A simple algorithm for the transport of gamma rays in a medium

F. Arqueros and G. D. Montesinos

Departamento de Fsica Atomica, Molecular y Nuclear, Facultad de Ciencias Fsicas,
Universidad Complutense, E-28040 Madrid, Spain
Received 29 November 2001; accepted 31 July 2002
A simple Monte Carlo algorithm for the simulation of the passage of rays of about 1 MeV in a
medium is presented. In this energy range the only relevant processes are Compton scattering and
photoelectric absorption. The algorithm allows the visualization of the photon tracks as well as the
calculation of many quantities of interest. Several problems for a layer and a cylinder are proposed.
In particular, the energy transferred to electrons in a water cylinder as a function of depth and the
line shape of a NaI scintillator is calculated. 2003 American Association of Physics Teachers.
DOI: 10.1119/1.1509416

I. INTRODUCTION the energy range of interest the probability of coherent scat-

tering or Rayleigh scattering, that is, the elastic scattering of
Understanding the effects of the passage of ionizing radia- a photon with the entire atom is very low. In addition, the
tion through matter is important in several fields, particularly energy threshold for a ray to be converted into an
those dealing with radiation protection and measurement. electronpositron pair is twice the rest mass of the electron
Therefore, the phenomenology of radiation-matter interac- 1.02 MeV. In the energy range around 1 MeV incoherent
tions are included in many undergraduate physics programs. scattering or inelastic scattering with an atom is dominant.
The interaction of an ionizing particle with a single atom In this process the photon transfers energy to an atomic elec-
follows well established laws. In particular, a ray can either tron which is ejected. At smaller energies the photon can also
be absorbed due to the photoelectric effect, scattered by an undergo a photoelectric absorption by an atom, which emits
atom, or converted into an electronpositron pair in the field an electron. The subsequent atomic de-excitation takes place
of an atom. However, the effects of a -ray beam passing by emission of x-ray photons and/or Auger electrons. In both
through a medium are usually the result of many individual processes, incoherent scattering and the photoelectric effect,
interactions because a photon can interact many times before the energy is transferred to atomic electrons which are
it is absorbed or escapes from the medium. In addition, most stopped in a very short distance in comparison to the mean
of the above processes release kinetic energy to electrons, free path of the rays.6
which also interact with the medium. Incoherent scattering can be approximately described by
The most reliable way to predict the effects of a -ray the Compton effect,3,4 that is, the elastic scattering of a pho-
beam is by a Monte Carlo simulation. Because fast comput- ton with a free electron Fig. 1. From conservation of energy
ers are available, many codes have been developed that allow and momentum, the following relations can be easily de-
a detailed simulation of the passage of radiation through mat- duced between the energy of the incident photon, E 0 , the
ter. For instance, the EGS4 code is successfully used in par- energy of the scattered photon, E, and the deviation angle of
ticle physics as well as in medical physics.1
the photon ,
Studying the structure of one of these simulation codes
would be very helpful for obtaining a deeper knowledge of E
2 1
the various processes and how they are related to each other. ,
Unfortunately, the professional programs are extremely com- mc 1 1cos
plex for an undergraduate student even at the user level.
However, in many cases it is possible to use simple algo- where the parameter is the incident energy in electron mass
rithms that can be understood and programmed by under- units, E 0 /mc 2 . The energy T transferred to the electron
graduate students. For example, Williamson and Duncan2 ranges from 0 at 0) to a maximum value of
have developed a simple Monte Carlo calculation for the
transport of nonrelativistic electrons which allows the stu- 2
T maxE 0 2
dent to visualize the electron tracks as well as do interesting 12
calculations. In this paper we propose a simple algorithm for
the simulation of the passage of rays in the energy range for backscattered rays at 180 degrees.
around 1 MeV through a medium. This energy range is very The probability distribution of is given by the well-
interesting because it corresponds to that of rays emitted known KleinNishina cross section3,4
by most radioactive sources. Some applications of the code
in radiation dosimetry and nuclear instrumentation will be d KN r 2e 1
discussed.
d 2 1 1cos 2

II. THE PROBLEM

1cos2
2 1cos 2
1 1cos , 3
The main physical processes associated with the transport
of rays through matter are coherent scattering, incoherent which integrated over the solid angle gives a total cross
scattering, the photoelectric effect, and pair creation.35 In section

38 Am. J. Phys. 71 1, January 2003 http://ojps.aip.org/ajp/ 2003 American Association of Physics Teachers 38
Fig. 1. Compton effect. A ray of energy E 0 is scattered by a free electron
reducing its energy to E. The electron is emitted with energy E 0 E.

