research article
A neutron track etch detector for electron linear accelerators in radiotherapy
Branko Vukovic1, Dario Faj2, Marina Poje1, Maja Varga1, Vanja Radolic1, Igor Miklavcic1, Ana Ivkovic2, Josip Planinic1
1	Department of Physics, University of Osijek; Osijek, Croatia
2	Clinical Hospital Osijek, Osijek, Croatia
Received 16 September 2009 Accepted 16 November 2009
Correspondence to: Branko Vuković, PhD, Department of Physics, University of Osijek, P.O. Box 125, HR-31000 Osijek, Croatia. Phone: +385 31 232 700; Fax: +385 31 232 701. E-mail: branko@fizika.unios.hr
Disclosure: No potential conflicts of interest were disclosed.
Background. Electron linear accelerators in medical radiotherapy have replaced cobalt and caesium sources of radiation. However, medical accelerators with photon energies over 10 MeV generate undesired fast neutron contamination in a therapeutic X-ray photon beam. Photons with energies above 10 MeV can interact with the atomic nucleus of a high-Z material, of which the target and the head of an accelerator consist, and lead to the neutron ejection. Results and conclusions. Our neutron dosimeter, composed of the LR-115 track etch detector and boron foil BN-1 converter, was calibrated on thermal neutrons generated in the nuclear reactor of the Josef Stefan Institute (Slovenia), and applied to dosimetry of undesirable neutrons in photon radiotherapy by the linear accelerator 15 MV Siemens Mevatron. Having considered a high dependence of a cross-section between neutron and boron on neutron energy, and broad neutron spectrum in a photon beam, as well as outside the entrance door to maze of the Mevatron, we developed a method for determining the effective neutron detector response. A neutron dose rate in the photon beam was measured to be 1.96 Sv/h. Outside the Mevatron room the neutron dose rate was 0.62 mSv/h. PACS: 87.52. Ga; 87.53.St; 29.40.Wk.
Key words: electron linear accelerator; photoneutron; track etch detector; neutron dose equivalent
Introduction	and the largest group has a spectrum, which can be
described by a Maxwellian distribution, with the
Nowadays, cobalt and caesium teletherapy machines	most probable energy, typically between 0.4 and 1
in medical radiotherapy are being replaced by linear	MeV; the second group of neutrons is produced by
accelerators.1 The great advantage of this new equip-	direct emission and is somewhat higher in energy
ment is that it has no attached radioactive source,	(up to 10-20% of the total neutrons in general).5 The
which makes them safer from the radiological point	mean energy of the neutron spectrum generated by
of view. However, medical accelerators with photon	the (Y,n) reaction is around a few MeV, but, at the
energies over 10 MeV generate undesired fast neu-	patient plane, neutrons have a more complex dis-
tron contamination in a therapeutic beam. Photons	tribution and a less mean energy. As beam energies
with energies above 10 MeV can interact with the	increase (>10 MeV), an undesirable photoneutron
atomic nucleus of a high-Z material, of which the tar-	dose also increases. Otherwise one can expect, that
get and the head of the accelerator consist, and lead	inside the area treated by linear accelerator, the neu-
to the neutron ejection. Consequently, this can in-	tron dose in a tissue will not exceed 1% of the pho-
crease the patient dose and pose a problem in room	ton dose.6 shielding dosimetry, which is essential for the quality assurance in radiotherapy.2,3
Neutrons are principally produced through giant	Material and methods dipole resonance in a nuclear reaction (g, n) between
photons and target nuclei.4 The giant resonance	The neutron dosimeter consisted of the LR-115
process produces two groups of neutrons; the first	track detector and boron foil BN-1 (Kodak-Pathe,
France) or 10B converter for reaction (n, a); it was calibrated on neutrons generated in the nuclear reactor of the J. Stefan Institute (Slovenia).7 Neutron irradiation was carried out in the thermal column of the TRIGA Mark II reactor where the neutron flux was O = 3.33x105 cm-2 s-1; for irradiation time, t = 240 s, we got the fluence, f = Ot = 7.99x107 cm-2.
The LR-115 detectors, 2x3 cm2, were etched in a 10% NaOH aqueous solution at 60 °C for 120 min and afterwards counted visually using a microscope of (10x16) magnification.
The irradiated neutron track etch detector had a net track density Dt = (63394 - B) @ 63370 cm-2, where the background, B, was 24 cm-2; the standard deviation of the Dt was sDt = 570. The response, rt, of the neutron dosimeter for thermal neutrons was:
10^ E(eV)
FIGURE 1. The total cross-sections of neutrons on boron (s/b) versus neutron energy (E/eV).8
rt = Dt/f = (7.930 ± 0.071) x 10-4
[1]
where the error was given as a standard deviation, srt = sDt/f = 7.1x10-6.
