You are on page 1of 10

ARTICLE IN PRESS

Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434


www.elsevier.com/locate/nima

Advanced simulation code for alpha spectrometry


T. Siiskonen, R. Pollanen
STUK Radiation and Nuclear Safety Authority, P.O. Box 14, FIN-00881 Helsinki, Finland
Received 15 March 2005; accepted 10 May 2005
Available online 6 July 2005

Abstract

A Monte Carlo code, known as advanced alpha-spectrometric simulation (AASI), is developed for simulating energy
spectra in alpha spectrometry. The code documented here is a comprehensive package where all the major processes
affecting the spectrum are included. A unique feature of the code is its ability to take into account coincidences between
the particles emitted from the source. Simulations and measurements highlight the importance of coincidences in high-
resolution alpha spectrometry. To show the validity of the simulated results, comparisons with measurements and other
simulation codes are presented.
r 2005 Elsevier B.V. All rights reserved.

PACS: 02.70.Uu; 29.40.Wk; 29.30.Ep

Keywords: Monte Carlo simulation; Alpha spectroscopy; Coincidences

1. Introduction photoelectric absorption and continuous slowing-


down of charged particles, can provide acceptable
Monte Carlo simulations have proven to be results. In many cases, simulation is the only
adequate tools for describing alpha-, beta- and practical way to explore the physics behind
gamma-particle transport, even in complex geo- observed phenomena.
metries. A great variety of computer codes have Alpha-particle spectrometry is a widely-used
been developed for particle transport, dosimetry, analytical method, for example in surveys of
particle physics and industrial applications. Dif- environmental radioactivity. The low activity of
ferent levels of sophistication exist among the the samples necessitates long counting times and a
codes, but even the simplest ones, which take into small sample-detector distance (SDD). The draw-
account Rutherford and Compton scattering, back of a small SDD is the possibility of
coincidence summing between the emitted alpha
Corresponding author. Tel.: +358 9 75988 318; particle and subsequent emissions from the
fax: +358 9 759 88 433. daughter nucleus. In addition, carefully designed
E-mail address: teemu.siiskonen@stuk. (T. Siiskonen). sample preparation techniques are essential, since

0168-9002/$ - see front matter r 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.nima.2005.05.045
ARTICLE IN PRESS

426 T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434

the alpha particles continuously lose their energy 2. Properties of the source and particle tracking
as they travel through matter. The energy loss
leads to degradation of the spectrum quality via Particle propagation through a material layer is
peak spreading, which increases with as the SDD determined by two physical processes: direction
is reduced. changes (scattering) and energy loss. The algo-
Simulations can be used to investigate the rithm for particle propagation in a given material
inuence of various phenomena on the spectrum layer proceeds as follows:
quality. The most important factors can be singled
out and the measurement setup can be optimised. (1) Emit a particle from a randomly selected point.
Moreover, unknown properties of the source, such (2) Calculate the distance, i.e. step length to the
as source density (or thickness) or source particle next scattering (or photoabsorption) event
properties, can be determined. This is important, using the cross-section data.
especially in the case of direct alpha spectrometry, (3) If the particle is charged, adjust the step if a
when radiochemical sample treatment is omitted. boundary of absorbing material is crossed.
The particle beam attenuation and interactions in Calculate the continuous energy loss during
basic research can also be examined. the step.
Many Monte Carlo simulation packages, such (4) If the particle energy is below the cut-off value,
as the TRIM package [1], the GEANT software stop tracking.
suite [2], and MCNP code [3], are suitable (5) If the particle crossed boundary of the material
for simulating the alpha particle behaviour in the layer, proceed to the next layer if one is
medium. However, these packages are not neces- present. Otherwise, stop tracking.
sarily optimal for alpha spectrometry simulations. (6) Determine the next direction vector, i.e.
More specic approaches to alpha-spectroscopic scattering angles.
simulations include the backscattering study (7) If the particle is a photon, determine the energy
of Ferrero et al. [4] and the investigation loss in the scattering or photoabsorption event.
of aerosol particles by Pickering [5]. Roldan (8) Goto (2).
et al. [6] examined the spectrum quality at a
small SDD. Here, characteristics of the source as well as the
The present Monte Carlo simulation code, particle tracking method, i.e. determination of the
known as Advanced Alpha-spectrometric SImula- scattering angles, are described. Calculation of the
tion (AASI), is designed to simulate alpha-particle energy loss of alpha particles, electrons and
energy spectra. It is intended to be a comprehen- photons is presented in the following sections.
sive simulation package where all the major
processes inuencing the energy spectrum are 2.1. Source
included. Samples of various types (aerosol
particles, thick samples, non-uniform samples, Particle emission can originate from a point or
etc.) are accommodated. Coincidences between from a nite-sized object. These objects, e.g.
the emitted particles are calculated using nuclide- aerosol particles, can be embedded in the source
specic decay data that are stored in a library le matrix. The composition of the source matrix and
prepared in extensible markup language, XML. the objects that emit radiation need not to be the
Although the code has so far been applied to the same. For example, alpha particles can be emitted
simulation of alpha particle energy spectra from from an aerosol particle located inside a glass-bre
environmental samples, it can also be used for lter. In the spectrum simulations the number of
other applications. The typical running time on a alpha particle emissions is given in the input.
1.6 GHz Pentium PC varies from seconds to a The thickness of the source can be subjected to
couple of minutes depending on the complexity of random uctuation that is assumed to follow a
the simulation problem. The code is written in Gaussian distribution with a user-given standard
Fortran 95. deviation s. To prevent impossibly large thick-
ARTICLE IN PRESS

