Experimental measurement and Monte Carlo assessment of Argon-41 production in a PET cyclotron facility

Physica Medica 31 (2015) 991–996

Contents lists available at ScienceDirect

Physica Medica j o u r n a l h o m e p a g e : h t t p : / / w w w. p h y s i c a m e d i c a . c o m

Original Paper

Experimental measurement and Monte Carlo assessment of Argon-41 production in a PET cyclotron facility Angelo Infantino a,*, Lorenzo Valtieri a, Gianfranco Cicoria b, Davide Pancaldi b, Domiziano Mostacci a, Mario Marengo b a b

Department of Industrial Engineering, Laboratory of Montecuccolino, University of Bologna, Via dei Colli 16, 40136, Bologna, Italy Medical Physics Department, University Hospital “S. Orsola-Malpighi”, Via Massarenti 9, 40138, Bologna, Italy

A R T I C L E

I N F O

Article history: Received 25 February 2015 Received in revised form 9 July 2015 Accepted 14 July 2015 Available online 26 September 2015 Keywords: Argon-41 Gamma-ray spectrometry Monte Carlo FLUKA Radiation protection Medical cyclotrons

A B S T R A C T

In a medical cyclotron facility, 41Ar (t1/2 = 109.34 m) is produced by the activation of air due to the neutron flux during irradiation, according to the 40Ar(n,γ)41Ar reaction; this is particularly relevant in widely diffused high beam current cyclotrons for the production of PET radionuclides. While theoretical estimations of the 41Ar production have been published, no data are available on direct experimental measurements for a biomedical cyclotron. In this work, we describe a sampling methodology and report the results of an extensive measurement campaign. Furthermore, the experimental results are compared with Monte Carlo simulations performed with the FLUKA code. To measure 41Ar activity, air samples were taken inside the cyclotron bunker in sealed Marinelli beakers, during the routine production of 18F with a 16.5 MeV GE-PETtrace cyclotron; this sampling thus reproduces a situation of absence of air changes. Samples analysis was performed in a gamma-ray spectrometry system equipped with HPGe detector. Monte Carlo assessment of the 41Ar saturation yield was performed directly using the standard FLUKA score RESNUCLE, and off-line by the convolution of neutron fluence with cross section data. The average 41Ar saturation yield per one liter of air of 41Ar, measured in gamma-ray spectrometry, resulted to be 3.0 ± 0.6 Bq/μA*dm3 while simulations gave a result of 6.9 ± 0.3 Bq/μA*dm3 in the direct assessment and 6.92 ± 0.22 Bq/μA*dm3 by the convolution neutron fluence-to-cross section. © 2015 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.

Introduction In nuclear medicine, research on advanced and innovative technological solutions is necessary to improve diagnostic and therapeutic procedures, while concurrently assuring safety and protection of patients, workers, and the general population. This assumes particular relevance in the use of particle accelerators in the medical field, since in this case also the environment and the population living in areas adjacent to the site might be involved in an accident, as long as planned and unplanned releases of radioactivity. Cyclotrons are used in nuclear medicine to produce short-lived radionuclides of biomedical interest, especially for Positron Emission Tomography, as well as in radiation oncology, hadron therapy. Knowledge of the radiation fields around these devices is necessary for the design of shielding, the classification of areas and the appropriate selection of safety systems; of specific relevance, is the assessment of the activation of the accelerator itself, building

* Corresponding author. Department of Industrial Engineering, Laboratory of Montecuccolino, University of Bologna, Via dei Colli 16, 40136, Bologna, Italy. Tel.: +39 051 2087 702; fax: +39 051 2087 747. E-mail address: [email protected] (A. Infantino).

