Professional Documents
Culture Documents
Preface
These notes are intended as support material for part II of the course. The quantum mechanics material
can mostly be found in the book Introduction to Quantum Mechanics, second edition by D.J. Griffiths
(Pearson), where in part II the remainder of chapter 7, chapter 8 and chapter 9 are new. Realize, however,
that a thorough understanding of the material in the earlier chapters is essential.
Piet Mulders
February 2016
Contents
1 Introduction
1.1 Basics in quantum mechanics .
1.2 The quantum nature of matter
1.3 Units . . . . . . . . . . . . . . .
1.4 Physics and mathematics . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
101
101
102
102
103
constituents of matter
From matter to atoms: atomic nuclei and electrons . . . . . .
From nuclei to nucleons: protons and neutrons . . . . . . . .
From nucleons to quarks . . . . . . . . . . . . . . . . . . . . .
Standard model of quarks and leptons and their interactions .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
201
201
203
205
206
3 Semiclassical methods
3.1 Bohr quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 WKB in the classical region; Griffiths section 8.1 . . . . . . . . . . . . . . . . . . . . .
3.3 Tunneling; Griffiths section 8.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
301
301
301
302
401
401
402
404
405
2 The
2.1
2.2
2.3
2.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
601
601
603
603
604
A Spin
A.1 Rotational invariance (extended to spinning particles)
A.2 Spin states . . . . . . . . . . . . . . . . . . . . . . . .
A.3 Why is ` integer . . . . . . . . . . . . . . . . . . . . .
A.4 Matrix representations of spin operators . . . . . . . .
A.5 Rotated spin states . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
701
701
702
703
704
704
by atoms
. . . . . .
. . . . . .
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
Introduction
1
1.1
101
Introduction
Basics in quantum mechanics
At this point, you should be familiar with the basic aspects of quantum mechanics. That means you
should be familiar with working with operators, in particular position and momentum operators that do
not commute, but satisfy the basic commutation relation
[ri , pj ] = i~ ij .
(1)
The most common way of working with these operators is in an explicit Hilbert space of square integrable
(complex) wave functions (r, t) in which operators just produce new functions ( 0 = O). The
position operator produces a new function by just multiplication with the position (argument) itself. The
momentum operator acts as a derivative, p = i~ , with the appropriate factors such that the operator
is hermitean and the basic commutation relation is satisfied. We want to stress at this point the nonobservability of the wave function. It are the operators and their eigenvalues as outcome of measurements
that are relevant. As far as the Hilbert space is concerned, one can work with any appropriate basis. This
can be a finite basis, for example the two spin states along a specified direction for a spin 1/2 particle
or an infinite dimensional basis, for example the position or momentum eigenstates. In general a basis
is a set of eigenstates of one or more operators, denoted by a bra in Dirac representation containing the
(relevant) eigenvalues to label the quantum states. Here the kets must contain a set of good quantum
numbers, i.e. a number of eigenvalues of compatible (commuting) operators.
==========================================================
Question: Why is it essential that the quantum numbers within one ket correspond to eigenvalues of
commuting operators?
==========================================================
The connection with wave functions uses the (formal) eigenstates of position or momentum operators, |ri
or |pi. The coordinate state wave function then is just the overlap of states given by the inner product
in Hilbert space, (r) = hr|i, of which the square gives the probability to find a state |i in the state
|ri. Similarly one has the momentum state wave function, (p)
= hp|i.
Some operators can be constructed from the basic operators such as the angular momentum operator
` = r p with components `i = ijk rj pk . The most important operator in quantum mechanics is the
Hamiltonian. It determines the time evolution to be discussed below. The Hamiltonian H(r, p, s, . . .)
may also contain operators other than those related to space (r and p). These correspond to specific
properties, such as the spin operators, satisfying the commutation relations
[si , sj ] = i~ ijk sk .
(2)
In non-relativistic quantum mechanics all spin properties of systems are independent from spatial properties, which at the operator level means that spin operators commute with the position and momentum
operators. As a reminder, this implies that momenta and spins can be specified simultaneously (compatibility of the operators). The spin states are usually represented as spinors (column vectors) in an
abstract spin-space, which forms a linear space over the complex numbers.
The most stunning feature of quantum mechanics is the possibility of superposition of quantum states.
This property is of course the basic requisite for having a description in terms of a linear space over the
complex numbers C.
Introduction
1.2
102
In these lectures we look at the quantum nature of matter. On the one hand this includes unraveling
matter for its constituents, and the search for the forces between these constituents. But it also includes
understanding the beauty and principles underlying the complexity. Often these are symmetry principles.
Depending on the domains of distances, velocities and energies one is considering one needs to employ
quantum mechanics and/or special relativity.
In quantum mechanics a special role is played by Plancks constant h, usually given divided by 2,
~ h/2
(3)
We already have choosen here two different, often used units for energy, the Joule (J) and the electronvolt
(eV). The first is the formal MKS unit to be used (1 J = 1 kg m2 s2 ), but in many applications in
condensed matter, atomic and molecular physics and subatomic physics one uses the eV, or powers
thereof1 . One eV is the energy obtained or needed when an elementary charge,
e = 1.602 176 462 (63) 1019 C,
(4)
is displaced over a potential difference of 1 V. Quantum effects become unimportant in the limit that the
product of energy time or equivalently momentum distance or angular momentum is much larger
than ~. More precisely formulated, when the action A ~. Then one is in the classical domain.
In special relativity a special role is played by the velocity of light c,
c = 299 792 458 m s1 .
(5)
In the limit that v c one reaches the non-relativistic domain. Schematically one has
Classical Mechanics
A~
vc
Quantum Mechanics
1.3
Special Relativity
vc
A~
Units
The choice of an appropriate set of units is often important, because physical sizes and magnitudes only
acquire a meaning when they are considered in relation to each other. This is true specifically for the
domain of atomic and molecular physics, nuclear physics and high energy physics, where the typical
numbers are difficult to conceive on a macroscopic scale. They are governed by a few fundamental units
and constants, which have been discussed in the previous section, namely ~ and c. In fact one can work
with less units by making use of fundamental constants such as ~ and c. For instance, the quantity c
is nowadays used to define the meter. We could as well have set c = 1. This would mean that one of
the two units, meter or second, is eliminated, e.g. because l/c has the dimension of time, one has 1 m
= 0.33 108 s or eliminating the second one would use that ct has dimension of length and hence 1 s
= 3 108 m. Using both ~ and c, all length, time and energy or mass units then can be expressed in
one unit and powers thereof, for which one can use energy, as shown in the next table. The power of E
determines what is referred to as the canonical dimension d of the quantity.
1 k = kilo = 103 ; M = mega = 106 ; G = giga = 109 ; T = tera = 1012 ; m = milli = 103 ; = micro = 106 ; n = nano
= 109 , p = pico = 1012 , f = femto = 1015 .
Introduction
103
quantity
time t
length l
momentum p
angular momentum `
energy E
mass m
area A
force F
G
velocity v
constructed quantity
t/~
l/(~c)
pc
`/~
E
mc2
A/(~c)2
F ~c
G/(~c5 )
v/c
dimension
(energy)1
(energy)1
(energy)1
(energy)0
(energy)1
(energy)1
(energy)2
(energy)2
(energy)2
(energy)0
d
-1
-1
1
0
1
1
-2
2
-2
0
The most appropriate energy unit depends on the domain of applications, e.g. the eV for atomic physics
the MeV or GeV for nuclear physics and the GeV or TeV for high energy physics. To convert to other
units of length or time we use appropriate combinations of ~ and c, e.g. for lengths
~c = 0.197 326 960 2 (77) GeV fm
(6)
This quantity can of course be used to eliminate the meter if one puts ~ = c = 1,
1 fm = 1015 m 5.068 GeV1 .
(7)
Remembering only two numbers, e.g. c 3 108 m/s and ~c 200 MeV fm 200 eV nm, it is
possible to do the conversions. Often, this can also be used to give reasonable orders of magnitudes.
Depending on the specific situation, of course masses come in that one needs to know or look up. Two
important masses are that of the electron me c2 = 0.510 988 902 (21) MeV and that of the proton mp c2
0.938 271 998 (38) GeV. Furthermore one encounters the strength of the various interactions. In some
cases like the electromagnetic and strong interactions, these can be written as dimensionless quantities,
e.g. for electromagnetism the fine structure constant
=
e2
= 1/137.035 999 76 (50).
4 0 ~c
(8)
For weak interactions and gravity one has quantities with a dimension, e.g. for gravity Newtons constant,
G
= 6.707 (10) 1039 GeV2 .
~c5
(9)
Having many particles, the concept of temperature becomes relevant. A relation with energy is established
via the average energy of a particle being of the order of kT , with the Boltzmann constant given by
k = 1.380 650 3(24) 1023 J/K = 8.617 342(15) 105 eV/K.
1.4
(10)
In order to illustrate the use of units and mathematics, consider as an example the radial Schrodinger
equation for an electron in the hydrogen atom. Considering only the (central) electric potential between
electron and a central charge Ze, one has the hamiltonian
H=
~2 2
Ze2
.
2m
4 r
| {z 0 }
+Vc (r)
(11)
Introduction
104
As should be familiar from elementary quantum mechanics, this hamiltonian and the angular momentum
operators `2 and `z commute. Thus one can use the eigenfunctions of the latter two operators, the
spherical harmonics Y`m (, ) and find that the 3-dimensional problem after writing
(r) =
uE` (r) m
Y` (, ),
r
2m dr2 | 2m r2 {z
}
(12)
(13)
Veff (r)
r0
with boundary condition uE` (0) = 0 or more precisely, uE` (r) r`+1 . First of all it is useful to make
this into a dimensionless differential equation for which we then can use our knowledge of mathematics.
Define = r/a0 with for the time being a0 still unspecified. Multiplying the radial Schrodinger equation
with 2m a20 /~2 we get
"
#
d2
`(` + 1)
e2 2m a0 Z
2m a20 E
2+
uE` () = 0.
(14)
d
2
40 ~2
~2
From this dimensionless equation we find that the coefficient multiplying 1/ is a number. Since we
havent yet specified a0 , this is a good place to do so and one defines the Bohr radius
a0
40 ~2
.
m e2
(15)
The stuff in the last term in the equation multiplying E must be of the form 1/energy. One defines the
Rydberg energy
~2
e2
1
m e4
ER =
=
=
.
(16)
2m a20
2 40 a0
32 2 20 ~2
One then obtains the dimensionless equation
"
#
d2
`(` + 1) 2Z
2+
u` () = 0
d
2
(17)
mc
~2
1
~c
1
ER =
=
a0
(18)
(19)
One thing to be noticed is that the defining expressions for a0 and ER involve the electromagnetic
charge e/ 0 and Plancks constant ~, but it does not involve c. The hydrogen atom invokes quantum
mechanics, but not relativity! To evaluate the expressions using our unit-analysis one of course can
2 Note that one often encounters the radial wave function R (r) = u (r)/r. The advantaage of working with u
n`
n`
n` is
that it satisfies precisely the one-dimensional Schr
odinger equation (with a boundary condition).
Introduction
105
introduce c afterwards in making estimates. Secondly the nonrelativistic nature of the hydrogen atom is
confirmed in the characteristic energy scale being ER . From Eq. 19 we see using = 1/137 that it is of
the order 104 105 of the restenergy of the electron, i.e. very tiny!
Next, we can turn to an algebraic manipulation program or a mathematical handbook to look for the
solutions of our dimensionless differential equation, which turn out to be Laguerre polynomials. We see
from this treatment that (using p n ` 1, a 2` + 1 and x 2Z/n) the solutions for hydrogen
are
1/2 s
`+1
(n ` 1)! Z/n 2Z
2Z
2Z
L2`+1
un` () =
e
(20)
n`1
n a0
2n (n + `)!
n
n
with eigenvalues (energies)
Z2
ER ,
(21)
n2
labeled by a principal quantum number number n, choosen such that the energy only depends on n. For
a given ` one has n ` + 1. Actually nr = n ` 1 is the number of nodes in the wave function.
En` =
Introduction
106
Exercises
Exc. 1.1: fundamental constants and units
(a) The (quantum mechanical) size of the hydrogen atom is of the order of the Bohr radius a0 . Reexpress this quantity in terms of the electron Compton wavelength
e = ~/me c and the fine structure
constant . Similarly express the (relativistic) classical radius of the electron, re = e2 /40 me c2
in the Compton wavelength and the fine structure constant.
(b) Calculate the Compton wavelength of the electron and the quantities under (a) using the value of
~c, and me c2 . This demonstrates how a careful use of units can save a lot of work. One does not
need to know ~, c, 0 , me , e, but only appropriate combinations.
(c) Use the value of the gravitational constant G/~c5 to construct a mass Mpl (Planck mass) and a
corresponding length rpl and give the value of the latter.
(d) Consider photons with wavelength of 500 nm. Calculate the frequency in Hz, the energy in J and
in eV and the wavenumber in cm1 (Recall the relations = c/, E = h). Calculate the Rydberg
energy in cm1 .
(solution)
(a-b) The basic connecting length is
~c
G
~
and rpl =
=
Mpl c
~G
.
c3
Numerically Mpl = 1.22 1019 GeV/c2 0.02 mg and rpl 1.6 1035 m.
in electronvolt per Tesla (eV/T ). You dont have to go to notes or books on electromagnetism, just
look at the expression and argue what must be the unit V/T in the SI system of units (kg, m, s, J =
kg m2 /s2 , etc.).
(solution)
15
Note that ~/m =
c, hence only using
m and c = 3 108 m/s, one gets
e = 386 x 10
e~
= e
e c = 5.788 105 e m2 /s = 5.788 105 eV/T,
2me
e~
p =
= 3.152 108 eV/T.
2mp
e =
Introduction
107
u(r) = E u(r),
2m dr2
2m r2
4 0 r
which writing = r/r0 can be rewritten as
2
~
d2
`(` + 1)
Z e2 1
u() = E u(),
2mr02
d 2
2
4 0 r0
Equating the factor multiplying dimensionless quantities fixes r0 . It is actually common to equate ~2 /mr02
to Ze2 /40 r0 , giving
r0 =
4 0 ~2
= a0 /Z,
Z me2
E0 =
~2
Z 2 e4 m
= Z 2 ER .
=
2
2mr0
32 2 20 ~2
u() = E u()
d
2
~2 `(` + 1)
~2 d2
+
+ T0 r.
2
2m dr
2m r2
What would be the magnitudes of characteristic lengths and energies for a quark-antiquark state
with masses of 300 MeV/c2 and an string tension T0 = 1 GeV/fm.
(solution)
(a) Harmonic oscillator: introducing = r leads to radial equation
2 2
~
d2
`(` + 1)
m 2 2
2+
+
u() = E u().