T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434 427

nesses, the resulting source thickness H s r is standard deviation of the efciency, is given in
limited to the input. Calculation of the geometrical detection
efciency is necessary, for example, in direct alpha
0pH s rpH ss (1)
spectrometry when radiochemical sample treat-
where r is the radial position inside the source, s is ment is omitted. Tracers cannot then be used for
a user-given parameter and H is the nominal quantitative activity determination.
(mean) thickness. The measurement setup, consisting of the
Coordinates of the source particles are sampled source, source backing, absorbing material layers
as described by Siiskonen and Pollanen [7]. For and the detector, can be plotted in a le for visual
sources with a random thickness and convex or inspection. Library routines for plotting were
concave sources, the vertical coordinate is sampled written by Kohler [8].
by the rejection method. Convex and concave
source shapes are described with a paraboloid of 2.2. Particle tracking
revolution. The user of the code supplies the
central and side thicknesses of the source. If the Electrons are tracked when they travel in the
source thickness is zero, all source particles lie on a source, in the source backing and in the detector,
plane. including its dead layer. Photons are only tracked
Source particles can have a spherical or elliptical inside the detector (including its dead layer). Alpha
shape. Spherical source particles can have a log- particles are tracked in the source backing for
normal size distribution. Inactive source particles backscattering studies. Otherwise, particles are
can be coated with a uniform layer of radioactive assumed to travel in straight paths. The tracked
material. In addition, a spherical shell of inactive particle is followed until it escapes the absorbing
material can be placed around a spherical source material or its energy falls below a cut-off value.
particle. When crossing a boundary between two adjacent
The distance of the source particles from the absorbing layers, the tracking step length is adjusted
source surface can be exponentially distributed so that the step does not cross the layer boundary.
inside the source matrix. This is a useful feature for Particle tracking starts with the sampling of the
investigating air lters in which radioactive aerosol initial emission coordinates. The initial emission
particles are accumulated. This option is only direction y0 ; f0 is chosen from a uniform
available for cylindrical sources without thickness distribution. Following the emission, the cosine
uctuations. The distance d i of a source particle i of the polar angle yn (see Fig. 1) of the tracked
from the source surface is obtained from particle is determined by
d i l ln xi (2) cos yn cos o cos yn1 sin o sin yn1 cos c (3)
i where yn is the polar angle after nth scattering, o is
where x is a random number between 0 and 1, and
l is the mean penetration depth given by the user. the scattering polar angle and c is the scattering
Another user-given parameter, f, determines the azimuthal angle. The cosine and sine of the
fraction of the particles to be distributed according azimuthal angle fn are given by
to Eq. (2). The rest, fraction 1  f , is distributed sin fn sinfn  fn1 cos fn1 cosfn  fn1
on the source surface (d i 0). Particles which
have d i larger than the source thickness penetrate  sin fn1 4
the source and are ignored. The total number of
emissions from the source is reduced accordingly. cos fn cosfn  fn1 cos fn1  sinfn  fn1
An average solid angle subtended by the  sin fn1 5
detector, the geometrical detection efciency, is where
calculated. This is the number of hits received by
the detector divided by the number of alpha sin o sin c
sinfn  fn1 (6)
particle emissions. The desired accuracy, the sin yn
ARTICLE IN PRESS