materials, and air during the routine use and the entire working life of the accelerator. In particular, the production, and consequent release of radioactive gases in the external atmosphere is an important aspect of the radiation protection of workers and reference person (as defined by ICRP publication 103 [1], par. 193). In a medical cyclotron facility, 41Ar (t1/2 = 109.34 m) is the most important air activation product, due to the relatively high secondary neutron flux during irradiation and according to the 40Ar(n,γ)41Ar reaction (Fig. 1) [2]. In the literature, some analytical assessments of the 41Ar production in air have been published, based on assumptions or experimental measurement of the neutron flux. A theoretical model of the production of 41Ar in a cyclotron vault and its release in atmosphere were studied in several papers by Birattari et al. [3,4], where hypothesis on the shape and the energy distribution of the neutron field were done. Gutermuth et al. [5] and Biju et al. [6] performed FLUKA Monte Carlo assessment of 41Ar concentration around proton accelerators in the 0.1–1 GeV energy range; in these works, Monte Carlo simulations were compared with results of analytical estimations, according to methods described in IAEA Report 283 [7] or NCRP Report 144 [8] respectively. An estimation of the production of 41Ar due to photoneutrons near a 15 MV linear accelerator was performed by Chao et al. [9], based on experimental

http://dx.doi.org/10.1016/j.ejmp.2015.07.146 1120-1797/© 2015 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.

992

A. Infantino et al./Physica Medica 31 (2015) 991–996

Figure 2. Sampling positions adopted during the measurement campaign within the bunker.

Figure 1. Cross section of the 40Ar(n,γ)41Ar reaction. This reaction has a nonnegligible cross section for fast neutrons and a significant value of 660 mbarn at 0.025 eV [2].

measurement of the neutron flux using Indium foils. Some experimental results are available as regards 41Ar release from nuclear reactors [10,11]. Bezshyyko et al. [12] made an evaluation on the production of 41Ar, based on analytical calculations and on a simplified Monte Carlo model of a self-shielded cyclotron, and finally Braccini et al. reported on experimental measurements of the induced radioactivity due to a 18 MeV proton beam directly extracted in air from a medical cyclotron [13]. However, no reports on direct measurement of 41Ar production in air around biomedical cyclotrons are currently available. In this work, we developed a sampling methodology for the measurement of 41 Ar inside the cyclotron vault. To compare the experimental measurements, a detailed Monte Carlo model of the bunker, the cyclotron and the target assembly, used in the University Hospital “S. Orsola-Malpighi”, Bologna (Italy) has been created using the FLUKA code. Materials and methods Experimental setup The cyclotron used in the irradiation tests and to which our simulations setup refers, is a PETtrace (GE Medical System), a compact cyclotron with vertical acceleration plane, capable of accelerating negative hydrogen and deuterium ions up to an energy of 16.5 and 8.4 MeV, respectively. The cyclotron is used for the routine production of Positron Emission Tomography (PET) radionuclides. The target system modeled was a GE assembly comprised of a silver chamber filled with [18O]-water to produce Fluorine-18 by the (p,n) reaction. The front of the target body is sealed with a 25 μm Havar™ foil, an alloy of cobalt (42.5%), chromium (20%), nickel (20%) and traces of manganese, molybdenum, iron and others. The inner dimensions of the bunker are: 650 cm by 535 cm with a height of 350 cm and 200 cm thick concrete walls. Data on dimensions and characteristics of the bunker, the cyclotron and its components were taken from technical sheets and project drawings of the site. The