The electron linear accelerator Siemens Mevatron 15 MV has been used as an X-ray radiotherapy source in the Clinical Hospital Osijek. The same accelerator was used to perform experiments for determining a dose equivalent of undesirable photoneutrons by using the neutron track etch detector.
Results and discussion
Linear accelerator neutrons - track detector in the beam
In order to determine a dose equivalent of photoneutrons produced by linear accelerator, operating in a photon production mode at an acceleration potential of 15 MV, we used the LR-114 track etch detector, which was positioned in the beam at 1 m from the accelerator head.
For fast neutrons with higher energy, a detector response should be lower; having considered a total cross-section of neutrons on boron, 10B (n, a) 7Li, depending on neutron energy (s(E), Figure 1)8, as well as a neutron fluence spectrum on energies by the linear accelerator (Figure 2)2, we divided neutron energy Ea (MeV) in the two areas as follows: the first area , 0 < Ea1 < 0.5, and the second one, 0.5 < Ea2 < 7.5.
Afterwards, we estimated the respective mean cross-sections at = 1000 b (for thermal neutrons), aa1 = 1.33 b and aa2 = 0.25 b (for energies Ea1 and Ea2, respectively) from the curve a(E) in Figure 1, and
figure 2. Neutron spectrum produced in 15 MV linear accelerator after crossing the tungsten head; fluence (f/n cm-2) versus neutron energy (E/MeV).2
10^ E(eV)
figure 3. Conversion coefficient (k/pSv cm2) from neutron fluence to personal dose equivalent versus neutron energy (E/eV).
we determined the neutron detector responses for the tow energy areas as follows: at/aa1 = rt/ra1 , st/aa2 = rt/ra2 , where the rt was the response that had already been calculated by using equation [1]. Thus we got ra1 = 1.06 x 10-6 and ra2 = 2.00 x 10-7, with the standard deviations of 9.1x10'9 and 1.8x10'9, respectively, and we were able to determine an average or effective value of the neutron detector response rae, but as a weighted or pondered mean.9
Therefore, we took the surfaces under the neutron fluence spectrum curve (Figure 2; f(E)) for the two energy areas as above and we got the relative surfaces: sa1 = 0.58 and sa2 = 0.42, those had the meaning of the relative frequencies in the calculation of the pondered mean (for instance, s1 is the ratio of the surface under the curve between 0 and 0.5 MeV to the total surface under the curve between 0 and 7.5 MeV). The mean or effective detector response	was:	rae	=	ra1sa1	+	ra2 sa2	=	(6.97	±	0.07)	x10-7.
For a measured detector density Da, the respective fluence was (like in equation [1]): fae = Da /rae. When the neutron fluence is known, a conversion coefficient (k) from neutron fluence to dose equivalent, depending on neutron energy, gives a personal dose equivalent (Ha), as follows:10 Ha= kae fae.
Having considered a great dependence of k on neutron energy (Figure 3), we calculated the average or effective kae for the two energy areas, like above, and we used the same relative frequencies sai (i = 1, 2); taking the average values of ka for the neutron energies Ea1 and Ea2 as ka1 = 200 and ka2 = 430 pSv cm2, respectively, we calculated the effective conversion factor as follows:
kae = ka1sa1 + ka2sa2 = 2964 pSv cm2.
Thus, the measured detector net density Da = (383.1 ± 0.04) cm-2 corresponded to the following dose equivalent:
Ha = kaefae = kae/rae Da= 425.5Da (^Sv) ,	[2]
Ha = (0.163 ± 0.002) Sv
The dose rate was calculated as a ratio of the dose equivalent and exposure time, t, or:
H = H/t , and for t = 5 min, we got the dose rate H = (1.96 ± 0.02) Sv/h.
The measurement errors were determined as variances or standard deviations for track densities in the following way (according to the Poisson distribution):
= s,-
2 + sr2 = Db + B ;
The dose equivalent variance was calculated as total differential of the function of the form like in equitation [2], which led to the following expression:
sH2 = (kD/r2)2sr2 + (k/r)2sD2 .
Linear accelerator neutrons - track detector behind the wall
The neutron track etch detector was positioned outside the entrance door to the accelerator maze. The neutron spectrum in the same position was measured by Schraube et al.11, hereby presented in Figure 4. We used the given neutron spectrum in the procedure for determining a neutron dose equivalent by the track detector, as above. Otherwise, neutron spectrum can vary depending on the wall construction of a room.