Fig. 3. Cross sections per NaI molecule summation of the cross sections of

the corresponding atoms. See the caption of Fig. 2 for labels.
1 2 1 ln 12
KN2 r 2e
2 12

tables would make our program more complicated. Instead,
ln 12 13 we have used the following approximation for the photoelec-
, 4
2 12 2 tric cross section:12
where r e 2.821015 m is the classical electron radius. If
we neglect the binding energy of the atomic electrons, the
PE 0 b 0
b1 b2

2
f Z , 5
cross section for incoherent scattering for an atom is given
by multiplying the KleinNishina cross section by the where is the fine structure constant, and 0 is given by
atomic number Z. The effect of the binding energy can be
taken into account by reducing the Z value by an amount that
increases with atomic number and decreases with both in-
0 4 r 2e Z 4
Z

. 6

creasing E 0 and . At about 1 MeV this correction is negli- The values of b 0 , b 1 , b 2 , and f ( Z) have been
gible for angles higher than a few degrees, even for heavy tabulated12 as a function of Z. Fourteen values of Z
elements.7 In Fig. 2 the value of Z KN is plotted versus E 0 ranging from 0.00 to 0.70 are sufficient to obtain an ap-
for lead along with the corresponding accurate incoherent proximate value of PE . Therefore, a simple algorithm for
cross sections5 for comparison. Figure 3 shows the cross sec- the calculation of the photoelectric cross section for any el-
tion per molecule for NaI that is (Z NaZ I) KN]. These ement can be carried out by implementing a very small table
figures show that in the energy range where incoherent scat- containing 14 sets of four parameters each b 0 , b 1 , b 2 , and
tering dominates, the correction is negligible, and in our al- f ( Z)]. In Figs. 2 and 3 we have plotted the values of PE
gorithm the KleinNishina formulas, Eqs. 3 and 4 have given by Eq. 5 together with the accurate one from Storm
been used. and Israel.8 We see that the Messel and Crawford approxi-
A rigorous treatment of the photoelectric effect leads to a mation in Ref. 12 underestimates the cross section for ener-
cross section PE that cannot be expressed by simple analyti- gies beyond the K edge. As will be discussed later, for heavy
cal formulas. Therefore, photoelectric cross sections are usu- elements, a lower photoelectric cross section leads to longer
ally tabulated for E 0 and Z. See, for instance, Refs. 5 and path lengths of the photons. However, for a particular ele-
8 11. The implementation in our simple algorithm of these ment or compound, it is possible to find a simple law that
provides an accurate value5,8 11 of PE in the energy range of
interest of our simulations see Sec. IV B.

III. THE SIMULATION

A. The ingredients
When a photon is in a medium, it traverses a certain dis-
tance s in a straight line before it interacts with an atom. As
is well known, the magnitude of s follows the statistical
distribution3
1 s/
p s ds e ds, 7

with a mean free path
1
s , 8
Fig. 2. Cross section per lead atom for various processes: Incoherent scat- n
tering dasheddotted line, Compton effect for a free electron multiplied by
Z solid line, photoelectric effect from tabulation Ref. 8 dashed line and where n is the number of atoms per unit volume, and is the
from approximation Ref. 12 dotted line, and pair creation Ref. 5 total cross section per atom, which in our simple model is
crosses. given by

