Optimal shielding design for bunkers of compact cyclotrons

used in the production of medical radionuclides

A. Facurea兲 and W. F. França
Comissão Nacional de Energia Nuclear, R. Gal. Severiano 90, Sala 409, 22294-900 Rio de Janeiro, Brazil
共Received 18 May 2010; revised 12 October 2010; accepted for publication 21 October 2010;
published 19 November 2010兲
Purpose: There are several options to consider in the design of a vault that will house a cyclotron
for radioisotopes production with regards to the door entrance. Alternatives are a direct-shielded
door, a simple maze, or a double-legged maze. In this work, the impact of the neutron and photon
doses at the vault entrance was evaluated for these options.
Methods: Monte Carlo simulations were carried out in order to assess photon and neutron ambient
dose equivalents. Simulations results were compared to experimental measurements taken inside a
vault with a direct-shielded door.
Results: The double-legged maze is the configuration that provides the higher degree of radiologi-
cal protection at the vault entrance. In addition, the fact of the location of the cyclotron target plays
an important role in the ambient dose equivalents.
Conclusions: The comparison performed between measurements and results of MCNP simulations
confirmed a favorable agreement; maze legs reduce the neutron energies at the entrances of the
vault. However, a degree of dose reduction similar to the one obtained for a two-legged maze can
be achieved with the addition of shielding against neutrons in the inner maze entrance to act as a
second door. A choice of a vault design is more evident by comparing the results of this
study. © 2010 American Association of Physicists in Medicine. 关DOI: 10.1118/1.3515458兴

Key words: cyclotrons, Monte Carlo, radiation shielding

I. INTRODUCTION higher than 30 MeV. More recently, other publications be-

came available with significant data, such as the IAEA TRS
In recent years, the increasing interest in the use of positron 283 共Ref. 5兲 and NCRP 144 共Ref. 6兲, but there is still a need
emission tomography 共PET兲 technique has promoted the ad- of basic information associated to the shielding of PET cy-
vance of nuclear medicine. The quality of the PET technique clotrons in the range of 10–20 MeV.
is strongly related to its capability of showing physiology In general, experienced cyclotron personnel prefer un-
information, making possible studies that are not feasible shielded cyclotrons because of the practicality to repair and
with typical nuclear medicine procedures. The most used
perform maintenance operations.7 Typically, in a 18F produc-
positron emitters for these studies are in general produced by
tion, the neutron ambient dose equivalent at 1 m of the cy-
compact cyclotrons as shown in Table I.1
clotron target can reach values from 200 to 400 mSv per
Due to their very short half lives, the cyclotron facility
proton integrated beam current 共␮Ah兲.8,9 As a result, an un-
shall operate near the hospital where the medical examina-
shielded cyclotron will require a dedicated concrete vault,
tion is conducted. The exception are the 18F compounds,
since for the typical bombardment current from 30 to
which can be transported to centers located over longer dis-
60 ␮A, the neutron ambient dose equivalents inside the bun-
tances. In addition to that, 18F compounds are the most used
radiopharmaceutical in the PET technique, especially the la- ker can reach 18 Sv/h.
beled tracer 18F-FDG 共a sugar molecule with a radioactive The choice of the appropriated design for a PET cyclotron
fluorine label that allows imaging of glucose metabolism in vault is complex and, despite the fact that concrete is rela-
brain, heart and tumors兲. tively inexpensive, the vault should be small to minimize
The most effective method for the production of 18F is via cost but big enough to carry out the required maintenance
enriched water, through the reaction 18O共p , n兲 18F,2 which is operations. Hence, concrete vault walls should provide ad-
possible with medium energy cyclotrons 共10–20 MeV兲. One equate shielding against the intense neutron flux during the
can predict that the presence of neutrons is substantial during bombardment.
the bombardment of the target; in consequence, the neutron With regard to the design of the access to the vault, there
source term resulting from this reaction is the key element are several reasonable alternatives. One option is a direct-
for the shielding calculations.3 shielded door and this heavy door has to have the same
Very few publications are available concerning the shield- thickness as the adjacent walls, which means that the door
ing calculations for PET cyclotrons; the main reference is must be hydraulically or electrically driven. The downside of
NCRP 51 共Ref. 4兲 that shows graphs and tables obtained direct-shielded door design is the failure of this driving
40–50 years ago and for cyclotrons operating at energies mechanism and accidents that may cause to the staff. Hence,

6332 Med. Phys. 37 „12…, December 2010 0094-2405/2010/37„12…/6332/6/$30.00 © 2010 Am. Assoc. Phys. Med. 6332
6333 A. Facure and W. F. França: Optimal shielding design for bunkers of compact cyclotrons 6333

TABLE I. Most used positron emitters and reactions employed for their pro-
duction 共Ref. 1兲.

