Published on 02 July 2015 on https://pubs.rsc.org | doi:10.

1039/9781782620198-00019

CHAPTER 2

Quantum Theory of Atom–Atom

Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Elastic Scattering

2.1 The Wave Picture of the Scattering Process

In the previous chapter we discussed how the motion of the two atoms in an
atom–atom scattering process could be represented in terms of the motion
of the center-of-mass of the combined system and the motion of one of the
atoms relative to the other. The same analysis applies in quantum mechanics
(see the treatment of the hydrogen atom problem in any quantum mechanics
text book).1–3 The wavefunction of the two atoms is then represented as a
product of the wavefunction of the center-of-mass and the wavefunction for
the motion of one atom relative to the other. Just as in Chapter 1, we may
ignore the motion of the center-of-mass of the system for all purposes other
than transforming from the center-of-mass to laboratory reference frames.
The relative motion of two colliding atoms approaching each other is rep-
resented as a plane wave in quantum mechanics. Such a plane wave has a
well defined linear momentum and, as a result of the uncertainty principle,
is therefore totally delocalized in space. Its wavefunction may be written in
the form
  12
km
jo (r, E) Z eik·r (2.1)
2ph2

where the subscript on jo indicates that this is the solution of the problem
in the absence of the interaction potential, k is the “wavenumber” and m is
the reduced mass of the colliding atoms.

RSC Theoretical and Computational Chemistry Series No. 7

Theory of Molecular Collisions
By Gabriel G. Balint-Kurti and Alexander P. Palov
C Gabriel G. Balint-Kurti and Alexander P. Palov, 2015

Published by the Royal Society of Chemistry, www.rsc.org

19
 View Online

20 Chapter 2

Unlike the wavefunction of a bound state, a plane wave or a scattering

wavefunction cannot be normalized in the conventional sense. The nor-
malization factor in eqn (2.1) is somewhat arbitrary. We discuss the prob-
lem in Appendix A. In this book we will consistently normalize scattering
wavefunctions on the energy scale.4
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

Following the normal convention, we consider the incident particle to be

a plane wave approaching from the negative Z direction. The target atom or
scattering center is located at the origin. The interaction potential is taken
to be spherically symmetric, and the azimuthal angle f therefore plays no
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

role in the collision. After the collision the particle is scattered into different
polar angles with a scattering amplitude of f (q). At large distances from the
scattering center the wavefunction therefore takes the form
 12 
eikr
 
km ikz
j(r, E) →∞
r^ e C f (q) . (2.2)
2ph2 r

The term eikz is the initial incident wave, while f (q) eikr represents the scat-
tered wave going radially outwards from the scattering center. The momen-
v
tum operator is Ki~ . If we operate with this on the incident wave we see
vz
that it has a momentum of ~k and consequently the incident flux of atoms
(i.e. the number of atoms crossing a unit area per unit time) corresponding
k~ ikz 2 k~
to the plane wave eikz is |e | Z . The solid angle sin q dq df subtends
m m
2
an area of r sin q dq df on the surface of a sphere of radius r centered on the
scattering center. As the collision is elastic, the radial momentum remains
unchanged by the collision and is equal to k~. The flux of atoms passing
through the small area r2 sin q dq df on the surface of the sphere arising
2
eikr k~ eikr
from the scattered wave f (q) is therefore f (q) r2 sin q dq df Z
r m r
k~
|f (q)|2 sin q dq df.
m
The “differential cross section” is defined as the flux of atoms scattered
into a given solid angle divided by the incident flux of atoms (see Chapter 1).
Note that any normalization factor cancels out in this definition of the cross
section. From the above definition we may write the differential cross section
in the form
k~
|f (q)|2 sin q dq df
m
s(U) dU Z . (2.3)
k~
m

We conclude that the differential cross section is given by

s(U) Z |f (q)|2 . (2.4)

 View Online

Quantum Theory of Atom–Atom Elastic Scattering 21

2.2 The Phase Shift

Close to the scattering center the wavefunction can be found by solv-
ing the time-independent Schrödinger equation. As with all second order
differential equations there are two linearly independent solutions to this
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

equation. We will mainly be interested in the physically reasonable or

“regular” solution. This is the one which tends to zero near the origin,
where the interaction potential becomes infinite. The full time-independent
Schrödinger equation is
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

~2
 
K V2 C V(r) j(r) Z Ej(r). (2.5)
2m
Expressing the Laplacian operator (V2 ) in spherical polar coordinates1–3,5,6
yields
~2 1 v 2 vj 1 v2 j
      
1 1 v vj
K r C sin q C C V(r)j Z Ej.
2m r2 vr vr r2 sin q vq vq sin2 q vf
(2.6)
If we multiply this equation by r2 we see that the Hamiltonian operator
becomes a sum of radial and angular parts. The Hamiltonian is said to be
“separable” in radial and angular coordinates and the wavefunction may
therefore be written as a product of radial and angular parts. The angu-
lar wavefunctions are the eigenfunctions of the orbital angular momentum
operator
1 v2
   
2 1 v v
K~ sin q C Y`m (q, f)
sin q vq vq sin2 q vf
Z `(` C 1)~2 Y`m (q, f) (2.7)
where Y`m (q, f) are the spherical harmonics.7,8 Utilizing eqn (2.7) we now
write the total wavefunction in the form
1
j`,m (r) Z c` (r)Y`m (q, f) (2.8)
r
where ` and m are both “good” quantum numbers, as in the case of the
motion of an electron in the hydrogen atom.2,3
1
The extra factor of is introduced so as to simplify the ensuing differential
r
equation in the radial coordinate, r. Substituting eqn (2.8) into eqn (2.6) we
obtain
~2 v2
 