428 T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434

z 3. Simulation of alpha particle behaviour in medium

3.1. Energy loss


n n1
Alpha particle energy loss is calculated as
described in Ref. [7], using the stopping power
parametrisation of Ziegler as described in Ref. [9].
The total stopping power is the sum of the
stopping power due to electrons, Se , and the
nuclear stopping power, Sn . In the energy region
of interest (below 10 MeV), Se is parametrised as
n B 1 1 1
(9)
O Se S low S high

n-1 where
Slow c1 E cp2 c3 E cp4 (10)

A and

Fig. 1. Coordinate system. The original particle direction is Shigh c5 E c 6 1


p lnc7 E p c8 E p . (11)
OA, the new direction is OB. The scattering polar angle is o
and the azimuthal angle is c. Laboratory polar angles are yn
Values of parameters c1 ; . . . ; c8 are tabulated in
and yn1 . The z-axis is parallel to the detector symmetry axis Ref. [9]. Here, E p is the energy of a proton moving
and points from the source towards the detector. at the same velocity as the alpha particle in
question. For composite materials, other parame-
trisations are also available [7].
An arbitrary number of absorbing material
and layers can be added between the source and the
detector. The user supplies the number of layers,
cos o  cos yn cos yn1
cosfn  fn1 . (7) their atomic and mass numbers, densities, thick-
sin yn sin yn1 nesses and standard deviations representing the
thickness uctuations. Alpha particles are assumed
All angles are in laboratory coordinates. The
to travel straight paths, except in the source
scattering angle o depends on the differential
backing where their path is tracked collision by
scattering cross-section. After the initial emission,
collision.
the particles undergo successive scatterings which
are assumed to be statistically independent.
Alpha particle scattering is calculated in the 3.2. Energy loss straggling
centre-of-mass frame. Before the determination of
cos yn , the scattering angle is transformed to Straggling of the alpha particle energy loss can
laboratory coordinates via be approximated by a Gaussian energy distribu-
tion. Although not strictly correct with thin
sin j absorbing layers,1 it gives a reasonable estimate
tan o ma (8)
cos j in many cases. Standard deviation sG of the
M Gaussian distribution depends on the maximum
energy transfer in one collision between the alpha
where j is the scattering polar angle in centre-
of-mass coordinates, ma and M are the 1
The asymmetric Vavilov (or Landau) distribution is better
masses of alpha particle and target atom, respec- with thin absorbers. However, computationally they are much
tively. more complex.
ARTICLE IN PRESS