ventilation rate inside the bunker is routinely fixed at 10 air changes per hour. To measure the activity concentration of 41Ar inside the cyclotron bunker, an extensive measurement campaign was performed: Marinelli beakers (typical total filling volume 1.36 dm3) were placed inside the bunker, during routine productions of 18F, in a series of marked positions, as reported in Fig. 2. Beakers 1, 2, 3, 4 were placed at a distance of 1, 2, 3, 4 m from the target respectively; beaker 5 was placed at 1 m while beakers 6 and 7 at 2 m from the target; beaker 8 was placed “in contact” with the target. Positions were marked on the floor, to be reproducible; all the beakers were placed at the same height (120 cm from the ground) as the target assembly used in the irradiation. Marinelli beakers were sealed to simulate the absence of the ventilation, as in the Monte Carlo model described in the following; given its really small thickness, the influence of the plastic walls of the beaker can be neglected, and the activation of the air inside a beaker can be assumed to be equal to activation of air outside the beaker. At the end of the irradiation, and after a waiting time of 10–20 minutes necessary to enter the bunker in safety conditions, the samples were removed, transferred to the gamma ray spectrometry laboratory (located on the same level at a distance of 25) and measured using an HPGe N-Type detector. The HPGe detector has a 30% relative efficiency and a resolution of 1.8 keV at 1332 keV. The high-resolution gamma-ray spectrometry system is based on digital electronics (Areva Canberra, distributed in Italy by TNE, Milan). The spectrometry system was calibrated in the 59–1836 keV range by means of a multi-radionuclide certified reference solution, obtained from an accredited Standardization Laboratory (Areva CERCA LEA, Pierrelatte Cedex, France). The calibration process was performed accordingly to the IEC 61452 standard [14], using the Genie 2000 software (Areva Canberra, distributed in Italy by TNE, Milan). A dual logarithmic polynomial efficiency curve was used. The method implemented in the software accounts for the propagation of the uncertainties in the calibration of the reference source (1–2% at 1 sigma level, depending on the peak in the mixture), in the tabulated yield (typically <1%), in the net peak area (<1% for calibration peaks) and in the interpolation of the curve (typically <3%). The calibration uncertainties therefore result of about 4–5% at 1 sigma level [15]. Samples were measured for 600–1800 seconds and all the

A. Infantino et al./Physica Medica 31 (2015) 991–996

experimental measurements were decay corrected to end of bombardment (EOB). Since the gamma ray spectrometer was calibrated using water solutions [14] while air samples were collected and measured, in order to take into account the different density and geometry, a gaseous Marinelli beaker source was modeled and its correspondent efficiency curve was evaluated using the software Labsocs [16,17], an optional package of the Genie2000 suite. A correction factor for efficiency, for the 1294 keV peak, was calculated and applied to water solution based efficiency calibration in the analysis of the experimental measurements. The Labsocs efficiency data points should be assigned 4.3% s.d. in the range 400 and 7000 keV [16]. The overall uncertainty in gamma-ray spectrometry analysis includes propagation of the above referenced uncertainties in calibration and uncertainties in the volume of the sample (typically 6%), in the position of the sample (1%), random errors in the peak area (2–5%) and the correction factor for the density of the sample (4.3%). The final uncertainty is of the order of 10–15% at one sigma, while the minimum detectable activity (variable as a function of acquisition time and peak energy in the spectra) resulted of the order of 5 Bq/dm3. FLUKA Monte Carlo model FLUKA is a well-known general purpose code to model particle transport and interaction with matter, covering an extended range of applications [18,19]. Thanks to the graphical interface Flair [20] it is possible to easily manage the writing of the input, the running of the simulations and the post-processing of the results. The geometry of our Monte Carlo model of the cyclotron, the target assembly and the cyclotron bunker was created using the solid modeler SimpleGeo [21] version 4.3 and exported to Flair (version 1.1-3 at the time of computation). Some of the data used for the creation of the model, like default settings, beam properties and irradiation profile, are based on preliminary models presented in Infantino et al. [22,23]. The model of the cyclotron includes the magnet and magnet poles (iron), the vacuum chamber (aluminum), the coils (copper) and the target filling stations panel (aluminum, lead and polyethylene) (Fig. 3). A 16.5 MeV proton beam, elliptically shaped in the perpendicular direction to beam direction, was simulated with a Gaussian distribution in energy (FWHMΔE = 0.08 MeV), according to the manufacturer specifications, and along both the x and y axis