`(` C 1)
K 2C c` (r) C V(r)c` (r) Z Ec` (r) (2.9)
2m vr r2
or alternatively

~ 2 v2 ~2 `(` C 1)
 
K c` (r) C V(r) C c` (r) Z Ec` (r). (2.10)
2m vr2 2m r2
 View Online

22 Chapter 2

The term in curly brackets in eqn (2.10) is called the “effective potential”.
 ~2 `(` C 1) 
At large r both the centrifugal potential and the interaction
2m r2
potential (V(r)) tend to zero. The general solution of eqn (2.9) is then a
combination of sines and cosines and can be written in the form
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

 
`p
→∞ A sin kr K
c(r) r^ C h`
2
1 n i(krK `p2 Ch` ) o
K eKi(krK 2 Ch` )
`p
ZA e
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

2i
i`C1 eKih` n Kikr o
ZA e C (K1)K(`C1) ei(krC2h` ) (2.11)
2
√
 2mE 
where k is the wavenumber k Z and h` is the phase shift which rep-
~
resents the difference in phase compared to the solution in the absence of a
potential.
The radial Schrödinger equation (eqn (2.9)) for free motion when V(r) Z
0 has two linearly independent solutions. These are called the spheri-
cal Riccati–Bessel functions9 ĵ` (kr) Z kr j` (kr) and n̂` (kr) Z kr n` (kr), where
j` (kr) and n` (kr) are the closely related spherical Bessel functions.6,10 The ex-
tra “r” in front of the spherical Bessel functions of the first (j` ) and second
1
(n` ) kinds arises from the fact that we introduced an extra factor of when
r
deriving eqn (2.9). The regular solution ĵ` (kr) goes to zero at the origin where
~2 `(` C 1)
the centrifugal potential becomes infinite. The irregular solu-
2m r2
tion n̂` (kr) becomes infinite as r → 0. At large r values the solutions behave as
below
 
`p
→∞ sin kr K
ĵ` (kr) Z kr j` (kr) r^
2
 
`p
n̂` (kr) Z kr n` (kr) r^→∞ K cos kr K . (2.12)
2

2.3 Partial Wave Expansion

A plane wave traveling in the positive Z direction can be expanded in terms
of the regular spherical Bessel functions

∞
X
eikz Z (2` C 1)i` j` (kr)P` (cos q) (2.13)
`Z0

where P` (cos q) are the Legendre polynomials.7,8

 View Online

Quantum Theory of Atom–Atom Elastic Scattering 23

Using eqn (2.12) and (2.13) in eqn (2.2) we obtain

  12 (X∞  
km ` 1
j(r, E) r^
→∞ (2` C 1)i
2ph2 kr
`Z0
)
eikr
 
`p
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

! sin kr K P` (cos q) C f (q)

2 r
  21 (X ∞  
km ` 1 1  i(krK `p2 ) Ki(krK `p
2 )

Z (2` C 1)i e K e
2ph2 kr 2i
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

`Z0
)
eikr
!P` (cos q) C f (q)
r
 21 (X ∞
1 (K1)`C1 n Kikr
  
km K(`C1) ikr
o
Z (2` C 1) e C (K1) e
2ph2 kr 2i
`Z0
)
eikr
! P` (cos q) C f (q) . (2.14)
r

In order to relate the phase shift of eqn (2.11) to the scattering amplitude
f (q) we need to expand the scattering amplitude in terms of the Legendre
polynomials
∞
X
f (q) Z f` P` (cos q). (2.15)
`Z0

It is this expansion which is called the “partial wave expansion” and the f`
constitute the partial waves or amplitudes.
Substituting eqn (2.15) into eqn (2.14) we obtain
 12 (X ∞
1 (K1)`C1
  
km
j(r, E) r^
→∞ (2` C 1)
2ph2 kr 2i
`Z0
∞
)
n
Kikr K(`C1) ikr
o X eikr
! e C (K1) e P` (cos q) C f` P` (cos q)
r
`Z0
 21 (X ∞
(K1)`C1 Kikr
 
km 1
Z (2` C 1) e
2ph2 r 2ik
`Z0 )
  
1 ikr
C (2` C 1) C f` e P` (cos q)
2ik
  21 (∞
km 1 X
Z (2` C 1)(K1)`C1 eKikr
2ph2 2ikr
`Z0 )
ikr

C ((2` C 1) C 2ikf` ) e P` (cos q) . (2.16)
 View Online

24 Chapter 2

Incoming and outgoing radial waves must have equal amplitudes be-
cause of the particle numbers conservation law. Consequently, comparing
eqn (2.16) to eqn (2.11) we can equate the relative coefficients of these waves
((2` C 1) C 2ikf` )
(K1)K(`C1) ei(2h` ) Z (2.17)
(2` C 1)(K1)`C1
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

and therefore
(2` C 1) ih`
e sin h` .
f` Z (2.18)
k
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Combining eqn (2.18) with eqn (2.4) and (2.15) provides an expression for
the differential cross section in terms of the phase shift
∞ 2
X (2` C 1) ih`
s(U) Z e sin h` P` (cos q) . (2.19)
k
`Z0

The integral cross section is obtained by integrating the differential cross

section over all solid angles.
The normalization condition for the Legendre polynomials is
Z p
2d``0
P` (cos q)P`0 (cos q) sin q dq Z (2.20)
0 2` C1
where d``0 is a Kronecker delta function (i.e. d``0 Z 1 if ` Z `0 but is zero
otherwise).
The integral cross section is therefore given by
Z
s Z s(U) dU
∞ 2
Z 2p Z p X (2` C 1) ih`
Z df e sin h` P` (cos q) sin q dq
0 0 k
`Z0
∞ 2
Z p X (2` C 1)
Z 2p eih` sin h` P` (cos q) sin q dq
0 k
`Z0
∞
X (2` C 1) (2`0 C 1) i(h` Kh0` )
Z 2p e sin h` sin h`0
k k
`Z0,`0 Z0
Z p
! P` (cos q)P`0 (cos q) sin q dq
0
∞
4p X
Z (2` C 1) sin2 h` . (2.21)
k2
`Z0

