Some Notes on Scattering

Eef van Beveren

Centro de Fı́sica Teórica
Departamento de Fı́sica da Faculdade de Ciências e Tecnologia
Universidade de Coimbra (Portugal)
http://cft.fis.uc.pt/eef

28 de Outubro de 2005
i
Contents

1 Introduction 1

2 Wave packets 3
2.1 One dimensional wave packet . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1 Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.2 The Heisenberg uncertainty relation . . . . . . . . . . . . . . . . . . 7
2.1.3 The time development of a wave packet . . . . . . . . . . . . . . . . 8
2.1.4 Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.5 Momentum, velocity and energy . . . . . . . . . . . . . . . . . . . . 11
2.1.6 Wave equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.7 Plane wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Three dimensional wave packet . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.1 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Velocity and wave equation . . . . . . . . . . . . . . . . . . . . . . 15

3 Scattering 17
3.1 Scattering of wave packets . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2 Differential scattering cross section . . . . . . . . . . . . . . . . . . . . . . 20
3.3 The retarded Green’s functions . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4 Lippmann Schwinger equation for the wave function . . . . . . . . . . . . . 25
3.5 Asymptotic behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.6 Initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.7 Spherical well . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.7.1 The effective range expansion . . . . . . . . . . . . . . . . . . . . . 31
3.7.2 One-delta-shell approximation . . . . . . . . . . . . . . . . . . . . . 32
3.7.3 A formal solution of the one-delta-shell potential . . . . . . . . . . . 34
3.7.4 Multi-delta-shell approximation . . . . . . . . . . . . . . . . . . . . 36
3.7.5 Detailed study of the delta-shell approximation . . . . . . . . . . . 38
3.7.6 A closer look at the X-matrices . . . . . . . . . . . . . . . . . . . . 39

4 Formal scattering theory 42

4.1 The M6 oller operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 The scattering operator S . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 The Green’s operator G0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.4 The LS equation in momentum space . . . . . . . . . . . . . . . . . . . . . 49
4.5 Relation Green’s operators and M6 oller operators . . . . . . . . . . . . . . . 51
4.6 The LS equation for the Green’s operators . . . . . . . . . . . . . . . . . . 53

ii
 4.7 The transition operator T . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.8 Relation T and S operators . . . . . . . . . . . . . . . . . . . . . . . . . . 55

5 Examples 58
5.1 The potential term . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.1.1 Local potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.2 Spherically symmetric local potentials . . . . . . . . . . . . . . . . . 60
5.2 Relation transition amplitude and T -matrix . . . . . . . . . . . . . . . . . 61
5.3 The delta-shell potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.4 The T -operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.4.1 The delta-shell potential in momentum space . . . . . . . . . . . . 68
5.4.2 The T -matrix elements for the delta-shell . . . . . . . . . . . . . . 68
5.4.3 Details of the ~
k integration . . . . . . . . . . . . . . . . . . . . . . 73
5.5 Meson-meson scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.5.2 The wave equation for the model . . . . . . . . . . . . . . . . . . . 76
5.5.3 The radial wave equation . . . . . . . . . . . . . . . . . . . . . . . . 77
5.5.4 Coupled channel wave equation with one delta shell . . . . . . . . . 78
5.5.5 Bound states and resonances . . . . . . . . . . . . . . . . . . . . . . 81
5.6 The T -matrix for the meson-meson model . . . . . . . . . . . . . . . . . . 84

6 Relativistic kinematics 90
6.1 Relativistic kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.1.1 π + π − → K + K ∗ − . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.1.2 Elastic Scattering in the center-of-mass system . . . . . . . . . . . . 93
6.1.3 Elastic Scattering in the lab system . . . . . . . . . . . . . . . . . . 95
6.1.4 Scattering in the forward direction . . . . . . . . . . . . . . . . . . 96
6.2 Crossing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.3 The physical region for elastic scattering . . . . . . . . . . . . . . . . . . . 98
6.3.1 The subthreshold crescent . . . . . . . . . . . . . . . . . . . . . . . 101
6.4 The amplitude for elastic πN scattering . . . . . . . . . . . . . . . . . . . 102
6.4.1 The t channel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.4.2 The transition-matrix elements . . . . . . . . . . . . . . . . . . . . 106

iii
Chapter 1

Introduction

<< Much of what we know about the fundamental forces of Nature has been learned
from scattering experiments, in which essentially a target is bombarded with a beam
of particles. Well-known examples are electron-positron, electron-proton and protron-
antiproton scattering. But many other beam and/or target particles have been and are
presently used in a rich variety of different types of experiments. The most famous perhaps
of all scattering experiments is the dispersion of alpha particles on gold atoms, performed
by Rutherford and his collaborators in the beginning of the 20th century.
Usually we know the nature of the particles used as projectiles, their energy and mo-
mentum, and perhaps their polarization. These particles are scattered by the target and
subsequently, at distances large compared to the size of the target, detected by devices
that may give us the intensity as a function of the direction of scattering.
From the theoretical point of view the most significant aspect of scattering processes
is that we are concerned with the continuous part of the energy spectrum. We are free to
choose the energy of the incident particles. Consequently, our interest is focused on the
prediction of intensities since they are the object of measurement, rather than inspecting
a Schrödinger equation for its eigenvalues.
Intensities, being measures of the likelihood of finding a particle at certain places, are
of course related to the eigenfunctions. But, the connection between eigenfunctions and
measured intensities is recondite and indirect. Scattering data can be compared with the-
oretical predictions only if we elucidate carefully the various stages of a scattering process.
Relating observed intensities to calculated wave functions or transition probabilities, is
the first problem of scattering theory.
In an idealized scattering experiment a single fixed scattering center is bombarded by
particles incident along the z-axis. It is assumed that the effect of the scattering center on
the particles can be represented by some kind of interaction, which is appreciably different
from zero only within a finite region of small dimension.
Let us imagine a collimated homogeneous beam of monoenergetic particles moving
toward the scatterer from a great distance. The width of the beam could be determined
by slits, which although quite narrow from an experimental point of view (millimeters
or even micrometers), are nevertheless very wide compared with the spatial extension of
the scattering region (Fermies to Ångstr6 oms). Experimentally, in the interest of securing
good statistics, which is the maximum number of counts in a given period of operation, it
is desirable to employ intense beams. Yet the beam density must be low enough so that
it can safely be assumed that the incident particles do not interact with one another.

1
 After scattering, the particles are detected at a great distance from the scatterer. The
detector subtends a cone of solid angle dΩ at the origin, where the target is placed, and
the particles scattered into this cone are counted. If I0 is the number of incident particles
per unit area and IdΩ the number of these scattered into the cone, then we define the
differential scattering cross section as

dσ I(ϑ, ϕ)
= . (1.1)
dΩ I0
This is the quantity which the experimentalist delivers to the theoretician. The latter
interprets the cross section in terms of probabilities calculated from a theory. >>
(From chapter 11 of ”Quantum Mechanics” by Eugen Merzbacher [1])
In this notes, which are not intended to be complete neither rigorous, we first study
the non-relativistic Schrödinger formalism in order to set up a formal framework. In the
second part we come to relativistic scattering, where not only high velocities, but also
particle creation plays an important role.

2
Chapter 2

Wave packets

In quantum mechanics one describes a particle by a wave packet. Conceptually a wave

packet may be first studied in one dimension. The generalization to three dimensions is
straightforward and will be dealt with in a separate section.

2.1 One dimensional wave packet

Let us begin by studying an example: We indicate the position parameter in one dimension
by x and the time by t. Let us assume that at the instant t = 0 a wave packet is given
by the following expression (Fourier expansion):
1 ∞ Z
ψ(x, t = 0) = √ dk ϕ(k)eikx . (2.1)
2π −∞
So, once the Fourier transform ϕ is specified we have an explicit example. The interpre-
tation of the wave packet ψ, shown in formula (2.1), is the usual, i.e. the probability P
to find the particle, described by ψ, at the instant t = 0 at the position x is given by the
square of the modulus of ψ(x, t = 0) according to:

P(x, t = 0) = |ψ(x, t = 0)|2 . (2.2)

The example we will study here for ϕ, defined in formula (2.1), is a function of k which
peaks around a certain value k = k̄ and which vanishes rapidly for large values of |k|.
For the sake of calculational simplicity we choose a Fourier transform which vanishes
everywhere at the k-axis except for a small interval where it has a constant value different
from zero, as is shown in figure (2.1a). In formula, the example for ϕ(k) depicted in figure
(2.1a), is represented by

√1
  
 , k − k̄  ≤ ∆k
1

2∆k
 2  
ϕ(k) = √ θ k − k̄ − (∆k)2 = (2.3)
2∆k   
0 , k − k̄  > ∆k

  

The wave packet ψ which according to equation (2.1) follows for the above choice of
Fourier transform ϕ can readily be determined to be given by

1 sin (∆k x)
ψ(x, t = 0) = √ eik̄x , (2.4)
π∆k x

3
 ϕ(k) |ψ(x, t = 0)|2
2∆k- ∆k
√1 π
2∆k

k
k̄ − ∆k k̄ k̄ + ∆k − π 0 + π x
∆k ∆k
(a) (b)

Figure 2.1: (a) The Fourier transform in one dimension of a wave packet which differs
from zero only in a small region of width 2∆k around a central value k = k̄. (b) The
corresponding wave packet at the instant t = 0.

the probability distribution of which expression is depicted in figure (2.1b).

For completeness, let us go through the calculations which lead from formula (2.3), for
our choice of Fourier transform, to expression (2.4): First notice that ϕ, given in formula
(2.3), is normalized according to
Z ∞ 1 2
Z k̄ + ∆k
dk |ϕ(k)| = = 1 . (2.5)dk
−∞ k̄ − ∆k 2∆k
This has as a consequence that the wave packet ψ, given in equation (2.1), is automatically
normalized, which expresses the fact that the probability to find the particle somewhere
along the x-axis equals 1 and can be seen from

Z ∞ Z ∞
2
dx |ψ(x, t = 0)| = dx ψ ∗ (x, t = 0)ψ(x, t = 0) =
−∞ −∞
( )( )
∞ 1 ∞ 1 ∞ 0
Z Z Z
= dx √ dk ϕ∗ (k)e−ikx √ dk 0 ϕ(k 0 )eik x
−∞ 2π −∞ 2π −∞

∞ ∞ 1 ∞ 0
Z Z Z
= dk dk ϕ (k)ϕ(k ) 0 ∗ 0
dx ei(k − k)x
−∞ −∞ 2π −∞
Z ∞ Z ∞ Z ∞
0 ∗ 0 0
= dk dk ϕ (k)ϕ(k ) δ(k − k) = dk ϕ∗ (k)ϕ(k)
−∞ −∞ −∞
Z ∞
= dk |ϕ(k)|2 = 1 . (2.6)
−∞

4
Performing the integral of formula (2.1) for the Fourier transform (2.3) is straightforward
as we see:

Z k̄ + ∆k
1 Z∞ ikx 1
ψ(x, t = 0) = √ dk ϕ(k)e = √ dk eikx
2π −∞ 2 π∆k k̄ − ∆k
1 1
 
= √ ei(k̄ + ∆k)x − ei(k̄ − ∆k)x
2 π∆k ix
1 sin (∆k x)
= √ eik̄x , (2.7)
π∆k x

which is right the function given in formula (2.4).

5
2.1.1 Momentum
In general a particle has velocity. So, we might wonder how velocity is represented by a
wave packet. However, before studying the particle’s dislocation in time we first determine
its momentum. The time development of the wave packet we leave for a subsequent
section.
The expectation value of momentum is in quantum mechanics defined by
( )
∞ ∂
Z
∗
hki = dx ψ (x) −i ψ(x) . (2.8)
−∞ ∂x
For the example (2.4) we obtain for the above expression the result

hki = k̄ . (2.9)
The calculations can most easily be performed by following the same steps as in formula
(2.6), i.e.

( )
∞ ∂
Z
∗
hki = dx ψ (x, t = 0) −i ψ(x, t = 0)
−∞ ∂x
( )( )( )
∞ 1 ∞ ∂ 1 ∞ 0
Z Z Z
= dx √ dk ϕ (k)e−ikx
∗
−i √ dk ϕ(k )eik x
0 0
−∞ 2π −∞ ∂x 2π −∞

( )( )
∞ 1 ∞ 1 ∞ 0
Z Z Z
= dx √ dk ϕ (k)e−ikx
∗
√ dk ϕ(k ) k eik x
0 0 0
−∞ 2π −∞ 2π −∞

∞ ∞ 1 ∞ 0
Z Z Z
= dk dk 0 ϕ∗ (k)ϕ(k 0 ) k 0 dx ei(k − k)x
−∞ −∞ 2π −∞
Z ∞ Z ∞ Z ∞
= dk dk 0 ϕ∗ (k)ϕ(k 0 ) k 0 δ(k 0 − k) = dk |ϕ(k)|2 k
−∞ −∞ −∞

Z k̄ + ∆k k
= dk = k̄ . (2.10)
k̄ − ∆k 2∆k
We find thus that the most probable value to be measured for the particle’s momentum,
or the average value for a repeated number of measurements, equals k̄ which is indeed
the central value of the k-distribution. As a consequence of this result one interprets the
integration variable k in the Fourier expansion defined in formula (2.1) as the momentum
of the Fourier component in the expansion. This has then moreover as a consequence that
the time development of each Fourier component is different and thus in general that the
wave packet tends to spread.

6
2.1.2 The Heisenberg uncertainty relation
By describing the motion of a particle by a wave packet, one introduces some uncertainty
in the particle’s position as well as some uncertainty in the particle’s momentum. For
instance, in the previous example as well the momentum distribution (2.3) as the position
distribution (2.4) have some spreading. From figure (2.1a) we learn that the uncertainty
in the momentum of the particle equals ∆k, i.e.

k = k̄ ± ∆k . (2.11)
Moreover, from figure (2.1b) we may estimate that the width of the probability distri-
bution equals about π/∆k, which amounts for the uncertainty in the particle’s position
to
π
∆x ≈ . (2.12)
2∆k
Consequently, for the product of the two uncertainties we obtain

1 1
∆k ∆x ≈ 2
π > 2
, (2.13)
for which we recognize the Heisenberg relation in units h̄ = 1.

7
2.1.3 The time development of a wave packet
The general expression for the time development of a wave packet is as follows:
1 Z∞
ψ(x, t) = √ dk ϕ(k) exp {i (kx − ω(k)t)} , (2.14)
2π −∞
where ω(k) is some function of momentum k.
In order to study the above expression (2.14), we assume that ω is linear in k, at least
for values where the Fourier transform ϕ is maximal, i.e.

ω(k) = ω̄ + ω̄ 0 (k − k̄) . (2.15)

Such choice for ω leads for the wave packet (2.14) to the expression

1 Z∞ n  h i o
ψ(x, t) = √ dk ϕ(k) exp i kx − ω̄ + ω̄ 0 (k − k̄) t
2π −∞
1 n 
0
o Z ∞
= √ exp i −ω̄t + ω̄ k̄t dk ϕ(k) exp {ik (x − ω̄ 0 t)}
2π −∞
n  o
= exp it −ω̄ + ω̄ 0 k̄ ψ(x − ω̄ 0 t, 0) , (2.16)

which implies that apart from a phase factor and a translation along the x-axis, the wave
packet at the instant t has the same form as the wave packet at the instant t = 0. For
the probability distribution one finds consequently
2
|ψ(x, t)|2 = |ψ(x − ω̄ 0 t, 0)| . (2.17)
From the latter formula we read that the central peak in the probability distribution,
which according to figure (2.1b) at t = 0 is found at the origin of the x-axis, is located at
the position x = ω̄ 0 t at the instant of time t. Consequently, for ω linear in k as in formula
(2.15), the wave packet moves with a constant velocity ω̄ 0 and hence represents a freely
moving particle. The general idea is depicted in figure (2.2).
Notice, that for ω linear in k as in formula (2.15), the wave packet does not spread
as time develops. A feature which at first sight also seems quite reasonable for a point
particle in the absence of external forces. However, as we will see later on in subsection
(2.1.5), in general ω is not linear in k as in formula (2.15). Hence, also quadratic and
higher order terms must be considered which in general leads to spreading.

8
 |ψ|2 t=0 |ψ|2 t=1

0 x 0 x
peak at x = 0 peak at x = ω̄ 0

|ψ|2 t=2 |ψ|2 t=3

0 x 0 x
peak at x = 2ω̄ 0 peak at x = 3ω̄ 0

Figure 2.2: The probability distribution (2.17) in space (x) of wave packet (2.16) for the
momentum distribution (2.3) at four different instances, t = 0, 1, 2 and 3. The position
of the maximum probability, which represents the most probable place where the particle
can be found, moves at constant velocity ω̄ 0 .

9
2.1.4 Energy
The expectation value of energy is in quantum mechanics related to the wave function’s
time development, and hence defined by
( )
∞ ∂
Z
∗
hEi = dx ψ (x, t) i ψ(x, t) . (2.18)
−∞ ∂t
For the example (2.3) and moreover under the assumption that ω is linear in k as in
formula (2.15), we obtain for the above expression the result

hEi = ω̄ . (2.19)
The calculations can most easily be performed by following the same steps as in formula
(2.8), i.e.

( )
∞ ∂
Z
hEi = dx ψ ∗ (x, t) −i ψ(x, t)
−∞ ∂t
Z ∞ Z ∞ 0
= dk dk 0 ϕ∗ (k)ϕ(k 0 ) ω(k 0 ) ei (ω(k) − ω(k )) t δ(k 0 − k)
−∞ −∞

Z ∞ Z k̄ + ∆k ω̄ + ω̄ 0 (k − k̄)
= dk |ϕ(k)|2 ω(k) = dk = ω̄ . (2.20)
−∞ k̄ − ∆k 2∆k
We find thus that the most probable value to be measured for the particle’s energy, or
the average value for a repeated number of measurements, equals ω̄, which is the central
value for the momentum distribution (2.3), since from expression (2.15) one has

ω(k = k̄) = ω̄ .

10
2.1.5 Momentum, velocity and energy
In the previous subsections we obtained for the wave packet representation (2.14) of a
point particle, in the approximation that ω is linear in k as in formula (2.15), the following
three results: The expectation values for the particle’s momentum and energy are given in
formulas (2.8) and (2.18) by respectively k̄ and ω̄ = ω(k̄); whereas the particle’s velocity
is given by ω̄ 0 . The latter quantity can more generally be written in the form (Taylor
expansion coefficient):
!
0 dω(k)
ω̄ = . (2.21)
dk k = k̄
Now, in nonrelativistic mechanics we have the following relations between velocity v,
momentum p, kinetic energy K and the particle’s rest mass:

p2
p = mv and K = . (2.22)
2m
In comparison, we would expect for the wave packet something similar. But, then we are
dealing with distributions. Suppose, however, that the uncertainty in momentum is very
small. In that case the variable k is for all Fourier components almost equal to its average
or expectation value k̄. The first of the two relations (2.22) would then translate into

k̄
ω̄ 0 =, (2.23)
m
whereas for the second relation one would expect

k̄ 2
E ≈ E(k̄) = . (2.24)
2m
This suggests that we may identify the time development function ω(k) defined in
formula (2.14), with the energy variable E(k) for each Fourier component. Formula
(2.24) suggests then the choice

k2
E(k) = . (2.25)
2m
A Taylor series expansion around the central value k = k̄ of the latter expression for the
k-depence of the energy variable E(k), gives us

d2 E
! !
dE   1  2
E(k) = ω(k) = E(k̄) + k − k̄ + 2 k − k̄ + · · ·
dk k = k̄ 2 dk k = k̄

k̄ 2 k̄   1  2
= + k − k̄ + k − k̄ (2.26)
2m m 2m
The above expansion is complete, since the higher order derivatives vanish. We may
moreover observe that in case ∆k  k̄ one has that the third term in (2.26) is indeed
much smaller than the second term. Hence, the assumption (2.15) is correct for such
cases.
By comparing formula (2.15) to formula (2.26) we find from the first term indeed
relation (2.24) and from the second term relation (2.23). Consequently, we may conclude
that for the kinematics of a classical point particle the choice (2.25) is perfect.

11
2.1.6 Wave equation
Once the k-dependence of the energy variable ω(k) = E(k) has been settled, the wave
equation follows immediatly. When, for instance, we determine the first derivative in t
for the wave packet (2.14), we find
∂ 1 ∞ Z
i ψ(x, t) = √ dk ϕ(k) E(k) ei (kx − E(k)t) . (2.27)
∂t 2π −∞
Similarly, when we determine its second derivative in x, we obtain

∂2 1 ∞
Z
dk ϕ(k) −k 2 ei (kx − E(k)t) .
 
2 ψ(x, t) =
√ (2.28)
∂x 2π −∞

Consequently, from the k-dependence (2.25) for E(k), we find for the wave packet (2.14)
the wave equation

∂ 1 Z∞
i ψ(x, t) = √ dk ϕ(k) E(k) ei (kx − E(k)t)
∂t 2π −∞

1 ∞ k 2 i (kx − E(k)t)
Z
= √ dk ϕ(k) e
2π −∞ 2m

1 ∂2
= − ψ(x, t) , (2.29)
2m ∂x2
for which we recognize the Schrödinger equation in units h̄ = 1 for a system without
interactions.
Notice, at this stage, that such wave equation just depends on our choice for the k-
dependence for E(k), in this case given by (2.25). Might we, for example, have preferred
a k-dependence for E(k) of the form
√
E(k) = k 2 + m2 , (2.30)
then we would have obtained for the wave equation of (2.14) the result

∂2 1 ∞
Z
dk ϕ(k) −E 2 (k) ei (kx − E(k)t)
 
2 ψ(x, t) =
√
∂t 2π −∞

1 ∞
Z
dk ϕ(k) −k 2 − m2 ei (kx − E(k)t)
 
= √
2π −∞

∂2
!
= − m2 ψ(x, t) , (2.31)
∂x2

for which we recognize the Klein-Gordon equation in units h̄ = 1 for a system without
interactions.
Consequently, the wave packet description in itself does not say anything about the
dynamics of the system. It is just a consistent way of describing quantum mechanically
the motion of a particle.

12
2.1.7 Plane wave
In the limit of vanishing ∆k, one obtains a particle with a very well defined momentum,
i.e. k̄, but with a constant and thus vanishing probability distribution along the x-axis
as can be seen, for instance by using formula (2.4), i.e.
s
1 sin (∆k x) ∆k ik̄x
√ eik̄x −→ e . (2.32)
π∆k x π
∆k → 0
The image of a particle being everywhere with the same probability and with a well-
defined momentum, applies well to a beam of particles. One speaks then of a plane wave
which has the form:

eik̄x
ψ(x) = √ . (2.33)
2π
The fact that a plane wave is not normalizable, can then be interpreted as describing the
infinite number of particles in the beam.

13
2.2 Three dimensional wave packet
The generalization of the one dimensional wave packet ( 2.14) to three dimensions is
straightforward. We define the position coordinates

x1 , x2 and x3 , (2.34)
and similarly the three components of momentum

k1 , k2 and k3 . (2.35)
and generalize wave packet ( 2.14) to a wave packet which describes a particle which
moves in three dimensions by

1 3/2
  Z
d3 k ϕ ~k exp i ~k · (~x − ~x0 ) − E ~k t
  n h   io
ψ(~x, t) = , (2.36)
2π
where ~x0 represents the ”position” of the particle at t = 0.