2m
d
2
2 2
To find a form for = 1/r0 one can equate the multiplicative factors in both terms (they multiply
dimensionless quantities), ~2 2 /m = m 2 /2 which gives 4 = m/~, leading to characteristic
length r0 = (~/m)1/4 and
~
d2
`(` + 1)
2
2+
+
u() = E u().
2
d
2
Note that the radial equation applies to the radial wave function with u(r) = 0. The simplest
solution (for ` = 0) to this equation is
u() = e
/2
Introduction
108
u() = E u().
2m
d
2
Equating the factors gives 3 = 2mT0 /~2 or r0 = (~2 /2mT0 )1/3 and
~2 T02
2m
1/3
`(` + 1)
d2
+
u() = E u().
2+
d
2
The simplest solution to this equstion is a shifted Airy function u() = Ai( ). The Airy
functions is actually the solution of Ai00 (z) z Ai(z) = 0 with zeros on the negative axis, the first
one at = 2.3381. The solution u() = Ai( ) is a solution that vanishes (as it should) at
= r = 0 with energy E = E0 , where E0 = (~2 T02 /2m)1/3 . The Airy function also pops up in the
semi-classical treatment of quantum mechanics problems (see Griffiths Chapter 8, Figure 8.8).
For the situation of a quark and an antiquark with masses of 300 MeV/c2 0.3 GeV and T0 = 1
GeV/fm 0.2 GeV2 we get (using the reduced mass m = 0.15 GeV),
1/3
1/3
1
1
2.5 GeV1 0.5 fm,
2mT0
0.06
1/3
2 1/3
0.04
T0
0.5 GeV.
E0
2m
0.3
r0
(solution)
Problem 7.7, taken from Griffiths. You have found that hHi > 13.6eV, which would appear to
indicate that there is no bound state at all, since it would be energetically favorable for one electron to
fly off, leaving behind a neutral hydrogen atom. This is not entirely surprising, since the electrons are
less strongly attracted to the nucleus than they are in helium, and the electron repulsion tends to break
the atom apart. However, it turns out to be incorrect. With a more sophisticated trial wavefunction (see
Griffiths, Problem 7.18) it can be shown that Egs < 13.6 eV, and hence that a bound state does exist.
Its only barely bound, however, and there are no excited bound states, so H has no discrete spectrum
(all transitions are to and from the continuum). As a result, it is difficult to study in the laboratory,
although it exists in great abundance on the surface of the sun.
201
In this section, we will be unravelling the matter following the pattern from matter/molecules atoms
atomic nuclei nucleons quarks. At present the latter appear as elementary particles characterized
by mass, spin and charge. Along the way also electrons and neutrinos pop up as elementary constituents.
MATTER
ELECTRON
ATOM
10
10
m
NEUTRINO
ATOMIC NUCLEUS
14
10
NUCLEON
proton/neutron
15
10
m
QUARK
up/down
18
< 10
2.1
The basic units of the matter around us are the atoms. They are found as the building blocks of molecules
or solids, bound in a variety of ways discussed in the sections on molecular physics and condensed matter
physics. The atoms are composed of the atomic nucleus with a positive charge +Ze, which is a multiple
of the elementary charge e. The atomic number Z characterizes the atom, Z = 1 for Hydrogen (H), Z = 2
for Helium (He), etc. In the atom the charge of the nucleus is neutralized by Z electrons, bound to the
nucleus via the electromagnetic force. The atoms can be organized in the Periodic Table. Its structure
is determined by the consecutive filling of the energetically lowest available electron orbits and the Pauli
exclusion principle.
# levels
n=3
n=2
n=1
3d
3p
3s
2p
2s
10
6
2
6
2
1s
nl
202
The mass of an atom is in essence determined by the atomic nucleus, consisting of Z protons, each
with a positive charge +e and N neutrons being neutral. Protons and neutrons (together called nucleons)
have similar masses,
mp = 1.672 621 58(13) 1027 kg = 938.271 998(38) MeV/c2 ,
(22)
(23)
The size of the atomic nucleus is tiny, of the order of 10 fm = 1014 m, as compared to the size of the
atom, which is of the order of 1
A= 0.1 nm = 1010 m. The atomic size is determined by the configuration
of the light electrons,
me = 9.109 381 88(72) 1031 kg = 0.510 998 902(21) MeV/c2
(24)
(about a factor 1836 smaller than the proton), orbiting the nucleus. Electrons can be freed from an atom
or additional electrons can be bound, leaving positive or negative ions, leading to ionic bounds. Electrons
can also be shared by atoms in covalent bounds.
For the mass of the atom one has introduced as standard the atomic mass unit (u). It is defined as
1/12 of the mass of the 12 C atom,
1 u = 1.660 538 73(13) 1027 kg = 931.494 013(37) MeV/c2 .
(25)
(26)
This is typically the number of atoms in a macroscopic sample. Without any mutual interactions, this
would lead for the electrons to Nav -fold degenerate levels as compared to a single atom. In solids, the
lifting of this degeneracy because of the e-e interactions and the interactions of electrons with neigboring atoms leads to the band structure. In particular in regularly layered structures such as cristals a
wonderful world of phenomena occurs of which superconductivity is one of the most well-known ones.
Depending on the occupation of these bands, the band gaps, the presence of impurities in the material and the temperature, completely different behavior emerges, e.g. in the conductivity (conductors,
203
semiconductors and isolators). The richness of phenomena ranging from plasmas (a hot but usually neutral collection of free atomic nuclei and electrons) in stars or fusion reactors, the cristalline structure of
matter extending to macroscopic sizes, superconductivity, macromolecules like DNA and all interactions
with light are described with electromagnetic interactions. For this one has a well-developed theoretical
framework. Quantum electrodynamics (QED) is a fully relativistic quantummechanical description for
the interactions between charges via exchange of photons (the light quanta). QED is the basis for obtaining the full (quantummechanical) Hamiltonian used in the Schrodinger equation to calculate the wave
functions of the electrons. Several techniques, ranging from smart approximations to extensive computer
calculations are employed to obtain an effective inter-atomic potential that is used in molecular physics
and chemistry. QED also underlies the Maxwell equations. These can be used to describe classically the
interactions of charges with electric (E) and magnetic (B) fields or at a semiclassical level the interactions
of charges/atoms with light (quantum optics).
2.2
As mentioned already, atoms are characterized by the atomic number Z corresponding to the number of
protons in the nucleus. Often there is more than one possibility for the number of neutrons (N ). Nuclei
with different number of neutrons for a given Z are called different isotopes for the same element. Nuclei
are denoted by the atomic symbol (H, He, etc.) and in order to distinguish the isotopes one adds as
a left superscript the total number of nucleons A = Z + N . One has for Hydrogen 1 H = p, 2 H = pn
(nucleus is called deuteron, atom is called deuterium) and 3 H = pnn (nucleus is called triton, atom is
called tritium). The isotope 1 H is most abundant. The 2 H abundance is only 0.015 %. The third isotope
3
H is not stable. For Helium one has two isotopes, 3 He and 4 He (-particle), the latter being the most
abundant one. The elements Hydrogen and Helium also make up the bulk of the (at least visible) matter
in the universe, roughly in the order 12 : 1 (atoms) or 3 : 1 (mass).
The density of nucleons in a nucleus is roughly constant. A roughly constant density implies a size
growing like
R = r0 A1/3 .
(27)
Experimentally the matter distribution (protons and neutrons) requires r0 1.2 fm, while the charge
distribution (protons have a size too!) is about r0 1.4 fm. This implies a nuclear density of about
1015 g/cm3 . Characteristic binding and excitation energies in nuclei are at the MeV level (compare this
with the eV-level for atoms), which for emission and absorption of photons means wavelengths as small
as 1012 m. Protons and neutrons in atomic nuclei are bound via a completely different force than that
in atoms, namely the strong force. To understand the main features of nuclear structure one employs a
(nonrelativistic) quantummechanical description with in the Hamiltonian an (effective) potential of which
Q/e = +1
NUCLEONS
Q/e = 0
proton
neutron
H=p
H = pn
3H = pnn
3
He = ppn
12C = 6p 6n
2
LEPTONS
electron neutrino
(atomic nuclei)
204
Z
N
110 days
10100 days
100 dgn 10 yr
1010.000 yr
naturally radioactive
stable
> 10.000 yr
the longest range part is coming from the exchange of pions with masses m 140 MeV, leading to a
Yukawa-like tail of the form V (r) exp(r/
)/r where
is the Compton wavelength ~/m c 1.4
fm, which is also the characteristic distance between nucleons in the nucleus.
As seen in the Z versus N plot for nuclei, stable nuclei lie mostly on or just above the Z = N diagonal
and there are many unstable nuclei with vastly different lifetimes. For the magic numbers indicated in
the figure, nuclei are particularly stable. An important decay mode for heavy nuclei is the emission of
-particles. The -decay is a nice example of quantummechanical tunneling through a potential barrier.
A second decay is -decay in which a neutron changes into a proton under emission of an electron and a
neutrino,
n p + e + e .
(28)
It is this decay that is also responsible for the instability of 3 H, decaying via 3 H 3 He + e + e .
-decay is the manifestation of yet another force, the weak force.
In processes like -decay, but also in interactions of photons or electrons with atoms, conservation laws
plays an important role. The most important conservation laws are the conservation of energy, momentum, and angular momentum. For energy and momentum the sum of all energies and the (componentwise) sum of all momenta in initial and final state is the same. For angular momentum the conservation
law applies to the total angular momentum of the initial and final state. This is made up from the spins
and the orbital angular momenta in a multi-particle or composite system.
Consider -decay as an example. For protons, neutrons and electrons, the spin is 1/2 (two spin states).
This implies that
X
X
integer J for even number of constituents
J=
`i +
si =
(29)
half-integer J for odd number of constituents
i
Q/e = +2/3
QUARKS
205
Q/e = 1/3
up
down
p = uud
n = udd
3 colors
(nucleons)
LEPTONS
electron neutrino
e
right
left
or
handed
handed
only
lefthanded
Quarks and leptons (of the first family); these quarks are the ones making up the protons and neutrons.
This led e.g. Pauli to postulate the existence of a neutral spin 1/2 particle, the neutrino, in -decay
(Eq. 28). Indeed, the neutrino was found and turned out to have spin 1/2 (although with only a lefthanded
state having ms = 1/2, antiparallel to the momentum).
Angular momentum is also very important in the study of transitions where photons are absorbed or
emitted. Photons have spin 1 (although with only the two states with ms = 1, parallel or antiparallel
to the momentum ). It leads to the selection rule J = 0, 1.
A last example we want to mention here is the Nitrogen atom. The nucleus 14 N contains 7 protons
and 7 neutrons. Including the 7 electrons the total angular momentum is half-integer3 . Indeed it turns
out that for the atom J = 1/2. Before one knew of the existence of the neutron and -decay, it seemed
natural to think of the 14 N nucleus as 14 protons and 7 electrons. In that case the atom would have
had an integer J, in disagreement with observation. This illustrates how simple combinatorial rules for
angular momentum provide strong discriminatory power.
2.3
To our present knowledge e and e are elementary particles without substructure. This is not true
for the proton and neutron. They have a substructure, e.g. evident from the magnetic moments of the
particles. Writing
e
i = gi
si ,
(30)
2mi
one has ge = 2, which was shown by Dirac to be (up to a very tiny correction) the natural value for an
elementary spin 1/2 particle. One has gp = 5.586 and gn = 3.826. In particular the latter is a surprising
result for a neutral particle. Proton and neutron turned out to be composed of two quark species, up (u)
and down (d) quarks with fractional charges Qu = +2/3 e and Qd = 1/3 e, having spin 1/2. In addition
to these quantum numbers, the quarks carry one of 3 color charges. One of the strongest indications
for the need of such an additional quantum number came from the existence of doubly-charged ++
3 Note that if we talk about total angular momentum of a composite system, we refer standard to its rest-frame. In other
frames one has to add orbital angular momentum
+ = ud
_
=ud
206
(pions)
up
down
up
down
QUARKS
e
electron neutrino
LEPTONS
ANTIQUARKS
e+
positron neutrino
ANTILEPTONS
particle which turned out to have spin J = 3/2. The natural explanation is a state consisting of three
u-quarks, which all are in an s-orbital (` = 0). Without an additional quantum number this would be in
contradiction with the Pauli principle, however. With the introduction of color everything is fine.
The color charge turned out to be more than just an additional quantum number. It is the source for
the strong interactions. The interactions between color charges are mediated by gluons and the theory has
been given the name Quantum Chromodynamics (QCD) because of the analogy with QED. The presence
of 3 colors leads to a more complex underlying structure with eight noncommuting charge operators being
3 3 matrices (in analogy to the three noncommuting spin operators being 2 2 matrices). Coupling
to each of these eight operators with the same strength one has eight gluons, whereas QED only has
one photon. The gluons actually have a color charge themselves, leading to a linearly rising potential
between color charges. This leads to the confinement of quarks in the nucleons. Only a color neutral
configuration requiring 3 quarks has a finite energy. Further substructure than the quarks has not (yet)
been found. Quarks are known to be smaller than 1018 m (about 103 times the size of the nucleons).
What has been seen is the possibility to excite the nucleon. The excitation energies are in the 100 MeV
range, to be compared with the nuclear MeV-range and the atomic eV-range. Because of the confining
potential, one does not have the situation that the mass (energy) of the composite system is less than
that of the constituents, as is the case for molecules, atoms and nuclei. In fact the masses of the up and
down quarks turn out to be in essence zero on the scale of the nucleon mass. The scales are set by the
size, RN 1 fm and ~c/RN 200 MeV. These excitation energies also play a role in the structure of
nuclei. One might wonder why nuclei dont like to be just a bag of quarks. The answer is simple, it is
energetically favorable to split into nucleons. Actually trying to squeeze six quarks together turns out to
give typically an additional energy of a few hundred MeV, visible as the repulsive core in the potential
between two nucleons.
2.4
The quarks and leptons all turned out to have corresponding antiparticles, of which the positron (e+ ) was
the first to be discovered. Antiparticles have opposite electric and color charges, but identical masses as
compared to the particles. Three quarks can form color neutral combinations (baryons) but also a quark
and an antiquark can form a color neutral particle, called mesons. The lightest of these are the pions.
Mesons of course must have integer spin, e.g. the pions have spin 0.
While the exchange of gluons produces the binding force of the quarks (and antiquarks) in baryons
(qqq) or mesons (q q), it is the exchange of mesons that produces the long-range (effective) interaction
responsible for the nuclear binding. Via the formation of quark-antiquark pairs, e.g. by colliding electrons
and positrons,
e+ + e q + q,
(31)
207
QUARKS
Family 1
up
down
LEPTONS
e
electron neutrino
charm
strange
muon
top
bottom
tau
Family 2
Family 3
+
ANTIQUARKS
neutrino
neutrino
+
ANTILEPTONS
+ FORCE PARTICLES
g
graviton
gravitation
photon
electromagnetism
Z0
W, Z bosons
+
weak force
G
8 gluons
The particle content of the Standard Model, quarks, leptons belonging to three families and the
force particles, responsible for
the different interactions.
strong force
one has found that other quark flavors exist besides up and down. There exist two other families of
quarks and leptons. Besides these new fermions, which in essence differ only from the first family by
mass, one has found 3 heavy bosons, the Z 0 and W -particles, which are responsible for the weak force,
e.g. in -decay.