T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434 429

particle and an atomic electron, E max , given where a0 is the Bohr radius. The effective
approximately by 2me c2 b2 g2 . The parameter g is charge of the alpha particle, zeff , is calculated
dened through gma c2 E, where E is the energy as described in Refs. [9,11]. The mean free
of the alpha particle. The deviation is given by [2] distance (step length) between the collisions, L, is
  sampled from
b2
s2G E max d 1  (12)
2 L sEN1 ln x (18)
where b is the alpha particle velocity in units of c. where N N A r=A is the atomic density and N A is
Parameter d is the average energy loss in the the Avogadro constant. The mean atomic spacing
material layer in question, N 1=3 is used as a step length if LoN 1=3 .
Z Moreover, if the energy loss between two succes-
d 0:0614 r dx keV. (13) sive collisions is more than ve percent of the
b2 A
alpha particle energy, the step length for the
Here, Z and A are the atomic and mass number of energy loss calculation is reduced until the loss is
the target, respectively, r is the material density in less than ve percent.
g=cm3 and dx is the distance travelled in micro- Angular deection in the scattering event from a
meters. potential (14) is given by
2Zx
3.3. Scattering in the source backing plate cos o 1  (19)
1Zx
Alpha particles can be tracked in the source and c 2px (the xs are independent random
backing plate. Screened elastic Rutherford scatter- numbers).
ing is used to determine the changes in the ight
direction. Between the elastic collisions, alpha
particles are assumed to lose their energy con- 3.4. Detector response to alpha particles
tinuously. The mean free distance between the
collisions is calculated from the potential Alpha particle energy loss in the detector dead
layer is treated as described in Section 3.1. When
e2 2Z r=a
V r e (14) the alpha particle hits the active volume of the
4p0 r
detector, all its remaining energy is assumed to be
where a is the screening radius, e is electron charge deposited. In other words, alpha particles are
magnitude, 0 is the permittivity of free space and r neither tracked nor is their energy loss calculated
is the radial distance. The resulting total cross- in the active volume of the detector. Instead, a
section is simplied solution is chosen which notably reduces
the calculation time.
Z2
sE p_ca2 (15) The properties of the detector are read from the
E 2 ZZ 1 user-prepared le. The parameters are the atomic
where a is the ne structure constant and screening and mass numbers of the detector material,
parameter detector radius and thickness, dead layer thick-
 2 ness, detector full-width at half-maximum
_ 1 (FWHM) and the parameters of the exponential
Z . (16)
a 8ma c2 E tailing function.
Measurements show that the detector response
The screening radius is given by [10]
to monoenergetic alpha particles is not Gaussian
0:885a0 [1214]. To take this into account, a double-
a q (17)
2=3 exponential tailing function can be added to the
zeff Z2=3 detector response. The resulting energy E is
ARTICLE IN PRESS

430 T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434

sampled from a distribution calculated by using Bethes formula [19]


 2 2 "
nR nE 0 E dE e e 2pN A Zr E e E e me c2 2 b2
PE e ln
1R 2
m dx 4p0 me c2 b A 2I 2 me c2
emE 0 E ; EpE 0 20  q 
1R
1  b2  ln 2 2 1  b2  1 b2
where E 0 is the energy of the incoming alpha
particle. The user supplies the parameters n, m and
 q2 #
1
R. They should be determined from the measure- 1  1  b2 24
8
ments of good quality (i.e. thin) sources at a large
SDD. Parameter RX0 is the ratio of the areas of where I 16Z0:9 (eV) is the average ionisation
the two exponential distributions. Typical values energy of the target atom [2]. We neglect
for Canberra PIPS with an area of 450 mm2 are bremsstrahlung and the production of delta
n 0:1 keV1 , m 0:02 keV1 and R 12:0. electrons and X-rays, since their inuence on the
FWHM of the Gaussian detector response is detected electron energy spectrum is small. The
14 keV. Convolution with the Gaussian detector mean distance between the collisions and the
response is done after the tailing. angular deection is calculated as for the alpha
particles, except that the maximum energy loss per
step is ten percent of the electron energy.
4. Simulation of electron and photon behaviour in When an electron hits the detector, its path is
medium followed through the dead layer and into the active
volume. In the dead layer, the user has an option
4.1. Electrons to have a partially-depleted region, where the
electron deposits part of its energy in the detected
Large-angle deections of the electrons result signal. The amount of energy deposited in the
from the screened elastic Rutherford scattering, formation of the signal, E s , is then given by
Eqs. (14) and (15), with Z replaced with z
E s DE e ; 0pzpdzdl (25)
ZZ 1=2, see, e.g. Ref. [15]. The screening dzdl
parameter Z can be chosen from three alternative
where DE e is the electron energy loss at depth z
models. Nigam et al. [16] suggested that
and dzdl is the dead layer thickness. In the active
Z 2=3 volume of the detector, the deposited energy is
ZN 5:448 (21) equal to the detected signal, E s DE e .
Ee
The number of backscattered and transmitted
where E e is the electron kinetic energy. Adesida et particles from the detector is calculated. A particle is
al. [17] tted electron scattering data in aluminium counted as backscattered when it escapes from the
and proposed that front side of the detector, and as transmitted when it
Z 2=3 escapes from other sides of the detector. For back-
ZA 2:61 . (22) scattering studies, a parallel electron beam hitting the
Ee
detector surface perpendicularly can be used.
Moliere [18] (see also [16]) concluded that

