ULER angles in this case are (0, π/2, π)
and we obtain
%
1 #(x) ∓1
onto the z-axis, with eigenvalues /2 and −/2, respectively. What about the states
|±1/2(x,y) ?
Here is a little exercise for the active reader, which can be solved with a few
lines of algebra (for complex quantities): Using (B.4), (B.16) and (B.17), show that
the spin states |α(y) and |β(y) are indeed eigenstates of the spin component in y
direction with eigenvalues ±/2:
Sy |α(y) = |α(y) and
Sy |β(y) = − |β(y) . (E.21)
2 2
And the same holds of course for the corresponding spin states pointing in ±x-
direction.
For many-particle systems we have to construct spin functions which account for
the fact that identical particles are indistinguishable. We have discussed this in the
main text on several occasions, see e.g. Table 6.1 in the context of fine structure and
Sect. 7.3 where specifically the He atom is treated. Let us take the latter example,
i.e. the two electrons in the He ground state (1s)2 1 S0 , or any singlet state of two
electrons in the continuum after ionization (with a symmetric spatial wave function).
The two spins of the electrons are anti-parallel and since we cannot distinguish
which of these two fermions has which spin we have to construct an antisymmetric
description of their joint state, according to (7.36)
0
where the numbers in brackets refer to electron (1) and electron (2), respectively.
Obviously, this state of the two electrons is inseparable, i.e. it cannot be written
as product of two wave functions of the individual electrons. One calls such states
entangled.
Equation (E.22) refers to a particular coordinate frame and |α and |β reflect
the spin components with respect to its z-axis. The total (spin) angular momentum
610 E Coordinate Rotation
operator is
S = S1 +
S 2 , and the singlet state (E.22) is constructed such that the
respective eigenvalues of S and
2
Sz are zero:3
2
Sz
χ00 = (
S2z )
χ00 = 0 and
S1z + S
χ00 = 0. (E.23)
One may now be curious what happens if we define this entangled, antisymmetric
spin state with respect to another axis, say the y-axis. This may be done in complete
analogy to (E.22) from spin states with angular momentum parallel and antiparallel
to the y-axis:
0
For reference we also communicate the rotation procedure (FANO, 1960) for the
real harmonics, real states and real tensor operators as defined in Appendix D. In
this case, the transformation is achieved by
# #
Tkq p θ , ϕ = Tkqp (θ, ϕ)Dkqp,q p (αβγ ) (E.25)
qp
3 The first relation is obvious, since each of the operators S1z and
S2z acts only onto one of the
states which have the eigenvalues ±/2, i.e. they compensate each other. The latter relation may
easily be verified as an exercise with the help of (B.8) and (B.17).
References 611
with p, p = ±1. The real rotation matrices are obtained by applying the complex
rotation (E.15) onto the defining equations (D.5) and finally back-transformation to
the real basis. One obtains:4
For k ≥ q, q > 0:
(E.26)
Dkqp,q p (αβγ )
⎧ k
⎪ k
⎪ (−1)q dqq (β) cos(αq + γ q ) + p dq−q (β) cos(αq − γ q )
⎪
⎨
if p = p
= (−1)−q
⎪
⎪ (−1)q p dqq
k (β) sin(αq + γ q ) + d k
q−q (β) sin(αq − γ q )
⎪
⎩
if p
= p
For q or q = 0: (E.27)
√ k cos(qα + q γ ) if p = p
Dkqp,q p (αβγ ) = (−1)−q+q 2dqq (β)
sin(qα − q γ ) if p
= p
specifically Dkqp,0+ (αβγ ) = Ckqp (β, α) (see Eq. (D.7))
References
B LUM, K.: 2012. Density Matrix Theory and Applications. Atomic, Optical, and Plasma Physics,
vol. 64. Berlin, Heidelberg: Springer Verlag, 3rd edn., 343 pages.
B RINK, D. M. and G. R. S ATCHLER: 1994. Angular Momentum. Oxford: Oxford University Press,
3rd edn., 182 pages.
E INSTEIN, A., B. P ODOLSKY and N. ROSEN: 1935. ‘Can quantum-mechanical description of
physical reality be considered complete?’ Phys. Rev., 47, 777–780.
FANO, U.: 1960. ‘Real representations of coordinate rotations’. J. Math. Phys., 1, 417–423.
H ERTEL, I. V. and W. S TOLL: 1978. ‘Collision experiments with laser excited atoms in crossed
beams’. In: ‘Adv. Atom. Mol. Phys.’, vol. 13, 113–228. New York: Academic Press.
H ORODECKI, R., P. H ORODECKI, M. H ORODECKI and K. H ORODECKI: 2009. ‘Quantum entan-
glement’. Rev. Mod. Phys., 81, 865–942.
R AIMOND, J. M., M. B RUNE and S. H AROCHE: 2001. ‘Colloquium: Manipulating quantum en-
tanglement with atoms and photons in a cavity’. Rev. Mod. Phys., 73, 565–582.
T ICHY, M. C., F. M INTERT and A. B UCHLEITNER: 2011. ‘Essential entanglement for atomic and
molecular physics’. J. Phys. B, At. Mol. Phys., 44, 192001.
4 See also H ERTEL and S TOLL (1978), Eq. (15) (some small typos there have been corrected here).
Multipole Expansions and Multipole Moments
F
One often needs suitable approximations for potentials which at large distances are
nearly symmetric around a centre, but not completely. For scalar as well as for vector
potentials, multipole expansions may be derived. We treat here the important case
of an electrostatic potential arising from a charge distribution localized in a small
region around the origin – as e.g. encountered when describing the interaction of
atomic nuclei with the electron charge cloud of an atom (see Fig. 9.12 in Chap. 9).
In the final subsection we shall also introduce a more general definition of multipole
tensor operators and multipole moments.1
We first note that the inverse distance between two points r and r in space may be
expanded for r r into a power series of x = r /r:
1 1 −1/2
= 1 − 2x cos θ + x 2 (F.1)
|r − r | r
1 3 1 5 3
= 1 + x(cos θ ) + x 2 cos2 θ − + x3 cos3 θ − cos θ · · ·
r 2 2 2 2
where θ is the angle between r and r . With (B.24) we identify the angle dependent
terms in brackets as L EGENDRE polynomials. With this result in mind, we commu-
nicate the L APLACE expansion which holds for r < r:
∞
r k
1
= Pk (cos θ )
|r − r| r k+1
k=0
1 The term “multipole moments” is used in the literature for a variety of slightly different defini-
tions. As far as necessary for distinction, we shall add some specification, indicated in (brackets).
∞
∞
r k ∗ r k (r) (r )
k
= C kq (θ, ϕ)C kq θ , ϕ = C · Ck (F.2)
r k+1 r k+1 k
k=0 q=−k k=0
∞
r k
k
= Ckqp (θ, ϕ)Ckqp θ , ϕ .
r k+1
k=0 q=0,p=±1
In the second line of (F.2) we have made use of the addition theorem (C.22) for
the renormalized spherical harmonics, with θ , ϕ and θ , ϕ being the angles of the
position vectors r and r , respectively. In the following equality we have made use
(r) (r )
of (C.21) and the scalar product notation (C.20) for the two tensors Ck and Ck .
The third line of (F.2) introduces the real (renormalized) spherical harmonics as
defined in Appendix D.3.
With these relations we write the electrostatic interaction energy dV12 (r, r ) of a
charge dq, located at a position r , with a point charge −e located at r
∞
1 −edq 1 −e (r) k (r )
dV12 r, r =
= Ck · r dq Ck . (F.3)
4πε0 |r − r| 4πε0 r k+1
k=0
For an extended electric charge with density ρ(r ), localized around the origin, the
charge element is dq = ρ(r )d3 r , and the total interaction energy with the point
charge −e is obtained from
−e ρ(r ) 3
V (r) = dV12 r, r = d r. (F.4)
4πε0 |r − r|
With (F.3) we may write this as a multipole expansion
∞
−e 1
k
V (r) = Qkq Ckq (θ, ϕ) or (F.5)
4πε0 r k+1
k=0 q=−k
∞
−e 1
k
in real form = Qkqp Ckqp (θ, ϕ), (F.6)
4πε0 r k+1
k=0 q=0,p=±1
where we have used in the second row for abbreviation the renormalized real solid
harmonics Ckqp as introduced in Appendix D.3.
These multipole expansions, derived from (F.3), are most appropriate if r r.
Thus, it is suitable e.g. for describing the field of an atomic nucleus as seen by an
electron far away from the nucleus (i.e. for ≥ 1), or for the far field of an atom in
a well defined polarization state (with ≥ 1).