993

(FWHMx = 0.55 cm, FWHMy = 0.47 cm), based on experimental measurements. A detailed model of the standard GE target for the production of 18F, according to the 18O(p,n)18F reaction, was realized, including the aluminum external body, the helium cooling flange and Havar foils, the target chamber in silver, and its contents of enriched 18 O-water. The model of the cyclotron was positioned within the model of the vault, reproduced on the basis of the original construction drawings and on measurements in the facility as built; all the dimensions were reproduced with an accuracy better than 1 cm. The walls of the bunker were simulated of standard Portland concrete, with a density of 2.35 g/cm3. A detailed model of the Marinelli beakers used in the experimental measurement campaign was created. Moreover, to assess the results in a larger volume, in order to obtain a lower relative uncertainty of the scored activation, several air regions, with a volume of 1 m3, were created and centered in the same position as the Marinelli beakers. The FLUKA “AIR” material used in the simulations was dry air (at sea level) with a composition, in fraction by weight, of [24]: C 0.000124, N 0.755267, O 0.231781 and Ar 0.012827. The set of defaults called NEW-DEFA was used as a basis for the simulations [22,23]: this set allows a good compromise, for most applications, between the activation of the main physical mechanisms and the CPU-time usage. However, to allow for a more accurate transport of radiation, some standard settings of NEW-DEFA were overwritten using specific FLUKA commands (cards). The transport threshold for protons was set to 1 MeV by the PART-THR card: when the energy of a particle becomes lower than the cutoff defined by PART-THR, and if such cutoff is lower than 100 MeV, the particle is not stopped but ranged out to rest and its kinetic energy is deposited uniformly over the residual range [18]. To achieve accurate results for residual nuclei production, the evaporation of heavy fragments and the coalescence mechanisms were activated. Finally, radiation decay was activated in “analogue mode” (meaning that the time evolution is calculated analytically and all daughter nuclei and all associated radiations are considered but at fixed times), and all the nuclides were simulated to be produced in their ground state [25]. An irradiation profile of 1 hour irradiation time and 1 μA extracted proton current was set and used for all the simulations. Two different types of score were used: RESNUCLE and USRTRACK. The RESNUCLE card allows scoring residual nuclei produced in inelastic interactions on a region basis: several cards were used to score the activity concentration of 41Ar at the EOB in the whole air volume (~120 m3), within the Marinelli beakers (~1 dm3) and in 1 m3-volumes. The USRTRACK score was used to assess the average differential neutron fluence distribution in energy in the air content of the vault, in an energy-binning basis. To reach a high statistic, 1010 primary particles were simulated using the most up-to-date version at the time of computation (FLUKA version 2011.2b.6). The simulations were performed on a remote cluster based on 8-physical processors Intel® Xeon® E5410 (@ 2.33 GHz) provided with 4 cores each, using half of the system resources (e.g. 16 cores). External cross sections method (ECSM)

Figure 3. Section of the FLUKA Monte Carlo model used in the simulations: The model reproduces one of the experimental setup adopted.

One of the main features of FLUKA is the possibility to calculate the residual nuclei produced through inelastic interactions in a defined region. To assess the radionuclide inventory, FLUKA bases its calculation on inelastic hadronic interaction models, except for low-energy neutrons, where tabulated cross sections are used [26]. In FLUKA the transport of neutrons with energies lower than a predefined threshold energy is performed by a multigroup algorithm. Actually, for neutrons with energy lower than 20 MeV, as in the case

994

A. Infantino et al./Physica Medica 31 (2015) 991–996

of interest, FLUKA uses a specific neutron cross section library in which the energy range is divided into 260 energy groups of approximately equal logarithmic width (31 of which are thermal), and containing more than 250 different materials. The energy groups are numbered in order of decreasing energy, i.e. group 1 corresponds to the highest energy [25]. In addition to direct assessment performed with the RESNUCLE score, the production yield was calculated also “off-line”, using ENDF/ B-VII.0 cross sections library [2] in combination with neutron fluence data obtained with FLUKA. The saturation activity can be written as

Asat = ∫

E1

E260

n40 Arφ ( E )σ (n,γ ) ( E ) dE

(1)

where n40Ar is the number of 40Ar atoms in the target material; φ ( E ) = d 2Φ (E) dE dt is the neutron flux distribution per unit of energy [cm−2 s−1 MeV−1] with Φ(E) the neutron fluence and σ(n,γ) is the cross section of the 40Ar(n,γ)41Ar reaction in [cm2]. Since with USRTRACK it is possible to score the differential neutron fluence distribution as a function of energy F(E) = ΔΦ(E)/ΔE in cm−2*GeV−1 per incident primary unit weight, we can transform the integral into a sum over the FLUKA 260 energy bins

Asat ≅ ρV ω40 Ar

NA 1 N P ∑ i=260 F ( Ei )σ (n,γ ) ( Ei ) ΔEi A40 Ar

Asat I

Beaker

Ysat ± SDOM [Bq/μA*dm3]