2.2.1 Example
Let us study here the generalization to three dimensions of a Fourier transform ϕ which
only differs appreciably from zero in a small area of momentum space surrounding a
central value ~k̄ , i.e. a wave packet which represents a particle with ”momentum” ~k̄ . In
practice we will study the generalization of example ( 2.3), given by

3
1
Y  2  2 
ϕ(k) = q θ ki − k̄i − ∆ki (2.37)
i=1 2∆ki

1
  
 √ , ki − k̄i  ≤ ∆ki
 
i = 1, and i = 2, and i = 3
2 2∆k1 ∆k2 ∆k3


=   
0 , ki − k̄i  > ∆ki i = 1, or i = 2, or i = 3
 

It represents a function ϕ ~k which vanishes everywhere in momentum space, except for

 

the interior of a box with sides of length 2∆k , 2∆k and 2∆k centered at ~k̄ .
1 2 3
The wave packet ψ, which according to equation ( 2.36) follows at the instant t = 0 for
the above choice ( 2.37) of Fourier transform ϕ, can, by performing three times the same
integration as shown in formula ( 2.7), readily be determined to be given by
 h i
~ 3
1 sin ∆ki xi − (~x0 )i
ψ(~x, t = 0) = eik̄ · (~x − ~x0 )
Y
q , (2.38)
i=1 π∆ki xi − (~x0 )i

the probability distribution of which expression has a large maximum centered around ~x0
as in the three dimensional generalization of figure ( 2.1b).

14
2.2.2 Velocity and wave equation
Let us suppose that for a classical particle in three dimensions serves the generalization
of the k-dependence for the energy E in one dimension as given in formula ( 2.25), i.e.

~k 2
E ~k
 
= (2.39)
2m
A Taylor series expansion of this expression around the central value ~k̄ gives us

3
!
∂E
E ~k = E(~k̄ ) +
  X  
ki − k̄i + · · ·
i=1 ∂ki ~k = ~k̄

~k̄ 2 ~k̄  
= + · ~k − ~k̄ + · · · (2.40)
2m m
When we restrict ourselves to the first two terms of this expansion, which for small values
of ∆k1 , ∆k2 and ∆k3 leads to a very good approximation, then we obtain, following a
similar calculus as in formula ( 2.16), for the probability distribution in coordinate space
at the instant t the result

~k̄ 2
  

|ψ (~x, t)|2 =
 
x−
ψ ~ t, 0 . (2.41)

 m 

From the latter formula we read that the central peak in the probability distribution,
~k̄
which at t = 0 is found at the position ~x0 , is located at the position ~x = ~x0 + m t at the
~k̄
instant of time t. Consequently, the wave packet moves with a constant velocity m and
hence represents a freely moving particle in three dimensions.
The wave equation which follows for the k-dependence of the energy given in formula
( 2.39), can be determined by a procedure similar to the one discussed in section ( 2.1.6),
and yields

∂ ∇2 ∂2 ∂2 ∂2
i ψ (~r, t) = − ψ (~r, t) where ∇2 = + + , (2.42)
∂t 2m ∂x12 ∂x22 ∂x32
for which one recognizes the three dimensional Schrödinger equation for a system without
interactions.

15
16
Chapter 3

Scattering

Let us consider the case in which a single fixed scattering center is bombarded by particles
incident along the z-axis. In non-relativistic Schrödinger theory it is moreover assumed
that the effect of the scattering center on the particles can be represented by a potential
function V (~r ), which is appreciably different from zero only within a finite region of
dimensions a. By limiting ourselves to scattering from a potential we specialize to the
case of elastic scattering, which is scattering without energy loss or gain by the projectile.
However, many of the concepts developed here will be found useful in the discussion of
inelastic collision processes and more general reactions.
For simplicity we shall suppose that the particles in the beam are all well represented
by very broad and very long wave packets and that, before they reach the neighborhood
of the scatterer, these packets can be described approximately by plane waves

i(kz − ωt) k2
e where ω = , (3.1)
2m
although strictly speaking the waves do not extend to infinity either in width or in length.
In this chapter we will follow the procedure of how to obtain the differential scattering
cross section from a given potential V , as outlined in chapter 11 of ”Quantum Mechanics”
by Eugen Merzbacher [1].

3.1 Scattering of wave packets

In the previous sections we only discussed wave packets which represent freely moving
particles. One might therefore be curious of how systems with interaction can be described
within this formalism.
Actually, there are two possibilities, which in a sense are equivalent, varying the k-
dependence of E(k), or varying the wave equations. As is usual, here we will represent
interaction by a ”potential” in the wave equation. For example, the extension of wave
equation (2.42) to a wave equation for a system with interaction, is then given by

∇2
!
∂
i ψ (~r, t) = Hψ (~r, t) = − + V (~r ) ψ (~r, t) . (3.2)
∂t 2m
The potential approach has the advantage that one can construct the wave equation
in close analogy with a similar classical system, for which one knows how to formulate

17
a potential function. But, as a consequence, one has then to solve the resulting wave
equation.
Here, we are interested in the scattering of particles from a target. So, we have an inci-
dent beam with a well-defined momentum, described by a plane wave, and outgoing waves
which describe the scattered particles. We assume that this situation is well described by
a potential which has a short range, i.e.:

V (~r ) = 0 for r = |~r| > a , (3.3)

where a describes the dimension of the scatterer.
Let us assume that at t = 0 an incident particle of the beam is found at a distance r0
away from the target and moving along the z-axis towards the target, which finds itself
at the origin of the coordinate system. We assume moreover that r0 is large with respect
to the dimension of the target. In formula this implies for the position ~r0 of the projectile
particle and its momentum ~k0 at the instant t = 0 that

~r0 = −r0 ẑ with r0  a and ~k0 = k0 ẑ (3.4)

Since at the position ~r0 the particle does not feel the potential, because of the conditions
(3.3) and (3.4), we may describe its motion by a freely moving wave packet of the form

d3 k   ~
Z
ψ (~r, 0) = ϕ ~k eik · (~r − ~r0 ) , (3.5)
(2π)3/2
where ϕ is a smooth function of narrow width, ∆~k, centered around a mean momentum
~k0 . We assume r0 to be so large that ψ at t = 0 has negligible probability density at the
origin.
In section (3.6) we will show that the plane wave functions in the expansion (3.5) for the
wave packet at t = 0 can be replaced by particular eigenfunctions ψ (+) of the Hamiltonian
H, defined in formula (3.2), such that it obtains the form

~
Z
d3 k ϕ ~k e−ik · ~r0 ψ (+) ~k, ~r
   
ψ (~r, 0) = . (3.6)

The retarded wave functions ψ (+) satisfy the stationary wave equation for the Hamiltonian
H which has be defined in formula (3.2), i.e.

Hψ (+) ~k, ~r = Eψ (+) ~k, ~r

   
, (3.7)
and, as we will see in section (3.5), have the following asymptotic behavior

 ikr
 
1  i~k · ~r ~k, r̂ e 
ψ (+) ~k, ~r
  
−→ e + f for r large , (3.8)
(2π)3/2 r
which differs from a plane wave at large r only by an outgoing spherical wave. The
function f is defined in formula (3.39).
Once it has been proven that the retarded wave functions ψ (+) exist and are eigenfunc-
tions of the Hamiltonian H, defined in (3.2), then one has automatically solutions for the
wave equation (3.2) by putting

18
 k2
Z
d3 k ϕ ~k exp −i~k · ~r0 − iEt ψ (+) ~k, ~r
     
ψ (~r, t) = where E = . (3.9)
2m
Since the Hamiltonian (3.2) operates in coordinate space it can easily be verified that the
above expression (3.9) is a solution of the wave equation (3.2), as

∂ d3 k
Z
~k exp −i~k · ~r0 − iEt E ψ (+) ~k, ~r
     
i ψ (~r, t) = ϕ
∂t (2π)3/2

d3 k
Z
~k exp −i~k · ~r0 − iEt H ψ (+) ~k, ~r
     
= ϕ
(2π)3/2

= H ψ (~r, t) . (3.10)

19
3.2 Differential scattering cross section
At t = 0 the contribution of the outgoing spherical wave is negligible. This fact will be
studied in more detail in section (3.6). In formula (3.4) we defined the kinematic initial
conditions for a projectile particle. In particular, we supposed that the particle is at a
distance r0 far away from the target at t = 0. Assuming that also the scattering detectors
are at a macroscopic distance, of the order of r0 , from the scatterer, we may conclude
that the broad pulse will be traveling through the position of the detectors after a time
r0 2mr0
T = 2 = , where as in eq’n (2.22) k0 = mv0 . (3.11)
v0 k0
When we examine the pulse at the position of the detectors, then ψ (+) can be represented
by its asymptotic expansion (3.8). But, since the phases have changed with time, the
outgoing spherical wave may no longer be neglegted. However, we can make an approxi-
mation based on the identity

k2 1 h~ 1 ~2
i2  2 
E ~k k0 + ~k − ~k0 k0 + 2~k0 · ~k − ~k0 + ~k − ~k0
     
= = =
2m 2m 2m
2 2
~k − ~k0 ~k − ~k0
 
~k0 ~k 2
~
= k· − 0
+ = ~k · ~v0 − E0 + , (3.12)
m 2m 2m 2m
where E0 = ~k02 /2m .
In order to be able to neglect the last term in expression (3.12), when E is substituted
into (3.9), we require that the arrival time (3.11) at the scatterer, T , although large,
should still satisfy the inequality
2
~k − ~k0


T 1 . (3.13)
2m
Now, since the relevant part of the ~k-integration in (3.9) comes from an interval of width
∆k around a mean value ~k0 (see formula (3.5)), we may, also substituting (3.11) for T ,
replace condition (3.13) by

(∆k)2 r0
1 . (3.14)
k0
This condition implies that the wave packet does not spread appreciable when it is dis-
placed by the macroscopic distance r0 .
Substituting the first two terms of (3.12) for E and the asymptotic expansion (3.8) for
(+)
ψ into the expression (3.9), we obtain the approximation

eikr
 
3
dk ~
Z
ϕ ~k exp −i~k · (~r0 + ~v0 t) + iE0 t eik · ~r + f ~k, r̂
  h i  
ψ (~r, t) ≈  .
(2π)3/2 r
(3.15)
~
Assuming, moreover, that f , unlike ϕ, is a slowly varying function of k, for (3.15) we can
write

20
 d3 k
Z
ψ (~r, t) ≈ eiE0 t ~k exp i~k · (~r − ~r0 − ~v0 t) +
  h i
ϕ
(2π)3/2

f ~k0 , r̂
 
d3 k
Z
+ eiE0 t ~k exp i kr − ~k · (~r0 + ~v0 t)
  n h io
ϕ (3.16)
.
r (2π)3/2

Since ϕ is appreciably different from zero only in an area of momentum space where
k̂ ≈ k̂0 , we can substitute at the domain of integration for (3.16) effectively

kr ≈ ~k · k̂0 r ,

and consequently, obtain the further approximation

d3 k
Z
ψ (~r, t) ≈ eiE0 t ~k exp i~k · [(~r − ~r0 ) − ~v0 t] +
  n o
ϕ
(2π)3/2

f ~k0 , r̂
 
d3 k
Z
+eiE0 t ~k exp i~k · r~k0 − ~r0 − ~v0 t (.3.17)
  n h  io
ϕ
r (2π)3/2

By comparing the above expression (3.17) to formula (3.5), we end up for (3.9) at the
position of the detectors with the approximation

f ~k0 , r̂
 

ψ (~r, t) ≈ ψ (~r − ~v0 t, 0) eiE0 t + ψ r k̂0 − ~v0 t, 0 eiE0 t .

 
(3.18)
r
Except for the phase factor exp (iE0 t), the first term on the righthand side of (3.18)
represents the initial wave packet displaced without change of shape, as if no scattering
had occurred. This situation describes those projectile particles in the beam which had
no interaction with the target constituents, and which consequently have no interest to
the scattering experiment.
The second term on the righthand side of (3.18) is a scattered spherical wave packet,
which describes the projectile particles which after interaction with the target move in
all possible radial directions. As a function of time t the scattered spherical wave packet
has the same form as the initial wave packet at time t = 0. This is indicated by the
time-dependent function ψ. The amplitude, however, is reduced by a factor r −1 , which
implies a reduction with a factor r −2 of the corresponding probability distribution as it
should be for a spherically expanding density. The amplitude, moreover, is modulated by
the angular amplitude f ~k0 , r̂ which depends exclusively on the direction of scattering,


not on the distance of the detector. Sensibly, the latter is called the scattering amplitude.
A detector outside the direction of the beam intercepts thus a radially expanding replica
of the initial wave packet, equal to the one seen by the scatterer at the origin, but reduced
~

in amplitude by a factor f k0 , r̂ /r. Now, let us assume that the detector has an area
given by

r 2 dΩ

21
perpendicular to the radial direction, and moreover that the probability distribution of
the particle under consideration flows through that area with an average velocity given by
v0 . Then, in an infinitesimally small interval of time between t and t + dt, all probability
contained in a volume given by

r 2 dΩ × v0 dt

flows through the detector. Consequently, the probability of observing the scattered
projectile particle at the detector in the time interval between t and t + dt follows from

 2
f ~
 
k0 , r̂    2 2
  2
− ~v0 t, 0  r 2 dΩ dt = v0 f ~k0 , r̂  dΩ ψ [r − v0 t] k̂0 , 0  dt.
  
v0   ψ r k̂0

 r 

(3.19)
Hence, the total probability for detecting it at the detector is

  2 Z +∞   2   2 Z +∞   2
 ~  ~
v0 f k0 , r̂  dΩ dt ψ [r − v0 t] k̂0 , 0  = f k0 , r̂  dΩ dξ ψ ξ k̂0 , 0  ,
     
−∞ −∞
(3.20)
where the limits of integration may be taken to be −∞ and +∞ with impunity, since ψ
describes a wave packet of finite length.
On the other hand, the probability that the incident particle will pass through a unit
area, located perpendicular to the beam in front of the scatterer, is
Z +∞   2
dξ ψ ξ k̂0 , 0  . (3.21)
 
−∞

Also here is no harm in extending the integration −∞ to +∞, since at t = 0 the wave
packet is entirely in front of the scatterer.
If the ensemble contains N particles, all represented by the same general type of wave
packet, then, using expression (3.20), we find that the number IdΩ of particles scattered
into the solid angle dΩ is given by
  2 N Z +∞  2
 ~ X 
I dΩ = f k0 , r̂  dΩ dξ ψi ξ k̂0 , 0  , (3.22)
  
i=1 −∞

whereas, using formula (3.21), one has that

N Z
X +∞   2
I0 = dξ ψi ξ k̂0 , 0  . (3.23)
 
i=1 −∞

gives the number of incident particles per unit area. Hence, by definition (1.1) we arrive
at the fundamental result
dσ   2
= f ~k0 , r̂  . (3.24)
 
dΩ

22
3.3 The retarded Green’s functions
Before we study the solutions ψ (+) of (3.7), we first study Green’s functions. The Green’s
function, G+ (~r, ~r 0 ), which suits us here is a solution of the following differential equation
1  2 
∇ + k 2 G (~r, ~r 0 ) = δ (3) (~r − ~r 0 ) , (3.25)
2m
which is, amongst other solutions, solved by
0
0m eik |~r − ~r |
G+ (~r, ~r ) = − . (3.26)
2π |~r − ~r 0 |
More correct would be to state that since equation (3.25) is a second order differential
equation, it has two independent solutions. Any linear combination of those two solutions
is a solution of equation (3.25). Below, we will see that the choice (3.26) gives us the
correct boundary conditions.
In order to verify that the above Green’s function (3.26) solves equation (3.25), we take
the case ~r 0 = 0 without loss of generality. First we determine the second derivative with
respect to x = x1 of expression (3.26) for r 6= 0, which gives

∂ 2 eikr k 2 3ik
( ) ( )
ikr ik 1 2 ikr 3
= e − +x e − 3 − 4 + 5 .
∂x2 r r2 r3 r r r

For y = x2 and z = x3 one obtains similar expressions. Their sum gives

ikr ( ) (
k 2 3ik
)
2e ik 1 3
∇ = 3eikr 2 − 3 + r eikr
2
− 3 − 4 + 5
r r r r r r

eikr
= −k 2 for r 6= 0 .
r
For r = 0 we integrate equation (3.25) over the interior of a sphere with radius R sur-
rounding the origin of coordinate space. For the righthand side we obtain trivially
Z n o
d3 r δ (3) (~r ) = 1 . (3.27)
sphere
The lefthand side of (3.25) takes some more work. Let us start with the second term, i.e.

ikr eikr
2e
Z Z Z R
3
d rk = dΩ r 2 dr k 2
sphere r 0 r
 
= 4π eikr (1 − ikR) − 1 . (3.28)

The r-integration can here most easily be obtained by applying the following identity
Z
∂ Z
rdr eikr = −i dr eikr .
∂k

23
Next we integrate the first term of the lefthand side of (3.25) by using the theorem of
divergence, leading to

ikr ikr
2e ~e
Z Z
3
d r∇ = ~ ·∇
dS
sphere r surface r

= 4π eikr (ikR − 1) . (3.29)

In the last step of formula (3.29) we used

ikr d eikr
~ = R2 dΩ r̂
dS and ~e
∇ = r̂ .
r dr r
When we put together the results (3.28) and (3.29), then we obtain for the integration on
the interior of a sphere with radius R of the lefthand side of formula (3.25) the result

1 eikr 
 
Z  
d 3 r ∇2 + k 2 − = 1 . (3.30)
sphere 4π r
Comparing the results (3.27) and (3.30) with equation (3.25) shows perfect agreement.
We may thus conclude that function (3.26) is a solution of the differential equation
(3.25). It is referred to in the literature as the retarded Green’s function.

24
3.4 Lippmann Schwinger equation for the wave func-
tion
The time-independent Schrödinger equation, which is related to the wave equation (3.2),
is given by
h i
−∇2 + 2mV (~r ) ψ (~r ) = k 2 ψ (~r ) . (3.31)
A formal solution to this equation is given by the following expression

~
eik · ~r
Z
~k, ~r d3 r 0 G+ (~r, ~r 0 ) V (~r 0 ) ψ (+) ~k, ~r 0
   
(+)
ψ = + , (3.32)
(2π)3/2
as, using formula (3.25), can most easily be seen by performing

1  2 1  2  ei~
k · ~r
∇ + k 2 ψ (+) ~k, ~r =
  
∇ + k2 +
2m 2m (2π)3/2
1  2
Z
∇ + k 2 G+ (~r, ~r 0 ) V (~r 0 ) ψ (+) ~k, ~r 0
  
+ d3 r 0
2m
Z
d3 r 0 δ (3) (~r − ~r 0 ) V (~r 0 ) ψ (+) ~k, ~r 0
 
= 0 +

= V (~r ) ψ (+) ~k, ~r

 
,

which leads exactly to equation (3.31).

Equation (3.32) is in fact an integral equation which substitutes the differential equation
(3.31) incorporating the correct boundary conditions (as we will see in section (3.6)). In
the literature it is referred to as the Lippmann Schwinger equation.
In order to see that the Lippmann Schwinger equation solves in principle equation
(3.31), we first simplify the expression (3.32) by defining

~
eik · ~r
φ ~k, ~r
 
= , (3.33)
(2π)3/2
which leads for (3.32) to
Z
ψ (+) ~k, ~r = φ ~k, ~r + d3 r 0 G+ (~r, ~r 0 ) V (~r 0 ) ψ (+) ~k, ~r 0
     
. (3.34)

In the latter expression we substitute ψ (+) in the righthand side, to find

Z
ψ (+) ~k, ~r = φ ~k, ~r +
   
d3 r 0 G+ (~r, ~r 0 ) V (~r 0 ) ×
  Z 
× φ ~k, ~r 0 + ~k, ~r 0 0
 
3 00 0 00 00 (+)
d r G+ (~r , ~r ) V (~r ) ψ .

25
Continuing this procedure leads to

Z
ψ (+) ~k, ~r = φ ~k, ~r + d3 r1 G+ (~r, ~r1 ) V (~r1 ) φ ~k, ~r1 +
     

Z Z
d3 r2 G+ (~r, ~r1 ) V (~r1 ) G+ (~r1 , ~r2 ) V (~r2 ) φ ~k, ~r2 +
 
+ d 3 r1
Z Z Z
d3 r3 G+ (~r, ~r1 ) V (~r1 ) G+ (~r1 , ~r2 ) V (~r2 ) G+ (~r2 , ~r3 ) V (~r3 ) φ ~k, ~r3 +
 
3 3
+ d r1 d r2

+··· . (3.35)

For a weak potential V this series converges and leads to a unique solution for ψ (+) .

26
3.5 Asymptotic behaviour
When we substitute the expression (3.26) for the retarded Green’s function into the Lip-
mann Schwinger equation (3.32), then we obtain

~ 0
(+) ~k, ~r
  eik · ~r m Z 3 0 eik |~r − ~r | 0 (+) ~

0

ψ = − d r V (~
r ) ψ k, ~
r . (3.36)
(2π)3/2 2π |~r − ~r 0 |
According to formula (3.3) the potential V vanishes at large distances, hence the domain
of the ~r 0 integral is located just around the origin within a sphere of radius a, i.e.

r0 < a . (3.37)
Consequently, at distances r large compared to the target size a, such that

ka2
 1 ,
r
which implies only very small variations in the phases due to the quadratic and higher
terms of the exponent, we may to first order in r 0 approximate

s
(2~r − ~r 0 ) · ~r 0
q q
|~r − ~r 0 | = (~r − ~r 0 )2 = r 2 + r 0 2 − 2~r.~r 0 = r 1 −
r2
s
r̂ · ~r 0 r̂ · ~r 0
( )
≈ r 1−2 ≈ r 1− ≈ r − r̂ · ~r 0 . (3.38)
r r

If, further, ~r 0 in the denominator of the integrand is neglected, which can be done since
|~r − ~r 0 | is smooth for r  a, we obtain for large r

~ r eikr ~ 
 
1 e ik · ~
ψ (+) ~k, ~r
 
≈ + f k, r̂ (r large)
(2π)3/2 r

where

√ Z 0
f ~k, r̂ = − 2π m d3 r 0 e−ikr̂ · ~r V (~r 0 ) ψ (+) ~k, ~r 0
 
and k = ~k  .
   
(3.39)
 

27
3.6 Initial conditions
Finally, it is left to be studied why the expansion (3.6) is correct. Using the explicit
expression for ψ (+) , given in formula (3.36), we obtain for (3.6)

Z
~
d3 k ϕ ~k e−ik · ~r0 ×
 
ψ (~r, 0) =

~
 
0
eik · ~r m eik |~r − ~r |
Z
V (~r 0 ) ψ (+) ~k, ~r 0 
 
× − d3 r 0
 
3/2 0
(2π) 2π |~r − ~r |

d3 k ~k e−i~k · ~r0 ei~k · ~r +

Z  
= 3/2
ϕ
(2π)

d3 k ~  −i~k · ~r0 m ik |~r − ~r 0 |

3 0 e
Z Z
(+) ~
 
0 0
− ϕ k e dr V (~
r ) ψ k, ~
r . (3.40)
4π 2π |~r − ~r 0 |

At t = 0, as afirmed in relation with formula (3.6), the second term in the righthand side
of (3.40) vanishes, since the first term at the righthand side already equals the expression
for the wave packet at t = 0 as given in formula (3.5). Now, since U = 0 for r 0 > a, it is
sufficient to show that

d3 k
Z
~k exp −i~k · ~r0 + ik |~r − ~r 0 | ψ (+) ~k, ~r 0 = 0
     
ϕ (3.41)
(2π)3/2
for r 0 < a. It may usually be assumed that in this integral the variation of ψ (+) with ~k
can be neglected. Moreover, ϕ is appreciably different from zero only for vectors ~k near
the direction of ~k0 . Hence, we may approximate

k |~r − ~r 0 | ≈ k̂0 · ~k |~r − ~r 0 |

and consequently, by comparing the result to formula (3.5), we find that the relevant part
of the lefthand side of relation (3.41) is nearly equal to

d3 k
Z
~k exp i~k · k̂0 |~r − ~r 0 | − ~r0
  n  o  
0
ϕ = ψ k̂ 0 |~
r − ~
r | , 0 .
(2π)3/2

The righthand side of this expression vanishes, because k̂0 |~r − ~r 0 | points to a position
behind the scatterer where at t = 0 the wave packet was assumed to vanish.