The total charge is given by
q = Q00 = ρ r d3 r . (F.9)
If only positive or only negative charges are involved, the dipole moments are made
to disappear by suitable choice of the origin, thus defining a charge centre. They
also disappear if both charges have the same centre (characteristic for atoms, unless
they are placed in an electric field, see Sect. 8.2.8). In contrast, if positive and neg-
ative charges with different centres are to be described (typical for polyatomic and
heteronuclear diatomic molecules) the dipole moment is finite.
With (F.8) and Table D.2 the real quadrupole moments are:
1 2
Q20+ = 3z − r 2 ρ r d3 r Q20− = 0
2
√ 3 √ 3
Q21+ = 3 z x ρ r d r Q21− = 3 zyρ r d r (F.11)
√ √
3 2 3
Q22+ = x − y 2 ρ r d3 r Q22− = x y ρ r d3 r .
2 2
616 F Multipole Expansions and Multipole Moments
The components (F.10) of the electric dipole moment D are zero for charges cen-
tred at the origin, while the quadrupole moments Q2qp according to (F.11) are fi-
nite for an anisotropic charge distribution. The renormalized real solid harmonics
C2qp (x, y, z) are found in Table D.2.
Closely related to the (charge) multipole moments just discussed, but without any
reference to the charge of the particles, one defines irreducible spatial multipole
tensor operators of rank k. They are constructed from the position vector and have
the components
using again the (renormalized) solid harmonics (D.11). They may also be written as
real quantities according to (D.5):
For explicit expressions we refer to Table D.2. General rules for tensor operators
have been laid out in Appendix C.
2 E.g.
following JACKSON (1999), in classical electrodynamics one usually expands in terms of the
standards spherical harmonics, multiplying the terms in the sum (F.7) by 4π/(2 + 1), so that the
(complex) multipole moments are given as
∗
qm = Ym (Θ, Φ)R ρ(R)d3 R.
F.3 Multipole Tensor Operators 617
They characterize the particle distribution (e.g. for the protons in an atomic nucleus,
for the electrons in an atom, or for electrons and nuclei in a molecule).
For a given angular momentum J of a quantum state the above definitions (F.14)
and (F.16) allow to construct multipole tensor operators and multipole moments of
rank 0 ≤ k ≤ 2J . They all have a well defined parity
kq (r) = (−1)k Qkq (r)
PQ (F.19)
but also a reflection symmetry in respect of the xy plane (see also Appendix D.4):
Most commonly used is the quadrupole tensor Q2 with its components Q2q =
C2q (r, θ, ϕ) according to Table D.2. Higher moments become increasingly com-
plicated. If the quantum state |J M of an object is known exactly, the components
of the quadrupole moment are obtained from the respective matrix elements of the
quadrupole tensor Q2 . These matrix elements can be derived from the W IGNER -
E CKART theorem (C.9):
J −M
√ J J 2
γ J M Q2q |γ J M = (−1) 2J + 1
J Qk J (F.22)
M −M q
with
J Q2 J =
J C2 J
γ J |r 2 |γ J . (F.23)
and if the radial wave functions Rγj (r) are known, the radial matrix element
r 2
is obtained from
2
r =
γ J |r |γ J = r 4 |Rγj |2 dr.
2
The expectation values of the quadrupole tensor, i.e. the (spatial) quadrupole
moments according to (F.18), may be related to the (charge) quadrupole moments
defined in Sect. F.2 by
Q2q = q
Q2q = r 2
J C2 J w(M)
J M20|J M. (F.31)
For a prolate (cigar shaped) charge distribution, dominated by low M states, (F.31)
leads to Q > 0, while an oblate (pancake like) distribution is characterized by large
|M| and Q < 0 (see e.g. Fig. 9.15).
In nuclear physics, the quadrupole moment of a nucleus (assuming cylindrical
symmetry around the z-axis) is
Note the factor 2eZ (overall charge eZ) as already introduced in (F.12): averaging
3
Tkq± = (Ji1 Ji2 . . . Jik )∇i1 ∇i2 . . . ∇ik Ckq± (r), (F.32)
i1 i2 ...ik =1
where ij = 1 to 3 stands for x, y and z. The angular momenta are given in a.u., i.e. in
units of . Applying this operation to the explicit expression for the (renormalized)
solid harmonics Table D.2 we obtain Table F.1.
Table F.1 Multipole tensor operators constructed from angular momenta for k = 1 to 3 (only the
zero component is given for the octupole moment k = 3); for reference we also communicate the
relations with the FANO and M ACEK (1973) orientation (O1− ) and alignment parameters (A1+
and A2+ )
k p = +1 p = −1
0 T00+ = 1 T00− = 0
1 T10+ = Jz T10− = 0
1 T11+ = Jx T11− = Jy = J (J + 1)O1−
T20+ = (3Jz2 − T20− = 0
2
2 J )
√ √ √
2 T21+ = 3(Jz Jx + Jx Jz ) = J (J + 1) 3A1+ T21− = 3(Jz Jy + Jy Jz )
√ √ √
2 T22+ = 3(Jx2 − Jy2 ) = J (J + 1) 3A2+ T22− = 3(Jx Jy + Jy Jx )
T30+ = (15Jz3 − 9
J Jz + 3Jz )
2
3 etc.
The Tk0 (= Tk0+ ) components are obviously diagonal in J and M. Explicitly, the
matrix elements for k = 1 are
M
J M|T10 |J M = M = √
J T1 J (F.33)
J (J + 1)
and describe the orientation of the state (nonvanishing angular momentum). The
latter equality follows from the W IGNER -E CKART theorem (C.9) and (B.55). The
matrix elements of T20 characterize the alignment (quadrupole moment or the
anisotropy) of the charge density distribution in a given state:
J M|T20 |J M = 3M 2 − J (J + 1) (F.34)
2[3M 2 − J (J + 1)]
=√
J T2 J .
(2J + 3)(2J + 2)2J (2J − 1)
The latter follows again from the W IGNER -E CKART theorem, this time using
(B.58).
The reduced matrix elements for rank k = 1 and k = 2 follow directly from (F.33)
and (F.33), respectively:
J T1 J = J (J + 1) and (F.35)
J T2 J = (2J + 3)(J + 1)J (2J − 1). (F.36)
For an arbitrary rank k one finds (see e.g. M ACEK and H ERTEL, 1974, and refer-
ences given there; note, however, that the factor (2J + 1)−1/2 is specific for the
B RINK and S ATCHLER notation of the reduced matrix elements):
k! (2J + k + 1)!
J Tk J = k . (F.37)
2 (2J + 1)(2J − k)!
We point out that these (general) multipole tensor operators and their expectation
values, the (general) multipole moments, may be finite also for odd rank – in contrast
Acronyms and Terminology 621
J Tk J
J M
Tkq |J M = J M Qkq |J M, (F.38)
J Qk J
with
J Qk J defined by (F.23). We emphasize that (F.38) also holds for the re-
spective expectation values (the multipole moments).
Often one is only interested in the expectation values (multipole moments) of
these tensors in an orbital angular momentum basis with sharp = (see e.g. Ap-
pendix D.2 in Vol. 2). We may then use the explicit expression (C.33) for the reduced
matrix elements of the renormalized spherical harmonics. These are nonvanishing
only for even k and we note that their sign is always negative – in contrast to (F.37)!
Thus, the multipole moments
Tkq (constructed from angular momenta) have the
opposite sign from
Qkq (which are built from spatial coordinates)!
Expectation values of these multipole tensor operators are widely used to char-
acterize anisotropy and orientation of atomic or molecular systems – in particular
if these cannot be described by pure states. Strictly speaking, they are measured in
units of k but one usually drops the unit. We shall refer to them as multipole mo-
ments as long as no confusion can arise with the (spatial) multipole moments con-
structed from position vectors.4 More details will be discussed in the context of the
density matrix in Chap. 9 and Appendix C, Vol. 2. In addition, we shall introduce
there yet another kind of irreducible tensor operators to characterize anisotropic
state populations, the so called statistical tensor operators. Their expectation values
are the state multipoles which again can be related by simple numerical factors with
any other irreducible representation of tensor operators, in analogy to (F.38).
References
B OHR, A. and B. R. M OTTELSON: 1998. Nuclear Structure, vol. 1: Single-Particle Motion. Sin-
gapore: World Scientific, reprint from 1969 edn., 471 pages.
B RINK, D. M. and G. R. S ATCHLER: 1994. Angular Momentum. Oxford: Oxford University Press,
3rd edn., 182 pages.
FALKOFF, D. L. and G. E. U HLENBECK: 1950. ‘On the directional correlation of successive nu-
clear radiations’. Phys. Rev., 79, 323–333.
FANO, U.: 1960. ‘Real representations of coordinate rotations’. J. Math. Phys., 1, 417–423.