39 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 39
 PEZ KN . 9 teractions Eq. 7. From Eqs. 12 and 13, it is easily
inferred that the random variable s is related to r by
The cross section for a molecule or compound can be ob-
tained by adding up the corresponding cross sections of the
atoms. ln r
s . 14
The interaction of the primary ray with the medium n
gives rise either to photoelectric absorption or to Compton
scattering with probabilities proportional to the correspond- This method also can be easily applied for the generation
ing cross sections Eqs. 4 and 5, respectively. In the first of ,
case all the energy is transferred to the medium at the posi-
tion of the interaction, while in the second case the photon is
2 r. 15
scattered in a direction given by the polar and azimuthal
angles and , respectively. The value of follows a uni-
form probability distribution between 0 and 2, and fol- Unfortunately, this direct method cannot be used for the
lows the probability distribution generation of . Although Eq. 10 can be integrated, can-
not be obtained analytically from r. In this case we can apply
d KN the acceptancerejection method.13 Lower and upper limits,
p d 2 sin d . 10
d K l and K u respectively, are defined for p(x). Also lower and
The energy of the scattered photon can be calculated by upper limits, x l and x u are defined for x. Two random num-
Eq. 1 and the amount of energy released to the medium is bers r x and r y are generated with uniform probability in the
TE 0 E. Assuming that the direction of the primary pho- intervals x l r x x u and K l r y K u , respectively. In other
ton in the lab coordinate system is given by the angles 1 and words, random points are generated in a rectangular area. If
1 , the direction of the scattered photon ( 2 , 2 ) can be r y p(r x ), the trial is accepted and the value xr x is used.
obtained as Otherwise, the value is rejected. It can be demonstrated that
the probability of x follows the p(x) function.13 This proce-
cos 2 cos 1 cos sin 1 sin cos , 11a dure is less efficient than the direct method in which all
cos cos 2 cos 1 random numbers generated by the computer are used, but the
cos 2 1 , 11b acceptancerejection method can be applied to any bounded
sin 1 sin 2 function. This method can be applied for the generation of
sin sin for Compton scattering.
sin 2 1 . 11c Finally the selection of the process, Compton scattering or
sin 2 photoelectric absorption, can be implemented by splitting the
In general, these relations can be applied for any subsequent 0,1 interval in two subintervals with a length proportional
Compton collision. to the corresponding cross sections. A ray moves inside the
medium until either it is absorbed by the photoelectric effect
or leaves the medium, that is, until the position of the next
randomly generated interaction occurs outside the medium.
B. The algorithm
An algorithm for the transport of rays in a medium for
simple geometries like a layer with infinite parallel faces or a C. The program
cylinder can be easily developed. For simplicity, we assume
that the medium consists of a single material that is de- The above algorithm has been implemented in
scribed by the constituent atom, molecule, or compound, and MATLAB14 by the authors. The conversion to other lan-
the density. guages such as Basic, Fortran, and C is straightforward. The
The simulation of the random processes is carried out by a flow diagram of the algorithm is shown in Fig. 4. The input
Monte Carlo method.13 The parameters that have to be ran- parameter are the number of chemical elements, number of
domly generated are s, , , and the selection of the pro- atoms of every species per molecule or compound, atomic
cesses photoelectric or Compton when an interaction oc- number, atomic weight of the atoms, density, and the energy
curs. of the rays. The program next computes the various cross
A simple method for the generation of a random variable y sections for the first interaction. Next, the path traversed be-
with a probability density p(x) defined in the interval (a,b) fore the first collision takes place is randomly generated us-
relies on the fact that r defined as ing Eq. 14. If the collision occurs inside the medium, the
kind of process is chosen. If it turns out to be Compton
rQ y a
y
p x dx 12
scattering, the direction and energy of the photon is updated,
and the final position is computed. If the photon is still inside
the medium, the next path is generated and the procedure is
is itself a random variable with a uniformly distributed prob- repeated. After every interaction the energy transferred to the
ability in the interval 0,1. Therefore y can be obtained from medium (T for a Compton scattering or E 0 for a photoelec-
yQ 1 r , 13 tric absorption and the corresponding coordinates are stored.
A trial ends either when the photon escapes from the medium
where r is a random number in the interval 0,1 that can be or a photoelectric absorption takes place.
provided by a computer. This method is feasible as long as We recommend that readers write their own code follow-
Q(y) is a function that can be inverted. In our case we apply ing the above discussion. However, the program developed
this method to generate the path traversed between two in- by the authors can be freely downloaded.15

40 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 40
 Fig. 4. Flow diagram of the algorithm
for the simulation of the passage of
rays through matter.

IV. APPLICATION TO SEVERAL PRACTICAL The program optionally shows the positions where the en-
CASES ergy has been deposited either by photoelectric absorption
or Compton scattering. Figure 5 lower panel shows the
The simple model can be used to visualize the -ray tracks corresponding result for a 1000 photon beam. This plot is
in the medium. Many interesting features of the passage of
very useful for the visualization of the spatial distribution of
rays through matter can be observed in this way. For ex-
ample, in Fig. 5 upper panel the graphical output of our the energy left in the medium, which is of much interest for
program for an Al layer 30.2 cm 5 thick corresponding to radiation dosimetry.
a 150 photons beam of 1 MeV rays is shown. In this case, The program also generates histograms corresponding to
the rays often suffer several Compton collisions before the angle and energy distributions of the photons escaping
they are absorbed. It is also clearly seen that backscattering from the medium. In addition, the spectrum of the energy
is the consequence of a number of collisions with high scat- transferred to the medium as well as its spatial distribution is
tering angles. Students can obtain a first impression of these calculated.
features, which strongly depend on the medium and the The predictions of the simple algorithm have been com-
-ray energy. pared with those of the EGS4 code. For light elements the

