You are on page 1of 7

Diffraction

and interference in QED


Andrew Daniels,1 Jordan Kupec,2 Timothy Baker,1 Eduardo V. Flores1



1Department of Physics, Rowan University, Glassboro, NJ
2Department of Physics, College of the Holy Cross, Worcester, Ma

We model a wire as a potential barrier and calculate diffraction from that potential using quantum
electrodynamics. We compare our results with classical Fraunhofer diffraction. We find general agreement
between the quantum and the classical results; however, we find that the classical approach overestimates the
wire radius. We consider an incoming electron beam diffracting from the potential. We also consider the case of
an incoming photon beam. For the photon case we only indicate the amplitudes that need to be evaluated
numerically. We study a case of interference in quantum electrodynamics: two beams in phase interfere and
diffract from the wire-like potential.



I. INTRODUCTION

Classical Optics has developed different tools to calculate diffraction for different setups [1]. On the other
hand, beyond path integral approaches for some particular examples [2], one cannot find in the literature
examples of diffraction calculations using the Feynman diagram approach in quantum electrodynamics
(QED). The reason is that QED deals with interactions among elementary particles and an obstacle such as a
wire contains a macroscopic number of elementary particles. However, if the macroscopic object could be at
least partially modeled by a static potential then the situation changes. In this paper we consider diffraction,
or more accurately scattering, of electrons and photons from a potential using Feynman diagrams [3,4].
Scattering of electrons from static potentials is a simple interaction at the lowest order in QED. Scattering
of photons from static potentials is also possible but it is complicated by the fact that photons only interact
indirectly with other photons. We model a wire as a cylindrical potential barrier. This model ignores
important surface properties. Nevertheless, the key aspect of this potential is its geometrical resemblance to
the wire.
It is possible to interpret a diffraction pattern as a probability distribution; this approach is probably the
easiest way to compare the results from QED with classical results. In QED the amplitude or transition matrix
may be obtained from Feynman diagrams. We multiply the absolute square of the amplitude by the number of
final states and obtain the transition rate. From the transition rate we directly obtain a probability
distribution. We compare the normalized probability distribution from QED with a normalized classical
diffraction distribution using Fraunhofer approach [1].
In QED we are normally interested in scattering and we pay less attention to particles that go through
undeflected. This is not the case in classical diffraction where we consider the full diffraction pattern.
However, using a semi-classical approach to diffraction it is possible to separate the photons that are
deflected from those that go through undeflected. We show this by first considering Babinets principle
applied to wire diffraction [1]. The undisturbed electric field ! of an original beam equals the electric field
! produced by the same beam in the presence of the wire plus the electric field that would be produced by
the beam in the presence of a hypothetical, complementary slit ! :

! = ! + ! .
(1)

Outside the beam the original field is zero, ! = 0, thus, according to Eq. (1) we have that the fields produced by
the wire and the slit are similar except for sign, ! = ! . We conclude that outside the beam, wire
diffraction and complementary slit diffraction are identical as may be seen in Fig. 1.

Intensity
2.0

1.5

1.0

0.5

-0.06

-0.04

0.02

-0.02

0.04

0.06

FIG. 1. Diffraction pattern produced by a wire and by its complementary slit. Both diffraction patterns are calculated using the
Fraunhofer approximation. The wire diameter is 17 m. The wire diffraction curve in dashed orange has been calculated using
a laser beam with Gaussian profile of radius 0.5 mm at a wavelength of 63310!! nm; it shows two dark spots at the edge of
the central beam [5]. The curve that shows slit diffraction is in solid blue. We note that outside the central beam both curves
are indistinguishable, as predicted by Babinets principle.


To see how the slit diffraction pattern represents wire diffraction within the beam we consider the following semi-
classical analysis. Photon number conservation requires that

! = !" + !" ,
(2)

where ! is the total number of photons, !" is the number of photons that go past the wire and !" is the number of
photons stopped by the dark wire. Since the wire and the slit are complementary constructs, the number of photons that
go through the hypothetical slit, !"!" , is the same as the number of photons that are stopped by the wire,

!"#$ = !" .
(3)

From Eq. (1) we obtain
!
!
= !! 2! ! + !! .
(4)

The number of photons that go pass the wire is proportional to the integral of the field intensity in Eq. (4) evaluated on a
far away surface, large enough to catch all the photons,

!" = ! 2

! ! + !"#$ .
(5)


Eq. (2) together with Eq. (3) and Eq. (5) results in the identity

!" =

! ! .

(6)

We note that ! , the field of the original beams, is non-zero in a small region entirely contained within the detector
range. Integrating Eq. (4) over region that covers just the original beam region the and using Eq. (6) we have an
expression for the number of photons in the main beam
= ! 2!" +

!! .
(7)


We note that the term

!! , where ! is the field produced by the slit, represents the number of photons diffracted

by the wire that fall within the beam. We conclude that !! accounts for wire diffraction outside ! = ! and
inside the original beam. Therefore, wire scattering effects from QED will be compared with classical slit diffraction.