FANO, U. and J. H. M ACEK: 1973. ‘Impact excitation and polarization of emitted light’. Rev. Mod.
Phys., 45, 553–573.
JACKSON, J. D.: 1999. Classical Electrodynamics. New York: John Wiley & Sons, 3rd edn., 808
pages.
M ACEK, J. and I. V. H ERTEL: 1974. ‘Theory of electron-scattering from laser-excited atoms’.
J. Phys. B, At. Mol. Phys., 7, 2173–2188.
Convolutions and Correlation Functions
G
In this appendix all distribution functions f (x) are normalized such that
+∞
f (x)dx = 1. (G.1)
−∞
The mathematical operation called convolution quantifies the overlap between two
mathematical functions f1 (x) and f2 (x) as one is reversed and shifted in respect of
the other. A convolution is defined as
∞
(f1 ∗ f2 )(x) := f1 (ξ )f2 (x − ξ )dξ, (G.2)
−∞
Thus, both procedures are identical if one of the functions is even, f (−x) = f (x),
which is the case in most of the physical applications.1 If f1 ≡ f2 one calls G(x) =
(f f )(x) an autocorrelation function.
1 An alternative method to solve (G.3) is given by F (f1 f2 ) = 2π F (f1 )F (f2 ) where F (f ) de-
scribes the F OURIER transform of f (see Appendix I).
S1
S2
x, ξ
Fig. G.1 Convolution scheme of two rectangular profiles as manifested by two slits S1 and S2 of
unequal size. Their respective transmission profiles are shown as full black and red dashed lines.
As slit S2 is shifted over slit S1 in x, ξ direction, the overlap (grey area) changes. The resulting
total transmission is shown as heavy red line (convolution)
the emission or absorption process and add to the broadening. The combination of
both effects is described by a convolution of the respective line profiles.
The physics involved in all these processes and examples will be discussed in
some detail in Sect. 5.1.1, and Chaps. 1, 2 in Vol. 2, and in other parts of these vol-
umes. In the present appendix we summarize a few useful mathematical expressions
which are needed to describe such measurements quantitatively. At the end one aims
of course at an intelligent de-convolution of the signal detected as a function of x,
i.e. one tries to extract the original profile f1 (ξ ) form the convoluted signal. This is
no trivial task unless the shapes of the two profiles involved are both well known.
Often one is content with just measuring the convolution and gleaning the width of
the profile f1 (ξ ) from known convolution formulas.
EA∗ (t)EB (t + δ)
g (1) (δ) = . (G.5)
[
|EA (t)|2
|EB (t)|2 ]1/2
2 The further treatment is independent of whether EA and EB relate actually to two different fields
or represent the same field at different positions in space.
626 G Convolutions and Correlation Functions
Correlation functions are e.g. used in state-of-the-art ultrafast science for the
determination of shape and duration of short laser pulses as to be detailed in Chap. 1,
Vol. 2. In such a measurement, the electric field amplitude EA (t) (or the intensity
∝ |EA (t)|2 ) of the pulse to be characterized is convoluted with a second light pulse
of well known profile EB (t) (or intensity ∝ |EB (t)|2 ). This is achieved by delaying
one pulse for a variable time δ in respect of the other. Both signals are multiplied
and integrated over a long time Tav . The signal is then recorded as a function of the
delay time δ and the result may finally be de-convoluted.
If EA (t) and EB (t) are identical (or just displaced in time) the first-order degree
of coherence (G.5) simply becomes the normalized autocorrelation function
E ∗ (t)E(t + δ)
E ∗ (t − δ)E(t)
g (1) (δ) = = = g (1) (−δ). (G.6)
|E(t)|2
|E(t)|2
The important symmetry with respect zero follows directly from the definition (G.4)
for A = B. If the shape of a pulse is already well known, it allows one to determine
directly the pulse duration (see below for special examples).
Note that here the 1/Tav factor cancels out in (G.5) and (G.6): these definitions
hold for CW fields as well as for a single pulses or a pulse sequence. In practice, the
averaging may not be that trivial, especially if E(t) is not given in closed analytical
form. For CW fields the average
. . . can be performed over times at a given po-
sition in space – or alternatively over the whole ensemble in space at a given fixed
time. One calls a system ergodic if the temporal average is equal to the average over
a representative ensemble – which is generally true for reasonable physical systems.
Usually the temporal average is more convenient to perform – while determining the
ensemble average (averaging over the whole space) would be a rather difficult task.
According to the ergodicity theorem,both averages are identical.
the convoluted FWHM thus being the geometric mean of both widths
x = 1.177w = 1.177 w12 + w22 = (x1 )2 + (x2 )2 . (G.13)
The hyperbolic secant squared is another, often used line profile, in particular for
laser pulse intensities. Normalized according to (G.1) it is
sech2 [(x − x0 )/w] 1
fH (x; w, x0 ) = = (G.15)
2w 2w cosh [(x − x0 )/w]
2
2
1 2
= ,
2w e(x−x0 )/w + e−(x−x0 )/w
√
The convolution is a not completely trivial integral. For the autocorrelation function
one finds (properly normalized)
1 (x/w) cosh(x/w) − sinh(x/w)
(fH fH )(x; w) = (G.17)
w [sinh(x/w)]3
The first equality is exact, the sech4 gives an excellent approximation, and the sech2
is still a good approximation for small |x| ≤ 1.7; in the far wings, however, it is
somewhat too high.
Another profile often encountered is the L ORENTZ profile, describing e.g. the natu-
ral line profile or collision broadening (see Sects. 5.1.1 and 5.1.3, respectively). We
write the L ORENTZ profile
2 γ 2 /4
fL (x; γ , x0 ) = (G.19)
γ π (x − x0 )2 + γ 2 /4
with a FWHM x = γ , (G.20)
2 γ 2 /4
(fL fL )(x; γ2 , x2 ) = (G.21)
γ π (x − (x1 − x2 ))2 + γ 2 /4
with the FWHM of the convolution x = γ = γ1 + γ2 .
Convolution of a L ORENTZ with a G AUSS profile is required e.g. for optical line
shapes of atoms or molecules in the gas phase if D OPPLER broadening and collision
broadening are on the same order of magnitude (or at low temperatures D OPPLER
and natural linewidth). This so called VOIGT profile
+∞
e−2ξ /w
2 2
2 1 γ
fV (x; σ, γ ) = (fG fL )(x; γ , 0) = dξ, (G.22)
π w 2π −∞ (ξ − x)2 + γ 2 /4
Acronyms and Terminology 629
supposed to be accurate to within 0.02 %, where ωL and ωD are the FWHM
of L ORENTZ and D OPPLER profile, respectively. One example of this convoluted
D OPPLER and collision profile for ωL = ωD is shown in Fig. 5.5. In this case
the formula gives ωV = 1.638ωL .
References
NIST-DLMF: 2013. ‘Digital library of mathematical functions: §7.19 Voigt functions’, NIST.
http://dlmf.nist.gov/7.19, accessed: 9 Jan 2014.
SWP 5.5: 2005. ‘Scientific work place’, Poulsbo, WA 98370-7370, USA: MacKichan Software,
Inc. http://www.mackichan.com/, accessed: 9 Jan 2014.
W EISSTEIN, E. W.: 2011. ‘Convolution’, Wolfram Research, Inc., Champaign, IL, USA. http://
mathworld.wolfram.com/Convolution.html, accessed: 9 Jan 2014.
W IKIPEDIA CONTRIBUTORS: 2014. ‘Voigt profile’, Wikipedia, The Free Encyclopedia. http://en.
wikipedia.org/wiki/Voigt_profile, accessed: 9 Jan 2014.
Vector Potential, Dipole Approximation,
Oscillator Strength H
In view of its clearness and compactness we use in this book – apart from a few ex-
ceptions – in general the dipole length approximation for describing the interaction
of electromagnetic waves with atoms and molecules. It is thus appropriate to justify
this approach in some detail. In addition, we shall provide the basis for some useful
concepts related to the dipole approximation.
p 2field 1
=
H + V (r) = p + eA)2 + V (r)
( (H.1)
2me 2me
p2 p·A
e e2 2
= + V (r) + + A (H.2)
2me me 2me
where the so called C OULOMB gauge of the field has been used:
∇ · A = 0. (H.3)
For simplicity we have restricted the discussion to a single active electron. One
easily generalizes (H.2) to multi-electron systems by replacing momentum p and
position vector r with p i and r i , respectively, for each electron i, and summing
over all electrons.