4.2. Photons
0:36 1:20aZ=b2
ZM a2 Z 2=3 (23)
tt 2
Photons are assumed to interact via photoelectric
where t E e =me c2 and me is the mass of the absorption and Compton scattering. Pair produc-
electron. tion is ignored, since we are interested in low-energy
Between the elastic collisions the electrons phenomena. The mean free distance is calculated
continuously lose their energy. The energy loss is from the total cross-section of the above-mentioned
ARTICLE IN PRESS

T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434 431

interactions. A random number is used to decide The total interaction cross-section is sE g


which interaction occurs at the interaction point. sP E g sC E g . The mean free distance is then
The total photoelectric absorption cross-section Lg E g sE g N1 ln x. (29)
is read from a text le and interpolated. Data were
obtained from the National Institute of Standards
and Technology database [20]. If data for the
element in question do not exist, an analytical 5. Treatment of coincidences
approximation for the cross-section [21]
  The majority of alpha emitters have a sign-
28 p 5 E I 7=2 2
sP E g aZ a0 (26) icant decay branch to excited states of the
3 Eg
daughter nucleus. The excited states decay by
This approximation overestimates the total cross- gamma-ray or conversion electron emission.
section for most elements, especially at low Since the lifetimes of the excited states are
energies (less than 100 keV). For example, the typically much shorter than the integration time
overestimation is approximately a factor of 10 in of the data acquisition electronics, pulse summa-
Si when E g is between 15 and 100 keV. As the tion between the alpha particle and particles
overestimation is quite large, the user should emitted by the daughter nucleus may occur.
supplement the photoelectric data library for the The summation is more pronounced when the
element in question, if possible. After the photo- SDD is small.
electric absorption, an electron is ejected in the The summation may lead to distortion
direction of the incoming photon, with E e E g . of the peak shape and, thus, may have an inu-
X-rays produced in this process are ignored. Their ence on nuclide identication. A good example
energy is often so small that the photon is is separation of 239Pu and 240Pu, which is
absorbed in the detector. difcult even in the case of a high-resolution
The differential cross-section for Compton detector and a sophisticated spectrum deconvolu-
scattering is tion code. Another example is the coincidence
dsC a_c2 1 summation of 241Am, whose main alpha
decay branch leads to third excited state of the
dO 2 me c E g 1  cos o2
2
( daughter 237Np, resulting in a clearly visible
E 2g 1  cos o2 bump on the high-energy side of the 241Am main