The 3 families of quarks and leptons, their antiparticles and the force-carrying particles form the
content of the socalled Standard Model of elementary particles. To complete the Standard Model and its
symmetries, the Higgs field is an essential ingredient and its signature, the Higgs particle also was found
a few years ago. All the particles in this three family scheme are elementary in the sense that they do
not appear to have substructure in the way as for nuclei and nucleons. Their sizes are smaller than what
can be measured at present, which is of the order of 1018 m.
208
Exercises
Exc. 2.1: magic numbers for atomic nuclei
In the periodic table for atoms, we know that filled shells of electrons play a crucial role, e.g. defining
which elements are noble gases, etc. Similarly for atomic nuclei there appear very stable nuclei with
numbers of protons (Z) or numbers of neutrons (N ) being 2, 8, 20, 28, 50, 82, 126. These numbers
can be easily accomodated in the nuclear shell model when one assumes the protons and neutrons to be
moving in an (effective) central potential with in addition a spin-orbit interaction of the form
Vso (r) = Aso (r) `s,
with Aso being positive. What is the effect of this interaction on an (n`) level in the central potential.
Find the level order that explains the magic numbers.
(solution)
E
magic #
126
1h
82
3s
2d
1g
2p
1f
2d 3/2
3s 1/2
1h11/2
2d 5/2
1g 7/2
1g 9/2
2p 1/2
1f 5/2
2p 3/2
1f 7/2
50
28
20
2s
1d
1d 3/2
2s 1/2
1d 5/2
1p
1p 1/2
1p 3/2
2
1s
nl
1s 1/2
nl j
209
much larger than typical nuclear energies of the order of a few MeVs. For protons or neutrons the energy
still is very well approximated by
p
E M c2 + p2 c2 Mp c2 + p2 /2Mp ,
(compare right and left side if you dont believe it). The result is p2 /2Mp p2 c2 /2Mp c2 0.5 MeV,
the right order of magnitude. Note that for nucleons v/c pc/E 0.1, which is relativistic but not
ultra-relativistic like for electrons.
Vss
V` (r) 2 V`s (r)
` +
`(s1 + s2 ) 2 s1 s2 .
~2
~2
~
The potential functions Vi (r) are positive, the central potential Vc thus being attractive (appearing with
a minus sign), etc. The orbital angular momentum ` is the relative angular momentum in the nucleonnucleon system.
(a) How do you find the right quantum numbers to label the energy eigenstates of the two-nucleon
system?
(solution)
The good quantum numbers to label the energy states are the eigenvalues of operators that commute with the Hamiltonian and in particular also the potential VN N . The relevant operators are
one of the components of `, s1 , s2 or components of the combinations S = s1 + s2 , J = ` + S, or
quadratic operators like `2 , etc. In this case the potential/Hamiltonian commutes with `2 , s21 , s22 ,
S 2 , J 2 and J z (but not with `z , s1z , or s2z ).
(b) Investigate for ` = 0, 1 en 2 in a table the possible configurations 2S+1 `, where the total spin is
built from s1 = 1/2 and s2 = 1/2 states. Give the degeneracy of the configurations, the possible
J-values and indicate which states are allowed for a proton-proton (pp) system and which ones are
allowed for a proton-neutron (pn) system. It is useful to realize that interchange of two nucleons
implies that the relative coordinate changes sign, thus the permutation symmetry of the orbital
part of the wave function is simply the same as the parity ()` .
`
...
S
...
symmetry
2S+1
`J
...
degeneracy
...
pp
forbidden/allowed
pn
forbidden/allowed
(solution)
The allowed configurations are
`
0
1
2
S
1
0
1
0
1
0
symmetry
+
+
+
2S+1
`J
S1
1
S0
3
P0,1,2
1
P1
3
D1,2,3
1
D2
3
degeneracy
3
1
9
3
15
5
pp
forbidden
allowed
allowed
forbidden
forbidden
allowed
pn
allowed
allowed
allowed
allowed
allowed
allowed
210
(c) Use perturbation theory to study the energies En`... , where . . . contains the quantum numbers
compatible with the Hamiltonian (channels). Indicate what the degeneracy of the levels is. Express
the energies in the expectation values of the radial potential functions, hVc i, etc. and picture the
order if hVc i hV` i hVss i > hV`s i. Which channel has the lowest energy?
(solution)
Applying perturbation theory we find
En`SJ = hVc i+hV` i `(`+1) 12 hV ssi (S(S + 1) 3/2)+ 21 hV`s i (J(J + 1) `(` + 1) S(S + 1)) .
with degeneracy in M (thus degeneracy 2J + 1) and energies
`
0
1
S
1
0
1
1
1
0
1
1
1
0
symmetry
+
+
+
+
+
2S+1
`J
S1
1
S0
3
P0
3
P1
3
P2
1
P1
3
D1
3
D2
3
D3
1
D2
3
degeneracy
3
1
1
3
5
3
3
3
3
5
energy
hVc i 14 hVss i
hVc i + 34 hVss i
hVc i + 2 hV` i
hVc i + 2 hV` i
hVc i + 2 hV` i
hVc i + 2 hV` i +
hVc i + 6 hV` i
hVc i + 6 hV` i
hVc i + 6 hV` i
hVc i + 6 hV` i +
1
4 hVss i
1
4 hVss i
1
4 hVss i
3
4 hVss i
1
4 hVss i
1
4 hVss i
1
4 hVss i
3
4 hVss i
2 hV`s i
hV`s i
+ hV`s i
3 hV`s i
hV`s i
+ 2 hV`s i
The lowest state (and actually the onlys one that exists as bound state) is the pn system in the 3 S1
state (deuteron).
(d) (optional) There is another interaction term (the tensor interaction) in the NN potential
3(s1 r)(s2 r)
V = . . . + Vt (r)
s1 s2 ,
r2
where r is the relative coordinate. For spin 1/2 particles the spin operators are (in spin-space)
s = ~/2. These matrices satisfy (a)(b) = ab + i(a b), thus (
r )2 = 1. Use this to
rewrite the interaction term in terms of the total spin operator S = s1 + s2 . Can you indicate
which of the above quantum numbers remain good quantum numbers including the Vt term?
(solution)
This interaction can be rewritten as
V = + Vt (r)
3(Sr)(Sr)
S2
r2
showing that of the above good set `2 no longer commutes. This leads for the deuteron to a mixing
of the 3 S1 and 3 D1 states (with probability of about 5%) which for instance is responsible for a
nonzero quadrupole moment of the deuteron.
211
(solution)
(a-b) The isospin states are built from basic states |pi = |1/2, 1/2i and |ni = ket1/2, 1/2 and give states
(antisymmetric).
The use of the generalized Pauli principle gives the same number of allowed states (the first two
entries of the Table are the answer to (a), the rest gives the answer for (b).
2S+1
`J
S0
3
S1
1
P1
3
P0,1,2
1
D2
3
D1,2,3
1
F3
3
F2,3,4
1
pp
yes
yes
yes
yes
nn
yes
yes
yes
yes
pn
yes
yes
yes
yes
yes
yes
yes
yes
(N N )I=0
yes
yes
yes
yes
-
(N N )I=1
yes
yes
yes
yes
(c) An interaction of the form V T 1 T 2 would give a splitting between the states of the form
hI, I3 |T 1 T 2 |I, I3 i =
1
(I(I + 1) 3/2) ,
2
with results -3/4 (for I = 0) and +1/4 (for I = 1), in complete analogy of a spin-spin interaction
between two spin 1/2 particles.
212
e
sp
2M
and n = gn
e
sn ,
2M
where M is the nucleon mass, sp and sn are the individual nucleon spin operators and gp 5.6
and gn 3.8. The magnetic moment operator of the atomic nucleus is found as the sum of the
nucleon magnetic moments,
X
e
A =
gi
si .
2M
i
and show that its expectation value for a nucleus can be written in terms of the expectation value
of the total spin S (You can do this for just a single spin state, and give the argument why it works
for all spin states),
e
hA i = gA
hSi.
2M
Calculate the gA -factors for 3 H and 3 He.
(solution)
Evaluating the magnetic moment operator for the z-component is easiest and gives
h3 He |Az |3 He i =
e~ gn
2M 2
while h3 He |Sz |3 He i =
~
.
2
213
We conclude gA (3 He) = gn . We dont need to check other components if we use the WignerEckardt theorem. Another way of seeing this, is to realize that we can reach all other components
by applying lowering and raising operators for spin (S = sp + sn , etc.), which will only affect
the magnetic quantum numbers. The result for 3 H is gA (3 H) = gp
1/6 (2 |d d u i |d d u i |d d u i) .
214
where mq is a quark (constituent) mass, which we as a first guess take to be mu = md = M/3. The
charges are eu = +2/3 e and ed = 1/3 e. What do you find for the nucleon magnetic moments,
i.e. calculate gp and gn in
e
hN i = gN
hsN i.
2M
(solution)
Evaluating the magnetic moment operator for the z-component is easiest and gives
e~
~
while hp |Sz |p i = ,
mq
2
2 e~
~
hn |N z |n i =
while hn |Sz |n i = .
3 mq
2
hp |N z |p i =
++
; 3/2, 1/2i =
=
|uu; 1, 1i|u i = |u u u i,
p
p
1/3 |uu; 1, 1i|u i + 2/3 |uu; 1, 0i|u i
p
1/3 (|u u u i + |u u u i + |u u u i) ,
etc. The states are indeed symmetric in space, spin-space, and if one chooses to work with that in
isospin space (where flavors u and d are taken to be quark quantum numbers). The states obtained
with a down quark are
|+ ; 3/2, 3/2i = |uu; 1, 1i|d i = |u u d i,
p
p
|+ ; 3/2, 1/2i =
1/3 |uu; 1, 1i|d i + 2/3 |uu; 1, 0i|d i
p
=
1/3 (|u u d i + |u u d i + |u u d i) ,
where one could also choose to include an additional symmetrization in isospin. If you are interested
you could calculate the g-factors, which are different for the different charge states, ++ , + , 0
and . These states have just as proton and neutron roughly the same mass, M 1235 MeV/c2 .
The four states form an isospin 3/2 multiplet in isospin space.
215
B[MeV]
9.0
7.5
56
240
where B is the nuclear binding energy. For instance for the deuteron this binding energy is 2.22 MeV or
about 1.1 MeV per nucleon. For 12 C the binding is about 12 7.93 MeV. The binding energies, usually
quoted per nucleon are roughly constant. Starting with Weisacker in 1936 attempts have been made to
understand the systematics of nuclear binding energies via an empirical mass formula. We will look at
some of the terms in this formula.
(a) Explain the most obvious terms
B = a1 A a2 A2/3 a3
Z2
A1/3
(solution)
The starting point is an average binding energy per particle,
B1 = a1 A.
A first correction to this is a correction because a nucleus has a finite size. Nucleons at the surface
dont experience the same force. This correction is assumed to be proportional to the surface
R2 A2/3 . For this you have to realize that for nuclei the sizes are roughly proportional to A1/3 ,
corresponding to a constant density of nucleons. One writes B2 = a2 A2/3 A second correction
is the electrostatic self-energy. For a uniform charge distribution one has
UQ =
Therefore one writes B3 = a3
3 Q2
Z2
1/3 .
5 R
A
Z2
A1/3
(b) A further correction comes from the fact that N and Z are not equal. A simple model to see what
is the expected dependence is to assume a Fermi gas model. Protons and neutrons in nuclei are
considered as two (independent) Fermi gases in a volume V , which have similar levels available
(the strong interactions are much stronger than electromagnetic effects). Thus for both types one
has a level
density n(E)dE = V d3 p/(2~)3 . For a spherically symmetric situation one obtains
n(E) V E. Integrate n(E) and E n(E) up to the Fermi energy EF to find the number of
particles N and the energy U of those particles and use this to show that the energy of Z protons
and N neutrons in a common volume V A is given by
U
Z 5/3 + N 5/3
A2/3
216
and expand this in the difference N Z to show that it leads to a correction term
B4 = a4
(A/2 Z)2
.
A
(solution)
EF
dE E 1/2 =
2 3/2
E ,
3 F
dE E 3/2 =
2 5/2
E ,
5 F
0
EF
3
5
giving the well-known value of EF for the average energy of particles in a Fermi gas. The equations
5/2
also tell us that the energy of the N particles is U V EF N 5/3 /V 2/3 . Thus we have for a
system of Z protons and N neutrons in a nuclear volume V A,
U
20 (A/2 Z)2
5 (N Z)2
Z 5/3 + N 5/3
+ ... = A +
+ ....
=A+
2/3
9
A
9
A
A
The expansion around Z = N shows as expected a term in the energy proportional to A, which
can be included into a1 and the correction B4 .
Additional information:
To complete the mass formula one needs to include the socalled pairing term distinguishing even-even,
even-odd and odd-odd cases. Empirically it has been found that pairing off spins between identical
fermions (protons or neutrons) lowers the energy. The correction is less important for heavy nuclei.
Z
even
even
odd
odd
even
odd
even
odd
# unpaired
spins
0
1
1
2
energy
lower
0
0
higher
correction
+a5 A3/4
0
0
a5 A3/4
Z2
(A/2 Z)2
a
+ .
4
A
A1/3
a2
17.81
0.0191
a3
0.7105
0.000 763
a4
94.80
0.098 55
a5
39
0.042
MeV
u
Semiclassical methods
301
Semiclassical methods
In this section we will treat applications of semi-classical methods in quantum mechanics. The theoretical
basis for this is discussed in Griffiths chapter 8 (The WKB approximation). You have to study in detail
sections 8.1 and 8.2, while you are encouraged to read section 8.3.
3.1
Bohr quantization
We do want to recall the idea of Bohr quantization. Quantization is imposed in an ad hoc way by requiring
` = n ~ with n being integer. This is actual not bad for large quantum numbers n where quantum
mechanics coincides with classical mechanics. For the electron in the atom one uses the condition that
the central force to bind the electron is provided by the Coulomb attraction,
e2
mv 2
,
=
r
40 r2
(32)
40 ~2
= n2 a 0
m e2
and En =
1
m e4
ER
= 2,
2
n 32 2 20 ~2
n
(33)
which turns out to give the correct (quantized) energy levels and also a good estimate of the radii. At
the classical level the Sommerfeld model of the atom even includes quantization conditions for treating
elliptical orbits.