Half life Energy

Radionuclide 共min兲 Reaction 共MeV兲
15 15
O 2.03 N共p , n兲 11
14
N共d , 2n兲 6
16
O共p , pn兲 ⬎26
b)
13
N 9.97 16
O共p , ␣兲 19 a)
13
C共p , n兲 11
11
C 20.3 14
N共p , ␣兲 11–17
18 18
F 110 O共p , n兲 11–17
nat
Ne共d , ␣兲 8–14

periodic inspections and preventive maintenance are more c) d)

X
often required and the possibility of stop production during
down time periods.
Another option is a less shielded door in a simple maze or FIG. 1. Some typical layouts for pet cyclotrons bunkers, designated in this
multilegged maze configuration. The downside of this design work as 共a兲 direct-shielded door, 共b兲 simple maze, 共c兲 intermediary maze,
is the availability of space when the cyclotron facility has to and 共d兲 double-legged maze.
be attached to an existing building.
If a vault is projected with a maze configuration, there is
note that the licensing process of a cyclotron facility is com-
a need to perform precise calculations in order to meet the
plex and includes the technical analysis of a variety of secu-
regulatory shielding criteria for the entrance. An inadequate
rity parameters such as radiation protection, physical protec-
maze design could result in high dose rates outside the vault,
tion, security, radiological emergency, and fire protection.
which can increase the cost and delay the operation of the
Although the use of borated polyethylene slabs proved that is
production of radionuclides.10
very effective in absorbing neutrons, its utilization must be
One can predict that a multilegged maze design requires a
tempered due to the fact of the fire hazard associated with its
larger footprint 共more concrete兲. The total cost of the facility
use. Therefore, polyethylene slabs must be encased in steel
is higher, including initial cost and decommissioning, if we
or other nonflammable sheathing.
take into account the induced activation of concrete due to
the neutron flux during the bombardment. The use of iron,
for example, even as a constituent of the concrete or present II. METHODOLOGY
in the structural components in the form of reinforcing bars, The Monte Carlo radiation transport code MCNP, version 5
should be avoided because of the induction of long-term ac- 共Ref. 11兲 and the Evaluated Nuclear Data File B-VI 共ENDF/
tivation products. Neutron-induced activation can be reduced B-VI兲 continuous energy neutron cross section library were
with the incorporation of boron within the walls, floor, and used to perform the calculations of neutron transport in the
ceiling of the vault or surrounding the target. This process cyclotron vaults.
can also affect the neutron dose rates at the bunker entrance, In order to simulate the neutron transport inside the cy-
due to the absorption of thermal neutrons, before they can clotron vaults, the neutron spectrum obtained from protons
diffuse back into the vault. Nevertheless, the final decision hitting the 18O target, presented by Bosko et al.,12 was used.
concerning vault layout and composition is a particular user According to the authors, this spectrum was obtained from
option and will depend 共besides the construction costs兲 on General Electric 共GE兲 PETtrace cyclotron source-term evalu-
the type of operations that are expected to be carried out ations. The neutron spectrum is presented in tabular form, in
inside the facility. Table II.
Figure 1 shows some typical layouts for PET cyclotrons The neutron source was considered as an isotropic source,
vaults bunkers simulated in this work. positioned at 200 cm from the left wall 共X direction兲. The
In this work, the Monte Carlo code MCNP5 共Ref. 11兲 was neutron ambient dose equivalent rates were adjusted to the
used in the assessment of shielding calculations for compact value of 300 mSv/h for a hypothetical proton current of
cyclotrons used in radionuclides production, operating with 1 ␮A. This mechanism of neutron emission and the neutron
negative ions 共H⫺兲 in the energy range of 10–20 MeV. The dose rates at 1 m from the source were based on the work of
impact of the vault design on the dose rates at the entrance Gallerani et al.,8 who performed several experimental mea-
was studied for various types of layouts, be they direct- surements inside a GE-PETrace bunker.
shielded doors and single or multilegged mazes vaults. The In this work, vaults simulated are those presented in Fig.
effect of positioning sheets of borated polyethylene, even 1. It was not practical to maintain the internal area of the
lining the walls or used as a second door in the inner maze vaults mazes constant at 800⫻ 800 cm2; for the double-
entrance, was also evaluated. However, it is important to legged maze 共due to area limitations兲, a small increase of 100