me c2 me c2 E g 1  cos o alpha peak.
) The probability of each alpha decay branch is
1 cos2 o 27 given in the nuclide library le, consisting of a
schema le and actual library in XML format. The
Fortran-XML interface is written by Markus [23].
neglecting the binding energy and momentum of The alpha decay branch for an individual decay is
the atomic electron. The photon scattering angle is selected using a random number. The nuclide
calculated from distribution (27) using the rejec- library le also contains decay routes of the excited
tion method by Brusa et al. [22]. The total cross states of the daughter nucleus. Each decay route
section for Compton scattering is has a known probability of occurrence (yield),
("  2 #
pa_2 me c2 me c2 decay type (gamma or conversion electron emis-
sC E g Z 12 2 sion), initial and nal state indices, and energy of
me E g Eg Eg
the emitted particle. The emission of a conversion
2E g me c2 1 m e c2 electron is associated with an X-ray, whose energy
 ln 2
4
me c 2 Eg is given in the library.
)
2 2
me c For each excited state of the decay route a
 . 28 random number is used to select the next decay
22E g me c2 2 channel (i.e., nal state and emitted particle). The
ARTICLE IN PRESS

432 T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434

route is followed until the ground state is reached. 6. Comparisons with measurements and other
For each emitted particle, the emission direction is simulations
sampled. After the conversion electron emission,
an X-ray is emitted before the cascade is followed Geometrical detection efciency and the alpha
further. We assume that each conversion electron particle energy loss were investigated by Siiskonen
is associated with X-ray emission. This is a and Pollanen [7]. They found an excellent agree-
simplication, since we overlook uorescence ment with earlier results and measurements. To
yields and Auger electrons. The approximation is further conrm the homogeneity of the emission
good for heavy elements, whose K-shell uores- point distribution inside a source, we compared a
cence yields are close to 100%. simulated alpha particle energy spectrum from a
If the particle deposits energy in the active thick sample with one obtained by numerical
volume of the detector, a coincidence is formed integration. In the comparison, a parallel alpha
and deposited energy is added to the alpha particle particle beam was considered (corresponding to a
energy. If cascade consists of n subsequent decays, very large SDD, polar angle y 0) in order to
the alpha particle can be in coincidence with mpn keep the numerical integration tractable. The
particles. Deposited energies of those m particles agreement between the simulated spectrum and
are then added to the alpha particle energy. the one from numerical integration is good (Fig.
The algorithm to calculate the coincidences 2). This conrms the homogeneity of the emission
proceeds as follows: point sampling, which is qualitatively also shown
in Ref. [24]. Equally good agreement was obtained
when the source was assumed to be spherical in
(1) Check that decay routes exist, i.e., transitions shape (results are not shown here).
are available for the present state. If no route is Electron backscattering can be used to examine
found, exit the loop. the quality of electron transport, since back-
(2) Use a random number to select the decay scattering is sensitively depend on continuous
route, i.e., decay type, line energy and nal energy loss and angular deections in elastic
state index. Rutherford scattering events. Electron backscat-
(3) If the emitted particle is an electron, follow it tering coefcients for various elements are com-
through the source and its backing. Determine pared in Table 1. The agreement between the
if the particle travels towards the detector. experimental values and those of the present work
(4) If the particle hits the detector, simulate the
deposited energy.
(5) If particle deposits energy to the detector, a 1
Monte Carlo
coincidence is formed. Add the deposited
0.8 Numerical integration
energy to the alpha particle energy. If the
Relative counts

particle is backscattered or transmitted, in- 0.6


crease corresponding counters.
(6) If the emitted particle is an electron, emit the 0.4
associated X-ray. Go to (3).
(7) Go to (1). 0.2

0
4 4.2 4.4 4.6 4.8 5
The lifetimes of the excited states are available in Energy (MeV)
the library le. However, when coincidences are
calculated, they are not taken into account. The Fig. 2. Comparison between simulated alpha particle energy
spectrum from a convex source (side thickness 0 mm, central
lifetimes are assumed to be short enough, com- thickness 2 mm, solid black line) with a spectrum obtained by
pared to the integration time of data acquisition numerical integration (dashed grey line) from the same source.
electronics, for a coincidence to be seen. Alpha particles were assumed to travel in parallel tracks.
ARTICLE IN PRESS

T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434 433

Table 1 10000
Comparison between experimental (Exp.) and simulated (MC) Measured
electron backscattering coefcients for various elements when Simulated
1000
E e 30 keV and for normal incidence