It is interesting to observe that the Bohr quantization condition not only gives the right characteristic
size (a0 ) and energy (R ) and the right power dependence on quantities like e2 , but what is more
surprising also the right power behavior of the quantum numbers (n, `). Note e.g. that the Bohr model
gives r n2 and (indeed) all the expectation values involving rp have a polynomial behavior in (n, `) of
order 2p. Energy has a behavior E 1/n2 . Replacing n (nr + ` + 1) even gives the exact energies
in terms of number of nodes nr (quantum number for radial excitations) and ` (quantum numbers for
orbital excitations). The WKB approach gives a more proper justification for the method.
3.2
This is the region in which the E V (x), thus the spatial region between the turning points in classical
mechanics. Here the time-independent Schrodinger equation is rewritten as
d2
2m
=
(E V (x)) (x).
(34)
dx2
~2
|
{z
}
k2 (x)=p2 (x)/~2
(35)
Semiclassical methods
302
(36)
in terms of two real functions, the amplitude A(x) and the phase (x) which satisfy
00
A A(0 )2 = k 2 A = A00 = A (0 )2 k 2 ,
p
2 A0 0 + A00 = 0
=
(A2 0 )0 = 0 = A = C/ 0 (x).
(37)
Dropping the A00 term (assuming the amplitude to be slowly varying) gives the WKB (Wentzel, Kramers,
Brillouin) approximation
Z
Z
C
exp i dx k(x) ,
(38)
(x) = dx k(x) and (x) = p
k(x)
with probability |(x)|2 = |C|2 /k(x). The general solution is a superposition of the two exponentials
(with different signs) or they can be written in terms of sin (x) and cos (x).
As applications we mention
One dimensional problems between two turning points ap
1 and a2 outside of which the potential
becomes very large. At the turning points k(a1 ) = 0, so k(x) (x) sin((x) (a1 )). Also at
the other turning point the wave function must vanish so (a2 ) (a1 ) = 0 or
Z a2
dx k(x) = n.
(39)
a1
3.3
Tunneling is the phenomenon that happens in the classically forbidden region E < V (x). The treatment
is completely analogous to the WKB method, except that the exponents now are real. Remember that
for E < V0 constant potential one has k 2 (x) = q 2 = 2m(V0 E)/~2 is constant and the solution is
simply
(x) = A eqx + B eqx = A0 sinh(qx) + B 0 cosh(qx).
(41)
For a not too fast changing potential one will have
(x) = p
C
q(x)
Z
exp dx q(x)
(42)
If the barrier is wide enough, only the decaying exponential survives because any finite probability will
require the amplitude multiplying the rising exponential to go to zero. Thus the transition (tunneling)
probability gives rise to a lifetime
Z r1
T T0 e2 with =
dr q(r).
(43)
r0
where one integrates over the classically forbidden region, where ~ q(r) =
Semiclassical methods
303
Excersises
Exc. 3.1: zero-binding limit
A finite potential well used to describe the interactions between nucleon has a finite number of bound
states. It is interesting to look at bound states with very small binding energies and their properties. The
simplest example of this kind is the deuteron being a very weakly bound state with binding energy of 2.2
MeV. We will investigate some of the properties. Calculate the minimum depth V0 of a (3-dimensional)
potential well with range a (i.e. V (r) = V0 for r a and zero outside) and calculate the value of V0 for
a deuteron with a potential range of a = 1.5 fm = 7.5 GeV1 .
(solution)
The radial Schr
odinger equation for a bound state is of the form
ra:
u(r) sin(Kr)
ra:
u(r) exp(qr)
2m
(E + V0 ),
~2
2m
with q 2 = 2 E.
~
K2 =
with
To match these solutions one sees that the interior solution needs to have at least a negative derivative at
r = a, thus Ka /2. Making the potential deeper, the lowest energy will be lower (it is of course also
negative). The minimal depth corresponds to E = 0 (hence the name zero-binding limit). The external
wave function is flat in that case (q = 0). One has
K 2 a2 =
2m
2
V0 a2 =
2
~
4
V0 =
2 ~2
.
8 ma2
r
rn =
~
n
m
r
and vn =
~
,
m
1
2
mvn2 +
1
2
krn2 = n ~.
Semiclassical methods
304
n2
~2
m T0
1/3
1/3
~ T0
m2
n2
~2 T02
m
and vn =
1
2
mvn2 + T0 rn =
3
2
T0 rn =
3
2
1/3
.
In this case the link to the exact result is not found from a simple replacement, but the replacement
n = 1.8 nr + ` + 1.376 gives a good approximation to the levels.
v=1
v
M=
Z
J=
L
L
dx
T0
,
1 v2
T0 x v
dx
1 v2
L
L
Use this to derive a relation between 0 and T0 and calculate the linear energy density T0 in GeV/fm.
(solution)
The integrals can easily be performed if we realize that v = x/L, thus
Z L
Z 1
1
T0
= T0 L
dv
= T0 L,
E=M =
dx
2
1v
1 v2
1
L
Z L
Z 1
T0 v x
v2
T0 L2
.
J=
dx
= T0 L2
dx
=
2
c2 1 v 2
1 v2
L
1
giving after elimination of L the desired result, 1/0 = 2 T0 . In this way one finds from the slope of the
J versus M 2 trajectory a measure for the string tension in mesons, T0 = 0.9 GeV/fm. This also works
for baryons with at the ends of the string having one and two quarks, respectively.
Semiclassical methods
305
dx
x
e
=
(3/2)
=
/2.
0
(solution)
(a) The logarithmic potential is interesting because the size parameter can be changed by an energy
shift, since ln(r/a) = ln(r) ln(a). This is also seen because the force balance gives
mv 2
V0
=
r
r
v=
p
V0 /m,
showing that v is independent of r. The quantization condition for mvn rn = n~ then fixes
s
~
~2
rn = n
=n
.
mv
mV0
and the energy becomes
s
2
e~2
~
1
1
= V0 ln
+ V0 ln n.
En = V0 + V0 ln n
2
mV0
2
mV0 a2
It exhibits a scale invariant result
En+1 En = V0 ln
n+1
n
.
(b) (see also Griffiths problem 8.13). The WKB method gives with for level En the turning point
En V0 ln(rn /a) = 0 or ln(rn /a) = En /V0 ,
Z rn r
2m
dr
(En V0 ln(r/a)) = (n 14 ).
~2
0
Thus one gets introducing ln(r/rn ) = x (x running from 0 to ),
r
Z rn r
Z
2mV0
2mV0
dr
(
ln(r/r
))
=
r
dx x ex = (n 41 ),
n
n
2
2
~
~
0
0
leading to
s
1
2~2
rn = (n )
4
mV0
and En+1 En = V0 ln
n + 3/4
n 1/4
Semiclassical methods
306
2
dx p
(x a)(b x) =
b a ,
x
2
is useful to obtain the following WKB result (with turning point improvements)
En` =
1
2
ER
2 .
p
+ `(` + 1)
(solution)
This is problem 8.14 from Griffiths.
Show the ln T a/ Eb behavior for -decay life times using the tunneling through a Coulomb potential
of the form V (r) = 2Ze2 /40 r from R1 to R2 , where R1 is the nuclear radius and R2 is related to the
energy of the emitted particle through E = 2Ze2 /40 R2 = 2Z~c/R2 . Take characteristic values
Z = 90 and A = 220 and calculate the coefficients in the tunneling exponent (see Griffiths or lecture
notes). To estimate T0 in T = T0 e2 one needs the velocity of the -particle in the nucleus. For this use
a typical value for the kinetic and potential energy of 50 MeV.
(solution)
The relevant integral for the survival probability is
r
r
Z R2 r
Z
2m
2mE R2
R2
=
dr
(Vc (r) E) =
dr
1
2
2
~
~
r
R1
R1
which is done in Griffiths and approximately gives
r
r
r
p
2mE R2
2 mc2
mc
R
R
ZR1
2
=
4
.
1 2
~2
2
E
~
Numerical estimates give Z~c 130 MeV fm, R1 r0 A1/3 7 fm, m c2 3.75 GeV, ~/m c (0.2
GeV fm)/(3.75 GeV) 0.05 fm. Numerically it gives
p
Z=90
The actual lifetime is then the characteristic time T0 = 2R1 /v multiplied by e2 . The characteristic time
is small. Using for the kinetic energy of the particle in the nucleus 50 MeV, we get p 600 MeV and
v 0.15 This gives T0 1013 m 3 1022 s. The precise valueof this time is not that important the
important behavior is the slope in the logarithmic ln T versus 1/ E plot coming from the first term in
(see Griffiths).
401
In this and the next sections of the notes, we treat Griffiths chapter 9, introducing, summarizing and
sometimes repeating some of the material presented in that chapter. We will also discuss a number of
applications, some of them are not in Griffiths.
4.1
Time dependence in quantum mechanics is described via an operator, namely the Hamiltonian H,
i~
= H ,
t
(44)
where = (t, . . .) and H is often built from other already known operators such as position x, momentum px in one dimension or r, p, orbital angular momentum ` = r p, spin s, etc. in three dimensions,
often for more than one particle, r 1 , r 2 , etc.
Two very important situations that we want to distinguish are the cases that H also changes in time
with a time-dependent Hamiltonian H(t) versus the case of a time-independent Hamiltonian.
(a) Time-independent Hamiltonian H0 : in that case one can first solve H0 E = EE , where E
does not depend on time. In general this Hamiltonian will have a set of eigenstates and eigenvalues
(energy spectrum), H0 (r, p, . . .)n = En n . Using the completeness of the states n one can write
X
(t) =
cn (t)n ,
(45)
n
cn (t) = cn ei En t/~ ,
(46)
(48)
(49)
The part H0 does not have explicit t-dependence, while the second part has a (possible) time-dependence.
For the system described by H0 eigenstates n and eigen-energies En are known. When doing timeindependent perturbation theory (i.e. the case that V is time-independent) one tries to express the true
eigenfunctions of H in the complete set n . The important observation is that because we now know
402
that any initial state (0), even if it would be a stationary state i of H0 , now has a more complex time
dependence than exp(iEi t/~), which is be the time dependence if V (t) = 0.
Starting with the known (time-independent) part H0 , one can use the completeness of the states n
to write
X
(t) =
cn (t) n eiEn t/~ .
(50)
n
Note that one could have absorbed the exponential time-dependence in cn (t), but not doing so is more
appropriate because the time-dependence of cn then solely is a consequence of V . Substituting the
expression for (t) in the full Schr
odinger equation,
i~
(51)
(52)
where Vpn = hp |V (t)|n i is the expectation value of the potential V between the (time-independent)
eigenstates of H0 , and pn = (Ep En )/~. As expected if V = 0, the righthand-side is zero and the
coefficients are time-independent, leaving in (t) only the trivial time-dependence of the stationary
states.
In the next subsection the above equations is solved for a simple two-state system. This is followed
by an often-used perturbative approach.
4.2
Consider a two-level system, with a and b being solutions of the unperturbed Hamiltonian with eigenvalues Ea and Eb , H0 a = Ea a and H0 b = Eb b . Using a matrix notation on the basis
1
0
a
and b =
(53)
0
1
the Hamiltonian H0 and the perturbation V can be written as
Ea 0
Vaa (t) Vab (t)
0
Vab (t)
H0 =
and V =
=
,
0 Eb
Vba (t) Vbb (t)
Vab
(t)
0
(54)
where we have restricted ourselves to the simpler situation that there is only a time-dependent off-diagonal
element. Most applications can be cast in this form anyway (see Exercises). The solution can always be
written as
1
0
ca (t) eiEa t/~
(t) = ca (t)
eiEa t/~ + cb (t)
eiEb t/~ =
,
(55)
0
1
cb (t) eiEb t/~
and with V off-diagonal the coefficients ca (t) and cb (t) obey the simple coupled set of equations of the
form
ca (t)
0
Vab (t) e+iab t
ca (t)
i~
=
cb (t)
Vab
(t) eiab t
0
cb (t)
0
v(t)
ca (t)
~
,
(56)
v (t)
0
cb (t)
representing simple coupled equations,
i ca (t) = v(t) cb (t),
(57)
(58)
403
(59)
coupling the magnetic moment proportional to its spin. For instance for an electron = (e/m) s
where s = (~/2) . For other (composite particles) the factor in general will be different. However,
for any spin 1/2 particle the spin operators can be represented by the Pauli matrices. Starting with a
constant magnetic field in (say) the z-direction, B 0 = (0, 0, B0 ), and using the matrix representation for
a spin 1/2 particle one has
B0
B0
1 0
H0 =
~ z =
~
.
(60)
0 1
2
2
The solutions are easily obtained,
a =
b =
1
0
0
1
B0
~,
2
B0
Eb = +
~,
2
with Ea =
(61)
with
(62)
If the system is in a spin-state along the z-direction, it will stay in this state. If it is in another direction,
it will start to oscillate with a frequency 0 = ba = (Eb Ea )/~ = B0 , known as the Larmor frequency.
Next consider the system in a circulating magnetic field in the x-y plane, superimposed on B 0 , B(t)
= B 0 + B 1 (t), where B 1 (t) = (B1 cos t, B1 sin t, 0). In that case
H
=
=
B0
B1
~ z
~(x cos t y sin t)
2
2
B0
B1
1 0
0
e+it
~
.
0 1
eit
0
2
2
(63)
B1 +it
e
2
and v(t) =
Vab (t) i0 t
e
= v1 ei(0 )t .
~
(64)
with v1 = B1 /2. The coupled equations in Eq. 56 can now simply be rewritten into a second order
differential equation for ca ,
ca i( 0 ) ca + v12 ca = 0.
(65)
This equation has two independent solutions of the form ei pt with
p=
with the Rabi flopping frequency r =
1
2
1
( 0 ) r
2
(66)
(67)
404
Starting off with ca (0) = 0 and |cb (0)| = 1, it is straightforward to check that
1
(68)
(70)
(69)
Thus, given an initial spin aligned parallel or antiparallel to the B 0 field, the probability for transition to
the other spin state shows oscillations with a frequency r , while the magnitude depends on the frequency
of the rotating perpendicular B 1 field, showing a resonance at = 0 . In that case the spin completely
flips from parallel to antiparallel and back with frequency r . At resonance the Rabi flopping frequency
is given by r = |v1 |. Away from resonance the flopping frequency increases and the amplitude of the
oscillations becomes smaller, amplitude squared being reduced to a factor 1/2 at = 0 2|v1 |.
4.3
In many case we are not able to exactly solve the time-dependent problem and we treat the problem
perturbatively. Writing, H = H0 + V (t), including a multiplicative factor to keep track of orders. In
the case of perturbation theory, we realize that in writing a solution of the form
(1)
cp (t) = c(0)
p (t) + cp (t) + . . . ,
(71)
(72)
Starting with cp (0) = pi , one immediately sees that the first two orders are given by
c(0)
p ( ) = pi ,
Z
1
(1)
dt Vpi (t) e+i pi t .
cp ( ) =
i~ 0
(73)
(74)
This can straightforwardly been extended and leads to the socalled time-ordered exponential. In first
order perturbation theory, the quantity |cp ( )|2 is the probability to find the system in the state p , which
(1)
(0)
means the probability for a transition i p. The first order result is valid if |cp ( ) + cp ( )|2 1.