The integral cross section can be also presented as a sum of partial integral
cross sections
X∞
sZ s` (2.22)
`Z0
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 25

where s` is given by

4p
s` Z (2` C 1) sin2 h` . (2.23)
k2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

2.4 Scattering of Identical Particles

When two identical atoms or molecules are scattered off each other, their
symmetry with respect to interchange must be considered. The situation
is exactly the same as that leading to the formation of ortho- and para-
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

hydrogen.2,3,5,11,12 If the particles have integer spin they are bosons and the
corresponding wavefunction must be symmetric with respect to their inter-
change. Fermions on the other hand have half integer spins and their wave-
functions must be odd with respect to their interchange. We have chosen to
illustrate atom–atom scattering using the scattering of two 20 Ne atoms. As
these are bosons we will discuss the effect of interchange symmetry on the
scattering for this case.
When the two atoms are interchanged the main effect is that their relative
position vector is changed in sign, i.e. r → Kr. In addition to changing the
sign of the Z component of the relative position vector the interchange of
the two atoms affects the scattering angle q in eqn (2.2). Figure 2.1 shows a
schematic illustration of the classical trajectories of two identical particles
colliding with each other in the center-of-mass reference frame. Only the ini-
tial and final velocities are illustrated. In Figure 2.1 (top) the solid red line
shows the trajectory of the incident particle and the solid blue line shows
that of the target. The incident particle is shown as being deflected through
an angle of q. In Figure 2.1 (bottom) the two particles have been interchanged
and we see that the effect is that the incident particle is now scattered into an
angle (pKq) as compared with the original direction of motion of the parti-
cles before interchange. Our conclusion is that the process of interchanging
the two indistinguishable particles is to change the angle of scattering in the
manner q → (pKq). The figure also illustrates the fact that the differential
cross section for the scattering of identical atoms will be symmetric about
q Z p/2, i.e. s(q) Z s(pKq).
If we denote the particle exchange or permutation operator by the symbol
P̂ we can write the symmetrized wavefunction as

1  
jsym (r, E) Z √ j(r, E) C P̂j(r, E)
2
1
Z √ [j(r, E) C j(Kr, E)] . (2.24)
2

1
The factor of √ is required to renormalize the wavefunction. Referring
2
back to eqn (2.2) we can now write the asymptotic form of the symmetrized
wavefunction as
 View Online

26 Chapter 2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.1 Schematic illustration of the trajectories of two colliding atoms in

the center-of-mass frame. (A) Before interchange of identical atoms.
(B) After interchange.

  12 
1 km  ikz 
→∞ √
j(r, E) r^ e C eKikz
2 2ph2
eikr

C [f (q) C f (p K q)] . (2.25)
r
The differential cross section is then found in the same way as before (see
the discussion leading to eqn (2.4)) and is found to be given by
1 2
s(U) Z |f (q) C f (p K q)| . (2.26)
2
Note the new wavefunction in eqn (2.25) contains flux incident from
both the negative and positive Z directions.11 As the incident particle and
the target particle are indistinguishable, both the incident particle wave,
approaching from negative z, and the target wave, approaching from pos-
itive z, now contribute to the scattering into the solid angle U. If we had
treated these two particle beams as independent beams, without insisting
that the wavefunction be symmetric with respect to exchange of the two
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 27

atoms, then we would have obtained the differential cross section expression
1
s(U) Z {|f (q)|2 C |f (p K q)|2 }, which as we will see leads to very different
2
results.
The scattering amplitude for the angle (p K q) can be expressed as (see
eqn (2.15))
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

∞
X ∞
X
f (p K q) Z f` P` (cos(p K q)) Z f` P` (Kcos q). (2.27)
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

`Z0 `Z0

The Legendre polynomials are symmetric or antisymmetric with respect to

the change of sign of their argument depending on their order

P` (Kcos q) Z (K1)` P` (cos q). (2.28)

Using eqn (2.18) along with eqn (2.27) and (2.28) we may write the total
scattering amplitude of eqn (2.26) as

∞
X (2` C 1)
eih` sin h` P` (cos q) C (K1)` P` cos q)


f (q) C f (p K q) Z
k
`Z0
∞
X (2` C 1) ih`
Z e sin h` 2P` (cos q) (2.29)
k
`Z0,2,4,...

where the odd terms have cancelled due to the parity property of the
Legendre polynomials.
Combining eqn (2.29) with (2.26) now yields an expression for the differ-
ential cross section in terms of the phase shift

2
∞
X (2` C 1) ih`
s(U) Z 2 e sin h` P` (cos q) . (2.30)
k
`Z0,2,4,...

If we consider scattering into an angle of q Z p/2 we can immediately see

a clear difference between the scattering predicted by the above equation
and that which would have arisen if we had not used a symmetric wavefunc-
tion. As P` (cos p/2) Z 0 for odd values of `, the cross section predicted by
eqn (2.30) will be a factor of 2 greater than that which we would have obtained
without invoking the use of symmetry.
We obtain the integral cross section by integrating over all solid angles.
Integration over f yields a factor of 2p. Using the orthogonality relationship
of the Legendre polynomials (see eqn (2.20)) the integral cross section is
 View Online

28 Chapter 2
Z
sZ s(U) dU
2
Z 2p Z p ∞
X (2` C 1) ih`
Z df 2 e sin h` P` (cos q) sin q dq
0 0 k
`Z0,2,4,...
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

∞ ∞
X X (2` C 1) (2`0 C 1) i(h` Kh0` )
Z 4p e sin h` sin h`0
k k
`Z0,2,4... `0 Z0,2,4,...
Z p
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

! P` (cos q)P`0 (cos q) sin q dq

0
∞
8p X
Z (2` C 1) sin2 h` . (2.31)
k2
`Z0,2,4,...