28
3.7 Spherical well
In this section we consider the example of S-wave (` = 0) scattering from a spherical well,
given by the following second order differential equation (in units h̄ = c = 1 and with the
choice 2µ = 1 for the reduced mass of the scattered particles)


2
! V0 , 0 < r < a
d


− 2 + V (r) − E u(r) = 0 where V (r) = . (3.42)
dr
0 , r>a



The general solutions of equation (3.42) are elementary:

For 0 < r < a one has linear combinations of

cos(κr) and sin(κr) , where κ2 = E − V0 . (3.43)

Whereas, for r > a one has linear combinations of

cos(kr) and sin(kr) , where k 2 = E . (3.44)

Taking moreover into account the boundary condition, u = 0, at the origin (r = 0), then
we obtain for a solution of equation (3.42), the form


 A sin(κr) , 0<r<a
u(r) = . (3.45)
B sin(kr + δ(E)) , r>a



The normalization constants A and B in formula (3.45) are determined by the boundary
conditions at r = a, i.e.
 
du  du 
= and u (r ↓ a) = u (r ↑ a) , (3.46)
dr r↓a dr r↑a
 

which leads to equations

Bk cos(ka + δ(E)) = Aκ cos(κa) and B sin(ka + δ(E)) = A sin(κa) . (3.47)

By elimination of the normalization constants A and B in formula (3.47), we obtain

cotg(ka)cotg(δ(E)) − 1
k = κcotg(κa) , (3.48)
cotg(δ(E)) + cotg(ka)
from which we can solve

k + κcotg(κa)cotg(ka)
cotg(δ(E)) = . (3.49)
kcotg(ka) − κcotg(κa)
Notice, that the righthand-side expression of formula (3.49) is real for real energy, E,
even for cases where κ2 < 0. In the latter cases one has
eiκa + e−iκa
κcotg(κa) = iκcoth(iκa) = iκ iκa ,
e − e−iκa

29
 δ δ

40 40

30 30

20 20

10 10

100 200 300 400 E 2 4 6 8 E

Figure 3.1: Phase shifts for a spherical well potential. For the depth, V0 , and the width,
a, of the well we have chosen here -8 and 0.5 respectively. The lefthand side picture shows
the phase shifts for energies up to 500. Whereas, the righthand side picture exhibits the
details for the phase shifts at low energies.

which is manifestly real for purely imaginairy κ.

A typical example for the phase shifts of a spherical well is depicted in figure (3.1).
From relations (3.47), also using formula (3.49), one may moreover determine the ratio
of the normalization constants A and B, i.e.
!− 1
A sin(ka + δ(E)) κ2 2
= = sin (κa) + 2 cos2 (κa)
2
. (3.50)
B sin(κa) k
The wave function (3.45) can then be obtained upto an overall normalization constant.
An example has been depicted in figure (3.2). The lefthand side picture shows the wave
function. Whereas the righthand side picture shows the continuation of each of the two
different functions (3.45), one for 0 < r < a and one for r > a, into the r-interval where
they do not coincide with the wave function. From the latter picture one observes more
easily that at the boundary of the spherical well, a = 0.5 here, both the wave function
and its derivative are continuous functions in r.
Notice that the function B sin(kr + δ(E)) which describes the part of the wave function
for r > a does not vanish at the origin, but instead results to B sin(δ(E)) for r = 0.
When E < V0 , then one may define the wave function in a slightly different way. Instead
of expressions (3.45) we prefer for imaginairy κ the following for the wave function

A0 (eiκr − e−iκr )



 2 , 0<r<a
u(r) = . (3.51)

B sin(kr + δ(E)) , r>a



This has no effect for the phase shifts, whereas for the ratio of A0 and B we obtain then

A0 sin(ka + δ(E))
= 1 iκr . (3.52)
B 2
(e − e−iκr )
An example has been depicted in figure (3.3).

30
 u u

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.5 1.0 1.5 r 0.5 1.0 1.5 r

Figure 3.2: Wave function for the spherical well for E = 0.1. The intensity, V0 , equals
−8.0, the width of the well, a, equals 0.5. The wave function is not normalized. We took
A = 1.

u u

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.5 1.0 1.5 r 0.5 1.0 1.5 r

Figure 3.3: Wave function for the spherical well for E = 2.0. The intensity, V0 , equals
8.0, the width of the well, a, equals 0.5. The wave function is not normalized. We took
A0 = 1.

3.7.1 The effective range expansion

We inspect the expression (3.49) for very low energies.
From formulae (3.43) and (3.44), we obtain

κ2 → κ20 = −V0 and k 2 → 0 , (3.53)

when E tends to zero. Hence, for the cotangents of ka and κa, we obtain for very low
energies the following approximations
1
cotg(ka) → and cotg(κa) → cotg(κ0 a) , (3.54)
ka
For expression (3.49) we find then

1 (ka)2 + aκ0 cotg(κ0 a)

cotg(δ(E)) → , (3.55)
ka 1 − aκ0 cotg(κ0 a)

31
or equivalently

1 aκ0 cotg(κ0 a) a
kcotg(δ(E)) → + k2 , (3.56)
a 1 − aκ0 cotg(κ0 a) 1 − aκ0 cotg(κ0 a)

The result (3.56) can be generalized to include a variety of interaction potentials and
for different partial waves (`):
1 1
k 2`+1 cotg(δ(E)) = − + R` k 2 + ... , (3.57)
a` 2

an expansion entirely in k 2 . The parameters a` and R` are in the literature refered to by

partial wave scattering length and effective range, respectively. Here, we will study some
properties of a0 , the S-wave scattering length for the spherical well potential.
By comparing formula (3.56) with formula (3.57), we find

aκ0 cotg(κ0 a) − 1
a0 = a . (3.58)
aκ0 cotg(κ0 a)

On analyzing the following expression

xcotg(x) − 1
F (x) = ,
xcotg(x)

we find that for reasonable values of a and V0 the partial S-wave scattering length a0 is
positive for positive V0 and negative for negative V0 .
To lowest order in k 2 one obtains for the exterior part of the wave function (3.45), the
expression
B sin(δ(E))
u(r > a) = B sin(kr + δ(E)) → (a0 − r) , (3.59)
a0
which vanishes for r = a0 . A positive zero of the wave function, hence positive a0 , can be
observed for the case of a positive potential well in Figure (3.3), where the exterior wave
function is continued into the interior (r < a)). For a negative potential well we observe
from Figure (3.2) that a zero of the exterior wave function takes place at negative values
of r, hence negative a0 .

3.7.2 One-delta-shell approximation

The system (3.42) may be approximated by a one-delta-shell potential, according to

d2
!
− 2 + aV0 δ(r − b) − E u(r) = 0 with 0 < b < a . (3.60)
dr
In the absence of any potential in both regions, 0 < r < b and r > b, the general solutions
of differential equation (3.60) are there linear combinations of

cos(kr) and sin(kr) , where k 2 = E . (3.61)

Hence, taking into account the boundary condition at the origin, we find the following
solution for (3.60)

32
 

 A sin(kr) , 0<r<b
u(r) = . (3.62)
B sin(kr + δ(E)) , r>b



The boundary conditions at r = b are different from those given in formula (3.46). By
twice integrating the differential equation (3.60) in a small domain, (b − , b + ), around
the delta-shell position r = b and letting  → 0, one finds
 
du  du 
− + + aV0 u(b) = 0 and u (r ↓ b) = u (r ↑ b) , (3.63)
dr r↓b dr r↑b
 

which leads to equations

−Bk cos(kb + δ(E)) + Ak cos(kb) + aV0 A sin(kb) = 0

and B sin(kb + δ(E)) = A sin(kb) . (3.64)

Elimination of the normalization constants, A and B, in formula (3.64), leads to an

expression for the cotangent of the phase shifts, given by

cos(kb) k
cotg(δ(E)) = − − . (3.65)
sin(kb) aV0 sin2 (kb)
A typical example for the phase shifts of a spherical delta shell is depicted in figure
(3.4).

δ δ

80 80

60 60

40 40

20 20

100 200 300 400 E 2 4 6 8 E

Figure 3.4: Phase shifts for the one-delta-shell potential as defined in formula (3.60). The
intensity of the potential is given by aV0 . We have chosen here the same values for a and
V0 , i.e. 0.5 and -8 respectively, as for the results depicted in figure (3.1). For the delta
shell radius we have chosen here b = 12 a.

From relations (3.64) one may also here determine the ratio of the normalization con-
stants A and B, i.e.

A sin(kb + δ(E))
= . (3.66)
B sin(kb)

33
The wave function (3.62) can then be calculated upto an overall normalization constant.
An example has been depicted in figure (3.5). The lefthand side picture shows the wave
function for E = 2.0, the righthand side picture for E = 10.
Notice that now the wave function has a sharp discontinuity in its derivative with
respect to r at the position of the delta shell. This corresponds to the first of the boundary
conditions given in formula (3.63), which says that the difference in the derivatives from
above r ↓ b and from below r ↑ b is proportional to the value of the wave function at
r = b.

0.5
0.2

0.5 1.0 1.5 r

0.0
0.0
0.5 1.0 1.5 r

-0.2 -0.5
E = 2.0 E = 10.0
u

Figure 3.5: Wave functions for the one-delta-shell potential for E = 2.0 and E = 10.0.
The intensity is given by aV0 /. The dashed line is the exact solution for the spherical well
(V0 = −8.0 and a = 0.5). The wave function is not normalized. We took A = 1.

3.7.3 A formal solution of the one-delta-shell potential

The solution (3.62) can formally be written in the form

cos(kr)  C 
   
u(r) sin(kr) − k
  = 


  , (3.67)
0
u (r) k cos(kr) sin(kr) D
where u0 (r) stands for du/dr and where



 C = A and D = 0 for 0<r<b
. (3.68)
C = B cos(δ(E)) and D = −kB sin(δ(E)) for r>b



Let us define the propagator, G (r2 , r1 ), which connects the values of the wave function
and of its derivatives at two different positions r1 < r2 , as follows
   
u (r2 ) u (r1 )
  = G (r2 , r1 )   . (3.69)
0 0
u (r2 ) u (r1 )
This propagator has the following properties

34
 G (r1 , r1 ) = 1 , G (r3 , r1 ) = G (r3 , r2 ) G (r2 , r1 ) and G −1 (r2 , r1 ) = G (r1 , r2 ) . (3.70)
When there is no potential, i.e. V (r) = 0, in the interval from r1 to r2 , then it is easy
to verify that a suitable choice for the propagator is given by

cos (kr2 ) cos (kr1 )

   
 sin (kr2 ) − k   sin (kr1 ) k
G (r2 , r1 ) =   . (3.71)

 
k cos (kr2 ) sin (kr2 ) −k cos (kr1 ) sin (kr1 )
Now, the boundary conditions (3.63) at r = b can be written as
    
u (r ↓ b) 1 0 u (r ↑ b)
  =    . (3.72)
u0 (r ↓ b) aV0 1 u0 (r ↑ b)
Hence, when we want to pass from one side of the delta shell to the other, i.e. from r1 < b
to r2 > b, then we may proceed as follows

   
u (r2 ) u (r ↓ b)
  = G (r2 , r ↓ b)  
u0 (r2 ) u0 (r ↓ b)
  
1 0 u (r ↑ b)
= G (r2 , r ↓ b)   
0
aV0 1 u (r ↑ b)
   
1 0 u (r1 )
= G (r2 , r ↓ b)   G (r ↑ b, r1 )   , (3.73)
aV0 1 u0 (r1 )
which upon the use of formulas (3.71) and (3.72), takes the form

cos (kr2 )
   
u (r2 ) sin (kr2 ) −
  = 
 k ×

0
u (r2 ) k cos (kr2 ) sin (kr2 )

1 + aV − aV20 cos2 (kb)

 
0
k sin(kb) cos(kb) k
× ×
 

aV0 sin2 (kb) aV 0

1 − k sin(kb) cos(kb)

cos (kr1 )  u (r1 ) 

 
sin (kr1 ) k
×  , (3.74)
 

0
−k cos (kr1 ) sin (kr1 ) u (r 1 )

Formula (3.74) can easily be generalized for the case of more delta-shells. It just leads to
repeated matrix multiplications as will be studied in section (3.7.4).
In the case of one delta shell at r = b one has expressions (3.68) for the coefficients of
the wave function in the two domains, 0 < r1 < b and r2 > b. This reduces formula (3.74)
to

35
 cos (kr2 )  B cos(δ(E)) 
 
 sin (kr2 ) − k   =
 
k cos (kr2 ) sin (kr2 ) −kB sin(δ(E))

cos (kr2 )
 
 sin (kr2 ) − k
=  ×


k cos (kr2 ) sin (kr2 )

1 + aV0 sin(kb) cos(kb) − aV20 cos2 (kb)

 
k k
× ×

aV

2 0
aV0 sin (kb) 1 − k sin(kb) cos(kb)

cos (kr1 ) cos (kr1 )  A 

  
sin (kr1 ) k sin (kr 1 ) − k
×  , (3.75)
  
 
−k cos (kr1 ) sin (kr1 ) k cos (kr1 ) sin (kr1 ) 0

or equivalently, to

1 + aV0 sin(kb) cos(kb) − aV20 cos2 (kb)

     
B cos(δ(E)) k A
  =  k    .
aV
 
−kB sin(δ(E)) 2
aV0 sin (kb) 1 − 0 sin(kb) cos(kb) 0
k
(3.76)
For the phase shift we obtain then

cos(kb) k
cotg(δ(E)) = − − , (3.77)
sin(kb) aV0 sin2 (kb)
which may be compared to formula (3.65).

3.7.4 Multi-delta-shell approximation

Let us consider an approximation of the potential (3.42) by more than one delta shell, i.e.
N
X
V (r) = V0 ∆n δ (r − bn ) , (3.78)
n=1

where 0 < b1 < b2 < . . . < bN < a and where ∆1 , ∆2 , . . ., ∆N represent the measures
which belong to the distribution {bn | n = 1, . . . , N }. Let us, moreover, define the following
matrices

1 + ∆V0 sin(kb) cos(kb) − ∆V2 0 cos2 (kb)

 
k k
X(b, ∆) =   , (3.79)
 
2
∆V0 sin (kb) 1− ∆V 0 sin(kb) cos(kb)
k
and

X = X (bN , ∆N ) · . . . · X (b2 , ∆2 ) · X (b1 , ∆1 ) . (3.80)

36
Then, in passing over all delta-shell positions from r1 < b1 to r2 > bN , we obtain for the
relations between the normalization constants, A and B, and the cosine and sine of the
phase shift the expression
    
B cos(δ(E)) X11 X12 A
  =    , (3.81)
−kB sin(δ(E)) X21 X22 0
which results for the cotangent of the phase shift in
X11
cotg(δ(E)) = −k . (3.82)
X21
Typical examples for the phase shifts of many-delta-shell approximations of the spher-
ical well are depicted in figure (3.6), respectively for two and for five equally spaced delta
shells. The exact result (3.49) is indicated in the figures by a dashed line. For the spacings
we used 12 d, 1 12 d, 2 21 d, . . . , (N − 12 )d, where d = a/N and N the number of shells. For all
measures we took d.

δ δ
50
40
40

30 30

20 20

10 10

100 200 300 400 E 100 200 300 400 E

Figure 3.6: Phase shifts for 2 delta shells (lefthand side picture) and for 5 delta shells
(righthand side picture). The shells are distributed equally spaced over the width, a, of
the well (see text). The intensity of each shell is given by aV0 /2 and aV0 /5 respectively.
For the depth, V0 , and the width, a, of the well we have chosen here the same values, i.e.
-8 and 0.5 respectively, as for the results depicted in figure (3.1). The dashed line is the
exact solution for the spherical well.

Notice from figure (3.6) that the approximation is already rather good for five delta
shells.
The wave functions can be calculated as follows. First, we define the pairs of coefficients
Ai and Bi which determine the wave function in the i-th interval, according to
!
cos(kr)
u(r) = Ai sin(kr) + Bi − for r ∈ [bi , bi+1 ] , (3.83)
k
In the zeroeth interval one has b0 = 0 and moreover, from relation (3.68),

A0 = A and B0 = 0 . (3.84)
Furthermore, we consider bN +1 → ∞.

37
 For the other intervals, by the definition given in formula (3.79), one has the relations
   
Ai Ai−1
  = X(bi , ∆i )   . (3.85)
Bi Bi−1
With the help of relation (3.85) one can calculate all coefficients in all intervals, hence
determine the wave function. An example has been depicted in figure (3.7).

u u
0.4 0.4

0.2 0.2

0.0 0.0
0.5 1.0 1.5 r 0.5 1.0 1.5 r

-0.2 -0.2

Figure 3.7: Wave function for 2 and for 5 delta shells at E = 2.. The shells are distributed
equally spaced over the well width a. The coupling of each shell is given by aV0 /2. The
dashed line is the exact solution for the spherical well (V0 = −8.0 and a = 0.5).

3.7.5 Detailed study of the delta-shell approximation

Let us assume that we dispose over functions φ(r) and G(r, r 0 ) which solve the following
equations

d2 d2
" # " #
− 2 − E φ(r) = 0 , and − 2 − E G(r, r 0 ) = −δ(r − r 0 ) . (3.86)
dr dr
Then we obtain a formal solution of the wave equation

d2
!
− 2 + V (r) − E u(r) = 0 , (3.87)
dr
by the expression
Z ∞
u(r) = φ(r) + dr 0 G(r, r 0 )V (r 0 )u(r 0 ) . (3.88)
0
This can most conveniently be seen, using formulas (3.86) and (3.88), by performing

d2 d2
! ! Z ∞ 
0 0 0 0
− 2 − E u(r) = − 2 − E φ(r) + dr G(r, r )V (r )u(r )
dr dr 0

Z ∞
= 0 + dr 0 [−δ(r − r 0 )] V (r 0 )u(r 0 ) = −V (r)u(r) ,(3.89)
0

38
which leads exactly to equation ( 3.87).
Now, the integral in equation (3.88) can be approximated by a sum
∞
X
u(r) = φ(r) + ∆n G (r, rn ) V (rn ) u (rn ) , (3.90)
n=0

where 0 < r1 < r2 < . . . and where ∆1 , ∆2 , . . . represent the measures which belong
to the distribution {rn }.
For a short-range potential the summation in formula (3.90) can moreover be truncated.
We obtain then
N
X
u(r) = φ(r) + ∆n G (r, rn ) V (rn ) u (rn ) . (3.91)
n=0

This can again be written in the form of an integral equation, as follows

Z ∞ N
0 0
∆n V (rn ) δ (r 0 − rn ) u(r 0 ) .
X
u(r) = φ(r) + dr G(r, r ) (3.92)
0 n=0

Hence, when we define

N
X
U (r) = ∆n V (rn ) δ (r − rn ) , (3.93)
n=0

then wave function (3.92) solves the wave equation (3.87) with V (r) substituted by U (r).
The resulting equation can be solved by the method that we have exposed in the previous
paragraphs.

3.7.6 A closer look at the X-matrices

Let us inspect here in more detail the X-matrices which are defined in formula (3.79). For
that purpose we define two linearly independent solutions, F and G, of equation (3.60)
for r 6= b, by

cos(kr)
F (r) = sin(kr) and G(r) = − , (3.94)
k
Notice that those functions have been defined such that their Wronskian has the following
property

W (F, G)(r) = F (r)G0 (r) − F 0 (r)G(r)

!
cos(kr)
= sin(kr) sin(kr) − (k cos(kr)) − = 1
k

and (3.95)

W (G, F )(r) = G(r)F 0 (r) − G0 (r)F (r)

!
cos(kr)
= sin(kr) sin(kr) − − (k cos(kr)) = 1 .
k

39
For formula (3.67) we obtain then
    
u(r) F (r) G(r) C
  =    , (3.96)
u0 (r) F 0 (r) G0 (r) D
with C and D as defined in formula (3.68). Whereas, for the propagator (3.71) we find

G0 (r1 )
   
F (r2 ) G (r2 ) −G (r1 )
G (r2 , r1 ) =     . (3.97)
F 0 (r2 ) G0 (r2 ) −F 0 (r1 ) F (r1 )
Using formulas (3.95), (3.96) and (3.97), we verify, for the case that there is no delta-shell
in the interval [r1 , r2 ], that

 
u (r1 )
G (r2 , r1 )   =
0
u (r1 )

G0 (r1 )
     
F (r2 ) G (r2 ) −G (r1 ) F (r1 ) G (r1 ) C
=      
F 0 (r2 ) G0 (r2 ) −F 0 (r1 ) F (r1 ) F 0 (r1 ) G0 (r1 ) D
 
F (r2 ) G (r2 )
=  ×
0 0
F (r2 ) G (r2 )

G0 (r1 ) F (r1 ) − G (r1 ) F 0 (r1 ) G0 (r1 ) G (r1 ) − G (r1 ) G0 (r1 )

  
C
×  
−F 0 (r1 ) F (r1 ) + F (r1 ) F 0 (r1 ) −F 0 (r1 ) G (r1 ) + F (r1 ) G0 (r1 ) D
      
F (r2 ) G (r2 ) 1 0 C u (r2 )
=      =   , (3.98)
F 0 (r2 ) G0 (r2 ) 0 1 D u0 (r2 )

which confirms formulas (3.69) and (3.71). Whereas for the case that there is only one
delta-shell, at r = b, in the interval [r1 , r2 ], we obtain, by also using formula (3.73), the
following result

   
1 0 u (r1 )
G (r2 , r ↓ b)   G (r ↑ b, r1 )   =
aV0 1 u0 (r1 )

G0 (b)
    
F (r2 ) G (r2 ) −G(b) 1 0
=     ×
0 0 0
F (r2 ) G (r2 ) −F (b) F (b) aV0 1

G0 (r1 )
    
F (b) G(b) −G (r1 ) F (r1 ) G (r1 ) C
×    
F 0 (b) G0 (b) −F 0 (r1 ) F (r1 ) F 0 (r1 ) G0 (r1 ) D
 
F (r2 ) G (r2 )
=  ×
0 0
F (r2 ) G (r2 )

40
 G0 (b)F (b) − G(b)aV0 F (b) + G(b)F 0 (b) G0 (b)G(b) − G(b)aV0 G(b) − G(b)G0 (b)
 

× ×
0 0 0 0
−F (b)F (b) + F (b)aV0 F (b) + F (b)F (b) −F (b)G(b) + F (b)aV0 G(b) + F (b)G (b)
  
1 0 C
×  
0 1 D
   
F (r2 ) G (r2 ) 1 − G(b)aV0 F (b) −G(b)aV0 G(b) C
=     .(3.99)
F 0 (r2 ) G0 (r2 ) F (b)aV0 F (b) 1 + F (b)aV0 G(b) D

Notice that the central matrix, given by

 
1 − G(b)aV0 F (b) −G(b)aV0 G(b)
X(b, a) =   . (3.100)
F (b)aV0 F (b) 1 + F (b)aV0 G(b)
does not contain derivatives of the functions defined in formula (3.94).