41 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 41
 Fig. 6. Kerma in a water cylinder of 10 cm radius and 50 cm length versus
radial distance for a 1 MeV -ray beam impinging perpendicularly to the
cylinder axis. The average energy transferred per unit mass, per photon for
radial distances of a 0.0 cm, b 0.8 cm, c 3.6 cm, and d 9.6 cm are
Fig. 5. A beam of 1 MeV -rays impinging on an Al layer of 30.2 cm 5 shown. The predictions of Eq. 21 stars are also shown.
width. The upper panel shows the tracks of the rays. The lower panel
shows the positions where a ray has deposited energy either by photoelec-
tric absorption or by Compton scattering.

according to Eq. 1. Hence, the value of E tr is given by

agreement is very good. However, for heavy elements like PE tr,KN
Pb, the simple algorithm overestimates the path length of E tr E 0 Z E0 , 18
photons because Eq. 5 underestimates the photoelectric
cross section in the energy range of the simulation see Figs. where
2 and 3. For light elements, incoherent scattering is domi-
nant and, therefore, the inaccuracy of Eq. 5 does not give
rise to any obervable effects.
tr,KN TT max

T0
T d KN
E 0 d
d. 19

The evaluation of kerma for water is of particular interest

A. Energy deposition in radiotherapy because water is the main component of
many biological tissues. As is well known,3,5 the attenuation
One of the most relevant quantities related to the effect of
coefficient for a molecule or compound is given by
the radiation on a medium is the amount of energy that is
absorbed and the corresponding spatial distribution. Consider i
a -ray beam traversing a medium along the z axis. The w , 20
i i i
beam fluence16 , the number of rays per unit area, de-
creases with path length z as where w i is the weight fraction of the atoms in the molecule
or compound. Extensive data on attenuation coefficients for
z 0 e z , 16 both atomic elements and compounds of interest in medical
where is the linear attenuation coefficient that is, the col- physics can be found in Refs. 5, 9, 10, and 17.
lision probability of a photon per unit path length given by Because of the exponential attenuation of with depth,
n see Eq. 7. the kerma due to primary collisions follows the relation
The quantity that most directly connects the description of
the radiation beam with its effects is the kerma, which is K 0 E e z . 21
defined as the kinetic energy transferred from photons to tr
electrons per unit mass in a volume element.16 In our case In the following we will calculate the kerma produced by
the energy transferred to electrons arises either from incoher- 1 MeV rays in a beam of 0.5 cm2 area in water. For this
ent scattering or photoelectric absorption. The unit of kerma case the mass attenuation coefficient equals 0.0707 cm2 g1
is 1 joule per kg. In radiation dosimetry this unit is called
the gray and is represented by G.) Eqs. 9 and 20. and E tr equals 0.440 MeV Eqs. 18 and
19. These values can also be found in Ref. 5. From these
The kerma K due to the primary collisions of a -ray beam
can be obtained from results, a kerma value at the entrance of the medium (z
0) of 0.0622 MeV/g that is 1.031011 G) per incident
K
E ,
tr
17
photon is inferred. In Fig. 6 the exponential attenuation of
the kerma along the beam path, Eq. 21, is shown. Unfortu-
nately, this simple calculation does not account for the en-
where the mass attenuation coefficient / is defined as the ergy transferred to the medium by Compton scattered rays
linear attenuation coefficient divided by the medium density see Fig. 5. However, our algorithm can easily obtain the
. E tr is the mean energy transferred per collision. In photo- energy that is released as kinetic energy to the atomic elec-
electric absorption almost all the energy is transferred to the trons in every collision, including those carried out by these
medium, while in a Compton scattering, T depends on secondary photons outside the beam path.