We divided neutron energy, Eb, in two areas as follows: the first area, 0 < Eb1 < 100 eV, and the second one, 100 eV < Eb2 < 37'^0 eV. Because the Eb1 was the area of thermal neutrons, we estimated the respective mean cross-sections at = abj = 120 b and ab2 = 14.75 b (for energies Eb1 and Eb2, respectively) from the curve a(E) in Figure 1, and we determined the neutron detector responses for the low energy areas as follows: sb1/ab2 = rb1/rb2 , where rb1 = rt was already assessed response by using equation [1]. Thus we got rb1 = 7.94x10-4 and rb2 = 9.75x10-5, and were able to determine the average or effective value of the neutron detector response rbe for neutrons in the energy area Eb.
As in the previous case, we took the surfaces under the neutron fluence spectrum curve (Figure 4; f(E)) for the two energy areas and we got the relative surfaces: sb1 = 0.18 and sb2 = 0.82, those had the meaning of the relative frequencies in the calculation of the pondered mean. The mean or effective detector response was: rbe = rb1sb1 + rb2 sb2 = 2.2x10'4.
For the measured detector density Db, the respective fluence is (like in equation [1]): fbe = Db /rbe. When the neutron fluence is known, a conversion coefficient (k) from neutron fluence to dose equivalent, depending on neutron energy, gives a personal dose equivalent (Ha), as follows: Hb= kbe fbe
Having considered a great depending k on neutron energy (Figure 3), we calculated the average or effective kbe for the two energy regions, like above, and we used the same relative frequencies sbi (i = 1, 2); taking the average values of kb for the neutron energies Eb1 and Eb2 as kb1 = 11.58 and kb2 = 9 pSv
cm2, respectively, we calculated the effective conversion factor as follows:
kbe = kb1sb1 + kb2sb2 = 9.45 pSv cm2.
Thus, the measured detector net density Db = 1.2 ± 0.01 cm-2 corresponded to the following dose equivalent:
Hb = kbefbe = kbe/rbe Db= 42.9x10-3 Db (^Sv),
Hb = (0.050 ± 0.0006) ^Sv.
The dose rate was calculated as a ratio of the dose equivalent and exposure time of 5 min, and we got the neutron dose rate outside the entrance door to the maze of the 15 MV Mevatron, H b = (0.62 ± 0.007) ^Sv/h. The linac room 1.7 m walls were constructed of barite concrete, with density of 3200 kg/m3.
Although the obtained neutron dose rate outside the accelerator room was 3165 times smaller than the neutron dose rate in the photon beam, the measured dose rate H b was not negligible from the aspect of personal dosimetry.
Some considerations of neutron energy attenuation after crossing the treatment room walls were performed for different concrete barrier thickness and materials. A neutron spectrum attenuation from the 15 MV linear accelerator, after passing a conventional 1 m concrete barrier, with density of 2260 kg/m3, was measured by Facure et al., hereby presented in Figure 5.2
Observing the neutron spectra in Figure 5 and Figure 4, one can notice a broad neutron energy area in Figure 5 (from 0.1 eV to 10 MeV), that contributes to the neutron dose outside the Mevatron treatment room; neutrons that crossed the 1 m concrete barrier had higher energies than those behind the 1.7 m concrete wall (with energy below 1 keV).
In order to compare the neutron detection parameters re and ke, we divided neutron energy Ec (MeV) of the spectrum (Figure 5) in the two areas as follows: first area Ec1 < 2 MeV, and the second one 2 MeV < Ec2 < 7.5 MceV.
Afterwards, we estimated the respective mean cross-sections at = ac1 = 0.619 b (for thermal neutrons, like above), ac2 = 0.203 b (for energies Ec1 , Ec2 , respectively) from the curve a(E) in Figure1, and we determined the neutron detector responses for the three energy areas as follows: sc1/ac2 = rc1/ rc2, where rc1 = rt = 7.94x10'4 was the response that had already been calculated by using the equation [1]. Thus we got rc2 = 2.6 x 10'4 and we were able
FIGURE 4. Neutron spectrum measured at position outside the entrance door to the maze of the 15 MV Mevatron.11
FIGURE 5. Neutron spectrum attenuation from a 15 MV linear accelerator, after concrete barrier of 1 m.
to determine the average or effective value of the neutron detector response rce, but as a weighted or pondered mean.
Therefore, we took the surfaces under the neutron fluence spectrum curve (Figure 5; f(E)) for the two energy regions, like above, and we got the relative surfaces: sc1 = 0.22, sc2 = 0.78, which had the meaning of the relative frequencies in the calculation of the pondered mean. The mean or effective detector response was: rce = rc1sc1 + rc2 sc2 = (3.78 ± 0.03) x10-4.