Medical Physics, Vol. 37, No. 12, December 2010

6334 A. Facure and W. F. França: Optimal shielding design for bunkers of compact cyclotrons 6334

TABLE II. Neutron source term used in the simulations. Neutron doses and neutron average energies at 1 m from
the target and at the vault door were simulated in order to
Neutron energy Neutron fluence
共MeV兲 共n/s兲 study the impact of the bunker design in the neutron energy
distributions. Tenth value layers 共TVLs兲 were assessed for
0–1 1.1⫻ 1011 the neutron source term employed for ordinary concrete and
1–2 1.0⫻ 1011 5% borated polyethylene. The tenth value distance 共TVD兲
2–3 5.9⫻ 1010
was also evaluated. This parameter gives the distance inside
3–4 2.9⫻ 1010
the maze that neutrons should cross, from the inner maze
4–5 1.4⫻ 1010
5–6 6.4⫻ 109
entrance up to the door, in order to reduce the doses to one-
6–7 2.9⫻ 109 tenth of its original value.
7–8 1.2⫻ 109 The position of cyclotron body and target play a role in
8–9 5.1⫻ 108 the neutron dose rates at the entrance, since the cyclotron
9–10 2.1⫻ 108 magnet itself can potentially act as a partial shield. In order
⬎10 1.1⫻ 108 to study this effect, the isotropic neutron source 共target兲 was
positioned behind an iron cylindrical structure, with 2.0 m
diameter and 0.50 m height. The neutron dose rates in a
simple maze vault configuration entrance were simulated,
cm in the X direction was opted. The height of the vaults was with and without the iron cylinder. The number of histories
set to 360 cm. Vault walls are 200 cm thick, except for the used 共NPS兲 in all cases was 7.5⫻ 108, which assures statis-
maze, which were set to 150 cm. Similarly, all the maze tical errors associated with the results to be less than 2%.
corridors and room entrances are 150 cm in width. In order to verify the validity of the results, a comparison
The effect of lining some bunkers walls with sheets of 5% was performed between MCNP simulations and measurements
borated polyethylene or used at the inner maze entrance for in a direct-shielded door vault that had neutron and photon
vaults shown in Figs. 1共b兲 and 1共c兲 was studied. The compo- area monitors inside the vault. The neutron detector used in
sition of this shielding material is described in Table III. the measurements was a 3He Ludlum detector model 42–30
The well-known variance reduction technique point detec- H and the photon ion chamber was the model 3–0070 from
tor 共F5 tally兲 belongs to the group of partially deterministic Rotem Industries. The location of point detectors in the
variance reduction methods implemented in MCNP5. For each simulation matches the real position of the detectors.
source particle and collision event, a deterministic estimation The comparison of the neutron and photon measurements
of the fluence contribution is made at the point detector. For with the simulation results was performed by selecting three
this work, the F5 tally was used in order to provide the different cyclotron runs, under the same conditions, for an
neutron fluxes at point detectors in 共X, Y, Z兲 coordinates. The average current of 30 ␮A.
DE and DF cards allow modeling an energy-dependent dose All the studied vaults were simulated without doors and
function that is a continuous function of energy from a table the point neutron detectors were located at 30 cm outside.
whose data points need not coincide with the tally energy bin The exception was the condition where experimental mea-
structure.11 Here, the DF and DE tally cards have been used surements are compared to simulation results. In this particu-
to convert the flux from units of neutrons/ cm2 to ambient lar case, aiming to reproduce the same condition of the mea-
dose equivalents 共pSv cm2兲, with the flux-to-dose conversion surements, the Monte Carlo simulations were conducted
factors obtained from ICRP Publication 74.13 considering the bunker as being completely closed.