4.4
405
Exercises
1+i
1i
|+iy +
|iy .
2
2
and we find a 50-50 division with hsy i = 0. The coefficients can be obtained by calculating
y h|(0)i.
(b) Given a (time independent) Hamiltonian H0 for a spin 1/2 particle with charge e and mass m in a
magnetic field along the z-axis,
eB
H0 =
sz 0 sz .
m
What are the energies and corresponding eigenstates for this Hamiltonian? Give the time dependence of the system with initial state as under (a).
(solution)
The energies and eigenstates are
E1 = +~0 /2 and |E1 i = |+iz =
1
0
,
ei0 t/2
1/2
e+i0 t/2
!
.
0
1
.
406
(c) Calculate the expectation values hsz i, hsx i as a function of the time for the full solution in (b) and
check if this agrees with what you found at time t = 0 in (a).
(solution)
Straightforward calculation of h(t)|si |(t)i gives
hsz i(t) = 0,
hsx i(t) =
~
~ i0 t
e
+ ei0 t = cos(0 t) .
4
2
(solution)
The Ehrenfest relations gives, using H = 0 sz ,
dhsz i
= h[sz , H]i = 0,
dt
dhsx i
i~
= h[sx , H]i = i~0 hsy i,
dt
dhsy i
i~
= h[sy , H]i = i~0 hsx i.
dt
i~
d i R t dt0
e 0
dt
va (t0 )
= va (t) ei
Rt
0
dt0 va (t0 )
407
These types, however, appear to be different from free neutrinos. In order to explain some recent
experiments, one makes the assumption that a muon-neutrino produced in the atmosphere is a linear
superposition of the two mass eigenstates 1 and 2 . Taking these as basisstates, we then have at time
t = 0,
cos
(0) =
.
sin
(a) Denoting the energies of the free neutrino states with E1 = ~1 and E2 = ~2 , show that the
probability for finding a muon neutrino at time t is given by
1 2
t .
P = 1 sin2 2 sin2
2
(solution)
One has at time t,
(t) =
cos ei1 t
sin ei2 t
,
and the probability to find at time t again a muon-neutrino is obtained as the absolute value squared
of the (complex) scalar product of |(t)i and | i,
h |(t)i2 = (cos
sin )
cos ei1 t
sin ei2 t
2
= 1 sin2 2 sin2 1 2 t .
2
(b) Approximate the energies for free neutrinos with momentum p with different masses in the limit
that the momentum p m1 c and p m2 c. Show that the energy difference E1 E2 m2 c4 /2E,
where m2 = m21 m22 .
(solution)
For neutrinos with apgiven momentum p, the p
energies are different because of different masses m1
and m2 , i.e. E1 = p2 c2 + m21 c4 and E2 = p2 c2 + m22 c4 . For the situation that m1 , m2 pc
one has E1 E2 , both roughly equal to E = pc with the difference being
E1 E2
m21 c4 m22 c4
m2 c4
,
2E
2E
(c) Give the probability of finding muon neutrinos as a function of distance travelled (L), their energy (E) and the mass difference (m2 ) assuming that they are ultra-relativistic (i.e. v c) and
determine the oscillation wavelength defined as
L
P = 1 sin2 2 sin2
.
(solution)
After travelling a distance L with approximately the speed of light c one has a survival probability
m2 c3 L
L
P = 1 sin2 2 sin2
= 1 sin2 2 sin2
4~ E
with
= 4
~c E
.
m2 c4
408
(d) Given an oscillation wavelength of the order of 1000 km for neutrinos with an energy of 1 GeV =
109 eV (note that ~c = 0.2 106 GeV m). Calculate the mass/energy difference m2 c4 between
the two neutrino mass eigenstates for that case.
(solution)
Numerically a wavelength = 1000 km corresponds for a neutrino with a typical energy of 1 GeV
to a mass squared difference of m2 c4 = (0.05 eV)2 .
imply that |ca |2 + |cb |2 = 1. Starting with ca () = 1(real), it is clear that cb (t) will become imaginary.
Assume ca (t) = cos(z(t)) and cb (t) = i sin(z(t)). The equation for z(t) becomes
z(t)
=
(t).
~
(t),
~
thus ca = cos(/~) and cb = i sin(/~) for t > 0. Thus Pab = sin2 (/~).
501
One of the most important applications of time dependent perturbation theory is its use in describing
scattering processes from a given initial state into a final state. In one dimension this are just transition
probabilities, in three dimensions the transition probability is expressed in what is called a cross section,
representing the transition probability per unit of time normalized to the flux (which turns out to be
an area as compared to just a number in one dimension). While in non-relativistic quantum mechanics
the potential determines the scattering amplitude and there are also ways to solve the problem without
perturbation theory (Chapter 10 of Griffiths), the relativistic treatment requires quantum field theory
and will usually follow the perturbative route outlined in this chapter.
5.1
Going back to the perturbative calculation in the previous section, one notes that even for a timeindependent interaction V , transitions will occur, if the initial state is not an eigenstate of the full
Hamiltonian, but only of H0 . For a time-independent interaction V , which we immediately generalize to
an interaction with a harmonic time dependence proportional to ei t , we can easily obtain the coefficient
(1)
cp . If V is sufficiently weak, we find the result in first order perturbation theory,
Z
|Vpi |
|Vpi |
i(pi )t
i(pi )t
(1)
e
dt e
=
cp ( ) =
i~ 0
~( pi )
0
2
|V
|
|Vpi |
pi
1 ei(pi ) =
sin(( pi ) /2) ei (pi ) /2 ,
(75)
=
~( pi )
~( pi )
and thus for p 6= i,
(1)
Pip ( ) =
(76)
The function
sin2 ( /2)
2
is for increasing times ever more strongly peaked around = 0. The value at zero is f (0) = 2 /4, the
first zeros are at || = 2/ . Since
Z
sin2 ( /2)
d
=
,
(77)
2
f () =
sin2 ( /2)
=
().
2
(78)
Then we find4
2
|Vpi |2 (Ep Ei Q)
~
where Q = ~, or for the transition probability per unit time,
(1)
Pip ( ) =
2
(1)
Wip Pip =
|Vpi |2 (Ep Ei Q)
~
4
(ax) =
1
|a|
(x)
(79)
(80)
502
Including a harmonic perturbation, Fermis golden rule implies Ef = Ei + Q, where Q is the energy
transferred to the system in the interaction.
Although the allowed final state is selected via the energy delta function, it is often possible that the
system can go to many final states, because we are dealing with a continuum. In that case one needs the
density of states f (E), where f (E) dE is the number of states in an energy interval dE around E. The
transition probability per unit time is then given by
Z
2
2
Wif dEp f (Ep )
|Vf i |2 (Ep Ei Q) =
|Vf i |2 f (Ei + Q)
(81)
~
~
(Fermis Golden Rule No. 2).
5.2
The quantummechanical treatment of a scattering problem is that of a particle (with mass m and incoming
momentum p) scattering in a given potential V (r). We assume that the particle is scattered into a final
state with momentum p0 . The latter is the result of a measurement with a detector with opening angle
d, located under an angle (, ) with respect to the incoming momentum.
V(r)
p
The number of scattered particles per unit time per solid angle, n(, ), is proportional to the incoming
flux jin , the number of particles per area per unit time,
n(, ) d = |j in | d(, ).
(82)
This is the definition of the differential cross section d, from which it should be immediately clear that
the unit of cross section indeed is that of an area.
Typically cross sections have something to do with the area over which the scattering potential acts,
which could be the area of a target particle as seen by the incoming particle, e.g. for proton-proton
scattering a characteristic cross section is 40 mb, where 1 barn = 1 b 1028 m2 . The number 40 mb,
indeed, is roughly equal to the area of a proton (with a radius of about 1 fm = 1015 m). Besides the
area of the target the cross sections will also depends on the strength of the interaction. For instance
electromagnetic interactions are typically a factor 100 or (100)2 smaller, e.g. p 100 b and ep 1b,
corresponding to the presence of the fine structure constant or 2 respectively, where = e2 /40 ~c
= 1/137. Weak interactions, e.g. neutrino-proton scattering, again have much smaller cross section in
the order of 102 pb, indicative for the weakness of the weak interactions.
You have most likely seen the calculation of the Rutherford cross section in classical mechanics. Just
look at Exercise 5.2.
Cross section in Born approximation
We use the result of time-dependent perturbation theory to obtain an expression for the cross section. The
unperturbed situation is simply the free case with H0 = p2 /2m, with as possible solutions, the incoming
particle in a plane wave, i (r) = 0 exp (i p r/~), with energy E = p2 /2m and the detected final state,
f (r) = 0 exp (i p0 r/~), with energy E 0 = p02 /2m. We keep the (arbitrary) normalization to show
503
that it is finally irrelevant. Allowing processes in which the energy of the scattered particle changes we
have Q E 0 E. The situation Q = 0 corresponds to the elastic scattering process, an energy release,
Q > 0 to an exothermic (inelastic) process and energy absorption, Q < 0, to an endothermic (inelastic)
process. The potential V is a perturbation that can cause transitions between the plane waves. Starting
with particles with momentum p, we want to know the number of particles with momentum p0 (of which
the direction with respect to p is given by the angles , ),
i
2 h
2
|hf |V |i i| (E 0 ) 0
.
(83)
n(, ) d =
~
E =E+Q
Fermis golden rule gives Wif = Wpp0 , which is the transition probability per unit time of a plane
wave i = p into a plane wave f = p0 Because there are many energy levels (characterized by p0 with
the same energy E 0 we need the second rule. In order to get d we need to know the flux J in the initial
state and the density of states (E 0 ) in the final state at the energy E 0 fixed by energy conservation. For
the initial plane wave state the flux is given by J = 0 v = 0 p/m. If you forgot, check Griffiths chapter
1 (e.g. Exercise 1.14 in Griffiths). The density of final state plane waves is
(p0 )d3 p0 =
1 d3 p0
= (E 0 ) dE 0 d0 .
0 (2~)3
With the flux and density of final states, one obtains (Exercise 5.4)
2
m 2 p0 Z
i
3
0
0
,
d r exp
(p p ) r V (r)
d(, ) = d
2
2~
p
~
E 0 =E+Q
(84)
(85)
(86)
one obtains the following expression for the differential cross section in the socalled Born approximation,
2
m 2 p0
d
V
(q)
(87)
=
,
d0
2~2
p
where ~q = p p0 is the momentum transfer in the process. For an azimuthally symmetric differential
cross section one uses d = d cos d = 2 d cos to obtain d/d. Integrating the differential cross
section over all angles one obtains the total cross section,
Z
d
(E) = d
(E, ).
(88)
d
Note that in the case of elastic scattering one has p0 = p and the momentum transfer squared becomes
~2 q 2
(89)
A dependence of the differential cross section (d/d)(E, ) on this combination is a test for the validity
of the Born approximation. This simple link of energy and scattering angle is in particular useful for
central potentials, V (r) = V (r), in which case the Fourier transform
Z
Z Z 1
V (q) =
d3 r V (r) exp(i qr) = 2
dr
d cos r2 V (r) ei qr cos
0
1
Z
4
dr rV (r) sin(qr),
(90)
=
q 0
indeed only depends on q = |q|.
504
dr r sin(qr) =
dx x sin(x) =
[sin qa qa cos qa]
V (q)
=
3
q
q
q3
0
0
4 V0
1
1
4
qa1
qa (qa)3 qa + (qa)3 + . . . =
V0 a3 ,
(91)
q3
3!
2!
3
leading for E 0 to
d
1
d
9
2m V0 a2
~2
2
a2
(92)
dr sin(qr)
(93)
diverges and we need to consider for instance the screened Coulomb potential, multiplied with exp(r).
In that case one obtains a Yukawa potential V (r) = V0 er /r and
Z
Ze2
1
Ze2 4
dr sin(qr) er =
.
(94)
V (q) =
40 q 0
0 q 2 + 2
The screening thus allows the calculation. It even allows without problem to take the limit 0. This
can be used to get the cross section for scattering an electron of a nucleus with charge Ze,
m 2
d
(E, ) =
d
2 ~2
Ze2
0
2
1
=
q4
Ze2
8 0 pv
2
1
.
sin (/2)
4
(95)
This result is known as the Rutherford cross section and is identical to the classical result.
Application: Behavior of cross sections near threshold
If the volume integral over the potential exists, one knows that V (0) is finite and one sees that for small
values of the momentum transfer one can write
r
E0
p0
=
.
(96)
(E)
p
E
Thus for an endothermic process (energy absorption or Q < 0) one has a threshold value for the incoming
energy, Ethr = |Q| and one has for E Ethr
p
(E)
E Ethr .
(97)
For an exothermic process (with energy release Q > 0) one can scatter for any (positive) energy E and
one has near E 0
1
(E) .
(98)
E
505
a+b
c1 + c2
d1 + d2 + d3
(elastic scattering)
(inelastic scattering)
(99)
Nevertheless, one can deal with these processes, at least the two two ones, by considering the problem
in the center of mass (CM) system. Considering two particles with momenta p1 and p2 and masses
m1 and m2 , for which the only translationally invariant interaction that is allowed must be of the form
V (r 1 r 2 ) = V (r) with r = r 1 r 2 the relative coordinate. Since the flux factor is just given by
J = 0 |v 1 v 2 | = 0 |p|/,
(100)
where p is the relative momentum and the reduced mass one sees that the collision of two particles
indeed can be described by considering the scattering of one particle with reduced mass having the
relative momentum p, scattering of the potential V (r).
p2
p
V(r)
p1
Notes:
Note that in the scattering of one particle in an external potential, there is no translation invariance, hence no momentum conservation, while for two particles with a potential depending on
the relative coordinate there is translation invariance. The latter requires conservation of the total
momentum P = p1 + p2 , but not of the relative momentum.
In the limit that one of the masses becomes very large, the light particles momentum and mass,
indeed, coincide with relative momentum and reduced mass, so one finds (consistently) that the
heavy particle can be considered as scattering center.
5.3
In the scattering of electrons from an extended charge distribution e(r) the interaction is described by
Z
e2
(r 0 )
V (r) = d3 r0
.
(101)
40 |r r 0 |
Here the charge distribution could be e.g. that of one of the electrons in an atom described by the
squared orbital wave function (r) = |n`... (r)|2 . In an atom the full potential would then involve a sum
over all electrons, taking care of all antisymmetrization issues and a contribution of the atomic nucleus
with its own charge distributions (in turn linked to the proton and neutron wave functions, with in turn
again internal charge distributions linked to quark structure).