41
Chapter 4

Formal scattering theory

In chapter (3) we discussed the dispersion of a wave packet from a target which is placed
at the origin of our coordinate system. In particular, we constructed a solution of the
time development operator H which at the instant of time t = 0 describes a wave packet
concentrated at the macroscopical distance r0 away from and moving towards the scat-
terer. At much later times this solution describes then a spherically expanding wave pulse
which passes through the detectors, placed at macroscopic distances also of the order of
r0 away from the target.
Here, we will follow a slightly different strategy. As in chapter (3), we concentrate on
elastic nonresonant scattering from a target situated at the origin of the coordinate system.
However, now we will concentrate on a single projectile particle. An incoming projectile
particle is characterized by its momentum, ~kin , long before scattering (t → −∞), whereas
an outgoing projectile particle is characterized by its momentum, ~kout , a long time after
scattering (t → +∞). Notice that, since we deal with elastic scattering, the energy of the
projectile particle is unchanged by the scattering process, which implies
2 2
~k ~k
 
in = out = k2 . (4.1)
We will moreover assume here that the projectile particle enters the interaction region
some time before t = 0 and leaves the interaction region some time after t = 0.
The time development of the wave function ψ (~r, t) is assumed to be described by a
Hamiltonian of the form

H = H0 + V , (4.2)
where the potential V is supposed to represent the interaction of the projectile particle
with the target and thus of short range, whereas H0 describes the time development of the
projectile particle outside the interaction region. Throughout these notes we will always
assume that H0 is simply the kinetic energy operator as in the Hamiltonian (3.2) and V
as given in formula (3.3). But in more sophisticated applications H0 may include part
of the interaction, provided that the solutions to the eigenvalue problem H0 Ψ = EΨ are
known.
In the following we wish to work in the momentum representation, for which the time
independent basis states are denoted by ~k , and which is associated to the coordinate
E


representation, for which the time independent basis states are denoted by |~ri, by the
following relation

42
 ~
D  E eik · ~r
~r ~k = , (4.3)

(2π)3/2
representing the plane wave approximation of a freely moving wave packet. When H 0 is
just given by the kinetic term of equation (3.2), then one has in the momentum represen-
tation the equation
 E  E k2
H0 ~k = E(k) ~k .
where E(k) = (4.4)
 
2m
which, comparing to formula (3.33), implies that we may identify
D  E
~r ~k = φ ~k, ~r
 
. (4.5)


4.1 The M6 oller operators

When a system is described by a wave equation of the form
∂
i χ (~r, t) = H χ (~r, t) , (4.6)
∂t
then its time evolution is given by

χ (~r, t) = e−iHt χ (~r, t = 0) , (4.7)

for any arbitrary initial wave function χ (~r, 0). This, because the wave function (4.7)
satisfies the wave equation (4.6), as can easily be seen from
∂
i χ (~r, t) = i(−iH) e−iHt χ (~r, 0) = H χ (~r, t) , (4.8)
∂t
independent of the choice for χ (~r, 0).
When we apply the above to the time development operators H0 and H, defined in
formula (4.2), for which we denote their solutions by φ and ψ respectively, then we have
the following solutions to the respective wave equations

φ (~r, t) = e−iH0 t φ (~r, 0) and ψ (~r, t) = e−iHt ψ (~r, 0) , (4.9)

for arbitrary initial wave functions φ (~r, 0) and ψ (~r, 0).
Now, since in the limit t → −∞ the time development operators H0 and H are supposed
to be equal, we may expect that there exist solutions to their respective wave equations
which in the same limit are equal, and similar for the limit t → +∞. Consequently,
we assume that there exists a solution φi (~r, t) of the wave equation related to H0 which
in the limit t → −∞ equals a solution ψi (~r, t) of the wave equation related to H, and
similarly φf (~r, t) and ψf (~r, t) in the limit t → +∞, i.e.

n o
lim ψi (~r, t) − φi (~r, t) = 0 ,
t→−∞
and (4.10)
n o
lim ψf (~r, t) − φf (~r, t) = 0 .
t→+∞

43
By the use of the expressions of formula (4.9) for the time development of respectively φ
and ψ, the above formulas (4.10) turn into

 
lim e−iHt ψi (~r, 0) − e−iH0 t φi (~r, 0) = 0 ,
t→−∞
and (4.11)
 
lim e−iHt ψf (~r, 0) − e−iH0 t φf (~r, 0) = 0 ,
t→+∞

which relations, upon multiplying by exp(iHt) from the left, give the following equations

ψi (~r, 0) = lim eiHt e−iH0 t φi (~r, 0)

t→−∞
and (4.12)
ψf (~r, 0) = lim eiHt e−iH0 t φf (~r, 0) .
t→+∞

At this stage it is convenient to introduce the M6 oller operators, defined by

Ω(+) = lim eiHt e−iH0 t and Ω(−) = lim eiHt e−iH0 t . (4.13)
t→−∞ t→+∞

After which definition we obtain for formula (4.12) the expressions

ψi (~r, 0) = Ω(+) φi (~r, 0) and ψf (~r, 0) = Ω(−) φf (~r, 0) . (4.14)

For the M6 oller operators one has the following property

H Ω(±) = Ω(±) H0 , (4.15)

which we will proof below for Ω(+) :
When, in general, a function of the time t tends to zero in the limit t → −∞, then its
derivatimes with respect to t also vanish in the same limit, hence for the first expression
in formula (4.11) one has

" #
∂

0 = lim i e−iHt ψi (~r, 0) − e−iH0 t φi (~r, 0)
t→−∞ ∂t
 
= lim e−iHt H ψi (~r, 0) − e−iH0 t H0 φi (~r, 0) ,
t→−∞

which relation, upon multiplying by exp(iHt) from the left, gives the following equation

H ψi (~r, 0) = lim eiHt e−iH0 t H0 φi (~r, 0) .

t→−∞

Substitution of formulas (4.13) and (4.14) leads to

H Ω(+) φi (~r, 0) = Ω(+) H0 φi (~r, 0) ,

which proofs formula (4.15) for Ω(+) , since the result is valid for an arbitrary state φi .
For the other M6 oller operator, Ω(−) , the proof is similar because of formula (4.11).

44
 The property (4.15) for the M6 oller operators has the following interresting consequence:
 E
The eigenstates ~k of the Hamiltonian H0 , as defined in formula (4.4), represent freely


moving particles with a well defined momentum ~k. Now, because of momentum conser-
vation, when a freely moving particle has momentum ~k at the instant of time t = 0, then
it has momentum ~k for all times. Hence, the wave function

e−iH0 t ~k
 E


describes a freely moving particle with a definite momentum equal to ~k. Consequently,
from their definition in formula (4.12) one has that the wave functions
 
 (±) ~
= Ω(±) ~k
 E E
ψ k (4.16)

are states which at t → ±∞ coincide with a particle of definite momentum ~k. Moreover,
because of the property (4.15), also using the eigenstate equation (4.4), one finds that the
wave functions (4.16) are eigenstates of the scattering Hamiltonian H, as

  E  E  E 
H ψ (±) ~k = H Ω(±) ~k = Ω(±) H0 ~k = Ω(±) E(k) ~k = E(k) ψ (±) ~k
 E  E
.
    

(4.17)
~
Now, since a particle enters the interaction region with momentum kin and leaves it
with momentum ~kout , following their definition in formula (4.16), during interaction its
state is changed
 
from ψ (+) ~kin to ψ (−) ~kout
 E  E
, (4.18)
 
 
 (+) ~  (−) ~
 E  E
where ψ kin and ψ describe respectively an incoming projectile particle
kout
with momentum ~kin and an outgoing projectile particle with momentum ~kout .

45
4.2 The scattering operator S
The matrix elements S ~kout , ~kin of the scattering operator S give the probability that
 

a projectile particle, which enters the interaction region with momentum ~kin , leaves this
region with momentum ~kout . By virtue of the definitions expressed in formula (4.18),
such transition probabilities are given by

S ~kout , ~kin ψ (−) ~kout ψ (+) ~kin
  D   E
= . (4.19)


Substitution of the definition given in formula (4.16), gives moreover

 h  
 (−) † (+) 
S ~kout , ~kin ~k Ω  ~kin
  i
= out  Ω
 . (4.20)
 E  E
Now, since ~kin and ~kout are both supposed to represent arbitrary elements of an
 

orthonormal basis of eigenstates of H0 , we may define

h i†
S = Ω(−) Ω(+) . (4.21)

 (−) ~
 E
The set of states ψ k
forms a complete basis of eigenstates of the Hamiltonian

H, and so does the set of states ψ (+) ~k . That they are eigenstates of H can be seen
 E

from formula (4.17), whereas their normalizations are given by

 h  
~k 0  Ω(±) † Ω(±)  ~k
  E
~k 0 ψ (±) ~k ~k 0 ~k = δ (3) ~k 0 − ~k
D    E i D  
(±)
ψ =   = , (4.22)

because the M6 oller operators are unitary as can easily be seen from their definition in
formula (4.13).
Any solution of the stationary
 E wave equation for H can be expanded in either basis.
 (+) ~
In particular, a state ψ k , which describes an incoming plane wave long before
  E
interaction, can be expanded in the set of basis states ψ (−) ~k , which describe outgoing


plane waves. Consequently, the matrix elements (4.19) of the S operator are just the
corresponding expansion coefficients.

46
4.3 The Green’s operator G0
The Green’s operator or resolvent, G0 (z), corresponding to the self-adjoint free Hamilto-
nian H0 is defined to be

G0 (z) = (z − H0 )−1 , (4.23)

where z is an arbitrary complex number and where we assumed that the inverse of (z − H0 )
exists. Its relation to the retarded Green’s function, shown in formula (3.26), is given by

G+ (~r, ~r 0 ) = lim h~r |G0 (E(k) + i)| ~r 0 i . (4.24)

↓0

Below we will demonstrate that relation (4.24) is correct.

To that aim, we first determine the following nasty integral
~0
Z
d3 k 0 eik · ~a
lim . (4.25)
↓0 (2π)3 k 02 − (k 2 + i)

We begin by setting up a system of spherical coordinates for the integration variable ~k 0 ,

such that the k30 -axis coincides with ~a. This then, with ϑ the angle between ~k 0 and ~a,
gives
~k 0 · ~a = k 0 a cos(ϑ) ,

and gives, moreover, for the integral (4.25) the expression

1 ik 0 a cos(ϑ)
0 e
Z 2π Z +1 Z ∞
02
lim dϕ d cos(ϑ) k dk 02 . (4.26)
↓0 (2π)3 0 −1 0 k − (k 2 + i)
The integrand does not depend on ϕ, so the related integration yields a factor 2π. For
the integral over cos(ϑ) we obtain
0 +1

+1 eik az  1
 
0 0 0
Z
dz eik az =  = eik a − e−ik a ,
−1 ik 0 a 
 ik 0 a
−1

after which (4.26) turns into

0 0
k 0 dk 0 eik a k 0 dk 0 e−ik a 
 
1 Z ∞ Z ∞
lim − . (4.27)
↓0 ia(2π)2  0 k 02 − (k 2 + i) 0 k 02 − (k 2 + i) 
When, next, we redefine the integration variable k 0 in the second term for −k 0 , then the
above equation (4.27) takes the form

0 0
 
1 Z ∞ k dk eik a
0 0 Z 0
k 0 dk 0 eik a 
lim + =
↓0 ia(2π)2  0 k 02 − (k 2 + i) 02 2
−∞ k − (k + i) 

0
1 +∞ k 0 dk 0 eik a
Z
= lim √ √  . (4.28)
↓0 ia(2π)2
 
−∞ k0 − k 2 + i k0 + k 2 + i

47
This integral can be turned into a complex contour integral by closing it, using a half
circle with infinite radius in the upper half complex plane. A positive infinite imaginary
part for k 0 in the exponent exp(ik 0 a) damps the contribution to the integral of this half
circle to zero. The only nonzero contributions for this counterclockwise closed contour
come from the residues of the poles in the upper half complex plane. Now, for small  we
may approximate
√ 


2
k + i ≈ k 1 + i 2 .
2k
√
Consequently, the pole at k = + k 2 + i lies in the upper half complex plane √ and thus
gives a contribution to the integral (4.28). Whereas the other pole, at k = − k 2 + i, is
in the lower half complex plane and hence does not contribute to the integral (4.28). The
relevant residue is readily determined and after taking the limit  ↓ 0, we end up with

2πi ika
e ika = 1 e (4.29)
ia(2π)2 4π a
for the integral (4.25).
After this exercise we return to the retarded Green’s function of formula (4.24). We
begin by the substitution of the definition of the
 EGreen’s operator (4.23) in expression
(4.24), and moreover apply completeness of the ~k -basis by inserting unity, i.e.


Z D   
d3 k 0 ~r ~k 0 ~k 0 [E(k) + i − H0 ]−1  ~r 0
ED E
G+ (~r, ~r 0 ) = lim .

↓0

Then we let the Green’s operator act to the left, to obtain

Z D   
d3 k 0 ~r ~k 0 ~k 0 [E(k) + i − k 02 ]−1  ~r 0
ED E
G+ (~r, ~r 0 ) = lim .

↓0

Substitution of formulas (4.3) and (4.4) gives moreover

~0
 
0
3 0
d k 2m eik · (~r − ~r )
Z
G+ (~r, ~r 0 ) = − lim .
 
(2π)3 k 02 − (E(k) + i2m)

↓0

Comparing this expression with the integral (4.25) and its solution (4.29), we find, after
taking ~a = ~r − ~r 0 and a = |~r − ~r 0 |, the result
0
0 m eik |~r − ~r |
G+ (~r, ~r ) = − .
2π |~r − ~r 0 |

which is indeed equal to formula (3.26).

To conclude this section we note that since [H0 ]† = H0 , it is easily seen that

[G0 (z)]† = G0 (z ∗ ) . (4.30)

48
4.4 The LS equation in momentum space
In section (3.4) we studied the Lippmann Schwinger equation for the solutions of the sta-
tionary Schrödinger equation (3.7). Here, we will rewrite
 formula
 E (3.32) in the momentum
 (+) ~
representation. To that aim we define an eigenstate ψ k of the Hamiltonian (4.2),
i.e. to
 the equation Hψ = E(k)ψ, which in coordinate space is related to the solution
(+) ~
ψ k, ~r of equation (3.31) by
D 
ψ (+) ~k, ~r ~r ψ (+) ~k
   E
= . (4.31)
Moreover, we define for a local potential V the following relation in the coordinate repre-
sentation

h~r |V | ~r 0 i = V (~r) δ (3) (~r − ~r 0 ) . (4.32)

First, we rewrite equation (3.32) to

~
eik · ~r
Z Z
~k, ~r d3 r 0 0 G+ (~r, ~r 0 ) V (~r 0 ) δ (3) (~r 0 − ~r 0 0 ) ψ (+) ~k, ~r 0 0
   
(+) 3 0
ψ = + dr ,
(2π)3/2

and then we substitute the relations (4.3), (4.24), (4.31), and (4.32), to obtain

D  D  E Z Z 
~r ψ (+) ~k ~r ~k + ~r 00 ψ (+) ~k
 E D ED  E
= dr 3 0
d3 r 00 h~r |G0 (E(k) + i)| ~r 0 i ~r 0 |V | ~r 00 ,
  

(4.33)
where the limit  ↓ 0 is implicitly understood. Next we remove two times the identy
operation which stems from completeness of the |~ri-basis, to end up with
D  D  E
~r ψ (+) ~k ~r ~k + ~r |G0 (E(k) + i) V | ψ (+) ~k
 E D  E
= ,
 

which leads to the following relation

  E 
 (+) ~
= ~k + G0 (E(k) + i) V ψ (+) ~k
 E  E
ψ k . (4.34)
 

This is the Lippmann Schwinger equation in momentum space. Like in the case of the
wave function (3.32) for which we wrote the perturbation series (3.35), one may also iterate
formula (4.34) in order to obtain (for convenience we write G0 instead of G0 (E(k) + i))

  E  E  E
 (+) ~
= ~k + G0 V ~k + G0 V G0 V ~k
 E
ψ k + ··· (4.35)
  

 E  E
= ~k + {G0 + G0 V G0 + G0 V G0 V G0 + · · ·} V ~k .
 

At this stage it is oportune to define the Green’s operator G(z) by

G(z) = G0 (z) + G0 (z)V G0 (z) + G0 (z)V G0 (z)V G0 (z) + · · · . (4.36)

This series can be summed, using the following expansion for operators A and B:

49
   −1  −1
(A − B)−1 = A 1 − A−1 B = 1 − A−1 B A−1
  2  3 
= 1 + A−1 B + A−1 B + A−1 B + · · · A−1 (4.37)

= A−1 + A−1 BA−1 + A−1 BA−1 BA−1 + A−1 BA−1 BA−1 BA−1 + · · ·

So, when we substitute for A the expression (z − H0 ), which implies, by formula (4.23),
A−1 = G0 (z), and for B the potential operator, V , we obtain for the righthand side of
formula (4.37) the righthand side of formula (4.36). Consequently, we discover for the
Green’s operator (4.36) the expression

G(z) = (z − H0 − V )−1 = (z − H)−1 , (4.38)

as a generalization of the definition (4.23), and hence for the series expansion of formula
(4.35) the relation
  E
 (+) ~
= lim [1 + G (E(k) + i) V ] ~k
 E
ψ k . (4.39)

↓0

To conclude this section we note that since H † = H, it is easily seen that

[G(z)]† = G (z ∗ ) . (4.40)

50
4.5 Relation Green’s operators and M6 oller operators
Let |φ0 i represent an arbitrary state in momentum space, and |ψ± i be defined by

|ψ± i = Ω(±) |φ0 i = lim eiHt e−iH0 t |φ0 i . (4.41)

t→∓∞

It is this definition which we would like to rewrite in terms of the Green’s operator.
First, we remember to write a function of the time t as the integral of its derivative
with respect to t. For |ψ− i we obtain then
  Z +∞
|ψ− i = eiHt e−iH0 t + dt eiHt iV e−iH0 t |φ0 i ,
t=0 0

where the derivative, also using formula (4.2), comes from

∂
 
eiHt e−iH0 t = eiHt (iH − iH0 ) e−iH0 t = eiHt iV e−iH0 t .
∂t
So, we arrive for the state |ψ− i of formula (4.41) at the expression
 Z +∞ 
|ψ− i = 1 + dt eiHt iV e−iH0 t |φ0 i , (4.42)
0
and similarly for the other state, |ψ+ i of formula (4.41), at the expression
 Z 0 
|ψ+ i = 1 + dt eiHt iV e−iH0 t |φ0 i . (4.43)
−∞

The integrals of formulas (4.42) and (4.43) are absolutely convergent, in which case
one may substitute the integrands for

 Z +∞ 
|ψ− i = 1 + i lim dt e−t eiHt V e−iH0 t |φ0 i
↓0 0
and (4.44)
 Z 0 
|ψ+ i = 1 + i lim dt e+t eiHt V e−iH0 t |φ0 i .
↓0 −∞

We can interpret the expressions (4.44) by noting that it differs from the formulas (4.42)
and (4.43) by a change in potential from V to V exp (− |t|), which expresses the fact
that the motion of any scattering orbit would be unchanged under such redefinition of
the potential for small . This is exactly what one would expect, since we know that
the projectile particle eventually moves so far away that the potential ceases to have any
effect. The result that one can do scattering theory with V replaced by V exp (− |t|) is
known as the adiabatic theorem.  E
When we insert unity, in the form of a complete set of states ~k , into the equations


(4.44), then we find

 Z Z +∞ 
dt e−t eiHt V e−iH0 t ~k ~k  |φ0 i
 ED 
3
|ψ− i = 1 + i lim dk
 
↓0 0
and (4.45)
 Z Z 0  ED 
dt e+t eiHt V e−iH0 t
~ ~ 
|ψ+ i = 1 + i lim d3 k k k  |φ0 i .
↓0 −∞

51
Now, by the property (4.4) we obtain

 Z Z +∞ 
dt e−i (E(k) − i − H) t V ~k ~k  |φ0 i
ED 
3
|ψ− i = 1 + i lim dk
 
↓0 0
and (4.46)
 Z Z 0  ED 
dt e−i (E(k) + i − H) t V
~ ~ 
|ψ+ i = 1 + i lim d3 k k k  |φ0 i .
↓0 −∞

After moreover performing the t integrations, the expresions (4.46) turn into

 Z  ED 
3 −1 ~ ~ 
|ψ− i = 1 + lim d k (E(k) − i − H) V k k  |φ0 i
↓0
and (4.47)
 Z  ED 
~ ~ 
|ψ+ i = 1 + lim d3 k (E(k) + i − H)−1 V k k  |φ0 i .
↓0

Here, we recognize the Green’s operator, G (E(k) ± i), defined in formula (4.38), substi-
tution of which leaves us with

 Z  
d3 k G (E(k) − i) V ~k ~k  |φ0 i
ED
|ψ− i = 1 + lim
↓0
and (4.48)
 Z  ED 
~ ~ 
|ψ+ i = 1 + lim d3 k G (E(k) + i) V k k  |φ0 i .
↓0

Note that the signs of the term ±i correspond to the subscripts ± of the state vectors
ψ± , and hence to the superscripts of the operators Ω(±) .
Finally, returning to the definitions (4.41), we end up with
Z  ED 
Ω (±)
= 1 + lim d3 k G (E(k) ± i) V ~k ~k  . (4.49)
 
↓0

When we apply the above relation (4.49), between Green’s and M6 oller operators, to
the case of the states ψ (±) , which are defined in formula (4.16), then we may perform the
~k-integration, as

 
 (±) ~
= Ω(±) ~k
 E E
ψ k
 Z    E
d3 k 0 G (E(k 0 ) ± i) V ~k 0 ~k 0  ~
ED
= 1 + lim k

↓0

 E Z 
= ~k d3 k 0 G (E(k 0 ) ± i) V ~k 0 δ (3) ~k 0 − ~k
E  
+ lim
 
↓0
   E
= 1 + lim G (E(k) ± i) V ~k . (4.50)

↓0

52
4.6 The LS equation for the Green’s operators
For the Green’s operators G0 , defined in formula (4.23), and G, defined in formula (4.38),
we may derive the following Lippmann Schwinger relations.
For the first relation we rewrite formula (4.36) as

G(z) = G0 (z) + G0 (z)V {G0 (z) + G0 (z)V G0 (z) + · · ·}

= G0 (z) + G0 (z)V G(z) . (4.51)

For the second relation we rewrite formula (4.36) as

G(z) = G0 (z) + {G0 (z) + G0 (z)V G0 (z) + · · ·} V G0 (z)

= G0 (z) + G(z)V G0 (z) . (4.52)

Those two equations constitute one of the essential corner stones of the stationary scat-
tering formalism.