TABLE III. Chemical composition of shielding materials considered in the III. RESULTS AND DISCUSSION
simulations 共relative weight contribution兲 共Ref. 14兲.
Figure 2 shows the transmission curve for the ordinary
5% borated polyethylene Conventional concrete concrete simulated in this work, taking into consideration an
共0.94 g / cm3兲 共2.26 g / cm3兲 a isotropic source emitting a neutron spectrum described in
Element 共%兲 共%兲 Table II. As a result, the TVL for ordinary concrete barriers
H 11.60 0.92 is 33 cm. For the 5% borated polyethylene used in the simu-
C 61.20 — lations, a TVL of 13 cm was noted.
Na — 1.71 In Fig. 3, results of the TVD inside a maze of the cyclo-
O 22.20 49.83 tron vault are shown. According to the simulations, a value
B 5.00 of 430 cm can be considered as a reasonable parameter to be
K — 1.92 used in shielding calculations for a simple maze, such as the
Al — 4.56 one depicted in Fig. 1共b兲.
Si — 31.58
The neutron spectrum at 1 m from the source, inside the
Ca — 8.26
standard vault shown in Fig. 1共a兲, is represented in Fig. 4. It
Fe — 1.22
is clear that a low-energy component is present; this is due to
a
Data from Ref. 14. neutron scattering in the walls.

Medical Physics, Vol. 37, No. 12, December 2010

6335 A. Facure and W. F. França: Optimal shielding design for bunkers of compact cyclotrons 6335

FIG. 2. Percent transmission versus thickness of ordinary concrete for the FIG. 4. Neutron spectra at 1 m from the source, inside a standard cyclotron
neutron source term studied in this work. bunker.

The weighted average neutron energy was calculated as

5共a兲, neutron doses at door entrance give the effect that a
follows:
second door is present in the vault. This “second door” re-
兺i=1
N
Ei⌽共Ei兲 duces the neutron doses by a factor of 3.4 in comparison
Ē = , 共1兲 with the simple maze configuration. The MCNP5 FMESH
兺i=1
N
⌽共Ei兲
card was used to obtain the data; this card allows the user to
where Ei is the neutron energy of the ith energy interval; define a mesh tally superimposed over the problem geom-
⌽共Ei兲 corresponds to its fluence obtained by MCNP calcula- etry, calculating the track length estimate of the particle flu-
tions and the sum symbol is extended to all N energy inter- ence, averaged over a mesh cell, in units of particles/ cm2.
vals of the spectrum. According to Eq. 共1兲, the average neu- Also, from the obtained data it can be concluded that a simi-
tron energy at 1 m from the source was found to be about lar degree of protection from a double-legged maze can be
1.25 MeV. attained if the vault is constructed with an extension leg, as
The simulation results for neutron and photon ambient illustrated in Fig. 1共c兲, forcing the neutrons to undergo more
dose equivalents at the vault entrances are presented in Table reflections before reaching the bunker door. For this particu-
IV, as well as the average neutron energies for each configu- lar bunker layout, similarly to the simple maze configuration,
ration. it is clear that the degree of dose reduction is improved by
It was demonstrated that for either neutrons or photons, a the addition of a borated polyethylene sheet in the inner
vault entrance with a double-legged maze configuration pre- maze entrance, to act as a second door.
dicts a better dose reduction. Now, when borated polyethyl- In the case of a directed-shielded door configuration 共no
ene is positioned at the inner maze entrance as shown in Fig. maze兲, the fact of adding a 5 cm borated polyethylene sheets
on the walls provides neutron doses reduction by a factor
of 2.
The direct-shielded door bunker results are too high in
comparison with the other vault layouts due to the inexis-
tence of any mechanisms of neutron interaction between the
source and the bunker entrance, in a different manner of
what happens in vaults that contain a maze. In the direct-
shielded door bunker, there is a possibility of direct neutrons
reaching the door and very high neutron doses are produced
at 1 m from the source.
However, when the simulation includes the cyclotron’s
magnets positioned between the target and the inner maze
entrance, an average neutron dose reduction of 25% was ob-
served in a simple maze vault. In this configuration, cyclo-
tron target position can significantly affect the neutron dose
rates at the vault entrance, due to the angular distribution of
the neutrons produced in the nuclear reactions, which will be
FIG. 3. Neutron ambient dose equivalent reduction as a function of maze forward-peaked.
length. Experimental neutron measurements inside the bunker