II. SINGLE BEAM CASE

Consider a beam of electrons travelling along the z-axis and a long wire centered at the origin along the y-
axis as in Fig. 2a. In QED the lowest order contribution to electron scattering in terms of Feynman diagrams is
shown in Fig. (2b).


a)
b)
!
!


x

!


z
!


Wire


FIG. 2. a) An electron moving along the z-axis is deflected by a wire located at the origin along the y-axis. b) This Feynman
diagram represents an electron scattering from an external field.


The amplitude for the process in Fig. 2b is represented by the second term in the matrix element [3]

!" = !"

! ! ! ! ! ,

(8)

where ! represents an incoming electron, ! represents the scattered electron, ! represents the external
field and the delta function stands for the particles that go through undeflected. The external field is


= 0 .
! =

0 0
(9)

We model the wire as a potential barrier:

0
=
.
0
>
(10)

The time independence of potential in Eq. (10) yields the delta function 2 ! . Similarly, independence
of the potential on the y-coordinate yields the delta function 2 !" !" . Using standard techniques [3]
!

we get the transition probability per particle !" ! ! , which is proportional to



!

+ ! ! + ! ! cos !
!! ,
+ !

(11)

where ! ! is a hypergeometric regularized function; this function is the Fourier transform of the potential in
Eq. (10). We note that if the initial spin is up and the final spin is down we get zero amplitude, thus, diffraction
from static potentials does not flip the electron spin.
Our interest is in diffraction at energies low compared to the electron mass. In this limit Eq. (11) becomes

!
!
= ! ! 2, ! ! sin!
.

2
(12)

To obtain C we integrate Eq. (12) and equate it to 1. We call ! / in Eq. (12) the transition probability
distribution. We compare Eq. (12) with Fraunhofer diffraction for the slit, which is proportional to

sinc ! sin .

We note that QED treats the wire as a 3D object while in the Fraunhofer approach the wire is a 2D object [6];
this might result in an overestimation of the wire radius [7] in Fraunhofer diffraction as shown in Fig. 3.

Intensity
1.2
1.0
0.8
0.6
0.4
0.2
-0.10

0.05

-0.05

0.10

FIG. 3. Classical and quantum wire diffraction. The orange dashed curve represents classical wire diffraction. The solid blue
curve represents quantum diffraction. The net area under each graph is the same. We note that first dark point for the
quantum case is shifted outward from the classical case, which indicates that the Fraunhofer approach overestimates the wire
diameter compared to the QED approach.


We note that if in the quantum approach we increase the wire radius by a factor of 1.21 the first dark
spots in Fig. 3 line up. Experimentally we have verified this in the lab; a direct measurement of the wire
diameter gives a value of 14 m while a measurement obtained through Fraunhofer diffraction gives 17 m,
which is 1.21 times larger than the actual value. Thus, QED provides the more realistic value.
Photon diffraction from a static potential is scattering of light by light [8,9]. The lowest level contribution
comes from a set of six Feynman diagrams [8,9,10] as the one in Fig. 4a.


!
!
a)
b)

x


z

!
!

/2 /2


!


FIG. 4. a) There are six Feynman diagrams for photon scattering from a static potential but they are all related to this diagram.
The photon interacts with the potential through a virtual electron loop. b) Notation: the incoming photon has 4-momentum !
and the scattered photon has 4-momentum ! . The scattering angle is . The wire is placed along the y-axis. The 4-momenta !
and ! are the momenta exchanged with the wire.

In QED the process in Fig. 4a is closely related to Delbruck scattering first observed in 1973 [11]. In
Delbruck scattering, a photon is scattered by the Coulomb potential of the nucleus of an atom. With some
modifications in notation, shown in Fig. 4b, we adopt the results for Delbruck scattering in Ref. 10. This is
possible because the authors of Ref. 10 approach to Delbruck scattering is general.
The amplitudes for scattering circular polarized photons from a static potential [9,10] can be written as

M!! = M!! =

! ! !

! !

+ ! ! ! + ! ! ! ,
(13)

M = M =

! ! ! [ ! ! ! ! ! ,

(14)

where the sign + (-) stands for right (left) circular polarized light, represents the Fourier transform of the
!
potential, and , ! , and ! are functions of the internal and external momenta in terms of rational,
logarithm and dilogarithm functions [10]. The magnitude square of the momentum of the two virtual photons
is ! and ! . The integral represents the virtual electron loop.
The Fourier transform, (), of the potential that represents the wire is the same as in Eq. (11). This
function is finite and positive for small relative to the electron mass; at higher values the function is small
and highly oscillatory. This means that only small values of the internal momentum in the integral in Eqs. (13)
and (14) are significant. For wire diffraction at energies small compared to the electron mass we use the
!
approximations given in Ref. 10 for the functions , ! , and ! ; however, these approximations contain
negative powers of , secant and cosecant functions of which are problematic when approaches 0 and
when is integrated from 0 to 2 respectively. Therefore, at any given order of external momentum, one
must make sure those singular terms in Eqs. (13) and (14) are canceled.

III. TWO BEAM CASE

In this section we explore wire diffraction when a wire is at the beam intersection as in Fig. 5. Two beams
in phase cross each other at an angle .