Counts
Element C Al Cr Au 100

Exp. [26] 0.144 0.482


Exp. [27] 0.060 0.155 0.270 0.521 10
MC [25] 0.042 0.124 0.231 0.502
MC, this work 0.053 0.142 0.253 0.511 1
5200 5250 5300 5350 5400 5450 5500 5550 5600
In all simulations, the screening model of Nigam et al. [16] was Energy (keV)
used, see Eq. (21).
Fig. 3. Measured (dots) and simulated (solid line) spectrum of
241
Am. The source-detector distance was 5 mm and a detector
is good. However, the results of the present work FWHM of 14 keV was used in the simulation. Detector tailing
parameters were n 0:1 keV1 , m 0:02 keV1 and R 12:0.
are higher than the simulated results of Gueor-
guiev et al. [25]. The difference could be explained
by different calculation of the continuous energy
loss. Gueorguiev et al. used different average 10000
ionisation energy and a three-point difference
Simulated
scheme. The convergence of the present back- 1000 Simulated, no coincidences
scattering calculations was ensured to two sig-
Counts

nicant digits. 100


To investigate the inuence of coincidences,
measured spectrum of 241Am is compared to the 10
simulation. The coincidence summing of alpha
particles with photons and electrons from 237Np is 1
5200 5250 5300 5350 5400 5450 5500 5550 5600
clearly visible as a bump above 5490 keV (Fig. 3).
Energy (keV)
This bump is absent at a large SDD. Since
experimental subshell conversion coefcients for Fig. 4. Simulated spectra of 241Am when coincidences are taken
237 into account (solid line, same as in Fig. 3) and when
Np were not available, the relative yields for
coincidences are ignored (circles). See the caption of Fig. 3
conversion electrons were set as follows: the yield
for simulation parameters.
of the La line was assumed to be 20% and that of
Lb was assumed to be 80%. Fig. 3 illustrates that
simulations are accurately able to explain the
effect of coincidence summing. When the coin- quality is not optimal for high-resolution alpha
cidences are ignored in the simulation, the result- spectrometry. The sources may be considerably
ing spectrum clearly disagrees with the thicker than those obtained from radio-
measurement (Fig. 4). chemical sample treatment and their thickness
may not be uniform. The source may even be an
aerosol lter in which radioactive materials are
7. Discussion deposited. This option is useful when the presence
of alpha particle emitting materials in the lter
The present Monte Carlo simulation code, must be identied rapidly, i.e. alpha particles are
known as AASI, is designed for simulating energy counted directly without prior radiochemical
spectra in alpha spectrometry. The code was sample manipulation [7]. This information may
originally developed for estimating the inuence be of utmost importance should a nuclear incident
of source characteristics on the alpha particle or malicious dispersal of radioactive material into
spectra, for example in the case when the source the environment occur.
ARTICLE IN PRESS