1 2 3 4 5 6 7 8 Total

2.44 ± 0.08 2.32 ± 0.09 2.42 ± 0.08 2.77 ± 0.11 3.58 ± 0.17 3.55 ± 0.16 3.79 ± 0.20 2.92 ± 0.14 3.0 ± 0.6

the results 19.8%; we consider thus practical to express in a synthetic form our results using an overall average, and its standard deviation, of the saturation activity of 41Ar produced in air, that results 3.0 ± 0.6 Bq/μA*dm3.

(2)

where ρ is the density of the target material; V is the volume of the target region; ω40Ar is the 40Ar mass fraction in air; A40Ar is the atomic weight of the isotope 40Ar; NA is Avogadro’s number; NP is the number of primary particles per second; and ΔEi is the width of the ith energy bin. Since the constraint of the FLUKA energy structure for neutrons, equation (2) was implemented in a script that first “adapted” the ENDF cross sections to the “Low Energy Neutrons” (LEN) structure; then the convolution was performed and 41 Ar activity concentration for 1 hour–1 μA irradiation was calculated. The script provides also the assessment of the uncertainty of the convolution, δAsat, calculated through the propagation of the independent uncertainties of the cross section data and the fluence. Finally, the saturation yield Ysat in MBq/μA was calculated as the ratio of the saturation activity, Asat, to the irradiation current, I.

Ysat =

Table 1 Results of the experimental measurements of the saturation yield of 41Ar, per one liter of air, in the different positions. For each beaker the weighted average, over 17 samples, and the standard deviation of the mean were calculated.

Figure 4. Typical spectrum obtained in gamma-ray spectrometry analysis of a Marinelli beaker: typical MDA are in the order of magnitude of 5 Bq/dm3.

(3)

Results Experimental results A total of 68 samples, distributed in various sets of 4 different positions in the bunker and during 17 irradiations were taken. For all samples, the only radionuclide identified was 41Ar. The average saturation yield of 41Ar, per one liter of air, is listed in Table 1 for all the positions, including the standard deviation of the mean (SDOM). A typical spectrum obtained through gamma-ray analysis is shown in Fig. 4. The presence of a peak at 511 keV (not always visible in all the samples) is due to the production of traces of 11C in the plastic walls of the beaker and in the Parafilm used to seal the beaker itself as confirmed by tests performed by placing opened beakers in the sampling positions. The values of saturation yield in air show some differences between the various sampling positions; a slight variability can be observed in the thermal neutron flux in the different positions (1–4), with a maximum in sampling position 4, close to a corner between two walls of the bunker (Fig. 5). Despite these differences, the values show a substantial uniformity, being the coefficient of variation of

Figure 5. Neutron spectra in the positions 1 to 4 within the cyclotron vault. The spectra were assessed using the USRTRACK score.

A. Infantino et al./Physica Medica 31 (2015) 991–996

Results of Monte Carlo simulations

Table 3 FLUKA assessment of the

Direct activation of the air within the cyclotron vault, obtained with the RESNUCLE score, can be represented as a bi-dimensional map (atomic vs mass number) of the radionuclides produced (Fig. 6). As the figure shows, traces of other radionuclides can be produced in air during irradiation (Table 2); however, the saturation activity of these radionuclides is at least two order of magnitude lower than that of 41Ar, and this only in the case of radionuclides with half-life of seconds or minutes. For all the other, longer lived radionuclides, the evaluated activity concentrations are 5 or more order of magnitude lower than 41Ar; their production can therefore be neglected. The 41Ar saturation yield obtained directly using the RESNUCLE card (already reported in Table 2), and indirectly by the neutron fluence-to-cross section convolution, both over the whole air volume (~120 m3), are compared in Table 3, together with the ratio of the simulated concentration to the experimental average. The ratio between the FLUKA 41Ar saturation yield, in both direct assessment (FD) and neutron fluence-to-CS convolution (FC), to the experimental measurements was calculated for Marinelli beakers and 1 m3-volumes in position 1–4 (Table 4). The results reported in Table 4 are substantially equal within the uncertainties; however, it can be observed a tendency in the results: the larger is the volume in which the simulation scoring is performed, the less is the ratio of FLUKA simulations to the experimental measurements.