2.5 The S and T Matrices

In this section we introduce the S and T matrices. These matrices are most
useful in the case of inelastic and reactive scattering. The discussion here
serves simply as an introduction.
The large distance, asymptotic form, of the scattering wavefunction (see
eqn (2.11)) may be written as
 
`p
→∞ A sin kr K
c(r) r^ C h`
2
1 Kih` n Ki(krK `p2 ) o
K e2ih` ei(krK 2 )
`p
Z KA e e
2i
1 n o
Z K A eKih` eKi(krK 2 ) K S` ei(krK 2 ) .
`p `p
(2.32)
2i
The S matrix is the ratio of the radial outgoing wave to that of the radial
incoming wave. The radial incoming wave is eKi(krK 2 ) and the correspond-
`p

ing outgoing wave is ei(krK 2 ) . These radial waves come from the asymptotic
`p

form of the spherical Bessel functions which are the solutions of the scatter-
ing problem in the absence of a potential. From eqn (2.32) we can identify
the S matrix as

S` Z e2ih` . (2.33)

An important property of the S matrix is its unitarity, i.e. S` S†` Z S†` S` Z 1. This
property is a reflection of the conservation of particle flux.
We will define the T matrix by†

S Z 1 K T. (2.34)

† The T matrix is sometimes defined by the alternative relationship S Z 1 K iT or S Z 1 K i2pT.

 View Online

Quantum Theory of Atom–Atom Elastic Scattering 29

Substituting this into eqn (2.32) we see that

 
`p
→∞
c(r) r^ A sin kr K C h `
2
1 n o
Z A eKih` (1 K T` )ei(krK 2 ) K eKi(krK 2 )
`p `p
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

2i    
`p 1
K T` ei(krK 2 ) .
`p
Z A eKih` sin kr K (2.35)
2 2i
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

The first term in the curly bracket is just the asymptotic form of the spheri-
cal Bessel function. This would be the entire wavefunction if there were no
potential. The second term is pre-multiplied by the T matrix and represents
the extra outgoing spherical wave resulting from the scattering caused by the
presence of the potential.
In the present elastic scattering case discussed in this chapter the S and
T are just complex numbers. We will see later that for inelastic and reactive
scattering they become full multi-dimensional matrices.
There is another matrix which needs to be included
 here
 for completeness.
 
`p `p
This is the K matrix. This matrix relates cos krK to sin krK in
2 2
the asymptotic form of the wavefunction (see eqn (2.35) and (2.32)). (Note
Mott and Massey11 refer to this matrix by the letter K and call it the reactance
matrix.) It is related to the S matrix by the equation (see ref. 13)

S Z (I C iK) (I K iK)K1 . (2.36)

2.6 Quasi-Bound States and the Lifetime Matrix

On occasions we will wish to discuss scattering resonances. They show up
as sharp peaks in cross sections and generally correspond to quasi-bound
states. These are states of the collision system which became trapped for
a relatively long period of time compared with the normal duration of a
collision. For atom–atom collisions such states come about through the co-
incidence of the scattering energy with a state of the two particle system
that forms a nearly bound quantum level within an inner well created by the
centrifugal potential (see the ` Z 15 curve in Figure 2.3). In such a case the
resonance is termed a “shape” resonance as it results entirely from the shape
of a single potential. In more complex situations, such as arise in atom–
molecule scattering, the resonances may have their origin in an exchange
of energy from relative translational motion to the internal excitation of the
molecule resulting in the system temporarily not possessing sufficient en-
ergy of relative motion to break up. Such resonances are termed “Feschbach”
resonances. Yet another type of resonance which we will encounter in Chap-
ter 6 is a reactive scattering resonance in which the motion of the whole
system is temporarily trapped in a cooperative motion in such a way that
no one atom can escape from the scattering complex.
 View Online

30 Chapter 2

In all cases the key feature of a resonance is the fact that the scattering
partners stay close to each other for a considerably longer period than would
be anticipated. The problem has been elegantly analyzed by Smith14 who
discusses the time the collision partners spend within the collision region
as compared with the time they would spend there in the absence of any
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

interaction between them. This leads him to define a “lifetime matrix”

dS
Q Z Ki~S∗ . (2.37)
dE
The eigenvalues of this matrix give the lifetimes of the states involved. As Q
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

is a Hermitian matrix all its eigenvalues will be real. Some may be negative as
they measure the lifetime compared to that expected without any interaction.
If we denote the eigenvalues of Q by qi , then a resonance is present when
2Eqi /~  1.
For the case of elastic atom–atom scattering when, there is just a sin-
gle “channel” and there are no inelastic or reactive processes, the S matrix
reduces to a single complex number as in eqn (2.33). Substituting this into
eqn (2.37) we obtain‡
dh
Q Z t Z 2~ (2.38)
dE
where t is the lifetime of the resonance at energy E. Some examples of shape
resonances are given in the section below.