The vector potential A relates to the electric and magnetic field vectors
∂
E(r, t) = − A(r, t) + ∇Vext and B(r, t) = ∇ × A(r, t) (H.4)
∂t
with Vext (r, t) describing an additional, external potential. If such a field is present,
one has to add Vext to V (r) in (H.1) and (H.2). Not included in this description are
of course all interactions related to the electron spin which follow from the D IRAC
equation, including spin-orbit interaction, as briefly discussed in Chap. 6.
In vacuum, we write the vector potential, electric field, and magnetic field for a
plane electromagnetic wave as:
A0 i(kr−ωt)
A(r, t) = ee + e∗ e−i(kr−ωt) = A0 e cos(kr − ωt) (H.5)
2
iE0 i(kr−ωt)
E(r, t) = ee − e∗ e−i(kr−ωt) = E0 e sin(kr − ωt) (H.6)
2
iB0 k i(kr−ωt) k
B(r, t) = × ee − e∗ e−i(kr−ωt) = B0 × e sin(kr − ωt). (H.7)
2 k k
The relations among the amplitudes are derived from (H.4)
ω
E0 = ωA0 , B0 = kA0 = A0 , and B0 = E0 /c. (H.8)
c
H.1.2 Intensity
In principle, the amplitude E0 (r, t) (and A0 and B0 ) as well as the averaged intensity
I (r, t) may still depend slowly on time t and position r. This is called SVE.1 More
about cycle averaged intensities is discussed in Sect. 1.3.1, Vol. 2. The electric field
is measured in units [E0 ] = V m−1 , the intensity in [I ] = W m−2 .
Note that all field quantities used here represent quasi-monochromatic waves:
their bandwidth is assumed to be much narrower than any atomic or molecular ab-
sorption line studied. The transition to a continuous spectrum is made by replacing
I → I˜(ω)dω, where the spectral intensity distribution I˜(ω) (or the spectral energy
density ũ(ω) = I˜(ω)/c) refer to the unit
& of angular frequency interval. Finally, one
has to integrate over all frequencies: . . . dω.
For a static, homogeneous external magnetic field B one derives the vector potential
by inverting (H.4):
1 1
A = − r × B = B × r. (H.12)
2 2
This may be verified by inserting this expression into (H.4).2
For such a constant B field the Hamilton operator (H.2) becomes
=
H
p2
+ V (r) +
e
L·B +
e 2 A2
, (H.13)
2me 2me 2me
e 2 A2 e2 2 2
= r B sin θ, (H.14)
2me 8me
where θ is the angle between the external B field and the position vector r of the
electron in the atom. We estimate for, say, 30 T and r = a0 a maximum value of ca.
5 × 10−8 eV. Thus, in spectroscopy this term only plays a role if extreme accuracy
is asked for or very high magnetic fields and very large orbital radii are involved
(i.e. high lying RYDBERG states). On the other hand, it is this very term which is
responsible for all diamagnetism of matter.
For the following considerations we derive an important relation between the matrix
elements of momentum and position. For electrons i and j the standard commuta-
tion rules for canonically conjugated observables (momentum and position) hold:
[xi , p
yj ] = 0 and [xi , p
xj ] = iδij , etc. (H.15)
a,
a,
b2 = [ b]
b + a,
b[ b] (H.16)
we can write
] = i p
[xi , H xi and [r, H ] = i p, (H.17)
me me
where r = r i and p= p i . Thus, for the matrix elements of
p between eigen-
states |a and |b of the Hamiltonian we derive
me ]|a = me (Wa − Wb )
b|r|a.
b|
p |a =
b|[r, H (H.18)
i i
With (Wa − Wb )/ = −Wba / = −ωba the sought-after relation between the matrix
elements of
p and r is:
b|
p |a = ime ωba
b|r|a. (H.19)
Before actually evaluating the matrix elements of the interaction in the Hamiltonian
(H.2) in detail, we have a second look at the term proportional to A2 , now for an
atom or molecule in an electromagnetic wave. Clearly, this term is a time dependent
additional energy in the Hamiltonian. By averaging A2 over one period we obtain
with (H.11) an estimate for its dependence on intensity I :
e2
A2 e2 I
Up = = . (H.20)
2me 2ε0 cme ω2
H.1 Interaction of the Field of an Electromagnetic Wave with an Electron 635
This is identical to the expression (8.139) discussed in Sect. 8.5.1 for the pondero-
motive potential Up . There, a completely classical picture of an electron oscillating
in the electric field of the wave is used to ‘derive’ it.
In respect of standard laser spectroscopy, the ponderomotive potential is usually
negligible. As described in Sect. 8.5.1, however, it leads to very interesting phe-
nomena if the objects studied are exposed to very intense electromagnetic radiation
which can be generated without difficulties by today’s short pulse lasers.
eik·r −ieik·r
D=
p·e= ∇·e (H.22)
ωme ωme
(with the dimension of a force MLT−2 ). For electromagnetic waves in the IR, VIS,
UV, and VUV spectral range the wavelength is usually very large compared to the
objects studied. We thus can assume k · r 1 and expand the exponential function
in (H.22):
1
D= (1 + ik · r + · · · )
p · e. (H.24)
ωme
In the electric dipole approximation (short: dipole approximation) only the first
term is taken into account and one speaks of E1 transitions. Thus, the dipole transi-
tion matrix elements of D between two eigenstates |a and |b is:
1
Dab =
a|
p |b · e = −i a|∇|b · e = i
a|r|b · e. (H.25)
ωba me ωba me
For the last equality we have made use of (H.19). Strictly mathematical, both forms
of
Dab are completely equivalent – if the wave functions used are exact solutions of
636 H Vector Potential, Dipole Approximation, Oscillator Strength
the unperturbed Hamiltonian. However, since we usually know the wave functions
only approximatively (except for some special cases such as the H atom) both for-
mulas (H.26) and (H.27) lead to slightly different results. Both variants are used in
the literature: they are called dipole velocity approximation (since
v = p /me ), and
the dipole length approximation, respectively.
The matrix elements of the interaction potential (H.21) are thus
i e 2I (ω)
ab (t) =
a|U
U (r, t)|b = −
a|∇|b · ee−iωt + e∗ e+iωt (H.26)
2 ωme cε0
N N
Dab = −i
a| ∇ i |b · e = i
a| r i |b · e. (H.29)
ωba me
i=1 i=1
The first and second equality refer to the dipole velocity and dipole length form
of the transition operator, respectively. They are identical if (and only if) the wave
functions are exact. The sum has to be carried out in principle over all electrons i,
with N being the total number of electrons in the system.
Several, slightly differing, quantities are used in the literature to characterize the
overall strength of a dipole transition between levels a and b with angular mo-
mentum quantum numbers ja ma and jb mb , and degeneracies ga = 2ja + 1 and
H.2 Line Strength and Oscillator Strength 637
gb = 2jb + 1, respectively. In principle, j may refer to the orbital (L) or to the to-
tal angular momentum (J ), whichever is appropriate. Note that we abbreviate the
quantum numbers γa ja ma := a, but we also use a and b for designating the respec-
tive energy levels – as long as no confusion can arise.
We essentially3 follow C ONDON and S HORTLEY (1951) and define the line
strength symmetric in respect of initial and final energy levels and as a sum over
all polarizations:
S(jb ja ) ≡ S(ja jb ) := |r ba |2 ≡
γb jb mb |r|γa ja ma
2
mb ma mb ma
(H.30)
=
b|x|a
2 +
b|y|a
2 +
b|z|a
2 .
mb ma
Using the W IGNER -E CKART theorem in the form (4.82) and the orthogonality re-
lations of the 3j symbols (B.42) we obtain the compact relations:4
2
S(jb ja ) = (2jb + 1)
γb |r|γa
jb C1 ja 2
2 (H.32)
= (2ja + 1)
γa |r|γb
ja C1 jb 2 ≡ S(ja jb ).
Ja La SC1 Jb Lb S2 using (C.46). Exploiting the orthogonality relation (B.67) for
the 6j symbols one obtains the relation
S(Jb Ja ) S(Lb La )
, (H.33)
2Jb + 1 2Lb + 1
Ja
where the equality sign holds in so far as the radial matrix element does not depend
on the individual fine structure levels J but only on the orbital quantum numbers L
and n – which is a good approximation for light atoms.
Here Dba is the dipole transition matrix element (H.29). In the second equality we
use explicitly the dipole length form for a single electron system, with r ba as de-
fined by (4.79). The oscillator strength is thus proportional to the transition proba-
bility (4.65). And since it is also proportional to the absorbed (or emitted) photon
energy Wb − Wa = Wba = ωba , its value is positive for absorption and negative for
emission.