Let us first start with Eq. 101 which hopefully shows that compositeness can be dealt with in a
straightforward way. In perturbation theory, we need the Fourier transform V (q) of the potential, so
506
lets calculate that. Remember that the exp(iqr) comes from initial and final state wave functions of
the scattering particle, so ~q is the momentum transfer in the scattering process. One finds
Z
Z
e2 (r 0 )
V (q) =
d3 r d3 r0 exp(i qr)
4 0 |r r 0 |
Z
Z
e2
(r 0 )
e2 4
F (q),
(102)
=
d3 s d3 r0 exp(i q(s + r 0 ))
=
4 0
|s|
4 0 q 2
where we have introduced the form factor
Z
F (q) =
d3 r0 exp(i qr 0 ) (r 0 )
(103)
which is the Fourier transform of the charge density (r 0 ). The result for the cross section is
d
=
d
m e2
2 0 ~2 q 2
2
p0
2
|F (q)| .
p
(104)
and shows how to determine the charge distribution via a scattering process as a deviation from the point
scattering result.
Form factors
From the measurements of form factors one obtains via the Fourier transform the charge density,
Z
F (q) = d3 r exp(i qr) (r).
(105)
As before, in discussing the potential in momentum space, one has for a spherically symmetric density,
Z
4
dr r (r) sin(qr).
(106)
F (q) =
q
For a spherical distributions it is trivial to find by expanding the exponential exp(i q r) = 1 + i q r 1
2
2 (q r) + . . ., that
1
F (q) = Q q 2 hr2 i + . . . ,
(107)
6
R 3
R
where Q = d r (r) is the total charge and hr2 i = d3 r r2 (r) is the charge radius squared. The
small-q behavior of a form factor can thus be used to determine the charge radius of an atom, a nucleus
or a nucleon. Some examples of form factors corresponding to specific densities are:
A uniform density has a Bessel function j1 as Fourier transform (see Eq. 91),
(r) = 0
for x R
F (q) =
3 j1 (qR)
.
qR
(108)
sin x cos x
x2
x
and
3 j1 (x)
1 2
1
x + ...,
x
10
so the charge radius of a uniform distribution is hr2 i = 53 R2 . Examples of uniform densities are the
nuclear densities, although agreement with data can be improved by introducing a smooth fall-off
at the edge.
507
2 er
4 r
F (q) =
q2
1
2
=
,
+ 2
1 + q 2 /2
(109)
which is called a monopole form factor. We have encountered this same Fourier transform already
earlier where we derived the momentum space screened Coulomb potential.
The form factor of the (normalized) exponential distribution is simply found by differentiation of
the Yukawa form factor with respect to , yielding
(r) =
3 r
e
8
F (q) =
1
(1 + q 2 /2 )
2,
(110)
which is called a dipole form factor. The charge density of the proton behaves in this way, with the
experimental formfactor having 2 0.71 GeV2 .
Finally, for a normalized Gaussian distribution
1
(r) = 0 e 2 r
/R2
F (q) = e 2 q
R2
(111)
508
Exercises
Exc. 5.1: comparison with exact time dependence
Note that the result in Fermis golden rule (Eq. 75) looks somehow different from Eq. 68, but in fact it is
an approximation that agrees in perturbation theory where |Vpi | ~|pi |. The only sizable growth of ca
(as a 2-level system) comes from the region where r is minimal (but this is where |pi | = |ab | = 0 ,
thus we are in the regime r || 0 ). Show that in that case both Eq. 68 and Eq. 75 are approximated
by
|Vpi | i(pi ) /2
cp ( ) =
e
,
~
(solution)
In the exact result, Eq. 68, one needs to approximate sin(ax)/x a for x 0.
Z
and =
R2
= R2 ,
4
which (indeed) just is the area of the object that the particles bounce off.
(b) For Coulomb scattering in classical mechanics one has an hyperbola as solution. There are various
ways (ranging from straightforward solving Newton to a subtle use of angular and radial degrees
of freedom and conservation of angular momentum) to find the relation between impact parameter
and angular difference between asymptotic directions,
b() =
Ze2
cot(/2).
80 E
Dont derive this result, but just use it to derive d/d and show that it can be written in the form
m 2 Ze2 2 1
d
(E, ) =
,
d
2 ~2
0
q4
where ~2 q 2 = |p p0 |2 = 4p2 sin2 (/2), i.e. the dependence on energy and angle is only through
one variable (the momentum transfer).
509
(c) What happens if you calculate the cross section integrated over all angles. Which angles are the
problematic ones and do you understand the reason.
(solution)
The differential cross section diverges for small angles (small q) and the integrated cross section is
infinite. This is easy to understand because even for very large impact parameters the Coulomb
potential is felt, even if the scattering angle becomes ever smaller. The total area is infinite.
m2
M R = m1 r 1 + m2 r 2
r
r1 = R +
M
m1
r = r1 r2
r2 = R
r
M
where M = m1 + m2 is the total mass. Furthermore it will be convenient to introduce the reduced mass
m m1 m2 /(m1 + m2 ). If there are more particles or confusion may arise, we can use subscripts R12 ,
etc.
= P /M and r = p/m and show that they satisfy
(a) Construct the absolute and relative momentum R
m1
P = p1 + p2
p1 =
P +p
M
p
p1
p2
m2
=
p2 =
P p.
m
m1
m2
M
Show that one can rewrite the kinetic part of the energy as
Ek
m1 v 21 +
1
2
1
2
p22
p21
+
2m1
2m2
m2 v 22
1
2
MV2+
1
2
m v2
P2
p2
+
.
2M
2m
Furthermore show that the flux to be used in scattering two particles |v1 v2 | reduces to the result
of scattering just one particle with reduced mass.
(solution)
The results follow from
=p +p =P
MR
1
2
and r =
p
p
p1
2 =
m1
m2
m
= .
m1
m2
510
(b) [You can skip this one if you want.] Show that the transformation is a cononical one. To save
yourself some writing do this for just one of the coordinates, (x1 , px1 ) and (x2 , px2 ) and their
transformation into (X, Px ) and (x, px ).
(solution)
In order to show this one needs to show that the Poisson brackets {x1 , p1 }P = 1 and {x2 , p2 }P = 1
with the transformation lead to {X, P }P = 1 and {x, p}P = 1, and the inverse. Here the Poisson
bracket is defined
F G F G
.
{F, G}P
x p
p x
(c) [Note: do everything just for one coordinate.] For a consistent quantum mechanical treatment
the relations between the individial momenta and the CM and relative momenta need also to be
true for the operators in quantum mechanics. Check this by studying the transformation between
/x1 , /x2 and /X and /x and see that you recover the transformations for the momentum
operators as for the momenta under (a).
(solution)
We get for instance
x
m1
+
=
+
,
=
x1
x1 X
x1 x
M X
x
etc. After multiplying with i~ we get the required result.
(d) [Note: do everything just for one coordinate.] Rather than doing the check under (c) one can also
check if CM and relative quantities behave well in quantum mechanics by checking the commutation
relations. Relate [X, P ] and [x, p] to [x1 , p1 ] and [x2 , p2 ] and show that if the individual particles
satisfy the canonical commutation relations ([x1 , x1 ] = [p1 , p1 ] = 0 and [x1 , p1 ] = i~ and similarly for
2) one finds the right commutation relations for CM and relative position and momentum operators.
(solution)
This is straightforward, for instance
M [X, P ] = [m1 x1 + m2 x2 , p1 + p2 ] = m1 [x1 , p1 ] + m2 [x2 , p2 ] = i~ (m1 + m2 ) [X, P ] = i~.
(a) Calculate for a plane wave i (r) = p (r) = 0 exp(i k r) the flux j i .
(solution)
The result is
ji =
~
( i (i ) i ) = 0 ~k/m = 0 p/m = 0 v.
2i m i
511
1 d3 p0
.
0 (2~)3
(solution)
The answer is
d3 p0 = (p0 )2 dp0 d0 = mp0 dE 0 d0
(E 0 ) =
1 m p0
.
0 (2~)3
(c) Show that the matrix element Vf i is proportional to the Fourier transform of V (r),
Z
Vf i = 0
d3 r V (r) exp(i q r) = 0 V (q).
Express q in p and p0 . Calculate this Fourier transform in two steps, first calculate the Fourier
transform for V (r) = V0 er /r and then make the right substitutions for V0 and .
(solution)
One needs q = (p p0 )/~ in V (q). The potential is a central potential and we can use
Z
Z
4
2 V0
V (q) =
dr rV (r) sin(qr) =
dr er (eiqr eiqr )
q 0
iq
0
2 V0
1
1
4 V0
=
.
= 2
iq
iq + iq
q + 2
(d) Combine all results to find d/d.
(solution)
The result is
2
m 2 p 0
d
V
(q)
=
.
d
2~2
p
from which one taking the limit 0 of the Yukawa potential with V0 = Ze2 /40 obtains the
Rutherford cross section
2
m 2 Ze2 2 1
Ze2
d
1
(E, ) =
=
.
(112)
4
2
4
d
2 ~
0
q
8 0 pv
sin (/2)
If you wonder why the expression is written with the combination pv rather than using pv = p2 /m
or pv = mv 2 , that is done because with pv it also represents the exact relativistic result.
er
r
f(q) =
4
.
q 2 + 2
512
(a) The Fourier transform of a Yukawa spatial distribution is a know as the monopole form factor.
Give the normalized Yukawa distribution (r) and its form factor F (q).
(solution)
For a normalized distribution F (q = 0) = 0, so since f(0) = 4/2 we have
(r) =
2 er
4 r
F (q) =
q2
1
2
=
.
+ 2
1 + q 2 /2
(b) Using parametric differentiation with respect to the parameter you can also get from the result
in (a) also the form factor and normalization of a normalized exponential distribution (r) er .
(solution)
Differentiating (d/d) the results for f (r) f(q) we get the pair
f (r) = er
f(q) =
3 r
e
8
F (q) =
(q 2
8
,
+ 2 )2
1
(1 + q 2 /2 )
2,
(c) For a proton the charge distribution is approximately a dipole form factor with 2 = 0.71 GeV2
(using units where ~ = c = 1. As discussed in this section, the charge radius of the charge
distribution is related to the behavior of the form factor near q 2 = 0. Give the relation for the
charge radius of the proton and determine its value in fm.
(solution)
The charge radius is related to the slope of the form factor,
dF (q 2 )
2
.
hr i = 6
dq 2 q2 =0
For the dipole form factor this gives hr2 i = 12/2 and
p
hr2 i 4.1 GeV1 0.81 fm.
601
In this lecture, we return to chapter 9 of Griffiths and we discuss the emission and absorption of radiation,
applied mostly to atoms (Griffiths 9.2). Relevant issues in the quantummechanical treatment are also
spontaneous emission and lifetimes as well as the selection rules for transitions (Griffiths 9.3).
6.1
Electromagnetic waves can be treated as external oscillating electric and magnetic fields. We take a
plane wave for the scalar and vector potential. If you now are worried because you have learned that
electromagnetism deals with real fields, remember that we can write cos(t) = (eit + eit )/2 and
indeed oscillating electromagnetic fields can both describe emission (taking Q = ~ away from the atom)
or absorption (dumping Q = ~ into the atom). Thus we take for the electric potential and the vector
potential
= (k,
) exp[i(k r t)],
(113)
(114)
with = |k|c, corresponding with the energy and momentum relation, E = |p|c, for a massless photon.
The corresponding behavior for the electric and magnetic fields can be obtained from the potentials5
E = E(k,
) exp[i(k r t)] with E(k,
) = ik (k,
) + i|k| A(k,
),
(115)
(116)
The radiation fields can actually be obtained from only the vector potential6 , for which we (smartly)
choose
0 (k, )
cE
exp[i(k r t)]
(117)
A = (k, )
i
and = 0. The vector will precisely be the polarization of the light. One has
0 exp[i(k r t)],
E = E
k
E0 exp[i(k r t)].
B=
|k|
(118)
(119)
This corresponds to the well-known picture of E and B fields oscillating 90 degrees out of phase in two
planes orthogonal to the direction of motion. The energy density of the wave is just 0 E02 /2.
With this radiation field we study the interaction of matter which for an electromagnetic field is given
by
Z
Hint = d3 r [(r) (r) j(r) A(r)] ,
(120)
where and j are the charge and current distribution. The dipole approximation is valid when the wave
length = 2/|k| is much larger than the typical size of the system, e.g. for light ( 6000
A) and
atoms (size 1 10
A). In that case one can restrict oneself to the first nontrivial term in
exp[i(k r t)] = ei t (1 + i k r + . . .).
5 Recall
(121)
that
E =
1 A
,
c t
B =A
6 One can use gauge invariance to change the potentials + /t and A A + with (t, r) and arbitrary
function.
602
One obtains
Hint
i
) (1 + i k r) j(r) A(k,
(1 + i k r) (r) (k,
)
h
i
) D E(k,
) B(k,
) + . . . ,
= eit Q (k,
= e
it
d3 r
(122)
(123)
where we have used Eqs 115 and 116. The charge and current distributions give rise to charge, electric
and magnetic dipole moments,
Z
X
Q = d3 r (r) =
qi ,
(124)
i
Z
D=
d3 r r (r) =
qi r i ,
(125)
X qi
`i ,
mi
i
(126)
Z
=
d3 r r j(r) =
(127)
hp |D |i i E0 ei (pi ) 1
,
i~
i (pi )
(128)
which gives as before rise to a delta function ( pi ). With being the positive photon frequency, this
can only describe absorption of a photon, ~ = Ep Ei > 0. For the real electromagnetic fields also the
complex conjugate solution must be considered, which gives the same result with . This gives
rise to a delta function ( + pi ) and describes the emission of a photon, ~ = ~pi = Ei Ep > 0.
The transition probability can be summarized by
(1)
Pip ( ) =
E02 ()
(129)
If one is not working with monochromatic light one has an integral over different frequencies . Instead of the intensity of the field E0 = E0 () one can use the number of incident photons N ()
(number/(areatime)). This number is determined by equating the energy densities in a frequency
interval d,
1
0 E02 () d = N () ~ d.
(130)
2
Integrating over the photon frequencies, one sees that the atom absorps or emits photons of the right
frequency leading to a transition rate
(1)
Pip =
|pi | N (|pi |) |hp |D |i i|2 .
0 ~
(131)
P
For electrons D = i e r i = e R. For unpolarized light is arbitrary and averaging gives a factor
1/3. In terms of the fine structure constant = e2 /4 0 ~c the averaged transition rate is
(1)
Wip = Pip =
4 2
c |pi | N (|pi |) |hp |R|i i|2 .
3
(132)
6.2
603
Spontaneous emission
Besides the stimulated transmission, there is also the spontaneous emission corresponding to the emission
spont
of a photon. This is a rate A = Wip
which is not proportional to N (), in contrast to Wip and
Wpi , which are equal and could be set to B N (). Griffiths shows in section 9.3.1 how to obtain this
rate using some basic thermodynamics. In short the argument is putting the loss-gain for the higher level
(say b) to zero,
dNb
= Nb A Nb B N () + Na B N () = 0
(133)
dt
(A and B are known as Einstein coefficients). This gives
Na
A
= N ()
1 .