2.7 Illustrative Examples: Ne + Ne and ArC + Ar

Scattering
A Lennard-Jones potential energy function15,16 is used as a model for the
Ne + Ne interaction. This is the same model as was used in Chapter 1. We
have written two different computer codes to evaluate the phase shifts. One
of these is based directly on the numerical solution of eqn (2.10) followed by
an analysis of the resulting wavefunction at large internuclear separations.17
This code uses the Numerov18 method to perform the integration. This ap-
proach does not yield the absolute value of the phase shift, but rather results
in phase shifts between Kp/2 and p/2. Careful examination of the formu-
lae shows that this uncertainty in the absolute values of the phase shifts is
not reflected in any ambiguity in the value of the cross section. The second
program we have written enables us to evaluate the absolute value of the
phase shifts. This code is based on the Variable Phase Approach19–21 and is
described in Appendix B.
The calculated phase shifts as a function of collision energy for a model
Ne + Ne collision system are shown in Figure 2.2 while Figure 2.3 shows the
corresponding effective potentials (see eqn (2.10)).

‡ Eqn (2.38) has an extra factor of 2 as compared with Smith’s paper. This arises from the fact
that Smith uses a nonstandard definition of phase shift.
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 31

Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.2 Phase shift as a function of the relative collision energy of two Ne atoms.
The phase shift, divided by p, is shown for four different values of the
orbital angular momentum `. The well depth of the potential is 3 Z
29.52 cmK1 .

Figure 2.3 Effective potential energy as a function of the inter-nuclear separation

of two Ne atoms. The effective potential energy is shown for four dif-
ferent values of the orbital angular momentum `. The well depth of the
potential is 3 Z 29.52 cmK1 .
 View Online

32 Chapter 2

The first thing to note is that the low energy values of the phase shifts on
the left hand side of Figure 2.2 are all different. They are respectively equal to
3p, 2p, p and 0. According to Levinson’s theorem11,22 this indicates that the
corresponding effective potentials (see Figure 2.3) support three, two, one
and zero bound states§ . Figure 2.3 illustrates the fact that as the orbital an-
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

gular momentum, `, increases the depth of the effective potential decreases

correspondingly. Thus for sufficiently large values of ` the effective potential
will not support any bound states and the low energy limit of the phase shift
will be zero.
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

At large collision energies the the repulsive part of the potential dominates
the behavior of the phase shift and it decreases steadily with increasing en-
ergy¶ .19,20 At low collision energies and for ` O 0, as the energy increases from
zero, the wavefunction progressively samples more of the attractive part of
the potential and this results in an increase of the phase shift.
Figure 2.4 shows the computed phase shift as a function of the orbital an-
gular momentum quantum number for three different energies. Although
the orbital angular momentum quantum number takes on only discrete inte-
ger values, for the purposes of visualization we have connected up the values
on the graph using a smooth curve. At low values of the angular momentum
the phase shift is negative. This indicates that the repulsive part of the poten-
tial is dominating the scattering. As the orbital angular momentum increases
the wavefunction is prevented from penetrating to small values of r and sam-
ples the repulsive part of the potential to a progressively smaller extent. So at
larger values of the orbital angular momentum the scattering is determined
to a smaller extent by the repulsive part of the potential and to a larger de-
gree by the attractive part of the potential. This leads to more positive values
of the phase shift. At very large values of the orbital angular momentum the
centrifugal potential prevents the wavefunction from sampling any part of
the potential and the phase shift tends to zero.

2.7.1 Differential and Integral Cross Sections

Figure 2.5 shows the differential cross section for Ne + Ne scattering where we
have treated the Ne atoms as distinguishable and have calculated the cross
sections by the direct application of eqn (2.19). Note that because we have not
taken account of the exchange symmetry in these calculations they should be
regarded as illustrative of general theory and not suitable for direct compari-
son with experimental results. The figure displays cross sections for the same
three collision energies as were shown in the figures of the classical deflec-
tion functions and cross sections in Chapter 1 (see Figures (1.7) and (1.9)).

§ The proof of Levinson’s theorem is complicated but in Appendix B a simple argument is set
out to support its reasonableness.
¶ For a Lennard-Jones 12-6 potential whose inner repulsive wall varies as A/r 12 the phase shift

varies as h ∼ KC k5/6 .
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 33

Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.4 The phase shifts divided by p as a function of the orbital angular momen-
tum quantum number `. The calculated phase shift is shown for three
separate energies.

Figure 1.7 shows that orbiting collisions are present for the two lower ener-
gies examined (E Z 3/2 and E Z 3), but not for the highest energy (E Z 23).
Both the classical and quantum cross sections display backward glories (at
q Z 180◦ ) for the two lower energies but only a forward glory (at q Z 0◦ ) for
the highest energy. The cross sections are plotted on a logarithmic scale to
permit the display of the large glory maxima. As discussed in Chapter 1, the
differential cross section becomes infinite for the classical glories, but we see
from Figure 2.5 that the quantum cross sections yield large but finite values
at the glory maxima.
The most noticeable differences between the classical and quantum cross
sections in Figure 2.5 are the oscillations present in the quantum cross
sections. In an attempt to clarify the origin of these oscillations we have
performed a calculation using only the repulsive part of the Lennard-Jones
potential. These calculations are shown in Figure 2.6. We see that while the
differential cross section computed using only the repulsive part of the po-
tential does show a small amount of oscillatory structure, this is far less
than that for the calculation using the full potential. The backward glory
scattering is also totally absent from the cross section computed with only
 View Online

34 Chapter 2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.5 Differential cross sections for model of Ne + Ne scattering ignoring sym-
metry and indistinguishability of Ne atoms. Also shown are the classical
cross sections.
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 35

Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.6 Differential cross sections for model of Ne + Ne scattering ignoring sym-
metry and the indistinguishability of Ne atoms. The figure shows cross
sections computed with the full Lennard-Jones potential and also using
just its repulsive part. See the Figure 1.5 caption for the Lennard-Jones
parameters.