For the particular case of linear polarization with elin z this gives a simple ex-
pression for the oscillator strength:
(opt) Wba
zba
2 me ωba
fjb mb ←ja ma =2 =2 |zba |2 . (H.35)
E h
a0
It may be further evaluated with (4.75) and the W IGNER -E CKART theorem in the
form (4.82):
2
|zba |2 =
γb jb mb |r0 |γa ja ma
2 (H.36)
2 ja 1 jb
= (2jb + 1)
γb |r|γa
jb C1 ja 2 .
−ma 0 mb
2 Wba
2
2
1
= |aq |2
γb |r|γa
jb mb |C1q |ja ma
. (H.38)
ga E h m m
a b q=−1
H.2 Line Strength and Oscillator Strength 639
Using (4.82), the orthogonality relation (B.42) for the 3j symbols, and q |aq |2 = 1
(unit vector e) the sums can be carried out:
2
Wba gb
r
1
1
(opt)
fjb ja =2
γ | |γ
jb C1 ja 2 |aq |2 (H.39)
E h ga
b a
a0 3
q=−1
2
(opt) 2 Wba gb
r
fjb ja =
γ b | |γ a
jb C1 ja .
2
(H.40)
3 E h ga a0
where we have exploited that here q = 0, and thus only contributions from ma = mb
arise. For an initial singly degenerate ns state (H.42) is reduced to (H.35). Equivalent
expressions are derived from (H.41) with (H.30) and the dipole moment D = −er:
2me ωba
2
γb jb mb |r|γa ja ma
(opt)
fjb ja = (H.43)
3ga m m
2
a b
2
2 Wba
r
8π 2 me Wba
=
γ j m
b b b | |γ a a a
=
j m |D ba |2 .
3ga Eh m m a0 3ga h2 e2 m m
a b a b
(H.44)
Note the factor 3 in the denominator in comparison to (H.42), compensating for the
equal contributions from the components of |r|2 = |x|2 + |y|2 + |z|2 .
We sum the oscillator strength according to (H.35) over all existing final states |b =
|γb jb mb and rewrite the expression suitably:
640 H Vector Potential, Dipole Approximation, Oscillator Strength
(opt)
2me ωba 2me ωba
fjb mb ←ja ma = |zba |2 =
a|z|b
b|z|a (H.45)
b b b
1
= ime ωba
a|z|b
b|z|a − ime ωab
a|z|b
b|z|a .
i
b
fjb mb ←ja ma =
a|z|b
b|pz |a −
a|pz |b
b|z|a
i
b b
(H.46)
1
a|zpz − pz z|a
=
a|zpz |a −
a|pz z|a = = 1,
i i
where we have exploited completeness 1 = b |b
b|. The last step realizes the
commutation rules (H.15) and normalization
a|a = 1.
We note that this derivation does not depend on the specific initial substate |a =
|ja ma since summation is always over all final states |b. Hence, (H.46) is also valid
(opt)
for the average oscillator strength fjb ja . Thus, we have derived the very important
T HOMAS -R EICHE -K UHN sum rule:
(opt)
fjb ja = 1. (H.47)
b
The oscillator strength is often used in atomic and molecular physics to characterize
dipole transitions. It allows one to compare the strength of different transitions in
(opt)
different atoms or molecules. For systems with one active electron fjb ja ≤ 1 strictly
holds, the classical reference being an oscillating electron which has an oscillator
strength = 1.
For systems with Ne active electrons one has to replace in (H.45) the single
e (i)
electron coordinate z by N i=1 z and obtains the sum rule:
(opt)
fjb ja = Ne . (H.48)
b
Finally, we point out that the summation just discussed must include the ion-
ization continuum if the basis is to be complete. In this context, continuum states
are usually normalized in the energy scale (see Appendix J). Correspondingly one
defines the optical oscillator strength density (OOSD), df(opt) /d, with the dimen-
sion Enrg−1 , where is the energy of the emitted electron in the continuum. Thus,
(H.48) includes summation over discrete states and integration for energies beyond
the ionization potential WI :
discrete ∞ (opt)
dfja
(opt) (opt)
fjb ja → fjb ja + d. (H.49)
WI d
b b
References
C ONDON, E. U. and G. S HORTLEY: 1951. The Theory of Atomic Spectra. Cambridge, England:
Cambridge University Press, 441 pages.
ISO 21348: 2007. ‘Space environment (natural and artificial) – process for determining solar
irradiances’. International Organization for Standardization, Geneva, Switzerland.
FOURIER Transforms and Spectral Distributions
of Light I
Various, slightly different notations can be found in the literature. We use the stan-
dard notations of modern physics with the F OURIER transform1
∞
(
X(ω) = F X(t) = X(t)e−iωt dt (I.2)
−∞
of the time dependent function X(t). Conversely, X(t) is recovered by the inverse
F OURIER transform
∞
1
(
X(t) = F −1 X(ω) = (
X(ω)e iωt
dω. (I.3)
2π −∞
1 In
√
other notations the factor 1/2π is applied symmetrically as 1/2π , or completely avoided by
using frequency ν instead of angular frequency ω = 2πν.
and correspondingly
∞
1 (∗ (ω)e−iωt dt.
X ∗ (t) = X (I.5)
2π −∞
By rewriting (I.1) as
∞ ∞
1
X(t) = X(τ ) e iω(t−τ )
dω dτ
−∞ 2π −∞
one identifies the expression in square brackets as the D IRAC delta function:
∞
1
δ(t − τ ) = F −1 e−iωτ = eiω(t−τ ) dω (I.6)
2π −∞
∞
1 iωc t
1
and equivalently δ(ωc − ω) = F e = e−i(ω−ωc )t dt. (I.7)
2π 2π −∞
= X(ω ( − ωc ) .
( − ωc )eiωt dω = F −1 X(ω (I.9)
2π −∞
Thus, the carrier frequency just shifts the F OURIER transform F[X(t) exp(iωc t)]
towards positive frequencies in respect of F[X(t)]. This formally also resolves the
problem one might envisage with negative frequencies in the spectral distribution
(
X(ω): since for any physically relevant distribution with a FWHM= γ one expects
ωc γ , there will be no contributions from negative frequencies in (I.2). An ana-
logue relation applies to a shift in time.
Another set of useful relations concerns F OURIER transforms of convolutions
(see Appendix G). The proof is similar to that for P LANCHEREL’s theorem.
I.1 Short Summary on FOURIER Transforms 645
∞ ∞
−iωt
F (X ∗ Y )(t) = X t Y t − t dt e dt
−∞ −∞
∞ ∞
−iωt
= X t e dt Y t − t e−iω(t−t ) dt
−∞ −∞
∞ ∞
= X t e−iωt dt Y t e−iωt dt .
−∞ −∞
F X(t)Y (t)
∞ ∞ ∞
1 iω t 1 iω t −iωt
= (
X ω e dω (
Y ω e dω e dt
−∞ 2π −∞ 2π −∞
∞ ∞
1 iω t
= (
X ω e dω
−∞ 2π −∞
∞
1
× ( ω − ω ei(ω −ω )t dω e−iωt dt
Y
2π −∞
∞ ∞ ∞
1 1
= ( ω Y
X ( ω − ω dω dω ei(ω −ω)t dt
2π 2π −∞ −∞ −∞
∞ ∞
1
= δ ω − ω X (ω Y ( ω − ω dω dω
2π −∞ −∞
∞
1
= ( ω Y
X ( ω − ω dω .
2π −∞
( Y
F −1 X(ω) ((ω) = (X ∗ Y )(t). (I.13)
646 I FOURIER Transforms and Spectral Distributions of Light
F
X(ω) = X ∗ (τ )X(τ + t)dτ or (I.16)
−∞
2 ∞
(X(ω)
= F X ∗ (τ )X(τ + t)dτ . (I.17)
−∞
±ωc t → ∓k c r ± ωc t .
2 Where necessary, we shall simply reintroduce the wave vector by replacing
Similarly, we keep the r dependence of the envelope in the back of our mind and set E ± (t) →
E ± (r, t) when required.
I.2 How Electromagnetic Fields are Written 647
3 Even (+ (ω))∗
= E
though (E + (t))∗ = E − (t), note that (E (− (ω), since
∞ ∗ ∞
+ ∗ + ∗ iωt
( (ω) =
E E + (t)e−iωt dt = E (t) e dt
−∞ −∞
∞
= (− (−ω).
E − (t)eiωt dt = E
−∞
648 I FOURIER Transforms and Spectral Distributions of Light
(
Note that in this terminology E0 (t) is a real quantity. E(ω) will also be real if and
only if E0 (t) is symmetric around its maximum, while E (+ (ω) is usually complex.
However, with these definitions E (+ (ω) typically has a pronounced resonance at
ω > 0 so that contributions from negative frequencies to the inverse FT (I.22) can
safely be neglected. In other words: artificial truncations of the limits of integrals
are not needed, since anyhow only positive frequencies contribute when working
with E(+ (ω). These relationships will be illustrated below for specific examples.