995

41

Ar saturation yield: total air volume.

Direct (RESNUCLE) Convolution (USRTRACK)

Ysat [Bq/μA*dm3]

Ratio FLUKA/Exp

6.9 ± 0.3 6.92 ± 0.22

2.2 ± 0.4 2.2 ± 0.4

Table 4 FLUKA assessment of the 41Ar saturation yield: ratio of the simulated concentrations for Marinelli beakers and 1 m3-volumes to experimental measurements. N/a means that for the direct assessment (built-in RESNUCLE score) the volume of integration enclosed in the Marinelli beakers was too small to obtain a statistically significant result. Actually, increasing the volume of integration (1 m3 volumes and furthermore the total air volume of about 120 m3) the results became consistent and the uncertainty decreased. Beaker

1 2 3 4

Marinelli Beakers

1 m3-volumes

FD/Exp

FC/Exp

FD/Exp

FC/Exp

n/a n/a n/a n/a

3.08 ± 0.16 3.03 ± 0.17 3.07 ± 0.16 2.89 ± 0.16

6±4 3.0 ± 1.0 2.6 ± 1.1 2.2 ± 0.7

2.60 ± 0.12 2.69 ± 0.14 2.63 ± 0.13 2.58 ± 0.13

This trend might be due to the very low reaction rate of the Ar(n,γ)41Ar reaction: actually, direct assessment in the Marinelli beakers (10−3 m3) did not produce any results while in the 1 m3volumes 41Ar was produced in a detectable quantity; the improved sampling makes possible a significant quantitative assessment. The same considerations apply to the results obtained from the fluenceto-cross section convolution; again, it is possible to see how by increasing the volume of integration, the ratio between FLUKA to experimental results decreases, and tends to the ratio obtained considering the whole air volume inside the cyclotron vault (~120 m3). 40

Conclusions

Figure 6. RESNUCLE bi-dimensional map of the radionuclides produced during irradiation (1 h–1 μA) in the air volume within the cyclotron vault.

Table 2 Saturation yield of the radionuclides produced in air during irradiation. Isotope

t1/2

Ysat [Bq/μA*dm3]

41

109.34 m 269 y 35.04 d 5.05 m 3.01e5 y 7.13 s 5730 y 20.20 ms 12.33 y

(6.9 ± 0.3)E+00 (5.6 ± 2.0)E-08 (1.6 ± 0.5)E-01 (1.1 ± 0.3)E-04 (9 ± 9)E-13 (8.0 ± 2.2)E-02 (1.345 ± 0.003)E-04 (6 ± 6)E-03 (4.42 ± 0.13)E-05

Ar

39Ar 37Ar 37

S

36Cl 16

N

14C 12B 3

H

In this work we reported on direct measurements of the production of 41Ar inside the vault of a cyclotron for PET radionuclides production; to the best of our knowledge, this is the first time that an extensive set of experimental data is published as regards this relevant aspect of radiation protection around a biomedical cyclotron. Sampling air within a bunker in irradiation conditions is a relatively complex task; we have addressed all main aspects, including consistency of the sampling, timing between sampling and gamma ray spectrometry analysis, correction of the efficiency calibration accounting for sample’s density. Individual measurements resulted affected by uncertainties of the order of 10–15% at 1 sigma level; being the results quite similar in the different sampling positions, it is possible to evaluate an overall average, within a variability of less than 20%. We consider these results satisfactory and useful, particularly to support the planning stage of new facilities and the choices regarding proper regulation of the ventilation system. We have also developed a detailed Monte Carlo model of our cyclotron, targetry and bunker, in order to perform simulations of air activation. As a first, relevant result, it is confirmed that, as expected, production of 41Ar is the only significant air activation process; in our experimental measurements, we have never identified other radionuclides (minimum detectable activity of the order of 5 Bq/dm3). Nevertheless, we only obtain an agreement within a factor of 2–3 between simulations and experimental results. A similar level of discrepancy has been observed in other attempts to model activation in materials not directly interested by the primary beam [5,6,26–28]. In our work we have used the standard composition of dry air [24]; we have not investigated the role of filtration and other air pre-treatment systems. Any eventual influence of the latter in varying