Medical Physics, Vol. 37, No. 12, December 2010

6336 A. Facure and W. F. França: Optimal shielding design for bunkers of compact cyclotrons 6336

TABLE IV. Neutron and photon ambient dose equivalents rates and average neutron energies at the entrance of the vaults studied in this work for 1 ␮A of
integrated proton current hitting the 18O target.

Neutron ambient dose equivalent Neutron average energy Photon ambient dose equivalent
Vault configuration 共␮Sv h−1兲 共keV兲 共␮Sv h−1兲

Directed-shielded door 4000.0 900.0 36.0

Directed-shielded door +5 cm of poly. on the walls 2000.0 317.3 13.8
Simple maze 113.0 53.8 3.5
Simple maze +5 cm poly. door on the inner maze
entrance 33.0 77.4 1.0
Intermediary maze 5.4 12.5 0.6
Intermediary maze +5 cm poly. door on the inner maze
entrance 1.8 15.0 0.3
Double-legged maze 0.3 5.1 0.03

were not constant as expected. Measurements showed other effects that may cause an erratic behavior, such as ra-
0.27⫾ 20%, 0.60⫾ 20%, and 2.18⫾ 20% mSv/h for 1 ␮A diofrequency and magnetic fields.
of integrated proton current, while the obtained result from
MCNP simulations was 4.85 mSv/h. These discrepancies in
neutron measurements can be attributed to the uncertainties
IV. CONCLUSIONS
and error terms associated with factors such as beam loss due
to gas stripping and collision with cyclotron internal compo- For the design of a vault that will house a cyclotron ac-
nents, nontarget reactions from foils, collimators, beam-line celerator, some factors might be considered. In many cases,
components, and the difficulties inherent to neutron detec- the direct-shielded door configuration is the preferred choice,
tion, such as low efficiency, dead time and pile up effects, but due to the door weight and size, the closing and opening
radiofrequency in unshielded detector, and magnetic field. process of the vault becomes an engineering problem and
Concerning the photon ambient dose equivalents, an av- can delay the production of radionuclides. On the other hand,
erage reading of 118.7⫾ 10% ␮Sv h−1 was observed in the vaults designed with mazes can improve personnel access
ion chamber, taking into account the three different cyclotron and movement, but accurate shielding calculation for bun-
runs and for 1 ␮A of integrated proton current, whereas the kers with mazes is complex; there is no deterministic ap-
result obtained from the MCNP simulations was proach for this type of situation.
91.8 ␮Sv h−1. Possible explanations for obtaining a lower In this work, neutron and photon ambient dose equiva-
photon dose rate value in the simulations is that the photon lents at the entrance of cyclotron vaults were simulated in
detector inside the bunker is also measuring the activation order to verify the neutron energy degradation due to the
compounds produced in the cyclotron metallic body, which reflections on the walls; it was found that the addition of
is a component that was not assessed in this work, and the maze legs reduces the neutron energies at the vault entrance.

FIG. 5. Mesh tally depiction of neutron ambient dose equivalents 共Sv/h兲 inside a cyclotron vault, considering 共a兲 the existence of a 5 cm thick borated
polyethylene door in the inner maze entrance or 共b兲 the absence of the borated polyethylene door.