If one inserts E ± (t) according to (I.22) into expression (I.19) for the cycle averaged
intensity and integrates over all times, one recovers P LANCHEREL’s theorem (I.8)
∞ ∞ ∞
ε0 c 1
+
2
I (t)dt =
E (ω) dω = I˜(ω)dω = F, (I.25)
−∞ 2 2π −∞ −∞
leading to the fluence F (i.e. the total pulse energy Wtot per unit area). With the
second equality we have defined the spectral intensity distribution:
ε0 c
(+
2 ε0 c
(
2
I˜(ω) = E (ω) = E(ω − ωc )
. (I.26)
4π 4π
The assumption that this indeed describes the spectrum of the radiation is by
no means as trivial as commonly assumed. Note, that this expression is not the
F OURIER transform of the intensity I (t)!
One may, however, rationalize this definition by remembering how optical (or
other) spectra of electromagnetic radiation are measured: Assume e.g. an interfero-
metric measurement (FABRY-P ÉROT interferometer, diffraction grating etc.), where
the signal S is generated by superposition of two (or many) amplitudes. A typical
interference pattern emerges from, say two wave fields E + (t) and E + (t + δ), which
have traversed different optical path length cδ (corresponding to a delay time δ). The
signal is then
S ∝ E − (t) + E − (t + δ) E + (t) + E + (t + δ) (I.27)
and only the interference term [in square brackets] is relevant for the measurement
of the spectrum. If such a measurement is made with short pulses, one obviously
integrates over all times (or sufficiently long times), thus an autocorrelation function
of the field is recorded. To obtain the spectrum one has to F OURIER transform this
signal which according to (I.17) leads to
1
(+
2 ˜
F E − (t) E + (t) = E (ω) ∝ I (ω).
2π
I.4 Examples 649
If a prism spectrograph is used a similar argument can be based on the wave optical
interpretation of the index of refraction.
Hence, the definition of the spectrum according (I.26) is justified. Proper normal-
ization is provided by the P LANCHEREL theorem in the form (I.25), which simply
states energy conservation: the fluence F is independent of whether we integrate in-
tensity over all times or spectral intensity distribution over all angular frequencies.
The units are [F ] = J m−2 , [I (t)] = W m−2 , and [I˜(ω)] = J s m−2 , respectively.
It is important to note that the above use of P LANCHEREL’s theorem is only pos-
sible for square integrable functions, i.e. for pulses of electromagnetic radiation with
finite duration. If we want to describe CW light sources the integration over all times
has to be replaced by an average over a sufficiently long time Tav . P LANCHEREL’s
theorem (I.25) must then be replaced by
Tav /2 ∞
1 ε0 c Tav /2
+
2
I (t)dt = E (t) dt = I˘(ω)dω = I, (I.28)
Tav −Tav /2 2Tav −Tav /2 −∞
where I˘(ω) is now the spectral intensity of the ensemble, measured in units [I˘(ω)] =
W s m−2 , while I is the average intensity with [I ] = W m−2 . More about station-
ary, quasi-monochromatic or chaotic light sources, their spectra and their coherence
properties will be found in Sect. 2.1, Vol. 2.
I.4 Examples
In the following we shall present the most commonly used temporal and spectral
profiles of light pulses. The F OURIER transforms collected here can be derived
by analytic integration, where necessary with the help of a suitable computer pro-
gramme (e.g. SWP 5.5, 2005). They are also found in standard textbooks or at
the web (e.g. W EISSTEIN, 2012; W IKIPEDIA CONTRIBUTORS, 2014). We write the
field envelope as
E0 (t) = E0 h t/|t0 | with h(0) = 1 and (I.29)
ε0 c 2
I0 = E (I.30)
2 0
where E0 is the maximum field amplitude, |t0 | a characteristic width, and I0 the
cycle averaged maximum intensity.
We point out that the spectral intensity distribution as defined in (I.26) refers to a
single pulse. It has the dimension [I˜(ω)] = J s m−2 . Integrated over all frequencies
it gives the fluence F per pulse, i.e. the total pulse energy per unit area:
∞ ∞
F= I (t)dt = I0 h2 (t/t0 )dt. (I.31)
−∞ −∞
are characterized by individual time constants τi and their individual spectra I˜i (ω).
One then has to average these spectra and to renormalize in respect of the average
duration τc of the pulses. One obtains an overall spectrum:
1 ∞
I˘(ω) = I˜i (ω) = w(τi )I˜i (ω)dτi . (I.32)
τc 0
This spectrum has now the dimension W s m−2 and the integration over all frequen-
cies gives the average, stationary intensity I of the beam.
Gaussian pulse shapes are probably the most popular ones, mainly due to their con-
venient mathematics. Starting from (I.18) we write the field envelope
2
t
E0 (t) = E0 h(t) = E0 exp − 2 (I.33)
τG
√
with a FWHM= 2 ln 2τG = 1.665τG . The corresponding cycle averaged intensity
(I.19) is now
t2
I (t) = I0 exp −2 2
τG
√
with a FWHM t1/2 = 2 ln 2τG = 1.177τG . (I.34)
2 ln 2
t1/2 ν1/2 = = 0.441. (I.40)
π
Another, quite popular distribution function for describing the time dependence of
short light pulses is the hyperbolic secant
t E0 2E0
E0 (t) = E0 sech = = (I.41)
τs cosh(t/τs ) et/τs + e−t/τs
√
with a FWHM of −2 ln(2 − 3)τs = 2.634τs . The corresponding intensity distribu-
tion is
2
2
I (t) = I0 sech (t/τs ) = I0 t/τ
2
, (I.42)
e s + e−t/τs
√
with a FWHM t1/2 = −2 ln(−1 + 2)τs = 1.763τs . (I.43)
The intensity distribution has the same temporal shape with I0 = ε0 cE02 /2. It is
convenient to directly evaluate the F OURIER transform (I.24) of Ei+ (t):
t0 +τr
(+ (ω) = E0
E exp i(ωc t − ωc ti − φi − ωt) dt
i
t0
exp[i(ωc − ω)τi ] − 1
= E0 τr exp −i(ωti + φi ) (I.51)
i(ωc − ω)τi
The FWHM of I˜i (ω) is given by ω1/2 = 5.566/τi or in frequency space ν1/2 =
0.886/τi so that the time-bandwidth product is
Another important case is a rectangular spectrum. Such a situation may arise, e.g.
if one spectrally filters a broad band light CW source with a sharp band-pass filter
(e.g. in stellar interferometry). The spectrum would then be
I.4 Examples 653
I /ω for ωc − 12 ω < ω < ωc + 12 ω
I˘(ω) = (I.55)
0 else,
where I is the average stationary intensity of the beam. The integral (I.3) can easily
be carried out in this case and one obtains the inverse F OURIER transform:
I iωc t ωt sin x
F −1 I˘(ω) = e sinc with sinc(x) = . (I.56)
2π 2 x
This formula turns out to be useful e.g. for understanding stellar interference spec-
troscopy (see Sect. 2.1.4, Vol. 2).
0 if x < 0
h(x) = (I.57)
exp(−x) if x ≥ 0
is used to describes an exponentially decaying field (oscillating at a carrier fre-
quency ωc ) and its cycle averaged intensity, respectively:
t
E(t) = E0 h and
2τe
t
I (t) = I0 h with FWHM t1/2 = τe ln 2 = 0.693τe . (I.58)
τe
( 2τe E0
E(ω) = , and
1 + 2iτe ω
1 I0 τe2 I0 γ 2 /4
I˜(ω) = = , (I.59)
π 1 + [2τe (ω − ωc )]2 πγ 2 γ 2 /4 + (ω − ωc )2
with a FWHM ω1/2 = γ = 1/τe . (I.60)
654 I FOURIER Transforms and Spectral Distributions of Light
In frequency space this gives ν1/2 = 1/(2τe π) = 0.159/τe , so that the time-
bandwidth product becomes
( 4τee E0
E(ω) = , or alternatively (I.66)
1 + (2τee ω)2
(+ (ω) = E0 e−iφ γ
E , (I.67)
γ 2 /4 + (ω − ωc )2
with an FWHM γ = 1/τee . The intensity spectrum is given by
2
˜ I0 γ
I (ω) = (I.68)
2π γ 2 /4 + (ω − ωc )2
√
with a FWHMω1/2 = 2 − 1γ = 0.644γ = 0.643/τee , (I.69)
We point out that (I.68) is not a L ORENTZ profile, and may, e.g. not easily be convo-
luted. We emphasize that in any case this is not a typical profile for characterizing a
real light pulse: it has a cusp at time zero and extremely wide tails, both in the time
and in the frequency domain.