996

A. Infantino et al./Physica Medica 31 (2015) 991–996

the average air composition, particularly for a trace element like argon, as well as the role of variable relative humidity, have not been assessed. It has to be noted that analytical approaches to calculate 41Ar production in air require a relatively complex series of calculations, based on rough approximations of the real geometry, neutron spectra and fluence distribution. Their accuracy cannot expected to be better than those obtained using Monte Carlo methods, based on a detailed description of the real problem. Even if our simulations have been carefully prepared their capacity to reproduce the results of experimental measurements is limited. Nevertheless, in our opinion the use of MC modeling in this field is a useful and powerful tool, since it allows for simultaneous assessment of a variety of quantities of interest in radiation protection like the dose distribution produced and transmitted by shielding; dose transmission in ducts and labyrinths; activation of the structure of the accelerator, of the lateral walls, of the soil underneath and of air [23]. This avoids the need for adopting a series of “ad hoc” approximate formulas. The MC model of an accelerator facility can be scalable, including only the components strictly necessary for a specific simulation, up to reproduce the real geometry in a very detailed way; this constitutes an elegant, unitary approach to the solution of several radiation protection aspects during the planning of a new facility. Calculation times, with modern computers, are acceptable; for our high statistics simulations (1010 primary particles) running times were about 20 hours using 16 cores. Finally, our simulations produced an overestimation, by a factor of 2, of the saturation yield of 41Ar produced in air. Further studies are necessary in order to improve the precision of Monte Carlo simulations, as regards activation of materials due to secondary radiation; however, the observed discrepancies being in terms of a cautious overestimation, our results confirmed the usefulness of FLUKA in the perspective evaluation of the radiological impact of new cyclotron installations. Acknowledgements The Authors would like to warmly acknowledge Prof. Emeritus D. Bollini for his continuous support and encouragement. References [1] ICRP. “The 2007 Recommendations of the International Commission on Radiological Protection”. ICRP Publication 103. Ann ICRP 2007;37(2–4). [2] Evaluated Nuclear Data File (ENDF) Database. . Database Version of April 13, 2015, Accessed 13/04/2015. [3] Birattari C, Bonardi M, Ferrari A, Silari M. Neutron activation of air by a biomedical cyclotron and an assessment of dose to neighbourhood populations. Radiat Prot Dosimetry 1985;14:311–19. [4] Birattari C, Ferrari A, Parnell CJ, Silari M. The environmental release of radioactive gases produced by neutron activation of air with the new Hammersmith cyclotron. Radiat Prot Dosimetry 1987;19:183–6. [5] Gutermuth F, Wildermuth H, Radon T, Fehrenbacher G. Radiation impact caused by activation of air from the future GSI accelerator facility FAIR. Radiat Prot Dosimetry 2005;115:437–40.