the repulsive part of the potential. The small amount of oscillatory structure
in the cross section calculated using only the repulsive part of the potential
may be thought of as arising from diffraction effects at pthe edges of a hard
sphere potential of radius approximately equal to R Z s/p (see Figure 1.5
and eqn (1.24)11,12,23 ). From the classical deflection functions (Figure 1.7) we
see that for the lower two energies shown there are multiple classical trajec-
tories that can lead to scattering at all angles. For the highest energy (E Z 23)
there is a rainbow angle at about 62◦ and we see from Figure 2.5 that for this
energy the oscillations for angles beyond 62◦ are damped out. This confirms
that the origin of the quantum oscillations comes from interference effects
between classical trajectories leading to scattering into the same angle.
The classical cross section for E Z 23 shows a rainbow maximum, but this
is not really noticeable in the quantum cross sections. The reason for this
arises from the weak interactions in the Ne + Ne case and consequently in
the very low collision energies we are examining. This leads to very marked
quantum oscillations which mask the typical signatures of rainbow scatter-
ing in this case. Figure 2.7 shows differential cross sections computed using
a Lennard-Jones potential with parameters derived for ArC C Ar scattering.24
Here we see the main rainbow maximum at around 83◦ . It is accompanied
by many so-called supernumary rainbows at smaller angles. There are many
rapid quantum oscillations underlying the rainbow maxima. Note that the
potential used (see caption to Figure 2.7) has a far deeper well and the masses
of the atoms are far greater.
The integral cross section for Ne + Ne, calculated without taking account
of the indistinguishability of the two Ne atoms, is shown in Figure 2.8. The
 View Online

36 Chapter 2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.7 Calculated differential cross section using potential parameters derived
for ArC C Ar scattering.24 The cross sections shows rainbow and super-
numary rainbow maxima. A Lennard-Jones potential with parameters
3 Z 1.25 eV and s Z 2.17 Å was used in the calculations.

Figure 2.8 Calculated integral cross section for a model of Ne + Ne scattering ignor-
ing the indistinguishability of the two Ne atoms. See caption to Figure 1.5
for the Lennard-Jones parameters.

cross section displays considerable structure. In all there are eight sharp
features superimposed on a broader peak centered at around 17 cmK1 .
Seven of the features give rise to sharp peaks and one to a small shoul-
der. All eight marked features arise from “shape resonances”, which are
associated with quasi-bound vibrational states trapped within the effective
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 37

potential (see Figure 2.3). The peaks are located at 2.25, 3.54, 4.96, 8.16,
11.21, 14.36, 17.71 and 21.35 cmK1 and arise from shape resonances asso-
ciated with orbital angular momenta ` Z 8, 9, 10, (11 & 13), (12 & 14), 15,
(16 & 14) and 17 respectively. Note that the shape resonance associated with
` Z 14 contributes to two different sharp peaks. We will discuss the ` Z 12
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

shape resonance and its contribution to the integral cross section in detail
below.
From Figure 2.8 the “shape resonances” are seen to be located on a broad
peak in what appears to be part of an oscillatory pattern. There is a minimum
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

at around 2.8 cmK1 , followed by the broad peak at 17 cmK1 and subsequently
by another minimum at 47 cmK1 . To understand the nature of these os-
cillations in the integral cross section we need to look at Figure 2.9 which
illustrates the dependence of the phase shifts on ` at three different scatter-
ing energies. It is seen that for a collision energy of 17 cmK1 , when the cross
section is at the maximum of an oscillatory peak, the phase shift reaches
3p
its maximum value of around for ` Z 12. The classical deflection func-
2
 1 
tion written as a function of ` `C ~ ≈ b ! k~ passes through zero
2
at the same value of `. We commented in Chapter 1 (see Figures 1.7 and
1.8 and the ensuing discussion) that the classical cross section displayed a
glory maximum when the deflection function passed through zero. From a
quantum mechanical perspective when the phase shift, plotted as a func-
tion of `, passes through a maximum and that maximum value is equal to
(2n C 1)p
so that sin2 (h` ) is also maximal, then several ` values will make
2

Figure 2.9 The phase shifts (solid lines) and the classic deflection functions (broken
lines) versus angular orbital number calculated for a model of Ne + Ne
scattering.
 View Online

38 Chapter 2

the largest possible contribution to the integral cross section at that energy,
leading to a glory maximum (see eqn (2.31)). In contrast, at energies of 2.8
and 47.12 cmK1 the cross sections have glory minima because the maximum
2np
phase shift in the phase shift ∼ ` plots occurs at a value close to h` ≈ .
2
2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

Thus a large number of ` values with a minimal, near zero, value of sin (h` )
contribute to the integral cross section and a glory minimum results. By the
same logic we expect that another two broad maxima in integral cross sec-
tion should exist, one for an energy higher than 47 cmK1 and another for an
energy lower than 2.8 cmK1 , when the phase shift maxima will be located
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

p 5p
around and . We recall that the maximum possible value of the phase
2 2
shift for Ne + Ne scattering is equal to 3p (see Figure 2.2) and occurs for k Z 0
and ` Z 0. This means that we can expect exactly three glory maxima for this
system.25
Figure 2.10 shows the integral cross section for ArC C Ar collisions. In this
case we see a much larger number of glory oscillations. The number of these
oscillations has been shown by Bernstein26 to be equal to the number of
bound states supported by the potential. This is just three for the Ne + Ne
case, but much larger for ArC C Ar scattering. “Shape resonance” features
are superimposed on some of the glory oscillations. One of these is shown in
the inset in Figure 2.10.