!

!!
x




2

z



2

Wire
!!


FIG. 5. Two beams with a constant phase difference intersect at a small angle . A thin wire is scanned across the beam
intersection. The scattering angle is .


Here we consider the electron case; the photon case is similar. The matrix element is the linear superposition:

!" =

! ! ! ! !! + !! ,

(15)

where ! ! , ! !!!!! . The superposition of amplitudes in Eq. (15) is interference in QED [12].
Using standard techniques [3] we obtain the transition probability per particle, which is proportional to

1
1
+ !

+ ! ! sin sin !
2
2

! !
! !
! !
+
+ cos cos sin sin ! !!
2
2

+ !

+ ! ! cos cos

.
(16)


where ! = ! ! 2, ! sin!

!
!

!
!

! and ! = ! ! 2, ! sin!

!
!

!
!

! . We note that in order to apply

arbitrary shifts to the interference pattern we have introduced the phase .


At energies much smaller than the electron mass Eq. (16) becomes

= ! ! + 2! ! cos + ! ! ,

(17)

where is a normalization constant. In Fig. 6 we plot Eq. (17) together with Fraunhofer diffraction for a slit
illuminated by the two intersecting beams; the wire is placed at the center of a bright fringe, = 0. Once
again Fraunhofer diffraction overestimates the radius of the wire.

FIG. 6. The quantum diffraction curve for the electron is in solid blue. Fraunhofer diffraction curve is in dashed orange. The
beams intersect at = 0.1. The wire is at the center of a bright fringe.


We now observe the effect of the wire as we scan it across the beam intersection. We note that changing
the phase is equivalent to scanning the wire across the beam intersection while keeping the wire at the
origin. In Fig. 7 we plot the probability distribution in Eq. (17) as a function of scattering angle and phase
difference . As we scan the wire across the beam intersection we observe the presence of constructive and
destructive interference. When the wire is at region of destructive interference the distribution reaches a
minimum due to a decrease of diffracted photons.

FIG. 7. The wire is scanned across the beam intersection of two beams that intersect at angle = 0.1. The angle is the
scattering angle. Looking along phase difference axis we see a periodic increase and decrease of diffracted light as we reach
a bright and dark fringe respectively.


Wire diffraction, at low energies, with two interfering photon beams should result in a probability
distribution similar to the electron case in Eq. (17). An analysis of the incoming and outgoing photon
momentum shows that results for the single beam of photons in Eqs. (13) and (14) may be used with one
!
!
!
modification; the scattering angle changes from , where the sign + is for the + beam and the sign -
!
!
!
is for the - beam in Fig. 5. For instance, the probability distribution when light changes circular polarization is

! !!
!
= M!! ! + M!! ! !!! ,


where M!! ! is the single beam amplitude in Eq. (14) to change a right handed photon into a left handed
photon evaluated at

VI. REFERENCES

!
!

!
!

and M!!

!
! is the same amplitude evaluated at !

1.
2.
3.

!
!

Hecht, E., and A. Zajac. Optics. Reading: Addison - Wesley, 1979. Print.
Beau, M., Feynman Integral and one/two slits electrons diffraction: an analytic study, arxiv.org/pdf/1110.2346
Bjorken, James D., and Sidney D. Drell. International Series In Pure And Applied Physics: Relativistic Quantum Mechanics. New
York City: McGraw - Hil, 1964. Print.
4.
Griffiths, David. Introduction to Elementary Particles. 2nd ed. Vol. 1. Weinheim: WILEY-VCH Verlag GmbH & KGaA, 2010. Print.
5.
Ganci, S., Fraunhofer diffraction by a thin wire and Babinets principle, Am. J. Phys. 73, 83 (2005)
6.
Martinez-Anton, J. C., Serroukh, I., & Bernabeu, E., Laser diffraction wire diameter measurements: correction of diffraction
models by interferometric calibration, Proceedings of SPIE - The International Society for Optical Engineering September
(1999)
7.
Vyas, K. & Rao Lolla, K., High resolution diameter estimation of microthin wires by a novel 3D diffraction model, Proc.
SPIE 8133, Dimensional Optical Metrology and Inspection for Practical Applications, 813311 (September 13, 2011)
8.
Karplus, R. & Newman, Non-Linear Interactions between Electromagnetic Fields, M. Phys. Rev. 80, 380 (1950)
9.
Karplus, R. & Newman, The Scattering of Light by Light, M. Phys. Rev. 83, 776 (1951)
10. Costantini, V., De Tollis, B. & Pistoni, G., Nonlinear Effects in Quantum Electrodynamics, Nuovo Cimento, 2, 733 (1971)
11. Jarlskog, G., Jnsson, L., Prnster, S., Schulz , H. D., Willutzki , H. J., & Winter, G. G., Measurement of Delbrck Scattering and
Observation of Photon Splitting at High Energies, Phys. Rev. D 8, 3813 (1973)
12. Veltman, M. J. G., Facts and Mysteries in Elementary Particle Physics, World Scientific Publishing Co. Pte. Ltd. (2003) p. 254.