I.5 Fourier Transform in Three Dimensions 655
&∞
with −∞ L̃(ω, γ )dω = 1. The inverse FT is given by (I.63) as
∞
1 1 −γ |t|/2+iωc t
F −1 L̃(ω, γ ) = L̃(ω, γ )eiωt dω = e (I.72)
2π −∞ 2π
with γ = 1/τee . We have set here φ = 0.
We finally can now supplement the proof for the additivity of linewidths when
convoluting two L ORENTZ profiles, as introduced by (G.21). According to (I.12) the
inverse FT of a convolution of two L ORENTZ profiles is proportional to the products
of their inverse FT:
γ = γ1 + γ2 . (I.75)
δ(x − x )
δ x − x = 2 δ θ − θ δ ϕ − ϕ . (I.81)
x sin θ
For problems with some kind of spherical symmetry (e.g. in atomic physics)
it is useful to rewrite (I.76) in spherical polar coordinates, with standard position
coordinates (r, θr , ϕr ) and coordinates in wave vector space (k, θk , ϕk ):
2π π ∞
( θk , ϕk ) =
X(k, X(r, θr , ϕr )e−ikr r 2 sin θr drdθr dϕr . (I.82)
0 0 0
∞
X(r) = X(r, θr , ϕr ) = Xm (r)Ym (θr , ϕr ), where (I.83)
=0 m=−
2π π
∗
Xm (r) = X(r, θr , ϕr )Ym (θr , ϕr ) sin θr dθr dϕr . (I.84)
0 0
By expanding also the plane wave (I.82) according to (J.13) one obtains the 3D
F OURIER transform in k space
∞
( = X(k,
X(k) ( θk , ϕk ) = 4π (−i) X m (k)Ym (θk , ϕk ), (I.85)
=0 m=−
Acronyms and Terminology 657
References
BADDOUR, N.: 2009. ‘Operational and convolution properties of two-dimensional Fourier trans-
forms in polar coordinates’. J. Opt. Soc. Am. A, 26, 1767–1777.
BADDOUR, N.: 2010. ‘Operational and convolution properties of three-dimensional Fourier trans-
forms in spherical polar coordinates’. J. Opt. Soc. Am. A, 27, 2144–2155.
SWP 5.5: 2005. ‘Scientific work place’, Poulsbo, WA 98370-7370, USA: MacKichan Software,
Inc. http://www.mackichan.com/, accessed: 9 Jan 2014.
W EISSTEIN, E. W.: 2012. ‘Fourier transform’, Wolfram Research, Inc., Champaign, IL, USA.
http://mathworld.wolfram.com/FourierTransform.html, accessed: 9 Jan 2014.
W IKIPEDIA CONTRIBUTORS: 2014. ‘Fourier transform’, Wikipedia, The Free Encyclopedia.
http://en.wikipedia.org/wiki/Fourier_transform, accessed: 9 Jan 2014.
Continuum
J
In the following we use the radial functions in the form u(r) = rR(r). In the con-
tinuous spectrum, following B ETHE and S ALPETER (1957) Eq. (4.11ff), they are
normalized in T scale by demanding
∞ T +T
druT (r) uT (r)dT = 1. (J.2)
0 T −T
Here T (k) is any function of the electron wavenumber k in the continuum, possibly
k itself. In the theoretical treatment of photoionization normalization in scale is
frequently used, with being the energy of the electron in the continuum. The in-
tegration over dT needs only to be over a small interval 2T around T , since all
other contributions cancel.1
With such normalized continuum functions one may represent any wave function
including the continuum as partial wave expansion (for simplicity written here only
for one electron systems):
1 Summation of (J.1) over all n states in the discrete spectrum would be the equivalent to this second
integration in (J.2). Since only one state contributes, the result is also =1.
ϕ(r, θ, ϕ) (J.3)
) ∞ *
1 ∞
= Ym (θ, ϕ) anm un (r) + dT (k)am (T )uT (r) .
r k=0
m n=+1
d/dk = k. (J.7)
To determine the dimension of uk (r), we write the units in the expression under
the integral in (J.2) [udrdk] = 1, or as dimensional equation u2 L1 L−1 = 1 so that
u = 1, i.e. uk (r) in k scale normalization is dimensionless. For clarity it might be
useful to explicitly write out the a.u., i.e. to replace r → r/a0 and Z/k → Z/(ka0 )
in (J.9), so that the expression becomes independent of the units used.
For normalization in scale we use (J.4)–(J.7) and obtain in a.u.:
2 dk Z
u (r) = sin kr − π/2 + ln(2kr) + σ + δ
π d k
2 Z
= sin kr − π/2 + ln(2kr) + σ + δ or explicitly (J.10)
πk k
2 me Eh /2 1/4 Z
= sin kr − π/2 + ln(2kr) + σ + δ .
πEh a0 M̄ ka0
(J.11)
The latter expression is independent of the units used. The prefactor (Eh a0 )−1/2
is again obtained from a dimensional analysis: normalized in scale from (J.2)
we have [u2 drd] = 1, so that u2 = Enrg−1 L−1 and u = Enrg−1/2 L−1/2 . Note that
for electron scattering the mass factor me /M̄ 1 disappears. Different authors use
slightly different notations and energy units (often without mentioning which).2
∞
∗
ψ (k) (r) = eik·r = 4π i j (kr) Ym (θk , ϕk )Ym (θr , ϕr ), (J.13)
=0 m=−
Here, ε = 2/Eh , i.e. energies are measured in RYDBERG units = Eh /2, lengths in a0 .
662 J Continuum
with the polar coordinates (r, θr , ϕr ) representing r, while (k, θk , ϕk ) gives the mag-
nitude and direction of the wave vector k, and j (kr) = u (kr)/(kr) are spherical
B ESSEL functions. They are solutions of the radial S CHRÖDINGER equation (2.110)
for vanishing potential:
2
1 d2 u k ( + 1)
− + u (r) = 0. (J.14)
2 dr 2 2 2r 2
As one easily verifies, the most simple ones are:
sin x sin x cos x
j0 (x) = sinc x = and j1 (x) = − . (J.15)
x x2 x
All others may in principle be derived from the recursion formula:
2 + 1
j+1 (x) = j (x) − j−1 (x). (J.16)
x
Asymptotically the following relations hold:
x /[(2 + 1)(2 − 1)(2 − 3) · · · ] for x
j (x) = . (J.17)
sin(x − π/2)/x for x
With the addition theorem (C.22) and the angle γ between k and r one may also
write (J.13) as:
∞
ψ (k) (r) = eik·r = (2 + 1)i j (kr)P (cos γ ). (J.18)
=0
The probability density |ψ (k) |2 of the plane waves (J.13) or (J.18) is given per
wavenumber to the third power (dimension L3 ) and per volume (L−3 ), so they are
overall dimensionless. They may be normalized in k scale (see e.g. OVCHINNIKOV
et al., 2004):
1
((k) (r) =
ψ eik·r =
r|k. (J.19)
(2π)3/2
To verify this normalization we first integrate over position space using (I.78):
k 2 |k 1 k = ψ((k)∗ (r)ψ
((k) (r)d3 r a03 δ ka0 − k a0 .
2 1
of states. Finally, with (I.80), integration over k spaces confirms the correct normal-
ization of (J.19):
3
k
k k d k = a03 δ ka0 − k a0 d3 k ≡ 1.
holds, and the 3D plane wave (J.19) normalized in energy scale becomes
ka0 ik·r
ψ( (r) = (2π)
(k) −3/2
e . (J.22)
a03 Eh
((k) (r)|2 is
Its dimension is now Enrg−1/2 L−3/2 so that the probability density |ψ
correctly given per energy and volume.
References
B ETHE, H. A. and E. E. S ALPETER: 1957. Quantum Mechanis of One- and Two-Electron Atoms.
Berlin, Göttingen, Heidelberg: Springer Verlag, 369 pages.
B URKE, P.: 2006. ‘Electron-atom, electron-ion and electron-molecule collisions’. In: G. W. F.
D RAKE, ed., ‘Handbook of Atomic, Molecular and Optical Physics’, 705–729. Heidelberg, New
York: Springer.
C OOPER, J. W.: 1962. ‘Photoionization from outer atomic subshells. A model study’. Phys. Rev.,
128, 681–693.
OVCHINNIKOV, S. Y., G. N. O GURTSOV, J. H. M ACEK and Y. S. G ORDEEV: 2004. ‘Dynamics of
ionization in atomic collisions’. Phys. Rep., 389, 119–159.