434 T. Siiskonen, R. Pollanen / Nuclear Instruments and Methods in Physics Research A 550 (2005) 425434

Later development of the code is focused on the [4] J.L. Ferrero, C. Roldan, M.L. Acena, E. Garca-Torano,
development of modelling of the detector and the Nucl. Instr. and Meth. A 286 (1990) 384.
effect of alpha-electron and alpha-photon coin- [5] S. Pickering, J. Aerosol Sci. 15 (1984) 533.
[6] C. Roldan, J.L. Ferrero, F. Sanchez, E. Navarro,
cidences on the measured spectra. The code can be M.J. Rodrguez, Nucl. Instr. and Meth. A 338
easily used for various alpha detectors provided (1994) 506.
that the detector response can be treated as the [7] T. Siiskonen, R. Pollanen, Appl. Radiat. Isot. 60 (2004)
convolution of a Gaussian part of a peak and a 947.
double-exponential low-energy tail (Eq. (20)). The [8] K.E. Kohler, PSPLOT library home page, retrieved in
November 2004, http://www.nova.edu/ocean/psplot.html.
comparison of measured and simulated spectra, [9] R.B. Firestone, Table of Isotopes CD-ROM, eighth ed., in:
especially in the case of a thin 241Am source, S.Y.F. Chu, V.S. Shirley (Eds.), Version 1.0, Wiley-
highlights the importance of coincidence phenom- Interscience, New York, 1996.
ena. When the source-detector distance is small [10] J. Lindhard, V. Nielsen, M. Scharff, Mat. Medd. Dan. Vid.
(less than approximately a few cm) the coinci- Selsk. 36, vol. 10, 1968, pp. 2.
[11] G. Amsel, G. Battistig, A. LHoir, Nucl. Instr. and Meth.
dences must be taken into account in unfolding the
B 201 (2003) 325.
spectra. Quantitative separation of nuclides such [12] A. LHoir, Nucl. Instr. and Meth. 223 (1984) 336.
as 239Pu and 240Pu in alpha spectrometry is [13] G. Bortels, P. Collaers, Appl. Radiat. Isot. 38 (1987) 831.
questionable if coincidences are neglected. Devel- [14] E. Steinbauer, G. Bortels, P. Bauer, J.P. Biersack, P.
opments in the simulation of electrons and Burger, I. Ahmad, Nucl. Instr. and Meth. A 339 (1994)
102.
photons in the medium also facilitate the use of
[15] T. Kijima, Y. Nakase, Radiat. Meas. 26 (1996) 159.
AASI in basic research. The detector response to [16] B.P. Nigam, M.K. Sundaresan, Ta-You Wu, Phys. Rev.
various particle beams can be examined. 115 (1959) 491.
Future development of the code will be ad- [17] I. Adesida, R. Shimizu, T.E. Everhart, J. Appl. Phys. 51
dressed in the construction of an appropriate (1980) 5962.
library data le and a user-friendly interface. The [18] V.G. Moliere, Z. Naturforschg. 2a (1947) 133.
[19] K.S. Krane, Introductory Nuclear Physics, Wiley, New
validity of the results, i.e. the agreement of York, 1988, pp. 196.
simulated and those obtained from other calcula- [20] M.J. Berger, J.H. Hubbell, S.M. Seltzer, J.S. Coursey, D.S.
tions or measurements, is veried here and in Zucker, retrieved in November 2004, available from http://
previous publications. However, validation is a physics.nist.gov/PhysRefData/Xcom/Text/XCOM.html.
continuous process not limited to this article only. [21] M.D. Scadron, Advanced Quantum Theory, second ed.,
Springer, New York, 1991, pp. 172.
An executable version of the code, the input les [22] D. Brusa, G. Stutz, J.A. Riveros, J.M. Fernandez-Varea,
and a library le needed for the execution of the F. Salvat, Nucl. Instr. and Meth. A 379 (1996) 167.
code are available from the authors upon request. [23] A. Markus, An XML-Fortran interface, retrieved in
October 2004, see http://xml-fortran.sourceforge.net/.
[24] R. Pollanen, T. Siiskonen, P. Vesterbacka, Radiat. Meas.
References 39 (2005) 565.
[25] Y.M. Gueorguiev, D.I. Ivanov, G.M. Mladenov, Vacuum
[1] J.P. Biersack, L.G. Haggmark, Nucl. Instr. and Meth. 174 47 (1996) 1227.
(1980) 257. [26] L. Reimer, C. Tollkamp, Scanning 3 (1980) 35 as cited by
[2] S. Agostinelli, et al., Nucl. Instr. and Meth. A 506 (2003) D.C. Joy, retrieved in December 2004, http://www.nap-
250. chan.com/bse/index.htm.
[3] J.F. Briesmeister, MCNP-A General Monte Carlo N- [27] H.E. Bishop, X-Ray Optics and Microanalysis, in: R.
Particle Transport Code, Version 4C (2000), Los Alamos Castaing, P. Deschamps, J. Philibert (Eds.), Hermann,
National Laboratory, Los Alamos, New Mexico, USA. Paris, 1965. pp. 153.