53
4.7 The transition operator T
In scattering theory it is convenient to introduce the transition operator, T (z), which is
defined in terms of G(z) and V as

T (z) = V + V G(z) V . (4.53)

When this expression is multiplied on the left by G0 (z) and moreover relation (4.51)
substituted, then we find

G0 (z) T (z) = G(z) V . (4.54)

Similarly, when expression (4.53) is multiplied on the right by G0 (z) and moreover relation
(4.52) substituted, then we find

T (z) G0 (z) = V G(z) . (4.55)

The most famous Lippmann Schwinger equation is the one for the T operator, which can
be obtained by substituting relation (4.54) into the definition (4.53) for the transition
operator, since then

T (z) = V + V G0 (z) T (z) . (4.56)

Furthermore, we note that the identity (4.40) for G(z) leads to a corresponding result
for T (z), as, assuming that V is Hermitean, is easily seen from formula (4.53) by

[T (z)]† = V + V [G(z)]† V = V + V G(z ∗ )V = T (z ∗ ) . (4.57)

An other interresting property of T (z) stems from the matrix element

~k 0 |V | ψ (+) ~k
D  E
,

which, following formula (4.50) can be casted in the form

lim ~k 0 |V {1 + G (E(k) + i) V }| ~k

D E
.
↓0

Consequently, by the use of the definition (4.53), we then find

~k 0 |V | ψ (+) ~k = lim ~k 0 |T (E(k) + i)| ~k

D  E D E
. (4.58)
↓0

54
4.8 Relation T and S operators
In order to establish a relation between the transition operator T and the scattering
operator S, we start from the definition (4.19)

S ~k 0 , ~k ψ (−) ~k 0 ψ (+) ~k
  D    E
= . (4.59)


To proceed, we determine for ψ (−) ~k , using their definitions (4.16), the following identity
 

   E  E
 (−) ~
− ψ (+) ~k = Ω(−) ~k − Ω(+) ~k
 E  E
ψ k ,
  

which by virtue of formula (4.50), equals

 E
lim {G (E(k) − i) − G (E(k) + i)} V ~k .

↓0

So, we find
D 
ψ (−) ~k  = ψ (+) ~k  + lim ~k  V † G† (E(k) − i) − G† (E(k) + i)
D   D   n o
,
  
↓0

which using the property (4.40), moreover assuming that the potential V is Hermitean,
can be written as
D 
ψ (−) ~k  = ψ (+) ~k  + lim ~k  V {G (E(k) + i) − G (E(k) − i)}
D   D  
.
  
↓0

This expression substituted in formula (4.59) gives

S ~k 0 , ~k
 
= (4.60)
      
0 
ψ (+) ~k ψ (+) ~k + lim ~k 0 |V {G (E(k 0 ) + i) − G (E(k 0 ) − i)}| ψ (+) ~k
 
.
↓0

The first term on the righthand side of equation (4.60) is handled by formula (4.22). Using
the definition of the Green’s operator (4.38), we find for the second term on the righthand
side of equation (4.60) the following

~k 0 V (E(k 0 ) + i − H)−1 − (E(k 0 ) − i − H)−1  ψ (+) ~k
D n o  E
,

which, because of relation (4.17), turns into


~k 0 V (E(k 0 ) + i − E(k))−1 − (E(k 0 ) − i − E(k))−1  ψ (+) ~k
D n o  E
=

~k 0 |V | ψ (+) ~k
n o D  E
−1 −1
= (E(k 0 ) + i − E(k)) − (E(k 0 ) − i − E(k)) .

Next, we use the following representation of the Dirac delta function:

2i 1 1
 
2πiδ(x) = lim 2 = lim − ,
→0 x + 2 →0 x − i x + i

55
which when substituted in our final expression, gives

−2πiδ (E(k 0 ) − E(k)) ~k 0 |V | ψ (+) ~k

D  E
. (4.61)
Inserting the results (4.22) and (4.61) into equation (4.60), leads to

S ~k 0 , ~k = δ (3) ~k 0 − ~k − 2πiδ (E(k 0 ) − E(k)) ~k 0 |V | ψ (+) ~k

    D  E
. (4.62)

Here, we may moreover substitute relation (4.58) for the transition operator, to end up
for (4.62) with

    D E
S ~
k 0, ~
k = δ (3) ~
k0 −~
k − 2πiδ (E(k0 ) − E(k)) ~
k 0 |T (E(k) + i)| ~
k .
(4.63)
This is one of the central results of time-independent scattering theory.
Formula (4.63) shows explicitly that the matrix elements of the S operator vanish when
E(k 0 ) does not equal E(k), even when the matrix elements of the T operator do not vanish
in that case. In other words, the matrix elements of S are on-shell, whereas the matrix
elements of T might be off-shell.

56
57
Chapter 5

Examples

In this chapter we will work out in some detail two examples of models for the description
of scattering by means of a potential in Schrödinger theory.

5.1 The potential term

The potential term in the configuration space wave equation takes in general the form
Z    
d3 r 0 V ~r , ~r 0 ψ ~r 0 , (5.1)
which gives then the wave equation
1 2
Z    
− ∇r ψ (~r ) + d3 r 0 V ~r , ~r 0 ψ ~r 0 = E ψ (~r ) , (5.2)
2µ
where E represents the total energy of the system.
In momentum space we define
0
d3 r −i~k · ~r d3 r 0 i~k · ~r 0 V ~r , ~r 0
  Z Z
0
V ~k , ~k
 
e = e , (5.3)
(2π)3/2 (2π)3/2
for which quantity we may study the following expression

d3 k 00 i~k 00 · ~r
Z Z    
00 0 0
e d3 k 0 V ~k , ~k φ ~k =
(2π)3/2

d3 k 00 i~k 00 · ~r d3 r 00 −i~k 00 · ~r 00 d3 r 0 ~k 0 · ~r 0  00 0  ~ 0
Z Z Z Z  
= e dk 3 0
e e i V ~r , ~r φ k
(2π)3/2 (2π)3/2 (2π)3/2
00
 
d3 k 00 i~k · ~r − ~r 00 

d3 r 00  d3 r 0 ~k 0 · ~r 0  00 0  ~ 0
Z Z Z Z  
= d3 k 0 e e i V ~r , ~r φ k
(2π)3/2 (2π)3/2 (2π)3/2
0
Z Z
d3 r 00 
00
 Z d3 r 0 i~k · ~r 0 V ~r 00 , ~r 0 φ ~k 0
   
= d3 k 0 (2π) 3/2 (3)
δ ~
r − ~
r e
(2π)3/2 (2π)3/2
0
d3 k 0 i~k · ~r 0 φ ~k 0
Z   Z  
3 0 0
= d r V ~r , ~r e
(2π)3/2

58
 Z    
= d3 r 0 V ~r , ~r 0 ψ ~r 0 . (5.4)

The result (5.4) may be substituted in the configuration space wave equation (5.2), to
give

d3 k ~k · ~r 1 2 ~  d3 k ~k · ~r
Z Z Z    
i i 0 0
3/2
e k φ k + 3/2
e d k V k , k φ ~k
3 0 ~ ~ =
(2π) 2µ (2π)

d3 k i~k · ~r φ ~k
Z  
= E e , (5.5)
(2π)3/2
or also

d3 k k2
( ! )
~
Z Z   
eik · ~r
0 0
φ ~k d k V ~k , ~k φ ~k
 
3 0
3/2
−E + = 0 , (5.6)
(2π) 2µ
which gives for the wave equation in momentum space the expression

k 2 ~ 
Z    
0 0
d3 k 0 V ~k , ~k φ ~k = E φ ~k
 
φ k + . (5.7)
2µ

5.1.1 Local potentials

For a local potential in the configuration space wave equation (5.2) one may select the
following general form
     
V ~r , ~r 0 = δ (3) ~r − ~r 0 V ~r 0 , (5.8)
since indeed
Z     Z      
d3 r 0 V ~r , ~r 0 ψ ~r 0 = d3 r 0 δ (3) ~r − ~r 0 V ~r 0 ψ ~r 0 =
Z Z Z
0 0 0 0
= dr δ(r − r ) dϑ δ(ϑ − ϑ ) dϕ0 δ(ϕ − ϕ0 ) V (r 0 , ϑ0 , ϕ0 ) ψ(r 0 , ϑ0 , ϕ0 )

= V (r , ϑ , ϕ ) ψ(r , ϑ , ϕ ) = V (~r ) ψ (~r ) . (5.9)

The momentum space analogon of the configuration space potential is defined in for-
mula (5.3), which, when applied to expression (5.8), gives

d3 r ~k · ~r Z d3 r 0 ~k 0 · ~r 0
  Z
V ~k , ~k
0 −i i
 
0
= e e V ~
r , ~
r
(2π)3/2 (2π)3/2
0
d3 r ~ d3 r 0 i~k · ~r 0 δ (3) ~r − ~r 0 V ~r 0
Z Z
e−ik · ~r
   
= 3/2
e
(2π) (2π)3/2
 
0
Z
dr 3 i −~k + ~k · ~r
= e V (~r ) . (5.10)
(2π)3

59
5.1.2 Spherically symmetric local potentials
For spherically symmetric local potentials in configuration space, of the form
   
V ~r , ~r 0 = δ (3) ~r − ~r 0 V (r) , (5.11)
one may moreover perform the angular integrations of formula (5.10), to obtain for the
momentum space potential the following

 
0
i −~k + ~k  r cos(ϑ)


0
 Z ∞ Z 2π Z +1
V ~k , ~k = (2π)−3 r 2 dr V (r) dϕ d cos(ϑ)e
0 0 −1

1 Z∞ 2
  
0
=

~ ~
r dr V (r) j0 −k + k  r
 (5.12)
2π 2 0
In the book [2] of Morse and Feshbach, in chapter 11 at page 1574, we find furthermore
the following expansion for the spherical Bessel function
∞
q !
0 0
j0 r k 2 + k − 2~k · ~k

02 j` (kr) j` (k 0 r) ,
X
= (2` + 1) P` k̂ · k̂ (5.13)
`=0

which leads for (5.12) to

∞
1 X
  Z ∞
0 0
V ~k , ~k

= (2` + 1) P ` k̂ · k̂ r 2 dr V (r) j` (kr) j` (k 0 r) . (5.14)
2π 2 `=0 0

60
5.2 Relation transition amplitude and T -matrix
The relation between the transition amplitude and the retarded wave functions is given
in formula (3.39), i.e.
√ Z 0
f ~k, r̂ d3 r 0 e−ikr̂ · ~r V (~r 0 ) ψ (+) ~k, ~r 0
   
= − 2π µ . (5.15)
In ket-notation and for a non-local potential one has

√ Z 0Z
f ~k, r̂ = − 2π µ d3 r 0 e−ikr̂ · ~r d3 r 00 V ~r 00 δ (3) ~r 0 − ~r 00 ψ (+) ~k, ~r 00
       

√ Z 0 Z
d r e−ikr̂ · ~r ~r 00 |Ω+ | ~k
D ED E
= − 2π µ 3 0
d3 r 00 ~r 0 |V | ~r 00

√ Z 0
d3 r 0 e−ikr̂ · ~r ~r 0 |V Ω+ | ~k
D E
= − 2π µ , (5.16)

where we also introduced the M6 oller operator Ω+ , defined in formula (4.13).

Next, we use formulas (4.3), (4.50) and (4.58) to obtain the result

√ Z 0 Z    
d r e−ikr̂ · ~r
 0
f ~k, r̂ ~ ~k 0 |V Ω | ~k
 
3 0 3 0 0 +
= − 2π µ dk ~r k

~ 0 r0
 
√ 0 e ik · ~
Z 
Z   0 
= − 2π µ d3 k 0 d3 r 0 e−ikr̂ · ~r ~k |V Ω+ | ~k

 (2π)3/2 

   
µ
Z i
Z ~k 0 − kr̂ · ~r 0 
  0 
= − 3 0 3 0
dk  dr e ~ 2 ~
k |T (k + i)| k , (5.17)
2π  


The integration in configuration space can, by the use of formula (19.68) of [1], be
handled as follows

 
0
Z i ~k − kr̂ · ~r 0  
3 0
dr e = (2π) δ 3 (3) ~k 0 − kr̂ (5.18)

∞
2
X
0
 0 δ(k − k 0 )
= 2π (2` + 1) P `0 r̂ · k̂ .
`0 =0
k2

So, after calculating the matrix elements of the T -operator, one still has to perform
integrations.
Now, let us assume that the matrix elements of T have the following genuine form in
the case of spherical symmetry
  ∞
~k |T (k 2 + i)| ~k 0 0

T` (k, k 0 ) .
X
= (2` + 1) P` k̂ · k̂ (5.19)
`=0

61
When we substitute both expansions (5.19) and (5.18) into formula (5.17), then we find

f ~k, r̂
 
= (5.20)
∞
δ(k − k 0 )
Z  0  0
(2` + 1) (2`0 + 1) d3 k 0 P` k̂ · k̂ T` (k, k 0 ) P`0 r̂ · k̂
X
= −πµ .
`,`0 =0 k2

Next, let us concentrate on the integration over the angles in k 0 -space, i.e.
Z  0  0
dΩk0 P` k̂ · k̂ P`0 r̂ · k̂ . (5.21)
0
Here, we may select the z-axis of the ~k -system in the direction of ~k. But, that does not
help much for the second term in the integrand of formula (5.21). However, by the use of
the addition theorem (see e.g. formula 9.79 of [1]) we find

Z  0  0
dΩk0 P` k̂ · k̂ P`0 r̂ · k̂ = (5.22)

+` 0
Z  0 4π X (`0 ) ∗
 
(`0 )
 0 
= dΩk0 P` k̂ · k̂ Y` 0 z k̂ · r̂ Y` 0 z k̂ · k̂
2`0 + 1 `0 z =−`0

+` 0
4π X (`0 ) ∗
  Z
(`0 )
= Y` 0 z k̂ · r̂ dΩk0 P` (Ωk0 ) Y`0 z (Ωk0 ) .
2`0 + 1 `0 z =−`0

Then, we substitute for P` (Ωk0 ) in the above expression (5.22) formulas (9.78) and (9.65a)
of [1], to find

Z  0  0
dΩk0 P` k̂ · k̂ P`0 r̂ · k̂ = (5.23)

0
s
+`
4π X (`0 ) ∗
  Z 4π (`) ∗ (`0 )
= Y` 0 z k̂ · r̂ dΩk0 Y0 (Ωk0 ) Y`0 z (Ωk0 ) ,
2`0 + 1 `0 z =−`0 2` + 1

which, after applying the orthogonality of spherical harmonics (see e.g. [1] formula 9.69),
leads to

Z  0  0
dΩk0 P` k̂ · k̂ P`0 r̂ · k̂ = (5.24)

0
s
+`
4π X (`0 ) ∗
  4π
= Y` 0 z k̂ · r̂ δ`,`0 δ0,`0 z
2`0 + 1 `0 z =−`0 2` + 1
s
4π (`0 ) ∗
  4π 4π  
= Y k̂ · r̂ δ`,`0 = P` k̂ · r̂ δ`,`0 .
2`0 + 1 0 2` + 1 2` + 1
On substitution of the result (5.24) into the expression (5.20) one finds

62
 ∞  Z ∞
f ~k, r̂
  
2
dk 0 T` (k, k 0 ) δ(k − k 0 )
X
= −4π µ (2` + 1) P` k̂ · r̂
`=0 0

∞  
2
X
= −4π µ (2` + 1) P` k̂ · r̂ T` (k) . (5.25)
`=0

For the relation between the transition amplitude and the partial wave phase shifts one
has the following formulas (see [1] in formulas 11.59 to 11.62):

∞
S` (k) − 1
f ~k, r̂
  X  
= (2` + 1) P` k̂ · r̂
`=0 2ik

1 X∞   e2iδ` (k) − 1
= (2` + 1) P` k̂ · r̂
k `=0 2i
∞
1 X
(2` + 1) P` k̂ · r̂ eiδ` (k) sin (δ` (k))
 
=
k `=0
∞
1 X   1
= (2` + 1) P` k̂ · r̂ , (5.26)
k `=0 cotg (δ` (k)) − i

from which one moreover may deduce, that for a spherically symmetric local potential
holds the relation

S` (k) − 1
−4π 2 µ T` (k) = . (5.27)
2ik

63
5.3 The delta-shell potential
Let us consider the following one-channel Schrödinger equation
( )
1 λ
2
−∇ + δ(r − a) ψ (~r ) = E ψ (~r ) for a > 0 . (5.28)
2µ a
Inside (r < a) and outside (r > a) the sphere of radius a the scattered particles feel no
interaction. Hence, their equation of motion inside and outside the sphere is given by

∇2
− ψ (~r ) = E ψ (~r ) for r < a and r > a , (5.29)
2µ
which is the equation of motion for a freely moving particle.
Since the delta-shell potential is spherically symmetric, we may assume that its solu-
tions have definite quantum numbers, ` and `z , for respectively angular momentum and
its z component. When we define the radial wave function u` (r) by

u` (r) (`)
Y`z (ϑ, ϕ) ,
ψ (~r ) = (5.30)
r
then we find for the radial partial wave equation the expression

d2
" #
1 ` (` + 1) λ
− 2 + 2 + δ(r − a) u` (r) = E u` (r) . (5.31)
2µ dr r a
At r = a one obtains the following boundary conditions:

   
1  d u` (r)  d u` (r)  λ
u` (r ↑ a) = u` (r ↓ a) and − + + u` (a) = 0 .
2µ dr r↓a dr r↑a a
 
 
(5.32)
The second boundary condition of formula (5.32), which implies that the derivative of the
wave function is discontinuous at r = a, is found by integrating wave equation (5.31) at
an infinitesimal interval including the point r = a, i.e.
a+ d2
( " # )
lim 1 ` (` + 1) λ
Z
↓0 dr − 2 + 2 + δ(r − a) u` (r) = E u` (r) ,
a− 2µ dr r a

thereby noticing that in the limit  ↓ 0 the integrals over u` (r) and u` (r)/r 2 vanish.
The first boundary condition of formula (5.32) is obtained by integrating the second
boundary condition, also at an infinitesimal interval which includes the point r = a.
Further boundary conditions are the usual: The radial wave function u` (r) vanishes
at the origin (r = 0) and must have oscillating behavior at infinity (r → ∞) since it
describes a freely moving scattered particle.
The differential equation of formula (5.31) is for λ = 0 solved by spherical Bessel (J)
and Neumann (N ) functions, the first of which behaves well at the origin. A precise
definition of those functions is given below.
Let us first recall that the differential equation

d2
!
1 ` (` + 1)
− 2 + u` = E u ` , (5.33)
2µ dr r2

64
is solved by the following two linearly independent functions

J` (kr) = k −` rj` (kr) and N` (kr) = k `+1 rn` (kr) , (5.34)

where the linear momentum k is defined by

k 2 = 2µE (5.35)
and where j` and n` are the spherical Bessel and Neumann functions, respectively, the
properties of which can be found in any textbook on Quantum Mechanics (see e.g. [1]
pages 194 ff). The solution J` (kr) satisfies the usual boundary conditions at the origin,
i.e. J` (r → 0) → 0.
For the solutions ( 5.34) one has the following Wronskian relation
" #
d N` (r) d J` (r)
W (J` (ka), N` (ka)) = J` (r) − N` (r) = 1 . (5.36)
dr dr r→a
A general solution of the Schrödinger equation (5.31), which moreover satisfies the
boundary conditions at the origin and at infinity, is given by



 A J` (kr) r<a
u` (r) = h i (5.37)
B J` (kr)k 2`+1 cotg (δ` (E)) − N` (kr) r>a




where A, and B are normalization constants, which are not independent because of the
boundary conditions (5.32).
From the first boundary condition of formula (5.32) we derive the following equation
h i
A J` (ka) = B J` (ka)k 2`+1 cotg (δ` (E)) − N` (ka) . (5.38)
Whereas, from the second boundary condition of formula (5.32) one obtains

( )
1 h
0 2`+1 0
i
0 λ
−B J` (ka)k cotg (δ` (E)) − N` (ka) + A J` (ka) + A J` (ka) = 0 ,
2µ a
(5.39)
where we wrote J 0 and N 0 for respectively dJ/dr and dN/dr.
By elimination of the normalization constants A and B in equations (5.38) and (5.39),
one obtains
( )
1 λ h i
W (J` (ka), N` (ka)) + J` (ka) J` (ka)k 2`+1 cotg (δ` (E)) − N` (ka) = 0 .
2µ a
Using furthermore the Wronskian relation (5.36), we arrive for the cotangent of the phase
shift at
N` (ka) a
k 2`+1 cotg (δ` (E)) = − 2 ,
J` (ka) λJ` (ka)
or, similarly, in terms of the usual spherical Bessel and Neumann functions

n` (ka) 1
cotg (δ` (E)) = − . (5.40)
j` (ka) λkaj`2 (ka)

65
 The transition matrix elements of the partial wave scattering operator, S` , defined by
(see e.g. [1], formula 11.67)

S` (k) = e2iδ` (E(k)) , (5.41)

are then readily determined by
(2)
cotg (δ` ) + i 1 − iλkaj` (ka)h` (ka)
S` (k) = = (1)
. (5.42)
cotg (δ` ) − i 1 + iλkaj` (ka)h` (ka)
(1) (2)
The spherical Hankel functions, h` and h` , of respectively the first and the second kind,
are defined by
(1) (2)
h` (z) = j` (z) + in` (z) and h` (z) = j` (z) − in` (z) . (5.43)
Notice that the limit λ → ∞ yields the usual expressions for scattering from an in-
finitely hard sphere of radius r = a, i.e.
(2)
n` (ka) h` (ka)
cotg (δ` (E)) = and S` (k) = (1)
, (5.44)
j` (ka) h` (ka)
since indeed the inner region of the delta-shell becomes inaccessible for scattering when
the coupling constant λ grows infinitely large.
For later use we also determine the on-shell matrix elements of the partial wave tran-
sition operator, T` , which is defined in formula (5.19). Here we obtain

λaj`2 (ka)
! !
1 1 1
T` (k) = − = . (5.45)
4π 2 kµ cotg (δ` ) − i 4π 2 µ 1 + iλkaj` (ka)h` (ka)
(1)

The relation of those quantities to the scattering amplitude is given in formula (5.25).
Here we find
∞
f (k, Ω) = −4π 2 µ
X
(2` + 1) T` (k) P` (Ω) . (5.46)
`=0

For the differential cross section, defined in formula (3.24), and the total cross section we
have moreover the relations

2
∞

dσ  2 X 
= |f (k, Ω)|2 = 4π 2 µ  (2` + 1) T` (k) P` (Ω) ,
 
dΩ 
`=0

∞
dσ 2 X
Z 
and σ = dΩ = 4π 4π 2 µ (2` + 1) |T` (k)|2 . (5.47)
dΩ `=0

66
5.4 The T -operator
From the Lippmann-Schwinger equation (4.56) we deduce for the matrix elements of the
T -operator the following relation

Z Z   D
0
d k p~ |V | ~k ~k 0 |G0 (z)| ~k ~k |T (z)| p~ 0
D E D E E
0 0 3 0 3
p~ |T (z)| p~ = p~ |V | p~ + dk ,
(5.48)
or equivalently

Z Z    
0
d k V p~ , ~k ~k 0 , ~k ; z T ~k , p~ 0 ; z
     
0 0 3 0 3
T p~ , p~ ; z = V p~ , p~ + d k G0 ,
(5.49)
which relation can formally be solved by iteration, i.e.

   
T p~ , p~ 0 ; z = V p~ , p~ 0 +
Z Z    
0
d k V p~ , ~k ~k 0 , ~k ; z V ~k , p~ 0
 
3 0 3
+ dk G0 +
Z Z Z Z      
000
+ dk 3 000
dk 3 00
dk 3 0 3
d kV p~ , ~k G0 ~k 000 , ~k 00 ; z 00
V ~k , ~k
0
×
 
~k 0 , ~k ; z V ~k , p~ 0
 
× G0 + ... . (5.50)

The Green’s operator G0 (z), corresponding to the self-adjoint free Hamiltonian H0 =

k 2 /2µ (see formula 5.7), is defined in formula (4.23). Here we find
2µ
     
0
~k 0 (z − H0 )−1  ~k ~k 0 ~k
  
G0 ~k , ~k ; z = = . (5.51)
2µz − k 0 2
By insertion of the Green’s function (5.51) in expression (5.50) and moreover take out
identity operations of the form
Z  
d3 k ~k ~k  ,
ED
1 = (5.52)
we may obtain

2µ
Z
d3 k V p~ , ~k ~k , p~ 0 +
       
0 0
T p~ , p~ ; z = V p~ , p~ + V (5.53)
2µz − k 2
2µ 2µ
Z Z    
0
d3 k V p~ , ~k ~k 0 , ~k ~ ~0 + ...
 