Figure 2.10 Calculated integral cross section for model of ArC C Ar scattering. See
caption to Figure 2.7 for the Lennard-Jones parameters.
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 39

2.7.2 Shape Resonances

The very sharp resonance at 4.96 cmK1 in Figure 2.8 is found to arise from
a quasi-bound state for the effective potential with ` Z 12. Figure 2.11 shows
the wavefunction for this resonance superimposed on the effective potential.
The plot of the wavefunction has been shifted upwards so that its zero value
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

lies at the energy of the resonance. We see clearly that the amplitude of the
wavefunction is very large in the region of the potential well but that it tun-
nels through the centrifugal barrier and “leaks” out. As the wavefunction has
no nodes within the well region, it is clearly the first and only quasi-bound
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

state supported by the effective potential. The phase shift for ` Z 12 is shown
by a red line in Figure 2.12 as a function of energy. We see clearly that as the
energy increases through the resonance at E Z 4.96 cmK1 the phase shift in-
creases sharply by p. The energy interval over which the phase shift changes
sharply is designated by G in the figure. This is the “width” of the resonance.
The blue line in Figure 2.12 shows the contribution to the integral cross sec-
tion (see Figure 2.8 and eqn (2.31)) of the partial cross section for ` Z 12 . We
see from the figure that the resonance contribution is sharply peaked just be-
low the energy of the quasi-bound energy level. The width of the resonance,
G, coincides with the energy region over which the resonance contributes
significantly to the cross section.
The S matrix is defined as S Z exp 2ih` (see eqn (2.33)), consequently if we
draw an Argand diagram, i.e. ReS versus ImS, of the S matrix then as the

Figure 2.11 Wavefunction for Ne + Ne shape resonance at 4.96 cmK1 . The shape res-
onance occurs for ` Z 12. The wavefunction is superimposed on a plot
of the effective potential. See caption to Figure 1.5 for the Lennard-Jones
potential parameters.
 View Online

40 Chapter 2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.12 The phase shift for `Z12 divided by p, for Ne + Ne scattering as a
function of collision energy in the vicinity of the shape resonance at
4.96 cmK1 is shown by a red line. The contribution of orbital angu-
lar momentum quantum number ` Z 12 to the integral cross section
is shown as a blue line.

Figure 2.13 An Argand diagram of the S matrix for relative collision energies around
the shape resonance at 4.96 cmK1 .
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 41

energy increases through the resonance the Argand diagram will map out a
circle going counter-clockwise with energy. This is illustrated in Figure 2.13.
This behavior may be generalized to more complex situations including
collisions involving more than two atoms.27,28
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

2.7.3 Exchange Symmetry

Section 2.4 discussed the formulae applicable in the case of the collision
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

of two indistinguishable atoms, such as Ne + Ne. In the figures displaced

above we have treated the two Ne atoms as distinguishable. This provided
a good illustration of the general formalism and of many aspects of atom–
atom scattering. In Figure 2.14 we see a comparison of the differential cross
sections computed when taking full account of the exchange symmetry and
when not taking it into account. The cross sections seem quite similar, but at
larger collision energies when a far greater number of partial waves (` values)
contribute, they become very noticeably different.16
Figure 2.15 compares the integral cross sections computed with and with-
out taking account of the exchange symmetry. In this case, when proper
account is taken of the exchange symmetry, there are fewer sharp shape res-
onance peaks because only even ` values contribute and consequently the
shape resonances arising from odd ` values are absent. The remaining peaks
are doubled in amplitude in accordance with eqn (2.31).

Figure 2.14 Comparison of differential cross sections for Ne + Ne scattering com-

puted both taking account of exchange symmetry of indistinguishable
atoms and ignoring this symmetry.
 View Online

42 Chapter 2
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Figure 2.15 Comparison of integral cross sections for Ne + Ne scattering computed

both taking account of exchange symmetry of indistinguishable atoms
and ignoring this symmetry.

2.8 Extraction of Interaction Potentials

from Crossed Beam Experiments
In the sections above we have illustrated the integral and differential cross
sections which may arise in the case of atom–atom scattering. The details of
the cross sections are very sensitive to the potential energy function govern-
ing the interaction of the collision partners and consequently measurement
and subsequent modeling of the cross sections provides a reliable method of
determining the interaction potential. Several reviews have been written on
this subject and the reader is referred to them.16,29–34

2.9 Low Energy Scattering and the Scattering Length

At very low energies the centrifugal potential creates an impenetrable barrier
for all orbital angular momentum quantum numbers ` O 0 (see eqn (2.10)
and Figure 2.3). Because of this only orbital angular momentum ` Z 0 con-
tributes to the scattering. Let us consider a hard sphere of radius a. The
wavefunction has an asymptotic form of sin(krCh0 ) (see eqn (2.11)). To find
the phase shift h0 we need to compare the wavefunction with that for free
motion (sin(kr)). The presence of the hard sphere moves the effective origin
outwards by a distance a leading to a phase shift of h0 Z Kka. So the equation
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 43

for the integral cross section becomes (see eqn (2.23))

4p
sZ sin2 h0
k2
4p
Z 2 sin2 (ka). (2.39)
k
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

In the limit k → 0 we may therefore write

lim 2
k→0 {s(k)} Z 4pa . (2.40)
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

Note that the quantum mechanical cross section is four times as large as the
classical cross section.
Here we can identify a as the scattering length or the range of the potential.
More generally, the scattering length, a, is defined as (see ref. 35)
1 lim {k cot (h (k))} .
K Z k→0 0 (2.41)
a
The concept of scattering length is widely used especially in low tempera-
ture physics of Bose–Einstein condensates.