(134)
B
Nb
Using the equilibrium condition and the number of photons in thermal equilibrium (Eq. 5.113 in Griffiths),
Na
eEa /kB T
2
1
= E /k T = e~/kB T and N () = 2 3 ~/k T
,
B
Nb
c e
e b B
1
we get
3
4 ip
spont
Wip
(135)
= 2 |hp |R|i i|2 ,
3 c
governed by the same transition matrix element and thus obeying the same selection rules.
6.3
Unstable states
In many circumstances one encounters unstable states, i.e. the probability P to find a system in a
particular state decreases in time,
P (t + dt) = P (t) (1 dt)
dP
= P (t),
dt
where is the decay rate or decay probability per unit time. The solution is
P (t) = P (0) e t = P (0) et/T ,
(136)
with T = 1/ ~/ the lifetime. The quantity is referred to as the width of a state. For the lifetime
spont
of an atomic state, we need the spontaneous emission rate, thus = Wip
. Note that the decay rates
are additive if there are more decay channels, lifetimes are not! For a decaying state we thus write
|n (t)i ei En t/~n t/2 .
We can expand a decaying state in eigenmodes according to
Z
d
ei En t/~n t/2 (t) =
c() ei t ,
2
(137)
(138)
=
=
i
+i (n +in /2)t
e
n + in /2
0
i
.
n + in /2
(139)
604
For unstable states the transition amplitude for emission or absorption of a photon is then proportional
to
Z
d10 d20 0
T () =
c ( ) c1 (10 ) 2 ( 10 + 20 )
2 2 2 2
Z
d 0 0
=
c ( ) c1 ( 0 + )
2 2
Z
Z
Z
0
0
d 0
dt1
dt2 ei ( 2 i2 /2)t2 e+i ( +1 +i1 /2)t1
=
2 0
0
Z
Z
=
dt1
dt2 e+i (2 +i2 /2)t2 e+i (1 +i1 /2)t1 (t1 t2 )
0
0
Z
i
dt e+i (12 +i12 /2)t =
=
,
(140)
12 + i12 /2
0
where 12 = 1 2 and 12 = 1 + 2 . Thus the line-intensity becomes instead of a delta-function
( |pi |) proportional to
1
I() |T ()|2
2 /4 ,
( 12 )2 + 12
or normalizing to the peak intensity
I() = I0
~2
212 /4
,
( 12 )2 + 212 /4
(141)
showing the reason for the name width. The quantity 12 is precisely the width of the peak at halfmaximum intensity, when plotting I as a function of the photon energy ~. The function is known as a
Lorentzian distribution or a Breit-Wigner distribution.
6.4
Specific matrix elements of the form h|O|i describe for instance transitions between states and .
For instance the absorption or emission of a photon in a state produces the state O|i. The probability
that this state is observed as a state |i is
2
Prob = |h|O|i| .
For particular operators O we can make statements on the transition being allowed for specific angular
momentum quantum numbers of the states (similar as illustrated for parity).
We first note that for any operator S commuting with the angular momentum operators ([J , S] = 0),
the quantum numbers are not changed, thus
hJ 0 M 0 |S|J, M i JJ 0 M M 0 .
(142)
[Jy , V ] = i~Vz ,
[Jz , V ] = ~V ,
(143)
or using J ,
[J+ , V+ ] = [J , V ] = 0,
[J , V ] = 2~ Vz .
(144)
605
These can just as in the analysis of angular momentum quantum numbers be used to show that
Jz V |J, M i = (M 1) ~ V |J, M i,
(145)
From this one obtains selection rules stating that one can have nonzero transition probabilities only for
specific changes of quantum numbers,
for Vz :
M = M M 0 = 0,
for V+ :
M = M M 0 = +1,
for V :
M = M M 0 = 1.
Because of the commutation relations with the J operators one refers to this as the spherical basis
V`m = V1m ,
Vx i Vy
V11 =
,
V10 = Vz .
(146)
2
These are chosen such that
[Jz , O`m ] = m ~ O`m
and
[J , O`m ] = ~
(147)
(148)
The quantity hJ 0 ||O` ||Ji is referred to as reduced matrix element. The most well-known applications
of the Wigner-Eckart theorem are its application to the multipole operators describing the interactions
of photons with matter, as well as selection rules in interactions between atomic nuclei and elementary
particles with different spin states. It reduces (2J + 1)(2` + 1)(2J 0 + 1) matrix elements to one reduced
matrix element and the use of (tabulated) Clebsch-Gordan coefficients.
Emission and absorption of light
For emission and absorption of light (photons) the relevant operator is the electric dipole operator, which
in essence
P is the position operator, or more precisely the position operator weighted by the charges,
D = i ei r i . The calculation of transition probabilities are given by the matrix elements between the
appropriate states. Since the dipole operator does not involve spin operators, we have [Di , Sj ] = 0 and the
spin wave function doesnt change in a dipole transition, giving rise to a spin selection rule: ms 1 = ms 2 ,
i.e.
s = ms = 0.
(149)
In fact the photon polarization determines which of the components of the position operator is the relevant
operator. The photon polarizations for linearly or circularly polarized light (directed in z-direction) are
linear :
x = x
and y = y,
circular : + = (
x + i y)/ 2 and = (
x i y)/ 2
(150)
(151)
For polarized light we thus precisely get the (spherical) representation of the position vector in terms of
the two spherical harmonics with ` = 1,
r
4
m
r m =
rY1 .
(152)
3
606
4
3
Z
dr r un1 `1 (r) un2 `2 (r)
1
d Y`m
(, ) Y1
1
2
(, ) Y`m
(, ).
2
You can already see from the -dependence of the spherical harmonics that the matrix element is proportional to
Z
d ei m1 ei m ei m2 = 2 (m2 + m m1 ),
giving rise to the selection rule
m` = 1,
each of these corresponding to a specific photon polarization. The integral for the -dependent part is
simple, but more general one can use the properties of the Y`m -functions to see what happens with the
full angular integration. One only gets a nonzero result if the addition of angular momenta |`2 , m2 i and
|1, m i can yield the final state |`1 , m1 i via the well-known angular momentum addition rules. The result
is simply proportional to the Clebsch-Gordan coefficient in this recoupling,
r
Z
4
h1|r |2i =
dr r un1 `1 (r) un2 `2 (r) C(1, m , `2 , m2 ; `1 , m1 ).
3
which is an explicit example of the Wigner-Eckart theorem, and of course precisely follows the general
framework pointed out above.
This leads besides the m-selection rule to |`| 1 Knowing the parity of the spherical harmonics one
immediately gets a parity selection rule, namely 1 2 = 1 or with = ()` , one is left with
` = 1.
(153)
Rotational invariance further requires that the sum of the total angular momentum in initial and final
state is conserved. This becomes relevant if the orbital angular momentum and spin of electrons and/or
atomic nuclei are coupled to a specific total angular momentum. In many cases the orbital angular
momentum then is no longer a good quantum number. Still, even when ` and s are coupled to j, or for
m
many particles L and S are coupled to J, the transition operator involves a simple Y1 , implying
J = 0, 1
(154)
607
Exercises
Exc. 6.1: commutation rules and selection rules
(a) Prove, starting with [xi , pj ] = i~ ij the commutation relations for ` = r p (or its components
`i = ijk xj pk )
[`i , xj ] = i~ ijk xk and [`i , pj ] = i~ ijk pk .
(b) Show that if [`i , Aj ] = i~ ijk Ak and [`i , Bj ] = i~ ijk Bk one has
[`i , AB] = 0
and
(solution)
(a) [`i , xj ] = imn [xm pn , xj ] = imn xm [pn , xj ] = i~ imn xm nj = i~ imj xm = i~ ijm xm and
similarly for pj .
(b) Straightforward, most easily done for a component and then generalized.
(c) [`2 , Az ] = [`2x + `2y + `2z , Az ] = `x [`x , Az ] + [`x , Az ]`x + `y [`y , Az ] + [`y , Az ]`y = i~(`x Ay Ay `x +
Ax `y + `y Ax ) = 2~2 Az + 2i~(Ax `y Ay `x ).
(d) Apply (c) to A = r ` and use that (r `) ` = 0 because r` = 0. This relation is used in
Griffiths to discuss selection rules for the quantum number `.
satisfying a |ni =
n|n 1i.
(a) Calculate the nonzero expectation values hn0 |x|ni, e.g. hn + 1|x|ni = . . ., etc.
(b) Use this to calculate for a charged particle contained in such a harmonic oscillator the spontaneous
spont
2
2
transition rate Wnn
0 and express it in e /0 as well as in = e /40 ~c.
608
(solution)
(a) We have
r
0
hn |x|ni =
~
hn0 |a + a |ni, thus hn 1|x|ni =
2m
n~
and hn + 1|x|ni =
2m
r
(n + 1)
~
.
2m
Note that the stimulated transition rates in a given level thus are not the same. Transitions towards
the state with higher occupation is favored (used in lasers).
(b) To calculate the spontaneous transition rate from n to n 1 we also need n,n1 = . Then we get
spont
Wnn1
=
2
ne2 2
~
2 n~ 2
=
=
.
2
2
3 mc
3
mc
60 mc3
The various expressions in particular the one in terms of e2 /0 does not have an ~, indicating the
problem is classical. Indeed this is the case (see Griffiths Example 9.1).
(solution)
(a) A dipole transition requires s = ms and ` = m` = 1, thus that is possible in this case
between 6 degenerate 2 P and 2 degenerate 2 S states.
(b) The splitting in a magnetic field gives
E(`sML MS ) = B B(m` + 2 ms ).
indicated in the figure.
ML
1
0
+
-1
0
-1
609
2MS
1
1
-+1
-1
-1
-1
610
Spin
A
A.1
701
Spin
Rotational invariance (extended to spinning particles)
As discussed earlier in addition to orbital angular momentum a system, elementary or composite can
have a spin, which may include internal orbital angular momentum. Hence spin is a vector operator s,
which is a part of the rotation operator. These (three) hermitean operators satisfy commutation relations
[si , sj ] = i~ijk sk ,
(155)
similar to the commutation relations for the angular momentum operator ` = r p. The spin operators
s commute with the operators r and p and thus also with `. Thats it. All the rest follows from these
commutation relations.
For the orbital angular momentum, we have seen the explicit link to spatial rotations, `z = i~ /.
That means that scalars (e.g. numbers) are not affected. Also operators that do not have azimuthal
dependence are not affected, which means for operators that they commute with `z . The various components of a vector do depend on the azimuthal angle and vector operators such as r or p do not commute
with `z . In fact scalar operators S and vector operators V satisfy
[`i , S] = 0
for a single particle without spin,
(156)
[`i , Vj ] = i~ ijk Vk
e.g. scalars S = r 2 , p2 , r p or `2 and vectors V = r, p or `. Including spin vectors s, this behavior
seems to be no longer true, e.g. [`i , sj ] = 0 and [`i , ` s] = i~ (` s)i . But with the correct rotation
operator
j ` + s,
(157)
we do get
[ji , S] = 0
[ji , Vj ] = i~ ijk Vk ,
for a single particle
(158)
not only for the above examples, but now also for the vectors s and j and scalars like s2 and ` s.
For a system of many particles the operators r, p and s for different particles commute. Their sums
L=
N
X
n=1
`n ,
S=
N
X
sn ,
J=
n=1
N
X
j n = L + S,
(159)
n=1
also satisfy angular momentum commutation relations [Li , Lj ] = i~ ijk Lk , [Si , Sj ] = i~ ijk Sk , and
[Ji , Jj ] = i~ ijk Jk , while only the operator J satisfies
[Ji , S] = 0
for an isolated system
(160)
[Ji , Vj ] = i~ ijk Vk
for any scalar operator S or vector operator V .
==========================================================
Exercise: Show that from [Ji , Aj ] = i~ ijk Ak and [Ji , Bj ] = i~ ijk Bk (satisfying Eq. 160) one finds
[Ji , A B] = 0,
i.e. if A and B are vector operators, AB is a scalar operator. Prove this for Jx without loss of generality.
==========================================================
Spin
702
An important property is that rotational invariance is one of the basic symmetries of our world, which
reflects itself in quantum mechanics as
Rotation invariance of a system of particles requires
[J , H] = 0,
(161)
P
where J = L+S = i (`i +si ). This is the (generalized, cf. Eq. ??) fundamental rotational symmetry
of nature for a system of many particles including spin!
Besides the behavior under rotations, also the behavior under parity is considered to classify operators.
Vector operators behave as P V P 1 = V , axial vectors as P A P 1 = +A, a scalar operator S
behaves as P S P 1 = +S, and a pseudoscalar operator S 0 behaves as P S 0 P 1 = S 0 . Examples of
specific operators are
vector
r
p
axial vector
`
s
j
scalar
r2
p2
`2
`s
pseudoscalar
sr
sp
The hamiltonian is a scalar operator. Therefore, if we have parity invariance, combinations as s r cannot
appear but a tensor operator of the form (s1 r)(s2 r) is allowed. Note, however, that such an operator
does not commute with `.
A.2
Spin states
As mentioned above, the commutation relations are all that defines spin. As an operator that commutes
with all three spin operators (a socalled Casimir operator) we have s2 = s2x + s2y + s2z ,
[si , sj ] = i~ ijk sk ,
(162)
[s2 , si ] = 0.
(163)
Only one of the three spin operators can be used to label states, for which we without loss of generality
(s)
can take sz . In addition we can use s2 , which commutes with sz . We write states m = |s, mi satisfying
s2 |s, mi = ~2 s(s + 1)|s, mi,
(164)
(165)
It is of course a bit premature to take ~2 s(s + 1) as eigenvalue. We need to prove that the eigenvalue
of s2 is positive, but this is straightforward as it is the sum of three squared operators. Since the spin
operators are hermitean each term is not just a square but also the product of the operator and its
hermitean conjugate. In the next step, we recombine the operators sx and sy into
s sx i sy .
(166)
(167)
[sz , s ] = ~ s ,
(168)
[s+ , s ] = 2~ sz ,
(169)
Spin
703
It is convention to define
p
s+ |s, mi = ~ s(s + 1) m(m + 1) |s, m + 1i
p
= = ~ (s m)(s + m + 1) |s, m + 1i
p
s |s, mi = ~ s(s + 1) m(m 1) |s, m 1i
p
= ~ (s + m)(s m + 1) |s, m 1i.
(170)
(171)
This shows that given a state |s, mi, we have a whole series of states
. . . |s, m 1i, |s, mi, |s, m + 1i, . . .
But, we can also easily see that since s2 s2z = s2x + s2y must be an operator with positive definite
p
eigenstates that s(s + 1) m2 0, i.e. |m| s(s + 1) or strictly |m| < s + 1. From the second
expressions in Eqs 170 and 171 one sees that this inequality requires mmax = s as one necessary state
to achieve a cutoff of the series of states on the upper side, while mmin = s is required as a necessary
state to achieve a cutoff of the series of states on the lower side. Moreover to have both cutoffs the step
operators require that the difference mmax mmin = 2 s must be an integer, i.e. the only allowed values
of spin quantum numbers are
s = 0, 1/2, 1, 3/2, . . . ,
m = s, s 1, . . . , s.