+ d3 k 0 2 V 2 V k ,p .
2µz − k 0 2µz − k

67
5.4.1 The delta-shell potential in momentum space
In formula (5.28) we introduced for the transition potential in configuration space an
expression which contains a radial delta-shell, i.e.
λ
δ(r − a) , (5.54)
2µa
where a represents the radius of the shell and λ the intensity of the potential.
In the language of formula (5.8) for potentials local in configuration space this translates
into a potential of the form
    λ
V ~r , ~r 0 = δ (3) ~r − ~r 0 δ(r 0 − a) . (5.55)
2µa
The momentum space analogon for spherically symmetric configuration space potentials
is defined in formula (5.14), which, when applied to expression (5.55), gives

∞
1 X  Z ∞ λ
 
0  0
V ~k , ~k = 2 (2` + 1) P ` k̂ · k̂ r 2 dr δ(r − a) j` (k 0 r) j` (kr)
2π `=0 0 2µa
∞
λa X  0 
= (2` + 1) P ` k̂ · k̂ j` (k 0 a) j` (ka) . (5.56)
4π 2 µ `=0

5.4.2 The T -matrix elements for the delta-shell

In formula (5.56) we found the momentum space potential V for the delta-shell config-
uration space potential (5.54). By substituting the righthand side of that equation in
formula (5.53), also using (5.13) in order to compactify the resulting expression, we may
obtain the T -matrix for that case, i.e.

    
T p~ , p~ 0 ; z p + p~ 0  a +
= λ̄ j0 −~

2
Z   2µ  
p + ~k  a
 ~
 
0
+ λ̄ d3 k j0 −~ j − k + p
~ a
 
0
2µz − k 2


2µ
Z Z      
3 0  0
d3 k 0 p + ~k  a
d3 k j0 −~ ~ ~
  
+ λ̄ j 0 
 − k + k  a ×
2µz − k 0 2


2µ  ~
  
0
× j 0 − k + p
~ a
2µz − k 2


+ ... , (5.57)

where λ̄ stand for

λa
λ̄ = . (5.58)
4π 2 µ

68
The Born term
The first term of the righthand side of formula (5.57) can be compared to the first term
of the on-shell expression (5.45) after expanding in λ. For the latter expansion we find

λaj`2 (pa)
  2 
(1) (1)
T` (p) = 1 − iλpaj` (pa)h` (pa) + iλpaj` (pa)h` (pa) + ... . (5.59)
4π 2 µ

Using formula (5.13) we obtain for the first term of the righthand side of formula (5.57)
the following
∞
  
0 λa X
λ̄ j0 −~
p + p~ a = (2` + 1) P` (p̂ · p̂ 0 ) j` (pa) j` (p0 a) . (5.60)

2
4π µ `=0
Hence, for the Born term of T` we find

(1) λa
T` (p, p0 ) = j` (pa) j` (p0 a) . (5.61)
4π 2 µ
On-shell, i.e. p0 = p, expression (5.61) is equal to the first term of (5.59).

The second order term

In order to perform the integration of the second term of the righthand side of formula
(5.57), we start by introducing the expansion (5.13), to obtain

Z   2µ  
p + ~k  a
 ~
 
0
d3 k j0 −~ j − k + p
~ a = (5.62)
 
0
2µz − k 2


∞
j` (ka) j`0 (ka)
Z    
(2` + 1)(2`0 + 1) j` (pa) j`0 (p0 a) d3 k P` p̂ · k̂ P`0 k̂ · p̂ 0
X
= 2µ .
`,`0 =0 2µz − k 2

The details of the ~k integration are studied in subsection (5.4.3). Using the definition
(5.80), we may write here

2µ
Z    
p + ~k  a
 ~
 
0
d3 k j0 −~ j − k + p
~ a = (5.63)
 
0
2µz − k 2


∞  
(2` + 1)(2`0 + 1) j` (pa) j`0 (p0 a) I` p~ , p~ 0 ; µ ; 1
X
= ,
`,`0 =0

which, by substitution of the result (5.87), gives

2µ
Z   
p + ~k  a  ~
  
d3 k j0 −~ j 0 − k + p
~ 0
a = (5.64)
2µz − k 2


∞
4π 2 µp
" #
(1)
(2` + 1)(2`0 + 1) j` (pa) j`0 (p0 a) −i δ`,`0 P` (p̂ · p̂ 0 ) j` (pa) h` (pa)
X
= .
`,`0 =0
2` + 1

69
Putting everything together amounts for the second term on the righthand side of formula
(5.57) to

2µ
Z    
p + ~k  a
 ~
 
2 0
λ̄ d3 k j0 −~ j − k + p
~ a =
 
0
2µz − k 2


∞
iλ2 pa2 X (1)
= − 2 (2` + 1) j`2 (pa) h` (pa) j` (p0 a) P` (p̂ · p̂ 0 ) , (5.65)
4π µ `=0

or equivalently, for the second order in λ term of T` ,

(2) iλ2 pa2 2 (1)

T` (p, p0 ) = − 2 j` (pa) h` (pa) j` (p0 a) . (5.66)
4π µ
This term may be compared to the second term of the on-shell expansion (5.59). We
find again perfect agreement.

The third order term

For the integrations of the third term of the righthand side of formula (5.57), we may, of
0
course, use the results of the previous subsection. We start then with the ~k -integration,
i.e.
2µ
Z      
0 0
3 0
+ ~k ~ ~
   
d k j0 −~
p a j 0 − k + k  a
  , (5.67)

2µz − k 2
for which we recognize the expression (5.62) after the substitution p~ 0 by ~k. Accordingly,
using formula (5.66), we write

2µ
Z      
0 0
3 0
+ ~k ~ ~
   
dk j0 −~
p a j 0 − k + k  a =
 
2µz − k 0 2


∞  
(1)
= −4iπ 2 pµ (2` + 1) j`2 (pa) h` (pa) j` (ka) P` p̂ · k̂
X
. (5.68)
`=0

Hence, we obtain for the third term of the righthand side of formula (5.57) the following

2µ 2µ
Z Z     
0  0  
~ ~ ~  ~

3 0 3
   0
dk dk j0 −~
p + k a j 0
− k + k  a j 0 − k + p
~ a
2µz − k 0 2 2µz − k 2
  

∞
2µ
Z  
 ~
  
(1)
= −4iπ 2 pµ (2` + 1)j`2 (pa)h` (pa) d3 kj` (ka)P` p̂ · k̂ 0
X
j 0 − k+ p
~ a
2µz − k 2

`=0

(5.69)

Here we introduce once more expansion (5.13) and moreover defintion (5.80), to obtain
for the ingration piece of formula (5.69), the result

70
 2µ
Z  
 ~
  
d3 k j` (ka) P` p̂ · k̂ j 0 − k + p
~ 0
a = (5.70)
2µz − k 2


∞
Z
3
  2µ  
(2`0 + 1) P`0 k̂ · p̂ 0 j`0 (ka) j`0 (p0 a)
X
= d k j` (ka) P` p̂ · k̂ 2
2µz − k `0 =0
∞  
(2`0 + 1) j`0 (p0 a) I` p~ , p~ 0 ; µ ; 1
X
= .
`0 =0

The final integration is given in formula (5.87). So, we find for the third term of the
righthand side of formula (5.57) the following expression

2µ 2µ
Z Z      
0 0  
p + ~k  a ~ ~  ~

0
λ̄3 d3 k 0 d3 k j0 −~
   
2 j 0
 − k + k a
2 j 0 − k+ p
~ a
2µz − k 0 2µz − k
  

∞
λ3 p 2 a 3 X (1) 2
= − 2 (2` + 1) j`3 (pa) h` (pa) j` (p0 a) P` (p̂ · p̂ 0 ) . (5.71)
4π µ `=0

or equivalently, for the third order in λ term of T` ,

(3) λ3 p 2 a 3 3 (1) 2
T` (p, p0 ) = − 2 j ` (pa) h ` (pa) j` (p0 a) . (5.72)
4π µ
This term may be compared to the third term of the on-shell expansion (5.59). We find
once more perfect agreement.

The (n + 1)-th order term

By comparing the previous results given in formulas (5.60), (5.65) and (5.71), we may
expect for the (n + 1)-th term of expansion (5.57) the following expression

 
T (n+1) p~ , p~ 0 = (5.73)
∞ in
(1)
h
= λ̄ (−iλpa)n (2` + 1) P` (p̂ · p̂ 0 ) j` (pa) j` (p0 a) j` (pa) h` (pa)
X
,
`=0

where λ̄ is as defined in formula (5.58).

This can easily be verified as we will show below. From expansion (5.57) we observe
for the relation of two subsequent terms the following identity

2µ
Z  
d3 k T (n) p~ , ~k
 ~
    
T (n+1) p~ , p~ 0 = λ̄ j 0 − k + p
~ 0
a . (5.74)
2µz − k 2


When we substitute here expression (5.73) and formula (5.13), also using definition (5.58),
then we find

71
   Z
T (n+1) p~ , p~ 0 = λ̄ d3 k λ̄ (−iλpa)n−1 × (5.75)

∞ in−1
(1)
X   h
× (2` + 1) P` p̂ · k̂ j` (pa) j` (ka) j` (pa) h` (pa) ×
`=0

∞
2µ  
(2`0 + 1) P`0 k̂ · p̂ 0 j`0 (ka) j`0 (p0 a) .
X
× 2
2µz − k `0 =0

Using furthermore the result (5.87), we obtain

   
T (n+1) p~ , p~ 0 = λ̄2 (−iλpa)n−1 −i 4π 2 µp × (5.76)
∞ in−1
(1) (1)
h
(2` + 1) P` (p̂ · p̂ 0 ) j` (pa) j` (p0 a) j` (pa) h` (pa)
X
× j` (pa) h` (pa) ,
`=0

which, by substitution of the definition (5.58), gives indeed expression (5.73).

The full matrix elements

From the above results one can read off the expression for the full sum of all terms of
formula (5.57) for the matrix elements of the T -operator, i.e.

  ∞  
T p~ , p~ 0 T (n) p~ , p~ 0
X
= (5.77)
n=1

∞ ∞ in
(1)
h
0 0
(−iλpa)n j` (pa) h` (pa)
X X
= λ̄ (2` + 1) P` (p̂ · p̂ ) j` (pa) j` (p a)
`=0 n=0

∞
λa X 0 j` (pa) j` (p0 a)
= (2` + 1) P ` (p̂ · p̂ ) .
4π 2 µ `=0 (1)
1 + iλpaj` (pa)h` (pa)
Notice that the above expression (5.77) agrees with formula (5.59) and also that it is
of the form (5.19). We may thus arrive at the scattering amplitude by substitution of the
on-shell (i.e. p0 = p) matrix elements in formula (5.25), which gives
∞
X j`2 (pa)
f (~p, r̂) = −λa (2` + 1) P` (p̂ · r̂)
(1)
. (5.78)
`=0 1 + iλpaj` (pa)h` (pa)
It is also interesting to notice that this way we may obtain an expression for the T -
matrix elements of the infinitely hard sphere, by taking the limit λ → ∞ in formula
(5.77), i.e.
∞
  i j` (p0 a)
T p~ , p~ 0 (2` + 1) P` (p̂ · p̂ 0 )
X
= − . (5.79)
4π 2 µp `=0
(1)
h` (pa)
A direct calculation of those matrix elements might be impossible, since expression (5.14)
becomes infinite in the limit of V (r) → ∞ for r < a.

72
5.4.3 Details of the ~
k integration
Let us study the following integration in momentum space

 
I` p~ , p~ 0 ; µ ; f` = (5.80)
Z     j` (ka) j`0 (ka)  2 
= 2µ d3 k P` p̂ · k̂ P`0 k̂ · p̂ 0 f` k
2µz − k 2
∞ j` (ka) j`0 (ka)  2 
Z     Z
= 2µ dΩk P` p̂ · k̂ P`0 k̂ · p̂ 0 k 2 dk f` k ,
0 2µz − k 2
where f` represents an arbitrary well-behaved function of linear momentum squared.
For the integration over the angles we may use formula (5.24), i.e.
4π
Z    
dΩk P` p̂ · k̂ P` (p̂ · p̂ 0 ) δ`,`0 .
P`0 k̂ · p̂ 0 = (5.81)
2` + 1
Hence, we must concentrate on the radial integration, i.e.
∞ j`2 (ka)
Z  
k 2 dk f ` k
2
. (5.82)
0 2µz − k 2
Integrals of this type appear (for example at page 46 of the Ph.D. thesis of George Rupp
[3]) in perturbative calculus for the meson-meson scattering model which we will discuss
next. The tric amounts to the following identity
(1)
Z ∞ j`2 (ka)   1 Z ∞ 2 j` (ka) h` (ka)  2 
k 2 dk f ` k 2
= k dk f` k , (5.83)
0 2µz − k 2 2 −∞ 2µz − k 2
by using the following properties for the spherical Bessel and Hankel functions (the Hankel
functions are defined in formula 5.43)

j` eπi ka = eπi` j` (ka) and eπi ka = e−πi` h` (ka) .

(1) (2)
   
h` (5.84)

(1)
For large imaginary part of the argument ka the function h` (ka) tends to zero. Therefore
one can close a complex contour by joining to the real k-axis a semicircle in the complex
upper half plane and determine the integration for the closed contour in the complex
k-plane. The part of the semicircle does not contribute.
Finally, we define

2µz = (p + i)2 , (5.85)

and take the limit  ↓ 0 after integration. The integral (5.83) turns into a contour integral
for which we may apply Cauchy’s residue theorem as follows

∞ j`2 (ka)
Z  
k 2 dk f ` k
2
=
0 2µz − k 2

73
 (1)
1 j` (ka) h` (ka)
I  
= lim k 2 dk f` k 2
↓0 2 (p + i − k)(p + i + k)

1 −p2 (1)
  iπp (1)
 
= 2πi j` (pa) h` (pa) f` p2 = − j` (pa) h` (pa) f` p2 . (5.86)
2 2p 2

Putting everything together, we find for the integration of formula (5.80) the following
result


0
 4π 2 µp (1)
 
I` p~ , p~ ; µ ; f` = −i δ`,`0 P` (p̂ · p̂ 0 ) j` (pa) h` (pa) f` p2 . (5.87)
2` + 1

74
5.5 Meson-meson scattering
For a second example we study a model for the scattering of interacting composite particles
(see for example [4]).

5.5.1 Introduction
Imagine in the far past a pair of composite particles heading towards the interaction
region, which, evidently, is also the centre of mass of this system. In the far future we
imagine a number of such objects flying away from the interaction region.
For the structure of our composite particles, which we will refer to as mesons in the
following, we have the following picture: Their elementary constituents we will call valence
quarks, denoted by q, and valence antiquarks, denoted by q̄. In this report we do not
consider different flavors, neither color, nor spin. Each meson consists of one valence
quark and one valence antiquark confined by a harmonic oscillator force. Equivalently,
one may represent a meson by a spring which at one endpoint is connected to a valence
quark and at the other endpoint to a valence antiquark.
Mesons are here assumed to interact by the phenomenon of quark pair creation, which
is the process in which one valence quark and one valence antiquark are created from the
vacuum. In that case the spring breaks up in order to form two mesons. In the opposite
process, where one valence quark of one meson annihilates with one valence antiquark of
the other meson, two springs fuse in order to form one meson. Other processes will not
be considered here.
Moreover, here we will restrict ourselves to final states with two mesons, i.e.

A + B −→ C + D . (5.88)
Consequently, we assume that in the interaction region two mesons may fuse to one meson,
or break up into two mesons. But, due to the restriction ( 5.88) one has always at most
two and minimally one meson in the interaction region. The latter situation can only exist
around the origin, which represents the centre of mass of the system. We will describe
this by the one-particle harmonic oscillator, which is equivalent to the relative motion of
one valence quark and one valence antiquark confined by a spring.
For the two-meson situation we assume that the quantum numbers are the only rele-
vant parameters of each of the two mesons, which hence can then be described by point
particles.
The only interactions we allow for, are spring fusion when there are two mesons in
the interaction region, and spring fission into two mesons when there is one meson in
the interaction region. The probability of those processes to occur will be given by the
distributions of the end points of the springs, or, equivalently, by the quark and the
antiquark distributions.
The wave function for such a system must describe the probability that we find the
system in the one-meson state, as well as the probability that we find the system in
the two-meson state. This can conveniently be done by defining a two-component wave
function, given by
 
ψc
  , (5.89)
ψf

75
where |ψc |2 represents the probability to find the system in the one-meson state and |ψf |2
the probability to find the system in the two-meson state. Evidently, because of the
above described assumptions and, in particular, because of assumption ( 5.88), the two
components of ( 5.89) must satisfy here the following condition:

|ψc |2 + |ψf |2 = 1 . (5.90)

5.5.2 The wave equation for the model

We assume that the system ( 5.89) can be described by a stationary 2 × 2 non-relativistic
wave equation of the following form (in units c = 1)
     
Hc U (~r ) ψc (~r ) ψc (~r )
    = E   . (5.91)
U (~r ) Hf ψf (~r ) ψf (~r )
The dynamics of the one-meson state is contained in Hc , whereas Hf describes the
dynamics of the two-meson state. The potential U describes the transitions from one
mode of the model to the other mode.
In the absence of spring fission or fusion processes, the transition potential U vanishes.
In that case the two channels decouple and the system is either described by a stationary
meson at the origin, or by a system of two non-interacting mesons. Hence, since the one-
meson state is assumed to be described by a valence pair of quark and antiquark confined
by a harmonic oscillator force, Hc must be of the following form:

∇2
Hc = − + mq + mq̄ + 1
µ ω 2 r2
2 c
, (5.92)
2µc
where mq and mq̄ represent the masses of respectively the quark and the antiquark,
where ω represents the oscillator frequency and where µc stands for the reduced mass of
the quark+antiquark system. Moreover, since the two-meson state describes a system of
two non-interacting mesons, Hf is just the free particle Hamiltonian, according to the
relation

∇2
Hf = − + M1 + M2 , (5.93)
2µf
where M1 and M2 represent the meson masses and µf the reduced mass of the two-meson
system.
Interaction is achieved by considering a non-zero transition potential U . We will assume
that such potential vanishes at the origin and at large distances and has some maximum at
an intermediate distance which is of the order of the size of a meson. A transition potential
U which has such characteristics can conveniently be approximated by a delta-shell, i.e.
1
U (~r ) = λV (r) where V (r) = δ (r − a) . (5.94)
2µc a
Here we will study the behavior of the solutions of wave equation ( 5.91) in the ap-
proximation ( 5.94) for the transition potential.

76
5.5.3 The radial wave equation
The 2 × 2 Hamiltonian operator of equation ( 5.91) is rotational invariant, as can be
concluded from the expressions ( 5.92), ( 5.93) and ( 5.94). Consequently, it is useful to
introduce angular momentum, L,~ and write

uc (r) (`)
 
 
ψc (~r )
 = 
 r Y`z (ϑ, ϕ) 
 .
 (5.95)
ψf (~r ) u (r)
 
f (`)
r Y`z (ϑ, ϕ)
Substitution of equation ( 5.95) into the wave equation ( 5.91), also using the relations
( 5.92), ( 5.93) and ( 5.94), leads to the following system of coupled differential equations
for the radial parts of wave function
     
hc λV (r) uc uc
    = E   , (5.96)
λV (r) hf uf uf
where

d2
!
1 ` (` + 1)
hc = − 2 + + mq + mq̄ + 1
µ ω 2 r2
2 c
,
2µc dr r2

d2
!
1 ` (` + 1)
hf = − 2 + + M1 + M2
2µf dr r2

and
1
V = δ (r − a) . (5.97)
2µc a

77
5.5.4 Coupled channel wave equation with one delta shell
In this section we study the solutions of the 2 × 2 radial wave equation with one delta
shell which is defined in formula ( 5.97). For r < a and for r > a one has to solve the two
uncoupled differential equations, equivalent to λ = 0, given by
     
hc 0 uc uc
    = E   . (5.98)
0 hf uf uf
At r = a one may obtain the following boundary conditions:
 !
d uc (r)  d uc (r) 
 
1 + 2µλ a uf (a) = 0
2µc − dr r↓a +


dr r↑a




 c

, (5.99)

 !
d uf (r)  d uf (r) 
  

λ 1 −
2µc a uc (a) + 2µf + = 0


dr  dr 


r↓a r↑a

by integrating wave equation ( 5.97) at an infinitesimal interval including the point r = a,

i.e.
     
lim
Z a+  hc λV uc uc 
↓0 dr     = E   ,
a− 
λV hf uf uf 

and

 uc (r ↑ a) = uc (r ↓ a)
, (5.100)

uf (r ↑ a) = uf (r ↓ a)
by integrating equation ( 5.99) also at an infinitesimal interval which includes the point
r = a.
Further boundary conditions are the usual: The wave functions uc and uf both vanish
at the origin. At infinity the wave function uc must be damped exponentially, since uc
describes a confined system, whereas, above threshold (E > M1 +M2 ), the wave function
uf must have oscillating behavior, since uf describes the scattered mesons.
Let us denote by Fc and Gc the solutions of the upper differential equation of formula
( 5.98) which respectively vanish at the origin and damp exponentially at infinity. Below
we give a precise definition of those functions.
Let us first recall that the differential equation

d2
" ! #
1 ` (` + 1)
− 2 + + m1 + m2 + 21 µω 2 r 2 u = E u , (5.101)
2µ dr r2
is solved by the following two linearly independent functions

1 (`+1)/2 1 2
e− 2 µωr φ −ν; ` + 32 ; µωr 2
  
F (r) = 
3
 µωr 2
Γ `+ 2

and (5.102)
`/2 1 2
e− 2 µωr ψ −ν; ` + 32 ; µωr 2
  
G(r) = − 12 Γ (−ν) r µωr 2 ,

78
where the radial quantum number ν is defined by the following relation
 
3
E = ω 2ν + ` + 2
+ m1 + m2 , (5.103)
and where the functions φ and ψ are defined by (see [5] pages 248 and 257)

∞
X (a)n z n
φ(a; b; z) = with (a)0 = 1 and (a)n+1 = (a + n)(a)n
n=0 (b)n n!

and (5.104)

Γ(1 − b) Γ(b − 1) 1−b

ψ(a; b; z) = φ(a; b; z) + z φ(a − b + 1; 2 − b; z) .
Γ(a − b + 1) Γ(a)

The solution F (r) is regular at the origin, but is irregular at infinity, except for some
special cases known as the harmonic oscillator spectrum, whereas G(r) behaves regular
at infinity but not in the origin.
For the solutions ( 5.102) one has the following Wronskian relation
" #
d Gc (r) d Fc (r)
W (Fc (a), Gc (a)) = Fc (r) − Gc (r) = 1 (5.105)
dr dr r→a

The lower differential equation of formula ( 5.98) is solved by spherical Bessel (J) and
Neumann (N ) functions, discussed in formulas (5.34ff).
A general solution of the Schrödinger equation ( 5.97), which moreover satisfies the
boundary conditions at the origin and at infinity, is given by

  

 Fc (r) Ac
r<a

  


  
 J` (kr) Af
uc (r)



  = (5.106)
uf (r)
 
Gc (r) Bc





r>a

  
  h i 
J` (kr) k 2`+1 cotg (δ` (E)) − N` (kr) Bf




where Ac , Af , Bc and Bf are normalization constants, which are not independent because
of the boundary conditions ( 5.99) and ( 5.100) and where k is defined by

k 2 = 2µf (E − M1 − M2 ) . (5.107)
From the boundary conditions ( 5.100) we derive the following equations

Fc (a) Ac = Gc (a) Bc ,
h i
J` (ka) Af = J` (ka) k 2`+1 cotg (δ` (E)) − N` (ka) Bf . (5.108)

Whereas, from boundary conditions ( 5.99) one obtains

79
 1 λ
(G0c (a)Bc − Fc0 (a)Ac ) = J` (ka) Af ,
2µc 2µc a
1 h 0 i  λ
J` (ka) k 2`+1 cotg (δ` (E)) − N`0 (ka) Bf − J`0 (ka) Af = Fc (a)Ac .
2µf 2µc a

(5.109)
When we multiply the first (respectively second) line of equation ( 5.109) by Fc (a) (re-
spectively J` (ka)) and then substitute according to the relation in the first (respectively
second) line of equation ( 5.108), then we obtain

Bc λ
W (Fc (a), Gc (a)) = J` (ka) Fc (a) Af ,
2µc 2µc a
Bf λ
− W (J` (ka) , N` (ka)) = J` (ka) Fc (a) Ac .
2µf 2µc a
Using furthermore the Wronskian relations ( 5.36) and ( 5.105), we find

Bc λ Bf λ
= J` (ka) Fc (a) Af and − = J` (ka) Fc (a) Ac . (5.110)
2µc 2µc a 2µf 2µc a
The relations ( 5.110) can be substituted into equations ( 5.108), after which follows easily
for the cotangent of the phase shift the expression:
" #−1
2`+1 N` (ka) µf 1 2
k cotg (δ` (E)) = + −λ2 J (ka) Fc (a)Gc (a) . (5.111)
J` (ka) µc a 2 `
Using moreover the relations ( 5.34) and ( 5.102) for the various functions which are
introduced in solution ( 5.106) and substitute furthermore the function A defined by
   
1 φ −ν, ` + 32 ; z ψ −ν, ` + 32 ; z
A (ν, `; z) = z ` + 2 e−z 
3
 , (5.112)
Γ `+ 2
we arrive at
" #−1
n` (ka) 1 2 µf
 
cotg (δ` (E)) = + 2
λ ka j`2 (ka) Γ(−ν)A ν, `; µc ωa 2
. (5.113)
j` (ka) µc
The scattering matrix follows by

cotg (δ` (E)) + i

S` (E) = e2iδ` (E) = (5.114)
cotg (δ` (E)) − i
µ (2)
1 + 1
2
iλ2 µf Γ(−ν)A (ν, `; µc ωa2 ) kaj` (ka) h` (ka)
c
= µ (1)
,
1 − 1
2
iλ2 µfc Γ(−ν)A (ν, `; µc ωa2 ) kaj` (ka) h` (ka)
where the spherical Hankel functions of the first and the second kind are defined in formula
(5.43).