References
1. L. I. Schiff, Quantum Mechanics, McGraw-Hill Book Co. Inc., New York,
1955.
2. L. Pauling and E. B. Wilson, Introduction to Quantum Mechanics, McGraw-
Hill Book Co., 1935.
3. H. Eyring, J. Walter and G. E. Kimball, Quantum Chemistry, John Wiley
and Sons Inc., 1st edn, 1944.
4. R. D. Levine, Quantum Mechanics of Molecular Rate Processes, Dover
Publications, Oxford, 2nd edn, 1999.
5. L. D. Landau and E. M. Lifshitz, Quantum Mechanics: Non-Relativistic
Theory, Butterworth-Heinemann, Oxford, 3rd edn, 1977 (Reprinted with
corrections 1991).
6. A. Messiah, Quantum Mechanics, North Holland Publishing Co., Amster-
dam, 1965.
7. A. R. Edmonds, Angular Momentum in Quantum Mechanics, Princeton
University Press, Princeton, New Jersey, 1960.
8. R. N. Zare, Angular Momentum, John Wiley and Sons, New York,
1988. See also list of corrections at web address: http://web.
stanford.edu/group/Zarelab/Errata.pdf.
9. M. Abromowitz and I. A. Stegun, Handbook on Mathematical Functions,
Applied Mathematics Series 55, National Bureau of Standards, 10th edn,
1972.
10. P. M. Morse and H. Feschbach, Methods of Theoretical Physics, McGraw-
Hill Book Co. Inc., New York, 1st edn, 1953. See Section 9.3.
11. N. F. Mott and H. S. W. Massey, The Theory of Atomic Collisions, Claren-
don Press, Oxford, 2nd edn, 1949.
 View Online

44 Chapter 2

12. H. S. W. Massey and C. B. O. Mohr, Free paths and transport phenomena

in gases and the quantum theory of collisions. I. The rigid sphere model,
Proc. R. Soc. A, 1933, 141, 434.
13. R. G. Newton, Scattering Theory of Waves and Particle, David and Charles,
2nd edn, 1982.
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

14. F. T. Smith, Lifetime matrix in collision theory, Phys. Rev., 1960, 118, 349.
15. D. M. Hirst, Potential Energy Surfaces, Molecular structure and dynamics,
Taylor and Francis, 1985.
16. F. Pirani, S. Brizi, L. F. Roncaratti, P. Casavecchia, D. Cappelletti and
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

F. Vecchiocattivi, Beyond the Lennard-Jones model: a simple and accu-

rate potential function probed by high resolution scattering data useful
for molecular dynamics simulations, Phys. Chem. Chem. Phys., 2008, 10,
5489.
17. R. B. Bernstein, Quantum mechanical (phase shift) analysis of differ-
ential elastic scattering of molecular beams, J. Chem. Phys., 1960, 33,
795.
18. B. Numerov, Note on the numerical integration of d2 x/dt2 Z f (x, t),
Astron. Nachr., 1927, 230, 359.
19. F. Calogero, Variable Phase Approach to Potential Scattering, Academic
Press, New York, 1967.
20. F. Calogero, Approximation for the phase shifts produced by repulsive
potentials strongly singular in the origin, Phys. Rev., 1964, 135, B693.
21. V. Babikov, Variable Phase Approach in Quantum Mechanics, Nauka,
Moscow, 1968 [in Russian].
22. N. Levinson, On the uniqueness of the potential in a Schrodinger equa-
tion for a given asymptotic phase, K. Dan. Vid. Selsk. Mat-Fys. Medd., 1949,
25, 1.
23. M. S. Child, Molecular Collision Theory, Academic Press, 1974. Corrected
Dover Publication, 1996.
24. D. C. Lorents, R. E. Olson and G. M. Conklin, Rainbow scattering for
ArC + Ar and XeC + Xe, Chem. Phys. Lett., 1973, 20, 589.
25. R. B. Bernstein, Semiclassical analysis of the extrema in the velocity de-
pendence of total elastic-scattering cross sections: relation to the bound
states, J. Chem. Phys., 1963, 38, 2599.
26. F. A. Morse and R. B. Bernstein, Velocity dependence of the differential
cross sections for the scattering of atomic beams of K and Cs by Hg, J.
Chem. Phys., 1962, 37, 2019.
27. R. D. Levine, M. Shapiro and B. R. Johnson, Transition probabilities in
molecular collisions: computational studies of rotational excitation, J.
Chem. Phys., 1970, 52, 1755.
28. B. C. Garrett, D. G. Truhlar, R. S. Grev, G. C. Schatz and R. B. Walker,
Reaction probabilities, resonances, and thermal rate constants for the
collinear reactions H + FH and D + FD on a low-barrier surface. Close-
coupling and tunneling calculations, variational transition-state theory,
and the unified statistical model, J. Phys. Chem., 1981, 85, 3806.
29. U. Buck, Elastic scattering, Adv. Chem. Phys., 1975, 30, 313.
 View Online

Quantum Theory of Atom–Atom Elastic Scattering 45

30. U. Buck, Inversion of molecular scattering data, Rev. Mod. Phys., 1974,
46, 369.
31. J. P. Toennies, W. Welz and G. Wolf, The determination of the H–He
potential well depth from low energy elestic scattering, Chem. Phys. Lett.,
1976, 44, 5.
Published on 02 July 2015 on https://pubs.rsc.org | doi:10.1039/9781782620198-00019

32. P. E. Siska, J. M. Parson, T. P. Schafer and Y. T. Lee, Intermolecular po-

tentials from crossed beam differential elastic scattering measurements.
III. He + He and Ne + Ne, J. Chem. Phys., 1971, 55, 5762.
33. M. Shapiro and R. B. Gerber, Extraction of interaction potentials from
Downloaded by University of Delaware on 4/21/2020 9:19:17 PM.

the elastic scattering amplitudes: an accurate quantum mechanical

procedure, Chem. Phys., 1976, 13, 235.
34. M. S. Child and R. B. Gerber, Inversion of inelastic atom–atom scattering
data: recovery of the interaction function, Mol. Phys., 1979, 38, 421.
35. T. y. Wu and T. Omura, Quantum Theory of Scattering, Prentice-Hall,
1962.