Index of Volume 1
theorem for F OURIER transform, 645 Density of states, 21–24, 238, 240
VOIGT profile, 628, 629 particle in a box, 105, 106
C OOPER minimum, 260, 529 Detailed balance
Coordinate rotation, 575, 605–611 principle of, 184
exercise, 207 Detuning, 229
Coordinate system Diamagnetism, 396–398
atomic vs. photon, 174 Dielectric function, 57
cartesic and polar, 108 Diffraction
electron vs. nucleus, 295 D EBYE-S CHERRER, 58, 59
Core electron, 148, 151 experiment, 60
Core potential, 149 He scattering, 60
alkali atoms, 165 image, 42
Correlation function LEED, 59
1st order, 625 Dimensional analysis, 119, 192, 554, 556, 661
Correspondence principle, 127 Dipol vector, 183
C OULOMB gauge, 631 Dipole approximation, 189, 631–640
C OULOMB integral, 356, 359, 365, 506 electric, 635
C OULOMB law, 50, 59 magnetic, 250–254
C OULOMB potential, 119, 121, 123 Dipole excitation, linear combination of states,
H atom, 121 217–225
screened, 148, 149 Dipole matrix element, 193
C OULOMB wave, 262 length approximation, 193, 636
Coupling
velocity approximation, 193, 636
jj , 301, 354, 360
Dipole operator, 594
LS, 301, 307, 354, 360, 361
magnetic field, 386
LS vs. jj , 512–514
multi-electron system, 264, 363
RUSSEL -S AUNDERS, 301, 354, 360, 361,
Dipole oscillator, classical, 182
508
Dipole radiation
break down, 362
angular characteristic, 203–212
spin-orbit, see Spin-orbit
Crystal lattice, 40 Dipole transition
C URIE constant, 396 amplitude, 193
C URIE’s law, 396 E1 transitions in the H atom, 201–203
Cyclotron frequency, 54, 332 in He, 362–365
electron, 79 length approximation, 636
Cylindrical capacitor, 52 matrix element, 190, 635
operator, 190, 193, 263
D selection rules for E1 transitions,
Damping constant, 229 196–203
DARWIN term, 305 selection rules for E1-transition, 202
D E B ROGLIE, 57, 89 velocity approximation, 636
wavelength, 57–59 D IRAC delta function, 644
Decay constant, 15 D IRAC equation, 79, 93, 296, 303, 333
Decay rate, 15 Direct product, 578
Degeneracy, 21, 22 Dispersion, 421
, 123 anomalous, 422
m, 111 close to resonance, 232, 233
, removal of, 137–146 normal, 420, 422
m, removal of, 130, 131 Dispersion relation, 131–134
Degree of coherence matter waves, 94
1st order, 625 Displaced terms
Degrees of freedom, 19 alkaline earth term schemes, 372
Density distribution, 125 C atom, 515
Density functional theory, 510–512 Ne atom, 518
668 Index of Volume 1
319, 342, 347, 349, 351, 371, 373, Nuclear quadrupole moment, 447, 449, 616
431, 514, 522, 525, 529, 551, 629 oblate or prolate, 478
NMR spectroscopy, 487–491 Nuclear radius, 29, 121
apparatus, 488 liquid drop model, 476
CW spectrum of ethanol, 488 Nuclear spin, 449
magic angle spinning (MAS), 490 eigenvalue equations, 449
occupation probability of levels, 489 Nuclear spin resonance, see NMR
N OBEL prize in chemistry Nucleons, 48
Richard R. E RNST (1991), 490
KOHN and P OPLE (1998), 510 O
F ENN, TANAKA, W ÜTHRICH (2002), 490 O atom, G ROTRIAN diagram, 516
N OBEL prize in physics Oblate, 477, 478, 619
Wilhelm C. RÖNTGEN (1901), 530 nuclear shape, 478
L ORENTZ and Z EEMAN (1902), 377 Observable, 97–99, 118
Joseph J. T HOMSON (1906), 50 commuting, 101
Albert A. M ICHELSON (1907), 278 non-commuting, 101
Max K. E. L. P LANCK (1918), 31 simultaneous measurement, 100, 101
Johannes S TARK (1919), 399 One electron cyclotron oscillator, 334
Albert E INSTEIN (1921), 27, 31 One particle problem, 117–134
Niels H. D. B OHR (1922), 64 One sided exponential distribution, 653
C ORNELL, K ETTERLE, W IEMAN (1925), One-loop QED effects, 323, 324
25 Operator, 96, 97, 100
Arthur H. C OMPTON (1927), 28 energy, 101
Louis DE B ROGLIE (1929), 57 momentum, 101
Werner K. H EISENBERG (1932), 100 position in space, 101
S CHRÖDINGER and D IRAC (1933), 90 simultaneous measurement, 101
Otto S TERN (1943), 70 Optical path difference, 40
Isidor I. R ABI (1944), 482 Orbital angular momentum, 71, 295, 299,
Wolfgang PAULI (1945), 22, 138 302
Max B ORN (1954), 88, 89 components, 110, 111
L AMB and K USCH (1955), 317 eigenfunctions, 109–113
T OMONAGA, S CHWINGER, F EYNMAN square, 111, 112
(1965), 79, 324, 534 Orbital energies, 509
R AMSEY, D EHMELT, PAUL (1989), 55, Orders of magnitude, 5–9
225, 288, 332, 482 energy scales, 7
G LAUBER, H ALL, H ÄNSCH (2005), 248, length scales, 5
319 time scales, 7
M ATHER, S MOOT (2006), 6, 7 Orthonormality relation, 96
E NGLERT and H IGGS (2013), 46 Oscillator strength, 238, 239, 636–640
N OBEL prize in physiology or medicine density, 256
L AUTERBUR, M ANSFIELD (2003), 490 sum rule, 239, 639, 640
Non-crossing rule, 391, 394
Non-local potential, 507 P
Non-stationary problems Pair production, 526
dipole excitation (E1), 169–225 Paramagnetism, 394–396
light matter interaction, 227–270 Parity, 593, 594
Non-stationary states, 186, 187 conservation in E1 transitions, 202
‘normal’ Z EEMAN effect, 128–131, 382–386 multi-electron systems, 594–603
Nuclear gN factor, 447, 449 Parity violation, 249
Nuclear magnetic moment, 447, 449, 485, 487, Particle detection, 75
488 Particle diffraction, 58–61
Nuclear magnetic resonance, see NMR C60 , 60, 61
Nuclear mass, 119 He atoms, 59, 60
energy correction, 68, 69 Particles and waves, 57, 64
674 Index of Volume 1
Symbols Benzene
12, 6 potential, 149 D6h point group, 262
H ÜCKEL orbitals, 277–285
A β (anisotropy) parameter, 356, 360, 551, 557
A BBE sine law, 679 B ETHE formula, 528–530, 534, 537, 538, 546
Adiabatic representation, 478–480 ionization, 537
A IRY diffraction pattern, 25 B ETHE integral, 526
A IRY disc, 26 B ETHE ridge, 546, 548, 549, 553
Alignment, 596, 604, 616, 619, 698–700 B ETHE surface, 545
angle, 36, 44, 605, 610, 611 Binary peak in (e, 2e) process, 553
parameter, 622, 623, 698 Birefringence, 38, 40, 41
Alkali halide potentials, 220–224 B IRGE -S PONER plot, 177
Ammonia maser, 1, 252 Black body radiation, 105
Amplification profile, 15 B OLTZMANN
Amplified spontaneous emission, 11 distribution, 155, 173, 335, 389, 591
Anharmonicity constant, 162 statistics, 158
Anti-S TOKES lines, 334 Bond order, 195
Anti-symmetrization, 521 Bonding orbital, 183
Antibonding orbital, 183 B ORN approximation
Ar2 TPES spectrum, 369 first order, elastic, 444–448
Arn clusters, 400 generalized oscillator strength (GOS),
Atomic form factor 530–534
inelastic, 527 inelastic collisions, 460, 461, 522, 525–530
AUGER electron, 533 integral inelastic cross sections, 534
AUTLER -T OWNES effect, 637–639 B ORN phase shift, 447
Autocorrelation function, 53–60 B ORN series, 553
Avoided crossing, 220–224, 263 B ORN -O PPENHEIMER approximation,
139–151
B collision processes, 458, 474
Baseline in interferometry, 87, 91 BPP, see Beam parameter product
Beam divergence, 21, 22 Branching ratio, 653
Beam parameter, 17–21
product, 26, 34, 35 C
Beam radius, 17–35, 650 C6 H +
4 absorption spectra, 328
Beam waist, 17–21 C2 DFWN spectrum, 354
and coherence volume, 87 Carrier envelope, 46
and lateral coherence radius, 91 phase, 46, 55, 188