80
5.5.5 Bound states and resonances
Bound states and resonances, which are described by singularities in the scattering matrix
( 5.114), can be computed from the condition

cotg (δ` (E)) = i , (5.115)

which in this case reads
µf  
(1)
1
2
iλ2 Γ(−ν)A ν, `; µcωa2 kaj` (ka) h` (ka) = 1 . (5.116)
µc

Approximate bound states and resonance positions

For small values of λ2 the solutions of equation ( 5.116) can be found approximately. The
reason is that then the bound states and resonances are close to the harmonic oscillator
eigenstates. Hence, the values of the radial quantum number, ν, are almost non-negative
integers, ν0 . We write

ν = ν0 + ∆ν . (5.117)
Now since A and the spherical Bessel function in ( 5.116) are slowly varying functions
of the energy, E, or, equivalently, the radial quantum number ν, the only rapid varying
candidate that is left, is

(−1)ν0 +1
Γ(−ν) ≈ . (5.118)
(ν0 )!∆ν
When we moreover substitute the following real and positive quantity

  (−1)ν0 µf  
B ν0 , `; µcωa2 = A ν0 , `; µc ωa2 (5.119)
(ν0 )! µc
    2
3
µf 1 Γ ` + ν0 +
2 φ −ν0 , ` + 32 ; µc ωa2
(µc ωa2 )` + e−µc ωa
2
= 2     ,
µc (ν0 )! Γ `+ 3
2

we arrive for the shift in the resonance energy at

(1)
 
∆E = 2ω∆ν ≈ −iλ2 ωB ν0 , `; µc ωa2 k0 a j` (k0 a) h` (k0 a) , (5.120)
where, following formulas (5.103) and (5.107), k0 is defined by
h   i
k02 = 2µf ω 2ν0 + ` + 3
2
+ mq + mq̄ − M1 − M2 . (5.121)
For further studies we may distinguish two different cases:
(1) E0 > M1 + M2 , and
(2) E0 < M1 + M2 .

81
 (1) Above threshold
When the nearby harmonic oscillator state has energy, E0 , which exceeds the threshold
energy, M1 +M2 (i.e. the sum of the rest masses of the scattered particles), then expression
( 5.120) has a negative imaginary part and a real part, according to
  n o
∆E ≈ λ2 ωB ν0 , `; µc ωa2 k0 a −i j`2 (k0 a) + j` (k0 a) n` (k0 a) . (5.122)
The resonance singularity of the scattering matrix which corresponds to this situation
is depicted in figure ( 5.1).

=m(E)
6
- <e(E)
threshold E0 cut
•
E = M 1 + M2 
 ∆E


• resonance position

Figure 5.1: The position of a resonance singularity in the complex energy plane with
respect to the harmonic oscillator state at the real energy axis.

Notice that the resonance singularity is in the lower half complex energy plane (second
Riemann sheet) as it should be.

82
 (2) Below threshold
When the nearby harmonic oscillator state has energy, E0 , lower than the threshold
energy, then expression ( 5.120) has only a real part. The reason is, that in that case
(k0 )2 is negative and hence k0 imaginary, i.e.

k0 = iκ0 (κ0 real) . (5.123)

Moreover, κ0 > 0, because, since it describes a system of bounded scattering particles,
the wavefunction ψf of the scattering channel must be damped exponentially, not explode.
Now in formula (48) of section 7.2.7 from [5], we find the following expansion:

  m
(1) π i

 
1 2
` X − 14 z 2 Γ(2` + 2m + 2)
izj` (z)h` (z) = z z i2 +
2
2 4 h 
3
m m! Γ ` + m + 2
Γ(2` + m + 2)
 m 
− 14 z 2 Γ(2m + 1) 
+ (−1)`
X
    , (5.124)
m [m!]2 Γ ` + m + 3
2
Γ −` + m + 1
2


which for z = iκ0 a is obviously real.

The bound state singularity of the scattering matrix which corresponds to this situation
is depicted in figure ( 5.2).

=m(E)
6
- <e(E)
bound state position E0 threshold cut
• •
∆E E = M 1 + M2

Figure 5.2: The position of a bound state singularity in the complex energy plane with
respect to the harmonic oscillator state at the real energy axis.

Notice that the bound state singularity is at the real axis of the complex energy plane,
as it should be.

83
5.6 The T -matrix for the meson-meson model
In configuration space we may write the non-relativistic 2 × 2 stationary matrix wave
equation (5.91), also using relations (5.92), (5.93) and (5.94), in the following form

∇2
!
λ
− r + mq + mq̄ + 1
µ ω 2 r2
2 c
− E ψc (~r ) = − δ (r − a) ψf (~r ) ,
2µc 2µc a

∇2
!
λ
− r + M1 + M2 − E ψf (~r ) = − δ (r − a) ψc (~r ) (5.125)
.
2µf 2µc a

In order to compactify the formulas, let us define the following operators: Hc , which
describes the confinement dynamics in the interaction region, Hf , which describes the dy-
namics of the scattered particles at large distances, and Vt , which describes the transition
interaction. In configuration space we define those operators by

∇2r
Hc = − + mq + mq̄ + 1
µ ω 2r2
2 c
,
2µc

∇2r
Hf = − + M1 + M2 , and
2µf

λ
Vt = δ (r − a) . (5.126)
2µc a
Hence, we obtain for the wave equation

(E − Hc ) ψc (~r ) = Vt ψf (~r ) and (E − Hf ) ψf (~r ) = [Vt ]T ψc (~r ) . (5.127)

In the equations (5.127) we must eliminate ψc , since it is vanishing at large distances

and thus unobservable. Formally, this can easily be done. We obtain then the following
relation

n o
ψf (~r ) = (E − Hf )−1 [Vt ]T ψc (~r )

= (E − Hf )−1 [Vt ]T (E − Hc )−1 Vt ψf (~r ) . (5.128)

By comparison of equation (5.128) with the usual expressions for the scattering wave
equations, we must conclude that the generalized potential, V , is here given by

V = [Vt ]T (E − Hc )−1 Vt . (5.129)

84
The Born term
In the momentum representation formula (5.129) looks like
 E  E
p | V p~ 0
h~ p | [Vt ]T (E − Hc )−1 Vt p~ 0
= h~ . (5.130)
 

The properly normalized eigensolutions of the operator Hc (5.126) corresponding to the

energy eigenvalue En` are in configuration space given by

3 √
h~r |n`m i = (µc ω) 4 Nn` Ym(`) (r̂) Fn` ( µc ω r) , where (5.131)
v
u  
3 
u 2Γ n+`+ 2  3
u
1 2
 
and Fn` (z) = z ` e− 2 z 1 F1 −n ; ` + ; z 2

Nn` = u i2 ,
2
t h 
 n! Γ ` + 3 

2

3
 
En` = ω 2n + ` + + mq + mq̄ ,
2
and where n = 0, 1, 2, . . .; ` = 0, 1, 2, . . .; m = −`, . . ., +`.

So, by letting the self-adjoint operator Hc act to the left, we may write

 E  E
h~p | V p~ 0 p | [Vt ]T |n`mi hn`m| (E − Hc )−1 Vt p~ 0
X
= h~
 
n`m

|n`mi hn`m|  E
p | [Vt ]T Vt p~ 0 .
X
= h~ (5.132)

n`m E − En`

Next, we insert several times unity to obtain

 E X Z Z Z Z
h~p | V p~ 0 = d3 r d3 r 00 d3 r 000 d3 r 0 (5.133)

n`m

1  ED ED  ED   ED  E
p |~r ih~r | [Vt ]T ~r 00 ~r 00 |n`m n`m ~r 000 ~r 000  Vt ~r 0 ~r 0 p~ 0 .
h~
E − En`
Two of the four integrations are trivial, since, as we have seen in section (5.1.1), the
non-local equivalent of the local transition potential takes the form
 E λ  
h~r | Vt ~r 0 = δ (r − a) δ (3) ~r − ~r 0 . (5.134)

2µc a
By inserting expression (5.134) into formula (5.133) and also substituting
p · ~r
ei~
h~r |~
pi = , (5.135)
(2π)3/2
we find

85
  E
 0 X Z Z
3
h~
p| V p
~ = dr d3 r 0 (5.136)
n`m

1 e−i~
p · ~r λ D  E λ e p 0 · ~r 0
i~
 0
δ (r − a) h~r |n`m i n`m ~r δ (r 0 − a) .
E − En` (2π)3/2 2µc a 2µc a (2π)3/2

Next, we observe that the radial parts of the two remaining integrations are also trivial,
because of the two delta-functions. We insert thus two times the expression given in
formula (5.131) for the harmonic oscillator eigenfunctions, to obtain

!2
(µc ω)3/2 λ
 E Z Z
 0 2
a2 dΩ a2 dΩ0
X
h~
p| V p
~ = Nn` (5.137)
(2π)3 2µc a n`m

1 p · ar̂ Y (`) (r̂) F (√µ ω a) Y (`) ∗ (r̂ 0 ) F ∗ (√µ ω a) ei~

e−i~ p 0 · ar̂ 0 .
m n` c m n` c
E − En`
For the integrations over the angles we introduce Bauer’s formula [7],

~
e−ik · ~r =
∗  
4π(−i)λ jλ (kr)Yµ(λ) (r̂) Yµ(λ) k̂
X
, (5.138)
λ,µ

and its complex conjugate relation

~
eik · ~r =
∗ 
4π(i)λ jλ (kr)Yµ(λ) (r̂) Yµ(λ) k̂
X
, (5.139)
λ,µ

in order to obtain
Z
dΩ e−i~
p · ar̂ Y (`) (r̂) = 4π(−i)` j (pa) Y (`) (p̂)
m ` m (5.140)

and

p 0 · ar̂ 0 Y (`) ∗ (r̂ 0 ) = 4π(i)` j (p0 a) Y (`) ∗ (p̂ 0 )

Z
dΩ0 ei~ m ` m (5.141)

Substitution of relations (5.140) and (5.141) into formula (5.137) gives us the following
expression

λ2 a 2 ω
s
 E
 0 ω X 2
h~
p| V p
~ = N (5.142)
2π µc n`m n`

1 ∗ √
Ym(`) (p̂) Ym(`) (p̂ 0 ) j` (pa) j` (p0 a) |Fn` ( µc ω a)|2 .
E − En`
The summation over the magnetic quantum number m can be performed by the use of
the addition theorem (see formula 5.22), to give

86
  E
p | V p~ 0
h~ = (5.143)


λ2 a 2 ω
s
ω X 2 1 2` + 1 √
= Nn` P` (p̂ · p̂ 0 ) j` (pa) j` (p0 a) |Fn` ( µc ω a)|2
2π µc n` E − En` 4π
 √ 2
∞ ∞ N 2 F µ ω a
λ2 a 2 ω
s
ω X

0 0
X n` n` c 
= (2` + 1) P` (p̂ · p̂ ) j` (pa) j` (p a) .
8π 2 µc `=0 n=0 E − En`

For the summation over the radial quantum number n the result may be looked up in [8]
at page 227. Using the definition of A which is given formula (5.112), one finds

 E
p | V p~ 0
h~ = (5.144)


∞
λ2 a 2 ω −Γ(−ν)A (ν, `; µc ωa2 )
s
ω X
= (2` + 1) P` (p̂ · p̂ 0 ) j` (pa) j` (p0 a) √ ,
8π 2 µc `=0 ω µc ωa2

where, following formulas (5.103) and (5.107), ν is related to p according to

h   i
p2 = 2µf ω 2ν + ` + 3
2
+ mq + mq̄ − M1 − M2 . (5.145)
The Born term (5.144) may be compared to the first order term of the result obtained
in formula (5.113), which by means of formulas (5.19), (5.25) and (5.26), leads to

1 1
T` (p) = − 2 (5.146)
4π µf p cotg (δ` (p)) − i
1 2 µf
1 2
λ µc pa j`2 (pa) Γ(−ν)A (ν, `; µc ωa2 )
= − 2
4π µf p 1 − i 1 λ2 µf pa j` (pa) h(1) 2
` (pa) Γ(−ν)A (ν, `; µc ωa )
2 µc

1 λ2 a j`2 (pa) Γ(−ν)A (ν, `; µc ωa2 )

= − 2
8π µc 1 − i 1 λ2 µf pa j` (pa) h(1) 2
` (pa) Γ(−ν)A (ν, `; µc ωa )
2 µ c

λ2 a j`2 (pa) Γ(−ν)A (ν, `; µcωa2 )

= − ×
8π 2 µc
(
µf (1)
 
× 1 + i 12 λ2 pa j` (pa) h` (pa) Γ(−ν)A ν, `; µcωa2 +
µc
!2 
µf (1)
  
+ i 21 λ2 pa j` (pa) h` (pa) Γ(−ν)A ν, `; µc ωa2 + ...  .
µc

We find perfect agreement.

87
The full matrix elements
For the higher order terms one needs paper and some patience. However, no further
difficulties will emerge than already exposed here in this chapter. One obtains then

 
T p~ , p~ 0 = (5.147)
∞
λ2 a X 0 j` (pa) j` (p0 a) Γ(−ν)A (ν, `; µc ωa2 )
= − 2 (2` + 1) P` (p̂ · p̂ ) µ (1)
8π µc `=0 1 − i 12 λ2 µfc pa j` (pa) h` (pa) Γ(−ν)A (ν, `; µc ωa2 )

Notice that the above expression (5.147) is indeed of the form (5.19). We may thus
arrive at the scattering amplitude by substitution of the on-shell (i.e. p0 = p) matrix
elements in formula (5.25) by

λ2 a j`2 (pa) Γ(−ν)A (ν, `; µc ωa2 )

!
T` (p) = − µ , (5.148)
8π 2 µc (1)
1 − i 21 λ2 µf pa j` (pa) h` (pa) Γ(−ν)A (ν, `; µcωa2 )
c

which, by the use of formula (5.27), may be compared to expression (5.114).

Finally, it is interesting to study the limit λ → ∞ in formula (5.147), i.e.
∞

0

λ→∞ i X 0 j` (p0 a)
T p~ , p~ −→ − 2 (2` + 1) P` (p̂ · p̂ ) (1) , (5.149)
4π µf p `=0 h` (pa)
to notice, by comparison to formula (5.79), that this way we may obtain the expression
for the T -matrix elements of the infinitely hard sphere. Indeed, in the limit of very large
coupling (λ) the structure of the resonance channel ψc becomes invisible. What remains
is the screening hard sphere.

88
89
Chapter 6

Relativistic kinematics

6.1 Relativistic kinematics

The total energy, Etotal , for a system of two non-interacting on-mass-shell particles of
masses m1 and m2 , which are freely moving with linear momenta respectively p~1 and p~2 ,
is given by the sum of the individual energies, E (~
p1 ) and E (~
p2 ) respectively, according
to
q q
Etotal = E (~
p1 ) + E (~p2 ) = p~1 2 + m1 2 + p~2 2 + m2 2 . (6.1)
In the center-of-mass frame, where p~1 = −~
p2 = p~, one has the following relations:

q q
2 2 2
s = (ECM, total ) = 2~ 2 2
p + m1 + m2 + 2 p~ + m1 2 p 2 + m2 2 ,
−~
 2   
p 2 − m1 2 − m2 2
s − 2~ = 4 p~ 2 + m1 2 p 2 + m2 2
−~ ,
 
p 2 = s2 − 2s m1 2 + m2 2 + m1 4 + m2 4 − 2m1 2 m2 2 ,
4s~

1 nh ih io
and p~ 2 = s − (m1 + m2 )2 s − (m1 − m2 )2 . (6.2)
4s
The Mandelstam variables, s, t and u, for the process

1 + 2 −→ 3 + 4 (6.3)
are defined by

s = (p1 + p2 )2 , t = (p1 − p3 )2 , u = (p1 − p4 )2 , (6.4)

or, alternatively, by using total four-momentum conservation which is given by

p1 + p 2 = p 3 + p 4 , (6.5)

one also has

s = (p3 + p4 )2 , t = (p2 − p4 )2 , u = (p2 − p3 )2 , (6.6)

90
 Notice that we use here the metric (+, −, −, −), which for s gives
s = (p1 + p2 )2 = (E (~ p2 ))2 − (~
p1 ) + E (~ p1 + p~2 )2 . (6.7)
In the the center-of-mass frame, where p~1 = −~
p2 , one obtains, moreover
s = (E (~ p2 ))2 = (ECM )2
p1 ) + E (~ , (6.8)
which equals the total invariant mass, as already anticipated in formula (6.2).
Furthermore, from their definition one observes that the Mandelstam variables (6.4)
are Lorentz invariant and hence invariants with respect to any Lorentz transformation.
By total momentum conservation 6.5, one deduces
s+t+u = 3p1 2 + p2 2 + p3 2 + p4 2 + 2p1 · (p2 − p3 − p4 )
= 3p1 2 + p2 2 + p3 2 + p4 2 − 2p1 2
= p1 2 + p2 2 + p3 2 + p4 2
= m1 2 + m2 2 + m3 2 + m4 2 . (6.9)
Consequently, for on-mass-shell processes s, t and u are not independent.
In Fig. 6.1 we visualize things for the center-of-mass frame.

3 p
~3
m3
1 p
~1 p
~2 2 CM ϑCM
m1 CM m2
m4
p
~4 4

Before collision After collision

Figure 6.1: Collision in the center-of-mass system. Before collision particle 1 and particle 2 move
towards their center of mass with equal and opposite three-momenta, p ~ 1 and p~2 respectively.
After collision particle 3 and particle 4 move away from their center of mass with equal and
opposite three-momenta, p~3 and p~4 respectively.

The angle between the direction of motion of the outgoing particle 3 and the direction
of motion of the incoming particle 1 is defined as the angle ϑCM of the scattering process
of formula (6.3) in the center-of-mass system. It has the following relation with the
Mandelstam variable t.
t = (p1 − p3 )2 (6.10)

= (p1 )2 + (p3 )2 − 2p1 · p3

= (p1 )2 + (p3 )2 − 2E (~
p1 ) E (~ p1 · p~3
p3 ) + 2~
q q
p1 )2 + (m1 )2 (~
= (m1 )2 + (m3 )2 − 2 (~ p3 )2 + (m3 )2 + 2 |~p1 | |~
p3 | cos (ϑCM ) .

91
6.1.1 π+π− → K +K ∗−
Consider a π + meson which annihilates with a π − meson, resulting in two outgoing Kaon
mesons, a K + meson and a K ∗ − meson. The π − meson is at rest in the laboratory,
whereas the π + meson has a total energy of 9.0 GeV. The Kaon meson comes out with
an angle of 60◦ with respect to the direction of the incoming pion, in the center-of-mass
system.
Given this information, we may determine the other kinematical quantities. For the
masses of the particles we take

mπ = 0.14 GeV , mK = 0.50 GeV , mK ∗ = 0.89 GeV .

Let us first determine the total invariant mass of the system.

√ q
s = 2m2π + 2Eπ+ mπ = 1.60 GeV .

With that result we may determine p~π 2 in the center-of-mass frame.

h i
(~pπ )2 = 1
4
s − 4m2π = 0.62 (GeV) .

Next, we can determine (~pK )2 in the center-of-mass frame, as follows

1 nh ih io
p K )2 =
(~ s − (mK + mK ∗ )2 s − (mK − mK ∗ )2 = 0.148 (GeV) .
4s
The linear momentum of K ∗ is in the center-of-mass frame opposite to p~K , of course.
√
Consequently, we can check our calculations by determining the total invariant mass s
after collision. This gives
√ q q
s = p~K + mK + p~K ∗ 2 + m2K ∗ = 0.63 (GeV) + 0.97 (GeV) = 1.6 (GeV) ,
2 2

which is indeed what we obtained for the sitution before collision.

Then we may determine t and u

t = (mπ )2 + (mK )2 − 2Eπ EK + 2 |~pπ | |~pK | cos (ϑCM ) = −0.437 GeV2 .

and

u = (mπ )2 + (mK ∗ )2 − 2Eπ EK ∗ + 2 |~ pK ∗ | cos (ϑCM ) = −1.04 GeV2 .

pπ | |~

We will denote the total momentum of a particle in the laboratory by q and its linear
momentum by q~. The π − meson is at rest, hence ~qπ− = 0. For the π + we have
q
E (~qπ+ ) = (~qπ+ )2 + m2π+ = 9 (GeV) ⇐⇒ ~qπ+ ≈ 9 (GeV) .