[6] Biju K, Sunil C, Sarkar PK. Estimation of 41Ar production in 0.1-1.0-GeV proton accelerator vaults using FLUKA Monte Carlo code. Radiat Prot Dosimetry 2013;157:437–41. [7] International Atomic Energy Agency (IAEA). Radiological safety aspects of the operation of proton accelerators. IAEA Technical Reports Series No. 283. Vienna: IAEA; 1988. [8] National Council on Radiation Protection and Measurements (NCRP). Radiation protection for particle accelerator facilities: recommendations of the National Council on Radiation Protection and Measurements (NCRP report; no.144). Bethesda: NCRP; 2003. [9] Chao JH, Liu WS, Chen CY. Estimation of Argon-41 concentrations in the vicinity of a high-energy medical accelerator. Radiat Meas 2007;42:1538–44. [10] Lauritzen B, Astrup P, Drews M, Jorgensen H, Mikkelsen T, Thykier-Nielsen S, et al. Atmospheric dispersion of argon-41 from a nuclear research reactor: measurement and modelling of plume geometry and gamma radiation field. Int J Environ Pollut 2003;20:47–54. [11] Mladin D, Mladin M, Toma A, Dulama C, Prisecaru I, Covaci S. Calculation of radioactive species transport in a TRIGA reactor. Nucl Eng Des 2013;262:29–38. [12] Bezshyyko OA, Golinka-Bezshyyko LO, Kadenko IM. Contribution of fast neutrons to argon activation in air at medical cyclotrons. Proceedings of the conference Current Problems in Nuclear Physics and Atomic Energy, Kyev; 2008. [13] Braccini S, Ereditato A, Nesteruk KP, Scampoli P, Zihlmann K. Study of the radioactivity induced in air by a 15-MeV proton beam. Radiat Prot Dosimetry 2014;163(3):269–75. [14] IEC. Nuclear instrumentation – measurement of gamma-ray emission rates of radionuclides – calibration and use of germanium spectrometers. IEC standard 61452. Geneva: IEC; 1995. [15] Zagni F, Cicoria G, Lucconi G, Infantino A, Lodi F, Marengo M. Monte Carlo modeling provides accurate calibration factors. Appl Radiat Isot 2014;94:158– 65. [16] Bronson FL. Validation of the accuracy of the LabSOCS software for mathematical efficiency calibration of Ge detectors for typical laboratory samples. J Radioanal Nucl Chem 2003;255:37–141. [17] Venkataraman R, Bronson F, Atrashkevich V, Field M, Young BM. Improved detector response characterization method in ISOCS and LabSOCS. J Radioanal Nucl Chem 2005;264:213–19. [18] Ferrari A, Sala PR, Fassò A, Ranft J. FLUKA: a multi-particle transport code. CERN-2005-10 (2005), INFN/TC_05/11, SLAC-R-773. Geneva: CERN; 2005. [19] Böhlen TT, Cerutti F, Chin MPW, Fassò A, Ferrari A, Ortega PG, et al. The FLUKA code: developments and challenges for high energy and medical applications. Nucl Data Sheets 2014;120:211–14. [20] Vlachoudis V. FLAIR: a powerful but user friendly graphical interface for FLUKA. In: Proc. Int. Conf. on Mathematics, Computational Methods & Reactor Physics (M&C 2009), Saratoga Springs, New York; 2009. [21] Theis C, Buchegger KH, Brugger M, Forkel-Wirth D, Roesler S, Vincke H. Interactive three dimensional visualization and creation of geometries for Monte Carlo calculations. Nucl Instrum Methods Phys Res A 2006;562:827–9. [22] Infantino A, Cicoria G, Pancaldi D, Boschi S, Fanti S, Marengo M, et al. Prediction of 89 Zr production using the Monte Carlo code FLUKA. Appl Radiat Isot 2011;69:1134–7. [23] Infantino A, Marengo M, Baschetti S, Cicoria G, Longo Vaschetto V, Lucconi G, et al. Accurate Monte Carlo modeling of cyclotrons for optimization of shielding and activation calculations in the biomedical field. Radiat Phys Chem 2015;doi:10.1016/j.radphyschem.2015.01.001 (in press). [24] National Institute of Standards and Technology (NIST). Standard Reference Database. . Online October 1998, Last update August 2005. Accessed 13/04/2015. [25] Fassò A, Ferrari A, Ranft J, Sala PR. FLUKA-2011 manual. Geneva: CERN; 2011. [26] Brugger M, Ferrari A, Roesler S, Sala PR. Calculation of radionuclide production cross sections with FLUKA and their application in high energy hadron collider studies. In: Proceedings of the International Conference on Nuclear Data for Science and Technology, Nice, France; 2007. [27] Vincke H, Theis C, Roesler S. Induced radioactivity in and around high-energy particle accelerators. Radiat Prot Dosimetry 2011;164:434–9. [28] Infantino A, Oehlke E, Mostacci D, Schaffer P, Trinczek M, Hoehr C. Assessment of the production of medical isotopes using the Monte Carlo code FLUKA: simulations against experimental measurements. Nuclear Instruments and Methods in Physics Research Section B, Manuscript submitted for publication, Unpublished Results; 2015.