Having considered a great depending k on neutron energy (Figure 3), we calculated the average or effective kce for the three energy areas, and we used the same relative frequencies sci (i = 1, 2, 3); taking the average values of kc for the neutron energies Ec1 and Ec2 as kc1 = 366 , kc2 = 433 pSv cm2, respectively,
we calculated the effective conversion factor as follows:
kce = kc1sc1 + kc2sc2 = 418 pSv Cm2.
Thus, since kc/rc = 25,75 kb/rb , and according to equation [2], one can see that neutrons of the given distribution in Figure 5 (behind the 1 m concrete barrier) contribute to a neutron dose 26 times more than neutrons from the distribution in Figure 4 (behind the 1.7 m concrete wall).
Conclusions
The neutron dosimeter, consisting of the LR-115 track etch detector and boron foil BN-1, was calibrated on thermal neutrons generated in the nuclear reactor of the J. Stefan Institute (Slovenia), and was applied to dosimetry of undesirable neutrons in photon radiotherapy with the Siemens Mevatron 15 MV electron linear accelerator.
Having considered a broad neutron spectrum of energies in the photon beam and high dependence of the track detector response on neutron energy, we divided the spectrum in the two energy areas, below and over 2 MeV. Afterwards we determined the detector responses (r) for the energy areas using corresponding cross-sections for neutron and boron, and then we calculated the pondered or effective response depending on surfaces under the respective neutron spectrum areas. Using the empirical curve k(E), we performed the similar procedure for determining an effective conversion coefficient (k) from neutron fluence to dose equivalent, depending on neutron energy (E).
The relative measurement errors made by track etching method with the LR-115 detector were about 1%. It is to mention that we took the data from the empirical curves (e.g. a(E), k(E) ) as average values without respective experimental errors.
The measurement of the neutron dose equivalent by the track etch detector, positioned outside the Siemens Mevatron 15 MV room (room wall of 1.7 m, density of 3200 kg/m3), gave the dose rate of 0.62 mSv/h.
Observing the neutron spectrum attenuation from 15 MV Mevatron behind the 1m concrete barrier, one was able to notice that the neutrons (Figure 5) had higher energies and contributed to a neutron dose 26 factor times more than neutrons from the distribution in Figure 4 (behind the 1.7 concrete wall).
References
1.	Strojnik A. In search of the shortest regimen: fractionation of a fully isoef-fective combination of hyperfractionated and hypofractionated treatment. Radiol Oncol 2008; 42: 170-2.
2.	Facure A, Falcio RC, Silva AX, Crispim VR, Vitorelli JC. A study of neutron spectra from medical linear accelerators. Appl Radiat Isotop 2005; 62: 6972.
3.	D Ryan C Rivest, Terence A Riauka, Albert D Murtha, B Gino Fallone. Dosimetric implications of two registration based patient positioning methods 203 in prostate image guided radiation therapy (IGRT). Radiol Oncol 2009; 43: 203-12.
4.	Ongaro C, Zanini A, Nastasi U, Rodenas J, Ottaviano G, Manfredotti C. Analysis of photoneutron spectra produced in medical accelerators. Phys Med Biol 2005; 45: 55-61.
5.	Kräl^k M, Turek K. Characterisation of neutron fields around high-energy x-ray radiotherapy. Radiat Prot Dosimetry 2004; 110: 503-7.
6.	Rivera JC, Falcio RC, deAlmeida CE. The measurement of photoneutron dose in the vicinity of clinical linear accelerators. Radiat Prot Dosimetry 2008; 130: 403-9.
7.	Izerrouken M, Skvarč J, Ilić R. A wide range personnel neutron dosimeter. Radiat Meas 2003; 37: 21-4.
8.	Lilley J. Nuclear Physics. Chichester: John Wiley & Sons; 2001. p. 196.
9.	Vuković B, Radolić V, Lisjak I, Vekić B, Poje M, Planinić J. Some cosmic radiation dose measurements aboard flights connecting Zagreb Airport. Appl Radiat Isotop 2008; 66: 247-51.
10.	Lopez Ponte MA, Castellani CM, Currivan L, et al. A catalogue of dosimeters and dosimetric services within Europe—an update. Radiat Protec Dosim 2004; 112: 45-68.
11.	Schraube H, Kneschaurek P, Schraube G, Wagner FM, Weitzenegger E. Neutron spectra around medical treatment facilities. Nucl Instrum Meth A 2002; 476: 463-7.