Medical Physics, Vol. 37, No. 12, December 2010

6337 A. Facure and W. F. França: Optimal shielding design for bunkers of compact cyclotrons 6337

It is evident that doses at the door entrance are reduced enna, Austria, 2008兲.
2
T. J. Ruth and A. P. Wolf, “Absolute cross section for the production of
when a two-legged maze configuration is designed and more F-18 via the 18O共p,n兲18F reaction,” Radiochim. Acta 26, 21–24 共1979兲.
reduction of doses is achieved when shielding is positioned 3
K. Sindano, P. Leleux, P. Lipnil, and J. Vanhorenbeeck, “Production of
inside the inner maze entrance 共second door effect兲. How- 14,15O, 18F and 19Ne radioactive nuclei from 共p,n兲 reactions up to 30
ever, the target position inside the bunker and the cyclotron MeV,” Phys. Rev. C 42共2兲, 748–752 共1990兲.
4
National Council on Radiation Protection Measurements, “Radiation pro-
body have to be considered in the dose rates assessment at tection design guidelines for 0.1–100 MeV particle accelerator facilities,”
the vault entrances. NCRP Report No. 51 共Bethesda, MD, 1977兲.
5
Vaults with typical maze widths and barrier thicknesses International Atomic Energy Agency, “Radiological safety aspects of the
were studied, despite the fact that the increase or reduction in operation of proton accelerators,” Technical Reports Series No. 283
共IAEA, Vienna, Austria, 1988兲.
the mazes’ length or width will have an impact on the neu- 6
National Council on Radiation Protection Measurements, “Radiation pro-
tron and photon doses at the entrance of the rooms. More- tection for particle accelerator facilities,” NCRP Report No. 144 共Be-
over, although in this work the source term was used consid- 7
thesda, MD, 2003兲.
ering a specifically manufacturer, it is expected that the International Atomic Energy Agency, “Cyclotron produced radionuclides:
Guidelines for setting up a facility,” Technical Reports Series No. 471
comparative results are maintained for other types of equip- 共IAEA, Vienna, Austria, 2009兲.
ments, operating in the same range of energy. Consequently, 8
R. Gallerani, G. Cicoria, E. Fantuzzi, M. Marengo, and D. Mostacci,
as an outcome of this study, the degree of protection that is “Neutron production in the operation of a 16.5 MeV PETrace cyclotron,”
Prog. Nucl. Energy 50, 939–943 共2008兲.
obtained at the entrance of the vaults housing a PET cyclo- 9
R. Méndez, M. P. Iniguez, J. M. Martí-Climent, I. Penuelas, H. R. Vega-
tron for different designs was estimated. Nonetheless it has Carrilo, and R. Barquero, “Study of the neutron field in the vicinity of an
to be emphasized that other factors should be taken into ac- unshielded PET cyclotron,” Phys. Med. Biol. 50, 5141–5152 共2005兲.
10
count, such as cost, manpower, regulatory issues, space, etc. R. D. Sheu, C. C. Chen, R. J. Sheu, C. H. Kao, and S. H. Jiang, “The
refined shielding design for the cyclotron room of the Buddhist Tzu Chi
General Hospital,” Radiat. Prot. Dosim. 115共1–4兲, 216–221 共2005兲.
ACKNOWLEDGMENTS 11
X-5 Monte Carlo Team, “MCNP—A general Monte Carlo N-particle
transport code, version 5, Volume I: Overview and Theory,” LANL Re-
The authors would like to thank CNPq 共Conselho Nacio- port No. LA-UR-03-1987 共Los Alamos National Laboratory, USA, 2003兲.
nal de Desenvolvimento Científico e Tecnológico兲 for fund- 12
A. Bosko, D. Zhilchenkov, and W. D. Reece, “GE PETtrace cyclotron as
ing this research and Hospital das Clinicas da Faculdade de a neutron source for boron neutron capture therapy,” Appl. Radiat. Isot.
Medicina da USP for providing the experimental data pre- 61, 1057–1062 共2004兲.
13
International Commission on Radiological Protection, Conversion Coef-
sented in this article. ficients for Use in Radiological Protection Against External Radiation:
ICRP Publication 74 共Pergamon, Oxford, 1995兲.
a兲 14
Electronic mail: facure@cnen.gov.br National Council on Radiation Protection Measurements, “Structural
1
International Atomic Energy Agency, “Cyclotron produced radionuclide: shielding design and evaluation for medical use of x-rays gamma rays of
Principles and practice,” Technical Reports Series No. 465 共IAEA, Vi- energies up to 10 MeV,” NCRP Report No. 49 共Bethesda, MD, 1976兲.

Medical Physics, Vol. 37, No. 12, December 2010