42 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 42
 Spectrometers based on a NaI scintillator are a very popular
device in nuclear physics laboratories, and because of their
low cost, they can be found in many undergraduate labora-
tories. The device consists of a NaI crystal attached to a
photomultipler that detects the light pulse generated by the
scintillator when a ray is absorbed. The energy absorbed in
the scintillator is converted into visible light by lumines-
cence, giving rise to a photomultiplier pulse height distribu-
tion that can be analyzed by a multichannel analyzer. In the
ideal case all the -ray energy is deposited, and the light
pulse has an amplitude proportional to the -ray energy. In
practice, the amplitude spectrum is broadened by several in-
strumental effects. See Ref. 4 for a detailed discussion on
the various broadening sources.
In a real detector, photons very often leave the scintillator
Fig. 7. Kerma in water same case as Fig. 6 versus radial distance for after suffering one or more Compton collisions thus depos-
depths of a 0.5 cm, b 15.5 cm, c 33.5 cm, and d 49.5 cm. iting an energy smaller than E 0 . This effect gives rise to the
well-known -ray line shape of inorganic scintillators,4
which consist of the so-called photopeak total absorption
Let us consider a -ray beam along the axis of a water together with a continuous spectrum in the energy interval
cylinder of radius 10 cm and length 50 cm. Because kerma is from 0 to the Compton edge given by Eq. 2.
the energy per unit mass, we need to divide the medium in Our program can be used for an approximate calculation
volume elements cells and then compute the amount of of the line shape of a NaI spectrometer. As mentioned, for
energy transferred in each cell. The smaller the cell, the heavy elements, our simple algorithm overestimates the path
higher the spatial resolution for the calculation of the kerma, length of a photon and consequently overestimates the num-
but also higher are the statistical fluctuations of the results. In ber of rays escaping from the scintillator, in particular
our case, each cell, is symmetrical about the cylinder axis through the side surface. To achieve a more accurate simu-
and is defined by the radial distance to the axis r and the lation, we have slightly modified the algorithm for this par-
depth z. The size of the cells is r0.4 cm and z ticular case. Instead of using the general formula of Messel
1.0 cm. Therefore, the volume of the cell is 2 rrz. In and Crawford Eq. 5, we have computed the photo-cross-
Fig. 6 we have plotted the energy transferred on the aver- section of NaI from the accurate data of Fig. 3 Ref. 5. From
age per unit mass, per photon, as a function of depth for Fig. 3 the cross section can be approximately obtained as
several radial distances 0.0, 0.8, 3.6, and 9.6 cm. As ex-
PE cm2 2.12103 E 2.89 E0.033 MeV ,
pected, Eq. 21 is in agreement with the r0 curve. Figure
22a
7 shows the energy transferred as a function of r for several
depths 0.5, 15.5, 33.5, and 49.5 cm, showing that a non- PE cm2 2.35103 E 2.78 E0.033 MeV .
negligible amount of energy is released outside the beam 22b
path. The effect of secondary rays in radiation dosimetry is
crucial, but unfortunately the contributing kerma cannot be Using the simple algorithm with Eq. 22 the spectrum of
predicted by simple formulas. Monte Carlo simulation is a the energy absorbed by the NaI scintillator in our student lab
very valuable tool for the corresponding calculation. has been calculated, and the result has been compared with
The quantity that is of more interest in radiotherapy and the experimental line shape. The NaI scintillator is cylindri-
radiobiology is the absorbed dose, which is the energy per cal with a 1.5 diameter and 2.0 height. The -rays are
unit mass actually retained in the medium.5 At this energy emitted by an isotropic source of 0.2 cm radius located at a
range the differences between the kerma and the absorbed distance of of 3.7 cm to entrance face of the scintillator. The
dose are due to the track length of the ejected electrons, energy of the -rays is 1.275 MeV, which corresponds to an
because the energy is actually deposited in the medium at a isolated peak of the nucleus 22Na with which the simulation
certain distance from the point where the photon interacts. will be compared. In principle, this energy is over the thresh-
For 1 MeV photons the energy of the most energetic elec- old for pair creation. However, as can be seen in Fig. 3, the
trons is 0.8 MeV Eq. 2. According to Ref. 5, the corre- corresponding cross section is negligible in comparison to
sponding range in water is about 0.33 cm. This range is Compton scattering.
smaller than our cell size and thus Figs. 6 and 7 nearly give Figure 8a shows the simulated spectrum consisting of a
the absorbed dose. At higher energies, however, the electron pronounced photopeak which contains 7.15% of the input
track is non-negligible and the spatial distribution of dose rays together with a continuum spectrum due to photons es-
does not equal that of kerma. caping after one Compton collision 14.7% efficiency or
In radiation therapy it is extremely important to determine more than one 8.13%. For comparison with the experimen-
the absorbed dose at any point of the irradiated volume. For tal data, the theoretical line shape in Fig. 8 has to be folded
an accurate simulation of the dose it would also be necessary into the experimental broadening function. The experimental
to simulate the tracks of the ejected electrons. shape of the photopeak was found to be Gaussian with a full
width at half-maximum weakly dependent on the energy 90
B. Response of a NaI scintillator KeV at around 1.2 MeV has been used for that purpose. If
we fold the spectrum of Fig. 8a with this Gaussian, we
Another interesting application of our simple algorithm is obtain the result shown in Fig. 8b which can be compared
the simulation of the response of a NaI scintillator to rays. with the experimental spectrum. The comparison has been