Thus for spin states with a given quantum number s, there exist 2s + 1 states.
A.3
Why is ` integer
Purely on the basis of the commutation relations, the allowed values for the quantum numbers s and m
have been derived. Since the angular momentum operators ` = r p satisfy the same commutation relations, one has the same restrictions on ` and m` , the eigenvalues connected with `2 and `z . However, we
have only found integer values for the quantum numbers in our earlier treatment. This is the consequence
of restrictions imposed because for ` we know more than just the commutation relations. The operators
have been introduced explicitly working in the space of functions, depending on the angles in R3 . One
way of seeing where the constraint is coming from is realizing that we want uni-valued functions. The
eigenfunctions of `z = i~ d/d, were found to be
Y`m (, ) ei m .
Spin
704
In order to have the same value for and + 2 we need exp(2 i m) = 1, hence m (and thus also `) can
only be integer.
For spin, there are only the commutation relations, thus the spin quantum numbers s can also take
half-integer values. Particles with integer spin values are called bosons (e.g. pions, photons), particles
with half-integer spin values are called fermions (e.g. electrons, protons, neutrinos, quarks). For the
angular momenta which are obtained as the sum of other operators, e.g. j = ` + s, etc. one can easily see
what is allowed. Because the z-components are additive, one sees that for any orbital angular momentum
the quantum numbers are integer, while for spin and total angular momentum integer and half-integer
are possible.
A.4
In the space of spin states with a given quantum number s, we can write the spin operators as (2s + 1)
(2s + 1) matrices. Let us illustrate this first for spin s = 1/2. Define the states
(1/2)
1
and |1/2, 1/2i (1/2)
.
1/2
0
1
Using the definition of the quantum numbers in Eq. 165 one
1/2
0
0 1
sz = ~
, s+ = ~
0 0
0 1/2
finds that
, s = ~
1
,
0
For spin 1/2 we then find the familiar spin matrices, s = ~/2,
0 i
0 1
1
x =
, z =
, y =
0
i 0
1 0
.
1
For spin 1 we define the basis states |1, +1i, |1, 0i and |1, 1i or
0
0
1
(1)
(1)
(1)
,
0
0
1
,
+1
1
0
1
0
0
0
1 0 0
0
0
0
,
s
=
~
sz = ~
0 0 1
0
2 0
,
s
=
~
2
0
2
0
0
0
0 0
0
0
,
2 0
A.5
e.g.
Instead of the spin states defined as eigenstates of sz , one might be interested in eigenstates of s n,
because one wants to measure it with a Stern-Gerlach apparatus with an inhomogeneous B-field in the
(s)
direction. We choose an appropriate notation like |n,
i or two component spinors ms (n),
shorthand
n
(1/2)
= + (n)
+1/2 (n)
(1/2)
= (n)
Spin
705
Suppose that we want to write them down in terms of the eigenstates of sz , given above, +/ (
z) =
=
/ . To do this we work in the matrix representation discussed in the previous section. Taking n
(sin , 0, cos ), we can easily write down
~
1
sin
cos
=
= ~ n
sn
(172)
.
sin cos
2
2
We find the following two eigenstates and eigenvalues
cos(/2)
=
+ (n)
sin(/2)
sin(/2)
=
(n)
cos(/2)
with eigenvalue
+ ~/2,
with eigenvalue
~/2.
The probability that given a state + with spin along the z-direction, a measurement of the spin along
the +n-direction
yields the value +~/2 is thus given by
2
+ (n)
= cos2 (/2).
Instead of explicitly solving the eigenstates, we of course also can use the rotation operators in quantum
mechanics,
= exp (i n
J) ,
U (, n)
(173)
where J is the total angular momentum operator referred to before in chapter one and in the section on
rotation invariance. The total angular momentum operator is the generator of rotations.
A simple example is the rotation of a spin 1/2 spinor. The rotation matrix that brings a spin up state
along the z-axis into a spin up state along a rotated direction with polar angle in the x z plane is7 is
U () = ei y /2 = cos(/2) I i sin(/2) y .
where I is the 2 2 unit matrix.
==========================================================
= U () Sz U 1 ().
Exercise: Check that U ()|1/2, 1/2i gives the rotated spin state above, while S n
==========================================================
To find the expansion of any rotated spinor in the original spin states one considers in general the
Euler rotations of the form
U (, , ) = ei Jz ei Jy ei Jz
Its matrix elements are referred to as the D-functions,
(j)
(j)
(174)
dsm ()
..
(s)
(s)
= dm0 m ()
m (n)
.
(s)
dsm ()
(175)
Spin
706
where dm0 m () are the d-functions. These are in fact just matrix elements of the spin rotation matrix
exp(i Jy ) between states quantized along the z-direction. Extended to include azimuthal dependence
it is necessary to use the rotation matrix ei Jz ei Jy ei Jz and the relevant functions are called
Dm0 m (, , ). For integer ` one has
r
4
(`)
Dm0 (, , 0) =
Y m (, ),
(176)
2` + 1 `
r
4
(`)
Dm0 (0, , ) = ()m
Y m (, ).
(177)
2` + 1 `
==========================================================
Exercise: Give the explicit matrix U (, , ) for j = 1/2.
==========================================================
707
B.1
We consider situations in which two sets of angular momentum operators play a role, e.g.
An electron with spin in an atomic (n`)-orbit (spin s and orbital angular momentum ` combined
into a total angular momentum j = ` + s). Here one combines the R3 and the spin-space.
Two electrons with spin (spin operators s1 and s2 , combined into S = s1 + s2 ). Here we have the
product of spin-space for particle 1 and particle 2.
Two electrons in atomic orbits (orbital angular momenta `1 and `2 combined into total orbital
angular momentum L = `1 + `2 ). Here we have the direct product spaces R3 R3 for particles 1
and 2.
Combining the total orbital angular momentum of electrons in an atom (L) and the total spin (S)
into the total angular momentum J = L + S.
Let us discuss as the generic example
J = j1 + j2.
(178)
(179)
which we can write down since not only [j 21 , j1z ] = [j 22 , j2z ] = 0, but also [j1m , j2n ] = 0. The sumoperator J obviously is not independent, but since the J -operators again satisfy the well-known angular
momentum commutation relations we can look for states characterized by the commuting operators J 2
and Jz , | . . . ; J, M i. It is easy to verify that of the four operators characterizing the states in Eq. 179,
[J 2 , j1z ] 6= 0 and [J 2 , j2z ] 6= 0 (Note that J 2 contains the operator combination 2j 1 j2 , which contains
operators like j1x , which do not commute with j1z ). It is easy to verify that one does have
[J 2 , j 21 ] = [J 2 , j 22 ] = 0,
[Jz , j 21 ] = [Jz , j 22 ] = 0,
and thus we can relabel the (2j1 + 1)(2j2 + 1) states in Eq. 179 into states characterized with the quantum
numbers
|j1 , j2 ; J, M i.
(180)
The basic observation in the relabeling is that Jz = j1z + j2z and hence M = m1 + m2 . This leads to the
following scheme, in which in the left part the possible m1 and m2 -values are given and the upper right
part the possible sum-values for M including their degeneracy.
M
j1 + j2
=
m1
- j1
m2
x
- j2
j1
j2
+
+
j1 - j2
708
1. Since |m1 | j1 and |m2 | j2 , the maximum value for M is j1 + j2 . This state is unique.
2. Since J+ = j1+ + j2+ acting on this state is zero, it corresponds to a state with J = j1 + j2 . Then,
there must exist other states (in total 2J + 1), which can be constructed via J = j1 + j2 (in
the scheme indicated as the first set of states in the right part below the equal sign).
3. In general the state with M = j1 + j2 1 is twofold degenerate. One combination must be the state
obtained with J from the state with M = j1 + j2 , the other must be orthogonal to this state and
again represents a maximum M -value corresponding to J = j1 + j2 1.
4. This procedure goes on till we have reached M = |j1 j2 |, after which the degeneracy is equal to
the min{2j1 + 1, 2j2 + 1}, and stays constant till the M -value reaches the corresponding negative
value.
Thus
Combining two angular momenta j1 and j2 we find resulting angular momenta J with values
J = j1 + j2 , j1 + j2 1, . . . , |j1 j2 |,
(181)
(182)
J=|j1 j2 |
B.2
Clebsch-Gordon coefficients
The actual construction of states just follows the steps outlined above. Let us illustrate it for the case of
combining two spin 1/2 states. We have four states according to labeling in Eq. 179,
|s1 , m1 i |s2 , m2 i :
1. The highest state has M = 1 and must be the first of the four states above. Thus for the labeling
|s1 , s2 ; S, M i
|1/2, 1/2; 1, +1i = | i.
(183)
2. Using S = s1 + s2 we can construct the other S + 1 states.
709
and thus
1
|1/2, 1/2; 1, 0i = | i + | i .
(184)
2
Continuing with S (or in this case using the fact that we have the lowest nondegenerate M -state)
we find
|1/2, 1/2; 1, 1i = | i.
(185)
3. The state with M = 0 is twofold degenerate. One combination is already found in the above
procedure. The other is made up of the same two states appearing on the right hand side in
Eq. 184. Up to a phase, it is found by requiring it to be orthogonal to the state |1/2, 1/2; 1, 0i or
by requiring that S+ = s1+ + s2+ gives zero. The result is
1
(186)
|1/2, 1/2; 0, 0i = | i | i .
2
The convention for the phase is that the higher m1 -value appears with a positive sign.
It is easy to summarize the results in a table, where one puts the states |j1 , m1 i |j2 , m2 i in the
different rows and the states |j1 , j2 ; J, M i in the different columns, i.e.
..
.
..
.
j1 j2
J
M
..
.
..
.
... ...
m1 m2
... ...
For the above case we have
1/2 1/2
+1/2
+1/2
+1/2
1/2
1/2
1/2
+1/2
1/2
1
1
1
1
0
0
0
1
q2
1
2
1
-1
1
q2
12
1
Note that the recoupling matrix is block-diagonal because of the constraint M = m1 +m2 . The coefficients
appearing in the matrix are the socalled Clebsch-Gordan (CG) coefficients. We thus have
X
|j1 , j2 ; J, M i =
C(j1 , m1 , j2 , m2 ; J, M ) |j1 , m1 i |j2 , m2 i.
(187)
m1 ,m2
Represented as a matrix as done above, it is unitary (because both sets of states are normalized). Since
the Clebsch-Gordan coefficients are choosen real, the inverse is just the transposed matrix, or
X
|j1 , m1 i |j2 , m2 i =
C(j1 , m1 , j2 , m2 ; J, M ) |j1 , j2 ; J, M i.
(188)
J,M
In some cases (like combining two spin 1/2 states) one can make use of symmetry arguments. If a
particular state has a well-defined symmetry under permutation of states 1 and 2, then all M -states
belonging to a particular J-value have the same symmetry (because j1 +j2 does not alter the symmetry.
This could have been used for the 1/2 1/2 case, as the highest total M is symmetric, all S = 1 states
are symmetric. This is in this case sufficient to get the state in Eq. 184.
We will give two other examples. The first is
+1/2
+1
1/2
+1/2
1/2
1
1
+1/2
1/2
710
3/2
+3/2
1
3/2
+1/2
1/2
+1/2
q
1
q3
2
3
3/2
1/2
1/2
1/2
3/2
3/2
q3
1
3
2
q3
1
3
q3
2
3
for instance needed to obtain the explicit states for an electron with spin in an (2p)-orbit coupled to a
total angular momentum j = 3/2 (indicated as 2p3/2 ) with m = 1/2 is
!
r
r
u2p (r)
1 1
2 0
(r, t) =
Y (, ) +
Y (, ) .
r
3 1
3 1
The second example is
11
+1
+1
+1
+1
+1
+1
1
1
0
1
2
+2
1
2
+1
1
+1
1
q2
1
2
2
0
1
0
0
0
2
1
1
1
2
2
q2
1
2
1
q6
2
q3
1
6
1
2
0
q
1
2
q3
13
q
1
3
1
q2
1
2
q2
12
1
This example, useful in the combination of two spin 1 particles or two electrons in p-waves, also illustrates
the symmetry of the resulting wave functions.
==========================================================
Exercise: Repeat the steps outlined for 1/2 1/2 = 0 1 for one of the cases above and calculate the
CG coefficients.
==========================================================
Exercise: You can find CG coefficients using ClebschGordan[{j1,m1},{j2,m2},{J,M}] in Mathematica. Check some examples from previous Tables and construct the wavefunction for the state
J = 3/2, M = 1/2 in terms of the eigenstates of j12 , j1z , j22 en j2z with j1 = 1 and j2 = 3/2.
==========================================================
Exercise: The table (given below) contains Clebsch-Gordan coefficients for some general cases. Verify
at least one of the entries in the first Table. Check the |l1 , l2 ; l, mi = |1, 1; 0, 0i state explicitly using our
1 1 calculation above and the Table.
==========================================================
711
m2 =
j1 +
1
2
j1
1
2
j1 +m+ 12
2j1 +1
1
2
m2 = 12
12
j1 m+ 12
2j1 +1
12
j1 m+ 12
2j1 +1
j1 +m+ 12
2j1 +1
12
12
B.3
m2 = 1
h
i 21
m2 = 1
m2 = 0
h
i 12
i 21
i 12
1
[j1 (j1 +1)] 2
i 12
i 12
i 12
i 12
The possibility of finding in the coupling j1 j2 the angular momentum j3 actually also implies that
j3 j2 contains j1 , or that j1 j2 j3 couples to zero. For that purpose one has introduced the 3j-symbol
j1
j2
j3
,
m1 m2 m3
which is just a notation (no matrix or so!). It has nice symmetry properties and is related to various
CG coefficients as C(j1 , m1 , j2 , m2 ; j3 , m3 ) or C(j1 , m1 , j3 , m3 ; j2 , m2 ), etc. The proportionality constants
turn
out to be independent ofj the magnetic quantum numbers mi , but do contain typically some factors
2j + 1 and sign factors () .
Similarly one finds that the recoupling of three momenta to a total momentum J, can be done in
different ways with as intermediate steps j1 j2 J12 or j1 j3 J13 . The recoupling between such
different basis states is independent of magnetic quantum numbers with overlap matrix elements being
proportional to the Wigner 6j-symbol, denoted
j2 j1 J12
.
j3 J J13
Again the precise form with several factors 2j + 1 is constructed to exhibit many nice symmetries under
the interchange of the angular momenta involved.
The recouplings of four momenta by first recoupling particular pairs are related through 9j-symbols,
denoted
j2 J12
j1
j3
j4 J34
.
J13 J24 J
Properties and definitions of these Wigner symbols can be found in e.g. Messiah.