Moreover, since ~qπ− = 0, we have for t the relation

q
t = (qπ− − qK ∗ )2 = m2π− + m2K ∗ − 2mπ− ~qK ∗ 2 + m2K ∗ ,

hence !2
2 m2π− + m2K ∗ − t
q~K ∗ = − m2K ∗ = 4.37 GeV2 .
2mπ−

92
Similarly q
2
u = (qπ− − qK ) = m2π− + m2K − 2mπ− ~qK 2 + m2K ,
hence !2
2 m2π− + m2K − u
~qK = − m2K = 4.66 GeV2 .
2mπ−
Also q
E (~qK ) = ~qK 2 + m2K = 4.68 (GeV) ,
and q
E (~qK ∗ ) = ~qK ∗ 2 + m2K ∗ = 4.46 (GeV) .
Finally, we determine the angle in the frame of the laboratory, of the direction of the
outgoing K meson with respect to the direction of the incoming π + meson.

t − m2π − m2K + 2E (~qπ+ ) E (~qK )

cos (θK,lab ) = = 0.9974 ,
2 |~qπ+ | |~qK |

corresponding to an angle of 4.1 degrees.

6.1.2 Elastic Scattering in the center-of-mass system

In elastic scattering the outgoing particles are identical to the incoming particles, which
at this level amounts to m3 = m1 and m4 = m2 . When we define the three-momenta
before and after collision by respectively p~ and p~ 0 , then we have
0
p~1 = −~
p2 = p~ and p~3 = −~
p4 = p~ , (6.11)

hence, by using formula (6.2),

1 nh ih io
p~ 2 = s − (m1 + m2 )2 s − (m1 − m2 )2
4s

02 1 nh ih io
and p~ = s − (m3 + m4 )2 s − (m3 − m4 )2 . (6.12)
4s
Since, moreover, m3 = m1 and m4 = m2 for elastic scattering, we have for the center-of-
mass three-momenta in that case
2
p~ 2 = p~ 0 . (6.13)
Substitution of the result (6.13) in expressions (6.8) and (6.10) gives the results
q q 2
2 2 2 2
s = (~p ) + (m1 ) + (~
p ) + (m2 )
q q
p )2 + (m1 )2 (~
= (~p )2 + (m1 )2 + (~p )2 + (m2 )2 + 2 (~ p )2 + (m2 )2
q q
= 2 (~ p )2 + (m1 )2 (~
p )2 + (m1 )2 + (m2 )2 + 2 (~ p )2 + (m2 )2 , (6.14)

93
and
q q
02
 
t = (m1 )2 + (m3 )2 − 2 p~ 2 + (m1 )2 p~ + (m3 )2 + 2 |~
p| p~ 0  cos (ϑCM )
 

 
= 2 (m1 )2 − 2 p~ 2 + (m1 )2 p 2 cos (ϑCM )
+ 2~

p 2 {−1 + cos (ϑCM )} .

= 2~ (6.15)

Furthermore
q q
02
u = (m1 )2 + (m4 )2 − 2 p~ 2 + (m1 )2 p~ + (m4 )2 + 2~
p1 · p~4
q q
= (m1 )2 + (m2 )2 − 2 (~
p )2 + (m1 )2 (~
p )2 + (m2 )2 − 2~
p 2 cos (ϑCM ) . (6.16)

As is obvious from the definitions of ϑCM in Fig. 6.1 and p~ in formula (6.11), one has

p~ 2 ≥ 0 and − 1 ≤ cos (ϑCM ) ≤ +1 , (6.17)

hence for the Mandelstam variables s (formula 6.14) and t (formula 6.15), we find

s ≥ (m1 + m2 )2 and t≤0 . (6.18)

94
6.1.3 Elastic Scattering in the lab system
In the laboratory system particle 2 is assumed to be at rest. This is visualized in Fig. 6.2.
We define the laboratory four-momenta by q1 , q2 , q3 and q4 , in order to distinguish from
the center-of-mass four-momenta. We study here again the case of elastic scattering,
which implies m3 = m1 and m4 = m2 .

3 q
~3
m3
1 q
~1 2 ϑlab
m1 m2
m4
4 q
~4

Before collision After collision

Figure 6.2: Collision in the laboratory system. Before collision particle 1 moves with three-
momentum q~1 towards particle 2 at rest (~q2 = 0) in the center of coordinates. After collision
particle 3 and particle 4 move away from the center of coordinates with three-momenta q~3 and
q~4 respectively.

Here, we obtain for the Mandelstam variables (6.4), which, as mentioned before, are
invariant under Lorentz transformations, hence the same for the laboratory system and
the center-of-mass system,

s = (q1 + q2 )2 = (m1 )2 + (m2 )2 + 2E (~q1 ) E (~q2 ) − 2~q1 · ~q2

= (m1 )2 + (m2 )2 + 2E (~q1 ) m2

t = (q1 − q3 )2 = 2 (m1 )2 − 2E (~q1 ) E (~q3 ) + 2~q1 · q~3

= 2 (m1 )2 − 2E (~q1 ) E (~q3 ) + 2 |~q1 | |~q3 | cos (ϑlab )

u = (q2 − q3 )2 = (m1 )2 + (m2 )2 − 2m2 E (~q3 ) . (6.19)

Also using formula (6.9), we find for t

t = 2m1 + 2m2 − s − u = 2m2 (E (~q3 ) − E (~q1 )) . (6.20)

One defines the kinetic energy of the incoming particle by

T1 = E (~q1 ) − m1 . (6.21)

At threshold, where ~q1 = 0, we obtain T1 = 0.

95
6.1.4 Scattering in the forward direction
When q~3 is in the forward direction, i.e. ϑlab = ϑCM = 0, it has no transversal component.
Hence, by total three-momentum conservation also ~q4 is in the longitudinal direction.
Furthermore, because of result (6.13), we have

p~1 = p~3 . (6.22)

As a consequence, also using formula (6.4)

t = (p1 − p3 )2 = (E (~p1 ) − E (~p3 ))2 − (~

p1 − p~3 )2 = 0 . (6.23)

Furthermore, we define the variable ν

s−u E (~q1 ) + E (~q3 ) t

ν = = = E (~q1 ) + , (6.24)
4m2 2 4m2
also using formulae (6.19) and (6.20). For scattering in the forward direction, where
t = 0, the variable ν represents the total energy of the incoming particle in the laboratory
system.
Notice that, because of total four-momentum conservation, one has

s − u = (p1 + p2 )2 − (p1 − p4 )2 = (p1 + p2 ) · (p3 + p4 ) − (p1 − p4 ) · (p3 − p2 ) =

= p1 · p4 + p2 · p3 + p1 · p2 + p4 · p3 = (p1 + p3 ) · (p2 + p4 ) ,

hence
s−u (p1 + p3 ) · (p2 + p4 )
ν = = . (6.25)
4m2 4m2

96
6.2 Crossing
In the following we study the elastic scattering of two different processes. In the first
process one scatters two particles a and b, whereas in the second process the antiparticle
ā is scattered with b. The four-momentum of the incoming antiparticle ā in the second
process is chosen opposite to the four-momentum of the outgoing particle a in the first
process, and vice-versa, i.e.

a (p1 ) + b (p2 ) −→ a (p3 ) + b (p4 ) (I)

(6.26)
ā (−p3 ) + b (p2 ) −→ ā (−p1 ) + b (p4 ) (II) .

It can be shown that the scattering-matrix elements for the two processes are equal.
Consequently, this represents a symmetry of the dynamical equations, which is called
crossing symmetry.
Here, it is of interest to compare the Mandelstam variables of the two processes. From
formula (6.4) we find for the first process

sI = (p1 + p2 )2 , tI = (p1 − p3 )2 , uI = (p1 − p4 )2 , (6.27)

whereas for the second process, also using formula (6.6), we find

sII = (−p1 + p4 )2 , tII = (−p1 + p3 )2 , uII = (p2 + p1 )2 . (6.28)

From formulae (6.27) and (6.28) we observe that the roles of s and u are interchanged

sII = uI , tII = tI , uII = sI , (6.29)

for process (6.26,II) with respect to process (6.26,I). One says the the process (6.26,II) is
represented by the u-channel of process (6.26,I).
Observe, moreover, that for the variable ν which has been defined in formula (6.24),
we have
νII = −νI . (6.30)

97
6.3 The physical region for elastic scattering
The variables ν and t are related to the variables p~ 2 and ϑCM as can be seen from
expressions (6.24) and (6.15). Physically allowed values for p~ 2 and ϑCM are given by

p~ 2 ≥ 0 and − 1 ≤ cos (ϑCM ) ≤ +1 . (6.31)

Hence, it has to be expected that for related variables the physical region, i.e. the
region of allowed physical values, is also restricted. Furthermore, p~ 2 and ϑCM are rather
independent. But, the dependence of ν and t is not very clear from their definitions.
In formula (6.18) we indicate the physical limits for the Mandelstam variables s and
t in the case of elastic scattering, i.e. m3 = m1 and m4 = m2 . Here we will study this
subject in more detail, moreover concentrating on the variables ν and t.
From equation (6.13) we learn that for elastic on-shell scattering one has for the total
energies of the four particles involved

E (~
p3 ) = E (~
p1 ) and E (~
p4 ) = E (~
p2 ) . (6.32)

We will frequently use the relations (6.13) and (6.32) in the following.
From the expressions (6.14), (6.16) and the definition of ν, which is given in formula
(6.24), we deduce
1 h 1 2
i
ν = E (~p1 ) E (~p2 ) + 2
p~ {1 + cos (ϑCM )} . (6.33)
m2
Except for the forward direction where cos (ϑCM ) = 1, we may use expression (6.15) in
order to substitute p~ 2 for t.

ν = (6.34)
"s s #
1 2 t t t {1 + cos (ϑCM )}
(m1 ) − (m2 )2 − − .
m2 2 {1 − cos (ϑCM )} 2 {1 − cos (ϑCM )} 4 {1 − cos (ϑCM )}

Furthermore, in the forward direction we have

1 h i
t = 0 and ν = p2 ) + p~ 2 ≥ m1 .
E (~p1 ) E (~ (6.35)
m2
From formula (6.34) we observe that a condition for ν being real, is given by

t ≤ (m1 )2 {1 − cos (ϑCM )} or t ≤ (m2 )2 {1 − cos (ϑCM )} , (6.36)

depending on whether m1 < m2 , or m1 > m2 . Let us assume that m1 < m2 . In that case
we find that for ν to be real, t can at most equal 4 (m1 )2 . However, this value is outside
the physical region, because of condition (6.18). Consequently, t can take any value from
t = 0 to t = −∞ in the physical region, as is expressed by relation (6.15). The related
physical values of ν can be found from formulas (6.34) and (6.35), by varying cos (ϑCM )
within the limits (6.31).
In order to make some graphical representations of the various formulas, we will con-
centrate now on πN elastic scattering. In figure (6.3) we have depicted the situation. The
physical region for ν and t in the s-channel process given in formula (6.3), is shaded. The

98
upper boundary is given by t = 0, corresponding to forward scattering, whereas the lower
boundary, corresponding to backward scattering, is given by
s s
1 t t
ν = (m1 )2 − (m2 )2 − . (6.37)
m2 4 4
We have chosen here m1 = 0.14 GeV and m2 = 0.94, which corresponds to πN elastic
scattering. The solid lines correspond to constant values for ϑCM .

ν/m1
1 2 3 4 5 6
0◦ 22.5◦
45◦
67.5◦

90◦
t/ (m1 )2

-50
112.5◦

135◦

157.5◦
-100
180◦

Figure 6.3: Elastic s-channel πN scattering (m 1 = mπ = 0.14 GeV, m2 = mN = 0.94 GeV).

The shaded area, which is understood to extend to infinity, represents the physically allowed
region for ν and t. The solid lines result from formula (6.34) for constant ϑ CM , the values of
which are indicated on the righthand side of the figure.

Once an amplitude, or any other physical quantity, is determined in terms of the

variables ν and t, one can in principle extend the values of those variables to outside
the physical region. Such procedure is sometimes advantageous, for example when the
amplitude has a zero for a certain set of values of those variables. In the latter case
one may then Taylor expand the amplitude in a small region around that zero value and
determine the first few coefficients of the expansion. When from experiment one has
enough information to achieve an extension of fitted formulas to the same region, one
may then compare models with ”experiment” in the unphysical region. Now at this point
we must notice that we can obtain a better performance of experiment if we could join
knowledge of different regions. For that reason we come back here on the issue of crossing,
which is discussed in section (6.2).
We have seen in formulae (6.29) and (6.30) that antipion-nucleon scattering is described
by interchanging the roles of s and u, or equivalently by substituting ν by −ν in the
expression of the amplitude. Consequently, the physical regions for the processes antipion-
nucleon and pion-nucleon are each other mirror images, reflected around the t axis, as is
represented in figure (6.4).
Now, since the π 0 is its own antiparticle, and the π − is the antiparticle of the π + ,
pions and antipions are the same particles. Hence, both physical regions represented in
figure (6.4), describe πN elastic scattering. Consequently, we may obtain experimental

99
knowledge of the scattering amplitudes for elastic πN scattering in both regions at the
same time.

ν/m1
-6 -5 -4 -3 -2 -1 1 2 3 4 5 6

antipion-nucleon pion-nucleon
ϑ
C

t/ (m1 )2
M
=
-50 18
0◦

-100

Figure 6.4: The physically allowed regions (shaded and assumed to extend to infinity), for
pion-nucleon and for antipion-nucleon elastic scattering.

In figure (6.5) we show how the continuation of the curve for ϑCM = 180◦ to values
outside the physical region, include the combination ν = 0 and t = 4 (m1 )2 . The region

t = 4 (m1 )2 ν/m1
1 2 3 4 5 6

ϑ
C
t/ (m1 )2

M
=
-50 18
0◦

-100

Figure 6.5: The combination ν = 0 and t = 4 (m1 )2 lies on the continuation of the curve for
ϑCM = 180◦ to outside the physical region.

between the axis t = 0 for −1 ≤ ν/m1 ≤ +1 and the curve cos (ϑCM ) = −1 is called the
subthreshold crescent for πN elastic scattering. It is an unphysical region. But, experiment
can be analytically continued into it from both sides.

100
6.3.1 The subthreshold crescent
In figure (6.6) we show in some detail the subthreshold crescent for πN elastic scattering.
This nonphysical region touches the two physical regions for πN elastic scattering in the

t/ (m1 )2

4
.0
−1
M
)= 3
ϑC .5
s( −0
co
)=
(ϑ CM
2
cos 0.0
)=
(ϑ CM
cos 1
) = 0.5
cos (ϑCM
cos (ϑCM ) = 1.0

-1 0 1
ν/m1

Figure 6.6: The region spanned by the line element −1 ≤ ν ≤ +1 at the t = 0 axis and the
curve given by the relation of formula (6.34) over the same interval for ϑ CM = 180◦ , is shown
here, as well as some curves for constant cos (ϑ CM ).

positions (ν = −1, t = 0) and (ν = +1, t = 0). The solid lines in figure (6.6) represent
the curves for constant cos (ϑCM ) at the values 1 (forward scattering), 1/2, 0 (transversal
scattering), -1/2 and -1 (backward scattering). The curves seem to cross the line ν = 0 at
t = 2 (m1 )2 {1 − cos (ϑCM )}. However, except for forward and backward scattering, this
is only approximately true because of the smallness of the pion mass with respect to the
nucleon mass, as can be observed from formula (6.34).

101
6.4 The amplitude for elastic πN scattering
Pions have isospin Iπ = 1 and nucleons IN = 12 . Consequently, the total isospin of the πN
system may be IπN = 21 or IπN = 23 . The related non-trivial Clebsch-Gordon coefficients
hI, Iz | πN i are given by
D  E D  E q
3 1
,  π+n = 3
, − 21  π − p = 1
,

2 2 2 3

D  E D  E q
3 1
,  π0p = 3
, − 21  π 0 n = 2
,

2 2 2 3

D  E D  E q
1 1
,  π+n = − 1
, − 21  π − p = 2
,

2 2 2 3

D  E D  E q
and − 1 1
,  π0p = 1
, − 21  π 0 n = 1
. (6.38)

2 2 2 3

Using the values of table (6.38) we obtain for instance

 E q  E q  E
3  0
− 12 1 2
 −
I = , Iz = = π p + π n

2 3 3

 E q  E q  E
 0
= 12 , Iz = − 12 2 1
 −
and I = − π p + π n , (6.39)

3 3

for which the inverse relations are

 E q  E q  E
 0 1 3
− 21 2 1
π n = 3
2, − 3
2, − 21
 E q  E q  E
2 3
− 21 1 1
− 21
 −
and π p = 3
2, + 3
2, . (6.40)

Hence, the transition-matrix element for the process π − p → π 0 n yields

D E
π 0 n |T | π − p = (6.41)
q D  q D  q  E q  E
1 3
= , − 21  − 2 1
, − 21  T 2 3
 , − 21 + 13  12 , − 21
  
3 2 3 2 3 2

√ D   E D   E
2 3
= 3 2
, − 21  T  32 , − 21 + 1
3
3
2
, − 21  T  12 , − 21 +
D   E √ D   E
2 1
− , − 21  T  32 , − 21 − 2 1
, − 21  T  12 , − 21 .
   
3 2 3 2

When, furthermore, we assume that the operator T conserves isospin, then the transition-
matrix elements for states with different total isospin must vanish. Hence, it that case
D E √ nD   E D   Eo
π 0 n |T | π − p = 2 3
, − 21  T  32 , − 21 − 1
, − 21  T  21 , − 12 . (6.42)
   
3 2 2

Moreover, strong interactions are independent of the isospin z component. Consequently,

we may define

T (2) =
1
D   E D   E
1
2
, + 21  T  12 , + 21 = 12 , − 21  T  12 , − 21 , (6.43)

102
and, similarly,

T (2) =
3
D   E D   E
3
2
, ± 23  T  32 , ± 23 = 32 , ± 21  T  32 , ± 21 . (6.44)

For the transition-matrix element (6.42) of the process π − p → π 0 n, we obtain then

√  
T (2) − T (2)
3 1
D E
π 0 n |T | π − p = 3
2
. (6.45)

The transition-matrix elements collected in table (6.1) are obtained following similar gym-
nastics for the other πN channels.

process transition-matrix element

T (2)
3
hπ + p |T | π + pi = hπ − n |T | π − ni
 
T ( 2 ) + 2T ( 2 )
3 1
+ + 0 0 1
hπ n |T | π ni = hπ n |T | π ni 3
 
2T ( 2 ) + T ( 2 )
3 1
− − 0 0 1
hπ p |T | π pi = hπ p |T | π pi 3
√  
T (2) − T (2)
3 1
2
hπ 0 n |T | π − pi = hπ 0 p |T | π + ni 3

Table 6.1: πN processes

103
6.4.1 The t channel
In section (6.2) we study s ↔ u crossing. Here we will pay attention to the interchange
of s and t. Thereto, we study the process

a (p1 ) + ā (−p3 ) −→ b̄ (−p2 ) + b (p4 ) (III) . (6.46)

The Mandelstam variables (6.4) for the process (6.46) are given by

sIII = (p1 − p3 )2 = tI , tIII = (p1 + p2 )2 = sI , uIII = (p1 − p4 )2 = uI , (6.47)

where sI , tI and uI are defined in formula (6.27).

For πN elastic scattering, the process (6.46) represent the inelastic process

ππ −→ N̄ N . (6.48)

The physical region for the process (6.48) is given by

t = tI = sIII ≥ (2mN )2 , s = sI = tIII ≤ 0 . (6.49)

In order to account for the condition (6.49) on t, we rewrite expression (6.34) in the
following form.

ν = (6.50)
"s s #
1 t t t {1 + cos (ϑCM )}
− (m1 )2 − (m2 )2 − .
m2 2 {1 − cos (ϑCM )} 2 {1 − cos (ϑCM )} 4 {1 − cos (ϑCM )}

The enveloppe of the physical region in the t channel is given by ϑCM = 180◦ , as in the
case of the s channel and the u channel (see figures 6.3 and 6.4). In figure (6.7) we depict
the three physical regions in the (ν, t) plane.

104
 250
π π̄ → N̄ N

◦
0
200 18
=
M
ϑC

150

t/ (m1 )2
100

50

ν/m1
-6 -5 -4 -3 -2 -1 1 2 3 4 5 6

π̄N → π̄N πN → πN

ϑC
M
=
-50

18
0
◦
-100

Figure 6.7: The physically allowed regions (shaded areas) in the (ν, t) plane for the three
channels related to elastic πN scattering (m 1 = mπ = 0.14 GeV, m2 = mN = 0.94 GeV). The
shaded areas, which are understood to extend to infinity, represent respectively the physically
allowed region for πN (lower right), for π̄N (lower left) and for ππ̄ → N̄ N (top). The solid lines
result from formulas (6.34) and (6.50) for constant ϑ CM .

105
6.4.2 The transition-matrix elements
In the t channel, where the scattering process is given by ππ̄ → N̄ N , we have to consider
total isospin 0 and 1. The corresponding transition-matrix elements are denoted T (+) for
isospin 0 and T (−) for isospin 1. From the crossing relations one has
   
T ( 2 ) + 2T ( 2 ) T (2) − T (2)
1 3 1 3
(+) 1 (−) 1
T = 3
and T = 3
. (6.51)

The reverse relations are

T ( 2 ) = T (+) + 2T (−) T ( 2 ) = T (+) − T (−) .

1 3
and (6.52)

Using the relations (6.52), we may express the processes (6.1) in T (+) and T (−) . This
is shown in table (6.2).

process transition-matrix element

hπ + p |T | π + pi = hπ − n |T | π − ni T (+) − T (−)

hπ + n |T | π + ni = hπ 0 n |T | π 0 ni T (+) + T (−)

hπ − p |T | π − pi = hπ 0 p |T | π 0 pi T (+)
√
hπ 0 n |T | π − pi = hπ 0 p |T | π + ni − 2T (−)

Table 6.2: πN processes

The process(es) we have under study is given by

π (q) + N (p) −→ π (q 0 ) + N (p0 ) . (6.53)

Hence, in terms of the momenta which are defined in formula (6.3), we define here

q = p1 , q 0 = p3 , p = p 2 and p0 = p4 . (6.54)

106
Bibliography

[1] Quantum Mechanics, Eugen Merzbacher, 2nd edition (Wiley, NewYork).

[2] Methods of Theoretical Physics, Philip M. Morse and Herman Feshbach, McGraw-Hill
book company. inc. (NY 1953).

[3] Spectra and Decay Properties of Pseudo-scalar and Vector Mesons in a Multichannel
quark Model, George Rupp, Doctoral Thesis, Nijmegen (1982).

[4] A low-lying scalar meson nonet in a unitarized meson model, by E. van Beveren,
T.A. Rijken, K. Metzger, C. Dullemond, G. Rupp and J.E. Ribeiro, Zeitschrift für
Physik C30, 615-620 (1986)

[5] Bateman Manuscript Project, Higher Transcendental Functions, Erdélyi et al., eds.,
vol.1 (McGraw-Hill, New York).

[6] General Mechanics, J.J. de Swart, lecture notes of the Center for Theoretical Physics
of the Nijmegen University, January 1981.

[7] W. Bauer, Journal für Mathematik LVI (1859), pp. 104-106; see also: G.N. Watson,
A treatise on the Theory of Bessel Functions, section 4.32.

[8] Handbook of Feynman path integrals, C. Grosche and F. Steiner, Springer Verlag
(Berlin, 1998).

107