43 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 43
 with a basic knowledge of programming. The program can
be used as a tool to assist students in becoming familiar with
the phenomenology of the interaction of gamma rays in mat-
ter. The dependence on the material properties and energy
can be easily tested.
We have used the simple algorithm for the calculation of
the energy deposited in a cylinder of water versus depth and
radial distance. We also simulated the line shape of a NaI
scintillator and compared the shape with a spectrum mea-
sured in the laboratory. Finally two problems dealing with
the effect of multiple Compton scattering are proposed in the
Appendix.

ACKNOWLEDGMENTS
We thank C. Alonso-Arias for useful preliminary calcula-
tions. Also G. D. Montesinos acknowledges an undergradu-
ate grant from the program Becas Colaboracion of MEC.

APPENDIX
We include here some suggested problems that can be
solved by using the simple algorithm discussed in this paper.
Fig. 8. Line shape of a NaI scintillator cylinder of 1.5 diameter and 2 Problem 1. a A pencil beam of 1 MeV rays impinges
height for rays of 1.275 MeV energy. a Simulation results for the spec-
trum of the energy absorbed by the scintillator. b Experimental line shape
perpendicularly on a layer of Al 0.2 thick the mean free
solid and theoretical line shape dashed obtained by folding the energy path for 1 MeV rays in Al is 6.02 cm. Plot the probability
spectrum of plot a with a Gaussian of 90 KeV width. distributions per unit solid angle of a photon to be transmit-
ted and backscattered, respectively, with an angle between
the beam direction and that of the escaping photon do not
restricted to energies higher than 0.7 MeV. At lower energies take into account those rays that cleanly traverse the layer
the spectrum is dominated by the profile of the very intense without interacting. Compare the results of the simple algo-
annihilation peak at 0.511 MeV. The experimental spectrum rithm with the predictions of the KleinNishina formula Eq.
shown in Fig. 8b has been obtained by subtracting the labo- 3 for single Compton scattering.
ratory background spectrum and has been normalized to the b Plot the probability of an interacting photon for escap-
intensity of the simulated photopeak. The spectrum shows ing forward or backward as a function of the output energy.
reasonable agreement between simulation dashed line and Calculate the energy bounds of a photon after a single Comp-
experiment solid line for the Compton continuum. The dis- ton scattering using Eq. 1 for both forward and backscat-
crepancy is due to secondary experimental effects not taken tered rays. Are the escaping photons bounded to these lim-
into account in the simulation. its? Why?
The efficiency of the light collection depends on the posi- c Repeat the calculations for 1.0 and 5.0 thicknesses.
tion of radiation interaction. On the other hand, the radioac- Problem 2. Repeat the calculations of Problem 1 for lay-
tive source is attached to a cylindrical holder with a hole ers of Pb mean free path of 1.25 cm. Check that the energy
through which the photons escape. Many rays hit the alu- of the escaping photons are basically bounded to the limits
minum hole walls, thus degrading the energy spectrum of the imposed by Eq. 1. Why?
beam. In addition, rays also interact with the front and side
1
faces of the scintillator as well as with the photomultiplier. W. R. Nelson, H. Hirayama, and D. W. O. Rogers, The EGS4 Code System
All these effects tend to smooth the sharp cutoff of the con- SLAC, Stanford, CA, 1985 and http://www.slac.stanford.
tinuum spectrum at the Compton edge energy. edu/egs/; GEANT, Application Software group, CERN Program library
2001 and http://www.info.cern.ch/asd/geant/
2
W. Williamson and G. C. Duncan, Monte Carlo simulation of nonrela-
V. CONCLUSIONS tivistic electron scattering, Am. J. Phys. 54, 262267 1986.
3
R. D. Evans, The Atomic Nucleus McGraw-Hill, New York, 1955.
We have discussed a simple algorithm for the simulation 4
G. F. Knoll, Radiation Detection and Measurement Wiley, New York,
of the passage of rays with energy around 1 MeV in a 1979.
medium. In this energy range the only relevant processes are 5
H. E. Johns and J. R. Cunningham, The Physics of Radiology Charles C.
incoherent scattering and photoelectric absorption. The Thomas, 1983.
6
KleinNishina formula accounts accurately for the cross sec- For energies below 1 MeV, electron radiation losses due to bremsstrahlung
are negligible. On the other hand, the fluorescence photons emitted after
tion of the former, while the latter can be approximately the photoelectric absorption have very low energy and thus they are ab-
described by an analytical formula. The energy at which the sorbed over a very short distance.
cross section for pair creation, pair , is significant in com- 7
D. T. Cromer, Compton scattering factors for spherically symmetric free
parison to Compton scattering is much higher than 1 MeV; in atoms, J. Chem. Phys. 47, 18921893 1967; D. T. Cromer, Compton
particular for light elements Al, for example pair KN at scattering factors for aspherical free atoms, ibid. 50, 4857 4859 1969.
8
E. Storm and H. I. Israel, Photon cross sections from 1 KeV to 100 MeV
about 15 MeV. Therefore, the algorithm is also applicable to for elements Z1 to Z100, At. Data Nucl. Data Tables 7, 565 681
energies well above 1 MeV. 1970.
The algorithm is simple enough to be understood and con- 9
E. F. Plechaty, D. E. Cullen, and R. J. Howerton, Tables and graphs of
verted to a computer program by an undergraduate student photon interaction cross sections from 10 eV to 100 GeV derived from

44 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 44
 LLNL evaluated photon data library, UCRL-50400, Vol. 6, Rev. 4, Parts 1996; W. H. Press, B. P. Flanner, S. A. Teukolsky, and W. A. Vetterling,
A and B, Lawrence Livermore National Laboratory, 1989. Numerical Recipes Cambridge University Press, New York, 1983; R. Y.
10
M. J. Berger et al., XCOM: Photon cross section database, available at Rubinstein, Simulation and the Monte Carlo Method Wiley, New York,
http://physics.nist.gov/xcom National Institute of Standards 1981; F. James, Monte Carlo theory and practice, Rep. Prog. Phys. 43,
and Technology, Gaithersburg, MD, 1999; M. J. Berger and J. H. Hubbell, 11451189 1980.
NIST X-ray and Gamma-ray Attenuation Coefficients and Cross Sections 14
D. Hanselman and B. Littlefield, The Student Edition of MATLAB Version
Database, NIST Standard Reference Database 8 National Institute of
5: Users Guide Prentice Hall, New York, 1997. See also http://
Standards and Technology, Gaithersburg, MD, 1987.
11 www.mathworks.com/products/matlab/.
See http://www.photcoef.com/2121.html for photon interac- 15
tion coefficients of the elements. http://top.gae.ucm.es/radiofisica/simulations.html
16
12
H. Messel and D. F. Crawford, Electron-Photon Shower Distribution Fundamental Quantities and Units for Ionizing Radiation, ICRU Report
Function Tables for Lead, Copper and Air Absorbers Pergamon, New 60 International Commission on Radiation Units and Measurements, Be-
York, 1970. thesda, MD, 1998.
13 17
H. Gould and J. Tobochnik, An Introduction to Computer Simulation http://physics.nist.gov/PhysRefData/XrayMassCoef/
Methods: Application to Physical Systems Addison Wesley, New York, cover.html

Tantalus Cup. The mythological Tantalus angered the gods, and was condemned to hang from the limbs of a tree over a pool of water. When he tried to
drink, the water level receded from his lips. Tantalus cup is used as a paradox in introductory physics. The unsuspecting user pours water into the cup, and
the water level rises evenly in the cup and the inside end of the syphon tube. All is well until the water level just starts to exceed the highest point in the
curving tube. At this point syphon action commences, and the cup empties down to the level of the bottom of the tube. This large example at Harvard
University is 27 cm high. Photograph and notes by Thomas B. Greenslade, Jr., Kenyon College

45 Am. J. Phys., Vol. 71, No. 1, January 2003 F. Arqueros and G. D. Montesinos 45